a letter from the president dear readers, we are pleased to announce the 13th edition of the columbia undergraduate science journal. cusj’s dual mission is to publish high quality student research in our journals and to promote scientific research among students and introduce students into the scientific research community. in both regards, we have expanded and improved our efforts, and i am incredibly proud of all we’ve accomplished. we grew a lot as an organization, doubling the size of our editorial board and creating a general body for all students at columbia to be able to be involved with the cusj community. general body members participated in a mentorship program, journal club meetings, and other social events throughout the year. we focused our efforts on building community and providing resources for stem students. we hosted events that focused on student and professor research, mental health, resume editing, graduate school, finding summer research opportunities, grant writing, science in the media, and much more. over 30 student researchers presented their original research at our annual symposium and heard from keynote speakers on our faculty advisory board hugh ducklow, gerard parkin, kyle mandli, marko jovanovic, and patricia lindemann about the importance of undergraduate research. on the editorial side, we started a blog for student-written articles that synthesize recent and novel research. we are accepting submissions for blog articles from everyone, so if you are interested please visit the submissions page. both of our journals received more than twice as many submissions as the previous year, and after revamping our editing process, the research that we are publishing is impressive, rigorous, and advanced beyond what we’ve ever seen. the columbia junior science journal published original papers by advanced high school students across the country, and in this edition of the columbia undergraduate science journal, there are articles by undergraduate researchers from some of the top universities in north america. thank you for your support of cusj and the student research community, and we look forward to another great year next year. micah gay president, cusj a letter from the editor-in-chief kenneth pham department of biological sciences, columbia university dear readers, as the academic year once again comes to a close, we are thrilled to present the 2018 columbia undergraduate science journal. this year, we have redoubled our efforts to focus our journal to highlight undergraduate research and scientific thought. to that end, we believe that our new editorial and formatting guidelines have improved the process and product. in addition, submissions to our other publication, the columbia junior science journal amazed us with the passion and excitement of high school students and the quality and incisiveness of their research. we look forward to continuing to improve this journal for future submitters. on campus, here at columbia university, cusj has vigorously continued to create and support an undergraduate research community. to that end, we have worked to broaden our reach and scope. for example, cusj, in co-sponsorship with the columbia university chapter of the american physician scientists association and the charles drew premedical society at columbia university, hosted a panel discussion with md, phd and md/phd students about graduate degrees in the sciences. we also maintained our glucose and caffeine study break series, hosted speakers such as bioethicist, dr. john loike, and had our annual spring symposium. our symposium is an open, accessible forum for a diverse array of researchers to present their work. we also continued our annual “how to get involved with undergraduate research?” event. cusj wants to encourage and support undergraduates in scientific research, especially those who are seeking their first experience. fundamentally, cusj will continue to do so while importantly https://journals.cdrs.columbia.edu/cusj/authors/kenneth-pham/ http://creativecommons.org/licenses/by/4.0/ keeping in mind the pressing inequities in access and opportunities, the powerful force of community and mentorship, and the role science plays in our society. it would be impossible to undertake this goal alone, and i have to thank the staff for their tireless dedication and hard work throughout the year. finally, i wanted to close my letter by sharing a passage from a cell commentary published by the lab of victor ambros, reflecting on their discovery of micrornas, for your consideration. for me, this quotation encapsulates so much about what makes science research so powerful and so meaningful as an intellectual endeavor, and i hope that it will inspire you too. “the intellectual backdrop motivating our effort to clone lin-4 (lee et al., 1993) had nothing to do with questions about noncoding rnas or antisense gene regulation. we were simply curious about an interesting worm mutant, and everything we found out about it was unexpected. we consider ourselves very lucky to happen to have chosen lin-4 to study. in fact, good fortune appeared at many steps before and during our lin-4 project, often through the contributions of other people” (lee et al., 2004). kenneth pham editor-in-chief lee, r., feinbaum, r. & ambros, v. a short history of a short rna. cell 116, s89-92, 81 p following s96 (2004). microsoft word document3 article volume 2, spring 2007 the effects of neuregulin on synapse formation in primary muscle/nerve co-cultures jessica walsh1, mary a mann2, gerald fischbach2 1 columbia college, columbia university, 2 department of pharmacology, columbia university abstract in development, motor neurons innervate maturing myotubes to form the neuromuscular junction (nmj). during this process, the nerve terminal secretes a protein called neuregulin (nrg1). nrg1 acts as a signal which increases the number of acetylcholine receptors (achrs) on the postsynaptic membrane. nrg1 binds to receptors on the surface of the muscle, known as erbb receptors. the binding of nrg1 causes the erbb receptor to auto-phosphorylate (fu, 1999). as a result, there is an increase in transcription of the gene for achrs, integral membrane proteins that respond to the binding of the neurotransmitter acetylcholine. in order to study the effects of nrg1 on early stages of synapse formation, we prepared co-cultures of dissociated muscle cells from postnatal day 1 (p1) mice and neural tube explants from embryonic day 11 (e11) mice. silicone chambers were created as a system for growing dissociated muscle cells and neuronal explants in co-culture (loeb, 1999). erbb inhibitor (pd 158780) was added to chambers prior to the formation of the nmj. after one week the tissue was fixed and stained to visualize the synapses. based on the results of two experiments, the chambers that were not treated with the inhibitor had an average of 10 times more achr (+) contacts. therefore, at the time point studied, it appears that nrg1 signaling through the erbb receptor tyrosine kinases is necessary for the formation of achr (+) receptor contacts at the motor terminal in the post-synaptic membrane. microsoft word cusj_spring_2015_journal.docx !cusj%|%volume%9%|%spring%2015% 8% addicted to adderall osman moneer1, jacqueline lawson2, julia joern3 & theresa mensah4 1department of chemistry, columbia university, new york, ny, usa; 2department of neuroscience, columbia university, new york, ny, usa; 3department of biomedical engineering, columbia university, new york, ny, usa; 4department of biochemistry, columbia university, new york, ny, usa 1 in college student’s everyday struggle of completing assignment after assignment, the au naturel approach doesn’t always cut it. substances are often used to bridge the gap to allow students to meet the ideal level of academic rigor while maintaining a desired lifestyle. in the realm of “study drugs,” caffeine plays a significant role, keeping students awake and alert, but adderall reigns supreme. intended to alleviate difficulties associated with attention deficit hyperactive disorder (adhd), adderall touts a greater focus and increased alertness. but, defining and diagnosing adhd is a fairly difficult and subjective process. the national institute of mental health states that those with adhd may experience symptoms such as “difficulty staying focused and paying attention, difficulty controlling behavior, and hyperactivity (over-activity),” but provides no explanation as to its causes. the lack of understanding surrounding adhd, given its unknown origins and vague symptoms yields issues in diagnosing adhd and determining who actually needs to use the drugs designed for it. because of this confusion, adderall is one of the most commonly used drugs by college students, who often lack even a prescription to justify their use. 30% of students abuse adhd medication such as adderall, and an estimated 81% of students believe adhd medication is not dangerous. students like adderall because it makes them “smarter,” and if negative consequences of the drug are this unclear, who wouldn’t want this additional focus? ethical considerations aside – what are we to make of this? how does adderall interact biochemically and is it detrimental to one’s health? adderall is a chemical compound made up of different portions of amphetamine salts. it is composed of ¼ racemic amphetamine aspartate monohydrate, ¼ dextroamphetamine saccharin, ¼ dextroamphetamine sulfate, and ¼ racemic amphetamine sulfate. amphetamines are psychostimulants and target neurotransmitters of the central nervous system. before we can understand how adderall works in the body, we need to know a few things about neurotransmitters. neurotransmitters are chemical messengers responsible for a range of everyday life functions. popular neurotransmitters include dopamine, serotonin, norepinephrine, epinephrine, histamine, melatonin, and endorphins. we don’t usually consider it, but, in reality, our actions are simply dictated by a series of biochemical interactions going on in our bodies. this gives drugs that can control and alter these interactions significant power in affecting behavior. 2 adderall primarily functions by targeting activity of the neurotransmitters dopamine and norepinephrine in the brain. dopamine is, in part, responsible for pleasures related to motivation, emotional states, and internal reward systems. norepinephrine is related to focus and alertness levels. it’s evident that these neurotransmitters might be implicated in adhd, and increasing amounts of them seems a reasonable approach to counteract issues related to one’s attention. if the main effect of adderall is an increase in attention capacities and greater mental stimulation – it’s not hard to see why it might be taken as a method of efficient studying. greater dopamine concentrations keep you studying longer by tricking your brain to feel as if it is being rewarded for doing an otherwise unpleasant task, and greater norepinephrine concentrations allow you to stay focused on the task at hand. supposing one doesn’t have adhd, is adderall really bad for you? if adderall makes you a better studier, why aren’t we all popping addys and acing our exams? every person has a different biochemical makeup and no drug is without side effects. though it boasts mental alertness, adderall’s common side effects include nausea, loss of appetite, dry mouth, anxiety, agitation, mood changes, sleep problems, headache, dizziness, and more. other side effects may include fast, pounding, or uneven heartbeats, numbness, tingling, and dangerously high blood pressure. side effects vary between people, and there’s an inherent risk in taking a drug without knowledge of how it might affect one’s body. additionally, the issue of addiction to such an enhanced state exists as a significant concern. because dopamine and norepinephrine levels are at such peaks, chronic use of adderall for an individual without adhd may result in addictive use of the substance. adderall usage for those without adhd, though it may likely enhance performance, is not worth it. the benefits don’t outweigh the risks, and there are reasonable alternatives such as exercise! don’t use adderall without a prescription. it just doesn’t add up. ! ! ! meet dr. joachim frank, nobel laureate cusj 2018 ________________________________________________________________________________ our editors vikas chelur cc’21 and maria trifas cc’21 interviewed dr. joachim frank in december 2017. dr. frank is a professor in columbia’s biological sciences department who recently won the nobel prize in chemistry in 2017 with jacques dubochet and richard henderson for their research in cryoelectron microscopy. here is a snippet of the interview where professor frank explains his research at different levels. vc (vikas chelur): could you give us a brief explanation of your research in cryo-electron microscopy, as it pertained to you winning the nobel prize from first how you’d explain it to a toddler? jf (joachim frank): molecules are very tiny, tiny things that we have all over our body. and they bounce against each other and sometimes they recognize each other and say hi. and we want to know exactly what they’re doing and how they do this and how they say hi and why they want to get together and how this all makes our body work. this particular method that we developed makes it possible to see these molecules with very, very super microscopes that are better than light microscopes. mt (maria trifas): what about for a middle schooler? jf: everybody knows that life processes are constituted by interactions of molecules. we have thousands of molecules of different kinds in each cell. and in a normal cell they function properly. in a deceased cell, there are many difficulties in these interactions, problems. and that’s how disease processes are developed. from that it follows that we need to have as complete a knowledge of all our proteins in the cells as possible. the existing technique for determining the structure of molecules to the atomic level (which means the highest resolution level) was by x-ray crystallography. for decades now, x-ray crystallography has been used to determine structures of biomolecules. and there are in fact more than a hundred thousand structures in the public database. they make possible what we call molecular medicine, because medicine nowadays is molecular medicine. it is based on the extensive knowledge of molecular structures. so, this whole database has a huge, huge gap. that gap exists because not all molecules can be crystalized. that’s essentially where electron microscopy jumps in. electron microscopy can visualize molecules in a particular method that the three of us (nobel laureates) got our prize for. it is a method of obtaining the structure of molecules from molecules in solution. and we call it single not because there is only one, but single because they are not attached to each other. so, we can simply look at many, many molecules that are randomly oriented in solution and on that basis, on the basis of the projection, we see the surface structure. there are two important methodologies involved. one of them is the way how all the information from the different projections have to be related to one another in order to find the orientations and then develop the threedimensional visualization. that technique i developed long before the ‘cryo’ part came along. so, i developed this with a very poor contrasting method that existed at the time which is called negative staining – so molecules were badly treated but they were still in some way propped up by the negative stain. and the images received their contrast essentially received their contrast as stain exclusion. so, one didn’t see the molecules, but one saw the stain, and what the molecules excluded. with this kind of visualization, i developed the algorithms, the computational techniques, to get all the way to the 3d structures. when i was just about finished with that, this other technique came along which allowed us to embed molecules in vitreous ice. that’s the contribution by jacques dubochet. the existing attempts to keep molecules in water by quickly freezing the sample in liquid nitrogen – that method doesn’t work, because on the interface between the grid and the liquid nitrogen, gas bubbles form, very quickly. these gas bubbles don’t transmit heat very efficiently, and that allows the ice to form crystals. crystalline ice is dangerous for molecules; it sort of destroys them because it expands. and besides, it also has its own structure which interferes with the structures that we are looking at. so vitreous ice is really an invention that came along in 1980. and then dubochet discovered, along with his mentor, that the formation of crystalline ice can be prevented by substituting ethane for liquid nitrogen. so now we have this little vessel of liquid ethane suspended in a bath of liquid nitrogen. and when the grid with the sample on it is plunged in there, then the ice that is formed is glasslike, and hence vitreous. so it is a very simple concept, but it is an ingenious idea, and all of a sudden, there was a way of freezing the molecules as if they were in liquid water, they stayed in liquid water. meet dr. joachim frank, nobel laureate 20 microsoft word cusj 35.docx ________________________________________________________________________________ cusj 2019 __________________________________________________________________________________________ 35 following our annual cusj symposium this year, we were extremely proud to announce three honorable mentions — nikita mullick, justin oh, and niyasha wijedasa! these three students impressed us with not only their research, but also their ability to effectively convey their complex projects in a coherent and digestible way. congratulations once again to the cusj symposium honorable mentionees! read about them below: nikita mullick nikita mullick is a third-year undergraduate biology b.s. student at emory university. she works in the jacob lab at emory vaccine center. she attended high school at university school of milwaukee in wisconsin where she did research in the benjamin lab at the cardiovascular research center of medical college of wisconsin and had a three-year research internship at biosource inc. in the jacob lab, nikita works to discover novel antiviral agents to fill gaps where vaccines and drugs are ineffective or nonexistent. host defense peptides (hdp), produced innately by animals, can target common motifs on pathogens to help protect the animal from infection and could be harnessed for use against viruses. the aim of her project is to find hdps that have antiviral activities to improve the breadth of peptide-based drugs. in this study, nine peptides that have sequence homology to a known king cobra hdp were discovered through genomic scanning of various species. toxicity against human erythrocytes and antiviral activity against two influenza viruses were analyzed. three new snake peptides and one wallaby peptide were found to be non-toxic and have shown significant antiviral activity in vitro. these candidates are being further investigated for use as antiviral therapeutics. justin oh justin oh is a junior in the academy for advancement of science and technology at bergen county academies (bca) in hackensack, new jersey. apart from doing research, justin is the founder and organizer of bca flash, an annual event where bca students teach steam topics to approximately 300 middle schoolers, copresident of his school’s model united nations club, and a scholastic art winning photographer. in terms of his research, he began to focus on ovarian cancer, one of the deadliest gynecologic cancers in women, beginning march 2017. the treatment used in his project is hydrangea root (hr), which is known to naturally treat urinary tract infections like kidney stones made of ca2+ ions. skov3 cells, the human ovarian cancer cell line, depend on ca2+ ions to thrive in the body. the hr can cause a decrease in epidermal growth factor-induced ca2+ influx, which originates from the trpc3 protein expression in skov3 cells. the reduction in ca2+ ions will cause the skov3 cells to enter autophagy due to the absence of calcium ion transfer and a decrease in atp levels. thus, the skov3 cells will eventually die in the presence of hr. by conducting this research, justin hopes to create an effective, natural treatment that can inhibit proliferation of ovarian cancer cells. niyasha wijedasa niyasha wijedasa is a rising senior on the pre-dental track. although she has just been introduced to research this past semester, she became so passionate about her work, making the countless hours spent in the lab enjoyable. her love for dentistry prompted her to pursue a research topic in dental tissue engineering. to combat the issue of enamel erosion, which is irreversible, she constructed a scaffold which would aid in the cellular proliferation, organization, and differentiation of primary gingival fibroblast cells in hopes that it could be used for further research in enamel regeneration. to devise this scaffold, hydroxyapatite, the major constituent of enamel, was extracted from fish scales. then it was integrated with polygalacturonic acid, a pectin derivative found in plant cell walls, to form a gelatinous matrix similar to the extracellular matrix of natural dental enamel. nanofibers were incorporated to serve as the backbone of the scaffold and increase biocompatability. lastly an amelogenin protein, responsible for the biomineralization of enamel, was combined with the scaffold to further mimic the properties of the dental enamel matrix. once the scaffold was complete, their biocompatibility was tested in conjunction with the primary gingival fibroblasts. figure 1. pictured are the three 2019 cusj research symposium honorable mentions – congratulations! on the 2019 cusj research symposium honorable mentions microsoft word 42-160-1-ed.docx 1 genome analysis cfhmm: heterogeneous tumor cnv classification by hidden markov aleksandar obradovic1,*, hongjian qi2 1department of computer science, columbia university azo2104@columbia.edu 2department of computer science, columbia university hq2130@columbia.edu abstract summary: we here develop and implement a clonal fraction hidden markov model (cfhmm), to leverage positional information in classifying tumor cnvs and their corresponding clonal fraction from log-ratio-normalized tumor/normal sequencing data. in simulated data, this approach shows accurate calling of cnvs for high-fraction mutations, and improvement in calling over a naïve clustering benchmark across the board, as well as useful purity estimation for dominant clones. availability and implementation: source code and documentation is freely available at https://github.com/7lagrange/fcnv implemented in r, with all major operating systems supported. contact: azo2104@columbia.edu supplementary information: additional tables and figures available at https://docs.google.com/document/d/1ohbjwaz20jxx3tc64basuzp wmpne_ybfwfmju9jb0mu/edit?usp=sharing 1 introduction copy number variations (cnvs) are duplications or deletions of genome segments, of length greater than one kilobase by convention, which occur normally in the genome but have also been highly implicated in tumor genomes (zhang). it is therefore important to accurately classify cnvs in tumor genome data. this can be done by modeling next-generation sequencing data, where a natural model to apply is the hidden markov model, with transition matrix of copy number or cnv states and emission of normalized read-counts (zhao). however, admixture of normal cells in a tumor sample (gusnanto) and heterogeneity of distinct clones within a tumor (oesper) complicate analysis, such that apparent copynumber for any given genome region may appear fractional and be misclassified. this, in addition to the loss of prior location-specific variability information that can occur after tumor/normal normalization, is an issue that must be addressed for improved application of a hidden markov approach to cnv classification. in our cfhmm model, we implement a 15-state model, with a commonly used 5-copy-number (0,1,2,3,4) set for tumor and extremes-removed 3-copy-number (1,2,3) set for normal cells, where a normal diploid state of 2 is the most common. we derive an average purity prior from raw log-ratios, from which a set of 15 strong emission distribution priors are constructed that preserve variability information for a given tumor/normal state-pair. *to whom correspondence should be addressed. cfhmm runs modified unsupervised viterbi training on the data to give posterior state classifications that, if accurate, may also provide clonal fraction information for each cnv mutation. 2 methods we evaluate the accuracy of our model on simulated wholegenome-sequencing data for which the hidden tumor and normal copy-number states are known. the tumor and normal genomes are segmented into kilobase-length bins for which copy number is generated by markov chain with adjustable parameters favoring remaining at the same copy number from one bin to the next and providing equiprobable transition to any other copy number. 60 million reads, corresponding to 10x deep sequencing in (myers) are randomly distributed into these bins with probability weighted by copy number. clonal fraction k in the tissue for a given bin is randomly selected from a parameter list of options, where we ran several tests with two-tumor-clone data and one with three-tumorclone data. log-normalization proceeds from read-count data as: (1) where, conversely: (2) the purity prior k for input into cfhmm is taken as the mean of values from equation 2, assuming normal state 2 and using only extreme log-ratios over the equation 1 values for pure (k=1) tumor state 3, or under the threshold for pure tumor state 1. with this prior, since the read-counts are poisson distributed, the lognormalized emission data can be normally approximated in a way that preserves variability information dependent on location along the genome as: (3) where p1 and p2 are tumor and normal state, respectively, and n is the number of reads. from here, a hidden markov model is created with a 15-state transition matrix and modified for continuousemission such that the emission matrix contains a mean and variance for each of the 15-states. the viterbi algorithm is then: (4) with probability of state m at position n captured in xm(n), state transition probability captured in t, and log-ratio emission distributions captured with mean and standard deviation in e. this is applied for classification in unsupervised training with a modified version of the standard iterative viterbi algorithm (durbin), where the maximization step in em for the continuous-emission-matrix e uses bayesian update (lynch) on the normal distributions: a.obradovic, h.qi 2 (5) after 15-state classification, a posterior purity estimate for each bin where tumor state is not equal to normal state can be derived from equation 2, and this distribution can be plotted to visualize clonal fraction, as well as k-means-clustered to group mutations belonging to the same clone sub-population. this is compared in simulated data to the known fractions and the accuracy is evaluated. for cnv-state classification, the 15-state model is compressed by grouping of equivalent states into a 6-state model (see table 1). table 1. state compression table. state 1 2 3 4 5 6 t/n 0/1, 0/2, 0/3 1/2, 1/3, 2/3 1/1, 2/2, 3/3, 4/4 3/2, 4/3 4/2, 2/1 3/1, 4/1 15-state tumor-normal model is reduced by posterior resolution of ambiguities to 6 distinct cnv states. 1-full deletion, 2-partial deletion, 3-normal, 4partial amplification, 5-homozygous amplification, 6-large amplification. as a benchmark for accuracy of cnv state classification, a naïve thresholding method is used, where thresholds are drawn at the log-ratio values from equation 1 with purity prior k and t/n states 0/2, 2/3, 4/3, 4/2, and 3/1, and a genome location is classified into one of the six compressed states as whatever interval of these thresholds its log-ratio falls inside of. there is no equivalent benchmark for posterior purity from the hmm model, as this is the primary innovation of our method, so it is evaluated on its own. 3 results simulated data was run for ten matched tumor-normal genomes, with all parameters but purity kept constant at default value. the first trial was homogeneous tumor admixed with normal cells at 0.9 clonal fraction. each of the next eight trials contained two equiprobable purity values, with a dominant 0.9 fraction clone and a secondary subset of fraction ranging from 0.85 to 0.5 at intervals of 0.05. a final heterogeneous tumor trial was run with three equiprobable mutation fractions of 0.9, 0.7, and 0.5. in terms of state classification accuracy, each trial showed correct classification for clones represented in 0.7 or higher proportion of the tumor sample with accuracy over 99.9%. this is significantly higher than the approximately 70% accuracy seen in the naïve thresholding benchmark. both methods, however, lose accuracy for rarer cnv mutations, although our cfhmm method remains more accurate than the benchmark across the board (see fig. 1). accuracy of posterior purity estimates (+/ 0.01 of true value) was also better for high-fraction cnvs, showing around 80% accuracy in all trials for the dominant clone and accuracy decreasing to 56% for 0.5 purity. posterior purity estimates show clear peaks at the true purities for k greater than or equal to 0.7, usefully depicting tumor heterogeneity, but retaining only the dominant-clone peaks for tumors with rarer admixed mutations, such as the 3-state trial tumor, where states 0.9 and 0.7 were well-described and classified with high accuracy, but 0.5 was not (see table 2). table 2. three-state 0.9, 0.7, 0.5 purity trial purity states hmm state accuracy thresholding state accuracy hmm purity accuracy .9 0.9995894 0.6706574 0.8174977 .7 0.9992959 0.5384039 0.7150379 .5 0.649813 0.3955961 0.4992132 accuracy is shown per purity state for the hmm classification, thresholding classification, and hmm-derived purity. areas for future work include the implementation of baum-welch rather than viterbi training for hmm learning (durbin), in order to ensure globally optimal solution, as well as the application of the algorithm to real matched cancer sequencing datasets for further validation. in addition, the breakdown of the method for lowfraction cnvs is a limitation than should be addressed, possibly by iterative exclusion of classified dominant mutation sites, so that the new prior on subsequent runs of the algorithm is pulled toward lower purity, so long as the limitation for the markov assumption presented by resulting holes in the data can be addressed. acknowledgements we wish to thank dr. itsik pe’er and the students of cbmfw4761 for providing feedback and advisement on project design. references durbin, richard, ed. biological sequence analysis: probabilistic models of proteins and nucleic acids. cambridge university press, 1998. gusnanto, arief, et al. "correcting for cancer genome size and tumour cell content enables better estimation of copy number alterations from next-generation sequence data." bioinformatics 28.1 (2012): 40-47. lynch, scott m. introduction to applied bayesian statistics and estimation for social scientists. springer science & business media, 2007. myers, gene. "whole-genome dna sequencing." computing in science and engineering 1.3 (1999): 33-43 oesper, layla, gryte satas, and benjamin j. raphael. "quantifying tumor heterogeneity in whole-genome and whole-exome sequencing data."bioinformatics 30.24 (2014): 3532-3540. zhang, nancy r. "dna copy number profiling in normal and tumor genomes." frontiers in computational and systems biology. springer london, 2010. 259-281. zhao, min, et al. "computational tools for copy number variation (cnv) detection using next-generation sequencing data: features and perspectives."bmc bioinformatics 14.suppl 11 (2013): s1. fig 1. cfhmm v threshold accuracy for dominant and rare clones. microsoft word cusj_spring_2015_journal.docx !cusj%|%volume%9%|%spring%2015% 10% mirror panel edge sensors for the schwarzschild-couder telescope jonathan sun1, brian humensky2 & daniel nieto castano2 1 department of computer science, columbia university, new york, ny, usa 2 department of physics, columbia university, new york, ny, usa 1 gamma-ray astrophysics and motivation: gamma-ray astrophysics is an exciting field of high-energy particle physics that has expanded dramatically in the past decade. the science of gamma-ray cosmology presents a way of looking at the most interesting parts of the universe. very high energetic particles in space indicate the existence of exotic and extreme physical conditions. looking for gammaray emissions allows us to map the universe, discovering high magnetic/electric fields, shock waves, and cataclysmic explosions. these emissions offer the only direct probe of the extreme conditions in these exciting phenomena [2]. figure 1: gamma-ray showers and cherenkov light very energetic particles are extremely difficult to detect. charged particles are affected by the incredibly strong magnetic fields of deep space, so when they reach earth it is impossible to deduce the origins of the particle. gamma rays are unaffected by these magnetic fields. a gamma ray is a packet of electromagnetic energy photons. they are the most energetic photons in the electromagnetic spectrum, and they are emitted from the nuclei of unstable radioactive atoms. they travel in straight lines allowing astronomical instruments to determine the origins of the high-energy particles. however, there is an inverse relationship between particle energy and particle flux; higher energies result in very little flux. space-based instruments are expensive and too small to properly detect higher energy particles. modern projects use ground-based instruments designed specifically to detect gamma rays. it does so through secondary radiation detection. the process is visualized in figure 1: a source emits a gamma ray. the gamma ray interacts with the nitrogen in the atmosphere, producing a shower of electrons, positrons, and other particles as the gamma ray decays. this 2 particle shower is seen as a radiation called “cherenkov light.” large optical reflectors in the ground-based telescope then image the cherenkov light onto a photomultiplier tube camera in the instrument. multiple telescopes imaging the same cherenkov radiation allow us to triangulate and pinpoint the origin of the particle shower with a 3d reconstruction of the shower in the sky [1]. development of the mpes: overview: the new cherenkov telescope array (cta) is a next generation global-project ground-based observatory for high-energy (30 gev—200tev) gamma ray astronomy. the new schwarzschild-couder telescopes introduce a novel imaging atmospheric cherenkov telescope design featuring a schwarzschild-couder aplanatic two-mirror optical system (fig 2). an integration of several tens of these new telescopes will yield an astronomical instrument accommodating a wider field-of-view, significantly improved angular imaging resolution, and expanded energy detection range. it is more economic and yields better image quality and background rejection. the new reduced plate scale design also makes it compatible with highly integrated cameras assembled from silicon photo multipliers. figure 2: the new sct design each sct plate is composed of several individual mirror panels. each mirror panel needs to be precisely adjusted and positioned in order for accurate viewings of the sky. traditionally, ground-based telescopes needed to be manually adjusted. the new telescope array plan renders this process extremely tedious. thus, the project of developing edge sensors for the mirror panels began. the general concept is seen in figure 3. three edge sensors are placed between all adjacent plates for six degrees of freedom, allowing for location knowledge of a singular mirror panel in 3d space. the edge sensors will measure the relative displacements between mirror panels as an input for the panel-to-panel alignment system. when the edge sensors give information to the alignment system that indicates a mirror panel is displaced, !cusj%|%volume%9%|%spring%2015% 11% 1 the system will command the actuators underneath that mirror panel to correct its position accordingly. the edge sensors are required for a positional resolution of just a few µm over an operational area of around 10mm by 10 mm. ! ! figure 3: mpes design component ! edge sensor design: the sensor design is very basic. a single sensor consists of a photo sensor and a corresponding light source fundamentally. the sensor presents a single optical axis defined by a laser beam, which is orthogonal to the photo sensor plane. the primary components of the edge sensor consist of an economic usb webcam and an economic laser diode module. these components are both adequate for precise measurements and economically efficient. the laser diode emits its light at the webcam. the webcam will record an image of the laser dot (fig. 4). the positioning of the overall mirror panel is thus calculated and measured via the position of the laser dot in the webcam image; a displaced dot infers a displaced mirror panel. ! figure 4: sample laser image positioned in-between the photo sensor and the light source are two components to assist the accuracy and precision of the sensor module. the first component is a collimator. it acts as a diaphragm and allows the laser to pass through a small pinhole (300 µm diameter) that narrows the laser beam. the second component is an opal glass screen to! diffuse the laser light just before the webcam.! 2 webcam testing and disassembly: the project ordered a total of 380 webcams to be used for the edge sensors. like the laser diodes, the quality of the webcams was expected to vary. consequently, the webcams needed to be opened and tested for resolution and image quality. the defective cameras would be returned. the webcams were tested using a resolution test chart (fig. 5). next, the cameras needed to be disassembled to install them in custom sensor housings in the sensor unit. figure 5: resolution test chart webcam image distortion: an issue became evident as the webcams were tested. many of the cameras exhibited a “fishbowl,” or distortion affect where the captured image was slightly warped. the severity of the warping varied from camera to camera, with some exhibiting barely any warping and some exhibiting significant warping. an experiment was performed to test this. three cameras were used, one exhibiting negligible warping, one exhibiting some warping, and one exhibiting severe warping. the lenses were permuted between the camera circuit bodies to test if the lenses caused the distortions (fig. 6). the results are surprising. the distortions did not follow the lenses as predicted. in figure 6, a similar distortion pattern remains consistent between the lens permutations on body c. this suggests that the photo sensor holder causes the distortion instead of the lens. environmental stress tests: the final project to be discussed is the project regarding environmental variables and their effects on edge sensor stability and performance. the primary environmental factor of concern is temperature–it seems to have the most significant effect on laser intensity. an alteration of laser intensity can have damaging impacts on the sensor accuracy. if the intensity is too high and the webcam exposure level does not account for this, the image becomes extremely oversaturated. the center is impossible to determine. if the image is too underexposed, it becomes very difficult to determine the center of the laser as well. the ultimate purpose is to visualize the relationship between the environmental stresses and image intensity. the project began with using data from a prototype edge sensor tested in the field last summer and an edge sensor tested in a laboratory. laser images taken and regular intervals over a period of several weeks were compiled with <.log> files detailing the time, temperature, and humidity at the! !cusj%|%volume%9%|%spring%2015% 12% 1 moment of each image capture. all together, this served as a database for the next step. a perl script was written to scour the database of images and environmental variables. this script needed to account for errors in the database such as missing images, inconsistent time intervals, missing times and environmental information, inconsistent orderings of images and scalar data, and images that were completely unusable. the script ran the images through another pre-constructed c++ function called “snapshot” that scanned the image and produced the image intensity and the x and y centroid values. this information was compiled with environmental information to produce a single <.log> file containing all of the information. figure 8: “extreme” stress tests data 2 a subsequent r00t macro (an extension of c++) was written to handle the <.log> file. the macro gathered all the data and produced graphs that mapped the relationships between environmental variables and sensor performance. the results are shown in figures 7 and 8. [editor’s note: figure 7 can be found on the next page.] stress tests analysis: the results are very interesting and inconclusive. the top rows of graphs (in both figures) show the data after the laser image is converted to grey-scale. the bottom set of graphs comes from the original colored images. the first two columns of graphs show the x-centroid position and y-centroid position respectively. each dot is an image taken at a unique time. the x-centroid is much less consistent than the y-centroid position. there also appears to be two ‘bars.’ the investigation into this phenomenon is ongoing. originally, it was hypothesized from preliminary testing that there would be an inverse, linear relationship between temperature and laser intensity. as the temperature became colder, the intensity was increase. however, the outdoor stress test results show no clear relationship in this manner. the spray of intensity levels is extreme and strange. the lab ‘extreme’ stress tests show a similar inconclusive result. the outlying and severely varied intensity values may have been due to faulty images in the database. some of the images appeared to be distorted in terms of color and intensity. the analysis is ongoing in the cta project. figure 6: lens permutation tests. images highlighted with a red border indicate lenses paired with their original bodies. !cusj%|%volume%9%|%spring%2015% 13% 1 acknowledgments: this research was supported by columbia university in the city of new york. dr. brian humensky and dr. daniel nieto of the nevis laboratories guided this research. the authors would like to thank the primary author’s research advisor dr. gasperov. references: [1] g humensky, brian. “some helpful slides.” [2] <http://veritas.sao.arizona.edu/about-veritas-mainmenu81/atmospheric-cherenkov-technique-and-veritastechnologies-mainmenu-87> [3] “sct-opt/mec design report for sct review.” man-po/120920. [4] arxiv:1307.4072 [astro-ph.im] [5] nieto, daniel. “development of the prototype sct mirror panel edge sensors.” cta consortium meeting chicago / may 28, 2013. [6] “resolution test chart.” <http://www.graphics.cornell.edu/~westin/misc/iso_12 233-reschart.pdf>. 2 figure 7: outdoor stress tests data columbia undergraduate science journal open-access publication | http://cusj.columbia.edu spring 2014 | volume 8 31 cherenkov telescope array mirror alignment report francis toriello1* 1the veritas group, department of physics, columbia university, new york, ny, usa introduction the cherenkov telescope array (cta) is a groundbased observatory for gamma rays with very high energy characteristics within the range of 10 gev to 100 tev. it is currently still under development. as the title of the project suggests, it takes advantage of the cherenkov radiation that is produced by the gamma ray’s interaction with the earth’s upper atmosphere, similar to the guiding physical principles of veritas. the primary differences are that cta will feature more telescopes and possess a larger energy range. cherenkov radiation will be collected by an array of dozens of optical telescopes of various sizes. these optical telescopes will be composed of hexagonal mirrors that must be aligned relative to each other so as to represent the shape of the telescope as a whole properly. it was in this area of research that my efforts were focused. the most recent proposed system with which to align the mirrors consists primarily of a laser and a webcam. sixty to three hundred snapshots are taken of the laser and the position of the laser’s centroid in each frame is measured and analyzed by a custom-written computer program executed via root. one to ten microns was the ideal resolution that was required of the system to meet the needs of this project. since the position of the laser’s centroid was given by pixel number, it was necessary to determine the ratio of pixels to millimeters via another root program described in more detail on the following pages. it was also necessary to ensure by using an opencv analysis program that the pixels were not saturated. as will be shown, the consequences of saturation are bothersome edge effects that produce erroneous results. materials and methods apparatus (basic setup) 1. standard webcams (various models) 2. voltage supply 3. laser diode 4. l.e.d. 5. iris 6. calorimeter 7. diffuser 8. actuators results the first task at hand was to determine if the pixels were saturating, since this was the prime suspect of the following odd results: copyright: © 2014 the trustees of columbia university, columbia university libraries, some rights reserved, toriello. received 1/1/2014. accepted 3/14/2014. published 3/14/14. *to whom correspondence should be addressed: francis toriello, columbia university, new york ny 10027, email: fet2105@columbia.edu. physics columbia undergraduate science journal open-access publication | http://cusj.columbia.edu spring 2014 | volume 8 32 the data shown above represent a series of three hundred successive photos of a static laser with an applied voltage of 4v and a calorimeter setting of 2mm. as may be seen, the centroid does not appear to be very well behaved. in fact, it appears as though it moved dramatically in the first 100 frames as may be seen in the upper right hand graph. as previously mentioned, the prime suspect for this odd result was pixel saturation. below is one of the photographic images of this 4v 2mm laser point rendered by the webcam itself. the pure white center represents clear evidence of saturation. the more disconcerting feature of this, however, is the ring of reflected light visible around the edge of the iris. these are edge effects due to saturation that cause the centroid to appear to move when it has, in fact, not done so. this problem was made most apparent after writing an opencv executed program that analyzed this very image and produced a histogram consisting of the amount of pixels versus the intensity values they maintain (see image below): as can be seen, there is an initial spike at zero intensity. this spike represents the pixels that make up the area outside the iris while the spike all the way to the right, at the 255 mark, represents the saturated pixels that can be found at the center and throughout the image. the ideal (and necessary) situation is to have no pixels saturating or otherwise stated at the 255 intensity mark in the above graph. at the time we were using a calorimeter that possessed a smallest possible aperture of 0.1mm. we found that, even with this smallest aperture and various voltage values, there was still saturation that, in effect, would produce erroneous results as shown previously. we were finally able to eliminate saturation by using a different calorimeter that possessed a smallest possible aperture of zero millimeters. we used the very same 4v laser except with an aperture of less than 0.1mm, achieving successful results shown below. physics columbia undergraduate science journal open-access publication | http://cusj.columbia.edu spring 2014 | volume 8 33 while the exact values accompanying the above image are admittedly difficult to decipher, it is clearly apparent in the first two upper left hand graphs that the laser’s centroid is much more defined and better behaved. in fact, the sigma value for the x value of centroid is 0.06657+/0.00215 pixels while the sigma value for the y value is 0.1029+/-0.0040 pixels. this discrepancy is bothersome since both the x and y value should be the same. however, this could be due to a slight tilt of the camera that would affect the geometry in such a way as to elongate the spot subtly in the y direction. the x sigma value, however, is auspicious as it falls within the few micron range that is required of the position resolution we are seeking. the next task involved determining the pixel to millimeter ratio so we could obtain a better sense of the positional resolution provided by the sigma values previously mentioned and whether or not the pixels are square. this required performing two linear scans in which the camera is moved by the actuators in increments of .5mm in the x direction and then once again in the y direction. the slope of the fitted line should yield the amount of pixels per millimeter and both slopes should be equal to one another if the pixels are indeed square (see images below). as may be seen, both linear scans produce an approximate slope of 48.31 pixels/mm showing that the pixels are square and that our x sigma value of 0.06657 pixels translates to approximately one micron while our y sigma value of 0.1029 pixels translates to approximately two microns, both well within the range of our desired positional resolution. the second graph of each scan (position vs. deviation from fit) gives us an idea of the mechanical uncertainty of the apparatus; that is the uncertainty introduced once the apparatus is set in motion. we see that in both physics columbia undergraduate science journal open-access publication | http://cusj.columbia.edu spring 2014 | volume 8 34 scans there is a spread amongst the values of approximately .5 pixels or 10 microns. of course, this is subject to change as the setup we were using is far from the configuration that will eventually be used in the field. once the pixel to millimeter ratio was determined, the next task required us to confirm the apparent positional resolutions provided by the sigma values. this was achieved by performing a scan in the x direction similar to that of the previous linear scan. however, the difference this time was that, instead of keeping constant the increments by which the camera moved, we instead halved the distance every time. if our positional resolution was indeed approximately one micron then, once the increments reach one micron and below, the program should fail to recognize any change in position (see image below). as may be seen in the second position vs. deviation from the fit graph, there is a point very close to two millimeters where we reach the limit of our positional resolution. that position is actually 1.999023188 mm, exactly the point where the change in position dips below one micron (9.765625 x 10^-4 mm) as was apparent in the table of data used to create these graphs. therefore these results show that the positional resolution achieved was, in fact, approximately one micron (with regards to the x value of the centroid). to get a better idea of the mechanical uncertainty, a series of diagonal scans were performed. just as the name implies, these were contemporaneous scans in both the x and y direction. we should see the same mechanical uncertainty of approximately 10 microns or .5 pixels that had been reflected in the linear scans. y-centroid behavior: x-centroid behavior: notice that, similar to the linear scan in the x-direction, the x-centroid behavior yields a mechanical uncertainty of approximately .5 pixels or 10 microns (given by the spread of points in the position vs. deviation from fit graph) while the physics columbia undergraduate science journal open-access publication | http://cusj.columbia.edu spring 2014 | volume 8 35 y-centroid behavior yields a mechanical uncertainty of approximately 3.5 pixels or 73 microns. this discrepancy is similar to the difference between the x sigma and the y sigma mentioned earlier. it was determined that this inconsistency could also be due to some very slight tilt of the camera that may affect the geometry of the how the laser beam impinges on the camera’s lens. the next task required testing other webcams of similar design in order to evaluate the robustness of the current configuration of the apparatus. for the sake of brevity i have excluded these results as they were very similar to previous ones. the last task we completed before the end of the semester was the evaluation of an alternate light source, a light emitting diode. due to the fact that the light emitted from an led is considerably more spread and less concentrated, it was necessary to build a makeshift housing unit in which we passed the light though two custommade calorimeters, both of which, of course, possessed an aperture of less than 0.1mm. we proceeded to taking sixty successive images of the led laser point and analyzing them in much the same fashion as we had done for the three hundred successive photos of the 4v laser diode. the results are no doubt promising as, much like the laser diode, a well-defined centroid is reflected in the analysis as can be seen in the first two upper left hand graphs. we get an xsigma value of 0.0666 +/0.0161 pixels or once again approximately one micron and a y-sigma value of 0.09032 +/0.01511 pixels or once again approximately two microns. discussion in our final analysis, the use of webcams while promising is not without its discrepancies, such as the difference between x and y positional resolutions given by their respective sigma values, and it reflected some mechanical uncertainties. it is of interest to me why, when the linear scans were not diagonal but done separately and dedicated to one direction at a time(x or y), the same mechanical uncertainty of about 10 microns was reflected by both the x and y scans while this did not occur for the diagonal scans. if the geometrical argument that i have proposed throughout this report is correct, then perhaps some corrective measure may be investigated during further research anticipated to be done in months to come. with regard to most recent developments concerning the leds, it would be interesting to perform some diagonal and linear scans to not only gain a more advanced understanding of how we may use these light sources, but also to see if these discrepancies persist, and in effect learn more about them and the subtleties of the system as a whole. physics microsoft word 102-327-1-sm.docx abstract—monitoring and diagnosis of power as well as distribution transformer is the most important process for extending the life of transformer.a transformer monitoring system is already developed by siemens which is based on scada system. but that system can’t be used in small scale industry because most of the industry lacks in automation. this paper presents the details of a monitoring system which is a standalone device equipped on every transformer individually and this device is able to detect phase failure, unbalanced voltage, and earth fault. apart from these, the device can also monitor the oil level and oil temperature. the main aim of research is to develop this product and make available on economic rates. index terms—power transformer, embedded systems, transformer protection i. introduction ower transformers are the important part of power system which are used to step up and down the power to make usable for electricity consumers. transformers are vulnerable to failure due to external fault or internal fault or general ageing of parts which can lead to severe damage and results in replacement of the parts due to which the consumer suffers from longer power outage as well as high replacement of the part; it can also cause explosion and fire which is threatening to lives and property. at present following methods are used for fault detection of transformer 1) frequency response analysis (fra) 2) dissolved gas analysis (dga) however, this method requires disconnecting the system and isolating the transformer which leads to interruption of supply. while dga is instead an online monitoring method, it has limitations of detecting faults which associated with partial electrical discharge [1]. there are different kinds of faults in a transformer; the internal fault is divided into two i.e. short circuit faults and incipient faults. the internal short circuit faults are caused by turn-toturn shorting or turn-to-earth shorting in the windings of transformer. the incipient fault is developed over time as the transformer insulation deteriorates. the factors affecting transformer failure and deterioration are the environment in which the transformer is being operated, maximum load current, lightening short circuits and switching surges. [2] in this paper the detailed working of the fault detection system is explained. this system continuously monitors the transformer in real time and compares the output of transformer with the reference values i.e. of a healthy transformer. this device can monitor the voltage values and determine whether it’s a phase failure or earth fault. apart from this, it can also monitor temperature of oil, moisture level and oil level of transformer. at present the problem within india is regarding to the monitoring and diagnosis of transformer. regular monitoring is not carried out by the employees. so to eradicate this problem an automatic and efficient transformer monitoring system is employed. this will help in reduction of duration of power failure. it will also help in overcoming future faults. ii. research objectives • to design a system this can compare the output of transformer with the reference output and conclude it with healthiness or unhealthiness of transformer. • to monitor the oil level of transformer in conservator. • to monitor the temperature of oil. iii. design methodology the system consists of 1). power supply 2). potential transformer 3). microcontroller avr fig. 1block diagram of system 4.1 power supply unit a power supply circuit consisting of a transformer along with rectifier diodes, where the leads of the secondary winding are connected to first terminals of rectifier diodes whose second terminals are connected to output of the power supply circuit. [3]there are basically two types of power supply i.e. linear and switched mode power supply. we are using a linear power supply here. power transformer monitoring system shreyansh likhar chameli devi group of institutions, electrical and electronics engineering p fig. 2linear power supply circuit 4.2 potential transformer p.t is an instrument transformer which usually transforms high voltages to a value easy to handle for relays and other instruments. it also insulates the metering circuit from primary high voltages. instrument transformers are special kind of transformers used to measure currents and voltages. fig. 3arrangement of potential transformer in this system three potential transformers are connected on low voltage side of transformer each is connected on phase and neutral. 4.3 temperature sensor a temperature sensor is a device that collects data which relates temperature from an object or environment and converts it to a form that can be understood by device. these sensors come in many types and are used for a wide range of applications.[4] we are using tmp36 temperature sensor which is a digital temperature sensor the tmp36 is low power consumption, precise temperature sensor. the celsius temperature is linearly proportional to voltage output provided by the sensor. it doesn’t require any calibration. its accuracy is ±1°c at +25°c and ±2°c over the −40°c to +125°c temperature. fig.4temperature sensor tmp36 4.4 microcontroller for carrying out a specific function microcontrollers are used.a microcontroller is a device with processor and amemory that can be used as an embedded system. most consumer products or machinery including mobile phones, automobiles and household appliances or computersare equipped with a programmable microcontroller.[5] fig.5atmel avr pin outs 4.5 conservator oil level indicator 4.5.1 magnetic oil gauge magnetic oil gauge is used indicate the oil level in conservator of power transformer. this is a mechanical device. magnetic oil gauge consists of mainly; float, bevel gear arrangement, indicating dial and terminals. we are using two ways for indicating oil level on device 1). connecting mog with device – there are three terminals on mog; two are phases and one is neutral. on one of the phase we are supplying single phase voltage to power the mog. the other phase along with the common neutral is connected to the relay of the device. the working of this process is such that a mog consists of magnetic switch which is eventually activated as the level of float falls which signals the relay to activate the alarm. fig.6connection diagram of mog 2). modifications in mog and conservatorthis method give the real time level of oil. we are using laser and ldr sensors to detect the level of oil and further which can be displayed by the device. fig.7side view of conservator showing the arrangement of laser and light sensor there are 5 sets of laser and light sensor which are installed circularly inside the conservator. the lasers are arranged in such a way that the float interrupt the laser penetrating the sensor. suppose the conservator is half filled as shown in the diagram above, the float will interrupt the laser penetrating the sensor which is further sensed by the sensor and shown on the device. in this way the level of oil can be indicated. fig.8front inside view of conservator showing the arrangement of laser and light sensor. 4.6 system development and connections the functions of the system are to check phase failure, unbalanced voltages and earth fault. apart from these it an also monitor the oil temperature and oil level. this can be carried out by connecting 3 potential transformers on the low voltage side of transformer to step down the voltage to a safe level which can be read by microcontroller. the microcontroller compares the output voltages of potential transformer to the reference values which is already fed in microcontroller. further according to that the microcontroller takes the decision about the healthiness of transformer which is then displayed on lcd. fig.9connection diagram of system temperature sensor is interfaced with the microcontroller to monitor the temperature of oil of transformer. the maximum values of oil are fed in microcontroller. if the oil temperature exceeds the safer value then the alarm is activated to alert the authorities. the output of magnetic oil gauge is connected to the relay which is then connected to microcontroller. it alerts the system if the level of oil is minimum in the conservator. the another modifications in the conservator tank for monitoring the real time level of oil in this 5 lasers are used to emit lights on light sensor. the level at which float of mog interrupt the laser penetrating the light sensor is defined as the level of oil in conservator. to display all this parameters lcd is used which is interfaced with the microcontroller. iv. conclusion in this research paper we have discussed how to detect faults of power transformer. this system provides an easy way to detect faults and monitor the transformer health to avoid future major shutdown. the power transformer fault detector provides the following advantages. • a method that continuously monitor the faults or failures. • wide range of features. • simple to use. the testing of power transformer fault detector is under process. after that the device will then progress to product development attached to the power transformers under the collaboration with any company. references [1]https://www.curtin.edu.au/research/local/docs/ip-commercialisation/powertransformer.pdf [2]nweke chisom. b (2014 august)using gsm to detect fault in microcontroller based power transformer [online] issn: 2321-9653 available: http://www.ijraset.com/fileserve.php?fid=768 [3]ole s. seiersen, “power supply circuit”, u.s patent 4899271 a,feb 6, 1990 [4]http://www.wisegeek.org/what-is-a-temperature sensor.htm#didyouknowout [5]http://www.futureelectronics.com/en/microcontrollers/microc ontrollers.aspx columbia undergraduate science journal open-access publication | http://cusj.columbia.com 50 cusjvolume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu sy m po si u m cusjcolumbia undergraduate science journalsymposium poster titles 51cusj volume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu sy m po siu m pathogenesis-related mutations in the t-loops of human mitochondrial trnas affect 3’ end processing and trna structure dmitri serjanov, kyla-gaye pinnock, louis levinger biochemistry/biophysics probing aromatic peptide aggregation using spectroscopy =hslu[puh�:�-lszlu��1\z[puh�*opu^vun��9\ls�+lzhtlyv��(kht�7yvä[ 9hpumhss�wh[[lyuz�pu�zwljpäj�ylnpvuz�vm�(mypjh delroy wills fluorinated solvent processing for organic optoelectronic applications ilana shimunov, marshall cox, ioannis kymissis design of low-cost thermophilic microbial fuel cell for sustainable wastewater treatment nathan winkler engineering designing, breadboarding, testing, troubleshooting, and constructing a life guard alarm unit john buoncora, dane burkett designing, constructing, programming, and debugging an autonomous robot gregory kissoon, john buoncora female energy supply lines: a study of firewood in northern ghana madeline russell foster water usage and availability in bongo’s communities: research leading to the development of an indigenous fluoride jessica marie friscia osteoblast precursor cells need the pyrophosphate transporter ankh in order to sense and respond to mechanical alen trubelja, kristen lee, christopher jacobs testing the fortracc model for short term forecasting of cloud movement and life cycle xiaoqian pan automated detection of gnra tetraloop prevalance using 3dna and python prerana pradhan*, mauricio esquerra, wilma k. olsen biochemistry/biophysics elucidating the interaction of lpa with model membranes evan minzter, rivkah rogawski* a comparison of methods to collect host-seeking, blood-fed and gravid mosquitoes in the surburban landscape lily c. hughes*, richard c. falco, thomas j. daniels biology analysis of nfkb protein subunits instimulated human macrophages steven kennedy*, sitharam ramaswami, ashish p. juvekar, subrata manna, ivana vancurova from egg raft to emergence: growth and survival of mosquito larvae in vernal pools jessie lanterman*, mike rubbo, rich falco the effect of rnai protein armitage on oskar mrna transport and localization in drosophila oocyte linda molla*, joseph t. cammarata*, irina e. catrina, diana p. bratu analysis of bortezomib-induced nuclear protein translocation in cutaneous t-cell lymphoma cells hung tran*, ashish p. jukevar, sitharam ramasswami ashish p. juvekar, subrata manna, ivana vancurova, subrata manna, ivana vancurova bone marrow degradation: correlations between lipid oxidation and caloric decline rates lily xie*, robert j. blumeschine, karen m. schaich anti-predator and social monitoring functions in blue monkeys kaitlyn m. gaynor*, marina cords evaluating nest protectors for turtle conservation: unintended consequences for the malaclemys terrapin shahriar rahman* engineering structural glass shells: the constructional and structional feasbility of a hyperbolic umbrella gregor j. horstmeyer*, sigrid adriaenssens, george scherer optical observations of thermoelectric effects of self-heated nanocrystalline niaz khan*, gokhan bakan, ali gokirmak, helena silva electrical properties of methyl based hybrid melting gels louis gambino*, andrei jitianu, lisa c. klein mathematical modeling and characterization of coiling perimodiolar mei yi cheung*, jian zhang, nabil simaan pressure distribution and forefoot geometry in ballet pointe shoescochlear electrode arrays antonia zaferious*, d.m. wooton phase-change oscillations and pulse generation in silicon microwaves adam cywar*, gokhan bakan, helena silva, ali gokirmak synthesis of hydroxyapatite coatings on calcite: developing a coating to protect marble from damage in acid rain jeffrey i. campbell*, sonia naidu, george w. scherer targeting nanoparticles to macrophages lila cheung* 2010 symposium chemistry anticarcinogenic and prooxidant properties of pomegranate juice extract loriel j. solodokin*, audrey canter, ahuva freilich, orli haken, chana g. ovits-levy, a.g. schuck, h. babich size-controlled growth of selenium nanoparticles on selfassembled kinetin nanostructures and their applicatior stacey n. barnaby*, ipsita a. banerjee abscisic acid based hydrogels as drug delivery vectors evan m. smoak*, stephen h. frayne, karl r. fath, ipsita a. banerjee implementation and analysis of passive online schemes for rogue access point detection in ethernet networks oscar a. perez*, bing wang psychology & neuroscience elucidating the toll like receptor signaling pathways of the immune jennifer deluty*, jeremy seto, stuart sealfon /\thuz�4ptpjrpun�(upthsz!�0twspjh[pvuz�mvy�:wljplz�:wljpäj� vocalization processing in human cortex william j. talkington, kristina rapuano*, chris frum, james w. lewis experiences and challenges of women combining academic careers and motherhood danielle l. auriemma*, tovah p. klein subcortical localization of ectopic cell cycle events in five alzheimer’s mouse models luming li*, timmy cheung*, jianmin chen, karl herrup medicine & health services redistribution of muscle mitochondria assocaited with vertebrate neuromuscular synaptogenesis mohammad gilani*, joav prives gadolinium-loaded micelles as intravascular contrast agent for micro-magnetic resonance angiography lindsay k. hill*, karen c. briley-saebo, moustafa doudai, asad baig, susan pun, brian j. nieman, daniel h. turnbull sb-t-1214 dramatically downregulates stem cell-related .lul�,_wylzzpvu�7yväslz�pu�4l[hz[h[pj�*vsvu�*hujly�*lss yan leyfman*, yuan wang, iwao ojima, galina i. botchkina imbibition of gases between silica and graphene kate e. noa*, milton w. cole, angela d. lueking physics performance of different jet clustering algorithms in reconstructing hadronic tops david krohn, cristina popa*, liantao wang graduate students using continuum robots for contact detection and estimation of contact location andrea bajo*, nabil simaan validation and benchmarking of computational fluid dynamic simulations for wind building response francesca ceccarini*, rene b. testa high frequency plateau responses recorded in gerbil auditory nerve tuning curves stanley huang*, wei dong, elizabeth olson design of steerable electrode arrays and optimal insertion path planning for robot-assisted cochlear implant surgery jian zhang*, j. thomas roland jr., spiros manolidis, nabil simaan variability of cloudiness and water vapor content over the caribbean using isccp data alma cabral reynoso automated software testing & deployment; the graphyte web toolkit carl chinatomby generation of mutated forms of leishmania major phosphatidylethanolamine methyltransferase lmpem1 subbhalakshmi dhalladoo, tongtong zhu effect of ganoderma lucidum on the migration of human breast cancer cells di wu, regina sullivan sumo proteins may regulate head reshaping, capacitation, and stress response in human sperm jordana schneider biology comparison of copper surface mediated toxicity in gram positive and gram negative bacteria phyo thu, janet long, nidhi gadura understanding the mechanisms of copper induced, lipid peroxidation mediated cell death in saccharomyces cerevisiae jasodra ramlall, nidhi gadura impact of plasma membrane unsaturated fatty acids levels on copper surface mediated cell death in e.coli bo zhi robert hong, nidhi gadura effect of co-enzyme on cancer cell proliferation and survival shawna benjamin, maria l cotrina, regina sullivan pro-oxidant and pro-apoptotic activities of olive fruit extract toward oral carcinoma cells. jennifer lazaros, erica michelle hasten, alyssa schuck, harvey babich effect of vegf on glucose transporter 1 during status epilepticus bobby thomas effect of carnitine on brain cell survival jose zhagnay, maria l cotrina effect of l-carnitine on breast cancer cell migration adam mcconnell, maria l cotrina, regina sullivan analysis of bortezomib-induced nuclear translocation of ikba in metastatic prostate cancer cells subrata manna, adeel zubair, ivana vancurova the effect of hif2 alpha on cell growth and cell survival cathy meng hao, yatrik shah chemistry density function theory study on the reaction of deprotonated tyrosine [tyr-h]with singlet molecular oxygen kallol mitra heavy metals in golf balls by x-ray fluorescence (xrf) spectrophotometry andre hugh smithson reactivity of tris (trimethylsilyl) phosphite in the synthesis of the nmustard bisphosphonic acid of bicine and phenylalanine luis vargas, marjorie morales, kevin chavez determination of copper content of the us penny by x-ray fluorescence, visible spectroscopy and gravimetric method sharda jagdeo concentration dependence of refractive index measured by a laser pointer: refractive index vs. concentration rebecca cho determination and applications of the van ‘t hoff factor using microscale freezing point determination measurements fathima nazmudeen, paris svoronos, mario balducci, pedro irigoyen an unusual porous and spherical micro-scale morphology for poly(o-toluidine) daniel lafaurie, david m. sarno 5huväilyz�jv�wvs`tlypalk�myvt�hupspul�huk�z\smvuh[lk�hupspul� for electrostatic interaction with gold nanoparticles engred vanegas, david m. sarno the probing zinctetracycline using vibrational spectroscopy jinnette tolentino feasibility of using electronic health record temperature dependence of refractive index determined by a esther yang a thermoresponsive multicolor colloidal photonic crystal based on polymeric core-shell microspheres saad syed, man kin chow synthesis and behavior of 6-bromoindigo xiaomei ye, sasan karimi, pedro irigoyen 2011 symposium microsoft word cusj_spring_2015_journal.docx !cusj%|%volume%9%|%spring%2015% 6% cras+faucibus+tempor+magna.+maecenas+dapibus+sodales+erat.+ aenean+sem+felis.+ our future in the stars helen yang1, briley lewis2 & abhinav seetharaman3 1department of biochemistry, columbia university, new york, ny, usa; 2department of physics, columbia university, new york, ny, usa; 3department of chemistry, columbia university, new york, ny, usa 1 after years of hard physical and mental training, they take their first steps to the shuttle, waving goodbye to all the spectators and earth. these astronauts, originally scientists, teachers, pilots, and engineers, each endured at least three years of rigorous professional training before even applying to an astronaut program. earning a bachelor’s degree is the minimum requirement for nasa positions, and the astronauts train beyond the classroom by swimming laps in a space suit to experience zero-gravity. a day in an astronaut’s life might start by climbing aboard the “vomit comet,” an aircraft that flies a parabolic path to simulate microgravity conditions. astronauts also accustom themselves to move and work in weightlessness at the neutral buoyancy lab in nasa’s johnson space center (jsc). behind the walls of this facility in houston, texas, these astronauts-in-training submerge themselves in a massive swimming pool while in clunky space suits. they navigate in full-scale underwater mockups of their shuttle and familiarize themselves with the life-size replica of the international space station (iss). after years of training, these space pioneers stand at the launch pad with 1000 jet aircraft pilot-in-command hours in their pockets, thoroughly acquainted with every module on the iss. nasa statistics claim that the launch shuttle sends our astronauts hurling into space at 18,000 mph, a speed nine times faster than the average rifle bullet. in just six hours, they arrive at the actual iss, which spans about the width of an american football field. the docking process is actually the most complicated component of their journey; the spacecraft cannot dock without entering the correct orbit at the correct time, and there is no room for a mistake that might send the spacecraft crashing into the iss. when the astronauts finally do make it onboard the iss, they’ll find themselves inside a leviathan weighing nearly one million pounds. the astronauts have more space than a six-bedroom house and are required to exercise in the station’s gymnasium. they might walk through the main central truss and look through the 360º bay window, and then visit laboratories where physicists attempt to detect dark matter and biologists study muscle atrophy in zebrafish. to prevent loss of muscle and bone mass, our astronauts engage in scheduled exercise and various spaceship repairs every day, leaving them only an hour or two of free time in the mornings. control center staff back on earth likewise cannot sit back and relax. orbital debris presents a constant, imminent danger to the wellbeing of the iss. station-crew and on-ground staff must do all that they can to protect this $150 billion flying space station from large debris while simultaneously conducting research and repairing the ship. 2 the iss is anything but permanent. our astronauts’ toil will amount to nothing if we cannot raise the funds necessary to keep the station in orbit. boeing predicts that the iss’s parts can hold up through 2028, but the bigger issue is finding funding to keep the station alive. nasa and its partners in russia, japan, canada, and other countries have committed to funding the station through 2020, but what its fate afterwards is uncertain. these countries are debating the question if the structure should be kept in orbit, allowing further research and providing a market for space transport companies like spacex and sierra nevada. alternatively, they could choose to give up the mammoth, perhaps by letting it spiral down into the south pacific for a watery death. fortunately for ardent iss supporters, there is some solace for the immediate future; nasa and white house officials announced plans to keep the station running till at least 2024. still, it is time to look beyond this station. as famous and celebrated as it may be, new avenues for exploration must be built after the iss becomes dysfunctional. as our society has look into our future in space, we have expanded our ideas, including plans to construct space colonies –stations with earth-like features that function as permanent settlements. if these colonies successfully sustain human life, there are plans to build a mega-station called ‘elysium’ – based off the movie – which could house a large portion of the human population. such stations might very well be essential to mankind’s future in the stars. there are also plans to construct a new nasa vehicle, the orion multi-purpose crew vehicle (mpcv), specifically for deep space exploration. expected to meet the constantly growing needs of human space exploration programs, orion may eventually carry astronauts to worlds far beyond earth’s orbit which no man has ever seen or set foot on. this century marks the dawn of a new era in exploration, and if we can muster the manpower and financial support, humankind will advance further than we’ve ever been from our home planet. buy signal from limit theorem yiqiao yin statistics department columbia university november 2, 2017 abstract this paper studies price actions in capital market as a random walk from limit theorems. through clear construction, we derive algorithms from a series of theorems to create standardized buy signals given a trader’s committed frequency to participate in the market. 1 introduction security prices follow random walk. although some scholars doubt the concept of efficient market hypothesis, there are fruitful amount of previous research exploring and studying this topic. some notable papers are by fama and french [2], [3], [4], and [5]. other scholars such as malkiel have also provided persuasive empirical evidence that we do observe data in favor of efficient market hypothesis [6]. an important contribution from yin (2017) [7] was the idea and theoretical notion of optimal level in security prices. their work raised a concept that the anomaly prices can be corrected which was a notion not yet discovered in the field of probabilistic price analysis. we took this notion as a foundation and further explore this field of security price random walk. we discovered that a series of constructions can be built to form standard normal distributions. we will further prove these results (see appendix) and develop a series of trade-able signals from these theories. the hunger for this type of work is necessary for the industry because conventional asset pricing models do not signal buyers when to involve in the market. moreover, for retail traders with a fixed trading frequency (assuming rational retail traders), it only makes sense for them to participate in the “low” prices consistent with their frequency. there is currently no models presenting us any algorithms in that sort. this motivates us to formalize these algorithms from theorems we developed and we aim to provide traders a consistent buy strategy so that one can trade, whatever strategy one trades, at a systematically low price. 2 theoretical framework in this section, we first present, in §2.1, the architecture of the theorems, which are discussed in §2.2. we formalize the theorems based on the notion of central limit theorems, which are proved in respect to the order of theorems in §5 appendix. continuing from the proved theorems, we provide the construction of a series of algorithms in §3. 2.1 architecture definition 2.1.1. for each company i at a time t, we observe a price, that is, pi,t (2.1.1) definition 2.1.2. sman = 1 n t−n∑ i=n pi,t−n (2.1.2) definition 2.1.3. let n be the same value from definition 2.1.2, denote eman = (pi,t − eman−1)×m+ eman−1 (2.1.3) while m = 2 n+1 . 2.2 theories theorem 2.2.1. for some n, suppose we have price by definition 2.1.1 and sma by definition 2.1.2, then we have pi,n − sman ⇒ χ (2.2.1) while χ is the stand normal distribution. theorem 2.2.2. let the distance between price and moving average to be d which is defined as di := pn − sman while i = n, and then we can consider di to be i.i.d. with edi = 0 and edi = σ2 ∈ (0,∞). then n∑ m=1 dm /( n∑ m=1 d2 m )1/2 ⇒ χ (2.2.2) while χ is the stand normal distribution. theorem 2.2.3. let the distance between price and moving average to be d which is defined as di := pn − sman while i = n, and then we can consider di to be i.i.d. with edi = 0 and edi = σ2 ∈ (0,∞). let sn = d1 + · · · + dn. let nn be a sequence of nonnegative integer-valued random variables and cn a sequence of integers with cn → ∞ and nn/cn → 1 in probability. then snn /σ √ an (2.2.3) where χ is a standard normal distribution. 1 buy signal from limit theorem [by yiqiao yin] §4 theorem 2.2.4. let the distance between price and moving average to be d which is defined as di := pn − sman while i = n, and then we can consider di to be i.i.d. with edi = 0 and edi = σ2 ∈ (0,∞). let sn = d1 + · · · + dn. let nt = sup{m : sm ≤ t}. then as t→∞, (µnt − t)/(σ2t/µ)1/2 ⇒ χ (2.2.4) while χ is the stand normal distribution. 3 algorithms this section we take the theorems above, from §2 theoretical framework, as given and we introduce a series of algorithms targeting buy signals. algorithm 3.0.1. given a buy frequency by an investor c, for all i in a stock pool of companies: step 1. observe price pt for each company step 2. store pi,t step 3. compute sman dn := pi,n − sman step 4. if dn ≤ c, print ”+1”; else, print ”0”. print a collection of ”+1” per company i per n. as the first algorithm in the section, it has a very intuitive understanding. one can simply observe price and computes its sma. then one needs to look at the difference between price and sma to know how often should he buy given that he has a fixed frequency. this is and will always be true because the difference of price and sma follows random walk, as stated in theorem 2.2.2 and proved in appendix. this means that this time-series difference we are looking at goes up or down but stay in the middle most often. such bell-shape curve can give as a precise probability distribution and we can mark down an exact price level given a frequency we want to participate in the market. algorithm 3.0.2. given a buy frequency buy an investor c, for all i in a stock pool of companies: step 1. observe price pt for each company step 2. store pi,t step 3. compute dn := pi,n − sman signaln := n∑ m=1 dm /( n∑ m=1 d2 m )1/2 step 4. if signaln ≤ c, print ”+1”; else, print ”0”. print a collection of ”+1” per company i per n. algorithm 3.0.2 takes algorithm 3.0.1 as a building block and expand the idea and we can normalized the distance (or difference) of summation of a series of distances by square root of its own value to construct buy signals. algorithm 3.0.3. given a buy frequency by an investor c, for all i in a stock pool of companies: step 1. observe price pt for each company step 2. store pi,t step 3. compute snn = d1 + · · · + dn, σ is the variance of di, self-normn := snn/σ √ an step 4. if self −normn ≤ c, print ”+1”; else, print ”0”. print a collection of ”+1” per company i per n. algorithm 3.0.2 brought up the notion of normalizing by square of its own value. it is also practical to normalize by itself, which is what algorithm 3.0.3 was attempting to do. algorithm 3.0.4. given a buy frequency buy an investor c, for all i in a stock pool of companies: step 1. observe price pt for each company step 2. store pi,t step 3. compute the mean µ and the variance σ, renewaln := (µnt − t)/(σ2t/µ)1/2 ⇒ χ step 4. if renewaln ≤ c, print ”+1”; else, print ”0”. print a collection of ”+1” per company i per n. besides notions of self-normalizing, we can also construct standard normal distribution by taking time, risk, and mean into consideration. such “renewal” process can be done without breaking the form of standard normal distribution. for traders who are interested in looking at more parameters, there is such freedom to do so. algorithm 3.0.5. given results from the above algorithms, that is, algorithms 3.0.1, 3.0.2, 3.0.3, and 3.0.4, run step 1. retreat (dn), (signaln), (self-normn), and (renewaln). step 2. each i, at any time t, compute buy : ∑ all signals := val(dn)+val(signaln) +val(self −normn) +val(renewaln) step 3. print(t); print(buy). that is, buy =  very heavy, b = 4 heavy, b = 3 not that heavy, b = 2 tiny, b = 1 do nothing, b = 0 page 2 buy signal from limit theorem [by yiqiao yin] §5 while b is a discrete time-series function of time t, i.e. b(t) : z→ {0, 1, 2, 3, 4}. 4 conclusion this paper starts with a strong motivation §1 and introduced the background of why we study security prices in such construction. next, we present a clear architecture, in §2, and a series of theorems under such building blocks. continuing with the results from theorems which we can collect empirically, we develop trade-able algorithms §3. in summary, we believe such algorithms can build a capital market with less risk (an anomaly corrector). for future reference, we believe our attempts also opened up a lot more potential research problems. for example, what would happen if everyone starts to use this strategy? another great question can be, what would be an ideal (although state-of-art) game plan after the algorithm tells traders to buy? in macro point of view, how would an economy perform in long run if one implements this strategy in a larger scale? what if an unknown outside monetary force enter the market and act as an anomaly, how would this algorithm deal with such situation? 5 appendix 5.1 proof of theorem 2.2.1 this is a relatively easy proof since the definition follow the premises of the central limit theorem. that is, we have pi,n and sman that are i.i.d.. then by c.l.t., pi,n−sman ⇒ χ while χ stands for standard normal distribution. q.e.d. 5.2 proof of theorem 2.2.2 from weak law we know that n∑ m=1 d2 m/nσ 2 → 1. also note y−1/2 s continuous at 1, then we have( σ2n / n∑ m=1 d2 m )1/2 → 1, in prob., see ?∑n m=1 dm σ √ n ( σ2n∑n m=1 d2 m )1/2 ⇒ χ · 1, from ? = χ notice that the ? is because in weak convergence, there is a theorem stated that xn ⇒ x∞ if and only if for every bounded continuous function g we have eg(xn)→ eg(x∞). since we discussed the continuity of function y−1/2 at 1, this line is valid. q.e.d. remark 5.2.1. from [1], section 2, the theorem stated the following. suppose xn ⇒ x, yn ≥ 0, and yn ⇒ c, where c > 0 is a constant, then xnyn ⇒ cx. 5.3 proof of theorem 2.2.3 from kolmogorov’s inequality we know p ( max (1−ε)cn≤m≤(1+ε)cn |sm − s[(1−ε)cn]| ) ≤ 2ε/δ2 if dn = snn /σ √ cn and yn = scn/σ √ cn, then it follows that lim sup n→∞ p (|dn − yn| > δ) ≤ 2ε/δ2,∀ε then we have p (|dn − yn| > δ) → 0 for each δ > 0, i.e., xn − yn → 0 in probability. this is because of the cnverging together lemma stated in weak convergence part of [1]. we state the theorem in remark below. q.e.d. remark 5.3.1. suppose xn ⇒ x and yn ⇒ c, where c is a constant then xn + yn ⇒ x + c. a useful consequence is that if xn ⇒ x and zn −xn ⇒ 0 then xn ⇒ x. 5.4 proof of theorem 2.2.3 from convergence theorem, we know that nt tµ → 1 so from theorem 2.2.3 we have sn − µnt σ √ t/µ → 0 then it is sufficient to show (sn − t)/ √ t→ 0 since it follows that (µnt−t)√ σ2t/µ ⇒ χ. we have given finite variance, that is, σ2 < ∞, so by d.c.t. (dominated convergence theorem), we have p ( max 1≤m≤2tµ ym > ε √ t ) = 2t µ p (y1 > ε √ t) = 2 µε2e(y 2 1 ;y1 > ε √ t)→ 0 which proves that (sn − t)/ √ t → 0 is true. hence, this completes the proof. remark 5.4.1. this is because of the converging together lemma stated weak convergence. please see remark 5.3.1. q.e.d. page 3 buy signal from limit theorem [by yiqiao yin] §5 references [1] durrett, rick, “probability: theory and examples ”, 4e. [2] fama, eugene f. (1992), “the cross-section of expected stock returns,” journal of finance, 47, 427-465. [3] fama, eugene f. (1993), “common risk factors in the returns on stocks and bonds,” journal of financial economics, 33, 3-56. [4] fama, eugene f. (1996), “multifactor portfolio efficiency and multifactor asset pricing,” journal of financial and quantitative analysis, 31, 441465. [5] fama, eugene f. (2006), “profitability, investment, and average returns,”, journal of financial economics, 82, 491-518. [6] malkiel, burton g. (2007), “reflections on the efficient market hypothesis: 30 years later,” the financial review, 40, 1-9. [7] yin, yiqiao (2017), “anomaly correction by optimal trading frequency”, cusj, page 4 introduction theoretical framework architecture theories algorithms conclusion appendix proof of theorem 2.2.1 proof of theorem 2.2.2 proof of theorem 2.2.3 proof of theorem 2.2.3 microsoft word document3 articlevolume 2, spring 2007 differentiation of two species of mullerian inhibitory substance (mis) andrew r bader1, douglas lim2, david maclaughlin3 1 columbia college, columbia university, 2 clarkson university, 3 pediatric research laboratory, massachusetts general hospital abstract muellerian inhibiting substance (mis), also known as anti-muellerian hormone, is a testicular protein responsible for the regression of the mullerian duct. in vivo and in vitro experiments using mis have shown it to inhibit the growth of various tumors, including certain ovarian cancers. two approaches to purifying the product have been used, immunoaffinity purification and chromatography. the purified products were tested for affinity to binding sites using the enzyme linked immunoabsorbent assay (elisa). the goal of this study was to test the theory that the two mis proteins identified, termed iap and q, were different on structural and molecular levels. the results of these investigations reveal the existence of two distinct types of mis differentiated by elisa which based on future study may have different impacts as cancer treatment agents. columbia undergraduate science journal open-access publication | http://cusj.columbia.edu spring 2014 | volume 8 22 classification of neocortical neurons using affinity propagation naureen ghani1*, rafael yuste2 1the fu foundation school of engineering and applied science, columbia university, new york, ny, usa; 2department of biological sciences, the kavli institute for brain science, columbia university, new york, ny, usa abstract determining the circuitry of the neocortex requires an understanding of its components, which in turn necessitates a classification scheme of neocortical neurons. the neocortex consists primarily of excitatory pyramidal neurons (~80% of neocortical neurons) and inhibitory interneurons (~20% of neocortical neurons). neocortical interneurons are ideally positioned to control circuit dynamics. gabaergic interneurons, in particular, largely contribute to the vast morphological and physiological variability of the cortex. however, much is still not known regarding these interneurons. here we use unsupervised learning methods, such as pca followed by k-means analysis, to create quantitative and unbiased classification schemes. we explored the use of affinity propagation, a novel exemplar-based method of cluster analysis, to quantitatively classify a diverse set of 337 neocortical interneurons. each neuron was characterized by wholecell recordings done by patch-clamping and complete 3d anatomical reconstructions. it was revealed that each interneuron subtype is characterized by a unique set of morphological and electrophysiological features that allowed for specialized function within the neocortical circuit. affinity propagation may then be used to classify neurons, the first step to reverse engineer a neocortical circuit. introduction developmental neuroscience is an emerging field that harnesses powerful computational and mathematical techniques to model the brain. the goal of our research is to build a classification scheme of all neocortical neurons in a major effort to determine the circuitry of the neocortex. the neocortex is the top layer of the cerebral hemispheres and is about 2 to 4 mm thick. it is composed of six layers, labeled i to vi (with i being the outermost layer and vi being the innermost layer). along with the archicortex and paleocortex, the neocortex is one of three components of the cerebral cortex, accounting for 76% of the volume of the cerebral cortex. the neocortex is one of the most significant components of the brain and is involved in higher mental functions such as perception, memory, imagination, language, and music. since the neocortex participates in such a vast array of computational tasks, it is often compared to a parallel computer, which can carry out multiple calculations simultaneously. the underlying principle of parallel computing is that large problems can be divided into smaller ones, which are then solved concurrently or “in parallel.” in other words, many computing elements are wired together and then operate in parallel. furthermore, for every computing element added, it is often necessary to add new wires from the new addition to existing elements. thus, the number of wires increases much more rapidly than the number of computers: 3 wires for 3 computers, 6 wires for 4 computers, 10 wires for 5 computers, and so on. extending this analogy to the brain, we would need a brain the size a football field to encompass billions of neurons (biological computing elements) that are all wired together. a real brain however selectively wires neurons together, omitting any unneeded wires (axons) for proper function. this simple mathematical analogy hints at many aspects of the brain. for instance, as mammals grow larger with evolution, the brain similarly increases in size, and the “cost” of wires grows severe. neuroanatomical studies have quantitatively shown that the number of copyright: © 2014 the trustees of columbia university, columbia university libraries, some rights reserved, ghani, et al. received 1/1/2014. accepted 3/14/2014. published 3/14/14. *to whom correspondence should be addressed: naureen ghani, columbia university, new york ny 10027, email: ng2410@columbia.edu. neuroscience columbia undergraduate science journal open-access publication | http://cusj.columbia.edu spring 2014 | volume 8 23 axons varies greatly region to region in the brain; some areas are heavily linked to others with massive axon tracts while others are far less strongly linked. many strides have been made in understanding the formation and function of neocortical circuits using detailed morphological and electrophysiological analyses of the neurons. each of these model circuits differ greatly in form and content, however, because they each have different theoretical and experimental considerations. for historical reasons, many early neocortical circuit models are based on data from cat and primary visual cortex. in recent years, the rat somatosensory cortex has grown in prominence as a tool for understanding the cortex. the idea of a canonical circuit has previously been applied to aspects of the vertebrate brain and spinal cord. it is highly relevant to questions of evolution, development, and homology of form and function. supporting this idea is the close similarities in basic organization across vertebrate brains. similarly, recognizable neuronal subtypes have been found across the neocortices of different mammalian species15. for these reasons, we hypothesize the existence of a canonical cortical microcircuit and it implements a relatively simple computation. to create a neocortical circuit model, a classification scheme must first be established to outline the components of the cortex. in the past, cell type classification was qualitative and led to inconsistent subtypes. in recent years, the standard method for classification of neurons has been quantified through unsupervised cluster analysis2, 4, 7, 8, 9, 11, 12. unsupervised cluster analysis is the classification of a set of data or objects into subsets (clusters) so that the data in each subset shares a common feature without any prior knowledge. one widely-used technique has been ward’s method with hierarchical clustering. hierarchical agglomerative clustering is a bottomup technique and begins by grouping the two “closest” cells as defined by the algorithm, and then continues to join the next “closest” cells and so forth. this method does not require specifying the number of clusters in advance and generates smaller clusters that may be helpful for discovery. however, one of the main disadvantages of hierarchical clustering is that once two cells are linked, they remain joined together in the final hierarchy. moreover, hierarchical clustering is susceptible to a chaining effect in which objects may be “incorrectly” assigned to a cluster at an early stage instead of being grouped in new clusters. additional clustering techniques include k-means analysis k-metoids analysis, and affinity propagation. in k-means, each cluster is represented by the center of the cluster and in kmetoids analysis, each cluster is represented by one of the objects in the cluster. affinity propagation is one of the most recently developed unsupervised clustering techniques. each data point is viewed as a node in a network, and realvalued messages are transmitted between the data points until a set of exemplars and corresponding clusters is determined by the algorithm. thus, at any point in time, the magnitude of each message reflects the current affinity that one data point has for choosing another data point as its exemplar, hence the name of “affinity propagation”6. in this work, we explore the application of affinity propagation to classification of neocortical neuronal subtypes. the algorithm was used to blindly classify a test dataset of four interneuron subtypes. the dataset included unlabeled cells in addition to known cells, the latter serving as a ground truth. the dataset is comprised of 67 morphological variables and 20 electrophysiological variables describing (1) parvalbumin-positive (pv+) basket cells (bc), (2) pv+ chandelier cells (chc), (3) somatostatinpositive (som+) martinotti cells (mc), and (4) som+ non-martinotti cells (non-mc) as previously described in research12. we found that affinity propagation generates a fair classification in separating these four known interneuron subtypes and may be a powerful classification tool in discovering or defining neuronal cell types. materials and methods preparation of brain slices acute brain slices were prepared from nkx 2.1, g42, or gin mice, with an average of 15 postnatal days (range p13 – p25). mice were immediately decapitated, the brain was removed and then immediately placed in a cold sucrose cutting solution (222 mm sucrose, 2.6 mm kcl, 27 mm nahco3, 1.5 mm nah2p4, 0.5 mm cacl2, 3 mm mgso4, bubbled with 95% o2, 5% co2). coronal slices of 300 μm thickness were cut using a vibratome and then transferred to a neuroscience columbia undergraduate science journal open-access publication | http://cusj.columbia.edu spring 2014 | volume 8 24 holding chamber at room temperature with oxygenated acsf (126 mm nacl, 3 mm kcl, 3 mm mgso4, 1 mm cacl2, 1.1 mm nah2po4, 26 mm nahco3, and 10 mm dextrose, bubbled with 95% o2, 5% co2). the slices were left to equilibrate with the room temperature for at least 30 minutes. slices were then transferred to a recording chamber with the perfusion of acsf bubbled with 95% o2, 5% co2. transgenic mouse lines to identify different types of interneurons, we used three transgenic mouse lines. first, we used the g42 line that labels pv+ cells2. pv+ cells are rapid spiking interneurons with basket or chc morphology. we are able to identify basket cells from chandelier cells by their distinctive morphologies and threshold spiking responses. in addition, the chandelier cells have specially shaped axon arbors, in which axon terminals form distinct arrays called “cartridges” that we can visualize via gfp17. the nkx2.1 line labels a population of interneurons that express the transcription factor nkx 2.1, which includes interneurons that migrate from the medial ganglionic eminence (mge), most notably chcs17. a significant proportion of the chc cells were found at the top of layer ii, close to the layer i border, in both the g42 and nkx 2.1 lines17. finally, we used the gin line to label som+ cells12. som+ cells are regular spiking interneurons with diverse morphology. in previous work, we determined three unique subtypes of som+ interneurons in gin mice based on morphology and physiology: martinotti cells and two novel subtypes12. as a result, we now distinguish between martinotti cells (mc) and two novel subtypes (non-mc). electrophysiology recordings brain slices were placed in a recording chamber at room temperature with a constant supply of oxygenated acsf. pipettes of 3-7 mω resistance were pulled from borosilicate glass. whole cell recordings of cells were obtained using patchclamping. only cells with a healthy resting membrane potential (between -55 and -80 mv) were selected for recording. electrophysiological analysis 20 variables were measured for each neuron by analysis of the recordings in matlab. the petilla terminology scheme was used to name each variable describing firing and passive properties1. see appendix 1 for descriptions. histological procedure neurons were filled with biocytin by a patch pipette. slices were kept overnight in 4% formaldehyde in 0.1m phosphate buffer (pb) at 4°c. slices were then rinsed three times for five minutes per rinse on a shaker in 0.1m pb. they were then placed in 30% sucrose mixture (30g sucrose dissolved in 50 ml ddh2) and 50 ml 0.24m pb per 100 ml) for 2 hours and then frozen on dry ice in tissue freezing medium. the slices were kept overnight in a -80°c freezer. after defrosting, the slices were rinses in 0.1m pbthree times, twenty minutes each, to remove tissue freezing medium. slices were then incubated in 1% hydrogen peroxide in 0.1m pb for thirty minutes to pretreat the tissue. they were then rinsed twice in 0.02m potassium phosphate saline (kpbs) for twenty minutes. afterwards, the slices were kept overnight in avidin-biotin-peroxidase complex. the slices were next rinsed three times in 0.02m kpbs ((0.7 mg/ml 3,3” diaminobenzidine, 0.2 mg/ml urea hydrogen peroxide, 0.06m tris buffer in 0.02m kpbs) ) for 20 minutes each. each slice was observed under a light microscope and then mounted onto a slide using crystal mount. three-dimensional neuron reconstruction and morphological analysis three-dimensional reconstructions of successfully filled and properly stained neurons were done using neurolucida software (microbrightfield). the neurons were viewed with a 100x oil objective on an olympus bx51 upright light microscope. differential interference contrast (dic) microscopy was employed to see otherwise invisible features of the sample. the neuron’s processes were traced manually while the program recorded the coordinates of the tracing, thus creating a three-dimensional reconstruction. in addition to the neuron, the pia and white matter were drawn. the neurolucida explorer program was used to measure 67 morphological variables of the reconstruction describing somatic, neuroscience columbia undergraduate science journal open-access publication | http://cusj.columbia.edu spring 2014 | volume 8 25 dendritic, and axonal properties. see appendix 2 for descriptions. affinity propagation affinity propagation is a clustering algorithm based on “passing messages” between data points6. it aims to combine the advantages of affinity-based clustering and model-based clustering. affinity propagation is similar to kmetoid clustering in that both algorithms output “exemplars” that either serve as representative data or the centers of the clusters. unlike kmetoid clustering, the exemplars are not chosen randomly in affinity propagation. affinity propagation takes an input of real-valued similarities between data points, where the similarity s(i,k) indicates how appropriate a data point is to serve as an exemplar for data point i. negative euclidean distance is used to measure similarity to minimize squared error: for points xi and xk, s(i,k) = −||xi − xk||2. the key advantage of affinity propagation is that it does not require the number of clusters to be specified beforehand, unlike k-means cluster analysis. instead, affinity propagation takes as input a real number s(k,k) for each data point k, such that data points with larger values of s(k,k) are more likely to be selected as exemplars. there are two types of messages exchanged between data points. the first type called the “responsibility” r(i,k), is sent from data point i to candidate exemplar point k, and represents how well-suited point k is to serve as the exemplar for point i, compared to all other potential exemplars for point i. the second type, called the “availability” a(i, k), is sent from candidate exemplar point k to point i, and reflects the compiled evidence for how appropriate it would be for point i to choose point k as its exemplar, taking into account the support from other points that point k should be an exemplar. r(i,k) and a(i, k) can be viewed as log-probability ratios. initially, the availabilities are initialized to zero: a(i, k) = 0. then, the responsibilities are computed as: availabilities will eventually fall below zero as points are assigned to other exemplars. this will decrease the effective values of the input similarities, removing candidate exemplars from the competition. whereas the competition is data-driven for responsibilities and all the candidate exemplars compete for the ownership of a data point, the availability update gathers evidence from data points as to which candidate exemplar would make a good exemplar. the availability a(i,k) is set to the self-responsibility plus r(k, k) plus the sum of the positive responsibilities candidate exemplar k receives from other points: self-availability a(k, k) reflects evidence that k is an exemplar based on positive responsibilities sent to candidate exemplar k from other points: to evaluate the quality of a clustering produced by affinity propagation, we considered two aspects: (1) the number of points that are correctly classified (2) the number of clusters with affinity propagation, we make the assumption that the identities of the exemplars are known, and so the number of correctly classified points may artificially increase. therefore, we compute the classification accuracy as the ratio between correctly classified points (excluding the exemplars) and the total number of points (excluding the exemplars). in addition, a low number of clusters is preferred. results database of three known interneuron subtypes we explored the use of affinity propagation to classify neocortical interneurons based on their morphological and physiological properties. in order to test the affinity propagation algorithm, we used a dataset where the identities of the neurons were known from previous studies12, 13, 17. more specifically, we used a physiology database that contained 337 interneurons distributed as: 57 somatostatinpositive cells (som+), 87 chandelier cells (chc), and 193 parvalbumin-positive cells (pv+). the morphology database consisted of 111 interneurons distributed as: 24 chc, 55 som+ and 32 pv+. lastly, there were 51 neurons in a neuroscience columbia undergraduate science journal open-access publication | http://cusj.columbia.edu spring 2014 | volume 8 26 database that consisted of both morphology and physiology variables, formed by an intersection of the two databases. its distribution was: 12 pv+, 16 som+ and 23 chc. affinity propagation classification of interneuron morphologies the analysis of the morphology database resulted in 2 clusters (figure 1). the first cluster consists of 84 neurons while the second cluster had 27 neurons. the first cluster had an exemplar of member 42, a som+ interneuron. it consists of: 22 chc, 32 som+ and 30 pv+. even though this cluster has an exemplar of som+ subtype, this cluster encompasses all the pv neurons with an accuracy of 93.75%. however, it only classifies 26.19% of the som+ interneurons correctly. the second cluster consists of: 2 cc, 23 som+ and 2 pv+. its exemplar is member 83, a som+ neuron. thus, 85.9% of the som+ neurons were correctly classified. this suggests that the second cluster with an exemplar of member 83 is truly representative of som+ neuronal subtype while the first cluster is representative of the pv subtype. figure 1: using 20 electrophysiological variables, 110 interneurons were classified into 2 clusters by affinity propagation where the exemplars are members 42 and 83. affinity propagation classification of interneuron physiologies the analysis of the physiology database revealed 3 distinct clusters (figure 2). the first cluster consisted of 57 som+ neurons with an exemplar of member 21, also a member of the som+ subtype. hence, 100% of the som+ neurons were correctly classified by the affinity propagation algorithm. the second cluster consisted of 8 pv+ and 87 cc neurons, where the exemplar was a part of the cc class (member 322). it correctly classified 90.80% of the cc neurons. the third cluster consisted of 142 cc and 51 pv+ and had an exemplar of 326, a member of the pv class. while the exemplar was representative of the pv+ subtype, only 26.42% of the pv+ neurons were correctly classified. figure 2: using 67 electrophysiological variables, 337 interneurons were classified into 3 clusters by affinity propagation where the exemplars are members 21, 322, and 326. affinity propagation classification of interneuron joint databases the analysis of the interneuron joint database resulted in 2 clusters (figure 3). the first cluster consisted of: 6 pv, 8 som, and 17 cc. the exemplar was member 4, a member of the pv subtype. it correctly classified a mere 19.37% of pv neurons. the second cluster consisted of: 6 pv, 8 som, and 6 cc. the exemplar was member 38, a member of the cc subtype. only 30% of the cc neurons were correctly classified. neuroscience columbia undergraduate science journal open-access publication | http://cusj.columbia.edu spring 2014 | volume 8 27 figure 3: using 67 morphological variables and 20 electrophysiological variables, 50 interneurons were classified into 2 clusters by affinity propagation where the exemplars are 4 and 48. discussion affinity propagation: an exploratory tool for classification of neural data in this study, we have explored the use of a new algorithm, affinity propagation, for the classification of neuronal data. we used a database of 337 neocortical gabaergic interneurons. interneurons previously identified served as a ground truth, and acted as a measure for how accurate the algorithm was. the data was based on a collection of morphological and physiological data for each neuron. the classification accuracy we found was 0.56 for the physiology database, 0.45 for the morphology database, and 0.40 for the combined morphology + physiology database. the accuracy consistently decreased with a smaller data set, containing less information on neurons. the affinity propagation algorithm is able to consistently distinguish somatostatin neurons as a unique class among all three databases. however, difficulty arose s when the algorithm is asked to differentiate between chandelier cells and parvalbumin cells. in the morphology database and the combined database, the cells were grouped into a single cluster consisting of large components of both chandelier cells and parvalbumin cells as opposed to two distinct clusters. after simplifying the database to include only chandelier cells and parvalbumin cells, while excluding somatostatin-positive cells, the affinity propagation algorithm is able to separate the cells into two clusters but with low accuracy (< 0. 50 for physiology database, the largest database was used). the inability of the affinity propagation algorithm to separate chandelier cells and parvalbumin cells may be due to a number of potential reasons. one reason is that the chandelier cells and parvalbumin cells are morphologically and physiologically similar to one another. recent research shows that chandelier neurons are a subset of gaba-ergic cortical interneurons that are said to be parvalbumincontaining and fast-spiking when innunostained16, thus distinguishing them from other gabaergic neurons, . even though chandelier cells are truly distinct from other gabaergic neurons based on morphology with their unique axonal arbors, the affinity propagation algorithm correctly picks up this assumption. our observations may also be explained by potential sources of error. with a smaller dataset, the affinity propagation algorithm greatly decreased in accuracy. an additional error found in the dataset was that some of the measurements of input resistance were incorrectly normalized. the values less than 1 for resistance are reported in giga ohms while the values greater than 20 are reported in mega ohms. despite the moderate success of affinity propagation algorithm, we have insufficient evidence to reject it as an exploratory tool for neuronal classification. with a large enough dataset encompassing a much greater number of interneuron subtypes, we may be able to improved classification by affinity propagation. such a dataset is in progress as we are manipulating the windows application programming interface (api) in an effort to automate the extraction of morphological data from neuroexplorer. we have automated the clicks necessary to obtain the data of 67 morphological variables from a single neuron but we would like to implement this for the entire extraction process. our goal is to create a dataset encompassing an estimated 1000 neurons of diverse interneuron subtypes, each characterized by an anatomical 3d reconstruction and whole-cell patch-clamp recording. one great issue with classification schemes of neocortical neurons is that many markers, often transcription factors, exist to label neuroscience columbia undergraduate science journal open-access publication | http://cusj.columbia.edu spring 2014 | volume 8 28 interneuron subtypes while there are few known markers to label pyramidal neurons. moreover, pyramidal neurons account for a much larger percentage of the neocortex compared to interneurons (~ 4x as much). in the future, we would like to create an algorithm that separates pyramidal neurons from interneurons and will continue to research ways to classify pyramidal neurons. affinity propagation and additional machine learning techniques have the potential to serve as powerful exploratory tools to build such classification schemes. references 1. ascoli, g.a.; alonso-nanclares, l.; anderson, s.a.; barrionuevo, g.; benavides-piccione, r.; burkhalter, a.; et al. (2008). petilla terminology: nomenclature of features of gabaergic interneurons of the cerebral cortex. nat. rev. neurosci. 9, 557–568. 2. cauli, b.; audinat, e.; lambolez, b.; angulo, m.c.; ropert, n.; tsuzuki, k.; et al. (1997). molecular and physiological diversity of cortical nonpyramidal cells. j. neurosci. 17, 3894–3906. 3. chattopadhyaya, b., di cristo, g., wu, c.z.; knott, g.; kuhlman, s.; fu, y.; et al. (2007). gad67-mediated gaba synthesis and signaling regulate inhibitory synaptic innervation in the visual cortex. neuron 54, 889–903. 4. defelipe, j. ; lópez-cruz, p.; benavidespiccione, r.; bielza, c.; larrañaga, p.; anderson, s.; et al. (2013). new insights in the classification and nomenclature of cortical gabaergic interneurons. nat. rev. neurosci. 14, 202–216. 5. douglas, r.j. and martin, k.c. (2004). neuronal circuits of the neocortex. annual reviews of neuroscience 27, 419-451. 6. frey, b.j. and dueck, d. (2007). clustering by passing messages between data points. science 315, 972–976. 7. helmstaedter, m.; sakmann, b.; and feldmeyer, d. (2009). l2/3 interneuron groups defined by multiparameter analysis of axonal projection, dendritic geometry, and electrical excitability. cereb. cortex 19, 951– 962. 8. karagiannis, a.; gallopin, t.; david, c.; battaglia, d.; geoffroy, h.; rossier, j.; et al. (2009). classification of npyexpressing neocortical interneurons. j. neurosci. 29, 3642– 3659. 9. karube, f.; kubota, y.; and kawaguchi, y.; (2004). axon branching and synaptic bouton phenotypes in gabaergic nonpyramidal cell subtypes. j. neurosci. 24, 2853–2865. 10. liu, j.h.; hansen, d.v.; kriegstein, a.r.; development and evolution of the human neocortex. cell. 2011 july 8. 146(i): 18-36. 11. ma, y., hu, h., berrebi, a.s., mathers, p.h., and agmon, a. (2006). distinct subtypes of somatostatin-containing neocortical interneurons revealed in transgenic mice. j. neurosci. 26, 5069–5082. 12. mcgarry, l.m.; packer, a.m.; fino, e.; nikolenko, v.; sippy, t.; and yuste, r. (2010). quantitative classification of somatostatinpositive neocortical interneurons identifies three interneuron subtypes. front. neural circuits 4:12. 13. packer, a.m.; and yuste, r. (2011). dense, unspecific connectivity of neocortical parvalbumin-positive interneurons: a canonical microcircuit for inhibition. j. neurosci. 31, 13260–13271. 14. oliva, a.a. jr.; jiang, m.; lam, t. ; smith, k.l., ; and swann, j.w.; (2000). novel hippocampal interneuronal subtypes identified using transgenic mice that express green fluorescent protein in gabaergic interneurons. j. neurosci. 20, 3354–3368. 15. ramón y cajal, s. (1911). histologie du systeme nerveux de l'homme et des vertebres. vol. 2. paris: maloine. 16. taniguchi h.; lu j.; huang z.j.; (2013). "the spatial and temporal origin of chandelier cells in mouse neocortex". science 339 (6115). 17. woodruff, a.; xu, q.; anderson, s. a.; and yuste, r. (2009). depolarizing effect of neocortical chandelier neurons. front. neural circuits 3:15. neuroscience columbia undergraduate science journal open-access publication | http://cusj.columbia.edu spring 2014 | volume 8 29 appendices appendix 1: electrophysiological variables. action potential properties measured from response to twice threshold, 500-ms current injection from first action potential (ap1) and second action potential (ap2). ap2 variables not listed as the same measurements were made for ap2 as listed for ap1. appendix 2: morphological variables. variables were extracted using the neurolucida explorer program by microbrightfield. neuroscience columbia undergraduate science journal open-access publication | http://cusj.columbia.edu spring 2014 | volume 8 30 appendix 3: using 20 electrophysiological variables, 337 interneurons were visualized by creating the scatter plot matrix above of the two principal components of the neuronal data. appendix 4: using 20 electrophysiological variables, a color map was made of 337 interneurons and hierarchical clustering was done, revealing 3 distinct clusters as did affinity propagation. neuroscience columbia undergraduate science journal open-access publication | http://cusj.columbia.edu spring 2014 | volume 8 31 cherenkov telescope array mirror alignment report francis toriello1* 1the veritas group, department of physics, columbia university, new york, ny, usa introduction the cherenkov telescope array (cta) is a groundbased observatory for gamma rays with very high energy characteristics within the range of 10 gev to 100 tev. it is currently still under development. as the title of the project suggests, it takes advantage of the cherenkov radiation that is produced by the gamma ray’s interaction with the earth’s upper atmosphere, similar to the guiding physical principles of veritas. the primary differences are that cta will feature more telescopes and possess a larger energy range. cherenkov radiation will be collected by an array of dozens of optical telescopes of various sizes. these optical telescopes will be composed of hexagonal mirrors that must be aligned relative to each other so as to represent the shape of the telescope as a whole properly. it was in this area of research that my efforts were focused. the most recent proposed system with which to align the mirrors consists primarily of a laser and a webcam. sixty to three hundred snapshots are taken of the laser and the position of the laser’s centroid in each frame is measured and analyzed by a custom-written computer program executed via root. one to ten microns was the ideal resolution that was required of the system to meet the needs of this project. since the position of the laser’s centroid was given by pixel number, it was necessary to determine the ratio of pixels to millimeters via another root program described in more detail on the following pages. it was also necessary to ensure by using an opencv analysis program that the pixels were not saturated. as will be shown, the consequences of saturation are bothersome edge effects that produce erroneous results. materials and methods apparatus (basic setup) 1. standard webcams (various models) 2. voltage supply 3. laser diode 4. l.e.d. 5. iris 6. calorimeter 7. diffuser 8. actuators results the first task at hand was to determine if the pixels were saturating, since this was the prime suspect of the following odd results: copyright: © 2014 the trustees of columbia university, columbia university libraries, some rights reserved, toriello. received 1/1/2014. accepted 3/14/2014. published 3/14/14. *to whom correspondence should be addressed: francis toriello, columbia university, new york ny 10027, email: fet2105@columbia.edu. physics columbia undergraduate science journal open-access publication | http://cusj.columbia.edu spring 2014 | volume 8 32 the data shown above represent a series of three hundred successive photos of a static laser with an applied voltage of 4v and a calorimeter setting of 2mm. as may be seen, the centroid does not appear to be very well behaved. in fact, it appears as though it moved dramatically in the first 100 frames as may be seen in the upper right hand graph. as previously mentioned, the prime suspect for this odd result was pixel saturation. below is one of the photographic images of this 4v 2mm laser point rendered by the webcam itself. the pure white center represents clear evidence of saturation. the more disconcerting feature of this, however, is the ring of reflected light visible around the edge of the iris. these are edge effects due to saturation that cause the centroid to appear to move when it has, in fact, not done so. this problem was made most apparent after writing an opencv executed program that analyzed this very image and produced a histogram consisting of the amount of pixels versus the intensity values they maintain (see image below): as can be seen, there is an initial spike at zero intensity. this spike represents the pixels that make up the area outside the iris while the spike all the way to the right, at the 255 mark, represents the saturated pixels that can be found at the center and throughout the image. the ideal (and necessary) situation is to have no pixels saturating or otherwise stated at the 255 intensity mark in the above graph. at the time we were using a calorimeter that possessed a smallest possible aperture of 0.1mm. we found that, even with this smallest aperture and various voltage values, there was still saturation that, in effect, would produce erroneous results as shown previously. we were finally able to eliminate saturation by using a different calorimeter that possessed a smallest possible aperture of zero millimeters. we used the very same 4v laser except with an aperture of less than 0.1mm, achieving successful results shown below. physics columbia undergraduate science journal open-access publication | http://cusj.columbia.edu spring 2014 | volume 8 33 while the exact values accompanying the above image are admittedly difficult to decipher, it is clearly apparent in the first two upper left hand graphs that the laser’s centroid is much more defined and better behaved. in fact, the sigma value for the x value of centroid is 0.06657+/0.00215 pixels while the sigma value for the y value is 0.1029+/-0.0040 pixels. this discrepancy is bothersome since both the x and y value should be the same. however, this could be due to a slight tilt of the camera that would affect the geometry in such a way as to elongate the spot subtly in the y direction. the x sigma value, however, is auspicious as it falls within the few micron range that is required of the position resolution we are seeking. the next task involved determining the pixel to millimeter ratio so we could obtain a better sense of the positional resolution provided by the sigma values previously mentioned and whether or not the pixels are square. this required performing two linear scans in which the camera is moved by the actuators in increments of .5mm in the x direction and then once again in the y direction. the slope of the fitted line should yield the amount of pixels per millimeter and both slopes should be equal to one another if the pixels are indeed square (see images below). as may be seen, both linear scans produce an approximate slope of 48.31 pixels/mm showing that the pixels are square and that our x sigma value of 0.06657 pixels translates to approximately one micron while our y sigma value of 0.1029 pixels translates to approximately two microns, both well within the range of our desired positional resolution. the second graph of each scan (position vs. deviation from fit) gives us an idea of the mechanical uncertainty of the apparatus; that is the uncertainty introduced once the apparatus is set in motion. we see that in both physics columbia undergraduate science journal open-access publication | http://cusj.columbia.edu spring 2014 | volume 8 34 scans there is a spread amongst the values of approximately .5 pixels or 10 microns. of course, this is subject to change as the setup we were using is far from the configuration that will eventually be used in the field. once the pixel to millimeter ratio was determined, the next task required us to confirm the apparent positional resolutions provided by the sigma values. this was achieved by performing a scan in the x direction similar to that of the previous linear scan. however, the difference this time was that, instead of keeping constant the increments by which the camera moved, we instead halved the distance every time. if our positional resolution was indeed approximately one micron then, once the increments reach one micron and below, the program should fail to recognize any change in position (see image below). as may be seen in the second position vs. deviation from the fit graph, there is a point very close to two millimeters where we reach the limit of our positional resolution. that position is actually 1.999023188 mm, exactly the point where the change in position dips below one micron (9.765625 x 10^-4 mm) as was apparent in the table of data used to create these graphs. therefore these results show that the positional resolution achieved was, in fact, approximately one micron (with regards to the x value of the centroid). to get a better idea of the mechanical uncertainty, a series of diagonal scans were performed. just as the name implies, these were contemporaneous scans in both the x and y direction. we should see the same mechanical uncertainty of approximately 10 microns or .5 pixels that had been reflected in the linear scans. y-centroid behavior: x-centroid behavior: notice that, similar to the linear scan in the x-direction, the x-centroid behavior yields a mechanical uncertainty of approximately .5 pixels or 10 microns (given by the spread of points in the position vs. deviation from fit graph) while the physics columbia undergraduate science journal open-access publication | http://cusj.columbia.edu spring 2014 | volume 8 35 y-centroid behavior yields a mechanical uncertainty of approximately 3.5 pixels or 73 microns. this discrepancy is similar to the difference between the x sigma and the y sigma mentioned earlier. it was determined that this inconsistency could also be due to some very slight tilt of the camera that may affect the geometry of the how the laser beam impinges on the camera’s lens. the next task required testing other webcams of similar design in order to evaluate the robustness of the current configuration of the apparatus. for the sake of brevity i have excluded these results as they were very similar to previous ones. the last task we completed before the end of the semester was the evaluation of an alternate light source, a light emitting diode. due to the fact that the light emitted from an led is considerably more spread and less concentrated, it was necessary to build a makeshift housing unit in which we passed the light though two custommade calorimeters, both of which, of course, possessed an aperture of less than 0.1mm. we proceeded to taking sixty successive images of the led laser point and analyzing them in much the same fashion as we had done for the three hundred successive photos of the 4v laser diode. the results are no doubt promising as, much like the laser diode, a well-defined centroid is reflected in the analysis as can be seen in the first two upper left hand graphs. we get an xsigma value of 0.0666 +/0.0161 pixels or once again approximately one micron and a y-sigma value of 0.09032 +/0.01511 pixels or once again approximately two microns. discussion in our final analysis, the use of webcams while promising is not without its discrepancies, such as the difference between x and y positional resolutions given by their respective sigma values, and it reflected some mechanical uncertainties. it is of interest to me why, when the linear scans were not diagonal but done separately and dedicated to one direction at a time(x or y), the same mechanical uncertainty of about 10 microns was reflected by both the x and y scans while this did not occur for the diagonal scans. if the geometrical argument that i have proposed throughout this report is correct, then perhaps some corrective measure may be investigated during further research anticipated to be done in months to come. with regard to most recent developments concerning the leds, it would be interesting to perform some diagonal and linear scans to not only gain a more advanced understanding of how we may use these light sources, but also to see if these discrepancies persist, and in effect learn more about them and the subtleties of the system as a whole. physics microsoft word cusj 21.docx ________________________________________________________________________________ cusj 2019 __________________________________________________________________________________________ 21 abstract — nilotinib (tasigna) is a drug used to treat chronic myelogenous leukemia (cml), a disease which currently affects over 100,000 people in the u.s. the drug is manufactured by novartis and has been one of its most profitable, bringing in a total revenue of well over $2 billion usd in 2016. novartis’s patent on tasigna is set to expire in 2023 and cheaper, generic versions of the drug will soon be in great demand. an assessment of novartis’s existing patent on the synthesis of tasigna indicates many drawbacks in the current process. namely, the novartis synthesis requires multiple days of reaction, nearly a dozen steps, multiple solvent switching processes, and gives very low yields. on the other hand, work published by buchwald and coworkers has shown an efficient method of synthesizing nilotinib in fewer than four steps, in less than 24 hours, and at >85% yields. unfortunately, the buchwald reaction has only ever been applied in a laboratory setting and has never been scaled to an industrial level. thus, it was the goal of this investigation to develop an industrially scaled, flow-based synthesis of the buchwald synthesis of nilotinib. for this investigation, our team assumed the role of a generic pharmaceutical company with a hypothetical goal of producing 15% of the u.s. market share of nilotinib (1150 kg of nilotinib/year). utilizing chemical reaction modeling software, chemcad, a chemical process for the large scale buchwald synthesis of nilotinib was successfully designed at the 1150 kg production capacity of the active pharmaceutical ingredient of nilotinib per year. economic analysis demonstrated the fiscal viability of this process with estimated gross revenue of >$250 million usd/year even with a predicted 60% drop in generic drug price factored in. this, we envision, would serve as an important improvement in the efficiency of industrial level synthesis of a life-saving cancer drug. i. introduction the use of tyrosine kinase inhibitors (tkis) to treat chronic myeloid leukemia (cml) has been one of the greatest breakthroughs in modern oncology. the advent and introduction of tkis into the clinic has extended 8year survival rates of patients with cml from <15% in the 1980s to nearly 90% in 2012 (kantarjian et al., 2012). even today, over 10,000 people each year are newly diagnosed with cml and the vast majority, if not all, of these patients will rely on tkis (national cancer institute, 2014). the patent for novartis’s original first-generation tki, known as gleevec (imatinib), expired in 2016. to mitigate its loss of profit and market share, novartis developed a second-generation tki known as tasigna (nilotinib), shown in fig. 1. tasigna is designed to be more potent in treating cml than its first-generation counterpart and also addresses many of the challenges of chemotherapy resistance that were faced by gleevec (huang et al., 2007). tasigna received its final fda approval to treat cml in 2010. however, novartis’s patent on tasigna is set to expire in 2023. this makes it a particularly opportune time to plan the development of generic versions of tasigna so that once novartis’s patent on the production of tasigna expires, generic production can begin immediately. in 2015 tasigna represented 12.8% of the current market share of all tki drugs for cml (bloudek et al., 2015). additionally, it is currently predicted that the incidence of cml will increase with an estimated prevalence of 112,000 patients in the us by 2020 (cortes et al., 2012). by the year that novartis’s patent on tasigna expires, it would be expected that there will be at least 140,000 patients with cml in the us in need of treatment. the cost on the patient to take tasigna for one year is approximately $100,000, highlighting the importance of quickly bringing generic versions of the drug to market once the novartis patent expires in order to reduce patient financial burdens. the large reliance of cml patients on tki drugs in addition to the prohibitively expensive price of current tasigna treatments highlights the great market potential and demand for developing a generic, cheaper version of tasigna and thus, will be the focus of our investigation. figure 1. structure of nilotinib. here, we outline our unique process design for an industrially scaled, flow-based process for the production of generic tasigna (nilotinib). the ultimate goal of this investigation will be to computationally develop a scalable synthesis which could potentially be used to produce large quantities of nilotinib with a greater degree of efficiency than the existing novartis method. thus, once the novartis patent over nilotinib production expires, efficient methods of producing generic tasigna can be implemented to rapidly capture the large and profitable market which tasigna currently holds. this will allow for the increased availability and lower price of a much needed drug to tens of thousands of cml patients. for this investigation, our team has taken the hypothetical role of a generic pharmaceutical company in planning the development and production of generic tasigna. as such, this investigation consists of four designing an industrially scalable, flow-based synthesis for the production of nilotinib: chemical process modeling and economic analysis justin m. cheung*, jessica hofflich, pablo henriquez, jian huang, and nafiul jami department of materials science & chemical engineering, stony brook university cusj 2019________________________________________________________________________________ __________________________________________________________________________________________ 22 distinct sections: (1) an assessment of the commercial efficacy of novartis’s existing production methods of nilotinib, (2) literature review assessing the efficiency and scalability of existing, laboratory level (small scale) synthesis methods of nilotinib, (3) selection of an existing small scale nilotinib production process and modeling its scaling to an industrial level using chemcad software, and (4) an economic analysis of the scaled process to assess its fiscal viability and profitability. the results of this investigation have the potential to provide a robust framework for the large-scale development of nilotinib upon the expiration of the novartis patent on tasigna. ii. methods a three-stage literature search was initially con ducted to address the following goals: firstly, market data on nilotinib drug pricing was collected in order to develop robust predictions of feasible generic pricing patterns as well as market share targets for our given scenario. this included a comprehensive search of articles pertaining to the prevalence of specific tki usage in cml patients, marketing of each of these drugs, and pricing before and after generic introduction. the literature search mainly utilized the nci (national cancer institute) database as well as pharmaceutical pricing literature. secondly, due to the lack of an existing published process flow diagram, all of novartis’s u.s. patents for the production of tasigna were collected and a predicted process flow chart of novartis’s current synthesis method was constructed based on the claims outlined in their filed patents. thirdly, all the currently published methods of synthesizing nilotinib at any scale were collected via literature search and analyzed for their efficiency across seven distinct criteria: (1) number of steps, (2) reaction temperatures, (3) reaction pressures, (4) reaction yields, (5) catalysts required, (6) solvents required, and (7) reaction time. once the most efficient of these methods was selected, process design modeling of the industrially scaled nilotinib synthesis was conducted using chemcad (computer-aided design) 7 software (chemstations). this allowed for the quantitative assessment and control of large flow rates of material, heat transfers, and separation efficiencies, allowing us to judge the success of the industrial scaling process. during the process of envisioning the design through chemcad, special considerations were made to ensure that the process remained “flow-based” eliminating the need for any manual handling of reaction intermediates and relying on a single series of reaction chambers. this was achieved using packed bed reactors (pbrs) in the model which allow for connections between parallel reactions and the use of solid catalysts without the need for interruption of the flow process. the chemcad software was also utilized to verify that the proposed reaction schema remained thermodynamically viable throughout and that the output products were high in purity. calculations of heat and material balances done via chemcad across each chemical reaction stage ensured that the reactions were well controlled, in the predicted state of matter, and exited each stage at the appropriate temperature and purity. lastly, economic analysis incorporated an assessment of the estimated costs of running the chemical production facility and raw material costs balanced against the predicted profits of the drug. reactor, piping, and distillation tower cost estimates were made through two american society of mechanical engineers (asme) certified reactor manufacturers: dominion tank & vessel co., inc. based in richmond, virginia and procedyne inc., based in new brunswick, nj. iii. results and discussion a. chemical process model design rationale & market scale assessment a recent study on patient cml drug preferences indicated that approximately 20% of all cml patients take tasigna (nilotinib) as opposed to other tki drugs (sanford et al., 2014). assuming that this trend continues, we can predict that the total cml patient pool taking nilotinib by the year 2023 is approximately 28,000 individuals. based on historical precedent, as an emerging generic pharmaceutical company, our goal would be to capture approximately 15% of the nilotinib market share in the us, translating to about 4200 patients. the food and drug administration (fda) approved regimen for treatment of cml with nilotinib requires a dose of 0.3 mg twice daily (or approximately 230 g of nilotinib per patient per year) (national cancer institute, 2013). thus, if we wish to pursue 15% of the nilotinib market share, production of at least 1150 kg nilotinib/year is the ultimate goal. the current market price of tasigna is ~$600 per gram (or $600,000 per kg). at current pricing, the estimated profit for 1150 kg/year of is $690 million/year. however, once nilotinib comes off patent and is generically produced, we can expect a sharp price drop. we developed our estimates of this change based on the price drop that gleevec (the 1st generation version of tasigna) underwent when it came off patent. in 2016, gleevec cost ~$146,000/year. since coming off patent, the cheapest generic gleevec option is ~$57,000/year (kantarjian, 2016). this is a 60% decrease in price from original to generic form of the first generation tki drug. we will assume that this pattern (60% drop in price) will also hold for tasigna/nilotinib, therefore, we can assume that tasigna’s generic will be priced at $240 per gram and that the estimated revenue for 1150 kg/year nilotinib production is $276 million/year. in summary, our generic pharmaceutical company will aim to take 15% of the nilotinib market share when the drug comes off patent in 2023, equating to about 4200 patients. to meet this demand, our goal is to design a chemical process for nilotinib production with a minimum capacity of 1150 kg per year. given, an expected 60% lower price of generic, the cost of our generic nilotinib will be ~$240/gram and our total yearly revenue is predicted to be $276 million. ________________________________________________________________________________ cusj 2019 __________________________________________________________________________________________ 23 critical assessment of the novartis synthesis for nilotinib novartis has not made a process flow sheet for its synthesis of nilotinib publicly available. to circumvent this, our team has utilized the steps outlined in novartis’s patent to create our own process flow sheet of the novartis synthesis shown in fig. 2 (breitenstein et al., 2004). fig. 2 overviews novartis’s patented process chemistry for the synthesis of nilotinib which is currently used for the industrial production of tasigna (deadman et al., 2013). from the process flow diagram, it can be seen that the production of nilotinib consists of three main parts. the first part (the top portion of the reaction scheme) is required to produce the carbon-nitrogen bond (fig. 2). the second part is the nucleophilic substitution which occurs concurrently to produce the required inter mediate in the bottom portion of the reaction scheme (fig. 2). with both parts completed, the third and final amide formation occurs during the last synthesis step to nilotinib formation (fig. 2). although this process chemistry for nilotinib synthesis is currently in use for its commercial production, there are four distinct drawbacks of novartis’s currently patented method. (1) during the production of the intermediate at the top of the reaction scheme (fig. 2), the use of nucleophilic substitution allows for the possible formation of two intermediate isomers. the formation of these two distinct isomers means that regioselectivity of the reaction becomes a great challenge since only one of thregioisomers is desired (deadman et al., 2013). as such, this would mean that novartis’s non-regioselective method for the synthesis of this particular intermediate would require a great deal of subsequent separation and purification to obtain only the desired regioisomer, steps that diminish efficiency and induce great time and monetary penalties on the production process. (2) from fig. 2 it is evident that the entirety of the novartis synthesis is an extremely lengthy multi-stage process. for instance, the reflux stage of the reaction takes over 68 hours, alone. (3) the process’s use of multi-stage batch reactors is highly unfavorable due to the time-consuming transfer requirements that will inevitably also require a great deal of manpower, further prolonging the total length of the reaction. in addition, separation procedures such as evaporation and filtration will likely have to be applied after each batch reaction since solvent switching will be required. these separation techniques will be clumsy and greatly reduce the degree of process control over the reaction. (4) from a hazard and operability (hazop) stand point, the overall novartis synthesis is particularly dangerous as it requires many steps and solvents. this presents a concern for health and safety, particularly since the use of multi-stage batch reactor processes will create a large degree of exposure risk to chemical plant workers who will have to directly deal with carcinogenic solvents such as dioxane and toluene during mass transfer operations. while novartis does not publicly disclose their production processes, this entire synthesis would take over 100 hours in reaction time and an estimated 50 hours in cleaning and separation downtime. therefore, novartis’s primary pathway method is an expensive and time-consuming process with undisclosed yields that are expected to require additional purification steps. review of the alternative nilotinib synthesis pathways although the novartis synthesis for nilotinib is functional, there has been much effort in developing alternative syntheses which mitigate the drawbacks present in the novartis synthesis pathway. the five known paths for nilotinib synthesis were compiled and are outlined in table 1. in addition to the aforementioned weaknesses to the novartis synthesis, the other syntheses (table 1) also feature several key weaknesses affecting their practicality and profitability. the buchwald synthesis requires many solid catalysts. however, this is balanced by the minimal number of steps and the significantly shorter reaction time required (24h) in comparison to the novartis method figure 2. process flow diagram of novartis’s synthesis pathway of nilotinib. cusj 2019________________________________________________________________________________ __________________________________________________________________________________________ 24 (>100h) (ueda et al., 2012). the chen synthesis has been found to have long reaction times with certain steps taking 68 hours to complete, while the total reaction time for the process was over three days. furthermore, the chen synthesis features a very low overall yield, around 40%. additionally, the copper catalyst used in step 1a of the chen synthesis is particularly slow and non-regioselective, resulting in the need for additional separation in order to isolate the key aniline intermediate, further complicating an already unwieldy process (chen et al., 2009). the teva synthesis likewise was found to be exceptionally long, taking over five days total. additionally, the many solvents used can complicate the unit operations involved due to the extensive solvent switching they necessitate (yeori et al., 2010). lastly, the ariad synthesis, although brief at only four steps, was reported to have be unscalable, as its final step has been found to give inconsistent yields at scales above 250 milligrams, resulting in impracticality at an industrial scale (deadman et al., 2013). table 1. a comparison of the five existing nilotinib syntheses according to their number of steps. the associated reaction temperatures, reaction pressures, step yields, solvents, catalysts, and reaction times are also noted with respect to each reaction stage. of the syntheses of nilotinib that were discussed in table 1, the buchwald synthesis seemed by far the most promising in terms of developing a scalable, flow-based synthesis process of nilotinib for the production of a generic brand. figure 3. reaction scheme for the buchwald synthesis of nilotinib, modified from ueda et al., 2012. the advantages of the buchwald synthesis that make it the most promising are its three steps taking approximately 24 hours of reaction time, fast and efficient compared to the other existing methods of nilotinib synthesis. more importantly, the buchwald synthesis is highly regioselective in the step where an important aniline compound is produced (highlighted in the red box in fig. 3). this gives the buchwald synthesis a distinct advantage over other methods of synthesizing nilotinib and more importantly, reduces the need for time-intensive regioselective purification at this initial step. lastly, the buchwald synthesis requires only four, readily purchasable starting reactants: (1) 3-amino-5-bromobenzotrifluoride (2) 4-methylimidazole (3) 4-pyrimidin-3-yl-pyrimidine-2-amine (4) methyl 3-bromo-4-methylbenzoate thus, we chose the buchwald synthesis to be the focus of our computational process scaling methodology. process design for an industrial scale buchwald synthesis of nilotinib our goal was to develop a scalable, flow-based synthesis of nilotinib which has greater time efficiency, safety, and yield than the existing novartis synthesis. we hoped to employ a flow-chemistry based synthesis of nilotinib. in contrast to batch production methods, a flowbased synthesis would involve the use of a single, continuous flowing stream wherein the reaction would take place in a series of controlled steps without the need for multiple batch reactors. as reported in previous literature, multi-step batch processes require process control and purification techniques which are highly complex in comparison to flow-based reactions (hopkin et al., 2010). as such, a general process flow diagram which illustrates the changes and modifications we wish to make to the existing buchwald synthesis is shown in fig. 4. ________________________________________________________________________________ cusj 2019 __________________________________________________________________________________________ 25 figure 4. envisioned process flow diagram for the flow-based buchwald synthesis of nilotinib using packed bed reactors. firstly, we re-designed the reaction scheme of the buchwald synthesis to take place under flow chemistry conditions (rather than a multi-stage batch series). this should allow for: (1) greater efficiency by reducing manual mass transfers between batch reactors, (2) greater degree of process control over the synthesis wherein controlling reaction flow rates will be the key concern, (3) greater degree of safety by minimizing the need for reactor transfers and thus eliminating unnecessary exposure to dangerous solvents, (4) greater scalability of the reaction to an industrial scale. the buchwald synthesis requires palladium and cesium solid-state catalysts at several stages. to address this challenge within the scope of the flow-based model and make the use of such catalysts feasible on an industrial scale, we have designed for the implementation of packed bed reactors (pbrs) into our process. this will allow for the large-scale use of the solid-state catalysts in the reaction under flow chemistry conditions while also facilitating catalyst recovery and cleaning. fig. 5 illustrates the final process flow sheet that was constructed in chemcad. it illustrates the components entering the pbrs in each stage of the reaction and distillation towers for separation. the model successfully converged with respect to the mass and energy balances and was able to produce nilotinib at a desired rate that would allow us to achieve the 1150 kg /year design basis goal. furthermore, according to the chemcad model, the final distillation tower of nilotinib production achieved over 98% separation, indicating the robust final purity of our desired nilotinib product. to produce our desired output of 1150 kg of nilotinib per year, we approximate that our plant will need to run for at least 350 days of the year. assuming constant production, a minimum output of 137 g/hour (0.263 moles) of nilotinib is required. due to the conversion rates of 90%, 95%, and 90% for each of the three stages of the buchwald synthesis, respectively, that translates to a basis of approximately 0.325 mol/hr for the starting reactants of both initial reactions, with mole equivalencies accounted for appropriately. a table detailing the mass balance from our reactor system designed in chemcad is shown in table 2. here, an output of 138 g/hr is shown, successfully achieving our design basis of 1150 kg per year. additionally, very little accumulation is shown, as evidenced by the near identical input/output masses. the energy balance from the reactor process, including energy added by each distillation tower, is also shown in table 3. figure 5. final process flow sheet for synthesis of nilotinib as constructed by chemcad. cusj 2019________________________________________________________________________________ __________________________________________________________________________________________ 26 table 2. chemcad report of overall mass balance. table 3. chemcad report of overall energy balance. similar to the mass balance, it is evident that an almost perfect energy balance is achieved, indicating the convergence success of our chemcad model. this mass and energy balance from the chemcad model serves to highlight that our industrial scaled synthesis of nilotinib using the buchwald synthesis was successful and that, computationally, there is strong evidence to indicate that this process can be scaled to effectively produce at least 1150 kg of nilotinib per year. given our flow-based design and the nature of the buchwald synthesis itself, this represents a great improvement with respect to the time, safety, and ease of nilotinib production as compared to the existing novartis methodology. b. process economics process economics of the scaled buchwald synthesis i. capital and operating costs to calculate profitability of the process, considered the capital cost of procuring and maintaining a manufacturing facility and the operating cost of the process based on our production design basis of 1150 kg of nilotinib. we estimated a small-scale manufacturing facility will cost ~ $2 million usd and an additional ~$2 million to have the facility cgmp certified for production. direct costs associated with production include employee cost, utility cost and waste management cost. we evaluated our employee needs based on existing industry data. based on estimations for a small-scale production facility and accounting for employee benefits and human resources cost, our estimation for employee cost was $1,100,000$1,400,000 usd per year. we have also predicted a budget of $50,000 for the annual maintenance and servicing of the equipment. to implement a system to dispose of the organic waste in an environmentally friendly way in accordance with epa and osha regulations, we estimate the cost of a waste management system for our design will be $200,000 on an annual basis. based on our energy and water needs, we will require a budget of $500,000 for yearly utility cost. ii. reactor, piping, and distillation tower costs the design of the synthesis process for nilotinib will require properly sized industrial reactors and piping. the pbrs must be capable of resisting the organic solvents and the catalysts that will be reacting inside the vessels. 316l grade steel is the pharmaceutical industry standard for contact with drug-based flow to avoid contamination. the total cost of three 0.15 m3 316l steel pbrs is $75,600. it is estimated from our process flow diagram that at least 300 feet of 316l steel piping at 1” piping size will cost $8,850. the thickness of the piping is appropriate for the expected flow rates and the pressure drops expected from the system and confirmed with chemcad. the distillation towers that our process will require for the separation and purification of products are expected to cost $55,100 in total. thus, the overall equipment cost is estimated to be ~ $140,000. this relatively low cost is achieved through the low process flow rate as the active ingredient will be produced in small amounts throughout the year and does not require enormous vessel capacities. iii. starting reagent costs we have also calculated the costs of purchasing the four starting material components of the buchwald reaction as shown in table 4. the total molar amount and subsequent mass of each reactant required was calculated based on the process production demands. the total estimated annual costs of purchasing the starting components of the buchwald synthesis is estimated to be around $5,500,000. table 4. material economics of starting material for nilotinib production. overall costs and final profit calculation for the first year of production, the overall process cost will be near $12 million, after which the overall costs drop to around $8 million as shown in table 5. considering the annual projections of a profit of $276 million dollars, capital and operating expenditure only accounts for 4.2% and 2.8% of revenue in the first and ________________________________________________________________________________ cusj 2019 __________________________________________________________________________________________ 27 second year, respectively. altogether, in balancing our predicted revenue with predicted capital/operating expenditures, yearly profits greater than $250 million are expected if the design basis criteria are met and 15% of the nilotinib market share can be captured. table 5. table of capital and operating costs. iv. conclusions in this investigation, we have described a new method for the large scale production of nilotinib, an important chemotherapy drug. current novartis practices of nilotinib synthesis require upwards of 8 steps and 97 hours of reaction time. by contrast, a recently developed approach by buchwald and coworkers allows for the synthesis of nilotinib in 3 steps and 24 hours. until now, this approach had only been designed for small scale (< 1 g) synthesis of nilotinib in laboratory settings. this investigation provides a framework for which the buchwald synthesis of nilotinib can be scaled up to produce 1150 kg per year much more efficiently than novartis’s current methods. we adopted the role of a generic pharmaceutical company seeking to produce nilotinib upon novartis’s patent expiration in 2023. our goal of capturing 15% of nilotinib’s market share allowed us to calculate our needed production to be 1150 kg/year with projected revenue of $276 million/year (assuming 60% price drop in generic drugs). using chemcad software, a new process flow diagram was constructed for the scaled production of nilotinib using the buchwald synthesis. our approach focused on the use of plug flow reactors in order to achieve a flow-based synthesis and improve the overall efficiency and safety of nilotinib production, minimizing the need for solvent switching or complex unit operations for mass transfers. our chemcad program successfully converged and mass and energy balances across the system were achieved. this serves to demonstrate that it is possible to industrial scale the buchwald synthesis while maintaining profitability. this we hope serves to establish an initial framework by which the generic form of a life saving medication can be produced safely and efficiently. acronyms asme – american society of mechanical engineers cad – computer-aided design� cml – chronic myelogenous leukemia cgmp – current good manufacturing practices epa – environmental protection agency fda – food and drug administration hazop – hazards and operability nci – national cancer institute osha – occupational safety and health administration pbr – packed bed reactor tki – tyrosine kinase inhibitor author information corresponding author *justin.cheung@stonybrookmedicine.edu author contributions all authors contributed equally to the design, modeling, analysis, and write up of the article. funding sources no funding sources to disclose. acknowledgment we would like to thank professor devinder mahajan for his guidance on this project. references [1] h. kantarjian, s. o’brien, e. jabbour, et al., and j. cortes, blood, 2012, 119(9), 1981–1987. http://www.bloodjournal.org/content/117/6/1822.lo ng?sso-checked=true [2] national cancer institute, cancer stat facts: leukemia chronic myeloid leukemia, surveillance, epidemiology, and end results program, 2014. https://seer.cancer.gov/statfacts/html/cmyl.html [3] w. huang, w. c. shakespeare, synthesis, 2007, 14, 2121-2124. https://www.thiemeconnect.com/products/ejournals/html/10.1055/s2007-983754 [4] bloudek, l. m., makenbaeva, d., & eaddy, m. am. health drug benefits, 2015, 8(9): 472–480. https://europepmc.org/abstract/med/26834933 [5] j. cortes, x. huang, h. kantarjian, cancer, 2012, 118(12), 3123-3127. https://onlinelibrary.wiley.com/doi/abs/10.1002/cn cr.26679 [6] d. sanford, r. kyle, a. lazo-langer, a. xenocostas, i. chin-yee, k. howson-jan, c. hsia, current oncology, 2014, 21(2), 241-249. http://www.currentoncology.com/index.php/oncology/article/view/173 6/1479 [7] national cancer institute, fda approval for nilotinib, national institutes of health, 2013. [8] h. kantarjian. the arrival of generic imatinib into the u.s. market: an educational event. the asco post, 2016. http://www.ascopost.com/issues/may-252016/the-arrival-of-generic-imatinib-into-the-usmarket-an-educational-event/ cusj 2019________________________________________________________________________________ __________________________________________________________________________________________ 28 [9] w. breitenstein, p. furet, s. jacob, p. w. manley, united states patent and trademark office, 2004, patent publication no. wo/2004/005281. https://patentscope.wipo.int/search/en/detail.jsf%3 bjsessionid=39aea7187733e9827dc0bbb00f07b9d3. wapp2?docid=wo2004005281&recnum=94&office=& prevfilter=%2526fq%253dof%253akr%2526fq%253 dicf_m%253a%252522c07d%252522%2526fq%253d paf_m%253a%252522novartis%252bag%252522 &sortoption=relevance [10] b. j. deadman, m. d. hopkin, i. r. baxendale, s. v. ley, org. biomol. chem., 2013, 11(11), 1766-1800. https://www.ncbi.nlm.nih.gov/pubmed/23247657 [11] y. chen, l. wang, h. zhou, c. wang, zhongguo yiyao gongye zazhi, 2009, 40, 401–403. https://www.tib.eu/en/search/id/tibkat%3a1823 55640/zhongguo-yiyao-gongye-zazhi/ [12] a. yeori, y. wang, j. li, j. zhu, r. lifshitz-liron, x, he, united states patent and trademark office, 2010, patent publication no. wo/2010/060074. https://patentscope.wipo.int/search/en/detail.jsf?do cid=wo2010060074 [13] s. ueda, m. su, s. l. buchwald, j. of the american chemical society, 2012, 134(1), 700-706. https://pubs.acs.org/doi/abs/10.1021/ja2102373 [14] m. d. hopkin, i. r. baxendale, s. v. ley, rsc chemcomm, 2010, 46, 2450-2452. https://pubs.rsc.org/en/content/articlelanding/2010 /cc/c001550d#!divabstract (treisman 2003) allowing us to experience a world of coherent, integrated objects instead of disembodied or wrongly-combined shapes, colors, motions, and sizes (treisman 1998). !e combined system of visual object binding and short-term memory is called “visual binding working memory”, or simply “binding memory.” it is what allows people to experience, for example, a ball "ying through the air as a single moving object, instead of as snapshots of identical balls tracing a path through space. binding memory is an indispensable part of biological motion detection. basic evidence for the existence of binding memory recent developments in binding memory research, elucidating its properties and limitations, promise to expand scienti#c understanding of biological motion detection. several experiments have con#rmed the premise of binding memory: that features of objects are #rst perceived separately before being bound together into a uni#ed object. for example, in one experiment, subjects were shown collections of squares with varying orientations that were laid out in a grid, and they were asked to recall features of the squares at particular grid locations after a short delay (bays et al., 2011). interestingly, subjects often failed to recall which features went with which object, rather than forgetting the object altogether. !is suggests that people perceive the separate features of an object individually, rather than as a single uni#ed object, and have separate categories for di$erent features (e.g. color, shape, and orientation). in a similar experiment, researchers started from a baseline of three objects with six di$erent features (wheeler et al., 2002). when they doubled the number of objects, there was no appreciable change in recall performance, but halving the number of features resulted in a signi#cant increase in performance recall. !is suggests that perception can be limited by the number of features, but not by the number of objects. !ese results were further supported by a similar study, where subjects were brie"y shown objects on a screen, followed by a blank interval, and were then asked to choose the object they had seen from an array of similar types of objects (alvarez et al., 2004). !eir capacity for remembering these objects varied widely, with a greater number of features to be recalled correlating with a decrease in performance. since the increase in the number of features but not the number of objects led to a decrease in memory performance, it is likely that that features of objects are held separately in short term memory and then subsequently integrated. !is is a limit of human binding memory: subjects’ performance in object-recall tasks depend on the number features the objects have. e!ects of old age and disease on binding memory recent studies have also explored the e$ects of old age and diseases such as autism spectrum disorders (asd) and schizophrenia on binding memory. for example, functional magnetic resonance imaging (fmri) measurements of brain activity were taken in asd patients as they performed a biological motion detection task (herrington et al., 2007). !e asd subjects showed less activity in the fusiform gyrus as compared to control subjects, an area previously implicated in the processing of visual feature binding. !ese results #t with those of another study in which autistic and normal children were asked to perform two tasks: one involving the detection of abstract shapes in a still image, and one involving the detection of biological motion (blake et al., 2003). performance for both groups was similar on the #rst task, but the autistic children performed far worse on the task with the simulated biological motion. !is supports the claim that autistic individuals generally have more trouble binding orientation and position information together through time, and thus have a harder time detecting biological motion. studies with another disease, schizophrenia, have also highlighted the critical importance of binding memory and biological motion detection by showing what happens when there is a binding memory de#cit. in a study with schizophrenic patients, visual working memory was tested for two features of drawings of familiar items: their location in a grid, and the identity of the item (burglena et al., 2003). !e test was done under two sets of circumstances: in the #rst, subjects were asked to simply recall the features independently. speci#cally, they had to say where on the grid an object had been displayed (regardless of which object), or what object had appeared (regardless of where). in the second set of circumstances, subjects were asked to recall an object and its location (a test of binding memory). although schizophrenic patients generally have impaired working memory, this experiment’s results showed that they had disproportionately poor performance in binding memory. yet another study explored the e$ects of age on binding (kessels et al., 2007). young adults (mean age of 25) and older adults (mean age of 66) were asked to view a series of familiar objects displayed on a grid. !ey were then asked to recall either just the objects, just the grid locations that were occupied, or both the object and the grid location in which it was displayed. even after taking into account the fact that memory generally declines with age, the older adults had disproportionately poor performance on the #nal task. !is suggests that binding memory performance declines markedly with age, and to an even greater degree than other kinds of memory. !e fact that autism and schizophrenia – broad and severe cognitive disturbances – correlate with binding memory problems underscores the relevance of this ability to normal human cognition (herrington et al., 2007, blake et al., 2003, burglen et al., 2003). :h�uhylhz�wkh�olwhudwxuh�wr�gdwh�rq�elrorjlfdo�prwlrq�ghwhfwlrq�dqg�lwv�lpsruwdqw�suhuhtxlvlwh��kxpdq�elqglqj�phpru\��%lqglqj� memory is the system for combining features like shape, color and size into coherent visual objects, and tracking these objects through time in short-term memory. this in turn makes possible biological motion detection, which is crucial to a wide range of human social activity. recent experiments have shown that binding memory is available from birth, declines with old age, and is improved by emotional arousal. furthermore, binding memory has been shown to be disturbed in neurological disorders such as autism and schizophrenia. these results promise to contribute to a greater understanding of biological motion detection in the future. columbia undergraduate science journal open-access publication | http://cusj.columbia.edu 23 cusjvolume 6spring 2012 columbia undergraduate sci j http://cusj.columbia.edu + 8 0 $ 1 �% ,2 /2 * < cusjcolumbia undergraduate science journal review articles 24cusj volume 6spring 2012 columbia undergraduate sci j http://cusj.columbia.edu + 8 0 $ 1 �% ,2 /2 * < biological motion detection humans have a specialized system called biological motion detection that responds specially to the movements of living creatures. more speci#cally, biological motion detection is the cognitive process whereby the perception of motion patterns characteristic of living organisms evoke a compelling impression that a living thing is being observed (johansson 1973). !is distinct neurological process is important for developing an understanding of the minds of others, and for functioning socially in general; human understanding of what is and is not a living thing is based on cues including motion – such as walking, speaking, or yawning – and social skills are predicated on the ability to readily discriminate between living and non-living things (blake et al., 2007). biological motion detection – which exists in a wide variety of animals including chickens (simion et al., 2008) and macaques (oram et al.,1994) – is functional in humans almost from birth (simion et al., 2008). when twoday-old infants were shown various rudimentary animations consisting of black dots moving around on a white background, they spent more time looking at the patterns that resembled an upright, walking animal than any of the others. because the animations used were so simple – comprised of just 13 dots and 23 looping frames of video – they are unintelligible jumbles when motionless. !e fact that the infants could discriminate di$erent moving patterns of objects from each other at birth indicates that they must already have some sort of functioning binding memory. !is result seems to #t with the fact that binding memory does not require the hippocampus, a brain area widely implicated in memory but not functional at birth. in trials with a patient with severe hippocampal damage, binding memory (as measured by colorand shape-recognition tasks described below) for the patient was no di$erent than that of control subjects (baddeley et al., 2010). !e fact that newborns can detect biological motion from the simplest patterns (simion et al., 2008) with an ability that does not depend, like so many other memory processes, on the hippocampus suggests that this is an important component of human cognition with highlyspecialized neural correlates. in typical humans, biological motion detection has been localized to a small area in the superior-temporal sulcus (sts) (grossman et al., 2000). !is brain region was found to be sensitive to motion typical of animate creatures, and showed greater activation even when the body plan was reduced to abstract, simpli#ed points of light moving in a pattern reminiscent of animal motion. despite the unitary, integrated nature of the perceptual experience of biological motion detection – and the ease with which the brain typically performs this task – this ability only arises through the coordinated e$ort of several complex systems in di$erent parts of the brain. as the points-of-light experiment demonstrated, the low-level components of biological motion are simply visual stimuli with certain shapes, orientations, motions and colors associated with them. to detect biological motion, the brain must appropriately process these components in concert. a closer look at how this processing works indicates that the brain perceives visual features (e.g. color, shape, and orientation) individually, and then uses short-term working memory to “bind” them together into a unitary experience, low-level visual processing allows humans to react to animate objects: binding memory as a key subsystem of biological motion detection andrew hamilton* copyright: © 2012 the trustees of columbia university, columbia university libraries, some rights reserved, hamilton. received 12/31/2012. accepted 2/6/2012. published 4/1/2012 *to whom correspondence should be addressed: ah2926@columbia.edu abstract e!ects of attention and emotional arousal in binding memory !e e"ects of attention and emotional arousal in the function of binding memory have been explored experimentally as well. in one study, subjects were shown shapes with distinctive colors and orientations on a computer screen (johnson et al., 2008). !ey were then asked to state either which features had been simply present (to test memory but not binding), or which features had gone with which shape (to test binding). performance was evaluated both in the presence and absence of distracting stimuli. interestingly, the presence of a distraction impaired performance equally for individual feature recall as well as bound object recall. in other words, the fact that the subjects’ being distracted did not speci#cally impede their feature binding any more than it did the processing of the features alone indicates that one need not pay attention to an object to properly process it together with all of its features. in another study, subjects were shown words of varying colors and levels of emotional signi#cance (e.g. “slaughter” or “emergency” for highly-emotional words, and “taxi” or “dormitory” for neutral words) (doerksen et al., 2001). !ey were then asked to recall either one separate feature of the object (e.g. the word), another feature (e.g. the color), or the bound object (e.g. the word and the color it was displayed in). unsurprisingly, subjects exhibited an enhanced recall for highly emotional words. however, a high emotional valence also enhanced recall for the associated color, suggesting a binding e"ect. in other words, emotional arousal enhances not only memory, but binding e$ciency as well. !ese are surprising strengths: binding memory can function even when people are not paying attention and can actually be improved by emotional arousal. future directions biological motion detection is so important that it becomes functional almost immediately out of the womb and underpins people’s basic ability to interact with each other (simion et al., 2008). more research into binding memory is needed to gain a greater understanding of biological motion detection and the corresponding elements of high-level social cognition. not enough is known about what other lower-level systems aside from feature binding are indispensible to biological motion detection. further research is also needed to determine the extent to which feature binding de#cits in pathological cases contribute to problems with biological motion detection, in order to learn more precisely just how much it depends on binding. !is might be accomplished by combining some of the experimental paradigms above to explore the e"ects of conditions like emotional arousal on binding and motion detection in subjects with neurological disorders. furthermore, more work is required to understand the neural correlates of binding memory in cases of these neurological diseases, particularly those with salient social implications, such as alzheimer’s, asd, and schizophrenia. advances in this area promise to not only enrich the understanding of how biological motion detection works in humans, but of the pathologies themselves. alvarez ga, cavanagh p. “!e capacity of visual short-term memory is set both by visual information load and by number of objects.” psychol sci 15:2 (2004) 106-111 baddeley a, allen r, vargha-kadem f. “is the hippocampus necessary for visual and verbal binding in working memory?” neuropsychologia 48 (2010) 1089–1095 bays m, wu e, husain m. “storage and binding of object features in visual working memory.” neuropsychologia 49(6), (2011), 1622-1631 blake r, shi"rar m. “perception of human motion.” annu. rev. psychol.58 (2007) 47–73 blake r, turner l, smoski mj, pozdol sl, stone wl. “visual recognition of biological motion is impaired in children with autism.” psychol sci. (2003), 14(2):151-7. burglena f, marczewski p, mitchell k, van der linden m, johnson m, danion j, salame p. “impaired performance in a working memory binding task in patients with schizophrenia.” psychiatry research 125 (2004) 247–255 doerksen s, shimura a. “source memory enhancement for emotional words.”emotion (2001), (i)1 5 – 11 grossman e, donnelly m, price r, pickens d, morgan v, neighbor g, blake r. “brain areas involved in perception of biological motion.” j. cogn. neurosci. 12:5 (2000), 711–720 herrington j, baron-cohen s, wheelwright s, singh k, bullmore e, brammer m, williams s. “!e role of mt+/v5 during biological motion perception in asperger syndrome: an fmri study.” research in autism spectrum disorders 1 (2007) 14–27 johansson g. “visual perception of biological motion and a model for its analysis”, perception psychophys. 14(2), (1973), 201–211. johnson j, hollingworth a, luck s. “role of attention in the maintenance of feature bindings in visual short-term memory.” j exp psychol hum percept perform. (2008), 34(1): 41–55. columbia undergraduate science journal open-access publication | http://cusj.columbia.edu 25 cusjvolume 6spring 2012 columbia undergraduate sci j http://cusj.columbia.edu + 8 0 $ 1 �% ,2 /2 * < kessels r. c., hobbel, d., & postma, a. “aging, context memory and binding: a comparison of ‘what, where and when’ in young & older adults”. intl j. neurosci 117:6 (2007), 795-810. oram mw, perrett di. “responses of anterior superior temporal polysensory (stpa) neurons to ‘biological motion’ stimuli”. j. cogn. neurosci. 6, (1994) 99–116. saiki j, miyatsuji h. “feature binding in visual working memory evaluated by type identi#cation paradigm.” cognition 102 (2007) 49–83. simion f, regolin l, bulf h. “a predisposition for biological motion in the newborn baby.” proc natl acad sci u s a. (2008), 105(2):809-13. treisman a. “consciousness and perceptual binding.” in a. cleeremans, ed., !e unity of consciousness: binding, integration, dissociation.oxford university press, 2003. treisman a. “feature binding, attention and object perception.” phil. trans. r. soc. lond. b (1998) 353, 1295-1306. wheeler m, treisman a. “binding in short-term visual memory.” journal of experimental psychology (2002), 131(1), 48–64 cusjcolumbia undergraduate science journal review articles 26cusj volume 6spring 2012 columbia undergraduate sci j http://cusj.columbia.edu + 8 0 $ 1 �% ,2 /2 * < references optical characterization of potential dampening in dendritic spine electrophysiology kim, samuel1, sakamoto, masayuki2, and yuste, rafael3 1columbia college, columbia university, new york, ny 10027, usa 2department of biological sciences, columbia university, new york, ny 10027, usa 3department of biological sciences, kavli institute for brain science, columbia university, new york, ny 10027, usa abstract this study aims to quantitatively characterize the electrophysiology of the dendritic spine as compared to that of its adjacent dendritic shaft, by imaging artificially induced backpropagating action potentials using a variety of different genetically encoded voltage indicators. we performed whole cell patch clamp and current injection recordings with simultaneous voltage imaging of neonatal mouse hippocampal neurons, which were transfected to express ‘arclight’ or one of two variants of ‘archaerhodopsin 3 (arch)’ known as ‘quasar1’ and ‘quasar2’. with arclight, we coupled electrophysiological current injection recordings with fluorescence imaging and compared the individual peak fluorescence change (∆𝑭/𝑭) value of each spine with the peak ∆𝑭/𝑭 value of its adjacent dendritic shaft in response to induced backpropagating action potentials. the results from arclight do not indicate a statistically significant dampening in membrane potential from the dendrite shaft across an adjacent spine, but do suggest a difference in membrane potential fluctuations between long pulse (100msec) and short pulse (20msec) current injections. with arch, we quantified the ∆𝑭/𝑭 values from current injection recordings and voltage imaging sessions of neuronal soma for ‘quasar1’ and ‘quasar2’. preliminary current injection and soma imaging results initially exhibited negligible signal-to-noise ratios of ∆𝑭/𝑭 in response to induced action potentials, possibly due to technical maladjustments. a subsequent, more general characterization of arch using voltage clamp and voltage step manipulations with quasar1 at different laser intensities indicates that quasar1 shows large changes in fluorescence at much weaker laser intensities (~10 mw) than was used for aforementioned current injection voltage imaging (~110 mw). we intend to further optimize and apply quasar1 and quasar2 to spine imaging, and subsequently investigate the difference in peak ∆𝑭/𝑭 values for different current injection pulse durations as suggested by arclight imaging data. i. introduction the prevailing view on consciousness speculatively characterizes the mind as ‘emergent’, meaning that it is unobserved on a microscopic level but emerges as a macroscopic property due to the collective interactions of its components, i.e., neurons. in order to understand the macroscopic workings of the brain, it inevitably follows that we must thoroughly decode the dynamic real-time communications that occur across the micro-circuitry between neurons 1 . considering the extent of electrophysiological variation that is present even on the level of an individual neuron, however, it would be useful to first characterize the elusive small-scale electrophysiological components of neuronto-neuron communication; one such component being the dynamics of membrane potential fluctuation in dendritic spines. dendritic spines are, as the name implies, small protoplasmic protrusions on the surface of the dendritic shaft of a neuron that are known to be postsynaptic contact points with the axons of presynaptic neurons 2 . previous studies have proposed and tested a number of hypotheses on the function and utility of these structures, such as linear integration of electrical input and prevention of dendrite potential saturation 2, 3,4 , biochemical compartmentalization and contribution to long term potentiation 5 , and preservation of the rapid time course of action potentials (aps) in affecting synaptic plasticity within a small time window 6 . in particular, we place interest in the suggested electro-physiological compartmentalization of spines in the course of a neuron’s processing of multiple inputs; in other words, the function of some sort of alteration of membrane potentials that flow between the shaft of a dendrite and its adjacent spine. the advent of techniques capable of direct measurement of potential on such a miniscule and rapid scale in dendritic spines, as opposed to indirect methods such as calcium imaging which utilizes delayed calcium dynamics as a proxy 7, 8 , is relatively new. voltage imaging, one of the particularly promising optical methods, utilizes voltage indicators, which usually consist of a complex of proteins that respond to fluctuations in electrical potential by undergoing a conformational change and altering their fluorescence emission. by embedding these indicators in the plasma membrane of a neuron, it is possible to make accurate measurements of membrane potential dynamics with fluorescence change as a proxy 9 . of particular interest in this study are two distinct genetically encoded voltage indicators known generally as ‘arclight’ and ‘arch (archaerhodopsin 3)’. arclight utilizes a complex between the voltagesensitive domain of ciona intestinalis’ voltage-sensitive phosphatase and a super ecliptic phluorin (a modified gfp), and has been shown to undergo ~1% to ~5% negative changes in fluorescence in response to individual action potentials in mammalian neurons 10 . variants of arch, on the other hand, are derived from a singlechannel bacterial rhodopsin (in other words, a light-activated ion channel) of halorubrum sodomense, and, while exhibiting negligible fluorescence at baseline, have been shown to undergo positive increases in fluorescence of much greater magnitudes (up to ~100%, depending on variant) per 100mv of membrane potential change under intense laser illumination 11,12 . hence, for the purposes of this study we adopt arclight as well as two variants of arch known as quasar1 and quasar2 as genetically encoded voltage indicators to closely track electrophysiological dynamics in individual neurons 13 . in examining spine electrophysiology, we primarily utilize artificially induced back propagating action potentials (baps) to elicit membrane potential fluctuations in dendritic spines. baps are known to be primarily passively diffusing changes in membrane potential that propagate in reverse of the conventional direction, from the axon hillock towards the dendritic tree, as a result of regular elicited action potentials 14 . thus by utilizing arclight and variants of arch to image the membrane potential fluctuations that occur from baps induced by a variety of current injections, we seek to quantify and test the electrical compartmentalization theory of dendritic spines. specifically, we address the potentially specialized treatment of and differential response to distinct electrical stimuli, through differences in the possible dampening of membrane potential from dendritic shaft to spine. ii. materials and methods a. hippocampal neuron plating and culture using 70% etoh and uv-sterilized stainless steel surgery tools, we euthanized neonatal mouse pups (p0-p1) via instantaneous decapitation and dissected hippocampi under a light microscope and stored them temporarily in a separate 15 ml falcon tube. all procedures following decapitation took place in a cold dissection solution, made from 160 mm sodium chloride, 5 mm potassium chloride, 1.1 mm magnesium sulfate, 2 mm magnesium chloride hexahydrate, 5 mm hepes, 5.6 mm glucose, 0.001g phenol red and titrated to ph 7.4. following dissection, isolated hippocampi were incubated at 37 ℃ for 20 min. on a shaker in a filtered dissociation solution consisting of lcysteine, calcium chloride, ethylenediaminetetraacetic acid (edta) and papain dissolved in dissection solution. the supernatant was subsequently removed and replaced with a filtered papain deactivation solution, which consisted of trypsin inhibitor and albumin dissolved in complete medium consisting of minimum essential medium (mem) w/o l-glutamine (invitrogen #11095-072), hi-glucose mem, serum extender, 100x (200 mm) lglutamine (sigma # g7513), and heat inactivated fcs (thermo fisher scientific hyclone products #sh30070.02hi). this was immediately replaced with regular complete medium, the hippocampal tissue further dissociated into individual cells via glass pipettes, and the obtained cells counted on a hemocytometer before being plated onto individual coverslips coated with poly-l-lysine in a 24-well plate (cell density 75k-120k per coverslip). following incubation overnight, the culture medium was completely replaced with neurobasal medium (nbm), which contained additives of b-27 supplement, lglutamine and penicillin/streptomycin solution. the medium was changed partially once a week. b. e. coli transformation and plasmid propagation dh5-alpha e. coli from invitrogen (www.invitrogen.com) stored in -80 ℃ were thawed, distributed into aliquots, and, after addition of a voltage indicator-encoding plasmid, were heat shocked at 42 ℃ for 45 s, followed by a 2 min. incubation on ice. the transformed cells were subsequently incubated in a ~37 ℃ water bath for 45 min., centrifuged and plated on lb agar plates, and incubated overnight at 37 ℃. a single colony of the transformed e. coli was grown in 5 ml lb broth with ampicillin, and incubated at 37 ℃ on a shaker for 8 hours. we then added a 1 ml aliquot to a 300 ml lbampicillin culture and incubated the cells overnight on a 37 ℃ shaker. the plasmid was isolated from incubated e. coli cultures according to the ‘endotoxin-free plasmid dna purification‘ user manual’s ‘nucleobond xtra maxi ef’ protocol by machery-nagel gmbh and co. kg, also accessible at www.mn-net.com. the purified plasmid was stored at -20 ℃. c. plasmid vector transfection transfections were performed 3-5 days after neuron plating. a mixture of the dna plasmid, calcium chloride, doubledistilled water, and 2x hepes buffered saline (hbs) was incubated at rt for 20 minutes (total 30 𝜇𝐿 for each coverslip). the entirety of the nbm for each culture was transferred to a 15 ml falcon tube and stored in a 37 ℃ incubator, and replaced with mem (www.invitrogen.com). each coverslip was subsequently transfected and incubated at 37 ℃ for 45 minutes. afterwards, the coverslips were washed three times with mem, and the previously collected nbm was re-applied to the coverslips. transfection was allowed to take place over about a 1-2 day period. d. electrophysiology: patch clamping and voltage imaging coverslips were taken from the incubator for patching starting 2-3 days post-transfection. the stage and coverslip were superfused at room temperature in oxygenated artificial cerebrospinal fluid (acsf) solution heated to ~37 ℃, diluted from a 10x acsf stock of 126 mm nacl, 26 mm sodium hydrogen carbonate, 1.145 mm sodium dihydrogen phosphate, 10mm dextrose, and 3 mm kcl, with additives of 1 m magnesium sulfate and 1 m calcium chloride. micropipettes (sutter instruments: http://www.invitrogen.com)/ http://www.mn-net.com/ http://www.invitrogen.com/ borosilicate glass capillaries) were pulled using a dmz-universal puller (zeitzinstruments) to 4-7mω resistance, and filled with an internal pipette solution of 140 mm k-gluconate, 5 mm kcl, 0.2 mm egta, 2 mm mg𝐶𝑙2 , 2 mm 𝑁𝑎2 atp, and 10 mm hepes, subsequently adjusted to ph 7.3 with koh. hippocampal neurons were whole-cell patch clamped using a brightfield/epifluorescence microscopy setup equipped with a mercury arclamp [for arclight (gfp): excitation filter 480/40 (460-500) nm, dm 495 nm, emission filter 535/550 nm. for arch (morange reporter): excitation filter 540/20 (530-550) nm, dm 570nm, barrier filter 590 nm. for arch (quasar1 and quasar2): emission filter 700/75) and a x40 (n.a. 0.8) or x60 (n.a. 0.9) objective setup mounted with an emccd camera (product: imageem (hamamatsu)) and four pipette manipulators. whole-cell patch clamping was performed using axon multiclamp 700b amplifiers (molecular devices), digitized at 10 khz with national instruments 6259 multichannel cards and recorded using custom software written using labview (national instruments). we used the multiclamp and packio softwares to hold voltage at -65 mv and apply and record current/voltage. for arclight, we performed current injection protocols, each protocol injecting 250 pa pulses for 10 times at regular intervals over a period of ~12 seconds, each pulse lasting 20 ms (short) or 100 ms (long). the packio recordings were coupled with fluorescent voltage imaging recordings that were performed under 570nm excitation. for arch, we initially also performed current injection protocols, each protocol injecting 250 pa pulses for 30 times at regular intervals over a period of ~33 seconds, each pulse lasting 20 ms (short) or 200 ms (long). in addition, for arch we performed voltage step manipulation protocols, which manipulated the membrane potential of a cell gradually by successive increments of 40 mv/500 ms voltage steps, inducing net voltage alterations ranging from -20 mv to +140 mv over the course of ~22 seconds. the packio recordings were coupled with fluorescent voltage imaging recordings that were performed under 642 nm ‘arch laser’ (coherent: obis) excitation. e. data processing as a preliminary check, we used imagej (nih) to process fluorescence recordings and derive fluorescence vs. slices (equivalent to time) plots. all subsequent data were derived via matlab (mathworks). iii. results a. criteria for data selection in analysis, we selectively excluded data that were problematic in one or more respects. foremost, neurons with aspiny dendrites, in other words dendrites with little to no dendritic spines, provided no data to work with. neurons that did possess spiny dendrites when viewed at 1 binning/300 msec exposure time but were too dim to show any spines to select during 4 binning/10 msec exposure time voltage imaging either due to progressive photobleaching, low baseline voltage indicator expression and fluorescence, or excessive background fluorescence due to high cell density were also excluded. finally, a number of cells, particularly during 20 msec current injection protocols, exhibited spontaneous firing even without stimulation; this disrupted the matlab program’s capability to synchronize the electrophysiology recordings with voltage imaging recordings. many 20msecpulse protocol recordings were excluded due to this phenomenon. though some of these omissions are contingent on technical limitations of our methods and may cause us to overlook some potentially viable data, we believe such overall necessarily excludes confounds and noise from our final results at the cost of a smaller sample size. b. arclight: current injection and shaftspine voltage imaging for exemplary ‘cell a’ we present the data from current injections and voltage imaging of a single representative shaft-spine pair from ‘cell a’. under bright-field microscopy at 60x magnification, cell a exhibits normal neuronal morphology very faintly, partially due to overpopulation and high cell density of the culture (fig. 1a). cell a as identified via epifluorescence microscopy of gfp under 570nm excitation (included in arclight; no additional marker was used) indicates very spiny dendrites, as well as appropriate localization of the genetically encoded arclight to the plasma membrane and dendritic spines (fig. 1b). cell a was whole-cell patch clamped and subjected to a current injection protocol that applied a total of 10 pulses, each of 250 pa magnitude and 100 ms duration, over the duration of ~12 seconds; neuronal membrane potential responses to 9 of the pulses are selected here for ideal visualization (fig. 2). for a selected region of interest (roi) (fig. 2a) indicating either a given dendritic shaft or its adjacent spine, we synchronized whole cell membrane potential recordings with local fluctuations in fluorescence and plotted both voltage and ∆f/f with respect to time for the duration of the simultaneous recordings. fluorescence changes (blue) coincide only very weakly with peak voltage fluctuations (green), presumably due to the low relative signalto-noise of arclight (fig. 2b). spike-triggered average (sta) data a b figure 1. arclight: ‘cell a’ a. cell a as seen by bright-field microscopy under 60x magnification, with micropipette visible. b. cell a as seen by epifluorescence microscopy of the gfp contained in arclight (no additional reporter was required). arclight (current injection): cell a spine a b c figure 2. arclight: current injection electrophysiology and δf/f of dendritic shaft vs. spine in ‘cell a’ a. the region of interest (roi) for δf/f comparison in a single dendritic shaftspine pair for cell a, indicated by the red boundaries as selected during data processing and analysis. b. whole cell electrophysiological recordings of voltage (green) and local fluctuations in δf/f (blue) in a single recording session of the aforementioned shaft-spine pair, synchronized on a time axis. c. spiketriggered average (sta) for the aforementioned shaft-spine pair. this represents the average δf/f (blue) of a specific roi, in response to each and every induced bap within a given train of current injections. the peak after frame=0 that spikes above the baseline fluorescence (red) was selected as the peak δf/f to be used in shaft vs. spine comparison. arclight (current injection): cell a shaft f lu o re sc en ce (a .u .) / v o lt ag e (m v ) time (s) time (s) frames frames δ f / presents an averaged representative ∆f/f response to one induced bap for the selected roi, based on voltage-fluorescence synchro-nizations; the data given indicates a ~5% peak change in fluorescence for a roi of the shaft and a ~4% peak change in fluorescence for its adjacent spine (fig. 2c). c. arclight: mean peak ∆𝑭/𝑭 of shaft vs. spine from the sta data (exemplified in fig. 2c) of multiple imaged shaft-spine pairs, we summed the peak ∆f/f of all shaft rois and of all spine rois, and compared the mean peak ∆f/f of the dendrite shaft (blue) to the mean peak ∆f/f of its adjacent spine (red) for long pulse (100 ms) and short pulse (20 ms) current injection protocols respectively (fig. 3). taking standard deviation into account, neither the long pulse nor the short pulse current injection/ voltage imaging results indicate a statistically significant dampening of peak ∆f/f from the dendrite shaft to its adjacent spine. however, the data does seem to suggest a potential difference in ∆f/f change between the long pulse and short pulse current injections, in that the 100 ms pulse injection protocol seems to potentially show greater dampening of ∆f/f in the spine, relative to the negligible ∆f/f difference in shaft to spine of the 20 ms protocol (fig. 3a, b). d. arch: current injection and soma voltage imaging for exemplary ‘cell b1’ and ‘cell b2’ as with arclight, we present the data from current injections and voltage imaging of representative neurons ‘cell b1’ and ‘cell b2’, respectively expressing the quasar1 and quasar2 variants of arch. under bright-field microscopy at 60x magnification, both cell b1 and cell b2 show normal, healthy neuronal morphologies (fig. 4a) in particular comparison to arclight-expressing cell a (fig. 1a). epifluorescence microscopy using the coexpressed morange reporter protein indicates sufficient expression for both cells, albeit relatively scarce dendritic spine distributions (fig. 4b). illumination under an arch-specific laser (642 nm) at 110 mw output indicates intense expression of the respective arch variants for each cell (fig. 4c). cell b1 and b2 were whole-cell patch clamped and both subjected to the same current injection protocol, which injected a total of 30 pulses, each 200 ms, over a duration of ~33 seconds. instead of a shaftspine pair, an roi of the soma illuminated by the arch laser was selected for analysis for both cells as a test of efficacy (fig. 5a). both cells responded to their respective current injections by firing a train of action potentials in response to each pulse (fig. 5b). voltage recordings corresponding to all 30 pulses for each cell were synchronized with fluorescence on a time axis; results do not visibly resolve individual fluorescence spikes (fig. 5c). the stas for both a b figure 3. arclight: mean peak δf/f of shaft vs. spine for long pulse (100msec) and short pulse (20msec) current injection recordings a. the mean peak δf/f sta with standard deviation (%) of shaft vs. spine in arclight voltage imaging of a total of 21 shaft-spine pairs, for the long pulse (100msec) current injection protocol. b. the mean δf/f sta with standard deviation (%) of shaft vs. spine in arclight voltage imaging of a total of 6 shaft-spine pairs, for the short pulse (20msec) current injection protocol. arclight (current injection) m ea n p ea k δ f /f shaft shaft spine spine m ea n p ea k δ f /f 100 ms injections 20 ms injections quasar1 in cell b1 and quasar2 in cell b2 show negligible signal relative to background fluorescence, presumably due to excessively high laser output and/or excessive noise (fig. 5d). a b c figure 4. arch: ‘cell b1’, ‘cell b2’ a. cell b1 (left) and cell b2 (right) as seen by bright-field microscopy under 60x magnification, with micropipettes visible. b. cell b1 (left) and cell b2 (right) expressing an morange reporter protein, as seen by epifluorescence microscopy. c. cell b1 (left) and cell b2 (right) expressing quasar1 and quasar2 respectively, as seen by epifluorescence microscopy coupled with 642nm arch laser illumination (110 mw). cell b1: quasar1 cell b2: quasar2 arch (current injection): cell b1, cell b2 morange morange quasar1 quasar2 cell b1: quasar1 cell b2: quasar2 a b c d figure 5. arch: current injection electrophysiology and δf/f of soma in ‘cell b1’ and ‘cell b2’ a. the region of interest (roi) for δf/f in soma for cell b1 (left) and cell b2 (right), indicated by the red boundaries. b. membrane potential fluctuations in cell b1 (left) and cell b2 (right) in response to a single long current pulse (200msec) injection. c. whole cell electrophysiological recordings of voltage (green) and local fluctuations in δf/f (blue) in a single recording session of soma for cell b1 (left) and cell b2 (right), synchronized on a time axis. d. spike-triggered average (sta) for cell b1 (left) and cell b2 (right). this represents the average δf/f of a specific roi, in response to each and every induced bap within a given train of current injections. arch (current injection): cell b1, cell b2 v o lt ag e (m v ) time (s) time (s) f lu o re sc en ce (a .u .) /v o lt ag e (m v ) time (s) time (s) δ f /f frames frames e. arch: voltage step manipulation and soma voltage imaging for exemplary ‘cell c’ to correct for the negligible signal given in figure 5d, we performed direct, large-scale voltage step manipulations of membrane potential and simultaneous quasar1 voltage imaging in exemplary ‘cell c’. cell c shows normal, healthy neuronal morphology under 60x bright-field microscopy (fig. 6a) as well as epifluorescence microscopy visualizing the morange reporter (fig. 6b). for quasar1 fluorescence, the arch laser output was decreased drastically from 110 mw to 10mw, and brightfield microscopy was temporarily turned off to eliminate noise; cell c is shown expressing quasar1 towards the peripheries of the range of laser-illumination (fig. 6c). we applied a voltage step protocol that manipulated the membrane potential of cell c by 40 mv/500 ms voltage step increments, for a net alteration ranging from -20 mv to +140 mv over a duration of ~22 seconds. four successive recordings were performed on cell c at laser outputs of 10 mw, 20 mw, 30 mw and 40 mw, and similar rois were selected from all four recordings for analysis and comparison (fig. 7a). preliminary ∆f vs. time plots of image stacks derived from voltage imaging recordings indicate the largest quasar1 signal at 10 mw laser output, with fluorescence fluctuations corresponding to individual voltage steps visible; subsequently higher intensities of laser output seem to generate excessive noise relative to arch signals, and the fluorescence ‘steps’ are rendered increasingly less visible towards 40 mw output (fig. 7b). fluorescence fluctuations a b c figure 6. arch: ‘cell c’ a. cell c as seen by bright-field microscopy under 60x magnification, with micropipettes visible. b. cell c expressing an morange reporter protein, as seen by epifluorescence microscopy. c. cell c expressing quasar1 (yellow arrow), as seen by epifluorescence microscopy coupled with 642nm arch laser illumination (10 mw) arch (voltage step): cell c morange quasar1 1 0 m w a b c figure 7. arch: voltage step electrophysiology and δf/f of soma in ‘cell c’ a. four distinct rois of cell c soma, selected in 4 different quasar1 voltage step manipulation and voltage imaging recordings using 4 different respective laser intensities (10mw, 20mw, 30mw, and 40mw). b. general fluorescence fluctuations of quasar1 in response to voltage step manipulations, as seen in cell c soma imaging using 4 different respective laser intensities. c. whole cell electrophysiological recordings of voltage (green) and local fluctuations in δf/f (blue) in a single recording session of soma for cell c, synchronized on a time axis. arch (voltage step): cell c f lu o re sc en ce / v o lt ag e (m v ) f lu o re sc en ce ( a. u .) time (s) frame s 2 0 m w 3 0 m w 4 0 m w f lu o re sc en ce ( a. u .) f lu o re sc en ce ( a. u .) f lu o re sc en ce ( a. u .) frame s frames frames f lu o re sc en ce / v o lt ag e (m v ) f lu o re sc en ce / v o lt ag e (m v ) f lu o re sc en ce / v o lt ag e (m v ) time (s) time (s) time (s) corrected for bleaching and synchronized with membrane potential fluctuation recordings on a time plot further demonstrate this trend; the characteristic parabola of ∆f as corresponding to the shape of the voltage waveform is rendered unrecognizable starting at 30 mw laser intensity (fig. 7c). iv. discussion we hypothesized a dampening of membrane potential from the dendritic shaft into its adjacent dendritic spine as a test of the theory of spine electrophysiological compartmentalization. arclight voltage imaging of induced baps in mouse hippocampal neurons does not in fact seem to indicate a statistically significant dampening of membrane potential in the spine relative to the dendrite shaft. there does, however, seem to be some suggestion of a difference in membrane potential alteration at the spine between electrophysiological dynamics induced by long (100 ms) and short (20 ms) current injections, in other words between a train of multiple baps in rapid succession vs. single baps (fig. 3). considering the nature of baps as primarily passively transmitted potentials, one would expect some amount of default deterioration as a bap progresses across a dendrite into a spine. however, results from this study’s arclight voltage imaging indicate not only that there is no immediately significant default deterioration (i.e., dampening) of potential, but also that a more prolonged stimulus (~100 ms current injection) inducing multiple baps may entail greater dampening than that caused by a small stimulus (~20 ms current injection) inducing a single bap. this seems to contradict any initial assumptions that a passively traveling membrane potential would be prevented from deteriorating, i.e. dampening, if it is ‘reinforced’ by multiple subsequent baps. rather, assuming our arclight imaging results are valid, our study may indicate that, in dendritic spines, weaker baps tend to be conserved while stronger, ‘reinforced’ baps are likely to be dampened, possibly in association with and/or manifested as phenomena such as long-term potentiation. caution must be taken, however, concerning the sheer number of possible neuronal subtypes and the consequent variability in neuron electrophysiology. literature indicates that inhibitory gabaergic interneurons, and even som interneurons which comprise a subcategory of inhibitory interneurons, may be divided into a number of subtypes according to their electrophysiology 15, 16 . the indiscriminate nature of this study in terms of neuronal subtypes suggests that certain trends may very well have been overlooked or overemphasized. we have thus started preparations for more precise imaging of spines, by optimizing two variants of arch, quasar1 and quasar2, reported to be superior to arclight in fluorescence and kinetics under optimized conditions, in voltage imaging of neuronal soma. our initial results from arch with current injections, as had been done with arclight, indicated nearly negligible signal-to-noise ratios for both variants (fig. 5d). subsequent voltage imaging of voltage step manipulations with quasar1, performed at reduced brightfield noise and lower laser intensities (~10 mw vs. the previous ~110 mw), indicates that, to utilize the sensitivity of arch in capturing spine electrophysiology, voltage imaging must be performed at reasonably low laser intensities (~10 mw) to prevent saturation. depending on further investigation of spine electrophysiology at a higher spatiotemporal resolution, it may be possible to more robustly demonstrate that dendritic spines discriminate between weak and strong stimuli. in conjunction with the results we have attained so far via arclight, this would lend support to the hypothesis arguing for the electrophysiological compartmentalization and specialization of dendritic spines; elucidation of a component crucial in the composition of larger-scale neuronal circuits and patterns. v. references [1] alivisatos, a., chun, m., church, g., greenspan, r., roukes, m. and yuste, r. (2012). [2] the brain activity map project and the challenge of functional connectomics. neuron 74, 970-974. [3] yuste, r. (2011), dendritic spines and distributed circuits. neuron 71, 772-781. [4] cash, s., and yuste, r. (1999). linear summation of excitatory inputs by ca1 pyramidal neurons. neuron 22, 383-394. [5] yuste, r., and denk, w. (1995). dendritic spines as basic functional units of neuronal integration. nature 375, 682-684. [6] koch, c., and zador, a. (1993). the function of dendritic spines: devices subserving biochemical rather than electrical compartmentalization. j. neurosci. 13, 413-422. [7] holthoff, k., zecevic, d., and konnerth, a. (2010). rapid time course of action potentials in spines and remote dendrites of mouse visual cortex neurons. j. physiol. 7, 10851096. [8] koester, h.j., and sakmann, b. (1998). calcium dynamics in single spines during coincident preand postsynaptic activity depend on relative timing of back propagating action potentials and subthreshold excitatory postsynaptic potentials. proc. natl. acad. sci. usa 95, 9596-9601. [9] grienberger, c. and konnerth, a. (2012). imaging calcium in neurons. neuron 73, 862885. [10] peterka, d., takahashi, h., and yuste, r. (2011). imaging voltage in neurons. neuron 69, 9-21. [11] jin, l., han, z., platisa, j., wooltorton, j., cohen, l., and pieribone, v. (2012). single action potentials and subthreshold electrical events imaged in neurons with a fluorescent protein voltage probe. neuron 75, 779-785. [12] kralj, m. j., douglass, a. d., hochbaum, d. r., maclaurin, d., and cohen, a. (2012). optical recording of action potentials in mammalian neurons using a microbial rhodopsin. nature methods 9:1, 90-95. [13] maclaurin, d., venkatachalam, v., lee, h., and cohen, a. (2013). mechanism of voltage-sensitive fluorescence in a microbial rhodopsin. pnas 110:15, 59395944. [14] akemann, w., mutoh, h., perron, a., rossier, j., and knopfel, t. (2011). imaging brain electric signals with genetically targeted voltage-sensitive fluorescent proteins. nature methods 7:8, 643-649. [15] stuart, g., spruston, n., sakmann, b., and hausser, m. (1997). action potential initiation and backpropagation in neurons of the mammalian cns. tins 20:3, 125-131. [16] ascoli, g. a., et al. [the petilla interneuron nomenclature group (ping)]. (2008). petilla terminology: nomenclature of features of gabaergic interneurons of the cerebral cortex. nat rev neurosci. 9:7, 557-568. doi:10.1038/nrn2402. [17] mcgarry, l. m., packer, a. m., fino, e., nikolenko, v., sippy, t., and yuste, r. (2010). quantitative classification of somatostatin-positive neocortical interneurons identifies three interneuron subtypes. frontiers in neural circuits 4:12, 1-19. abstract this study aims to quantitatively characterize the electrophysiology of the dendritic spine as compared to that of its adjacent dendritic shaft, by imaging artificially induced back-propagating action potentials using a variety of different geneticall... i. introduction ii. materials and methods v. references paper title (use style: paper title) characterizing intracellular localization and chromatin remodeling role of chd6 in human fetal development and in glioblastoma noah basri and nadejda tsankova icahn school of medicine at mount sinai departments of pathology and neuroscience friedman brain institute abstract— chd6, a member of the chd family of dna-binding, chromatin-modifying proteins, has been shown to be upregulated in glioblastoma. using immunofluorescence the intracellular region to which chd6 localizes can be determined. through the use of chromatin immunoprecipitation (chip), candidate genomic loci enriched for chd6 can be determined. immunofluorescence and chip results for chd6 using a current protocol are presented. notably, immunofluorescence of fetal germinal matrix and adult subventricular zone showed principally cytoplasmic localization of chd6, contrary to expectation. chd6 chip showed partial enrichment in certain genes of possible significance to glioblastoma in fetal tissue relative to the positive control. a higher sample size is needed to test and evaluate the significance of these findings. the results presented here reveal new insights into the changes in the epigenetic landscape contributing to glioblastoma and the manipulation of chd6 expression could be considered for treating specific incidents of glioblastoma. i. introduction the set of proteins that constitute epigenetic regulators include so-called ‘writers,’ ‘editors’ and ‘readers’ of dna methylation or histone post-translational modifications5. whereas ‘writers’ are responsible for de novo modifications to dna or dna-bound histone proteins and ‘editors’ are responsible for chemically altering or reverting the modification placed by ‘writers’, ‘readers’ modulate and mediate interaction between proteins and protein complexes, including the transcriptional machinery5. dysfunction or deregulation of epigenetic ‘writers,’ ‘editors’ or ‘readers,’ is widely implicated in stem cell activation and tumorigenesis5. among the epigenetic ‘readers’ is a subset of chromodomain-containing proteins, the chromodomain helicase dna-binding, or chd, family of chromatin modifiers1,4,5,8,12. the chd proteins are distinguished from other chromodomain-containing chromatin readers by their tandem chromodomains1,4,5,8, which have been shown to interact with methylated lysine residues correlated with either repressed transcription or active transcription (h3k27me3 and h3k4me3)1,4,11,12. within the larger family of chd proteins, are three subfamilies distinguished by their varying functional or conserved domains. class i chd proteins contain chd1 and chd2. the class i chd proteins are the best characterized in terms of function and structure,4,8,12. chromatin immunoprecipitation (chip) experiments, which utilize antibodies to selectively precipitate proteins bound to specific genomic regions, have been conducted to elucidate the dna regions enriched for chd112. class ii chd proteins are distinguished by their dual phd fingers, or plant homeodomains, and are composed of chd3, chd4 and chd51,4,8,12. class iii chd proteins are composed of chd6, chd7, chd8 and chd91,4,8,12. they are distinguished from the other two classes by their sant (switching-defective protein 3, adaptor 2, nuclear-receptor corepressor, transcription factor iii b) and brk (brahma and kismet) domains, which lie close to the proteins’ c-terminus1,4,12. the precise function of the brk domain is unknown4; however, it has been shown to interact with the transcriptional repressor ctcf1,8, potentially coordinating chromatin loops7. of considerable interest in the third chd class is the poorly characterized nucleosome-remodeling protein chd6. though its precise mechanism of action has not been established8,12, chd6 has been reported to be upregulated in glioblastoma3,10. as a result, insight into the intracellular localization of chd6 as well as its enrichment in genetic loci of interest to glioblastoma, such as olig22, egfr (epidermal growth factor receptor)2 and gfap (glial fibrillary acidic protein)2 using immunofluorescence and chip would provide a more holistic understanding of chd6 expression and the epigenetic landscape of glioblastoma pathology. immunofluorescence and chip experiments were performed to characterize the intracellular localization and chromatin-remodeling role of chd6. notably, immunofluorescence of fetal germinal matrix and adult subventricular zone showed principally cytoplasmic localization of chd6, contrary to expectation. through the use of chip, candidate genomic loci enriched for chd6 were elucidated. chd6 chip showed partial enrichment at the promoter of egfr, one gene of possible significance to glioblastoma2, in fetal and glioblastoma tissues, relative to a presumed positive control locus where chd6 has been previously demonstrate to bind8. however, higher sample size is needed to test and evaluate the significance of these findings. ii. methods a. immunofluorescence paraffinized tissue sections (4 micrometers) previously fixed in 10% formalin were deparaffinized in xylene. sections were rehydrated gradually through washes in graded alcohols: wash in 100% ethanol twice for 10 minutes each, then 95%, 75%, 50% ethanol for 5 minutes each then deionized water for 5 minutes. slides were boiled in citrate-based buffer (ph=6) for 5 minutes then cooled on bench top for 30 minutes. specimen was blocked in blocking buffer for 60 minutes using 0.25% tritonx and 1% normal donkey serum. cells were incubated with anti-gfap (rat) and chd6 (rabbit) primary antibodies overnight at a concentration of 1:100. sections were washed 3 times in pbs. sections were then incubated for 1 hour with fluorochrome-conjugated secondary antibodies. secondary antibodies for rat (pink fluorescence) and rabbit (green fluorescence) were used. following three more pbs washes, specimens were incubated with dapi nuclear stain (blue fluorescence) for 5 minutes. sections were examined using confocal microscopy. b. chip and qpcr analysis dna binding of chd6 was assessed via chromatin immunoprecipitation (chip) assays. frozen post-mortem brain tissue or neurosurgical de-identified specimens (adult cortex n=3, fetal germinal matrix n=3, glioblastoma n=3) were minced manually using razor blades in a petri dish. minced tissue was suspended in 1400ul 1% formaldehyde fixative and rotated for 10 minutes at room temperature. formaldehyde fixation was quenched with 93.3ul 2m glycine (125mm final concentration) and rotated for 5 minutes at room temperature. chromatin was sheared to 150-600bp9 and subjected to immunoprecipitation with antibodies against chd6 or igg (rabbit)2,9 as negative control (12-370, millipore, 5 mg) after pre-incubation with magnetic dynabeads from rabbits (11203d, invitrogen). ten percent of the lysate was saved as “input” for normalization analysis. the chromatin-associated dna was purified using the qiaquick pcr purification kit (28104, qiagen, valencia, ca) and quantified by qpcr (stratagene mx3000p) using primers designed within the egfr promoter (egfrprom), zic1, a reputed gene enriched for chd6, olig2, and b2m, a housekeeping gene (appendix 1). the data was analyzed as percent input recovery relative to zic1, the presumed positive control. melting curves were analyzed to ensure specificity of the primers. the efficiency for all primers was calculated, deemed appropriate, and accounted for in the analysis2. iii. results a. chd6 localizes to the cytoplasm in fetal germinal matrix and adult subventricular zone immunofluorescence staining of fetal germinal matrix showed principally strong cytoplasmic localization of chd6 (figure 2). there was high cell abundance in the fetal tissue compared to the adult tissue, consistent with the findings by samuelsen et.al6. chd6 staining appeared comparatively stronger in the fetal germinal matrix than in adult subventricular zone. gfap, stained in pink, was clearly expressed in the radial glia projecting out of the germinal matrix. notably, chd6’s principally cytoplasmic localization seemingly conflicts with its reputed function as a dna-binding protein and its hypothesized nuclear localization12. however, strong cytoplasmic staining could have obscured nuclear staining, and chip studies indicate the presence of locusspecific nuclear chd6 presence. further experimentation on chd6 is required to determine whether it is natively found in the nucleus or if it translocates to dna in response to some stimulus, as is seen in other transcription factors14. intriguingly, there is remarkable chd6 signal seen in the ependymal layer of the subventricular zone (figure 1), which has not been previously demonstrated. previous chip experiments have shown enrichment of chd6 in cf-pac, a pancreatic adenocarcinoma cell line, and caco2, a colon cancer cell line8. interestingly, pancreatic ductal cells (from which pancreatic adenocarcinoma arises), enterocytes, and ependymal cells have simple columnar structure17,18. enterocytes and ependymal cells also have a similar brush border on their apical surface17. whether these commonalities are related to chd6 expression remains to be verified. b. chd6 may be enriched at the egfr promoter in fetal tissue and glioblastoma chip was performed on adult cortex (n=3), fetal germinal matrix (n=3) and glioblastoma (n=3) tissue types to assess for in vivo binding of chd6 at genomic loci of potential interest in fetal development and gliomagenesis. the enrichment of chd6 in each of the genomic loci of interest was calculated as percent of input recovery, and it was normalized to percent input enrichment of chd6 at zic1, the positive control locus (figure 4). the expectation was to see enrichment of chd6 at gene loci, which are dysfunctionally upregulated in figure 2 note the principally cytoplasmic localization of chd6 and strong chd6 signal. fetal germinal matrix under the confocal microscope. gfap fluoresces pink, chd6 is green and dapi is blue. figure 1 adult subventricular zone under the confocal microscope. note cytoplasmic localization of chd6 and robust chd6 signal in the ependymal layer, as indicated by the white arrow. gfap fluoresces pink, chd6 is green and dapi is blue. glioblastoma, such as at olig2 and egfr, since the chromodomains of chd6 bind methylated lysine residues associated with active transcription11. this preliminary data reveal greatest chd6 enrichment at the egfr promoter in fetal and glioblastoma samples, and at the olig2 promoter in fetal samples, relative to zic1. unfortunately, significance cannot be established in this data since the qpcr amplification of igg samples yielded more than one amplicon product. iv. discussion the immunofluorescence staining of both fetal germinal matrix and adult subventricular zone showed principally cytoplasmic localization of chd6. this contradicts the hypothesis that chd6 is strictly relegated to dna-binding in the nucleus. one possible explanation for these results is that most chd6 could natively reside in the cytoplasm but translocate to the nucleus in response to binding of a cofactor. lathrop et. al. report that chd6 precipitates in complex with the transcription factor nrf2, which is trafficked to the nucleus in response to cellular oxidative stress13,14. analysis of the primary amino acid sequence of chd6 reveals several potential nuclear localization signals (nls)19. one possible mechanism for chd6’s counterintuitive intracellular localization could be that chd6’s nls sequences are obscured by protein folding but are subsequently revealed following binding of a cofactor. upon the emergence of the nls sequences, importin, the protein responsible for protein import into the nucleus, could shuttle chd6 to the genetic loci where chd6 exerts its action as a chromatin remodeler. both the immunofluorescence and chip experiments relied on primary antibodies effective at recognizing chd6. prior to the immunofluorescence and chip, no experiments validating the antibody and proving its efficacy in precipitating chd6 were performed. furthermore, although no published literature has elaborated on the finding that chd6 is enriched at zic1, zic1 was used as the positive control locus in the chip experiment. sancho et. al., independently reported that zic1 would act as an effective positive control. in designing primers for zic1, the ucsc genome browser was consulted for candidate loci that could be enriched for chd6. candidate loci were determined using enrichment of transcription factor ctcf as proxy for chd6 since sancho et. al. reported ctcf acting in complex with chd68. to date, there has been no brain tissue chip-seq data for chd6 upon which to base zic1 as a positive control more accurately. operating under the conjecture that chd6 is enriched at zic1, based on the assertion by sancho et. al. is an assumption in this experiment. another limitation of the data is the lack of a negative control. since chd6 is a chromatin remodeler that colocalizes with rna polymerase ii8, an effective negative control would be loci of constitutively heterochromatic regions of dna, such as telomeres or centromeres. since these loci are constitutively heterochromatic, they are not transcribed and therefore enrichment of rna polymerase ii and chd6 would be unlikely in those regions. the qpcr amplification of the igg samples yielding more than one amplicon could suggest the presence of primer dimers on the melting curve due to very little dna pulled down in the immunoprecipitation. as a result, the experiment must be repeated with higher sample size in order to establish significance and make a strong assertion based on the chip data. future experiments will seek to titrate parameters for and effectuate a chip protocol using mnase, as opposed to sonication, for chromatin fragmentation. more chip experiments on adult cortical, fetal and glioblastoma tissue isolated from postmortem brain are needed to confirm or deny the results presented here. also, immunofluorescence on epileptic brain specimens might shed further light on the question of chd6’s translocation to the nucleus. epileptic seizures are characterized by failure to reuptake glutamate released into the synapse, resulting in prolonged and uncontrolled neuronal depolarization15. this behavior is highly taxing on neurons and often results in a state of cellular figure 3 enrichment of chd6 binding is represented as percent input, relative to zic1, the positive control locus. loci assessed include, egfr promoter (green), b2m (purple), and olig2 (blue). e n ri ch m en t o f im m u n o p re ci p it at ed d n a o v er 2 0 % i n p u t oxidative stress called excitotoxicity16. examining such oxidatively stressed tissue through immunofluorescence could enhance the understanding of chd6’s intra cellular localization. going forward, experiments should seek to elucidate the coordination between chd6 and ctcf8 in the formation of loops within chromatin7. these loops signify the topological landscape of chromatin and are as impactful to transcription as the molecular epigenetic landscape of histones and dna7. to further the body of knowledge around chromatin loops and a potential role for chd6 in catalyzing their formation, coimmunoprecipitation (co-ip) of chd6 should be performed. if chd6 precipitates in complex with ctcf and cohesin, it can be determined whether chd6 has a correlative role in coordinating chromatin loops. the above data provide evidence that suggests a role for chd6 in binding chromatin at actively transcribed loci, such as the egfr promoter or olig2 in fetal cortical and glioblastoma tissue. the results presented here reveal new insights into the changes in the epigenetic landscape contributing to glioblastoma. manipulation of chd6 expression could be considered for treating specific indications of glioblastoma by decreasing transcription of inappropriately expressed genes such as olig2 and egfr; however, further research is needed to fully understand chd6’s role in gliomagenesis. acknowledgment i would like to thank dr. tsankova and dr. tome-garcia whose reagents, support and assistance were critical to performing the immunofluorescence and chip. i would also like to thank parsa erfani for his expertise and guidance throughout the lysate preparation and statistical analysis. thank you to dr. walsh for collaborating on the experimental design, providing the chd6 antibody and reporting zic1 as a positive control for chd6 chip. references [1] allen et. al. (2007) solution structure of the brk domains from chd7. j. mol. biol. 371, 1135–1140. [2] erfani, p., tome-garcia, j., canoll, p., doetsch, f., & tsankova, n. (2015). egfr promoter exhibits dynamic histone modifications and binding of ash2l and p300 in human germinal matrix and gliomas. epigenetics, 496-507. [3] forbes, et. al. cosmic: exploring the world's knowledge of somatic mutations in human cancer. nucl. acids res. (28 january 2015) 43 (d1): d805-d811 first published online october 29, 2014 doi:10.1093/nar/gku1075 [4] micucci joseph a., sperry ethan d., and martin donna m. chromodomain helicase dna-binding proteins in stem cells and human developmental diseases. stem cells and development. april 15, 2015, 24(8): 917-926. doi:10.1089/scd.2014.0544. [5] plass c, pfister sm, lindroth am, bogatyrova o, claus r, lichter p. mutations in regulators of the epigenome and their connections to global chromatin patterns in cancer. nat rev genet. 2013;14:765–80. doi: 10.1038/nrg3554. [6] samuelsen, g., bogdanović, n., laursen, h., graem, n., larsen, j., & pakkenberg, b. (2011). total cell number in fetal brain. image analysis & stereology, 19(1), 35-38. doi: http://dx.doi.org/10.5566/ias.v19.p35-38 [7] sanborn et. al. chromatin extrusion explains key features of loop and domain formation in wild-type and engineered genomes. pnas 2015 112 (47) e6456-e6465; published ahead of print october 23, 2015, doi:10.1073/pnas.15185521 [8] sancho et. al. chd6 regulates the topological arrangement of the cftr locus. hum. mol. genet. (2015) 24 (10): 2724-2732 first published online january 28, 2015 doi:10.1093/hmg/ddv032 [9] schoppee bortz, p. d., & wamhoff, b. r. (2011). chromatin immunoprecipitation (chip): revisiting the efficacy of sample preparation, sonication, quantification of sheared dna, and analysis via pcr. plos one, 6(10), e26015. http://doi.org/10.1371/journal.pone.0026015 [10] uhlén m et. al. a human protein atlas for normal and cancer tissues based on antibody proteomics. mol cell proteomics. 2005 4(12):1920-32. pubmed:16127175 doi: 10.1074/mcp.m500279-mcp200 [11] yap, k. l., & zhou, m.-m. (2011). structure and mechanisms of lysine methylation recognition by the chromodomain in gene transcription. biochemistry, 50(12), 1966–1980. http://doi.org/10.1021/bi101885m [12] zentner ge, tsukiyama t, henikoff s (2013) iswi and chd chromatin remodelers bind promoters but act in gene bodies. plos genet 9(2): e1003317. doi: 10.1371/journal.pgen.1003317 [13] m.j. lathrop, l. chakrabarti, j. eng, et al., deletion of the chd6 exon 12 affects motor coordination, mamm. genome 21 (2010) 130–142. [14] jaramillo mc, zhang dd. (2013). the emerging role of the nrf2-keap1 signaling pathway in cancer. genes dev 2013; 27: 2179–2191. [15] chapman ag (april 2000). "glutamate and epilepsy". j. nutr. 130 (4s suppl): 1043s–5s. pmid 10736378 [16] fujikawa, dg (2005). "prolonged seizures and cellular injury: understanding the connection.". epilepsy & behavior : e&b. 7 suppl 3: s3–11. doi:10.1016/j.yebeh.2005.08.003. pmid 16278099. [17] keller, t. c. s. and mooseker, m. s. 2011. enterocyte cytoskeleton: its structure and function. comprehensive physiology. 209–221. [18] bottin-grapin, anne. 2004. ductal cells of the pancreas. international journal of biochemistry and cell biology. 504-510. [19] kosugi s., hasebe m., tomita m., and yanagawa h. (2009) systematic identification of yeast cell cycle-dependent nucleocytoplasmic shuttling proteins by prediction of composite motifs. proc. natl. acad. sci. usa 106, 10171-10176. i. introduction ii. methods a. immunofluorescence b. chip and qpcr analysis iii. results a. chd6 localizes to the cytoplasm in fetal germinal matrix and adult subventricular zone b. chd6 may be enriched at the egfr promoter in fetal tissue and glioblastoma iv. discussion acknowledgment references columbia undergraduate science journal open-access publication | cusj.columbia.edu spring 2013 | volume 7 columbia undergraduate science journal 4 gem cell line establishment, characterization, and analysis daniel chester1* 1abate-shen lab, columbia university, new york ny 10027 abstract the use of in-vitro cell lines in conjunction with in-vivo mouse models is a common practice in the study of cancer. culturing cells with the same genotypes as the specific strains of mice being used in studies can be very important and useful in supplementing the research and findings in-vivo. the purpose of this particular study was to develop, characterize, and analyze different cell lines in order to utilize them in conjunction with mice in the lab. we have successfully established two cell lines from two different mouse models. we determined that the two different genetic strains of mice generate two different types of cells when grown in culture that accurately reflect the source strains. the purpose of this project was to adapt a protocol meant for establishing human breast tumors to mouse prostate models, and then to test if the correct cells had been isolated for future use. western blotting and other testing verified the correct cell type, therefore this project succeeding in developing a method for establishing more cell lines in the future. introduction the united states reported approximately 220,000 new cases and 32,000 deaths due to prostate cancer in 2010 [1]. prostate cancer is the most commonly diagnosed cancer in men and is the second highest source of cancer-related deaths in men in the united states. although men diagnosed with localized prostate cancer have a 5-year survival rate of nearly 100%, men with advanced prostate cancer face a much more difficult challenge [1]. as a result, it is important to develop novel treatments for patients with advanced prostate cancer. for many years, research in the abateshen lab has focused on generating a series of genetically engineered mouse (gem) models to evaluate drug therapies for prostate cancer. this project is part of a larger one aimed at evaluating drugs targeting the androgen receptor signaling pathways that are often used by prostate cancer cells as positive feedback loops promoting their growth and metastasis [2] [3]. this project takes advantage of the gem models that are based on perturbations of molecular pathways that are frequently deregulated in human prostate cancer. no model is perfect, and every experimental system has its own limitations. mice breed relatively quickly and to an extent can be used to reflect human disease, but it can be expensive and very time consuming to develop strains with the necessary genotype in large enough quantities. additionally, it takes 618 months for a mouse to develop prostate tumors. to choose a gem strain to test a drug in terms of response and resistance, it would be beneficial to know which strains are susceptible and respond most effectively to treatment and which strains do not. drug screening in cell lines attempts to address this problem by creating a more efficient and accurate process of deciding which drugs to test on which gem models. with a functioning bank of cell lines representative of each of the different strains held in the lab, any new drug can be tested on all or a selected few representative cell lines at any given point of time. this will reduce turnaround times compared to mouse models. copyright: © 2013 the trustees of columbia university, columbia university libraries, some rights reserved, sorrento, et al. received 1/1/2013. accepted 2/1/2013. published 1/1/13 *to whom correspondence should be addressed: abate-shen lab, columbia university, new york ny 10027, email: djc2162@columbia.edu columbia undergraduate science journal open-access publication | cusj.columbia.edu spring 2013 | volume 7 columbia undergraduate science journal 5 in vitro models are, however, not witho ut their limitations. there is always the fact that cells are dependent on their microenvironment, and drug response can be different in these models. therefore it can sometimes be very challenging to extrapolate from the results of in vitro work back to the biology of the whole organism. however, there are several additional reasons to develop mouse cell lines. a comprehensive collection of developed cell lines gives you the capability to ask questions and come up with answers quickly. using shrna, it is simple and quick in culture to knock out the expression of a certain gene of interest, and observe the effects of that on the growth of cancer cells. there is also the possibility of using the cell lines for injection into nude mice for xenograft studies. this would give an idea of how a tumor, given certain genetic alterations from the gene knockouts performed, would grow in vivo, albeit in a mouse without an immune system. having and being able to manipulate cell lines reflecting our mouse models opens up many interesting experimental possibilities. an understanding of the inducible nature of these mouse models is important for understanding this paper. nkx3.1 is a prostatespecific homeobox gene whose haploinsuffiency or loss predisposes to prostate cancer in humans as well as in mouse models. it codes for a recombinase that, when activated, cuts out other segments of dna in specific places. it can either cut out stop sequences, promoting oncogene activation, or entire tumor supressors, promoting unchecked cancer growth. these alterations are reflected in the mouse models, and the goal of this research is to develop strains that maintain these genetic characteristics. this project is novel in that no other cell lines exist with these exact modifications, and that now we have a repeatable process for establishing them. the methods for establishing the cell lines as well as making sure they were the right ones follows, as well as a discussion of their implications. methods gem models the genetically engineered nkx3.1creert2/+; ptenflox/flox; kraslsl/+ mice (npk mice) developed in the lab as well as nkx3.1creert2/+; ptenflox/flox; hi-myc (npm mice) were used in this study. the nkx3.1 creert2 allele simultaneously inactivates nkx3.1 and drives tamoxifen-dependent cre-mediated recombination in adult prostate epithelium. both strains have the nkx3.1creert2/+ allele, which is heterozygous for nkx3.1 and expresses cre-ert2 under the control of the nkx3.1 promoter. these mouse models also contain a conditional allele for pten (ptenflox/flox) having loxp sites flanking exon 5. the npk mice have a lox-stop-lox kras allele expressing an inducible krasg12d allele (figure a). the npm mice express a transgene coding for the human c-myc specifically in the prostate (figure 1b). for induction of cre activity, tamoxifen is delivered by oral gavage (100mg/kg) for 4 consecutive days, to mice at 3 months of age. the inducible cre allele also allows us to target gene deletion in the prostate epithelium of adult mice. figure 1: strategies for inducible recombination in the prostate [3] a. diagram illustrating npk mouse recombination b. diagram illustrating npm mouse recombination columbia undergraduate science journal open-access publication | cusj.columbia.edu spring 2013 | volume 7 columbia undergraduate science journal 6 cell line establishment we adapted the protocol for culturing breast tumor cells from [4] to create our own prostate cancer cell line cultures. the steps, illustrated below, were as follows: 1) prostate tumors were obtained from the npk and npm mouse models as well as metastases from lung and liver from the npk mice. each successful culture came from one mouse, but three npk and three npm mice were used, one final culture coming from each group. tissues were harvested in dmem media supplemented with 10% fetal bovine serum (fbs), 2mm glutamine, 10mm hepes, 10ng/ml cholera toxin, 0.5µg/ml hydrocortisone, 5µg/ml insulin, and 5ng/ml epidermal growth factor (egf). 2) the tissues were subsequently minced into small pieces. 3) the pieces were differentially centrifuged, and 4) placed into three different cultures labeled “organoid,”   “epithelial,”   and   “mesenchymal”   based on differential centrifugation (figure 2). cells were put onto primary culture plates with the supplemented dmem media. 5) after several passages, the cells were then transferred to regular plates, and grown on rpmi media supplemented with 10% fbs. 6) once the cells were homogenous in morphology and could survive 2-3 passes, the cells  were  considered  “established.”  and  frozen. figure 2: schematic illustration of the cell separation procedure with the representative phase table 1: description of the marker used to characterize the cell lines. cell line characterization cell lines were taken for protein quantification once they were established and 20 70 percent confluent. protein was quantified using the bradford method and 15 micrograms of proteins per gel lane was used. the table below describes each marker tested for and the reasoning behind it (table 1). proliferation rate the proliferation rate of cells was also measured to see the relative rates of growth. cells were plated at 50,000 cells per well for npk cells and 100000 cells per well for npm cells in 6-well plates in rpmi media with 10% fbs and counted every day for five consecutive days results cell line establishment npm and npk cell lines were successfully established in culture from the protocol outlined above. five different strains were established from the npm and npk tumors: npk primary prostate tumor (npk-pt), npk lung metastasis (npk-lm), npk hepatic metastasis (npk-hm), npm primary tumor (organoid, npm-pt1), npm primary tumor (epithelial, npm-pt2) figure 1. the difference between the organoid and epithelial cells of origin for the npm cells is that they were taken from different stages of the differential centrifugation process and grown separately (figure 3). protein marker actin expressed in all cells, used to normalize western androgen receptor expressed in prostate cells cytokeratin 5 basal marker, expressed in epithelial cells of basal origin cytokeratin 8 luminal marker, expressed in epithelial cells of luminal origin e-cadherin epithelial cell marker phospho akt and total akt evidence of mtor activation, consequence of pten deletion phospho erk and total erk evidence of kras activation phospho s6 and total s6 evidence of mtor activation, consequence of pten deletion vimentin expressed in stromal cells columbia undergraduate science journal open-access publication | cusj.columbia.edu spring 2013 | volume 7 columbia undergraduate science journal 7 figure 3: pictures of the different cell lines, at 20x magnification. a. npk-pt cells in culture, b. npk-lm cells in culture, c. npk-hm cells in culture, d. npm-pt1 cells in culture, e. npm-pt2 cells in culture proliferation rate the doubling times of the npk pt, lm, and hm cells were found to be around 17 hours. the doubling time for the npm cells was found to be 27 hours (figure 4). figure 4: growth curves of npk-pt, npk-lm, npk-hm, and npm cells. western blot western blotting of five samples (npkpt, npk-lm, npk-hm, npm-1 and npm-2) was conducted. the npk cells all show high cytokeratin 8 (ck 8), as well as expression of phosphos6 and phosphoakt. they do not appear to express ck5 at high levels and their ecadherin expression is not very high especially compared to the npm cells. all npm cells show high expression of ck 8 and e-cadherin as well as expression of phospho-akt and phospho-s6. npm cells do not express cytokeratin 5 at high levels (figure 5). figure 5: western blot analyses using total protein extracts prepared from the different cell lines, npk-pt, npk-lm, npk-hm, and npm cells. discussion in the speirs et al. paper, they experienced a rate of success with breast tissue of around 30% with the full establishment of cell lines. we experienced a very similar success rate of around 33% from our own attempts. it is important that we now have a protocol for setting up cell lines from our mouse models, and my future research will focus on setting up lines for all the models as well as characterizing them in the ways described previously. after the cell lines were established, it became clear that their morphologies were 0 24 48 72 96 0 1000 2000 3000 4000 npk-pt npk-lm npk-hm npm-pt time (h) ce ll n um be r ( x1 03 ) columbia undergraduate science journal open-access publication | cusj.columbia.edu spring 2013 | volume 7 columbia undergraduate science journal 8 quite different. the cell lines from the npm clones appeared to be more epithelial in nature, as shown below when compared to breast epithelial cells from [5] and the kras tumors cells look more fibroblastic. previously it was shown in the lab that kras tumors contain emt (epithelial mesenchymal transition) cells and western blot supports this (figure 6) [3]. figure 6: comparison of experimental cells with known epithelial cells and cells undergoing emt. a. npm-pt1 cells b. breast epithelial cells c. breast cells undergoing emt d. npk-pt cells the graph showing cell proliferation reflects the in vivo findings that kras mice have especially aggressive tumors that grow very quickly, while the npm mice have slower growing tumors [7]. the western blots helped confirm that both the npm cells were indeed epithelial cells with pten deletion. the presence of e-cadherin in all the cells signifies that they are at least partly epithelial in nature, and the fact that it is expressed to higher levels in the npm cells is consistent with their more epithelial morphology. as npk cells did not express ecadherin as strongly, it can be said that they are not as epithelial in nature, supporting the idea that they are undergoing emt. the fact that they all express phosphor-s6 as well as phosphor-akt indicates that the akt/mtor pathway has been activated in response to pten deletion. when prostate cancer cells take to culture, it has been shown that they become more luminal as opposed to basal in nature, and that occurred in these cells as well. ck8 (a luminal marker) was strongly expressed in all cells while ck 5 (a basal marker) was not as highly expressed [6]. the western blot for ar, vimentin, and phosphor-erk will be performed as part of future research. ar would confirm the prostate nature of the cells, vimentin would indicate how mesenchymal they are in nature (which could confirm the suspicion that the kras tumors are emt in nature), and phosphor-erk would show activation of kras. in the future, myc also will be tested to make sure that npm cells express myc. further research will include the establishment and characterization of other cell lines from others gem models that show different genetic alterations, as well as drug screening, xenograft studies, and gene knockdown studies, as discussed in the introduction. this project provides a protocol and a basis for the future development and characterization of additional cell lines. references 1. vishnu p and tan ww. update on options for treatment of metastatic castrationresistant prostate cancer. onco targets ther. 2010; 3:39-51. 2. kinkade cw, castillo-martin m, puzio-kuter a, yan j, foster th, gao h, sun y, ouyang x, gerald wl, cordon-cardo c and abate-shen c. targeting akt/mtor and erk mapk signaling inhibits hormone-refractory prostate cancer in a preclinical mouse model. j clin invest. 2008; 118(9):3051-3064. 3. floc'h n, kinkade cw, kobayashi t, aytes a, lefebvre c, mitrofanova a, cardiff rd, califano a, shen mm and abate-shen c. dual targeting of the akt/mtor signaling pathway inhibits castration-resistant prostate cancer in a genetically engineered mouse model. cancer res. 2012; 72(17):4483-4493. 4. speirs v, green ar, walton ds, kerin mj, fox jn, carleton pj, desai sb and atkin sl. shortterm primary culture of epithelial cells derived from human breast tumours. british journal of cancer. 1998; 78(11):1421-1429. columbia undergraduate science journal open-access publication | cusj.columbia.edu spring 2013 | volume 7 columbia undergraduate science journal 9 5. sendurai a. mani, wenjun guo, mai-jing liao, elinor ng. eaton, ayyakkannu ayyanan, alicia y. zhou, mary brooks, ferenc reinhard, cheng cheng zhang, michail shipitsin, lauren l. campbell, kornelia polyak, cathrin brisken, jing yang, robert a. weinberg. the epithelial-mesenchymal transition generates cells with properties of stem cells. cell 16 may 2008 (vol. 133, issue 4, pp. 704-715). 6. xin l, lukacs ru, lawson da, cheng d and witte on. self-renewal and multilineage differentiation in vitro from murine prostate stem cells. stem cells. 2007; 25(11):27602769 7. wang xs, shankar s, dhanasekaran sm, ateeq b, sasaki at, jing x, robinson d, cao q, prensner jr,yocum ak, wang r, fries df, han b, asangani ia, cao x, li y, omenn gs, pflueger d, gopalan a, reuter ve, kahoud er, cantley lc, rubin ma, palanisamy n, varambally s, chinnaiyan am. characterization of kras rearrangements in metastatic prostate cancer. cancer discovery. june 2011, 35-43. http://www.ncbi.nlm.nih.gov/pubmed?term=wang%20xs%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=shankar%20s%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=dhanasekaran%20sm%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=dhanasekaran%20sm%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=ateeq%20b%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=sasaki%20at%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=jing%20x%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=robinson%20d%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=robinson%20d%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=cao%20q%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=prensner%20jr%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=yocum%20ak%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=wang%20r%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=wang%20r%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=fries%20df%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=han%20b%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=asangani%20ia%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=cao%20x%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=li%20y%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=li%20y%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=omenn%20gs%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=pflueger%20d%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=gopalan%20a%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=reuter%20ve%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=reuter%20ve%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=kahoud%20er%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=cantley%20lc%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=rubin%20ma%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=rubin%20ma%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=palanisamy%20n%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=varambally%20s%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=chinnaiyan%20am%5bauthor%5d&cauthor=true&cauthor_uid=22140652 http://www.ncbi.nlm.nih.gov/pubmed?term=chinnaiyan%20am%5bauthor%5d&cauthor=true&cauthor_uid=22140652 dear readers, founded in 2006, the columbia undergraduate science journal was created by a team of graduate students who wanted to provide a collegiate platform for undergraduates to publish their scholarship and to educate students about the academic publication process. with guidance from advisors and faculty, the editorial board strove to produce a professional-level, open-access science journal that provides participants with a valuable experience in publication. since then our board has shifted to being solely composed of undergraduates. while still maintaining our core editorial standards, the journal has evolved to more explicitly try and foster a supportive student research community on our morningside campus. it is this notion of community that can help us cope and process the unfamiliar circumstances of a global pandemic. in quarantine, student researchers can reflect on the utility of science and its capacity to contain and eradicate this crisis. the collegiate scientists of today should recognize this moment for its affirmation in the power of collaboration and community. recall this moment as you go onto make the next breakthroughs in the natural and applied sciences. this year, the columbia undergraduate science journal formalized our intentions for community building, made inroads in connecting previous authors and alumni of our publication, and developed our editorial process further with open journal infrastructure. i was impressed by the quality of our submissions this year and i want to thank each of the authors for their cooperation in the compilation of our journal. i am incredibly proud of what we have accomplished this year. thank you all for your support in the growth of the student research community. the columbia undergraduate science journal editorial board is proud to present the 14th annual columbia undergraduate science journal. congratulations to the authors, thank you to our readers, and enjoy! sincerely, jason mohabir president cusj letter 5 treating neurodegenerative disorders: towards detailed pathophysiology and precision medicine cusj 2018 ________________________________________________________________________________ abstract — the following paper reviews current literature on disease etiology, pathophysiology, and therapeutic options of neurodegenerative disorder (nds). we first discuss current directions of disease management and treatment, and highlight technical hindrances of current gene therapy-based treatments as well as significant ethical, regulatory, and efficacy-based concerns present in current stem cell treatment options for alzheimer’s, parkinson’s and huntington’s disease. we then suggest a narrower focus for future research on identifying molecular, pathological, and biomarker-based disease progression that can in turn inform the development of novel approaches in developing nd therapies. introduction neurodegenerative diseases (nds) encompass a wide range of acute and chronic conditions characterised by the loss of neuron and glial cells in the brain or spinal cord. in acute conditions, due to stroke or spinal cord injury, different types of neurons within a restricted brain region die over a short amount of time. in chronic conditions, there is either a selective neuronal loss of specific sub-populations (such as dopamine neurons in parkinson’s or medium spiny neuron loss in huntington’s disease), or widespread degradation of many neuronal types over several years. for both conditions, no cures currently exist. current clinical approaches adopted can only target symptomatic treatment and disease management, while clinical applications of novel therapies developed for nds have been limited. this paper reviews the current literature on both implemented and developing therapies for alzheimer’s (ad), parkinson’s (pd), and huntington’s (hd) disease, three of the most common and debilitating chronic nds. the symptoms, affected brain regions, and some molecular changes have been identified for several nds, some of which are summarized in table 11. novel therapies have been developed in animal models to target these observed pathological and molecular changes, but significant ethical, economic, regulatory, safety, and etiological weaknesses still hinder their widespread clinical application. within the context of the two main developing therapies examined, gene and stem cell therapy, specific focuses for research are suggested. major milestones in research that must 1 table 1 based on: http://www.nhs.uk/conditions; http://www.neural.org.uk/store/assets/files/20/original/neuronumbers.pdf; be reached before clinical application begin with better understandings of disease initiation and progression. molecular pathological classification through the identification of biomarkers can provide the greatest cross-methodological benefits in furthering effective and personalized treatment of nds. fundamentally, a more thorough understanding of underlying nd pathophysiology can be accomplished through biomarker identification, stem cell-based disease modelling, and genetic research. combined, this knowledge will help develop more personalized and ultimately more effective novel therapies for treating neurodevelopmental disorders. part 1: current approaches: disease management current clinical approaches to addressing ad, pd, and hd all center on disease management and symptom alleviation. the use of pharmacological agents, physical therapy, and caregiving aid is efficient and affordable, but are ultimately only forms of palliative care rather than effective treatments. four acetylcholinesterase inhibitors (donepezil, galantamine, rivastigmine, tacrine) and one nmda receptor antagonist (memantine) constitute the five main medications currently used to provide temporary symptomatic relief to ad patients (desai & grossberg, 2005). the first four enhance cholinergic function by reducing rates of acetylcholine breakdown (as reduced ach levels is related to memory and cognitive decline in ad), while memantine reduces adrelated excitotoxicity (observed in ad & related to neuronal death). further behavioural, psychosocial, and caregiving provisions are used to improve quality of life for ad patients. current therapies for pd are arguably the most welldeveloped and effective among these three discussed nds. drugs such as levodopa or dopamine agonists used in earlier stages of care can restore some motor function, but their efficacy decreases dramatically over time. deep brain stimulation (dbs) and rehabilitation techniques provide further symptomatic therapies for pd (mizuno, 2014). treatments for hd, like those for ad and pd, are only able to reduce symptom severity. various pharmacological treatments are available to reduce chorea. as the disease progresses, physical, speech, and language therapies are used to rehabilitate physical and cognitive function (busse et al., 2012). https://www.ncbi.nlm.nih.gov/books/nbk1352/; https://neurodiscovery.harvard.edu/challenge treating neurodegenerative disorders: towards detailed pathophysiology and precision medicine katie gu stanford university 10 ________________________________________________________________________________ cusj 2018 gaps in current treatment techniques techniques currently available to treat neurodegenerative diseases are able to provide temporary symptomatic relief, but are critically unable to halt the rapid deterioration and progression of neurodegeneration. though the complete causes of ad and pd have not been fully characterised, neurofibrillary tangles, -amyloid plaques, decreased synaptic function, and cortical atrophy observed in ad and the loss of dopaminergic neurons characteristic of pd can be addressed by gene therapy and stem cell-based therapies. these two therapies can also be applied to treat the root causes of hd, which has been attributed to an expanded trinucleotide repeats, the mhtt protein, and consequent neostriatal neuronal loss. however, their widespread clinical application has been limited by underdeveloped understandings of disease etiology and mechanisms of therapeutic action. insights into the molecular mechanisms of nds lie at the crux for the development of novel targets of disease-modifying therapies. therefore, efforts to further gene and stem cellbased therapies (discussed in parts 2 & 3) will benefit from an increased focus on biomarker research (discussed in part 4) to elucidate disease pathophysiologies. part 2: gene therapy2 the theory behind the use of gene therapy to treat neurodegenerative disease is simple: deliver a transgene that either corrects or replaces a defective gene to cure a complex disease using a transfer vector. however, the numerous technical difficulties associated with gene therapy are daunting in complexity: the transfer vector must be easy to produce, safe for patients and the environment, target specific cells efficiently, and not trigger a deleterious immune reaction; simultaneously, the appropriate transduced transgene must be expressed at the optimal level for an optimal duration (sarkis & mallet, 2005). all these factors that must be taken into consideration can only be optimized with increased knowledge of disease etiology, which can be mediated through biomarker research. gene therapy trials gene therapy vectors are either viral or non-viral. currently, the most common viral vectors are adeno-associated viruses (aavs) and lentiviruses. both infect non-dividing and dividing cells, but lentiviral vectors integrate into host genomes while aavs do not. thus, lentivirus integration can confer long-term, stable transgene expression, but poses a risk of integrational mutagenesis. non-viral vectors (mostly used for hd treatment) have more localized effects, require higher 2 this section is based on readings from o’connor & boulis (2015), as accredited in works cited dosage levels, and confer only transient gene expression that is normally insufficient to fully treat chronic nd. viral techniques a series of viral vector-mediated gene therapy trials has been conducted for ad and pd, summarized in the table below. in these trials, aav2-mediated transgenes were injected into patients’ brains, but treatment results either fail to meet or achieve the same level of therapeutic benefit as current standards of care. it is important to emphasize that the efficacy of gene therapy depends critically on the therapeutic gene delivered; this can only be improved by better understandings of the etiology of neurodegenerative diseases. non-viral techniques huntington’s is a genetically acquired disease with a known etiology (cag repeats in the iti5 huntington gene), for which non-viral gene therapy techniques carry significant therapeutic potential. carroll et al. (2011) have developed asos that selectively bind to exonic and intronic snp (single nucleotide polymorphisms) sites on the mutant huntingtin protein (mhtt), thus selectively silencing gene expression of mutated regions on the mhtt. the experiment used primary cells from hdpatients with transgenic mouse lines, and the developed asos achieved potent and selective allele-specific knockdown of mhtt both in vitro and in vivo. this presents the most exciting potential treatment for huntington’s, as aso delivery to adult cns can be achieved through simple intrathecal or intracerebroventricular infusion rather than viral delivery. further, aso dosing can be precisely controlled or stopped when necessary. kordasiewicz et al. (2012) have shown that transient infusion of asos catalysing rnase h-mediated degredation of mhtt in mouse and non-human primate models delays disease progression and sustains hd disease reversal that outlasts aso-mediated mhtt knockdown. thus, clinical trials should be conducted to test the efficacy of gene therapy either through direct aso infusion, or combined with ipsc-mediated transplantation (correcting disease mutations of patient-derived ipscs with asos, then transplanting corrected npcs). low efficacy in gene therapies promising proof-of-principle concepts have been presented in phase i trials, which have shown that gene therapy techniques targeting ad & pd are safe and welltolerated. however, therapeutic efficacy in further clinical trials remains weak, in that observed improvements failed to exceed placebo effect or current standards of care for ad and 11 cusj 2018 ________________________________________________________________________________ pd. non-viral aso-mediated silencing of mhtt expression has proved efficacious in hd-animal models but have not yet moved to clinical trials. in sum, success by design meets failure of efficacy within these gene therapy studies. moving forward, future research must focus first on expanding choices of possible transgenes for delivery, as improvements in vectors and delivery methods depend crucially on efficacy of the transgene delivered. greater understanding in disease etiology will help identify these gene targets. a significant area of development lies in the identification of biomarkers of nds, which would not only improve gene therapy techniques, but also allow for earlier diagnoses, continuous motoring of the progression of therapeutic effects, and the development of precision medicine. part 3: stem cell therapy3 stem cell and ipsc research has significantly furthered neurodegenerative disease modelling, diagnosis, and transplantand drug-based therapies. a plethora of therapies developed in animal models have failed to translate into effective clinical trials due to differences in mammalian genomes and embryonic development (begley & ellis, 2012). human-based stem cell research generates not only potential nd therapies, but also presents clearer pathophysiological models for those diseases. in particular, the development of induced pluripotent stem (ips) cells has presented a unique approach for studying signalling pathways, growth control, and disease mechanisms in previously inaccessible human brain tissue. the main types of stem cells used for neurodegenerative therapies are embryonic, mesenchymal, progenitor, and ips cells (singh et al., 2016). embryonic stem cells are pluripotent and hold excellent potential to restore damage caused by brain injury or neurodegeneration; however, their ability for unlimited self-renewal poses a high risk for tumour formation postengraftment, limiting the range of their clinical application. mesenchymal stem cells are multipotent and immunomodulating (derived from autologous source; do not trigger a host immune response), and thus find a relatively widespread application for neurodegenerative treatment. however, their use in genetic diseases, wherein autologous sources contain the same genetic predisposition to the disease, is limited. neural progenitor cells (npc) and neural stem cells (nsc) are derived from either fetal or adult neural tissue, carry limited differentiation and tumour-formation potentials, but are difficult to isolate and thus lower in availability. ipscs, generated from somatic 3 this section is based on readings from singh et al. (2016), kumar et al. (2016), lunn et al. (2011), goldman (2005), and chmielnicki & goldman (2005), accredited in works cited tissue and reprogrammed using tfs into embryonic stem cell-like states, offer novel sources of autologous cellular therapy. application of the appropriately selected type of stem cell to treat neurodegenerative diseases is mediated through two primary mechanisms. first, stem cells can replace or stabilize neuronal networks that experience both widespread or specific subpopulation-based neuronal loss. second, stem cells can provide environmental enrichment through neurotrophic support to prevent further neuronal degradation. these mechanisms will be discussed in the context of specific neurodegenerative diseases below. following this discussion, i’ll emphasize that the focus of stem cell research should first lie in exploiting its means of modelling disease pathophysiology and providing an in vitro screening tool for screening efficacy and safety of drugs, then on its direct therapeutic applications. i’ll then highlight the potential ipsc research holds in future developments of precision medicine. alzheimer’s complementing current pharmacological ad treatments that regulate neurotransmitter activity (section 1.1), stem cell (sc)-based therapies for ad can also help restore degeneration and loss of cholinergic neurons, and increase levels of brain-derived neurotrophic factor (bdnf) and nerve growth factor (ngf) that are lowered in ad. in mouse models, combinatorial upregulation of bdnf with npc transplants promoted neurogenesis and protected neuronal function. thus, npc-mediated stem cell-based therapy may offer an enhanced approach to preventing adassociated degradation of cholinergic synapses and neurons (tuszynski, 2007). wang et al. (2006) compared transplants of escderived nsc to transplants of esc alone into mouse ad models. as suggested in section 3.0, esc-transplanted mice developed teratomas (escs: unlimited self-renewal capabilities, high risk of tumour formation), while transplanted nscs successfully differentiated into stable cholinergic neurons, improving memory and learning. genetically modified nscs for bdnf expression has been shown in many rodent studies to reduce ad symptoms (lindvall & kokaia, 2010). in these mouse models, increased hippocampal synaptic density, increased cognitive function, and enhanced cell-cell synaptic communication, as associated with increased bdnf expression (blurton-jones et al., 2009; li et al., 2009; xuan et al., 2008). ngf has also been implicated in preventing neuro-degeneration and reducing amyloid toxicity 12 ________________________________________________________________________________ cusj 2018 (tuszynski et al., 2007). in addition to aav2-mediated gene therapy techniques discussed in section 2.1.1, autologous fibroblasts genetically modified to express ngf implanted into patient forebrains have been found to slow rates of cognitive decline in an open-label phase 1 clinical trial (tuszynski et al., 2005). however, fibroblasts are immobile and cannot slow degeneration across large brain regions. transplanted stem cells can efficiently migrate and release growth factors to many damaged sites, but their use in ad treatment has yet to be fully explored. another technique to increase ngf release to treat ad is with encapsulated cell biodelivery (ecb) implants, for which the first clinical trial was conducted in 2008 by the danish company nsgene. encapsulated retinal pigment epithelial cells releasing ngf were implanted into the basal forebrains of six ad patients (wahlberg et al., 2012). at 12 months, implants were retrieved and low but persistent ngf secretion was detected in half of the subjects. thus, both ecb implantation and stem cell-based gene therapy may provide future methods for enhancing ngf expression to counteract cholinergic neuronal death. although nsc transplantations into the adult forebrain of patients may be able to increase cholinergic and cognitive aptitude, they are unable to target the main molecular cause of the disease: accumulation of β-amyloid plaques (blurtonjones et al., 2009). nonetheless, the ability of nsc transplants to enhance cholinergic function, synaptic transmission, and neuronal function is suggested by animal studies and replicated by initial clinical trials (tuszynski, 2007). combining growth factor overexpression with npcs’ ability to integrate across many neural networks thus presents an exciting potential for limiting ad pathogenesis. continued research must be conducted before established sc-based treatments can be developed for ad therapy. parkinson’s earlier open-label studies of human fetal ventral mesencehalic tissue (hfvm) transplants in parkinson’s patients yielded successful long-term pd symptomatic relief, but these results failed to replicate in subsequent placebo-controlled, double-blind trials (brundin et al., 2010). success has been observed in using genetically modified escs and ipscs to derive neurons with the dopaminergic (da) phenotype, but these techniques have not yet been translated into clinical studies. clinical trials using hfvm tissue transplants in the 1990s (madrazo et al., 1988; lindvall et al., 1990) found impressive restoration of dopaminergic neurotransmission and motor-symptom alleviation. in these open-label clinical studies, pet scans showed long-lasting evidence of functionally-integrated dopaminergic neurons forming from transplanted tissue, restored da release in vivo, and reactivation of cortical motor areas. however, the design of these open-label trials contains significant bias and thus deserve cautious interpretation. further, the occurrence of recurring graft-induced dyskinesias (gids) in consequent double-blind trials, in addition to heavy ethical concerns raised over the use of fetal tissue as a therapeutic source, has heavily impeded the clinical development of this form of hfvm-based therapy (brundin et al., 2010). instead, the focus has been shifted to escs and ipscs: mouse model studies have found that overexpression of nurr1, fgf-8, and shh signalling in mouse escs and ipscs has successfully induced midbrain dopaminergic phenotypes and improved functional recovery (kim et al., 2003; cooper et al., 2010). a major concern for esc clinical transplants, however, is the risk for tumour formation observed in animal models (lindvall & kokaia, 2010). as pd patients have normal life expectancies, even a slim risk of tumour formation is unacceptable. initial suggestions of hfvm transplant-mediated pd therapy have been significantly hindered by ethical and methodological issues, while the induction of escs and ipscs into dopaminergic neurons is cumbersome, potentially nonhomogeneous (may contain traces of undifferentiated cells that can lead to teratomas), and currently unsubstantiated by clinical studies. thus, progress made in stem cell-based pd treatments have been slower than for ad. however, ipsc-derived dopaminergic neurons have provided critical insights into the pathophysiology of pd, by providing a sc-based human disease model. abnormalities in mitochondrial and dopamine homeostasis, as well as mutations in parkin and pten-induced putative kinase 1 (pink1) previously noted in animal models have been substantiated by similar data from studies using pd patient-derived ipsc neurons (van laar et al., 2010; cai et al., 2012). though stem cell technology has yet to develop clinically-relevant pd therapies, it has proved effective in modelling parkinson’s. in particular, mutation-bearing ipscs has helped study pd physiology, elucidate underlying signalling cascades modulating the disease, identify pd biochemical markers, and screen efficiency of therapeutic compounds. huntington’s applications of stem cell techniques in hd studies have paralleled those for ad and pd. more specifically, fetal striatal grafting into hd patients have provided transient symptom improvements, but its mechanisms are poorly understood and its use is limited by ethical constraints (bachoud-lévi et al., 2000). both ipscs and escs have been 13 cusj 2018 ________________________________________________________________________________ used to establish differentiation protocols for medium spiny neurons (msn)— these transplantations have shown successful neural circuit development and integration, but are currently restricted to animal models (viegas et al., 2011; precious & rosser, 2012). initial clinical trials using fetal-derived tissue, ipscs, or escs as sources of msns have observed reduced motor and cognitive dysfunction in hd patients. in addition, decreased neuronal loss was observed in hd rodents with npc transplants engineered to overexpress gdnf, as compared to control mice with unmodified nsc transplants, emphasizing the potential of stem cell-based therapies to provide environmental enrichment and protect endogenous neurons (pineda, 2007). for ad, pd, and hd, similar trends in stem cell-based therapies can be observed. early cell-based therapies using fetal ventral mesencephalic tissue (for pd) and fetal striatal tissue (for hd) transplants provided initial support for successful symptomatic relief. however, widespread clinical applications of human fetal tissue transplants has been limited by significant ethical and scarcity issues associated with the nature of the tissue source. subsequent clinical trials have failed to consistently replicate initial data and observed deleterious side effects (teratoma formation, gids, immunorejection). thus, the focus of stem cell therapies has shifted to escs derived nscs induced to certain cell fates or gene expressions (increased bdnf/ngf expression for ad, dopaminergic phenotypes for pd, and msn for hd); however, these induced populations are usually not pure, long-term graft survival is generally low, and any es cells escaping in vitro differentiation carries the potential or yielding teratomas and tumours following transplantation. above all, the development of effective stem cell-based therapies is ultimately hindered by our current lack of knowledge on the mechanisms of action of transplanted stem cells, and how stem cell proliferation, differentiation, survival, and migration are controlled in a pathological environment. the mechanisms of cellular protection provided by sc-mediated trophic support are also not fully understood (mcbride et al., 2004). because current potential risks outweigh validated benefits in their therapeutic applications, research on human stem cells and patientderived ipscs should first be focused on disease modelling and used for screening drugs and other therapeutics. successful development of clinical therapies can begin once more precise disease pathophysiologies and mechanisms of sc action post-transplantation are identified. in sum, the use of stem cells for transplantation has generated enormous ethical, technical, efficacy, and safetyrelated debates. it is true that sc replacement therapy can stabilize and regenerate entire networks of neuronal loss and prevent further neurodegeneration by providing environmental and trophic factors against toxic, neurodegenerative factors. however, with current literature, the greatest functionally validated potential of scs lies in their unique ability to model disease pathophysiology and mechanisms of action. patient-derived ipsc cells may be a particularly tremendous treatment option for personalized medicine in the future, but for now serve as prominent tools for understanding disease etiology on a highly personalized level, as they carry the patient’s genotype, disease mutations, and also account for environmental influences on the individual’s disease. part 4: precision medicine & biomarkers in discussing future developments of neurodegenerative therapies, it is critical to emphasize the importance of elucidating the molecular etiologies of nds. pathologically altered proteins and protein aggregations are fundamental characteristics of neurodegenerative diseases and should be of paramount focus for further research. i’ve alluded to the importance of biomarker research in aiding developments of novel therapies during part 2 & 3’s discussion of gene therapy and stem-cell based treatments. this is because molecular classifications of nds can detail the nature of pathologically dysfunctional proteins and correlate corresponding clinical symptoms to genetic abnormalities. ongoing biomarker research has already identified extracellular deposits of a fibrils and aggregation of abnormally-phosphorylated tau proteins in ad, the presence of lewy body deposits, altered dopamine transporters in pd, and protein misfolding and aggregation associated with mhtt proteins in hd (kovacs, 2016). thus, the ability of biomarker data to identify core disease pathology has already been validated, and therapies targeting these protein abnormalities (through immunotherapy, gene therapy, sc-transplants) have yielded promising results in animal and some clinical models. however, the most exciting potential of biomarker research lies in its ability to characterise patient-specific environment-disease dynamics and gene variations, and consequently provide personalized diagnoses. in the emerging era of precision medicine, the use of biomarkers (detected in body fluids or pet imaging) can provide earlier and more accurate nd diagnosis, and their identified protein dysfunctions can be targeted in translational research using novel therapies. looking forward, biomarker research will contribute greatly to developing pathophysiological disease models, improving immunotherapy, gene therapy and sc-based transplant methods, and moving us closer to personalized 14 ________________________________________________________________________________ cusj 2018 cures for neurodegenerative diseases. similarly, stem cell research can be applied to experimentally model nds. patient-derived ipscs are especially well-suited to uncover specific disease pathophysiologies as they carry the patient’s genotype and bear patient-specific disease mutations and gene-environment interactions. thus, i believe that research focusing on biomarkers, patientderived ipscs, and genetic studies are currently most necessary and potentially most beneficial for current experimental and clinical endeavours for nd cures. table 1: overview of neurodegenerative disorders disorder ~incidence (total in us) brain region affected molecular changes symptoms alzheimer’s disease (ad) 5,000,000 hippocampus, amygdala, neocortex, basal forebrain · neurofibrillary tangles (hyperphosphorylation of tau proteins) · β-amyloid plaques (amyloid peptide aggregates) dementia (cognitive, psychiatric, behavioural) amyotrophic lateral sclerosis (als) 30,000 motor neurons · motor neuron degeneration in spinal cord, brain stem, primary motor cortex difficulty walking, weakness in limbs/arms, slurred speech, muscle cramps and twitching parkinson’s disease (pd) 1,000,000 substantia nigra · dopaminergic neuron loss from nigrostriatum and substantia nigra pars · build-up of lewy bodies, α-synuclein aggregates tremors, bradykinesia, impaired balance and posture, loss of automatic movements huntington’s disease (hd) 30,000 basal ganglia, neostriatum · trinucleotide (cag) repeat in iti5 huntington gene · loss of medium spiny neurons (msn) · toxicity of mhtt: protein misfolding & aggregation muscle problems (rigidity, dystonia), involuntary jerking movements, abnormal eye movements spinal muscular atrophy (sma) 9,000 motor neurons mutations in smn1 gene muscular atrophy, balance and coordination deficits table 2: current therapies for ad, pd, & hd disease pharmacological treatments other treatments alzheimer’s disease · acetylcholinesterase inhibitors (donepezil, galantamine, rivastigamine, tacrine) · nmda receptor antagonist (memantine) · caregiving · behavioural therapy (reduce problem behaviours) · psychosocial interventions (emotion and cognition regulation) parkinson’s disease dopamine replacement (levodopa, carbidopa, selegeline) · deep brain stimulation (of thalamus, globus pallidus, subthalamic nucleus) · physical therapy huntington’s disease chorea treatment (tetrabenazine, benzodiazepines, neuroleptics) · caregiving · physical therapy · speech and language therapy 15 cusj 2018 ________________________________________________________________________________ table 3: viral vector-mediated gene therapies (ad & pd) disease study / trial codes gene therapy method of delivery current status alzheimer’s disease rafii et al., 2014 (nct00876863) aav2-ngf - ngf: nerve growth factor // dysregulated in ad, neuroprotector/cellular therapy efficacy injection into basal forebrain · phase 1 successful*, therapy well-tolerated, ngf-induced axonal sprouting observed · phase ii in progress *success defined by meeting safety requirements and showing functional improvement parkinson’s disease lewitt et al., 2011 (nct00643890) aav2-gad - gad: glutamic acid decarboxylase // ratelimiting biosynthetic enzyme for gaba, reduces excitoxicity injection into subthalamic nucleus phase i & ii successful, but symptom improvements achieved were not better than current standards of care parkinson’s disease bartus et al., 2011 (nct00400634; nct00985517) aav2-nrtn - nrtn: neurturin // delivers support to dopaminergic neurons injection into striatum & putamen · phase i successful, observed increased nrtn expression · phase ii showed no improvement in clinical outcome over controls parkinson’s disease azzouz et al., 2002 (nct00627588; nct01856439) lentivirus-th, aadc, gch1 (*exciting tricistronic vector: cells transduced express all 3 enzymes) - th: tyrosine hydroxylase / aadc: aromatic amino acid dopa decarboxylase / gch1: gtp cyclohydrolase i // three genes encode rate-limiting dopamine biosynthesis enzymes injection into striatum & putamen · phase i (open label) successful, improved motor function correlating with increasing dose · currently optimizing delivery method before phase ii (double-blinded randomized trial) parkinson’s disease muramatsu et al., 2015 (nct02418598) aav2-aadc - (as discussed above) injection into putamen · phase i safety trial successful, stable aadc expression 4 years postdelivery · phase ii recruiting, in progress parkinson’s disease (nct00229736) aav2-aadc intrastriatal infusion · phase i safety trial successful 16 ________________________________________________________________________________ cusj 2018 table 4: overview of current literature on sc-based nd therapies disease stem cell applications goal of sc-based therapies current therapeutic progress alzheimer’s disease · nscs · escs · restore loss of cholinergic neurons · increase bdnf & ngf trophic factors to promote neurogenesis · successful basal forebrain grafts of ngfsecreting fibroblasts (open-label trial, tuszynski, 2007) · development of encapsulated cell biodelivery implants to increase ngf release (nsgene) · no current clinical trials testing sctransplants in ad parkinson’s disease · escs · restore nigrostriatal dopaminergic neuronal loss · successful clinical open-label hfvm transplants (lindvall et al., 1990; madrazo et al., 1988) — provided long-term symptomatic relief, but failed to replicate in subsequent double-blind trials (which observed gids) · successful animal studies using escs & ipscs to derive da phenotype, not yet translated to clinical trials due to risk of tumour formation and technical difficulties huntington’s disease · msns · restore medium spiny neuron loss · successful clinical fetal striatal grafting (bachoud-lévi et al., 2000)— provided transient symptom improvements, but limited by ethical constraints & poorly-understood mechanism of action · successful animal studies using escs & ipscs to restore msn loss, not yet translated to clinical trials 17 cusj 2018 ________________________________________________________________________________ works cited [1] azzouz m., martin-rendon e., barber r.d., mitrophanous k.a., carter e.e., rohll j.b., kingsman s.m. (2002). multicistronic lentiviral vector-mediated striatal gene transfer of aromatic l-amino acid decarboxylase, tyrosine hydroxylase, and gtp cyclohydrolase i induces sustained transgene expression, dopamine production, and functional improvement in a rat model of parkinson's disease. j. neurosci., 22 (23), 10302–10312. [2] bachoud-lévi a-c, rémy p, nǵuyen j-p, brugières p, lefaucheur j-p, bourdet c, baudic s, gaura v et al (2000) motor and cognitive improvements in patients with huntington’s disease after neural transplantation. lancet 356(9246):1975–1979 [3] bartus, r.t., brown l., wilson, a., kruegel, b., siffert, j. properly scaled and targeted aav2-nrtn (neurturin) to the substantia nigra is safe, effective and causes no weight loss: support for nigral targeting in parkinson's disease. neurobiol. dis., 44 (2011), 38–52. [4] blurton-jones, m., kitazawa, m., martinez-coria, h., castello, n. a., müller, f.j., loring, j. f., et al. (2009). neural stem cells improve cognition via bdnf in a transgenic model of alzheimer disease. proc. natl. acad. sci. u.s.a. 106, 13594–13599. [5] brundin, p., barker, r., parmar, m. (2010). neural grafting in parkinson’s disease: problems and possibilities. progress in brain research, vol. 184, chapter 14. [6] busse, m., khalil, h., brooks, s., quinn, l., rosser, a. (2012). practice, progress and future directions for physical therapies in huntington’s disease. journal of huntington’s disease. 1(2), 175-85. [7] carroll, j. b., warby, s. c., southwell, a. l., doty, c. n., greenlee, s., skotte, n., et al. (2011). potent and selective antisense oligonucleotides targeting single-nucleotide polymorphisms in the huntington disease gene/allele specific silencing of mutant huntingtin. mol. ther. 19, 2178– 2185. [8] chmielnicki, e. & goldman, s. stem cells and cell-based therapy in neurodegenerative disease. in beal, m., lang, a., ludolph, a. (eds.), neurodegenerative diseases: neurobiology, pathogenesis and therapeutics (329-40). cambridge university press. [9] cooper o, hargus g, deleidi m, et al. (2010). differentiation of human es and parkinson’s disease ips cells into ventral midbrain dopaminergic neurons requires a high activity form of shh, fgf8a and specific regionalization by retinoic acid. mol cell neurosci. 45:258– 266. [10] desai, a. & gossberg, g. (2005). diagnosis and treatment of alzheimer’s disease. neurology. 64(12), s34-9. [11]ebert ad, barber ae, heins bm, svendsen cn. ex vivo delivery of gdnf maintains motor function and prevents neuronal loss in a transgenic mouse model of huntington’s disease. experimental neurology. 2010; 224:155–162. [12] goldman, s. (2005). stem and progenitor cell-based therapy of the human central nervous system. nature biotechnology. 23(7), 862-70. [13] li g, peskind er, millard sp, et al. (2009). cerebrospinal fluid concentration of brain-derived neurotrophic factor and cognitive function in nondemented subjects. plos one. 4:e5424. [14] lindvall o, brundin p, widner h, rehncrona s, gustavii b, frackowiak r, leenders kl, sawle g et al. (1990). grafts of fetal dopamine neurons survive and improve motor function in parkinson’s disease. science. 247(4942):574–577 [15] lindvall, o., kokaia, z. (2010). stem cells in human neurodegenerative disorders— time for clinical translation? journal of clinical investigation. 120(1), 29-31. [16] kim t. e., lee h. s., lee y. b., hong s. h., lee y. s., ichinose h., et al. . (2003). sonic hedgehog and fgf8 collaborate to induce dopaminergic phenotypes in the nurr1-overexpressing neural stem cell. biochem. biophys. res. commun. 305, 1040–1048. 10.1016/s0006291x(03)00879-9 [17] kordasiewicz, h. b., stanek, l. m., wancewicz, e. v., mazur, c., mcalonis, m. m., pytel, k. a., et al. (2012). sustained therapeutic reversal of huntington’s disease by transient repression of huntingtin synthesis. neuron 74, 1031–1044. [18] kovacs, g. (2016). molecular pathological classification of neurodegenerative diseases: turning towards precision medicine. int. j. mol. sci. 17(189), 1-21. [19] kumar, a., narayanan, k., chaudhary, r., mishra, s. 2016. current perspective of stem cell therapy in neurodegenerative and metabolic diseases. mol. neurobiol., online. [20] sarkis, c. & mallet, j. (2005). gene therapy. in beal, m., lang, a., ludolph, a. (eds.), neurodegenerative diseases: neurobiology, pathogenesis and therapeutics (329-40). cambridge university press. [21] singh, s., srivastava, a., srivastava, p., dhuriya, y. 2016. advances in stem cell researcha ray of hope in better diagnosis and prognosis in neurodegenerative diseases. frontiers in molecular biosciences. 3(72), 1-10. [22] tuszynski, m. h. (2007). nerve growth factor gene therapy in alzheimer disease. alzheimer dis. assoc. disord. 21, 179–189. 18 ________________________________________________________________________________ cusj 2018 [23] tuszynski, m. h., thal, l., pay, m., salmon, d. p., bakay, r., patel, p., et al. (2005). a phase 1 clinical trial of nerve growth factor gene therapy for alzheimer disease. nat. med. 11, 551–555. [24] mizuno, y. (2014). recent research progress in and future perspective on treatment of parkinson’s disease. integr med int. 1(2), 67-79. [25] rafii, m.s. et al. (2014). a phase1 study of stereotactic gene delivery of aav2-ngf for alzheimer's disease. alzheimers dement., 10, pp. 571–581 [26] mcbride jl, behrstock sp, chen ey, et al. human neural stem cell transplants improve motor function in a rat model of huntington’s disease. j comp neurol. 2004; 475:211–219. [27] o’connor, d. & boulis, n. (2015). gene therapy for neurodegenerative diseases. trends in molecular medicine. 21(8), 504-11. [28] precious sv, rosser ae (2012) producing striatal phenotypes for transplantation in huntington’s disease. exp biol med 237(4):343–351 [29] lewitt, p. et al. (2011). aav2-gad gene therapy for advanced parkinson's disease: a double-blind, shamsurgery controlled, randomised trial. lancet neurol., 10, pp. 309–319 [30] xuan ag, long dh, gu hg, et al. (2008). bdnf improves the effects of neural stem cells on the rat model of alzheimer’s disease with unilateral lesion of fimbria-fornix. neurosci lett. 440:331–335. [31] viegas p, nicoleau c, perrier al (2011) derivation of striatal neurons from human stem cells. prog brain res 200:373–404 [32] wahlberg, l. u., lind, g., almqvist, p. m., kusk, p., tornøe, j., juliusson, b., . . . linderoth, b. (2012). targeted delivery of nerve growth factor via encapsulated cell biodelivery in alzheimer disease: a technology platform for restorative neurosurgery. journal of neurosurgery, 117(2), 340-347. doi:10.3171/2012.2.jns11714 19 microsoft word cusj 11.docx ________________________________________________________________________________ cusj 2019 __________________________________________________________________________________________ 11 abstract — the ability to detect the presence of many neurodegenerative diseases during the early stages has been done with limited success. this article will briefly explore biochemical characteristics of alzheimer’s disease (ad), and current methods for detecting ad. these methods will be evaluated against how accurate and invasive these tests are as well as the time required to conduct one of these tests. as well the innovations made for detecting other neurodegenerative diseases and how these methods could be applied for detecting ad in the early stages. how a diagnostic test based off discussed detection principles will also be detailed in addition to the theoretical creation of a fluorescent assay that could be used as a detection method for ad. i. introduction alois alzheimer first diagnosed what would later become alzheimer’s disease (ad) in 1908 (alzheimer et al., 1991). alzheimer’s first patient stated that she felt “that she had lost [herself]” and claimed that she was possessed by demons (alzheimer et al., 1991). alzheimer, not willing to accept this, believed that his patient was suffering from a form of dementia. the disease was discovered after an autopsy was conducted on the patient, after which alzheimer saw the physical renditions of ad. ad is primarily characterized through groups of matter glommed together that alois alzheimer identified as plaques and tangles. since the discovery of the disease, new methods of diagnosing ad have been developed, but each with varying degrees of success. the more invasive direct methods of detecting ad have shown to have a greater accuracy; however, diagnosing the presence of a specific neurodegenerative disease post-mortem provides no tangible benefit to the patient and others suffering from ad. in this area of research, there is a great benefit in creating earlier methods of detection for ad as well as methods that are less invasive. the disease that was discovered over a century ago still eludes physicians and researchers in how to diagnose and treat this condition. this report will focus on evaluating current methods in the diagnosis of ad and future tests that could indicate the disease. these tests will range from physiological exams and imaging methods such as mri which are less invasive, to more invasive methods such as obtaining cerebrospinal fluid for further analysis. the type of ad that will be focused on throughout this report involves late-onset ad. there have been genetic populations which have shown to have a large presence of individuals with ad in very localized geographic regions (sherrington et al., 1995). the mechanisms by which ad manifests itself in late and early onset ad differ widely. often in early-onset populations, there is a large accumulation of plaques and tangles in the individual as early as 20 years old (sherrington et al., 1995). comparatively, mechanisms by which late-onset ad, and the many factors that go into the propagation of it, shows manifestation around the age of 65. generally, individuals with early-onset ad have shown to have proportionally smaller brains than individuals who experience late-onset ad (sherrington et al., 1995). additionally, determining the difference between alzheimer’s and other types of dementia is important in determining how to better diagnose ad. dementia is a more general term used to describe symptoms that impact memory and recall as well as a decrease in the performance of everyday activities, while ad is a specific form of dementia which has shown to get significantly worse over time and affects memory, language, and thought. neither alzheimer’s nor dementia are considered to be a normal part of the aging process. ii. biochemistry of ad plaques and tangles in ad are created by amyloidbeta and tau aggregates, respectively (ballatore et al., 2007; hardy et al., 1992). beta-amyloid results from the amyloid precursor protein when it undergoes fragmentation (hardy et al., 1992). the presence of amyloid beta protein fragments in a healthy brain is relatively low as these fragments are broken down and removed. when large amounts of protein fragments exist in the brain for prolonged periods of time, they can clump and form plaques (hardy et al., 1992). these plaques can block signaling between cells and are known to cause immune responses that can result in the destruction of nerve cells (hardy et al., 1992). tangles are the result of twisted fibers inside of brain cells due to a protein called tau (ballatore et al., 2007). in normal brains, tau forms the structure of the microtubule which allows for the transport of important substances such as food from one nerve cell to another (ballatore et al., 2007). in ad, tau begins to aggregate, causing the collapse of the microtubule structure (ballatore et al., 2007). this leads to twisting which ultimately leads to a disruption of transportation in the cell, resulting in cell death (ballatore et al., 2007). the production of amyloid beta and tau aggregates have been seen to present in large quantities many years before the diagnosis of ad (ballatore et al., 2007; hardy et al., 1992). usually, tau tangles and amyloid beta plaques begin production in the hippocampus which is responsible for short-term memory, implicating short-term memory loss as the first most immediate sign of ad (hyman et al., 1984; ballatore et al., current methods and future research in the diagnosis of alzheimer's disease gaganpreet jhajj1 2 3 * and mark farinas4 1 department of chemistry, university of alberta, edmonton ab t6g 2g2, canada 2 neuroscience and mental health institute, university of alberta, edmonton ab t6g 2e1, canada 3 department of biochemistry, university of alberta, edmonton ab t6g 2r3, canada 4 department of medical microbiology & immunology, university of alberta, edmonton, ab t6g 2e1, canada cusj 2019________________________________________________________________________________ __________________________________________________________________________________________ 12 2007; hardy et al., 1992). from this point, the plaques and tangles invade other areas of the brain, causing a wide host of other symptoms. iii. cognitive exams physiological exams are diagnostic in the late stages of ad and dementia. by this point, the patient often has little chance to deal with the buildup of plaques and tangles and the long-term damage that has already been caused to various parts of the brain. modern physiological examinations may involve simple recall tasks that are tracked over a large period of time to aid in determining the progression of the disease. as the disease progresses through different parts of the brain, individuals can see a loss in speech production, motor skills, and changing temperaments. ad eventually leads to the degeneration of key parts of the brain that control basic functions such as heart rate. diagnosis of ad is often accompanied by individuals forgetting both short and long-term memories which is diagnostic as different memories are stored in different regions of the brain and this can determine how far ad has progressed. in a clinical and familial setting, patients who stop recognizing close relatives and friend were shown to be deep into the progression of ad. one of these tasks involves drawing a clock and filling in the numbers. the clock drawing task shown in figure 1 is part of a larger assessment called the montreal cognitive assessment (moca) (price et al., 2011). figure 1. a comparison of the moca clock drawing task on individuals with ad, vad (vascular dementia), and pdd (parkinson’s disease dementia). figure adapted from price et al., 2011. this test consists of six categories, the first of which is a recall task, and the second of which is a task involving drawing a clock and a specific time, as well as drawing a three-dimensional cube (price et al., 2011). combined, the several cognitive domains that are tested on the moca create a scoring guide out of 30 in which an individual who has a score over 26 is considered to be normal (freitas et al., 2013). individuals with mild cognitive impairment scored an average of 22 and individuals with ad scored approximately 16 (price et al., 2011). various studies have affirmed the role of the moca as a diagnostic assessment for ad and mci; however, it should be noted that there is detection primarily due to the late stage progression of ad in patients (freitas et al., 2013; price et al., 2011). as well, the moca being a specialized assessment looking for cognitive impairments, it is often not administered to patients who may have ad unless there is a strong indication that there is a presence of ad. this is usually the case when the disease has already progressed notably far that it is known there is some form of cognitive impairment. there is an immense need to diagnose ad earlier, to attempt to increase the quality of life for individuals who may be living with ad and increase their life spans. iv. current laboratory tests in a specific study using radiolabeled particles, researchers were able to determine the levels of tau and amyloid beta present in the brain through a variety of imaging techniques (johnson et al., 2012). imaging techniques have their own limitations in the diagnosis of ad, as it is often hard to distinguish specifically which neurodegenerative condition one might have merely through non-invasive imaging techniques such as mri, ct, and pet scans (johnson et al., 2012). additionally, it is not possible for certain individuals to undergo certain methods of imaging which may be due to factors such as metal implants that make mri not feasible and other psychological issues that may prevent someone from undergoing an mri such as being uncomfortable with the loud humming of the electromagnet. there have been studies looking at the usage of a blood test for the detection of ad; however, this is not as indicative as using cerebrospinal fluid (csf), a more valuable diagnostic indicator of ad (o'bryant et al., 2016). the usage of a blood test however, does allow for positive indications of the potential presence of ad allowing patients to see more advanced diagnostic procedures such as csf or further imaging (o'bryant et al., 2016). obtaining csf for analysis for ad is one of the most invasive methods of diagnosing ad (hansson et al., 2006). getting access to csf involves a complex procedure where many patients may not be willing to undergo as it can cause potential complications and discomfort. a more invasive method would be the use of a brain biopsy to diagnose ad. biopsies come with their own associated risks and can cause unprecedented damage to the brain, resulting in why current methods in diagnosing ad are very limited in accuracy and early stage detections. when evaluating the use of a diagnostic test, several key factors should be considered. factors such as accuracy, cost, and time required to run a test should all be considered. the detail in the amount of information a test can provide should also be evaluated alongside of how invasive the test is. a matrix was constructed using this as a framework, and it is seen below in figure 2. ________________________________________________________________________________ cusj 2019 __________________________________________________________________________________________ 13 figure 2. comparison of six detection methods for late stage ad. the effectiveness of six methods of detection for latestage/end-stage ad were compared relatively using the factors of accuracy, invasiveness, and time required for tests. a is the 3d overview of the graph while b to d shows in each 2d plane the relative performance of each method to one another. the values used for this graph quantitatively are arbitrary but provide a qualitative comparison between available methods. image generated with (geogebra, 2018). v. current and future research the current limitations in the detection of ad stem from the lack of accuracy in less invasive methods coupled with the fact that most of these methods are good at detecting late-stage ad. an ideal test would be able to detect precursors such as amyloid through an obtained sample such as blood. thioflavin t (tht) and congo red can both interact and bind to fibrils which allow for the detection of the accumulation of amyloid. tht has been shown to increase in fluorescence when binding to aggregated amyloid occurs. using properties of diagnostic dyes and their derivatives, new tools and tests can be developed to detect the wide array of differently formed and shaped amyloid plaques. applications of assays using tht fluorescence to detect prion diseases has been shown to work in diseases such as creutzfeldt jakob disease (cjd) (orrù et al., 2017). using a process called real-time quaking-induced conversion (rt-quic) researchers were able to use csf and nasal brushings to achieve a diagnostic sensitivity greater than 90% (orrù et al., 2017). rt-quic was also used to diagnose parkinson's disease (pd) through obtained csf to quantify α-synuclein a major component of lewy bodies in pd (groveman et al., 2018). a schematic of the process used in rt-quic can be seen below in figure 3. this technology has the potential to be applied to the detection of ad. figure 3. a schematic overview of the rt-quic detection method. figure adapted from orrù et al., 2017. using similar tests discussed in the previous literature for the detection of neurodegenerative conditions such as pd and cjd, there is future potential for the detection of plaques using tht (groveman et al., 2018; orrù et al., 2017). a laboratory test could be developed in an attempt a b cusj 2019________________________________________________________________________________ __________________________________________________________________________________________ 14 to improve the detection of conformers of amyloid in obtained samples. this proposed method of detection aims to result in detecting ad during early stages of disease progression by identifying plaques while working towards using a less invasively obtained sample such as blood. initially, an ad diagnostic test similar to that of cjd and pd would require the usage of csf but the goal of this proposed bench test would be to improve the detection of ad in at least one of the domains of accuracy, time, or decrease invasiveness. vi. discussion traditional methods of obtaining samples for amplification involve the use of harsh detergents which may result in loss of potential sample. using seeding, the formation of amyloid fibrils can be accelerated (groveman et al., 2018). rapid techniques for the quantification of primary markers of ad is needed for a more rapid diagnosis. by using the cerebrospinal fluid (csf) in new detection techniques, it is possible to identify and amplify specific biomarkers pertaining to a specific neurodegenerative condition. rt-quic has used αsynuclein (αsyn) seeds to provide a higher diagnostic accuracy when diagnosing pd (groveman et al., 2018). the study used csf from 29 synucleinopathy cases and 31 controls, which included 16 samples from patients with ad (groveman et al., 2018). the diagnostic accuracy of this test was over 80% (groveman et al., 2018). the csf was obtained by carrying out a lumbar puncture, and approximately 20 ml of fluid was used for the rt-quic procedure (groveman et al., 2018). αsyn seeds promoted amyloid fibril formation, which was then seen by the fluorescence of thioflavin t (tht) (groveman et al., 2018). tht is able to bind to amyloids at two sites with a high density (bs1 and bs2) and one at lower density (bs3). these binding sites are lined with hydrophobic and aromatic residues, as seen in figure 4 (reinke and gestwicki, 2011). figure 4. representation of the different binding sites on the amyloid fibrils. figure obtained from reinke and gestwicki, 2011. when tht binds to the β-amyloid fibrils, the fluorescence increases due to the change in carbon-carbon interactions between the aniline and the benzothiazole in tht, as seen in figure 5 (reinke and gestwicki, 2011). figure 5. chemical structures of tht and bta-1 to show the major system of a benzothiazole ring system attached to a functionalized aniline. figure obtained from reinke and gestwicki, 2011. when bound to fibrils, steric hindrance prevents rotation and allows the tht to easily transition into an excited state (reinke and gestwicki, 2011). using an amine functionalized microplate, a phosphate buffered saline (pbs) solution would be added and incubated on the microplate for a period of approximately 45 minutes (interchim.fr, 2018). afterward, the plate would be washed with the coupling buffer two times, and then the plate would be blocked with solution (interchim.fr, 2018). the formation of amide bonds using 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (edc) generates bonds between carboxylic acids and amines (andersen and denmark, 2018). edc allows for the formation of a o-acylurea by activating the carboxylate. using n-hydroxysulfosuccinimide (sulfo-nhs), a hydrolysis resistant ester can be formed as under aqueous conditions the compound can undergo hydrolysis which greatly limits the obtained yield (andersen and denmark, 2018). the activated ester undergoes a reaction with sulfonhs, which allows for the formation of a stable succinimidyl activated ester. this reaction can be seen in figure 6 (andersen and denmark, 2018). figure 6. reaction scheme for the immobilization of dnplabelled tri-peptide. figure obtained from andersen and denmark, 2018. ________________________________________________________________________________ cusj 2019 __________________________________________________________________________________________ 15 using these, tht cross-linked plates would allow for the capture of tau aggregates as well as amyloids and seeds in their native forms without degrading signal strength through the use of harsh detergents. then, using recombinant monomers of amyloidogenic proteins, it would be possible to look for amplification. the amplification of certain amyloid proteins and a lack of others could aid in distinguishing between neurodegenerative diseases, if no cross seeding is present. the study mentioned previously has had similar success when methods were applied to the detection of cjd (groveman et al., 2018). the binding interactions between tht and the tau fibrils also need to be understood to provide a more accurate baseline reading for the extent of aggregation present in the samples. the seeding assays initially will use samples that have come from patients confirmed of having ad and seeing how this method can amplify signals from a known source. as used in the method for both pd and cjd, csf will ideally be used to run the rt-quic procedure. when compared to previous seed amplification methods, the rt-quic procedure allowed for more rapid quantification of αsyn seeds in the csf (orrù et al., 2017). this is all done by maintaining high degrees of sensitivity and specificity in diagnosing pd. the large success of this established protocol gives a promise for similar methods to be used for the detection of primary markers of ad (groveman et al., 2018; orrù et al., 2017). early stage detection of primary markers of alzheimer’s disease could provide immense health benefits for patients. in order for effective treatment of ad, methods such as the rt-quic protocol could be adapted and optimized for the quantification of tau seeds (orrù et al., 2017). the csf used in the rt-quic procedure resulted in a diagnostic accuracy greater than 80%, which is higher than conventional detection methods (groveman et al., 2018). investigating further the formation of fibrils in specific tauopathies would be needed to ensure this method has a greater accuracy. effective methods to detect the formation of primary ad markers would allow for timely treatment to slow the progression of this disease. vii. conclusion the early stage detection of alzheimer’s disease would provide the best health outcome for patients. the large populations in developing nations entail that there will be a rapidly increasing diagnosis of alzheimer’s in the coming years. the best way to decrease the burden on the health care system and allow patients to live normal lives for longer periods of time would be to slow the development of plaques and tangles. however, to provide correct treatment, there need to be effective early-stage diagnostic methods which currently there are a few that are proven to be reliable and non-invasive in nature. an ideal test would detect ad early, be minimally invasive, be accurate, and be able to be conducted rapidly. innovations in any of these areas would greatly increase ad diagnosis and treatment. innovations in the detection of cjd and pd could translate to future methods in detecting ad. potential benchtop tests involving tht and its binding interactions with amyloid could be developed in the near future using detection methods similar to the rt-quic detection method. author information corresponding author gaganpreet jhajj* gjhajj@ualberta.ca abbreviations alzheimer’s disease (ad) thioflavin t (tht) cerebrospinal fluid (csf) creutzfeldt jakob disease (cjd) real-time quaking-induced conversion (rt-quic) vascular dementia (vad) parkinson’s disease dementia (pdd) references alzheimer, a., förstl, h., and levy, r. (1991). on certain peculiar diseases of old age. history of psychiatry 2, 7173. ballatore, c., lee, v., and trojanowski, j. (2007). taumediated neurodegeneration in alzheimer's disease and related disorders. nature reviews neuroscience 8, 663-672. freitas, s., simões, m., alves, l., and santana, i. (2013). montreal cognitive assessment. alzheimer disease & associated disorders 27, 37-43. groveman, b., orrù, c., hughson, a., raymond, l., zanusso, g., ghetti, b., campbell, k., safar, j., galasko, d., and caughey, b. (2018). rapid and ultrasensitive quantitation of disease-associated αsynuclein seeds in brain and cerebrospinal fluid by αsyn rt-quic. acta neuropathologica communications 6. hansson, o., zetterberg, h., buchhave, p., londos, e., blennow, k., and minthon, l. (2006). association between csf biomarkers and incipient alzheimer's disease in patients with mild cognitive impairment: a follow-up study. the lancet neurology 5, 228-234. hardy, j., and higgins, g. (1992). alzheimer's disease: the amyloid cascade hypothesis. science 256, 184-185. hyman, b., van hoesen, g., damasio, a., and barnes, c. (1984). alzheimer's disease: cell-specific pathology isolates the hippocampal formation. science 225, 11681170. johnson, k., fox, n., sperling, r., and klunk, w. (2012). brain imaging in alzheimer disease. cold spring harbor perspectives in medicine 2, a006213-a006213. o'bryant, s., edwards, m., johnson, l., hall, j., villarreal, a., britton, g., quiceno, m., cullum, c., and graffradford, n. (2016). a blood screening test for alzheimer's disease. alzheimer's & dementia: diagnosis, assessment & disease monitoring 3, 83-90. orrù, c., groveman, b., hughson, a., manca, m., raymond, l., raymond, g., campbell, k., anson, k., kraus, a., and caughey, b. (2017). rt-quic assays for prion disease detection and diagnostics. prions 185-203. cusj 2019________________________________________________________________________________ __________________________________________________________________________________________ 16 price, c., cunningham, h., coronado, n., freedland, a., cosentino, s., penney, d., penisi, a., bowers, d., okun, m., and libon, d. (2011). clock drawing in the montreal cognitive assessment: recommendations for dementia assessment. dementia and geriatric cognitive disorders 31, 179-187. reinke, a., and gestwicki, j. (2011). insight into amyloid structure using chemical probes. chemical biology & drug design 77, 399-411. sherrington, r., rogaev, e., liang, y., rogaeva, e., levesque, g., ikeda, m., chi, h., lin, c., li, g., and holman, k. et al. (1995). cloning of a gene bearing missense mutations in early-onset familial alzheimer's disease. nature 375, 754-760. direct imaging of exoplanets with project 1640 briley lewis1 under the guidance of rebecca oppenheimer american museum of natural history central park west at 79th street, new york, ny 10024, usa cusj spring 2017, submitted april 10, 2017 abstract project 1640 is a suite of instrumentation and software focused on high-contrast imaging of exoplanets, probing the parameter space of companion size > 1mj and distance 5-50 au around the host stars. the instrument consists of an apodized lyot coronagraph, with a mach-zender interferometer and an integral field spectrograph, forming data cubes of dimensions right ascension, declination, and wavelength. p1640 is operated at palomar observatory in southern california, in conjunction with their palm-3000 adaptive optics system. data reduction models out remaining speckle noise using principle component analysis and produces a residual cube, which can then be manually inspected for possible companions. for this summer project, data reduction using the karhunen-loéve image projection (klip) algorithm was completed on many of the survey stars and inspected, in an effort to search for more candidates. at this time, two possible candidates have been found, and are awaiting further confirmation. keywords: data analysis, exoplanets, instrumentation, planetary systems, undergraduate 1. introduction in the past two decades, exoplanet science has blossomed as a field, exciting both the scientific community and the general public. the first exoplanet, 51 pegasi b, was discovered in 1995 with the radial velocity method (mayor & queloz 1995), marking the start of a new age of “planet hunting.” kepler was then launched in 2009, an entire nasa mission devoted to discovering planets based on the transit method, and to date is has revealed over 5000 candidate planets and 3000 confirmed planets (e.g. borucki et al. 2010, and numerous others). despite the success of these techniques, it was not until 2008 that we truly saw an exoplanet by collecting light from the object itself (marois et al. 2008; oppenheimer & hinkley 2009; traub & oppenheimer 2011). direct imaging of exoplanets is a feat of engineering and physics; in order to achieve the contrast needed to resolve a planet orbiting a nearby star, the instrumentation must overcome the overwhelming brightness of the star and the noise created by diffraction and scattering, both from within the optical system and from the atmosphere above. despite these challenges, it is possible to image exoplanets, and even to extract their spectra to better understand their composition. using all of these techniques, we now have tools to answer many questions, such as: how are planetary systems formed? what types of planets are there in the universe? how common are these planets? are there any other habitable planets? how are brown dwarfs and planets related? how can we classify sub-stellar objects? project 1640 at the american museum of natural history aims to answer some of these questions. started in 2008, this project is a 99-night survey at the palomar 200-inch hale telescope, seeking to discover and directly image brown dwarfs and large exoplanetary companions (hinkley et al. 2008; beichman et al. 2010; hinkley et al. 2011; oppenheimer et al. 2012). the parameter space being probed by this instrument is generally a distance bll2124@columbia.edu 1 department of astronomy, columbia university, 550 w 120th st, new york, ny 10027, usa from the host star on the order of 5-50 au (where au means “astronomical unit,” the distance between the earth and the sun), and of a companion mass of one jupiter-mass (mj) or greater. over 200 possible target stars were chosen based on the following criteria: nearby, for higher spatial resolution; young, so that planetary companions are still self-luminous from heat of formation; and bright, since the instrumentation operates best with a greater flux of photons from the star (beichman et al. 2010; oppenheimer & hinkley 2009). of these target stars, about 150 have been observed by the team and are being processed to find possible stellar/sub-stellar or planetary companions, and even some previously undiscovered binary systems. after eight years, the survey recently came to a close as of the june 2016 observing run at palomar observatory, and the group is now in the process of sifting through all the data and computing statistics to come to broader conclusions about the zoo of exoplanets in the universe, including ideas about how to classify these objects. 1.1. why is it so difficult to see an exoplanet? the primary difficulty in directly studying exoplanets lies in the fact that planets, even young, self-luminous ones, are very faint compared to the stars they orbit and in extremely close angular proximity. for example, if we were to look at our own solar system from a distance of 10 parsecs, jupiter would be about 10−8 to 10−12 times fainter than the sun (depending on the wavelength of observation) and only 0.5 arcsec away. thus both angular resolution and suppression or control of the light from the star are required to see an exoplanet (oppenheimer & hinkley 2009; traub & oppenheimer 2011). to understand the difficulty of this, imagine first how a telescope functions. without atmospheric distortion or imperfections in optics, a star viewed through a circular aperture, such as that of a telescope, does not appear as a single point. instead, its “point spread function” (psf) follows the airy function. the psf is the angular probability distribution of where a photon will strike the mailto:bll2124@columbia.edu 2 b. lewis image plane of an optical system, essentially describing the image formed. the airy function, in essence, is the fourier transform of the entrance aperture’s illumination pattern (in this case a simple, uniformly lit circle). it takes the shape of concentric “airy rings” around the circular, bright core of the source, whose angular width in radians is roughly λ/d, where λ is the wavelength of light and d is the diameter of the telescope(nyquist 1928). the airy function is simply a result of diffraction of light by the telescope, an insurmountable aspect of physics (at this time). for reference, λ/d is roughly 40 milli-arseconds (mas) for a 5-m telescope observing at λ = 1 µm. in the example of viewing our solar system and trying to detect jupiter from 10 pc away, that angular size, the limit to the resolving power of the telescope, is a mere 12 resolution elements from the sun. the airy rings are at that separation only about 10−4 times the brightness of the sun. thus with simple imaging techniques, jupiter would be completely swamped by the psf of the sun by a factor of about a million. simply having a perfect, reasonably sized telescope with a standard camera behind it does not permit the direct study of exoplanets (oppenheimer & hinkley 2009; traub & oppenheimer 2011). furthermore, most real telescopes also have a secondary mirror suspended above the primary entrance aperture that creates a central hole in the circular aperture. this modifies the airy function in a way that brightens the airy rings. this secondary mirror has to be supported physically, usually with beams of metal that cross the aperture in a symmetric “x” pattern. these are called “spiders” and they also induce additional, extremely bright diffraction in the form of spikes of light emanating from the central core of the image. these diffraction “spikes” have a brightness dependent on the width of the support spiders. diffraction spikes typically have a brightness of 10−3 that of the star, but they are localized in the image only in the directions of the spiders themselves (typically n-s and e-w for an equatorial mount telescope). in addition to diffraction, other real-world problems make imaging exoplanets even harder. these problems can be binned into one piece of jargon: “speckles.” speckles come from (1) imperfections in the wave front of light entering the telescope and (2) imperfections in the optics of the telescope and whatever instrumentation is placed behind the telescope. speckles are both dynamic and quasi-static. the atmosphere causes speckles that are stochastic and highly variable in time; while the speckles from imperfect optics are quasi-static, only affected by changes in the optical surfaces themselves, such as sagging when the telescope is rotated and the gravity vector changes, or expansion due to a change in temperature. this speckle noise currently poses one of the most significant challenges to high-contrast observations (oppenheimer & hinkley 2009). speckles appear in an image as bright dots of light, as their name suggests, roughly of the same size as λ/d scattered about the image in a seemingly random pattern (fig. 1). unfortunately, speckles do not behave like a typical gaussian noise source that can be dealt with through simply collecting more data and averaging it together (as one does with long-exposure faintsource detection in fields such as extragalactic astronomy). rather, they exhibit statistical behavior that requires that they be controlled or eliminated in order to overcome their contribution to the background of light that a real object is being observed against (hinkley et al. 2007; crepp et al. 2011; fergus et al. 2014). fortunately, the speckles have two properties that allow for control and elimination. first, because speckles are an optical effect due to an imperfect wave front of light used to form the final image (c.f. §2.3), one can attempt to detect and correct these imperfections optically. second, also because they are an optical effect, their size and location in the image follows the λ/d relation mentioned above. at longer wavelengths they are larger and farther away from the star in the image, and at shorter wavelengths they are closer and smaller. fig. 1, demonstrates this effect. a real object on the sky does not change its position simply because of the wavelength of light it is being observed with. thus, one can discern a real object from speckles simply because it does not move when viewed in images taken at different wavelengths (fig. 1). figure 1. four example coronagraphic images at different wavelengths from 1.0 to 1.8 µm. speckles on the red end are further away from the star. on the blue end the same speckles are closer. the real object at top right does not move relative to the star. thus real objects can be discerned from speckles. 2. project 1640 instrumentation in order to image exoplanets, project 1640 employs several techniques to overcome the difficulties described in §1.1. for the purposes of a quick introduction, these techniques include (1) apodization and coronagraphy to control diffraction, (2) very high-order adaptive optics and precision wave front sensing to control both atmospheric and quasi-static speckles and (3) a specialized imaging device that takes images at 32 different wavelengths at once to permit discerning real objects from speckles due to the motion of speckles as a function of wavelength. a schematic of this complex suite of instrumentation is shown in fig. 2. first, we can significantly reduce transmission of the host star’s light by adding an apodized pupil and a lyot coronagraph, blocking the core of the star. wave front correction is then dealt with using adaptive optics and and interferometer, which will be further discussed in §2.3. an integral field spectrograph (ifs) is added before the detector, which aids in post-processing removal of remaining speckle noise, and also allows for low-resolution spectra to be taken from the companion objects, providing information about their compositions and more direct imaging of exoplanets 3 (hinkley et al. 2011; oppenheimer et al. 2012). figure 2. summary of optical system of gemini planet imager, the southern counterpart to project 1640, which has the same components. adapted from macintosh et al. (2008). section numbers here refer to documentation in macintosh et al. (2008), not in this paper. we describe each subsystem below by first describing control of diffraction, then control and elimination of speckles. in discussing diffraction, we will ignore speckles and assume a diffraction-limited psf in order to describe how to suppress diffracted light in an ideal optical path. it is important to note that there are a great number of other techniques for high-contrast imaging, including speckle nulling, interferometry, and other forms of coronagraphy; the instrumentation described here is specific to project 1640, and is closely related to that of the gemini planet imager (macintosh et al. 2014) and the eso’s sphere project at the very large telescope (beuzit et al. 2006). 2.1. apodized pupil to improve upon this system, a graded transmission function can be added to the telescope entrance pupil (the leftmost opening in fig. 3) to “soften” the effects of diffraction. this “apodization” reduces the brightness of both the airy rings and speckles, allowing the instrument to achieve higher contrasts overall (soummer 2005; soummer & aime 2004; soummer et al. 2003b,a; oppenheimer & hinkley 2009). the apodization optic used in project 1640 is shown at bottom-left of the top panel of fig. 3. 2.2. lyot coronagraph further control of diffracted light is achieved with coronagraphy. two masks are used to obscure starlight in the lyot coronagraph, eponymous with its inventor (lyot 1939). the first mask is placed in the focal plane, known as the “occulting spot” or “focal plane mask.” this mask blocks out the light from the core of the star, similar to holding your thumb over a bright light to block it from your vision. the second mask is placed at the pupil plane, where it blocks some of the bright, diffraction of the focal plane mask, as well as diffraction from the secondary mirror and the spider support structure; this is the “lyot stop” or “pupil plane mask.” the optical schematics of coronagraphy are shown in fig. 3. light from a planet in the field of view is not blocked, even with the presence of the occulting spot, due to the fact that light from a planet would be coming from off the central axis. that is, it is coming in from a different spatial location from the star, and thus takes a slightly different path than the on-axis star, avoiding the occulting spot as shown in fig. 3 (bottom panel). figure 3. top: a diagram of a lyot coronagraph. airy rings are visible in the “modified pupil” image. adapted from oppenheimer & hinkley (2009). bottom: path of both “on-axis” and “off-axis” light through the coronagraph, from kenworthy (2016). 2.3. palm-3000 adaptive optics as mentioned earlier, the wave front of a star’s light is not an ideal simple flat plane due to atmospheric turbulence, temperature differences near the telescope, and aberrations in the optics. there are imperfections, referred to as “wave front deformations,” which can significantly lessen resolution and cause speckles, reducing the contrast we can achieve with our instrument. though the data quality is inherently somewhat dependent on the natural seeing conditions, adaptive optics systems allow for some correction of this wave front, and can bring the telescope and instrument significantly closer to diffraction-limited seeing. adaptive optics, using a natural guide star, works by sensing the input wave front from the target star and quickly changing the shape of a deformable mirror to counteract the deformations, ideally producing a flat wave front as shown in fig. 4 (oppenheimer & hinkley 2009). the hale telescope’s new system, palm-3000 (dekany et al. 2013), has 3388 actuators controlling the two deformable mirrors, allowing them to be adjusted 2000 times per second to quickly process the incoming wave front and correct it. one of the mirrors, with around 200 actuators, deals with low-order corrections, the larger deformations in the wave front, whereas the other mirror has a much higher density of actuators for high-order corrections, the much finer details of the wave 4 b. lewis front (dekany et al. 2013). additionally, there is a small tip/tilt mirror, dealing with the lowest-order corrections; it adjusts using a proportional-integral-derivative loop controller to simply keep the star at the center of the image, allowing the ao system to focus on more detailed corrections. figure 4. top: diagram of the basic concept of adaptive optics. adapted from veicht (2016). bottom-left: simulation of input wavefront corrupted by the atmosphere, the image that results in below. bottom-mddle: shape imposed on deformable mirror. bottom-right: corrected wave front and below that the corrected image (courtesy j. lloyd). to improve upon palm-3000, project 1640 additionally uses a mach-zender interferometer, referred to as the cal (calibration) interferometer (zhai et al. 2012; vasisht et al. 2012). the light rejected by the coronagraph, as well as part of the beam that will eventually travel to the ifs detector, are both processed by this interferometer, sensing any remaining deformations in the wave front down to 1 nanometer. cal sends this information back to the palm-3000 adaptive optics system, which then corrects for this deformation. since cal is placed after the coronagraph in the optics, it allows for corrections within the system, greatly improving the image and thus the achievable contrast (oppenheimer et al. 2012; zhai et al. 2012; vasisht et al. 2012). since we have control over the deformable mirror, it is also possible to purposefully impose deformations onto the wave front, creating an image in the final science exposure. this is how “grid spots” on the data are created; these grid spots are a sinusoidal perturbation inserted into the wave front, which creates four identical images of the star outside the focal plane mask with known locations and brightness (see fig. 5, left). the intersection of these four spots details the location of the star, which would otherwise be unknown when the star is occulted, allowing us to perform astrometry calculations and some photometry on the data if a companion is found (sivaramakrishnan & oppenheimer 2006; marois et al. 2006; oppenheimer & hinkley 2009). in fact, the project demonstrated relative astrometry between alcor and its newly discovered companion at a level of a few mas, allowing for parallactic motion to be used to show, within months, that the companion is indeed orbiting alcor (zimmerman et al. 2010). previously companion confirmation relied upon showing that the two objects have common proper motion across the sky over years or more. 2.4. integral field spectrograph the last physical part of the project 1640 instrument (aside from computers, electronics and software) collects the data. project 1640 has a detector with a 3.8 x 3.8 arcsecond field of view, placed after an integral field spectrograph (ifs). the ifs contains an array of lenslets, which break up the incoming light into spectra, allowing us to take 32 simultaneous images in different wavelengths. within this 2040 x 2040 pixel space on the detector, there are approximately 37,000 spectra, spanning the wavelength range of 995-1798 nanometers. these spectra are extracted from the raw data to form data cubes, with dimensions of right ascension, declination, and wavelength (zimmerman et al. 2011; oppenheimer et al. 2012). the range of the detector falls in the near infrared, including the y, j, and h bands. project 1640 is aptly named, as 1640 nanometers falls within the detection range of our instrument, and in fact is the wavelength at which the system optimally suppresses a star’s light. surveys for direct imaging of exoplanets, such as this, are conducted in the infrared because the planets we are able to image would still be glowing in the infrared from the heat of their formation. this is why our survey stars chosen are young (beichman et al. 2010; oppenheimer et al. 2012; david & hillenbrand 2015). 3. data collection at palomar observatory project 1640 has completed its 99 nights of observations over a span of 3 years at palomar observatory’s 200-inch (5.1-m) hale telescope in southern california. this reflecting telescope, constructed in the 1940s, has been a workhorse of modern astronomy since, with notable achievements such as the discovery of the first quasar, 3c 273 (schmidt 1963). palomar observatory is also where project 1640’s own r. oppenheimer discovered the first methane brown dwarf, gliese 229b, in 1995, using a coronagraph on the 60-inch telescope and a spectrograph on the 200-inch telescope (nakajima et al. 1995; oppenheimer et al. 1995, 1998; oppenheimer 2014). as a part of this summer project, i had the opportunity to participate in the june 2016 observing run with the project 1640 team at palomar, which was incidentally the last official run of the survey. multiple collaborators, from johns hopkins, caltech, nasa jpl, columbia university, palomar observatory, and the american museum of natural history, were also present, allowing me direct imaging of exoplanets 5 figure 5. examples of project 1640 data. left: one λ slice of a raw data cube with visible gridspots. right: one λ slice of a residual data cube after processing using klip. the opportunity to learn from each about a different aspect of this project. this experience also allowed me to participate in and directly learn about modern observational astronomy, from set-up of the instrument to actual data collection and observation logs. 4. data processing even with the complicated optical instrument described above, that is still not enough to achieve the contrasts necessary to image an exoplanet or sub-stellar companion since speckles remain in the image. in many normal astronomical contexts, this extra light would be negligible; yet, when most of the starlight has been removed, speckles become prominent features in the image, as seen in the raw data in fig. 5, left. this leads to the question how can we possibly remove speckle noise? it is mostly random, unable to be accurately modeled by a physical process. instead, we turn to the speckles themselves, using their properties and presence in one image of a data set to model them out of the other images in the data set. one important property to note is their radial movement; as speckles are mostly a result of diffraction of wave front distortions through the optics, they appear to move radially outwards with wavelength as described in §1.1. this feature is what allows spectral differential imaging (sdi), used in project 1640 to aid in modeling out speckles, and also makes it possible to discern planets within the residual noise (oppenheimer & hinkley 2009). spectral differential imaging exploits the radial movement of speckles with respect to wavelength (see fig. 1). this is why project 1640 uses the ifs to obtain simultaneous images at multiple wavelengths, so we can discern the speckles from a real object. if we scale and align all the slices of a single data cube, so that the grid spots coincide, the speckles will also be in similar positions. a planet, on the other hand, will be in different positions in each cube, since it is stationary in the original images. an algorithm can then model out the speckles, which are present in the same place in each scaled image, without modeling out the planet. 4.1. project 1640 data pipeline the method used to model speckles is referred to as psf subtraction, since the goal is to reconstruct the point spread function of the star, including the speckles, and then subtract it from the science image, thus revealing the psf of a fainter companion object. the klip (karhunen-loéve image projection) algorithm is an implementation of principle component analysis, similar to the s4 and loci (locally optimized combination of objects) algorithms that also reconstruct psfs for subtraction (soummer et al. 2012; fergus et al. 2014). post-processing of project 1640 data is mainly handled by both s4 and klip; however, for this project, i have been reducing data with klip. though it may seem redundant to reduce data twice, it is actually a powerful tool for confirming the validity of a candidate object. since each algorithm has different strengths and shortcomings, if a candidate is visible in both outputs, that is a strong indication that it is not simply a product or error of the modeling. before processing data with klip, it is necessary to visually inspect each cube to ensure that it is of the quality needed for this survey. since palomar observatory schedules specific nights for observation, instead of queue observing, some data was taken in worse seeing conditions than desired. less than 1.4 arcsecond seeing is optimal; however, data with higher seeing values is not entirely excluded if the entire data set is under those conditions. science exposures for this project must be fully occulted by the mask, greater than 180 seconds exposure time, with clear grid spots, and in acceptable seeing conditions. such stringent requirements are necessary since any of these factors can prevent us from seeing a faint companion. since klip also requires precise grid spots for scaling and aligning of cubes prior to running the speckle-modeling algorithm, it is necessary to fit spots to the image with an algorithm, and then visually inspect those spot fits to ensure they are correct. with spot fitting completed, the klip algorithm is able to scale and align the frames of the data cubes so that the grid spots coincide, then proceed with the principle component analysis speckle modeling. 4.2. karhunen-loéve image projection (klip) with an input of scaled and aligned data cubes, klip performs psf subtraction on each cube individually, using one frame of the cube as the target image and the rest of the frames as reference images. in order to re-create the speckles of the psf to subtract them accurately, klip computes principle component analysis, including a karhunen-loéve transform, for which it is named. in essence, principle component analysis is the process of creating an orthonormal basis that can represent the speckle noise in the data, in order to rebuild an image of that noise that is as similar to the original as possible (e.g. minimizing the difference between the image and the model). to start, the algorithm separates the wavelength slices of the input cubes into a target image t (n), and a set of reference images rk(n). a whole data set (multiple cubes, generally around 10) is loaded for klip, so all the slices at one particular wavelength become the “target” images and the rest of the wavelengths are references. for example, an input cube has dimensions (32, 250, 250). if we input two cubes into klip, these dimensions become (64, 250, 250). one is chosen as the target image, so then we have one set of data with dimensions (250, 250) and another, the reference images, with dimensions (63, 250, 250). all images of the same wavelength as the target are taken out of the reference set, leaving the reference set in this case as (62, 250, 250). some wavelengths are intentionally left out of 6 b. lewis the reference image set to reduce contamination, so then that leaves a set of around (54, 250, 250) (j. aguilar, personal communication, july 28, 2016). keep in mind that the target image can contain an astronomical signal (e.g. the point spread function of a companion) in addition to the noise, so the target image can be represented as the linear addition of the noise and an astronomical signal; our reference images are assumed not to contain the astronomical signal, implying they are images of only the noise that is to be modeled out. in order to model more accurately, klip segments these images into search areas, slicing the image into a specified number of annuli and sections of angle φ (soummer et al. 2012). for these search areas, the karhunen-loéve transform is computed in order to create an orthonormal basis of eigenimages, which are basically the reference images rp weighted by the eigenvectors of the basis ck and normalized by the eigenvalues λk. the index p tracks the reference image number, n is the pixel index within the image, and k is the index of the eigenvector. a number up to k eigenvalues and eigenvectors are included in the computation. essentially, we take a certain cutoff number of eigenimages, which are the “principle components” that make up the most of the reference image; these eigenimages are weighted by eigenvalue as to how important they are in the reference image in question. zk then is the weighted sum of reference images for a given eigenvector of the basis k. smaller features are not modeled, since they are less important, e.g. less weighted by eigenvalue. this is part of why this process isn’t entirely exact, and cant get rid of all speckles in a given image. (zk)kl = 1√ λk k∑ p=1 ck(ψp)rp(n) additionally, given an infinite number of references, it would be possible to find a basis that exactly represents the input data; however, we are limited to at most the p = 31 slices of each cube, not including the frame selected as a target image, so our basis is only an approximation. this process can be visualized for the first four eigenimages of an arbitrary reference image in figure 6. figure 6. reconstructed image with its eigenimages, weighted by eigenvalue. (image from a talk by václav hlavaĉ) however, not all eigenimages should necessarily be included fewer modes (e.g. number of eigenimages included) can lead to under-modeling, not subtracting enough of the noise to see a faint object, but too many kl modes can lead to over-modeling, subtracting out any signal from the objects we are searching for (soummer et al. 2012). to compensate for this uncertainty, klip is run with a variety of parameters (annuli, angular sections, kl modes) over each image. this produces multiple outputs, including a “median-combined” image, which averages the residual outputs of all kl-mode options for a given set of annuli and angular sections. following computation of zk for each eigenvector, t is projected onto zk for all pixels n in the image, giving a measure of the similarity between the two images. the weighted sum of reference images itself is multiplied by this product, weighting it by importance in the final sum for the reconstructed psf i(ψ0). this is repeated for each eigenvector k in the basis. i(ψ0) = k∑ k=1 < t, (zk)kl > (zk)kl(n) given this model, psf subtraction can finally be done, creating a final residual image f (n) (soummer et al. 2012). f (n) = t (n)− iψ0(n) 4.3. planetary detection in order to detect planets and other companion objects in the data, it is necessary to again exploit the radial motion of the speckle noise. when viewing the residual data cubes after using klip, or other psf subtraction algorithms, the remaining speckles that have not been modeled out generally still appear to move. an exoplanet, on the other hand, would not share that property of radial motion, and can be seen as the one fixed object in the image. attempts have been made at automating this data searching with an algorithm, but at this time the human eye is still superior. 4.4. signal-to-noise ratio maps in addition to visual inspection of raw data cubes, signal-to-noise ratio (snr) maps can be useful for more easily distinguishing a candidate object from the background. in our residual data cubes, the pixel values should approximately follow a gaussian, with the distribution centered at zero, due to details of the klip algorithm that produces the processed data cubes. additionally, the pixel values tend to fall off radially in the residual cubes, due to the fact that the majority of noise from the host star is concentrated at the center, where the host star is physically located in the image. these properties can be harnessed to judge the significance of a detection; essentially, we are trying to see where the planetary signal falls on this gaussian of pixel values. if the pixels fall within one standard deviation, they are likely from this same distribution of the noise. for example, if they are three standard deviations away (a “3 sigma detection”), then we can judge that they may arise from a different distribution altogether and are therefore significant. by looking at standard deviations, we can judge the confidence with which we can claim a detection; a one-sigma detection implies that there is an approximately 68% chance the object is not from the noise. three-sigma detections are more accepted as valid, however, as they imply a 99.7% chance the object is not a part of the noise. signal-to-noise ratio maps are one way of judging the significance of a detection based on the standard deviation of the pixel values in the image. we assume that direct imaging of exoplanets 7 0 20 40 60 80 100 120 140 0 20 40 60 80 100 120 140 h band snr map 0 20 40 60 80 100 120 140 0 20 40 60 80 100 120 140 j band snr map 0 20 40 60 80 100 120 140 0 20 40 60 80 100 120 140 h and j bands snr map 0 20 40 60 80 100 120 140 0 20 40 60 80 100 120 140 all bands snr map 0.0 1.5 3.0 4.5 6.0 7.5 9.0 10.5 0.0 0.8 1.6 2.4 3.2 4.0 4.8 5.6 6.4 7.2 0.0 1.5 3.0 4.5 6.0 7.5 9.0 10.5 0.0 1.5 3.0 4.5 6.0 7.5 9.0 10.5 test figure 7. example snr map, for a ”test” object. companion is clearly visible near upper right limb of the star. the noise is the one-sigma value of the pixel value distribution, and then divide all pixels in the image by this value to calculate the “signal-to-noise ratio”. in theory, a significant detection should stand out as a bright spot in this sort of map, where the signal is significantly higher than the noise. with a value seen over this threshold, we could claim a one-sigma detection. one sigma is not a particularly high confidence level, but is good enough for a preliminary look at the data, which is all this technique aims to accomplish. to claim a more conclusive detection, we would need to reach towards the three to five sigma level and use more complex methods. 5. results approximately 40 out of the 150 survey stars have been processed with klip as a part of this summer project. much of the data processed is marginal, in seeing or weather conditions that are not ideal, or with less than a full set of data (considered to be 60 minutes or more of observation time). one prospective candidate object has been discovered, very close to our detection limits2. it only appears in four pixels in the residual data cubes processed with klip, from an observing run in july 2015, and is visible in multiple wavelength bands. there is currently only one epoch of project 1640 data on this star, and a second epoch of observation was completed in august at the w.m. keck observatory in hawaii with some of our collaborators; however, this data was of marginal quality, and was not able to give us additional information. this star is on the list for re-observation in p1640’s final follow-up run at palomar observatory in april 2017. additionally, a second possible companion has been discovered around a different star. though originally noticed in a reduction processed with the s4d algorithm, it is also now seen in a reduction done with klip as a part of this summer project. it is visible in multiple wavelength bands throughout the data cube, and appears to be fairly large, especially in comparison with the other possible companion mentioned above. there is only one 2 names of these objects are omitted from this paper, as more work is needed before we announce any possible discovery. papers in preparation are not noted since we are still in the early stages, but are in progress for some candidates. figure 8. first possible companion (circled) around an occulted star in a klip-processed data cube. figure 9. second candidate companion (circled). left: slice of the residual data cube, processed with s4. right: slice of the residual data cube, processed with klip. epoch of p1640 data on this star as well, from april 2015. to follow up on this object, other members of the team are extracting a spectrum from this object using the s4s algorithm. if the spectrum indicates it is a viable candidate, we plan to re-observe when it is visible again in april 2017. lastly, a third possible companion has been discovered around yet another star. first observed in july 2015, this object was observed again in october 2016. both epochs of data, as well as the signal-to-noise ratio maps for each, clearly show the presence of a large bright spot at least 1σ above the noise. based on preliminary astrometric estimates, this object is moving opposite the direction of expected motion of a background star, indicating it may be gravitationally bound to its host star. with more detailed calculations, we hope to further constrain some of its orbital parameters to ensure that it is a physically plausible system. a spectrum is in preparation and should provide more detail on this object as well. 6. discussion based on reductions of other data, there is a possibility the first candidate is an artifact of either the detector itself or the klip reduction process. similar bright pixels have appeared in multiple data sets at nearly the same coordinates and in similar slices of the cube, but not exactly identical. additionally, we have not yet been able to see this same possible companion in reductions with s4 (example shown in fig. 11), so a spectrum will have 8 b. lewis figure 10. third candidate companion (circled). left: slice of the residual data cube, processed with s4. right: slice of the residual data cube, processed with klip. to be obtained through other methods, such as aperture photometry, to investigate this issue further. the lack of detection in s4 does raise doubt about the credibility of this object, but that is not a strong enough indication to completely disregard this object as a false positive. the second and third candidates, however, seem greatly promising, given their clear visibility in both data reductions and in the signal-to-noise ratio maps. spectra are required for further work on these objects, however, and a second epoch of data is needed for candidate two and will hopefully be obtained at the april 2017 observation run at palomar. figure 11. example of four spectra extracted using the usual s4s methods of project 1640. (oppenheimer et al. 2013) 7. conclusion though there are many possible pitfalls in the complicated, new world of direct imaging, the possibility of three new directly imaged planets is extremely exciting. few planets have yet to be discovered by direct imaging, though the technique has been proven to work by imaging known systems, such as that of hr8799. as the technology improves, this technique seems promising for unveiling important and interesting information about extrasolar systems. these new direct imaging surveys such as project 1640, gemini planet imager, sphere, and others mark a new age of astronomy and high-contrast imaging, which is the result of years of innovation in instrumentation, mechanical engineering, optical systems, data processing algorithms, and more. though so far we have not discovered as many planets with direct imaging as originally expected, those that have been revealed through all methods of detection topple the preconceptions of what a planetary system is expected to look like. hot jupiters, circumbinary planets, and many other objects already discovered represent the great diversity of planets, and are simply the tip of the iceberg for what exists for scientists to discover. the next few decades should be a time of great excitement and revolutionary thought, answering questions about how solar systems form and planets migrate, how to define the wide range of sub-stellar objects in existence, and even how to find prospective habitable planets and life on other worlds. this paper is based on observations obtained at the hale telescope, palomar observatory. this work was funded through the national science foundation research experiences for undergraduates program. the author would also like to thank the rest of the project 1640 team for their support and guidance over the course of this internship, especially rebecca, jonathan, emily, ricky, aaron, and statia, as well as all those who made the 2016 amnh reu possible. facilities: hale (project 1640) references beichman, c. a., krist, j., trauger, j. t., et al. 2010, pasp, 122, 162 beuzit, j.-l., feldt, m., dohlen, k., et al. 2006, the messenger, 125, 29 borucki, w. j., koch, d., basri, g., et al. 2010, science, 327, 977 crepp, j. r., pueyo, l., brenner, d., et al. 2011, apj, 729, 132 david, t. j., & hillenbrand, l. a. 2015, apj, 804, 146 dekany, r., roberts, j., burruss, r., et al. 2013, apj, 776, 130 fergus, r., hogg, d. w., oppenheimer, r., brenner, d., & pueyo, l. 2014, apj, 794, 161 hinkley, s., oppenheimer, b. r., brenner, d., et al. 2008, in society of photo-optical instrumentation engineers (spie) conference series, vol. 7015, society of photo-optical instrumentation engineers (spie) conference series, pp. 701519–701519–10 hinkley, s., oppenheimer, b. r., soummer, r., et al. 2007, apj, 654, 633 hinkley, s., oppenheimer, b. r., zimmerman, n., et al. 2011, pasp, 123, 74 kenworthy, m. 2016, phd thesis lyot, b. 1939, mnras, 99, 580 macintosh, b., graham, j. r., ingraham, p., et al. 2014, proceedings of the national academy of science, 111, 12661 macintosh, b. a., graham, j. r., palmer, d. w., et al. 2008, in presented at the society of photo-optical instrumentation engineers (spie) conference, vol. 7015, society of photo-optical instrumentation engineers (spie) conference series, 701518–701518–13 marois, c., lafrenière, d., macintosh, b., & doyon, r. 2006, apj, 647, 612 marois, c., macintosh, b., barman, t., et al. 2008, science, 322, 1348 mayor, m., & queloz, d. 1995, nature, 378, 355 nakajima, t., oppenheimer, b. r., kulkarni, s. r., et al. 1995, nature, 378, 463 nyquist, h. 1928, transactions of the american institute of electrical engineers, volume 47, issue 2, pp. 617-624, 47, 617 oppenheimer, b. r. 2014, in astrophysics and space science library, vol. 401, 50 years of brown dwarfs, ed. v. joergens, 81 oppenheimer, b. r., & hinkley, s. 2009, ara&a, 47, 253 oppenheimer, b. r., kulkarni, s. r., matthews, k., & nakajima, t. 1995, science, 270, 1478 oppenheimer, b. r., kulkarni, s. r., matthews, k., & van kerkwijk, m. h. 1998, apj, 502, 932 direct imaging of exoplanets 9 oppenheimer, b. r., beichman, c., brenner, d., et al. 2012, society of photo-optical instrumentation engineers (spie) conference series, 8447 oppenheimer, b. r., baranec, c., beichman, c., et al. 2013, apj, 768, 24 schmidt, m. 1963, nature, 197, 1040 sivaramakrishnan, a., & oppenheimer, b. r. 2006, apj, 647, 620 soummer, r. 2005, apj, 618, l161 soummer, r., & aime, c. 2004, in advancements in adaptive optics. edited by domenico b. calia, brent l. ellerbroek, and roberto ragazzoni. proceedings of the spie, volume 5490, pp. 495-503 (2004)., ed. d. bonaccini calia, b. l. ellerbroek, & r. ragazzoni, 495–503 soummer, r., aime, c., & falloon, p. e. 2003a, in eas publications series, ed. c. aime & r. soummer, 93–105 soummer, r., aime, c., & falloon, p. e. 2003b, a&a, 397, 1161 soummer, r., pueyo, l., & larkin, j. 2012, apj, 755, l28 traub, w. a., & oppenheimer, b. r. 2011, direct imaging of exoplanets (university of arizona press), 111–156 vasisht, g., cady, e., ligon, e. r., et al. 2012, pasp, submitted veicht, a. 2016, phd thesis, columbia university zhai, c., vasisht, g., shao, m., et al. 2012, in society of photo-optical instrumentation engineers (spie) conference series, vol. 8447, 259 zimmerman, n., brenner, d., oppenheimer, b. r., et al. 2011, pasp, 123, 746 zimmerman, n., oppenheimer, b. r., hinkley, s., et al. 2010, apj, 709, 733 abstract introduction why is it so difficult to see an exoplanet? project 1640 instrumentation apodized pupil lyot coronagraph palm-3000 adaptive optics integral field spectrograph data collection at palomar observatory data processing project 1640 data pipeline karhunen-loéve image projection (klip) planetary detection signal-to-noise ratio maps results discussion conclusion 2017-letter-from-the-editors dear readers, after a year of developments, we are proud to present the spring 2017 columbia undergraduate science journal. the current edition of the journal features topics from all over the natural sciences. this year, our main focus regarding the journal was to extend the reach of the journal to universities all over the world. as the cusj’s mission is to encourage students to involve themselves in scientific discovery, we saw no need to restrict our scope to only columbia university. we received submissions from diverse areas such as biochemistry, astrophysics, and electrical engineering. our team also published the second issue of the columbia junior science journal, a journal meant to introduce high school researchers to the world of research publication. after making changes to the previous columbia research scholars journal to better reflect the mission of the journal, we were proud to publish many of the incredible submissions sent in from high schoolers from around the united states, and invite several of them to our annual spring symposium. in addition to producing the two research journals for high school and college students, the cusj also seeks to foster the undergraduate research community at columbia university. this year, we worked hard to further this goal by hosting several events on campus. in the fall, the cusj hosted an event to assist undergraduate students in the process of soliciting a research position. in the spring, the cusj hosted dr. george yancopoulos, founder of regeneron and chief scientific officer, and held the annual symposium. this year’s the cusj spring research symposium involved a welcome address from dr. gasperov, the undergraduate science research adviser at columbia university, as well as a poster session where students presented on their research. the awards of excellence this year went to nicholas page from rutgers university and sarah lundell from fordham university. this spring, the cusj also established the cusj colloquium, a bi-weekly undergraduate speaker series where students from an array of scientific disciplines presented their research in a comfortable peer setting. students discussed topics in research subjects including two-dimensional materials, exoplanets, biofilms, viral binding, ophthalmology, and photosynthetic origins of life. our publications, the cusj and cjsj, and the events we hosted this year could not have happened without the hard work of our editorial team and the assistance of our faculty board and advisors. natalie kolba, editor-in-chief osman moneer, president microsoft word cusj_spring_2015_journal.docx !cusj%|%volume%9%|%spring%2015% 7% the genetics of human behavior katherine kim1 & david streid2 1department of biochemistry, columbia university, new york, ny, usa 2department of biology, columbia university, new york, ny, usa behavioral disorders arise from environmental, lifestyle, and genetic factors. past studies have shown evidence for the hereditability of several major behavioral neuropsychiatric disorders, such as schizophrenia, depression, and bipolar disorder. in these cases, certain genetic defects are passed down from parental generations and increase an offspring’s risk of inheriting a specific disorder. while neuropsychiatric disease susceptibility cannot be attributed solely to genetics, it is important to study how one’s genetic makeup can affect various facets of human behavior. uncovering this link between genes and behavior could lead to the discovery of new biological factors involved in the development of highly prevalent neurological responses and disorders. a recent study in behavioral genetics has shown that there may be a genetic basis for irrational phobias. it highlights the possibility that phobias are a form of inherited defense mechanism passed down through familial genes. in this study, researchers dias and ressler from the emory school of medicine subjected mice to fear conditioning by exposing them to the scent of chemical acetophenone, which smells like cherry blossoms, before administering electric shocks to the mice. offspring of these mice (which were not exposed to the same conditioning as their parents) showed fearful responses to the odor of acetophenone, even when smelling it for the first time. this demonstrated that they had acquired a phobia of the chemical odor. structural abnormalities were also discovered in the olfactory bulbs of the offspring mice. upon sequencing the mice’s sperm dna, dias found that the gene encoding m71, an odo receptor activated by acetophenone, was methylated in the conditioned parental and direct offspring generations. however, it is unknown whether this epigenetic alteration in sperm dna was responsible for the offspring’s heightened odor sensitivity. it is possible that different biological mechanisms worked in conjunction to translate the inherited ancestral experiences to irrational phobias in the offspring. other studies in behavioral genetics have shown that some neuropsychiatric disorders are less heritable—or have a weaker genetic component—than others. for example, while genes may account for more than half of the risk for certain neuropsychiatric disorders, such as schizophrenia or bipolar disorder, the hereditability of anxiety and depression appear to be lower. according to dr. pine at the cold spring harbor laboratory, approximately 30-50% of the risk for anxiety and depression is genetic, while the other 50% to 70% of the risk may be attributed to environmental factors, such as substance use, stress, diet, and childhood experiences. anxiety disorders are the most common form of mental illness in the u.s., affecting 18% of the total population. depression is also common, with around 10% of americans experiencing a major depressive disorder at some point in their lives. despite the high prevalence, genetic disposition for anxiety and depression is weak when compared to other neuropsychiatric disorders. as scientists, we must determine why this is the case. is it due to a difference in the number of gene defects? for example, are there less genetic variations linked to anxiety and depression than to other more heritable diseases? or are depression/anxiety genes less evolutionally conserved? only by answering these questions can we get a firm understanding of the genetic root of these conditions and develop ways to prevent or fight the disorders. we must examine the gene defects themselves. perhaps, in behavioral disorders with relatively low heritability, the gene variations only minimally disrupt the major pathways of the brain. in such cases, it would be wise to study non-genetic factors that trigger the behavioral response. in addition, a psychodynamic treatment approach – alleviating a patient’s mental tension with the help of a psychiatrist—may be more helpful than invasive medical procedures. on the other hand, personalized medicine, such as gene therapy, may be the best option for treating significantly inheritable disorders, like schizophrenia. through advancements in gene testing, doctors are able to conduct pre-symptomatic diagnostic tests to see the risk for patients with a family history of inherited neurological disorders. tests can detect abnormalities, which may include missing or heavily altered sections of a gene, or genes that are inactive or lost, in dna or rna samples of patients. in other cases, a test may detect excessive rna from a single gene, indicating that it is overexpressed in the body. identifying and fixing these problematic sequences in the genetic code requires extensive knowledge of the human genome. physicians providing personalized medicine must take into account a patient’s genetic makeup to determine the best form of targeted treatment for an illness. through genetic research, we are slowly beginning to unravel the biological basis for many neuropsychiatric disorders. understanding the role of genes in highly prevalent neurological responses, like anxiety and phobias, is crucial for designing effective treatments tailored to patients who are suffering these conditions. specifically, by identifying the genetic markers associated with inheritable neuropsychiatric diseases, we can analyze a patient’s risk of disease inheritance and responsiveness to existing medical treatment. this knowledge will make a powerful impact on the medical community and the future of medicine. columbia undergraduate science journal open-access publication | http://cusj.columbia.com 20 cusjvolume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu n eu ro sc ie n c e cusjcolumbia undergraduate science journalresearch articles 21cusj volume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu m o lec u la r bio lo g y identifying serum-inducible sequence elements on the mouse zfp36 promoter max shutran1,*, ron prywes 2,* 1tufts university school of medicine, boston, ma 02110, 2department of biological sciences, columbia university, new york, ny 10027 apuj�äunly�wyv[lpu�����a-7����pz�hu�95(�ipukpun�wyv[lpu�[oh[�kv^uyln\sh[lz�[ol�[yhuzsh[pvu�vm�jly[hpu�nlul�wyvk\j[z�i`� inducing the degradation of mrna. zfp36 is an immediate-early gene: its transcription is activated directly following appropriate cellular stimuli, without any need for intervening protein synthesis. prominent mrna targets of zfp36 are tumor uljyvzpz�mhj[vy����;5-�����lwpklyths�nyv^[o�mhj[vy��,.-��huk�j�mvz��(z�klmlj[z�pu�[ol�j�mvz�huk�,.-�wh[o^h`z�oh]l�illu� consistently found in cancers, zfp36 is thought to be a tumor-suppressor. zfp36 also is responsible for the suppression vm�;5-����np]pun�p[�hu�hkkp[pvuhs�yvsl�pu�[ol�jvu[yvs�vm�[ol�puåhtth[vy`�ptt\ul�ylzwvuzl��;o\z��z[\k`pun�[ol�yln\sh[pvu� huk�hj[pvu�vm�a-7���th`�`plsk�puzpno[z�pu[v�tljohupztz�\uklys`pun�jhujly�huk�puåhtth[vy`�z`ukyvtlz��;v�l_htpul�[ol� regulation of zfp36, two conserved regions upstream of the start of transcription were assayed for their ability to confer serum induction. using a luciferase reporter gene system in hela cells, we show that neither region examined is induced by serum. key words: electron transfer, photosynthesis, quinones, uv-vis spectroscopy immediate-early genes are characterized by immediate transcriptional activation upon appropriate stimulation, including treatment of cells with mitogen-containing serum, without any need for intervening protein synthesis (johansen, 1995). as they are among the !rst genes to be activated in many signaling pathways, immediate-early genes provide a gateway to study the cellular response. zinc !nger protein 36 (zfp36) is an immediate-early gene product that regulates the expression of some genes by destabilizing their mrna. "e protein binds to au-rich sequences on mrna and catalyzes removal of the polyadenylated tail, reducing the half-life of the rna (blackshear, 2002). "e zinc !nger motif is common in dna and rna binding proteins, though zfp36 is unusual in that it has a cys3-his zinc-coordinating domain, as opposed to the more common cys2-his2 domain (tchen, 2004). "is slight di#erence in peptides might subtly impact the nucleotide binding dynamics of zfp36. notable genes regulated in part by zfp36 include tumor necrosis factor-ơ� (tnf-ơ), epidermal growth factor (egf) and c-fos, a transcription factor that promotes proliferation (amit, 2007) mice de!cient in zfp36 develop an excess of tnf-ơ, and the overabundance of this cytokine results in autoimmune complications such as arthritis (lai, 1998). as egf and c-fos are both implicated in proliferation, zfp36 is also thought to act as a tumor-suppressor (amit, 2007) "e zfp36 gene is regulated in several ways. tchen and colleagues et al. have shown that the zfp36 gene is mitogen-inducible via protein kinase p38 (tchen, 2004). "is result is consistent with qpcr data demonstrating that zfp36 is induced (upregulated) strongly by serum after one hour (figure 1). however, the upregulation seems to be independent of serum response factor (srf), the transcription factor that primarily mediates the serum response (prywes, unpublished data). lai and colleages showed that there are important regulatory elements downstream of the start of zfp36 transcription, and at least one of those regulatory elements is on an intron (lai, 1995; lai, 1998). given that zfp36 is induced by serum one hour after induction, we hypothesized that there may be additional serum-dependent regulatory elements upstream of the start of transcription. we identi!ed two conserved regions in the promoter region of mouse zfp36 and used a luciferase reporter gene assay to test the ability of each region to confer serum induction in hela cells. however, we ultimately ascertained that neither region was able to confer serum induction of zfp36 in hela cells. "is study and other studies of the regulation and action of zfp36 will provide vital information about this protein, and further research will play a signi!cant role in gaining a better understanding of the biological pathways for proliferation and in$ammation. introduction copyright: © 2009 the trustees of columbia university, columbia university libraries, some rights reserved, albert, et al. received dec. 29, 2009. accepted feb. 8, 2009. published april 5, 2009. *to whom correspondence should be addressed: columbia university, department of biology. 813b fairchild center. neew york, ny 10027 max.shutran@tufts.edu mrp6@columbia.edu the connections to appear unclear. by stereotaxically injecting virus we can also be more accurate and consistent with future injections into the mouse brain. different orientations of slicing must be tested in order to find the optimal orientation for preserving connections. once the connection from ca1 to the ec is clearly established, we also plan to examine the connections projecting from the ec to ca1. this anatomical data will contribute to the success of future electrophysiological studies. i would like to thank dr. siegelbaum for giving me the amazing opportunity to perform research in his lab and rebecca piskorowski for her helpful advice, constant guidance, and patience. i would also like to thank justine barry for her frequent and important support, guidance, and instruction, sebastian thuault for one of his mouse brains, and the rest of the siegelbaum lab. buzsaki, g. 1989. two-stage model of memory trace formation: a role for “noisy” brain states. neurosci 31: 551-570. de curtis, m., d. paré, and r.r. llinás. the electrophysiology of the olfactory-hippocampal circuit in the isolated and perfused adult mammalian brain in vitro. hippocampus. 1991 oct;1(4):341-54. eichenbaum, h., a.p. yonelinas, and c. ranganath. “the medial temporal lobe and recognition memory.” annu. rev. neurosci. 2007.30:123-152. eldridge, l.l., b.j. knowlton, c.s. furmanski, s.y. bookheimer, s.a. engel. nat neurosci. remembering episodes: a selective role for the hippocampus during retrieval. 2000 nov;3(11):1149-52. franklin, keith b. j., and george paxinos. the mouse brain in stereotaxic coordinates, third edition. new york: academic, 2007. godement, p., j. vaselow, s. thanos, and f. bonhoeffer. (1987) a study in developing visual systems with a new method of staining neurons and the processes in fixed tissue. development, 101: 697-713. heredia, m., m. santacana, and f. valverde. a method of using dii to study the connectivity of cortical transplants. journal of neurosci methods. 36 (1991) 17-25. honig, m.g. and r.i. hume (1985) interactions between sympathetic preganglionic neurons and sympathetic ganglion neurons in vitro. neurosci. abstr. 11:98. hyman, b.t., g.w. van hoesen, a.r. damasio, and c.l. barnes. alzheimer’s disease: cell-specific pathology isolates the hippocampal formation. science 225: 11681171. johnston, daniel, and david g. amaral. “hippocampus.” the synaptic organization of the brain. ed. gordon m. sheperd. 4th ed. new york: oxford up, 1998. print. naber, p. a., f. h. lopes da silva, m. p. witter. reciprocal connections between the entorhinal cortex and hippocampal fields ca1 and the subiculum are in register with the projections from ca1 to the subiculum. hippocampus. 11 (2001) 99-104. o’keefe, john. a review of the hippocampal place cells. prog neurobiol 13: 419–439, 1979. tamamaki, n. and y. nojyo. (2003) preservation of topography in the connections between the subiculum, field ca1, and the entorhinal cortex in rats. journal of comparative neurology, 353: 379-390. van groen, t., i. kadish, and j. m. wyss. species differences in the projections from the entorhinal cortex to the hippocampus. brain res bull. 2002 feb-mar 1;57(34):553-6. van groen, t., p. miettinen, and i. kadish. the entorhinal cortex of the mouse: organization of the projection to the hippocampal formation. hippocampus. 2003;13(1):133-49. vindal-sanz, m., m.p. villegas-pérez, g.m. bray, and a.j. aguayo. (1988) persistent retrograde labeling of adult rat retinal ganglion cells with carbocyanine dye dii. exp. neurol., 102:92-101. witter, m.p., floris g. wouterlood, pieterke a. naber, and theo van haeften. anatomical organization of the parahippocampal-hippocampal network. ann n y acad sci. 2000 jun;911:1-24. zola-morgan s., l.r. squire, and d. g. amaral. human amnesia and the medial temporal region: enduring memory impairment following a bilateral lesion limited to field ca1 of the hippocampus. j neurosci. 1986 oct;6(10):2950-67. references (jruv^slknltlu[z columbia undergraduate science journal open-access publication | http://cusj.columbia.com 22 cusjvolume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu m o le c u la r bi o lo g y cusjcolumbia undergraduate science journalresearch articles 23cusj volume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu m o lec u la r bio lo g ypriate ph, dna binds weakly to the calcium phosphate particles, and cells which endocytose the precipitate may take up the genetic material. each transfection comprised a reporter plasmid, an rlsv40 plasmid, and the non-luciferase plasmid pcan, thus keeping the total dna quantity constant at approximately 1 µg/µl. !e rl-sv40 plasmid carries the enzyme renilla luciferase, which is coupled to a promoter from the sv40 virus. since renilla luciferase can be measured independently of "re#y luciferase, and the sv40 promoter is expressed constitutively, this plasmid was used to control for variable transfection e$ciencies. in one experiment, the alternate renilla plasmid rl-tk was used, which has a tyrosine kinase promoter in place of the sv40 promoter. cells were plated onto 24-well plates the evening prior to transfection at a density of 6.5 x 104 cells per well, in dulbecco’s modi"ed eagle medium (dmem) containing 10% ncs. transfection mixtures were prepared using 1.75 µl of 2.5 m calcium chloride, the appropriate plasmids, and water to a total volume of 16 µl. !e reporter plasmids were added at a concentration of 125 ng per well, and the renilla plasmid at a concentration of 66 ng per well. for each reporter gene, four identical wells were prepared to allow for duplicates. each transfection mixture had a total volume of 64 µl. to each tube, 64 µl of hbs (a ph-adjusted mixture containing soluble phosphate salts) was added. !e tubes were vortexed brie#y and allowed to sit for 10 minutes at room temperature to allow the calcium phosphate to precipitate. !en 32 µl of each mixture was added to the appropriate well. !e following morning, the media was replaced with 10% ncs in dmem. serum induction and luciferase assay on the afternoon of the day following transfection, the media in all of the wells were changed to 0.2% neonatal calf serum (ncs). !e following morning, the cells in half of the wells were induced by adding 125 µl of pure ncs, raising the serum concentration to 20%. !ree hours after serum induction, all cells were washed with sterile pbs and treated with 100 µl of passive lysis bu%er, which keeps cytosolic proteins intact during lysis. !e cells were left at room temperature for 20 minutes on a slowly rotating shaker. a bioluminescence detector was used to assay luciferase activity. two reagents were used for the assay: luciferase materials and methods plasmid construction two plasmids, pzfp-2791-fgl3 and pzfp-1835fgl3, were constructed from a plasmid called pzfp-3.7fgl3 that had been made previously. !e insert was excised with two di%erent enzyme combinations: kpni and hindiii in one case, and kpni and bglii in the other. !is resulted in linearized plasmids that carried an ampicillin resistance gene, a minimal promoter from the human cfos gene, and a "re#y luciferase gene. !e two promoter regions of interest were obtained from pcr ampli"cation of mouse genomic dna. !e primers delineated regions from -2591 to -1810 and from -1835 to -901 base pairs relative to the start of transcription, with regions chosen based on mammalian conservation patterns (figure 3). each primer was designed with overhangs to insert the restriction sequences into the pcr products. each pair of plasmid and pcr product was ligated at 14 ˚c for four hours using the t4 dna ligase enzyme. !e resultant plasmids were transformed into e. coli for ampli"cation. sequencing revealed that the pzfp1835-fgl3 plasmid had only a single-base change in the c-fos minimal promoter region. !e pzfp-2590-fgl3 sequence showed an extra, non-contiguous piece of mouse zfp36 dna that had been inserted due to the presence of an internal bglii site in the mouse gene. !is piece was excised and the plasmid ampli"ed to give the correct sequence. !e other plasmids transfected were pcyr61, a control for which moderate serum induction was expected (babic et al.), pofgl3, which contains only the c-fos minimal promoter and luciferase, and two other plasmids (pzfp3.7-fgl3 and pzfp-932-fgl3), which incorporate regions of the mouse promoter at -3700 and -932 bases respectively. transfection transfection was carried out using a calcium phosphate protocol. in this method, a "ne calcium phosphate precipitate introduces genetic material into cells. at the appro� figure 1 9lz\s[z�myvt�x7*9�mvy�j�mvz��tpn���huk�a-7����pu�/l3h�jlssz� � figure 2 .yhwo�pss\z[yh[pun�[ol�puk\j[pvu�vm�lhjo�ylwvy[ly�nlul�i`�zly\t��;ol�yh^�kh[h�jvuzpz[lk�vm�k\wspjh[l�whpyz�vm������huk�����zly\t�jvukp[pvuz��np]lu�hz�yh[pvz�vm�3(9�[v�:.�s\tpulzjlujl��;ol�]hs\lz�^lyl�jhsj\sh[lk�i`�uvythspapun�lhjo�kh[h�zl[�[v�vul�vm�[ol������yh[pvz��[olu� h]lyhnpun�[vnl[oly�[ol�k\wspjh[lz�huk��pm�h]hpshisl��v[oly�kh[h�zl[z�myvt�kpmmlylu[�kh[lz��5v�zpnupäjhu[�zly\t�puk\j[pvu�pz�zllu��l_jlw[�pu�[ol� a-7������jvukp[pvu��^opjo�\zlk�93�;2�pu�wshjl�vm�93�:=���hz�[ol�ylupssh�wshztpk� columbia undergraduate science journal open-access publication | http://cusj.columbia.com 24 cusjvolume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu m o le c u la r b io lo g y cusjcolumbia undergraduate science journalresearch articles 25cusj volume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu m o lec u la r b io lo g y activating reagent (lar) and stop & glow reagent (sg). lar is the substrate of the !re"y luciferase enzyme, and sg is the substrate of renilla luciferase. each well was measured in two steps. first, 5 µl of cell lysate is added to 25 µl of lar, and the luminescence was measured. second, 25 µl of sg was added to the mixture and the luminescence is measured again. when analyzing the data, each measurement was normalized by computing the ratio of lar to sg luminescence. zfp36 mrna was signi!cantly induced by serum quantitative pcr was run on cdna prepared from hela cells that were stimulated with serum for durations between 0 and 120 minutes (figure 1). zfp36 showed an induction pattern similar to c-fos, with strongest induction between 30 minutes and 1 hour. mig6, a gene that is not induced by serum, was used as a negative control. none of the reporter genes tested was signi!cantly induced by serum figure 2 shows a summary of the results of the luciferase assays performed on transfected hela cells. none of the reporter genes showed an appreciable increase in expression upon induction with 20% serum. minimal serum induction was observed when a di#erent renilla plasmid was used (see discussion). $e data do not support the hypothesis that conserved regions between -2150 and -901 confer serum induction of zfp36. $is result was reproducible through several rounds of transfections. $is suggests that the observed serum induction of zfp36 is mediated entirely by elements that are downstream of the start of transcription. one issue that arose in the luciferase data was the inconsistency of renilla measurements. $e sg values tended to vary, and previous experiments using a similar protocol indicated that renilla itself may be induced by serum. to test if serum induction of the renilla gene was a#ecting the results, an alternate renilla gene with a tyrosine kinase promoter (rl-tk) was substituted for rl-sv40 in one trial. $e calculated data do indeed show that pzfp-3.7fgl3 was induced slightly less than threefold in this trial, but both the lar and sg measurements from this trial were so di#erent from those in other trials that no conclusions could be drawn from this result. $e great variation in lar measurements indicates a non-trivial relationship between two transfected genes: that is, the presence of one gene in the transfection mixture can a#ect the e%ciency of the transfection of other genes. another possible confounder could have been the plasmid preparations themselves, as calcium phosphate transfection is ph-sensitive. using a di#erent transfection method could more conclusively resolve this issue. nevertheless, the signi!cant serum induction observed in the qpcr results does not appear to be mediated by any of the upstream regions that were examined. in conclusion, neither region tested was induced by serum when expressed on a reporter gene in hela cells. future studies may further investigate the downstream regulatory elements characterized by lai et al., or attempt to !nd other cellular stimuli that regulate zfp36. amit, ido, ami citri, tal shay et al. “a module of negative feedback regulators de!nes growth factor signaling.” nature genetics. 2007, 39(4): 503-512. babic, alexander m, maria l kireeva, tatiana v kolesnikova & lester f lau. “cyr61, a product of a growth factor-inducible immediate early gene, promotes angiogenesis and tumor growth.” proceedings of the national academy of sciences united states of america. 1998, 95(11): 6355-6360. blackshear, perry j. “tristetraprolin and other ccch tandem zinc-!nger proteins in the regulation of mrna turnover.” biochemical society. 2002, 30(6): 945-952. cen, bo, ahayla selvaraj & ron prywes. “myocardin/ mkl family of srf coactivators: key regulators of immediate early and muscle speci!c gene expression.” journal of cellular biochemistry. 2004, 93:74-82. johansen finn-eirik & ron prywes. “serum response factor: transcriptional regulation of genes induced by growth factors and di#erentiation.” biochimica et biophysica acta. 1995, 1242:1-10. lai, wi s, michael j $ompson & perry j blackshear. “characteristics of the intron involvement in the mitogeninduced expression of zfp-36.” $e journal of biological chemistry. 1998, 273: 506-517. lai, wi s, michael j $ompson, gregory a taylor, yi lui & perry j blackshear. “promoter analysis of zfp-36, the mitogen-inducible gene encoding the zinc finger protein tristetraprolin.” $e journal of biological chemistry. 1995, 270: 25266-25272. tchen carmen r, matthew brook, jeremy saklatvala & andrew r clark. “$e stability of tristetraprolin mrna is regulated by mitogen-activated protein kinase p38 and by tristetraprolin itself.” $e journal of biological chemistry. 2004, 279: 32393-32400. discussion and conclusion results � figure 3 upstream zfp36 promoter region, showing mammalian conservation regions. from uscs genome biogenetics, http://genome.ucsc.edu/ references microsoft word cusj 29.docx ________________________________________________________________________________ cusj 2019 __________________________________________________________________________________________ 29 abstract — this paper presents an agent-based model of a stock market in which investors trade based on heterogeneous and changing beliefs. the model extends that from (goodman, 2016), in which each agent trades based on the theory of dynamic investment in (merton, 1969), with his own belief on the return of the stock. the baseline model implies that optimistic investors buy stocks from pessimistic investors. then, the model is extended such that the beliefs of agents are time-varying: each agent adjusts his belief differently with the arrival of new information. the key findings from the simulation of the extended model are that (1) trade volume increases with higher heterogeneity of agents' response to news arrivals and (2) return volatility decreases with more positive or more negative average level of response to news arrivals. i. introduction stock prices are determined by trading, which arises from differences in opinions among investors. if all investors had the same expectations on stock returns, no trades would occur, because everyone would seek to trade in the same direction. thus, it makes sense to study the stock price as a representation of heterogeneous beliefs on stock returns materialized through trading. figure 1. time series of analyst forecasts of amazon's eps from i.b.e.s. it is also reasonable to think that these beliefs on stock returns change over time, with the influx of new information. more specifically, figure 1 illustrates the analysts’ amazon's earnings per share (eps) forecast from 2010. figure 1 reveals that the opinions are time-varying and heterogeneous across analysts. this may suggest that investors make different conclusions from the same set of information. this motivates modeling the heterogeneity of beliefs as an accumulation of the idiosyncratic response to the identical set of randomly arriving news information. there is a large body of literature regarding empirical evidence on the effect of heterogeneous beliefs on asset price dynamics, many of which including (diether, malloy, & scherbina, 2002) (chen, hong, & stein, 2002) (park, 2005) (berkman, dimitrov, jain, koch, & tice, 2009) (qu, starks, & yan, 2003), document a negative relationship between belief dispersion and the mean returns of stocks. (goetzmann & massa, 2005) and (yu, 2011) find that dispersion in beliefs is negatively correlated with future returns. on the other hand, (avramov, chordia, jostova, & philipov, 2009) show that such negative relation is limited to worst-rated firms, while (doukas, kim, & pantzalis, 2006) find that the relationship is positive. there are also papers that study the effect of belief dispersion on return volatility and trading volume. for instance, (banerjee, 2011) finds evidence that stock volatilities increase in belief dispersion and (goetzmann & massa, 2005) finds positive relationship between trading volume and belief dispersion. to study how interactions among investors with heterogeneous and changing beliefs affect stock prices, i apply an agent based model (abm) approach. the abm approach offers several advantages that representative agent models do not. firstly, an abm allows its agents to have heterogeneous beliefs, not just heterogeneous utilities. while representative agent models allow for heterogeneous utilities of agents, which means that agents can have different preferences and thus different objectives, all of its agents must agree on the return of the assets. the representative agent theorem requires the assumption that all agents should have homogeneous beliefs. however, abm operates without that assumption, a property which i find useful for studying markets in which the beliefs on asset returns are heterogeneous across investors. secondly, agents in an abm follow a simple behavioral rule, instead of optimizing with the full knowledge of the state of the world. thirdly, the key dynamics in an abm are generated endogenously from the interaction among the agents. in relation to abm and economics, (westerhoff & franke, 2012) illustrates the usefulness of abm in designing economic policies with their example of technical traders, fundamental traders, and a central authority model to study the impact of simple intervention strategies on asset price dynamics. (lengnick, 2011) constructs an abm for business cycles in an economy and illustrates that an abm can reproduce many stylized facts without the strict assumption of rationality. (lengnick, 2011) also finds that the aggregate behavior generated by this abm is not equal to the results of a microeconomic optimization by the representative agent. a more empirical application was conducted by (baptista, hinterschweiger, farmer, low, & uluc, 2016), who developed an abm for the uk housing market to examine the effect of macroprudential policies on key agent based trading model of heterogeneous and changing beliefs jaehoon jung* nyu courant institute of mathematical sciences, 251 mercer st #801, new york, ny 10012 cusj 2019________________________________________________________________________________ __________________________________________________________________________________________ 30 housing market indicators. their results imply that a larger buy-to-let sector might amplify house price cycles and lead to higher price volatility. these papers show that the abm can be economically useful in reproducing stylized facts without strong assumptions on agents or equilibriums; the abm also generates non-standard aggregate behavior that is markedly different from that of a representative agent. in this paper, i extend the baseline model from (goodman, 2016) to construct a model in which agents trade based on heterogeneous and changing beliefs. in the baseline model from (goodman, 2016), each agent trades based on his beliefs and the theory of dynamic investment in (merton, 1969), in which an agent maximizes his power utility by holding a fixed proportion of his wealth in risky assets. this baseline model implies that optimistic investors buy stocks from pessimistic investors. i then extend the model by having the agents adjust their beliefs in a manner that differs from randomly arriving news information. by simulating the extended model, i find that the trade volume increases with the higher heterogeneity of the agents' response to the news arrivals. i also find that the return volatility decreases with a more positive or negative average response level to news arrivals. ii. model the baseline model from (goodman, 2016) uses the result from (merton, 1969) to define the trading behaviors of the agents. an agent maximizes power utility v(t) by allocating his wealth w(t) between a riskless asset and a risky asset, the value of which evolves with mean ! and volatility σ. !"#$%[' ( ) ] subject to dw = rw t dt + µ − r x t dt + σx t d56 the solution to this problem is to keep a fixed proportion of his wealth in risky assets. this proportion # is a function of the risky asset's excess return µ − r, return volatility σ, and the agent's risk aversion parameter γ. 8 9 = :((9), where : = =-? (@ab)cd this proportion # determines the agents' trading behavior: each agent will buy or sell stocks to maintain that proportion of wealth in stocks. for instance, if the price of the risky asset goes up while # does not change, the agent will sell the risky asset to maintain the proportion of his wealth in the risky asset. in the baseline model, each agent trades based on the merton proportion and there are n such agents in the market. an important assumption of this model is that each agent has his own belief: agent k has beliefs ef and gf and risk aversion parameter hf. it is further assumed that the total number of stocks in the market does not change and that there is only one risky asset (stock) and one riskless asset (cash). from the assumptions of the model, the following identities hold: first, the wealth of each agent is the sum of her cash and stock value: (f 9 = if 9 + jf 9 k(9) then, the wealth of the economy can be expressed as: (f 9 l fm@ = (if 9 + jf 9 k 9 ) l fm@ second, in equilibrium, each agent has the optimal ratio of risky assets. (f 9 :f 9 = 8f 9 = jf 9 k(9) the wealth of the economy at equilibrium can thus be expressed as: (f 9 l fm@ = 1 :f 9 jf 9 k(9) l fm@ third, total number of stocks remains invariant: jf(9@) l fm@ = jf(9o) l fm@ let n denote the total number of stocks in the market. j = jf(9) l fm@ fourth, total cash in the economy remains invariant: if(9@) l fm@ = if(9o) l fm@ fifth, from 3 and 4, i find that the wealth of the economy is increased only by the rise in stock price: (f(9o) l fm@ (f 9@ l fm@ = j(k 9o -k 9@ ) consider a simple case in which only one of the agents, agent 1, increases his e@ and thus his :@. when only agent one increased its e@ and thus �@, he would not be able to transact at current price if all the other agents are content with their portfolios. therefore, agent 1 should keep bidding higher prices to buy stocks until he achieves his new proportion :@ of his wealth in stocks. when agent 1 pushes the price up from s to k + pk, wealth of agent k, (f, will increase by d(f = jfpk. as :f has not changed for q > 1, the change in the optimal stock value for agent k would be p8f = :fjfps. if agent k buys ________________________________________________________________________________ cusj 2019 __________________________________________________________________________________________ 31 !"# number of stocks to rebalance to the new optimal allocation, the value of his stock holdings would change from "#$ to ("# + !"#)($ + !$). therefore, the following equation holds for each agent other than agent 1: !(# = *#!+# = *#"#!$ = "# + !"# $ + !$ − "#$ = "#!$ + $!"# rearranging the equation above, the number of stocks that agent k should buy is: !"# = − "# 1 − *# $!$ this suggests that in response to agent 1's optimism, all the other agent will sell stocks (assuming that *# ≤ 1)to maintain their merton proportion of their wealth in stocks. by separating the variables, i get: !"# "# = −(1 − *#) !$ $ by solving this equation and plugging in the initial condition, i derive the number of stocks that agent k will hold in equilibrium: "# 0 = "#(1 2 1 3 ) 4567(3) for 8 ≠ 1 as total number of stocks remains invariant, the number of stocks owned by the agent 1 who changed his views can be expressed as: "4 0 = ""#(0)( $ 0 $ 0 ) 4567(3) < #=> from the identities and the stock holdings derived above, i now derive the new equilibrium price given that only agent 1 revised its µ4 and thus *4. at the new equilibrium, both identities 1 and 2 should hold: +#(0) < #=4 = (@# 0 + "# 0 $(0) < #=4 = 1 *# 0 "# 0 $(0) < #=4 as total cash in the economy is invariant, (@# 0 + "# 0 $ 0 ) < #=4 = @# 0 < #=4 + "#(0)$(0) < #=4 = 1 *# 0 "# 0 $(0) < #=4 as total number of stocks is invariant, @#(0) < #=4 + "$ 0 = 1 *# 0 "# 0 $(0) < #=4 now, i separate agent 1 from others: @# 0 < #=4 + "$ 0 = 1 *4 0 "4 0 $ 0 + 1 *# 0 "# 0 $(0) < #=> plugging in the stock holdings of agents in new equilibrium, @# 0 < #=4 + "$ 0 = 1 *4 0 ""# 0 $ 0 $ 0 4567 3< #=> $ 0 + 1 *# 0 < #=> "# 0 $ 0 $ 0 4567 3 $(0) by dividing up the sum, i get: @# 0 < #=4 + "$ 0 = 1 *4 0 "$ 0 − 1 *4 0 "# 0 �0 $ 0 4567 3< #=> $ 0 + 1 *# 0 < #=> "# 0 $ 0 $ 0 4567 3 $(0) by rearranging the terms, i get the following equation for s(t): 1 − 1 *4 0 "$ 0 + 1 *# 0 1 *4 0 < #=> "# 0 $ 0 4567 3 $(0)67 3 − @# 0 < #=4 = 0 i confirm that the equation above is consistent with the simulation. iii. model extension now, consider a more realistic case in which agents adjust their beliefs based on the arrival of news information. i assume that news information arrives through poisson process and the time between each of them is exponentially distributed. each agent's reaction to news information can be decomposed into two parts: the common response a3 and agent-specific response b#,3. d# 0 + e0 -d# 0 = a3 + b#,3 cusj 2019________________________________________________________________________________ __________________________________________________________________________________________ 32 the common response can be considered as the original news information shared by all agents; i will denote it by !". i assume that !" is normally distributed with mean µ$ and variance %$&. in this setting, µ$ would be the average level of response of agents to the news information and %$& would be the variability among the news. !"~((µ$, %$&) then, for a given original new information, each agent reacts differently as would actual investors in the market. i assume that this idiosyncratic response, denoted as ,-,", is distributed normally with mean µ. and variance %.&. as i want this term to capture only the effect specific to a certain agent, i have the mean µ. equal to 0 and set %.& as the parameter that determines the level of heterogeneity among agents' reactions to news information. !"~((0, %.&) iv. results i simulated the extended model with news information. starting from an identical distribution of wealth, stock holdings, and beliefs for all agents, i had the news information arrive in exponentially distributed intervals. figure 2. sample stock price path generated from simulation. figure 2 shows one of the stock price paths generated by the simulation. the simulated stock price seems to become more volatile as heterogeneity across agent beliefs increases with the dispersion in reaction to news information accumulating over time. figure 3. trade volume increases in %.& (dispersion in agent response to new information). figure 3 shows that the trade volume increases with dispersion in reaction to news information, which is the variance in agent-specific response (%.&). this result is in line with my expectation, because this means that agents trade more when they respond to news information more differently. thus, in this model, heterogeneity in beliefs gives rise to trading in the market. figure 4. return volatility decreases in µ$ (average level of response to new information). figure 4 shows that return volatility decreases with more positive µ$, which means that the average level of response by agents to new information is more positive. i also find that return volatility decreases with more negative µ$, showing no asymmetry. this result is also within my expectation as my model implies that the stock price would be less volatile if all the investors are becoming more optimistic about the stock. ________________________________________________________________________________ cusj 2019 __________________________________________________________________________________________ 33 figure 5. trade volume decreases in news arrival frequency. figure 5 shows that trade volume decreases with more frequent news arrivals. this result is counterintuitive as i expect the heterogeneity in beliefs to accumulate faster with more frequent news arrivals. as of now, i do not have a clear explanation as to why agent trade less if news information arrives more frequently. v. future work as the second extension to the baseline model, i aim to add an extrapolative component to agent beliefs. !" # + %# = 1 − )" !" # + *+ + ,",+ + )"!+ where )" denotes the degree to which belief of agent k reflects extrapolated market return !+. this would make the model more realistic, because it would be more reasonable that investors also adjust their beliefs based on the realized return, not just on news information. it would also make the belief dynamics more endogenous, since each agent's beliefs will be influenced by the realized return, which is determined by other agents' beliefs. the addition will make this abm more valuable as the beliefs of agents will be affected by interaction (trading) with other agents with different beliefs, instead of relying solely on the arrival of news information, which is exogenous in this model. vi. conclusion in this paper, i construct a model in which the stock price is determined by investors who trade based on their heterogeneous and changing beliefs. by simulating this model, i find that trade volume is positively correlated with dispersion in investors' response to news arrivals, suggesting that investors trade more when they have more different opinions. this result is consistent with the empirical evidence documented in (goetzmann & massa, 2005). however, i do not find a clear positive relationship between dispersion in reaction to news and return volatility, which was found in (banerjee, 2011). moreover, i also find that the return volatility is negatively correlated with more positive (or more negative) average level of response by investors to news arrivals, which means that the stock price is less volatile if all investors are becoming more optimistic (pessimistic). finally, i find that trade volume is negatively correlated to news arrival frequency, contrary to my expectation that more frequent news arrival would accelerate the divergence in agent beliefs and thus increase trade volume. the results from simulations of this model suggests that the model is capable of replicating a number of intuitive results, such investors trading more when they disagree and returns being less volatile when investors are generally more optimistic or more pessimistic, with only simple, reasonable assumptions. i believe that this model can be used as a framework for studying the effect of heterogeneous beliefs on market dynamics under more specific or complex conditions. for example, i could extend the model such that investors adjust their beliefs according to the actual trading price. then, i would be able to identify the effect of the extrapolative component by comparing the result with that from simulations without the component. in summary, this paper contributes to the literature of heterogeneous beliefs by showing that abm could be an alternative theoretical approach and providing a simple framework. author information corresponding author *email: jj1419@stern.nyu.edu funding sources received funding from the summer undergraduate research experience program by the math department at courant institute of mathematical sciences. acknowledgments i am grateful to my advisor jonathan goodman for his guidance and mentorship. all errors are my own. references [1] j. goodman, "stock prices in metronia," unpublished paper, 2016. [2] r. c. merton, "lifetime portfolio selection under uncertainty: the continuous-time case," the review of economics and statistics, 1969. [3] k. b. diether, c. j. malloy and a. scherbina, "differences of opinion and the cross section of stock returns," journal of finance, 2002. [4] j. chen, h. hong and j. c. stein, "breadth of ownership and stock returns," journal of financial economics, 2002. [5] c. park, "stock return predictability and the dispersion in earnings forecasts," journal of business, 2005. [6] h. berkman, v. dimitrov, p. c. jain, p. d. koch and s. tice, "sell on the news: differences in opinion, and trading activity," journal of financial economics, 2009. cusj 2019________________________________________________________________________________ __________________________________________________________________________________________ 34 [7] s. qu, l. starks and h. yan, "risk, dispersion of analyst forecasts and stock returns," working paper, 2003. [8] w. n. goetzmann and m. massa, "dispersion of opinion and stock returns," journal of financial markets, 2005. [9] j. yu, "disagreement and return predictability of stock portfolio," journal of financial economics, 2011. [10] d. avramov, t. chordia, g. jostova and a. philipov, "dispersion in analysts' earnings forecasts and credit rating," journal of financial economics, 2009. [11] j. a. doukas, c. kim and c. pantzalis, "divergence of opinion and equity return," journal of financial and quantitative analysis, 2006. [12] s. banerjee, "learning from prices and the dispersion in beliefs," review of financial studies, 2011. [13] f. westerhoff and r. franke, "agent-based models for economic policy design: two illustrative examples," working paper, 2012. [14] m. lengnick, "agent-based macroeconomics a baseline model -," working paper, 2011. [15] r. baptista, m. hinterschweiger, j. d. farmer, k. low and a. uluc, "macroprudential policy in an agentbased model of the uk housing market," staff working paper no. 619, 2016. summer undergraduate research fellowship. bounoutas, a., and chal!e, m. 2007. touch sensitivity in caenorhabditis elegans. p"ugers arch. 454, 691–702. brenner, s. 1974. #e genetics of caenorhabditis elegans. genetics 77, 71–94. chal!e, m., & au, m. 1989. genetic control of di$erentiation of the caenorhabditis elegans touch receptor neurons. science 243, 1027 – 1033. finney, m., ruvkun, g., & horvitz, h.r. 1988. #e c. elegans cell lineage and di$erentiation gene unc-86 encodes a protein with a homeodomain and extended similarity to transcription factors. cell 55, 757 – 769. goodman, miriam b. “mechanosensation.” wormbook (may 2008): 9 june 2011. hobert, oliver, and heiner westphal. “functions of limhomeobox genes.” trends in genetics 16.2 (2000): 75-82. trends in genetics. 21 jan. 2012. kamkin, andre glebovich, and irina s. kiseleva, eds. mechanosensitive ion channels. new york: springer, 2007. #e tavernarakis lab. 3 aug. 2011. mitani, s. 1995. genetic regulation of mec-3 expression implicated in the speci!cation of the mechanosensory neuron cell types in caenorhabditis elegans. dev. growth di$er., 37, 551 557. mitani, s., du, h., hall, d. h., driscoll, m., & chal!e, m. 1993. combinatorial control of touch receptor neuron expression in caenorhabditis elegans. development 119, 773 – 783. stewart gw, argent ac, dash bc, 1993. stomatin: a putative cation transport regulator in the red cell membrane. biochim biophys acta 1225,15–25 #acker c, rose am, and sheps ja. “caenorhabditis elegans dpy-5 is a cuticle procollagen processed by a proprotein convertase.” cellular and molecular life sciences 63.10 (2006): 1193-204. topalidou, irini, alexander van oudenaarden, and martin chal!e.” caenorhabditis elegans aristaless/arx gene alr-1 restricts variable gene expression.” proceedings of the national academy of sciences 108.10 (2011): 4063-4068. topalidou, irini, and martin chal!e. “shared gene expression in distinct neurons expressing common selector genes.” proceedings of the national academy of sciences 108.48 (2011): 1-6. pnas. 25 feb. 2012. way, j.c. and chal!e, m. 1988. mec-3, a homeobox-containing gene that speci!es di$erentiation of the touch receptor neurons in c. elegans. cell 54, 5–16. way, j. c., & chal!e, m. 1989. #e mec-3 gene of caenorhabditis elegans requires its own product for maintained expression and is expressed in three neuronal cell types. genes dev. 3, 1823 – 1833. wu j, duggan a, chal!e m. inhibition of touch cell fate by egl44 and egl-46 in c.elegans. genes dev 2001; 15, 789-802. zhang s, arnadottir j, keller c, caldwell ga, yao ca, chal!e m (2004) mec-2 is recruited to the putative mechanosensory complex in c. elegans touch receptor neurons through its stomatin-like domain. curr biol 14, 1888 –1896. zhang, y., ma, c., delohery, t., nasipak, b., foat, b.c., bounoutas, a., bussemaker, h.j., kim, s.k., chal!e, m., 2002. identi!cation of genes expressed in c. elegans touch receptor neurons. nature 418, 331–335. columbia undergraduate science journal open-access publication | http://cusj.columbia.edu 15 cusjvolume 6spring 2012 columbia undergraduate sci j http://cusj.columbia.edu % ,2 /2 * < references breast cancer affects one in eight women per year, and 70% of patients with stage iv breast cancer develop metastases in bone, fdxvlqj�olih�wkuhdwhqlqj�vlgh�hiihfwv��:h�h[dplqh�wkh�xvh�ri�dq�h[sodqw�v\vwhp�wr�plplf�euhdvw�fdqfhu�phwdvwdvlv�wr�erqh�zlwkrxw� confounding cell types and to determine the role that osteocytes, the bone mechanosensing cells, may have in the development of metastasis. using the explant system, a custom cell seeder and sterile cell culture techniques, we introduced metastatic mda-mb-231/ *)3�fhoov�wr�d�wkuhh�glphqvlrqdo�erqh�pdwul[�zlwk�rvwhrf\whv�rqo\��&rqirfdo�lpdjlqj�frq¿upv�wkdw�euhdvw�fdqfhu�fhoov�zhuh��lq�idfw�� successfully seeded onto bone cores, mimicking metastasis. though additional experiments will be necessary to determine the importance of breast cancer-osteocyte interactions, this study shows that the explant system is a viable methodology for studying breast cancer in bone. cancer is a devastating disease that is responsible for thirteen percent of deaths worldwide and has a$ected countless families and individuals throughout the world (cancer, world health organization). it a$ects 1 in 8 women every year (u.s. breast cancer statistics). breast cancer originates from the inner lining of the lobules that supply the milk ducts in the breast (wolf et al., 2003). stage iv breast cancer is metastatic, meaning that it is violent and transcends the host organ (the breast) and spreads to a secondary site. #e cancer that metastasizes is still considered breast cancer. it has been reported that up to 70% of stage iv breast cancer patients will experience some form of metastasis of breast cancer to bone (roth et al., 2009). patients often experience pathological fractures, intense pain, hypercalcemia, and various nervous compression complications (zhang et al., 2010). #ese devastating e$ects are caused by an imbalance of bone remodeling, which involved the interactions of the three main bone cell types. bone is comprised of three types of cells: osteocytes (ocy), osteoblasts (ob) and osteoclasts (ocl). osteocytes are the primary mechanosensing cells in bones (burger et al., 1995). #ey regulate the activity of osteoblasts and osteoclasts. osteocytes are “trapped” in the mineralized bone matrix, and are thus they are thought to have only signaling functions, both intercellular and intracellular. after osteocytes sense a mechanical load, that load is transduced into a chemical signal is sensed by the cells. #is stress is translated into a biochemical signal that is communicated to the osteoblasts and osteoclasts, the bone forming and bone resorbing cells (burger et al., 1995). osteoblasts synthesize the bone matrix, which is subsequently deposited and calci!ed to become bone mineral. when osteoblasts secrete too much matrix, they become stuck in the bone, and as a result they completely di$erentiate into osteocytes (saladin, 2007; buckwalter et al. 1995). osteoclasts, on the other hand, resorb bone. to accomplish this, they use their “ruf"ed” membrane (as shown in figure 1) to create a seal around bone and then pump enzymes and hydrochloric acid to degrade the matrix (saladin, 2007; buckwalter et al. 1995). #ere are two types of metastasis: osteolytic and osteoblastic. osteolytic metastases break down bone and are the most common type of metastasis for breast cancer. osteoblastic lesions, characterized by excess bone formation, a$ect 15-20% of patients. mixed types also exist, wherein the patient experiences unnecessary bone excess as well as dearth (zhang et al., 2010). #e large majority of stage iv breast cancer cases end in metastasis to bone because bone has high levels of growth factors that breast cancer uses to survive. we can see that bone is a likely candidate for breast cancer metastasis due to the presence of “transforming growth factor b (tgfb), insulin-like growth factors i and ii (igf), !broblast growth factors (fgfs), platelet-derived !e veri"cation of a novel explant system used to determine the role of osteocytes in the breast cancer vicious cycle sonia bansal*, genevieve n. brown, x. edward guo %rqh�%lrhqjlqhhulqj�/derudwru\��'hsduwphqw�ri�%lrphglfdo�(qjlqhhulqj��&roxpeld�8qlyhuvlw\��1hz�<run�1<� copyright: © 2012 the trustees of columbia university, columbia university libraries, some rights reserved, bansal, et al. received 12/22/2012. accepted 1/31/2012. published 4/1/2012 *to whom correspondence should be addressed: bone bioenjlqhhulqj�/derudwru\��'hsduwphqw�ri�%lrphglfdo�(qjlqhhulqj�� columbia university, new york ny. sb3218@columbia.edu abstract introduction cusjcolumbia undergraduate science journal research articles 16cusj volume 6spring 2012 columbia undergraduate sci j http://cusj.columbia.edu % ,2 ( 1 * ,1 ( ( 5 ,1 * growth factors (pdgfs), bone morphogenetic proteins (bmps), and calcium” which make up the microenvironment of bone cells (zhang et al., 2010). !us, the bone microenvironment is particularly suitable for cancer cell survival. when breast cancer cells metastasize to bone, they "rst proliferate from the primary site, using proteases such as urokinase plasminogen activator (upa), matrix metalloproteinases (mmps) and mitogen-activated protein kinases (mapk) which degrade surrounding matrix proteins, letting the cancer cells migrate more easily. additionally, cadherin-11 and n-cadherin expression has been shown to have a strong link to bone metastases (zhang et al., 2010). next, cancer cells must migrate to the bones. studies suggest that chemokines may home breast cancer cells to bones. !e cells must adhere to the bones via the glycoproteins in the extracellular matrix (mundy, 2002). !e bvb3 integrin has been implicated as a key protein in this step. lastly, the cells must proliferate. !is proliferation occurs via a “vicious cycle” (figure 2) between tumor growth and bone remodeling which utilizes the osteoblasts (bone-forming) and osteoclasts (boneresorbing) in the bone (zhang et al., 2010). !e bone remodeling cycle is normally stable, but following the introduction of metastatic breast cancer cells, this cycle creates a positive feedback loop. first, the breast cancer cell produces parathyroid hormone-related protein (pthrp) that signals osteoblasts to increase the amount of receptor activator of nuclear factor b ligand (rankl). additionally, osteoblasts will decrease the amount of osteoprotegrin (opg), which is a decoy receptor for rankl (guise, 2000). as a result of an increase in the amount of rank ligand and less opg to competitively inhibit the activity of receptor activator of nuclear factor b (rank) on osteoclasts, more primary osteoclasts are activated to become osteoclasts. !e ratio of rankl to opg determines the rate of osteoclast activity. when the bone is degraded, transforming growth factor-b (tgf-b) is released. tgf-b stimulates the production of pthrp, completing the positive feedback loop. !is is generally referred to as the “vicious cycle” between breast cancer and the bone environment (mundy, 1997). most studies assume that proliferation of breast cancer cells is due to osteoclast and osteoblast action and with no other cell types involved. !is phenomenon appears to explain the predominance of osteolytic lesions in breast cancer patients (burger et al., 1995). however, studies have shown that osteocytes also produce opg and rankl when regulating bone modeling (bonewald, 2011). !ere are some who argue that osteocytes can degrade bone matrix through a rapid and ephemeral mechanism called osteocytic osteolysis (teti and zallone, 2009). !ese links lead us to believe that cancer cells may a#ect osteocytes, which in turn affect the other cell types. it is possible that the accepted “vicious cycle” has another component: osteocytes. !is would suggest that preventing osteocytes from being a#ected by cancer cells may stop the vicious cycle, because then the osteoclasts and osteoblasts would not have to be blocked from cancer cells. in order to explore the interactions between breast cancer cells and osteocytes, we utilize a trabecular bone explant system. it is a tractable and novel system that contains only a cleaned bone core and no confounding cell types, ensurcolumbia undergraduate science journal open-access publication | http://cusj.columbia.edu 17 cusjvolume 6spring 2012 columbia undergraduate sci j http://cusj.columbia.edu % ,2 ( 1 * ,1 ( ( 5 ,1 * ing that any variables created by cells’ signaling processes are controlled. however, other cell types can be added to the bone cores as this system is seedable, meaning that we can add precise amounts of whichever cells we wish to examine. although this study does not require it, this system also allows for long-term culture by incorporating a perfusion chamber which prevents the bone cores from dying due to lack of nutrition (chan et al., 2009). imaging and 3-d bone reconstruction allow for the determination of how breast cancer cells distribute themselves on the bone and a#ect bone volume fraction. figure 3 shows dendriticlike signaling processes that breast cancer cells develop in 3-d culture. !ese chains are only visible on a 3-d sca#old and are indicative of proliferation (wolf et al., 2003). it is clear that the mechanism of the metastasis of breast cancer cells to bone must be evaluated, but this would require a system in which a bone has osteoblasts and osteoclasts, with the variables being the presence of osteocytes and breast cancer cells. !is experiment would have to appropriately model metastasis and then indicate a loss or gain in bone. !rough this study, we demonstrate a methodology that can be expanded and extrapolated to a larger scale experiment in order to model the breast cancer vicious cycle with the appropriate cell types. preparation of bone cores !e tarsal-metatarsal joint was harvested from a calf under sterile conditions. specimens were sacri"ced a day prior to the procedure. !is joint is used due to the $at articulating surface found between these two joints, which is conducive to drilling. cores were obtained by drilling into the articular cartilage and through trabecular bone using a starlite diamond-tipped brill bit. drilled cores were 7 mm in diameter (see figure 4). !e cores were lightly cleaned with an interplak dental water pick running phosphate bu#ered saline (pbs), a saline solution containing nacl and nak. !e cores were then cut using an isomet low-speed saw to a height of 7 mm, avoiding the articular cartilage on top and the less dense bottom half of the cores. !e cores were cleaned again thoroughly with the water pick running pbs before undergoing a serial trypsin treatment, which involved the cores being incubated in 3-4 ml of trypsin at 37oc for 8 minutes and then being thoroughly cleaned with pbs via the water pick. !is treatment was replicated three times. !e trypsin treatment was done to remove the surface cells from the cores. by completely eradicating the surface cells on bones, the cores are free of variable cell types. twelve cores were harvested and cultured in supplemented media, which consisted of a-minimum essential medium (a-mem), 10% cusjcolumbia undergraduate science journal research articles 18cusj volume 6spring 2012 columbia undergraduate sci j http://cusj.columbia.edu methods figure 1 schematic depicting the types of cells in bones. the image depicts the osteocytes, trapped in the matrix of the bone, osteoblasts, on the surface, and the multiqxfohdu�rvwhrfodvw��zklfk�yld�wkh�uxiàhg�phpeudqh�� resorbs bone. $gdswhg�iurp�kwws���frxuvhv�zdvklqjwrq� hgx�frqm�ehvv�erqh�erqh�fhoov�sqj� figure 2 schematic of the “vicious cycle” pathway. the tumor cell produces pthrp, increasing the amount of rankl and decreasing the amount of opg. this activates more osteoclasts, which degrades bone, releasing 7*)²ȕ��frpsohwlqj�d�srvlwlyh�ihhgedfn�orrs� [adapted from steeg ps, theodorescu d, nat clin 3udfw�2qfro�����@ figure 3&rqirfdo�lpdjlqj�ri�0'$�0%����*)3�fxowxuhg�lq�rq�d�erqh� core. the image highlights the dendrite-like connections made ehwzhhq�fhoov�rq�d���'�prgho��6fdoh�edu� �����0�� >0loohu������@� figure 4&ohdqhg�erqh�fruh��phdvxulqj���pp�lq�khljkw�dqg�gldphwhu� [adapted from chan, me, lu, xl, huo, b, baik, ad, chiang, v, hw�do����������$�7udehfxodu�%rqh�([sodqw�0rgho�ri�2vwhrf\wh� 2vwhreodvw�&r�&xowxuh�iru�%rqh�@ % ,2 ( 1 * ,1 ( ( 5 ,1 * fetal bovine serum (fbs), and 1% penicillin-streptomycin (pen-strep). creation of experimental groups !e cores were randomly assigned to four experimental groups, “osteocytes and breast cancer” (ob), “osteocytes” (o), “dead osteocytes and breast cancer” (db) and “dead osteocytes” (d) (see figure 5). half the cores were put through a repeated freeze-thaw technique in order to kill the osteocytes still in the cores. !is was done to see if mineralized bone, but not necessarily active osteocytes, a"ects breast cancer cell activity. half of the dead osteocyte and live osteocyte groups were seeded with mda-mb-231/gfp stage iv breast cancer cells via a custom cell seeder created in the lab on day 0. !e bone cores were stuck onto needles and submerged in a solution containing 5x104 cells for each core. !e loader was placed on a stir plate with a magnetic stir #y for an hour to facilitate uniform adheration of the cancer cells. figure 6 shows a schematic of the cell seeder. !is process simulated breast cancer in bone. confocal microscopy !e cores were initially stained with cell tracker red and incubated for 45 minutes in order to stain the viable cells in the bone cores before being seeded. !e cores were imaged using a leica confocal microscope on day 0 after initial seeding, and on day 4 using confocal microscopy. !e confocal imaging on day 0 indicates that the cancerous cells do exist on the bone matrix. figures 7.1 and 7.2 depict cores that were seeded with breast cancer cells, and the green #uorescence indicates that the cells are on the core. in contrast, figures 7.3 and 7.4 only show the osteocytes in the lacunae of the bone core. !e subsequent set of images on day 4 shows a much higher number of cells on the bone core, as indicated by figures 8.1 and 8.2. figures 8.3 and 8.4 again indicate that there are no breast cancer cells in these two experimental groups. !e confocal imaging on day 0 indicates that the cancerous cells do exist on the bone matrix. figures 7.1 and 7.2 depict cores that were seeded with breast cancer cells, and the green #uorescence indicates that the cells are on the core. in contrast, figures 7.3 and 7.4 only show the osteocytes in the lacunae of the bone core. !e subsequent set of images on day 4 shows a much higher number of cells on the bone core, as indicated by figures 8.1 and 8.2. figures 8.3 and 8.4 again indicate that there are no breast cancer cells in these two experimental groups. columbia undergraduate science journal open-access publication | http://cusj.columbia.edu 19 cusjvolume 6spring 2012 columbia undergraduate sci j http://cusj.columbia.edu % ,2 ( 1 * ,1 ( ( 5 ,1 * confocal imaging of the bone cores confocal imaging of the bone cores on days 0 and 4 indicate a simulated metastatic process. !e $rst step to metastasis is the adheration of foreign, cancerous, metastatic cells into a new environment. !e images, taken on day 0, indicate that the cells were added in an e"ective amount that attaches to the bone core appropriately. figures 7.1 and 7.2 show this adherence and verify that we were able to force adheration of cancerous cells onto the matrix. additionally, the negative controls are veri$ed in figures 7.3 and 7.4, which do not have green #uorescence. after this, day 4 imaging indicates that the breast cancer cells were sustained on the bone cores as they maintained viability and also expanded in quantity on the bone cores. figures 8.1 and 8.2 show the proliferation of the cells, as there are many more cells that e"ectively take over the lacunae. when looked at in conjunction with figures 7.1 and 7.2, it is clear that the cancerous cells are thriving on day 4. !ese images suggest the proliferation of the cancerous cells, indicating that metastasis was e"ectively mimicked in this system. cusjcolumbia undergraduate science journal research articles 20cusj volume 6spring 2012 columbia undergraduate sci j http://cusj.columbia.edu % ,2 ( 1 * ,1 ( ( 5 ,1 * figure 5 visual representation of the experimental design groups. first, we have cores with live osteocytes and no breast cancer cells, as a control. the variable condition is the core with live osteocytes with breast cancer cells. we also included two more groups, both with dead osteocytes. this was because we wanted to see if the mineralized bone, not necessarily the active osteocytes, affected the way breast cancer cells acted. figure 6 schematic of the cell seeding technique. bone cores were stuck on the needles and suspended in the media. adheration was facilitated by the stir bar.[adapted from chan, me, lu, xl, +xr��%��%dln��$'��&kldqj��9��hw�do����������$�7udehfxodu�%rqh� ([sodqw�0rgho�ri�2vwhrf\wh�2vwhreodvw�&r�&xowxuh�iru�%rqh�@ figure 7'd\���frqirfdo�lpdjlqj�ri�erqh�fruhv��,pdjhv�wdnhq�rq�/hlfd�&rqirfdo�plfurvfrsh��5hg�àxruhvfhqfh�lv�&hoo� 7udfnhu�5hg�dqg�vkrzv�rvwhrf\whv��ghdg�dqg�dolyh���zkloh�juhhq�àxruhvfhqfh�lqglfdwhv�*)3�wdjjhg�0'$� 0%������7khvh�lpdjhv�zhuh�wdnhq�wr�hqvxuh�wkdw�wkh�&hoo�7udfnhu�5hg�vwdlq�glg�vwdlq�rvwhrf\whv��zklfk�lv�fohdu� through the masses of red signals near the lacunae, and to ensure that the breast cancer cells were properly vhhghg�rqwr�wkh�erqh�fruhv��6fdoh�edu� �����p� results discussion modi!cations to be made as the results indicate that the methodology is sound, we must look to future studies and the next step in the larger scale study. !e harvest technique, while sterile (given that there were no infections), was time consuming and would bene"t by becoming more e#cient, so instead of 8 cores being harvested at a time, we could harvest 32 at a time. day 4 confocal imaging shows a decreased cell tracker red signal, and a longer culture would require a more stable imaging technique. a longer culture would provide for more complicated culture conditions involving all three bone cell types. using more bone cell types would require more than two di$erent %uorescent stains, relying on the use of the lab’s four-color microscopy technology. explant system as a viable in vitro model for study !e qualitative (confocal images) data suggests that the explant system is a good method to simulate metastasis of breast cancer, proving that it can be used to further study columbia undergraduate science journal open-access publication | http://cusj.columbia.edu 21 cusjvolume 6spring 2012 columbia undergraduate sci j http://cusj.columbia.edu % ,2 ( 1 * ,1 ( ( 5 ,1 * figure 8'd\���frqirfdo�lpdjlqj�ri�erqh�fruhv��,pdjhv�wdnhq�rq�/hlfd�&rqirfdo�plfurvfrsh��5hg�àxruhvfhqfh�lv�&hoo�7udfnhu�5hg� dqg�vkrzv�rvwhrf\whv��ghdg�dqg�dolyh���zkloh�juhhq�àxruhvfhqfh�lqglfdwhv�*)3�wdjjhg�0'$�0%������7kh�&hoo�7udfnhu� red stain begins to fade at this point. these images were taken to ensure that the breast cancer cells were properly prolifhudwlqj�dfurvv�wkh�erqh�fruh��zklfk�lv�fohdu�jlyhq�wkh�qxpehu�ri�fhoov�suhvhqw�rq�wkh�fruhv��6fdoh�edu� �����p� this topic. !e study presents a starting point for future experiments, as it demonstrates our methodology is viable and e#cient. future studies would rely on the use of µct tomography to determine the bone volume/total volume, also known as the bone volume fraction (bvf). bvf is a good measure of the amount of bone mineral present in a sample as it determines the space of bone mineral versus total space of the core. di$erences in bvf between seeded and nonseeded cores would indicate cancer-induced lesions. !e incidence of these lesions would determine the extent of a variable cell’s role on the vicious cycle. in the future, the study would start to include other cell types, creating di$erent experimental groups such as a core with osteoclasts, osteoblasts and breast cancer cells, but no osteocytes. !is condition would simulate a bone without osteocytes, and the resulting bvf, as compared to a bone core with all three cell types and breast cancer, would indicate how osteocytes a$ect bone cell activity in the presence of breast cancer cells. !e future implications of this experiment are promising, as it is the "rst step in determining the individual role of each cell type; determining "nally how responsible osteocytes are for metastatic breast cancer spreading to bones. sb thanks xeg for allowing her to work in bone bioengineering lab and gnb for guidance and teaching her the cell techniques. sb was funded by the columbia department of biological sciences from the summer undergraduate research fellowship. bonewald, lf (2011). !e amazing osteocyte. journal of bone and mineral research 26 229-238. buckwalter ja, glimcher mj, cooper rr, et al. bone biology part i: structure, blood supply, cells, matrix, and mineralization. j bone joint surg am 1995; 77: 1256-75. burger, eh, klein-nulend, j, van der plas, a, and nijweide, pj (1995). function of osteocytes in bone--their role in mechanotransduction. j nutr 125 2020s-2023s. “cancer.” world health organization from http://www.who.int/mediacentre/factsheets/fs297/en/. chan, me, lu, xl, huo, b, baik, ad, chiang, v, et al. (2009). a trabecular bone explant model of osteocyte-osteoblast co-culture for bone mechanobiology. cell mol bioeng 2 405-415. guise ta. molecular mechanisms of osteolytic bone metastases. cancer 2000; 88 (s12): 2892-8. mundy gr. metastasis to bone: causes, consequences and !erapeutic opportunities. nature reviews cancer 2002; 2: 584-93. mundy, gr (1997). mechanisms of bone metastasis. cancer 15 15461556. saladin ks. anatomy and physiology: !e unity of form and function. new york: mcgraw-hill 2007. teti, a, and zallone, a (2009). do osteocytes contribute to bone mineral homeostasis? osteocytic osteolysis revisited. bone 44 11-16. wolf, k, mazo, i, leung, h, engelke, k, von andrian, uh, et al. (2003). compensation mechanism in tumor cell migration. !e journal of cell biology 160 267-277. “u.s. breast cancer statistics.” breastcancer.org from http://www.breastcancer.org/symptoms/understand_bc/statistics.jsp. zhang, y, ma, b, and fan, q (2010). mechanisms of breast cancer bone metastasis. cancer lett 292 1-7. cusjcolumbia undergraduate science journal research articles 22cusj volume 6spring 2012 columbia undergraduate sci j http://cusj.columbia.edu % ,2 ( 1 * ,1 ( ( 5 ,1 * acknowledgments references conclusions columbia undergraduate science journal open-access publication | http://cusj.columbia.com 42 cusjvolume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu v ir o lo g y cusjcolumbia undergraduate science journalresearch articles 43cusj volume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu v ir o lo g y overall survival rates. preventing glioma invasion has the potential to convert this highly malignant tumor into a focal disease, which could then be e!ectively treated with focal therapies, such as radiation and surgery (madsen, 2003; gillespie, 1999). invasion and migration of malignant gliomas tissue invasion by malignant gliomas is a multi-step process. central to this is the ability of transformed cells to crawl through the extracellular matrix. "e initial step requires receptor-mediated adhesion of tumor cells to matrix proteins, followed by a second phase of degradation of the matrix by tumor-secreted proteases. "e ability of invasive glioma cells to navigate these diverse anatomic structures and molecular substrates raises the question of whether speci#c mechanisms and phenotypes of invasive cells are involved in dispersion following speci#c pathways (gillespie, 1999). gliomas typically invade the brain by migrating long distances through white matter tracts and by in#ltrating cortex and subcortical gray matter structures. when migrating through brain white matter, gliomas move in a manner remarkably similar to that previously described for neural progenitor cells. given the submicrometer size of the extracellular spaces that are present in the tightly packed neuropil of the brain, most other motile cells would likely #nd it impossible to migrate because there would be no room to allow a lamellipodium to form. "e unique form of motility shared by gliomas and neural progenitors re$ects adaptations that these cells have to make to a uniquely challenging environment with small e!ective pore sizes (beadle, 2008). one of these adaptations is in how glioma cells use myosin ii, the major source of cytoplasmic contractile force (gillespie, 1999). myosin ii is absolutely required for migration in the brain, where its major role is to push the bulky nucleus and cell body through the small pores found within the brain matrix. "is feature distinguishes glial cells from other cancers because other cancers do not require myosin ii for the collapse of their nucleus. "e progenitor-like motility of glioma cells represents a method of migration that is speci#cally adapted for traversing the densely packed neuropil of the mature brain (beadle, 2008). poliovirus poliovirus, among other human pathogenic viruses, is being studied increasingly along with the development of novel therapeutic agents against malignant glioma. poliovirus, the causative agent of poliomyelitis, is classi#ed within the enterovirus genus of the family picornaviridae. poliovirus is composed of a rna genome and a protein capsid. "e genome is single-stranded positive-sense rna genome that is about 7500 nucleotides long, while the viral particle is about 30 nanometers in diameter with icosahedral symmetry. because of its short genome and its simple composition, poliovirus is widely regarded as the simplest signi#cant virus. first isolated in 1909 by karl landsteiner and erwin popper, poliovirus has become one of the most well-characterized viruses, and has become a useful model system for understanding the biology of rna viruses. "ere are three serotypes of poliovirus, pv1, pv2, and pv3, each with a slightly di!erent capsid protein. "e capsid proteins de#ne cellular receptor speci#city and virus antigenicity. pv1 is the most common form encountered in nature, however all three forms are extremely infectious. wild polioviruses can be found in approximately 10 countries. pv1 is highly localized to regions in india, pakistan, afghanistan, and egypt, but following outbreaks of poliomyelitis in 2003–2004 it remains widespread in west and central africa. wild poliovirus type 2 has probably been eradicated; it was last detected in october 1999 in uttar pradesh, india. wild pv3 is found in parts of only #ve countries (nigeria, niger, pakistan, india, and sudan) (merrill, 2004). poliovirus, like other picornaviruses, is characterized by a single positive strand genomic rna. infection occurs via the fecal-oral route; meaning that one ingests the virus, which then replicates within the alimentary tract. in 95% of cases only a primary, transient presence of the virus in the bloodstream occurs (called a viremia) and the poliovirus infection is asymptomatic. in about 5% of cases, the virus spreads, and replicates in other sites such as brown fat, the reticuloendothelial tissues, and muscle. "is sustained replication causes a secondary viremia, and leads to the development of minor symptoms such as fever, headache and sore throat. paralytic poliomyelitis occurs in less than 1% of poliovirus infections. paralytic disease occurs when the virus enters the central nervous system (cns) and replicates in motor neurons within the spinal cord, brain stem, or motor cortex, resulting in the selective destruction of motor neurons; leading to either temporary or permanent paralysis and, in rare cases, to respiratory arrest and death (pliaka, 2007). "ere are two key mechanism by which poliovirus evades the immune system. first, it is capable of surviving the highly acidic conditions of the gastrointestinal tract, allowing the virus to infect the host and spread throughout the body via the lymphatic system. second, because it can replicate very quickly, the virus overwhelms the host organs before an immune response can be mounted. in addition, there are more than two such mechanisms; additional ones include inhibition of rna and protein synthesis, and inhibition of protein secretion, both of which interfere with the ifn respsonse. fully assembled poliovirus leaves the con#nes poliovirus receptor cd155 over-expression e!ect on migration in c6 glioma cells zahra bhaiwala1,*, christine marshall2 1columbia college, columbia university, new york, ny 10027, 2phillips academy, andover, ma, 01810 abstract cell migration plays a key role in brain cancer invasion, an early step in metastasis, and proteins that regulate migration are often up-regulated in tumor cells. the poliovirus receptor cd155 has recently been shown to affect migration levels of select malignant glioma strains, fueling the exploration of treating brain cancer with oncolytic virus recombinants. in the beginning phase of experimentation, we analyzed the migratory behavior of control c6 gliomas through two rounds of [yhuzäs[ly�hzzh`z��0u�vykly�[v�l_wsvyl�[ol�m\uj[pvu�vm�*+����pu�nspvth�tpnyh[pvu��^l�[olu�jvuk\j[lk�[^v�v]ly�l_wylzzpvu� experiments with the use of a full-length cdna expression vector and compared the resulting migration rates to control kh[h��>l�mv\uk�[oh[�hu�h]lyhnl�vm��������jvu[yvs�jlssz�jyvzz�v\y�[yhuzäs[lyz�hm[ly�����ov\yz��huk�hu�h]lyhnl�vm�������jlssz� [yhuzmlj[lk�^p[o�[ol�7=9�l_wylzzpvu�]lj[vy��6\y�ylz\s[z�ylålj[lk�h�������pujylhzl�pu�jlss�tpnyh[pvu�k\l�[v�hu�pujylhzlk� wylzlujl�vm�[ol�wyv[lpu�*+�����0u�vykly�[v�jvuäyt�[oh[�*+����^hz�l_wylzzlk�pu�[ol�yh[�*�z��^l�z\izlx\lu[s`�jvuk\j[lk� two western blots: one comparing control c6s and transfected c6s, and the other running control c6s next to a human u87 glioma control cell line. the blots establish that cd155 is indeed present in both of these human and rat strains of glioma, and that our over-expression was successful. key words: glioblastoma multiforme; tumor migration; poliovirus receptor cd155; protein over-expression. introduction after the human brain completes its development soon after birth, the vast majority of its cells enter the g0 phase, in which they never divide again. one exception to this rule is when a brain tumor develops as a result of abnormal, unregulated growth of cells. "e atypical brain cells re-enter the cell cycle because of alterations in any of a large number of genes that control cell division and growth. astrocytomas and oligodendrogliomas are types of gliomas that are the most common primary tumors of the adult brain. primary brain tumors arise from cells of the brain itself rather than traveling, or metastasizing, to the brain from another location in the body like other known forms of cancer. tumors are generally classi#ed in four grades, with grade 1 being the most benign and grade 4 being the most malignant. signs that the tumor is growing rapidly include cells undergoing division (mitosis), the presence of newly-formed blood vessels (angiogenesis), and evidence that the tumor is outgrowing its blood supply (necrosis). typically, malignant gliomas show an area of central necrosis surrounded by a highly cellular rim of viable tumor. malignant gliomas have remained resistant to therapy and are capable of spreading long distances within the brain (madsen, 2006). in glioblastoma multiforme (gbm), the most aggressive form of brain cancer, tumor cells disperse so extensively that common treatment approaches such as surgical resection or radiation therapy are not e!ective in checking progression. unfortunately, signi#cant in#ltration of normal brain matter is not limited to the most malignant of tumors. "e invasive behavior expressed by all levels of malignant gliomas has limited the e!ectiveness of local therapies and contributes to the high mortality rate seen in these tumors. "e position of the tumor is also crucial. brain stem gliomas are particularly di%cult to treat, regardless of their grade. "e brain stem is such a complicated and delicate part of the brain that completely removing the tumor is highly unlikely. unfortunately, high doses of radiation are not recommended for treatment, as this may cause too much damage to the normal brain stem (giese, 2003). having an agent that blocks migration is key to managing glioblastoma multiforme. interventions to control the spreading of glioblastoma multiforme have the potential to slow the clinical course of the disease and improve copyright: © 2011 the trustees of columbia university, columbia university libraries, some rights reserved, porter, et al. received jan. 2, 2011. accepted jan. 26, 2011. published april 1, 2011. *to whom correspondence should be addressed: columbia college, columbia university, new york, ny 10027. zsb2104@columbia.edu columbia undergraduate science journal open-access publication | http://cusj.columbia.com 44 cusjvolume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu v ir o lo g y cusjcolumbia undergraduate science journalresearch articles 45cusj volume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu v ir o lo g y materials and methods somal entry site (ires) within the 5’ untranslated region of the viral genome, which is active in cells of neuronal origin and allows translation of the viral genome without a 5’ cap. non-transductional targeting involves altering the genome of the virus so it can only replicate in cancer cells. !is can be done by either transcription targeting, where genes essential for viral replication are placed under the control of a tumor-speci"c promoter, or by attenuation, which involves introducing deletions into the viral genome that eliminate functions that are dispensable in cancer cells, but not in normal cells. gromeier et al. replaced the normal poliovirus ires with a rhinovirus ires, altering tissue speci"city (gromeier, 2000). !e resulting pv1(ripo) virus was able to selectively destroy malignant glioma cells, while leaving normal neuronal cells untouched. although the poliovirus itself has been demonstrated in previous studies to have an oncolytic e#ect in certain types of gliomas, it is still somewhat unclear what the actual role of the receptor cd155 might have, if any, in this relatively new discovery of cancer treatment through viral infection. in glioblastoma, the most aggressive form of brain cancer, tumor cells disperse so extensively that current treatment approaches such as surgical resection or radiation therapy have little e#ect in checking progression. !e survival of patients with malignant gliomas, and a median survival time of patients with malignant gliomas is still anticipated as a bleak 12 to 14 months. invasive cells remaining after the surgical resection signi"cantly contribute to the demise of the patient. any successful treatment will have to treat the invasive portion of the tumor and the core lesion. speci"cally targeting invasive glioma cells remains an interesting concept because invasiveness must be a common characteristic of most tumor cells remaining after surgical resection. because it has been increasingly shown that glioma invasion is regulated by distinct trigger mechanisms, downstream e#ector molecules of the invasion process represent the best treatment targets. it has been previously found that cd155/pvr was highly expressed in both u87 human glioma cells and primary glioblastoma tumor tissue, and that inactivation of cd155/pvr reduced cell migration in vitro (sloan, 2005). !ese "ndings suggest a novel role for cd155/pvr in regulating motility and has prompted me to explore this function further in the c6 rat glioma cells we have here in lab by conducting an over-expression experiment to mirror the relationship that has been found in the knockdown experiments. cell culture c6 rat glioma cell lines (canoll lab, columbia university) were cultured in dulbecco’s modi"cation of eagle’s medium (dmem) with 4.5 g/l glucose, l-glutamine, and sodium pyruvate supplemented with 10% fetal bovine serum and 1x penn strep (mediatech). cells were incubated at 37°c in a napco series 8000 wj co2 incubator under a humidi"ed 5% co2 atmosphere. cells were passaged every seven days onto fresh poly-l-lysinecoated t75 $asks, and medium was changed every three days. transfection of cdna expression vector we obtained glycerol stocks of e. coli transformed with a 4.395 kb pcmv-sport6 expression vector (image id: 3902226, american type culture collection), which used a 50 µg/µl ampicillin marker and was stored inde"nitely at -80°c. dna was isolated according to standard plasmid preparation procedures and absorbance was measured at 260 nm via spectrophotometer (biomate). one day prior to transfection, the cells were trypsinized from the t75 $ask and added to 10 ml of growth medium. !e mixture was then spun in a tabletop centrifuge (2000rpm for 5 minutes) and resuspended in 5 ml of antibiotic-free growth medium. cells were triturated and then counted by inserting 10 µl into a hemocytometer and approximately 5x105 cells were plated into 60 mm tissue culture dishes containing 3 ml of growth medium. in two separate sterile eppendorf tubes, the appropriate amount of dna (based on absorbance value) and 20 µl of lipofectamine 2000 reagent were diluted in 50 µl of serum-free medium. both tubes were mixed gently and incubated for 5 minutes at room temperature. !e tubes were combined, incubated for another 20 minutes and the entire mixture was added to the transfection dishes. 6 hours after transfection, the media was changed to regular growth media (dmem + 10% fbs + p/s). cells were allowed to grow for 48 hours prior to trans"lter assays. transfilter migration assay in preparation for the migration assay, the c6 cells were trypsinized from the tissue culture dishes, added to 5 ml growth medium, and spun at 1600 rpm for 5 minutes. cells were resuspended in 1-2 ml of serum-free medium and kept on ice. 3µm pore fluoroblok trans"lters (bd biosciences) were coated with 10% type 1 rat tail collagen (bd biosciences) in sterile de-ionized water at 37°c for 1 hour. 500 µl of serum-containing growth medium was added to the bottom of each well beneath the "lter to stimulate migration in a single direction (top to bottom). cells were counted and the number was adjusted so that 5x105 cells were added to the top of each trans"lter in a 200 µl mixture. cells were left in the incubator to run on the "lters for 5.5 hours. post-migration, cells were "xed in 4% paraformaldeof its host cell 4 to 6 hours following initiation of infection in cultured mammalian cells. !e mechanism of viral release from the cell is unclear, but each dying cell can release up to 10,000 polio virions (merrill, 2004). cd155/pvr !e cellular life cycle of poliovirus is initiated by binding to the cell surface receptor cd155. !e virion is taken up via endocytosis, and the viral rna is released. translation of the viral rna occurs by an ires-mediated mechanism. !e polyprotein is then cleaved, yielding mature viral proteins. !e positive-sense rna serves as template for complementary negative-strand synthesis, producing double-stranded replicative form (rf) rna. many positive strand rna copies are produced from the single negative strand. !e newly synthesized positivesense rna molecules can serve as templates for translation of more viral proteins, or can be enclosed in a capsid to form progeny virions. lysis of the infected cell results in release of infectious progeny virions. !e presence of cd155 is thought to de"ne the animals that can be infected by poliovirus. cd155 has only been thought to be on the cells of humans, higher primates, and old world monkeys. poliovirus is however strictly a human pathogen, and does not naturally infect any other species. cd155 is a type i transmembrane glycoprotein in the immunoglobulin superfamily. commonly known as poliovirus receptor (pvr) due to its involvement in the cellular poliovirus infection in primates, cd155’s normal cellular function is in the establishment of intercellular adherens junctions between epithelial cells. like many other receptor molecules used by picornaviruses, cd155 is a long, highly glycosylated, single-span surface molecule. while its predominant function is related to cellular adhesion and activation, cd155 also activates natural killer cells and has been reported to play a role in cell motility and tumor cell invasion. it consists of 3 successive ig-like domains (d1, d2, and d3), a transmembrane domain and a cterminal cytoplasmic domain. !e human cd155 gene is expressed in 4 splice variants (ơ, ơ, ƣ, and dž) of which 2 variants (ơ and ƣ) are lacking the transmembrane domain and are released from the cell subsequent to their expression. !e cd155ơ and cd155dž isotypes di#er only in their cytoplasmic domains and can both function as pv receptors zhang, 2008). cd155 protein binds speci"cally to the extracellular matrix component vitronectin. con"rming the activity pattern of the cd155 promoter in the developing spinal cord, morphogenic factors active in the $oor plate and notochord—the transcription factors sonic hedgehog (shh) and its downstream gli e#ectors—strongly activate the cd155 promoter and induce cd155 expression. both shh and gli transcription factors have been implicated in the oncogenesis of neuroectodermal tumors. !us, the role of shh and gli transcription factors in cd155 gene regulation suggested that cd155 expression may occur ectopically in neuroectodermal malignancies. evidence for cd155 expression in neuroectodermal tumors stems mainly from studies of neuroectodermal tumor cell lines that are susceptible to oncolytic poliovirus-based agents (solecki, 2002). oncolytic applications viral oncolysis has been recently recognized as a new development in the treatment of malignant glioma. oncolytic viruses must speci"cally target tumor cells, a property related to the fact that tumor cells often have aberrant innate immune reponses. !e "rst reported incidences of viral oncolysis, over a century ago, were due to nonintended exposure to naturally occurring viruses or after administration of live attenuated vaccine strains. in the last twenty years, new prospects for genetically manipulating viruses have opened possibilities for increasing the tumor speci"city and lowering the toxicity of oncolytic viral agents. !ese e#orts have given rise to oncolytic adenoviruses, herpesviruses, reoviruses, vesicular stomatitis virus, and most recently, polioviruses. !e antineoplastic e#ects of oncolytic viruses are subject to multifaceted interactions with the host cell. !e primary prerequisite for viral oncolysis is expression of cellular receptors mediating viral entry in malignant cell types. in preparation for clinical applications using oncolytic viruses, analysis of viral receptor expression in the target tissue is highly desirable in order to select tumor types and patients most likely to respond favorably to therapeutic intervention. cell adhesion molecules of the immunoglobulin superfamily are aberrantly expressed in malignant glioma. amongst these, the human poliovirus receptor cd155 provides a molecular target for therapeutic intervention with oncolytic poliovirus recombinants. analysis of 6 cases has indicated that cd155 over-expression is commonly associated with high-grade malignant glioma (hgl). cd155 expression levels in tumor tissues corresponded to those in primary tissue cultures derived from the tumors. furthermore, cd155 expression in primary glioma explant cultures was equivalent to that found in established glioma cell lines used in preclinical evaluations of oncolytic poliovirus recombinants. poliovirus has been genetically modi"ed through insertion of regulatory sequences derived from human rhinovirus type 2 to selectively replicate within and destroy cancerous cells. successful oncolysis with the use of poliovirus infection depends directly on the presence of cd155 in targeted tumors (sloan, 2005). poliovirus has a plus-strand rna genome, the translation of which depends on a tissue-speci"c internal ribocolumbia undergraduate science journal open-access publication | http://cusj.columbia.com 46 cusjvolume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu v ir o lo g y cusjcolumbia undergraduate science journalresearch articles 47cusj volume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu v ir o lo g y hyde at 4°c for 20 minutes. after 3 washes with phosphate-bu!ered saline (pbs) at room temperature, the cells were stained with the nuclei stain hoechst (calbiochem/emd) at a 1000x dilution of 10µg/10µl for 30 minutes at 4°c. "e cells were washed again 3 times with pbs and then plated onto glass coverslips for analysis. data analysis "e number of nuclei on the bottom of each of both control #lters was counted in two low power #elds (25x) of a zeiss axioplan microscope and averaged. "e nuclei on the top of the #lters was also counted in two low power #elds and averaged in order to further assess the rate of migration in normal c6 cell lines. lysates lysis bu!er (ripa bu!er) was prepared (1 m ph 8 tris, 5 m nacl, 20% np40, 10% sds, distilled h20 up to 50 ml) and stored at 4°c. 5x running bu!er (7.5 g tris, 47 g glycine, 25 ml 10% sds, distilled h20 up to 500 ml), 5x transfer bu!er (9.5 g tris, 47 g glycine, distilled h20 up to 500 ml), and 10x tbs (6 g tris, 44 g nacl, dh20 up to 500 ml, ph 7.4) were prepared and kept at room temperature. 0.5 ml of edta was added to 4.5 ml of lysis buffer to make a 10x stock of protease inhibitor cocktail. "e cocktail was then diluted 1:10 in lysis bu!er (50µl to 450µl). cells were washed twice with pbs and 250µl of the lysis bu!er mixture was added directly to the tissue culture dish. cells were scraped and pipeted to an epindorph tube, vortexed for 10 seconds and returned to ice, and then vortexed for 30 seconds. cells were then spun for ten minutes at 1600 rpm and the supernatant was pulled o! into a new epindorph. "e lysates were snap-frozen with liquid nitrogen and stored at -80°c until quanitation. western blot analysis cd155 expression levels were determined by using a western blot procedure. 24 µl protein lysate was mixed with 8 µl 4x sample bu!er (20 ml 40% glycerol, 4 g sds, .01 g .02% bromophenol-blue, 5 ml beta-mercaptoethanol (bme), 45ml dh20) and boiled for ten minutes. 10% stacking and separating acrylamide gels were prepared according to page recipe calculator (chang bioscience) and placed into the chamber (vwr), followed by the addition of 1x running bu!er up to the brim of the apparatus. "e protein mixture, along with prestained standard (biorad) was then loaded into the wells and run at 110v for approximately two hours. after the gel was run, it was placed directly into 1x transfer bu!er. "e nitrocellulose was prewet in sterile water to activate and the gels were subjected to overnight electrophoretic transfer (.03amp) following standard procedures. membranes were rinsed in sterile water and then blocked with 50 ml of tris-bu!ered saline tween-20 with 0.25 g bsa (tbstb) + 5% milk for one hour on the shaker at room temperature. after soaking, the blots were treated with anti-necl-5 rat monoclonal antibody clone 1a88 (gift of dr. yoshimi takai, stored at -80°c) diluted 1:1000 in tbstb and incubated overnight at 4°c. after the removal of the primary antibody and 3 consecutive ten-minute washes of tbst, the last with tbstb + 5% milk, the membranes were treated with secondary antimouse igg-alkaline phosphatase antibody produced in goat (sigma, stored at 4°c) diluted 1:10,000 in tbstb and set on the shaker for 1 hour. after three thorough rinses of the blots with tbst, they were treated with bm purple (roche) until the appearance of protein bands— after which they were stored at 4°c in tbst. with two successful western blots (see figures 5 and 6), we were able to formulate a comprehensive picture of c6 glioma migration in$uenced by transfection of pvr. to determine if increased synthesis of cd155 affects the migration of c6 glioma cells, we captured two images of each control and transfection well, as well as one #eld each of the cells remaining on the top of the #lter (see figures 1, 2, 3). it was not immediately prevalent through qualitatively assessing cell numbers that a substantial increase in cell migration occurred between the control and transfected cells, but a detailed count revealed an increase. "e nuclei counts for the control #elds were 315, 355, 260, and 287, respectively. "e numbers of cells transfected with the pvr expression vector that transversed the matrix were 327, 367, 332, and 392. "e average of migrating control cells was 304.25, in comparison with the transfection average of 354.5. our data re$ect an average percent increase of 16.5%. by conventional standards according to a statistical t-test (two-tailed p-value of 0.0957), this di!erence is not considered to be remarkably signi#cant. it is important to note, however, that the second round of trans#lter assays yielded a 32.4% increase in glioma migration as a result of the transfection (see figure 4). in the #rst western blot, which compared cd155 protein levels in c6 control cells to c6 cells transfected with the pvr over-expression vector, we could con#rm from the location of the bands (around 40 kda, just like the mass of cd155) that the antibody did recognize the figure 1 c6 control cells remaining on top of the äs[ly�hm[ly�h�����ov\y�tpnyh[pvu�wlypvk� c6 control cells at the bottom of äs[ly�hm[ly�����ov\y�tpnyh[pvu�wlypvk�� figure 2 pvr-transfected cells at the bottom of äs[ly�hm[ly�����ov\y�tpnyh[pvu�wlypvk� figure 3 figure 4 migrating c6 cells: control vs transfection with pvr expression vector avg. of assay 1 control: 335 tfect: 347 avg percent increase: 3.5% avg of assay 2 control: 273.5 tfect: 362 avg percent increase: 32.4% avg of all control: 304.25 tfect: 354.5 avg percent increase: 16.5% two-tailed p value = 0.0957 image content dna absorbance (µg/ul) nuclei count _______________________________________________________________________ _ control 1a 315 control 1b 355 control 2a 260 control 2b 287 tfect 1a .3 327 tfect 1b .3 367 tfect 2a .4 332 tfect 2b .4 392 results columbia undergraduate science journal open-access publication | http://cusj.columbia.com 48 cusjvolume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu v ir o lo g y cusjcolumbia undergraduate science journalresearch articles 49cusj volume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu v ir o lo g y human protein and that it is in fact present in the cells. because the band on the right, the transfected band, is so much darker, we can also assume that our over-expression last term was successful and that any impact on migration we saw was directly correlated to the up-regulation of cd155 in the c6 cells (figure 5). in the second western blot, which compared c6 and u87 control glioma cells, we can observe two distinct, dark bands (again located around 40 kda) (figure 6). !is is a strong indication that both c6 and u87 glioma cells naturally contain comparable and substantial levels of cd155 protein. we were attempting to mirror the results of a reputable knockdown study conducted by the department of physiology at tufts university school of medicine, (boston, massachusetts) and the department of neurobiology at yale university school of medicine (new haven, connecticut). !is study showed that a knockdown of cd155 by fali in u87mg cells resulted in a signi"cant (16 to 22%) decrease in trans-well migration (sloan, 2005). because the sirna expression vector was not available to conduct a knockdown experiment, our study would be used to perhaps mirror the relationship that has been found previously in the knockdown of cd155. we expected that over-expression of cd155 would signi"cantly increase migration based on the expression of protein levels indicated by western blot analysis. if the over-expression were successful, we would expect that the knockdown that we were unable to conduct out of lack of a rnai expression vector would also decrease migration. despite our predictions earlier in the year, there did not appear to be a tangible visible di#erence between the control pictures and the transfected pictures for this experiment as a whole. one may notice that in general the transfected cell "elds were clearer, and that the control "elds contained a lot of processes (especially in trans"lter assay 1) (figure 3). !is could be possibly interpreted as the presence of cd155 speeding up the growth of cells. if a knockdown in u87 glioma decreased migration of cells, we could infer that cd155 possibly enhances the tumor and may serve as a catalyst for tumor growth. cd155 does not appear to a#ect the shape or size of the cells. without the western blots, which were conducted towards the end of our study, we were not able to detect whether or not the transfection substantially increased the presence of cd155 or by how much. !is led us to question whether or not cd155 was actually present in c6 cells. we supposed that pursuing this inquiry would allow us to make further distinction between c6 cells and human strain glioma. at the very least, we would "nd our results to con"rm an expected di#erence between rat and human models. we predicted that increased cd155 expression in glioma cells was important for migration of the tumor cells in vivo. while our results indicate promise, only two successful experimental replicates were conducted and the variation observed was large—precluding a sound conclusion on the role of cd155 in migration. an additional control, a transfection with a plasmid encoding another protein, would be needed to add to future experiments in order to determine if increased synthesis of any other protein would have the same e#ect on migration. we would also hope to obtain the sirna in order to knock down pvr, which was not available at the time. !e challenge we faced during this experiment was the fact that cd155 is thought to be a uniquely human protein, because poliovirus only naturally infects humans. while mice have been genetically engineered to produce references pvr, we had no predictions as to whether the c6 rat glioma cells would express cd155 at all. cd155 analysis has never been carried out before in c6 gliomas. furthermore, the primary rat monoclonal antibody we used for our western blots was expected to react with necl5, the mouse origin of the protein—and not with the human origin cd155. we were essentially testing to see whether our rat gliomas do naturally express the human protein, and if the mouse antibody would recognize the human homolog. since the blots con"rm that cd155 is indeed present in c6 cells, in an amount comparable to that present in u87 human cells, we can con"rm that the mouse and human homologs of the protein are similar. indeed, cd155 and necl-5 are referred to in combination within scienti"c dialogue and have virtually the same structure. yet if the antibody treats cd155 and necl-5 as the same protein, why is the poliovirus itself speci"c to humans? not only do our "ndings shed light on the strict nature of speci"city characteristic to viruses, but these results suggest that cd155 has more than one function besides being the human receptor for poliovirus, and in concurrence with our assay counts, does e#ect migration on some level. although some of our counts were not statistically signi"cant, what we do see is that the second assay with a higher concentration of dna had an overall much larger percentage di#erence than the control. if we know that the c6 cells contain the human protein and their migration is in$uenced by it, we can now potentially use rats as a model for cd155/pvr oncolytic treatment in humans. a study done by the department of molecular genetics and microbiology at the state university of new york (stony brook, ny) demonstrated for the "rst time that highly attenuated poliovirus recombinants can infect and propagate in cell lines derived from malignant gliomas and, most interestingly, halted tumor progression and eliminated tumors in athymic mice. !ey proposed that susceptibility of these malignant cells to poliovirus may be mediated by expression of the cd155 gene in glial neoplasms (gromeier, 2000). as our understanding of the biology of brain cancer progresses, new knowledge about tumorigenesis and tumor biology can be used to diagnose, treat, and prevent this type of cancer. our experiment may help to support the concept that oncolytic poliovirus recombinants may be the next possible treatment for malignant glioma. beadle c, assanah mc, monzo p, vallee r, rosenfeld ss, canoll p. “!e role of myosin ii in glioma invasion of the brain.” mbc: vol. 19, issue 8, 3357-3368. august 2008. burgoyne am, phillips-mason pj, brady-kalnay sm, et al. “proteolytic cleavage of protein tyrosine phosphatase mu regulates glioblastoma cell migration.” cancer research: 6960-6968. august 18, 2009. giese a, et al. “cost of migration: invasion of malignant gliomas and implications for treatment.” american society of clinical oncology: 1624-1635. 2003. gillespie gy, soroceanu l, manning tj, gladson cl, rosenfeld ss. “glioma migration can be blocked by nontoxic inhibitors of myosin ii.” cancer research: 2076-2082. may 1999. gromeier m, lachmann s, rosenfeld mr, et al. “intergeneric poliovirus recombinants for the treatment of malignant glioma.” proc natl acad sci usa: 6803–6808. june 6, 2000. madsen sj, sun ch, tromberg bj, cristini v, de magalha n, hirschberg h. “multicell tumor spheroids in photodynamic !erapy.” lasers in surgery and medicine: 555–564. 2006. merrill mk, bernhardt g, sampson jh, wikstrand cj, bigner dd, gromeier m. “poliovirus receptor cd155-targeted oncolysis of glioma.” neuro-oncology: 208-217. 2004. pliaka v, dedepsidis e, kyriakopoulou z, papadopoulou i, levidiotou s, markoulatos p. “use of mutational pattern in 5’-ncr and vp1 regions of polioviruses for molecular diagnosis.” molecular and cellular probes: vol. 21, issue 4, 267-275. august 2007. sloan ke, stewart jk, treloar af, matthews rt, jay dg. “cd155/pvr enhances glioma cell dispersal by regulating adhesion signaling and focal adhesion dynamics.” cancer research: 10930–10937. december 1, 2005. solecki dj, gromeir m, mueller s, bernhardt g, wimmer e. “expression of the human poliovirus receptor/cd155 gene is activated by sonic hedgehog.” !e journal of biological chemistry: 277, 25697-25702. april 30, 2002 zhang p, mueller s. “crystal structure of cd155 and electron microscopic studies of its complexes with polioviruses.” pnas: vol. 105, no. 47, 18284-18289. november 2008. figure 5 western blot ac6 control vs c6 transfection with pvr western blot bc6 control vs u87 controlfigure 6 discussion microsoft word cusj_spring_2015_journal.docx !cusj%|%volume%9%|%spring%2015% 27% polynomial-time matrix-based method of determining subset sum solutions aubrey alston1 1department of computer science, columbia university, new york, ny, usa 1 introduction: the subset sum problem is a well-known member of the npcomplete complexity class: given a set of integers a and some constant c, is there some subset of a which sums to c? there have been no known algorithms or methods to solve the value-unbounded, general-case subset sum problem in polynomial time. a naive algorithm performs in exponential time by cycling through the possible subsets of a until it has either seen all subsets or found one that sums to c. a common pseudo-polynomial algorithm employs a dynamic programming method whose complexity is polynomial with respect to the length of the set and the range of the inputs, o(n(m-n)), where n is the length of the set and m-n is the range of inputs; however, this solution is not truly polynomial, as it is polynomial with respect to m-n, which is exponential in its number of bits. approximate algorithms exist which can be modified to find exact solutions; however, they too degrade to being exponential in the number of bits required to represent elements in the set. in contrast to pre-existing algorithms, the method described here does not concern itself with the various subsets that exist within the input set, but rather searches the solution space of a set of linear constraints when applied to an input set to deduce if a solution can exist; this method is a strategy which may be employed to find solutions satisfying the constraints of the subset sum problem in time polynomial with respect only to the length of the input, having generalcase applicability on the basis of universally occurring properties in sets satisfying the problem. conventions, definitions, and properties: given a set a of n greater than four elements and an instance of subset sum for a constant c, satisfied by a subset s of length greater than 2 (the algorithm first catches trivial cases for s of length 1 or 2), index a as follows: the strategy outlined will conform a to a set of linear constraints which will reveal a subset sum-satisfying subset if one exists. to this end, define a subset membership vector m specific to a such that the ith value in m is 1 if the ith element of a is an element of a given subset s of a summing to c, 0 otherwise. if such a subset s exists, m exists and encodes s within a. 2 if s exists, the four following linear constraints surrounding a, s, and m will be satisfied: 1. s sums to c. 2. s has finite length t. 3. there exists an index r for which the rth element of a is or is not in s. 4. there exists an index s for which the sth element of a is or is not in s. using the above constraints, an underdetermined system z can be constructed in the parameters of the constraints listed: the algorithm given below explores the solution spaces of a polynomial number of forms of z to construct the characteristic membership vector m for some subset s of a summing to c if one exists. the convention for the determining solution space of z(a, t, r, vr, s, vs) is to first form an equivalent set representation a’ by interchanging index 3 of a with index r, interchanging index 4 with index s in a, and solving z(a’, t, 3, vr, 4, vs) for m’ equal to m with likewise index permutations using matrices. to represent the solution space of z(a’, t, 3, vr, 4, vs), the outlined algorithm follows the convention of expressing solution space with respect to a particular solution of the system and any linear combination of the null space of the multiplier of z(a’, t, 3, vr, 4, vs). !cusj%|%volume%9%|%spring%2015% 28% 1 the followed solution convention is simple forward elimination, followed by back-substitution. the particular solution is chosen such that all free variables (the rank of the multiplier of the system is 4; given its representation, the free variables are a’5, . . . a’n) are assumed to be zero. the null space is then composed of n-4 special solutions each respectively assuming one unique mi, i = 5, . . ., n to be 1, all other mj to be 0. this convention then allows m’ for any a’ to be expressed as follows: (10) the prescribed convention thus provides a representation of the solution space that may be explored to determine if a valid m’ satisfying subset sum exists for a given z(a’, t, 3, vr, 4, vs): each vector of the null space can be contributed to m’ one or zero times, and, together, the sum of the particular solutions and applicable vectors of the null space will take b1 and b2 (the non-zero and non-one values in the particular solution unique to the values of the first two elements and the chosen vr and vs) to 0 or 1. for any solution space for which this applies, the resulting m’ will be a vector of 0s and 1s encoding the membership of s. due to the form found as a result of convention of this method, the first four elements of a’ are referred to as the current window, the first two elements are the balance elements, and the second two elements are the pivot elements. the algorithm below attempts to reveal and exploit solution space properties which may exist universally among all sets for some window configuration if the set satisfies subset sum to form m’. if the found properties are, in fact, universal, the algorithm outlined is an exact, general-case algorithm for the subset sum problem. in order to determine, view, and exploit these properties, the algorithm utilizes a construct which will be called a directional contribution table. a directional contribution table is a tabulation of the contributions of the elements of the null space towards bringing the balance values of the particular solution towards 0 or 1. a directional contribution table d tabulates the contribution of a given vector in the null space of z towards paired balance targets and is defined with respect to the solution space (as expressed in the previously given convention) of a system s(z) and given target values of the balance points within the particular solution. 2 (11) for a given set a satisfying subset sum, there appears to exit a window configuration for which in z(a’, t, 3, vr, 4, vs) and d(s(z), t1, t2), t, vr, vs, t1, and t2 apply to an extant s, characterized by specific properties within d. the following (possibly non-exhaustive) property has been determined and is employed by the algorithm to determine m’ encoding s within a’: (a) if the length of s is 4, m’ is the exact solution of s(z) when the elements of the window are the elements of s and membership variables are set appropriately. if the length of s is 5, m’ is the exact solution of s(z) plus the vector representing the column of d for which d1,i and d2,i are both 1 when membership variables are set appropriately. in all other cases for a set a of length greater than four, m’ may be formed by taking the vectors represented by each column i for which the absolute value of d3,i is less than one. justification of properties: within the parameters of the convention listed above, manual algebraic reduction in the general case yields the following closed forms for the variables of the particular solution of s(z), the null space of s(z), and values within the directional contribution table: ( (17) (19) !cusj%|%volume%9%|%spring%2015% 29% 1 property (a) is to say that an instance of a subset s of a summing to c exists under the following constraints: (1) the length of s is four, and when the elements of s are set as the window of a’, m’ is found encoding s within a. assume all members of s are the current window of a’, and set membership to t1 = 1, t2 = 1, vr = 1, vs = 1. thus, the particular solution of s(z) for this configuration is encoding s and showing (1) to be true. the above also extends to show that (1) is true for any instance of s having length less than four for which all elements of s are within the window of a’ and membership is set appropriately. the length of s is 5, and when all but one element of s is set as the window of a’, m’ encoding s within a’ is found by adding to the particular solution the vector of the null space corresponding to d1,r = d2,r = 1. assume four members of s are the current window of a’, and set membership to t1 = 1, t2 = 1, vr = 1, vs = 1. 2 solving for the values in the directional contribution table: showing (2) to be true. the above may also be generalized for any s having length less than or equal to five for which all but one element of the subset exists within the window and membership is set appropriately. (3) in all other cases, s of length t >5 of a exists if and only if there exists a window configuration and membership assignment t1, t2, vr, vs such that there exists a set i, length t’ = t t1 t2 vr vs satisfying d3,ij <1 and σd1,ij= σd2,ij = 1 for all j in the range of the length of i. suppose that such a window exists. it is given that and using the closed forms derived earlier in this section: !cusj%|%volume%9%|%spring%2015% 30% 1 repeating the same process as was used in the justification of (1), m’ is derived encoding s within a’: thus, if such a window exists, a s of a’ of length t exists and is encoded by the derivable m’. suppose s of length t of a’ exists. let call window w = {a’1, a’2, a’3, a’4} and define membership of the window as follows: under this membership assignment, 2 likewise, thus, if s of a exists, σd2, ij = σd1,ij = 1. by (3), all d3,ij for the window must satisfy expanding the variables and simplifying, for δ1δ2 >0, and for δ1δ2 <0, it can be shown that the above is satisfied for some window configuration by showing that a tighter contained constraint is also satisfied for some window: summing the d3,ij, !cusj%|%volume%9%|%spring%2015% 31% 1 the numerator and denominator of this expression are both polynomial with respect to t’ (the numerator is of degree 1, and the denominator of degree 2); given that t’ >1, all coefficients of t’ in the numerator appear as a product in the denominator, and all non-t’ constants in the numerator appear in the denominator, the result is bounded by (-t’, t’). given that the above is bounded, viewing the second constraint, assert that thus meaning noting that it can be seen that a window configuration composed of the minima and/or maxima (and/or points whose values are set according to these) with membership appropriately set of si can be made among the set (or possibly created) to satisfy both these (and thus the primary) constraints. additionally, note that if an initial ordering of the set is enforced in which the set is ordered by the absolute value of its elements (increasing or decreasing), enumerating and swapping all possible windows of a in a pair-wise order (1 with 2, 1 with 3, . . ., 2 with 3; followed by pair 1 with pair 2 .. . . pair 2 with pair 3) creates such a set ordering for which a window may satisfy in the same window assignment as the previous constraint. this shows, then, that if s of length t of a exists, there exists a window configuration and membership assignment t1, t2, 2 vr, vs such that there exists a set i, length t’ = t t1 t2 vr vs satisfying d3,ij <1 and σd1,ij = σd2,ij = 1 for all j in the range of the length of i. algorithm & complexity: available upon request to cusj. discussion: a simple implementation of the algorithm was written in java using 64-bit long integers and tested for accuracy to reveal efficacy of the algorithm. to test and explore the algorithm’s performance, a driver was implemented which generates random sets of integers of a given length n, having range -2n2 to 2n2 and tests whether the set satisfies subset sum for c equal either to the sum of a randomly chosen subset or a random number unrelated to the set using (a) a conventional exponential algorithm and (b) the subsetsum routine for n permutations of the set. under the parameters of this test, failure occurs when the output of (b) differs from that of (a). for each n from 1 to 20, 1,000,000 such sets were generated and used to test the implementation of the algorithm. following all 20,000,000 trials, the success rate was 100%, exhibiting precisely 0 failures. to further explore general-case applicability of the algorithm, a reduction to subset sum from 3-sat, another np-complete problem, was implemented and tested, also exhibiting precisely 0 failures over all trials. the method can be reduced to an o(n4) approximation method by only using the subset of possible window configurations represented by shifting the entire set to the left (wrapping the element of the first index to the last position in the set) n times and repeating the constrain procedure. performing the same testing as was performed on the exact method as was outlined in section 2.2, yielding a success rate of 99.95% over 20,000,000 trials. the fact that such accuracy is yielded from a derived approximation which performs in the somewhat practical time of o(n4) gives reason to believe that such a method may call into question the reliability of the assumption that “practically many” naturally occurring instances of npcomplete problems cannot be solved in polynomial time. conclusion: an algorithm has been made that reduces the conditions under which a given set satisfies the stipulations of the subset sum proposition to a set of linear relationships, answering question of whether a set satisfies subset sum may be answered in a number of steps strongly polynomial with respect to the length of the input. following the justification, implementation and exploration of applications, as well as testing of this algorithm, a rate of accuracy and observed applicability was found that calls to question the reliability of assumption that “practically many” naturally occurring instances of np-complete problems cannot be solved in polynomial time. columbia research symposium the columbia undergraduate science journal takes great pride in having the honor of hosting the annual columbia spring research symposium. this year, the 2021 symposium was held virtually for the first time, and we are happy to have celebrated undergraduate research and retained the lively spirit of the event despite the unprecented circumstances. below are the winning presentations selected by our esteemed faculty judges! first place: "can we use next-generation gravitational wave detectors for terrestrial precision measurements of shapiro time delay?" by andrew sullivan abstract: shapiro time delay is an effect predicted by einstein's theory of general relativity whereby the travel time of light is delayed as light passes by massive objects. shapiro time delay is related to the parameterized post-newtonian formalism parameter γ, which quantifies spacetime curvature produced by a unit mass. consequently, the measurement of shapiro time delay can be used as a method of measuring the accuracy of the theory of general relativity. to date, all measurements of shapiro time delay have been conducted in space over astronomical scales. we propose an experiment that will allow shapiro time delay measurements to be conducted on earth, in which we use a rotating mass unit and a next-generation gravitational wave detector. with this scheme, we find that shapiro time delay and γ may be measured with sub-percent precision. this is the most precise scheme proposed for measuring shapiro time delay on earth to date. about the author: "andrew is a junior physics major at columbia university from yonkers, new york. andrew has performed research with columbia's experimental gravity group for the last two years and his research interests lie in the field of gravitational physics and gravitational wave astronomy. andrew hopes to obtain a phd in physics and become a professional researcher. for fun, andrew enjoys watching baseball and running." 62 second place: "neural oscillations as predictors of second language learning" by victoria ogunniyi the full body of this work can be found on page 39 of this issue of the columbia undergraduate science journal. about the author: "victoria ogunniyi is a third-year undergraduate student majoring in neuroscience with a minor in professional writing at the university of illinois at chicago. after college, she plans to apply to medical school and pursue a residency in psychiatry. in her free time, she enjoys improving her fictional writing skills and hopes to one day become a published novelist." 63 third place: "probing the statistical relationship between binary black hole mergers and active galactic nuclei hosts" by amanda beck about the author: "my name is amanda beck, and i am a brazilian junior at columbia university, columbia college, majoring in astrophysics. i am mainly interested in high energy astrophysics, and anything that deals with relativity and statistical analysis, as well as stem education, but am open to any field of research. i plan on obtaining a phd in astronomy or astrophysics, and engage in research and teaching. my favorite pastime is reading, specially fantasy or sci-fi!" abstract: since 2015, ligo/virgo has detected many binary black hole merger gravitational wave signals. identifying the origins of these is key to discover more about these mergers. rare host galaxies, like agn, present a favorable environment for these events due to the possible dynamical interactions in their accretion disks. in this project we will probe the statistical relationship between bbh mergers and agn hosts by analyzing the overlap in localization, as outlined in bartos et. al. 2017. to do that, we developed a python-based framework that can get the volume overlap between agn catalogs and ligo/virgo 90% probability density volume of bbh mergers. it can be used to establish the fraction of bbh gw detections that come from agn and to inform real-time em follow-up. fourth place: "the diagnosis of median arcuate ligament syndrome and postural orthostatic tachycardia syndrome and the effect upon the presentation of clinical depression" by jessica eddy abstract: median arcuate ligament syndrome is a rare, congenital condition where the diaphragm sits too low and the median arcuate ligament crushes the celiac artery. postural orthostatic tachycardia syndrome often presents as a co-occurring condition. the presentation of these conditions impacts the solar plexus, celiac ganglion, autonomic nervous system dysregulation, and neuropathy. my research was conducted in order to see if patients with median arcuate ligament syndrome and orthostatic intolerance as a comorbidity present with higher levels of clinical depression. furthermore, if the severity of the chronic condition increased, would clinical depression increase, as well? this research utilized methods such as peak systolic velocities from color duplex ultrasound technology, tilt table results to test for orthostatic intolerance, and the beck depression inventory-ii to seek to understand the connection between median arcuate ligament syndrome and orthostatic intolerance caused by dysregulation of the autonomic nervous system, which works in some level of conjugation with the solar plexus, sympathetic nervous system, dopamine and serotonin pathways, and dopamine receptor agonists and antagonists. about the author: "my name is jessica eddy. i am from grand rapids, michigan and am currently studying at the university of oxford. my current research includes the impact of rare vascular/gastrointestinal diseases and the role of endothelial cells in hiv latency. i plan to attend medical school for a joint md/phd and hope to research genetic biomarkers in rare vascular conditions. for fun, i love dancing and own my own dance studio, running with my puppy, and traveling the world!" 64 figure: first place poster from andrew sullivan 65 figure: second place poster from victoria ogunniyi figure: third place poster from amanda beck 66 figure: fourth place poster from jessica eddy 67 a letter from the editor-in-chief dear readers, the 2015–16 academic year saw many great developments at the columbia undergraduate science journal. most notably, our team brought the new columbia research scholars journal from concept to first issue. the mission of this new journal is to connect high school researchers from around the world, so that they can meet fellow scientists within their respective fields and share their results. this year, we were amazed to see a diverse array of high-quality research projects, from plant-microbial fuel cells to visual adaptation in schizophrenia. the final selection of papers includes work in biology, chemistry, environmental engineering, computer science, and physics. in this year’s cusj spring symposium, held on april 10th, we were excited to welcome many undergraduate researchers in the new york area as well as some of our most distinguished high school crsj contributors. for the first time, we held a poster contest designed to reward great communication skills and novel research. the awards of excellence this year went to justin cheung and andrew saydjari. this year’s edition of the cusj includes four papers from columbia university undergraduates in the areas of biomedicine and computational genomics. in the past, our policy has been to publish only research conducted by columbia undergraduates. in order for the cusj to grow, we have decided to review submissions from undergraduates anywhere in the world. the cusj, crsj, and our successful events this year were made possible by the dedicated and meticulous work of our editing team and ceo helen yang. we all felt privileged to examine so many insightful research projects, and our new editors gained significant experience in reviewing submissions for two different journals. i am confident that next year’s board will uphold our rigorous review standards and successfully take on new ventures in our mission to bring science to the public. –sean ballinger letter from the editor-in-chief dear readers, it is with great pleasure that the editorial board of the columbia undergraduate science journal presents the 15th edition of our publication. since 2006, cusj has aimed to create a space to highlight the accomplishments of undergraduate scientists and to foster the academic growth of students across all universities. to this end, this cycle saw a record number of submissions, and i am proud to witness the rapidly growing number of undergraduates gaining experience and accessing research opportunities. this edition of cusj features research and review articles that traverse disciplines in their scope. starting with our featured cover story on epithelial barrier integrity, this issue also explores topics from the microwave imaging of pipes to neural oscillations in second language proficiency. i am ecstatic to see the wide variety among these papers, and hope that you will find them as insightful as our editorial board did. coming from undergraduates working at universities located around the world, we hope that you appreciate the importance of their diverse voices. in a year where our organization had to move to a virtual format, producing an issue filled with outstanding research was more difficult than ever, and i would like to show my appreciation to everyone who was involved in the process. this includes but is not limited to the cusj editorial board, the cusj editorial committee, our faculty advisory board, our advisors at columbia libraries, our peer reviewers, research mentors, and all students who submitted their work. in particular, thank you to our published authors, who demonstrated an unwavering commitment to the quality of their papers and who were always a joy to work with despite the lengthy editing process they were subjected to. it has been an honor and a privilege to serve as the 2020-2021 editor-in-chief of cusj. i look forward to seeing the direction of the next editorial board and the future achievements of everyone involved. thank you for reading cusj! sincerely, isabella leite editor-in-chief columbia undergraduate science journal 3 changes in nr2b-nmda receptor expression relate to individual differences in adaptability to food restriction-stress ________________________________________________________________________________ cusj 2018 abstract — the eating disorder anorexia nervosa has an extremely high mortality rate of 5.9% (arcelus et al. 2011) and a poorly understood neurobiology. research indicates the animal model activity based anorexia (aba) may provide insight to the neural correlates of its behavioral and physical markers, hyperactivity and weight loss. previously, female aba rodents were shown to have increased levels of the nr2b subunit of nmda receptor (nmdar), an excitatory glutamate receptor, in stratum radiatum of dorsal hippocampal ca1. increased nr2b-nmdar levels may increase excitability in hippocampus, an anxiety-regulating brain region, possibly increasing anxiety –like behavior (chen et al. 2017) such as hyperactivity (wable et al. 2015). here we tested whether adolescent female mice exposed to a second food restriction (aba2) also demonstrate the behavioral and physical markers of aba. nr2bnmdars in aba2 and control mice were immunolabeled for electron microscopy and quantified at excitatory synapses of ca1 at various pre-, post-, and peri-synaptic regions. electron microscopy was chosen to selectively observe nr2b-immunoreactivity at distinct preand post-synaptic compartments because nmdars can operate differently depending on how closely they are anchored to the synaptic junction immediately apposed to presynaptic release site (hardingham and bading 2010). we found that aba2 animals had more nr2b-labeling in presynaptic terminals and postsynaptic cytoplasm than controls. increased nr2blabeling at postsynaptic density correlated with increased running activity during food access and weight loss, suggesting that nr2b-nmdars at postsynaptic density may mediate vulnerability to aba2. increased levels of nr2b-labeling at extra-synaptic membrane of dendritic spines correlated with less running during aba2, suggesting nr2b at this site may mediate aba resilience. the results indicate localized nr2b expression correlates with adaptability to second exposure to food restriction and support a neurobiological basis for aba. i. introduction anorexia nervosa (an) is a psychiatric eating disorder that can be characterized by restriction of food intake, extreme weight loss, distorted body image and excessive exercise (national eating disorders association 2016). exercise-based hyperactivity, which is evident in 4080% of an patients is not only unhealthy, but also compulsive and compensatory for calorie intake (holland et al. 2014), thereby maintaining the disorder (holtkamp et al. 2004). etiology ranges from socio-cultural factors, such as the media’s construction of the ideal body type, to trauma, personality traits and other mental health conditions (bemis 1978, schmidt et al. 1997); for instance, symptoms of anxiety and restricted food intake may actively contribute to hyperactivity and worsen symptoms of an (holtkamp et al. 2004). familial and twin genetic studies have also revealed susceptible chromosomal loci and candidate genes involved in an pathophysiology, providing evidence for both a biological and experiential bases to the disorder (pinheiro et al. 2009). the disorder typically manifests amongst the adolescent female population and has the highest mortality rate of any psychiatric disease (arcelus et al. 2011). treatment commonly consists of psychotherapy or counseling but can be challenging and with poor outcomes, since about 41% relapse within a year of recovery (carter et al. 2012). medications such as antidepressants and antipsychotics may be prescribed but improve symptoms by targeting pathways involved with depression and bipolar disorder rather than an itself (american psychiatric association). understanding the neurobiological causes of an and how they may contribute to an pathogenesis and progression is therefore critical for developing an alternate route for effective and targeted therapy for an patients. the neural correlates of anorexia nervosa may be explored using activity-based anorexia (aba), the animal model for the disease (chowdhury et al. 2015). built upon the physiological premise that experiencing hunger induces foraging behavior, the model manipulates rodent’s food restriction and access to a running wheel to induce reduced (voluntary) food intake, hyperactivity, and extreme weight loss (chen et al. 2017). for rodents that demonstrate aba, implementing food restriction increases voluntary wheel running such that the rodent chooses to run even during times of food availability and a hungered state (chen et al. 2017). what is intriguing about the model is that rodents experience behavioral symptoms such as hyperactivity in the absence of exposure to the sociocultural factors often ascribed to human etiology of the an, suggesting that a neurobiological change may be sufficient for exacerbating the behavioral symptoms in rodents that are analogous to those of an. invoking aba in rodents thus provides a route to explore vulnerability to an. age, weight, and sex, for instance, are factors established to affect the extent of wheel running and weight loss in aba. animals who are older, heavier, and male tend to demonstrate less hyperactivity and weight loss than younger, lighter or female counterpartstrends that parallel manifestation of an in patients (boakes and dwyer 1997, pare et al. 1978), who are typically adolescent females. experimental evidence reveals distinct neural changes that occur in conjunction with aba to explain these trends. the hippocampus is a highly plastic region that has not yet reached maturity during adolescence in rodents; for instance, the dentate gyrus (dg), which is part of the hippocampal formation dorsal to the ventral ca1 region, changes in nr2b-nmda receptor expression relate to individual differences in adaptability to food restriction-stress danielle mendonca1*, yi-wen chen, katherine varela, chiye aoki1 1 1center for neural science, new york university, 4 washington place, room 809, new york, ny, 10003, usa. cusj 2018 ________________________________________________________________________________ is still undergoing neurogenesis, migration and maturation during puberty and adolescence (stagni et al. 2015). the ongoing development of the hippocampus during this time makes it a particularly vulnerable target for experience-dependent plasticity in response to environmental stressors. coincidentally, adolescent female rats that underwent acute induction of the aba paradigm demonstrated suppressed non-neuronal cell proliferation in the dg and surrounding areas (barbarich-marsteller et al. 2013). while the direct consequence of this abnormality has yet to be discovered, it is well known that normal development of the hippocampus is important for learning and memory formation. however, the hippocampal formation is especially relevant for aba experiments because it regulates stress and anxiety (shen et al. 2007), a condition symptomatic or comorbid with an (kaye et al. 2004). maladaptive alterations to the hippocampus can therefore be detrimental to stress and anxiety management, contributing to the behavioral symptoms of aba. in response to stress, there are elevated levels of glutamate, the excitatory neurotransmitter, in the hippocampus (kazi and oomenn 2014). when glutamate binds nmda receptors on the post-synaptic membrane, and if the membrane potential is sufficiently depolarized, there is an influx of calcium into the cell that activates intracellular signaling cascades that may activate receptor trafficking pathways and exocytotic events. higher concentrations of glutamate mediate increased excitability in the hippocampus, which strongly correlates with stressinduced anxiety (shen et al. 2007). for animals that are vulnerable to aba, there is a significantly higher concentration of nmdars within regions of dorsal hippocampus (chen et al. 2017), in addition to an altered dendritic arborization of ca1 pyramidal cells in dorsal hippocampus (chowdhury et al. 2014). the experience-dependent changes in ca1 spines in response to aba can therefore be used to study vulnerability to aba. the dendrites of ca1 branch together within stratum radiatum (sr), receiving inputs from ca3 schaffer collaterals and entorhinal cortex (ec), and sending outputs to the ec layer v and subiculum (olypher, lytton and prinz 2012). for adolescent female mice that have undergone one round of aba (aba1), there is a marked increase in levels of the nr2b subunit of the nmda receptor in ca1, thereby mediating increased excitability and anxiogenesis within hippocampus (chen et al. 2017). functional nmdars are comprised of four subunits: two subunits of nr1 and two subunits of nr2a and/or nr2b. this experiment implemented a second exposure to food restriction to the wheel-acclimated mice after seven days of recovery from aba1. although both nr2a and nr2b are nmda subunits of interest, only the relationship between nr2b levels and aba2 behaviors was considered here. rodents who are exposed to food restriction once demonstrate the behavioral and physical markers of aba. here we test whether or not adult female mice with a second exposure to food restriction also demonstrate the behavioral and physical markers of aba and have a redistribution of nr2b-nmda at hippocampal ca1. using immunocytochemistry and post-embed gold (peg) labeling for electron microscopy, nr2b-nmda receptors were visualized at axo-spinous synapses within sr of dorsal ca1 of aba2 animals and age-matched controls. electron microscopy was chosen for visualization of the peg-labeled nr2bs because it allows visualization of both presynaptic and post-synaptic regions of the ca1 junction and differentiation between axospinous excitatory synapses verses axo-dendritic shaft inhibitory synapses. because the location of the nr2b containing nmdar dictates its function (hardingham and bading 2010, petralia et al. 2010), it was important to quantify peg-particles across various synaptic regions; the subcellular synaptic compartments analyzed included at the postsynaptic density, near the postsynaptic density, postsynaptic cytoplasm, post-extrasynaptic membrane, at/near presynaptic membrane, presynaptic cytoplasm, preextrasynaptic membrane and the synaptic cleft. the hypothesis was that neural circuitry modifications in the hippocampus do not get restored after aba1, and there are therefore increased levels of nr2b-nmda in ca1 pyramidal cells during and after aba2. higher levels of nr2b-nmda can mediate increased excitability in the hippocampus, leading to increased anxiety-like behavior and increased hyperactivity, mediating vulnerability to aba. we found that aba2 animals demonstrated significantly higher levels of nr2b-immunoreactivity within the postsynaptic cytoplasm, presynaptic cytoplasm and pre-extrasynaptic membrane than agematched controls. aba2 animals also demonstrated individual differences in nr2b levels that correlated to adaptability to aba2, such as extent of wheel running activity and weight loss. this provides evidence that a second exposure to food restriction also correlates with the behavioral and physical markers of aba. ii. methods animals mice used for this experiment were handled according to the institutional animal care and use committee of new york university (animal welfare assurance no. a311701). there were a total of 15 mice used in the experiment, 8 in the aba condition and 7 age-matched controls. the control animals did not experience any food restriction. all mice were female, bred and housed within the nyu animal facility, and with a c57bl/6j strain background (see chowdhury et al. 2013, chen et al. 2016). the aba1 and aba2 paradigm is as described in chowdhury et al. (2013), except we used 2 hours of food restriction from 7pm-9pm, during the first two hours of the dark cycle. the aba paradigm begins with a baseline period starting at p36, when animals are acclimated to the running wheel in individual cages and have normal access to food (see fig. 1a). starting at p41, the animals are subject to aba. this involves continuous access to the running wheel in conjunction with food restriction. animals in aba only have access to food during the hours of 7-9pm, during the first two hours of the dark cycle (see fig. 1b). the animals’ dark cycle begins at 7pm and terminates at 7am, during which time the lights are kept off. at the end of each food access period, animals were moved to a new cage and the amount of food consumed and the wheel count recorded. wheel counts were 2 ________________________________________________________________________________ cusj 2018 recorded at these intervals: 1am-7am (during the dark cycle), 7am-1pm, 1pm-7pm (faa), 7pm-9pm (food access) and 9pm-1am. this allowed for comparison of wheel activity during distinct periods of the day and a summation of wheel counts to calculate total activity during aba. after completion of aba1, animals were given a recovery period until p51 in order to restore a healthy body weight (see fig. 1a). from p51-p55, animals underwent a second baseline period with continuous running wheel access to reacclimate to the wheel. aba2 began on p55 through p59 and followed the same protocol as for aba1. the paradigm was culminated with euthanasia by perfusion on p60. wheel activity and weight were monitored throughout this time. figure 1: schematics of the aba paradigm. (a) timeline outlining sequence of events for aba2 mice from birth (p0) to perfusion (p60). during aba1 and aba2, mice had continuous access to a running wheel but limited access to food. (b) typical day of aba: mice received food only from 7pm-9pm, during the first two hours of the dark cycle that continues until 7am. wheel running is monitored throughout aba; food anticipatory activity is seen and measured from 1pm to 7pm. brain tissue preparation aba animals and age-matched controls were euthanized on p60 (after the second aba), using a urethane based anesthesia (i.p. 34%, 0.15ml/20g), at a time between 12pm and 4pm (chen et al. 2016). animals were then transcardially perfused using a buffer containing 4% paraformaldehyde (pfa). fixation with glutaraldehyde was not conducted until post-embed tissue processing for the purpose of antigen retention (chen et al., 2016). after extraction, brains were post-fixed in a pfa fixative for 9h, and sectioned on a coronal plane using a vibratome (40!m). tissue sections were stored at 4°c in 0.01m phosphate buffer, 0.9% saline (buffer ph = 7.5) and 0.05% sodium azide (pbs-azide). sections containing dorsal hippocampus were then selected for post-embed processing. antibodies primary antibodies used in the experiment are previously described (aoki et al. 2009a, b, and chen et al. 2017). the antibody used against nr2b (cat. #06-600, millipore (previously available from upstate biotechnology)) was developed in rabbits and targets the last 20 amino acids of the rodent nr2b c-terminus. antibody specificity was confirmed with western blots on rat brain microsome, which revealed a single band at ~180kd corresponding to the molecular weight of nr2b (chen et al. 2017) and no indication of cross-reactivity with nr1 or nr2a subunits of the nmdar (rinaldi et al. 2007). the aba2 and con tissue was also labeled with anti-nr2a primary antibody for a portion of the experiment that will be considered in the near future. the secondary antibody used to detect the primary nr2b antibody was anti-rabbit igg, produced in donkey, conjugated to 10nm colloidal gold particles (cat. #25705, electron microscopic sciences). immunocytochemistry procedures for targeting nr2b using immunocytochemistry are as previously described in aoki et al. 2000, 2003, 2009a, 2009b, fujisawa and aoki 2003). processing for immunolabeling was done on two vibratome sections (randomly selected) per animal that were post fixed with 2% glutaraldehyde in pbs for 10 mins. these sections were then processed using the osmium-free procedure for electron microscopy and embedded in plastic. the plastic-embedded blocks were then ultrathin sectioned, producing five em grids per section. two of the grids were used for nr2b labelling, two for nr2a, and one as a no primary antibody control. post-embed immunogold labelling was conducted as described previously (aoki et al. 2009b). the grids were incubated with anti-nr2b (10ug/ml, millipore) overnight and anti-rabbit igg secondary antibody (1:100, ems) for 1 hour at room temperature. electron microscopy visualization and imaging of immuno-labeled sections was conducted using a jeol 1200xl transmission electron microscope and amt xr80 camera system and software (boston, ma).�typically, one grid, randomly picked from pairs of immunolabelled grids, was selected per animal for imaging. the grids contained anatomical markers including ca3, dentate gyrus and hippocampal fissure that were used to locate the ca1 region. a second grid was only used in cases when the tissue was torn over the region of interest, making it challenging to capture good images for quantitative analysis. all images captured were within sr of ca1 in dorsal hippocampus, and the synapses accounted for were asymmetric axo-spinous, identified using a thick post-synaptic density opposite to a presynaptic terminal containing vesicles. the identity of the animals, either aba2 or control, was not revealed until after image capturing and peg quantification to prevent bias from seeking higher or lower levels of nr2bimmunoreactivity during image capturing. images were analyzed for peg-particle quantification using imagej (nih). 28 figures figure 1. schematics of the aba paradigm. a timeline outlining sequence of events for aba2 mice from birth (p0) to perfusion (p60). during aba1 and aba2, mice had continuous access to a running wheel but limited access to food. b typical day of aba: mice received food only from 7pm-9pm, during the first two hours of the dark cycle that continues until 7am. wheel running is monitored throughout aba; food anticipatory activity is seen and measured from 1pm to 9pm. p0 p25 p36 p41 p51 p55 p60 p44 aba1 aba2 birth wean baseline 1 (acclimation) baseline 2 (reacclimation) recovery aba paradigm schematic +2hr food +2hr food perfusion running activity schematic 1am 1pm 7am 7pm 9pm food access food anticipatory activity (faa) 24 hr aba cycle lights off (dark cycle) a b 3 cusj 2018 ________________________________________________________________________________ peg-particle analysis and statistics because each nr2b-immunoreactivity was analyzed by quantifying peg particles at distinct subcellular compartments (see fig. 2a, b) pre-extrasynaptic, precytoplasmic, at/near presynaptic membrane, cleft, at/near postsynaptic density, post-extrasynaptic, post cytoplasmicat axo-spinous synapses in sr of dorsal ca1. peg particles that were clustered in groups of 2 or more particles that were less than 20nm apart were quantified as a single nr2b label. peg particle analysis was conducted in two ways: using animals as independent units and using synapses as independent units. figure 2: electron micrograph images from aba tissue to demonstrate categorization of immuno-labeled nr2b subunits at distinct axo-somatic synaptic compartments. numerically annotated images contains a bar to scale to 200nm. (a) a presynaptic axon terminal is indicated by “a.t.” and postsynaptic dendritic spine by “d.s.”. nr2b-immunoreactivity is observed through peg immunolabelling, as indicating by numbers 1-9, categorized accordingly: (b) 1 at postsynaptic density, 2 in cleft, 3 in postsynaptic cytoplasm, (c) 4 near postsynaptic density, 5 and (d) 6 at pre-extrasynaptic (pre-es) membrane, 7 at presynaptic cytoplasmthe cluster of two pegs at 7 was counted as a single labeled nr2b subunit, since the pegs were < 20nm apart(e) 8 at post-extrasynaptic (post-es) membrane, (f) 9 at presynaptic membrane. (g) schematic to show how pegs were categorized at distinct locations within excitatory synapses. using synapses as independent units compares group differences in nr2bimmunoreactivity between the aba2 group and the control group using mann-whitney u tests (see fig. 3). approximately 200 synapses were analyzed per animal, and sorted into groups of 10 (per-10-synapses) to assess for the level of peg particles and proportion of nr2b-immunoreactivity present across the subcellular compartments per tissue. the sums of the pegs per 10 and per animal (200 synapses) or sums of synapses immunolabeled per 10 and per animal were then pooled together. nr2b-immunoreactivity at each subcellular synaptic site in aba2 and control animal tissue was compared using mann-whitney u tests for nonparametric data because the data was not normally distributed. using each animal as an independent unit compares how individual differences in nr2bimmunoreactivity correlate with wheel running activity and change in body weight (see figs. 4, 5). the per-10-synapses sorting was repeated 20 times to obtain a sum of the pegs or immunolabeled synapses per 200 synapses. the sum of the pegs per animal (or 200 synapses) or sum of synapses immunolabeled was then used to determine significant individual differences in aba2 tissue immunolabeling and behavioral symptoms of aba. nr2b-immunoreactivity was related to wheel running activity and weight loss by conducting a pearson’s correlation analysis on prism (la jolla, calif., versions 7.0). the p value used to determine statistical significance was p< 0.05. iii. results aba2 evokes a distribution of nr2b in sr of hippocampal ca1 that is different from controls and aba1 animals animals that experienced a second aba (aba2) demonstrated a distribution of nr2b-pegs that was different from age-matched control animals. nr2b detection by peg immuno-labeling revealed that aba2 animals had a greater proportion of synapses with peglabeled nr2b in the pre-synaptic cytoplasm (mannwhitney u (mwu)= 6515, p= <.0001, fig. 3a) and postsynaptic cytoplasm (mwu= 8749, p= 0.003, fig. 3a) relative to age-matched controls. aba2 animals also had higher levels of nr2bimmunoreactivity, as quantified by the number of peg particles, within the postsynaptic cytoplasm (mwu= 9091,p=0.0023, fig. 3b), presynaptic cytoplasm (mwu= 6900, p= <0.0001, fig. 3b), and preextrasynaptic membrane (mwu= 9716, p= 0.0255, fig. 3b). while aba2 and control animals did not demonstrate a significant difference in nr2bimmunoreactivity at other subcellular presynaptic sites, there was still an overall 29 figure 2. electron micrograph images from aba tissue to demonstrate categorization of immunolabeled nr2b subunits at distinct axo-somatic synaptic compartments. numerically annotated images contains a bar to scale to 200nm. (a) a presynaptic axon terminal is indicated by “a.t.” and g a b c d e f 4 ________________________________________________________________________________ cusj 2018 increase in nr2b-pegs within presynaptic terminals (mwu = 6821, p= <0.0001, fig. 3b) compared to controls. figure 3: the distribution of nr2b in sr of hippocampal ca1 is different in aba2 tissue compared to control tissue. mannwhitney u test for nonparametric data was conducted to determine significant differences in nr2b-immunoreactivity. significant differences are marked with an asterisk (*). (a) group comparisons in proportion of synapses labeled with nr2b using synapses as independent units. (b) group comparisons in levels of nr2b-pegs using synapses as independent units. (c) group comparisons in proportion of synapses labeled with nr2b using animals as independent units. (d) group comparisons in levels of nr2b-pegs using animals as independent units. when nr2b-peg levels were compared using each animal as an independent unit, a significant difference in immunoreactivity relative to controls was only seen in the presynaptic cytoplasm, since aba2 animals had a greater proportion of terminals labeled (mwu= 0, p= 0.003, fig. 3c) and more abundant nr2b-immunoreactivity (mwu = 1, p= 0.0006, fig. 3d) within this region. although nr2b labeling anywhere in the presynaptic spine was significantly greater (mwu= 5, p= 0.0059, fig. 3d) between aba2 animals and controls, the animals did not exhibit a significant differences in nr2b distribution anywhere in the postsynaptic spine. these results additionally indicate that nr2b distribution for aba2 animals was different from aba1 animals’. previous findings conducted by chen et al. 2017 that considered how the aba induction affects nr2b distribution in sr of hippocampal ca1 found significantly greater nr2b-immunoreactivity at both the presynaptic axon terminals and postsynaptic spines, relative to control animals’ (see chen et al. 2017 fig 3b). the new findings suggest that experiencing aba2 results in a loss of the upregulated postsynaptic nr2b present during aba1, since there is no consistent significant difference between aba2 and control animals within postsynaptic regions (fig. 3c, 3d). increased nr2b-immunoreactivity at the postsynaptic density correlates with individual differences in running activity during food access and increased weight loss after aba2 running during food access is an indication of hyperactivity and self-starvation in aba because the animal is choosing to exercise rather than eat its food. aba2 animals demonstrated individual differences in running activity during food access. for animals that ran more during the feeding hours of the last three days of aba2, there was a higher number of nr2b-pegs at the postsynaptic density (psd) (r= 0.902, p = 0.005, fig. 4a). interestingly, there was no significant correlation between nr2b-immunoreactivity at the postsynaptic density and running activity within the reacclimation period, which occurred prior to aba2 (data not shown). this may suggest the role of nr2b at the psd involved not in general locomotion, but instead when there is a conflict between an innate urge to forage (physical activity) versus feed. if upregulated nr2b at the psd has a role in exacerbating physical activity during aba2, specifically during times of food availability, animals with more nr2b at the psd should also demonstrate increased weight loss after aba2 due to their voluntary food restriction. the aba2 animals’ weight loss (%) at perfusion was used to indicate the weight lost after aba2 relative to the animals’ weight before beginning aba2. indeed, individual differences in aba2 animals’ weight loss (%) at perfusion correlated positively and significantly with nr2b-pegs at the psd (r= -0.89, p= 0.018, fig. 4b). no other behavioral data during aba2 30 postsynaptic dendritic spine by “d.s.”. nr2b-immunoreactivity is observed through peg immunolabelling, as indicating by numbers 1-9, categorized accordingly: (b) 1 at postsynaptic density, 2 in cleft, 3 in postsynaptic cytoplasm, (c) 4 near postsynaptic density, 5 and (d) 6 at preextrasynaptic (pre-es) membrane, 7 at presynaptic cytoplasmthe cluster of two pegs at 7 was counted as a single labeled nr2b subunit, since the pegs were < 20nm apart(e) 8 at postextrasynaptic (post-es) membrane, (f) 9 at presynaptic membrane. (g) schematic to show how pegs were categorized at distinct locations within excitatory synapses. a preat prene ar preex tra sy na pti c precy top las mic clef t pos t-a t pos t-n ea r pos t-e xtr as yn ap tic pos t-c yto pla sm syn ap tic pr e syn ap tic po st syn ap tic any whe re pre any whe re po st any whe re pre po st extr as yn pr e p os t 0 2 4 6 8 peg count n um be r o f p e g -p ar tic le s pe r 1 0 sy na ps es con aba * * * * *b 31 figure 3. the distribution of nr2b in sr of hippocampal ca1 is different in aba2 tissue compared to control tissue. mann-whitney u test for nonparametric data was conducted to determine significant differences in nr2b-immunoreactivity. significant differences are marked with an asterisk (*). a group comparisons in proportion of synapses labeled with nr2b using synapses as independent units. b group comparisons in levels of nr2b-pegs using synapses as independent units. c group comparisons in proportion of synapses labeled with nr2b using animals as independent units. d group comparisons in levels of nr2b-pegs using animals as independent units. preat prene ar preex tra sy na pti c precy top las mic clef t pos t-a t pos t-n ea r pos t-e xtr as yn ap tic pos t-c yto pla sm syn ap tic pr e syn ap tic po st syn ap tic any whe re pre any whe re po st any whe re pre po st extr as yn pr e p os t 0 50 100 150 synapse count per animal n um be r o f s yn ap se la be le d by p e g p er 2 00 s yn ap se s con aba * * * c preat prene ar preex tra sy na pti c precy top las mic clef t pos t-a t pos t-n ea r pos t-e xtr as yn ap tic pos t-c yto pla sm syn ap tic pr e syn ap tic po st syn ap tic any whe re pre any whe re po st any whe re pre po st extr as yn pr e p os t 0 50 100 150 200 peg count-animal n um be r o f p e g la be le d pe r 2 00 s yn ap se s con aba * * * d 5 cusj 2018 ________________________________________________________________________________ correlated significantly with nr2b-pegs at the psd (data not shown), suggesting nr2b at the psd may independently increase susceptibility to aba2 by increasing anxiety, which promotes physical activity during food access, thereby increasing weight loss. figure 4: pearson correlation results for nr2bimmunoreactivity at the postsynaptic density (psd) and individual differences in aba vulnerability. (a) running during food access, measured from 7pm9pm (the hours of food availability), during the last three days of aba2 correlated significantly and positively with nr2b at the psd. (b) weight loss at perfusion, relative to the animals weight before aba2, was greater in animals that had increased nr2b at the psd, indicating increased vulnerability to aba with increased nr2b at the psd. black triangles indicate data from aba2 tissue; grey triangles indicate data from age-matched controls (con) tissue. pearson correlation r values are marked with two asterisks (**) to indicate p<0.01. post-extrasynaptic nr2b-immunoreactivity in relationship to running activity during aba2 individual differences in running activity over the course of aba2 (km) correlated negatively with levels of nr2b-pegs at the post-extrasynaptic (post-es) plasma membrane: animals that demonstrated increased hyperactivity had a smaller proportion of synapses containing nr2b-immunoreactivity at the post-es membrane (r= -0.88, p=0.0088, fig. 5a) and fewer nr2bpegs at the post-es membrane (r= -0.87, p= 0.011, fig. 5b). this suggests post-es nr2b may have a role in mediating aba resilience, since animals with greater nr2b levels in this region demonstrate less hyperactivity over the course of aba2. figure 5: pearson correlation results for nr2bimmunoreactivity at the post-extrasynaptic (post-es) membrane and individual differences in aba2 vulnerability. (a) a greater proportion of synapses were labeled with nr2b-pegs at the post-es membrane and (b) a higher level of post-es nr2bpegs were present in aba2 animals that ran less over the course of the aba2. (c) nonfood anticipatory activity and (d) food anticipatory activity during aba2 correlated negatively with post-es nr2b-immunoreactivity. black triangles indicate data from aba2 tissue; grey triangles indicate data from agematched controls (con) tissue. pearson correlation r values are marked with one asterisk to indicate 0.05>p>0.01 but two asterisks (**) to indicate p<0.01. although post-es nr2b-immunoreactivity did not correlate with running activity during baseline, aba1, or the reacclimation period (data not shown), there was a significant correlation between post-es nr2b-pegs and food anticipatory activity (faa, 1pm-7pm, preceding the feeding hours) during aba2 (r=-0.77, p=0.04, fig 5c) as well as non-faa activity (7pm-1pm) during aba2 (r=0.79, p=0.03, fig. 5d). food anticipation during aba typically results in an innate increase in wheel running behavior (mistlberger 1994). the findings suggest higher levels of post-es nr2b levels make aba animals less susceptible to hyperactivity during aba2, even when anticipating food. post-es nr2b may therefore have a role in protecting animals form aba2-induced hyperactivity. however, post-es nr2b pegs did not correlate with running activity during times of food availability, nor did it correlate with weight loss (data not shown). iv. discussion functional significance of increased presynaptic nr2bimmunoreactivity in aba2 animals compared to controls presynaptic receptors are known to have a role in enhancing synaptic efficacy by modulating neurotransmitter release through cascadic signaling pathways (diguid and smart, 2009). presynaptic nmdars (prenmdars) are not classical hebbian receptors because their activation via glutamate binding and depolarization arises presynaptically (bidoret 2009, dore et al. 2017) instead of postsynaptically. yang et al. (2011) found that entorhinal cortical nr2b-nmda presynaptic autoreceptors had a role in tonically facilitating glutamate release in chronically epileptic rats (yang et al., 2006), suggesting that enhancement of glutamate release from presynaptic terminals can be modulated by nr2b-nmda autoreceptors. additionally, dore et al. (2017) outlined the prenmdar’s roles in both evoked and spontaneous neurotransmitter release (berretta and jones 1996, sjöström et al., 2013), with and without calcium+dependency (kunz et al., 2013). if hippocampal prenr2bnmdars function similarly, in that they modulate both tonic and evoked glutamate release, the concentration of glutamate would be elevated within the synaptic cleft, exciting adjacent neurons. the level of nr2b-nmdars within presynaptic terminals in sr of hippocampal ca1 of aba2 animals was significantly higher than basal levels in control animals. increased nr2b-nmdars may thus function to enhance vesicular release of glutamate, and contribute to increased 32 figure 4. pearson correlation results for nr2b-immunoreactivity at the postsynaptic density (psd) and individual differences in aba vulnerability. a running during food access, measured from 7pm9pm (the hours of food availability), during the last three days of aba2 correlated significantly and positively with nr2b at the psd. b weight loss at perfusion, relative to the animals weight before aba2, was greater in animals that had increased nr2b at the psd, indicating increased vulnerability to aba with increased nr2b at the psd. black triangles indicate data from aba2 tissue; grey triangles indicate data from age-matched controls (con) tissue. pearson correlation r values are marked with two asterisks (**) to indicate p<0.01. 0 1 2 3 4 5 6 0 10 20 30 running during food access last 3 days of relapse (km/2hr) n r 2b a t p s d pe r 2 00 s yn ap se s conpost-at post-at con r = -0.902** a b -30 -25 -20 -15 -10 0 10 20 30 weight loss at perfusion (%) n r 2b a t p sd pe r 2 00 s yn ap se s con-post-at post-at con r = -0.89** 33 figure 5. pearson correlation results for nr2b-immunoreactivity at the post-extrasynaptic membrane and individual differences in aba2 vulnerability. a a greater proportion of synapses were labeled with nr2b-pegs at the post-es membrane and b a higher level of post-es nr2bpegs were present in aba2 animals that ran less over the course of the aba2. c nonfood anticipatory activity and d food anticipatory activity during aba2 correlated negatively with post-es nr2b-immunoreactivity. black triangles indicate data from aba2 tissue; grey triangles indicate data from age-matched controls (con) tissue. pearson correlation r values are marked with one asterisk to indicate 0.05>p>0.01 but two asterisks (**) to indicate p<0.01. 0 10 20 30 0 5 10 15 20 running during relapse (km) n r 2b -la be lle d po st -e s m em br an e pe r 2 00 s yn ap se s con-post-extrasynaptic post-extrasynaptic con r = -0.88** 0 10 20 30 0 5 10 15 20 running during relapse (km) po st -e s nr 2b p er 2 00 s yn ap se s con-post-extrasynaptic post-extrasynaptic con r = -0.87* 0 1 2 3 0 10 20 30 faa vs post-es peg faa in relapse (km/ 6hr) p os t-e s n r 2b p er 2 00 s yn ap se s con-post-es post-extrasynaptic con r = -0.77* 0 10 20 0 5 10 15 20 non faa vs post-es peg non-faa in relapse (km) po st -e s n r 2b p er 2 00 s yn ap se s con-post-es post-extrasynaptic con r = -0.79* d c a b 6 ________________________________________________________________________________ cusj 2018 excitability in the hippocampus that may be occuring in conjunction with aba2. interestingly, there were limited correlations between pre-nr2b and individual differences in aba2 vulnerability that could be made. this suggests that while aba tissue had significantly higher pre-nr2bimmunoreactivity compared to controls, pre-nr2b nmdas does not have a large role in role in the two measurements we made of aba vulnerability – wheel running and weight loss. there may be other measurements which we did not make, such as corticosteroid levels, with which prenr2bimmunoreactivity correlates. functional significance of increased nr2b-immunoreactivity at the postsynaptic density correlating with increased running activity during food access a higher concentration of nr2b-nmda receptors receiving glutamate at the postsynaptic density correlated with increased running during food access during aba2. increased running during food access reduced food intake during the two hours of food availability, directly increasing weight loss, which also correlated positively with increased nr2bimmunoreactivity at the psd. these individual differences in running during food access in aba2 mice are a trademark behavior of aba: the amount of energy metabolized for exercise is greater than the animal’s caloric intake, ultimately leading to starvation and death. while the mechanism underlying hyperactivity despite low energy levels has yet to be determined (kim 2012), hyperactivity may also be an indication of increased stress, and stress induced anxiety (shen et al. 2007). in the absence of aba, sibold et al. (2011) found that stress significantly increased wheel-running activity in mice already acclimated to the wheel. there is evidence to suggest hyperactivity in mice serves as an anxiolytic to relieve stress, a behavior that can be paralleled to human cases of anorexia nervosa, where patients who exercise excessively also demonstrate higher levels of anxiety and report exercise to be anxiolytic (peñas-lledó et al. 2002). while contradictory evidence suggests that hyperactivity in rodents may serve to increase anxiety instead (garrett et al. 2012), the correlation between mice in the aba-paradigm and increased anxiety relative to controls is consistent, since aba mice have been shown to demonstrate increased anxiety-like behaviors in the elevated plus maze after food restriction (wable et al. 2015). functional nr2b-nmdars at the post-synaptic density can increase rate of cell firing, not only increasing excitability but also increasing local concentrations of glutamate to sustain excitability. this can sustain anxiety-like behaviors in mice and contribute to the hyperactive behaviors demonstrated by aba and aba2 mice, which engage in highenergy-expending exercise despite having low-energy levels due to low caloric (food) intake. future studies involving the aba paradigm should examine the role between nr2b at the postsynaptic density and increased ca1 cell excitability in aba or aba2 mice that demonstrate increased anxiety and hyperactivity. post-extrasynaptic nr2b-nmda as a facilitator of aba resilience aba2 animals demonstrated increased nr2b-nmda at the post-extrasynaptic membrane that correlated with decreased running during aba2, including reducing food anticipatory activity (faa) and non-faa wheel running. this suggests post extrasynaptic nr2b (post-es-nr2b) may provide resilience to aba2. whereas post-es-nmda has been shown to promote neuronal apoptosis in cultured neurons (hardingham and bading 2010), chen et al. (2017) suggest that post-es-nmda role may have more to do with synaptic, spine and dendritic pruning as evident in adolescent animals that demonstrate voluntary exercise (chowdhury et al 2014). post-es-nr2b activation may also evoke long-term depression (ltd) in hippocampal ca1 (liu et al 2013), suggesting post-esnr2bs may play a role in reducing excitability within the post-synaptic neuron. if upregulated post-es-nr2b functions to depress synaptic efficacy via ltd and reduce ca1 excitability, this explains how post-es-nr2b can protect aba-vulnerable animals from exercising excessively during aba2 and also suppress overall activity, as seen with reduced faa in aba2 animals that have increased post-es nr2b. interestingly, the aba2 findings for post-synaptic nr2b-immunoreactivity differ from those found after only one aba. chen et al. (2017) noted no significant correlation between post-extrasynaptic nr2b levels and weight loss (see table 4 of chen et al. 2017). this can suggest either a downregulation of nr2b from the post-es membrane or a redistribution of nr2b from the post-es membrane to another postsynaptic site that only occurs after the experience-dependent changes facilitated by aba1. immunodetection of nr2b-nmda is not an indication of functionality. in the absence of electro-physiological experiments on neural tissue from aba2 animals, the ratio of functional to non-functional nr2b-nmdars, and the true change in excitability of the hippocampus due to an increase in nr2b-nmda, are not known. while the possibility of non-functional nr2b-nmda present at cytoplasmic and extrasynaptic sites cannot be ruled out, there is still evidence for an upregulation, or downregulation, of nr2b at these sites. moreover, while some differences between nr2b-immunoreactivity in aba1 (from chen et al. 2017) and aba2 were noted, it is important to consider that the aba1 animals were perfused at a younger age of p44 than aba2 animals that were perfused at p60. this means that age is likely to have also contributed to differences in nr2b distribution between the two aba groups, which can be determined by doing an age-matched aba1 paradigm. v. conclusions animals that experience two episodes of food restriction in the presence of a running wheel exhibited increased nr2b-immunoreactivity in ca1 neurites that positively correlated with activity-based anorexia behaviors such as hyperactivity (wheel running) and weight loss, both of which are analogous to behaviors symptomatic of anorexia nervosa. animals that demonstrated aba had differential levels of nr2b 7 cusj 2018 ________________________________________________________________________________ nmda that correlated with how adaptable they were to stress evoked by food restriction. the findings of elevated nr2b-nmdars across preand post-synaptic regions correlating with ababehaviors, after a second exposure to food restriction, indicate a route for therapy through nr2b antagonism. future studies should examine how nr2bselective nmda antagonists such as ifenprodil (chennard and menniti 1999) administered through an implantable infusion into ca1 of dorsal hippocampus may alleviate aba severity. vi. author information corresponding author *ddm316@nyu.edu vii. acknowledgements thank you to dr. chiye aoki for providing her scientific mentorship throughout this project, for teaching me how to use the electron microscope, for ensuring accuracy in my data collection, and for assistance in preparing this manuscript. the responsibility of em image capturing, nr2b-peg quantification and analysis was shared between katherine varela and me, with dr. aoki overseeing our accuracy. thank you to yi-wen chen for providing assistance with data analysis and with graph generation using prism. the animals used in this study were handled by gauri wable before 2013, and the peg immunolabeling for electron microscopy was conducted by dr. aoki and yi-wen chen in the summer of 2017. viii. references [1] american psychiatric association. practice guidelines for eating disorders. am j psychiatry. 1993; 150:212. [2] aoki c, fujisawa s, mahadomrongkul v, shah pj, nader k, erisir a (2003) nmda receptor blockade in intact adult cortex increases trafficking of nr2a subunits into spines, postsynaptic densities, and axon terminals. brain res 963(1–2):139–149 [3] aoki c, kojima n, sabaliauskas n, shah l, ahmed th, oakford j, ahmed t, yamazaki h, hanamura k, shirao t (2009a) drebrin a knockout eliminates the rapid form of homeostatic synaptic plasticity at excitatory synapses of intact adult cerebral cortex. j comp neurol 517(1):105–121 [4] aoki c, lee j, nedelescu h, ahmed t, ho a, shen j (2009b) increased levels of nmda receptor nr2a subunits at preand postsynaptic sites of the hippocampal ca1: an early response to conditional double knockout of presenilin 1 and 2. j comp neurol 517(4):512–523 [5] aoki c, rodrigues s, kurose h (2000) use of electron microscopy in the detection of adrenergic receptors. methods mol biol 126:535–563 [6] arcelus j, mitchell aj, wales j, nielsen s (2011) mortality rates in patients with anorexia nervosa and other eating disorders. archives of general psychiatry, 68(7), 724-731. [7] barbarich-marsteller nc, fornal ca, takase lf, bocarsly me, arner c, walsh bt. et al. (2013) activitybased anorexia is associated with reduced hippocampal cell proliferation in adolescent female rats. behav brain res, 236, pp. 251-257 [8] bemis km (1978). current approaches to the etiology and treatment of anorexia nervosa. psychological bulletin, 85(3), 593-617. [9] berretta n, jones rsg (1996) tonic facilitation of glutamate release by presynaptic nmethyl-d-aspartate autoreceptors in the entorhinal cortex. neuroscience 75(2): 339-344 https://doi.org/10.1016/0306-4522(96)00301-6. [10] bidoret c, ayon a, barbour b, casado m (2009) presynaptic nr2a-containing nmda receptors implement a high-pass filter synaptic plasticity rule. proc natl acad sci 106:14126– 14131.doi:10.1073/pnas.0904284106 pmid:19666514 [11] boakes ra, dwyer dm. (1997), weight loss in rats produced by running: effects of prior experience and individual housing. q j exp psychol b. 50:129–148. [12] carter jc, mercer-lynn kb, norwood sj, bewellweiss cv, crosby rs, woodside db, olmsted mp. (2012) a prospective study of predictors of relapse in anorexia nervosa: implications for relapse prevention. psychiatry research, 200 (2-3):518523, https://doi.org/10.1016/j.psychres.2012.04.037. [13] chen y-w, wable gs, chowdhury tg, & aoki c. (2016). enlargement of axo-somatic contacts formed by gad-immunoreactive axon terminals onto layer v pyramidal neurons in the medial prefrontal cortex of adolescent female mice is associated with suppression of food restriction-evoked hyperactivity and resilience to activity-based anorexia. cerebral cortex. 26(6): 2574-2589. doi: 10.1093/cercor/bhv087. epub, 2015, may 15. [14] chen yw, actor-engel h, sherpa ad, klingensmith l, chowdhury tg, aoki c (2017) nr2aand nr2bnmda receptors and drebrin within postsynaptic spines of the hippocampus correlate with hunger-evoked exercise. brain struct funct 222:2271. doi: 10.1007/s00429016-1341-7 [15] chenard bl, menniti fs. (1999) antagonists selective for nmda receptors containing the nr2b subunit. curr pharm des. 5:381–404. [16] chowdhury tg, barbarich-marsteller nc, chan te, aoki c (2014) activity-based anorexia has differential effects on apical dendritic branching in dorsal and ventral hippocampal ca1. brain struct funct 219(6):1935–1945. doi:10.1007/s00429013-0612-9 [17] chowdhury tg, chen yw, aoki c. (2015) using the activity-based anorexia rodent model to study the neurobiological basis of anorexia nervosa. j. vis. exp. (104), e52927, doi:10.3791/52927. [18] chowdhury tg, wable gs, sabaliauskas na, & aoki c (2013). adolescent female c57bl/6 mice with vulnerability to activity-based anorexia exhibit weak inhibitory input onto hippocampal ca1 pyramidal cells. neuroscience, 241, 250– 267. http://doi.org/10.1016/j.neuroscience.2013.03.020 [19] dore k, stein is, brock ja, castillo pe, zito k, jespersjöström p.(2017). unconventional nmda receptor signalling. journal of neuroscience 37 (45) 10800-10807; doi:10.1523/jneurosci.1825-17.2017 [20] duguid ic, smart tg. presynaptic nmda receptors. in: van dongen am, editor. biology of the nmda receptor. boca raton (fl): crc press/taylor & francis; 8 ________________________________________________________________________________ cusj 2018 2009. chapter 14. available from: https://www.ncbi.nlm.nih.gov/books/nbk5275/ [21] fujisawa s, aoki c (2003) in vivo blockade of nmethyl-d-aspartate receptors induces rapid trafficking of nr2b subunits away from synapses and out of spines and terminals in adult cortex. neuroscience 121(1):51–63 [22] garrett l, lie dc, hrabé de angelis m, wurst w, hölter sm (2012) voluntary wheel running in mice increases the rate of neurogenesis without affecting anxietyrelated behaviour in single tests. bmc neuroscience, 13, 61. http://doi.org/10.1186/1471-2202-1361 [23] hardingham ge, bading h (2010) synaptic versus extrasynaptic nmda receptor signalling: implications for neurodegenerative disorders. nat rev neurosci 11(10):682– 696. doi: 10.1038/nrn2911 [24] holland la, brown ta, keel pk (2014) defining features of unhealthy exercise associated with disordered eating and eating disorder diagnoses. psychol sport exerc. 15(1) doi: 10.1016/j.psychsport.2013.10.005 [25] holtkamp k, hebebrand j, herpertz-dahlmann b (2004) the contribution of anxiety and food restriction on physical activity levels in acute anorexia nervosa. int. j. eat. disord., 36: 163–171. doi:10.1002/eat.20035 [26] hudson ji, hiripi e, pope hg, kessler rc (2007) the prevalence and correlates of eating disorders in the national comorbidity survey replication. biological psychiatry, 61(3), 348–358. [27] kaye wh, bulik cm, thornton l, barbarich n, masters k, and the price foundation collaborative group (2004) comorbidity of anxiety disorders with anorexia and bulimia nervosa. american journal of psychiatry 2004 161:12, 2215-2221 [28] kazi ai, oomenn a (2014) chronic noise stressinduced alterations of glutamate and gammaaminobutyric acid and their metabolism in the rat brain. noise & health 16(73): 343-349. [29] kim sf (2012). animal models of eating disorders. neuroscience, 211, 2–12. http://doi.org/10.1016/j.neuroscience.2012.03.024 [30] kunz pa, roberts ac, philpot bd (2013) presynaptic nmda receptor mechanisms for enhancing spontaneous neurotransmitter release. journal of neuroscience, 33 (18): 7762-7769; doi: https://doi.org/10.1523/jneurosci.2482-12.2013 [31] liu dd, yang q, li st (2013) activation of extrasynaptic nmda receptors induces ltd in rat hippocampal ca1 neurons. brain res. bull., 93:10-16 [32] mcquail ja, beas bs, kelly kb, simpson kl, frazier cj, setlow b, bizon jl (2016) nr2a-containing nmdars in the prefrontal cortex are required for working memory and associated with age-related cognitive decline. the journal of neuroscience, 36(50), 12537–12548. http://doi.org/10.1523/jneurosci.233216.201 [33] mistlberger, re (1994) circadian food-anticipatory activity: formal models and physiological mechanisms. neuroscience biobehav rev. 18(2):171-95. [34] national eating disorders association (2016) “anorexia: overview and statistics.” [35] olypher av, lytton ww, prinz aa (2012) input-tooutput transformation in a model of the rat hippocampal ca1 network. frontiers in computational neuroscience, 6, 57. http://doi.org/10.3389/fncom.2012.00057 [36] pare wp, vincent gp, isom ke, reeves jm (1978) sex differences and incidenc of activity-stress ulcer in the rat. psychol rep. 4: 591–594. [37] penas-lledo e, vaz leal fj, waller g. (2002) excessive exercise in anorexia nervosa and bulimia nervosa: relation to eating characteristics and general psychopathology. int j eat disord.31:370–5. [38] petralia rs, wang yx, hua f, yi z, zhou a, ge l, stephenson fa, wenthold rj (2010) organization of nmda receptors at extrasynaptic locations. neuroscience 167(1):68–87. doi: 10.1016/j.neuroscience.2010.01.022 [39] pinheiro, ap, root t, & bulik, cm (2009) the genetics of anorexia nervosa: current findings and future perspectives. international journal of child and adolescent health, 2(2), 153–164. [40] rinaldi t, kulangara k, antoniello k, markram h (2007) elevated nmda receptor levels and enhanced postsynaptic long-term potentiation induced by prenatal exposure to valproic acid. proc natl acad sci usa 104(33):13501–13506. doi: 10.1073/pnas.0704391104 [41] schmidt u, tiller j, blanchard m, andrews b, & treasure j (1997) is there a specific trauma precipitating anorexia nervosa? psychological medicine, 27(3), 523-530. [42] shen h, gong qh, aoki c, yuan m, ruderman y, dattilo m, williams k, smith ss (2007) reversal of neurosteroid effects at alpha4beta2delta gabaa receptors triggers anxiety at puberty. nat neurosci 10(4):469–477 [43] sibold js, hammack se, falls wa (2011) c57 mice increase wheel-running behavior following stress: preliminary findings. percept mot skills. oct; 113(2): 605– 618. doi: 10.2466/06.16.20.pms.113.5.605-618 [44] sjöström pj, turrigiano gg, nelson sb (2003) neocortical ltd via coincident activation of presynaptic nmda and cannabinoid receptor.s neuron, 39(4): 641 – 654 [45] stagni f, giacomini a, guidi s, ciani e, & bartesaghi r (2015). timing of therapies for down syndrome: the sooner, the better. frontiers in behavioral neuroscience, 9, 265. http://doi.org/10.3389/fnbeh.2015.00265 [46] wable gs, min jy, chen yw, aoki c (2015) anxiety is correlated with running in adolescent female mice undergoing activity-based anorexia. behavioral neuroscience; 129(2): 170-182. doi:10.1037/bne0000040. [47] yang j, woodhall gl, jones rs (2006) tonic facilitation of glutamate release by presynaptic nr2bcontaining nmda receptors is increased in the entorhinal cortex of chronically epileptic rats. j neurosci 26(2):406–410. doi: 10.1523/jneurosci.4413-05.2006 9 a letter from the editor-in-chief dear readers, 2014 proved to be an instrumental year for the growth and total revitalization of the columbia undergraduate science journal. our associate editors, ceo and myself looked forward to recruiting a new editorial board made up of underclassmen and building strong relationships with these editors through weekly workshops and event planning meetings. i am especially proud of our board’s commitment to building strong relationships; what was once a journal that lacked mentorship opportunities has now blossomed into a tight-knit community of aspiring researchers who value the power of quality writing. a highlight of our events this past year was our spring undergraduate research symposium, which was held on april 20th, 2014. 22 research posters were presented by students from colleges across the northeast region. it was delightful to feel the passion from each of these students as they talked about their research, and we look forward to hosting the symposium again on april 26th, as we have expanded the event to include 30 posters. looking at this year’s journal, i am also proud to see the various authors and research that we have been blessed to exhibit. for the first time, our editors have been given the opportunity to publish in the journal through our editorials section. we are thrilled to provide these fresh perspectives on scientific research to our audience and hope to continue this practice in future journals. in addition, the multi-page papers in this journal showcase the breadth of scientific research that our organization promotes. i sincerely hope that you enjoy the articles. all in all, the columbia undergraduate science journal has made major progress over the past year. “stability leads to legitimacy” has been our mantra this academic year, and our e-board has provided the stability needed to give meaningful experiences for our editorial board and a quality publication for our readers. thank you for your continued support. it has been an honor to serve as the editor-in-chief of the columbia undergraduate science journal for this past year. – ritish patnaik microsoft word sentiment analysis election paper cusj (v2) sentiment analysis of tweets to gain insights into the 2016 us election tim hamling, ankur agrawal department of computer science, manhattan college, new york abstract — social media use is at an all-time historic high for the united states, so we considered one popular social media platform, twitter, and tried to see if we could predict how a group of people felt about an issue by only using posts from social media. for our research, we looked at tweets that focused on the 2016 united states presidential election. using these tweets, we tried to find a correlation between tweet sentiment and the election results. we wrote a program to collect tweets that mentioned one of the two candidates, then sorted the tweets by state and developed a sentiment algorithm to see which candidate the tweet favored, or if it was neutral. after collecting the data from twitter and comparing it to the results of the electoral college, we found that twitter sentiments corresponded with 66.7% of the actual outcome of the electoral college. the overall sentiment of all tweets collected leaned more positively towards donald trump than it did for hillary clinton. using the data that was collected, we also looked at how different geographical locations affected a candidate’s popularity, analyzed what issues were most prevalent in tweets, and looked at the ratio of a state’s population versus the number of tweets gathered. i. introduction in united states politics, 2016 was an important year because it was an election year for the nation. the two presidential candidates were donald trump for the republican party and hillary clinton for the democratic party. as the days grew closer to november 8th, election day, many news outlets believed that the race between the two candidates was extremely close, with sites such as fivethirtyeight reporting that the election could be a toss-up for either candidate [1]. the united states’ presidential elections work using an electoral college, meaning that each state has a certain number of electors who will vote for a candidate. voters cast their vote for these electors, meaning that popular vote does not always decide the outcome [2][3]. this makes predicting the election’s outcome difficult, so leading up to election day, political scientists looked to polling data as well as computer analysis programs to try and see which candidate was poised to win. the nature of the election, as well as the two candidates themselves, was widely discussed across both the united states and other countries in the world [4]. debates and conversations about the candidates and their policies most certainly happened verbally, but people also took to social media to voice their opinions about the election. this is no surprise, especially due to the current usage of social media sites such as facebook, with 1.79 billion monthly active users [5], and twitter, with around 313 million monthly active users [6]. in addition to these platforms, the rising popularity of sites such as snapchat, with 150 million daily active users [7], and instagram, with 500 million monthly active users [8], also contributed to social media conversations. analyzing social media users’ activity to determine what people are talking about is quite easy because many people, when posting to social media, make their posts and activity public for anyone to see. twitter is a social media service that allows users to post “tweets” to the site [9]. these can either be viewed publicly by anyone who wishes to see, or can be made private so that only people who have been allowed to follow a user can see that person’s tweets. regardless of the privacy, one of twitter’s features is that each tweet is limited to 140 characters, which includes whitespace characters, non-ascii text, or links to web pages or images [10]. while this may seem like an odd restriction, it was adopted to work in conjunction with sms messaging services, and has stuck around because it forces users to get to the point and only focus on providing important details in their tweet. due to the quick and concise nature of tweets, some classify the social media site as a news media site [11]. per research done by the research group statista, as of quarter 3 in 2016, twitter has approximately 317 million monthly users. the monthly user base has been rising since twitter’s creation in 2006, but has slowly leveled off around the 300 million user mark since early 2015 [12]. another key feature of twitter is its use of hashtags, represented by a pound sign (#) preceding a word or phrase. these are used within tweets by a user to identify different topics or keywords present within the tweet. then, other users could search other instances of that specific hashtag to see other tweets that match that topic. for example, during breaking news stories, hashtags are helpful because they allow a user to search and filter for tweets that discuss that news story. as twitter has evolved, it is now possible for users to search by keywords that are not marked with a hashtag. this allows others to search for a work and receive a listing of all tweets that mention or use that word. in this study, we set out to see if we could predict the results of the 2016 united states presidential election by looking at tweets that mention either candidate, and analyzing them to determine an overall sentiment. this sentiment could be leaning in favor or against either candidate. for this study, we sought to collect any tweets that were posted in the days leading up to the united states election that openly mentioned either of the two candidates by name. we then aimed to map these tweets, along with their sentiments, to locations, either within the united states or from foreign countries. for locations within the united states, we mapped each tweet to one of the states in the us, which was determined based on a user’s location description tag. finally, we collected the total number of tweets from a state and looked at the ratio of positive/negative tweets for each candidate to determine how that state, overall, feels about each candidate. these results would then be checked against the official results of the election, and we could then see if the way a state voted in the electoral college matched the prediction we made about it based on that state’s tweet sentiment. sentiment analysis of twitter data is not a new field. pang and lee worked on creating algorithms to facilitate opinion mining and word analysis in their 2008 study [13]. researchers such as go, huang, and bhayani conducted a study in 2009 to train a sentiment algorithm to detect a tweet’s positivity about a certain subject by using emoticons [14]. pak and paroubek expanded upon this in 2010 by using the subjectivity and objectivity of words in conjunction with a tweet’s structure to create a classifier that could use collected data to determine a tweet’s sentiment [15]. we decided to create our own sentiment algorithm that used a collection of words, each with its own sentiment value, to analyze tweets about the two candidates. we also used the data collected to see what issues were most discussed in the week leading up to the election. using a list of popular political issues during elections [16], we searched through our collection of tweets to see if any of these issues were being mentioned, and counted which issues were discussed most frequently. these totals were then mapped to each state, so we could then see which states talked about which issues most, and try to find correlations between the issues being discussed and the results of the election in that state. ii. methods our methodology composed of collecting tweets and associated metadata from twitter and performing a sentiment analysis on each tweet. we developed a few algorithms and implemented them using the java and perl programming languages to access the tweets, clean them, perform a sentiment analysis and aggregate the results. to collect data, we developed an algorithm that implemented the twitter4j library [17]. this algorithm searched through all publicly posted tweets that mentioned, by name, either of the two presidential candidates, and stored all the tweets that matched these criteria in text files. our search words were “clinton”, “hillary”, and “trump”. we did not include “donald” as a search word because it is a common name and could result in too many results that did not relate to the election. when we stored the tweets in the files, we included the tweet’s text, the handle/username of the poster of the tweet, the timestamp of when the tweet was posted (using est), the location provided by the user’s twitter bio, and a number representing which state (if any) corresponded to this location. these state numbers were determined using a database, explained in the next paragraph. each tweet in the file, and the tweet’s corresponding info, was separated from the next tweet by a series of dashes (-----). an example of how the tweet is laid out is shown in fig. 1. line numbers were included in this figure to aid in readability. data from twitter was collected for the week leading up to the us presidential election on november 8th. we separated the data by the date the tweet was posted, so we ended up with eight files containing tweets, one file for each day from november 1st to november 8th. figure 1. example layout of a tweet in our data file with line numbers added in manually after collecting the data, we began to analyze the tweets. first, we developed an algorithm to determine the state that the tweet corresponded with. to sort tweets, we looked at the location tag provided by the tweet’s user. we parsed it for any mention of a us state, either by full name or abbreviation. if a state was found, the tweet was given a number representing that state. for example, alabama was given a value of 00, alaska got 01, arizona got 02, and so on. washington dc was also included, and was given its own value. in addition, if we detected a location origination from mexico, canada, or great britain we gave that tweet its own unique location number. these countries were included in our analyses so that we could see what people outside of the us were tweeting in regards to the election. if no us state or outside country was detected by name, then we parsed the location line for a town name. we used a database provided from simplemaps.com to match town names to states [18]. the database includes all unites states towns and their matching states. if our algorithm detected multiple states associated with one town name, then we excluded that dataset. any tweets that did not match with any state or did not have a location line were given a state value of -1 to signal that it was a tweet without a location. we still ran our algorithm on these tweets to see what results it would come up with. the sentiment analysis of our algorithm was only run on tweets that exclusively mentioned either candidate; tweets that mentioned both candidates were excluded from this portion of the algorithm. sentiment was determined using the sentiment wordmap file from sentiwordnet with some modifications and additions to words associated with the election [19]. this sentiment file has a positive or negative score associated with each word. for positive words, values ranged from +0.0625 to +1.0; for negative words, values ranged from -0.0625 to -1.0. for example, a somewhat positive word like “helpful” would receive a value of +0.125 from the sentiment algorithm, or a strong negative word like “unhappy” would receive a value of -0.25. neutral words like “a” or “the” were given neutral values of 0.0. since these words would not affect the score, we decided to remove all neutral words from the file to improve the efficiency and 1 @katastrophy1951 trump only wants to take care of the girl and make sure she is not hurt anymore. said he's worried about her not his star. 2 katastrophy1951 3 1 nov 2016 05:46:02 gmt 4 seattle ,wa. 5 46 6 ---- speed of our algorithm. when our algorithm encountered a word in a tweet that was not in our sentiment file, it defaulted to giving it a neutral value of 0. another function we implemented was recognition of negation words like “not” or “don’t.” our algorithm would detect these words as negations, and flip the value of the next word. so, while “good” may return a value of +0.375, “not good” would return -0.375 because the algorithm recognized “not” as a negation word. in addition to these functions within our algorithm, we wanted to further refine the sentiment word file. we went through the file by hand and removed or modified words that had inaccurate values. we also did sample testing on tweets from our data to see if the values of any words needed to be changed. for example, the word "investigation" was given a negative score because, in the context of the election, an investigation is never something good. sometimes, words had to be added to the file. the word “wikileaks” was not included in the file originally, but we decided to add it and give it a negative value because it is normally associated with a scandal or something negative. other words were completely removed from the sentiment file. for example, the word “trump” was taken out of the file. this was done because this verb had a positive value in the file, but since “trump” is the name of one of the candidates, we had to remove it to prevent erroneous data. another aspect of our algorithm included scanning a tweet for different hashtags or phrases that would automatically indicate it as a positive or negative tweet. if a tweet included “#maga”, “crooked hillary”, “lock her up”, or “#trumptrain” then we knew for sure that it would be a positive tweet for trump and a negative tweet for hillary. on the other side, if a tweet contained “i’m with her”, “madame president”, “dump trump”, or “stronger together” then we could know for sure that it would be a positive tweet for clinton, and a negative tweet for trump. hashtags and key phrases were collected from top-hashtags, which aggregated data from twitter, facebook, instagram, and tumblr [20]. to further refine our results, we looked to remove any duplicate tweets. sometimes, multiple news organizations tweet a link to the same article, and use that article’s headline as the content within the tweet. this can be seen in this tweet by @collectedn that states “the campaign tour: pop musicians get on the bus (mostly clinton’s) https://t.co/b6cpxuz6nf” [21] and this tweet by @thetablenz that states “the campaign tour: pop musicians get on the bus (mostly clinton’s) https://t.co/srx84kvxpa” [22]. both are linking to an article published by the ny times, and use the same headline as the tweet body. the only difference is the url that the tweet links to. when our algorithm collected tweets, it would save the entire tweet body, which includes any urls. to remove duplicate tweets, such as these posts about news articles, we decided to remove any urls from the tweets so that we could compare the raw text of each tweet. we then got rid of tweets that contained identical text, leaving us with only one copy. this improved our results by preventing repetitions within our data. using the sentiment file in addition to our updates, our algorithm would take a tweet and break it apart into words. next, it would give each word a value per the sentiment file. after each word had a score, the algorithm summed up all the scores to determine the overall sentiment of the tweet. we counted tweets that had a score above 0.0 as positive tweets, tweets that had a score below 0.0 to be negative tweets, and tweet with a score of 0.0 as a neutral tweets. an algorithm was also developed to see what topics were most widely discussed throughout our collection of tweets. to do this, we looked at topics discussed during the 2016 election, using research from political science sites such as fivethirtyeight [23] and pew research center [24], as well as polls conducted by the new york times and the washington post [25]. we settled on 10 different issues: economy, education, election problems, environment, foreign policy, guns, healthcare, immigration, social, and trade. the full list of words we used to cover each topic is shown in fig. 2. figure 2. list of keywords used when searching for different topics within tweets iii. results table 1 shows our results from the data that we collected. tweets from the 50 us states are listed first, followed by tweets from washington dc, then canada, then great britain, and finally mexico. tweets that did not include locations are listed under n/a. each state’s electoral outcome is shown in the far-right column. these results were gathered from the new york times [26]. there are two us states that do not give all their electoral votes to the winner of that state, and these two states economy: job market, minimum wage, stimulus, economy, taxpayers education: common core, tuition, fafsa election: voter fraud environment: greenhouse, global warming, climate, pipeline, drilling, fracking, nuclear, solar foreign policy: syria, isis, israel, patriot act, surveillance, al qaeda, iran, middle east guns: regulation, no-fly, concealed, carry law, bernardino, pulse healthcare: obamacare, medicare, medicaid, health care, vaccination, affordable care immigration: mexico, muslim, boarder wall, immigrant, refugee, xenophobia, daca, dapa, h1b social: parenthood, abortion, lgbtq, racism, feminism, homosexual, gender, religion, gay marriage trade: tpp, nafta, import, export, trade, business table i. correlation between tweet sentiment by state and electoral college result are maine and nebraska. in these states, congressional districts decide several electoral votes while the remaining are given to whichever candidate gets more votes in that state [27]. in the 2016 election, maine was the only state to split its votes in this manner. it awarded three electoral votes to hillary clinton, who won the popular vote of the state as well as a congressional district, and 1 electoral vote to donald trump for winning a congressional district [28]. the remaining 49 states and washington dc cast their electoral votes entirely for either donald trump or hillary clinton. state clinton tweets % pos (clinton) % neg (clinton) trump tweets % pos (trump) % neg (trump) twitter results elect. college correct prediction total 2810051 39.6% 45.7% 4044162 43.2% 42.2% trump trump yes al 13931 37.2% 48.9% 14650 45.7% 39.4% trump trump yes ak 3404 45.7% 39.7% 2924 42.8% 42.7% clinton trump no az 19111 37.3% 49.4% 20333 44.7% 42.9% trump trump yes ar 6272 39.1% 46.6% 6917 44.5% 41.6% trump trump yes ca 103818 40.6% 45.7% 155246 42.6% 44.0% trump clinton no co 18434 39.8% 47.0% 23128 42.4% 44.1% trump clinton no ct 10515 41.0% 45.0% 15397 42.1% 44.0% trump clinton yes de 1613 41.7% 43.5% 2641 40.0% 45.4% clinton clinton no fl 84085 37.6% 48.3% 108133 44.6% 40.9% trump trump yes ga 27977 39.3% 47.4% 35206 44.6% 41.5% trump trump yes hi 4267 37.4% 49.5% 5298 48.4% 38.8% trump clinton no id 3429 34.5% 53.0% 4454 43.2% 43.9% trump trump yes il 24821 41.5% 45.0% 35979 42.7% 44.1% trump clinton no in 37409 40.2% 45.6% 56576 42.5% 43.4% trump trump yes ia 5808 41.1% 45.3% 8149 43.5% 43.3% trump trump yes ks 6244 39.9% 45.7% 8788 43.4% 43.9% trump trump yes ky 10280 37.0% 50.1% 12817 45.5% 41.4% trump trump yes la 18136 35.5% 44.7% 25244 37.6% 40.6% trump trump yes me 7152 38.6% 45.5% 10395 41.4% 42.9% trump clinton no md 19105 42.8% 44.7% 12763 44.4% 41.8% trump clinton no ma 29621 43.4% 43.1% 19251 42.0% 44.4% clinton clinton yes mi 23349 40.4% 44.3% 33935 43.8% 41.7% trump trump yes mn 10004 42.7% 43.4% 17207 44.5% 42.0% trump clinton no ms 5679 37.5% 47.8% 6033 44.0% 39.5% trump trump yes mo 14574 38.7% 48.4% 17740 44.7% 42.4% trump trump yes mt 3697 34.6% 50.5% 4050 42.9% 43.1% trump trump yes ne 3992 39.5% 45.5% 5586 43.5% 42.3% trump trump yes nv 13664 38.1% 48.8% 17184 44.6% 42.6% trump clinton no nh 4704 40.3% 45.8% 6300 44.7% 41.8% trump clinton no nj 24578 39.3% 47.3% 35938 42.5% 43.8% trump clinton no nm 4596 39.0% 48.7% 6127 43.7% 43.4% trump clinton no ny 87549 42.6% 43.2% 136087 42.7% 44.1% clinton clinton yes nc 28432 41.0% 45.8% 37555 44.7% 42.6% trump trump yes nd 1196 38.1% 49.2% 1338 47.8% 40.7% trump trump yes oh 29317 41.1% 45.2% 39320 43.7% 42.6% trump trump yes ok 9847 39.9% 45.4% 10429 44.5% 41.5% trump trump yes or 14058 40.4% 47.2% 21570 43.1% 44.9% trump clinton no pa 32602 40.9% 45.5% 45773 44.8% 41.8% trump trump yes ri 2812 40.5% 46.7% 4472 42.9% 45.0% trump clinton no sc 13313 38.7% 47.2% 15496 44.3% 40.3% trump trump yes sd 1204 41.4% 44.3% 1512 45.2% 40.1% trump trump yes tn 16916 38.8% 48.4% 19025 46.0% 40.1% trump trump yes tx 80770 37.6% 48.2% 98815 44.4% 42.0% trump trump yes ut 4984 42.0% 43.3% 7027 44.2% 40.7% trump trump yes vt 1726 44.0% 44.2% 2653 42.5% 47.3% clinton clinton yes va 26620 38.1% 49.2% 32515 44.3% 42.8% trump clinton no wa 21189 41.5% 45.1% 29288 42.4% 44.7% trump clinton no wv 4753 38.8% 47.6% 5544 45.4% 41.6% trump trump yes wi 11657 41.8% 44.2% 15866 43.4% 42.8% trump trump yes wy 1254 41.1% 48.8% 1638 46.5% 41.9% trump trump yes dc 34016 45.0% 39.3% 52744 43.0% 41.8% clinton clinton yes can 43087 40.2% 44.4% 74136 41.6% 44.1% trump n/a n/a gb 51657 44.7% 41.0% 98721 44.3% 41.5% clinton n/a n/a mx 6751 19.7% 25.2% 11444 19.0% 28.2% clinton n/a n/a n/a 1720072 36.5% 42.8% 2546805 38.8% 40.0% trump n/a n/a the majority of maine’s electoral votes went to hillary clinton, and because she won that the popular vote there as well, we marked maine as being won by hillary clinton. to understand if our results matched with those of the electoral college, we compared the percentage of positive tweets for each candidate per state to one another, as well as the percentage of negative tweets for each candidate per state. a candidate’s “favorability” was found by subtracting the percentage of negative tweets for that candidate in a state from the percentage of positive tweets for that candidate in a state. each candidate’s favorability value per state was compared, and the candidate with the higher favorability in that state was marked under the “twitter results” column. from table 1, we can see that our sentiment showed more tweets being positive towards trump and negative towards clinton. overall, there were only five states plus washington d.c. that had a higher favorability for clinton than trump. the five states were alaska, delaware, massachusetts, new york, and vermont. four of these five states were in the north-eastern region of the united states. in addition, alaska, which was the only non-atlantic state to support clinton on twitter over trump, ended up casting its electoral votes for donald trump in the election. table 1 also shows which states tweeted most, or had the highest percentage of positive or negative tweets. for both candidates, california and new york were the two states that tweeted about each candidate the most. it makes sense for california to have the most total tweets about each candidate since it is the state with the highest population in the united states. we decided to see if the population of a state had any effect on the tweet count coming from that state, so we decided to plot the total number of tweets from a state, and then compared it to the state’s population. data was gathered using 2016 us census counts [29], and results are shown in table 2. these results are listed in order of descending population. fig. 3 shows that there is a correlation between state population and tweet count. the linear growth of the graph shows that there is a somewhat constant ratio between state population and tweet count. table ii. comparison of state population to tweet counts state population total tweets ca 39250017 184588 tx 27862596 164855 fl 20612439 171634 ny 19745289 120151 pa 12802503 57423 il 12801539 54138 oh 11614373 57294 ga 10310371 56409 nc 10146788 51781 mi 9928301 47927 nj 8944469 51198 va 8411808 47809 wa 7288000 40300 az 6931071 48732 ma 6811779 46537 tn 6651194 54325 in 6633053 51983 mo 6093000 33679 md 6016447 30762 wi 5778708 30091 co 5540545 28438 mn 5519952 23317 sc 4961119 27244 al 4863300 32067 la 4681666 28416 ky 4436974 24338 or 4093465 23905 ok 3923561 20362 ct 3576452 16323 ia 3134693 10792 ut 3051217 10663 ms 2988726 11951 ar 2988248 19936 nv 2940058 19908 ks 2907289 10840 nm 2081015 8588 ne 1907116 8745 wv 1831102 8182 id 1683140 7696 hi 1428557 8971 nh 1334795 11856 me 1331479 9964 ri 1056426 6509 mt 1042520 5310 de 952065 2817 sd 865454 2400 nd 757952 4600 ak 741894 37420 dc 681170 35742 vt 624594 2980 wy 585501 1254 figure 3. graph plotting state population vs tweet count, with line of best fit in addition to looking at the 50 states and washington d.c., we also looked at how other countries discussed the two candidates. canada had far more tweets that discussed trump than those that discussed clinton, about 74,000 compared to 43,000. however, the positive and negative percentages were about the same between each candidate, with 40% to 41% of tweets being positive for clinton and trump respectively, and 44% of tweets being negative for both. great britain was similar, having about 51,000 tweets about clinton compared to 98,000 tweets about trump. the positive percentages for clinton and trump were both around 44%, while clinton had 0 50000 100000 150000 200000 250000 0 20000000 40000000 60000000 a slightly lower negative percentage of 40% compared to trump’s 41%. in mexico, the positive and negative percentages were both lower than anywhere else, but this is likely since many of these tweets were written in spanish, and our algorithm used an english dictionary to operate. states that had a calculated favor that matched their electoral college votes were marked with a green highlight under the final column, “correct prediction” in table 1. out of the 50 states, plus washington dc, the sentiment on twitter for 34 states correctly aligned with the electoral college’s outcome for that state, resulting in an accuracy of about 66.7%. we wanted to see if the geographical location of a state affected how it voted in the election. to start, we had to divide the 50 states into regions. we looked for different maps that showed various ways to section off the united states, and finally settled on a map with the following five regions: northeast (me, vt, nh, ri, ct, nj, ny, pa), southeast (md, de, dc, wv, va, ky, tn, nc, sc, ar, la, ms, al, ga, fl), midwest (nd, sd, mn, wi, mi, ne, ia, il, in, oh, ks, mo), southwest (az, nm, ok, tx), and west (ak, hi, wa, or, id, mt, wy, ca, nv, ut, co). after combining state results from each state by region, the results are shown in table 3. in addition to the results in table 1, we were also able to see what issues were most discussed by each state leading up to the election. in table 4, we show the percentage of issues discussed by states that were, per the electoral college vote, won by donald trump. in table 5, we show the percentage of issues discussed by states that hillary clinton won. column one lists the abbreviation for each state, and column two lists the total number of tweets from that state that mentioned one of the issues we searched for. from there, the remaining columns each list the percentage of that total that relate to each issue. from table 4, we can see that out of the 30 states that donald trump won in the election, the most frequently discussed issues were immigration and foreign policy. trade issues and social issues were also discussed often, as well as gun issues, healthcare, and environmental issues. finally, economic issues and election issues were discussed the least often, and education was barely discussed in any tweets. table 5 shows that out of the 21 states that hillary clinton won in the election, the most frequently discussed issues were immigration and foreign policy, which matches the discussion trends for states that voted for trump. following this, foreign policy, social issues, and trade issues were the next most discussed topics. environment, guns, and healthcare were tweeted about less often than these, while economic and election issues were talked about even less frequently. finally, education issues were talked about very minimally throughout our tweets. this data shows that, overall, the candidate that a state voted for had little impact on which issues were most talked about by that state. table iii. division of twitter results by geographic location table iv. issue discussion percentage on states that donald trump won in the electoral college state total economy education election environment guns healthcare immigration foreign social trade al 1531 5.0% 0.3% 3.6% 7.1% 10.3% 8.0% 21.2% 17.1% 15.6% 11.9% ak 352 2.8% 0.0% 1.7% 10.8% 9.9% 7.7% 20.7% 16.5% 15.1% 14.8% az 2954 3.9% 0.3% 3.0% 7.8% 9.3% 8.4% 21.4% 17.7% 14.4% 13.9% ar 787 3.0% 0.4% 3.6% 9.8% 11.4% 8.1% 18.9% 17.3% 16.9% 10.5% fl 13178 3.8% 0.2% 3.0% 7.7% 9.1% 8.5% 20.9% 18.5% 12.9% 15.4% ga 3719 3.7% 0.1% 2.6% 6.9% 11.5% 8.7% 20.4% 15.9% 14.6% 15.5% id 430 2.6% 0.0% 1.9% 7.0% 7.4% 9.3% 23.5% 14.4% 19.5% 14.4% in 5680 3.5% 0.3% 2.3% 7.9% 9.7% 6.0% 18.9% 22.7% 17.1% 11.6% ia 681 3.1% 1.0% 3.2% 9.8% 9.8% 8.1% 22.9% 13.5% 16.3% 12.2% ks 873 2.2% 0.1% 1.9% 6.8% 11.5% 5.6% 18.2% 21.0% 20.0% 12.7% ky 1224 2.6% 0.2% 3.6% 7.0% 10.2% 8.4% 19.9% 18.3% 15.7% 14.1% la 2226 3.1% 0.2% 2.7% 7.7% 11.8% 6.9% 23.4% 17.8% 14.4% 11.9% mi 3277 2.8% 0.3% 1.6% 8.2% 9.9% 7.2% 16.9% 16.1% 16.3% 20.6% ms 567 3.0% 0.5% 3.0% 6.7% 10.9% 10.2% 18.0% 18.0% 18.0% 11.6% mo 1871 3.4% 0.1% 2.7% 6.6% 11.5% 10.6% 19.5% 16.5% 17.0% 12.1% mt 452 2.9% 0.0% 3.1% 6.4% 9.1% 6.6% 21.9% 22.8% 13.1% 14.2% ne 462 4.5% 0.6% 3.2% 10.2% 10.6% 6.7% 21.2% 14.1% 16.9% 11.9% nc 3757 4.4% 0.5% 2.4% 6.5% 13.7% 8.8% 19.1% 15.3% 16.4% 12.8% nd 123 0.0% 0.0% 0.0% 16.3% 18.7% 3.3% 22.0% 17.1% 10.6% 12.2% oh 3547 4.1% 0.4% 2.8% 12.2% 10.8% 7.1% 17.7% 13.5% 17.5% 14.0% region clinton tweets % pos (clinton) % neg (clinton) trump tweets % pos (trump) % neg (trump) total 2810051 39.6% 45.7% 4044162 43.2% 42.2% n.east 201259 41.7% 44.4% 276266 42.9% 43.6% s.east 311128 39.3% 46.6% 387283 44.0% 41.3% midwest 169575 40.7% 45.3% 241996 43.4% 42.9% s.west 114324 37.8% 48.2% 135704 44.4% 42.2% west 192198 40.3% 46.2% 271807 42.9% 43.8% ok 1033 3.8% 0.4% 2.5% 8.5% 9.7% 7.6% 21.4% 16.4% 18.2% 11.6% pa 4761 4.1% 0.3% 2.6% 9.2% 8.9% 9.1% 20.4% 14.7% 19.3% 11.4% sc 1573 4.6% 0.3% 2.2% 6.3% 8.9% 9.3% 19.0% 17.9% 20.9% 10.7% sd 158 5.1% 0.0% 1.9% 7.0% 15.8% 9.5% 21.5% 11.4% 10.1% 17.7% tn 1983 3.8% 0.1% 2.7% 7.2% 10.8% 9.7% 20.0% 17.2% 15.3% 13.2% tx 10846 4.3% 0.3% 3.1% 7.2% 9.6% 8.0% 22.2% 17.4% 15.3% 12.5% ut 614 2.4% 0.3% 2.1% 10.9% 10.9% 8.5% 19.4% 15.3% 17.6% 12.5% wv 541 4.4% 0.2% 1.8% 5.4% 13.1% 10.5% 20.0% 14.4% 16.5% 13.7% wi 1484 4.9% 0.5% 3.7% 10.3% 12.7% 9.6% 17.4% 13.2% 15.3% 12.5% wy 180 3.9% 0.0% 1.1% 9.4% 8.3% 8.9% 26.7% 18.3% 8.9% 14.4% table v. issue discussion percentage on states that hillary clinton won in the electoral college state total economy education election environment guns healthcare immigration foreign social trade ca 17593 4.1% 0.2% 2.4% 9.6% 9.3% 7.6% 20.4% 17.1% 15.2% 14.0% co 2593 2.9% 0.2% 2.5% 10.9% 9.3% 7.4% 19.5% 16.7% 16.6% 14.1% ct 1313 4.4% 0.5% 2.6% 7.9% 10.7% 8.5% 17.6% 17.0% 15.8% 14.9% de 1633 2.6% 0.1% 1.0% 7.2% 6.7% 5.3% 29.3% 28.8% 11.8% 7.1% hi 547 4.6% 0.4% 4.6% 8.8% 8.8% 8.6% 19.4% 19.7% 13.5% 11.7% il 3356 3.9% 0.2% 1.9% 9.7% 10.3% 9.1% 19.3% 15.6% 17.3% 12.7% me 900 2.1% 0.3% 1.0% 10.1% 12.0% 8.2% 19.2% 17.3% 16.6% 13.1% md 1823 3.9% 0.2% 2.4% 8.9% 10.3% 6.3% 22.2% 13.5% 18.7% 13.7% ma 2894 4.5% 0.3% 2.0% 10.4% 9.8% 7.6% 21.1% 15.3% 16.7% 12.3% mn 1561 4.2% 0.2% 2.5% 9.4% 10.5% 7.2% 21.7% 12.6% 14.3% 17.3% nv 2180 2.9% 0.5% 1.8% 7.2% 9.7% 7.9% 20.9% 23.8% 10.9% 14.3% nh 850 4.5% 0.1% 2.2% 10.4% 7.9% 5.8% 14.6% 10.8% 10.5% 33.3% nj 3768 3.1% 0.3% 2.0% 8.2% 9.5% 7.8% 21.7% 17.8% 15.8% 13.8% nm 923 2.7% 0.0% 5.4% 8.7% 8.5% 4.4% 30.6% 14.1% 12.6% 13.1% ny 16730 4.0% 0.2% 1.8% 9.4% 9.2% 8.3% 21.9% 16.1% 14.8% 14.3% or 2080 4.0% 0.4% 2.8% 10.6% 10.1% 7.4% 20.4% 13.9% 16.7% 13.7% ri 398 6.5% 0.3% 2.0% 9.0% 13.3% 7.5% 18.6% 12.1% 14.8% 15.8% vt 294 3.4% 0.7% 0.7% 18.4% 7.8% 11.9% 13.9% 9.9% 22.1% 11.2% va 3336 3.6% 0.2% 2.3% 8.2% 10.2% 7.6% 19.3% 20.3% 15.0% 13.3% wa 2982 3.0% 0.2% 2.6% 11.0% 9.9% 6.0% 19.8% 17.8% 15.6% 14.1% dc 6757 3.7% 0.1% 1.8% 10.9% 8.4% 11.5% 22.1% 14.3% 14.6% 12.5% iv. dicussion the goal of this study was to determine if the sentiment on twitter for the 2016 election matched with the electoral results of the election. our results showed us that results from twitter sometimes matched, but were not always accurate with the results of the electoral college. of the 50 states plus washington dc, 34 states had results from twitter that matched their electoral college results. of the states that twitter correctly predicted, the only state that supported hillary on twitter but voted for trump was alaska. there were 29 states, alabama, arizona, arkansas, florida, georgia, idaho, indiana, iowa, kansas, kentucky, louisiana, michigan, mississippi, missouri, montana, nebraska, north carolina, north dakota, ohio, oklahoma, pennsylvania, south carolina, south dakota, tennessee, texas, utah, west virginia, wisconsin, and wyoming, all voted for trump in the election and showed more support on twitter for trump as well. there were five states, including washington dc, that backed hillary clinton in our finding as well as in the election. these states were delaware, massachusetts, new york, vermont, and washington dc. this leaves 17 states that voted for clinton yet supported trump more on twitter: california, colorado, connecticut, hawaii, illinois, maine, maryland, minnesota, nevada, new hampshire, new mexico, oregon, rhode island, virginia, and washington state. when looking at states that had differing results on twitter, there were more states that voted for clinton yet supported trump on twitter than states that voted for trump yet supported clinton on twitter. of the 30 states that trump won in the election, twitter sentiments matched for 96.7% of them. for clinton’s 20 states plus d.c., twitter sentiments matched 28.6% of the state results. comparing these percentages shows that, while overall the twitter sentiments lined up with state results 66.7% of the time, most of this percentage came from states that supported donald trump on twitter as well as in the election. these results point to twitter being either more positive towards donald trump, more negative towards hillary clinton, or both. in fact, donald trump was discussed in tweets far more often than hillary clinton was; there was not a single state that tweeted about hillary clinton more than donald trump. although this does not correlate to an increase in positive tweets for trump, it could certainly be a factor. a final piece of evidence is that, per our tweet analysis, donald trump had a higher total favorability rating than hillary clinton. the percentage of positive tweets for trump was 43.47% while clinton had a lower percent, with 40.03% of tweets about her being marked as positive. hillary clinton’s negative tweet percentage was 46.03%, which is a bit higher than donald trump’s 42.71%. by looking at the most discussed issues, we can see that issues relating to the two candidates was by far the most discussed out of any of the topics. this is likely because, if someone is going to be tweeting about a candidate, there is a high chance that they will be talking about something related to that candidate, such as an issue or scandal revolving around the candidate. following this, immigration was the next most discussed topic, which again makes sense because it was a very heavily discussed topic during the election. news sources such as usa today claimed “immigration at front of 2016 presidential race” and our data shows this is likely true per twitter [31]. there are many reasons for why the correlation is not fully accurate and leaned heavily towards trump as opposed to clinton. firstly, our sentiment algorithm is not perfect when used to analyze the word structure. it works by summing up values of individual words, but cannot detect the sentiment from complex word clauses or phrases consisting of multiple words. one example of this is the following tweet by twitter user @buttonlol who tweeted “another example of above the law… hitlerary [sic] clinton…” [32]. although this tweet is understood to be negative towards clinton by the usage of the phrase “above the law,” the three words that make up that phrase carry little to no connotation on their own, and thus our sentiment algorithm would be unable to classify it as negative. in addition, we only dealt with tweets that exclusively mentioned a candidate. if a tweet were to mention both trump and clinton, our algorithm could not identify who the subject was or who to give the sentiment score to, so we discarded all these tweets. if we could include these tweets and isolate which candidate the tweet was focusing on to attach a sentiment, the numbers for both candidates would certainly be changed. whether this would balance out the results or further increase the divide is unknown, but it would certainly make the results much more accurate. sarcasm is another weakness of our algorithm, and when testing, we saw that it was impossible for our algorithm to detect sarcasm at this stage. one example of sarcasm is a section from a tweet by twitter user @followingleads who tweeted “…nice unbiased piece on trump today…” [33]. out of context, it appears this tweet is supporting an article posted about trump. however, this tweet was posted as a sarcastic response to an anti-trump journalist. this context can be determined by looking at who the tweet was directed at, as well as reading into the rest of the tweet, where @followingleads writes that this is sarcastic praise. a computer would not be able to determine this sarcasm. in addition, since tweets can be posted by anyone and have no quality or content assurance, spelling and grammar is not guaranteed to be fully accurate, so if key words were misspelt, they could be ignored or incorrectly analyzed by our algorithm. one example is the following tweet by user @komptonmusic who tweeted “i'm suddenly voting hilary [sic]” [34]. in this tweet, the user misspelt hillary clinton’s first name, which means that our searching algorithm did not pick it up at all. if the algorithm had seen the spelling error and collected the tweet, then it would have most likely been counted as an additional positive tweet for clinton. looking at related studies shows similar trends. one study by kunal jagtap [35] showed that, per tweets collected for 5 days, donald trump was tweeted about negatively more than hillary clinton was. however, other algorithms and systems that used a wider field of data came up with different results. an ai system named “mogla”, created by sanjiv rai [36], took in data from numerous social media platforms, such as google, facebook, twitter, and youtube, and used this information to come up with a conclusion. rai’s system found that people were engaging more with content related to donald trump than they were with content from hillary clinton, and in the past three elections, the candidate with more engagement ended up winning the election. from this, rai’s ai determined that donald trump would win the election over hillary clinton. in our study, we found that there was a total of 4,044,162 tweets originating from the united states that exclusively mentioned donald trump, and a total of 2,810,051 tweets originating from the united states that exclusively mentioned hillary clinton. though most predictions showed clinton winning the election, trump ended up with the victory. one of the main reasons for why twitter was not able to accurately predict these election results is because it is not a good sample of the population. results from twitter are only showing the sentiments of those who are actively using the platform. not every american uses twitter, and in addition, discussions about the candidates on twitter do not always correlate to how people vote. v. conclusion twitter is used globally by millions of users, and discussions on the social media platform can sometimes be used to see what the public’s opinion is regarding a certain issue. after collecting tweets about the us presidential election and analyzing them to determine how people viewed the two candidates, we saw that the sentiment according to twitter was somewhat accurate. if sentiment analysis algorithms are improved and further developed, they could be used to predict election results in the future. references [1] n. silver, “election update: the how-full-is-this-glass election,” last modified november 2, 2016, http://fivethirtyeight.com/features/election-update-the-how-full-isthis-glass-election/ [2] “what happens if the us presidential election is close?” reuters, last modified november 4, 2016, http://www.hindustantimes.com/world news/factbox-what-happens-if-the-u-s-election-is-close/storyzpb0gvmqtvrkakqtcymzik.html [3] j. uscinski, “the real presidential election is in december when the electoral college votes,” last modified november 7, 2016, http://www.miamiherald.com/opinion/op-ed/article113171073.html [4] s. reich, “what will the us presidential election mean for europe?” http://blogs.lse.ac.uk/europpblog/2016/11/01/trump-clinton-natoeurope/ [5] “number of monthly active facebook users worldwide as of 3rd quarter 2016 (in millions)” https://www.statista.com/statistics/264810/number-of-monthly-activefacebook-users-worldwide/ [6] “twitter usage / company facts” twitter inc., https://about.twitter.com/company [7] “number of daily active snapchat users from march 2014 to june 2016 (in millions)” https://www.statista.com/statistics/545967/snapchat-app-dau/ [8] “number of monthly active instagram users from january 2013 to june 2016 (in millions)” https://www.statista.com/statistics/253577/number-of-monthly-activeinstagram-users/ [9] “getting started with twitter,” twitter inc., https://support.twitter.com/articles/215585 [10] “posting a tweet,” twitter inc., https://support.twitter.com/articles/15367 [11] h. kwak, c. lee, h. park, s. moon, “what is twitter, a social network or a news media?” acm pp. 591-600, 2010. [12] “number of monthly active twitter users worldwide from 1st quarter 2010 to 3rd quarter 2016 (in millions)” https://www.statista.com/statistics/282087/number-of-monthly-activetwitter-users/ [13] b. pang and l. lee, “opinion mining and sentiment analysis.” foundations and trends in information retrieval, vol 2, pp. 1-135. 2008. [14] a. go, l. huang, and r. bhayani, “twitter sentiment classification using distant supervision.” the stanford natural language processing group, 2009, https://wwwcs.stanford.edu/people/alecmgo/papers/twitterdistantsupervision09. pdf [15] a. pak and p. paroubek, “twitter as a corpus for sentiment analysis and opinion mining.” lrec, vol. 10, pp. 1320-1326. 2010. [16] “big {political} data,” https://www.isidewith.com/polls [17] “introduction” twitter 4j. http://twitter4j.org/en/index.html [18] “us zip codes database,” simplemaps, http://simplemaps.com/data/us-zips [19] sentiwordnet, http://sentiwordnet.isti.cnr.it/ [20] “hashtags for #election2016 in instagram, twitter, facebook, tumblr, ello,” top-hashtags. https://tophashtags.com/hashtag/election2016/ [21] @collectedn. twitter post. november 3, 2016, 6:45 pm. https://twitter.com/collectedn/status/794309570759946241 [22] @thetablenz. twitter post. november 3, 2016, 11:24 pm. https://twitter.com/thetablenz/status/794379871342313472 [23] “the big issues of the 2016 campaign,” five thirty-eight, last modified november 19, 2015. https://fivethirtyeight.com/features/year-ahead-project/ [24] “top voting issues in 2016 election,” pew research center, last modified july 7, 2016. http://www.people-press.org/2016/07/07/4top-voting-issues-in-2016-election/ [25] “problems and priorities,” polling report. polls conducted september 5-8 and october 28-november 1. http://pollingreport.com/prioriti.htm [26] “presidential election results,” the new york times, http://www.nytimes.com/elections/results/president [27] “what is the difference between the winner-takes-all rule and proportional voting, and which states follow which rule?” frequently asked questions, national archives and records administration, https://www.archives.gov/federal-register/electoral-college/faq.html [28] “maine results,” the new york times, http://www.nytimes.com/elections/results/maine [29] unites states census bureau, https://www.census.gov/quickfacts/table [30] “immigration at front of 2016 presidential race,” usa today, http://www.usatoday.com/story/news/politics/elections/2015/05/15/im migration-2016-presidential-race/27360717/ [31] @buttonlol. twitter post. october 23, 2016, 7:42 pm. https://twitter.com/buttonlol/status/790337640222560256 [32] @followingleads. twitter post. november 3, 2016, 7:58 pm. https://twitter.com/followingleads/status/794328035071705088 [33] @komptonmusic. twitter post. november 3, 2016, 7:51 pm. https://twitter.com/komptonmusic/status/794326138742972416 [34] k. jagtap, "2016 usa presidential election twitter sentiment analysis & topic modelling." linkedin pulse, april 5, 2016. accessed december 5, 2016, https://www.linkedin.com/pulse/2016usa-presidential-election-twitter-sentiment-analysis-jagtap [35] a. kharpal, “trump will win the election and is more popular than obama in 2008, ai system finds.” cnbc, october 28, 2016. accessed october 28, 2016. http://www.cnbc.com/2016/10/28/donald-trump-will-win-theelection-and-is-more-popular-than-obama-in-2008-ai-systemfinds.html microsoft word cusj 17.docx ________________________________________________________________________________ cusj 2019 __________________________________________________________________________________________ 17 abstract — in this experiment, we create and evaluate graphite-based 3d printing composite filament in hopes of determining its effectiveness as a conductive or semiconductive 3d printing material. with 3d printable materials becoming more relevant in materials construction, the development of a 3d printable transistor would be an invaluable invention. our results are from experiments we ran in our 3d printing lab with custom-made filaments using graphite powder in addition to the standard (non-conductive) pla. these results show that is indeed possible to create graphite infused filament with conductive properties. carbon is also a potential substitute for a semiconductor when in its single-layered graphene form. however, the attributes of the filament might not make a suitable replacement for conductive filament. these results show that graphite might have a place in a 3d printable transistors for the future. i. introduction with 3d printing becoming more popular, the need for a 3d printable transistor rises. being able to have a semi-conductive filament could lead to this development. the applications for a 3d printable transistor would impact space exploration the most. if astronauts could create any parts or devices they need in space, the need to bring extra parts would disappear. in turn, rockets would weigh less because the engineers would not have to pack copious back up devices. they would merely bring the filament needed to construct all the desired materials. companies like nasa and spacex have already put research into making a 3d printer that is functional in a zero-gravity environment. a 3d printer is the only equipment necessary to utilize the semi-conductive filament. therefore, further investment into the machines necessary to utilize this filament would not be required. thus, the 3d printable transistor would be available for use immediately, which helps cut down time to market. before discussing how the filament is created, some terminology that will be used in the paper must be explained as prerequisite knowledge. first, the most important is valence band. all atoms have what is known as a valence band, which is the highest energy level occupied by electrons. figure 1. this is the subatomic structure for a silicon atom. it has the first two energy levels filled, the blue and red bands. the green band is the valence band. it needs 4 electrons to be filled. silicon atoms have a unique subatomic structure that contains 14 electrons (figure 1). its valence band is indicated by green in figure 1. every atom wishes to either remove or fill the valence band. if the atom is closer to filling its valence band than losing it, it will try to get more electrons to fill it. otherwise, it will give away those electrons when it bonds with another element. the number of electrons needed to fill the valence band is determined by which band is being filled. silicon in specific needs eight electrons. this means silicon could either lose or gain four electrons. it is this property that makes silicon semi-conductive and a primary component in transistor fabrication. the experiments performed replaced silicon with graphite as our semi-conductive component. graphite is primarily made of carbon, which is in same periodic column as silicon. this means it possess a similar valence band structure to silicon. it is for this reason that many other labs have done research into using pla combined with graphite[2-3]. second, rho and resistivity are both important words to know. resistivity is a value that is unique to each material. it denotes the resistance of that material relative to its area and length. rho is used to denote resistivity in the equation (1). a stands for area, l is short for length, and r stands for resistance. length and area are the dimensions of the material being used. resistance determines how much electrical current will be prevented from flowing through a material when attached to an electrical source. (1) r = (ρ*l)/a third, pla is a type of plastic used in 3d printing. it is an important material used in the experiments. pla beads were used in the research. all filaments are made using beads. ii. methods experimental design this experiment utilized both qualitative and quantitative data. the first few trials used graphite and pla that was measured based on their mass. the last trials the materials were measured out based on volume instead of mass. in both the case of mass and volume, a set ratio was picked between graphite and pla before they were mixed and placed inside the filament extruder. the temperature and motor speed for each data point was picked and then kept constant during the filament extrusion. after the filament was extruded, qualitative analysis of the filament was taken before measuring its quantitative features. the quantitative data consisted of brittleness, mass, resistance, and conductivity. graphite based semi-conductive 3d printing filament matthew mann rutgers university ccr 240* *this is not the author’s present address cusj 2019________________________________________________________________________________ __________________________________________________________________________________________ 18 measurements/calculations the device used to make filament was a desktop filament maker as seen below. the structure of a desktop filament maker consists of a hopper for transferring materials into the device and a heating element for melting the plastic (figure 2). figure 2. diagram of a filament extruder. the device used to measure brittleness was a protractor. filament would be bent until it broke. the angle at the point of breaking determined how brittle a sample was. the device used to measure conductivity and resistance was a generic multimeter. the multimeter was set to continuity and had alligator connectors attached to the probes. the other end of the alligator connectors was connected to the filament sample. this read the resistance with an error of 0-1 ohms, which was caused by resistance of the alligator connectors. iii. results silicon could not be used in filament creation because it hardens when heated. during the first trials, larger sample sizes were used and measure the materials using a mass balance. we were able to find a ratio, based in mass, that created filament that was conductive. each test run was made with a ratio between pla and graphite. for example, one test run was made with 2 parts pla and the ratios tested were 2:1, 1:1, and 2:3. table 1 lists the actual amounts that were used. table 1. this table describes different trials. the first column describes the ratio between column two and three. the fourth column says if the filament produced was conductive. ratio pla(g) graphite(g) conductive 2:1 35.283 17.644 no 1:1 35.045 34.519 no 2:3 28.345 42.297 yes although the 2:3 ratio produced conductive filament it did not create it in a consistent manner. it also caused the extruder to clog with graphite. i observed that the reason this occurred was because larger amounts of pla and graphite. in other words, the concentration was not kept consistent once the materials entered the chamber of the desktop filament extruder. it was determined that the best way to handle this was to start measuring the plastic and graphite by volume instead of mass. this helps determine how much material needs to be added to the extruder to fill the chamber. also, the size of the samples was made smaller as well. this prevents pockets of graphite and pla from forming more frequently. the next five tests taken with these changes were better and produced consistent results. the data for these experiments can be seen below. the first two columns represent a ratio between the pla and graphite. the cumulative volume of this ratio had to be around 2-3 tsps for the results to be consistent. table 3 describes the ratios of pla and carbon used in the first two columns. the third column describes the temperature the materials were heated at while they were pushed through the filament extruder. column four is the brittleness of the filaments produced. this quality was measured by bending the filament until it broke. the value is how many degrees the filament bent until it broke (table 3). the sixth column is the conductivity of the samples. these results are indicated by a n(no) or a y(yes). the last column is the speed at which the filament was pushed out. these values range from 0-100 and indicate the percent of power being used. table 2. this table describes more trial runs. cond. is an abbreviation for conductivity. temp. is an abbreviation for temperature. pla carbon temp. (c) brittle. (°) cond. motor speed 1 1 195 35 n 5 1 2 195 18 n 5 1 2 195 18 y 10 2 5 195 12 n 5 1 3 175 2 y 5 qualitative observations were excluded from table 3, so i could delve into more detail here. the first, second, and fourth samples produced a filament whose texture was consistent. they were brittle and reflected some light. all samples were a black color, which was caused by the graphite. the third and fifth trials consistently produced conductive filament, but only in bursts. each successful result would be interrupted by a batch of hot mushy filament. mushy in this context is supposed to refer to a filament that does not take the shape of filament. rather it is more like a non-newtonian fluid, which in this case could also be compared to ketchup. this is why during the fifth experiment the temperature was turned down. this allowed those short bursts of non-conductive filament to retain a filament shape. the decrease in temperature had no effect on the conductive samples that were produced. it is important to note that regardless of the ratio placed inside the mixer, any conductive filament that was made had the same appearance. all of them were extremely smooth and brittle (figure 3). figure 3. conductive graphite filament sample. in table 3, the data shows that each sample had a different resistance and mass value. this data and two ________________________________________________________________________________ cusj 2019 __________________________________________________________________________________________ 19 key equations can be used to prove the difference in atomic structure between samples. the following explains how reached this conclusion. each equation has a reference number next to it for easy reference. the first equation was mentioned back in the introduction as equation (1). first the density of the sample is determined (2), m represents mass, l represents the length, and a represents the cross-sectional area. (2) ! = $ % × ' => ) = $ * × ' equation (2) is arranged so that length is found, so this is can be combined with and substituted into equation (1) to obtain (3). (3) + = , × $* × '=> . / = , × $* × 0 we will use a variant of (3) with a generalized subscript i representing each sample of table 3, named (4). (4) ./ = ,1 × $1 *1 × 01 now two versions of equation (4), assuming cross sectional area is constant due to the nature of the experiment, can be created with substitutions from table 3: (5) ,× $*× 0= ,2 × $2 *2 × 02 now (5) can now be manipulated to create (6). (6) 0× $2 02 × $= ,× *2,2 × * most samples are of a similar volume and similar mass. thus, we can assume the density of all samples are equal. line (7) represents this assumption. (7) *2 *= 1 using the above we can create the equation (8), (8) 0× $2 02 × $= ,2, which provides a way to study the percent difference of the resistivity of the samples rho values, against some base sample, here sample 1 of table 3. the percent difference was calculated relative to sample one, which is why sample 1 does not have a value associated with the fourth column of table 3. the results are shown in table 3, shown to the right. table 3. the table shows the resistance and mass for each piece of conductive filament. the third column is the percent difference between the resistivity of the sample 1 and the sample in the corresponding row. samples resistance mass % diff 1 59 0.104 n/a 2 80 0.146 4 3 184 0.366 11 4 443 0.327 139 5 352 0.27 129 6 123 0.373 42 7 179 0.288 9 8 165 0.776 63 9 172 0.243 25 10 97 0.288 41 11 184 0.223 45 12 85 0.298 50 13 86 0.263 42 14 63 0.237 53 15 87 0.284 46 16 137 0.218 11 17 374 0.211 212 18 76 0.234 43 19 37 0.167 61 20 64 0.202 44 21 43 0.206 63 22 84 0.173 14 23 67 0.217 46 24 107 0.197 4 25 135 0.206 16 along with quantitative analysis, table 3 helps confirm the earlier assumption that the density between filament samples is the same. a visual assessment of the filaments in table 3 confirms that the length and mass have a linear relationship. this further supports the idea that the densities are the same because the linear relationship between length and mass is consistent with the density equation, (3). furthermore, the seventh and tenth filament have the same mass, but different resistances. the group has included a picture to demonstrate the similar volumes. figure 4. a picture comparing the length of two conductive graphite filament samples. visual confirmation shows that they are similar in size. note: the bottom piece was broken after having the mass and resistance measured. the top piece is sample 10. the bottom one is 8. these two samples are clearly of similar length. despite their similar volume their resistances are dissimilar. carbon makes up several different materials such as graphite and diamond. however, even though both of the substances are made up of carbon, they have vastly different properties. the reason the two materials have different properties is because of the carbon atom structure. cusj 2019________________________________________________________________________________ __________________________________________________________________________________________ 20 figure 5. atomic structure of graphite. the atoms form layers of graphite atoms. figure 5 gives a visualization of the carbon atom structure in graphite. this structure is known to be conductive. the relevance of this information is that it offers insight into the difference between conductive filament samples. based on this evidence, it is reasonable to believe that the less conductive samples stray further away from the traditional graphite structure. the cause could be from either the inclusion of non-conductive plastic or the way graphite and pla is mixed. the first cause needs no explanation. the second cause however could use some more detail. when the graphite is mixed in a heated chamber and pressed through a tight hole. it is possible that the atoms rearrange into a structure that while similar to the original graphite structure, are still different. other tests and setups confirm that graphite’s structure within pla can change depending how it is pushed through [2-3]. this could lead to a different resistivity, which explains the percentage difference in certain samples. iv. discussion the results show that semi-conductive graphite filaments are possible. however, the filament can only be made in short bursts. usually the extruder ends up clogging, which stops the filament from being made. the clogging is caused by the limitations of the filament extruder. the device is not powerful enough to properly mix or push the high graphite to pla ratio without clogging. the only way for the filament to be made and not clog the extruder would be for the filament to be followed by non-conductive filament. the latter provides the pressure necessary to push the former through the extruder. in addition, it is difficult to maintain the proper concentration of pla and graphite. even if using small amounts of materials at a time were 100% effective at maintaining the ratio, it still would require the constant attention of the user to achieve the desired results. v. conclusion overall, the results of these experiments help verify the difficulties of using graphite as a silicon substitute in 3d printing. in the future, it might be prudent to find a way to more effectively mix the components before pushing them through the extruder nozzle. this would allow larger amounts of pla and graphite to be added at once and the consistency issue with the produced filament. various experiments have confirmed it is possible to create consistent results or improve certain qualities with different setups or procedures [1,6-7]. author information corresponding author 20 vliet drive hillsborugh, nj 08844 (908) 448-9458 author contributions conceived and designed analysis, collected data, contributed data, performed analysis, wrote paper funding sources none acknowledgments advisor: prof. d.p.birnie, iii revisor/editor: steven coulter references [1] goncalves, j; p lima; b krause; p potschke; u lafont; jr gomes; cs abreu; mc paiva; ja covas: electrically conductive polyetheretherketone nanocomposite filaments: from production to fused deposition modeling. polymers. 2018; 10(8). doi: 10.3390/polym10080925. [2]. jia, yc; h he; y geng; b huang; xd peng: high through-plane thermal conductivity of polymer based product with vertical alignment of graphite flakes achieved via 3d printing. composites science and technology. 2017; 145:55-61. doi: 10.1016/j.compscitech.2017.03.035. [3] kim, m; jh jeong; jy lee; a capasso; f bonaccorso; sh kang; yk lee; gh lee: electrically conducting and mechanically strong graphene-polylactic acid composites for 3d printing. acs applied materials & interfaces. 2019; 11(12):11841-8. doi: 10.1021/acsami.9b03241. [4] lebedev, sm; os gefle; et amitov; dy berchuk; dv zhuravlev: poly(lactic acid)-based polymer composites with high electric and thermal conductivity and their characterization. polymer testing. 2017; 58:241-8. doi: 10.1016 j.polymertesting.2016.12.033. [5] mortazavi, b; f hassouna; a laachachi; a rajabpour; s ahzi; d chapron; v toniazzo; d ruch: experimental and multiscale modeling of thermal conductivity and elastic properties of pla/expanded graphite polymer nanocomposites. thermochimica acta. 2013; 552:10613. doi: 10.1016/j.tca.2012.11.017. [6] shemelya, c; a de la rosa; ar torrado; k yu; j domanowski; pj bonacuse; re martin; m juhasz; f hurwitz; rb wicker; b conner; e macdonald; da roberson: anisotropy of thermal conductivity in 3d printed polymer matrix composites for space based cube satellites. additive manufacturing. 2017; 16:18696. doi: 10.1016 j.addma.2017.05.012. [7] zhu, dc; yy ren; gx liao; sl jiang; fh liu; jj guo; gj xu: thermal and mechanical properties of polyamide 12 graphene nanoplatelets nanocomposites and parts fabricated by fused deposition modeling. journal of applied polymer science. 2017; 134(39). doi: 10.1002/app.45332. ©2024 li, darwin. this is an open-access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. ascending arousal network connectivity in disorders of consciousness: a diffusion mri study darwin li1 1upper canada college, toronto, ontario, canada keywords: diffusion-weighted imaging, neuroimaging, ascending arousal network, disorders of consciousness, mri abstract: the human brain gives rise to a great variety of conscious experiences. patients with disorders of consciousness, a state characterized by the dissociation of awareness and wakefulness, are particularly noteworthy. this study attempts to find key biomarkers of the disorder of consciousness state and discover key regions of the brain that govern consciousness. the focus is on the ascending arousal network—a network of nodes and edges representing connections from the brainstem to subcortical (thalamus, hypothalamus, basal forebrain) nuclei and reaching the cerebral cortex. previous studies using animal models have demonstrated a high prognostic value of the ascending arousal network in relation to consciousness. this study conducts a diffusion tensor imaging analysis and generates a tract count plot to illustrate differences in connectivity between (n=6) healthy controls and (n=6) patients with chronic disorders of consciousness. each region of interest was isolated to investigate its specific role and impact on consciousness. a principal component analysis was performed to assess the separability of the two cohorts. the results found each of the regions of interest to be significantly (p<0.05) disrupted in patients with disorders of consciousness. they contributed equally to the linear separability of the two cohorts. this is consistent with previous research and hints at the importance of the ascending arousal network in governing consciousness. these changes are likely associated with the many pathological deteriorations associated with an impaired cognitive state, such as neuronal loss, gliosis, and the degeneration of white matter tracts that connect critical areas of the brain involved in consciousness. introduction the mechanisms of consciousness are characterized by the dynamically changing interplay of neural activities underpinning two factors: human awareness and wakefulness.1 a myriad of previous studies have attempted to isolate the precise neurobiology and function of the human brain that governs consciousness. yet, given the brain is an immensely complex system, this realm of neuroscience remains poorly understood. an increasingly promising biomarker of the conscious state is the structural connectivity within the ascending arousal network (aan). knowledge of the aan used to be largely based on past studies using animals with experimentally induced lesions.2,3,4 recent research reveals similarities between the human aan and that of animals. a mapped neuroanatomic connectivity of the human aan revealed definite nuclei connections that mediate wakefulness. the aan is now believed to consist of neuronal projections from the brainstem that extend to subcortical (thalamus, hypothalamus, basal forebrain) nuclei before reaching the cerebral cortex. thalamic pathways are a key relaying center through which brainstem neurons pass to the cerebral cortex, and the thalamus helps to integrate and modulate these nuclei interactions.5 columbia undergraduate science journal vol. 18, 2024 li the hypothalamus, along with the basal forebrain further mediates wakefulness by regulating various autonomic functions and circadian sleep-wake cycles. in the end, the aan is a network of nodes representing gray matter nuclei, beginning with arousal (wakefulness) pathways in the brainstem and connecting to awareness networks in the cerebral cortex.5,6,7 the human brain gives rise to a great diversity of conscious experiences based on the varying extents of awareness and wakefulness. in this spectrum, patients in pathological or pharmacological coma experience the least conscious state. they are characterized by a loss of both awareness and wakefulness to the surrounding environment, usually as a result of severe brain injuries.1 among those who survive, 44-45% remain in a disorder of consciousness (doc), such as a vegetative state (vs) or minimally conscious state (mcs).8 these are unique instances where awareness and wakefulness are disassociated. patients with doc have their eyes wide open and are awake but are not aware of and cannot voluntarily interact with their surrounding environment.1 this is supported by several functional mri-based approaches, which found preserved cognitive processing and willful modulation of brain activity in doc patients, indicating that behavioral unresponsiveness might not always equate to a complete absence of cognitive function.9,10 this unique dissociation of awareness and wakefulness allows studies to isolate awareness from wakefulness and identify specific correlates of each. for instance, there has been a postulation that the default mode network is compromised in doc patients due to a loss of awareness while structures integral to wakefulness appear to be maintained.11 the prognostic importance of the aan has become particularly evident in comatose patients. acute damage to the brainstem is a hallmark of these patients, likely a consequence of a lesion centered in the upper pons or in the midbrain which disrupts aan pathways.12 consistent with this, traumatic coma has met the criteria for a disconnection syndrome with a complete disconnection of brainstem arousal nuclei from previously mentioned subcortical nuclei.7,8 in addition, connectivity between the temporal lobe and the medial thalamus and rostral brainstem seems to be linked to the loss of consciousness associated with temporal lobe epilepsies.13 a disassociation of awareness and wakefulness is detected during seizures, although this is a much more transient example.1 here, we further test the aan as a regulator of consciousness using a cohort (n=6) of patients suffering from chronic doc following traumatic brain injury (tbi) and comparing them with a cohort of (n=12) healthy controls. we use previously published high-angular resolution diffusion imaging (hardi) data to evaluate aan connectivity in each cohort.8 hardi captures the architecture of white matter fibers in the brain and offers a detailed representation of neural pathways including key intersecting and diverging regions. with the hardi data, we conducted diffusion tensor imaging (dti) and tractography analysis to determine differences in aan connectivity between patients with chronic post-traumatic doc and healthy controls. finally, we performed a principal component analysis (pca) to highlight the axes accounting for the most variance between the cohorts. methods patients previously published hardi data of six patients with chronic doc was obtained from openneuro dataset 003367.14 among them, three were diagnosed with vs and three with mcs on the coma recovery scale-revised (crs-r, total = 6.5) following tbi. similarly, hardi data of 12 healthy controls with no recorded psychiatric or neurological disorders were attained. details of the hardi sequence can be found in previous research.14 note specifically the 2 mm isotropic resolution, 60 diffusion-encoding directions, and contrast columbia undergraduate science journal vol. 18, 2024 li of b = 2,000 sec/mm2. processing all preprocessing was done using fsl and mrtrix3. brain extraction (0.2 thresholds), eddy current, and bulkhead motion correction were performed. linear registration was used to transform each patient’s diffusion space to montreal neurological institute (mni) 152 t1 1 mm space. to guide the tractography, fiber response functions were estimated using the tournier algorithm, and fiber orientation distributions (fods) were computed. a white matter mask was also generated to confine the whole-brain tractography. finally, diffusion tensor imaging (dti) parameters were calculated specific to the entire brain and aan regions of interest (rois). metrics such as fractional anisotropy (fa), mean diffusivity, axial diffusivity, and radial diffusivity were derived. regions of interest a combination of previously published rois, the harvard ascending arousal network atlas, and the harvard-oxford cortical and subcortical structural atlases included within fsl were used.14 the thalamus, hypothalamus, and basal forebrain were taken from publicly available rois. the cortical lobes were generated from the harvard-oxford cortical and subcortical structural atlases, and brainstem rois from the harvard ascending arousal network atlas. all rois matched the mni 152 t1 1 mm space. tractography whole brain tractography was first generated, guided by a white matter mask. three million streamlines were launched and dynamic seeding based on the fods was used. then, a second tractography was created only including tracts connecting aan brainstem rois to the thalamus, hypothalamus, and basal forebrain. here, a threshold of 100 mm was set as the maximum length of streamlines. otherwise, default parameters were used. finally, a voxel-wise map representing the number of streamlines passing through each voxel in the aan tractography was generated, normalized by the total number of streamlines launched. the exact path of the streamlines within each voxel was accounted for and the results were set to unsigned 32-bit integers. finally, tractspecific analyses of dti parameters were performed, and the benjamini-hochberg method was used to control false discovery rates. principal component analysis to assess the separability of chronic doc patients from healthy controls, a pca was conducted in r studio with connectivity in each aan roi as the dimensions. principal components 1 (pc1) and 2 (pc2) were projected on the same axes along with the eigenvectors for each roi to explain the major patterns in the data while preserving the maximum variance. the associations of each roi with pc1 and pc2 were determined by computing their respective loadings using the formula l = v ⋅√λ, where l is the loading, v is the component of the eigenvector of the roi on the axis of the pc, and λ the eigenvalue of the pc. results to visualize aan connectivity between controls and chronic doc patients, we generated normalized tract plots of streamlines passing between all subcortical aan structures. the axial and sagittal views are shown (figure 1). figure 1. differences in aan connectivity between healthy controls and patients with chronic disorders of consciousness. columbia undergraduate science journal vol. 18, 2024 li connectivity was measured by the group-sum value of the number of streamlines between all subcortical aan structures (brainstem to thalamus, hypothalamus, and basal forebrain), normalized by the total number of streamlines launched (3m). with tracts from the brainstem extending fully into the thalamus and hypothalamus regions, controls exhibit much more robust and widespread connectivity than chronic doc patients. the latter’s lack of tracts indicates greatly reduced aan connectivity, with narrower areas of activation, fewer pathways, and reduced communication between all aan rois. to determine connectivity in each aan roi, a boxplot was created with the tract count of each control and doc subject (figure 2). in all of the hypothalamus, basal forebrain, thalamus, and brainstem regions, we observed that the normalized tract counts appear to be significantly (p<0.05) lower in the chronic doc cohort compared to the controls. figure 2. fa values in aan rois of healthy controls and patients with a chronic doc, p<0.05. this whiskers-only box plot shows connectivity, measured by the fractional anisotropy values obtained from diffusion tensor imaging analysis. the differences in all four box plots are significant for the threshold p<0.05. in all four box plots and for both cohorts, the extended whiskers indicate a wide distribution of data. for healthy controls, the variability could reflect natural differences in brain architecture and connectivity that occur within a normal population. on the other hand, the variability in the chronic doc group likely represents the diverse impacts of various pathologies associated with disorders of neural connectivity. the box plots for the hypothalamus and thalamus show a notably lower median tract count in chronic doc patients compared to controls, with the median line of the doc group visibly closer to the first quartile than the third. this shift in the median indicates that the majority of chronic doc patients have a lower connectivity profile compared to the median of the control group. furthermore, the presence of a wider range in the chronic doc group’s data suggests that some patients retain a degree of connectivity closer to healthy norms, while others are significantly more affected. finally, we performed pca with the aan roi tract count values to assess the separability of chronic doc patients from healthy controls (figure 3). figure 3. principal component analysis of structural connectivity among aan rois in healthy controls and patients with disorders of consciousness. the vectors represent the loadings of each aan roi onto the principal components. the first principal component explains 94.09% of the variance in the data while the second explains 2.89%. the third and fourth principal components account for 2.43% and 0.58%, respectively. the orange dots represent chronic doc patients while blue are the controls. pc1 = principal component 1, pc2 = principal component 2. with the controls clumped on the right y-axis and doc patients spread out on the left, the results demonstrate clear linear separability between these two cohorts. the diffuse nature of the doc patient data, in contrast to columbia undergraduate science journal vol. 18, 2024 li the controls, mirrors the wide range of the box plots. the dominance of pc1, explaining 94.09% of the total variance, suggests structural connectivity differences within the aan heavily contribute to the total variance in the data. furthermore, we calculated the absolute loadings of each aan roi to determine each roi’s influence in separating chronic doc patients from healthy controls (table 1). roi pc1 loading pc2 loading hypothalamus 0.9580 0.2833 thalamus 0.9906 -0.0333 basal forebrain 0.9634 -0.1686 brainstem 0.9684 -0.0782 table 1. contributions of each aan roi to pc1 and pc2. loadings of pc1 and pc2 for aan rois were extracted from the pca results and computed using the formula l = v, where l is the loading, v is the component of the eigenvector of the roi in the axis of the pc, and the eigenvalue of the pc. higher absolute loadings indicate a stronger contribution of the roi to the respective pc. the similar loadings on pc1 demonstrate approximately equal contributions from each roi. it follows that connectivity within the aan in doc patients is very distinct from healthy controls, with the former exhibiting significantly fewer tracts and experiencing crucial disruptions in all aan rois. on pc2, the hypothalamus is notably higher than all other rois. this holds little weight given pc2 accounts only for 2.89% of the total variance. the importance of pca is simply its demonstration that there is clear separability between the two cohorts based on aan connectivity, with each aan roi contributing relatively equally to this variance. discussion this study demonstrated that connectivity within the aan is significantly disrupted in the chronic doc cohort and that this separates these patients from the healthy controls. great disruption in all subcortical structures (hypothalamus, basal forebrain, thalamus) and the brainstem was found. these results are consistent with previous research analyzing connectivity differences between these groups, as well as studies aiming to find neural correlates of consciousness. the disruptions observed in the rois within the aan in patients with doc in this study can be attributed to the specific pathophysiological changes associated with chronic doc states. in chronic doc, the brain often undergoes significant structural changes that can lead to impaired connectivity. these changes may include neuronal loss, gliosis, and the degeneration of white matter tracts that connect critical areas of the brain involved in consciousness.15 this disruption is determined by the decrease in the number of intact white matter tracts, which could be attributed to neural damage or degeneration caused by the underlying pathology of chronic doc. tract counts refer to the number of neural connections, or white matter tracts, that are identified between the different regions of the brain. these tracts are made up of axons, long, thread-like parts of a nerve cell where impulses are conducted from the cell body to other cells.16 it follows that tracts are pathways for neural signals to travel and communicate between. higher tract counts have been found to be correlated with better cognitive and functional outcomes. furthermore, tract counts are a common biomarker to assess the integrity of the brain's white matter. the wide distribution of tract counts within both groups, but especially among chronic doc patients, raises important questions about the heterogeneity of the disorder. in doc patients, this variability may be due to differences in the etiology of the condition, the extent of brain damage, the duration of the disorder, or individual variations in the capacity for neural plasticity and recovery. often, a result of cognitive impairments such as docs are axonal injuries, which are characterized by widespread damage to the white matter tracts of the brain.17 understanding the sources of this columbia undergraduate science journal vol. 18, 2024 li variability is critical for developing personalized approaches to treatment and care for doc patients. moreover, the results suggest that some chronic doc patients retain a level of connectivity that approaches that of healthy controls. this finding opens several lines of inquiry: it may indicate the potential for recovery in some individuals, or it could reflect the presence of alternate neural pathways that compensate for lost connections. further research is needed to explore these possibilities and to determine the clinical significance of these retained connections. potentially, tract counts can be considered biomarkers for the recovery of doc patients. the investigation of the aan has great value given the significant differences in connectivity in each aan roi. the hypothalamus is particularly susceptible to disruption in doc patients due to its central role in maintaining the body's internal balance and arousal states. the hypothalamus regulates autonomic and endocrine functions, including the sleepwake cycle.5 lower tract counts in the hypothalamus may be indicative of damage or dysfunction in this region, which could be a result of traumatic brain injury, or hypoxic events, in addition to a doc. such disruptions can directly impair the hypothalamus's ability to contribute to arousal mechanisms, thereby affecting the patient's ability to maintain consciousness. the specific vulnerability of the hypothalamus to disruption in doc patients could also be attributed to its extensive network of connections with other brain regions involved in consciousness, including the thalamus, basal forebrain, cerebral cortex, and brainstem. in the thalamus, lower tract counts observed in doc patients may be related to its role as the primary relay station for sensory information to the cortex. the thalamus's involvement in consciousness extends to its participation in thalamocortical loops, which are essential for cognitive functions, including awareness.5 previous research has identified the thalamus as a critical neural correlate of dexmedetomidine-induced unconsciousness, with decreased rates of glucose and cerebral blood flow.18 in addition, similar studies investigating mild cognitive impairments show a disrupted thalamus white matter anatomy.19 damage to the thalamus is quite commonly seen in neuropsychiatric disorders and could stem from a variety of causes, including a direct injury, degeneration of connecting white matter tracts, or secondary effects of cortical damage.20,21,22 the vulnerability of the thalamus in doc patients highlights the importance of intact sensory processing pathways for the maintenance of conscious awareness. the brainstem is critical for its involvement in the reticular activating system, which regulates arousal and wakefulness.5 lower tract counts in doc patients here may be because of its susceptibility to direct injury or as a consequence of secondary brain injury mechanisms such as edema or increased intracranial pressure. the brainstem's role in basic life-sustaining functions also makes its disruption common in diffuse brain injuries.12 the first limitation of this study is rooted in the static nature of the box plots and tractography, providing a visualization of connectivity but not capturing dynamic changes over time. longitudinal studies would be beneficial to understand how connectivity patterns in doc patients evolve and whether they correlate with changes in clinical status. secondly, the normalization of tract counts allows for comparison between groups but may mask absolute differences in connectivity that could be clinically relevant. approaches that involve additional analyses complementing the normalization are recommended as future studies. the implications of these findings extend beyond the scientific understanding of chronic doc. these results can inform clinical practices. from this study, the aan is a potential biomarker for the diagnosis and prognosis of doc. given this information and the role each aan roi columbia undergraduate science journal vol. 18, 2024 li plays in governing consciousness, therapeutic interventions aimed at specifically enhancing connectivity in the regions of the aan may arise. furthermore, there may also be a need for individualized treatment plans, hinted by the great variability in connectivity observed in the doc cohort. a personalized medicine approach to treat doc patients targeting specific aan connections or neural pathways may have strong potential benefits in the path to recovery. conclusion the results presented in this study contribute to a growing body of evidence that suggests structural connectivity is fundamentally altered in chronic doc. these alterations have significant implications for the clinical management of the disorder and for our theoretical understanding of consciousness. further research should first apply more rigorous statistical techniques and machine learning on a larger dataset to validate the findings of the current study and analyze the connectivity of aan rois in junctions rather than treating them as independent entities. author information corresponding author darwin li darwin.li@ucc.on.ca author contributions darwin li conceptualized the study, collected and analyzed data, and wrote the manuscript. references [1] s. laureys, the neural correlate of (un)awareness: lessons from the vegetative state. trends in cognitive sciences 9, 556–559 (2005). [2] d. b. lindsley, j. w. bowden, h. w. magoun, effect upon the eeg of acute injury to the brain stem activating system. electroencephalography and clinical neurophysiology 1, 475–486 (1949). [3] d. h. smith, et al., immediate coma following inertial brain injury dependent on axonal damage in the brainstem. j neurosurg 93, 315–322 (2000). [4] p. m. fuller, d. sherman, n. p. pedersen, c. b. saper, j. lu, reassessment of the structural basis of the ascending arousal system. j comp neurol 519, 933–956 (2011). [5] b. l. edlow, et al., neuroanatomic connectivity of the human ascending arousal system critical to consciousness and its disorders. j neuropathol exp neurol 71, 531–546 (2012). [6] g. aston-jones, s. chen, y. zhu, m. l. oshinsky, a neural circuit for circadian regulation of arousal. nat neurosci 4, 732–738 (2001). [7] b. l. edlow, et al., disconnection of the ascending arousal system in traumatic coma. j neuropathol exp neurol 72, 505–523 (2013). [8] s. b. snider, et al., ascending arousal network connectivity during recovery from traumatic coma. neuroimage clin 28, 102503 (2020). [9] a. m. owen, m. r. coleman, detecting awareness in the vegetative state. annals of the new york academy of sciences 1129, 130–138 (2008). [10] m. m. monti, et al., willful modulation of brain activity in disorders of consciousness. n engl j med 362, 579–589 (2010). [11] a. vanhaudenhuyse, et al., default network connectivity reflects the level of consciousness in noncommunicative brain-damaged patients. brain 133, 161–171 (2010). [12] j. parvizi, a. r. damasio, neuroanatomical correlates of brainstem coma. brain 126, 1524– 1536 (2003). [13] h. blumenfeld, j. taylor, why do seizures cause loss of consciousness? neuroscientist 9, 301–310 (2003). [14] s. b. snider, et al., disruption of the ascending arousal network in acute mailto:darwin.li@ucc.on.ca columbia undergraduate science journal vol. 18, 2024 li traumatic disorders of consciousness. neurology 93, e1281–e1287 (2019). [15] m. pievani, et al., assessment of white matter tract damage in mild cognitive impairment and alzheimer’s disease. hum brain mapp 31, 1862– 1875 (2010). [16] a. fornito, a. zalesky, e. t. bullmore, eds., “chapter 1 an introduction to brain networks” in fundamentals of brain network analysis, (academic press, 2016), pp. 1–35. [17] f. b. mesfin, n. gupta, a. hays shapshak, r. s. taylor, “diffuse axonal injury” in statpearls, (statpearls publishing, 2024) (march 1, 2024). [18] o. akeju, et al., disruption of thalamic functional connectivity is a neural correlate of dexmedetomidineinduced unconsciousness. elife 3, e04499. [19] t. alderson, et al., disrupted thalamus white matter anatomy and posterior default mode network effective connectivity in amnestic mild cognitive impairment. front. aging neurosci. 9 (2017). [20] h.-l. s. wang, c.-l. rau, y.-m. li, y.-p. chen, r. yu, disrupted thalamic resting-state functional networks in schizophrenia. front. behav. neurosci. 9 (2015). [21] c. schleifer, et al., dissociable disruptions in thalamic and hippocampal resting-state functional connectivity in youth with 22q11.2 deletions. j. neurosci. 39, 1301–1319 (2019). [22] d. t. jones, f. j. mateen, c. f. lucchinetti, c. r. jack jr, k. m. welker, default mode network disruption secondary to a lesion in the anterior thalamus. archives of neurology 68, 242–247 (2011). introduction results corresponding author author contributions references columbia undergraduate science journal open-access publication | http://cusj.columbia.com 26 cusjvolume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu o n c o lo g y cusjcolumbia undergraduate science journalresearch articles 27cusj volume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu o n co lo g y relationships between oncologist gender, participatory decision making, anxiety and breast cancer care kevin xu1,2,*, allyson weseley2 1columbia college, columbia university, new york, ny 10027; 2 roslyn high school, roslyn heights, ny 11577 abstract the discovery that adjuvant treatment (chemotherapy and radiotherapy) after lumpectomy contributes to high survival rates has been a groundbreaking development in modern breast cancer care. participatory decision-making (pdm) is a communication style that involves active interactions and discourse between doctors and patients. although the receipt of adjuvant therapy for various types of cancers has been found to be linked to differences in patient-physician communication, few studies have explored whether this relationship exists in breast cancer treatment. the present study, one vm�[ol�äyz[�l_wspjp[s`�[v�l_htpul�kltvnyhwopj�huk�wz`jovzvjphs�mhj[vyz�[oh[�th`�il�ylsh[lk�[v�hkq\]hu[�[ylh[tlu[�yh[lz�� surveyed a sample of inner-city breast cancer patients (n = 105) about the type of care they received, assessing concomitant levels of participatory decision making (pdm), and anxiety. the results demonstrated that patients who indicated higher pdm tended to have lower levels of anxiety (p < .01). breast cancer patients who saw female oncologists were more likely to receive adjuvant treatment than breast cancer patients who saw male oncologists (p < .05). contrary to popular belief, patients reported equivalent levels of pdm and anxiety regardless of their physician’s gender or their receipt of adjuvant treatment. the surprisingly low adjuvant treatment rates (45%) in this sample suggest possible system failures in breast cancer care and indicate that more attention needs to be given to public education on the importance of receiving follow-up treatment in breast cancer care. key words: adjuvant treatment; anxiety; gender; lumpectomy; participatory decision making a!ecting more than 1.3 million people each year, breast cancer is the second most widespread cancer worldwide and kills more than 40,000 americans annually (jemal et al., 2003). a lumpectomy (breast conservation surgery) is the most common treatment for breast cancer and consists of the removal of a localized lump of cancerous growth on the breast. adjuvant treatment is de"ned as the receipt of radiotherapy or chemotherapy after lumpectomy. clarke et al. (2005) and the early breast cancer triallists’ collaborative group’s (2005) studies were amongst the "rst to "nd that adjuvant treatment after lumpectomy not only reduced breast cancer recurrence, but also signi"cantly raised breast cancer survival rates. numerous studies (lantz et al., 2005; morris et al., 2000; national cancer policy board, 1999) have indicated that mastectomy, as opposed to adjuvant treatment, is “over-used,” which has been “labeled as a problem in terms of ‘over-treatment.’” lantz and colleagues’ (2005) study posited that the use of mastectomy in place of adjuvant treatment may be an indication of breast cancer patients’ “lack of decision involvement or control.” participatory decision-making (pdm) is a recommended communication style that involves active interactions and dialogue between doctors and patients and is indicative of high degree of patients decision involvement in their health care (cooper-patrick et al., 1999). several studies have shown that female patients have better communication with female physicians than with male physicians, leading some to speculate (hershman et al., 2008; lurie et al., 1993) whether female breast cancer patients exhibit higher pdm and lower anxiety with female oncologists than with male oncologists. past research has found that female patients undergoing breast cancer screening indicated higher pdm and felt less anxious when they had a female doctor than a male doctor (lurie et al., 1993). lurie and colleagues explained their "ndings by suggesting that female doctors at an obstetrics-gynecology clinic may be more e!ective in dealing with female patients’ emotional stress and gender-speci"c concerns pertaining to breast health. other studies have also provided evidence that female breast cancer patients copyright: © 2011 the trustees of columbia university, columbia university libraries, some rights reserved, purushothaman, et al. received jan. 2, 2011. accepted jan. 31, 2011. published april 1, 2011. *to whom correspondence should be addressed: kyx2001@columbia.edu introduction materials and methods may feel more comfortable talking with female oncologists and may feel less nervous regarding their medical condition than with male oncologists (roter et al., 1991; stevens et al., 1996). it has also been ascertained that cancer patients who report better communication with their oncologists tend to have less anxiety (meeuwesen et al., 1991; stevens et al., 1996). in an attempt to respect patients’ rights and to address their anxiety, medical communication in recent years has become progressively less “paternalistic” by encouraging patients, instead of their doctors, to make decisions about their medical care. #is increasingly patient-centered type of patient-physician dynamic has been found to contribute to increased levels of pdm (guadagnoli & ward, 1998; #ompson, 2007). given that research on determinants of the use of breast cancer adjuvant treatment remains a relatively new "eld, no research to date has examined di!erences in pdm and anxiety between breast cancer patients who received adjuvant treatment and those who did not receive adjuvant treatment. understanding pdm may give doctors an opportunity to learn about patients’ comorbidities and take patients’ personal concerns into consideration before deciding on a medical treatment. in addition, few studies have examined factors that are related to adjuvant treatment rates, perhaps due to the recent nature of clarke et al.’s (2005) "ndings on the tangible bene"ts of adjuvant treatment on breast cancer survival. we hypothesized that: 1. compared to breast cancer patients with male oncologists, patients with female oncologists would indicate: (a) higher pdm, (b) higher rates of adjuvant treatment, and (c) less anxiety. 2. compared to patients who did not receive adjuvant treatment, patients who received adjuvant treatment would indicate: (a) higher pdm and (b) lower anxiety. #e present study seeks to examine whether the gender of patients’ oncologists is related to patient anxiety, breast cancer adjuvant treatment rates, and the quality of patient-physician communication. participants during the fall of 2008, the american cancer society (acs) patient database was utilized to randomly select 240 breast cancer patients to be surveyed through the mail. #e sample for this study consisted of 105 female breast cancer patients who had been diagnosed with and treated for the disease in the past all participants n = 105 n % age group 40-54 48 45.7 55-69 41 39.0 70-90 16 15.2 race caucasian 52 49.5 african american 25 23.8 hispanic 16 15.2 asian 10 9.5 other/mixed 2 1.9 borough of residence brooklyn 76 72.4 staten island 29 27.6 breakdown of treatment lumpectomy by itself 50 47.6 mastectomy followed by chemotherapy 6 5.7 mastectomy followed by radiotherapy 2 1.9 mastectomy by itself 0 0 lumpectomy followed by chemotherapy 15 14.3 lumpectomy followed by radiotherapy 32 30.5 1 table 1 demographic and clinical characteristics of study participants o n co lo g y columbia undergraduate science journal open-access publication | http://cusj.columbia.com 28 cusjvolume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu cusjcolumbia undergraduate science journalresearch articles 29cusj volume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu o n co lo g y two years, a response rate of 44%. all participants were diagnosed with breast cancer in stages i-iii and were thus eligible for adjuvant treatment. previous studies on pdm and anxiety of breast cancer patients have used similar sample sizes (e.g., hall et al., 1993; stewart, 1984). all participants resided in the new york city boroughs of brooklyn or staten island, new york, which have some of the highest rates of advanced cancer staging in the united states (susan g. komen for the cure, 2009). all participants were members of the american cancer society who attended support groups in the past or worked with the american cancer society’s patient and family services at local hospitals. as seen in table 1, participants’ ages ranged between 40 and 90; the average age of the participants was 58 years. all participants had medical insurance and had incomes ranging from $30,000 to $60,000 per year. procedure prior to conducting this study, the procedure was approved by the institutional review board of the roslyn union free school district and also was approved by the directors of both american cancer society sites. participants’ responses in the survey were anonymous to protect their privacy. because data were collected through the american cancer society, which o!ers counseling and support services to those su!ering from cancer, patients who experienced possible stress while completing the survey had access to appropriately trained professionals. preceding data collection, a pilot study was conducted at the staten island american cancer society (acs) by the investigators in this study. "e acs sta! distributed the patient medical care questionnaire to 30 breast cancer patients involved in the acs breast cancer support programs. after completing the survey, participants were asked to give feedback on the comprehensibility of the survey’s items. because participants indicated that several items contained medical jargon that made parts of the survey di#cult to understand, the language of the patient medical care questionnaire was simpli$ed and was expert-reviewed by one of the authors of the rochester participatory decision-making scale. "e revised patient medical care questionnaire was then distributed to members of the acs sta! in brooklyn and staten island, who mailed the questionnaire to potential participants, and the researchers visited the acs sites weekly to monitor the data collection process. to increase the response rate, a stamped, addressed return envelope was provided to participants. completed surveys were mailed to the researchers between august 2008 and november 2008. materials "e $rst section of the survey, participants indicated the gender and race of their oncologists and themselves and whether they had received a lumpectomy, radiotherapy, chemotherapy, or mastectomy. because breast cancer adjuvant treatment is most commonly de$ned as adjuvant radiotherapy or chemotherapy following lumpectomy (clarke et al.. 2005; early breast cancer triallists’ collaborative group, 2005), only participants who received a lumpectomy preceding their receipt of adjuvant treatment were classi$ed as having received adjuvant treatment. in the data analysis, participants who received a mastectomy were not classi$ed as having received adjuvant treatment. in the second section of the questionnaire, participants completed a pdm and anxiety scale. to minimize response bias, items from the two scales were mixed throughout the survey. all items were answered on a 6-point bipolar scale ranging from 1 (indicating that a speci$c event “almost never” occurred) to 6 (“almost always” occurred). "e pdm scale consisted of 13 items (e.g., “"ere are times when i don’t understand my oncologist’s medical language”, see appendix). seven items were modeled after the rochester participatory decision-making scale (shields et al., 2005). we created six additional items to measure cross-cultural medical communication and delays in medical treatment since past research has noted that these are crucial components of pdm (cooperpatrick et al., 1999). six items on the pdm scale were reverse-scored. to assess the scale’s internal reliability, the researchers calculated a cronbach’s alpha. "e pdm scale had a cronbach’s alpha of .82 on this study’s sample. "e pdm scale was then scored by taking the mean of the 13 items. "e anxiety scale consisted of four items, two of which were modeled after items on the hospital anxiety and depression scale (snaith, 2003), and the other two items were created for this study. an example of an item from the anxiety scale is “even if i try not to worry, i still feel frightened about my medical condition.” "e scale had a cronbach’s alpha of .70 on this study’s sample. "e anxiety scale was scored by taking the mean of the four items after correcting for reverse-scoring. data analysis statistical tests were run on the aggregate survey scores of each patient as opposed to the scores from each question separately. a chi-square test was used to examine hypothesized di!erences in adjuvant treatment rates between breast cancer patients of male oncologists and patients of female oncologists. while this study did not hypothesize a correlation between pdm and anxiety, a correlation coe#cient was calculated to quantify the extent of association between pdm and anxiety. independent sample t-tests were used to examine the di!erences in the pdm and anxiety of patients who received versus those who did not receive adjuvant treatment. t-tests were also run to analyze the di!erences in the pdm and anxiety of patients of male oncologists and those of female oncologists. because the hypotheses were all directional, all statistical tests were one-tailed. "e cuto! for statistical signi$cance was p=0.05. statistical package for social sciences version 16.0 (spss 16.0) was utilized to analyze the data. 37 51 0 10 20 30 40 50 60 70 80 90 100 male female oncologist gender s ta n d a r d c a r e r a te ( % ) figure 1 (�z[h[pz[pjhss`�zpnupäjhu[�kpmmlylujl�^hz�mv\uk�il[^llu�[ol���vm�mlthsl�vujvsvnpz[z»�wh[plu[z��� ����������^ov�yljlp]lk�hkq\]hu[� [olyhw`�huk�[ol���vm�thsl�vujvsvnpz[z»�wh[plu[z�yljlp]lk�hkq\]hu[�[ylh[tlu[���������������_õ� ���$�������w�#���� 1 2 3 4 5 6 1 2 3 4 5 6 pdm a n x ie ty figure 2 /pnoly�why[pjpwh[vy`�kljpzpvu�thrpun�^hz�mv\uk�[v�il�zpnupäjhu[s`�jvyylsh[lk�^p[o�sv^ly�sl]lsz�vm�hu_pl[ �̀�y�$��������w�#����� columbia undergraduate science journal open-access publication | http://cusj.columbia.com 30 cusjvolume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu o n co lo g y cusjcolumbia undergraduate science journalresearch articles 31cusj volume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu o n co lo g y relationship of oncologist gender to adjuvant treatment as seen in table 1, 47 of all 105 participating breast cancer patients received adjuvant treatment (lumpectomy followed by either radiotherapy or chemotherapy). fifty of the 105 participants received a lumpectomy with no follow-up treatment. about one-half -52% (50/97) -of patients who received a lumpectomy did not receive adjuvant treatment. a chi-square test revealed that oncologist gender was associated with the likelihood of patients’ receiving of adjuvant treatment, x!(96) = 3.41, p < .05. as seen in figure 1, patients of female oncologists tended to receive adjuvant treatment at higher rates (19/37, 51%) than did patients of male oncologists (22/60, 37%). relationship of pdm to anxiety higher levels of pdm were associated with lower levels of patient anxiety. "e results of a linear correlation are seen in figure 2, which shows that patients who indicated higher pdm with their physicians tended to report lower levels of anxiety related to their medical treatment, r = -0.38, p < .01. relationship of oncologist gender to anxiety and pdm overall, participants indicated high levels of pdm with an average rating of 4.50 out of 6.00. as seen in figure 3, the results of an independent samples t-test show that patients of male and female oncologists reported similar levels of pdm, t(94) = 0.64, p =.52. as seen in figure 4, contrary to hypotheses, patients of male and female oncologists reported similar levels of anxiety t(103) = -1.42, p = .17. relationship of adjuvant treatment to anxiety and pdm contrary to predictions, an independent samples ttest indicates that there were no signi#cant di$erences in pdm between patients who received adjuvant treatment (m = 4.46) and those who did not (m = 4.53), t(94) = -0.43, p = .62 (figure 5). as seen in figure 6, patients who received adjuvant treatment reported marginally less anxiety (m = 3.71) than those who did not receive adjuvant treatment (m = 4.10), t(103) = 1.57, p = .09. only 48% of participating breast cancer patients received adjuvant treatment. researchers have found that patients from urban areas are not only 54% more likely to be diagnosed with breast cancer but also are more likely to face complications in breast cancer treatment and higher mortality than patients elsewhere, re%ective of low adjuvant treatment rates (ayanian et al. 1993; fiscella et al., 2000). "e largest group of participants who did not receive adjuvant treatment (48%, n = 51) received a lumpectomy, indicating that patients failed to receive follow-up treatment. a recent study examined disparities in adjuvant treatment and provided a potential explanation for this low adjuvant treatment rate (bickell & cohen, 2008). "ey attributed patients’ failure to obtain follow-up treatment for breast cancer to a systems failure in which oncologists fail to comply with referrals and communicate with other specialists (e.g., surgeons, radiologists) in the cancer treatment #eld. as expected, breast cancer patients of female oncologists were more likely to receive adjuvant treatment than patients of male oncologists. recent literature has demonstrated that female oncologists may be more pro#cient at communicating with their fellow medical professionals of either gender, ranging from nurses to specialists (levinson & lurie, 2004). "us it may be that female oncologists’ greater emphasis on teamwork and cooperation with their peers may contribute to fewer systems failures and therefore higher rates of follow-up treatments. given the study’s correlational nature, it is possible that female and male oncologists, in fact, recommend adjuvant treatment at equal rates. breast cancer patients may be adhering to physician recommendations more frequently if their physician is a female, as opposed to male. alternatively, previous studies have noted that female oncologists tend to be younger than male discussion results 4.56 4.46 1 2 3 4 5 6 male female oncologist gender p d m figure 3 5v�zpnupäjhu[�kpmmlylujlz�^lyl�mv\uk�il[^llu�[ol�tlhu�7+4�zjvylz�vm�thsl�vujvsvnpz[z»�wh[plu[z��4�$�������huk�mlthsl�vujvsvnpz[z»� wh[plu[z��4�$��������[� ���$�������w�$���� 3.72 4.07 1 2 3 4 5 6 male female oncologist gender a n x ie ty figure 4 5v�zpnupäjhu[�kpmmlylujlz�^lyl�mv\uk�il[^llu�[ol�hu_pl[`�vm�thsl�vujvsvnpz[z»�wh[plu[z��4�$�������huk�[ol�hu_pl[`�vm�mlthsl�vujvsvnpz[z»�wh[plu[z��4�$��������[������$��������w�$����� 4.534.46 1 2 3 4 5 6 yes nostandard care p d m figure 5 5v�zpnupäjhu[�kpmmlylujlz�^lyl�mv\uk�il[^llu�[ol�7+4�vm�wh[plu[z�^ov�yljlp]lk�hkq\]hu[�[ylh[tlu[��4�$�������huk�wh[plu[z�^ov�kpk� uv[�yljlp]l�hkq\]hu[�[ylh[tlu[��4�$��������[� ���$���������w�$����� columbia undergraduate science journal open-access publication | http://cusj.columbia.com 32 cusjvolume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu o n co lo g y 33cusj volume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu cusjcolumbia undergraduate science journalresearch articles oncologists and may be more familiar with new developments in breast cancer treatment that emphasize the importance of adjuvant treatment(lurie et al., 1993; mechanic et al., 2001). because the present study gathered data directly from patients who did not know their doctors’ ages, this variable was not examined. however, despite controlling for oncologist age, one study (cooperpatrick et al., 1999) found that female oncologists are more likely to o!er breast cancer adjuvant treatment than their male counterparts. still, to investigate this possibility in future studies, data should be gathered on oncologists’ ages. as expected, participants who indicated high levels of pdm tended to report less anxiety. a wealth of research (e.g., cooper-patrick et al., 2008; stevens et al., 1996) has found that patients who have the opportunity to actively participate with their physicians in their health care tend to feel less anxious about their medical condition. "is #nding challenges the “paternalistic” model of patient-physician communication by highlighting a potential emotional bene#t of pdm. however, it is important to note that this #nding is correlational. whereas most research suggests that pdm leads to lower anxiety, it is possible that high anxiety leads to lower pdm or that both pdm and anxiety are confounded by a third factor such as breast cancer patient age. contrary to predictions, patients of female oncologists indicated equivalent levels of anxiety as patients of male oncologists. breast cancer patients experience a great deal of anxiety and believe that their oncologists’ technical skill is the most important aspect of their medical care (stevens et al., 1996; wiggers, et al., 1990). despite any di!erences patients may perceive between male and female oncologists, they may view them similarly in terms of technical skill, contributing to similar anxiety ratings. in addition to #nding no di!erences in pdm between patients who received adjuvant treatment and those who did not, we also found that patients of female oncologists indicated equivalent levels of pdm as patients of male oncologists. female patients in general tend to ask their oncologists more questions and exhibit more interest in their medical treatment than do male patients (elderkin"ompson & waitzkin, 1999). however, it has been suggested that oncologists, regardless of their genders, may be sensitive to female breast cancer patients’ desire for a highly communicative style of care, contributing to equivalent levels of participatory decision making (cooper-patrick et al., 1999). participants who received adjuvant treatment indicated marginally lower levels of anxiety relative to those who did not. because the #nding is correlational, the causal agent is unclear. it is possible that patients who receive adjuvant treatment may have lower levels of anxiety because they are receiving a higher quality of cancer care. alternatively, patients with lower levels of anxiety may be more likely to obtain follow-up treatment in breast cancer care and complete their medical treatment. limitations and further study although this study drew from a random sample of breast cancer patients generated by the american cancer society’s siebel database, all participants were active acs members whose ideas and views may not be representative of the general population of breast cancer patients. "e study’s overall response rate of 44% is an issue of concern because low response rates typically result in unrepresentative samples. however, it should be noted that this study had a comparable response rate to similar studies surveying breast cancer patients (e.g., cyran et al., 2001; frank & clancy, 1993). "e participants’ high pdm suggests that they may not be re$ective of the population of breast cancer patients in the new york city boroughs. urban cancer patients tend to face problems in medical care that are characteristic of low pdm, including short medical visits and long wait times (mehta et al., 2008). "e generalizability of our study’s #ndings was also limited in that all the participants’ were members of the american cancer society, and all were insured. because an optional mail survey was utilized, volunteer bias may have impacted the results. participants who were more likely to respond may have felt stronger positive or negative feelings towards their oncologists and their medical treatments. in an era in which medical treatment can be a!ected drastically by the insurance providers of the patients, it would be valuable to investigate potential associations of patients’ insurance coverage with adjuvant treatment rates and pdm. whereas past research has found insurancebased discrepancies in breast cancer screening (ayanian et al., 1993), studies have yet to examine associations between breast cancer adjuvant treatment and insurance. in addition, due to #ndings that patients and physicians of the same race to have higher pdm than those of di!erent races (e.g., murray-garcia et al., 2000), future studies should examine potential di!erences in adjuvant treatment rates and anxiety between race concordant versus discordant patients and physicians. "is is one of the #rst studies to date to examine psychosocial and demographic factors that may be related to adjuvant treatment rates. contrary to the assumptions of past studies (e.g., johnson et al., 1988; lurie et al., 1993), this study found that pdm between oncologists and patients is not related to adjuvant treatment. "e surprisingly low adjuvant treatment rates (48%) in this sample suggest possible system failures in breast cancer care and indicate that more attention needs to be given to educating people about the importance of follow-up treatment in breast cancer care. in addition, this study’s #nding that breast cancer patients who select female oncologists are more likely to receive breast cancer adjuvant treatment highlights the need for more research to probe the cause of di!erences in adjuvant treatment rates. ayanian, j, kohler, b, abe, t, epstein, a (1993). "e relation between health insurance coverage and clinical outcomes among women with breast cancer. new england journal of medicine 329 326-331. bickell, n, and cohen, a (2008). understanding reasons for underuse: an approach to improve quality of care and reduce disparities in breast cancer treatment. mt. sinai journal of medicine 72 23-30. clarke, m, collins, r, darby, s, davies, c, elphinstone p, et al. (2005). e!ects of radiotherapy and of di!erences in the extent of surgery for early breast cancer on local recurrence and 15-year survival. lancet 366 2087-2106. cooper-patrick, l, gallo, j, gonazlez, j, vu, h, powe, n, et al. (1999). race, gender and partnership in the patient-physician relationship. "e journal of the american medical association 282 583-589. cyran, e, crane, l, and palmer, l (2001). physician sex and other factors associated with type of breast cancer surgery in older women. annals of surgery 136 185-191. elderkin-"ompson, v, and waitzkin, h (1999). differences in clinical communication by gender. journal of general internal medicine 14 112-121. fiscella, k, frank, p, gold, m, and clancy, c (2000). addressing socioeconomic, racial, and ethnic disparities in health care. journal of the american medical association 283 2579-2584. guadagnoli, e, and ward, p (1998). patient participation in decision-making. social science medicine 47 329-339. frank, p, and clancy, c (1993). physician gender bias in clinical decision-making: screening for cancer in primary care. medical care 31 213-218. 4.13.71 1 2 3 4 5 6 yes no standard care a n x ie t y figure 6 7h[plu[z�^ov�yljlp]lk�hkq\]hu[�[ylh[tlu[�ohk�thynpuhss`�sv^ly�sl]lsz�vm�hu_pl[`��4�$�������[ohu�why[pjpwhu[z�^ov�kpk�uv[�yljlp]l�hkq\]hu[�[ylh[tlu[��4�$��������[������$�������w�$��� � conclusion references o n co lo g y low energy surface acoustic waves (saw) have been shown to reduce bio!lm formation on urinary catheters in rabbits (hazan et al., 2006). recently, the fda has approved saw as a safe method to enhance wound healing (howell-taylor et al, 2008). low-energy saw devices generate and spread vibrational energy at frequencies of 100 to 300 khz and amplitudes of 300 to 800 nm. "e waves are propagated uniformly and in all directions, covering the entire system (hazan et al., 2006). we hypothesized that the e#ectiveness of saw was based on an underlying immunological mechanism since it showed clinical use in wound and bone healing, processes which involve immune cells. while there is no current research on the impact of saw on immune cell function, recent unpublished studies in our lab have shown that saw signi!cantly improves the capacity of human neutrophils, which are innate immune cells, to kill bacteria within a three-dimensional in vitro !brin gel environment. based on these results, it was hypothesized that saw enhanced bacterial killing in a tissue-like matrix by facilitating chemotaxis of neutrophils; the mechanical vibrations induced by saw may cause activation of neutrophil through an unidenti!ed mechanoreceptor. since saw appeared to enhance neutrophil function, we wanted to examine if other immune cells could also be stimulated by saw. t-cells constitute an important arm of the adaptive immune system and are ubiquitously involved in defense against cancerous cells. we extended our study to examine whether saw has an e#ect on t-cells’ ability to kill melanoma tumor cells. cutaneous melanoma is a type of skin cancer. each year, the incidence of cutaneous melanoma continues to escalate, with over 50,000 new cases being diagnosed within the united states in one year. approximately 1020% of all such melanomas are located in the head and neck region and mortality is growing at a very high rate, second only to that of lung cancer. in most cases, excision of the tumor is the best treatment option available to patients (rigel et al, 2000). "e application of saw can offer a treatment alternative perhaps not for serious cases of melanoma in which lesions must be immediately excised, columbia undergraduate science journal open-access publication | http://cusj.columbia.com 34 cusjvolume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu o n c o lo g y cusjcolumbia undergraduate science journalresearch articles 35cusj volume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu ca n c er bio lo g y surface acoustic waves cause net reduction in mouse and human melanoma growth in vitro komal kothari1,*, jona zumeris2,, harold jacob2, john loike3* 1department of biological sciences, columbia university, new york, ny 10027, 2nanovibronix, nesher, israel 36603; 3department of physiology and cellular biophysics, columbia university college of physicians and surgeons, new york, ny 10032. . low-energy surface acoustic waves (saw) have been shown to be effective in curing urinary catheter infections and in enhancing neutrophil killing of bacteria. here, we examined whether cellular immunity, the killing of melanoma tumor cells by t-cells, can be enhanced with the application of saw. from the preliminary data, it appears that saw is capable of reducing mouse and human melanoma cell growth independently of t-cell activity. in the b16 mouse melanoma model, nyv^[o�pu�:(>�[ylh[lk�jvsshnlu�äiypu�nlsz�^hz�ylk\jlk�i`�����h[�+h`���huk�i`�����h[�+h`���^olu�jvtwhylk�[v�nyv^[o� pu�jvu[yvs�jvsshnlu�äiypu�nlsz��(�7yvwpkp\t�0vkpkl�hzzh �̀�^opjo�å\vylzjlz�klhk�jlssz��z\nnlz[lk�[oh[�[ol�hwwspjh[pvu�vm� saw kills b16 cells directly, since the saw-treated gels showed almost twice the percentage of dead cells (44.25%) as compared to the control gels (26.09%). in addition, saw-treated a375 cells, or human melanoma cells, showed apwyv_pth[ls`�h�����ylk\j[pvu�pu�nyv^[o�pu�jvsshnlu�äiypu�nlsz�pu�jvtwhypzvu�[v�nyv^[o�pu�jvu[yvs�nlsz��;�jlss�rpsspun�huk� migration experiments performed with b16s in the presence or absence of saw showed that saw seems to adopt an independent mechanism in killing b16 melanoma cells, while enhancing t-cell chemotactic capacity. our future experiments involve studying the effect of saw on other types of dermatological cancers and understanding the underlying mechanisms of saw in reducing tumor cell growth. key words: surface acoustic waves, melanoma, t-cells. abstract introduction copyright: © 2011 the trustees of columbia university, columbia university libraries, some rights reserved, thomas, et al. received jan. 2, 2011. accepted jan. 26, 2011. published april 1, 2011. *to whom correspondence should be addressed: department of physiology and cellular biophysics, college of physicians and surgeons, columbia university, new york, ny 10032. krk2122@columbia.edu hall, j, epstein, a, deciantis, m, and mcneil, b (1993). physicians’ liking for their patients: more evidence for the role of a#ect in medical care. health psychology 12 140-146. hershman, d, buono d, mcbride r, tsai w, joseph k, et al. (2008). surgeon characteristics and the receipt of adjuvant radiotherapy in women with breast cancer. journal of the national cancer institute 100 199-206. jemal, a, murray, t, and "um, m (2003). cancer statistics. ca: a cancer journal for clinicians 53 5-26. johnson, j, nail, l, lauver, d, king k, and keys, h (1988). reducing the negative impact of radiation therapy on functional status. cancer 61 46-51. lantz, p, janz, n, fagerlin, a, schwartz k, liu, l, et al. (2005). satisfaction with surgery outcomes and the decision process in a population-based sample of women with breast cancer. health services research 40 745-768. levinson, w, and lurie, n (2004). when most doctors are women: what lies ahead? annals of internal medicine 141 471-474. lurie, n, slater, j, mcgovern, p, ekstrum, j, quam, l, et al. (1993). preventive care for women – does the sex of the physician matter? "e new england journal of medicine 329 478-482. mechanic, d, mcalpine, d, and rosenthal, m (2001). are patients’ o$ce visits with physicians getting shorter? "e new england journal of medicine 344 198-204. meeuwesen, l, schaap, c, and van der staak, c (1991). verbal analysis of doctor-patient communication. social science medicine 32 1143-1150. mehta, r, liang, l, karve, a, hernandez, a, rumsfeld, j, fonarow, g, et al. (2008). association of patient case-mix adjustment, hospital process performance rankings, and eligibility for !nancial incentives. journal of the american medical association 300 1897-1903. morris c, cohen r, schlag r, wright, w (2000). increasing trends in the use of breast-conserving surgery in california. american journal of public health 90 281-4. murray-garcia, j, selby, j, schmittidiel, j, grumbach, k (2000). racial and ethnic di#erences in a patient survey. medical care 38 300-310. national cancer policy board. institute of medicine and national regional council. in: ensuring quality cancer care. hewitt m, simone jv, editors. washington, dc: national academy press; 1999. roter, d, lipkin, m, korsgaard, a (1991). sex di#erences in patients’ and physicians’ communication during primary care medical visits. medical care 29 1083-1093. shields, c, frank, p, fiscella, k, meldrum, s, and epstein, r (2005). rochester participatory decision-making scale (rpad): reliability and validity. annals of family medicine 3 436-442. snaith, r (2003) hospital anxiety and depression scale. health and quality of life outcomes 2 1-4. stevens, f, courtens, a, crebolder, h, and philipsen, h (1996). longitudinal study on quality of life and social support in cancer patients. cancer nursing 19 162-169. stewart, m (1984). patient characteristics which are related to the doctor-patient interaction. family practice 1 30-6. susan g. komen for the cure (2009). new york city community pro!les. retrieved october 13, 2009 from the susan g. komen for the cure new york city af!liate website: http://www.komennyc.org/site/docserver/2009_community_pro!le_report_final_for_web. pdf?docid=3341. early breast cancer triallists’ collaborative group (2005). e#ects of chemotherapy and hormonal therapy for early breast cancer on recurrence and 15-year survival: an overview of the randomized trials. lancet 365 1687717. "ompson, a (2007). "e meaning of patient involvement and participation in health care consultations. social science and medicine 64 1297-1310. wiggers, j, donovan, k, redman, s, and sansonfisher, r (1990). cancer patient satisfaction with care. cancer 66 610-616. 5 letter from the president dear readers, i am proud to announce the publication of the 2021-22 edition of the columbia undergraduate science journal! the columbia undergraduate science journal is our original namesake publication, dedicated to sharing scientific knowledge of the highest quality of scholarship resulting from undergraduate scientific research. the need for accurate scientific research and reporting is ever-present, especially now more than ever before. a significant issue that plagues society today is misinformation. opinions uninformed by science present a danger to society we need not look any further than the still-ongoing covid-19 pandemic for an example of the consequences of unscientific misinformation. a society misinformed miscalculates. so, in a world of chaotic opinion, we must remain steadfast in our pursuit of scientific knowledge. although scientific facts should inform our opinions, it is important to remember that science itself is not a list of facts, but rather a process of discovery. students all over the world contribute to this collective process as scientific researchers. through scientific research we explore nature, discover truths, and debate ideas in our relentless pursuit of knowledge. the columbia undergraduate science journal is a platform for student scientists to share their knowledge so that we may become a better informed society. it was an honor to read, edit, and review submissions made to our journal. i am grateful to our columbia undergraduate science journal team and editorial board for their significant contributions to our scientific review process. i am especially grateful to kazi rabbe, editor-in-chief of the columbia undergraduate science journal, whose leadership made this publication a success. i would also like to thank the columbia undergraduate science journal faculty advisory board, a group of esteemed columbia university professors whose support ensures publication of the highest quality of scholarship. congratulations to our authors, and thank you to our readers! arjun kudinoor president, chief editorial officer columbia undergraduate science journal protein-polymer complex coacervates as synthetic membrane-less organelles hansen tjo1*, nicholas zervoudis2, allie c. obermeyer2 1department of chemical engineering, university of massachusetts amherst 2department of chemical engineering, columbia university in the city of new york complex coacervation, protein engineering, soft matter abstract: in solution, oppositely-charged macromolecules undergo charge-mediated liquid-liquid phase separation into a complex coacervate phase – a dense, macromolecule-rich liquid. in nature, the basis for intracellular compartmentalization in the formation of membrane-less organelles has been shown to follow similar complexation principles, where charged proteins represent the ionic species. we seek to capture the spatiotemporal tunability properties of such organelles for enzymatic reactions in vitro. however, the in vitro formation and deformation of protein-based coacervate microenvironments as a nanoreactor is a limiting factor. here, we prescribe high-precision turbidimetry coupled with optical microscopy, to characterize the phase behavior of binary protein-polymer complexes between the weak anionic enzyme glucose oxidase (gox) and four different synthetic polycations as functions of composition and ionic strength. establishment of conditions ideal for coacervate formation in each protein-polymer system informed ph titration experiments on characterizing self-assembly regulation. the results from this study will help inform the design of novel coacervate microenvironments for industrial enzyme cascades and elucidate the role of associative phase separation in cellular evolution. introduction selectivity and specificity are an enzyme’s key characteristics [1, 3, 4]. as biological catalysts, enzymes increase rates of reaction such that greater amounts of biological product may be obtained under less time. they also have been used to improve chemical processes in industries from food, agriculture, and petroleum in addition to reducing energy costs and operation time [4]. it is difficult to synthetically match the efficacy and specificity of enzymes as biochemical systems have had millions of years to evolve [4, 16]. while advents in directed evolution and protein engineering attempt to circumvent the time required for natural evolution, amongst other approaches to improve enzyme activity, there is also particular interest to achieve similar goals by optimizing an enzyme’s surroundings [1, 4, 16]. this strategy is inspired by metabolic reactions, such as those that make up cellular respiration, where the spatiotemporal efficacy of enzymes is enhanced by its surrounding biological environment [1, 4]. with recent reports highlighting the complex functions of biological condensates in vivo (signaling, reaction networks etc.), we took a biomimetic approach in constructing a stable and responsive enzymatic compartment [1-5]. using de novo liquid-liquid phase separated synthetic organelles, these microenvironments may be the bridge to advance how enzymatic power is harnessed industrially. complex coacervation, an example of associative liquid-liquid phase separation, describes how oppositely charged polyelectrolytes phase separate into a coacervate phase – a dense, polyelectrolyte-rich phase with potential applications in biomolecular encapsulation – and a dilute phase, the supernatant [1]. because the coacervate phase compartmentalizes both enzymes and substrates within the same microenvironment, enzymes can perform their catalytic functions with greater spatiotemporal efficacy [1-4]. these de novo systems are especially advantageous due to their tuneability by a variety of parameters: ph, charge stoichiometry, ionic strength, mixing order, and others [1, 2]. the use of proteins or other charged biomacromolecules as coacervating macro-ions allows further structural modulation through ionic tagging and supercharging [6, 7, 9]. complex coacervate systems typically involve binary mixtures of oppositely charged components. to simplify coacervate formation, a net charged enzyme is chosen such that it constitutes one of the electrostatic components for complexation [10, 11]. in particular, the weakly anionic enzyme glucose oxidase (gox) was used in conjunction with four structurally distinct polycations: poly(4-vinyl nmethyl pyridinium iodide) (qp4vp), poly(allylamine hydrochloride) (pah), poly(ethyleneimine) (pei), and poly(1-vinyl imidazole methyl iodide) (pvi). in sum, we sought to elucidate the effects of mixing order, salt concentration, and finally ph on the phase behavior of four different gox-polycation systems: gox-qp4vp, gox-pah, gox-pei, gox-pvi. 21 methods the phase behavior of four sets of binary gox-polycation systems were studied. the four sets of polymer-protein mixtures consisted of poly(4-vinyl n-methyl pyridinium iodide) (qp4vp), poly(allylamine hydrochloride) (pah), polyethylenimine (pei), and poly(1-vinylimidazole methyl iodide) (pvi) as the polycations, with glucose oxidase (gox) as the anionic charged protein. additionally, we investigated the effects of salt on coacervate formation via the addition of sodium chloride (nacl). finally, we utilized ph titrations to explore how complex coacervation can be regulated by solution ph. sample preparation: glucose oxidase from aspergillus niger was purchased from sigma aldrich (g2133). a stock solution of 2 mg/ml gox was prepared in 10 mm tris at ph 7.4. application of beer’s law was used to determine true concentrations of gox via its absorbance at 280 nm in a 4 ml quartz cuvette. polymer solutions were diluted from liquid stocks of 5 mg/ml; relevant stoichiometric calculations using the molar equivalency equation were used to determine the requisite volumes of 10 mm tris needed to dilute appropriated samples to a mass concentration of 2 mg/ml. each polycation solution was subsequently adjusted to a ph of 7.4. mixing ratios: four mixing ratios of gox/polymer were investigated: 88% gox/12% polymer, 84% gox/16% polymer, 80% gox/20% polymer, 76% gox/24% polymer. such values were determined from preliminary data indicating an optimum mixing ratio range for gox at roughly 80%. data results for variations in mixing ratios were obtained through turbidimetry analysis and optical microscopy. to exclude external ionic strength contributions in studying mixing ratio effects, salt species were absent in all mixtures. salt effects: the effects of added salt on system phase behavior were examined via turbidimetry analyses and optical microscopy. sodium chloride (nacl) concentration was varied from 25 mm to 50 mm on all polymer/protein systems. this salt range was predicted to be conducive for liquid-liquid phase separation in gox-polymer systems based on preliminary work. for the gox-pei system, no form of phase separation was observed at all salt concentrations. thus, further planned investigations on the gox-pei system with salts were abandoned. ph titration: the effects of ph on phase behavior reversibility was examined via turbidimetry analysis, based on absorbance readings from an uv-vis spectrophotometer, and a ph probe at constant ambient temperature (25 ℃). 1 m hydrochloric acid (hcl) was used as the titrant; a 1 cm stirbar at roughly 500 rpm was used to ensure consistent solution mixing throughout the procedure. each gox-polycation system was set at its experimentally-determined optimum mixing ratio and salt concentration: 88% mixing ratio, and 50 mm nacl for all systems with the exception of goxpah, which was set at 175 mm nacl. turbidimetry analyses: turbidimetry analyses were done to investigate the effects of mixing ratios and salt concentration. each sample was prepared in triplicate in tissue culturetreated polystyrene 96-well half-area plates (corning), followed by incubation at room temperature for 3 h. using a plate reader (tecan infinite m200 pro), the absorbance of the mixture was taken at a wavelength (λ = 600 nm) to monitor scattering of the phase separated mixture. each sample had an invariant volume of 50 µl allowing for appropriate absorbance measurements and physical mixing by a tecan infinite m200 pro plate reader (10 s of orbital shaking). finally, the absorbance was converted into turbidity using the following set of relationships: 𝜏 = 100 −%𝑇 (1) %𝑇 = 10(./0) (2) where: • 𝜏 is the turbidity of the solution and an indicator of the extent of phase separation present within the sample. • 𝑇 is the transmittance of the solution as a function of mixture absorbance. • 𝐴 is the measured absorbance of the mixture. optical microscopy: all four protein-polymer samples were prepared in triplicate, followed by individual well examination with optical microscopy using an evos fl auto 2 inverted fluorescence microscope (invitrogen). each sample, controlled at a volume of 50 μl, were formulated in an optically clear 384-well plate (nunc) and then underwent 3 h incubation period at room temperature to maximize the degree of liquid-liquid phase separation taking place in the wells (preliminary data indicated that samples tended to favor figure 1. chemical structures of the four cationic polymers used in this work. structures were drawn using the chemdraw prime software. 22 precipitation following initial mixing before eventual transition into a coacervate phase). all optical microscopy images were taken under 20x objectives with transmitted light. results and discussion initial experiments investigated the effects of mixing ratio of protein to polymer using gox and four different polycations (qp4vp, pah, pei, pvi). preliminary data suggested each binary systems’ tendency to phase separate at all four preselected mixing ratios; however, certain mixing ratios resulted in greater extents of phase separation as indicated by their relative turbidity magnitudes. nevertheless, this initial assumption did not hold as no phase separation was observed with pei at all mixing ratios. however, we hypothesized that additions of salt may facilitate phase separation through external charge compensations and increasing the total possible conformations of electrostatic associations given the presence of non-polyelectrolyte ions [1, 2, 6]. thus, pei was kept for subsequent experiments on the effects of salt addition. turbidity data as a function of mixing ratio was plotted in such a way that optimum mixing ratios for each gox-polycation system can be determined. although ranging in value from 0 – 100%, turbidity is typically employed as a qualitative indicator in characterizing phase behavior: for example, turbidity values above 20% usually suggest (but do not guarantee) the presence of phase separation whereas values nearing 0% imply the absence of phase separation. however, the presence and exact nature of phase separation, be it liquid-liquid or liquid-solid phase separation, can only be verified through optical microscopy. nevertheless, turbidity is useful as a continuous measure for when a system exhibits the same morphology under different conditions, and therefore can indicate the conditions most conducive to the desired phase behavior. in this paper, for example, turbidity is used to determine mixing ratios that best drives phase separation. out of the four mixing ratios investigated, the two mixing ratios most favoring liquid-liquid phase separation will be selected for subsequent investigations involving salt and ph. it was expected that maximum complexation, as determined by peaks in relative turbidity, for a given gox-polycation system would lie around mixing ratios of 84% and 88% (although turbidity values were generally higher at a mixing ratio of 92%, optical microscopy showed greater tendencies for precipitation). indeed, all gox-polycation systems except for gox-pei exhibited coacervate formation at such mixing ratios as determined via optical microscopy (data not shown). phase separation absence in gox-pei also explains the low turbidity magnitudes (below 20% at all mixing ratios). we predicted the absence of phase separation in the gox-pei system to be due to the lack of entropic gains from bound counter-ion release in electrostatic interactions within the system as a result of pei being too weakly charged. the larger relative sample standard deviations with gox-pvi and gox-pei suggests possible systematic inaccuracies: e.g., bubble formation interference at select mixing ratios, but the consistency of turbidity maximums coupled with optical microscopy images (data not shown) at mixing ratios of 84% and 88% supported the selection of these particular mixing ratios for subsequent experiments. while the quantitative limitations of turbidity must be considered, this may also suggest the dominance of (charged) protein-polymers system’s electrostatic interactions driving macro-phase separation on turbidity readings as opposed to the strengths of individual components. salt effects based on the two selected mixing ratios (88% and 84%) from the previous experimental section, the effects of salt (nacl) concentration on gox-polycation phase behavior were investigated (fig. 3). fig. 3c illustrates the characteristic binodal curve phase diagram in phase separating polyelectrolyte systems [1, 8, 12, 14]. in this investigation, however, we are more concerned with determining salt concentrations that promoted liquid-liquid phase separation for the two mixing ratios: 84% and 88%. turbidimetry analysis was used to provide a relative measure for the degree of phase separation across all gox-polycation systems and facilitate comparison across the two different salt concentrations. the two chosen salt concentrations were 25 mm and 50 mm based on preliminary data suggesting that this resulted in liquid-liquid phase separation in the gox-polycation systems being studied. at both mixing ratios of 84% and 88%, all gox-polycations that exhibited liquid-liquid phase separation in the absence of salt (fig. 2) continued to do so with increasing salt concentration. although gox-pvi and gox-qp4vp systems underwent a decrease in turbidity with the addition of salt from 25 mm to 50 mm at constant mixing ratio of 84% (fig. 3b), optical microscopy still showed liquid-like morphologies. given that there were still no indicators of phase separation in gox-pei with the addition of salt, as suggested by low turbidity magnitudes below 20% (figs. 3a, figure 2. complex coacervation of glucose oxidase (gox) enzyme with a palette of synthetic polycations. mixtures were prepared across a select range of macromolecule mixing ratios informed from a more thorough investigation of the protein’s phase behavior. error bars describe the standard deviation of each triplicated data point (n = 3). as a control, all turbidity values shown have had 10 mm tris turbidity reference values subtracted. 23 3b), and confirmed by optical microscopy, the system was excluded from subsequent experiments on the effects of ph on coacervate formation. as an exception amongst the phase-separating systems, pah did not demonstrate the desired formation of biomolecular condensates at both salt concentrations of 25 mm and 50 mm based on optical microscopy (data not shown). instead, its turbidity values consistently above 20% for 25 mm and 50 mm salt concentrations (figs. 3a, 3b) were shown to be a result of precipitate formation, which furthers evidence for the gox-pah system’s greater relative propensity for liquid-solid phase separation at the current salt concentration range. for gox-pah, low concentrations of salt ions may not provide sufficient charge compensations to favor liquidliquid phase separation, so much as it is strengthening electrostatic interactions via increasing possible coulombicdriven conformations. conversely, at higher salt concentrations, its charge screening effects would work towards dampening such electrostatic interactions to reduce the entropic gains from bound counter-ion release and drive coacervation [2, 6, 12]. an alternative explanation as to the goxpah system’s propensity for liquid-solid phase separation at low salt concentrations could be due to kinetic trapping effects as is prevalent in solid phases [2, 8, 12]. next, experiments with the gox-pah system involving higher salt concentrations were necessary to identify a “minimum” salt concentration at which the system undergoes complex coacervation instead of precipitation. thus, we conducted an additional salt titration exploring gox-pah phase behavior at higher salt concentrations: 150 mm to 200 mm nacl at 25 mm intervals. samples were analyzed via turbidimetry (data not shown) and optical microscopy (fig. 3d). while the possibility to utilize a different salt as per the hofmeister series e.g. kbr was considered, ultimately liquid-liquid phase separation was observed at a nacl concentration of 150 mm, with the optimum salt concentration for gox-pah determined to be 175 mm (fig. 3d), thereby negating the need to use a different salt species and maintaining consistency across all gox-polycation systems. ph effects building on preceding work determining ideal mixing ratio and salt compositions conducive to coacervate formation, the goal of this section is to effectively explore coacervate figure 3. the phase behaviors of multiple gox-polycation mixtures as functions of salt (nacl) concentration. goxpolycation mixtures were prepared using the two optimum mixing ratios for complex coacervation from fig. 2 (84% and 88%). both turbidity (λ = 600 nm) and optical microscopy were used to confirm liquid-liquid phase separation. error bar values represent sample standard deviation; n = 3 for all data points. a, turbidity versus salt concentration at constant mixing ratio of 88%. b, turbidity versus salt concentration at constant mixing ratio of 84%. all turbidity values shown in figs. 3a and 3b have had 10 mm tris turbidity reference values subtracted; results from fig. 2 were also included as no-salt controls. c, binodal phase boundary of complex coacervate systems where both charged polymers and charged proteins facilitate phase separation; φ denotes ‘phase’. arrow points to increasing two-phase region with increasing macromolecular charge density or patterning, demonstrating phasic tuneability with charge-associated parameters. d, optical microscopy images of gox-pah phase behavior (88% mixing ratio) provide sufficient qualitative evidence of liquid-liquid phase separation at salt concentration ranges beyond those in figs. 3a and 3b. 24 self-regulation by ph. all gox-polycation systems were set at an initial ph of ~9.5 in addition to their optimum compositions (specified in the experimental section), and a fixed volume of acid was titrated into the solution to slowly decrease the solution ph (fig. 4b). reductions in turbidity as ph decreases marks the dissolution of the coacervate microenvironment. since gox is characterized as a weak polyelectrolyte and has an isoelectric point (pi) of roughly ~4.2, protonation of gox below its pi reduces its net negative charge. thus, the coacervate microenvironment formed by gox and a polycation would deform as liquid-liquid phase separation dissipates from weakening electrostatic interactions. this is evident from the turbidity of each system being lowest at ph values below 4.2, the isoelectric point of gox (fig. 4b). however, the periodicity of the trends, notably with the gox-pah and gox-pvi systems, was unexpected (fig. 4b). we hypothesize possible induced charging effects on prolonging liquid-liquid phase separation given the uneven anisotropy and charge patchiness of gox [5, 8]. it is possible the coacervate microenvironment may persist at some of the lower ph ranges due to such effects, which may explain why each system undergoes a transition towards a turbidity maximum at the same ph value of 6 (indicating gox as the limiting factor) following a local minimum at ph 8~ due to the initial decrease in ph. ultimately, we hope to demonstrate ph-mediated regulation of our coacervate microenvironments. throughout the experimental section, we have shown how coacervate microcompartments can be formed with polymer-enzyme complexes. further, we can tune the interaction strength and morphology via mixing ratio, salt, and ph. due to many enzymatic reactions being affected by system ph, demonstrating the microenvironment’s ability to form and dissolve reversibly in response to ph represents the primary goal of our work on protein-polymer synthetic nanoreactors [4, 16]. it is important to stress that these results are building blocks towards this objective. conclusion an enzyme-polymer complex coacervate system was investigated on the basis of its ability to capture the complexities seen in biological condensates. the phase behaviors of multiple gox-polycation systems as functions of mixing ratios and ionic strength were investigated. we were interested in each gox-polycation system’s propensity to undergo liquidliquid phase separation, and the ionic stability of their formed coacervate microenvironments. gox phase separated upon mixing with the polycations qp4vp, pah, and pvi; but did not phase separate with pei. for the gox-pah system, the addition of salt was needed to screen existing charge such that the entropic gains favoring liquid-solid phase separation may be suppressed. the gox-pei system did not undergo any type of phase separation both with and without the addition of salt. it is also worth remarking that the same species of salt, nacl, may be used to induce liquidliquid phase separation for all phase separating gox-polycation systems, suggesting the salt species’ versatility for driving complexation. ph titrations were used to investigate the dynamics of the formed coacervate microenvironments for each gox-polycation system; turbidimetry suggests that all gox-polycation systems that phase separated at a ph of ~9.5 no longer phase separated upon reaching ph < pi of gox. we suggested the possibility of induced-charging effects as an explanation for the periodicity in phase behavior of gox-pvi as ph decreases, where the anisotropy of gox and presence of charge patches prolonged phase separation despite non-ideal ph [5, 10, 11]. nevertheless, while the exact phase behavior of gox-polycation systems at low ph were not determined as with optical microscopy, due to methodology limitations, the global minimums in turbidity at low ph compared to relative maximums at high ph strongly suggested dissolution of coacervates and dissipation of liquid-liquid phase separation. this crucial finding will inform future work on demonstrating the reversibility of coacervate formation in gox-polycation systems such that these microenvironments can be made smart and self-regulated to mimic the complexity of condensates found in cells. the establishment of ideal parameters at which various gox-polycation systems undergo liquid-liquid phase separation provides a foundation for coacervate microenvironment formation for use in advanced synthetic nanoreactor design and elucidation of cellular compartmentalization phenomena. figure 4. the effects of ph on complex coacervation in multiple gox-polycation systems. each gox-polycation mixture was set at ph of roughly 9.5, followed by titration of hydrochloric acid (hcl) until dissolution of complexation. turbidity (λ = 600 nm) was used to indicate absence of phase separation upon convergence towards a minimum value after multiple ph titrates. a, schematic depicting ph-induced dissolution of complex coacervation via protonation of gox, which reduces its net negative charge. b, measured turbidity of gox-polycation systems as a function of ph. 5th order polynomials were plotted to visualize turbidity trends with ph and to guide the eye (a 5th order polynomial represented the lowest-degree polynomial containing the local extremas that track the data) 25 author information corresponding author *htjo@umass.edu author contributions hansen tjo performed the experiments, analyzed data, and produced the figures and the manuscript. nick zervoudis devised the experimental plan and reviewed the manuscript. dr. allie obermeyer reviewed the manuscript. funding sources this work was supported by the fu foundation school of engineering and applied sciences at columbia university and funding from the columbia chemical engineering reu program. acknowledgments the authors would like to acknowledge members of the obermeyer group for helpful discussions. abbreviations gox – glucose oxidase hcl – hydrochloric acid nacl – sodium chloride pah – poly(allylamine hydrocholoride) pei – poly(ethylenimine) pvi – poly(1-vinyl imidazole methyl iodide) pi – isoelectric point qp4vp – poly(4-vinyl n-methyl pyridinium iodide) references [1] l.p. bergeron-sandoval, n. safaee, s.w. michnick, mechanisms and consequences of macromolecular phase separation. cell 165(5),1067–1079 (2016). [2] w.c. blocher, s.l. perry, complex coacervate-based materials for biomedicine. wiley interdisciplinary reviews: nanomedicine and nanobiotechnology 9(4), doi:10.1002/wnan.1442 (2016). [3] m. castellana, m. z. wilson, y. xu, p. joshi, i. m. cristea, j. d. rabinowitz, z. gitai, n. s. wingreen, enzyme clustering accelerates processing of intermediates through metabolic channeling. nature biotechnology 32(10), 1011–1018 (2014). [4] j.-m. choi, s.-s. han, h.-s. kim, industrial applications of enzyme biocatalysis: current status and future aspects. biotechnology advances 33(7),1443–1454 (2015). [5] f. comert, p.l. dubin, liquid-liquid and liquid-solid phase separation in protein-polyelectrolyte systems. advances in colloid and interface science 239, 213– 217 (2017). [6] c.s. cummings, a.c. obermeyer, phase separation behavior of supercharged proteins and polyelectrolytes. biochemistry 57(3), 314–323 (2017). [7] j. horn, r. a. kapelner, a.c. obermeyer, macroand microphase separated protein-polyelectrolyte complexes: design parameters and current progress. polymers 11(4), 578 (2019). [8] p. jha, p. desai, j. li, r. larson, ph and salt effects on the associative phase separation of oppositely charged polyelectrolytes. polymers 6(5):1414–1436 (2014). [9] r. a. kapelner, a.c. obermeyer, ionic polypeptide tags for protein phase separation. chemical science 10(9), 2700–2707 (2019). [10] l. li, s. srivastava, m. andreev, a.b. marciel, j. j. d. pablo, m. tirrell, phase behavior and salt partitioning in polyelectrolyte complex coacervates. macromolecules 51(8), 2988–2995 (2018). [11] a.c. obermeyer, c.e. mills, x.-h. dong, r. j. flores, b.d. olsen, complex coacervation of supercharged proteins with polyelectrolytes. soft matter 12(15), 3570–3581 (2016). [12] s.l. perry, phase separation: bridging polymer physics and biology. current opinion in colloid & interface science 39, 86–97 (2019). [13] s.l. perry, y. li, d. priftis, l. leon, m. tirrell, the effect of salt on the complex coacervation of vinyl polyelectrolytes. polymers 6(6), 1756–1772 (2014). [14] d. priftis, m. tirrell, phase behaviour and complex coacervation of aqueous polypeptide solutions. soft matter 8(36), 9396–9405 (2012). [15] d. priftis, n. laugel, m. tirrell, thermodynamic characterization of polypeptide complex coacervation. langmuir 28(45), 15947–15957 (2012). [16] e. ricca, b. brucher, j.h. schrittwieser, multi-enzymatic cascade reactions: overview and perspectives. advanced synthesis & catalysis 353(13), 2239–2262 (2011). 26 columbia undergraduate science journal open-access publication | http://cusj.columbia.edu spring 2014 | volume 8 4 effect of fasudil on glioma cell migration christine wang1*, athanassios dovas2, benjamin amendolara3, peter canoll2 1columbia college, columbia university, new york, ny, usa; 2department of pathology and cell biology, columbia university, new york, ny, usa; 3department of neurosurgery, columbia university, new york, ny, usa abstract gliomas are tumors arising from glial cells that are highly infiltrative within the brain. their extensive migratory patterns greatly limit the treatment options and prognoses for patients. it has been shown that glioma cells undergo migration in a unique manner that relies heavily on non-muscle myosin ii. since functionality of non-muscle myosin ii requires the phosphorylation of its regulatory light chain by enzymes like rho-kinase, we have treated glioma cells with the rho-kinase inhibitor, fasudil, and tested its effects. the results indicated that fasudil is a potent inhibitor of migration and also accordingly decreases regulatory light chain phosphorylation in glioma cells. this suggests that fasudil could be a good potentia l candidate for future in vivo studies and contribute to current therapeutic options for the treatment of gliomas. however, residual cell motility and phosphorylation despite treatment with fasudil also suggest that more work must be done to assess the importance of other pathways in glioma cell migration. introduction gliomas are brain tumors that arise from glial cells and malignant gliomas are associated with poor prognoses due to their prolific migration within the brain [1]. though the tumor cells often do not migrate outside of the central nervous system, they spread across white matter tracts and are highly infiltrative [1]. in patients, this extensive migration makes it difficult for a complete surgical tumor resection, which can eventually cause recurrent tumors and disease progression [6]. this makes gliomas particularly hard to treat and limits available therapeutic options [6]. glioma cells exhibit a rather unique migratory pattern due to the tightly packed and constraining physical environment they are exposed to within the brain parenchyma which forces them to migrate through sub-micrometer spaces [1]. consequently, cell migration occurs in two steps where the cytoplasm first protrudes outward and the cell body follows in saltatory fashion [4]. interestingly, this type of migration closely resembles that of glial progenitors [4]. non-muscle myosin ii (nmii) has been shown to be particularly important in glioma cell migration and plays a key role in the second step, in which it forces the large nucleus and cell body through small spaces within the brain [1]. nmii consists of three distinct isoforms (a, b, c), defined by their heavy chain, but only nmii a and nmii b seem to be involved in cell migration [13]. three distinct genes encode the respective heavy chains, and the three isoforms share 6080% of the same amino acid sequence identity [8]. all nmii molecules have two globular heads and a helically coiled tail and assemble into bipolar filaments through interactions between their coiled-coil tails [11]. each head consists of an amino-terminal portion of a heavy chain, one essential light chain and one regulatory light chain, and the tail consists of the carboxyl terminal portions of the two heavy chains [11]. when nmii regulatory light chain (rlc) is phosphorylated, actin-dependent mg2+-atpase motor activity that is associated with the globular head domain increases. this induces the unidirectional movement of the motor domain on its associated actin filament resulting in actin filament contractility [8, 9]. though originally it was thought that only the ca2+-dependent myosin light chain kinase (mlck) phosphorylated the regulatory light chains (rlcs), it was later found that a significant amount of rlc phosphorylation occurs through a ca2+ independent pathway [5]. this pathway is now known to be mediated by a small gtpase, rhoa, copyright: © 2014 the trustees of columbia university, columbia university libraries, some rights reserved, wang, et al. received 1/1/2014. accepted 3/14/2014. published 3/14/14. *to whom correspondence should be addressed: christine wang, columbia university, new york ny 10027, email: cjw2146@columbia.edu. biology columbia undergraduate science journal open-access publication | http://cusj.columbia.edu spring 2014 | volume 8 5 which had previously been associated with many cellular activities including cell motility, cytokinesis, regulation of stress fibers and formation of focal adhesions [5]. activated gtprhoa can activate rho-associated protein kinase (rock), which then phosphorylates nmii rlcs directly and also indirectly by inactivating myosin phosphatase [5]. both mlck and rock can mono-phosphorylate the serine 19 site or diphosphorylate the serine 19 and threonine 18 sites on the rlc [5]. regardless, previous studies have demonstrated promising results in which inhibition of just rock can efficiently inhibit cell motility [1]. in the beadle et al. study, the treatment of cells with y-27632—a potent inhibitor of rock—proved to be an efficient inhibitor of cell migration and motility. however, no studies have tested what the effect of the drug, fasudil, also a rock inhibitor, will have on glioma cell migration. since fasudil is already an fdaapproved drug, it is valuable to evaluate it further and determine whether or not it has any effect on migratory patterns. materials and methods cell culture mgpp3 (pdgf+/pten−/−/p53−/−) and ptyb (ptenfl/fl/yfp+) glioma cells were used in this study. these cells were isolated from gliomas generated in mice [10]. cells were grown in bfp medium (dmem containing 2.5% fbs, n-2 supplement, 10 ng/ml fgf, 10 ng/ml pdgf-b, and antimycotic-antibiotic supplement) at 37oc, 5% co2 in a humidified atmosphere. mtt (3-(4,5-dimethylthiazol-2-yl)-2,5diphenyltetrazolium bromide) assay mgpp3 cells (1x103) in 100 μl/well were plated in a 96-well plate for 12 hours with 3 wells dedicated to each of the 13 treatment conditions. treatment conditions included no drug and the indicated concentrations of either fasudil or y27632. 20 μl of reagent (celltiter 96 aqueous one solution reagent) was added to each well and the plate was incubated for 6 hours with reagent. the reagent of the assay is a yellow tetrazole that is reduced to purple formazan in living cells [12]. consequently, wells that appeared more purple in hue contained more viable cells. absorbances were recorded at 490 nm using a 96-well plate reader. transwell migration assay mgpp3 cells (1x105) in a total volume of 200 μl 10% fetal bovine serum (fbs) were seeded on top of 3 μm fluoroblok transfilters (bd biosciences, san jose, ca). 500 μl of 10% fbs was added to the bottom well. a plate was set up with three transfilters per condition—control, 100 μm fasudil and 500 μm y-27632. for conditions that require drug treatment, drug was added both above and below the transfilter membrane. the entire plate containing all transfilters was incubated for 6 hours at 37°c. after incubation, both sides of the transfilter were washed three times with phosphate-buffered saline (pbs). cells were then fixed with 4% paraformaldehyde for 15 minutes on ice. after another three washes in pbs, cells were stained with 4,6-diamidino-2phenylindole (dapi) for 20 minutes. the transfilter membranes were removed and mounted on slides. images of the bottom part of the filters were acquired on a nikon widefield-inverted microscope using a 4× objective and the nuclei within five fields per membrane were counted and averaged together. phosphomyosin staining ptyb cells (3.4x104) were plated on polyl-lysine coated coverslips overnight. cells were then treated with either fresh bfp media, bfp supplemented with 60 μm fasudil, or bfp supplemented with 50 μm y-27632 for three hours. cells were fixed with 4% pfa and stained overnight at 4°c with phospho-serine 19 antibody (ps19, mouse antibody; cell signaling technology) or phosphor-threonine 18-serine 19 antibody (pt18ps19, rabbit antibody; cell signaling technology). later, cells were incubated at room temperature with alexa-568 conjugated secondary antibodies (invitrogen). 
 images were acquired on a nikon widefield-inverted microscope using a 20× objective under identical excitation conditions. images were analyzed for the integrated density of the signal using imagej (national institute of health), and results were imported into microsoft excel and values were normalized to the control condition (bfp only), n=3. statistical tests significance was determined by student's t-test (paired, two-tailed unless otherwise __________ biology http://en.wikipedia.org/wiki/dihttp://en.wikipedia.org/wiki/dihttp://en.wikipedia.org/wiki/thiazole http://en.wikipedia.org/wiki/phenyl http://en.wikipedia.org/wiki/tetrazole http://en.wikipedia.org/wiki/purple http://en.wikipedia.org/wiki/formazan columbia undergraduate science journal open-access publication | http://cusj.columbia.edu spring 2014 | volume 8 6 indicated) and a p<0.05 was considered significant. results in order to test the motility of untreated and treated cells and confirm fasudil would inhibit glioma cell migration to a significant degree, transwell migration assays (tma) were the assay of choice. transfilters provide 3 μm pore sizes comparable to sizes of pores that exist in the brain through which glioma cells would need to migrate [1]. figure 1 demonstrates the relative size of the glioma cells compared to the small pore that it must migrate through. the tmas evaluate chemokinetic migration as either side of the filter membrane contains media supplemented with growth factors that act as chemoattractants. this approach mimics a native environment similar to that which exists in the brain. figure 1: glioma migration through 3 μm transwell membranes. representative image of the top of a transfilter membrane stained with ps19-rlc antibody showing the relative sizes of the control glioma cell and the 3 μm pore. purple color represents cell nuclei; green color represents phosphorylation at serine 19 site; black dots represent 3 μm pores through which cells migrated. in order to identify non-toxic drug concentrations for the tma experiments, we conducted a dose response assay, specifically a colorimetric mtt assay. from this assay, we determined that 100 μm fasudil is the highest concentration that can be used without affecting cell survival (figure 2). cells treated with equivalent amounts of y-27632 demonstrated more cell viability although cells tolerate both inhibitors at high concentrations (figure 2). 500 μm was chosen as the designated y-27632 concentration because previous studies used 50 μm y-27632 as an established, effective concentration for inhibition and a tenfold increase would create an extreme concentration point that could help further elucidate the effect of fasudil [1]. figure 2: mtt assay demonstrating cell viability for different concentrations of rock inhibitors. dose response curves for mgpp3 cells treated with either fasudil or y-27632 for 6 hours. each point on the curve represents the average of five trials for each condition and the error bars represent standard error. cell survival (y-axis values) was normalized to the absorbance levels measured in cells treated with only h2o, pvalues for concentrations greater than 100 μm determined by a student’s t-test (unpaired, onetailed) are labeled in the corresponding color above the points. as figure 3 shows, 100 μm fasudil reduces trans-membrane migration of cells to 25.36% compared to the control condition (p=0.009). 500 μm y-27632 also significantly reduces cell migration to 17.14% (p=0.0055). on the other hand, the difference in motility between fasudil-treated cells compared to y-27632-treated cells is just under p=0.05 (p=0.04985), despite the much higher dosage of y-27632. biology columbia undergraduate science journal open-access publication | http://cusj.columbia.edu spring 2014 | volume 8 7 a. effect of rock inhibitors on cell migration through 3 μm transwell membranes. c) fasudil-treated d) y27632 b) control cells cells treated cells figure 3: cell migration in transwell migration assays. (a.) percent migration was normalized to the number of migrated control cells, error bars represent standard error of the averages of the nuclei count from five representative fields of each filter at 4x magnification and p-values are listed above the bars. representative images of the bottom portion of transfilters used in tma experiments after mgpp3 cells were incubated for 6 hours at 37°c with either 10% fbs and were (b.) uninhibited (control) cells, (c.) fbs+100 μm fasudil-treated cells, or (d.) fbs+500 μm y-27632-treated cells. cells were fixed and stained with dapi. each experimental condition was assessed in triplicate for a total n=3. aside from observing the difference rock inhibitors can make on migration, it is also important to ascertain whether the predicted pathway is operating as expected when inhibition occurs. since phosphorylated rlcs (phosphorlcs) are expected to be responsible for much of the nmii-dependent contractions within glioma cells, fasudil and y-2632 inhibition of rock should decrease motility after decreasing rlc phosphorylation. in order to quantify this observation and corroborate this with previous findings, ptyb cells were stained with the same ps19 and pt18ps19 antibodies after treatment with 60 μm fasudil (figure 4a,b) or 50 μm y-27632 (data not shown) which were concentrations used in other studies [1]. representative images were captured using a wide-field microscope and integrated density of the signal was calculated (figure 4a,b). the results show a 47.99% decrease in phosphorylation of the serine 19 site when cells were treated with 60 μm fasudil compared to control (figure 4d;; p=0.006). similarly, 50 μm y27632 demonstrated a 41.93% decrease in monophosphorylation (figure 4d; p=0.0046). c. phosphorylation of cells w/ pt18ps19-rlc biology columbia undergraduate science journal open-access publication | http://cusj.columbia.edu spring 2014 | volume 8 8 d. phosphorylation of cells w/ ps19-rlc figure 4: phospho-rlc staining. representative images of ptyb cells treated with bfp media, or bfp+60μm fasudil for 3 hours before fixing and staining with dapi and the primary antibodies (a.) pt18ps19-rlc or (b.) ps19-rlc. underlying blue color represents dapi staining of nuclei; red color represents phospho-rlc. integrated density of the (c.) diphosphorylated-rlc signal or (d.) monophosphorylated-rlc signal was measured from 20 cells per condition, averaged and normalized to control ptyb cell phosphorylation. error bars represent standard error of measurements for each condition and p-values are listed above the corresponding bars. after measuring the integrated density of cells incubated with the di-phosphorylated antibody, the results demonstrated 60 μm fasudil decreased phosphorylation to 57.9% of the control (figure 4c; p=0.02). on the other hand, 50 μm y-27632 decreased phosphorylation to 55.82%, which was not significant on a paired ttest (p=0.095; figure 4c), but was significant on an unpaired t-test (p=0.039). fasudil, therefore, effectively decreases the monoand diphosphorylation of the rlcs. discussion through the various transwell migration assays, the results indicate that chemokinetic motility is indeed inhibited, though not completely abolished, when glioma cells are treated with 100 μm fasudil. these results do yield promising implications for fasudil as its efficacy suggests it could make a good inhibitor of cell migration in in vivo studies as well. the concentrations used in the transwell experiments as well as the mtt assay can help inform the drug dosages that could be administered to a living system for maximal inhibition and minimal side effects. it is surprising, however, that the negative control of 500 μm y-27632-treated cells produced results that represented a statistically significant difference from that of fasudil but did not represent a large difference even though the y27632 concentration had been purposefully and dramatically increased. these results suggest that even at a much higher concentration of rock inhibitors, glioma cell migration cannot be completely inhibited. preliminary results of tma experiments of cells treated with a heightened concentration of 680 μm fasudil corroborate with this finding (5.5% residual migration, n=1). similarly, the ptyb cells stained with phospho-rlc antibodies demonstrated cells treated with fasudil or y-2763 exhibited a lower level of total phosphorylation (figure 4). the phospho-rlc images also demonstrate a clear change in cell morphology after treatment with fasudil or y-27632. cells begin to form long, thin spindle-like extensions and lose their more rotund cell shape (figure 4). this suggests that decrease in nmii rlc phosphorylation and the decrease in nmii activity is qualitatively related to the cytoskeleton structure of glioma cells. in addition, the nmii activity decrease potentially is related to the decrease in focal adhesions and stress fiber formation, as detailed in other studies [2]. in images of cells stained with pt18ps19, however, there was evidence of non-specific nuclear staining that most likely overestimated the phosphorylation signal during quantification. future experiments can be done to corroborate the results shown in figure 4 by measuring phosphorylation levels through other methods such as western blots using antibodies that can detect monoor di-phosphorylation. decreased rock activity after treatment with fasudil could also be assessed through western blots with a rock antibody like ser1366, antibodies for an alternate substrate of rock such as lim kinase, through a quantitative rock activity assay [3]. it is also possible, however, that the residual phosphorylation and migration across transfilters is due to rlc phosphorylation from biology columbia undergraduate science journal open-access publication | http://cusj.columbia.edu spring 2014 | volume 8 9 other pathways. one important pathway to consider is the mlck mediated route, which is also known to phosphorylate rlcs at the same sites as rock. this could be tested by assessing phosphorylation and migration after treating cells with both fasudil as well as an mclk inhibitor such as ml-7 [7]. finally, it is important to note that though fasudil is an effective drug target in short-term or acute treatment, we still do not know if it will work after a more long-term treatment. it is possible that cells will eventually compensate for rock inhibition. this information will be particularly important for future in vivo studies or clinical trials because it can greatly influence how the drug is delivered, how often it is delivered, or whether it should be administered along with another drug. subsequent experiments can be done to test motility of glioma cells after prolonged periods of treatment with fasudil. further experiments in motility could also be conducted with fasudil that has been incubated for periods of time at biological temperatures to test the biostability of the drug and to see if its efficacy decreases with prolonged incubation. these experiments also have very important implications for future in vivo or clinical studies. thus, though fasudil is a very promising drug, more work must be done in order to elucidate the prolonged effects of fasudil and determine the cell’s precise response to the drug and its potential as a cancer drug in living systems. references 1. beadle, c., m. c. assanah, p. monzo, r. vallee, s. s. rosenfeld, and p. canoll (2008). "the role of myosin ii in glioma invasion of the brain." mol biol cell 19, no. 8: 3357-68. 2. chihara, k., m. amano, n. nakamura, t. yano, m. shibata, t. tokui, h. ichikawa, r. ikebe, m. ikebe, and k. kaibuchi (1997). cytoskeletal rearrangements and transcriptional activation of c-fos serum response element by rho-kinase. j. biol. chem. 272:25121-25127. 3. chuang, h. h., c. h. yang, y. g. tsay, c. y. hsu, l. m. tseng, z. f. chang, and h. h. lee. "rock ii ser1366 phosphorylation reflects the activation status." biochem j 443, no. 1 (apr 1 2012): 145-51. 4. farin, a., s. o. suzuki, m. weiker, j. e. goldman, j. n. bruce, and p. canoll. "transplanted glioma cells migrate and proliferate on host brain vasculature: a dynamic analysis." glia 53, no. 8 (jun 2006): 799-808. 5. fukata, y., m. amano, and k. kaibuchi. "rho-rho-kinase pathway in smooth muscle contraction and cytoskeletal reorganization of non-muscle cells." trends pharmacol sci 22, no. 1 (jan 2001): 32-9. 6. giese, a. "glioma invasion--pattern of dissemination by mechanisms of invasion and surgical intervention, pattern of gene expression and its regulatory control by tumor suppressor p53 and proto-oncogene ets-1." acta neurochir suppl 88 (2003): 153-62. 7. gillespie, g. y., l. soroceanu, t. j. manning, jr., c. l. gladson, and s. s. rosenfeld. "glioma migration can be blocked by nontoxic inhibitors of myosin ii." cancer res 59, no. 9 (may 1 1999): 2076-82. 8. heissler, s. m., x. liu, e. d. korn, and j. r. sellers. "kinetic characterization of the atpase and actin-activated atpase activities of acanthamoeba castellanii myosin-2." j biol chem 288, no. 37 (sep 13 2013): 26709-20. 9. kamm, k. e., l. c. hsu, y. kubota, and j. t. stull. "phosphorylation of smooth muscle myosin heavy and light chains. effects of phorbol dibutyrate and agonists." j biol chem 264, no. 35 (dec 15 1989): 21223-9. 10. sonabend, a. m., j. yun, l. lei, r. leung, c. soderquist, c. crisman, b. j. gill, et al. "murine cell line model of proneural glioma for evaluation of anti-tumor therapies." j neurooncol 112, no. 3 (may 2013): 375-82. 11. tan, j. l., s. ravid, and j. a. spudich. "control of nonmuscle myosins by phosphorylation." annu rev biochem 61 (1992): 721-59. 12. van meerloo, j., g. j. kaspers, and j. cloos. "cell sensitivity assays: the mtt assay." methods mol biol 731 (2011): 237-45. 13. vicente-manzanares, m., x. ma, r. s. adelstein, and a. r. horwitz. "non-muscle myosin ii takes centre stage in cell adhesion and migration." nat rev mol cell biol 10, no. 11 (nov 2009): 778-90. biology mechanosensory abnormal) and unc-86 (unc for uncoordinated) are two genes that encode transcription factors (mec-3 and unc-86, respectively), which form heterodimers (unc-86::mec-3, a complex with increased dnabinding speci!city) that regulate trn development in c. elegans (goodman, 2008). mec-3 encodes an lim-type homeodomain protein (where lim stands for the proteins lin-11, isl-1 and mec-3) with an evolutionarily conserved role in neuronal di"erentiation, migration, and morphogenesis (hobert and westphal, 2000). lim-type protein expression in the trns, flps, and pvds activates the downstream transcription of genes vital for trn, flp, or pvd function, respectively, when its promoters bind to the unc-86::mec-3 heterodimer (figure 2 displays the combinatorial action of these genes). mrn di"erentiation is dependent on the combinatorial action of several genes: unc-86, mec-3, egl-44, egl-46 and lin-14. unc-86 encodes a pou-type (derived from the names of three transcription factors: the pituitary-speci!c pit-1, the octamer-binding proteins oct-1 and oct-2, and the neural unc-86 from c. elegans) homeodomain protein (from a highly conserved family of eukaryotic transcription factors) expressed in 59 cells, which is a key regulator in the development of cell lineages into trns, and activates expression of the mec-3 gene (finney et al., 1988). #e development of trns requires unc-86 to produce appropriate trn lineages, and mec-3 for trn di"erentiation (way and chal!e, 1989). in unc-86 mutants, the correct lineage of trn precursors is disrupted, which prevents the six trns from being made. mec-3 mutants contain cells with the potential to become trns, but lack adequate differentiation and the typical features of trns. lin-14 is a gene that encodes a protein vital for regulating postembryonic cell division timing and acts as a switch between trn and pvd di"erentiation: in the presence of lin-14, avm and pvm di"erentiation occur, while the absence of lin-14 gives rise to pvds. #e combined action of these genes ultimately restricts the expression of trn fate to the six neurons observable in wild-type animals (mitani et al., 1993). trns detect gentle touch to the body in a manner dependent on the expression of mec-3. in trns, combined action of mec-3 and unc-86 activates various other mec genes (mec-1, 2, 4, 7, 8, 9, 10, 12, 14, 15, 17, 18) (chal!e and au, 1989) and the alr-1 gene (topalidou et al., 2011), which de!ne trn fate. #ese genes comprise the mechanoreceptor channel complex, encoding parts of the extracellular matrix, tubulins for the trn speci!c 15-proto!lament microtubules, and other proteins of unidenti!ed function. transduction of the touch stimulus itself is accomplished via proteins encoded by several mec genes (mec-2, 4, 6, 10) and unc-24 (chal!e and bounoutas, 2007). flp neurons express the gene sto-5, which encodes a stomatin-like protein that regulates ion permeability (stewart et al., 1993). in the absence of mec-3, sto-5 is no longer expressed in flps, indicating that sto-5 expression in flps depends on mec-3 (topalidou and chal!e, 2011). although all flps and trns are regulated by mec-3, they have di"erent functions, which may be linked to how mrns express their speci!c traits and acquire distinct cell fates (way and chal!e, 1988) (figure 3 demonstrates the central role of mec-3 in trn and flp di"erentiation). since both mec-3 and unc-86 are also expressed in flp neurons and sto-5 is only expressed in flps, there must be other co-factors that promote or block trn fate. although mec-3 is needed for sto-5 gene expression in flps (characteristic of flp cell fate), flps do not express the alr-1 and other mec genes characteristic of trns. it is known that mec-3 activates the sto-5 gene directly (topalimechanosensation, or converting mechanical forces of external touch into electrical signals, is easily studied in caenorhabditis elegans (c. elegans), a transparent nematode with a well-understood nervous system and a fully sequenced genome. the mechanovhqvru\�v\vwhp�phgldwhv�jurzwk�dqg�ghyhorsphqw�lq�prvw�rujdqlvpv��d�odfn�ri�ru�ghihfw�lq�phfkdqrvhqvru\�vwlpxodwlrq�fdq�ohdg�wr� ghyhorsphqwdo�ghod\�lq�fkloguhq��uhvxowlqj�lq�ghyhorsphqwdo�glvrughuv�vxfk�dv�dxwlvp�vshfwuxp�glvrughuv��$6'v��dqg�$wwhqwlrq�'h¿flw� disorder (add). studying touch in the animal model c. elegans can elucidate the underlying neuronal mechanisms of touch and their developmental effects. in c. elegans, mechanoreceptor neurons (mrns) detect touch, with three types that require the gene mec-3 (one of the lim-homeodomain transcription factors) for their proper differentiation: touch receptor neurons (trns) (gentle touch), flp qhxurqv��wls�ri�wkh�khdg�wrxfk��dqg�39'�qhxurqv��kduvk�wrxfk���:kloh�wkhvh�wkuhh�051v�gliihuhqwldwh�e\�wkh�vdph�wudqvfulswlrq�idfwru�� each cell type’s differentiation produces distinct morphology and express different genes. flps exclusively express the sto-5 gene, while trns express the mec-17 gene. the genes that determine and distinguish flp vs. trn differentiation are currently limited to egl-44, egl-46 and alr-1. to identify additional genes needed for the differential expression of trns and flps, the wild-type strain (tu3813), with the flp neurons labeled with sto-5p::gfp and the trns labeled with mec-17p::rfp, was mutated and animals zlwk�dowhuhg�h[suhvvlrq�ru�qhxurqdo�pruskrorj\�zhuh�lvrodwhg��:kloh�ghihfwv�lq�erwk�h[suhvvlrq�dqg�pruskrorj\�ri�wkh�)/3v�dqg� trns were observed in six animals, only one animal’s mutation affecting trn appearance was true-breeding. this morphological mutant exhibits abnormal trn axon morphology, due to an autosomal recessive mutation. the mutant strain will be sequenced to identify the mutated gene, and the role of this gene in trn and flp differentiation. columbia undergraduate science journal open-access publication | http://cusj.columbia.edu 9 cusjvolume 6spring 2012 columbia undergraduate sci j http://cusj.columbia.edu % ,2 /2 * < cusjcolumbia undergraduate science journal research articles 10cusj volume 6spring 2012 columbia undergraduate sci j http://cusj.columbia.edu % ,2 /2 * < touch sensitivity via converting detected mechanical forces into electrical signals is a vital survival mechanism in most organisms, a behavior known as mechanosensation. to understand normal function of the mechanosensory system and its complex neuronal circuitry, defective genes can be studied in animals that are de!cient in mechanosensation (touch insensitive). touch insensitive mutants were !rst developed in the nematode caenorhabditis elegans (c. elegans), which led to the discovery of a mechanosensory complex with the necessary components to allow certain cells to detect touch to the body. mechanosensation is well studied as a basic behavioral component in c. elegans since the organism’s entire genome has been mapped. investigating mechanosensation is much more challenging in complex eukaryotes because the mechanosensing cells and their scarce transducing molecules are more di$cult to isolate and observe (chal!e and bounoutas, 2007). development of touch sensitivity can be studied in c. elegans by analyzing genes expressed in a specialized set of 30 neurons that detect mechanosensation: mechanoreceptor neurons (mrns). #is experiment investigates expression mutants (with altered neuronal touch receptor-speci!c gene expression) and morphological mutants (with defects on genes that a"ect neuronal outgrowth, number, or branching patterns) of two types of mrns (touch receptor neurons (trns) and flp neurons) in order to better characterize mrn di"erentiation. trns are an important type of mrns because they exhibit the greatest electrophysiological response to mechanical stimuli (kamkin and kiseleva, 2007). two other types of mrns in c. elegans are pvd neurons, a pair of posterior interneurons that mediate harsh touch, and flp neurons, a pair of ciliated mechanosensory neurons that detect touch at the very tip of the head. collectively, trns (six cells), flps (2 cells) and pvds (2 cells) comprise one-third of all mrns (goodman, 2008) (figure 1). several genes have been identi!ed as necessary for differentiation of the six touch receptor neurons (alml, almr, plml, plmr, avm and pvm). mec-3 (mec for identifying factors that control mechanoreceptor neuron development in c. elegans alexis tchaconas 1, ��0duwlq�&kdo¿h1 1department of biological sciences, columbia university, new york, ny 10027 copyright: © 2012 the trustees of columbia university, columbia university libraries, some rights reserved, tchaconas, et al. received 12/15/2012. accepted 1/31/2012. published 4/1/2012 *to whom correspondence should be addressed: 1018 fairchild center, columbia university, new york, ny 10027, e-mail: aet2129@columbia.edu abstract introduction figure 1 mechanosensory neurons in c. elegans figure 2 combinatorial mrn differentiation dou and chal!e, 2011), but it is not known why sto-5 is not expressed in the trns; this experiment aims to !nd genes that are involved in this process. past studies have isolated egl-44 and egl-46 as two genes that restrict trn fate in the flp neurons: mutations of either gene cause the flp neurons to transform into cells that resemble trns (mitani et al., 1993) (wu, duggan and chal!e, 2001). however, the genes that restrict flp fate in other cells remain unknown. we hypothesize that another gene or network of genes is restricting sto-5 expression to flp neurons and is not allowing mec gene expression in the flps (which inhibits trn fate). one morphological mutant was identi!ed through this study, exhibiting irregularly wavy trn axonal processes. general methods and strains to identify genes needed for the di"erential expression of trns and flps, wild-type animals (tu3813 strain) were mutated. #ese c. elegans strains were maintained on op50 seeded agar plates at 20oc as outlined by brenner (1974). #e expression of sto-5 and mec-17 was observed in the wild-type strain using $uorescent protein tags: sto-5 was tagged with green $uorescent protein (sto5p::gfp) and mec-17 was tagged with red $uorescent protein (mec-17p::rfp). sto-5 expression was used for flp analysis since it is characteristic of flp di"erentiation, and mec17 was used for trn observations since it is needed for sustained trn di"erentiation. since c. elegans are transparent, these $uorescently tagged genes are visible under a dissecting microscope. gene expression was monitored under a dissecting microscope, looking for any changes in expression upon mutagenesis (figure 4 & figure 5 are images from the dissecting microscope of the $uorescent protein tagged flp and trn cells in a wild-type animal). #e wild-type animals were synchronized to the same stage (l4) prior to mutagenesis using adecontaminating solution (20% bleach). fluorescence imaging #e appearance of flps and trns in the wild-type strains were observed and characterized via a stereo-$uorescence dissecting microscope (leica mz12) powered by a uv light source (kramer scienti!c corporation) coupled with the x-cite 120 fluorescence imaging system. axiovision 4.8.2 software was used to further characterize flps and trns. ems mutagenesis mutagenesis was performed on the wild-type tu3813 strain with a standard protocol involving ethyl methanesulfonate (ems) (brenner, 1974): two to three plates of healthy tu3813 l4 animals (approximately 100-300 animals) were mutagenized, and 30 of the progeny animals (p0) were picked and plated individually, and stored at 25oc for three days. after three days, four animals from each p0 plate were plated individually (f1 animals) onto new plates and then stored at 25oc for three days (producing f2 animals). #ese f2 plates were manually screened three days later for any apparent abnormalities in flp or trn expression. f2 animals with noticeable abnormalities in either set of cells were picked individually onto new plates and their progeny were observed from subsequent generations to determine if the mutation is heritable (truebreeding) (figure 6 schematically outlines the mutagenizing process). mutant screens via !uorescence imaging #e sto-5 promoter was expressed with gfp (sto-5p::gfp), and the mec-17 promoter with rfp (mec-17p::rfp) to observe and compare flps and trns among the wild-type and mutant strains. #e $uorescent constructs are fucolumbia undergraduate science journal open-access publication | http://cusj.columbia.edu 11 cusjvolume 6spring 2012 columbia undergraduate sci j http://cusj.columbia.edu % ,2 /2 * < figure 3 unc-86 and mec-3 are both required for trn and flp differentiation, although they express different genes upon differentiation. sions of promoters to gfp and rfp injected and integrated into the wild-type tu3813 animals. f2s were !rst screened abnormal expression of the gfp and rfp markers, indicating abnormal gene expression. #en, the morphology of flps and trns was characterized in identi!ed mutant strains, noting the shape (axonal branching patterns), position, and changes in the number of flps or trns. any animals with abnormal traits in these cells were isolated and screened for the abnormality over several generations. if the isolated mutant consistently produced progeny with the same abnormality, the mutation was further investigated through genetic crosses. characterizing mutants via genetic crosses #e isolated mutants were crossed with the n2 wild-type strain and then with the dpy-5 mutant strain to determine the nature of the mutation: whether it was dominant or recessive, and x-linked or autosomal. two successive crosses were set up: the !rst was a cross between three hermaphrodite mutants and seven n2 (wild-type) males, and the second was a cross between seven of the male mutant progeny and three dpy-5 hermaphrodites. dpy-5 mutants are characterized by a short, “dumpy” body (dpy phenotype) due to the disruption of the dpy-5 gene, which encodes a cuticle procollagen (#acker, sheps and rose, 2006). #e dpy-5 animals were used to distinguish hermaphroditic self-progeny (dpy/dpy genotype with dpy phenotype) from hermaphroditic cross progeny (m/dpy genotype with normal phenotype, since dpy-5 is a recessive mutation). all animals crossed were in the l4 stage, a standard procedure used in order to enable mating between males and hermaphrodites and to minimize hermaphroditic self-progeny (brenner, 1974). six mutant strains exhibit abnormal flp and trn features seven mutageneses were performed on the tu3813 strain, yielding 710 observable f1 mutant animals. 32 candidate mutants were isolated from the mutant screens, six of which produced heritable flp and/or trn defects. one mutant is true-breeding for abnormal trn morphology among the six viable mutants, two mutants exhibited abnormalities in flp neuron quantity and appearance. observed morphological mutations included multiple and vertically moving flp neurons, and abnormal trn process morphology (wavy, crossed, or shortened). expression mutations involved simultaneous flp and trn expression when observed in the gfp channel on the dissecting microscope. over successive generations, only one mutant (4a1) retained its mutant phenotype (‘true-breeding’). #e mutation is particularly observable along the processes of alm and plm neurons, producing irregularly wavy axons in these animals that are noticeably distinct from the straight axons in wild-type animals (figure7-12 display side-by-side comparisons of tu3813 wild-type trn morphology and 4a1 mutant trn morphology from the anterior, midbody, and posterior sections of the body). cusjcolumbia undergraduate science journal research articles 12cusj volume 6spring 2012 columbia undergraduate sci j http://cusj.columbia.edu % ,2 /2 * < materials and methods figure 4 sto-5p::gfp expression in wild-type tu3813 animals figure 5 phf���s��uis�h[suhvvlrq�lq�zlog�w\sh�78�����dqlpdov figure 6 growth timeline and observation scheme for a standard ems mutagenesis results figure 7 anterior view of tu3813 wild-type (wt) animal (left to right: alm process, alml). the alm process is straight as it extends from the cell body. to-5p::gfp expression in wild-type tu3813 animals 4a1 mutation characterization via genetic crosses !e "rst cross between hermaphrodite mutants and n2 (wild-type) males yielded male o#spring and hermaphrodite o#spring without the mutant phenotype (wild-type). !e second cross between the male mutant progeny and dpy-5 hermaphrodites also produced wild-type male and hermaphrodite o#spring (figure 13 summarizes the cross). columbia undergraduate science journal open-access publication | http://cusj.columbia.edu 13 cusjvolume 6spring 2012 columbia undergraduate sci j http://cusj.columbia.edu figure 8 anterior view of 4a1 mutant (avm), with irregularly wavy alm process (left to right: alm process, avm, almr). the alm process is wavy as it extends from the almr cell body !is study isolated genes that disrupt the function or development of gene expression, or morphology in the trns and flp neurons (by isolating an observed phenotype and then identifying the responsible gene). based on the speci"c abnormality observed in a mutant, the normal function of the mutated gene can be inferred from characteristics in its absence, and its role in mrn di#erentiation can then be established. !e seven mutageneses performed enabled screening of 710 f1 animals, yielding 32 candidate mutants, only one of which bred true. typically, there is a 1/2000 probability of "nding a mutant via mutagenesis, so an appropriate sample size would be screening 10,000 f1 animals, and their f2 progeny (brenner, 1974). !e f1 population size in our study was relatively small, and thus should be increased in the future by mutagenizing and screening more animals. as mentioned above, this work has isolated one viable mutant via ems mutagenesis (4a1). !is mutant has displayed a consistent, morphological mutant phenotype in every successive generation, indicating that it is a truebreeding mutant. !e 4a1 mutant phenotype, which involves irregularly wavy trn morphology (along the alm and plm processes), suggests that the induced mutation may have disrupted the axonal development of the alm and plm neurons. although none of the putative mutants with flp abnormalities appeared to be true-breeding mutants, the only true-breeding mutant identi"ed exhibits interesting trn abnormalities. while "nding a trn or flp morphological mutant was a secondary goal of the project (the primary goal was to "nd expression mutants), the mutant nonetheless appears to have a signi"cant mutation that may reveal more about mrn functionality. !us, the 4a1 mutant is worthy of further exploration and genetic characterization. !e results of the cross between 4a1 mutant hermaphrodites and n2 males, followed by the cross of the male o#spring and dpy-5 hermaphrodites indicate that the mutation is autosomal recessive. !e male o#spring only inherited one x chromosome, from the mutant hermaphrodite. since the cross with 4a1 mutant hermaphrodites and n2 males produced male o#spring that did not exhibit the mutant phenotype, the mutation is not x-linked, which is further veri"ed by the next cross. furthermore, from both crosses the o#spring do not show a mutant phenotype, which indicates that the mutation is recessive. an additional cross will be performed to recon"rm the mutant allele’s mode of inheritance, by crossing dpy-5 (dpy/dpy) hermaphrodites with n2 (+/+) males. !e heterozygous male o#spring (dpy/+) will then be crossed with 4a1 mutant hermaphrodites (m/m), producing two types of heterozygous progeny (m/dpy and m/+). since a recessive hermaphroditic allele was crossed with a male to produce heterozygote mutant o#spring in the second generation, the cross further clari"es whether the mutation is dominant or recessive (based on the second cross male progeny’s phenotype – if all males are mutant, the mutation is dominant; if the males are wildtype, the mutation is recessive). future studies will continue to follow the 4a1 mutant strain, speci"cally to determine on which chromosome the mutation is located. first, additional two-step crosses will be performed with mutants that have a known chromosomal location (i.e. dpy-5, with an autosomal recessive mutation on chromosome i). such crosses will indicate which chromosome the 4a1 mutation is on, based on the progeny phenotypes. !en, complementation tests will be performed to determine whether the mutation is on a previously isolated gene, or an entirely unexplored gene. in addition, if it is a novel mutation for mrn di#erentiation (based on the chromosome and gene on which it is located), then the mutant strain will have its genome fully sequenced. genomic sequencing of a 4a1 mutant animal can be compared with the wild-type genome in order to identify the gene that contains the mutation. additional mutageneses will also be performed (increasing f1 sample size) to "nd more genes potentially involved in the combinatorial regulation of mrn di#erentiation. once these genes have been identi"ed via genetic sequencing, they can be fused with gfp to further characterize their normal expression. with an understanding of the normal and abnormal function of these combinatorial genes, each gene’s role in flp and trn di#erentiation can be determined. other experiments can be performed to determine how these genes mechanistically function in the de"nition of mrn development or function. !e irregularly wavy trn processes observed in the 4a1 mutant could be due to a defect in axonal attachment. another researcher in our lab has found a mutant with a cuticle attachment defect (from a mutation on the gene mec-5), so the 4a1 mutant will be compared to this other mutant’s phenotype. such comparisons can determine phenotypic similarities between mutants, and potentially identify genes that have a combinatorial interaction with one another in mrn di#erentiation. once the combinatorial mechanisms regulating di#erentiation of the touch neurons is better understood, more insight into mechanosensory systems in higher organisms can be gained, due to the cross-organismal homology of touch. we thank the chal"e lab (dr. irini topalidou, dr. emalick njie, dr. charles keller, xiaoyin chen, chaogu zheng, yushu chen, and ana pozo). aet was funded by the columbia department of biological sciences from the cusjcolumbia undergraduate science journal research articles 14cusj volume 6spring 2012 columbia undergraduate sci j http://cusj.columbia.edu discussion figure 9 mid-body view of a tu3813 wt animal’s plm process displays its normally straight morphology. figure 10 mid-body view of a 4a1 mutant’s plm process displays an irregularly wavy morphology throughout its midsection figure 11 3rvwhulru�ylhz�ri�d�78�����:7�dqlpdo¶v�3/05�/�dqg�qrupdoo\� vwudljkw�3/0�surfhvv��ohiw�wr�uljkw��3/0�surfhvv��3/05�/�� % ,2 /2 * < figure 12 4a1 mutant with irregularly wavy posterior process (plm) (left to right: pvm, plm process), exhibiting an apparent deviation from the wt process morphology (in figure 11). figure 13 4a1 crosses for mutational characterization symbol key acknowledgments % ,2 /2 * < summer undergraduate research fellowship. bounoutas, a., and chal!e, m. 2007. touch sensitivity in caenorhabditis elegans. p"ugers arch. 454, 691–702. brenner, s. 1974. #e genetics of caenorhabditis elegans. genetics 77, 71–94. chal!e, m., & au, m. 1989. genetic control of di$erentiation of the caenorhabditis elegans touch receptor neurons. science 243, 1027 – 1033. finney, m., ruvkun, g., & horvitz, h.r. 1988. #e c. elegans cell lineage and di$erentiation gene unc-86 encodes a protein with a homeodomain and extended similarity to transcription factors. cell 55, 757 – 769. goodman, miriam b. “mechanosensation.” wormbook (may 2008): 9 june 2011. hobert, oliver, and heiner westphal. “functions of limhomeobox genes.” trends in genetics 16.2 (2000): 75-82. trends in genetics. 21 jan. 2012. kamkin, andre glebovich, and irina s. kiseleva, eds. mechanosensitive ion channels. new york: springer, 2007. #e tavernarakis lab. 3 aug. 2011. mitani, s. 1995. genetic regulation of mec-3 expression implicated in the speci!cation of the mechanosensory neuron cell types in caenorhabditis elegans. dev. growth di$er., 37, 551 557. mitani, s., du, h., hall, d. h., driscoll, m., & chal!e, m. 1993. combinatorial control of touch receptor neuron expression in caenorhabditis elegans. development 119, 773 – 783. stewart gw, argent ac, dash bc, 1993. stomatin: a putative cation transport regulator in the red cell membrane. biochim biophys acta 1225,15–25 #acker c, rose am, and sheps ja. “caenorhabditis elegans dpy-5 is a cuticle procollagen processed by a proprotein convertase.” cellular and molecular life sciences 63.10 (2006): 1193-204. topalidou, irini, alexander van oudenaarden, and martin chal!e.” caenorhabditis elegans aristaless/arx gene alr-1 restricts variable gene expression.” proceedings of the national academy of sciences 108.10 (2011): 4063-4068. topalidou, irini, and martin chal!e. “shared gene expression in distinct neurons expressing common selector genes.” proceedings of the national academy of sciences 108.48 (2011): 1-6. pnas. 25 feb. 2012. way, j.c. and chal!e, m. 1988. mec-3, a homeobox-containing gene that speci!es di$erentiation of the touch receptor neurons in c. elegans. cell 54, 5–16. way, j. c., & chal!e, m. 1989. #e mec-3 gene of caenorhabditis elegans requires its own product for maintained expression and is expressed in three neuronal cell types. genes dev. 3, 1823 – 1833. wu j, duggan a, chal!e m. inhibition of touch cell fate by egl44 and egl-46 in c.elegans. genes dev 2001; 15, 789-802. zhang s, arnadottir j, keller c, caldwell ga, yao ca, chal!e m (2004) mec-2 is recruited to the putative mechanosensory complex in c. elegans touch receptor neurons through its stomatin-like domain. curr biol 14, 1888 –1896. zhang, y., ma, c., delohery, t., nasipak, b., foat, b.c., bounoutas, a., bussemaker, h.j., kim, s.k., chal!e, m., 2002. identi!cation of genes expressed in c. elegans touch receptor neurons. nature 418, 331–335. columbia undergraduate science journal open-access publication | http://cusj.columbia.edu 15 cusjvolume 6spring 2012 columbia undergraduate sci j http://cusj.columbia.edu % ,2 /2 * < references breast cancer affects one in eight women per year, and 70% of patients with stage iv breast cancer develop metastases in bone, fdxvlqj�olih�wkuhdwhqlqj�vlgh�hiihfwv��:h�h[dplqh�wkh�xvh�ri�dq�h[sodqw�v\vwhp�wr�plplf�euhdvw�fdqfhu�phwdvwdvlv�wr�erqh�zlwkrxw� confounding cell types and to determine the role that osteocytes, the bone mechanosensing cells, may have in the development of metastasis. using the explant system, a custom cell seeder and sterile cell culture techniques, we introduced metastatic mda-mb-231/ *)3�fhoov�wr�d�wkuhh�glphqvlrqdo�erqh�pdwul[�zlwk�rvwhrf\whv�rqo\��&rqirfdo�lpdjlqj�frq¿upv�wkdw�euhdvw�fdqfhu�fhoov�zhuh��lq�idfw�� successfully seeded onto bone cores, mimicking metastasis. though additional experiments will be necessary to determine the importance of breast cancer-osteocyte interactions, this study shows that the explant system is a viable methodology for studying breast cancer in bone. cancer is a devastating disease that is responsible for thirteen percent of deaths worldwide and has a$ected countless families and individuals throughout the world (cancer, world health organization). it a$ects 1 in 8 women every year (u.s. breast cancer statistics). breast cancer originates from the inner lining of the lobules that supply the milk ducts in the breast (wolf et al., 2003). stage iv breast cancer is metastatic, meaning that it is violent and transcends the host organ (the breast) and spreads to a secondary site. #e cancer that metastasizes is still considered breast cancer. it has been reported that up to 70% of stage iv breast cancer patients will experience some form of metastasis of breast cancer to bone (roth et al., 2009). patients often experience pathological fractures, intense pain, hypercalcemia, and various nervous compression complications (zhang et al., 2010). #ese devastating e$ects are caused by an imbalance of bone remodeling, which involved the interactions of the three main bone cell types. bone is comprised of three types of cells: osteocytes (ocy), osteoblasts (ob) and osteoclasts (ocl). osteocytes are the primary mechanosensing cells in bones (burger et al., 1995). #ey regulate the activity of osteoblasts and osteoclasts. osteocytes are “trapped” in the mineralized bone matrix, and are thus they are thought to have only signaling functions, both intercellular and intracellular. after osteocytes sense a mechanical load, that load is transduced into a chemical signal is sensed by the cells. #is stress is translated into a biochemical signal that is communicated to the osteoblasts and osteoclasts, the bone forming and bone resorbing cells (burger et al., 1995). osteoblasts synthesize the bone matrix, which is subsequently deposited and calci!ed to become bone mineral. when osteoblasts secrete too much matrix, they become stuck in the bone, and as a result they completely di$erentiate into osteocytes (saladin, 2007; buckwalter et al. 1995). osteoclasts, on the other hand, resorb bone. to accomplish this, they use their “ruf"ed” membrane (as shown in figure 1) to create a seal around bone and then pump enzymes and hydrochloric acid to degrade the matrix (saladin, 2007; buckwalter et al. 1995). #ere are two types of metastasis: osteolytic and osteoblastic. osteolytic metastases break down bone and are the most common type of metastasis for breast cancer. osteoblastic lesions, characterized by excess bone formation, a$ect 15-20% of patients. mixed types also exist, wherein the patient experiences unnecessary bone excess as well as dearth (zhang et al., 2010). #e large majority of stage iv breast cancer cases end in metastasis to bone because bone has high levels of growth factors that breast cancer uses to survive. we can see that bone is a likely candidate for breast cancer metastasis due to the presence of “transforming growth factor b (tgfb), insulin-like growth factors i and ii (igf), !broblast growth factors (fgfs), platelet-derived !e veri"cation of a novel explant system used to determine the role of osteocytes in the breast cancer vicious cycle sonia bansal*, genevieve n. brown, x. edward guo %rqh�%lrhqjlqhhulqj�/derudwru\��'hsduwphqw�ri�%lrphglfdo�(qjlqhhulqj��&roxpeld�8qlyhuvlw\��1hz�<run�1<� copyright: © 2012 the trustees of columbia university, columbia university libraries, some rights reserved, bansal, et al. received 12/22/2012. accepted 1/31/2012. published 4/1/2012 *to whom correspondence should be addressed: bone bioenjlqhhulqj�/derudwru\��'hsduwphqw�ri�%lrphglfdo�(qjlqhhulqj�� columbia university, new york ny. sb3218@columbia.edu abstract introduction cusjcolumbia undergraduate science journal research articles 16cusj volume 6spring 2012 columbia undergraduate sci j http://cusj.columbia.edu % ,2 ( 1 * ,1 ( ( 5 ,1 * columbia undergraduate science journal open-access publication | cusj.columbia.edu spring 2013 | volume 7 columbia undergraduate science journal 18 the evolving role and nature of gene mutations in the neuropathology of autism spectrum disorders alexis tchaconas1* 1department of biological sciences, columbia university, new york, ny 10027 abstract recent research increasingly suggests that autism spectrum disorders (asds) can be caused by genetic factors. asds are one of the fastest growing neurodevelopmental disorders, encompassing a spectrum of disorders marked by difficulties with social interaction, communication (verbal and non-verbal), and unusual, repetitive behaviors. the etiology, or specific factors that cause a disorder, are is relatively uncertain for asds. consequently, viable treatment options for asds have received greater attention among autism researchers – in particular, the neurological consequences of genetic mutations found in people with asds. in the past year, the literature has presented many novel treatments to address this promising neurobiological etiology of asds. the current trajectory of autism research, supported by a wealth of studies connecting genetic mutations in neural substrates to the core symptoms of asds, suggests a greater appreciation for and understanding of the genetic complexity that underlies asds. additionally, results of asd twin studies have encouraged consideration of environmental factors that may act as triggers for gene mutations associated with asds. genetic and environmental factors are increasingly accepted as joint contributors to the etiology of asds, rather than isolated factors strictly regulated by nature or nurture, respectively. while there has been substantial progress on the geneticneurobiological front of asd research in the past decade, there is a burgeoning avenue of geneticenvironmental asd research. with impending changes to the definition of asds in the newest edition of the diagnostic and statistical manual (5th edition) in may 2013, it is likely that these multifactorial etiologies of asds will receive even greater attention in the field. keywords: autism spectrum disorders; etiology; de novo mutations; copy number variations; epigenetics introduction autism spectrum disorders autism spectrum disorders (asds) are a group of neurodevelopmental disorders with a markedly growing prevalence: more children are diagnosed with asds each year in the united states than are diagnosed with aids, cancer, and diabetes combined (autism speaks, 2012). asds affect 1 in 88 children in the united states and are four times more prevalent in males than they are in females 1 in 54 boys affected compared with 1 in 252 girls affected (mmwr, 2008). prevalence estimates of asds have increased significantly in recent years, from 1 in 150 in the year 2000 to 1 in 88 in the year 2008 (see figure 1), yet it is unclear whether this is a true increase or merely a reflection of expanded diagnostic criteria (fombonne, 2009). regardless, the measured increase has encouraged researchers to investigate possible etiologies of these disorders, paying particular attention to genetics given the sex difference in prevalence statistics. the core deficits of asds include delayed speech development and communication skills (verbal and non-verbal), impaired social interaction, and restricted, repetitive behaviors (apa, 1994). ‘autism’ first received its own classification in the diagnostic copyright: © 2013 the trustees of columbia university, columbia university libraries, some rights reserved, sorrento, et al. received 1/1/2013. accepted 2/1/2013. published 1/1/13 *to whom correspondence should be addressed: department of biological sciences, columbia university, new york ny 10027, email: aet2129@columbia.edu mailto:aet2129@columbia.edu columbia undergraduate science journal open-access publication | cusj.columbia.edu spring 2013 | volume 7 columbia undergraduate science journal 19 and statistical manual of mental disorders, third edition (dsm-3) in 1980 (apa, 1980). when the dsm-4 was updated in 1994, “autistic disorder” was subsumed under a new umbrella category of “pervasive developmental figure 1: identified prevalence of autism spectrum disorders, 2000 – 2008. figure 1 is adapted from the center for disease control (cdc)’s morbidity and mortality weekly reports from 2000-2008. (http://www.cdc.gov/ ncbddd/autism/imagesautism_data_graphic2012.jpg) disorders” (pdd), which included four other disorders: asperger's disorder, childhood-onset disintegrative disorder (codd), rett syndrome, and pdd-not otherwise specified (pdd-nos). although each of these conditions shared some features, the dsm-4 attempted to differentiate these different conditions with respect to age of onset, symptom severity, prognosis, and associated features (apa, 1994). even though these disorders are now generally known as asds, they were never labeled this way in the dsm-4. efforts have been made to officially list these disorders under the single name ‘autism spectrum disorder’ in the newest edition of the dsm-5, due for release in may 2013. this revision is mainly influenced by inconsistent diagnoses in time and among physicians, which may be minimized by consolidating asds into ‘asd’. i hypothesize that the semantic changes in diagnostic criteria will clarify the controversy between actual asd prevalence versus overdiagnosis from broad diagnostic criteria, especially when data reflecting identified prevalence rates under the more narrow dsm-5 definition of asd are available (tchaconas & adesman, 2013). the diagnostic criteria for “asd” in the dsm-5 will reflect several major changes in the specific criteria that qualify a diagnosis (see figure 2). columbia undergraduate science journal open-access publication | cusj.columbia.edu spring 2013 | volume 7 columbia undergraduate science journal 20 figure 2: dsm-5 proposed revisions of diagnostic criteria for autism spectrum disorders (asds). figure 2 is original (featured in tchaconas & adesman, 2013), and the information is adapted from the asd criteria in american psychiatric association’s dsm5 development (apa 2011). (http://www.dsm5.org/proposedrevision/pages/proposedrevision.aspx?rid=94) the three domains previously defining asds in the dsm-4 are consolidated into two domains in the dsm-5: “social and & communication deficits” and “fixed interests & repetitive behaviors” (apa, 2011). social and communication deficits are combined to improve diagnostic specificity without compromising sensitivity because they are considered clinically inseparable domains. furthermore, numerical severity levels will be added to enhance diagnostic specificity (see figure 3). http://www.dsm5.org/proposedrevision/pages/proposedrevision.aspx?rid=94 columbia undergraduate science journal open-access publication | cusj.columbia.edu spring 2013 | volume 7 columbia undergraduate science journal 21 figure 3: dsm-5 proposed revisions of severity level specification for autism spectrum disorders (asds). figure 3 is original (featured in tchaconas & adesman, 2013), and the information is adapted from the asd criteria in american psychiatric association’s dsm-5 development (apa 2011). (http://www.dsm5.org/proposedrevision/pages/proposedrevision.aspx?rid=94) data analyses suggest that these dsm-5 coredeficit criteria will yield the highest crossdiagnostic consistency yet for people with autism (jabr, 2012). as displayed in figure 2, dsm-5 asd diagnostic criteria will now require children to exhibit all three symptoms in social communication and interaction (a) and two of four symptoms in repetitive behavior domain (b); the symptoms must present in early childhood (c) and limit everyday functioning (d). though it is unclear how these changes will impact autism prevalence rates, the most thorough case analysis of the effect of the revised criteria on patients currently diagnosed with asds suggests that 91% of these patients will retain their diagnosis under these criteria (huerta et al. 2012). the actual diagnostic efficacy of these revisions will play an integral role in the success of etiological research, as diagnostic criteria dictate the core symptoms of autism around which researchers will design their future studies. until recently, the core symptoms of asds were largely untreatable, as most research focused on alleviating the disruptive behavioral symptoms rather than on targeting the core deficits. in recognizing the neurobiological component of the disorder, current research into genetic factors influencing neurodevelopment has created a promising model for asd brain pathology. the extensive neurogenetic autism research over the past few years has supported the development of asd treatments that can target the unique neural circuitry that results from asd neuropathology. since genes code for proteins and proteins are responsible for specific functions in their area(s) of expression, dysfunctional genes responsible for neurological function in patients with and asd are of particular interest in the neurobiological front of autism research. this review highlights the current advances in asd etiological research, with a focus on the wellestablished neurobiological advances and our growing knowledge of environmental contributions, and how these advances are shaping future research efforts in light of the evolving definition of asd. http://www.dsm5.org/proposedrevision/pages/proposedrevision.aspx?rid=94 columbia undergraduate science journal open-access publication | cusj.columbia.edu spring 2013 | volume 7 columbia undergraduate science journal 22 the genetic basis of asds interest in the genetic heritability of asds originated with the first twin study reporting a significant difference between monozygotic and dizygotic twins (folstein & rutter, 1977); subsequent studies reported even higher monozygotic concordance rates and nearly 0% dizygotic concordance (szatmari, 2011). in the past decade, genetic susceptibility to asds has continued to be a prime focus among researchers, with studies reporting asd concordance rates as high as 90% in monozygotic twins and as low as 10% in dizygotic twins (mendelsohn & schaefer, 2008). these polarized concordance rates have implied a strong genetic component to asds, attributing almost all variances in phenotypic expression to heritable factors (szatmari, 2011). this pattern of inheritance is exhibited in the single-gene disruption known as fragile x syndrome, the most common inherited cause of asds and intellectual disability (aap, 2012), attributed to mutation on the fmr1 gene (henderson et al., 2012). however, nearly two years ago the results of the largest asd twin study, the california autism twin study, found asd concordance rates to be much lower than previously reported (hallmayer et al., 2011). based on the total of 192 twin pairs born between 1987 and 2004, with at least one twin affected by either autism or a milder asd, asd was 55% attributable to the twins’ shared environmental factors, a much greater percentage than previously predicted (hallmayer et al., 2011). the california twin population was ethnically, demographically, and socioeconomically diverse, with a fairly large sample size, providing convincing evidence for the environmental contribution to asd etiologies. despite these findings, the genetic component of asds cannot be simply rejected; instead, the data revise the genetic explanation of asds to include environmental factors that work with, or possibly even trigger, gene mutations linked to asds. the new twin study evidence by revealing that asd concordance rates for dizygotic twins were previously underestimated, has thus shifted the field’s focus toward potential prenatal and postnatal environmental triggers for gene mutations (stoltenberg et al., 2010). although asd etiologies were once divided into separate genetic or environmental etiologies, there is now a growing acceptance of epigenetics– the theory that environmental factors influence gene expression to produce mutations in a nonmendelian manner. it is believed that this combined etiological hypothesis can explain the presence of these non-heritable, de novo gene mutations increasingly observed in asd populations (sanders et al., 2012; gilman et al., 2011; sebat et al., 2007). de novo gene mutations and copy number variations most known mutations without an identified genetic etiology attributed to asds are de novo gene mutations: rare genetic variations that are found in neither of the individual’s parents. the risk for de novo mutations has been associated with increased parental age, one of many environmental factors that may account for the spontaneous appearance of these mutations (sanders et al., 2012). it also appears that the more de novo mutations found in the same gene, the greater the chance that these are asd risk-associated mutations (see figure 4). columbia undergraduate science journal open-access publication | cusj.columbia.edu spring 2013 | volume 7 columbia undergraduate science journal 23 figure 4: identification of multiple de novo mutations in the same gene reliably distinguishes risk-associated mutations. figure 4 is adapted from sanders et al., 2012. (a) reflects a simulation experiment modeling the probability of observing two independent nonsense/splice site de novo mutations in a single brain-expressed gene among unrelated subjects. (b) is the application of the simulation described in (a) to predict of the number of genes found to carry two or more nonsense/splice site de novo mutations for a sample of varied size (indicated on the x-axis). (c) is a repeat of the simulation in (a) for non-synonymous de novo mutations. three or more independent nonsynonymous de novo mutations in a brain-expressed gene offer significant evidence for association with asd, though this sample is limited by its size and heterogeneity models. a common type of de novo mutation involves copy number variations (cnvs), or an abnormal number of copies of a dna segment (i.e. a deletion or duplication of a gene) that often causes dysfunction of the gene’s normal protein. fernandez et al.’s initial implication of a cnv on the contactin4 gene has led to research linking other contact in family candidate genes vital for the functioning of the central nervous system (i.e. cntnap2) to small populations with asds (bakkaloglu, 2008). gilman and colleagues’ analysis of de novo variants suggests that asd is a disorder of ‘perturbed synaptogenesis’, based on their model of a large biological network of genes affected by rare de novo cnvs in asd, known as netbag, which contains genes primarily related to synapse development, axon targeting, and neuron motility (gilman et al. , 2011). their model supports that theory that networks underlying complex human phenotypes – i.e., asd can be identified by a network-based functional analysis of rare genetic variants (see figure 5). columbia undergraduate science journal open-access publication | cusj.columbia.edu spring 2013 | volume 7 columbia undergraduate science journal 24 figure 5: an overview of network-based analysis of genetic associations (netbag) for identifying significant and functionally related gene networks with de novo mutations. figure 5 is adapted from gilman et al., 2011. (a) is a background network of human genes constructed, with nodes denoting genes and edges indicating the probability of two genes producing the same genetic phenotype, (b) demonstrates one or two genes selected from each of de novo cnv region to form a cluster. the genes are mapped to the probability network and a combined score is computed for each cluster based on the interactions between its genes, (c) exhibits the greedy search procedure used to identify the cluster of maximal score, (d) determines the significance of the cluster with the maximum score by comparing it to the distribution of maximal scores from randomly selected genomic regions with similar numbers of genes. de novo mutations appear to produce the brain pathologies and core symptoms of asds in some cases. given that new gene mutations are found each year in people with asds, there are likely hundreds of genes linked to asds that have yet to be found. accordingly, it may be more effective to identify the neural circuits, or networks of genes, involved in functions often impaired in the brains of those with asds. the identification of asd-specific neural circuits, coupled with neuroimaging of people with asds performing specific tasks representative of the disorder’s core deficits, can perhaps enable the mapping of asd neuropathology in the brain. in the coming years, it will also be crucial to determine whether de novo gene mutations versus inherited gene mutations implicated in asd etiology induce identical or distinct brain pathologies. while treatment-based genetic studies are limited by the presently sparse representation of mutations in small asd subpopulations, investigation of these genes and their proper functions opens a valuable pathway for understanding the complex mechanisms that propagate asds. this has led to the concurrent growth of asd neuroimaging studies as a means of analyzing the effects of columbia undergraduate science journal open-access publication | cusj.columbia.edu spring 2013 | volume 7 columbia undergraduate science journal 25 asd-related gene mutations in their respective neural substrates. neuroimaging reveals the neuropathology of asds among the increasingly sophisticated neuroimaging technologies available today, functional magnetic resonance imaging (fmri) is an effective research tool for isolating and studying frequently disrupted neural systems connected to asds (aap, 2012). fmri is a technique that measures brain activity by detecting changes in blood oxygenation and blood flow which is a proxy for neural activity. it is a more helpful technique than the previously used neuroimaging techniques. it is thought that brain abnormalities found consistently among patients with asds are a product of underlying genetic mutations that influence the expression of key proteins in the brain, and thus result in inefficient neural circuitry, especially with neuronal migration and cortical organization (sowell & bookheimer, 2012). current studies are using neuroimaging to isolate these problem regions in the brains of asd patients as opposed to the brains of normal controls, noting differences in gene expression, brain architecture, and behaviors linked to asds (anagnostou & taylor, 2011, wolff et al., 2012). for example, language skills in children with autism who exhibit language impairment was compared to the language skills in age-matched controls by presenting these children with recordings of their parents talking to them while measuring each child’s brain activity during the task (lai et al., 2011). differential activation in autistic versus control subjects upon passive speech stimulation was localized to the superior temporal gyrus (stg), a brain region that regulates language; this observed differential activation could potentially implicate fmri as an objective detector of language impairment in children with autism, and perhaps a general asd diagnostic test (lai et al., 2011). in addition to fmri, recent studies using diffusion tensor imaging (dti, a functional neuroimaging technique that measures water diffusion within a tissue) have identified a distinct white-matter fiber tract maturation pattern discernible in high-risk infants who eventually develop an asd or associated symptoms (wolff et al., 2012). the results indicate that abnormal development of the white-matter pathways in infant brains may precede asd symptom onset in the first year of life and can potentially be used as an early detection test for asds from 6 to 24 months. clinicians, who currently lack a medical test to diagnose autism, hope that identification of asd brain biomarkers will soon allow for earlier, more accurate asd-risk detection and that neuroimaging will evolve to eventually provide an objective diagnostic test. while careful analysis of the structural and volumetric measures from brain mris has yet to identify a consistent pattern of early brain development in children with asds, functional neuroimaging techniques have used these volume abnormalities to uncover atypical structural connectivity in the asd brain (anagnostou & taylor, 2011). therefore, functional neuroimaging techniques have emerged as highly informative neuroimaging tools for asd etiological research that will ultimately be used to relate imaging findings to behavior and the underlying genes, and the expression of these genes, which produce these neurobiological abnormalities. collectively, these neuroimaging studies suggest that a critical, preclinical period exists during which anatomical brain abnormalities that are found in neurodevelopmental disorders such as asds, do manifest before the associated behavioral symptoms. responding to these asd biomarkers with early behavioral interventions, once they can be more clearly defined, could presumably minimize asd severity and improve the child’s prognosis. though it must be further explored in the future, identifying biomarkers for asds during the critical neurodevelopmental period would be an invaluable tool for early detection and preventive interventions that could minimize the disorder’s severity. columbia undergraduate science journal open-access publication | cusj.columbia.edu spring 2013 | volume 7 columbia undergraduate science journal 26 neuropharmacological interventions genetic investigations of asds have led researchers to develop experimental medical treatments that target these genetic abnormalities in the brain. the neurobiology research-centered pharmaceutical company seaside therapeutics recently developed arbaclofen, an investigational drug for patients with fragile x syndrome that has functioned as a disease-modifying drug in preclinical mice models (henderson et al., 2012) due to significant improvements in social interaction and engagement observed in clinical trials (berry-kravis et al., 2012). the causative mutation of fragile x syndrome is in the fmr1 gene and involves a loss of the protein that regulates synaptic protein synthesis, which disrupts the function of the inhibitory neurotransmitter gaba. the drug targets this mutation via postnatal activation of the gabab receptors. the drug increases inhibitory neurotransmitter function which can reverse the pathological synaptic overstimulation caused by fragile x syndrome. arbaclofen was applied in vitro to the hippocampus of mice lacking the fmr1 protein, and was found to reduce the severity of fragile x. the positive results in these mice models led to ongoing human clinical trials this last year which showed, the potential of being the first viable drug to treat the core symptoms of fragile x in humans. in phase 2 of the clinical trials, researchers observed statistically significant improvements in social avoidance, a core symptom of fragile x syndrome (berry-kraviset al., 2012). given that arbaclofen was well tolerated by humans in this double-blind, placebo-controlled, clinical trial, it has the potential, with further testing, to become the first fda-approved medicine to treat core deficits of fragile x syndrome. by logical extension, if arbaclofen can improve the social functioning of children with comorbid asd and fragile x, it may also be effective in other asd populations or lead to the development of other effective medications. enzyme replacement therapy is another experimental treatment that has advanced to the level of clinical trials in the u.s. the treatment targets children with autism who present biomarkers suggesting enzyme deficiencies. these enzyme deficiencies then result in an inability to digest protein, thereby compromising amino acid production. a proprietary high-protease enzyme replacement formulation (cm-at) has just been reported to be significantly more effective than a placebo in treating core and noncore symptoms of children aged 3–8 years with autism in a randomized, double-blind, phase iii clinical trial (curemark, 2011, 2012). another potential therapeutic approach focuses on patients with autism, intellectual disability, and seizures because of a defect of branched chain amino acid (bcaa) metabolism. patients with this rare mutation have decreased levels of bcaas. novarino et al. (2008) used a genetically engineered ‘knockout mice’ model to demonstrate that these abnormal brain amino acid profiles and neurobehavioral deficits can be corrected with dietary supplementation, or the addition of foods rich in the depleted amino acids into a patient’s diet. the researchers also used skin samples from patients with this gene defect and converted them into neural stem cells; these neural cells functioned normally in the presence of an environment rich in the depleted amino acids, suggesting that this one very rare form of autism, at least, may be treatable (novarino et al., 2012). certain medications already on the market may be helpful in treating children with irritability related to an asd. a recent pilot study indicates that oral administration of nacetylcysteine (nac), an antioxidant and a glutamatergic modulator, displayed efficacy in treating disruptive symptoms in children with autistic disorder (hardan et al., 2012). glutamatergic dysfunction and redox imbalance are potential links to irritability in some forms of asd, originating from genetic disruptions (hardan et al., 2012). this 12-week doubleblind, randomized, placebo-controlled pilot investigation found that nac was helpful in managing irritability and was relatively well columbia undergraduate science journal open-access publication | cusj.columbia.edu spring 2013 | volume 7 columbia undergraduate science journal 27 tolerated. although there were benefits from nac treatment, the study is limited by its small sample size and narrow age range, so it should be repeated with a larger population and a broader age range to assess the feasibility of using nac as a routine treatment option for disruptive behaviors from autistic disorder and related disorders of repetition and compulsion (i.e. obsessive compulsive disorder). similarly, preliminary results of a large-scale, placebo-controlled, double-blind study suggest that nasally-administered oxytocin, a naturally occurring substance produced in the brain to regulate social abilities, improves the socialcommunicative core deficits of asds in children and adolescents, aged 7 to 18 years. fmri detected increased brain activity in regions known to process social information following oxytocin treatment, behaviorally translating to heightened social interaction (peart, 2012). as before, additional data demonstrating sustained core symptom improvements over longer time periods following administration of these therapeutics are needed to determine whether such interventions are safe and effective for use in individuals with asds. substance name disorder targeted therapeutic mechanism stage in clinical trials relevant studies arbaclofen fragile-x increase inhibitory neurotransmitter (gaba) activity that is disrupted by an fmr1 gene mutation. completed phase ii henderson et al., 2012; berry-kravis et al., 2012 high-protease enzyme replacement formulation (cm-at) asd correcting enzyme deficiencies that compromise amino acid production. completed phase iii curemark, 2011, 2012 branched chain amino acid (bcaa) metabolism enrichment rare form of autism with intellectual disability and seizures, induced by defect in bcaa metabolism. dietary supplementation with foods rich in the depleted amino acids. n/a – pilot study novarino et al., 2012 n-acetylcysteine (nac) autistic disorder (specifically, its disruptive symptoms) alleviating glutamatergic dysfunction and redox imbalance from genetic disruptions. n/a – pilot study hardan et al., 2012 oxytocin (nasallyadministered) asd (specifically, the social-communicative core symptoms) increases brain activity in regions processing social information. n/a – ongoing preliminary investigation preliminary results from ilanit gordon and kevin pelphrey 2012; peart, 2012 figure 6: summary of potential asd neuropharmacological interventions. figure 6 is an original figure. future directions results of genetic research of asds over the past decade suggest that mutations in hundreds of genes vital to the brain’s growth and circuitry may underlie the development of asds. the wealth of studies indicating that gene mutations cause some asd cases points to the growing body of literature supporting a genetic basis of asd neuropathology. accordingly, drug development studies will continue to search for ways to correct aberrant protein functioning resulting from the plethora of gene mutations columbia undergraduate science journal open-access publication | cusj.columbia.edu spring 2013 | volume 7 columbia undergraduate science journal 28 already positively connected to asd neuropathology. specifically, researchers are currently working to identify commonly disrupted molecular pathways in the brains of patients with asds, as these findings can lead to highly specific treatments. regardless of whether the exact cause of asds remains unknown, treatments targeting these disrupted neurobiological pathways can still correct the core deficits of asds, assuming that such disruptions are an etiology rather than a consequence of asds. furthermore, environmental factors that are increasingly believed to trigger asdrelated gene mutations will continue to be explored, especially in the case of de novo mutations that cannot be explained by typical mendelian inheritance patterns, and the new data reporting higher concordance rates among dizygotic twins with asds. twin studies in the future will likely analyze twins’ shared environment prior to the second year of life, when signs of asds typically manifest themselves (stoltenberg et al., 2010). more asd-related gene mutations will undoubtedly be discovered in the coming years; the challenge for scientists will be to consolidate the scattered mutations observed into specific asd etiologies, and to connect them to their potential environmental triggers. in this regard, it is likely that many asds will ultimately be classified as multifactorial genetic disorders. references 1. american academy of pediatrics (aap) (2012). autism toolkit, physician fact sheet. 2. american psychiatric association (apa) (1980).diagnostic criteria for autistic disorder. in: diagnostic and statistical manual of mental disorders. 3rd edition (dsm-iii). washington, dc: american psychiatric publishing. 3. american psychiatric association (apa) (1994). autistic disorder, pdd-nos, asperger’s disorder, rett’s disorder, and childhood disintegrative disorder. in: diagnostic and statistical manual of mental disorders.4th edition (dsm-iv). washington, dc: american psychiatric publishing. 4. american psychiatric association (apa) (2011).‘autism spectrum disorder.’ proposed revision apa dsm5. http://www.dsm5.org/proposedrevision/p ages/proposedrevision.aspx?rid=94#. 5. anagnostou, e, taylor, mj (2011). review of neuroimaging in autism spectrum disorders: what have we learned and where we go from here. molecular autism 2 1–9. 6. autism speaks – what is autism? (2012). http://www.autismspeaks.org/whatautism/ treatment. 7. berry-kravis, em, hessl, d, rathmell, b, zarevics, p, cherubini, m, et al. (2012). effects of stx209 (arbaclofen) on neurobehavioral function in children and adults with fragile x syndrome: a randomized, controlled, phase 2 trial. science translational medicine 4 152ra127. 8. curemark (2011).curemark llc reports positive phase iii results of cm-at in children with autism. http://www.curemark.com/news/2011/12/ 07/curemark-llc-reports-positive-phase-iiiresults-of-cm-at–in-children-with-autism/. 9. curemark (2012). a phase iii randomized double blind placebo controlled trial of cmat in children with autism. in: clinicaltrials.gov [internet]. bethesda, md: national library of medicine (us). 2009 – [cited 29 october 2012]. available from http://clinicaltrials.gov/show/nct00881452 . nlm identifier: nct00881452. 10. folstein, s, rutter, m (1977). infantile autism: a genetic study of 21 twin pairs. journal of child psychology and psychiatry 18 297–321. 11. fombonne, e (2009). epidemiology of pervasive developmental disorders. pediatric research 65 591–598. 12. gilman, s, iossifov, i, levy, d, ronemus, m, wigler, m, vitkup, d (2011). rare de novo variants associated with autism implicate a large functional network of genes involved in formation and function of synapses. columbia undergraduate science journal open-access publication | cusj.columbia.edu spring 2013 | volume 7 columbia undergraduate science journal 29 neuron 70 898-907. 13. hallmayer, j, cleveland, s, torres, a, phillips, j, cohen, b, et al. (2011). genetic heritability and shared environmental factors among twin pairs with autism. archives of general psychiatry 68 1095– 1102. 14. hardan, ay, fung, lk, libove, ra, obukhanych tv, nair, s, et al.(2012).a randomized controlled pilot trial of oral nacetylcysteine in children with autism. biological psychiatry 71 956–961. 15. henderson, c, wijetunge l, kinoshita, mn, shumway, m, hammond, rs, et al. (2012). reversal of disease-related pathologies in the fragile x mouse model by selective activation of gabab receptors with arbaclofen. science translational medicine 4 152ra128. 16. huerta m, bishop s, duncan a, et al. (2012). application of dsm-5 criteria for autism spectrum disorder to three samples of children with dsm-iv diagnoses of pervasive developmental disorders. american journal of psychiatry 169 105664. 17. lai, g, schneider, h, schwarzenberger, j, hirsch, j (2011). speech stimulation during functional mr imaging as a potential indicator of autism. radiology 260 521-530. 18. mendelsohn, nj, schaefer, gb(2008).genetic evaluation of autism. seminars in pediatric neurology 15 27–31. 19. mmwr morbidity and mortality weekly report (2008). prevalence of autism spectrum disorders – autism and developmental disabilities monitoring network, 14 sites, united states, 2008. morbidity and mortality weekly report 61 1–18. 20. novarino, g, el-fishawy, p, kayserili, h, meguid, na, scott, em, et al. (2012). mutations in bckd-kinase lead to potentially treatable form of autism with epilepsy. science 338 394–397. 21. peart, kn (2012). oxytocin improves brain function in children with autism. yale news. 22. sanders, sj, murtha, mt, gupta, ar, murdoch, jd, raubeson, mj, et al.(2012). de novo mutations revealed by whole-exome sequencing are strongly associated with autism. nature 485 237-241. 23. sebat, j, lakshmi, b, malhotra, d, troge, j, lese-martin, c, walsh, t. et al. (2007). strong association of de novo copy number mutations with autism. science 316 445-449. 24. sowell, er, bookheimer, sy (2012). promise for finding brain biomarkers among infants at high familial risk for developing autism spectrum disorders. american journal of psychiatry 169 551–553. 25. stoltenberg, c, schjølberg, s, bresnahan, m,hornig, m, hirtz, d,et al. (2010). the autism birth cohort: a paradigm for geneenvironment-timing research. molecular psychiatry 15 676–680. 26. szatmari, p (2011). is autism, at least in part, a disorder of fetal programming? archives of general psychiatry 68 1091– 1092. 27. tchaconas, a, adesman, a (2013).autism spectrum disorders: a pediatric overview and update. current opinion in pediatrics 25 130-144. 28. wolff, jj, gu, h, gerig, g, elison, jt, styner, m, et al. (2012). differences in white matter fiber tract development present from 6 to 24 months in infants with autism. american journal of psychiatry 169 589–600. dear readers, it is with great excitement that i am able to present to you the cusj volume 14 2020 edition. since our very first volume in 2006, it has been the mission of cusj to provide an accessible and professional, open access platform for undergraduates to communicate their research advances in various scientific fields. we cusj editors believe, as a community of young scientists and writers, that it is absolutely essential for researchers of all ages and levels of experience to have the opportunity to share their ideas and learn from their peers. creating a space to facilitate that discussion is exactly what cusj aims to achieve not only with each volume, but also with our outreach to the public and numerous research oriented events. covid-19 has posed an incredibly difficult and tragic challenge to communities all over the world, scientists and non-scientists alike. with the shutdowns of universities accompanied by sudden changes in employment status and housing, the perseverance and determination shown by both our authors and editing team has been a true testament to their character, work ethic, and resilience. the hard work evident in each article of this volume, from experimental design to data interpretation, is a display of the talent and strength of our authors, and their ability to push on through an unprecedented global emergency. these undergraduates have shown that even during the most trying of times, while physically isolated, we can all still come together and communicate through science. volume 14 of cusj is diverse in scientific fields represented, ranging from biochemistry and the development of enzymatic nanoreactors for industrial synthesis, to the social science of shame in teens. our cover features a graph from author aidan reddy’s cutting-edge investigation into x-ray absorption spectroscopy data processing, in which he develops a new method for spectra correction. his work should allow for more accurate analysis of various materials’ atomic properties, advancing the fields of both materials science and physics. it has been a pleasure to serve as the editor-in-chief of cusj for the 2019-2020 academic year. my sincere thank you to everyone involved in the creation of this volume for their passion and commitment. julia parsley editor-in-chief letter 3 4 letter from the president dear readers, on behalf of the editorial board of the columbia undergraduate science journal, i am proud to present the 15th edition of our undergraduate publication. over the past fifteen years we have published a wide-variety of work produced by undergraduate student-scientists. from our first edition, which featured innovative work on cell-cell communication in bone degeneration, to our present edition, which includes novel insights on the gut microbiome’s contribution to epithelial barrier permeability, it is evident not only that science has made leaps, but that our young scientists have as well. as the countable infinity of scientific knowledge left to learn grows ever smaller, i am amazed by the remarkable curiosity of undergraduate student-scientists, who tackle seemingly insurmountable questions with vigor and drive. the entire cusj enterprise has taken great strides over the past fifteen years. as one of the first undergraduate peer-reviewed scientific journals, we paved the way many years ago for the creation of scientific publications at other universities. now, on our 15th anniversary, cusj spearheaded the formation of and currently leads the national undergraduate consortium of science journals, uniting more than 20 undergraduate scientific publications in an effort to ensure high ethical standards. at our now-annual national undergraduate conference on scientific journalism, we gathered more than 500 students from 34 countries to visualize the way forward for young scientists. as we move into the next fifteen years of cusj, i look forward to the continued sharing of scientific pursuits and common values. i would like to thank all students who submitted to cusj, past and present. it is your commitment to thoughtful experimentation and thorough questioning that drives the young scientific world. i would like to specifically commend the authors published in our 15th edition, who have demonstrated an admirable commitment to scientific achievement and collaboration in this unusual time. i am grateful to you, your research mentors, your parents, and all who supported you for making cusj possible. it has been a great honor to serve as the president and chief editorial officer of the columbia undergraduate science journal for the 2020-2021 academic year. thank you to the editorial team, our selected peer reviewers, and our esteemed faculty advisory board for ensuring the growth and prosperity of the young scientific community. congratulations to all authors, and thank you to our readers! arya rao president, chief editorial officer columbia undergraduate science journal 55 the effects of apobec3g’s cytidine deaminase activity on retroviral evolution mary-benedicta obikili1 1department of microbiology and immunology, university of illinois at chicago college of medicine, chicago, il, usa abstract: apolipoprotein b editing complex (apobec3/a3) genes are found in mammalian cells. in primates, there are 7 apobec3 genes, namely, 3a, 3b, 3c, 3de, 3f, 3g, and 3h. previous research has shown that a3 proteins help to inhibit viral infection via their cytidine deaminase activity. however, it has also been found that a3 proteins could lead to viral evolution, where retroviruses such as hiv (human immunodeficiency virus), acquire beneficial mutations that enable them to overcome the antiviral activity of a3 proteins, gain resistance to certain drugs used for treating viral infections and escape recognition by the immune system. this paper is a review article summarizing the role of a3g on viral infection and evolution, and the potential impact viral evolution could have in treatment of retroviral infections such as hiv. keywords: apobec3g, cytidine deamination, retroviral evolution introduction apolipoprotein bediting complex 3 (apobec3/a3) genes encode enzymes that prevent viral infection through their cytidine deaminase activity [1]. there are seven apobec3 genes found in humans, namely: a3a, a3b, a3c, a3de, a3f, a3g, and a3h. most of these proteins play a role in inhibiting viral infection, however, a3g, a3f, a3d, and a3h are best known for their antiviral role [2, 3]. a3 proteins can be found in human and mouse dendritic and myeloid cells, and they are usually expressed at different levels in hematopoietic cell populations, which include b cells, cd4+ and cd8+ t cells [1]. when a host cell is infected, interferons, and other chemokines and cytokines are released and stimulate the expression of a3 proteins in other cells which help to fight off the infection [1]. it has been found that a3 proteins, specifically a3g, is the main antiviral protein in cd4+ t cells. therefore, when a cell is infected, a3g levels increase, hence inhibiting viral replication by introducing more g to a hypermutations in the virus [1]. these a3 proteins are packaged into the virion, which lead to viral inhibition through their cytidine deaminase activity. the cytidine deaminase activity of a3 proteins refers to the conversion of cytosine to uracil in single-stranded dna (fig. 1a). during the process of retrovirus infection, the virion enters the cell cytoplasm and uncoats its viral rna (fig. 1b). the viral enzyme reverse transcriptase then uses the rna as a template for making dna. however, during the process of reverse transcription, the a3 protein deaminates the cytidine residues of the (-) strand vi© 2021 obikili. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. columbia undergraduate science journal vol. 15, 2021 obikili. 56 ral dna to uracil residues, resulting in g-to-a mutations in the viral coding (+) strand dna. in other words, because cytidine is converted to uracil, which reverse transcriptase reads as thymidine, guanosine is replaced with adenine in the viral coding strand dna. after the synthesis of a double stranded dna molecule, the viral dna goes into the host cell’s nucleus and becomes integrated in the chromosomes (fig. 1b). the a3-induced mutations inhibit the synthesis of virus proteins and virions because they inactivate the virus, which prevents it from infecting and replicating in other cells. although apobec3 proteins help inhibit viral infection through their cytidine deaminase activity, a3d, a3f, a3g, and a3h may also play a role in viral evolution, where viruses such as hiv gain beneficial mutations that enable them to overcome the antiviral activity of a3 proteins [5, 6]. this paper will highlight the role of a3 proteins, specifically a3g, on retroviral evolution. retrovirus replication viruses are infectious pathogens that require a host cell for their growth and replication. alfigure 1: the process of cytidine deamination. (a) the enzyme apobec3, which is a cytidine deaminase, deaminates cytidine to uridine in the single-stranded retroviral dna before its integration into the host cell (source: national university of singapore, faculty of science, special programme in science and igem 2018, yuhui deborah fong (used with permission)) (b) the virus enters the cell. after uncoating of the rna genome, the viral enzyme, reverse transcriptase makes a double-stranded dna copy of the genome, which then enters the nucleus and integrates into the host chromosomes. during reverse transcription, a3 enzymes (red ball), which are packaged in the virion deaminates cytidine to uracil in the single-stranded viral dna before its replication and integration into the chromosomes, resulting in g-to-a changes in the viral genome. the red dna in the nucleus is the integrated viral dna. figure 2: retrovirus life cycle this shows the process of retroviral integration and replication in the host cell. (figure obtained and used with permission from openstax microbiology textbook [7]) (access at https://openstax.org/books/microbiology/pages/1-introduction) a b columbia undergraduate science journal vol. 15, 2021 obikili 57 though they do not have the ability to grow on their own, once inside a cell they use its machinery to grow and multiply and have the potential to cause diseases in the host organism. retroviruses are viruses that use their viral rna-dependent enzyme, reverse transcriptase, to synthesize viral dna in a host cell which is then inserted into the chromosomal dna (fig. 1b). hiv is a lentivirus that infects cd4+ t cells, macrophages, and dendritic cells in humans [1]. the process of retrovirus infection is shown in fig. 2. once a retrovirus encounters a host cell, it binds to a cell surface receptor, hence enabling its fusion with the cell. after it fuses with the cell, it uncoats its rna genome and enzymes, such as reverse transcriptase (rt), which then synthesize a new copy of the viral dna from the rna template. then the viral dna enters the nucleus and the viral enzyme integrase, along with several host proteins, splices the viral dna into the cell’s genome. cellular rna polymerase is then used to transcribe the viral dna into mrna which is then translated into viral protein. the new virions made are assembled at the host membrane and bud off from the cell membrane. the final stage consists of the retrovirus protease enzyme cleaving the precursor proteins into mature viral proteins leading to the maturation of the virus and the start of a new cycle [8]. retroviral evolution the cytidine deaminase activity of a3 proteins may also lead to retroviral evolution, where the virus gains beneficial mutations that help the virus to counteract a3 antiviral activity [5, 6]. a3g has been shown to preferentially deaminate cc motifs, resulting in ag motifs in the coding strand, whereas the other a3 proteins have been shown to preferentially deaminate tc motifs, resulting in aa motifs [5, 6]. however, hiv-1 encodes a protein the viral infectivity factor (vif), which leads to the degradation of a3 proteins in virus-producing cells [5]. vif has two motifs that bind to a3 proteins: drmr which binds to and degrades a3f and a3d, and the other motif, yrhhy, which binds to a3g, leading to its degradation [5]. fig. 3 shows the structure of the vif complex [5]. when vif binds to a3, it prevents it from entering the budding virion by targeting it for degradation. it does this by binding to cbfb co-factor and cullin 5 e3 – ubiquitin ligase (elob-eloc-cul5), via the sites diagrammed in fig. 3b, which then targets the a3 protein for proteasomal degradation [5]. figure 3: the structure of vif (a) shows the two vif motifs that bind to a3 proteins. drmr (red) binds to a3d and a3f, whereas yrhhy (blue) binds to a3g (b) shows the vif complex. the blue and red correspond to the vif binding motifs in a where vif binds to a3d, a3f and a3g, respectively. the yellow, cyan and pink colors represent the vif protein, cbfb co-factor and cul5 ligase respectively (fig. obtained from sato k. et al. 2014, with permission to reprint from plos pathogens). viral evolution can occur when there are defective (mutant) vif proteins. the defective proteins do not cause complete a3 degradation and therefore allow some a3 proteins to enter the virion, resulting in mutations in the viral genome. since a3g prefers to deaminate tggg – tagg motifs, it often converts the codon encoding for the amino acid tryptophan (tgg) into a stop codon (tag), which renders the virus defective and unable to synthesize its proteins. this prevents it from `assembling at the host cell’s membrane and infecting other cells. sato et al. found that a3 proteins lead to mutations in the env region of the hiv-1 genome, which codes for the viral glycoprotein [5]. this protein is important for enabling the virus to bind to cellular receptors and fuse with the host cell’s membrane. mutations caused by columbia undergraduate science journal vol. 15, 2021 obikili. 58 a3 proteins inhibit the virus from synthesizing its glycoprotein, which in turn prevents it from infecting the cell. these defective viruses may later get degraded. occasionally, the resulting g-to-a mutations may also be sub-lethal (depending on the levels of a3g present in the cell, which may be due to genetic factors, such as different polymorphisms in a3g genes in humans [9], and the context of the tg motif), resulting in missense mutations. in other words, lower levels of a3g expression, would result in more sub-lethal mutations, compared to higher levels of a3g expression, which would mostly result in lethal mutations. moreover, other a3 proteins such as a3d and a3f, respectively prefer deaminating the ga – aa and gaa – aaa motifs [5], resulting in sub-lethal missense mutations. these missense mutations may result in viral evolution and diversification. for example, the sato study showed that a3 mutations in env also changed the co-receptor specificity of the glycoprotein, enabling it to preferentially infect different cell types. it has been found that some a3 proteins may indirectly contribute to retroviral evolution, where the cytidine deaminase activity of the a3 protein helps the virus to acquire mutations that will make it more fit. studies by kim et al., sato et al., and fourati et al. showed that retroviruses, such as hiv, can sometimes gain beneficial mutations that enable them to replicate better [3, 5, 6]. sato et al., fourati et al., and borzooee et al., also demonstrated that a3 proteins such as a3d, a3f, a3g and a3h can sometimes lead to retroviral evolution [3, 5, 10]. it was also shown that a3 proteins can mutate viruses so that they can use different receptors on the cell surface for entry into the host cell [5]. moreover, in the sato study [5], where humanized mice models were used to study different vif mutants that were initially mutated via a3-independent means, it was found that a3d and a3f highly suppressed one mutant, a3g highly suppressed another mutant, and the other was a double mutant, that was suppressed by a3d, a3f, and a3g [5]. it was demonstrated that although a wild type vif protein can target a3 for proteasomal degradation, a defective vif protein, could be suppressed by a3 proteins. however, sometimes these a3-dependent mutations may benefit the virus. in the study, they found that a3f and a3d led to more viral diversification compared to a3g (for reasons previously described). these studies suggest that a3 proteins could benefit the virus by leading to mutations that will help the virus to overcome a3’s antiviral activity, and evade antiretroviral drugs detection, leading to viral escape. this will enable them to continue to undergo replication and infection in additional cells. in other words, the a3 proteins assist the virus by enabling them to gain diversity, which in turn may help the virus to acquire mutations that make it gain drug resistance. drug resistance drug resistance refers to the ability of a pathogen to overcome the effects of a drug. this usually occurs when a certain drug is continuously used to treat a particular infection, and due to constant exposure of the disease pathogen to the drug, the drug becomes less effective, usually because the pathogen acquires drug-resistance gene mutations. to minimize the possibility of disease pathogens gaining resistance, multiple drugs could be used for the treatment of that disease. this is usually seen in the treatment of certain diseases, such as hiv, where a combination of different drugs (such as highly active antiretroviral therapy, haart) are used simultaneously to prevent the emergence of drug-resistant virus. however, g-to-a mutations caused by a3 proteins in the retrovirus may cause the retrovirus to gain mutations that enable it to become drug-resistant. as shown in the study by fourati et al., peripheral blood mononuclear cells (pbmcs) and other body fluids/tissues obtained from 30 haart-treated patients were examined for possible drug-resistance mutations. it was found that a3-dependent hypermutations in addition to viral recombination could lead to viral evolution [6]. columbia undergraduate science journal vol. 15, 2021 obikili 59 additionally, in a study by mulder et al., it was found that g-to-a hypermutations caused by a3g reduces the viral infectivity and could also lead to viral diversification [11]. in this study, it was demonstrated that defective vif mutants with suboptimal anti-a3g activity could lead to the production of proviruses resistant to the drug lamivudine (3tc) (which inhibits the retroviral reverse transcriptase) before drug exposure. although the proviruses were not able to replicate, when they recombined with wild-type hiv-1, they produced competent viruses that were resistant to the drug 3tc [11]. in a similar study by hernandez et al., it was found that sublethal mutations induced by a3g could lead to viral diversification. they infected humanized mice with different variants of hiv (that were mutated via a3-independent means), and they found that the wild type hiv strain had a higher fitness compared to the mutant strains prior to its treatment with 3tc [12]. however, when they started the treatment with the drug, they found that the mutant strain was less susceptible to the antiretroviral drug’s effect because of its defectiveness. in other words, because the mutant hiv strain is less fit, the drug has a decreased effect on it compared to the wild type, hence the mutant strains become resistant to the drug [12]. thus, these studies show that sublethal mutations caused by a3g, increase viral diversification which could make the virus gain resistance to certain drugs. these mutations could also enable the virus to undergo immune escape. immune escape immune escape refers to a situation where a pathogen evades detection by the host’s immune system. in the case of retroviruses, this immune escape can be aided by a3 proteins [1, 10, 13]. as mentioned earlier, a3 proteins may sometimes cause sublethal mutations, which could help the virus to avoid detection by cytotoxic t cells (ctl), which kill infected cells [1, 10, 13]. these sublethal mutations may alter ctl epitopes or lead to mutations that alter peptide degradation and its presentation by the human leukocyte antigen (hla) proteins on the cell surface [10, 13]. hla proteins are human major histocompatibility complex (mhc) proteins that present peptides to ctls, thereby allowing recognition of infected cells [10, 13]. in a study by grant and larijani, it was found that a3g/a3f could cause mutations in hiv peptides that prevented the virus from being detected by ctls, hence enabling it to escape the host cell’s immune response [13]. similarly, a study by borzooee et al. showed that a3g-dependent mutations could lead to decreased binding affinity of hla to the peptide epitopes, which also enables viral immune escape [10]. in this study, it was observed that the sequences that encode ctl epitopes in hiv are rich in a3g-preferred deamination motifs, which preferentially mutate the c in ccc, tcc, and acc [10]. this could result in missense mutations that alter ctl epitopes. thus, these mutations could help the virus evade detection by the host cell's immune response, which enables it to continue to spread in an individual. these studies show that although a3g could result in lethal mutations in the virus, sublethal mutations could cause viral immune escape, where ctls are no longer able to recognize infected cells to target them for killing. this could lead to natural selection and viral evolution, where the viruses are able to acquire beneficial mutations that enable them to become more fit and harmful/difficult to treat. conclusion in conclusion, many studies have shown the effects of the cytidine deaminase activity of the a3 genes on retroviral evolution. apobec3 proteins are known to inhibit viral infection through their cytidine deaminase activity [1, 2, 3, 14]. however, lower expression of a3 proteins in cells could also lead to viral evolution [3, 5, 6]. when a3 proteins such as a3g and a3f are expressed at lower levels, they are likely to introduce missense mutations in the virus, which instead of inhibiting the virus, helps it obcolumbia undergraduate science journal vol. 15, 2021 obikili. 60 tain beneficial mutations that allow it to become more fit and able to evade the cells’ immune response [10, 13] and develop drug resistance [6, 11, 12]. also, many studies show that hiv can evolve to overcome a3’s cytidine deaminase activity by encoding vif's variants that more effectively target a3 for proteasomal degradation [1, 3, 5, 14]. moreover, fourati et al. and mulder et al., both showed that in addition to a3-dependent mutations, viruses could also evolve via recombination, where defective viruses after recombination gain drug resistance [6, 11]. these studies collectively suggest that a3 proteins play an important role in the host cells’ immune response and help the cells to combat viral infection. although, these studies highlight the importance of a3 proteins, they also show the limitations of its use in preventing viral infection, since hiv-1 vif protein inhibits a3 proteins’ antiviral activity. future studies should focus on developing drugs that would inhibit hiv-1 vif’s ability to bind and degrade a3 proteins, or enable a3 proteins, specifically a3g, to overcome vif activity, so that they could better inhibit retroviral infection and evolution. author information corresponding author *mary-benedicta obikili mobiki2@uic.edu funding sources this research was supported by the nih/niaid grant r01ai085015 and the l@s ganas project (hsi stem grant from dept. of education p031c160237). competing interests the author declares no competing financial and non-financial interests. acknowledgments i would like to thank l@s ganas research fellowship and the chancellor’s undergraduate research award (cura) for funding my research. i would also like to thank the entire ross lab, for their support during my research. i especially want to thank dr. karen salas briceno, for all her mentorship and guidance during my research. finally, i am especially grateful to dr. susan r. ross, who provided me with immense knowledge and guidance during my time at her lab. i am truly grateful for the opportunity of doing research at the ross lab. references [1] stavrou s, ross sr (2015) apobec3 proteins in viral immunity. the journal of immunology 195(10):4565–4570. [2] armitage ae, et al. (2014) possible footprints of apobec3f and/or other apobec3 deaminases, but not apobec3g, on hiv-1 from patients with acute/early and chronic infections. journal of virology 88(21):12882– 12894. [3] kim e-y, et al. (2014) human apobec3 induced mutation of human immunodeficiency virus type-1 contributes to adaptation and evolution in natural infection. plos pathogens 10(7). doi:10.1371/journal.ppat.1004281. [4] sousa mm, krokan he, slupphaug g (2007) dna-uracil and human pathology. molecular aspects of medicine 28(3-4):276–306. [5] sato k, et al. (2014) apobec3d and apobec3f potently promote hiv-1 diversification and evolution in humanized mouse model. plos pathogens 10(10). doi:10.1371/journal. ppat.1004453. [6] fourati s, et al. (2014) differential impact of apobec3-driven mutagenesis on hiv evolution in diverse anatomical compartments. aids 28(4):487–491. [7] parker n, schneegurt m, tu a-ht, forster bm, lister p (2016) microbiology (openstax, houston, tx). columbia undergraduate science journal vol. 15, 2021 obikili 61 [8] herron jc, freeman s (2014) evolutionary analysis (pearson). 5th ed. [9] an p, et al. (2004) apobec3g genetic variants and their influence on the progression to aids. journal of virology 78: 11070–11076. [10] borzooee f, joris kd, grant md, larijani m (2019) apobec3g regulation of the evolutionary race between adaptive immunity and viral immune escape is deeply imprinted in the hiv genome. frontiers in immunology 9. doi:10.3389/fimmu.2018.03032. [11] mulder lcf, harari a, simon v (2008) cytidine deamination induced hiv-1 drug resistance. proceedings of the national academy of sciences 105(14):5501–5506. [12] hernandez mm, et al. (2019) impact of suboptimal apobec3g neutralization on the emergence of hiv drug resistance in humanized mice. journal of virology 94(5). doi:10.1128/jvi.01543-19. [13] grant m, larijani m (2017) evasion of adaptive immunity by hiv through the action of host apobec3g/f enzymes. aids research and therapy 14(1). doi:10.1186/s12981-0170173-8. [14] cadena c, et al. (2016) the effect of hiv1 vif polymorphisms on a3g anti-viral activity in an in vivo mouse model. retrovirology 13(1). doi:10.1186/s12977-016-0280-y. misdiagnosis of community-acquired pneumonia is an important clinical problem, leading to a high rate of mortality. diagnoses are typically conducted using two-dimensional chest x-rays, which have shown to be time-consuming and inaccurate. in an effort to improve the current diagnostic method, we utilized micro-computed tomography (microct) and image analysis software to develop a diagnostic algorithm that can quantitatively assess the severity of pneumonia in mice. we believed this method would provide more immediate, precise, and accurate diagnoses as opposed to the qualitative assessments done by radiologists at present, because microct provides opportunities for non-invasive radiographic endpoints for pneumonia studies. a quantitative scoring of previously obtained computed tomography (ct) scans of pneumonia infected and control mice lungs was developed with a semi-automated image segmentation algorithm. at the endpoint of 168 hours, each of the mice was categorized as either a) a saline (control)-injected mouse (total=13), a pneumonia-injected survivor (total=11), or a pneumonia-injected non-survivor (total=11). three comparison tests were then completed, including saline vs. all pneumonia injected mice, pneumonia survivors vs. pneumonia non-survivors, and all survivors (both saline & pneumonia) vs. pneumonia non-survivors. in all three comparisons, the semi-automated algorithm was better able to distinguish between the different groups than radiologists using two-dimensional chest x-rays of the mice’s lungs, with p-values of 0.001, 0.039, and 0.001 for the semi-automated algorithm, and 0.004, 0.581, 0.058 for the radiologists, respectively. key words: community-acquired pneumonia, computed tomography columbia undergraduate science journal open-access publication | http://cusj.columbia.com 4 cusjvolume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu b io m ed ic a l en g in ee r in g cusjcolumbia undergraduate science journalresearch articles 5cusj volume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu b io m ed ic a l en g in eer in g in the united states, pneumonia is the sixth leading cause of death and the number one infectious disease killer (m. s. niederman, 1998). !e disease is an in"ammation of one or both lungs caused by an infection from bacteria, viruses or fungi. !e infection causes the alveoli of the lungs to become in"amed and #lled with "uid, which leads to symptoms such as cough, fever and respiratory breathing di$culties (jelic, 2005). often times, pneumonia occurs as a secondary infection when the immune system of a person is already weakened due to prior infection, such as an upper respiratory tract infection. !is primary infection causes in"ammation in the inner lining of airways that leaves the patient susceptible to the secondary infections such as pneumonia (boone, 2004). pneumonia can be classi#ed according to the population a%ected. hospital-acquired pneumonia is acquired when a patient breathes germs during a hospital stay for another illness. people are most prone to hospital-acquired pneumonia while on a mechanical ventilator, since potentially pneumonia-causing bacteria and viruses may be blown directly into the lungs. !e most common type of pneumonia is community-acquired. community-acquired pneumonia is acquired outside of hospitals and other health care settings, with about 5.6 million people getting infected every year in the usa and 1.1 million requiring hospitalization (m. s. niederman, 1998). community-acquired pneumonia is an important clinical problem, with high rates of misdiagnosis and mortality. current methods to diagnose pneumonia rely on two dimensional (2d) chest x-rays, which are known to have low sensitivity early in the course of pneumonia (mohd). radiologists typically score six lung zones (upper, middle, and lower, on the right and left sides) for each patient on a scale of 0 to 4, such that zero is normal, and the maximum possible abnormal score is 24 for the combined zones; 0 represents 0% pneumonia involvement, 1 represents up to 25% involvement, 2 represents up to 50%, 3 represents up to 75%, and 4 represents up to 100% (armbrust, 2005). !ese chest x-rays may take days to diagnose the severity of pneumonia, in which time immunocompromised patients, such as patients with hiv/aids, cancer, diabetes, or sickle cell anemia, may reach a severity beyond curing (smergal, 2008; stuart, 2008). for example, immuct based semi-automated method for pneumonia severity in mice ansh johri1*, lewis hsu 2 1deparment of biomedical engineering, columbia university, new york, ny 10027; 2national institutes of health, national heart, lung, blood institute, washington, dc 20010. copyright: © 2011 the trustees of columbia university, columbia university libraries, some rights reserved, johri, et al. received dec. 31, 2009. accepted feb. 8, 2010. published april 1, 2011. *to whom correspondence should be addressed: center for cancer and blood disorders, west 4-600. children’s national medical center111 michigan ave., n.w. washington, dc 20010 aj2381@columbia.edu abstract introduction nocompromised patients with pneumonia have a mortality rate of 12% (mohd). furthermore, radiologists are often inconsistent with their diagnoses; two radiologists may judge the severity of pneumonia in patients very di%erently, leading to possible misdiagnosis (l hsu, 2007). !us, imaging techniques to evaluate pneumonia earlier and with more accuracy would be important diagnostic tools for clinicians. imaging information could also be used to guide decisions on the clinical care needed, such as whether to hospitalize or to treat the patient at home, thus improving pneumonia diagnosis. in order to address these limitations of inaccuracy, inconsistency, and delayed diagnosis, a di%erent diagnostic method is required. computed tomography (ct) scans use x-rays that pass through the specimen and are received by sensors on the other end. denser portions of the specimen result in a reduced amount of radiation received by the other end, since the specimen hinders the radiation. !is disparity in densities, or attenuation, can be reconstructed to produce a 3d image with di%erent grayscale values (figure 1). houns#eld units (hu) are grayscale values that correspond to the density of each voxel. in the houns#eld scale, -1000 represents air, 0 represents water, and 1000 represents bone density. notably, "uid or pus in the alveolar sacs would be approximately 0 hu, normal lung alveoli have a mixture of air and tissue reading near -500 hu, and voxels in lung with a mixture of air and "uid would be between -500 and 0 hu. ct scans, which can visualize the entire lung as opposed to the 2d projection scans in a chest radiograph, might have the sensitivity to assess the severity of pneumonia as early as 24 hours after onset. !is earlier timeframe for treatment would allow immunocompromised patients to receive immediate treatment, thus decreasing their mortality rate. since ct scans provide a more detailed depiction of the lung, they are potentially more accurate than the current chest x-ray method. finally, by developing a semi-automated method that uses ct scans to diagnose the severity of pneumonia, more precise diagnoses can be conducted, since the procedure is more automated and less prone to human error (muller, 2006). !e purpose of this particular research was to develop an algorithm to measure the severity of pneumonia in mice through micro-computer tomography (microct) scan analysis and test its e%ectiveness through comparison with radiologists’ diagnosis. microct works in the same way as a regular ct scanner, but is typically used to image smaller specimens, such as rodents, as opposed to human beings. !ere were three goals for the image analysis algorithm. !e #rst was to achieve high reproducibility in repeated analysis of the same microct scan. current methods typically involve having two radiologists independently score the chest x-rays; the #nal score is then the average of the independent scores. !e second goal was to achieve higher accuracy using image segmentation algorithm to quantify the amount of pneumonia in the lungs. !is is di%erent from current methods which require radiologists to qualitatively assess multiple images of pneumonia. !e quanti#cation would be done by loading the ct scans in an imaging software, and determining the voxel distribution in order to compare densities. finally, the project aimed to increase e$ciency in diagnosis. a semi-automated computer algorithm would allow more measurements to be taken in a smaller amount of time than with current methods, without special expertise in radiology. a computer would automatically calculate the severity of the pneumonia, which, under current circumstances, would be done by a radiologist. in order to use the computer algorithm method, it would be necessary for the radiologist to have some basic skills, however. !e #rst is the ability to use amira, the software used in this paper. !e second is knowledge of basic lung anatomy, such as the location of the trachea, stomach bubble, and mediastinum. finally, the radiologist would need the ability to use a quantitative diagnosis performed by the computer to give the correct treatment to the patient. we hypothesized that in vivo microct scans of mice with early bacterial pneumonia could be scored quantitatively by semi-automated imaging methods, with good reproducibility and correlation with the bacterial dose inoculated, pneumonia survival outcome, and radiologists’ scores previously obtained. materials and methods !e project used microct scans to evaluate a murine model of bacterial pneumonia through image analysis by semi-automated segmentation and comparison of results to figure 1 �+�9ljvuz[y\j[lk�3\un��<utvkpälk�����(u�pthnl�vm�[ol�s\un�hm[ly�yljvuz[y\jtion, containing the trachea, mediastinum, and the stomach bubble, which are subsequently removed since they cannot be infected by pneumonia. columbia undergraduate science journal open-access publication | http://cusj.columbia.com 6 cusjvolume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu b io m ed ic a l en g in ee r in g cusjcolumbia undergraduate science journalresearch articles 7cusj volume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu radiologists’ interpretation. !e following steps were used in the research to prepare mice with pneumonia, and then determine and validate the severity of their pneumonia: 1. inoculation of mice with di"erent severities of pneumonia bacteria and acquisition of ct scans (l hsu, 2007) 2. radiologists’ diagnosis of pneumonia (l hsu, 2007) 3. pneumonia diagnosis using semi-automat ed diagnostic algorithm !e #rst two steps were done prior to this project, whereas the third step was conducted speci#cally in this research, and performed by one individual, with three trials for each mouse. mercury computer systems’ amira 4.1 software was used to perform image analysis of the ct scans. amira can perform image segmentation, 3d visualization, and other image processing (schimel, 2007). since the severity of pneumonia can be assessed by the volume distribution of materials within the lung due to in$ammation, it was expected that more lung volume distribution in the range of high attenuation for greater severities of pneumonia would be observed (l hsu, 2007). after loading the stack of typically near 400 axial slices for 3d reconstruction in amira, the lung was reconstructed by selecting houns#eld units -510 to 0 (muller, 2006). non-lung anatomical structures overlapping or in the vicinity of the lungs in the mice ct scans were excluded from the lung reconstruction because these structures are not involved with pneumonia infection (figures 2a, 2b, and 3) (iwaki, 2001). !e non-lung components that overlapped with the lungs were primarily the stomach bubble, trachea, and mediastinum (armbrust, 2005). !e algorithm excluded each of these components systematically using the image manipulation features of amira. all subsequent non-3d images were axial crosssections of the upper mouse body. !e stomach bubble is a pocket of air that lies right below the lung. !e contents in this bubble fall in the houns#eld unit range of -510 to 0 and also border the lung. !erefore, when a region in the lung within this range was clicked, the stomach bubble was included in the selection as well. !e stomach bubble was manually removed, as shown in figures 2a and 2b. like the stomach bubble, the trachea is attached to the lung and gets selected in the -510 to 0 range. to remove it, the trachea was cut o" at the carina, the point at which the two bronchi join to become the trachea. !e same procedure used in the stomach bubble removal was then used for trachea removal (figure 3). !e mediastinum is the anatomic region comprising a group of structures between the lungs that includes blood vessels and the esophagus. it was at times hard to distinguish between lung and mediastinum, making separation of mediastinum di%cult in many scans. however, since the mediastinum accounts for such a small percentage of the lung volume, its e"ect on the voxel quanti#cation was negligible, essentially eliminating human error. !e procedures for removing of stomach bubble and trachea were used to remove the mediastinum as well. at this point, the trachea, stomach bubble, and mediastinum had been removed from the lungs. !erefore, the image reconstructed contained only those voxels that could be a"ected by pneumonia. !is image was now ready for segmentation (figure 4). !e lung was segmented into eight regions using the following steps. !e density ranges of each of the eight materials, named well-aerated lung, b, c, d, e, f, g, and h were created, with houns#eld values of -510 to -350, -350 to -300, -300 to -250, -250 to -200, -200 to -150, -150 to -100, -100 to -50, and -50 to 0, respectively. !ese materials were entered into the “label voxel” tool, generating these 8 distinct density ranges within the entire ct scan. figure 5 illustrates this procedure: !e materials “wellaerated lung,” “b,” and “c,” encapsulate all voxels inside the entire scan within the density ranges of -510 to -350, -350 to -300 and -300 to -250, respectively (figure 5). !e tool can only analyze three materials at a time, but this had a negligible e"ect on the time it took to do the quanti#cation analysis because the analysis is almost instantaneous.. we were only interested in the volume of each material within the lung, not the entire scan, which includes bone, muscle, fat, fur, and other tissues. it should be noted that the units of the volume measurements are irrelevant, since we eventually determined the percentage distribution of each material within the lung. using amira, we were able to measure how much of each of the eight materials was present in the lung volume. in table 1, the volume of each of the #rst three materials can be seen: outside the lung (exterior), inside the lung (lung), and in the whole scan (total) (table 1). essentially, the semi-automated method was broken down into the three following steps. !e #rst was the isolation of the lung from the rest of the ct scan. !e second was the removal of extraneous anatomical features, such as the stomach bubble, trachea, and mediastinum, which cannot get a"ected by pneumonia and would therefore skew our calculations. !e #nal step involved #nding the percentage of each of the lung materials (well-aerated, b, c…). !e voxel distribution was analyzed for each of the survival groups. in the animal facility, 24 mice inoculated with bacteria developed symptoms of pneumonia, 11 of which died by the endpoint of 168 hours. !e 11 mice inoculated with saline had few or no signs of pneumonia, and none died. no mice died during the actual microct scan. after monitoring the mice for 7 days, each of these 35 mice was assigned to one of three experimental groups based on survival outcome and inoculation: (1) pneumonia survivor (total = 13), (2) pneumonia nonsurvivor (total = 11), and (3) saline-inoculated control (total = 11). in the development of a semi-automated method, three trials were conducted for each of the 35 mice, and the coe%cient of variation (cv) was calculated to evaluate reproducibility. !e semi-automated segmentation was reproducible, with the trials for each microct scan resulting in the same segmentation volumes within a coe%cient of variation of 2%. although all three trials were done by the same researcher, it is doubtful that there was much bias. !e procedure to do the quanti#cation is very structured and standard. if another researcher performed the experiment, assuming that the researcher followed the exact same procedure, there should be a negligible amount of variation, since all the quanti#cations are automatically done by the computer, and are therefore human-independent. grouping the microct segmentation results showed the expected #ndings for the three experimental groups (table 2). in the lower ranges of attenuation, saline-inoculated mice had the greatest percentage of lung results figure 2a axial lung slice (manual exclusion of stomach bubble) a projection view of amira being used to eliminate the stomach bubble from the lung. figure 2b 3d reconstructed lung (manual exclusion of stomach bubble) view of amira selecting stomach bubble, which will then be removed from the lung. figure 3 3d reconstructed lung (manual exclusion of trachea) a 3d view of amira selecting the trachea, which will then be removed from the lung. figure 4 3d reconstructed lung (manual exclusion of trachea) a 3d view of amira selecting the trachea, which will then be removed from the lung. b io m ed ic a l en g in eer in g columbia undergraduate science journal open-access publication | http://cusj.columbia.com 8 cusjvolume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu b io m ed ic a l en g in ee r in g cusjcolumbia undergraduate science journalresearch articles 9cusj volume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu b io m ed ic a l en g in eer in g volume, followed by pneumonia survivors, then pneumonia non-survivors. for example, the material with the lowest attenuation, “well-aerated lung” (houns!eld units -510 to -350), on average, made up 18.3% of lung volume in the saline-inoculated group, 10.9% of pneumonia survivors, and 8.2% of pneumonia non-survivors. on the other side of the density spectrum, the mice that eventually died of pneumonia had the highest percentages of lung volume in the higher ranges of attenuation, followed by pneumonia survivors, and !nally the salineinoculated group. table 2 presents the p-value distribution from a t-test¬ for each of the eight density ranges, for the following three experimental group comparisons: (1) salineinoculated vs. all pneumonia infected, (2) pneumonia survivors vs. pneumonia non-survivors, and (3) all survivors vs. pneumonia non-survivors. comparisons (1) and (3) were signi!cantly di"erent (pvalue < 0.05) for all the materials except for material d (table 2, highlighted yellow). #is may indicate the density range of material d (-250 to -200) includes regions of the lung that are not a"ected by pneumonia, since the three experimental groups did not show signi!cant di"erence in their average percentages for this material. excluding material d, the pneumonia survivors vs. pneumonia non-survivors comparison, however, showed no statistical signi!cance except in material h (table 3, highlighted orange). material h, as can be seen by table 4, showed far greater signi!cance by the semi-automated method than by the radiologists’ scores, thus indicating that the semiautomated method was indeed more reproducible than the radiologists’ method (table 4). all the materials showed more signi!cance than the radiologists’ interpretations; however, only h had a p-value under 0.05 for all three group comparisons. a qualitative assessment of the scans was also made. figures 6-8 show frontal pictures of three mice, each from the saline, pneumonia survivors, and pneumonia nonsurvivors groups, respectively (figures 6, 7, 8). #e red color represents voxels in the highest third density range of the lung, the orange color represents voxels in the middle third density range, and the blue color represents the voxels in the lower third of the density range. as it can be seen, the saline mouse had the greatest amount of orange and least amount of red, the pneumonia survivor group had less orange and more red, and !nally, the pneumonia non-survivor group had the least orange and most red. preliminary statistical analysis of the semi-automated segmentation of microct detected differences between the groups of pneumonia survivors versus pneumonia deaths. there was good reproducibility of the semi-automated segmentation, with less than 2% variability with repeated application of the methods. radiologists’ average scores provided statistically significant differences between the mice inoculated with pneumonia vs. the mice inoculated with saline, but not between the groups of pneumonia survivors vs. pneumonia non-survivors, nor between all surviving mice vs. pneumonia deaths. this comparison suggests that the semi-automated segmentation may provide a better method for quantitative scoring of pneumonia severity by ct scans, compared to scoring by radiologists. only a few other studies have applied quantitative image analysis to ct scans of pneumonia in animal models (m amigoni, 2008). for example, one examined another type of bacterial pneumonia and used different methods to score the pneumonia. another examined acid-aspiration pneumonia to score the lung injury. their studies did not examine severity in terms of survival, and did not compare the image analysis to scoring by radiologists. previous applications of quantitative image analysis of lung ct scans have focused on different clinical problems, rarely pneumonia: (1) emphysema, in which lungs have abnormal pockets of low attenuation, (2) lung tumors, which have much higher attenuation than normal lungs, and (3) normal and abnormal physiologic distribution of aeration and blood flow, often using tracer materials to detect flow (ritman, 2008). tracers to detect blood flow include xenon, intravenous iodinated contrast, and microspheres, but since such tracers were not used in this project, the problem of distinguishing normal lung vs. high-attenuation pneumonia involvement was more difficult. clinical use of ct imaging is widespread throughout the world because of the wealth of information that ct scans provide about abnormal fluid, tumors, aeration, etc. clinical high resolution ct (hrct) is now starting to be used to diagnose pneumonia in patients, but is rarely used for community-acquired pneumonia, which affects more people than any other type of pneumonia (jelic, 2005). although the spatial resolution of microct is better than clinical hrct on an absolute scale (60 microns vs. 500-1000 microns), the spatial detail is better with hrct when considering the anatomic size of mouse lungs vs. human lungs (apex to diaphragm ~3cm vs. ~50cm, alveoli 80 microns vs. 210 microns, respectively). thus, it can be concluded that transferring the technique figure 5 materials “well-aerated lung,” “b,” and “c” the beginning of the segmentation process. the three densities with lowest attenuation well-aerated, b, and c are shown in an axial view of the lung. note that these materials spl�v\[zpkl�vm�[ol�s\un��huk�t\z[�il�kpzylnhyklk�myvt�[ol�x\hu[päjh[pvu� volume of interest well-aerated lung b c exterior 720.2 174.7 165.4 lung 60.20 61.95 75.21 total 780.4 236.7 240.6 table 1 volume of interest calculated by amira software the percentages that each of the three groups had for the 8 density ranges. the pneumonia non-survivors had the greatest percent distribution in the higher lung density ranges, indicating a pneumonia-affected lung. the saline mice, however, had a greater percentage towards the lower density ranges, the opposite of the pneumonia non-survivors. table 2 average lung distribution for each group volumes of 3 materials, out of the total 8: a) outside the lungs (exterior), b) inside the lungs (lung), and c) in the entire scan (total). comparison wellaerate d lung (-510: -350) b b ( (350: -300) c c ( (300: -250) d d ( (250: -200) e e ( (200: -150) f f ( (150: -100) g g ( (100: -50) h h * ( (-50: 0) p-value from t-test of radiologists scores 1) saline vs. all pneumonia infected 0.018 0.00 0 0.00 0 0.06 5 0.00 0 0.00 0 0.00 0 0.00 1 0.004 2) pneumonia survivors vs. pneumonia nonsurvivors 0.375 0.29 8 0.31 7 0.94 6 0.67 2 0.62 2 0.29 6 0.03 9 0.581 3) all survivors vs. pneumonia nonsurvivors 0.032 0.00 7 0.01 0 0.44 8 0.04 3 0.03 1 0.00 9 0.00 1 0.058 table 3 7�]hs\l�+pz[ypi\[pvu�mvy�lhjo�4h[lyphs��hz�jhsj\sh[lk�i`�[�[lz[�*vtwhypzvu���:[h[pz[pjhss`�zpnupäjhu[�hipsp[`�mvy�th[lyphs�/�[v�kpz[pun\pzo�il[^llu�:hspul�]z��(ss�7ul\monia, pneumonia survivors vs. pneumonia non-survivors, and all suvivors vs. pneumonia non-suvivors. radiologists were not as successful in distinguishing [olzl�nyv\wz��4h[lyphs�+�^hz�\zlslzz�pu�kpz[pun\pzopun�hu`�vm�[ol�nyv\wz�hwhy[���4h[lyphs�/�^hz�zpnupäjhu[s`�kpmmlylu[�mvy�[ol�[oyll�nyv\wz��w�#��� discussion columbia undergraduate science journal open-access publication | http://cusj.columbia.com 10 cusjvolume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu b io m ed ic a l en g in ee r in g cusjcolumbia undergraduate science journalresearch articles 11cusj volume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu b io m ed ic a l en g in eer in g from mouse to humans should make the algorithm more accurate due to increased relative resolution. the current hrct technique, though an improvement to the more common 2d chest x-ray method, may still provide insufficient detail for quantitative analysis of a 3-dimensional reconstruction because it generally captures only a few 2d slices (uchiyama, 2003). newer techniques of human spiral ct will capture enough information for the 3d reconstruction as used in our algorithm. the principal limitation of this method was the accuracy of manually isolating the lung “volume of interest” from the rest of the mouse. it was sometimes difficult to separate the lung from nearby structures, such as the mediastinum. in the future, iodinated or other contrast materials may help more finely define these anatomical structures, until the algorithm is applied to hrct, which provides finer resolution and, therefore, clearer anatomical structure contrast. this way, it would be easier to isolate the lung during the steps using amira. however, the extraneous materials have a relatively small percentage in the original lung constructed, so there is still a little room for error in terms of having a slightly poor exclusion of these anatomical features. it was found that on average, even if quantifications were done having all three principal anatomical features still in the volume of interest, there would be a coefficient of variation of less than 3%, which is still well under the statistically significant threshold of 5%.the removal of the anatomical features is for lung infection consistency and improved precision. another limitation of this animal study was the small sample size; we intend to extend this technique to another group of mice with pneumonia. finally, these studies examined only one strain of bacteria and mice, with the inclusion of antibiotics and supportive care. extending these techniques to other mice will require additional validation, but may help to provide a non-invasive endpoint for studies with experimental pneumonia in transgenic animals. this method of quantitative assessment of pneumonia severity by ct has potential for application in clinical trials in community-acquired pneumonia, as well as other lung diseases. armbrust, lj, derek a. mosier (2005). correlation of results of pulmonary computed tomography and pathologic findings in mice with pasteurella-induced pneumonia. american journal of v e t e r i nary research 66 835-838. boone (2004). small-animal x-ray dose from microct. moecular imaging 3 149-158. iwaki, t (2001). a color atlas of sectional anatomy of the mouse). jelic, s (2005). pneumonia explained what is pneumonia? how is pneumonia contracted. the new york times company. l hsu, es, r summers, d koziol (2007). computed tomography scans to assess severity of bacter ia l pneumonia in a mouse model. m amigoni, gb, s. masson (2008). lung injury and recovery in a murine model of unilateral acid aspiration: functional, biochemical, and morphologic characterization. anesthesiology 108 1 0 3 7 1046. m. s. niederman, jjm, a n unger, a kumar (1998). the cost of treating community-acquired pneumonia. clin ther. mohd, n a discrimination method for the detection of pneumonia using chest radiograph. computerized medical imaging and graphics 160166. muller, nl (2006). pulmonary infection: basic concepts. imaging of pulmonary infections 1-17. ritman, el (2008). micro-computed tomography of the lungs and pulmonary-vascular system. proceedings of the american thoracic society 2 477480. schimel, d (2007). steps for amira procedure for segmentation to quantify and visualize lung volume in well-aerated lung and bacterial pneumonia. uchiyama, y (2003). quantitative computerized analysis of diffuse lung disease in high-resolution computed tomography. medical physics 30. references comparison groups p-value from t-test of percent lung by semi-automated segmentation in material h p-value from t-test of radiologists scores saline vs. bacteria 0.001 0.004 pneumonia survivors vs. pneumonia deaths 0.039 0.581 all survivor vs. bacterial deaths 0.001 0.058 table 4 *vtwhypzvu�vm�4pjyv*;�x\hu[päjh[pvu�i`�4h[lyphs�/�[v�x\hspäjh[pvu�i`�yhkpvsvnpz[z���(�jvtwhypzvu�vm�[ol�hipsp[`�vm�[ol�zltp�h\[vth[lk�tl[ovk�[v�[lss�nyv\wz�hwhy[�]lyz\z� radiologist ability. scoring by the semi-automated method is far more precise than the scoring by the radiologists. figure 6 example of a saline mouse lung the percentages that each of the three groups had for the 8 density ranges. as it can be seen, the pneumonia non-survivors had the greatest percent distribution in the higher lung density ranges, indicating a pneumonia-affected lung. the saline mice, however, had a greater percentage towards the lower density ranges, the opposite of the pneumonia non-survivors. figure 7 example of a pneumonia-survivor mouse lung frontal view of a pneumonia survivor mouse’s lung. the lung has far more red shade than the saline mouse, indicating its relative high density. figure 8 example of a pneumonia non-survivor lung frontal view of a pneumonia non-survivor mouse’s lung. the lung is almost completely shaded red, showing that the mouse has a severely pneumonia-affected lung. i’m upset, not ashamed: an investigation into adolescent shame acknowledgement olivia putnam1 , keely lake2 1department of psychology, northwestern university, evanston, il 2wayland academy, beaver dam, wi shame, emotion, psychology, negative affect, mood disorder, self-conscious emotion abstract: shame is a powerful and acutely painful “master emotion” that is strongly correlated with maladaptive behaviors and a host of psychological symptoms. concerningly, the affect remains under-researched and difficult to identify or address in a clinical setting. this may be caused, at least in part, by shame’s intrinsically hidden nature, which drives people to deny the emotion and express it through other means. this study aimed to understand the degree to which people fail to acknowledge their own shame and the psychological and behavioral implications of this shame. participants completed both a self-report measure of shame and an empirical assessment of internalized shame, as well as measures of shame coping methods and emotional regulation. as expected, results showed no significant correlation between participants’ self-rated shame and measured shame. we also saw a significant correlation between assessed internalized shame and use of shamecoping methods as well as difficulty in emotion regulation – specifically, difficulty with clarity of emotion, acceptance of emotions, and strategies for coping with emotions. these findings indicate that people struggle to acknowledge their own shame and also speak to the maladaptive, dysregulated ways people manage their shame. recognizing shame as a powerful emotion with implications in psychiatric disorders and understanding the factors that prevent people from acknowledging their own shame may help improve treatment for those who struggle with the emotion and reduce the likelihood that they will engage in maladaptive coping behaviors. introduction shame is one of the most under-researched emotions. whereas emotions like sadness, anger, and nervousness have undergone decades of psychological research and have come to be seen as the underlying feelings behind widespread clinical diagnoses like anxiety or depression, shame research within the field of psychology was virtually nonexistent until the last two decades and remains remarkably limited today. this lack of research is particularly troubling when it comes to shame because the affect is highly maladaptive. unsurprisingly, as a result, it is highly correlated with addiction, depression, violence, aggression, bullying, suicide, and eating disorders. in recent years, it has also become an emerging component of ptsd (taylor, 2015). a brief history of shame conceptualization one especially influential early premise for defining shame, and differentiating it from guilt, is early anthropologists’ focus on public vs. private transgressions (e.g., benedict, 1946). anthropologists commonly distinguished shame based on the situations they believed elicited it. more specifically, shame was conceived as a "public" emotion, arising from public exposure and disapproval of some transgression in societal rules and norms. guilt, on the other hand, was described as a more "private" experience arising from self-inflicted criticism and regret. however, more recent empirical research has failed to support this public/private distinction (tangney, marschall, rosenberg, barlow & wagner, 1994; tangney, miller, flicker & barlow, 1996). one such example is a study conducted in 1992 asking participants to describe three guilt-inducing events and three shameinducing events. a systematic analysis of the social context of these events found that shame and guilt are equally likely to be experienced in the presence of others (tangney, et al., 1992). "solitary" shame experiences were equally as common as "solitary" guilt experiences. even more to the point, “the frequency with which others were aware of the respondents' behavior did not vary as a function of shame and guilt” (tracy, 2011). this led to a new conceptualization of shame which remains today: shame as holistic negative self-concept. helen block lewis, renowned psychologist and pioneer of this shame-understanding, asserted that while guilt involves a negative evaluation of a specific behavior, shame involves a negative evaluation of the global self ("i did something bad" vs. "i am bad") (1971). though the distinction may appear inconsequential, this contrasting emphasis on the self “sets the stage for very different emotional experiences and very different patterns of motivation `and subsequent behavior” (tracy, 2011). the two emotions, for instance, produce distinct * 27 “action tendencies.” shame is commonly accompanied by attempts to deny, hide from, or escape the experiences that elicit shame, while guilt typically leads to “reparative action” – confessing, apologizing, undoing. this difference in internal conceptualization of the self and subsequent “action tendencies” is part of what ultimately makes shame maladaptive. while guilt can be painful and overwhelming, it is generally limited to the guilt-inducing action or experience. shame consumes the entire self, leaving experiencers with a globally-negative self-conception (“i am a terrible person”). this negative self-concept is not only painful and distressing, but it also feels irreparable. a person can correct a behavior, but one’s fundamental essence seems permanent. this sense of futility drives much of shame’s maladaptivity. rather than embracing adaptive behaviors like apologies or changes in behavior, which increase psychosocial success, shame-experiencers tend to recede and hide from the shame-inducing event. often, this leads people to isolate themselves socially, withdraw from activities that potentially remind them of the shame, and engage in anhedonic-behaviors. in other cases, this avoidance manifests as anger or hostility, as experiencers attempt to “turn the tables” on others to avoid their own shame, or as risky behavior (i.e. substance abuse) which many use a distraction from their shame (ellison, 2006). in instances where shame-experiencers have in fact committed some wrongdoing (the determination of which is of course subjective) the failure to take the expected “reparative action” can lead to social conflict (tangney, stuewig & mashek, 2007). for those who experience shame about events for which reparative action is not typically expected (i.e. being the victim of sexual assault, mental illness, minor mistakes or failures), shame can cause dissociation, debilitate people from talking about their experience, and limit much-needed processing of their own emotions and/or trauma (taylor, 2015). either way, these maladaptive responses to shame led researchers to “consistently report a positive relationship between proneness to shame and a host of psychological symptoms, including depression, generalized anxiety and social anxiety, low self-esteem, ptsd, eating disorder symptoms, cluster c personality disorders, suicidal behavior and self-injurious behavior, and substance abuse” (tracy, 2011). shame acknowledgement in the context of shame’s maladaptive consequences, the lack of research into the affect becomes dangerous. without comprehensive research into shame and its implications in psychological disorders, we cannot develop evidence-based treatments for shame-related disorders or, more importantly, adjust treatments for preexisting mental health disorders in which shame plays a more important role than previously-realized. one roadblock that commonly hinders the development of these treatments or the confrontation of shame within a therapy setting is the fact that shame often goes unacknowledged by the experiencer (mcgonigal, 2016). as terry f. taylor ph.d. writes in a review article of peritraumatic shame, “shame...is a virtually invisible, ubiquitous part of everyday life. because the experience of shame is often considered to be painful and disempowering, and because recognition of shame in itself can be felt as shameful… shame remains unacknowledged and is expressed as avoidant behavior” (2015). this instinct to hide one’s shame “makes it difficult to recognize internally when it happens,” let alone acknowledge out loud (luoma, 2012). concerningly, this tendency among people not to acknowledge their own shame also makes it difficult to study the affect, as it renders self-report measures unreliable. present study the consensus that shame characteristically goes unrecognized has never been scientifically reviewed. further, researchers have not studied whether the degree to which people report their own shame correlates with their mental health in other capacities. our research attempts to fill this gap. like most shame-studies before it, we utilize assessments intended to empirically measure participants’ levels of shame. in addition to these assessments, however, we also use an assessment of emotional affect that asks participants to self-rate the frequency with which they experience different emotions, “ashamed” being one them. comparisons of participants’ scores on the shame assessments and their self-reported level of shame will act as quantified measurements of how well they acknowledge their own shame. this research will not only test the assumption of shame’s unidentified nature but may also provide some insight into the prevalence of unacknowledged shame and how it affects people’s mental health. methods participants and procedure the assessments were administered to a sample of 54 adolescent students between the ages of 14 and 17 (58% female, 42% male, m = 16.3 years old, sd = 1.3, 55% american). participants were recruited from the student population at wayland academy, a small boarding high school with international students. as an incentive for participation, students were offered a small amount of extra credit in their science and math classes. participants were each given a battery of psychological assessments including the positive and negative affect scale (panas), the internalized shame scale (iss), the compass of shame scale (css), and the difficulty in emotion regulation scale (ders). participants were assured that their responses would be anonymous and confidential. assessments were given in a quiet, distraction free room. assessments positive and negative affect scale: (panas; watson and clarke, 1998) the most commonly used measure of affect in scholarly research, the panas is comprised of 10 negative affects (afraid, upset, distressed, jittery, nervous, ashamed, guilty, irritable, hostile) and 10 positive affects (enthusiastic, interested, determined, excited, in28 spired, alert, active, strong, proud, attentive). participants use a 0-4 likert scale to rate the frequency with which they tend to experience each affect. for the purposes of this research, the “ashamed” item was used to measure self-rated shame because, unlike empirical assessments of shame, it requires participants to explicitly endorse the word “ashamed.” internalized shame scale: (iss; rosario and white, 2006) the internalized shame scale, a 30-item questionnaire, is the most widely used empirical measure of shame across psychology and sociology research. the assessment has two subscales that are intended to be reported separately: a 24-item shame scale and a 6-item self-esteem scale. the shame scale attempts to tease apart different experiences of shame to create a holistic measure of the affect. importantly, the iss does not actually use the word shame, because shame can itself be a shameful thing to admit. there are four identified cutoffs: a score of 50 or higher indicates problematic levels of shame, a score of 60 or higher indicates possible depression and/or other emotional or behavioral problems, a score of 70 or higher indicates a high probability of depression and/or other emotional or behavioral problems. compass of shame scale: (css; ellison, 2006) because shame is an emotion that commonly goes unacknowledged both internally and outwardly, it is often expressed through other emotions or behaviors. the compass of shame scale recognizes this tendency and assesses the maladaptive ways people cope with shame. the four coping methods that it identifies are “attack self,” “attack others,” “withdraw,” and “avoid.” this scale is of particular importance because it acknowledges shame as a fundamental source for many other maladaptive, unhealthy behaviors and is the first of its kind to assess and quantify these shame-based behaviors. additionally, the test questions are situational rather than experiential. participants cannot always recognize feelings as shame, but they can often identify situations that produce those shame feelings. difficulty in emotional regulation scale: (gratz, 2004) this assessment represents one of the most popular, comprehensive and well-established measures of emotion regulation and is widely used in both clinical and nonclinical settings. the questionnaire assesses five primary components of emotional regulation: emotional awareness, emotional clarity, emotional acceptance, impulse control, ability to engage in goal-directed behavior while experiencing negative emotions, and ability to use situationally appropriate emotion regulation strategies flexibly to modulate emotional responses as desired. results self-rated shame vs. measured shame to investigate the relationship between self-rated shame and measured shame, pearson correlations were computed between ratings from the panas item “ashamed” and scores from the internalized shame scale. as expected, there was no significant correlation between self-rated shame and assessed “true” shame. the correlation and significance are displayed in figure 1. figure 1. pearson correlation between panas “ashamed” score and internalized shame score. this graph shows the pearson correlation between panas “ashamed” rating and iss scores, r = 0.0806, p > 0.05 to visualize how shame acknowledgment relates to shame’s maladaptivity, an anova test for the significance of differences in mean panas “ashamed” ratings among internalized shame cutoff groups – shame w/in normal limits, problematic levels of shame, possible indicator of mental health disorders, and likely indicator of depression/mental health disorder -was performed and is displayed in figure 2. an anova test for the differences in panas “ashamed” ratings between iss cutoff groups was computed and is displayed in figure 2. figure 2. analysis of variance test of panas “ashamed” score by iss cutoff groups. this graph shows the mean panas ashamed ratings reported by individuals in different iss cutoff groups, f ratio = 0.903, p > 0.05 as expected, and in line with our other findings, there was no significant difference in self-rated shame (panas “ashamed”) between among iss cutoff groupings (i.e. all participants reported similar levels of sham, no matter how much shame an individual was actually experiencing). 29 identifying shame as negative affect in order to explore whether participants may identify their internalized shame as other negative emotions, pearson correlations were computed between panas negative affect item scores and internalized shame scale scores. both are displayed in table 1. table 1. pearson correlations between panas negative affect subscores and internalized shame scale score. this table shows the results of pearson correlations between various panas negative affect subscores and iss score. there was also a significant positive correlation between panas negative affect subscale total score and internalized shame score (r = 0.534, p < 0.01). implications of shame to investigate shame’s relationship with other emotional and behavioral problems, pearson correlations were computed between iss scores and scores on the individual compass of shame scales and the difficulty in emotion regulation scale (figure 3). strong correlations between the iss and the withdrawal and attack self scales were expected and obtained. both correlations were significantly stronger than the iss correlations with the avoidance and attack other scales. all correlations were significant. category differences male/female, age, and continent of origin differences were assessed for both the internalized shame scale and the panas “ashamed” rating. women tended to have higher internalized shame scores than men (t = 1.76, p < 0.05). there was no significant difference, however, between men and women for the panas “ashamed” rating. there was also no significant correlation between age and internalized shame score or the panas “ashamed” rating. continent of origin did not produce any significant differences in iss score or panas “ashamed.” figure 3. pearson correlations between css subscores and iss scores. a, pearson correlation between css attack self subscore and iss score, r = 0 .75, p < 0.001 b, pearson correlation between css withdraw subscore and iss score, r = 0 .72, p < 0.001 c, pearson correlation between css avoidance subscore and iss score, r = 0.41, p < 0.01 d, pearson correlation between css attack others subscore and iss score, r = 0.27, p < 0.05 30 discussion the lack of correlation between iss score and panas “ashamed” rating, as displayed in figure 1, indicates that there is no relationship between self-rated shame and empirically-measured “true” shame. this is further demonstrated by the analysis of variance displayed in figure 2, which shows that there is no significant difference in selfrated shame between cutoff categories of “true” iss shame. essentially, even those who experience shame at an intensity high enough to indicate depression or other mental health disorders tended to describe their experience of shame as “rare.” a paired t-test comparing average iss shame and panas shame affirmed this underreporting phenomenon in individuals, with nearly 74 percent of participants reporting lower self-rated panas shame than “true” iss shame. importantly, there was a significant positive correlation between iss score and panas negative affect subscale score, suggesting that while people struggle to accurately identify their shame, they may describe it broadly as negative affect. this point is reiterated by the fact that, with the exception of “distressed,” the individual panas negative affect items had weak or insignificant correlations with iss score and none had stronger correlations than the negative affect subscale score. this indicates that people are not calling shame by another name, but instead use negative umbrella terms, or a variety of different emotion words, to imprecisely describe the feeling. these results both confirm a common understanding that shame often goes unacknowledged and also underline the difficulty of identifying and properly addressing shame. this pattern becomes especially meaningful in the context of internalized shame’s negative implications. iss score was found to have positive correlations with all css scales: attack self, withdraw, avoid, and attack others. given their naturally internalized nature, the “withdraw” and “attack self” coping mechanisms had a stronger relationship with internalized shame than “avoid” or “attack other.” internalized shame also showed a strong positive relationship with ders score and with the lackof-clarity, non-acceptance, and strategies subscales specifically. this fits well with the findings about unacknowledged shame, as clarity and acceptance are both components of emotional acknowledgment, and are necessary for strategic management of one’s emotions. the male/female differences on the iss reflect those of previous studies, with women tending to experience more internalized shame than men. also similar to adult studies, age was not a significant factor in iss score, suggesting that experience of shame does not change significantly during adolescent development. given that this comes from a cross-sectional review rather than a longitudinal one, however, the accuracy of this conclusion is limited. continent of origin, which has never before been studied in relationship to internalized shame, produced no significant difference. the accuracy of this conclusion, however, may be compromised by the small and varying number of participants within each continent group. conclusion the comparisons of self-rated shame and empirically measured “real” shame confirm both our hypothesis and a larger long-held public understanding that people hesitate to acknowledge their shame. this finding is important not only because it is the first of its kind to quantitatively validate that informal understanding, but, more significantly, because it speaks to the extent of the disparity. more than one third of participants experienced internalized-shame with a frequency associated with depression and other clinical disorders, yet the great majority of these participants rated their own experience of shame as rare. this is problematic because, if people cannot acknowledge shame as a component of their emotional distress or mental illness, then these problems become much more difficult to address within a clinical or intrapersonal context. this lack of acknowledgment becomes especially concerning if the shame centers around a specific, potentially-traumatic event (i.e. sexual assault). if shame prevents a person from speaking up about and working through such an experience, then symptoms can worsen dramatically. as time goes on, this can also become a self-fulfilling prophecy of sorts because the longer one avoids their shame the more internalized it becomes and, consequently, the more inhibiting it becomes. given that shame is a powerful emotion, however, it cannot be entirely ignored or suppressed. in fact, this study’s findings indicate that people may be able to recognize the emotion broadly as emotional distress. however, without specific identification and management, shame is often expressed through maladaptive coping mechanisms. the correlation between internalized shame and each pole of the shame-coping scale suggests that, rather than addressing shame head on, shameexperiencers tend to avoid the emotion through problematic behaviors such as: risk-taking and distraction, which can develop into substance abuse and have been shown to be severely maladaptive; withdrawal from social interaction, often a symptom of depression; excessive selfcriticism, also related to mental illness; or attacking others, an instinct associated with aggression and potential violence. though they vary in commonality, each pole is maladaptive in its own right—an attempt to ignore, hide from, wallow in, or push back one’s shame without ever truly acknowledging it. it is also important to note that these individuals coping mechanisms are not orthogonal and in fact tend to converge. this tendency to cope with shame is similarly reflected by the significant correlation between internalized shame and difficulty in emotional regulation. even more to the point, the ders subscales that had the strongest relationships with shame were lack-of-clarity, nonacceptance, and strategies. this indicates that people with high levels of internalized shame have significant trouble identifying their emotions, acknowledging their emotions without guilt or embarrassment, and coping with their emotions effectively. this not only speaks to the troubling ways people manage their shame but also to the dysregulatory nature of shame itself, which can make it 31 more difficult for a person to manage any of their emotions. the implication of these findings is three-fold: tools for shame recognition must be better integrated into both clinical and intrapersonal settings, shame must be better accounted for within diagnostic criteria, and more research into the affect must be conducted. it is imperative that mental health practitioners recognize that patients are unlikely to forthrightly acknowledge or report their shame and that these practitioners are trained in how to identify this underlying shame and address it without causing the patient to shut down or react with anger. it is also an unfortunate truth that, because the perception of shame in others “can also evoke a discomforting emotion in the observer,” it may fail to be addressed in therapy, with the therapist remaining in an “unconscious collusion with the patient” to ignore the shame (taylor, 2015). additional shame-specific training and wider acknowledgement of the emotion’s role in psychopathology may help mediate this problem an important step in helping clinicians (and, in fact, any individual) better identify and work with shame is better accounting for the emotion in diagnostic criteria. despite the fact that the emotion is strongly correlated with many different mental illnesses and social-emotional problems, shame is rarely listed as a symptom in the dsm iv and is generally relegated to the “associated features” of a disorder. not only does this lack of representation reinforce shame’s hidden nature, but it also fails to account for the way shame’s role in a disorder can change the way it must be treated. perhaps most importantly, more scientific research must be conducted into shame. since shame’s psychological conceptualization as holistic self-blame was established, only a handful of scientific studies into the affect have been conducted and, even fewer yet have investigated shame’s role in psychiatric disorders. if we are to help people recognize and cope with their internalized shame, we must first understand what prevents that acknowledgment and which tools are the most effective in fostering it. author information department of psychology, northwestern university evanston, il, usa references [1] h. l. bash, a. papa, shame and ptsd symptoms. psychological trauma: theory, research, practice, and policy. 6(2), 159–166 (2014). [2] r. benedict, the chrysanthemum and the sword, (houghton mifflin harcourt, 1946). [3] p. m. del rosario, r. m. white, the internalized shame scale: temporal stability, internal consistency, and principal components analysis. personality and individual differences. 41(1), 95–103 (2006). [4] j. elison, r. lennon, s. pulos, investigating the compass of shame: the development of the compass of shame scale. social behavior and personality: an international journal. 34(3), 221–238 (2006). [5] k. l. gratz, l. roemer, multidimensional assessment of emotion regulation and dysregulation: development, factor structure, and initial validation of the difficulties in emotion regulation scale. journal of psychopathology and behavioral assessment. 26(1), 41–54 (2004). [6] h. b. lewis, shame and guilt in neurosis, (international universities press, 1974). [7] j. luoma, j. lejeune, association for contextual behavioral science. (2005). available at: https://contextualscience.org/acbs [8] k. mcgonnigal, r. d. siegel, breaking the power of shame by engaging it. (2016). available at: https://nextlevelpractitioner.nicabm.co/week40/ [9] l. a. medrano, m. trogolo, construct validity of the difficulties in emotion regulation scale: further evidence using confirmatory factor analytic approach. abnormal and behavioural psychology. 2(2), 1000117 (2016). [10] p. michl, et al, neurobiological underpinnings of shame and guilt: a pilot fmri study. social cognitive and affective neuroscience. 9(2), 150–157 (2012). [11] d. sznycer, et al, cross-cultural differences and similarities in proneness to shame: an adaptationist and ecological approach. evolutionary psychology. 10(2), 147470491201000 (2012). [12] k. tangey, w. fischer, self-conscious emotions: shame, guilt, embarrassment, and pride, (guilford press, 1995). [13] j. p. tangney, recent advances in the empirical study of shame and guilt. american behavioral scientist. 38(8), 1132–1145 (1995). [14] j. p. tangey, "shame and guilt in interpersonal relationships" in self-conscious emotions: the psychology of shame, guilt, embarrassment, and pride, j. p. tangey, k. w. fisher, eds. (guilford press, 1995), pp 114-139. [15] j. p. tangney, conceptual and methodological issues in the assessment of shame and guilt. behavior research and therapy. 34(9), 741-754 (1996). [16] j. p. tangney, s. a. burggraf, p. e. wagner, "shame-proneness, guilt-proneness, and psychological symptoms" in self-conscious emotions: the psychology of shame, guilt, embarrassment, and 32 pride, j. p. tangey, k. w. fisher, eds. (guilford press, 1995), pp. 343-367. [17] g. taylor, pride, shame and guilt: emotions of selfassessment. philosophical books. 28(1), 40–41 (1987). [18] t. f. taylor, the influence of shame on posttrauma disorders: have we failed to see the obvious? european journal of psychotraumatology. 6(1), 28847 (2015). [19] j. tracy, j. p. tangney, m. r. leary, handbook of self and identity, (guilford press, 2012). [20] d. watson, l. a. clark, a. tellegen, development and validation of brief measures of positive and negative affect: the panas scales. journal of personality and social psychology. 54(6), 1063–1070 (1988). [21] m. r. zaslav, shame-related states of mind in psychotherapy. j psychother prac res. 7(2), 154–166 (1998). 33 template for electronic submission to acs journals 35 © 2023 robbins, jocelyn & yuan, melinda. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. inside the nucleon: tomographic interpretations and universality of gpds with ddvcs jocelyn robbins1, melinda yuan2, marie boër3 1university of colorado, boulder 2columbia university in the city of new york 3virginia polytechnic institute and state university keywords: generalized parton distributions (gpds), double deeply virtual compton scattering (ddvcs), nucleon structure, quantum chromodynamics (qcd), tomographic interpretations abstract: the goal of double deeply virtual compton scattering (ddvcs) experiments is to better understand the internal structure of the nucleon. previous attempts to resolve the internal structure of nucleons have resulted in electromagnetic form factors and parton distribution functions for elastic scattering and deep inelastic scattering processes, respectively. generalized parton distributions (gpds) are the latest attempt to unify these models of nucleon structure. the gpds of ddvcs give us ability to investigate off of the diagonal where x ̸= ±ξ. the main goal of our analysis is to determine the best experimental setup in order to deduce the kinematic variables on which gpds depend from the lab observables. the effectiveness of our data collection in the laboratory is by determined the physical kinematics, q2, q′2,t, xi,ϕlm, ϕcmv , and θcmv . we can then run ddvcs experiments and collect data on observables to improve upon the current models for gpds of the nucleon. introduction the fundamental aim of particle physics is to discover and understand matter down to its smallest possible constituents. the discovery of the quark in 1964 revolutionized this process, breaking apart nucleons, which were previously believed to be elementary particles[6]. these subatomic particles were detected using scattering, a process in which a high energy particle beam is directed at a stationary object, resulting in collisions between the particles in the beam and object. data is collected from these collisions, such as cross sections or scattering byproducts, which is then used to infer information about the internal structure of the object. the model that we are attempting to explore is generalized parton distributions, currently the most detailed model of nucleon structure to exist. gpds are a hybrid of its predecessors, form factors (produced through elastic scattering) and parton distributions (produced through deep inelastic scattering)[8]. the particular scattering process that we are investigating is double deeply virtual compton scattering (ddvcs), in which an electron beam is scattered off a proton, exchanging a virtual photon in the process. the outgoing virtual photon will then decay into a detectable muon-antimuon pair as seen in figure 1[7]. columbia undergraduate science journal vol. 17, 2023 robbins, yuan, et. al. 36 the current reactions studied are timelike compton scattering (tcs) and deeply virtual compton scattering (dvcs). both of these collisions include one virtual photon, with tcs containing an outgoing virtual photon and dvcs an incoming virtual photon. in ddvcs experiments, both the initial and final photons are virtual. the virtuality of the photons impacts the matrix elements that describe the reaction. the matrix element refers to the probability amplitude of finding a quark at a space-time point in a nucleon, then finding the same quark at another space-time point in the nucleon which has now changed its momentum [5]. the involvement of 2 space-time points means that the matrix element is non-local, and the differing momenta of the initial and final nucleon makes the matrix element non-forward. gpds depend on the following kinematic variables: q2, the virtuality of the exchanged photon, t, the momentum transfer to the nucleon, bjorken x (xbj), the fraction of the total nucleon momentum, and three reaction angles. these kinematics be used to calculate ξ and ξ’, components of light cone frame momentum, using the following equations [3]. using ξ and ξ’, the longitudinal momentum transfer fraction of incoming spacelike photon, -2(ξ−ξ’), and the longitudinal momentum transfer fraction of outgoing timelike photon, (2ξ’−ξ), can be calculated. these momentum transfers make ddvcs unique because in a ddvcs reaction, ξ ̸= ξ’ unlike tcs and dvcs reactions where ξ = ±ξ’ [3]. when ξ ̸= ξ’, xbj and ξ dependence can be decoupled, allowing access to ‘off-diagonal elements’ of gpd regions. currently, with tcs and dvcs, only elements on the diagonal xbj = ±ξ are accessible due to the dependence of x and ξ on each other. however, x and ξ decoupling enables newly accessible regions of gpds, which will allow for a more detailed picture of the distribution of nuclear forces inside a nucleon as well as the determination of parton transverse densities. such investigation will add to the current model of gpds and facilitate more detailed tomographies of the nucleon. another focus of the ddvcs experiments is the question about the universality of gpds. under current theories, gpds are presumed to be universal, which means that the calculations of gpds will be the same regardless the experiment used to measure them. however, there is no concrete experimental proof to support this conclusion. ddvcs reactions have the unique ability to simultaneously study spacelike (defined as q2 > q’2) and timelike regions (defined as q’2 < q2) in order to determine the two regions result in the same leading order and twist. the agreement of results between the spacelike and timelike regions would then in turn provide support of the universality of gpds figure 1: ddvcs e− + e−→µ+ + µ−[2] columbia undergraduate science journal vol. 17, 2023 robbins, yuan, et. al. 37 methods in order to simulate ddvcs collisions, we used the event generator deepgen, developed by dr. marie boër. these event generators, written in c++ and run through root, generate simulated particle collisions events given a certain set of parameters, such as luminosity, beam energy, and phase space. it can be altered depending on the type of collision it is replicating. the version that we used, deepgen 5.0, simulates deep exclusive photoand electroproduction of lepton pairs and photons, including ddvcs as well as dvcs and tcs. the parameters of the event generator were set to match experiments at hall c of jefferson lab with an 11 gev electron beam. the generator creates equal weighted events and uses a montecarlo simulation technique to weight the events with an n-differential cross section[4]. the weights are multiplied by a normalization factor, where l is luminosity and ∆ω is the dimension of the phase space. there are five weighting options; total unpolarized, ddvcs, bethe-heitler (bh), ddvcs/bh, and beam spin asymmetry. the beam spin asymmetry weighting provides insight into how difficult it will be to measure the asymmetric polarization of the electron beam. simulated events with the total unpolarized weight are proportional to the number of measured events in the physical experiment. the total unpolarized weight is calculated with the ddvcs weight and bh weight, wtot = |wddvcs + wbh|2. analyzing the data with ddvcs and bh weighting allows their contributions to be distinguished. the bh contribution is precisely known due to it’s dependence on qed calculations and proton form factors[1]. it describes the hard (known) region of ddvcs reactions. the ddvcs contribution, on the other hand, describes the soft region of the reactions and can not be calculated. the ratio of ddvcs and bh weighting allows for insights into where the so-called ‘new physics’ can be found. areas where ddvcs/bh is large are promising in terms of information about the gpds in the previously inaccessible regions. figure 2: illustration of non-local non-forward matrix element[5] figure 3:the ddvcs and bh contributions to the total unpolarized weight[2] columbia undergraduate science journal vol. 17, 2023 robbins, yuan, et. al. 38 after generating the simulated and weighted ddvcs events, we analyzed the data to better understand how the experimental kinematics variables relate to each other. in addition to the variables of t, ξ, ξ′ displayed in figure 1 above, the relationship between the various angles in the particle scattering, shown below, were explored. the experiment takes place on three planes, with the far left plane being created by the initial and final electrons from the electron beam, the middle plane by the photons and the nucleon, and the plane on the far right by the muons. the notation cm indicates the measurements are taken from the center of mass frame. the angles of interest are θcm, the angle between the incoming and outgoing electrons, ϕl, the angle between the plane of the electrons and the plane of the photons, and ϕcm, is the angle between the plane of the muons and the plane of the photons. the last angle, θcm, is the angle the scattered muon makes from parallel. discussion while the earlier graphs were created to better understand the relevant kinematic relationships, the later graphs explored how events can be interpreted. the first set of graphs (figure 5) traces out the amount of events that exists within a certain angle range (corresponding to the angles in figure 4). the ϕ angles have a range of π and the θ graph measures entries between 0 and π. figure 5: θcm (upper left), ϕcm (upper right), and ϕl (lower) figure 6: ’ both of the center of mass angles (ϕcm and θcm appear largely symmetric without many irregularities. however, the ϕl graph displays unexpected asymmetries, which may arise due to relations between all three angles and warrants more exploration. the next figure, which compares ’, provides important information about events that need to be excluded from the data set and events that might be useful for physical interpretation. the events along the timelike and spacelike cut line are events that must be excluded. the line corresponds with = 1 which makes the data difficult to interpret. figure 4:collision vectors and angles columbia undergraduate science journal vol. 17, 2023 robbins, yuan, et. al. 39 figure 7: ξ vs ξ’ figure 7 is a comparison of ξ and ξ’ with boundaries that define later data cuts in q2 vs q’2. by comparing the ξ and ξ’, we are granted the ability to determine how far ’off the diagonal’ we can measure our events. the diagonal, in this case, is defined as x = ±ξ and has previously restricted the investigation of gpds in the erbl and dglap regions. the cuts that are made focus on areas where measured events can be physically interpreted. the cuts on the ξ vs ξ’ graph correspond to selected q2 and q’2 bands (figure 8). the graphs are also restricted in terms t, between −0.15 gev and −0.55 gev. t must be constrained because while there are more events measured between −0.55 gev and −1.05 gev, when the momentum transfer becomes too large, approximations are rendered invalid and a physical interpretation loses meaning. thus, the graphs must be constrained in t. the value of these graphs are derived from the fact that q2 and q’2 cannot be measured outright. the transfer of the virtual photon’s momentum occurs in the soft region, which is not physically accessible with our current technology, but can still be calculated from the measured kinematics. using the q2 vs q’2 selected bands and the bjorken x hypothesis, the structure of the proton can be determined. the bjorken x hypothesis states that, for point-like particles, as the limit q2 → ∞, gpds lose dependence on q2. hence, if the gpd demonstrates dependence on q2, it’s structure cannot be classified as pointlike.[9] summary and outlook double deeply virtual compton scattering is the next step when it comes to accessing and analyzing gpds. the incoming and outgoing virtual photons make ddvcs unique from previous experiments, allowing access to regions of generalized parton distributions that were previously unreachable. this is done through taking advantage of the fact that in ddvcs, ξ ̸= ±ξ’ which allows access figure 8: q2 vs q’2 selected bands, spacelike (left) and timelike (right) columbia undergraduate science journal vol. 17, 2023 robbins, yuan, et. al. 40 beyond the x = ±ξ diagonal gpds are currently restricted to. ddvcs also provides opportunity to compare measurements of gpds in spacelike and timelike regions simultaneously, which can be used to evaluate the validity of gpd universality. the deepgen event generator provided events that could be applied to different weighting systems and analyzed. boundaries were created for q2 vs q’2 using ξ vs ξ’, allowing for greater insight into proton gpds as well as an understanding of the extent to which measurements can be taken ’off of the diagonal.’ the future of this project requires more data analysis and planning for the jefferson lab proposal. the kinematics regarding the bjorken x hypothesis contains a vast domain of interpretation that has potential to be explored further. additionally, more detailed data analysis can be done by with regards to the relationship between the experimental angles and the other kinematics variables, as this study lacks the proper angular corrections and acceptance cuts from q2 and other values. looking forward, there exists a great deal more investigation into ddvcs to be done, as it appears to be an untapped realm when it comes to uncovering more physics regarding proton gpds. author information corresponding author melinda yuan: my2740@columbia.edu funding sources this work was made possible by the national science foundation under grant no. phy2149165. acknowledgments we would like to thank our mentor, dr. marie boër for the opportunity to take part in her research and for her guidance through the research process. we acknowledge the outstanding support from the national science foundation, the virginia tech physics department, and the virginia tech center for neutrino physics. abbreviations ddvcs: double deeply virtual compton scattering gpds: generalized parton distributions tcs: timelike compton scattering bh: bethe-heitler references [1] c. adloff. “measurement of deeply virtual compton scattering at hera”. in: physics letters b 517.1 (sept. 2001), pp. 47–58. issn: 03702693. doi: 10.1016/s03702693(01) 00939-x. arxiv: hep-ex/0107005. url: http://arxiv.org/abs/hep-ex/0107005 (visited on 07/12/2022). [2] marie boër. double deeply virtual compon scattering: measuring ddvcs in hall c, why and how? url: https://indico.jlab.org/event/546/contributions /9973/ attachments/7924/11128/collabmeetinghallac_jun e2022_mboer.pdf (visited on 07/12/2022). [3] alexandre camsonee, mark k. jones, and jian-ping chen. measurement of double deeply virtual compton scattering in the di-muon channel with the solid spectrometer. https://www.jlab.org/.url:https://www.jlab.org/ exp_prog/proposals/15/loi12-15-005.pdf (visited on 07/12/2022). [4] deepgen event generator. solid wiki. url: https://solid.jlab.org/wiki/index. php/deepgen_event_generator#normalizati on (visited on 07/12/2022). columbia undergraduate science journal vol. 17, 2023 robbins, yuan, et. al. 41 [5] m. guidal. “generalized parton distributions : experimental aspects”. in: nuclear physics a 699.1 (feb. 2002), pp. 200–209. issn: 03759474. doi: 10.1016/s03759474(01)01494-4. arxiv: hep-ph/0108143. url: http://arxiv.org/abs/hep-ph/0108143 (visited on 07/12/2022). [6] cian o’luanaigh. fifty years of quarks. cern. jan. 17, 2014. url: https://home.cern/ news/news/physics/fiftyyears-quarks (visited on 07/12/2022). [7] b. pire, l. szymanowski, and jakub wagner. “timelike compton scattering from jlab to rhic and lhc energies”. in: proceedings of science (july 10, 2012). [8] a.v. radyushkin. “15 years with gpds”. in: osti (). url: https://www.osti.gov/ servlets/purl/1093878. [9] mark thomson. particle physics. 2011. url: https://www.hep.phy.cam.ac.uk/ ~thomson/partiiiparticles/handouts/handout _6_2011.pdf (visited on 07/18/2022). columbia undergraduate science journal open-access publication | cusj.columbia.edu spring 2013 | volume 7 columbia undergraduate science journal 10 characterizing the biomechanical and biochemical properties of mouse uterine tissue eli mondragon1*, kyoko yoshida2, m.s. kristin myers2, ph.d 1biomedical engineering department, columbia university, new york, ny 10027 2mechanical engineering department, columbia university, new york, ny 10027 abstract for a successful pregnancy, the uterus and the cervix work together as a biomechanical structure to protect the fetus until term. during gestation, typically 37 weeks, the uterus undergoes a growth transformation to accommodate the growing fetus and to prepare for labor. this uterine growth is characterized by an increase of its wet weight, elastin content, and collagen content. then at parturition, the uterus must contract while the cervix ripens and dilates to allow the passage of the fetus. the transformation mentioned above is believed to be responsible for the contractions, and any deviations from the expected biochemical transformation put both the mother and baby in danger. the goals of this study are to quantify and compare the biochemical and biomechanical properties of uterine tissue from normal and abnormal mouse models of pregnancy. this study utilizes anthrax toxin receptor 2 knock-out mice (antxr2 -/-), which exhibit an accumulation of collagen in the cervix and uterus as a result of a defect in the maintenance of their extracellular matrix (ecm). uterine tissues from nonpregnant antxr2 -/and non-pregnant wild type mice (antxr2 +/+) were tested. tissue samples were tested for collagen content, collagen crosslink strength (i.e. collagen extractability) and were subjected to tensile mechanical testing. results from the biochemical assays revealed that the antxr2 -/uterine samples had significantly higher levels of collagen. it was also revealed that collagen extractability was region-dependent. lastly, mechanical testing proved that antxr2 -/uterine tissue is mechanically stronger than antxr2 +/+ (peak stress 0.078 mpa and 0.04 mpa). this study presents one of the first attempts to correlate the biochemical makeup of the uterus to its biomechanical properties. key words: anthrax toxin receptor 2; extracellular matrix introduction the united states has an annual preterm birth (ptb) rate of approximately 12%, making it the highest among developed nations [5]. although the causes of ptb remain elusive, cervical insufficiency (ci) is a leading mechanism of ptb [5]. ci is the premature softening of the cervix in the absence of uterine contractions. in addition, a cervix that fails to dilate/ripen at term can be fatal to both the mother and child. these   two   medical   conditions,   a   “soft”   cervix   and  a  “stiff”  cervix,  highlight  the  importance  of   the   cervix’s   dual   function. a need for a better understanding of cervical ripening has led to multiple studies on the biomechanical and biochemical properties of cervical tissues [6, 810]. however, for a successful pregnancy, the cervix and uterus must work together to retain the fetus during full gestation. subsequently, at parturition, uterine contractions and cervical ripening occur to allow for the delivery of the fetus through the birth canal. an understanding of the biochemical and biomechanical relationship in uterine tissue is therefore equally as important as the cervix in building finite element models, reducing both the ptb rate and neonatal deaths. future finite element models will help investigate how the uterus and copyright: © 2013 the trustees of columbia university, columbia university libraries, some rights reserved, sorrento, et al. received 1/1/2013. accepted 2/1/2013. published 1/1/13 *to whom correspondence should be addressed: biomedical engineering department, columbia university, new york ny 10027, email: em2787@columbia.edu mailto:em2787@columbia.edu columbia undergraduate science journal open-access publication | cusj.columbia.edu spring 2013 | volume 7 columbia undergraduate science journal 11 the cervix work together during gestation, and will require the input of material properties that can only be acquired through experimentation. the mechanical properties of the uterus have been previously studied [2,3,11]. early work done on human myometrium by conrad, et al. has shown that pregnant tissue is more compliant than non-pregnant tissue [3]. more importantly, it has been shown that the connective tissue framework, not the muscle cells of the uterus, is responsible for the mechanical behavior of the organ [2]. although conrad, et al. conclude that the extracellular matrix  (ecm)  is  crucial  for  a  tissues’  mechanical   integrity, biochemical analysis was not performed. other studies have investigated the ecm of uterine tissue, but these were qualitative [12-14]. briefly, peters, et al. worked with capillary morphogenesis protein-2 knock-out mice (cmg2 -/-) and reeves, et al. with antxr2 -/mice. similar to the antxr2 -/ mouse model, cmg2 -/mice are characterized by the accumulation of collagen in the cervix and uterus. both researchers used h&e (hematoxylin   and   eosin)   and   masson’s   trichrome stains to reveal the accumulation of collagen in the uterus of their respective mouse model [12,13]. spiess, et al. showed the distribution of different types of collagen (i, iii, and v) in the endometrium of pregnant mice using immunofluorescence and confocal imaging [14]. zhao, et al. also investigated the ecm of mouse uterine tissue, but their results are more quantitative. zhao, et al. showed that both pregnant relaxin knock-out mice (rlx -/-) and pregnant wild type mice (rlx +/+) contain 6% of dry weight of collagen in their uterine tracts [16]. these studies on the biochemical makeup of mouse uterine tissue had no biomechanical information [12-14,16]. previous studies have correlated biochemical makeup to biomechanical properties in other load-bearing soft tissues, but few studies have correlated biochemical variables to uterine mechanical properties. in particular, analysis of mutant and non-mutant mice achilles tendon has shown that alteration of the ecm manifests in changes to its biomechanical strength [7]. this study aims to connect the biochemical composition of the uterus to its mechanical properties. due to the difficulty in attaining human uterine tissue, mouse tissue was used throughout this research. the advantage of using mouse models is twofold. they provide timed pregnancy samples as well as an opportunity to investigate unique genetic models that may alter the maintenance of the ecm. for example, antxr2 /mice experience a loss in both longitudinal and circular myometrial cell layers, causing inadequate uterine contractions for delivery [13]. as the antxr2 -/mice age, the cervices and uteruses thicken and accumulate collagen due to a defect in ecm homeostasis. reeves, et al. have shown that the consequence of faulty ecm maintenance in antxr2 -/mice, morphologically different reproductive tracts, can be seen once the mice reach sexual maturity (3 months) [13]. the accumulation of collagen also leads to failed parturition. the nature of these effects, the abundance of collagen, provides the opportunity to investigate the relationship between the biochemical and biomechanical properties of the tissue (i.e. structure-function properties). to determine uterine structure-function properties, collagen content, collagen extractability (relative strength of the collagen network), and tensile strength of uterine tissue specimens from both antxr2 -/and antxr2 +/+ non-pregnant mice were measured. antxr2 -/ uterine tissue was found to have significant higher levels of collagen than antxr2 +/+. specific regions of antxr2 -/uterine tracts had significantly lower levels of extractabilities than antxr2 +/+ tissues, indicating antxr2 -/has a stronger collagen network (25 ± 5 % and 45 ± 4 %). lastly, antxr2 -/uterine tissue displayed stiffer properties than antxr2 +/+ (peak stress 0.078 mpa and 0.04 mpa). columbia undergraduate science journal open-access publication | cusj.columbia.edu spring 2013 | volume 7 columbia undergraduate science journal 12 methods sample preparation all procedures followed the nih guide for the care and use of laboratory animals. all mice sacrificed were between 3-4 months old and were in the diestrus stage of their estrous cycle. the uterine tracts of 7 non-pregnant antxr2 +/+ and 4 non-pregnant antxr2 -/were dissected from sacrificed mice. the two uterine tracts of the mouse were labeled left (l) and right (r), with the junction being evenly divided between the two (figure 1a). l horns were used for mechanical testing and the r horns were used for biochemical assays. once dissected and labeled, the uterus was separated at the junction of the two horns. the ovaries were left intact to determine orientation. both horns were sliced open to expose the inner lining. while the l horn was immediately stored away in a -80 ºc freezer, the r horn was divided into regions a, b, c, and d, where region a is the furthest from the ovary (figure1b). the samples were then stored in a -80 ºc freezer. figure 1: a uterine tract of antxr2 +/+ is divided into horns r and l. b division of an antxr2 +/+ r horn into regions a, b, c and d biochemical analysis all tissue samples were homogenized prior to testing by flash freezing and pulverizing each sample using liquid nitrogen. the wet weights of all regions after pulverization were measured. the dry weights of all regions were measured after freeze-drying. the hydration of a sample is defined as the fraction of the difference between wet weight and dry weight over the wet weight. the equation for hydration is (1) all regions, a through d, were assayed for collagen content using a standard hydroxyproline assay described in myers, et al [8]. briefly, a colorimetric procedure determined hydroxyproline content. collagen content was then obtained by using a mass ratio of hydroxyproline to collagen of 1:7.47, and normalizing by the dry weight of the tissue. collagen extractability was measured for regions a and d. the collagen extractability of a tissue determines the solubility of the collagen network. it was measured by extracting the pulverized tissue in 0.5 m acetic acid with 0.5 mg ml-1 of pepsin (150 µl mg-1) for 3 days at 4 ºc. after centrifuging the samples for 1 hour at 15,000g, the tissue pellet and the supernatant were separated. the hydroxyproline content was measured in both the supernatant and pellet. the extractability of a sample is defined as the fraction of hydroxyproline in the supernatant compared to the total amount of hydroxyproline. the equation for extractability is (2) biomechanical analysis all mechanical tests were conducted on a universal mechanical testing machine (instron micro tester 5948, norwood, ma) equipped with a 5 n load cell and a custom fluid chamber. two perpendicular ccd cameras (point grey grasshopper, gras-50s5m-c 75mm f/4 lens) were used to capture tissue deformation. immediately before testing, the l horns were thawed in pbs for approximately 30 minutes. samples were clamped into custom-made tensile jaws using sand paper and cyanocrylate to ensure a tight grasp. all samples were oriented such that the outer lining faced the same front camera. forceps were used to avoid b a columbia undergraduate science journal open-access publication | cusj.columbia.edu spring 2013 | volume 7 columbia undergraduate science journal 13 misalignment and the formation of wrinkles. prior to loading, each samples was airbrushed with india ink to produce a speckle pattern for tracking purposes (figure 2a). once the ink dried, each sample was loaded under a regimen that included three load-unload cycles to 5% engineering strain with a 5-minute reequilibration time between cycles. these cycles were followed by four ramp-hold tests. the four engineering strain levels were 10%, 15%, 20%, and 30%, each was held for 45 minutes. an engineering strain rate of 0.5% s-1 was used for all loading. see figure 2b for an illustration of the full loading regimen. during testing, images of the tissue deformation were taken at a frequency of 0.2 hz. figure 2: a. speckle pattern on uterine horn b. loading regimen for tension test results biochemical composition the hydration levels for both antrxr2 +/+ and antxr2 -/mice uterine samples were 76 ± 11% and 75 ± 10%.   a   student’s   t-test revealed that were no significant differences in hydration levels between regions in antxr2 +/+ and antxr2 /mice (p>0.05). in addition, hydration levels were not significantly different between antxr2 /and antxr2 +/+ samples. (figure 3a). it was found that collagen makes up 29 to 37% of antxr2 -/uterine tract (n=4), and only 12 to 17% of antxr2 +/+ uterine   tract   (n=7).   a   student’s   ttest showed that the collagen content levels in all regions of the antxr2 -/samples were significantly higher than their antxr2 +/+ counterparts (p<0.05) (figure 3b). collagen extractability is region specific in the case of antxr2 -/-. a   student’s   t-test also showed that region d of the uterine tract from antxr2 -/mice had significantly lower levels of extractability than region d of antxr2 +/+ samples (figure 3c) (p<0.05). the differences in extractability between regions a and d from antxr2 +/+ and antxr2 -/samples were not significant. figure 3: a uterine hydration levels were not significantly different between antxr2 -/and antxr2 +/+, b collagen content in all regions was significantly higher in antxr2 -/ mice, c region d for antxr2 -/was significantly less extractable than region d of antxr2 +/+ mice. a b a b c columbia undergraduate science journal open-access publication | cusj.columbia.edu spring 2013 | volume 7 columbia undergraduate science journal 14 biomechanical characteristics tensile tests were performed on whole uterine horns (l horns). figure 4a shows the averaged stress response of the tissue samples during the three load-unload cycles. there was a trend towards higher stress peaks for all three cycles for antxr2 -/samples (n=3) compared to antxr2 +/+ samples (n=3), but they did not reach significance (p=0.1, 0.9, 0.18,  student’s  t-test).  the  tissues’  stress-strain curve displayed non-linear behavior. in the case of the antxr2 -/samples, the first cycle, shown in light blue, was the stiffest peaking at 0.02 ± 0.01 mpa. the samples became more compliant with subsequent cycles, the peaks dropping to 0.018 ± 0.008 and 0.016 ± 0.008 mpa, respectively. this was the case for both the loading and unloading portions. this is usually referred   to   as   “preconditioning”,   and   it   has   been seen in other soft tissues [8]. preconditioning, or softening, was not seen for the antxr2 +/+ samples (n=3) and all cycles peaked around 0.007 ± 0.002 mpa. figure 4b shows the averaged stress response for both antxr2 -/(n=3) and antxr2 +/+ samples (n=3) to three ramp-hold cycles. there is a trend toward higher peaks and equilibrium levels for the antxr2 -/samples. briefly, at 10% strain a peak stress of 0.035 ± 0.01 mpa and an equilibrium stress of 0.0164 ±0.003 mpa were found for the antxr2 -/ samples. for the antxr2 +/+ samples, peak stress response was 0.018 ± 0.002 mpa and equilibrium stress response was 0.012 ±0.003 mpa. at 15% strain, the antxr2 -/samples had a peak and equilibrium stress response of 0.051 ± 0.007 mpa and 0.029 ± 0.002 mpa, respectively. for the antxr2 +/+ samples, peak and equilibrium stress response were equal to 0.027 ± 0.003 mpa and 0.019 ± 0.006 mpa. lastly, at 20% strain, the antxr2 -/samples had a peak and equilibrium stress response of 0.078 ± 0.008 mpa and 0.046 ± 0.002 mpa. the antxr2 +/+ samples had peak and equilibrium stress response of 0.04 ±0.01 mpa and 0.03 ± 0.01 mpa. antxr2 -/samples had significantly higher peak stress responses at 15 and 20% strain   (p<0.05,   student’s   t-test). figure 4c shows a table with the peak and equilibrium stress responses seen in both the antxr2 -/and antxr2 +/+ samples. figure 4: a graph of averaged stress response for antxr2 / and antxr2 +/+ samples for load-unload tests, b graph of averaged stress response for antxr2 -/ and antxr2 +/+ samples for ramp-hold tests at 10, 15, and 20% strain, c table of the averaged stress peak and equilibrium values for antxr2 -/ and antxr2 +/+ samples for ramp-hold tests at 10, 15, and 20% strain. strai n leve l (%) avg stress peak antxr2 -/ (mpa) avg stress peak antxr2 +/+ (mpa) avg stress equilibrium level antxr2 -/(mpa) avg stress equilibrium level antxr2 +/+ (mpa) 10 0.035 ± 0.01 0.018 ± 0.002 0.0164 ±0.003 0.012 ±0.003 15 0.051 ± 0.007 0.027 ± 0.003 0.029 ± 0.002 0.019 ± 0.006 20 0.078 ± 0.008 0.04 ±0.01 0.046 ± 0.002 0.03 ± 0.01 c a b columbia undergraduate science journal open-access publication | cusj.columbia.edu spring 2013 | volume 7 columbia undergraduate science journal 15 discussion the purpose of this study was to quantify the biochemical and biomechanical properties of uterine tissue from normal (antxr2 +/+) and abnormal (antxr2 -/-) mouse models of pregnancy. uterine tracts from 3-4 month antxr2 -/mice were demonstrated to have altered biochemical composition and biomechanical properties compared to antxr2 +/+ mice. in addition, the differences in biomechanical properties were correlated to altered biochemistry. few studies have correlated specific biochemical variables to uterine mechanical properties. here, the accumulation of collagen, a load bearing protein, is correlated to the increase in the mechanical strength of the tissue. despite the increase in its tensile strength, the uterus of antxr2 -/mice cannot produce adequate contractions for birth. the altered uterus cannot contract, highlighting the importance of the   uterus’   structure-function properties. limitations were encountered and alterations were made during protocol development. here the motivations behind protocol alterations, the significance of the experimental results, and future endeavors are presented. biochemical the collagen content of these antxr2 -/ mice has been measured to be higher than their antxr2 +/+ counterparts through qualitative means, but here quantitative analysis in the form of percent of dry weight is presented [13]. averaging at 33 ± 7% per dry weight, antxr2 -/ uterine tracts were found to have twice as much collagen than antxr2 +/+ tracts (15 ± 5% per dry weight). previous work by zhao, et al. showed that mice contain around 8% of dry weight of collagen in their uterine tracts [16]. taking into consideration the difference between protocols, breed, age, and the estrous stage in which the mice were sacrificed, reaching 15% of collagen is acceptable. knowing   a   tissue’s   collagen   content   is   useful, but a more detailed analysis of the existing collagen network is needed. the extractability of the tissue measures the solubility collagen network is in a solvent, and is an improved measure of the mechanical quality of the uterine collagen. regions a and d were chosen for extractability analysis because it was easy to guarantee that samples would not overlap with each other. a lower extractability, as seen in region d of antxr2 -/(25 ± 5 %), signifies a stronger network than region d of antxr2 +/+ (45 ± 4%). it can be inferred that the accumulation of collagen is responsible for extractability differences seen between region d of antxr2 -/and antxr2 +/+. extractability was found to be region specific in antxr2 -/uterine tissue. regions d, closest to the ovaries, were found to be less extractable than region a of antxr2 -/-. there are a few reasons why samples from region a of antxr2 -/mice (36 ± 13%) were not significantly less extractable than samples from region a of antxr2 +/+ mice (43 ± 5%). it was noticed that the relatively large variability in region a of antxr2 -/was caused by one specific sample. using   grubb’s   test   showed   that   the   individual   value was close to being an outlier (p=0.09). if this individual value is excluded from the samples, region a from antxr2 -/is found to be significantly less extractable than region a from antxr2 +/+ (p=0.012). note that the particular sample is included in figure 3c. another explanation for the higher extractability value of region a from antxr2 -/tracts could be due to inclusion of the junction. the junction, or the area where the two uterine horns meet, could have inherently different extractability properties. a final explanation could be the close proximity of region a to the cervix. the collagen fiber architecture of the cervix is different than the uterus’.  the  ultrastructure  of   region a could be a combination between the upper regions of the uterus and the cervix, resulting in these extractability values. biomechanical for the load-unload test, antxr2 -/ samples had stiffer (higher averaged stress response) than antxr2 +/+ samples. in addition, antxr2 +/+ tissues displayed no difference in columbia undergraduate science journal open-access publication | cusj.columbia.edu spring 2013 | volume 7 columbia undergraduate science journal 16 their averaged peak stress response between the three cycles (0.007 mpa all three), meaning pre-conditioning was not observed. it should be noted that antxr2 +/+ tissues tend to be very compliant, and at 5% strain levels, the difference between cycles could be below our instrument’s  sensitivity. for the ramp-hold test, antxr2 +/+ had lower peak and equilibrium stress response than antxr2 -/-, meaning antxr2 +/+ uterine tissue is more compliant. the 45 minutes in the ramp-hold test was not enough time for the tissues to equilibrate and reach their steady value. in initial tests of antxr2 +/+ uterine tissues,   “drifting”   was   seen   during   the   45   minutes following the 10 and 15% strain levels. this led to a seemingly increasing stress reading during resting periods instead of reaching a steady value. this was not seen at 20% strain. however, limiting the initial gauge length of the samples to 4 cm eliminated the drifting. despite the drifting, the antxr2 +/+ samples never reached the stress values for antxr2 -/samples. future work biochemical in conclusion, uterine tracts from antxr2 -/mice are stiffer than antxr2 +/+ tracts due to the accumulation of collagen. the change   seen   in   the   uterus’   structure-function properties leads to failed pregnancies. having biochemical data on these non-pregnant tissues has laid the foundation for future work, and will be essential when comparing to pregnant tissues. it is known that the uterus increases in wet weight, elastin, and collagen content during pregnancy [2]. this transformation has not been proven in antxr2 -/mice. future work will analyze the biochemical makeup of pregnant uterine tissue from both antxr2 -/and antxr2 +/+ mice. in the future, the extractability of regions b and c will be investigated in order to get a complete understanding of the uterine tracts. in addition, the sample preparation protocol will no longer include the junction since it is suspected to affect the extractability of region a. avery, et al. describes a more in-depth analysis of the collagen network [1]. briefly, identifying and determining the concentration of the different types of collagen cross-links provides insight into collagen turnover. future work will focus on analyzing collagen cross-links using the protocol described by avery, et al [1]. biomechanical the 45 minutes in the ramp-hold test was not sufficient for the tissues to equilibrate and reach their steady value. consequently, testing times in the future will be increased. weiss, et al. revealed that the uterus has zones of anisotropy [15]. a layer immediately surrounding the uterine cavity has collagen fibers oriented in a circular fashion [[15]. outside this layer there are no easily identifiable holistic structures. mechanical testing on the uterine tracts was performed in a longitudinal orientation. due to the anisotropy found in both human and mouse uterine tissue, future mechanical tests will be performed in a different orientation. uterine samples were airbrushed to produce a speckled pattern. using correlated solution vic 2-d dic software, the anisotropy of the uterus will be investigated. vic 2-d software will find local strain values and material properties that will be used in future finite element models to investigate the interaction between the uterus and the cervix of the mouse in gestation. references 1. avery, et al (2009). quantitative determination of collagen cross-links. methods in molecular biology (clifton, n.j.) 522: 103-121. 2. conrad, et al. (1966) passive stretch relationships in human uterine muscle 3. conrad, et al. (1966) stress relaxation in human uterine muscle. american journal of obstetric and gynecology 95.2: 254-265. columbia undergraduate science journal open-access publication | cusj.columbia.edu spring 2013 | volume 7 columbia undergraduate science journal 17 4. fata, et al. (2000) cellular turnover and extracellular matrix remodeling in female reproductive tissues: functions of metalloproteinases and their inhibitors. cellular and molecular life sciences, cmls 57.1: 77-95. 5. lawn, et al. (2012) save the children, who. born too soon: the global action report on preterm birth. geneva. 6. mahendroo, m. (2012) cervical remodeling in term and preterm birth: insights from an animal model. 7. mikic, b. (2000) gdf-5 deficiency in mice alters the ultrastructure, mechanical properties and composition of the achilles tendon. journal of orthopaedic research 19.3: 365-371. 8. myers, et al (2008) mechanical and biochemical properties of human cervical tissue, acta biomaterialia 4.1: 104-116. 9. obara, et al. (2001) changes in molecular weight of hyaluronan and hyaluronidase a activity in uterine cervical mucus in cervical ripening. acta obstetricia et gynecologica scandinavica 80.6: 492-496. 10. oxlund, et al. (2010) cervical collagen and biomechanical strength in non-pregnant women with a history of cervical insufficiency. reproductive biology and endocrinology 8.1: 92-102. 11. pearsall, g.w. and roberts, v.l. (1978) passive mechanical properties of uterine muscle (myometrium) tested in vitro. journal of biomechanics 11.4: 167-176. 12. peters de. capillary morphogenesis protein2 is required for mouse parturition by maintaining uterine collagen homeostasis. biochemical and biophysical research communications. 2012-06; 422:393397. 13. reeves et al (2011) anthrax toxin receptor 2 functions in ecm homeostasis of the murine reproductive tract and promotes mmp activity, in press at plos one. 14. spiess k. distribution of collagen types i, iii, and v in pregnant mouse endometrium. connective tissue research. 2007-01;48:99108. 15. weiss, et al. (2006) three-dimensional fiber architecture of the nonpregnant human uterus determined ex vivo using magnetic resonance diffusion tensor imaging. the anatomical record. part a, discoveries in molecular, cellular, and evolutionary biology 288a.1: 84-90 16. zhao l. collagen studies in late pregnant relaxin null mice. biology of reproduction. 2000.9;63:697-703. tive disorders and the prevention of stroke (chiaradonna, 2012; huang et al., 2011; dhawan et al., 2011). currently, there are only two known glutaminase inhibitors: bis-2-(5-phenylacetamido-1,2,4-thiadiazol-2-yl)ethylsul!de (bptes) and 6-diazo-5-oxo-l-norleucine (don) (hartwick, 2011). however, neither compound crosses the blood-brain barrier nor is neuroactive; thus, neither is suitable for pharmacotherapy of schizophrenia. "e purpose of this study was to screen for pag inhibitors in a library of 58,000 neuroactive compounds. robotic processing with a #uorescence-based assay was used to assess inhibition of pag activity in vitro. a secondary assay was used to con!rm pag-speci!city of apparent inhibitors, and subsequent kinetic analysis of the top tier of these hits was used to select two leads for further chemical optimization. primary screen a library of 58,000 neuroactive compounds was purchased from chembridge. compounds were supplied at a concentration of 10-3 m in dmso and were stored in 384well plates at -80oc. human pag was synthesized and puri!ed by our lab using a plasmid provided by professor norman curthoys (colorado state university) as previously described in kenny et al. (2003). "e #uorescence assay used for the screen was optimized from a commercially available kit (amplex red, invitrogen) and was used to measure pag activity in the presence of each library compound. glutamate, the product of pag catalysis, was converted to a #uorescent compound called resoru!n by the addition of a series of kit enzymes, including glutamate oxidase (go), horseradish peroxidase (hrp), and the #uorogenic substrate amplex red (figure 1). because the #uorescent molecule resoru!n is dependent on glutamate production and glutamate production is the product of pag catalysis, the rate of change in #uorescence is a measurement of pag activity. potent inhibitors of pag would signi!cantly decrease the #uorescence compared to controls without inhibitor. "e assay was carried out in a 384-well plate, with 80 µl of total liquid volume per well. 320 wells were carried out under experimental conditions, where 20 µl of amplex red solution (itself containing 8 ml of tris-hcl reaction bu$er, 60 µl of go, 25 µl of hrp, and 80 µl of amplex red reagent), 40 µl of pag solution (itself 2 µl of synthesized pag in 32 ml of tris-phosphate bu$er), 0.8 µl of one library compound (originally 10-3 m, but diluted to 10-4 m in well), and 20 µl of 40 µm glutamine solution were pipetted into each well. while kit enzyme concentrations were the same as those used in the amplex red kit protocol (so that all kit enzymes were in excess), optimum glutamine and pag concentrations were determined according to michaelis-menten kinetics. "e concentration of glutamine was selected based on the previously determined value of the km of pag-catalyzed glutamine deamination. km is de!ned as the concentration of substrate (glutamine) required to reach one half of the maximal reaction velocity (vmax) for a particular enzyme-catalyzed reaction (copeland, 2000). control conditions were included in the remaining 64 wells of each plate. positive control wells, representing 100% pag activity (0% pag inhibition), contained pag, glutamine, and amplex red solution. negative control wells, representing 0% pag activity (100% inhibition), contained glutamine and amplex red solution. 75%, 50%, and 25% inhibition controls were also used, obtained by creating a dilution series of the positive control. additionally, amplex red solution only controls were used as an additional negative control. in control wells, where necessary, tris-hcl reaction bu$er was added to bring wells to the standard volume. outlined in figure 2 below is a typical assay plate setup. fluorescence measurements were taken at baseline (immediately after glutamine was added to each well), and after a 60-minute incubation at 37 oc, by an envision #uorescence reader (perkin elmer). pipetting was done by a minitrak robotic liquid handling system (perkin elmer) for speed and e%ciency. to calculate percent inhibition of pag, go, and hrp, the average negative control #uorescence was subtracted from the experimental group #uorescence and divided by the average positive control #uorescence. subtracting percent activity from 100% gave percent inhibition. in the brain, phosphate-activated glutaminase catalyzes the recycling of glutamine back to glutamate for excitatory neurotransmission. in mice, genetic knockdown of phosphate-activated glutaminase has been shown to confer resilience to schizophrenia-like symptoms, suggesting that inhibition of glutaminase may have therapeutic potential for the pharmacotherapy of schizophrenia. as there are no known neuroactive inhibitors of glutaminase, i.e. inhibitors that get into the brain, high-throughput screening of a library of 58,000 qhxurdwlyh�frpsrxqgv�zdv�frqgxfwhg�xvlqj�d�àxruhvfhqfh�edvhg�dvvd\��wklv�vfuhhq�lghqwl¿hg�����srwhqwldo�joxwdplqdvh�lqklelwruv�� $�vhfrqgdu\�vfuhhq�zdv�fduulhg�rxw�wr�dvvhvv�vshfl¿flw\��wklv�\lhoghg����klwv���8vlqj�d�nlqhwlf�dqdo\vlv��wzr�ohdgv�zlwk�orz�plfurprodu� activity were found. columbia undergraduate science journal open-access publication | http://cusj.columbia.edu 3 cusjvolume 6spring 2012 columbia undergraduate sci j http://cusj.columbia.edu p s y c h ia tr y cusjcolumbia undergraduate science journal research articles 4cusj volume 6spring 2012 columbia undergraduate sci j http://cusj.columbia.edu p s y c h iatr y schizophrenia is a complex and diverse neuropsychiatric disorder that a$ects approximately one percent of the world population. symptoms typically present in schizophrenia are placed into three categories: positive, such as hallucinations, negative, such as paucity of speech and social withdrawal, and cognitive, such as conceptual disorganization and de!cits in working memory. "e treatment of schizophrenia has focused on targeting dopamine hyperactivity with anti-psychotic drugs. "e dopamine hypothesis, which attributes the symptoms of schizophrenia to hyperactive dopaminergic signal transduction in the striatum and, more broadly, the mesolimbic pathway, rose to prominence with the discovery that many common antipsychotic drugs, such as chlorpromazine and clozapine, work by binding to postsynaptic dopamine receptors, notably d2 receptors, and interfering with dopamine signal transduction (creese, 1976; seeman, 1975; van rossum, 1966; matthyse, 1973). however, the dopamine hypothesis does not fully explain schizophrenia. one problem is that these dopamine receptor antagonists are largely ine$ective in treating the negative and cognitive symptoms of schizophrenia, which suggests that dopaminergic hyperactivity may be unrelated to this class of symptoms (davis, 1991; javitt, 2010; kellendonk, 2006). moreover, a signi!cant number of patients with schizophrenia do not respond to these drugs, which also may be due to the involvement of other neurochemical pathways (davis 1991). more recently, abnormal glutamate transmission has been implicated in the pathophysiology of schizophrenia. glutamate is the major excitatory neurotransmitter in the central nervous system. under normal conditions it is produced and maintained at a low extracellular concentration through the glutamate-glutamine cycle (sanacora, 2008). "e importance of glutamate transmission in schizophrenia and psychosis was realized with the discovery that drugs like phencyclidine (pcp) cause symptoms that are nearly indistinguishable from those of schizophrenia by acting as an antagonist to n-methyl d-aspartate (nmda) receptors to which glutamate is an agonist. since then, many studies have shown that processes that modulate normal glutamatergic signaling may have bene!cial e$ects on positive, negative, and cognitive symptoms of schizophrenia (javitt 2010; vinson and conn 2011; gaisler-solomon et al., 2009). in particular, mice haploinsu%cient for phosphate-activated glutaminase (pag) show a reduction in glutamatergic synaptic transmission, are behaviorally quite normal, and are resilient to the e$ects of pro-psychotic drugs. pag is an enzyme that catalyzes the presynaptic conversion of glutamine to glutamate in neuronal mitochondria, "ese !ndings suggest that the reduction of glutamate release through the genetic reduction of pag confers resilience to a schizophrenia-like phenotype in mice, and that the inhibition of pag may have therapeutic potential in the pharmacotherapy of schizophrenia (gaisler-solomon et al., 2009). pag inhibition also appears to have therapeutic potential in several diseases beyond schizophrenia, such as in the treatment of cancer, and hiv-associated neurocogniidenti!cation of inhibitors of phosphate-activated glutaminase for the pharmacotherapy of schizophrenia cristina sorrento 1,3,* , stephen rayport 1,3 , andra mihali 1,3 , genevieve kaunitz 1,3 , shreya subramani 1,3 , joanne macdonald 2 , francine katz 2 departments of psychiatry1 and medicine2, columbia university, new york, ny 10032 departments of molecular therapeutics3, new york state psychiatric insitute, new york, ny 10032 copyright: © 2012 the trustees of columbia university, columbia university libraries, some rights reserved, sorrento, et al. received 1/1/2012. accepted 2/1/2012. published 1/1/12 *to whom correspondence should be addressed: departments of psychiatry and medicine, columbia university, new york, ny 10032 cristinasorrento75@gmail.com abstract introduction figure 1 7kh�vwhsv�ri�wkh�àxruhvfhqfh�edvhg�dvvd\�iru�vwxg\lqj�wkh�dfwlylw\� ri�skrvskdwh�dfwlydwhg�joxwdplqdvh��3$*����5hvrux¿q�àxruhvcence is a measure of glutamate concentration and, therefore, 3$*�dfwlylw\��0rgl¿hg�iurp�0f(our\�hw�do���������� methods varying concentrations. additionally, !uorescence readings were taken every 10 minutes so that reaction rates could be calculated based on the slope of the graph of !uorescence versus time. graphpad prism 4 was used to transform the !uorescence data to "t a characteristic michaelis-menten curve and to calculate the ic50 value for the inhibitors. ic50 is a measure of inhibitory power; it is the concentration of inhibitor at which the reaction rate catalyzed by a particular enzyme is halved. #e steps in the screening of the chembridge cnsset library are detailed in figure 3. a visual basic loop program was used to rank in descending order and in terms of percent inhibition the 320 library compounds on each of the 180 plates. because of variability in !uorescence values of the controls in di$erent plates, results of the primary screen were sorted based on their dynamic range. #e dr of !uorescence for a particular plate was de"ned using the control !uorescence values: dr = 100% * (positive control negative control) (positive control) plates were then placed into three categories: good secondary screen to select true inhibitors of pag, it was necessary to eliminate the hits that were inhibitors of the kit enzymes go and hrp. to eliminate go/hrp inhibitors, another resoru"n assay was designed for this purpose. #is assay was similar to the original glutamine-glutamate amplex red assay except that glutamine was replaced by the go/ hrp substrate glutamic acid, and pag was not included. because the conversion of glutamate to resoru"n is much faster than pag-catalyzed glutamine deamination, it was necessary to determine optimal substrate and enzyme concentrations that would produce a linear increase in product over a given time frame. based on a preliminary kinetics assay, it was determined that replacing the 40 µm glutamine with 10 µm glutamic acid and maintaining previous kit enzyme concentrations produced a reaction where the concentration of product (resoru"n) increased linearly with time (and did not saturate). once the glutamic acid assay parameters were determined, the primary screen assay and glutamic acid assay were carried out in parallel on the primary screen hits. percent inhibition of pag, go, and hrp (primary screen) and of go and hrp (glutamaic acid assay) were determined for each hit, and percent inhibition of pag only was calculated by subtracting go/hrp inhibition from pag/ go/hrp inhibition. ‘hit’ kinetics analysis of the compounds that were picked up in the secondary screen, the ten with the highest percent pag inhibition were selected for kinetic analysis and purchased from chembridge. 0.1 m aliquots of each compound were prepared with dmso and stored at -80oc. from these stock solutions, 10-fold dilution series ranging from 10-3 m to 10-8 m were prepared for each compound. #e same assay that was used in the primary screen was used for the kinetics analysis experiment, except that the inhibitors were used in columbia undergraduate science journal open-access publication | http://cusj.columbia.edu 5 cusjvolume 6spring 2012 columbia undergraduate sci j http://cusj.columbia.edu p s y c h ia tr y cusjcolumbia undergraduate science journal research articles 6cusj volume 6spring 2012 columbia undergraduate http://cusj.columbia.edu figure 2 a typical plate from the primary screen. figure 4 reaction rate vs. concentration plotted using prism software. ic��� values are displayed in the table to the right figure 3 $�vfkhpdwlf�gldjudp�rxwolqlqj�wkh�vfuhhqlqj�surfhvv�iru�wkh�vhohfwlrq�ri�3$*�vshfl¿f�lqklelwruv p s y c h iatr y trial 1 2 3 4 average ± se ic50 (um) 5.482 3.647 4.655 3.303 4.272 ± 0.572 r2 0.951 0.982 0.944 0.956 (dr=95-50%), low or extreme (dr<50% or dr>95%), and poor dynamic range (dr!0%). for plates with a good dynamic range, compounds that exhibited at least 90% inhibition were selected. for plates with a low or extreme dynamic range or non-existent dynamic range, compounds were selected if they stood out in the plate distribution. representative distributions of compound inhibition data for three plates in each of the three categories are shown in figure 4. for plates with a good dynamic range, there were typically very few or no compounds with percent inhibition above 90%; however, for plates with a low or extreme dynamic range or a poor dynamic range, there were signi"cantly more compounds whose percent inhibition was greater than 90%, and in fact, there were a signi"cant number of compounds with percent inhibition greater than 100%. #e hits picked up by this process were tested in the secondary screen to eliminate inhibitors of go and hrp. #e compounds that had the least impact on go and hrp were ranked based on their selective inhibition of pag. from this list of compounds, the top 10 were selected for kinetic analysis. for 2 of these 10 compounds, the inhibition curve "t michaelis-menten kinetics. plots for these 2 compounds, 3d2 and 8o, of reaction velocity versus inhibitor concentration are shown below, and figure 5 indicates the ic50 values 3d2 in four independent experiments. #e average ic50 value for 3d2 was 4.272 ± 0.572 µm. as indicated by the regression coe$cients of each experiment, the michaelis-menten model of enzyme kinetics was applicable to the activity of 3d2. based on its average ic50 value, inhibitor 3d2 appears worthy of further investigation. because combinatorial chemistry can be used to optimize inhibitory power by up to three orders of magnitude, it appears that optimization of 3d2 particularly would produce a compound that would inhibit pag in the nanomolar range (jorgensen, 2009). at concentrations lower than 10-4 m, 8o also exhibits similar dose-dependent behavior, but because of its unusual behavior for concentrations greater than this, where reaction velocity appears to increase with greater concentrations of inhibitor, it is necessary to subject 8o to further kinetic scrutiny. for both compounds, before they are to be optimized, it is necessary to con"rm their inhibitory power through alternative assays and to increase the number of hits to be optimized. #e organization of the primary screen experiments into three categories based on the dynamic range of control %uorescence enabled comparison between plates and appropriate selection of hits. for plates with a good dynamic range, there were few or no compounds with greater than 90% inhibition. however, for plates with an extreme (low or high) or non-existent dynamic range, there were many compounds with greater than 90% inhibition and even some well above 100%. #is result most likely does not indicate a greater number of hits for these two categories of plates; indeed, the >100% inhibitors suggests that the large number of apparent hits with percent inhibition >90% is an artifact of an abnormal dynamic range. #erefore, the division of the plates into the three categories and the selection of outliers from the two categories of plates with abnormal values for dynamic range were used to correct this artifact so that only outliers were selected from every plate. #e secondary screen was designed to eliminate nonspeci"c inhibitors and false positives from the screen. in testing for inhibition of glutamate oxidase and horseradish peroxidase, the secondary screen was able to exclude compounds that were potential inhibitors of these kit enzymes. in addition to this secondary assay, it is necessary to con"rm the inhibitory power of 3d2 using another assay without the use of amplex red; one such assay involves the direct determination of glutamate as a product of pag catalysis. from the ic50 curve of 8o, it appears that for concentrations of inhibitor greater than 10-4 m, the reaction rate is actually higher than it is for lower concentrations; otherwise, for concentrations below 10-4 m, 8o appears to adhere to the dose-dependent pattern normally seen with good inhibitors. a likely explanation for this unusual pattern is that 8o is itself %uorescent and enhances the %uorescence signal at high concentrations. one caveat to michalis-menten kinetics determinations is that the value of the ic50 measurement is in itself limited; ic50 depends on the concentration of substrate, enzyme, and inhibitor, as well as other experimental conditions. ki, the dissociation constant of the inhibitor to the enzyme, is a more valuable measurement because it is an intrinsic thermodynamic quantity independent of substrate concentration, but dependent on enzyme and inhibitor concentrations. for this reason, ki measurements are more replicable across di&erent experimental methods (cer, 2009). #ere are ways of measuring ki based on ic50 values, one columbia undergraduate science journal open-access publication | http://cusj.columbia.edu 7 cusjvolume 6spring 2012 columbia undergraduate sci j http://cusj.columbia.edu p s y c h ia tr y figure 5 ic���ydoxhv�ghwhuplqhg�iru��'��zlwk�fruuhvsrqglqj�5 � values. �2�zdv�qrw�whvwhg�rq����������� of which is the cheng-pruso& equation: ki= ic50/(1+[s]/km ) however, this equation is used to calculate ki for inhibitors known to be competitive (copeland, 2000). another method that can be used to determine ki is through the use of the lineweaver-burk plot, which graphpad prism 4 can generate. #ese plots are obtained by varying both inhibitor and enzyme concentration, and can both elucidate the nature of the inhibitor (competitive, noncompetitive, or uncompetitive) and be used to calculate ki (copeland, 2000). knowing the nature of the inhibitor will allow prioritization of the leads for optimization. noncompetitive inhibitors will be favored for optimization over competitive inhibitors. competitive inhibitors, binding to the enzyme’s active site, are less likely to be speci"c to pag, as they are also likely to bind to the active sites of other enzymes in the glutamate cycle. #erefore, since pagspeci"city is one of the selection criteria, and competitive inhibitors are unlikely to be pag-speci"c, non-competitive inhibitors will be preferentially selected for optimization. once these leads are optimized, they will be assessed for drug-like character and their suitability in the pharmacotherapy of schizophrenia. for example, pag activity measurements in mice administerd the leads, electrophysiological recordings of their brain slices, and behavioral tests analyzing their response to amphetamine-induced hyperactivity can be used to assess the value of these leads in pharmacotherapy. additionally, adme-tox (absorbtion, distribution, metabolism, excretion, and toxicity) analysis will be necessary in assessing suitability for drug development. cs thanks am and gk for their guidance and encouragement and sr for his mentorship, as well as all the collaborators for their support and advice. cer, r. z., mudunuri, u., stephens, r., and lebeda, f. j. (2009) nucleicacids res. 37, w441–w445 copeland, robert allen. enzymes: a practical introduction to structure, mechanism, and data analysis. new york: wiley, 2000. chiaradonna, f., r.m. moresco, c. airoldi, d. gaglio, r. palorini, f. nicotra, c. messa, and l. alberghina. “from cancer metabolism to new biomarkers and drug targets.” biotechnology advances 30.1 (2012): 30-51. creese, i., d. burt, and s. snyder. “dopamine receptor binding predicts clinical and pharmacological potencies of antischizophrenic drugs.” science 192.4238 (1976): 481-83. davis, kenneth l., rene s. kahn, grant ko, and michael davidson. “dopamine in schizophrenia: a review and reconceptualization.” am j psychiatry 148.11 (1991): 1474-486. dhawan, j. “a new look at glutamate and ischemia: nmda agonist improves long-term functional outcome in a rat model of stroke.” future neurol. 6.6 (2011): 823-34 gaisler-salomon, inna, gretchen m. miller, nao chuhma, sooyeon lee, hong zhang, farhad ghoddoussi, nicole lewandowski, stephen fairhurst, yvonne wang, agnès conjard-duplany, justine masson, peter balsam, rené hen, ottavio arancio, matthew p. galloway, holly m. moore, scott a. small, and stephen rayport. “glutaminase-de"cient mice display hippocampal hypoactivity, insensitivity to pro-psychotic drugs and potentiated latent inhibition: relevance to schizophrenia.” neuropsychopharmacology 34.10 (2009): 2305-322. hartwick, e. w., and n. p. curthoys. “bptes inhibition of hga(124-551), a truncated form of human kidneytype glutaminase.” j enzyme inhib med chem (2011). huang, y., l. zhao, b. jia, l. wu, y. li, n. curthoys, and j. c. zheng. “glutaminase dysregulation in hiv-1-infected human microglia mediates neurotoxicity: relevant to hiv1-associated neurocognitive disorders.” journal of neuroscience 31.42 (2011): 15195-5204. javitt, daniel c. “glutamatergic #eories of schizophrenia.” isr j psychiatry relat sci 47.1 (2010): 4-16. jorgensen wl. acc chem res. 2009;42:724–733. kellendonk c, simpson eh, polan hj, malleret g, vronskaya s, winiger v, moore h, kandel er. transient and selective overexpression of dopamine d2 receptors in the striatum causes persistent abnormalities in prefrontal cortex functioning. neuron. 2006;49:603–615. matthysse s. antipsychotic drug actions: a clue to the neuropathology of schizophrenia.federation proc. 1973;32:200–208. mcelroy, k. “implementation of a continuous, enzyme-coupled fluorescence assay for high-#roughput analysis of glutamate-producing enzymes.” analytical biochemistry 284.2 (2000): 382-87. sanacora, gerard, carlos a. zarate, john h. krystal, and husseini k. manji. “targeting the glutamatergic system to develop novel, improved #erapeutics for mood disorders.” nature reviews drug discovery 7.5 (2008): 426-37. cusjcolumbia undergraduate science journal research articles 8cusj volume 6spring 2012 columbia undergraduate sci j http://cusj.columbia.edu p s y c h iatr y discussion acknowledgments references 16 identification of lncrnas as therapeutic targets in chronic lymphocytic leukemia simay dolaner1,2, harpreet kaur2, mohit mazumder2, julia panov2,3, elia brodsky2 1department of molecular biology and genetics, bahcesehir university, istanbul, turkey 2pine biotech, new orleans, usa 3tauber bioinformatics research center, haifa, israel abstract: chronic lymphocytic leukemia (cll) is a type of blood cancer that has a very heterogeneous biological background and diverse treatment strategies. however, a small part of this malignancy may disappear without receiving any treatment, known as “spontaneous regression”, which occurs as a result of a poorly investigated mechanism. exposing the underlying causes of this condition can lead to a novel treatment approach for cll. in this article, we applied in-silico analysis on total rna expression data from 24 cll samples to determine possible regulatory mechanisms of spontaneous regression in cll. these were first selected by comparing spontaneous regression with progressive samples of cll at the transcriptional level using two unsupervised machine learning algorithms, i.e., principal component analysis (pca) and hierarchical clustering. subsequently, the deseq2 algorithm was used to scrutinize only statistically significant (adjusted p-value < 0.01) rna transcripts that can differentiate both conditions. here, at first, we have elucidated 870 significantly differentially expressed protein-coding genes that were involved in the biogenesis and processing of rna. consequently, these findings led our study to investigate non-coding rna, and 33 long non-coding rnas (lncrnas) were found to be significantly differentially expressed among two conditions based on differential gene expression analysis. further, our analysis in the current study suggested lncrnas, ptpn22-as1, pcf11-as1, syngap1-as1, prrt3-as1, and h1fx-as1 as potential therapeutic targets to trigger spontaneous regression. ultimately, the results presented here reveal new insights into spontaneous regression and its relationship with non-coding rnas, particularly lncrnas. keywords: chronic lymphocytic leukemia, spontaneous regression, therapeutic targets, lncrnas, transcriptomics © 2021 dolaner, kaur, mazumder, panov, brodsky. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. introduction according to the american cancer society statistics, leukemia is the second leading blood cancer, with roughly 60,000 new cases identified in 2020 [1]. leukemia has different subtypes, which are classified in the context of the tumor origin [2]. the most common variety in adults is chronic lymphocytic leukemia (cll), which is a lymphoid malignancy due to failed apoptosis and aggressive proliferation of mature b cells [3]. these cells circulate through the blood as non-proliferating cells or arrested cells in the g0/g1 phase of the cell cycle and may affect the function of normal cells in other organs [4]. columbia undergraduate science journal vol. 15, 2021 dolaner et. al. 17 cll has a highly diverse biological and clinical background for each patient that determines the stage of the disease [5]. although there are several stages of cll classified according to their genetic background and b cell number, one of the most intriguing concepts is known as spontaneous regression, which is the disappearance of the tumor over time either without any treatment or with treatment that is categorized as insufficient to have an impact on the tumor [6]. spontaneous regression can be seen in 1-2% of all cll patients, and it is a phenomenon that is poorly understood [7]. in this process, cells that proliferate uncontrollably are transmitted to the quiescent state so that the tumor disappears partially or completely with time [6]. spontaneous regression is not a common feature of cancer cells and is regulated by mechanisms that are not well-understood. if such mechanisms can be determined, target biological molecules that have a specific role in disease progression can be identified and manipulated in vitro. current strategies aim to inhibit bcr signaling, which is crucial for the survival of the b-cells, and chemokine signaling that creates survival signals and attracts leukemic cells to communicate with its microenvironment [5]. moreover, activation of apoptosis pathways via blocking bcl2 activity, an anti-apoptotic protein, which is highly expressed in leukemic cells, is included in the aforementioned strategies [8]. although these types of targeted therapies improve the outcome of the patients, the heterogeneity of the leukemia microenvironment reinforces the necessity of new targets. as the targeted therapies may have an impact on tumor surroundings and affect the other cells found in the tumor microenvironment, triggering spontaneous regression mechanisms may improve the strategies as well as patient outcome. in cll, the most frequent chromosomal abnormalities and somatic mutations on the protein-coding region of the genome have been distinguished as a result of genomic sequencing, and disrupted cellular pathways are identified using next-generation sequencing of mrna expression [9]. in spite of this progress, nearly 20% of cll cells do not show chromosomal abnormalities or genomic variation. therefore, researchers recently shifted their focus to the non-coding region of the genome and regulatory rna molecules, especially long non-coding rnas, which are deregulated in many cancers [10]. long non-coding rnas (lncrnas) are a subgroup of non-coding rnas which are longer than 200 nucleotides and encompass thousands of diverse transcripts in humans [11]. there are approximately 100,000 known lncrnas, and this quantity is expanding each year with the new studies [12]. lncrnas play a significant role in gene regulation, controlling multiple cellular mechanisms involved in tumor progression. they are involved in epigenetic regulation of gene expression via histone modification, dna methylation, or acetylation. specifically, these epigenetic regulations may include recruitment of histone remodeling complexes, interaction with histone methyltransferases and demethylases to regulate dna methylation, or histone acetyltransferases and deacetylases to modulate the acetylation [13]. furthermore, lncrnas may regulate gene expression at the transcriptional level by recruiting transcription factors [14]. moreover, lncrnas can produce hybrids or act as scaffolds through interaction with proteins to regulate expression at the post-translational level, including regulation of phosphorylation and ubiquitination [15]. in tumor development, lncrnas can serve as either tumor suppressors, oncogenes, or even both at the same time for some cancer types [16]. analysis of lncrna expression patterns had led to the identification of putative biomarkers such as hotair, h19, and dleu1/2 [17]. specifically, hotair serves as an oncogene by inducing invasiveness and metastasis in several cancers via recruiting a demethylase [13]. since this lncrna is highly expressed in aggressive tumors, it is considered as a biocolumbia undergraduate science journal vol. 15, 2021 dolaner et. al. 18 marker and a possible therapeutic target for many cancers [18]. dleu1/2 epigenetically regulate the tumor suppressors and are deleted in cll cells, which results in the progression of the cll. this allows it to serve as a biomarker in the diagnosis [19]. on the other hand, h19 has a dual role in different cancers by stimulating distinct mechanisms through transcription factors, which makes it a target that needs to be studied separately for each cancer [13]. furthermore, lncrnas are known to play an important role in cell differentiation and tissue specificity [20]. however, characterization may be compelling because lncrnas are transcribed in different loci and localized distinctly. more importantly, lncrna expression is generally tissue-specific and can be detected under certain conditions [21]. as the lncrnas are differentially expressed, their roles and activities can be identified in disease conditions. due to the diverse functions of lncrnas, novel studies are concentrated on the identification of lncrnas as therapeutic targets. as the expression of these molecules is tissue and disease-specific, this specificity makes them excellent targets compared to protein-coding genes [22]. in the scope of this project, the trigger mechanism behind the spontaneous regression process is investigated at the transcriptomic level to identify a pattern of lncrna expression that would explain cell “decision” by comparing cll tissue at the spontaneous regression and the progressive states. methods dataset the dataset of this project was generated by kwok et al. (2020) and published as a bioproject on the ncbi with the accession number prjna535508 (supplementary notes 1) [6]. transcriptome data contains raw reads of rna sequencing from the illumina nextseq 550 platform by using paired-end sequencing. in this study, the authors compared multiple samples from multiple subtypes of chronic lymphocytic leukemia that can be seen in table i. in our project, spontaneous regression and progressive states were chosen for further analysis to investigate expression variation specifically involved in the regression mechanism in cll cells. table i. dataset of the bioproject rna-seq raw data processing raw reads were used to construct an rna sequencing pipeline that contains pre-processing using trimmomatic, mapping of reads on reference transcriptome with bowtie2-t, and quantification using rna-seq by expectation-maximization (rsem) algorithms with t-bioinfo server (supplementary notes 1) as represented in figure 1. briefly, in order to get gene expression levels, duplicated sequences which resulted in pcr amplification were removed by pcr clean considering best coverage. then, trimmomatic was used to remove adaptor sequences and poor-quality data at the end of the sequence. mapping of these clean reads on transcriptome was performed by using the bowtie2-t algorithm based on the reference genome (grch38). bowtie2-t is an option of the bowtie, which is a mapping algorithm and aligns short reads according to the seed approach [23]. finally, for the quantification of gene expression, the rsem algorithm was used with fpkm normalization to obtain the gene expression table. columbia undergraduate science journal vol. 15, 2021 dolaner et. al. 19 figure 1: rna-seq data processing pipeline to generate an rna expression table on t-bioinfo server exploratory analysis exploratory analysis facilitated the examination of variation between all samples, including healthy and diseased patients, in order to select a comparison parameter for further analysis. thus, the visual outputs were used to determine the patterns. data was explored by using principal component analysis (pca), which is a dimensionality reduction technique that discerns the variability between the samples [24]. pca was performed twice, both for all the samples and for the spontaneous regression and the progressive samples to be able to observe the improvement on principle components. hierarchical clustering, which finds patterns among the samples using similarity measures [25], made it possible to understand the clustering aspects of spontaneous regression and progressive samples based on their gene expression, especially after the selection of statistically significant genes, which will be explained in the following section. differential gene expression analysis the differential gene expression analysis was conducted by contrasting spontaneous regression with progressive cll samples. separate studies were performed for protein-coding and non-protein coding transcripts to understand the mechanisms involved in spontaneous regression. the differential gene expression (dge) pipeline, which includes pre-processing, mapping with hisat2, rna expression quantification using htseq, and differential gene expression analysis with deseq2 algorithm was constructed by using the t-bioinfo server (figure 2). here, initially, pcr cleaning was performed by considering coverage to get rid of the duplicated sequences generated via pcr amplification. to eliminate the adaptors and bad quality sequences from the data, the trimmomatic algorithm was used. next, the hisat2 was utilized for the mapping of sequence reads to the reference genome (grch38) by taking into account the splice junctions. then, the htseq algorithm quantified the gene expression through overlapping reads and generated a gene expression table in the form of count values. finally, differential gene expression analysis was performed by using the deseq2 algorithm which gives the expression differences between two groups using the shrinkage estimators [26]. deseq2 provides results, which include p-value, log 2-fold change, and adjusted p-value. then, an adjusted p-value or false discovery rate (fdr) that is a standard statistical value utilized for multiple testing correction, is calculated to eliminate the false-positive results [26, 27]. the t-bioinfo server uses a onestep approach for dge analysis and combines several methods. although deseq2 can be used for both normalization and statistical analysis, the t-bioinfo server provides an additional approach, which includes gene set enrichment analysis (gsea) to discover related pathways and processes [28]. selection of significant genes the significant genes were identified by their p-adjusted values and log2 fold change. the determined threshold for the adjusted p-value was 0.01 and the log2 fold change was ±1 in non-coding rnas and ±1.5 in protein-coding genes. columbia undergraduate science journal vol. 15, 2021 dolaner et. al. 20 figure 2: differential gene expression analysis pipeline using the t-bioinfo server gene ontology analysis the enrichment analysis was done via using the enrichr platform for the protein-coding genes [29]. respectively, the go and kegg pathways were considered for the observed upregulated and downregulated genes in spontaneous regression samples. data visualization the gene expression patterns were observed by heatmaps. furthermore, pca and h-clustering were repeated with the selected significant rna transcripts to make a comparison as before and after, and see an improvement on the basis of variance. visualization was performed independently for protein-coding genes and non-protein coding genes. results variation is detected between spontaneous regression and the progressive samples the exploratory analysis using the pca revealed that there was a variation between the spontaneous regression and the progressive state. based on figure 3a, it can be seen that healthy samples and the diseased samples are well separated, which means there is significant variation between these groups. when the spontaneous regression and progressive samples were examined in a specific scatter plot, it was obvious that there was an improvement in principal components, and the two groups were clearly separated from each other (figure 3b). after the observed variation between the two groups, the reason for this difference and its impacts could be investigated at the level of gene regulation. figure 3. principal component analysis, (a) healthy samples versus diseased samples (b) spontaneous regression (red) versus progressive samples (purple) revealed pathways involved in biogenesis and processing of rna based on deseq2 analysis, expression levels of 870 protein-coding genes were identified as significantly different between the two groups with p-adjusted values (<0.01) and log2 fold change (+/1.5) (supplementary table 1). the heatmap of the protein-coding genes shows the gene expression patterns between spontaneous regression and progressive samples (figure 4a). it is obvious that most of the genes were upregulated in spontaneous regression while they downregulated in progressive ones. to understand the biological importance of pathways, gene ontology analysis was performed. interestingly, although there were different pathways in upregulated genes, many represented biological pathways involved in biogenesis and processing of rna, and mrna translation (figure 4b). in addition, identified pathways also included ribosome-related pathways, which imply translation of protein-coding rnas is affected. the same concept could be detected with columbia undergraduate science journal vol. 15, 2021 dolaner et. al. 21 figure 4. data visualization of protein-coding genes (a) gene expression patterns of protein-coding genes. heatmap exposes the downregulated and upregulated genes among the samples (b) gene ontology analysis of protein-coding genes. pathway analysis shows the downregulated and upregulated pathways in the spontaneous regression samples, respectively. arrows indicate the rna related pathways (c) representation of principal component analysis with 3d scatter plots. first pca shows the separation of spontaneous regression (sr) and progressive (p) samples before the selection of the protein-coding genes, second pca represents the separation of sr and p samples after the selection of the protein-coding genes, and last pca indicates the true variability source considering four outliers (d) hierarchical clustering results as dendrograms. each dendrogram point out the related scatter plots that are placed above and reveal the clustering perspective. red boxes indicate the sr samples, and the other ones are the p samples a b c d columbia undergraduate science journal vol. 15, 2021 dolaner et. al. 22 the downregulated genes as presented in figure 4b. consequently, these findings led this research to investigate the non-coding rnas, which was the second branch of this study. there was a diversity among the spontaneous regression samples according to the first principal component, spontaneous regression had a lot of variability within its own group (figure 4c). this variation could be observed from the dendrograms in figure 4d. correspondingly, spontaneous regression samples had two diverse groups. the clinical data of these samples, therefore, needed to be examined to interpret them as outliers or true variability sources. however, no difference was detected between the four outliers and the other samples. this meant the difference had to be in their biological background. that is why these four diverse samples were selected in comparison with progressive samples to see the most significant difference between the two groups in the context of expression patterns. as a result, it can be seen in figure 4c that there was a significant improvement with the well-separated two groups after the selection of true variability source. additionally, better separation could also be observed from the dendrogram in figure 4d that four samples and the progressive samples have distinct branches compared to the remaining twelve samples. identified non-coding rnas were novel transcripts based on differential gene expression analysis, 33 lncrnas, which are significantly expressed, were identified with the specific parameters (supplementary table 2). each lncrna was investigated by their ensembl id and identified as novel transcripts. when the pca was repeated with selected lncrnas, clear separation could be seen among the spontaneous regression and the progressive samples (figure 5a). even though there was no dramatic improvement in principal components, hierarchical clustering results were highly distinctive. it can be seen in figure 5b that the two groups cluster separately in the dendrogram, which implies these significant genes might predict tumor characteristics. to be able to observe the gene expression pattern of the lncrnas, a heatmap was generated. as shown in figure 5c, most of the genes were upregulated in progressive samples compared to spontaneous regression. so far, pathway analysis of differentially expressed protein-coding genes in spontaneous regression samples has demonstrated that global gene expression in these samples might be modulated at the transcriptional and post-transcriptional level, and non-protein coding genes have shown that lncrnas are differentially expressed in these samples. to make a biologically relevant interpretation, the analysis will regard lncrnas identified as statistically most significant. discussion although spontaneous tumor regression is rare, it is a phenomenon that can be observed in some types of cancer, such as neuroblastoma, renal cell carcinoma, lung cancer, lymphoma, and leukemia [30]. to be able to increase the occurrence rate of this mechanism, detailed studies should be conducted, and the related pathways should be examined. in summary, we used rna-seq datasets from cll patients containing a total of 16 spontaneous regression and 8 progressive state samples. first of all, we looked for variation and pattern among these two groups by using pca and h-clustering based on gene expression. after the detection of variation, we hypothesized that the reason for this difference should be at the gene expression level. therefore, we applied differential gene expression analysis by comparing spontaneous regression and progressive tumor states. we identified 870 protein-coding genes which were significantly differentially expressed and also were associated with rna-related pathways based on gene ontology analysis. depending on these findings, we also investigated non-coding rnas and columbia undergraduate science journal vol. 15, 2021 dolaner et. al. 23 figure 5. data visualization of non-coding rnas (a) representation of principal component analysis with 3d scatter plots. initial pca indicates the separation of spontaneous regression (sr) and progressive (p) samples before the selection of the non-coding rnas and second pca shows the separation of sr and p samples after the selection of the non-coding rnas (b) hierarchical clustering results as dendrograms. each dendrogram specify the related scatter plots that are located above and expose the clustering perspective. red boxes demonstrate the sr samples, and the other ones are the p samples (c) gene expression patterns of non-coding rnas. heatmap represents the downregulated and upregulated genes between the samples a b c columbia undergraduate science journal vol. 15, 2021 dolaner et. al. 24 identified significantly expressed 33 lncrnas. this paper has highlighted that new therapeutic strategies may involve lncrnas to trigger the spontaneous regression phenomena in cancer cells. since the lncrnas can regulate the important cellular mechanisms during cancer progression, identification of disease-specific lncrnas may enlighten the way of new treatment strategies. therefore, a detailed literature review for each lncrna and their sense mrnas had been done. even though the detailed explanation can be found in supplementary table 2, identified lncrnas are involved in several pathways including tumor growth, metastasis, cell survival, and regulation of the tumor microenvironment such as the immune system. previous studies showed that identified two lncrnas, the prrt3-as1 and h1fx-as1, were focused on cancer progression [31, 32, 33, 34]. both lncrnas were upregulated in the progressive state compared to spontaneous regression, which indicates that these lncrnas were functioning as tumor enhancers in cll cells. li et al. (2020) showed that the lncrna prrt3-as1 is upregulated in prostate cancer and targets the pparγ gene by binding its 3’ end, which leads to regulation of the akt/mtor signaling pathway [31]. the same study also revealed that the down-regulation of prrt3-as1 inhibits prostate cancer progression by regulating cell proliferation, migration, and apoptosis. moreover, knocking down this lncrna triggers autophagy via the mtor pathway [31]. another study suggested that prrt3-as1 is estimated as an immune-related lncrna involved in ppar signaling and can be used as a potential target in glioblastoma [32]. in light of these findings, it can be said that the prrt3-as1 functions as a tumor promoter gene in prostate cancer and glioblastoma. since it is revealed that this lncrna is highly expressed in the progressive state of cll, further studies may clarify the related pathways and novel targets. moreover, downregulation of this lncrna might promote spontaneous regression by repressing cell proliferation and inducing apoptosis. according to shi et al. (2020), lncrna h1fx-as1 is downregulated in cervical cancer, and low expression is correlated with poor prognosis and linked to tumor size as well as metastasis. in silico analysis predicted that the possible target of the h1fx-as1 was the mir324-3p, and it was binding and regulating the dact1 [33]. furthermore, the same research included overexpression studies, which revealed that the high expression levels of this lncrna significantly reduced the proliferation and invasiveness, and activated the apoptosis pathways. h1fx-as1, therefore, is identified as a tumor suppressor gene in cervical cancer [33]. on the contrary, high expression levels of lncrna h1fx-as1 are associated with a poor prognosis in gastric cancer. it is predicted that this lncrna was related to epithelial to mesenchymal transition (emt) and metastasis pathways [34]. additionally, in-silico prediction analysis indicated several targets of the h1fx-as1 lncrna, such as h1fx, copg1, and mir6826 that can be used as potential therapeutic targets in gastric cancer [34]. since h1fx-as1 is upregulated in the progressive state of cll, the precise roles of this lncrna should be examined for cll cells. besides, considering the studies on gastric cancer, downregulation of h1fx-as1 may trigger the spontaneous regression through inactivation of metastasis. as the other 31 lncrnas were novel transcripts, several examples could be considered by examining the sense protein-coding genes of the lncrnas to reveal the significance and possible functions of these lncrnas. thereby, the specified lncrnas and their sense protein-coding genes can be targeted for further analysis and treatment strategies. the first one is the lncrna ptpn22-as1 that was upregulated in the spontaneous regression samples. it is known that the b cell receptor signaling is vital for the growth and survival of the leukemic cells [35]. these signaling pathways can be stimulated with diverse tyrosine kinases as well as phosphatases. ptpn22 is a protein tyrosine phosphacolumbia undergraduate science journal vol. 15, 2021 dolaner et. al. 25 tase specifically expressed in immune cells. it can function as an enhancer or suppressor of bcr and tcr signaling by regulating phosphorylation status [36]. according to studies, the ptpn22 gene is upregulated in cll cells, and this upregulation results in attenuation of the apoptosis signals produced by the bcr and stimulation of the akt activity that generates a survival signal [35]. this regulation reveals that the cancer cells which express autoreactive bcrs can escape from apoptosis by upregulating the ptpn22 gene. therefore, downregulation of this gene through the lncrna ptpn22-as1 at multiple levels may trigger the spontaneous regression in cll by eliminating autoreactive b cells in the context of immune tolerance. so, if the upregulation of this lncrna can be induced by an external factor, it could be used as a therapeutic target for cll cells. the next lncrna would be pcf11-as1, and it is also upregulated in spontaneous regression. alternative polyadenylation (apa) leads to transcription of diverse isoforms at the rna 3’ end that affects the functioning of encoded proteins. this mechanism needs multicomponent protein complexes, and one of the protein complexes is called cfii [37]. this complex comprises the pcf11 gene that is a cleavage and polyadenylation factor subunit, and regulates the transcription termination and rna 3’ end maturation. studies show that this gene particularly regulates the alternative polyadenylation of the genes associated with the wnt pathway as an oncogene in neuroblastoma cells [38]. accordingly, the downregulation of this gene results in diminished cell growth as well as invasiveness. interestingly, one of the studies indicates that spontaneously regressed neuroblastomas express the pcf11 gene at low levels compared to highly progressive neuroblastomas [39]. basically, downregulation of this gene at the epigenetic or post-transcriptional level through the lncrna pcf11-as1 may induce the spontaneous regression in cll via ceasing the cell growth. thus, the upregulation of this lncrna can be used as a therapeutic strategy. another lncrna is the syngap1-as1 which is upregulated in progressive samples. mutant ras genes stimulate the gtp-bound state which constitutively activates ras signaling in metastatic cells [40]. rasgaps inactivate ras signaling by converting active gtp-bound into its inactive state [41]. syngap1, also known as rasa5, is a member of the rasgap family and functions as a tumor suppressor gene [42, 43]. studies proved that the rasa5 gene is epigenetically disrupted in multiple cancer types by promoter methylation, and gain of function assays exposed that rasa5 expression leads to a reduction in metastasis by regulating emt and cell stemness [43]. as it is known that the lncrnas can regulate gene expression at the epigenetic level through methylation, this epigenetic silencing may be due to the lncrna syngap1-as1. so, the upregulation of the syngap1-as1 can knock down this gene at the epigenetic level by methylating the promoter region leading to aggressive tumor progression in cll patients. even though further studies are essential, targeting this lncrna may also trigger spontaneous regression through metastasis and cell stemness pathways, and enhance our understanding of this mechanism. as a result, the deduction that can be made from these exemplary lncrnas is that the detailed examination of the identified novel transcripts can be used to reveal the mechanisms that lead to spontaneous regression, as well as to identify these lncrnas as targets for small molecule inhibitors or activators so that the spontaneous regression mechanism can be triggered in other cancer types. conclusion these findings suggest that regulation through the lncrnas might have a major role in cells’ fate, and their detailed examination can enlighten the way of discovery of the possible therapeutic targets. in prospective studies, each lncrna should be investigated to perceive their functional pathways by over-expression and suppression studies in various cancer types columbia undergraduate science journal vol. 15, 2021 dolaner et. al. 26 in order to understand their specific roles. additionally, the following study which considers these findings should combine the protein-coding genes and non-protein coding genes. since these two groups revealed significant results independently, their combination can lead to future selection and determination of particular pathways which affected each other. this study had certain limitations that need to be overcome to conduct more efficient studies. spontaneous regression samples had diverse biological backgrounds, which increased the demand for the detailed investigation of these samples. furthermore, it was problematic to identify the significant pathways through differential gene expression algorithms due to this variability. finally, as the amount of the samples was limited, more comprehensive research is required to verify these findings and make a distinctive interpretation. author information corresponding author *simay dolaner (college of engineering and natural sciences, department of molecular biology and genetics, bahcesehir university, istanbul, turkey) simay.dolaner@bahcesehir.edu.tr author contributions simay dolaner performed the data analysis and produced the figures and manuscript. dr. harpreet kaur provided technical assistance with analysis and pipelines. elia brodsky and dr. mohit mazumder added project direction and provided guidance. dr. julia panov provided expert feedback for the project. competing interests the authors declare no competing financial and non-financial interests. supplementary documents supplementary notes 1, which includes external sources and related links mentioned in the methods section, can be found in the following link. https://www.dropbox.com/s/mlo17y1yjdxfekm the list of the significantly differentially expressed protein-coding genes can be found in supplementary table 1 through the following link. https://www.dropbox.com/s/zc1oaryr3zyhxlf supplementary table 2, which contains the name, ensembl id, and the possible functions of the 33 significant lncrnas, can be found in the following link. https://www.dropbox.com/s/4p67jf4d54if2b8 acknowledgments this research was performed within the framework of a research fellowship program and was supported by pine biotech. data processing and visualization pipelines were run with the t-bioinfo server. i would like to thank elia brodsky, dr. harpreet kaur, dr. mohit mazumder, and dr. julia panov for their expertise and guidance throughout the project. abbreviations cll: chronic lymphocytic leukemia ncbi: national center for biotechnology information bcr: b cell receptor tcr: t cell receptor pca: principal component analysis h-cluster: hierarchical clustering go: gene ontology kegg: kyoto encyclopedia of genes and genomes lncrna: long non-coding rna prrt3-as1: antisense to prrt3 h1fx-as1: antisense to h1fx ptpn22-as1: antisense to ptpn22 pcf11-as1: antisense to pcf11 syngap1-as1: antisense to syngap1 emt: epithelial to mesenchymal transition references [1] american cancer society. cancer statistics center. https://cancerstatisticscenter.cancer. columbia undergraduate science journal vol. 15, 2021 dolaner et. al. 27 org. 2020 [2] a. a. ferrando, c. lópez-otín, clonal evolution in leukemia. nature medicine, 23(10), 1135–1145. doi:10.1038/nm.4410 (2017). [3] m. v. haselager, a. p. kater, e. eldering, proliferative signals in chronic lymphocytic leukemia; what are we missing? frontiers in oncology, 10(), 592205–. doi:10.3389/ fonc.2020.592205 (2020). [4] g. fabbri, r. dalla-favera, the molecular pathogenesis of chronic lymphocytic leukaemia. nature reviews cancer, 16(3), 145–162. doi:10.1038/nrc.2016.8 (2016). [5] d. l. longo, j. a. burger, treatment of chronic lymphocytic leukemia. new england journal of medicine, 383(5), 460–473. doi:10.1056/nejmra1908213 (2020). [6] m. kwok et al., integrative analysis of spontaneous cll regression highlights genetic and microenvironmental interdependency in cll. blood, 135 (6): 411–428. doi: https://doi. org/10.1182/blood.2019001262 (2020). [7] j. delgado, f. nadeu, d. colomer, e. campo, chronic lymphocytic leukemia: from molecular pathogenesis to novel therapeutic strategies. haematologica, 105:xxx doi:10.3324/ haematol.2019.236000 (2020). [8] t. j. kipps, m. y. choi, targeted therapy in chronic lymphocytic leukemia. cancer journal (sudbury, mass.), 25(6), 378–385. https:// doi.org/10.1097/ppo.0000000000000416 (2019). [9] f. bosch, r. dalla-favera, chronic lym-phocytic leukaemia: from genetics to treatment. nat rev clin oncol, 16, 684–701 https:// doi.org/10.1038/s41571-019-0239-8 (2019). [10] l. fabris, j. juracek, g. calin, non-coding rnas as cancer hallmarks in chronic lymphocytic leukemia. interna-tional journal of molecular sciences, 21(18), 6720–. doi:10.3390/ijms21186720 (2020). [11] j. fernandes, et al., long non-coding rnas in the regulation of gene expres-sion: physiology and disease. non-coding rna, 5(1), 17. https://doi.org/10.3390/ncrna5010017 (2019). [12] l. nobili, d. ronchetti, e. taiana, a. neri, long non-coding rnas in b-cell malig-nancies: a comprehensive overview. on-cotarget, 8(36), –. doi:10.18632/oncotarget.17303 (2017). [13] m. aprile, v. katopodi, e. leucci, v. cos-ta, lncrnas in cancer: from garbage to junk. cancers, 12(11), 3220. https://doi. org/10.3390/cancers12113220 (2020). [14] m. gourvest, p. brousset, m. bousquet, long noncoding rnas in acute myeloid leukemia: functional characterization and clinical relevance. cancers, 11(11), 1638. https:// doi.org/10.3390/cancers11111638 (2019). [15] x. zhang, et al., mechanisms and functions of long non-coding rnas at multi-ple regulatory levels. international jour-nal of molecular sciences, 20(22), 5573. https://doi. org/10.3390/ijms20225573 (2019). [16] m. a. diamantopoulos, p. tsiakanikas, a. scorilas, non-coding rnas: the riddle of the transcriptome and their perspectives in cancer. annals of translational medi-cine, 6(12), 241. https://doi.org/10.21037/atm.2018.06.10 (2018). [17] j. gao, f. wang, p. wu, y. chen, y. jia, aberrant lncrna expression in leuke-mia. journal of cancer, 11(14), 4284–4296. https:// doi.org/10.7150/jca.42093 (2020). columbia undergraduate science journal vol. 15, 2021 dolaner et. al. 28 [18] q. tang, s. s. hann, hotair: an onco-genic long non-coding rna in human cancer. cellular physiology and biochem-istry, (), 893–913. doi:10.1159/000490131 (2018). [19] y. chi, d. wang, j. wang, w. yu, j. yang, long non-coding rna in the pathogene-sis of cancers. cells, 8(9), 1015. https://doi. org/10.3390/cells8091015 (2019). [20] j. iwakiri, g. terai, m. hamada, computa-tional prediction of lncrna-mrna inter-actions by integrating tissue specificity in human transcriptome. biology direct 12, 15 https://doi. org/10.1186/s13062-017-0183-4 (2017). [21] g. m. cruz-miranda, et al., long non-coding rna and acute leuke-mia. international journal of molecular sciences, 20(3), 735. https://doi.org/10.3390/ijms20030735 (2019). [22] g. arun, s. d. diermeier, d. l. spector, therapeutic targeting of long non-coding rnas in cancer. trends in molecu-lar medicine, 24(3), 257–277. https://doi.org/10.1016/j. molmed.2018.01.001 (2018). [23] b. langmead, s. salzberg, fast gappedread alignment with bowtie 2. nat meth-ods 9, 357–359 (2012). https://doi.org/10.1038/ nmeth.1923 [24] j. lever, m. krzywinski, n. altman, points of significance: principal component analysis. nature methods, 14(7), 641–642 (2017). doi:10.1038/nmeth.4346 [25] n. altman, m. krzywinski, points of sig-nificance: clustering. nature methods, 14(6), 545–546 (2017). doi:10.1038/ nmeth.4299 [26] m. i. love, w. huber, s. anders, moderat-ed estimation of fold change and disper-sion for rna-seq data with deseq2. ge-nome biol 15, 550 (2014). https://doi. org/10.1186/s13059-014-0550-8 [27] y. benjamini, y. hochberg, controlling the false discovery rate: a practical and powerful approach to multiple testing. journal of the royal statistical society: series b (methodological), 57: 289-300 (1995). https://doi. org/10.1111/j.2517-6161.1995.tb02031.x [28] a. subramanian, et al., gene set enrich-ment analysis: a knowledge-based approach for interpreting genome-wide ex-pression profiles. proceedings of the na-tional academy of sciences, 102(43), 15545–15550. doi:10.1073/pnas.0506580102 (2005). [29] m. v. kuleshov, et al., enrichr: a compre-hensive gene set enrichment analysis web server 2016 update. nucleic acids re-search, 44(w1), w90–w97. https://doi.org/10.1093/ nar/gkw377 (2016). [30] p. ghatalia, c. j. morgan, g. sonpavde, meta-analysis of regression of advanced solid tumors in patients receiving placebo or no anti-cancer therapy in prospective trials. critical reviews in oncolo-gy/hematology, 98(), 122–136. doi:10.1016/j.critrevonc.2015.10.018 (2016). [31] l. fan, h. li, w. wang, long noncoding rna prrt3-as1 silencing inhibits pros-tate cancer cell proliferation and promotes apoptosis and autophagy. experimental physiology, (), ep088011–. doi:10.1113/ep088011 (2020). [32] x. li, m. yutong, immune-related lncrna risk signatures predict survival of idh wildtype and mgmt promoter unmethylated glioblastoma, biomed re-search international, vol. 2020, article id 1971284, 11 pages, https://doi.org/10.1155/2020/1971284 (2020). [33] x. shi, et al. a newly identified lncrna h1fx-as1 targets dact1 to inhibit cervi-cal columbia undergraduate science journal vol. 15, 2021 dolaner et. al. 29 cancer via sponging mir-324-3p. can-cer cell int 20, 358 https://doi.org/10.1186/s12935020-01385-7 (2020). [34] e. delshad, et al. in-silico identification of novel lncrnas with a potential role in di-agnosis of gastric cancer, journal of bio-molecular structure and dynamics, (), 1–9. doi: 10.1080/07391102.2019.1624615 (2019). [35] r. negro, et al., overexpression of the au-toimmunity-associated phosphatase ptpn22 promotes survival of antigen-stimulated cll cells by selectively activat-ing akt. blood, 119(26), 6278–6287. https://doi. org/10.1182/blood-2012-01-403162 (2012). [36] r. cubas, et al., autoimmunity linked protein phosphatase ptpn22 as a target for cancer immunotherapy. journal for im-munotherapy of cancer, 8(2), e001439–. doi:10.1136/ jitc-2020-001439 (2020). [37] s. w. yang, et al., a cancer-specific ubiq-uitin ligase drives mrna alternative pol-yadenylation by ubiquitinating the mrna 30 end processing complex, molecular cell, 77(6), 1206–1221.e7. https://doi.org/10.1016/j. molcel.2019.12.022 (2020). [38] j. nourse, s. spada, s. danckwardt, emerging roles of rna 3′-end cleavage and polyadenylation in pathogenesis, di-agnosis and therapy of human disorders. biomolecules, 10(6), 915. doi:10.3390/biom10060915 (2020). [39] a. ogorodnikov, et al., transcriptome 3′end organization by pcf11 links alterna-tive polyadenylation to formation and neu-ronal differentiation of neuroblasto-ma. nature communications, 9(1), 5331. https://doi.org/10.1038/ s41467-018-07580-5 (2018). [40] m. h. berryer, et al., decrease of syngap1 in gabaergic cells impairs in-hibitory synapse connectivity, synaptic in-hibition and cognitive function. nature communications, 7, 13340. https://doi.org/10.1038/ncomms13340 (2016). [41] o. maertens, k. cichowski, an expanding role for ras gtpase activating proteins (ras gaps) in cancer. advances in bio-logical regulation, 55(), 1–14. doi:10.1016/j. jbior.2014.04.002 (2014). [42] m. berryer, et al. decrease of syngap1 in gabaergic cells impairs inhibitory syn-apse connectivity, synaptic inhibition and cognitive function. nature communica-tions 7, 13340 https://doi.org/10.1038/ncomms13340 (2016). [43] l. li, et al., tumor suppression of ras gtpase-activating protein rasa5 through antagonizing ras signaling perturbation in carcinomas. iscience, 21, 1–18. doi: 10.1016/j.isci.2019.10.007 (2019). 7 introduction dengue is a mosquito-borne viral infection mostly mediated by aedes albopictus. the virus has 4 distinct serotypes (denv 1-4), which are members of the flaviviridae family, genus flavivirus. dengue virus (denv), namely denv 1-4, spreads remarkably well in tropical regions where high temperatures are maintained all year long. the global temperature increase has resulted in a higher number of tropical and subtropical regions. previously safe regions have become targets of the denv, increasing the number of people at risk [1]. these opinions are supported by statistics discovering dengue virus antibody using phage display technology ryeogyeoung yoon1, yoonchang kim2, somin ahn3, lawrence kim4 © 2022 yoon, kim, ahn, kim. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 1columbia university, new york, ny, usa 2duke university, durham, nc, usa 3williams college, williamstown, ma, usa 4georgetown university, washington, dc, usa yntoab biotech, seongnam, south korea keywords: dengue virus (ns1), antibody discovery, phage display, elisa screening, scfv abstract: dengue virus (denv) is a type of flavivirus transmitted globally by the aedes aegypti mosquito and is the causal agent of dengue fever. due to global warming and rising temperatures worldwide, the rate of exposure to denv is now reaching unprecedented levels, with over 400 million people getting infected each year. despite worldwide attention to denv, there is not currently an effective treatment or vaccine. here, we focused on developing an antibody for the ns1 protein of the denv, which enables earlier detection of disease and could further decrease the rate of death. our study utilized a method more time-saving and cost-effective than the traditional method of extracting antibodies from the blood or b-cells of infected animals that show immunity: phage display. this technique allowed us to derive monoclonal antibodies (mabs) from the genome library and select denv antibodies that bind to antigen proteins. in this study, we used phage display to select ninety-six potential antibodies from three phases of panning, which then went through indirect elisa screening for optical density (od) confirmation (od > 1.5). sixteen of the high optical density antibodies were sent to a dna sequencing analysis facility where fifteen of them were revealed to be the same 1g1 strain. by understanding how antibodies are discovered using phage display technology, we hope to advance the field with reevaluation of current vaccines and the development of novel ones. columbia undergraduate science journal vol. 16, 2022 8 gathered by well-known organizations such as the centers for disease control and prevention (cdc). their statistics reveal a dramatic increase in the incidence rate of dengue in recent decades with over 400 million infections, 100 million people showing symptoms, and 22,000 deaths each year [2]. while 22,000 deaths alone are not a statistic to be ignored, we must also watch out for the symptoms of dengue fever. most people who are infected by denv recover within 2 -7 days, accompanied by mild symptoms including nausea, vomiting, and rash. however, severe dengue can result in severe internal bleeding and organ impairment [2]. in order to prevent such symptoms and casualties, it is imperative to diagnose the disease quickly and precisely. currently, most diagnoses is based on reverse-transcription pcr (rt-pcr), a technique that enables the identification of a patient’s serotype of denv. in addition to diagnosing, we must also find a cure or vaccine for dengue. the vaccine, dengvaxia, is currently in use; however, in 2017, the manufacturer announced that even those who have received the vaccine may be at risk of developing severe dengue if they become infected after being vaccinated [3]. as such, finding a more effective vaccine or treatment against denv has become more important than ever. conventional methods of vaccine development require the injection of the virus’s antigens into experimental animals such as mice and rabbits. scientists wait for the animals to develop an immune reaction and then extract either the animal’s antibodies against the virus from their blood or b cells, which develop the antibodies [4]. due to the duration of the antibody development in the animal's immune system, this method is time-consuming (usually about 6 months when using rats). furthermore, we cannot disregard the ethical implications of animal testing. therefore, in our study, we offer an alternative –phage display technology– in the discovery of an antibody for the non-structural protein 1 (ns1) antigen, a biomarker for denv. phage display technology refers to a bioengineering technique that allows in vitro selection of monoclonal antibodies (mabs) by displaying peptides or antibodies on the surface of bacteriophages. since the whole process occurs outside a living organism in a short timeframe, it is more ethical and time-effective compared to conventional methods [5]. the procedure consists of two main parts: antibody discovery and antibody production. the antibody discovery phase begins with the construction of the phage display library. once the library is created, the panning process begins, which exposes the library to the target protein the ns1 protein. unbound phages are washed away, and those that show affinity for the protein are recovered by elution. the recovered phages are used to infect new host cells for amplification. repeated panning in the phage display technique allows for the selection of only those that maintain a high affinity for the protein, resulting in high specificity of the antibodies. once the final panning is complete, bounded antibodies go through elisa screening with a secondary antibody containing an anti-ha tag. only those with optical density (od) values over 1.5 are considered valuable. the plasmid of these antibodies is separated and sent to sequencing analysis. the antibody production phase began with the cloning of eligible colonies. consistent human antibody fragments produced through the phage display technique can be re-cloned into igg expression vectors for the generation of antibodies in the human body. because it maintains compatibility with the human immune system, it should be highly efficacious and safer compared to the conventional method. 9 over the course of this project, we focused on developing an antibody for the earth-threatening dengue virus with a more productive ‘phage display’ technique. discovering the specific sequence of antibodies for the ns-1 protein of dengue virus has the potential of manufacturing helpful vaccines or treatments. methods 1. antigen coating, panning, rescue immune tubes containing a coating buffer and protein were prepared prior to the experiment. the panning and rescuing steps were executed three times each. phages and a 3% skim milk 1xpbs solution were added to the tubes. after washing and eluting, the solution was transferred to a tube containing tris-cl and bacterial strain tg1. in order to estimate the number of infected bacteria, the cultured bacteria were smeared on a 90mm lb agar plate containing ampicillin. the remaining infected bacteria were centrifuged and smeared on a 150mm lb agar plate before overnight incubation. after counting the number of bacteria grown in the 90mm plate, a culture medium was added to the 150mm plate. the mixture was then inoculated in a sb-ampicillin culture medium. helper phage and kanamycin were also added before overnight incubation. to prepare for the third and final panning step, the final medium was centrifuged and added to a tube containing 5xpeg. the supernatant was then filtered and used for panning and rescuing. 2. seeding and induction the bacteria grown in the 90mm plate were counted. single colony bacteria were picked from the 90mm plate and seeded into a 96 well plate containing the sb-ampicillin culture medium. after overnight incubation, the single colony bacteria stock was transferred to another 96 well plate containing the sb-ampicillin culture medium and placed in the deep freezer. iptg, a molecular mimic of allolactose, was added before overnight incubation. 3. screening and seeding to prepare for screening, periplasmic sup was collected. the incubated single colony bacteria were centrifuged. after removing the supernatant, 1xtes buffer was added to the bacteria pallet, creating a concentration difference. this was kept in ice to extract periplasm figure 1. phage display cycle [6] columbia undergraduate science journal vol. 16, 2022 10 using osmosis. after a 0.2xtes buffer was added and kept in ice, the mixture was centrifuged to collect the periplasmic sup from e.coli in a form of supernatant. in order to prepare elisa screening, the protein-coated 96 well plate was blocked with 3% skim milk. 3 columns of the plate were coated with protein and skim milk, while another 3 columns were blocked by skim milk only (for the control). the periplasmic sup was added to each well (6 columns). after 3 times of washing with a wash buffer, a tmb substrate solution was added. tmb reaction was stopped by adding sulfuric acid, and the result was compared with the control. the optical density (o.d) of each well was recorded with an elisa reader (450nm). the stocks of the wells with positive results (1.5 < o.d) were transferred into different 15ml tubes. 4. dna prep and sequencing the incubated stocks were transferred to 1.5ml tubes, which were then centrifuged. the pallet was filtered and resuspended with p1 buffer (resuspension buffer) [7]. p2 buffer (lysis buffer) was added and mixed by flipping the tubes upside down several times, and it was repeated with p3 buffer (neutralization buffer) [7]. after centrifuging the mixture, the supernatant was transferred into miniprep tube columns. with these columns, the mixture was centrifuged 5 times (first: the mixture itself; second: with ew buffer; third: with pw buffer; fourth: the resultant mixture; fifth: with new tube columns and dw buffer). in order to sequence the dna collected, they were quantified with a take3 device. these samples were sent to macrogen and sequenced. using the online website, expasy, the sequenced dna was translated into an amino acid sequence. the website gave several amino acid sequences depending on the starting base to be read as a codon. the frame containing the linker (gqss … gggg) and h6 tag was chosen as the correct sequence. results titering results from panning and seeding in 90mm plate to estimate the number of colonies after the 3rd panning, we counted the number of single colonies in the 90mm plate. table 1 shows the number of colonies counted on each plate (total 4). we divided the section of the plate into eighths and counted each section. assuming each section maintains a similar number of colonies, we multiplied the number of colonies counted in each section by 8. analyzing the elisa result in order to test the presence of targeted antibodies, we performed elisa screening on the samples of periplasmic sup. the result of the elisa test was presented in tables 2a to 2d, each with separate columns for antibody and for 3% skim milk. we compared the columns for antibody to the columns for 3% skim milk which were controls. after the analysis of the data from the elisa reader (450nm wavelength), we collected several positive results (blue and yellow cells from table 2) with an anomaly (c6 in table 2a). each table is performed by different people. stocks with a number greater than 0.2 are highlighted in blue with the gradient (darker blue means the value is greater). blue samples indicate positive results which means the antibodies are present. 8 yellow stocks were randomly selected for sequencing.the anomaly could have resulted from contamination during pipetting or unsuccessful washing. based on these results, we randomly chose 8 stocks that gave positive results (table 2) to quantify the dna of antibodies before sequencing. 8 randomly selected stocks are colored in yellow in tables 2a to 2d. columbia undergraduate science journal vol. 16, 2022 11 table 1. titering results of four plates after the 3rd panning process table 2. cultured antibody elisa test result under 450nm wavelength a) b) columbia undergraduate science journal vol. 16, 2022 12 dna quantification before sequencing the stocks that have absorbance ratios of 260 nm/280 nm above 1.8~2 were considered to have sufficient amounts of dna for successful sequencing [8]. with the take3 micro-volume plate, the absorbance ratio (260nm/280nm) of each sample was collected (table 3). samples were randomly selected (8 stocks which are yellow cells in figure 2). the highlighted number indicates the concentration of dna. there were variances in the value of the dna quantities per microliter. it was expected to have no variance as the samples were prepared with the same method and the environment. these variances may indicate that the samples were not pure or the sequencing was unsuccessful. however, because the observances at 260 nm were greater than those at 280 nm, resulting in high dna to protein ratios (greater than 2), we assumed that the dna samples were sufficiently pure to be sequenced. c) d) columbia undergraduate science journal vol. 16, 2022 13 analyzing the sequencing results after the dna quantification, we requested humanizing genomics macrogen to sequence the 8 samples. the purpose of single chain fragment variable (scfv) sequencing was to understand the different types and positions of amino acids within the created antibody. after obtaining the plasmid of the single colony antibodies, each dna strand was sequenced using the 5’-, 3’-oligonucleotide of a scfv. the final order of the scfv was determined after reading the sequence from both directions and combining the common regions. (abi3730xl, sanger-method sequencing, phred score ≥20). in order to analyze the results of the sequencing, we used an expasy(https://web.expasy.org/ translate/) that translated the dna sequence in our raw data (figure 2) into corresponding amino acids. from the results, we determined the accurate reading frame (figure 3) by identifying the h6 tag and the linker. within the scfv, there is a light chain variable region followed by a linker and a heavy chain variable region. as heavy and light chains of the antibodies were linked with linker protein (gggssggggs) at heavy chain 6, the presence of the linker and h6 tag on the amino acid sequence indicates that we have successfully discovered our targeted antibody. table 3. dna quantification results of 8 antibody stocks randomly selected figure 2. raw dna sequence columbia undergraduate science journal vol. 16, 2022 14 discussion we demonstrate here the successful discovery of an antibody for the non-structural protein 1 (ns1) antigen, a biomarker of denv. our discovery employed phage display technology, which enabled in vitro selection of monoclonal antibodies by presenting antibodies on the surface of bacteriophages. at large, our results indicate that discovered antibodies are highly specific to the targeted ns1 antigen. scfv sequencing confirmed our accuracy with the identified h6 tag and linker on our amino acid sequence. the high specificity of the antibody was derived from the elisa technique and the quantity of dna in the cultured colony was verified through dna quantification prior to sequencing. clones that had an od value above 1.5 in the elisa technique were only allowed in cultivation and colonies that maintained an absorbance ratio (260nm/280nm uv light) above 2 were considered pure enough to be tested. due to an increasing number of denv patients concurrent with global warming, the necessity for therapeutics to overcome this crisis is on the rise. the ultimate goal of our study was to aid the development of effective vaccines or treatments for denv; thus, enabling earlier detection of the disease and further lowering the rate of death. clearly, further studies are warranted to elucidate the immune parameters that could suggest a solution to overcome the limitation of the current vaccine (dengvaxia). unlike the current vaccine, the novel vaccine derived from our ns1 protein antibody is expected to significantly lessen the possibility of re-infection of dengue variants. in prospective studies, the sequence of the antibody should be investigated further to figure out the adequate sequence of dna that could potentially neutralize the targeted antigen. we thereby anticipate yielding efficient antibody therapeutic from the antigen-specific neutralizing sequence. author information corresponding author *ryeogyeoung yoon ry2410@columbia.edu competing interests the authors declare no competing financial and non-financial interests. figure 3. accurate translated reading frame columbia undergraduate science journal vol. 16, 2022 15 acknowledgements we thank yntoab for providing us with an equipped laboratory. abbreviations denvdengue virus 1g1histone h3 antibody mabsmonoclonal antibodies ns1nonstructural protein 1 scfvsingle-chain variable fragment odoptical density elisaenzyme-linked immunoassay iptgisopropyl β-d-1-thiogalactopyranoside references [1] guzman, m., halstead, s., artsob, h. et al.(2010).dengue: a continuing global threat. nat rev microbiol 8, s7–s16 . https://doi. org/10.1038/nrmicro2460 [2] centers for disease control and prevention. (2019). about dengue: what you need to know. cdc 24/7: saving lives, protecting people. https://www.cdc.gov/dengue/about/index.html [3] world health organization. (2021). dengue and severe dengue. who press, geneva, switzerland. https://www.who.int/news-room/ fact-sheets/detail/dengue-and-severe-dengue [4] s.kellermann, l.green.(2002). antibody discovery: the use of transgenic mice to generate human monoclonal antibodies for therapeutics. current opinion in biotechnology. 13 (6)., 593597,. [5] j.bábíčková, l’.tóthová, p.boor, p. celec. (2013). in vivo phage display a discovery tool in molecular biomedicine. biotechnology advances.,31 (8), 1247-1259,. [6] proteogenix. phage display cycle: phage display library, panning/biopanning and validation by elisa. proteogenix.,schiltigheim, france. https://www.proteogenix.science/antibody-production/phage-display-services/ [7] dna-spin plasmid dna purification kit from intron biotechnology [8] olson, n. d., & morrow, j. b. (2012). dna extract characterization process for microbial detection methods development and validation. bmc research notes, 5, 668 columbia undergraduate science journal vol. 16, 2022 letter from the editor-in-chief dear readers, i, along with the rest of the cusj board, are ecstatic to present you with the 16th volume of our publication. since its inception in 2006, cusj has strived to promote scientific journalism among undergraduate students and help foster a deeper appreciation and enhanced engagement for everyone in the science community. in this regard, i hope this issue will do exactly that. this issue encompasses a wide range of topics in hopes that you as a reader will be able to resonate and immerse yourself with the research presented such that you, too, feel provoked to challenge, corroborate, and contribute to the deeply enriching discussions that these articles naturally invite. this publication is a culmination of a year of hard work from the cusj team. specifically, i would like to thank the editorial board for their input in reviewing and selecting the articles presented in this issue as well as the cusj committee who were always so dependable throughout the year and were responsible for editing the articles alongside the authors. without them, this publication would not have been possible. in addition, i would like to thank our president, arjun kudinoor. throughout the year, arjun was always able to provide great feedback to improve our processes and was a guiding hand in ensuring that this specific publication lives up to, if not exceeds, the quality that has come to be expected from those before it. i am truly excited for what is to come for the future of cusj given arjun and the rest of the board's dedication to improving and setting up a foundation for cusj to suceed in the long-term, which is especially apparent with the inaugural publication of the columbia scientist this may led by aaron jackson. lastly, i would like to thank the cusj faculty advisory board, whose support has helped us greatly in allowing us to present you with research of the highest standard. and with that, it is my honor to present you with the 2021-2022 edition of cusj. it has been a privilege serving as cusj's editor-in-chief and i cannot wait to see what is to come! thank you so much for reading and supporting cusj! kazi rabbe editor-in-chief, cusj columbia undergraduate science journal 4 microsoft word cusj_spring_2015_journal.docx !cusj%|%volume%9%|%spring%2015% 14% are veterans subjected to social indifference? staff sergeant (usa, ret.) allen j. powell1 1college of liberal arts, pennsylvania state university, state college, pa, usa ! 1 introduction: social indifference is a difficult thing to prove. there have been many studies on the subject and often the conclusions lead us to believe that social indifference was a primary factor. yet indifference can be one of many underlying contributing causes that may lead to the end results, and therefore it would be a fallacy to argue its existence based on the actions of a large cohort simply because such generalized assumptions could have any number of different causes contributing to an end result. nevertheless, indifference is a very real problem. bennett (1998) conducted a study on indifference as it applied to politics and national responsibilities. his study showed that a greater exposure to higher education among today’s younger generations has not produced a birth cohort very interested in public affairs or inclined to expose itself to political media. as a result, today’s youth are poorly informed about political affairs at home and abroad. (bennett, 1998) bennett’s supporting contributions to his study included stein (1983), who stated “in a state of astonishing ignorance, young americans may well not be prepared for even the most basic national responsibility – understanding what the society is about and why it must be preserved.” (stein, 1983). stein believed that young people ignorant to political affairs are not prepared to continue the society because they basically do not understand the society enough to value it. as bennett discovered through the use of his research, which was conducted using numbers from past and present pew and nes polls, the majority of americans below the age of 30 cared more about sports and entertainment than they did about news, politics or foreign affairs, so much so, that the vast majority actually go out of their way to avoid coming into contact with political matters or affairs of state of any kind. (bennett, 1998) bennett concluded that the less people pay attention to media stories about public affairs, the more politically ignorant they are as a result. furthermore, the primary contributor to this was noted to be social indifference. as a whole, the majority simply did not care about the subject matter enough to want to pay attention to it. (bennett, 1998) while that may seem an obvious remark, the question remains as to whether those results remain consistent today, and whether the advancement of the information age has affected the decline in public affairs interests bennett showed. another relevant study on indifference – as it may apply to veterans and those with disabilities – is calderbank’s (2000) study on social indifference towards the abuses suffered by 2 individuals with disabilities. calderbank discussed whether the forms of abuse experienced by disabled people resulted from either vulnerability of the individual, or as a consequence of social attitudes towards disabled people as a whole. (calderbank, 2000) furthermore, calderbank claimed that the evidence suggests that individuals with disabilities are socially and personally devalued as both human beings and citizens, which is a contributing factor to the existence of indifference towards the issue. (calderbank, 2000) calderbank’s conclusions were that the sheer absence of studies conducted on social disability issues – the lack of research literature on the topic – directly indicates a lack of both social awareness and concern by society as a whole for the topic, and to which leads to further perpetuation and compounding of the issues. (calderbank, 2000) as a result, we are led to believe indifference is the root cause, even though there may be many other contributing factors. in both these studies, indifference was concluded as the primary underlying root cause for the issues existence. while the evidence may support the conclusions in one way or another, and there is no argument that indifference is a severe issue for any social issue, we are left with possibilities that other social factors might remain the true cause. so in the attempt to identify social indifference towards a sociological issue, we instead need to clearly identify a number of social factors supported by sociological theories that can be tied to indifference from several different perspectives, and to where they can be used to isolate indifference as the most predominant underlying cause for each factors existence. therefore, in the search for generalizable social indifference towards a particular social matter, three important questions are required: how is it that indifference is the primary root cause affecting the matter, what are the contributing factors, and are the findings consistent in similar matters. in the effort to answer the research question at hand – are veterans subjected to social indifference – a clear understanding of how multiple social theories impact social causes is essential for the isolation of indifference as the primary root cause. military service members frequently do jobs that are arguably the most dangerous jobs in the world. war is physically, psychologically, and spiritually damaging. veterans who are injured or wounded – either physically or psychologically – will suffer from physical and/or psychological disabilities that directly impacts their quality of life. while most physical disability treatments have received major improvements over the years in medical science, psychological disabilities have been historically far more difficult to treat. for example, an amputee may suffer the loss of a limb, but medical technology in the field of prosthetics has made significant advancements !cusj%|%volume%9%|%spring%2015% 15% 1 towards restoring lost mobility or use. on the other hand, the traumatic effects of depression, anxiety, and pts(d) come alive in expressions of fear, sleepless nights, nightmares, concentration issues, anger management issues, isolation, substance abuse, and even suicide, and while there does exist today a large array of drugs, treatments, and therapies designed to help with these issues, modern science has yet to find cures for psychological traumas. veterans are consistently subjected to a lifetime of issues – physical and psychological – unique to the veterans community. (wolfe, et al., 1999; riggs, byrne, weathers, and litz, 2005; dobie, et al., 2004) while there are a large host of services and organizations that exist today to help and aid veterans exclusively with many of these issues, the majority of services are designed for actively serving service members, and either stop or are reduced upon the veterans exit from service. unfortunately, however, many of these traumas last for the rest of the individual’s life. they are both life altering and pose significant hardships and challenges at restoring some manner of a veterans quality of life, and many do require a lifetime of continued care and unique services to some degree. (casarett, et al., 2008) furthermore, veteran reintegration into civilian society – or within some cultural aspect of a society – has historically been a unique challenge for veterans, and many veterans are never fully able to reintegrate upon return from in-theater service. (sayer, et al., 2010) it is because veterans issues exist that a large number of organizations and charities have been created to attempt to address veterans issues and even help to provide legislation for them. yet, like most charities and service organizations, meaningful support in the form of monetary charity and volunteer work are often the only ways to provide their services and achieve their goals. unfortunately, social indifference has become the limiting factor in many support causes and campaigns’ ability to succeed. instead, what we are finding is a society of individuals willing to offer token support (words of support, a hand shake, a “like” or a “share” on social media) simply to make themselves feel like they have accomplished the provision of support, while confirming such notions because others see them as supporting a cause (symbolic self-completion and slacktivism). social rules state that people within a society should be proud of their veterans and support them (both in the effort to defend the nation, and in the veterans time of need), and most choose to do so with words of “thanks” and/or a handshake. however, when confronted with the option to support a cause with meaningful support, the majority choose to simply “like” it (either by acknowledging its existence, or by clicking a like button online), and fewer will “share” it (either through display of a pin or a wristband, by telling others about the cause, or by clicking on a share button online), but fewer still will offer up the meaningful support requested by the campaign (either through donations or volunteering), and an even smaller number will offer up an array of reasons as to why they cannot provide meaningful support. 2 it is here where social rules might actually be responsible for the creation of symbolic self-completion, and the spread of slacktivism as a result, simply because social indifference exists towards the cause from the beginning. if individuals are indifferent towards the issues of concern, but social rules make them feel obligated to offer support of some kind, then slacktivism is the best way for the individual to show that support while providing as little actual support as they possibly can in the process. choosing token support instead of meaningful support is the easiest path towards achieving selfcompletion, thus making them feel like they provided real support by confirming it publically, and therefore obeying the social rules. in the end, the rules were followed, no actual support was provided, the individuals feel good about themselves, and yet the cause is left unfulfilled. with the presence of symbolic self-completion and slacktivism, and an understanding of social rule systems theory, we are more able to rule out other possibilities and identify social indifference as a primary root cause. supporting theories: social rule systems theory social rule systems are used to examine all levels of human interaction. (burns and flam, 1987; giddens, 1984; goffman, 1974; harré, 1979; and lotman, 1975) they provide more than potential constraints on action possibilities, they also generate opportunities for social actors to behave in ways that would otherwise be impossible; for instance, to coordinate with others, to mobilize and to gain systematic access to strategic resources, to command and allocate substantial human and physical resources, and to solve complex social problems by organizing collective actions. in guiding and regulating interaction, social rules grant behavior recognition, show characteristic patterns, and make such patterns understandable and meaningful for those who share in the rules. actors adhere to and implement rule and rule systems to varying degrees. (flam and carson, 2008) compliance with, or refusal to comply with, particular rules are complicated cognitive and normative processes. typically, there are diverse reasons for rule compliance, however, the key factors to highlight in the case of social indifference are: interest factors and instrumentalism, and identity and status. in the case of interest factors and instrumentalism, actors may advocate rules to gain benefits or to avoid losses. (mirowsky, ross, and willigen, 1996; woodbine, 2002) in the case of identity and statuses, adherence to rules – and commitment to their realization – may be connected to an actor’s identity, role, or status, and the desire to represent the self as identified by, or committed to, the set of rules. (nghe & mahalik, 2001; waterman & goldman, 1976; mclean & pratt, 2006). social rule systems theory is important in the identification of indifference because social rules guide the behavior of both individuals and large groups or populations of people. they are responsible for the underlying development, construction, maintenance, and evolution of culture, and not only include a continuously adapting set of norms, but a list of taboo’s for !cusj%|%volume%9%|%spring%2015% 16% 1 continuously adapting set of norms, but a list of taboo’s for which to avoid. finally, social rules contribute to the prevailing sense of identity, social statuses, interests, and instrumentalism factors that lead to the socially indifferent behavioral patterns that make it difficult to create interest in social causes, prompt social action to create change, and generate social willingness to provide for meaningful participation and support. symbolic self-completion symbolic self-completion is the idea that people define themselves as musicians, athletes, etc. by use of indicators of attainment in those activity realms, such as possessing a prestige job, having extensive education, or whatever is recognized by others as indicating progress toward completing the self-definition. (wicklund and gollwitzer, 1981). the self-completion idea postulates that when important symbols—indicators of self-definition—are lacking, the person will strive towards further, alternative symbols of the self-definition. symbolic self-completion provides individuals with the easiest way to express their need to be seen by others as the person they want to be – without actually being that person in real life – through the use of third-party confirmation. (fishback, dhar & zhang, 2006) an example of this would be the sharing of a charity campaign through social media, which is seen by an individual’s connections, and therefore creates the image that the individual actually supports that charitable cause. this then affirms the individual’s identity as a charitable person, despite the fact that a “share” is equal to token support and not meaningful support, and therefore, the individual isn’t really charitable at all, nor have they provided support for that cause. symbolic self-completion is relevant to include here because it supports the scapegoat loophole in social rules that state people should be charitable, despite the fact that the individual maintains indifference towards the social cause, and it allows for the creation of slacktivism as a by-product. slacktivism slacktivism may be a new word as far as social science is concerned, however, its existence is hardly new at all. slacktivism is the idea that an individual would rather give token support than meaningful support of any kind because they feel that token support is just as supportive and meaningful as meaningful support, and their self-image of fulfilling the social obligation of support is already fulfilled through a showing of that initial act of token support. (kristofferson, white & peloze, 2014) (gollwitzer, sheeran, michalski & seifert, 2009). slacktivism can be directly contributable to the failure of a charitable campaign for a few reasons. if individuals display an initial public token degree of support for a cause, they will be less likely to subsequently provide more meaningful support in the future because 1): their desire to create a positive self-image in both their own eyes, and the eyes of others has been fulfilled (impression management); and (2): 2 because public observability has kept them in-line with their previously displayed values (the desire to remain consistent with one’s own values). (kristofferson, white and peloze, 2014) this particular research is highly valuable to society today due to the large (and growing) number of charitable causes around the world – who are each attempting to provide tangible, meaningful action to fix social and world issues of all kinds – but whose campaigns for tangible, meaningful support receive more token support than meaningful support, ultimately leading to failure in the campaign. as such, slacktivism progresses the identification of indifference as an underlying social problem. it shows in one way (out of many) how society’s failure to be willing to offer anything of value or importance is governed by their idea of already completing that task through symbolic gestures of support, despite the fact that such token support fails to provide any meaningful support at all. this makes slacktivism a key aspect in the search for, and confirmation of, the existence of indifference towards a social issue because it provides the ability for indifferent individuals to fulfill social rules without actually participating. case studies: before beginning to analyze the case studies used to answer the research question at hand, a comparative example where social indifference has been clearly identified must first be established. to do this, the method will be to use the supporting theories in a way that isolates social indifference as the primary root cause by removing other possible factors. additionally, the comparative example used must also act as a control. it cannot be tied directly to veteran’s causes – as indifference directed towards veterans is what is being looked tested for in this study – but must be of a similar nature, such as a charity that directly seeks to impact a significant social cause. as such, the als ice bucket challenge will be used as the comparative example because of the following similarities: 1. it is an established and well-known charity that directly seeks to impact a major social cause, and as such, it is able to provide a similarity for reference in the following case studies covering campaigns that seek to support veterans charities that directly impact significant social causes for veterans. 2. it has nothing to do with veterans, and therefore can act as a control. 3. it gives the ability to establish the method of stripping away alternative possibilities, thus leaving social indifference behind as the primary root cause, which can be repeated for the following case studies. als ice-bucket challenge: 2014 introduction: in the summer of 2014, the “als ice-bucket challenge” became widely popular as an internet fad, and while there is no doubt that the social phenomena raised awareness of als, the numbers indicate a disproportionate level of participation versus donations. !cusj%|%volume%9%|%spring%2015% 17% 1 background information: amyotrophic lateral sclerosis (als)—also referred to as motor neuron disease (mnd), charcot disease, and, in the united states, lou gehrig's disease—is a neurodegenerative disease with various causes. it is characterized by muscle spasticity, rapidly progressive weakness due to muscle atrophy, and difficulty in speaking, swallowing, and breathing. individuals diagnosed with the disease typically live no longer than 6 to 8 months. als is the most common of the five motor neuron diseases. the disorder causes muscle weakness and atrophy throughout the body due to the degeneration of the upper and lower motor neurons. individuals affected by the disorder may ultimately lose the ability to initiate and control all voluntary movement, although bladder and bowel function and the muscles responsible for eye movement are usually spared until the final stages of the disease. cognitive function is generally spared for most patients, although some (about 5%) also develop frontotemporal dementia. a higher proportion of patients (30–50%) also have more subtle cognitive changes, which may go unnoticed, but are revealed by detailed neuropsychological testing. infrequently als coexists in individuals who also experience dementia, degenerative muscle disease, and degenerative bone disease as part of a syndrome called multisystem proteinopathy. sensory nerves and the autonomic nervous system are generally unaffected, meaning the majority of people with als will maintain hearing, sight, touch, smell, and taste throughout the duration of the disease. significance: the als ice-bucket challenge was a widespread internet social media-based phenomena designed to prompt individuals to participate through a “challenge” that increased social pressure to prompt further participation for the purpose of generating significant monetary donations to the alsa for funding of research efforts to find a cure for als. the phenomena turned into a social fad as symbolic self-completion created slacktivism resulting from generalizable social indifference towards the cause itself. results: forbes, the bbc, and time each gave consistent figures showing the growth of awareness, participation, number of individual donations, and total amount donated throughout the summer of 2014. as cited by each, by the end of august, the social media giant, facebook, indicated more than 2.4 million ice-bucket challenge videos had been uploaded, and over 30 million interactions – views, likes, comments, and shares – were reported. on instagram – a facebook subsidiary – over 3.7 million videos were uploaded as well, and even the wikipedia page dedicated to als received more than 2.7 million visits from 1 august to 27 august 2014. (townsend, 2014; diamond, 2014; stampler, 2014) finally, while the alsa gained over $100 million in donations covering the time period from the start of the ice-bucket challenge phenomena to the end of august, the amount of new donors attributed to the awareness created by the campaign itself was estimated at close to 740,000. (steel, 2014) 2 discussion: if we use the facebook numbers alone, out of 30 million interactions, 1 out of every 12 resulted in a video contributing to the perpetuation of the challenge, while 1 in 40 actually resulted in a donation of any kind. that means that only 0.0246 percent of all individuals who both saw and interacted with the challenge over facebook in any way ended up donating. why is this important? social rule systems theory states that individuals will comply with social rules to maintain or follow a norm, to show their compliance and gain acceptance, to confirm identity or protect identity, or simply to avoid loss. (mirowsky, ross, and willigen, 1996; woodbine, 2002) (nghe and mahalik, 2001; waterman and goldman, 1976; mclean and pratt, 2006) the percent of individuals that provided meaningful support was 0.0246, while the percent that chose to avoid giving meaningful support of any kind was 0.9754. out of the total, 0.3 percent was willing to offer a form of token support, and there are confirmed cases of overlap where those who offered token support followed up with meaningful support. however, even if we assumed 100 percent of all who offered token support also offered meaningful support, the total figure would still only come out to 0.3246, meaning that 0.6754 percent yet still refused to offer any form of support at all. $100 million divided by 740,000 comes out to a figure of $135.14 per donor, but that would also leave out the large number of celebrities and wealthy who donated large sums themselves – charlie sheen, for example, openly donated $10,000. (als, 2014) it would also leave out individuals who already donate on a consistent basis, and those who donated without interaction via the als ice-bucket challenge over social media. finally, these figures only account for interaction and participation figures from one social media platform, meaning that – out of the $100m total figure – adding in the combined figures from all other sources of the challenge that resulted in a donation would make that 0.0246 figure even smaller. in reality, the “1 in 40” figure could be closer to “1 in 100” or “1 in 500”… but that is just speculation. we may not know what the actual figure would end up as, however the evidence shows it is worse than “1 in 40”. finally, it must be understood that all of these figures – in fact, the entire als ice-bucket challenge campaign itself – where gathered from online interactions through the use of social media. therefore, it would be safe to assume that the vast majority of interactions came from social media users who could obviously afford internet access in one way or another. it would not be appropriate to assume that all those who maintain some form of internet access suffer from financial difficulties either. therefore, given the evidence based on percentages of interactions versus meaningful support, it is safe to say that the majority of people who saw the campaign ultimately were not motivated to follow-up with meaningful support. this indicates indifference towards the cause because if the individuals did care about it, they would have been motivated to offer meaningful support to some degree regardless of personal circumstances. !cusj%|%volume%9%|%spring%2015% 18% 1 case use for comparative analysis and control: a large number of widely known charities have used everything from celebrities to fund raisers, from advertisements to social media campaigns to attempt to spread awareness of their cause, gain donations, and prompt participation. the als ice-bucket challenge was different only in the sense that it was not sponsored or created by the alsa, but instead became a social phenomenon due to random faddish trends. as the evidence shows, 30 million interactions over a single social media platform proves a vast majority of individuals were only willing to offer token support, meaning that slacktivism and symbolic self-completion are directly in play. additionally, social rules would dictate that it is socially acceptable – and therefore good – to contribute to a charity and socially unacceptable – and therefore frowned upon – to not donate to or refuse to donate to help a cause such as als. therefore, with a near completely online social media campaign, and a figure of less than 0.0246 percent of actual meaningful support, factors other than social indifference can be removed as possibilities. social rule systems theory, combined with the presence of symbolic self-completion, and resulting slacktivism, make it clear that individuals only provided token support simply to avoid the loss of their selfimage or to gain a positive self-image at the confirmation of third parties. even though they all possessed enough financial capabilities to afford an internet connection of some kind, the overwhelming vast majority refused to offer any amount – not even a single $1 – of meaningful support. despite the trend over the summer of 2014, the als icebucket challenge was still the victim of social indifference, and potential donations were lost due to the loopholes of slacktivism and symbolic self-completion. simply put, the number of potential donations that could have been received as a result of the phenomenon suffered greatly due to social indifference to the cause itself. the isolation and identification of generalizable social indifference towards the als ice-bucket challenge, combined with the applied use of social theories that allow for the isolation of social indifference as a root cause for the failure of the campaigns achievement of its goals, develops the method to test for indifference as a cause in subsequent campaign efforts. further, identification of the subject of indifference can be isolated through the use of comparative analysis across several campaigns who, 1: maintain only that single subject in common, and 2: show consistency in the identification of indifference as the primary root cause for each campaign failure. generalizability can also be established as a result due to, 1: identification of the subject of indifference, and 2: identification of consistency of indifference directed by a cohort towards that specific subject across many varied cases. in the following case studies, the same social theories – social rule systems theory, symbolic self-completion, and slacktivism – will be used in consistent applied methods to, 1: establish the presence of social indifference towards the 2 campaign, 2: subsequently identify the subject of social indifference within each campaign, and 3: establish generalizability of social indifference towards the subject itself. to accomplish this, 2 large-scale and interconnected campaigns that focused exclusively on the provision of military specific content to build a large cohort that identified with the such content, and then subsequently campaigned directly toward the same cohort to generate monetary funding for veterans causes, will be used to establish the presence of social indifference. next, an additional three independent veterans charity campaigns that each maintain the idea of helping to aid disabled veterans through the use of meaningful support as the sole similarity between them, will be used to identify the subject said social indifference is direct toward. finally, analysis across all campaigns will be used to test for generalizability of the results. 1) military photography: march 2013 – july 2014 introduction: two australian media professionals started military photography in march of 2013. the company was a subsidiary under the control of carbine hq that served two purposes. one, to provide a media service entirely dedicated to photographic media that covered military forces around the world, and two, to receive monetary funding from sales designed to generate funding for charities that helped and aided disabled combat veterans. to acquire such funding, military photography generated a large fan base through the use of its facebook page, and then marketed items for sale directly to that fan base whereby 40% of all profits from sales went straight to help fund veterans charities that directly supported disabled veterans and their families. results: within 12 months of the start of its facebook page, military photography had grown a fan base of 118,000 followers worldwide. in addition, this fan base was consistent in active participation with the page, meaning that comments, shares, likes, views, emails, and requests where a daily norm. the pages email inbox routinely registered 1,000+ messages a week, while facebook analytic figures showed over 500,000 interactions every week, and the page was continuing to grow on a daily basis. this offered a large online presence for which to campaign to. by march of 2014, the page had become so engaged that the team had grown to include 8 individuals. three media professionals from australia (including the two founding members), the first international administrator was a retired u.s. army service member and professional photographer from the united states, a retired military photographer from canada became the fifth member, the sixth was a currently serving combat camera soldier from belgium, and the seventh and eighth where also united states service members and professional photographers. the international team went to work improving the pages content and growing the followership. however, after the sales campaign started, analytic figures showed consistent near-zero engagement figures for any post tied to the campaign. !cusj%|%volume%9%|%spring%2015% 19% 1 the more the page campaigned to offer items for sale for the purpose of generating funding to help disabled veterans, the less active the engagement became. eventually, the pages engagement had lowered below the 25,000 per week mark, then plummeted below the 10,000 per week mark, and the followership was being lost as well, lowering below the 100,000 mark by may 2014. the only type of posts that received any engagement at all were the posts that portrayed photos containing military content – which were completely free to view by all page fans – while any post that was a part of the campaign effort to generate funding to aid disabled veterans was nearly completely ignored. by july 2014, facebook analytic figures show engagement levels below the 5,000 mark per week, and by august, the page was shut down. military photography and carbine hq were subsequently dissolved and the assets divided amongst the two cofounders. the two cofounders went their separate ways – one founder retained ownership of the military photography facebook page – and after several months, by october of 2014, the page was restarted and today has rebuilt a fan base of more than 110,000 followers. discussion: the original military photography facebook page existed for 15 months, with a cohort of 118,000 individuals worldwide, and consistent interaction figures averaging 500,000 per week. this indicates that the 118,000 cohorts shared in the premise of military photography as a whole, and identified with military photography to a large extent. yet, as soon as military photography initiated its campaign to provide meaningful support for veterans through the use of items for sale, both the interaction levels and the total cohort numbers began to decline. loss of interaction numbers can be directly attributed to individual wall posts on the military photography facebook page. posts that had anything to do with the marketing campaign were largely ignored, and therefore, every time a post that contained information about the campaign or its mission to provide support for veterans was posted on the facebook page, a loss in interaction numbers resulted. yet, while that may indicate indifference by the cohort towards the campaign, the fact that said cohort were yet still members of a page dedicated to showcasing all facets of the world’s militaries would naturally lead to the belief that the cohort actually did care about veterans to some extent. after all, why become a fan of a page dedicated solely toward showcasing the world’s armed services if you didn’t support veterans? that question therefore brings us to the next problem the page faced, a loss in followership. the analytic numbers showed a slightly fluctuating but steady decrease in followership over time, starting with the launch of the marketing campaign. with each new campaign post, the total number of fans who were members of the page decreased as a result. not only were fans ignoring the posts, many ended up leaving the page to avoid seeing them. this would be consistent with bennett’s (1998) study on social 2 indifference towards political affairs and political media whereby individuals actually went out of their way to avoid coming into contact with the subject of indifference. social indifference directed at disabled veterans then can be isolated when social rules and symbolic self-completion are applied. what are seen in the numbers from military photography are 118,000 individuals who wanted free access to that visual media, but didn’t actually care about the intent to help disabled veterans. once posts began marketing items for sale in a campaign specifically designed to provide monetary support for disabled veterans, interaction numbers decreased and followership slowly declined. the cohort ignored posts that had anything to do with the campaign to provide funding for disabled veterans and many left the page purposely to avoid exposure to it. 2) military media: may – july 2014 introduction: started on the 1st of may 2014, military media was a paid subscription news service that focused exclusively on military related news and media from around the world that featured an international team of service members, veterans, and experienced media professionals as contributors. the monthly subscription fee was $5 usd/aud/cad/eur per month; however, the real purpose behind the service was its selling point. forty percent of all profit was donated directly to veterans charities, and research organizations that provided services for disabled veterans and research for pts(d). military media was hosted by the internet based independent news service provider beacon, and in order for military media to launch and officially receive financial funding from beacon, the service needed 800 individual subscriptions by june 30th, 2014. military media was a subsidiary under the control of carbine hq and controlled by the same eight veteran and media professionals that staffed military photography, while a contributing staff consisted of a worldwide network of veterans and media professionals who lived in key locations around the world. military media was created as a response to the increasing demand from military photography fans. the fans consistently sent in emails to military photography’s account expressing the desire for more content, more writing, more stories, etc. however, the international administrators for military photography were all volunteers and did not have the time, resources or funding to provide the fans the amount of content they asked for. military media was created to answer that demand. the team would start the subscription-based media service and still provide 40 percent of all subscription fees directly to support veterans charities. the sales campaign military photography had already tried for a year was a failure, but military media gave the fans exactly what they wanted while still fulfilling the original mission of providing continuous funding to veterans charities. since military media was controlled by the same staff as military photography, the new service already commanded a 118,000 strong cohort from which to market to, and to which a large portion had previously already expressed interest in. as the !cusj%|%volume%9%|%spring%2015% 20% 1 service gained subscriptions, the military photography page would only be used for minimal postings and teaser posts to continue to gain traffic to the military media site. since the service was now subscription based, individuals had to be paying members in order to gain exclusive access to everything military media had to offer, to which 60 percent of all profit went to cover costs of operations, while 40 percent went directly to support veterans charities. results: campaign efforts to gain the minimum 800 subscriptions, which were required in order to receive funding, started with the initial launch of the project on may 1st. the team had used the data from military photography’s facebook analytic profile to target the military photography audience towards the military media site, and subsequently subscribe to the service. advertising remained consistent throughout the 61-day period and used posts to the military photography page that maintained an open and honest approach in marketing the new service. the ads fully disclosed the reasons military media was created, the fact that the service was new, and the requirement to gain a minimum of 800 initial subscriptions before the official launch of the service would take effect on july 1st, 2014. yet the largest aspect of the advertising campaign remained the fact that the service would donate 40 percent of all profits directly to veterans charities. since the military photography fan base consistently sent the pages administrators messages asking for more photos, more content, and more military related news and stories, and due to the fact that military media was specifically designed to provide exactly that, the team believed the initial sign up would take only a week. the adds for the military media campaign consistently maintained a reach of over 500,000 weekly, this gave the campaign effort an estimated 4 million reach over the entire 61-day period. however, the interaction figures consistently remained below 1,000 per week, and after 61 continuous days of campaigning efforts with a target audience of 118,000 cohorts, only 34 individuals signed up. as a result, military media never reached the initial goal of 800 subscriptions, the service failed before it ever started, and was shutdown by july 1st. discussion: over a 61-day period, 34 individuals signed up out of a total campaign reach of 4 million people, equating to a subscription percentage of 0.0000085% and giving the campaign a subscription rate of 1 person per 117,647 people. if the percent of engagement were restricted to the 118,000 cohorts to whom the service was marketed directly towards, then that figure would become 0.000288, and the campaigns success rate in receiving meaningful support from the target audience would become 1 in 3,470. this would logically represent a more accurate figure, however, it is not possible to assume those numbers are restricted to the 118,000 cohorts because interactions that resulted in a sign-up for the service could have come from outside of that 118,000. these figures are on the extreme side of being representative of the als ice bucket challenge figures. in fact, if the 2 “faddish” aspect of the als ice bucket challenge might never have occurred, the 1 in 40 estimate made earlier might actually end up closer to the 1 in 3,470 rate shown in this case. this would be due to either removing or limiting the role of symbolic self-completion. as individuals would no longer feel pressured to conform to social pressures forced on them by the popularity of the challenge, the byproduct of slacktivism would be removed as a result. this is exactly what is being seen here with the results from military media. the service was marketed towards the military photography cohort, who expressed overwhelming social indifference towards it. however, because it was never popular to begin with, and there was no added pressure to participate, it was easy for individuals to get around the social rules. individuals were able to preserve their identity confirmation by either ignoring the campaign or leaving the cohort all together. furthermore, because symbolic selfcompletion did exist within the military media campaign (due to the small numbers of likes and share’s) it shares the same model as the als ice bucket challenge, and indifference is therefore able to be isolated and confirmed as the predominate cause for the services ultimate failure. the service was marketed as a way to provide what the cohort wanted and used to help veterans as a result. however, the cohort never cared about the service to begin with. after all, why pay for something if you think you can already get it for free? the military media case is a byproduct itself from the military photography case. while these two cases do work together to show indifference existed from a large cohort and that indifference was directed towards the campaigns, at this point, there is only a loose connection established between social indifference and veterans. it very well could be the case that the social indifference expressed was towards the idea of paying for a list of unpopular products offered by military photography, or for a media service from military media that provided a more enhanced version of the free service already provided by military photography. yet, in both cases all marketing to the cohort made it clear that the products and the services were designed to help provide continuous funding to support disabled veterans; and in both cases, indifference has been isolated as the primary cause for ultimate failure. therefore, it is necessary to use additional independent case studies to test whether that connection is consistent. through the use of several other independent cases, each of which also places the provision of monetary funding support to help veterans as their top priorities, combined with the isolation of indifference in each case as the primary cause for failure, it is possible to confirm generalizable social indifference exists towards veterans, veterans social issues, and providing meaningful support for veterans causes. as such, three additional independent social media-based veterans charity campaign case studies, that do not share any relationship outside of the single aspect of campaigning to provide monetary support for veterans, will be used to further isolate social indifference in connection with veterans as a whole, and ultimately test for the presence of generalizability. !cusj%|%volume%9%|%spring%2015% 21% 1 3) the words for warriors collection: book 1, “13 years of service”: november 2013 – present introduction: the words for warriors project was created for the purpose of generating monetary funding for disabled veterans charities through the sale of a book called “13 years of service: my personal collection of military bizarreness”, written by retired german army master sergeant per m. griebler. the book was a collection of short stories – reflective personal accounts from master sergeant griebler’s time in service – of a humorous nature, written with the intent that the book could be read randomly and spontaneously, and to which remained relatable for service members around the world. the book also contains a large number of small personal quotes from many of master sergeant griebler’s personal military friends – from several different nations’ military services – that he has met throughout his time in service. the book was written over a period of one year, from early 2012 until the summer of 2013. it was published on november 24th, 2013, and made available worldwide via amazon the same day. marketing for the book focused on the projects primary purpose. eighty percent of all proceeds from each sale went directly to support a small number of disabled veteran’s charities in the united states, germany, and australia. today, the book is still available for sale worldwide on amazon in nearly every major country, is available in both english and german, and holds a 5 star rating. the paperback version is listed for $9.95 usd, and the kindle version is listed for $4.49 usd. results: the words for warriors project campaign to market the book started on november 24th, 2013, and still continues to this day. the campaign started a facebook page dedicated to marketing the book, and by august of 2014, was absorbed under the umbrella of ruck headquarters – a veteran owned small business – in order to continue operating. in 13 months – starting from december 1st, 2013, through december 31st, 2014 – the words for warriors project campaign gained a total reach of 23,082. during that same period, however, interaction numbers totaled 3,090, and a total of 298 books have been sold worldwide (appendix a). discussion: 298 books sold out of a total reach of 23,082 equals a success rate of 0.0129. if that number were restricted to the interaction figures only, then the success rate would increase to 0.096. from the time the book was first published, until december 31st, 2014, 13 months later, 1 in 7.4699 was willing to offer some form of token support while only 1 in 77.456 was willing to buy the book. since the book was about veterans short stories relatable to other veterans, it is possible to say that any individual who expressed interest in the book – through either token support or by buying the book – might relate to its content. this would indicate that at least 3,090 individuals out of the total who saw the campaign found the book directly relatable to some extent. at the same time, the book itself was only marketed on a facebook page and exclusively focused on 2 the books purpose of generating monetary funding for veterans charities. this would indicate that the entire 23,082 cohorts remain consistent with the examples found in both the military photography and military media cases. next, since the marketing campaign focused exclusively on informing the potential buyer that the books purpose was to raise money for veterans charities, it is possible to rule out a misinterpretation as to the intention of the books purpose for sale. the total cohort was introduced to the book as both a military relatable subject and a way to help support veterans charities by purchasing it, because the campaign made explicit mention that 80% of all proceeds went directly to fund charities that supported disabled veterans. the figures indicate the audience maintained an interest in the book as a relatable subject only, but were overwhelmingly uninterested in offering any amount of meaningful monetary support for veterans charities by refusal to purchase the book. finally, since the entire campaign was carried out over internet based social media platforms, it is possible to rule out a lack of financial ability to afford a book under $10 usd. like the previous cases, all interactions for this campaign came from social media users who could obviously afford internet access in one way or another. it would not be appropriate to assume that all those who maintain some form of internet access suffer from financial difficulties either. therefore, given the evidence based on percentages of interactions versus meaningful support, it is safe to say that the majority of people who saw the campaign ultimately were not motivated to follow-up with meaningful support. this indicates indifference towards the cause because if the individuals did care about it, they would have been motivated to offer meaningful support to some degree regardless of personal circumstances. as seen in the previous cases, as well as in the control case, social rule systems theory, combined with the presence of slacktivism and symbolic self-completion, make it clear that individuals only provided token support simply to avoid the loss of their self-image or to gain a positive self-image at the confirmation of third parties. even though the entire cohort possessed enough financial capabilities to afford an internet connection of some kind, the overwhelming vast majority refused to offer meaningful support for a campaign they fully understood was purposely designed to provide monetary support for veterans charities. these factors allow for the removal of other possible contributing possibilities for refusal to provide meaningful support, leaving only social indifference behind as the largest contributing factor to the failure of the campaign itself. 4) ruck headquarters – operation nijmegen: july 2014 introduction: operation nijmegen is an annual non-profit ruck marching event whereby ruck headquarters staff and affiliates participate in the vierdaagse (4 days) event – a 100+ mile ruck march that takes place over a period of 4 days and is held every july in the city of nijmegen, the netherlands !cusj%|%volume%9%|%spring%2015% 22% 1 – for the purpose of raising funding for disabled military veterans charities, and to promote awareness of veterans disabilities, pts(d), and veterans organizations and programs. the event is both recorded and streamed live – via social media platforms – with the ruck headquarters team conducting interviews with fellow veterans from around the world who are also participating in the march, and at the same time, openly discussing veterans issues, charities, and organizations, all while the team completes the march themselves. ruck headquarters staff and affiliated personnel have participated in the march since 2010 in order to raise monetary donations for veterans charitable organizations. results: from june 1st, through august 31st, 2014, the operation nijmegen 2014 campaign gained a total reach of 1.969 million individuals worldwide via facebook using several facebook pages – ruck hq, military photography, vixens 4 veterans, enlist me, and words for warriors – as campaign platforms in order to reach a larger cohort. the majority of those individuals reached were from north america, europe, and australia, and in order to maintain interest within those targeted areas, specific veterans charities that already maintain a large presence within those areas were chosen for both information and interest posts, as well as for the ultimate donation of collected funds. during that period, the campaign had a total of 354,036 individual interactions – including shares, and likes – and finished with a total donation figure of $1,020 usd raised over the crowd funding platform go fund me, from a total of 16 individuals, with 3 of those individuals making repeat donations throughout the period. (appendix b) discussion: using the numbers from the analytic figures alone, 16 donors out of a total reach of 1.969 million equals a success rate of 0.000008126. if the numbers were limited to interactions only, the rate then comes to 0.1798. that means that 1 out of every 5.56 individuals who saw the campaign offered up some form of token support, while 1 out of every 123,062.5 was actually willing to offer up meaningful support. the campaign was only marketed via internet social media platforms and services. operation nijmegen 2014 maintained a presence on facebook, twitter, youtube, and the ruck hq website, however, all marketing was limited to the ruck hq facebook page alone, and shared across other military and veteran facebook themed pages from there for the purpose of increasing the campaigns total reach. additionally, the only facebook pages used for all shared marketing were those that exclusively focused on military content and veterans support causes. since the entire 1.969 million analytic figures came exclusively from the ruck hq facebook page, it is safe to assume that all those who interacted with the campaign found it relatable to some degree. these numbers therefore remain consistent with the examples found in both the military photography and military media cases. next, since the campaign focused exclusively on informing 2 all viewers of its intended purpose – to raise money for veterans charities – it is possible to rule out a misinterpretation as to the intention of the campaign itself. the total cohort was introduced to operation nijmegen 2014 as a way to help support veterans charities by encouraging monetary donations, all of which went directly to fund charities that supported disabled veterans. the figures indicate the audience maintained an interest in operation nijmegen 2014 as a subject, but were overwhelmingly uninterested in offering any amount of meaningful monetary support for veterans charities by refusal to donate. finally, as seen with previous cases, since the entire campaign was carried out over internet based social media platforms, it is possible to rule out a lack of financial ability to make a monetary donation of even $1 usd for the purpose of supporting veterans charities. like the previous cases, all interactions for this campaign came from social media users who could obviously afford internet access in one way or another. it would not be appropriate to assume that all those who maintain some form of internet access suffer from financial difficulties either. therefore, given the evidence based on percentages of interactions versus meaningful support, it is safe to say that the majority of people who saw the campaign ultimately were not motivated to follow-up with meaningful support. this indicates indifference towards the cause because if the individuals did care about it, they would have been motivated to offer meaningful support to some degree regardless of personal circumstances. as seen in the previous cases, as well as in the control case, social rule systems theory, combined with the presence of slacktivism and symbolic self-completion, make it clear that individuals only provided token support simply to avoid the loss of their self-image or to gain a positive self-image at the confirmation of third parties. even though the entire cohort possessed enough financial capabilities to afford an internet connection of some kind, the overwhelming vast majority refused to offer meaningful support for a campaign they fully understood was purposely designed to provide monetary support for veterans charities. these factors allow for the removal of other possible contributing possibilities for refusal to provide meaningful support, leaving only social indifference behind as the largest contributing factor to the failure of the campaign itself. 5) vixens 4 veterans 2015 wwii style pin-up calendar: 2014 introduction: the vixens 4 veterans project was a disabled veterans charity monetary funding campaign designed by mark greenmantle of mark greenmantle photography, and co-creator brittany jean of patriot pinup, whose aim was to develop and produce a full size 2015 wwii style pin-up calendar (featuring famous wwii aircraft) that would be marketed and sold worldwide for the purpose of generating monetary funding to help support disabled veterans charities that assisted disabled veterans, but more specifically aided those suffering from pts(d). initial planning for the project started in december 2013 and the projects production lasted !cusj%|%volume%9%|%spring%2015% 23% 1 10 months until october 1st, 2014. the vixens 4 veterans facebook page was setup on january 16th, 2014, and project development, planning, contacts list, model talent casting, and photo shoot location development took place from february 2014 through june 2014. the photo tour initiated in mid-june through mid-july in australia, and then took place across the united states from july 21st through august 19th. final production took place from august 22nd through official launch for sale on october 1st, 2014. the marketing point of the project was specific to its end goal. one hundred percent of all profits went directly to support a small number of disabled veteran’s charities in the united states, and australia, each of which specifically aided veterans with pts(d). today, the vixens 4 veterans 2015 calendar is still available for sale worldwide via woocommerce on the vixens 4 veterans website for $20 usd. results: as of december 31st, 2014, more than 500,000 individuals have seen the vixens 4 veterans 2015 calendar campaign, which still continues to this day. over a 12-month period – starting from january 1st, 2014, through december 31st, 2014 – the vixens 4 veterans campaign gained a total organic reach of 689,040. during that same period, however, total interaction numbers gained only reached 69,685. from the calendar’s official launch date for sale on october 1st, 2014, through the end of the year on december 31st, a total of 78 calendars have been sold worldwide. (appendix c) discussion: 78 calendars sold after a total reach of 689,040 equals a success rate of 0.000113. if the sales figures were restricted entirely to the interaction numbers, the rate would increase to 0.001119. out of the total reach for the project, only 1 in 9.88 was willing to offer some form of token support for the project, and only 1 in 8,722 was willing to actually buy a calendar for the purpose of supporting veterans. again, figures such as the ones represented here remain consistent with all previous cases as the success rate indicates a clearly disproportionate level of interaction versus the total amount of meaningful support gained. next, since the campaign focused exclusively on informing all viewers of its intended purpose – to raise money for veterans charities – it is possible to rule out a misinterpretation as to the intention of the campaign itself. the total cohort was introduced to buying the campaign as a way to help support veterans charities by encouraging all whom saw the campaign to buy the calendar. the campaign made it clear that 100% of all profit went directly to fund charities that supported veterans with pts(d). the figures indicate the audience maintained an interest in vixens 4 veterans only so far as the content it provided via its facebook page, but were overwhelmingly uninterested in offering any amount of meaningful monetary support for veterans charities by refusal to buy the calendar. finally, as seen with all previous cases, since the entire campaign was carried out over internet based social media platforms, again it is possible to rule out a lack of financial ability to offer meaningful support for the cause. like the 2 previous cases, all interactions for this campaign came from social media users who could obviously afford internet access in one way or another. it would not be appropriate to assume that all those who maintain some form of internet access suffer from financial difficulties either. therefore, given the evidence based on percentages of interactions versus meaningful support, it is safe to say that the majority of people who saw the campaign ultimately were not motivated to follow-up with meaningful support. this indicates indifference towards the cause because if the individuals did care about it, they would have been motivated to offer meaningful support to some degree regardless of personal circumstances. as seen in the previous cases, as well as in the control case, social rule systems theory, combined with the presence of slacktivism and symbolic self-completion, make it clear that individuals only provided token support simply to avoid the loss of their self-image or to gain a positive self-image at the confirmation of third parties. even though the entire cohort possessed enough financial capabilities to afford an internet connection of some kind, the overwhelming vast majority refused to offer meaningful support for a campaign they fully understood was purposely designed to provide monetary support for veterans charities. these factors allow for the removal of other possible contributing possibilities for refusal to provide meaningful support, leaving only social indifference behind as the largest contributing factor to the failure of the campaign itself. discussion: the data and discussion from each case study is able to accomplish the removal of factors other than indifference through the application of several different interconnected social theories that all share indifference as a common factor combined with data analysis to show commonality between individual cases. analysis of the data shows consistent refusal to the provision of meaningful support in every case, while the application of sociological theories (showing signs of either the presence of them or lack of their presence for specific reasons) shows that indifference remains consistent across all cases. due to the fact that each case maintains only one similarity between the others (campaigns to generate monetary support for veterans charities), and that all cases coincide directly with the results of the control, it is possible to say that generalizable social indifference not only exists, but is targeted directly towards the subject of veterans as a whole. conclusions: the results of this study prove that generalizable social indifference does exist directed towards veterans, and therefore, the answer to the question “are veterans subjected to social indifference” is, yes. not only are veterans subjected to social indifference on a large scale, but also, that indifference encompasses the subject of veterans as a whole. the term “society” itself does not necessarily imply that every single member within remains the same, however, it does imply the overwhelming majority, and as such, generalizability can be applied due to consistency. the reality of the issue is, that – as stated earlier military service !cusj%|%volume%9%|%spring%2015% 24% 1 members frequently do jobs that are arguably the most dangerous jobs in the world, and war is physically, psychologically, and spiritually damaging. veterans who are injured or wounded – physically or psychologically or both – as a result of their service, will suffer from physical and/or psychological disabilities that directly impact their quality of life. as a result of their service to a nation, veterans consistently face a lifetime of issues unique to veterans. however, the society itself simply does not care about its veterans, the life issues veterans face, and are indifferent towards veterans as a whole. state of the social issue: there exists a very serious problem within society, a culture full of empty words and zero action to back them up. how many times have we heard the words “thank you for your service” or “we support the troops”? these words have become commonplace throughout the public forum. there is no end to the number of people who, if they came across a veteran or current service member, would offer a handshake followed by these words. however, “thank you for your service” and “we support the troops” have become the source of a great amount of hatred throughout the veterans communities. after hearing these two phrases for more than a decade now, many veterans are filled with a silent rage if their utterance came from anyone except another veteran. veterans hate those words because many feel they are empty and meaningless. as the research from fishbach, et al. (2006) points out; members of society are expected to say those things. there exists a certain social-psychological rule that states, if you come in contact with a veteran, you must offer “support” of some kind. (wicklund and gollwitzer, 1981) to not do it, would be more than unacceptable, and it will be followed by social backlash of varying degrees. (fishback, dhar, and zhang, 2006) following the social rules then, the average person would offer up the bare minimum forms of token support, while at the same time, refusing to offer any form of meaningful support, because this allows the individual to follow the social rules while giving as little as possible. (kristofferson, white and peloze, 2014) therefore, the actions the average person takes upon encountering a veteran or current service member are largely motivated by making themselves look good to the eyes of all other bystanders and said veteran or current service member. (fishbach, et al., 2006) it is considered socially acceptable to be proud of a veterans service; likewise, it is therefore not socially acceptable to not be thankful for their service to the nation. according to this rule, society should be thankful and proud of its veterans, and as such, there exists the very first set of a long list of underlying social obligations. to refuse to acknowledge and be respectful and thankful for a veteran’s service is a social crime. it’s expected that you offer up something to show that thanks, but given the chance, the average person would never offer a single thing of value if they could get away with it. the need to conform to the social rules – due to the fear of image loss and/or 2 social backlash that comes from refusal is where slacktivism was born. slacktivism offers society a way to confirm to itself that they care about social issues – even if they don’t – through third party confirmation and self-completion, thereby obeying the social rules in the process. it is because veterans are aware that society in general really does not really care about them that “thank you for your service” and “we support the troops” is so hurtful to veterans on a personal level, and why veterans have grown to hate token support for veterans and veterans issues so much. this study improves upon the current understanding of social indifference by providing a method of isolation and to show how indifference directly impacts a social issue. at the same time, it also provides ground level work for further studies on both direct and indirect effects caused as a result of social indifference towards the subject. for example, the possibility that social indifference might be tied to potentially damaging effects, either direct or indirect, to related social issues. social indifference may act as a contributing factor for compounding veterans issues such as homelessness, unemployment, lack of public education/understanding of veterans and veterans issues, lack of public willingness to self-educate, veterans suicide, attitudes, psychological factors, etc. as such, the understanding of the damaging effects of social indifference is key to understanding the state of the social issue itself. therefore, this study’s contribution may find importance in the development of new methods to confront the social issue and identify possible methods to prompt social change. state of the science review: today, hundreds of organizations and charitable causes exist whose missions are to help and aid veterans through the use of services and legislation that directly impacts veterans issues. there exist a large array of social issues unique to veterans, many of them revolving around physical and psychological disabilities that impact a veteran’s quality of life for a lifetime. these organizations and charities exist to help with these issues, and often the only way for them to operate is through the use of meaningful support. yet meaningful support is consistently rare and difficult to obtain. after careful review of past figures, it seems as if social indifference may be the root cause, however indifference is difficult to prove because any number of other contributing causes may ultimately be the largest contributing factor. as such, in order to prove social indifference exists, indifference must be isolated as a factor so as to show it is the most predominate contributing factor to the end result, therefore proving its existence as a root cause. to do this, the use of social rule systems theory, combined with supporting evidence from theories like symbolic self-completion and slacktivism, and case study analysis covering a number of real-life social campaigns and veterans support projects, allows for the identification of a number of underlying social factors that indicates a direct link to social indifference. this multipronged approach therefore isolates indifference as a primary cause. this process is then applied in several independent case studies in an attempt to discover if indifference was the root cause in each case. the confirmation !cusj%|%volume%9%|%spring%2015% 25% 1 of indifference in the cases would thus confirm generalized social indifference exists towards veterans issues and veterans as a whole. the adherence to social rules is a complicated affair to say the least. social rules are constantly evolving, and many may even become contradicting as a result, however, research leaves little doubt that social rules are at the source of our actions and define our behavior as a society. understandably then, social rules play a large part in the creation of certain social phenomena like symbolic self-completion and slacktivism. individually, people may be indifferent to a particular social issue – for whatever reason – yet because it may be socially unacceptable for the individual to not care about a cause, they are forced to find ways to obey the social rules without actually obeying the social rules. therefore, we can safely conclude that indifference is the root cause behind “thank you for your service” and “we support the troops” (token support), because social rules obligated individuals to offer such forms of support publically for the purpose of either personal gain, or to prevent a loss of some kind (public scrutiny). slacktivism provides that loophole, allowing indifferent individuals to affirm publically that they followed the social norm and making themselves believe they adhered to the rules (through self-completion), while at the same time, doing as little as possible in the process. after understanding how this process is possible, generalizability is easy to confirm simply by running the numbers tied to meaningful support versus token support received for a number of social causes. unfortunately, the reality is that you simply cannot force an individual to care about a particular social issue or cause. this means that combating social indifference – even after identifying it as the root cause – may be close to impossible. the only way might be to change the social rules themselves, yet social rules are constantly evolving on their own, and there is no guarantee that efforts to change them for the benefit of a particular social issue will be successful. is a society truly proud of its veterans? how do you measure such a question? the easiest way to find out would be to measure the amount of meaningful support. individuals that truly care about something will take great care of it. as such, does society support its veterans in a meaningful way? the research would strongly suggest that the majority does not. individuals would rather not provide support of any kind simply because they could not care less about veterans or the issues veterans face as a result of war, yet since social rules say such actions are “taboo”, indifferent individuals turn to slacktivism and symbolic self-completion as loophole methods to get around that social taboo. 2 references: • als association (2013). return of organization exempt from income tax 2013. alsa.org. • als association (2014). the als association debunks fake news article that went viral. alsa.org. • als association (2014). financial information. alsa.org. • als ice bucket challenge (2014). charlie sheen als ice bucket challenge. youtube.com. • bennett, s. e. (1998). young americans’ indifference to media coverage of public affairs. ps: political science and politics, 31(3), 535-541. • burns, t. r., and flam h. (1987). the shaping of social organization: social rule system theory with applications. london: sage publications. • calderbank, r. (2000). abuse and disabled people: vulnerability or social indifference? disability & society, 15(3), 521-534. • casarett, d., et al. (2008). important aspects of end-oflife care among veterans: implications for measurement and quality improvement. journal of pain and symptom management, 35(2), 115-125. • diamond, d. (2014). the als ice bucket challenge has raised $100 million – and counting. forbes.com. • dobie, d. j., et al. (2004). posttraumatic stress disorder in female veterans association with self-reported health problems and functional impairment. jama internal medicine, 164(4), 394-400. • fishbach, a., dhar, r., zhang, y. (2006). subgoals as substitutes or complements: the role of goal accessibility. journal of personality and social psychology, 91(2), 232-242. • flam, h., and carson, m. (2008), rule system theory: applications and explorations. new york: peter lang. • giddens, a. (1984). the constitution of society. oxford: polity press. • goffman, e. (1974). frame analysis: an essay on the organization of experience. cambridge, mass.: harvard university press. • gollwitzer, p. m., sheeran, p., michalski, v., and seifert, a. e. (2009). when intentions go public. does social reality widen the intention-behavior gap? association for psychological science, 20(5), 612-618. • harré, r. (1979). social being. oxford: blackwell. • kristofferson, k., white, k., and peloza, j. (2014). the nature of slacktivism: how the social observability of an act of token support affects subsequent prosocial action. journal of consumer research, 40(6), 1149-66. • lotman, j. (1975). theses on the semiotic study of culture. lisse, netherlands: peter de ridder. • mclean, k. c., and pratt, m. w. (2006). life’s little (and big) lessons: identity statuses and meaning-making in the turning point narratives of emerging adults. developmental psychology, 42(4), 714-722. • military photography (2014). military photography. ! !cusj%|%volume%9%|%spring%2015% 26% 1 • mirowsky, j., ross, c. e., and willigen, m. v. (1996). instrumentalism in the land of opportunity: socioeconomic causes of emotional consequences. social psychology quarterly, 59(4), 322-337. • nghe l. t., and mahalik, j. r. (2001). examining racial identity statuses as predictors of psychological defenses in african american college students. journal of counseling psychology, 48(1), 10-16. • riggs, d. s., byrne, c. a., weathers, d. s., and litz, b. t. (1998). the quality of the intimate relationships of male vietnam veterans: problems associated with posttraumatic stress disorder. journal of traumatic stress, 11(1), 87-101. • ruck headquarters (2014). operation nijmegen 2014. • ruck headquarters (2014). ruck headquarters. • sayer, n. a., et al. (2010). reintegration problems and treatment interests among iraq and afghanistan combat veterans receiving va medical care. psychiatric services, 61(6), 589-597. • stampler, l. (2014). this is how many ice bucket challenge videos people have posted on facebook. time.com. • steel, e. (2014). ice bucket challenge donations for a.l.s. research top $41 million. nytimes.com. • townsend, l. (2014). how much has the ice bucket challenge achieved? bbc news magazine. • vixens 4 veterans (2014). vixens 4 veterans 2015 calendar. • vixens 4 veterans (2014). vixens 4 veterans. • waterman, a. s., and goldman, j. a. (1976). a longitudinal study of ego identity development at a liberal arts college. journal of youth and adolescence, 5(4), 361-369. • wicklund, r. a., and gollwitzer, p. m. (1981). symbolic self-completion, attempted influence, and self-deprecation. basic and applied social psychology, 2(2), 89-114. • wolfe, j., et al. (1999). relationship of psychiatric status to gulf war veterans’ health problems. psychosomatic medicine, 61(4), 532-540. • woodbine, g. f. (2002). moral choice in an agency framework and related motivational topologies as impacted by personal and contextual factors for financial institutions in china. curtin university school of accounting. • words for warriors (2014). the words for warriors collection: book 1 – 13 years of service: my personal collection of military bizarreness. • words for warriors (2014). words for warriors. ! 2 appendix: a) facebook insights data export for words for warriors: http://www.facebook.com/wordsforwarriors.org (available upon request from journal) a. first column represents the “date” b. second column represents “daily engagement: the total number of people who engaged with the page. engagement includes any click or story created. (unique users)” c. third column represents “daily reach: the total number of people who have seen any content associated with the page. (unique users)” b) facebook insights data export for ruck hq – operation nijmegen: http://www.facebook.com/ruckhq (available upon request from journal) a. first column represents the “date” b. second column represents “daily engagement: the total number of people who engaged with the page. engagement includes any click or story created. (unique users)” c. third column represents “daily reach: the total number of people who have seen any content associated with the page. (unique users)” c) facebook insights data export for vixens 4 veterans: http://www.facebook.com/vix4vets (available upon request from journal) a. first column represents the “date” b. second column represents “daily engagement: the total number of people who engaged with the page. engagement includes any click or story created. (unique users)” c. third column represents “daily reach: the total number of people who have seen any content associated with the page. (unique users)” d) woocommerce sales report data export for vixens 4 veterans: http://www.facebook.com/vix4vets (available upon request from journal) a. first column represents the “month” b. second column represents “the total number of items sold during same month” microsoft word cusj 04.docx cusj 2019________________________________________________________________________________ __________________________________________________________________________________________ 4 abstract — amazonian tropical forests are critical to global carbon cycling and sequestration, and in direct danger from deforestation. in order to contribute to limited existing literature on the carbon sequestration potential of secondary forest ecosystems and their aboveground biomass (agb), we established permanent 0.2-ha plots in a primary and a secondary forest near iquitos, loreto, peru. we measured diameter at breast height (dbh), tree height, and wood density for trees ≥10 cm dbh and took the diameter of lianas at 30 cm shoot extension, then used published allometric equations to estimate agb and compare it between forests. trees within the primary forest plot had a significantly greater mean dbh and higher mean wood density, as well as a greater overall agb than trees within the secondary forest. agb was calculated to be 322.05 mg/ha for the primary forest and 51.17 mg/ha for the secondary forest. sequestered carbon quantities were 151.36 mg/ha and 24.05 mg/ha, respectively. higher estimates of stored carbon within the primary forest are attributed to oldgrowth trees with large dbh values and increased wood density, and discrepancies between our carbon estimates for the secondary forest and past estimates for the same site suggest the need to focus more research and attention on allometric equation use. the results of this study provide a potential incentive for carbon sequestration funding to be awarded to the primary forest property studied and establish a foundation for future estimations of the carbon storage capacities of tropical secondary forests. i. introduction tropical forests play a fundamental role in the global carbon cycle by sequestering up to 40% of the world’s terrestrial biomass-bound carbon (pan et al. 2011). more than half of the carbon in tropical forests is found in the neotropics (ngo et al. 2013). however, the carbon storage potential of these forests can be truncated by deforestation and ecological degradation. an estimated 1.6 to 2.4 pg of carbon is released into the atmosphere from tropical forest clearing each year, representing 20-29% of global anthropogenic carbon emissions (naughton-treves 2004). in peru, forests are being cleared at a rate of 0.4% per year, releasing enormous amounts of stored carbon (naughtontreves 2004). after a forest has been disturbed, it will regenerate as a secondary forest and accumulate carbon in its aboveground and belowground tissues as it grows. lightdemanding ruderal species grow quickly and form wood with lower specific gravity than later successive, shadetolerant species that require time to establish and are not a significant fraction of secondary forest composition (ramanantoandro et al. 2016, gao et al. 2015, king et al. 2006). wood density is directly related to above ground biomass (agb), and therefore secondary forests tend to have less accumulated biomass than primary forests. carbon storage is often estimated as half of total agb (day et al. 2013, chave et al. 2005), so it can be assumed secondary forests store less overall carbon than primary forests. the proportion of tropical forests that are secondary is projected to increase continually due movement of populations toward urban centers (thomlinson et al 1996). abandonment of cleared sites may also contribute to increased secondary forest cover. this may be the case in regions of the amazon whereland-use practices continue to be characterized by reliance upon fallow periods (e.g. marquart et al. 2013, naughton-treves 2004). though secondary forests contain an estimated 60% less biomass and 19% less total stored carbon than primary forests (ngo et al. 2013), they exhibit slightly higher regeneration rates. naughton-treves (2004) found that the growth rate of secondary forests in the peruvian amazon was as high as 11.47 mg/ha per year, compared to chave et al. (2001)’s estimate of primary forest growth of 2-4 mg/ha per year, indicating that tropical secondary forests should not be overlooked for their carbon storage potential. significant attention in the literature has been given to the c storage abilities of tropical primary growth forests (chave et al. 2001, baker et al. 2004, ramanantoandro et al. 2016). to date, however, there has not been a comparable number of studies on the c storage potential of regenerating secondary forests in tropical areas. this is also the case in amazonia specifically, partially because much of the existing literature on amazonian agb has been compiled based only on trees ≥ 10cm dbh, and secondary forests are often excluded from these calculations entirely, due to their low abundance of large trees (e.g., baker et al. 2004, chave et al. 2001). further study of secondary forest carbon storage will help ensure they are properly valued for their ability to offset the effects of climate change and protected from further deforestation. proper evaluation of carbon storage in both secondary and primary forests could also improve the ability to use redd+ incentives to protect tropical forests. redd+ stands for “reducing emissions from deforestation and degradation” and aims to develop policies that add high financial value for carbon stored in intact forests (kricher 2011). it serves as an international incentive to reduce deforestation. to help offset global carbon emissions, redd+ is focused on maintaining carbon storage within aboveground biomass and carbon storage capacities of a western amazonian primary and secondary growth forest lindsay finnell1,3 and bailey j. nordin2,3* 1 drake university, des moines, ia, usa lindsay.finnell@drake.edu� 2 columbia university in the city of new york, ny, usa bjn2122@columbia.edu 3 center for amazon studies school for field studies, loreto, peru ________________________________________________________________________________ cusj 2019 __________________________________________________________________________________________ 5 the tropical forests of developing countries by reducing land use change of natural ecosystems (pan et al. 2011). forest concessions can apply for redd+ funding based on calculations of their total carbon storage capacity. to ensure the equitable distribution of redd+ funds, it is imperative that the carbon storage of different ecosystems is accurately estimated. in this study, we estimated and compared agb and carbon storage capacities for a section of primary terra firme forest and a section of secondary terra firme forest in the northeastern peruvian amazon (loreto, peru). with the objective of refining carbon estimates and improving the local understanding of land-use history, we established a permanent plot system to measure biomass at two locations: the universidad cientifica del peru conservation concession (ucp), a primary forest sites, and sfs center for amazon studies (cas), a secondary forest site. we also hope to contribute to the information bank needed to garner redd+ funding at these two sites. given that the trees at the two sites are found within a primary and secondary forest of different ages with significant differences in species composition, we expected that the agb of each site would reflect these differences. we predicted that the trees at ucp would have greater dbh and greater average height than those at cas. since faster growing, pioneer tree species found more commonly in secondary forests disturbed recently like those at cas may have less dense wood than older, slower-growing trees in the primary forest at ucp, we also predicted greater wood densities and more overall agb to be found at ucp. ii. methods study site we established permanent forest plots at two locations in the amazon basin: secondary broadleaf terra firme forest at cas that was cleared for mixed agricultural purposes 16-20 years ago, and primary terra firme forest at ucp that has never been cleared for agriculture. cas is located at km 54 of the iquitos-nauta road within the loreto region of peru (18 m 668840 9535047), and ucp is located two hours up the itaya river from the town of cahuide (iquitos-nauta road km 57) (18 m 651648 9528215) (figure 1). cas has 69.5 forested hectares; 22.0 are 15-20 year-old secondary regrowth and the remaining 47.5 are older forest (>20 years-old) (l. marshall, unpublished data). ucp has an estimated 10,500 hectares of forested area (h. portocarrero, personal communication). data collection at both sites took place between the 5th and 15th of november 2018, which is roughly the end of the dry season and the beginning of the rainy season in loreto. field data collection at each location, we established a 20 x 100-m plot composed of five contiguous 20 x 20-m subplots, according to rainfor field manual standards (phillips et al. 2016). these 20 x 20-m subplots are the replicate unit for plot-level variables (tree density and agb) in this study. within each of these subplots, measurements were taken and recorded for three variables: diameter at breast figure 1. study area overview: field data collection took place at school for field studies center for amazon studies (sfs cas) and universidad cientifica del peru’s conservation concession (ucp). the route from cas to ucp includes a short drive, a boat ride up the itaya river, and a hike of nearly 2 km, shown in purple. aerial imagery sourced from google earth (2018). height (dbh) in cm, height in m, and wood density in g/cm3. dbh was taken for each tree ≥ 10 cm, and for each liana ≥ 10 cm diameter at any point below 2.5 m vertical height using a standard dbh tape. a clinometer was used to estimate height for each tree. we measured the angle in degrees from observer’s eye level to the top of the tree (a) and the angle from eye level to the ground (b), and a bosch glm80 laser rangefinder to measure the horizontal distance from tree to person measuring height (x) in meters. wood density was measured by drilling into each tree and collecting the wood shavings in a marked bag. at each tree drilled, the hole was measured for depth and width in order to calculate the green volume of wood extracted. the dry weight of the wood was determined by weighing the wood shavings after drying them at 104°c for a minimum of 24 hours. all trees were measured for dbh and height, but due to time constraints, only 48 out of 83 trees at cas and 67 out of 150 trees at ucp were measured for wood density. in final agb calculations, a site average density was used for the trees that were not directly measured for wood density, as recommended by chave et al. (2005). agb calculation tree height was calculated in meters using the following trigonometric equation: tree height = x(tan(a) + tan(b)) wood density was estimated by calculating the green volume of the drilled hole as a cylinder, then dividing the resulting number by dry weight. agb for each non-palm tree was calculated in kg using the following equation from chave et al (2005): abgest = exp( -2.557 + 0.940*ln(ρ(d^2)h)) where ρ = wood specific gravity, d = dbh, and h = total cusj 2019________________________________________________________________________________ __________________________________________________________________________________________ 6 tree height. this equation was chosen based on chave et al. (2005)’s analysis of allometric equations to measure agb, since in our study region, evapotranspiration exceeds rainfall for less than a month per year (rivasmartinez 1994), thereby classifying our study sites within what chave et al. defines as “wet” forest. agb for lianas was calculated in kg using the following allometric equation: ln(total biomass) = -7.114 + 2.276*ln(d) where d = diameter at 30cm shoot extension. this equation is based on gehring et al. 2004’s evaluation of liana agb for primary and secondary forests of the amazon. agb for all palms (family: arecaceae) was estimated in kg using the following equation: agb^0.25 = 0.55512(dmf d^2 hstem)^0.25 where dmf = dry mass fraction, d = dbh, and hstem= stem height. this allometric equation is based on goodman et al. (2013)’s assessment of the allometry of amazonian palms. we used goodman et al.’s mean dry mass fraction, 0.370, for all palm stems. tree, liana, and palm agb were summed for both study sites to get an estimate of total agb per plot. mean agb values per hectare were then estimated by calculating the mean agb for each 20 x 20-m subplot, and multiplying the high, low, and mean subplot values by 25. our subplot estimates of agb and hectare estimate of agb were converted from kg/ha into mg/ha and multiplied by the carbon conversion rate of 0.47 to obtain an estimate of carbon storage per hectare and the total carbon storage at each site (day et al. 2013). a carbon mass fraction ~50% of total agb has been used in other studies of amazonian terra firme forests (chave et al. 2005) our mg/ha estimates of agb and carbon were converted into site-wide equivalencies of total carbon for both cas and ucp by multiplying by the total forested area at each site. when calculating total location agb for cas, we used the ucp estimate of agb for the areas of older forest at cas and the cas estimate for younger secondary regrowth. statistical analysis past3 (hammer et al. 2001) analysis software was used to compare the mean values of dbh, height, and wood density using standard t-tests. differences in dbh, height, and wood density between sites all had unequal variance. trees surveyed were grouped into four dbh size classes: dbh ≥10 – 20 cm, dbh >20 – 30 cm, dbh >30 – 50 cm, and dbh >50 cm. wood density was then compared between these size classes using an anova test. densities of trees and agb per subplot were also compared between sites using t-tests. a p-value ≤0.05 was considered statistically significant. iii. results in total, 228 trees (cas = 83 and ucp = 145) were surveyed and tagged within our plots. overall, trees of smaller size classes (dbh of 10-30cm) were more common at both sites, and more trees of larger size classes (dbh ≥ 30cm) were encountered at ucp (figure 2). no lianas were surveyed at cas, and 5 lianas fitting our dbh criteria were surveyed at ucp. four palms were measured at cas and 2 at ucp, and a total of 3 trees surveyed were not included in agb calculations because they were either dead or leaning too severely to estimate height. figure 2. size class distributions: number of individual trees surveyed in each dbh size class at both cas and ucp. 188 out of 228 trees had a dbhof 10-30cm, and 29 out of the 40 trees surveyed with a dbh≥ 30cm were found at ucp. the trees measured at ucp were found to have, on average, a greater dbh (cas = 18.9cm and ucp = 22.3cm; t-value = 2.231, df = 236.04, p-value = 0.027) and a higher wood density (cas = 0.328g/cm^3 and ucp = 0.732g/cm^3; t-value = 9.953, df = 112.98, p-value = 3.89e^17) (figure 3). no significant difference was found in wood density between size classes, however (f3,110= 1.05, pvalue = 0.373). mean tree height also did not exhibit significant differences between cas and ucp (tvalue = 1.571, df =149.13, p-value = 0.118) (figure 3). figure 3. mean dbh, height, and wood density: comparisons of mean values (± s.e.) calculated for tree dbh (cm), tree height (m), and wood density (g/cm^3) between cas secondary forest and ucp primary forest plots. the asterisk indicates that significant differences were found between the two sites at the critical p-value of 0.05. the total agb estimated within our 20 x 100-m plot at 0 20 40 60 80 100 dbh ≥ 10cm dbh 2030cm dbh 3050cm dbh >50cm nu m be r o f i nd iv id ua ls cas ________________________________________________________________________________ cusj 2019 __________________________________________________________________________________________ 7 ucp was greater than the total estimated at cas (table 1). per hectare, the ucp primary forest plot was also estimated to store more agb than the cas secondary forest plot, with a mean agb estimate of 322.05 mg/ha (95% conf. interval: 236.04-408.06) (range: 254.67433.19mg/ha) at ucp and 51.17 mg/ha (95% conf. interval: 31.44-70.90) (range: 37.71-61.75mg/ha) at cas. carbon estimates for each forest in mg/ha are also reported in the table, as well as estimates of the total agb and total carbon stored in the forested area of each location. density of trees at ucp was also higher per subplot, with a per hectare mean of 725 (95% conf. interval: 613.08-836.92), whereas the mean number of trees per hectare at cas was 415 (95% conf. interval: 350.63479.37). table 1. estimated agb and carbon: quantities of agb and carbon estimated within the 20x100m plots, calculated quantities of agb and carbon in mg/ha, and the quantity of agb and stored carbon calculated for the total area of cas (69.5 forested ha) and ucp (10,500 ha). site plot agb (mg) mean agb (mg/ha) total site agb (mg) plot carbon (mg) mean carbon (mg/ ha) total site carbon (mg) cas 10.55 51.17 16,423 4.86 24.05 7,718 ucp 64.41 322.0 3,381,525 30.27 151.3 1,589,317 iv. discussion the objective of our study was to contribute to existing literature on the agb, carbon sequestration capacities, and regenerative potential of secondary terra firme forests in the amazon basin. with these goals in mind, we tested the hypotheses that we would observe a greater mean tree dbh, wood density, and tree height within the primary forest at ucp, contributing to higher overall agb estimates and carbon storage than at the secondary forest at cas. with the exception of height, which did not vary significantly between sites, all the predicted differences were observed. we will first address the contributions of increased dbh and wood density to the biomass at ucp, followed by sources of variation concerning tree height and tree density, and finally, we will address the implications of the disparity between our agb estimations for cas and those of a previous study of the area as well as the implications of the greater carbon sequestration observed at ucp. dbh according to chave et al. (2005), the variables most important to predicting the agb of an individual tree are trunk diameter, wood density, and tree height, in that order. therefore, it is reasonable to assume that the significantly higher mean dbh measured at ucp directly influenced the site’s greater agb estimate and carbon storage capabilities. primary forests tend to have greater dbh measurements than secondary forests of the same forest type, simply because they have had more time to grow and accumulate biomass, although secondary forests are able to increase average dbh more rapidly (ngo et al. 2013). though we did not measure every stem located within our plots, we have reason to believe that the trees surveyed did give us a close estimation of actual agb. based on studies conducted in amazon forests, more than 80% of total agb is found in trees greater than 10cm in diameter (baker et al. 2014). in addition, trees less than 10 cm dbh only contribute an estimated 2% of total agb; the remaining quantities can be found in lianas, dead wood, and leaf litter (chave et al 2001). thus, our decision to omit smaller stems from our field measurements is supported by existing literature as a time-efficient way to accurately sample a large area of land. it is important to note the significance of especially large trees to agb. chave et al. (2001) found that trees greater than 70 cm dbh disproportionately contribute to agb estimates, with roughly 2.5% of trees surveyed contributing 36% 39% of biomass. our results were not nearly as dramatic, perhaps because we only measured two trees with dbh greater than 70 cm, but those two trees (1.3% of trees surveyed) still accounted for nearly 6% of the total agb at ucp. though forests quickly regenerate and accumulate biomass after clearing, the carbon contained within old-growth trees of such huge stature cannot simply be replaced by the rapid carbon sequestration of saplings. the disproportionate biomass of ancient and enormous trees in old-growth forests emphasizes the importance of incentivizing the preservation of these forests through redd+ funding. wood density studies of forests in temperate regions report that dbh alone is sufficient to estimate biomass (chave et al. 2005). indeed, the only other existing biomass estimate for cas was calculated using an allometric equation based solely on dbh (marshall 2018, unpublished data). however, it should be emphasized that many temperate models are based on simpler forests with dominant tree species, whereas tropical forests are more complex and require models that utilize more variables to estimate agb (chave et al. 2005). in order to refine our agb and carbon estimates, we chose a model that accounts for wood density and tree height in addition to dbh. wood density is a measurement that reflects the amount of biomass per unit volume of tree trunk and is significant in calculating agb (ramanantoandro et al. 2016). it has been concluded that fast-growing, lightdemanding species have lower wood densities than slower-growing, shade-tolerant species; there is a trade-off between the volume of wood produced, facilitating rapid growth toward the canopy, and its resulting density, which provides structural support and protects against breakage (baker et al. 2004, ramanantoandro et al. 2016). therefore, the greater wood densities observed at ucp suggests a species compositional difference within the mature forest, which probably contains more shadetolerant, slow growth trees as opposed to the higher number of light-demanding pioneer species in a secondary cusj 2019________________________________________________________________________________ __________________________________________________________________________________________ 8 forest such as cas. the importance of individual tree wood densities in estimating agb may not seem as significant in old-growth forests that are mainly composed of hardwood species with a narrow range in wood densities, but baker et al. (2004) has shown that ignoring variations in wood density results in poor prediction of overall agb. the landscape of this study is a mosaic of species and differing wood densities, and baker’s assertion here implies that, on the landscape level, it is important to account for the spatial variation of species composition and wood density. for tropical trees, species compositions appears to be more important than size class in influencing wood density. indeed, we found no correlation between dbh size class and wood density at either of our study sites. thus, our study supports previous findings in madagascar that tree diameter had no relationship to wood density within or between tree species, despite the variation of wood densities between species (ramanantoandro et al. 2016). height since we did not observe a significant difference in heights between our two sites, our data tends to agree with the assertion that tree height on its own is not a good estimator of agb (chave et al. 2001). though height on its own is not significantly important to agb, when used in conjunction with dbh it becomes d2h, which chave et al. (2001) asserts is the best estimator of agb. indeed, the standard error in estimation of biomass decreases from 19.5% to 12% when height is included in the equation (chave et al. 2005); we therefore chose to measure and include tree heights in our study. our results support previous findings that after trees have grown to a certain height, they focus most of their energy on growing outward rather than growing upward (da silva scaranello et al. 2012). this helps explain our lack of significant difference between tree height at cas and ucp despite the greater dbh of ucp trees. it can be assumed that there is a threshold past which vertical growth does not confer greater benefits to trees, and that trees at cas and ucp have both reached that point and have begun to invest more energy into increasing girth. again, the old-growth trees at ucp have merely had a longer lifespan to accrue biomass and increase in diameter, resulting in greater total agb. tree density we were not able to find much mention in literature of tree density differences between primary and secondary forests. this is indicative of a lack of available information concerning lower secondary forest tree density of trees ≥ 10 cm dbh. perhaps the comparison of tree density between forest types is not worth scientific attention because it can be assumed that primary forests will contain more trees of larger size classes than secondary forests, based merely on time required for recruitment and growth. carbon storage and agb our carbon estimates of 151.36 mg c/ha for ucp and 24.05 mg c/ha for cas are both slightly lower than average carbon storage values for primary forests found in the peruvian amazon by nebel et al. (2001) (220.2 mg c/ha) and previous estimates for the cas property by marshall (2018) (138.8 mg c/ha). however, agb and carbon storage estimates vary throughout the primary, wet tropical forests of central and western amazonia, ranging from 74.5 mg c/ha to 203.9 mg c/ha in certain areas (kauffman et al. 2009). even within a given tropical evergreen forest, the total sequestered carbon values derived from agb estimates can vary by as much as 117.5 mg c/ha across a local landscape (kauffman et al. 2009). ucp’s carbon estimate, therefore, falls within a normal range for primary terra firme forests of the amazon, and certainly warrants the area’s continued conservation and consideration for redd+ funding. just as the diversity of structure and composition among trees in tropical ecosystems represents a potential source of variation in calculating carbon quantities across the landscapes, the choice of equation used to estimate agb can also affect the predicted quantities (chave et al. 2005). this is in part because the broad applicability of each allometric equation is limited by the diversity and composition of the original data set used to derive it. for example, the chambers et al. (2001) equation previously used to estimate biomass at the cas site is known to consistently give the highest predictions of agb of any amazonian allometric equation, and was developed using only data from forest plots in the central amazon, where tree wood density is reportedly higher than in the western amazon (baker et al. 2004). local topographical variation within the cas site might also account for some of the variation between our estimates and marshall’s, and sampling a larger area could produce more representative agb values. tree densities, for example, were different between the region we sampled and the area sampled in the previous study (our range was 325-450 as compared to 500-575 trees per hectare) (l. marshall, unpublished data). regardless, the agb and carbon estimates derived using the chambers et al. (2001) equation are likely an overestimate of actual values for the cas property, as they were calculated without taking into account the unique wood density values and height/diameter relationships of this region that lead to variations in allometry across amazonia. it is important to note that even the chave et al. (2005) equation we used tends to overestimate agb to some degree (0-5%) when averaged across an entire site. nonetheless, equations such as the one chosen for our study, which estimate agb using forest type (dry, moist, or wet) as a predictive measure, represent a significant improvement over others because models without this measure reliably overestimate agb by as much as 50% (chave et al. 2005). we also believe that, had we used a different equation to estimate agb that did not include wood density in the calculations, our agb and carbon storage values would have been even greater overestimates. when we input our dbh and height ________________________________________________________________________________ cusj 2019 __________________________________________________________________________________________ 9 measurements into the chambers et al. (2001) used by marshall (2018), our agb and carbon estimates in mg/ha nearly tripled (our cas range changed from 17.72-29.02 mg c/ha to 56.74-121.20 mg c/ha). while these increased values are still smaller than marshall’s estimate of 138.8 mg c/ha for the cas property, the parameters and assumptions of the equations being used explain some of this inconsistency. differences in equation used also help contextualize the disparity between marshall’s (2018) total carbon estimate for the cas property (9,650.2 mg c) and ours (7,718.84mg c). this indicates that in many cases agb estimates are highly dependent on the allometric equation applied, and emphasizes the importance of applying an equation which allows for the use of more site-specific variables. v. conclusion in order to improve agb estimates for our study area, we recommend either calculating the individual wood density of every tree sampled or generating information on species composition in the area to improve the accuracy of wood density averages and thereby agb estimations (day et al. 2013). we also recommend the sampling of a larger area to account for local topographical variation, since our range of potential agb values calculated from inter-subplot variation was relatively large (ucp range: 254.67-433.19mg/ha, and cas range: 37.71-61.75mg/ha). any additional carbon estimates generated for secondary forests of the region would contribute significantly to existing research. the agb and carbon storage estimations for the ucp primary forest concession place it within the range of other highly productive primary tropical forests of the amazon basin. however, although our data supports the consensus that primary forests store more agb and sequester more carbon than secondary growth forests through increased parameters such as tree dbh, height, and wood density, this does not indicate that mature forests should be the only conservation priority. while old growth forests like ucp tend to store more of their carbon above ground, secondary growth forests like cas store only about 38.1% of forest carbon in agb, with the rest being stored primarily soil (ngo et al. 2013). our carbon storage estimates for cas, therefore, may be significant underestimates of total ecosystem carbon storage. more research on below ground biomass in western amazonia would contribute substantially to estimations of carbon storage in the secondary forests of the region. moreover, existing research on regenerating tropical ecosystems states that secondary forests, which already act as substantial carbon sinks, increase their capacity to store carbon with increasing time since disturbance (mukul et al. 2016). the potential carbon sequestration abilities of both primary and secondary terra firme forests of the western amazon, therefore, should not be undervalued. author information lindsay finnell, drake university, des moines, ia, usa: lindsay.finnell@drake.edu *bailey nordin, columbia university in the city of new york, ny, usa: bjn2122@columbia.edu (717) 830-4730� 604 w 114 st, new york, ny 10027 acknowledgments we would like to express the sincerest gratitude for the academic support, hard work in the field, and friendship of dr. laura morales, eliza amber, henry hurt, molly sayles, christian perez, and jose david urquiza, as well as all of the staff and students at universidad cientifica de peru and the center for amazon studies. we thank the staff and students of the school for field studies, peru center for amazon studies for making this research possible. references baker, t., o. phillips, y. malhi, s. almeidas, l. arroyo, a. di fiore, t. erwin, t. killeen, s. laurance, w. laurance, s. lewis, j. lloyd, a. monteagudo, d. a. neill, s. patino, n. c. a. pitman, j. n. m. silva, and r. vasquez martinez. 2004. variation in wood density determines spatial patterns in amazonian forest biomass. global change biology 10: 545– 562. chambers, j.q., j. dos santos, r.j. ribeiro, n. higuchi. 2001. tree damage, allometric relationships, and above-ground net primary production in central amazon forest. forest ecology and management 152:73-84. chave, j., c. andalo, s. brown, m. a. cairns, j. q. chambers, d. eamus, h. folster, f. fromard, n. higuchi, t. kira, j. p. lescure, b. w. nelson, h. ogawa, h. puig, b. riera, and t. yamakura. 2005. tree allometry and improved estimation of carbon stocks and balance in tropical forests. oecologia 145: 87–99. chave, j., b. riera, and m. dubois. 2001. estimation of biomass in a neotropical forest of french guiana: spatial and temporal variability. journal of tropical ecology 17: 79-96. da silva scaranello, m. a., l. f. alves, s. a. vieira, p. b. de camargo, c. a. joly, and l. a. martinelli. 2012. height-diameter relationships of tropical atlantic moist forest trees in southeastern brazil. scientia agricola 69: 26-37. day, m., c. baldauf, e. rutishauser, and t. c. h. sunderland. 2013. relationships between tree species diversity and above-ground biomass in central african rainforests: implications for redd. environmental conservation 41: 64–72. gao, y., j. cheng, z. ma, y. shao, and j. su. 2014. carbon storage in biomass, litter, and soil of different plantations in a semiarid temperate region of northwest china. annals of forest science 71: 427-435. cusj 2019________________________________________________________________________________ __________________________________________________________________________________________ 10 gehring, c., s. park, and m. denich. 2004. liana allometric biomass equation for amazonian primary and secondary forest. forest ecology and management 195: 69-83. goodman, r. c., o. l. phillips, d. c. torres, l. freitas, s. t. cortese, a. monteagudo, and t. r. baker. 2013. amazon palm biomass and allometry. forest ecology and management 310:994-1004. hammer, o., d. harper, and p. ryan. 2001. past: paleontological statistics software package for education and data analysis. paleontologia electonica 4: 9. <http.//folk.vio.no/ohammer/past/> kauffman, j. b., r. f. hughes, and c. heider. 2009. carbon pool and biomass dynamics associated with deforestation, land use, and agricultural abandonment in the neotropics. ecological applications 19:1211-1222. king, d., s. j. davies, s. tan, and n. s. md. noor. 2006. the role of wood density and stem support costs in the growth and mortality of tropical trees. journal of ecology 94:670-680. kricher, j. 2011. tropical ecology. pages 325, 556-57. a. kalett, editor. princeton university press, princeton, new jersey. marquardt, k., r. milestad and l. salomonsson. 2013. improved fallows: a case study of an adaptive response in amazonian swidden farming systems. agriculture and human values 30:417–428 marshall, l. 2018. carbon storage inventory of above ground biomass in a western amazonian secondary growth forest. unpublished work. directed research report. sfs peru, spring 2018. mukul, s. a., j. herbohn, and j. firn. 2016. tropical secondary forests regenerating after shifting cultivation in the philippines uplands are important carbon sinks. scientific reports 6: 22483. naughton-treves, l. (2004). deforestation and carbon emissions at tropical fontiers: a case study from the peruvian amazon. world development 32.1:173190.�nebel, g., j. dragsted, and a. s. vega. 2001. litter fall, biomass and net primary production in flood plain forests in the peruvian amazon. forest ecology and management 150:93-102. ngo, k. m, b. l. turner, h. c. muller-landau, s. j. davies, m. larjavaara, n. f. n. hassan, and s. lum. 2013. carbon stocks in primary and secondary tropical forests in singapore. forest ecology and management 296:81-89. pan, y., r. a. birdsey, j. fang, r. houghton, p. e. kauppi, and w. a. kurz. 2011. a large and persistent carbon sink in the world’s forests. science 120: 916. phillips, o., t. baker, t. feldpausch, and r. brienen. 2001. rainfor field manual for plot establishment and remeasurement. amazon forest inventory network 2016: 1-27. <www.rainfor.org/en/manuals.> qgis development team. 2017. qgis geographic information system. open source geospatial foundation project. <http://qgis.osgeo.org.> ramananantoandro, t., m. f. ramanakoto, g. l. rajoelison, j. c. randriamboavonjy, and h. p. rafidimanantsoa. 2016. influence of tree species, tree diameter and soil types on wood density and its radial variation in a mid-altitude rainforest in madagascar. annals of forest science 73: 11131124. rivas-martinez, s. 1994. iquitos climate data. worldwide bioclimatic classification system phytosociological research center. www.globalbioclimatics.org. thomlinson, j.r., m.i. serrano, t. del m. lopez, t.m. aide, j.k. zimmerman. 1996. land-use dynamics in a postagricultural puerto rican landscape. biotropica, 28:525. &kdudfwhul]dwlrq &ruuhfwlrq ri 6hoi�$evruswlrq 'lvwruwlrq lq ;$1(6 $lgdq 3� 5hgg\ dqg $sxuyd 0hkwd 6/$& 1dwlrqdo $ffhohudwru /derudwru\� ���� 6dqg +loo 5g� 0hqor 3dun� &$ ����� 7klv lv qrw wkh dxwkru¶v suhvhqw dgguhvv ;$1(6� ;�5$< $%62537,21 63(&7526&23<� 1($5 ('*( 6758&785( $%675$&7� ;�ud\ devruswlrq vshfwurvfrs\ �;$6� lv dq h[shulphqwdo whfkqltxh xvhg wr sureh wkh dwrplf surshuwlhv ri pdwhuldov� ;$6 phdvxuhphqwv shuiruphg lq ioxruhvfhqfh prgh duh vxemhfw wr vr�fdoohg ³vhoi�devruswlrq´ glvwruwlrq� hvshfldoo\ zkhq wkh vdpsoh xqghu lqyhvwljdwlrq lv wklfn ru frqfhqwudwhg zlwk uhvshfw wr wkh dwrp ri lqwhuhvw� +huh zh lqyhvwljdwh wkh ehkdylru ri vhoi�devruswlrq dqg suhvhqw vhyhudo vxjjhvwlrqv wr dyrlg dqg pdqdjh vhoi�devruswlrq wkurxjk h[shulphqwdo ghvljq� 3duwlfxoduo\� zh ilqg wkdw lw lv hdvlhu wr fruuhfw iru yhu\ wklfn vdpsohv wkdq iru vdpsohv ri lqwhuphgldwh wklfnqhvv� 6hoi�devruswlrq glvwruwlrq qhfhvvlwdwhv d fruuhfwlrq ri wkh phdvxuhg vshfwuxp gxulqj gdwd surfhvvlqj� :h suhvhqw d whfkqltxh iru fruuhfwlqj vhoi�devruehg vshfwud lq wkh ;$1(6 uhjlrq wkdw lv hiihfwlyh iru doo vdpsoh wklfnqhvvhv dqg h[shulphqwdo jhrphwulhv� ,1752'8&7,21 ;�ud\ devruswlrq vshfwurvfrs\ �;$6� lv d srsxodu whfkqltxh xvhg wr lqyhvwljdwh wkh dwrplf surshuwlhv ri pdwhuldov� ;$6 vshfwud lqglfdwh wkh devruswlrq frhiil� flhqw� µe�(�� ri d sduwlfxodu hohphqw dqg hgjh dv d ixqf� wlrq ri wkh hqhuj\ ri wkh lqflghqw ;�ud\� 7kh vwuxfwxuh ri wkhvh vshfwud uhyhdo hohfwurqlf dqg jhrphwulf surshu� wlhv ri d sduwlfxodu hohphqw zlwklq d vdpsoh� ;$6 vshf� wud pd\ eh glylghg lqwr wzr sulpdu\ uhjlrqv� ;�ud\ de� vruswlrq qhdu�hgjh vwuxfwxuh �;$1(6�� zklfk surylghv sulpdulo\ hohfwurqlf lqirupdwlrq (h�j� r[lgdwlrq vwdwh)� dqg ;�ud\ devruswlrq ilqh vwuxfwxuh �;$)6�� zklfk sur� ylghv sulpdulo\ jhrphwulf lqirupdwlrq (h�j� frruglqd� wlrq qxpehu)� ;$6 phdvxuhphqwv duh iuhtxhqwo\ wdnhq lq wudqv� plvvlrq prgh� lq zklfk rqh frpsduhv wkh wudqvplwwhg lqwhqvlw\ dqg lqlwldo lqwhqvlw\ ri wkh ;�ud\ ehdp wr gh� ulyh wkh devruswlrq frhiilflhqw ri wkh vdpsoh ri lqwhuhvw µe�(�� dffruglqj wr wkh uhodwlrqvkls µt�(� !t oq( i0i )� zkhuh 7 lv wkh vdpsoh wklfnqhvv� i0 lv wkh lqlwldo ;�ud\ lqwhqvlw\� i lv wkh lqwhqvlw\ ri wkh ;�ud\ diwhu wudyholqj wkurxjk wkh pdwhuldo� dqg µt�(� lv wkh wrwdo devruswlrq frhiilflhqw ri wkh vdpsoh� 7kh edfnjurxqg frpsrqhqw ri µt�(�� µb(e)� zklfk lv wkh devruswlrq frhiilflhqw gxh wr hohphqwv dqg hgjhv rwkhu wkdq wkh rqh ri lqwhuhvw� lv vxewudfwhg iurp µt�(� wr \lhog µe�(�� µe�(� lv wkhq gl� ylghg e\ µe0+ "µe(e0+�� zkhuh e0+ lv dq hqhuj\ mxvw deryh wkh hgjh� wr rewdlq µn�(�� wkh qrupdol]hg yhuvlrq ri µe(e)� 7udqvplvvlrq prgh phdvxuhphqwv uhtxluh wkdw wkh vdpsoh devrue d vljqlilfdqw sruwlrq ri wkh lqwhqvlw\ ri wkh lqflghqw ;�ud\ dqg wkxv wkdw wkh vdpsoh eh vxiil� flhqwo\ wklfn (odujh 7) ru rswlfdoo\ frqfhqwudwhg (odujh µt(e)) >�@� $qrwkhu frpprq ;$6 whfkqltxh zklfk grhv qrw uhtxluh wkhvh frqglwlrqv lv ioxruhvfhqfh prgh� ,q ioxruhvfhqfh prgh� rqh phdvxuhv wkh ioxruhvfhqfh hplwwhg iurp wkh vdpsoh dv d uhvxow ri wkh devruswlrq ri lqflghqw ;�ud\ skrwrqv wr lqihu µt�(�� )oxruhvfhqfh phdvxuhphqwv� xqolnh wudqvplvvlrq phdvxuhphqwv� fdq eh wdnhq iurp pruh wklq dqg rswlfdoo\ gloxwh vdpsohv� ,q ioxruhvfhqfh prgh� wkhuh lv d olqhdu uhodwlrqvkls eh� wzhhq wkh udwlr ri wkh phdvxuhg ioxruhvfhqfh lqwhqvlw\ wr wkh lqflghqw ;�ud\ lqwhqvlw\ dqg wkh devruswlrq frhi� ilflhqw ri wkh wdujhw hohphqw dqg hgjh� (if/i0)(e) # µe(e)� +rzhyhu� wklv rqo\ krogv zkhq wkh vdpsoh xq� ghu lqyhvwljdwlrq lv vxiilflhqwo\ wklq ru gloxwh zlwk uh� vshfw wr wkh hohphqw ri lqwhuhvw� )ru wklfnhu dqg pruh frqfhqwudwhg vdpsohv� wkhuh lv d qrqolqhdu uhodwlrqvkls ehwzhhq wkh ioxruhvfhqfh if i0 (e) dqg devruswlrq µe(e) vshfwud� dqg wkhlu vwuxfwxuhv gliihu dv d uhvxow� 7kh ydul� dwlrq ehwzhhq wkh ioxruhvfhqfh dqg devruswlrq vshfwud wkdw rffxuv iru vxiilflhqwo\ wklfn vdpsohv lv uhihuuhg wr dv vhoi�devruswlrq� 2qh vkrxog qrw uhdg wrr pxfk lqwr wklv � 9 qdph� ,w lv zlgho\ uhjdughg dv d plvqrphu� exw qrqhwkh� ohvv d zhoo�hvwdeolvkhg whup� ,w lv srvvleoh wr dyrlg vhoi�devruswlrq wkurxjk h[� shulphqwdo ghvljq� prvw vwudljkwiruzdugo\ wkurxjk wkh fkrlfh ru suhsdudwlrq ri d vdpsoh wkdw lv vxiilflhqwo\ wklq ru gloxwh zlwk uhvshfw wr wkh hohphqw ri lqwhuhvw� +rz� hyhu� iru ydulrxv sudfwlfdo uhdvrqv� wklv lv qrw dozd\v ihd� vleoh� ,q wklv fdvh� lw lv qhfhvvdu\ wr fruuhfw wkh phdvxuhg ioxruhvfhqfh vshfwuxp gxulqj wkh gdwd�surfhvvlqj vwdjh wr uhiohfw wkh vwuxfwxuh ri wkh ghvluhg vshfwuxp� µn�(�� $ qxpehu ri vhoi�devruswlrq fruuhfwlrq dojrulwkpv kdyh ehhq suhvhqwhg lq wkh sdvw >�� �� �@� 9dulrxv fruuhf� wlrq dojrulwkpv dsso\ wr gliihuhqw uhjlrqv ri wkh vshf� wuxp �;$1(6 ru ;$)6�� dqg hdfk kdv lwv rzq olpl� wdwlrqv� )ru lqvwdqfh� d srsxodu fruuhfwlrq dojrulwkp iru wkh ;$1(6 uhjlrq� )/82� lqfoxghg lq wkh srsx� odu ;$6 gdwd surfhvvlqj surjudp $7+(1$� dvvxphv wkdw wkh uhohydqw vdpsoh lv dw ohdvw d fhuwdlq wklfnqhvv� zklfk zh zloo glvfxvv ixuwkhu� 7kxv� wkhuh uhpdlqv d qhhg iru lpsuryhg vhoi�devruswlrq fruuhfwlrq phwkrgv zlwk juhdwhu dffxudf\ dqg zlghu dssolfdelolw\� +huh zh zloo h[dplqh wkh ghshqghqfh ri vhoi� devruswlrq rq ydulrxv h[shulphqwdo sdudphwhuv dqg surylgh h[shulphqwdo ghvljq vxjjhvwlrqv iru dyrlglqj dqg frqwuroolqj vhoi�devruswlrq� $gglwlrqdoo\� zh zloo suhvhqw d fruuhfwlrq phwkrg iru wkh ;$1(6 uhjlrq zklfk pd\ eh dssolhg wr d udqjh ri vdpsoh wklfnqhvvhv dqg h[shulphqwdo jhrphwulhv� :h zloo glvfxvv lwv hiil� fdf\ dqg olplwdwlrqv� 7kh h[shulphqwdo gdwd vhw xvhg wkurxjkrxw lv ri dk! hplvvlrq olqh ri d frsshu vdpsoh dqg zdv wdnhq e\ $sxuyd 0hkwd dqg 5\dq 'dylv� )ljxuh �� 6fkhpdwlf 'ldjudp ri ([shulphqw� ,q vwdqgdug jhrphwu\� ! = ± = 45� (yhq lq qrqvwdqgdug jhrphwulhv� w\slfdoo\ ! + " = 90� 0(7+2'6 &kdudfwhul]dwlrq ri 6hoi�$evruswlrq :h ehjlq zlwk wkh ioxruhvfhqfh ixqfwlrqdo� d zhoo� hvwdeolvkhg htxdwlrq wkdw ghvfulehv wkh udwlr ri ioxruhv� fhqfh lqwhqvlw\ wr wkh lqwhqvlw\ ri wkh lqflghqw ;�ud\� (if/i0)(e)� rqh zrxog phdvxuh jlyhq dq h[shulphqwdo jhrphwu\ dqg fkhplfdo frpsrvlwlrq ri d vdpsoh >�� �@� 7kh ixoo ghulydwlrq ri wklv htxdwlrq lv jlyhq lq dsshqgl[ $� if i0 = ( ! 4# )$(e) µe(e) µt(e) + µf vlq(") vlq(#) ! 1! e!(µe(e)+µb(e)+µf vlq(!) vlq(") ) t vlq(!) " ��� +huh� µf "µt(ef ) lv wkh wrwdo olqhdu devruswlrq frhiil� flhqw dw wkh ioxruhvfhqfh hqhuj\ ri wkh wdujhw hgjh �ef �� *hqhudoo\� wkh olqhdu devruswlrq frhiilflhqw µ ri dq duel� wudu\ pdwhuldo lv jlyhq e\ µ = # i µim%i zkhuh µim lv wkh pdvv devruswlrq frhiilflhqw dqg %i lv wkh ghqvlw\ ri d vshflhv zh zlwklq wkh pdwhuldo� $oo rwkhu sdudphwhuv duh ghilqhg lq wkh lqwurgxfwlrq ru lq )lj� ,� -xvw dv µt(e) lv edfnjurxqg�vxewudfwhg dqg qrupdo� l]hg wr rewdlq µn(e)� if i0 (e) lv edfnjurxqg�vxewudfwhg dqg qrupdol]hg wr rewdlq (if/i0)n (e)� ,q sduwlfxodu� d olqh lv ilwwhg wr wkh suh�hgjh gdwd wr rewdlq (if/i0)b(e)� dqg wklv olqh lv vxewudfwhg iurp wkh hqwluh vshfwuxp� 7khq wkh uhvxowdqw vshfwuxp lv glylghg e\ (if/i0)0+ = (if/i0)(e0+) wr qrupdol]h� $v lv dssduhqw lq )lj� ,,�d�� wkh wklfnhu d vdpsoh lv� wkh pruh vhoi�devruehg lwv vshf� wuxp zloo eh� :h fdq ghilqh wklfnqhvv pruh suhflvho\ lq whupv ri dwwhqxdwlrq ohqjwk� ghilqhg dv latt " 1 µt � ru wkh ghswk zlwklq d vdpsoh dw zklfk wkh lqwhqvlw\ ri wkh lqflghqw ;�ud\ zloo uhdfk ��h ri lwv lqlwldo ydoxh� 7kh dwwhqxdwlrq ohqjwk ri frsshu dw lwvk hgjh lv$ 3.87µp� 4xdolwdwlyho\� vhoi�devruehg (if/i0)n (e) ;$1(6 � 10 vshfwud glvsod\ hqkdqfhg suh�hgjh ihdwxuhv dqg gdps� hqhg srvw�hgjh rvfloodwlrqv �)lj� ,,�d��� 7kh uhdvrq iru wklv ehfrphv dssduhqw jlyhq d sorw ri wkh gliihuhqfh eh� wzhhq d qrupdol]hg devruswlrq vshfwuxp dqg ioxruhv� fhqfh vshfwuxp �)lj� ,,�e��� :h vhh wkdw wkhuh lv dq doprvw vlqxvrlgdo udlvlqj ri ydoxhv ri wkh ioxruhvfhqfh vshfwuxp frpsduhg wr wkh devruswlrq vshfwuxp ehorz µn � (fruuhvsrqglqj wr wkh hgjh dqg suh�hgjh� dv zhoo dv wurxjkv ri wkh srvw�hgjh rvfloodwlrqv) dqg dv\ps� wrwlf orzhulqj ri ydoxhv deryh µn � �fruuhvsrqglqj wr shdnv ri wkh srvw�hgjh rvfloodwlrqv�� �d� �e� )ljxuh �� %hkdylru ri 6hoi�$evruswlrq 'lvwruwlrq iru 9dulrxv 7� d� 6lpxodwhg ioxruhvfhqfh vshfwud iru ydulrxv vdp� soh wklfnqhvvhv frpsduhg wr ruljlqdo� xq�vhoi�devruehg gdwd ri frsshu k hgjh� 6hoi�devruehg vshfwud vkrz lqfuhdvhg ydoxhv ehiruh dqg dorqj wkh hgjh� 7kh ghjuhh ri vhoi�devruswlrq lqfuhdvhv zlwk wklfnqhvv� krzhyhu� t = 10latt \lhogv d yluwxdoo\ lghqwlfdo vshfwuxp wr t = %� e� 'liihuhqfh ehwzhhq fruuhfw devruswlrq frhiilflhqw� �µn�� dqg vlpxodwhg� qrupdol]hg ioxruhvfhqfh vshfwuxp� if/i0n � dv d ixqfwlrq ri µn iru ydulrxv wklfnqhvvhv� 1rwlfh wkdw wkh ioxruhvfhqfh vshfwuxp lv doprvw vlqxvrlgdoo\ udlvhg ehorz wkh hgjh dqg dw srvw�hgjh wurxjkv� fruuhvsrqglqj wr µe(e) < µe0+� zlwk wkh juhdwhvw ghfuhdvh rffxuulqj durxqg ��� ri wkh zd\ xs wkh hgjh� $gglwlrqdoo\� lw lv dqg pruh�ru�ohvv olqhduo\ lqfuhdvhg iru srvw�hgjh shdnv� fruuhvsrqglqj wr µe(e) > µe0+� /dvwo\� qrwh wkdw� iru wklqqhu vdpsohv� vhoi�devruswlrq lv prvw vljqlilfdqw iru srvw�hgjh shdnv� dqg iru wklfnhu vdpsohv� vhoi�devruswlrq lv prvw vljqlilfdqw iru wkh suh�hgjh dqg srvw�hgjh wurxjkv� %\ shuiruplqj d iluvw�rughu 7d\oru vhulhv dssur[l� pdwlrq rq wkh h[srqhqwldo whup lq ���� zh ilqg wkdw if i0 # µe�(� zkhq wkh txdqwlw\ >(µt(e)+µf vlq(") vlq(#) ) t vlq(") @<< �� fruuhvsrqglqj sk\vlfdoo\ wr yhu\ wklq �vpdoo /� ru rs� wlfdoo\ gloxwh (vpdoo µt) vdpsohv� 1! e!(µt(e)+µf vlq(!) vlq(") ) t vlq(!) $ 1! (1! (µt(e) + µf vlq(!) vlq(") ) t vlq(!) ) = (µt(e) + µf vlq(!) vlq(") ) t vlq(!) ��� if i0 = ( ! 4# )$(e) µe(e) µt(e) + µf vlq(") vlq(#) ((µt(e)+µf vlq(!) vlq(") ) t vlq(!) ) = ( ! 4# )$(e) t vlq(!) µe(e) & if i0 # µe(e) ��� :khq wklv lv wkh fdvh� wkh qrupdol]hg ioxruhvfhqfh vshf� wuxp lv wkh vdph dv wkh qrupdol]hg devruswlrq vshfwuxp� wkhuh lv qr vhoi�devruswlrq� krzhyhu� zkhq wkh vdpsoh lv qrw yhu\ wklq ru gloxwh� wklv iluvw�rughu dssur[lpdwlrq qr orqjhu krogv� dqg� dv lv dssduhqw iurp ���� wkh uhodwlrq� vkls ehwzhhq µe dqg if/i0 ehfrphv pruh frpsolfdwhg� ,q rughu iru wklv iluvw�rughu dssur[lpdwlrq wr eh ydolg� wkh vdpsoh pxvw eh h[wuhpho\ wklq� $ vdpsoh hyhq ��� latt fdq h[klelw vljqlilfdqw vhoi�devruswlrq (vhh )lj� ,,)� ,q idfw� zh fdq suhglfw wkh ghjuhh ri vhoi�devruswlrq glv� � 11 wruwlrq wkdw d jlyhq h[shulphqwdo vshfwuxp zloo glvsod\� :h fdq ghilqh wkh ghjuhh ri vhoi�devruswlrq glvwruwlrq dv wkh gliihuhqfh ehwzhhq d qrupdol]hg� edfnjurxqg� vxewudfwhg� fruuhfw devruswlrq vshfwuxp� µn� dqg d qru� pdol]hg� edfnjurxqg vxewudfwhg� vhoi�devruehg ioxr� uhvfhqfh vshfwuxp� (if/i0)n � dqg fdoo wklv gliihuhqfh 6$'� ,q rughu wr gr vr� zh zloo wdnh dgydqwdjh ri uh� vxowv xvhg iru wkh vhoi�devruswlrq fruuhfwlrq glvfxvvhg odwhu rq �vhh dsshqgl[ %�� 6$' = µn[1! (µe0+ + &)[1! e!(µnµe0++$)l] (µnµe0+ + &)[1! e!(µe0++$)l] ] ��� ,q rughu iru 6$' wr eh ohvv wkdq ���� wkurxjkrxw wkh zkroh vshfwuxp ri wkh frsshu vdpsoh xvhg lq wklv zrun� fruuhvsrqglqj wr glvwruwlrq ri ohvv wkdq ��zkhuh wkh shufhqwdjh lv uhodwlyh wr � rq wkh qrupdol]hg vshf� wuxp� 7 zrxog qhhg wr eh ����µp� ru ohvv wkdq ���� ri dq dwwhqxdwlrq ohqjwk� 7klv h[dpsoh ghprqvwudwhv wkdw wkh lpsruwdqfh ri gloljhqw vdpsoh suhsdudwlrq wr dyrlg vhoi�devruswlrq fdqqrw eh xqghuvwdwhg� 8qohvv wkh h[� shulphqwhu wdnhv vshfldo fduh wr suhsduh dq h[fhswlrq� doo\ wklq vdpsoh� vljqlilfdqw vhoi�devruswlrq glvwruwlrq zloo rffxu� 7kh htxdwlrq iru 6$' ���� lv qrw lqyhuwleoh lq d forvhg dojheudlf irup iru / " t/ vlq(!)� krz� hyhu� rqh fdq hdvlo\ hydoxdwh wkh h[suhvvlrq qxphulfdoo\ wr ghulyh wkh pd[lpxp vdpsoh wklfnqhvv wkdw zloo nhhs 6$' xqghu d fhuwdlq ydoxh� vd\ ����� ,w lv jhqhudoo\ ehvw wr fdofxodwh 6$' iru kljkhvw wkh ydoxh ri wkh qrupdo� l]hg fruuhfw devruswlrq vshfwuxp� µn� rqh zrxog h[� shfw wr ilqg� zklfk fruuhvsrqgv wr wkh kljkhvw shdn lq wkh µn vshfwuxp� 7klv lv ehfdxvh� dv ghprqvwudwhg lq )lj� ,,�e�� vhoi�devruswlrq iru wklq vdpsohv lv juhdwhvw dw wkh kljkhvw ydoxh ri wkh vshfwuxp� 2i frxuvh� dq h[� dfw pd[lpxp ydoxh iru µn fdqqrw eh nqrzq zlwkrxw dq devruswlrq vshfwuxp wr uhihuhqfh� exw d uhdvrqdeoh jxhvv fdq xvxdoo\ eh pdgh� )ru prvw devruswlrq vshf� wud� µnmax < 1.5� vr fdofxodwlqj 6$' iru µn ��� zloo \lhog d ydoxh kljkhu wkdq rqh vkrxog dfwxdoo\ hqfrxqwhu� krzhyhu� iru vshfwud zlwk pruh h[djjhudwhg srvw�hgjh rvfloodwlrqv� fdofxodwlqj 6$' iru d odujhu ydoxh ri µn zloo eh qhfhvvdu\� )ru dq lqgh[ ri wkh ghjuhh ri vhoi� devruswlrq zklfk lv frpsdudeoh iurp h[shulphqw wr h[� shulphqw� lw pd\ eh xvhixo wr fdofxodwh 6$' iru d w\slfdo ydoxh ri µnmax� vd\ ���� �d� �e� )ljxuh �� %hkdylru ri 6$'� d� 6$' dv d ixqfwlrq ri µb0/µe0 iru 7 ��µe0 (lq rwkhu zrugv� rqh dwwhqxdwlrq ohqjwk iru wkh hohphqw ri lqwhuhvw)� µf �� dqg µn ���� 7klv ghprqvwudwhv wkdw� wkh odujhu wkh edfnjurxqg devruswlrq frhiilflhqw lq sursruwlrq wr wkdw ri wkh hohphqw dqg hgjh ri lqwhuhvw� wkh ohvv vljqlilfdqw vhoi�devruswlrq glvwruwlrq lv� e� 6$' dv d ixqfwlrq ri µe0 iru 7 �� µf �� dqg µn ���� 7klv ghprqvwudwhv wkdw d juhdwhu frqfhqwudwlrq ri wkh hohphqw ri lqwhuhvw uhvxowv lq pruh vljqlilfdqw vhoi�devruswlrq� 6$' surylghv dqrwkhu xvhixo lqvljkw� e\ lqfuhdvlqj µb(e)� zh fdq ghfuhdvh 6$' iru d jlyhq µe(e) dqg t � dv ghprqvwudwhg lq )lj� ,,,� ,q rwkhu zrugv� li zh gloxwh wkh dwrp ri lqwhuhvw lq d ghqvh pdwul[ ri khdylhu dwrpv �dqg pdlqwdlq vdpsoh wklfnqhvv�� zh fdq ghfuhdvh wkh dprxqw ri vhoi devruswlrq� 7klv hiihfw lv dovr dssduhqw lq ���� ,i µb�(� dqg µf duh odujh� wkh yduldwlrq lq µe zloo uhvxow lq uhodwlyho\ ohvv fkdqjh lq wkh ydoxh ri wkh h[� srqhqwldo whup� $gglwlrqdoo\� yduldwlrqv lq µe(e) zloo uhvxow lq juhdwhu yduldwlrqv ri wkh ydoxh ri wkh iudfwlrqdo whup dv d zkroh� ehfdxvh wkh ghqrplqdwru zloo ydu\ ohvv� 3k\vlfdoo\� dq lqfuhdvh lq µb(e) zloo fruuhvsrqg wr dq lqfuhdvh lq µf (e) ehfdxvh zloo eh pruh dqg�ru khdylhu dwrpv lq wkh pdwul[� zklfk zloo dovr devrue ioxruhvfhqfh skrwrqv� 7kxv� gloxwlqj wkh hohphqw ri lqwhuhvw lq d pd� wul[ ri khdylhu dwrpv zloo ghfuhdvh vhoi�devruswlrq hyhq � 12 pruh wkdq )lj� ,,,�d� lpsolhv� &rqyhuvho\� vhoi�devruswlrq glvwruwlrq zloo eh prvw vljqlilfdqw zkhq µb(e) dqg µf (e) duh plqlpl]hg ± wkdw lv� zkhq wkh vdpsoh lv frqfhqwudwhg zlwk wkh hohphqw ri lqwhuhvw� ,i zh dvvxph wkdw rxu vdpsoh lv yhu\ frqfhq� wudwhg zlwk wkh hohphqw ri lqwhuhvw dqg wkxv µb(e) = µf = 0� wkhq� lq rughu iru 6$' ' ���� iru µn ���� 7 pxvw eh ohvv wkdq ����� latt $ (1/12)latt� ,i µn ���� wkhq 7< 0.27latt� %hfdxvh wklv pd[lpxp ydoxh ri 7 uhtxluhg wr dyrlg vhoi�devruswlrq ydulhv vljqlilfdqwo\ iurp fdvh wr fdvh� wkhuh lv qr suhflvh uxoh ri wkxpe uh� jduglqj wkh pd[lpxp wklfnqhvv zlwk zklfk wr suhsduh \rxu vdpsoh� +rzhyhu� lw lv w\slfdoo\ rq wkh rughu ri kxqguhgwkv wr whqwkv ri dq devruswlrq ohqjwk� )ru wklfn vdpsohv� lw lv sduwlfxoduo\ dgylvdeoh wr dyrlg kljk frqfhqwudwlrq zlwk uhvshfw wr wkh hohphqw ri lqwhuhvw ehfdxvh vhoi�devruswlrq fdq ehfrph yhu\ vljqli� lfdqw� 6shflilfdoo\� µf = 0, µt(e) = µe(e) & if = i0( ! 4# )$(e)[1! e!µt(e) t vlq(!) ] ��� ,i wkh vdpsoh lv yhu\ wklfn� wkh h[srqhqwldo whup zloo dssurdfk ]hur dqg wkh ioxruhvfhqfh vshfwuxp zloo h[� klelw plqlpdo vwuxfwxuh� pdnlqj fruuhfwlrq txlwh gliil� fxow� 7kh orzhu µb(e) lv lq uhodwlrq wr µe(e)� wkh kljkhu wkh suhylrxvo\�glvfxvvhg fds rq vhoi�devruswlrq glvwru� wlrq lq wkh wklfn olplw ehfrphv� 7kh juhdwhu wkh wklfnqhvv ri rxu vdpsoh lv� wkh ohvv dffxudwh wkh iluvw�rughu dssur[lpdwlrq ��� ehfrphv� dqg wkxv vhoi�devruswlrq glvwruwlrq lqfuhdvhv� +rzhyhu� df� fruglqj wr wkh ioxruhvfhqfh htxdwlrq� wkhuh lv d pd[l� pxp dprxqw ri vhoi devruswlrq wkdw fdq rffxu gxh wr vdpsoh wklfnqhvv� 7klv lv looxvwudwhg lq )lj� ,,�e�� 7kh uhdvrq iru wklv lv wkdw� iru vxiilflhqwo\ odujh 7� zh pd\ dssur[lpdwh wkh ioxruhvfhqfh htxdwlrq wr vdwlvi\ wkh 7 ( % olplw� \lhoglqj d pxfk vlpsohu dqg dojheudlfdoo\� lqyhuwleoh htxdwlrq >�@� olp t"# $ if = i0( ! 4# )$(e) µe(e) µt(e) + µf vlq(") vlq(#) [1!e!(µt(e)+µf vlq(!) vlq(") ) t vlq(!) ] % = i0( ! 4# )$(e) µe(e) µt(e) + µf vlq(") vlq(#) ��� 7kh shufhqw huuru ri wkh t ( % dssur[lpdwlrq iru d sduwlfxodu h[shulphqw lv vlpso\ wkh pdjqlwxgh ri wkh h[srqhqwldo whup lq ���� 7hpsrudulo\ qhjohfwlqj µf dqg dvvxplqj vwdqgdug jhrphwu\ �! = " ����� lw zloo wdnh d vdpsoh ri dssur[lpdwho\ (! oq(0.01)/ vlq(45) $ 3.3)latt iru wkh t ( % wr eh dffxudwh zlwklq ��� +rz� hyhu� li zh dwwhpsw wr dsso\ d fruuhfwlrq wr wklv vshfwuxp dvvxplqj wkdw wkh t ( % olplw lv vdwlvilhg� rxu fru� uhfwhg vshfwuxp pd\ eh ³ryhufruuhfwhg´ pruh vhyhuho\ wkdq ��� 7kxv� dv d jhqhudo uxoh ri wkxpe iru vwdqgdug jhrphwu\� lw lv dgylvdeoh wr hqvxuh d vdpsoh wklfnqhvv ri ' (! oq(0.001)/ vlq(45) $ 4.9)latt ehiruh dvvxp� lqj wkh t ( % olplw� $gglwlrqdoo\� dv zh kdyh vhhq� vljqlilfdqw vhoi�devruswlrq pd\ rffxu hyhq iru d vdpsoh zlwk d wklfnqhvv rq wkh rughu ri plfurqv� 2qhpxvw kdyh pruh suhflvh nqrzohgjh ri d wklq vdpsoh¶v wklfnqhvv wr surshuo\ fruuhfw iru vhoi�devruswlrq² dqrwkhu uhdvrq wr lqwhqwlrqdoo\ vdwlvi\ wkh t ( % wkurxjk wkh fkrlfh ri d wklfn vdpsoh� 7klv surylghv d xvhixo lqvljkw iru h[shulphqwdo gh� vljq� lw lv hdvlhu wr fruuhfw d ioxruhvfhqfh vshfwuxp wdnhq iurp d vdpsoh ri vljqlilfdqw wklfnqhvv wkdq lw lv iru d vdpsoh ri lqwhuphgldwh wklfnqhvv� 2qh fdq xvh wklv htxdwlrq dv wkh edvlv iru d vwudljkwiruzdug fruuhf� wlrq dojrulwkp� 'dqlho +dvnho¶v )/82� d srsxodu vhoi� devruswlrq fruuhfwlrq dojrulwkp iru ;$1(6� grhv suh� flvho\ wkdw >�@� ,i d vdpsoh lv yhu\ wklfn� rqh fdq xvh )/82 ru wkh fruuhfwlrq dojrulwkp zh zloo suhvhqw wr fruuhfw ioxruhvfhqfh vshfwud zlwkrxw suhflvh nqrzohgjh ri wkh vdpsoh¶v wklfnqhvv� 7kxv� d yhu\ wklfn vdp� soh pd\ eh suhihudeoh wr rqh ri lqwhuphgldwh wklfnqhvv� zklfk zloo uhtxluh wkh h[shulphqwhu wr kdyh suhflvh nqrzohgjh ri lwv wklfnqhvv lq rughu wr fruuhfw wkh vshf� wuxp dffxudwho\� )ru wklfn vdpsohv� lw lv sduwlfxoduo\ dgylvdeoh wr dyrlg kljk frqfhqwudwlrq zlwk uhvshfw wr wkh hohphqw ri lqwhuhvw� ru� vlploduo\� gloxwlrq lq oljkwhu hohphqwv� ,q wklv fdvh� vhoi�devruswlrq fdq ehfrph yhu\ vljqlilfdqw� ,q sduwlfxodu� $ µf = µb(e) = 0 % & if i0 n = ( ! 4# )$(e)[1!e!µt(e) t vlq(!) ] ��� ,i wkh vdpsoh lv yhu\ wklfn� wkh h[srqhqwldo whup zloo ds� surdfk ]hur dqg wkh ioxruhvfhqfh vshfwuxp zloo h[klelw plqlpdo vwuxfwxuh� pdnlqj fruuhfwlrq txlwh gliilfxow� $gglwlrqdoo\� dv zh ghfuhdvh wkh lqflghqw dqjoh !� zh lqfuhdvh wkh dprxqw ri vhoi�devruswlrq� dv vkrzq lq )lj� 9,� ,qwxlwlyho\� zh fdq xqghuvwdqg d ghfuhdvh lq lqflghqw dqjoh dv dq lqfuhdvh lq wkh hiihfwlyh wklfnqhvv ri wkh vdpsoh� ,q wkh fdvh ri qrupdo lqflghqfh� jud]lqj h[lw jhrphwu\ �! ���� " �� zh qrwlfh wkdw if ehfrphv sursruwlrqdo wrµ�(�� dqg wkxv wkhuh lv qr vhoi devruswlrq� � 13 $vvxplqj ! + " ���� olp ""90 if i0 = ( ! 4# )$(e) µe(e) µe(e) + µb(e) + µf ( vlq(90) vlq(0) = %) & 1 ! e!(µe(e)+µb(e)+µf ( vlq(90) vlq(0) =#)) t vlq(0) ' & ( ! 4# )$(e) µe(e) µe(e) + µb(e) + µf ( vlq(90) vlq(0) = %) # µe ��� 7khuhiruh� lw lv dgylvdeoh wr shuirup ioxruhvfhqfh phd� vxuhphqwv lq wklv jhrphwu\ li dw doo srvvleoh� �d� �e� )ljxuh �� %hkdylru ri 6hoi�$evruswlrq 'lvwruwlrq )ru 9dulrxv !� d� 6lpxodwhg ioxruhvfhqfh vshfwud iru ydulrxv lqflghqw dqjohv frpsduhg wr ruljlqdo� xq�vhoi�devruehg gdwd ri frsshuk hgjh� zlwk wkh frqglwlrq wkdw !+" ���� 7kh ghjuhh ri vhoi devruswlrq lqfuhdvhv dv ! ghfuhdvhv� 1rwlfh wkdw iru ! = 89.9� wkhuh lv yluwxdoo\ qr vhoi�devruswlrq� e� 'liihuhqfh ehwzhhq fruuhfw dqg vlpxodwhg vshfwud� 6hoi�$evruswlrq &ruuhfwlrq 'hulydwlrq 7kh edvlv ri wklv fruuhfwlrq phwkrg lv wr xvh qxphu� lfdo fdofxodwlrq dqg wdexodwhg gdwd wr zrun durxqg wkh idfw wkdw wkh ghvluhg txdqwlw\� µe�(�� pd\ qrw eh dojh� eudlfdoo\ lvrodwhg lq ���� :h ehjlq zlwk d voljkwo\ uhiru� pxodwhg ioxruhvfhqfh htxdwlrq zlwkrxw frqvwdqw pxowl� solfdwlyh idfwruv� zklfk ehfrph xqlpsruwdqw xsrq qru� pdol]dwlrq� :h dovr pdnh wkh dssur[lpdwlrq wkdw $(e) lv frqvwdqw ryhu wkh ;$1(6 uhjlrq� zklfk lv dffxudwh ehfdxvh $(e) grhv qrw ydu\ vljqlilfdqwo\ dv d ixqfwlrq ri hqhuj\ iru d jlyhq fkhplfdo vshflhv� if i0 n # µe(e) µt(e) + µfg [1! e!(µt(e)+µfg)l] ��� +huh� l " t/ vlq(!)� g " vlq(!)/ vlq(")� 7dnlqj wdexodwhg gdwd iurp >�@� zh pd\ fdofxodwh� µf = ( i µm0i(e0+)%i ���� µb0 = ( i µm0i(epreedge)%i ���� µt0+ = ( i µm0i(e0+)%i ���� zkhuh zh lv dq hohphqw zlwklq wkh vdpsoh� µn µe�µe0+ lv wkh vshfwuxp wkdw wklv fruuhfwlrq zloo xowlpdwho\ ghulyh� 5hfdoo� lw lv wkh qrupdol]hg� edfnjurxqg�vxewudfwhg� dqg fruuhfwhg devruswlrq vshf� wuxp� 8qohvv wkhuh lv dqrwkhu dwrp lq wkh vdpsoh zlwk dq hgjh dw dq hqhuj\ lq wkh ;$1(6 hqhuj\ uhjlrq ri wkh hohphqw xqghu lqyhvwljdwlrq (d vfhqdulr zklfk vkrxog eh dyrlghg li dw doo srvvleoh)� µb�(� zloo eh qhduo\ frq� vwdqw ryhu wkh ;$1(6 uhjlrq� dqg wkxv zh fdq dssur[l� pdwh µb(e) $ µb0 = µb(ebeforeedge)� zh pd\ wkhq fdofxodwh µe0+ = µt0+ ! µb0� :h pd\ dovr ghilqh & " µb0++µfg� $gglwlrqdoo\� zhpd\ gurs wkh h[solflw �(� ghshqghqfh ri whupv zlwklq wkh h[suhvvlrq� *lyhq � 14 wkhvh dphqgphqwv� zh kdyh� if i0 n # µnµe0+ µnµe0+ + & [1! e!(µnµe0++$)l] ���� 7r qrupdol]h� zh hydoxdwh wkh uljkw kdqg vlgh ri wkh htxdwlrq dw µn = 1� ru� htxlydohqwo\� µe = µe0+� :h wkhq glylgh wklv ydoxh e\ � dqg fdoo lw ³1´ iru qrupdo� l]dwlrq idfwru� :h wkhq pxowlso\ wkh uljkw kdqg vlgh e\ 1� $w wklv srlqw� zh pd\ vhw wkh uljkw dqg ohiw kdqg vlghv ri wkh htxdwlrq wr eh htxdo ehfdxvh wkh\ duh erwk qru� pdol]hg� n " 1 µe0 µe0++$ [1! e!(µe0++$)l] ���� if i0 n = nµnµe0+ µn + & [1! e!(µnµe0++$)l] ���� 1h[w� zh fdofxodwh if i0 n (e) iru d odujh qxpehu ri ydoxhv ri wkh ilqdo� qrupdol]hg devruswlrq vshfwuxp µn ehwzhhq ��� dqg �� 7klv udqjh vkrxog vxiilfh iru prvw vshfwud rqh frxog uhdvrqdeo\ hqfrxqwhu� exw� ri frxuvh� wkh udqjh pd\ eh hdvlo\ dgmxvwhg iru dw\slfdo vshfwud� :h qrz kdyh d pds 6$&� if i0 n (e) )(µn(e)� zkhuh 6$& lv dq dfurq\p iru ³vhoi�devruswlrq fruuhfwlrq´� zklfk zh fdq xvh wr fruuhfw rxu vhoi�devruehg vshfwuxp� 2qhpd\ uhsodfh hdfk gdwd srlqw lq d vhoi�devruehg vshf� wuxp� )�(�� zlwk wkh ydoxh ri µn(e) zklfk fruuhvsrqgv wr wkh qhduhvw ydoxh iru )�(� lq wkh vhw ri ydoxhv lq wkh pds 6$& wkdw kdv ehhq fdofxodwhg� 7kh uhvxow lv wkh fru� uhfwhg vshfwuxp� µn�(�� 3xw vlpso\� wklv phwkrg zrunv durxqg wkh odfn ri forvhg�irup lqyhuwlelolw\ ri wkh iox� ruhvfhqfh htxdwlrq wr ilqg dq �doprvw� h[dfw lqyhuvh qxphulfdoo\ xvlqj wdexodwhg ydoxhv dqg dssur[lpdwlrqv zklfk duh w\slfdoo\ yhu\ vwurqj ryhu wkh ;$1(6 uhjlrq� 7r whvw wkh hiihfwlyhqhvv ri wklv fruuhfwlrq phwkrg� zh dssolhg lw wr gdwd wkdw zh vlpxodwhg xvlqj wkh ioxr� uhvfhqfh htxdwlrq dqg s\wkrq� ,q sduwlfxodu� zh wrrn if/i0(e) gdwd froohfwhg e\ 5\dq 'dylv dqg $sxuyd 0hkwd rq wkh k! hplvvlrq olqh ri d frsshu vdpsoh� dv� vxphg wr qrw eh vhoi�devruehg� :h ilwwhg wklv gdwd vhw wr wdexodwhg olqhdu devruswlrq frhiilflhqw gdwd iru frs� shu iurp 1,67 >�@� :h wkhq xvhg wkh uhvxowdqw vshf� wuxp dv µt(e) lq wkh vlpxodwlrqv� )ru wkh vlpxodwlrqv� zh rewdlqhg µb(e) wkurxjk d suh�hgjh olqhdu uhjuhvvlrq ilw wkh wkh suh�hgjh gdwd ri wkh µt(e) fxuyh� :h fdo� fxodwhg µe(e) xvlqj µe(e) = µt(e) ! µb(e)� :h vxewudfwhg wkh edfnjurxqg ri wkh vlpxodwhg e\ shuirup� lqj d olqhdu uhjuhvvlrq ilw rq suh�hgjh gdwd dqg vxewudfw� lqj wklv olqh iurp p\ hqwluh gdwd vhw� :h wkhq iodwwhqhg p\ gdwd e\ shuiruplqj d olqhdu uhjuhvvlrq ilw rq srvw� hgjh gdwd dqg vxewudfwlqj wklv olqh iurp gdwd diwhu wkh hgjh� $ sro\qrpldo ilw wr srvw�hgjh gdwd zrxog eh pruh vxlwdeoh iru vshfwud zlwk ohvv�olqhdu srvw�hgjh vwuxfwxuhv� /dvwo\� zh qrupdol]hg e\ glylglqj wkh hqwluh vshfwuxp e\ (if/i0)0+� wkh ydoxh dw zklfk wkh srvw�hgjh iodw� whqhg �exw qrw \hw qrupdol]hg� gdwd frqyhujhg� $oo gdwd surfhvvlqj� vlpxodwlrqv� dqg sorwwlqj lq wklv zrun zh uh grqh xvlqj 3\wkrq� $ gdwd�surfhvvlqj surjudp vxfk dv 6,;3$&. ru $7+(1$ frxog eh xvhg iru wkhvh lqlwldo vwhsv� $v h[klelwhg lq )lj� 9 dqg 9,� wkh fruuhfwlrq phwkrg zrunv xqlirupo\ zhoo dfurvv gliihuhqw ydoxhv ri t dqg !� $gglwlrqdoo\� lw lv jhqhudoo\ urexvw djdlqvw huuruv lq xvhu�ghwhuplqhg ydoxhv ri 7 dqg !� +rz� hyhu� dv ! lqfuhdvhv� ru 7 ghfuhdvhv� wkh fruuhfwlrq eh� frphv ohvv urexvw djdlqvw plvphdvxuhphqwv� dqg wkh xvhu pxvw nqrz wkhlu ydoxhv zlwk juhdwhu dffxudf\ wr fruuhfw zhoo� )lj� 9 dqg 9, dovr vkrz wkdw wkh gliihu� hqfh ehwzhhq ruljlqdo dqg fruuhfwhg vshfwud lv juhdwhvw dw wkh ulvlqj hgjh ihdwxuh ($ 8985ev ) dqg wkh iluvw shdn ($ 9000ev )� lqglfdwlqj wkdw wkh fruuhfwlrq lv ohdvw hi� ihfwlyh lq wkhvh uhjlrqv ri wkh vshfwuxp� :h qrwh wkdw wklv fruuhfwlrq lv qrw dssolfdeoh wr wkh ³jud]lqj�lqflghqfh´ uhjlph� zkhuhlq ! ' !c� zkhuh !c lv wkh vdpsohv fulwlfdo dqjoh iru wrwdo h[whuqdo uhiohfwlrq� ,q wklv uhjlph� ��� qr orqjhu dssolhv� � 15 �d� �e� �f� �g� �h� �i� )ljxuh �� %hkdylru ri fruuhfwlrq iru ydulrxv 7� d�f� dqg h ghprqvwudwh wkh fruuhfwlrq rq vlpxodwhg vshfwud ri ydulrxv wklfnqhvvhv� 7 1 10 latt� 7 �latt� dqg 7 ��latt uhvshfwlyho\� 7kh ruljlqdo� xq�vhoi�devruehg vshfwud duh frpsduhg wr wkh vlpxodwhg vshfwud iru ydulrxv 7 dqg wkh fruuhfwhg yhuvlrqv ri wkhvh vlpxodwhg vshfwud� e� g� dqg i �fruuhvsrqglqj wr d� f� dqg h� vkrz wkh gliihuhqfh ehwzhhq wkh ruljlqdo� xq�vhoi�devruehg vshfwud dqg wkh fruuhfwhg vshfwud �lq eoxh�� dv zhoo dv wkh gliihuhqfhv ehwzhhq wkh ruljlqdo xq�vhoi�devruehg vshfwud dqg wkh fruuhfwhg vshfwud� zkhuh wkh fruuhfwlrqv duh shuiruphg zlwk lqfruuhfw vdpsoh wklfnqhvv lqsxw� zlwk ���� lq juhhq dqg ���� lq uhg� 7klv ghprqvwudwhv wkh hiihfwlyhqhvv ri wkh fruuhfwlrq dqg lwv urexvwqhvv djdlqvw plvphdvxuhg vdpsoh wklfnqhvvhv �7�� � 16 �d� �e� �f� �g� �h� �i� )ljxuh �� %hkdylru ri fruuhfwlrq iru ydulrxv !� d�f� dqg h ghprqvwudwh wkh fruuhfwlrq rq vlpxodwhg vshfwud ri ydulrxv lqflghqw dqjohv� ! ��� ! ���� dqg ! ��� uhvshfwlyho\� 7kh ruljlqdo� xq�vhoi�devruehg vshfwud duh frpsduhg wr wkh vlpxodwhg vshfwud iru ydulrxv ! dqg wkh fruuhfwhg yhuvlrqv ri wkhvh vlpxodwhg vshfwud� e� g� dqg i �fruuhvsrqglqj wr d�f� dqg h� vkrz wkh gliihuhqfh ehwzhhq wkh ruljlqdo� xq�vhoi�devruehg vshfwud dqg wkh fruuhfwhg vshfwud lq eoxh� dv zhoo dv wkh gliihuhqfhv ehwzhhq wkh ruljlqdo� xq�vhoi�devruehg vshfwud dqg wkh fruuhfwhg vshfwud� zkhuh wkh fruuhfwlrqv duh shuiruphg zlwk lqfruuhfw !� zlwk ���� lq juhhq dqg ���� lq uhg� 7klv ghprqvwudwhv wkh dffxudf\ ri wkh fruuhfwlrq dqg lwv urexvwqhvv djdlqvw plvphdvxuhg !� � 17 ',6&866,21 &21&/86,21 :h kdyh lqyhvwljdwhg wkh ehkdylru ri vhoi�devruswlrq glvwruwlrq� $gglwlrqdoo\� zh kdyh jdlqhg d ihz lqvljkwv zklfk pd\ eh zruwk wkh frqvlghudwlrq lq ghvljqlqj d ioxruhvfhqfh�prgh ;$6 h[shulphqw� ,ghdoo\� d vdpsoh zloo eh vxiilflhqwo\ wklq vxfk wkdw vhoi�devruswlrq glvwru� wlrq pd\ eh dyrlghg dowrjhwkhu� :h kdyh ghprqvwudwhg wkdw wklv wklqqhvv lv w\slfdoo\ kxqguhgwkv wr whqwkv ri dq dwwhqxdwlrq ohqjwk� ,i rewdlqlqj wklv ghjuhh ri wklqqhvv lv qrw srvvleoh� lw lv suhihudeoh wr fkrrvh d vdpsoh ri > 4.9latt wkdq lw lv wr fkrrvh rqh ri lqwhuphgldwh wklfn� qhvv� 7klv lv ehfdxvh� lq rughu wr fruuhfw d ioxruhvfhqfh vshfwuxpphdvxuhg iurp d vdpsoh ri lqwhuphgldwh wklfn� qhvv� d pruh suhflvh nqrzohgjh ri 7 lv uhtxluhg wkdq lv uhtxluhg wr fruuhfw d vshfwuxp iurp d vdpsoh ri vlj� qlilfdqw wklfnqhvv� ,i lw lv nqrzq wkdw wkh vdpsoh 7 lv > 4.9latt� qr ixuwkhu nqrzohgjh ri 7 lv uhtxluhg wr fru� uhfw wkh vshfwuxp dffxudwho\� 7kh fruuhfwlrq zh kdyh suhvhqwhg lv edvhg rq wkh zhoo�hvwdeolvkhg ioxruhvfhqfh htxdwlrq ���� dqg vlplodu fruuhfwlrq phwkrgv kdyh ehhq suhvhqwhg ehiruh� prvw qrwdelolw\ e\ ,lgd dqg 1rpd >�@� krzhyhu� wklv fruuhfwlrqphwkrg lv xqltxh lq wkdw lw wdnhv lq dqg sxwv rxw qrupdol]hg� edfnjurxqg�vxewudfwhg gdwd� $gglwlrqdoo\� lw qrupdol]hv e\ µe0+� zkhuhdv ,lgd dqg 1rpd¶v fruuhfwlrq qrupdol]hv e\ µe0max >�@� 7kh fruuhfwlrq phwkrg zh suhvhqw lv vwurqj� exw wkhuh lv urrp wr lqfrusrudwh pruh sk\vlfdo ghwdlo iru lp� suryhphqw� )ru lqvwdqfh� lw grhv qrw h[solflwo\ dffrxqw ioxruhvfhqfh ghwhfwru jhrphwu\ dqg ydu\lqj dwwhqxdwlrq iurp sl[ho wr sl[ho ri wkh lqflghqw ;�ud\ lqwhqvlw\ rq wkh sdwk ehwzhhq wkh lqwhqvlw\ prqlwru dqg wkh vdpsoh dqg dv dwwhqxdwlrq ri wkh ioxruhvfhqfh ;�ud\ lqwhqvlw\ rq wkh sdwk iurp wkh vdpsoh wr wkh ioxruhvfhqfh ghwhf� wru� krzhyhu� wklv pd\ eh uhphglhg� wkurxjk wkh xvh ri wkh 6h$))ox; vriwzduh sdfndjh� zklfk fdq dgmxvw phd� vxuhg vshfwuxp wr dffrxqw iru wkhvh dwwhqxdwlrq idfwruv >�@� 7klv dgmxvwhg vshfwuxp fdq wkhq eh fruuhfwhg xv� lqj wkh phwkrg zh kdyh suhvhqwhg deryh� krzhyhu� dv %rrwk dqg %ulgjhv qrwhg lq >�@� wkh hiihfw ri lqwhjudwlrq ri ��� ryhu wkh vrolg dqjoh ri wkh ghwhfwru lv uhodwlyho\ lqvljqlilfdqw lq ! + " = 90 jhrphwulhv� $owkrxjk wklv fruuhfwlrq phwkrg zloo qrw eh shu� ihfw lq doo fdvhv� zh eholhyh wkdw wkhuh lv olwwoh urrp iru lpsuryhphqw eh\rqg d fruuhfwlrq zklfk wdnhv lqwr df� frxqw ioxruhvfhqfh ghwhfwru jhrphwu\ dqg ehdp dwwhq� xdwlrq iurp vrxufhv rwkhu wkdq wkh vdpsoh dv lq 6h$)� )/x;� dqg wkhq lqyhuwv ��� dv lq wkh fruuhfwlrq phwkrg suhvhqwhg khuh >�@� 0rylqj iruzdug� wkhuh uhpdlqv d qhhg iru d ;$1(6 fruuhfwlrq lq wkh jud]lqj lqflghqfh uhjlph� $87+25 ,1)250$7,21 &ruuhvsrqglqj $xwkru $lgdq 5hgg\� &roxpeld 8qlyhuvlw\ lq wkh &lw\ ri 1hz <run� 1<� 86$� dlgdq�uhgg\#froxpeld�hgx )xqglqj 6rxufhv 7klv zrun zdv vxssruwhg lq sduw e\ wkh 8�6� 'hsduwphqw ri (qhuj\� 2iilfh ri 6flhqfh� 2iilfh ri :runirufh 'h� yhorsphqw iru 7hdfkhuv dqg 6flhqwlvwv �:'76� xqghu wkh 6flhqfh 8qghujudgxdwh /derudwru\ ,qwhuqvklsv 3ur� judp �68/,�� $&.12:/('*(0(176 :h wkdqn $sxuyd 0hkwd dqg 5\dq 'dylv iru wkhlu vxssruw dqg jxlgdqfh� 5()(5(1&(6 >�@ $� ,lgd� 7� 1rpd� &ruuhfwlrq ri wkh 6hoi�$evruswlrq (iihfw lq )oxruhvfhqfh ;�5d\ $evruswlrq )lqh 6wuxfwxuh� -sq� $sso� 3k\v� ��� ��������� ������� >�@ &�+� %rrwk� )� %ulgjhv� ,psuryhg 6hoi�$evruswlrq &ruuhfwlrq iru )oxruhvfhqfh 0hdvxuhphqwv ri ([� whqghg ;�5d\ $evruswlrq )lqh 6wuxfwxuh� 3k\vlfd 6fulswd� 7��� ���±��� ������� >�@ '� +dvnho� )/82� &ruuhfwlqj ;$1(6 iru 6hoi� $evruswlrq lq )oxruhvfhqfh 0hdvxuhphqwv� ������� >�@ -� *rxorq� &� *rxorq�*lqhw� 5� &ruwhv� -�0� 'xerlv� 2q h[shulphqwdo dwwhqxdwlrq idfwruv ri wkh dpsolwxgh ri wkh (;$)6 rvfloodwlrqv lq devruswlrq� uhiohfwlylw\ dqg oxplqhvfhqfh phdvxuhphqwv� -rxu� qdo gh 3k\vltxh� �� ���� ������� ������� >�@ -� -dnohylf� -�$� .lue\� 0�3� .ohlq dqg $�6� 5rehuw� vrq� *�6� %urzq dqg 3� (lvhqehujhu�)oxruhvfhqfh 'hwhfwlrq ri (;$)6� 6hqvlwlylw\ (qkdqfhphqw iru 'loxwh 6shflhv dqg 7klq )lopv� 6rolg 6wdwh &rppx� qlfdwlrqv ��� ������� ������� >�@ 1dwlrqdo ,qvwlwxwh iru 6wdqgdug dqg 7hfk� qrorj\� 3k\vlfdo 0hdv� /derudwru\� 'dwd iurp ´;�5d\ )rup )dfwru� $wwhqxd� wlrq� dqg 6fdwwhulqj 7deohv� $ydlodeoh dw kwwsv���sk\vlfv�qlvw�jry�3k\v5hi'dwd�))dvw�kwpo�irup�kwpo� >�@ 5� 0� 7uhyrudk� &� 7� &kdqwohu� 0� -� 6fkdonhq� 6roylqj 6hoi�$evruswlrq lq )oxruhvfhqfh� ,8&u�� ������� ������� �� 18 >�@ :�+� 0f0dvwhu� 1� .huu 'ho *udqgh� -�+� 0do� ohww� -�+� +xeeho� &rpslodwlrq ri ;�5d\ &urvv 6hf� wlrqv� $wrplf 'dwd dqg 1xfohdu 7deohv� �� ������� ������� �� 19 $33(1',; $� '(5,9$7,21 2) )/825(6&(1&( (48$7,21 :h ehjlq zlwk %hhu¶v /dz� zklfk ghvfulehv wkh h[sr� qhqwldo ghfd\ ri wkh lqwhqvlw\ ri wkh lqflghqw ;�ud\ dv lw wudyhov wkurxjk d vdpsoh dv d uhvxow ri devruswlrq� 6hh )lj� , iru jhrphwu\� di(x) dx = !i(x)µt(e) ���� ) x sin(!) 0 1 i(x) di(x) = ) x sin(!) 0 !µt(e)dx ���� i(x) = i0e !µt(e)x sin(!) ���� +huh� ,�[� lv wkh lqwhqvlw\ ri wkh ;�ud\ dw d jlyhq ghswk� µ lv wkh olqhdu devruswlrq frhiilflhqw� htxlydohqw wr# i µi = # i µmi%i� zkhuh zh lv dq hohphqw zlwklq wkh pdwhuldo� µmi lv lwv pdvv devruswlrq frhiilflhqw dqg %i lv wkh ghqvlw\ ri d wkh hohphqw zlwklq d pdwhuldo� dqg i0 lv wkh lqwhqvlw\ ri wkh ;�ud\ sulru wr hqwhulqj wkh pdwh� uldo� 7kh devruswlrq ri d skrwrq grhv qrw qhfhvvdulo\ uhvxow lq wkh uhohdvh ri d ioxruhvfhqfh skrwrq� )ru lq� vwdqfh� lw pd\ uhvxow lq wkh uhohdvh ri dq $xjhu hohfwurq lqvwhdg� 7kh sruwlrq ri devruswlrq hyhqwv uhvxowlqj lq ioxruhvfhqfh pd\ eh ghilqhg dv $(e)� d ixqfwlrq ri hq� huj\� ,qwxlwlyho\� wkh wrwdo ioxruhvfhqfh� iftot(x) pxvw eh htxdo wr iabs(x)� wkh wrwdo devruehg lqwhqvlw\ pxowl� solhg e\ $(e)� iftot(x) = $(e)iabs = $(e)(i0 ! i(x)) = $(e)i0(1! e! µt(e)x sin(!) ) & diftot(x) dx = $(e)i0 µt(e) sin(!) e! µt(e)x sin(!) ���� :khq d skrwrq lv uhohdvhg iurp dq dwrp� lw pxvw wudyho wkurxjk wkh exon ri wkh pdwhuldo ehiruh duulylqj dw iox� ruhvfhqfh ghwhfwru� 2q wklv mrxuqh\� ioxruhvfhqfh ;� ud\v dwwhqxdwh dffruglqj wr %hhu¶v /dz� $gglwlrqdoo\� d jlyhq hgjh ri d jlyhq dwrp zloo ioxruhvfh skrwrqv ri d fkdudfwhulvwlf hqhuj\�ef ex �eb� zkhuhex lv wkh hq� huj\ ri wkh lqfrplqj ;�ud\ skrwrq dqgeb lv wkh elqglqj hqhuj\ ri wkh devruelqj hohfwurq� )oxruhvfhqfh skrwrqv duh hplwwhg lvrwurslfdoo\� $ ioxruhvfhqfh ghwhfwru grhv qrw hqwluho\ vxuurxqg d pdwhuldo� vr lw grhv qrw ghwhfw doo ri wkh ioxruhvfhqfh hplwwhg iurp wkh vdpsoh� ,qvwhdg� lw ghwhfwv d sruwlrq ri ! 4% ri doo wkh ioxruhvfhqfh skrwrqv hplwwhg� zkhuh ! lv wkh vrolg dqjoh lq vwhudgldqv durxqg wkh vdpsoh zklfk wkh ioxruhvfhqfh ghwhfwru fryhuv� 7dn� lqj wklv lqwr dffrxqw� zh ilqg� dif (x) dx = ! 4# $(e)i0 µt(e) sin(!) e! µt(e)x vlq(!) e! µfx vlq(") ���� :khuh if lv wkh phdvxuhg ioxruhvfhqfh lqwhqvlw\ dqg µf µt(ef ) lv wkh wrwdo olqhdu devruswlrq frhiilflhqw dw wkh hqhuj\ ri wkh ioxruhvfhqfh skrwrqv� $ jlyhq hgjh ri d jlyhq dwrp zloo ioxruhvfh skrwrqv ri d fkdudfwhulvwlf hqhuj\� ef ex � eb� zkhuh ex lv wkh hqhuj\ ri wkh lqfrplqj ;�ud\ skrwrq dqg eb lv wkh elqglqj hqhuj\ ri wkh devruelqj hohfwurq� :h pxvw qrz lqwhjudwh ryhu wkh hqwluh ghswk ri wkh pdwhuldo� )lqdoo\� zh lqwhjudwh ryhu wkh hqwluh ghswk ri wkh pdwhuldo� ) t 0 dif (x) = ) t 0 ! 4# $(e)i0 µt(e) vlq(!) e!(µt(e) vlq(!) + µf vlq(") )xdx ���� if (t )! if (0) = if = ! 4# $(e)i0 µt(e) vlq(!) 1 !(µt(e) vlq(") + µf vlq(#) ) [e!(µt(e) vlq(!) + µf vlq(") )t!1] & if = ! 4# $(e)i0 µt(e) µt(e) + µf vlq(") vlq(#) [1!e!(µt(e)+µf vlq(!) vlq(") ) t vlq(!) ] ���� $33(1',; %� '(5,9$7,21 2) 6$' 6$' " µn ! if i0 n ���� :h wkhq frqvlghu ���� dqg ����� \lhoglqj� if i0 n = µnµe0+ µnµe0++$ [1! e!(µnµe0++$)l] µe0+ µe0++$ [1! e!(µe0++$)l] = (µe0+ + &)[1! e!(µnµe0++$)l] (µnµe0+ + &)[1! e!(µe0++$)l] & 6$' = µn[1! (µe0+ + &)[1! e!(µnµe0++$)l] (µnµe0+ + &)[1! e!(µe0++$)l] ] ���� �� 20 16 introduction glioblastomas (gbm) are grade iv gliomas and are the most lethal primary malignancy of the central nervous system (cns). despite comprising 54% of all gliomas and 16% of all primary brain tumors, gbm lacks a cure and only has a handful of treatments [1]. the lack of curative treatment for gbm can be attributed to several factors, some of which being high gbm resistance rates, rapid angiogenesis, and vast spread [10]. studies have demonstrated that, unlike non-tumorigenic cells – which have extracellular ph (phe) of approximately 7.4 – gbm have acidic phe ranging between 6.7-7.1 [3]. this acidic extracellular ph has recently been demonstrated to activate proteases (such as serine proteases, plasminogen activators, and metalloproteinases). these proteases degrade the gbm extracellular environment, explaining why they have been indicated to be involved in gbm growth, invasion, angiogenesis and metastasis [2, 4]. given the importance of ph in gbm progression, this study aimed to focus on ph regulation as a means of gbm treatment. a major regulator of physiological ph is the vacuolar (h+)-atpase (v-atpase). v-atpases are ubiquitously expressed and can be found in lysosomes, exploring the “cascade effect” of aebsf on reduced glioblastoma proliferation mehwish mirza © 2022 mirza. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. columbia university, new york, ny, usa keywords: glioblastoma, aebsf, v-atpase, cascade effect, metalloproteinases, cancer proliferation, extracellular matrix disassembly in cancer abstract: glioblastomas (gbms) are one of the deadliest cancers, and yet still lack treatment. in this study, serine protease inhibitor aebsf is shown to decrease gbm viability and proliferation through a series of intertwined steps. by decreasing the expression of atp6v1a--the first step of the “cascade effect”--aebsf creates an unfavorable environment for protease activity. such “unfavorability” is proven by a decline in expression of proteases mmp9 and elane upon gbm treatment with aebsf. increasingly higher concentrations of aebsf are shown to decrease the percentage of cell viability, proliferation, and metastasis. this is explained through biological process enrichment analyses, which indicate that mmp9 and elane play a role in the disassembly of the extracellular protein network. this disassembly of the extracellular environment enables tumorigenic spread and viability. the results of the viability and proliferation assays, in combination with those of the indirect elisas, enrichment analyses, and scratch assays, underscore an intricate and profound sequence of events that constitute the “cascade effect”. 17 endosomes, secretory vesicles, clathrin-coated vesicles, and the plasma membrane [4]. consisting of 13 subunits in mammalian cells, the primary function of these proton pumps is to regulate both intracellular and extracellular acidification [5]. v-atpases comprise two domains: the v0 and v1 domains. the peripheral v1 domain consists of eight subunits (a–h), all of which assist in atp hydrolysis. the membrane-embedded v0 domain--consisting of subunits a, c, c″, d, and e--is responsible for proton translocation [6]. in this study, subunit a of the v1 domain (atp6v1a) is closely examined due to its cytoplasmic location and ubiquitous nature across the body [14, 15]. this study explores the sequence of events in the “cascade effect” perpetuated by 4-benzenesulfonyl fluoride hydrochloride (aebsf). aebsf, also known as aebsf hydrochloride, is an irreversible serine protease inhibitor that has previously been indicated to inactivate serine proteases such as chymotrypsin and proteinase k [11]. aebsf has been shown in literature to have a potential regulatory effect on the secretion of antitumor effector molecules in leukemia [12]. aebsf has also been shown to reduces dengue virus infection via decreased cholesterol synthesis [13]. in this study, aebsf was studied under the context of gb, proliferation and metastasis. the question sought to be answered is, “what are the chronological events that occur in the reduction of gbm metastasis perpetuated by aebsf?” the findings in this study may: 1) provide more insight into the mechanics of gbm metastasis, 2) highlight the role of serine proteases in gbm progression, and 3) provide greater insight into such proteases’ mechanisms of action, thereby clarifying the link between ecm disassembly and gbm metastasis. methods cell culture t98g human glioblastoma cells were obtained from the american type culture collection (crl-1690; atcc, manassas, va, usa). cells were grown in dulbecco's modified eagle medium (dmem) supplemented with 1% 5000 u/ml penicillin-streptomycin (thermofisher, waltham, ma, usa) and 10% fetal bovine serum (fbs) (gibco, waltham, ma, usa). cells were incubated at 37 °c with 5% co2 and 100% humidity. cell culture images were taken with a nikon eclipse ts100 light microscope at 100x magnification. figure 1. created by author columbia undergraduate science journal vol. 16, 2022 18 reagent preparation aebsf (solubility 50mg/ml h2o, molar mass: 239.69 g/mol) was dissolved in h2o. treatment of cells with aebsf cells were seeded at ~10^6 cells in t25 flasks for 24 hours (incubated at 37°c with 5% co2 and 100% humidity). cells were subsequently treated with the appropriate aebsf concentrations--125µm, 500µm, or 2000µm--for 48 hours. preparation of lysates for enzyme-linked immunosorbent assay (elisa) phosphate buffered saline (pbs) (thermofisher, waltham, ma, usa), 1x cell lysis buffer (cell signaling technology, danvers, ma, usa), and protease inhibitor cocktail (sigma-aldrich, st. louis, mo, usa) were used to lyse 10^6 cells/ml from each treatment group. cells were incubated at 37°c with 5% co2 and 100% humidity with trypsin for 3 minutes until detachment. cells were subsequently neutralized with dmem and transferred to 15ml centrifuge tubes. tubes were gently inverted and cells were counted using vicell xr (beckman coulter, indianapolis, in, usa). tubes were subsequently centrifuged at 1000 rpm for 7 minutes at room temperature, after which the pellets were poured off and the pellet was broken. chilled pbs was added to the tubes, which were subsequently re-centrifuged at 1000 rpm for 7 minutes at 25°c. the residual supernatant was poured out and the pellet was broken up. the appropriate volume of lysis buffer, calculated using the following equation, was added to each tube and incubated on ice for 10 minutes: [viable cell count (cells/ml) × volume of cell suspension prior to first round of centrifugation (ml)] ÷ 106 cells/ml lysates from the tubes were subsequently transferred to chilled eppendorf tubes. the tubes were micro-centrifuged at 13000 rpm for 15 minutes at 4°c. the supernatants were stored at -80°c. indirect elisa 100µl/well of lysates were transferred at 4 replicates per treatment group to a 96-well plate. lysates were left to incubate for 24 hours at 4°c and then removed by blotting. 200µl/ well of blocking buffer (thermofisher, waltham, ma, usa) with 1% bovine serum albumin (bsa) (seracare, milford, ma, usa) was added and left to incubate for 10 minutes at 25°c. cell proliferation assay cells were seeded at 62,500 cells/well 24 hours prior to treatment in a 6-well plate. cells were subsequently treated for 48 hours, supernatants were collected, and cell proliferation was analyzed using a synergy machine. trypan blue exclusion viability assay cells were seeded at ~40,000 cells/ flask 24 hours prior to treatment. cells were subsequently treated for 48 hours, supernatants were collected, and cells were trypsinized and counted using vicell xr (beckman coulter, indianapolis, in, usa). reverse docking reverse docking tool chemmapper was used to predict the protein targets of aebsf. the smiles (simplified molecular input line entry system) notation of aebsf, obtained from pubchem, was inputted into chemmapper. subsequently, the “target navigator” service was selected and the 3d similarity method shaft was chosen. the default similarity threshold of 1.2 was selected, and the bioactivity database chembl was then chosen [7-9]. columbia undergraduate science journal vol. 16, 2022 19 data analysis statistical and computational biology analyses were performed using microsoft excel. an unpaired, two-tailed t-test was performed in excel, and a p-value < 0.05 was considered statistically significant. means and standard deviations are taken for all data and compared as % of control. results aebsf reduces t98g proliferation aebsf was first demonstrated to reduce proliferation in t98g cells. as hypothesized, aebsf reduces proliferation in a statistically significant manner, demonstrating the efficacy of this treatment (fig. 1a). aebsf decreases t98g viability aebsf was also demonstrated to reduce viability in t98g cells. specifically, higher concentrations of aebsf were shown to decrease the number of viable cells and increase the number of dead cells (fig. 2). as hypothesized, aebsf reduces viability in a statistically significant manner, once again demonstrating the efficacy of this treatment (fig. 2). plasma membrane v-atpase subunit a (atp6v1a) protein expression is enhanced in glioblastomas and decreased upon aebsf treatment v-atpase subunit a (atp6v1a) plays a vital role in the primary function of the overall proton pump--acidification of the extracellular environment. through an indirect elisa, it is demonstrated that atp6v1a protein expression decreases with increasing aebsf concentration (fig. 3). aebsf targets serine proteases and metalloproteinases aebsf is a well known serine protease inhibitor. although it is known that aebsf inhibits serine proteases such as trypsin and chymotrypsin, shape-screening tool chemmapper was utilized to predict its protein targets with the highest confidence scores. chemmapper is fundamentally rooted in the “chemical similarity principle”, which states that structurally similar compounds have similar bioactivies [7]. thus, it can be said that a higher similarity score between the query ligand (aebsf in this study) and the database ligands indicates a higher likelihood that a given protein target of the database ligand is also a target of aebsf, the query ligand. thus, as previously discussed, a similarity threshold of 1.2 (on a scale of 2.0) was set in order to filter out protein targets of database ligands with low-similarity scores. using chemmapper, it was predicted that aebsf targets 6 serine proteases (fig. 4a) and 2 metalloproteinases (fig. 4b). serine proteases and metalloproteinases targeted by aebsf primarily play a role in extracellular matrix disassembly and cell migration in glioblastomas the eight protein targets predicted by chemmap were plugged into “search tool for the retrieval of interacting genes/proteins” (string) to carry out an enrichment analysis of the biological processes that they are involved in. the top five biological processes, along with their false discovery rates (fdrs) and strength values, were analyzed (fig. 5). columbia undergraduate science journal vol. 16, 2022 20 figure 1. 48-hour treatment with 4-benzenesulfonyl fluoride hydrochloride (aebsf) reduces proliferation in t98g cells. a: bars are means ± stdev (n=4). *significantly different by t-test at p<0.05 comparing data to that at vehicle control (cells and h2o). **significantly different by t-test at p<0.01 comparing data to that at vehicle control. ***significantly different by t-test at p<0.001 comparing data to that at vehicle control. b: light microscopy images of t98g treated with 125µm, 500µm, or 2000µm aebsf at 100x. figure 2. 48-hour treatment with 4-benzenesulfonyl fluoride hydrochloride (aebsf) reduces viability in t98g cells. columbia undergraduate science journal vol. 16, 2022 21 figure 3. atp6v1a elisa on t98g cells treated with 4-benzenesulfonyl fluoride hydrochloride (aebsf) for 48 hours. 48-hour treatment with aebsf reduces atp6v1a protein expression in t98g. figure 4. reverse docking analysis of 4-benzenesulfonyl fluoride hydrochloride (aebsf) using chemmapper. z-scores and similarity scores of each predicted target are displayed. the similarity score ranges from [0, 2]. the closer the score is to 2.0, the higher the potential of pharmacological association there is between aebsf and the database ligands of the outputted protein targets ^similarity score ranges between 1.2-1.4 ^^similarity score ranges between 1.4-1.6 #z-score<0.5 ##z-score<0.1 ###z-score<0.01 a: aebsf targets serine proteases chymotrypsin-c, trypsin-1, trypsin-2, neutrophil elastase, urokinase-type plasminogen activator, and prothrombin. b: aebsf targets matrix metalloproteinase-9 (mmp9) and matrix metalloproteinase-14 (mmp14). columbia undergraduate science journal vol. 16, 2022 22 figure 5. biological process enrichment analysis of aebsf serine protease (chymotrypsin-c, trypsin-1, trypsin-2, neutrophil elastase, urokinase-type plasminogen activator, prothrombin) and metalloproteinase (mmp9, mmp14) targets. search tool for the retrieval of interacting genes/proteins (string) was utilized for this analysis. false discovery rates (fdrs) and strength values are displayed; p = 7.61e-1. five biological processes with the highest strength values are displayed. aebsf downregulates elane and mmp9 protein expression following the reverse docking and biological processes enrichment analyses, neutrophil elastase (elane) and matrix metalloproteinase-9 (mmp9) expression were selected to be analyzed. these two proteins were selected to be studied, as in the reverse docking analysis, elane and mmp9 were demonstrated to have the highest similarity scores, which was previously established to be a good indicator of the likelihood that a given protein target of the database ligand is also a target of aebsf, the query ligand (fig. 4a, 4b). hence, serine protease elane and metalloproteinase mmp9 expression were analyzed through indirect elisas. both elane and mmp9 protein expression were demonstrated to decrease with increasing aebsf concentrations (fig. 6a, 6b). columbia undergraduate science journal vol. 16, 2022 23 figure 6. mmp9 and elane elisas on t98g cells treated with 4-benzenesulfonyl fluoride hydrochloride (aebsf). bars are means ± stdev (n=4). a: mmp9 elisa on t98g cells treated with aebsf for 48 hours. 48-hour treatment with aebsf reduces mmp9 protein expression in t98g. b: elane elisa on t98g cells treated with aebsf for 48 hours. 48-hour treatment with aebsf reduces elane protein expression in t98g. columbia undergraduate science journal vol. 16, 2022 24 discussion this project sought to explore the sequence of events of the “cascade effect” perpetuated by aebsf in gbm. through an indirect elisa, it was demonstrated that serine protease inhibitor aebsf decreases atp6v1a expression, thereby activating elane and mmp9 proteases. this, in turn, was indicated to stimulate ecm degradation and overall gbm metastasis and migration (fig. 7). it was demonstrated that higher aebsf concentrations had a greater effect on the studied parameters. however, the optimal aebsf dosage remains to be determined and will be studied in the future. in the future, the mechanisms of elane and mmp9-mediated ecm disassembly will be studied. this may give greater insight into these proteases’ mechanisms of action, and possibly further clarify the link between ecm disassembly and gbm metastasis. the findings in this study may provide more insight into the mechanics of gbm metastasis, highlighting the role of serine proteases and tumor metastasis. author information corresponding author *mehwish mirza (academy for medical science technology, department of cellular biology research, bergen county academies, hackensack, nj, usa) mehwish.mirza05@gmail.com author contributions mehwish mirza performed the data analysis and produced the figures and manuscript. dr. german sabio provided technical assistance with analysis and pipelines. dr. german sabio and donna leonardi added project direction and provided guidance. elisa huang and christine chow provided expert feedback for the project. competing interests the authors declare no competing financial and non-financial interests. references [1] tamimi af, juweid m. epidemiology and outcome of glioblastoma. in: de vleeschouwer s, editor. glioblastoma [internet]. brisbane (au): codon publications; 2017 sep 27. chapter 8. doi: 10.15586/codon.glioblastoma.2017. ch8. https://www.ncbi.nlm.nih.gov/books/ nbk470003/ [2] kato, y., ozawa, s., miyamoto, c., maehata, y., suzuki, a., maeda, t., & baba, y. (2013). acidic extracellular microenvironment and cancer. cancer cell international, 13(1), 89. doi: 10.1186/1475-2867-13-89. https://cancerci.biomedcentral.com/articles/10.1186/1475-2867-13-89 [3] ward c, meehan j, gray me, murray af, argyle dj, kunkler ih, et al. the impact of tumour ph on cancer progression: strategies for clinical intervention. explor target antitumor ther. 2020;1:71-100. doi: 10.37349/ etat.2020.00005. https://www.explorationpub. com/journals/etat/article/10025 [4] park, k. c., dharmasivam, m., & richardson, d. r. (2020). the role of extracellular proteases in tumor progression and the development of innovative metal ion chelators that inhibit their activity. international journal of molecular sciences, 21(18), 6805. doi: 10.3390/ijms21186805. https://www.ncbi.nlm. nih.gov/pmc/articles/pmc7555822/ [5] l. shannon holliday, "vacuolar h+-atpase: an essential multitasking enzyme in physiology and pathophysiology", new journal of science, vol. 2014, article id 675430, 21 pages, 2014. doi: 10.1155/2014/675430. https://downcolumbia undergraduate science journal vol. 16, 2022 25 [6] cotter, k., capecci, j., sennoune, s., huss, m., maier, m., martinez-zaguilan, r., & forgac, m. (2015). activity of plasma membrane v-atpases is critical for the invasion of mdamb231 breast cancer cells. the journal of biological chemistry, 290(6), 3680–3692. doi: 10.1074/jbc.m114.611210. https://www.ncbi. nlm.nih.gov/pmc/articles/pmc4319033/ [7] jiayu gong, chaoqian cai, xiaofeng liu, xin ku, hualiang jiang, daqi gao and honglin li. chemmapper: a versatile web server for exploring pharmacology and chemical structure association based on molecular 3d similarity method. bioinformatics. 2013, 29 (14), 1827-1829. https://pubmed.ncbi.nlm.nih. gov/23712658/ [8] xiaofeng liu, hualiang jiang, honglin li. shafts: a hybrid approach for 3d molecular similarity calculation. 1. method and assessment of virtual screening. j. chem. inf. model. 2011, 51 (9), 2372-2385. https://pubs.acs.org/ doi/abs/10.1021/ci200060s [9] weiqiang lu, xiaofeng liu, xianwen cao, mengzhu xue, kangdong liu, zhenjiang zhao, xu shen, hualiang jiang, yufang xu, jin huang and honglin li. shafts: a hybrid approach for 3d molecular similarity calculation. 2. prospective case study in the discovery of diverse p90 ribosomal s6 protein kinase 2 inhibitors to suppress cell migration. j. med. chem. 2011, 54, 3564-3574. https://pubs.acs. org/doi/10.1021/jm200139j [10] calinescu aa, kauss mc, sultan z, al-holou wn, o'shea sk. stem cells for the treatment of glioblastoma: a 20-year perspective. cns oncol. 2021 jun 1;10(2):cns73. doi: 10.2217/cns-2020-0026. epub 2021 may 19. pmid: 34006134; pmcid: pmc8162173. https://pubmed.ncbi.nlm.nih.gov/34006134/ [11] biochemica. (2022). aebsf hydrochloride biochemica. panreac applichem itw reagents. retrieved from https://www.itwreagents.com/italy/en/product/aebsf+hydrochloride+biochemica/a1421 [12] nakabo y, pabst mj. lysis of leukemic cells by human macrophages: inhibition by 4-(2-aminoethyl)-benzenesulfonyl fluoride (aebsf), a serine protease inhibitor. j leukoc biol. 1996 sep;60(3):328-36. doi: 10.1002/ jlb.60.3.328. pmid: 8830789. https://pubmed. ncbi.nlm.nih.gov/8830789/ [13] sreelatha l, malakar s, songprakhon p, morchang a, srisawat c, noisakran s, yenchitosomanus pt, limjindaporn t. serine protease inhibitor aebsf reduces dengue virus infection via decreased cholesterol synthesis. virus res. 2019 oct 2;271:197672. doi: 10.1016/j. virusres.2019.197672. epub 2019 aug 3. pmid: 31386864. https://pubmed.ncbi.nlm.nih. gov/31386864/ [14] françois jouret, céline auzanneau, huguette debaix, ge-hong sun wada, chrystel pretto, etienne marbaix, fiona e. karet, pierre j. courtoy, olivier devuyst jasn nov 2005, 16 (11) 3235-3246; doi: 10.1681/asn.2004110935. https://jasn.asnjournals.org/content/16/11/3235#:~:text=the%20 vacuolar%20h%2b%2datpase%20(,of%20 intercalated%20cells%20(ic). [15] sarkar, j., wen, x., simanian, e. j., & paine, m. l. (2016). v-type atpase proton pump expression during enamel formation. matrix biology: journal of the international society for matrix biology, 52-54, 234–245. https:// doi.org/10.1016/j.matbio.2015.11.004 columbia undergraduate science journal vol. 16, 2022 6 microbiome composition and circadian rhythm disruption alters epithelial barrier integrity elisha pinker1, timur tuganbaev2 1columbia university, new york, ny, usa 2lab of dr. eran elinav; department of immunology, weizmann institute of science, israel abstract: the intestine is home to one of the most complex ecological communities, termed the human gut microbiome. the gut microbiome modulates a wide range of human diseases from diabetes to neurological disorders to cancer. separating the host and the gut microbiome is the epithelial barrier. the intestinal epithelium serves as an adaptive interaction hub between the host and microbiome that plays an important role in deciding the outcome of host-microbiome interactions. regulation of epithelial barrier permeability to ions, nutrients and microbiome metabolites is known to be a tightly controlled process on the host side. while previous studies have identified the microbiome as a factor for epithelial permeability, the exact mechanisms through which it mediates remains less clear. here, we show that alterations in microbiota composition by treatment with antibiotics selectively targeting specific members of the microbiome community impacts the permeability of the intestine. additionally, modulating the microbiome through other methods such as altering diet composition shows changes in permeability of the epithelial barrier. as daily feeding rhythm entrains diurnal fluctuations in microbiome, we have set out to measure how epithelial barrier permeability disrupts the clock. we have discovered that the permeability of the intestinal epithelial barrier exhibits circadian rhythms in mice. disruption of these rhythms, through jet lag or genetic deficiencies in circadian machinery, alters epithelial barrier integrity. together, these findings provide evidence that disruptions in circadian rhythms as well as alterations in microbiome composition have direct consequences in intestinal permeability, and that microbiome might serve as a tool in regulating epithelium permeability. introduction the human body is colonized by a complex community of microorganisms, termed the human microbiome, that includes bacteria, archaea, viruses and fungi [1]. this ecological community exists primarily on mucosal sites, like the gastrointestinal tract, skin, lungs and urogenital system [2]. this community is such where the microbiome bacteria outnumber eukaryotic cells in the entire human body 1.3:1 [3]. the microbiota is responsible for the regulation of many physiological processes, including digestion, absorption [4], host metabolism [5], the maturation and function of the immune system [4], and even host behavior and cognitive function [6]. however, when the balanced microbial composition is disrupted, dysbiosis can cause intestinal diseases [7]. modulating © 2021 pinker, tuganbaev. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. keywords: circadian rhythms, intestinal permeability, microbiome 7 columbia undergraduate science journal vol. 15, 2021 pinker et. al. the microbiome through the diet is one way to induce dysbiosis. treatment with high-fat diet (hfd), containing 60% of caloric energy derived from fat (a model of a “western diet”), is one such example where the balanced gut microbial community is altered to the point of dysbiosis resulting in metabolic dysfunction consequently inducing disease, such as obesity and type 2 diabetes as well as exacerbating inflammatory bowel diseases (ibd) [7]. but, when treated with broad-spectrum antibiotics, obesity and glucose intolerance in mice fed with hfd is abrogated [8]. previous studies show, that when microbiome from mice fed hfd were transplanted into germ-free mice (gf: free of microorganisms) significantly more weight is put on by these mice compared to gf mice transplanted with microbiome from lean donor mice fed nc (normal chow) [9]. additionally, gf condition in mice causes significant immune system deficiency [7]. tools like gnotobiotic animal models (gf mice) and broad-spectrum antibiotics, have enabled the discovery that microbiota contributions to health and disease span multiple organ systems. it has previously been shown that the mammalian intestinal microbiota composition exhibits diurnal oscillations both in terms of its composition and metabolic activity [8]. the circadian clock is the top of a hierarchical internal clock system that is responsible for the circadian biological processes ranging from dictating the rhythmic gene expression in organs to behavior, metabolism and immunity [10][11]. when working with circadian rhythms, scientists devised zeitgeber times (zt) which measure time on a 24-hour light/dark cycle, so zt12 and zt0 denote “dusk” and dawn”, respectively. the molecular mechanisms behind the circadian clocks in mammals is orchestrated under a transcriptional autoregulatory feedback loop comprised of the ‘core’ clock genes clock and bmal1, responsible for encoding activators, and per1, per2, cry1 and cry2, responsible for encoding repressors [12]. a disruption of the circadian clock in humans, environmentally or by gene deficiency, is associated with an array of diseases such as, obesity and diabetes [13] [14], cancer [15], cardiovascular disease [16], ibd (inflammatory bowel disease) exacerbation [17] and susceptibility to infection. one common feature among the diseases associated with circadian rhythm disruption, is that they appear to be triggered or promoted by inflammatory processes. desynchronized interaction between mucosal organs and a complex community of microorganisms colonizing them have emerged as a potential source of inflammation underlining diseases promoted by circadian clock disruption. the source of this inflammation can be located at the gut epithelium [18]. the intestinal epithelium is comprised of a single-cell layer constituting the largest and most important barrier against the external environment [19]. acting as a selectively permeable barrier, the intestinal epithelium can permit the absorption of nutrients, electrolytes and water, while simultaneously defending against toxins [19]. the epithelium maintains an effective barrier through the help of protein-protein networks forming complexes like tight junctions [19]. a compromised epithelial barrier allowing the passage of toxins, antigens and bacteria into the blood stream creates what is commonly known as “leaky gut” [20]. there is therefore an urgent need to better identify the players behind the disruption and upkeep of intestinal barrier function, to devise methods to combat negative barrier alteration. in this project we studied different factors affecting permeability in the epithelial barrier, specifically in the small intestine, the part of the intestine responsible for the majority of nutrient absorption [21]. recent studies on type 2 diabetes and obesity have elucidated the link between microbiome composition and small intestine permeability; however, the mediators of these effects have yet to be determined [22]. over the course of this project, we demonstrated the consequences that genetic and environmental circadian disruption have on gut barrier integrity in mice. furthermore, we 8 columbia undergraduate science journal vol. 15, 2021 pinker et. al. found evidence to suggest that altering the microbiome, either with diet intervention or antibiotic treatment, can disrupt the diurnal oscillations and alter permeability. understanding the key players behind circadian disruption and intestinal barrier maintenance has potential in therapeutic and preventative medicine interventions. methods mice under normal conditions, mice had strict lightdark cycles lasting 12 hours each. lights were turned on either at 11:30 am or 11:30 pm, zt0 or zt12 respectively. for the induction of jet lag, mice were shifted between both control light conditions every 3 days for 3 weeks at a consistent but arbitrary time. jet lag mice were moved between the different light schedules every 3 days for 3 weeks. for consistency, jet lag mice were only experimented on at 1:30 pm, which was synchronized with the zts of the control group (i.e., zt14 of jet lag mice corresponded to zt14 of control mice, as all mice were exposed to the same light-dark conditions at the onset of sample collection). hfd mice were placed in zt2 and zt14 light-dark cycles to measure two time points upon sacrifice. these mice were fed a strict hfd for 3 weeks. mice receiving antibiotic treatment, either received exclusively vancomycin (1 g/l), ampicillin (1 g/l), neomycin (1 g/l), and metronidazole (1 g/l) or a combination of all four in drinking water for 3 weeks. per 1/2/(ko) and per1/2-/+ (wt) were housed under normal conditions and sacrificed at zt 14. fecal microbiome transplantation (fmt) microbiome samples were collected and stored in eppendorf tubes containing 20% glycerol. fecal samples were homogenized and stored in dry ice before long-term storage at -80° c. for the microbiome transplantation, 0.1g of thawed fecal sample was mixed in 1ml sterile pbs and 200 μl of the diluted sample was gavaged per mouse. hfd microbiome samples were taken from mice fed ad libitum hfd for 4 weeks. measuring permeability with fitc-dextran food was removed from mice 2 hours before gavage (starting at 9 am). the solution was made with 4kda fluorescein isothiocyanate (fitc)-dextran dissolved in a phosphate buffered saline (pbs) with a concentration of 40 mg/ml. at 11 am, mice were gavaged with 200 μl dextran. after 2.5 hours (1:30 pm, zt14), blood was drawn and centrifuged to recover the serum. the fluorescence of the serum was quantified at an excitation wavelength of 485 nm and emission wavelength of 533 nm using a microplate reader. ussing chamber the ussing chamber is an instrument used to measure epithelial resistance which is a measure of epithelial barrier permeability. the procedure followed was in accordance with the manufacturer’s instructions (warner instruments, p2300). small intestine tissue is excised and immediately mounted in the chamber. current is then applied to the system and voltage clamp recordings were taken. small intestine was carefully cleaned and handled minimally to prevent microfracturing the tissue and compromising the results. additionally, samples were mounted within three hours of tissue excise to limit cell death. prr stimulation measures degree of microbial presence at systemic sites prr reporter cell lines were obtained from invivogen (hek-blue tlr (toll-like receptors) and nlr (nod-like receptors) reported cell lines): tlr2, tlr3, tlr4, tlr5, tlr7, tlr9, nod1, nod2. extracts from spleen, liver, and serum were homogenized and added to reporter cell lines incubated with kel-blue detection medium according to manufacturer’s instructions. detection is a measurement of specific binding indicating extent of microbial leakage out of the gut. statistical analysis data is presented as the mean ± standard deviation. mann-whitney u-tests were per9 columbia undergraduate science journal vol. 15, 2021 pinker et. al. formed to compare groups. anova with post hoc test was used for comparison between multiple groups with a bonferroni correction. p < 0.05 and q < 0.1 were the threshold for being considered significant. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. results state of small intestinal microbiome determines epithelial barrier function with a gf mouse model, we initially set out to establish the role a microbiome has on the permeability of the small intestine epithelial barrier. mice underwent fecal transplantation of either normal chow (nc) microbiome or high fat diet (hfd) fed donor specific pathogen free (spf) microbiome, while the control group received phosphate buffered saline (pbs) (figure 1a). small intestine samples were taken and epithelial resistance was measured with the ussing chamber. interestingly, mice with hfd microbiome showed increased permeability while mice with pbs (no microbiome) showed reduced permeability (figure 1b). together, this data suggests that microbiome shaped by diet is an active influencer over intestinal permeability. we next set out to further analyze which subsets of bacteria within the microbiome community are most influential in barrier function modulation. with an antibiotic murine model, we could target specific subsets of bacteria to determine the impact the presence or lack thereof has on epithelial permeability in the intestine. we treated 5 groups of wt mice, fed with normal chow, with selective antibiotics that target specific members of microbiome: metronidazole (anaerobic bacteria), vancomycin (gram-positive bacteria), ampicillin (broad spectrum), neomycin (gram-negative bacteria), or a combination of all four for a duration of 3 weeks delivered in their drinking water (figure 1c). the group containing all four antibiotics is a crude simulation of the germ-free mouse model. after 3 weeks, a fitc-dextran assay was performed, which entails orally administering mice with fluorescent dye coated 4 kda beads and measuring fluorescence in the peripheral blood, at zt14, as a proxy for epithelial barrier permeability. significant reduction in optical density (od) was observed in four of the antibiotic groups compared to the control (figure 1d). additionally, vancomycin treated mice had significantly higher od than neomycin and metronidazole groups (figure 1d). the lack of a significant reduction in permeability in the combination treatment, containing all four antibiotics, was likely due to outliers. two days after blood was taken, and fluorescence was no longer detectable, mice were sacrificed, at zt14, and small intestine samples were removed for the ussing chamber measurements. only the jejunum region of the small intestine was used for ussing chamber. all five antibiotic groups showed significantly reduced permeability and vancomycin, again, showed higher permeability than two of the other antibiotic groups (figure 1e). together, these results uncover that the state of microbial composition directly alters the barrier integrity in the small intestine and the entire system. additionally, vancomycin-treated groups experienced the smallest reduction in permeability suggesting, that gram-positive bacteria play the smallest role in intestinal epithelial barrier function compared to other bacterial subtypes in the gut microbiome. genetic and environmental disruption of circadian clocks impacts epithelial barrier integrity it has been shown that a functional circadian clock of a host is required for normal diurnal rhythmicity in microbiota composition and function ‎[8]. building off our previous observations, the next step is to examine if circadian clock disruption will likewise compromise the intestinal epithelial barrier. testing of per1/2-/mice, genetically deficient in a functional host clock, has previously shown to disrupt diurnal rhythmicity in microbiota composition ‎[8]. with our own group of per1/2-/mice, we ran a fitc-dextran assay. 10 columbia undergraduate science journal vol. 15, 2021 pinker et. al. the od for per 1/2-/was significantly higher (figure 2a) demonstrating that circadian clock disruption results in increased intestinal permeability, a feature of a “leaky gut” syndrome. environmental clock disruptions such as chronic jet lag and shift work are behavioral patterns associated with an increased risk for obesity, diabetes and cardiovascular disease [23][18][14][16][24]. it has been previously established that a loss of microbiota oscillations and dysbiosis is associated with jet lag in mice [8]. so, we set out to learn if these losses of oscillations and dysbioses, in jet lag mice, also impact barrier function in the intestinal epithelial barrier. prior to our jet lag experiment, we performed a 24-hour circadian experiment on wt mice measuring intestinal permeability with two independent methods – one measuring the electrical permeability of the epithelium to ions (the ussing chamber assay) and one measuring the concentration of microbial products in peripheral blood (toll-like receptor (tlr) assay) (figure 2b, figure 2c). our findings demonstrate that just like microbial composition behaves in diurnal oscillations, so too does permeability. from this data we chose zt2 and zt14 as the lowest and highest points of permeability respectively and prepared control groups at each of these time points. this makes sense due to the nocturnal behaviors of mice where zt2 is during the daylight when mice are least active while zt14 is during the night is when mice are most active. we then set out to mimic the situation of shift work and chronic jet lag by using a jet lag model in which mice were exposed to a 12-hour time shift every 3 days. using a circadian cabinet, which allows us to individually control the light cycles of each compartment, a group of mice were initially started in one of two a b c ed figure 1. state of small intestinal microbiome determines epithelial barrier function. (a) schematic summarizing transplantation of microbiota from hfd, nc and pbs into germ-free recipient mice. (b) ussing chamber recording of small intestines from gf mice; measuring current (ma) as a measurement of permeability. (c) schematic showing a murine antibiotic treatment setup. (d) fitc (fluorescein isothiocyanate)-dextran recovered from the serum of antibiotic treated wt mice and control. (e) ussing chamber recording of small intestines from antibiotic treated wt mice and control. means ± sd are plotted. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 by mann-whitney u test. 11 columbia undergraduate science journal vol. 15, 2021 pinker et. al. compartments (zt2 or zt14) with light cycle’s 12-hours apart. every 3 days for 3 weeks the cages of mice were switched between the two compartments until they were “jet lagged”. the mice were then bled for fitc-dextran assay at the same time, zt2 and zt14, depending on the compartment. however, if the mice were jet lagged properly, and their circadian clock was disrupted, the zt time at which blood was drawn would not matter. the results showed that the od from fitc-dextran between jet lag groups zt2 and zt14 were non-significant (figure 2d). additionally, compared to the controls, the jet lag groups’ ods was closer to control zt14 (highest measured permeability) than control zt2 (lowest measured permeability). when we excised small intestine samples, for ussing chamber assay, we observed similar trends: non-significant permeability measurements between jet lag groups (sign of successful jet lag) and overall permeability levels aligned with zt14 control levels (figure 2e). altogether, these data suggest that chronic environmental (jet lag model) or genetic disruption (per1/2-/-) of host light-dark cycles exhibits significant alterations in intestinal epithelial barrier function and as a result increasing barrier permeability. this has relevance for frequent fliers traveling across different time zones and people who perform shift work. additionally, our data suggests an explanation for clinical observations, finding shift workers, with irregular sleep/wake schedules, to experience exacerbated inflammatory bowel disease (ibd) due to the effect endogenous misalignment of circadian rhythms has on intestinal homeostasis ‎[26]. metabolic syndrome induced by hfd results in circadian disruption and intestinal barrier dysfunction metabolic syndromes, like obesity and diabetes, have impacted hundreds of millions of people worldwide and consequently taken the lives of millions annually ‎[25]. previous studies have uncovered the associations metabolic syndromes have with the functionality of the intesfigure 2. genetic and environmental disruption of circadian clocks impacts epithelial barrier integrity. (a) fitc (fluorescein isothiocyanate)-dextran recovered from the serum of per1/2-/and wt. (b) ussing chamber recording of small intestines from naïve wt over 24 hours. arrows mark the lowest (zt2) and highest (zt14) points of permeability. (c) prr stimulation by spleen, liver, serum extracts from naïve wt mice. arrows mark the lowest (zt2) and highest (zt14) points of permeability. (d) fitc (fluorescein isothiocyanate)-dextran recovered from the serum of jet lag mice and their respective controls. (e) ussing chamber recording of small intestines from jet lag mice and their respective controls. means ± sd are plotted. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 by mann-whitney u test. figures 2d and 2e: 2-way anova test. a b c d e 12 columbia undergraduate science journal vol. 15, 2021 pinker et. al. tinal barrier, specifically its state of permeability [27]. the study highlights the implications an influx of immune-stimulatory microbial ligands, leaking out of a permeable intestinal barrier can have, such as increased inflammation and risk of infection [27]. however, the tissue under scrutiny was the colon. in our studies, we focused our attention on the small intestine, which one could argue is of more significance due to it being the primary site of nutrient absorption [21]. in order to mimic human dietary habits, which would predispose one to a metabolic syndrome, we fed mice with an hfd for 3 weeks. in a previously referenced chart (figure 1b), where germ-free mice received fecal transplantation of an hfd microbiome we saw a significant increase in small intestine permeability than mice with a normal diet (nc) microbiome, suggesting that a microbial composition associated with an hfd diet causes intestinal barrier dysfunction. previous studies linked circadian rhythm disruption to metabolic syndrome when they found enhanced weight gain and exacerbated glucose intolerance in jet lag mice fed hfd [8]. in an attempt to determine if an hfd would alter permeability in a circadian context we set out to perform another permeability experiment on hfd mice but under circadian conditions. we fed two groups of mice with hfd for 3 weeks. utilizing the circadian cabinet, each group was placed in different compartments and light-dark cycles were set 12-hours apart at the previously observed lowest and highest times of permeability, zt2 and zt14 respectively (figure 3a). fitc-dextran assay was run and blood was drawn at zt2 and zt14 depending on which compartment the mice were in. from the fitc-dextran data we see no significant difference in od between hfd diet groups which suggests that the altered diet disrupted circadian rhythms (figure 3b). additionally, we see od levels of hfd groups to be at levels closer to control group zt14 reaffirming previous data showing increased permeability in hfd mice (figure 3b). when the ussing chamber is performed on small intestine samples, we likewise see no significance between hfd groups and overall hfd permeability levels correlate figure 3. metabolic syndrome induced by hfd results in circadian disruption and intestinal barrier dysfunction. (a) schematic showing the typical configuration of a circa-dian cabinet. (b) fitc (fluorescein isothiocyanate)-dextran recovered from the serum of hfd mice at different zt times and their respective controls. (c) ussing chamber recording of small intestines of hfd mice at different zt times and their respective controls. means ± sd are plotted. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 by mann-whitney u test. figure 3b and 3c: 2-way anova test. a b c 13 columbia undergraduate science journal vol. 15, 2021 pinker et. al. more strongly with zt14 than zt2 (figure 3c). together, these observed results draw attention to a serious issue. metabolic syndromes, which are usually onset by poor diets such as those high in fat, are only exacerbated by the same stimuli that caused the development of the disease. our data signifies metabolic syndromes’ role in both disrupting circadian rhythms and heightening intestinal barrier permeability that can lead to increased inflammation and infection. discussion & conclusion in this study, we describe that the mammalian intestinal barrier function displays diurnal oscillations and is dependent on microbiome composition. if microbiome composition is distorted, as in the case of antibiotic treatment or hfd, we see alterations in intestinal barrier integrity as well as disruptions in the circadian rhythms of epithelial permeability. while previous studies sought to understand the consequences, diurnal disruption has on microbial composition, here we found direct evidence of the role microbial composition plays in intestinal permeability. next, we wanted to study the direct implications distorting circadian rhythms, either with genetic clock deficiency or time-shifted induced jet lag, have on intestinal barrier function. we not only found that these conditions significantly impaired the circadian rhythms of permeability, but also, increased overall permeability resulting in greater pathobiont exposure and inflammation. lastly, we found that inducing metabolic syndromes via hfd intervention disrupts circadian rhythms in the intestinal barrier as well as causes significant barrier dysfunction resulting in increased epithelial permeability. altogether, these studies have uncovered numerous future paths. while this study has broadly implicated microbiome composition as a mediator of gut permeability, in order to learn more about the key bacterial players in permeability mediation, we can perform sequencing on the microbial compositions found in our antibiotic treated groups to reveal the unique bacteria having the biggest impact on reducing or increasing permeability. understanding which bacteria to promote and which to remove in the microbiota could be pivotal in suppressing symptoms in patients with ibd and metabolic syndromes, to name a few, caused by leakages in the intestinal epithelial. exploiting antibiotic or probiotic treatments to target these bacteria could help combat diseases exacerbated by gut leakiness. lastly, our observations showing intestinal permeability functioning under diurnal oscillations has applications from drug consumption to nutritional/dietary strategies. further unravelling the roles, regulators and mechanisms of intestinal permeability could serve to one day aid therapies for illnesses affected by the “leaky gut” syndrome. author information corresponding author *elisha pinker ep2858@columbia.edu funding sources columbia college class of 1939 fellowship competing interests the authors declare no competing financial and non-financial interests. acknowledgments e.p. would like to thank columbia college and the weizmann institute of science for supporting this research. references [1] marietta e, rishi a, taneja v. immunogenetic control of the intestinal microbiota. immunology. 2015;145(3):313-322. doi:10.1111/ imm.12474. [2] neish as. mucosal immunity and the microbiome. annals of the american thoracic society. 2014;11(suppl 1):s28-s32. doi:10.1513/annalsats.201306-161mg. 14 columbia undergraduate science journal vol. 15, 2021 pinker et. al. [3] sender r, fuchs s, milo r. revised estimates for the number of human and bacteria cells in the body. plos biology. 2016;14(8):e1002533. doi:10.1371/journal. pbio.1002533. [4] clemente, j.c., ursell, l.k., parfrey, l.w., and knight, r. (2012). the impact of the gut microbiota on human health: an integrative view. cell [5] sommer, f., and backhed, f. (2013). the gut microbiota—masters of host development and physiology. nat. rev. microbiol [6] hsiao, e.y., mcbride, s.w., hsien, s., sharon, g., hyde, e.r., mccue, t., codelli, j.a., chow, j., reisman, s.e., petrosino, j.f., et al. (2013). microbiota modulate behavioral and physiological abnormalities associated with neurodevelopmental disorders. cell [7] shi n, li n, duan x, niu h. interaction between the gut microbiome and mucosal immune system. military medical research. 2017;4:14. doi:10.1186/s40779-017-0122-9. [8] thaiss c.a., zeevi d., levy m., zilberman-schapira g., suez j., tengeler a.c. transkingdom control of microbiota diurnal oscillations promotes metabolic homeostasis. cell. 2014;159(3):514–529 [9] de groot pf, frissen mn, de clercq nc, nieuwdorp m. fecal microbiota transplantation in metabolic syndrome: history, present and future. gut microbes (2017) [10] dibner, c., schibler, u., and albrecht, u. (2010). the mammalian circadian timing system: organization and coordination of central and peripheral clocks. annu. rev. physiol. [11] keller, m., mazuch, j., abraham, u., eom, g.d., herzog, e.d., volk, h.d., kramer, a., and maier, b. (2009). a circadian clock in macrophages controls inflammatory immune responses. proc. natl. acad. sci. [12] takahashi js. transcriptional architecture of the mammalian circadian clock. nature reviews genetics. 2017;18(3):164-179. doi:10.1038/nrg.2016.150. [13] buxton, o.m., cain, s.w., o’connor, s.p., porter, j.h., duffy, j.f., wang, w., czeisler, c.a., and shea, s.a. (2012). adverse metabolic consequences in humans of prolonged sleep restriction combined with circadian disruption. sci. transl. med [14] fonken, l.k., workman, j.l., walton, j.c., weil, z.m., morris, j.s., haim, a., and nelson, r.j. (2010). light at night increases body mass by shifting the time of food intake. proc. natl. acad. sc [15] savvidis c, koutsilieris m. circadian rhythm disruption in cancer biology. molecular medicine. 2012;18(1):1249-1260. doi:10.2119/molmed.2012.00077. [16] scheer, f.a., hilton, m.f., mantzoros, c.s., and shea, s.a. (2009). adverse metabolic and cardiovascular consequences of circadian misalignment. proc. natl. acad. sci [17] swanson, gr., burgess, hj. (2017). sleep and circadian hygiene and inflammatory bowel disease. gastroenterology clinics of north america.46(4):881893. doi:10.1016/j.gtc.2017.08.014 [18] koch s, nusrat a. the life and death of epithelia during inflammation: lessons learned from the gut. ann. rev. pathol. 2012;7:35–60. doi: 10.1146/annurev-pathol-011811-120905. [19] groschwitz kr, hogan sp. intestinal barrier function: molecular regulation and disease pathogenesis. the journal of allergy and clinical immunology. 2009;124(1):3-22. 15 columbia undergraduate science journal vol. 15, 2021 pinker et. al. doi:10.1016/j.jaci.2009.05.038. [20] mu q, kirby j, reilly cm, luo xm. leaky gut as a danger signal for autoimmune diseases. frontiers in immunology. 2017;8:598. doi:10.3389/fimmu.2017.00598. [21] kiela pr, ghishan fk. physiology of intestinal absorption and secretion. best practice & research clinical gastroenterology. 2016;30(2):145-159. doi:10.1016/j. bpg.2016.02.007. [22] chakaroun rm, massier l, kovacs p. gut microbiome, intestinal permeability, and tissue bacteria in metabolic disease: perpetrators or bystanders? nutrients. 2020;12(4). doi:10.3390/nu12041082 [23] archer, s.n., laing, e.e., mo¨ ller-levet, c.s., van der veen, d.r., bucca, g., lazar, a.s., santhi, n., slak, a., kabiljo, r., von schantz, m., et al. (2014). mistimed sleep disrupts circadian regulation of the human transcriptome. proc. natl. acad. sci [24] suwazono, y., dochi, m., sakata, k., okubo, y., oishi, m., tanaka, k., kobayashi, e., kido, t., and nogawa, k. (2008). a longitudinal study on the effect of shift work on weight gain in male japanese workers. obesity [25] obesity and overweight. https://www. who.int/news-room/fact-sheets/detail/obesity-and-overweight. accessed 20 dec. 2020. [26] nojkov, borko, et al. “the impact of rotating shift work on the prevalence of irritable bowel syndrome in nurses.” the american journal of gastroenterology, vol. 105, no. 4, apr. 2010, pp. 842–47. pubmed central, doi:10.1038/ajg.2010.48. [27] thaiss ca, levy m, grosheva i, zheng d, soffer e, blacher e, et al. hyperglycemia drives intestinal barrier dysfunction and risk for enteric infection. science. 2018;359:1376–1 columbia undergraduate science journal open-access publication | http://cusj.columbia.com 12 cusjvolume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu n eu r o sc ie n c e cusjcolumbia undergraduate science journalresearch articles 13cusj volume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu n eu ro sc ien c e lesion studies, eeg studies fmri studies, and electrophysiology studies indicate that the medial temporal lobe (mtl) plays a large role in learning and memory (johnston, 1998; o’keefe, 1979; elridge, 2000; buzsaki, 1989; eichenbaum, 2007). there is some debate as to how different types of memory are created within certain regions of the mtl. it is understood that the hippocampus, part of the mtl, plays a central role in forming new memories. the observation of patient h.m., who suffered from epilespy in his mtl, best exemplifies the importance of the hippocampus in forming new memories. due to the severity of the epilepsy, his hippocampi were surgically removed, which damaged the entohinal cortex (ec). following the surgery, h.m. experienced anterograde amnesia, he could no longer store any new long term memories. however, the rest of his cognitive abilities stayed intact. he still had full control over his working (short-term) and procedural memories (johnston, 1998). the results of the h.m. study have been replicated in other studies in which hippocampal damage also results in anterograde amnesia (johnston, 1998; zola-morgan, 1986). alzheimer’s disease, a disease in which people are unable to make new memories, has been found to target the hippocampus and the ec (johnston, 1998). it is hypothesized that the hippocampus becomes functionally disconnected from the rest of the brain in alzheimer’s disease, perhaps suggesting the importance of the ec as a relay center and a connective intermediate (hyman, 1984). previous studies in rats have shown that a major source of input to the hippocampus is from the cortical layer of the ec through the perforant pathway (pp) and that outputs of the hippocampal areas ca1 and subiculum project to the ec. this means that there are many major connections traveling from the ec to the hippocampus and then many connections from the hippocampus to the ec (johnston, 1998; van groen, 2003). this connectivity is thought to contribute to processing of sensory information. neurons in layer ii of the ec comprise the pp and terminate in the dentate gyrus and ca3 region of the hippocampus, and neurons in layer iii project to ca1 and the subiculum. axons from ca1 and the subiculum project back to the ec, returning to the original cortical area that relayed the information to the ec (johnston, 1998; witter, 2000; de curtis, 1991). one study in rats proposed that dye injection in ca1 resulted in terminal fibers labeled in layers v-vi of the ec (naber, 2001). these loops are thought to be essential for the formation of long-term memories (johnston, 1998; witter, 2000; de curtis, 1991). although the mouse brain is similar to the wellcharacterized rat brain, there are some differences (van groen, 2002). in order to further explore the formations and circuitry in mice, we intend to label neurons from the ec to the dg and ca3, as well projections lipophilic tracer dii used for neuronal tracing in the fixed hippocampal formations of mice jessica greenberg1*, rebecca piskorowski2, and steven siegelbaum2 1department of biological sciences, columbia university, new york, ny 10027, 2center for neurobiology and behavior, department of neuroscience, columbia university, and howard hughes medical institute, new york, new york 10032 copyright: © 2011 the trustees of columbia university, columbia university libraries, some rights reserved, rastogi, et al. received jan. 11, 2010. accepted feb. 8, 2010. published april 1, 2011. *to whom correspondence should be addressed: jtg2112@columbia.edu abstract both the hippocampus and the entorhinal cortex are involved in memory formation. it is thought that they work together to create a loop that is involved in the formation of long term memories. the entorhinal cortex sends projections to the ca1 älsk�vm�[ol�opwwvjhtw\z�huk�[ol�*(��älsk�zlukz�wyvqlj[pvuz�ihjr�[v�[ol�lu[vyopuhs�jvy[l_���/v^�l_hj[s`�[ol`�jvtt\upjh[l� is unknown. in order to do electrophysiological studies it is necessary to know in which orientation to cut the brain to preserve the most connections between these two areas. here we show that dii crystal application can be used to successfully shils�ruv^u�äily�wh[oz�myvt�[ol�klu[h[l�n`y\z�[v�[ol�*(��älsk���;olu�^l�ilnpu�^p[o�zhnp[[hs�huk�ovypavu[hs�zspjlz�vm�[ol�iyhpu� [v�zll�ov^�^lss�jvuulj[pvuz�il[^llu�[ol�*(��älsk�huk�[ol�,*�hyl�wylzly]lk� 2l`�^vykz!�opwwvjhtw\z��lu[vyopuhs�jvy[l_[��tltvy �̀�äily�wh[oz� introduction from the ec to ca1 and then back to the ec. knowing how this connectivity is preserved in in vitro mouse models (fixed brain slices) will allow for the development of a protocol for future electrophysiological experiments on the connectivity between the entorhinal cortex and hippocampus. these studies require acute, unfixed slices that contain preserved loops between the ec and hippocampus, so understanding how to slice the brain to preserve connections is essential. to inspect the connectivity in the hippocampus and ec specifically, we used the lipophilic tracer dii. it was first introduced in 1987 by godement et al. as a new fluorescent tracer. it was originally used to label neurons in fixed slices from mouse brains and chicken embryos (godement, 1987). dii is an especially reliable dye since it has no discernable effect on the survival of neurons and causes them to fluorescence for long periods of time (honig, 1985; vindal-sanz, 1985). it seems that, dii has not previously been used to label circuitry in the hippocampal formation of mice. a similar experiment in japan at the fukui medical school examined the entorhinal-hippocamus-entorhinal circuit of rats by means of in vivo injections of dii into the rhinal sulcus. this study proposes that the ca1 projection, rather than the subicular projection, is the main projection that feeds back information from the hippocampus to the enthorhinal cortex. their data support the idea that the connections between the subiculum, ca1 field, and ec is reciprocal and that the entorhinal input to the ca1 field is fed back to the same cortical column directly through the ca1 field or by the subiculum (tamamaki, 2003). in this study, we show that the application of dii crystals can be used to successfully label known fiber paths from the dg to the ca3 field. we also show that orienting the slices sagitally and horizontally is useful for demonstrating how well connections between the ca1 field and the ec are preserved. these findings have great potential to benefit future studies of brain circuitry and neuronal pathways. obtaining brain samples adult male mice were anesthetized with a mix of ketamine (40 mg/ml) and xylosine (2.8 mg/ml), 50 microliters/10g of mouse. a pump was used to deliver 25 ml of filtered pbs and then 20 to 25 ml of 4% pfa to the heart and blood vessels. the excised brain was fixed for a minimum of one night in 15 to 20 ml of 4% pfa and kept at 4°c. a series of washes were carried out to clear each brain of pfa. first a 10 minute wash with 0.1m glycine and then 3x10 minute washes with filtered pbs. sectioning the brain a vibratome series 1000 sectioning system was used to slice the brain in the orientation of choice. the brain was sliced in several different orientations: dissected hippocampi were sectioned transversely along the temporal axis, whole brains were cut coronally, sagittally, and horizontally. the first slices were transverse slices of only the hippocampus 400 microns thick. later, sagittal slices of the entire brain were sectioned, at 50, 200 and 300, and 400 microns thick. horizontal and coronal sections were cut at a thickness of 400 microns. materials and methods � ��� ��� � ��� ��� � ��� ��� figure 1a figure 1b figure 1 1a, 1b, and 1c are all transverse slices stained with a dii crystal, imaged under a confocal microscope, and stitched together with adobe photoshop. the arrows wvpu[�[v�[ol�+.�huk�[ol�tvzz`�äilyz�[yh]lspun�\w�[v�*(����;ol�zspjlz�pu��(�huk��)�ohk�k`l�hwwsplk�^p[o�[ol�ullksl�z`ypunl�tl[ovk�huk�^lyl�puj\ih[lk�mvy�[oyll� kh`z�h[����klnyllz�*�huk�[olu�wshjlk�h[����klnyllz�*�\u[ps�ilpun�tv\u[lk���;ol�zspjl�pu��*�ohk�k`l�hwwsplk�^p[o�tpjyväs�[pw�tl[ovk�huk�^hz�puj\ih[lk�h[���� degree c for 2 days and then placed in a 4 degree c room until mounting. figure 1c columbia undergraduate science journal open-access publication | http://cusj.columbia.com 14 cusjvolume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu n eu ro sc ie n c e cusjcolumbia undergraduate science journalresearch articles 15cusj volume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu n eu ro sc ien c e application of the dye a wild heerbrugg stereo-dissection microscope was used in the application of the dii to the dg in the transverse slices and ca1 in the sagittal slices and to the ca1 in whole brain mounts. after the dye was applied, these slices were incubated at 30°c for a period between 2 and 4 days and then placed in at 4°c before mounting. in whole mounts, the brain was coronally cut from the rostral side until -2.30 mm rostral from bregma.17 then under the dissection microscope, dii was applied to the ca1 region of one side of the brain. the brain was then fixed to a silicon plate, which was glued down to the bottom of a beaker. the beaker was filled with pbs with 0.1% azide and incubated at 30 °c for 1-2 weeks. after incubation the brain was sectioned at 50 or 400 microns and then placed in a 4°c room before mounting. methods for dye application previously have not been well developed. the first method we applied was to use the tips of two syringe needles to push the dii into the slice. one needle was used to pick up the crystal from the dye jar and the other was used to slide the crystal off the first needle and onto the slice. then one or both of the needles were used to push the crystal underneath the surface of the slice. a second method employed was to use the tip of an eyelash to apply the dye. two small hairbrushes were made, consisting of a single eyelash applied to the end of a glass tube with superglue. the last and most successful method was to use the tips of microfils, which are nonmetalic syringe needles usually used to fill micropipettes, to apply the dye. the microfils were cut down so that they were approximately 1 cm away from their base and then they were attached to syringes to make them easier to use. imaging the slices the slices were mounted under coverslips onto microscope slides with mounting media. images were obtained using the ziess lsm 700 scanning confocal microscope with 1.25x, 5x, 10x, and 20x objectives. slices were imaged in sections, and these sections were then pieced together using image j and adobe photoshop software. new method the final method of labeling the circuitry was stereotaxically to inject the live mouse in the ca1 field with a lentivirus expressing gfp, at +/-1.9 (x), -2.2 (y), -1.5 (z1), -1.3 (z2) at least two weeks before being perfused and stored in 4% pfa. these slices were cut at 50 mm. the mossy fibers are clear and well known projections that arise from the dg and run to the ca3 field. two of the three dye application methods successfully labeled these projections in transverse slices of the fixed mouse hippocampus. the first method used syringe needles and the second method used the microfil tips. both application of the dye and preservation of the slice were easier with the second method, but either method would successfully stain the mossy fibers, as long as the crystal was successfully submerged in the dentate gyrus under the surface of the slice. figures 1a-1c show images of transverse slices. only the mossy fibers originating from the dg are labeled, along with the dg itself. therefore, the dye can be used to label specific projections. consistent with other experiments, it appears that the dye travels both retrogradely and anterogradely, but not transynaptically (herredia, 1991). figures 1a and 1b both display slices in which dye was applied with the needle syringe method. these slices were incubated for three days at 30 degrees c and then placed at 4 degrees c until being mounted. figure 1c shows a slice in which the dye was applied with microfil tip method. this made applying the dye easier and more accurate. this slice was incubated at 30 degree c for 2 days and then placed in a 4 degree c room until mounting. figure 2 shows a magnification of the mossy fibers traveling from the edge of the dg to field ca3. the dii crystals were applied using the syringe needle method. this slice was incubated for 4 days at 30 degrees c and then placed at 4 degrees until being mounted. the images show that both methods of applying the dye work well. the results also support the hypothesis that incubation time can vary without greatly impacting the effect of the dye. the fibers labeled after 2 days of incubation do not vary significantly from the fibers labeled after 4 days. the successful labeling of these projections allows these transverse slices to serve as controls. sagittal slices of the whole brain were cut in an attempt to preserve connections between the hippocampus and the ec. to test if and where these connections were preserved and where, dii labeling was used. in most sagittal sections, there were two ca1 sites, therefore, dye was placed in two locations in the hippocampus of each slice. these slices were 200 and 300 microns thick, had dii applied with the microfil tip to both the upper (dorsal) and lower (ventral) ca1 regions, and were incubated for 4 days at 30 degrees. one of the 200 microns slices, corresponding to image 127/128 (lateral 3.12mm/3.25mm) of the brain atlas, is � ��� figure 2� ��������;opz�pz�h�zlj[pvu�vm�h�[yhuz]lyzl�zspjl���;ol�tvzz`�äilyz�myvt�[ol�klu[h[l� n`y\z��+.��jslhys`�wyvqlj[�[v�[ol�*(��älsk���;ol�hyyv^�wvpu[z�[v�[ol�tvzz`�äilyz��;ol�+p0� crystals were applied using the syringe needle method. this slice was incubated for 4 kh`z�h[����klnyllz�*�huk�[olu�wshjlk�h[���klnyllz�\u[ps�ilpun�tv\u[lk� � � � � � figure 2a figure 2b figure 2c figure 3 ;olzl�hyl�pthnlz�vm�h�����tpyjvu�zhnp[[hs�zspjl�^p[o�[^v�+p0�jy`z[hsz���-pn\yl��(�pz�hu�pthnl�vm�[ol�[vw�vm�[ol�zspjl���;ol�^op[l�hyyv^�pz�^olyl�[ol�+p0�jy`z[hs�^hz� placed in ca1 and the blue arrow point to labeling in the top of cent. figure 3b is a close up of the connection between the two areas. figure 3c is the lower half of the slice and a white arrow shows where the dii crystal was placed and the blue arrow points to the cent. � � � � � �figure 4a figure 4b figure 4������������������������;olzl�pthnlz�hyl�vm�����tpjyvu�[opjr�zhnp[[hs�zspjlz�huk�zov^�shilspun�pu�[ol�*,u[���-pn\yl��(�zov^z�shilspun�pu�[ol�\wwly�ylnpvu�vm�[ol�*,u[�huk�än\yl��)�zov^z� labeling farther down in the cent. both labeling is light and a little hard to see, but it is clear under the microscope. the arrows point to where the dye is labeled in the cent. results columbia undergraduate science journal open-access publication | http://cusj.columbia.com 16 cusjvolume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu n eu ro sc ie n c e cusjcolumbia undergraduate science journalresearch articles 17cusj volume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu n eu ro sc ien c e shown in figure 3 (franklin 2007). this slice clearly has dye traveling out of the ca1 region and labeling in the ec. there is terminal labeling in the very upper region (most dorsal) of the caudomedial entorhinal cortex (cent) traveling from the top (dorsal) ca1 region as well as some light labeling from the lower (ventral) ca1 region to the dorsolateral entorhinal cortex (dlent) up into the cent. two other sagittal slices that were 400 microns thick also showed similar labeling. figure 4a shows one of these two slices, which had dye crystals placed in the upper (dorsal) ca1 region only and had an incubation time of 7 days. both slices had labeling in the same region as the slice previously mentioned. a third 400 micron slice had slightly different labeling that is more ventral in the cent shown in figure 4b. figure 5 shows labeling in the upper (most dorsal) region of the cent close to where the dye was placed in the upper (dorsal) ca1 region and also near the lower (ventral) ca1 region in the dlent and in the adjacent cent. since there were no crystals placed in the lower (ventral) region in ca1, this labeling must have come from the dye put in the upper (dorsal) ca1 region. in the first whole mount, dye was placed on the left side of the brain and the brain was cut sagittally into 100 micron thick slices. light labeling can be seen not only on the same side the dye was on, but also on the other side of the brain, confirming that the connections projecting from ca1 travel contralaterally. projections on the same side of the brain appear to travel to two places, the upper region of the cent as in previous sagittal slices, as well as the lower region of the dlent as also seen in previous sagittal slices. figure 6 displays this connectivity in a slice that is comprable to figure 130 in the mouse brain atlas (lateral 3.44mm). contralaterally, the projections terminate only in the upper region of the cent, as shown in figure 7, which approximately correlates to figure 128/129 in the atlas (lateral 3.25mm/3.36mm). in a second whole mount, dye was placed on the left side of the brain and the brain was cut horizontally into slices that were 100 microns thick. labeling can be clearly seen on the same side of the brain in the dorsolateral entorhinal cortex (dlent) in at least five of the slices. this labeling is shown in figure 8, which appears to correlate to altas figures 145 to 147 (bregma -3.96mm to -3.60mm). in the third whole mount, dye was placed on the left side of the brain and the brain was cut sagittally into slices that were 400 microns thick. in these slices, it is very clear that the fibers from the corpus callosum are turning in to the ec. it appears as though subiculum fibers from ca1 use the same track as fibers of the corpus callusum. after the corpus callosum ends, there appear to be tiny bright dots and thick fibers on the surface of the slice. because the corpus collusum connects the two halves of the brain, this data indicate that fibers are traveling contralaterally. this data can be seen in figure 9. in figure 10, a sagittal section from a gfp stereotaxic injection in ca1 is shown. in this figure, labeling is very clear in layer ii/iii and v/vi of the ec. however, no axonal projections can be seen, only small dots. mossy fibers of the hippocampus were clearly labeled using dii, running from the dentate gyrus to the ca3 field in 400 micron thick transverse slices. this labeling of well-defined and studied projections served to show how to best apply the dye and treat the slices to develop clear staining of pathways. only mossy fibers originating from the dg are labeled, which is significant because it indicates that the dye can be used to label specific projections. experimentation with sagittal slices was carried out in order characterize the relatively undefined and unstudied pathways running from the ca1 field of the hippocampus to the ec and from the ec to the ca1 field. so far, the dye has been placed in the ca1 field to observe how the projections run to the ec. several of the 300 micron sagittal slices appear to be promising, but further analysis of the slices is needed. after this analysis, the brains will be cut at a variety of different angles in an attempt to preserve the most connections between ca1 and the ec. applying dye with microfil tips worked better than the dye application using needles or the eyelash brush. the shortened microfils were rigid enough to apply the dye and also thinner and therefore more accurate than the needles. the microfils also proved to be less electrostatic and were less destructive to the slice. the eyelash brush proved to be completely ineffective at applying crystals underneath the surface of the slice, as it was not rigid enough. thirteen 400 micron thick transverse slices had dii crystal applied with needles. four of the slices were in� � � � � ��� � � figure 5a figure 5b figure 5c figure 5d figure 5 ;opz�zhnp[[hs�zspjl�ohk�vul�+p0�jy`z[hs�puzly[lk�pu�[ol�\wwly�*(��ylnpvu��[opz�pz�zllu�pu�än\yl��*���-pn\yl��*�zov^z�zvtl�shilspun�pu�[ol�\wwly�*,u[�ylnpvu�thyrlk� i`�[ol�hyyv �̂��-pn\yl��(�huk��)�zov^�shilspun�pu�[ol�tpk�ylnpvu�vm�[ol�*,u[�hz�^lss�hz�pu�[ol�+3,u[��iv[o�wshjlz�hyl�thyrlk�i`�hyyv^z����;olyl�^lyl�uv�jy`z[hsz� wshjlk�pu�[ol�sv^ly�ylnpvu�pu�*(��zv�[opz�k`l�t\z[�oh]l�jvtl�myvt�[ol�k`l�w\[�pu�[ol�\wwly�*(��ylnpvu���-pn\yl��+�pz�h�jsvzl�\w�vm�[ol�jvuulj[pvuz�il[^llu� projections from the upper ca1 region and the mid region of the cent. � � � � figure 6a figure 6b figure 6 ;opz�pz�h�����tpjyvu�[opjr�zhnp[[hs�zspjl�myvt�h�^ovsl�tv\u[�^olyl�[ol� k`l�^hz�puzly[lk�pu[v�*(��^opsl�[ol�iyhpu�^hz�z[pss�pu[hj[���/lyl�[olyl� is labeling in the upper region of the cent and the dlent. this is a slice from the same side of the brain that the dye was placed on. figure 7a figure 7b figure 7� ������;olzl�hyl�iv[o�����tpjyvu�[opjr�zhnp[[hs�zspjlz�myvt�h�^ovsl�tv\u[�myvt� the opposite side that the dye was put on. this shows there are contralateral connections from ca1 to the ec. both slices are labeled in the upper cent region. discussion columbia undergraduate science journal open-access publication | http://cusj.columbia.com 18 cusjvolume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu n eu ro sc ie n c e cusjcolumbia undergraduate science journalresearch articles 19cusj volume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu n eu ro sc ien c e cubated for 4 days and nine were incubated for 3 days. the first four slices all showed mossy fiber staining from the dg to ca3. of the nine, four showed some mossy fiber staining, but only two of those showed mossy fibers running all the way to ca3. all of the slices showed staining of the dg. four slices had dii applied with the eyelash brush. one was incubated for three days and the others for two days. because it was hard to get the crystal beneath the surface with this method, not one of the slices displayed mossy fiber staining. because the crystals stayed on top of the surface of the slice and could easily move around when the slice was put back into a well of pbs, the staining was random and diffuse. four slices had dii applied with the microfil tip and were incubated for two days. of these one was completely unsuccessful, one was very successful sending stained projections from the dg to ca3, and two were partially successful, clearly staining the dg and the beginning of the mossy fibers. to ensure that the dye had enough time to travel, it was important to leave the slices in incubation for as long as possible, though a significant difference between the slices incubated for two days and those incubated for four days was not seen. the clear labeling of the mossy fibers seems to be more dependent on getting the dye underneath the surface of the slice and into the correct location in the dg. although we originally wanted slices to be as thin as 50 microns in order to see as much detail as possible, such slices were too thin to apply dii crystals. any attempt at application caused these slices to tear immediately. applying the dii crystals to the 200-400 micron thick sagittal slices proved to be fairly successful. sometimes it was difficult to observe the exact pathway of the projections because the diffusion area of the dye was so large. clear projections can be seen to go from the upper ca1 region to the upper region of the cent. additionally, dye in the lower ca1 region appears to travel through the dlent and up into the cent. in total, five successful whole mounts were made. two were cut horizontally and three were cut sagittally. the first sagittal one appeared to have ipsilateral and contralateral projections, the ipsilateral projections went to the dlent and cent and the contralateral projections only went to the cent. in the horizontal orientation, whole mount light labeling can be seen in the ipsilateral dlent. more analysis of the whole mount slices and images still remains to be carried out. the whole mount method, though more technically challenging, can hopefully provide clearer results than applying the dye after slicing because it allows the dye to run throughout the brain and does not severe connections. another method of labeling connections was used to show connections between the ca1 field and the entorhinal cortex. a gfp lentivirus was injected in the ca1 field of a live mouse stereotaxically, at +/-1.9 (x), -2.2 (y), -1.5 (z1), -1.3 (z2) relative to bregma. preliminary analyses of the gfp labeling of this brain to be promising. there seems to be labeling in layers ii/ iii and v/vi; however, it is unclear which projections are contralateral because both sides of the mouse’s brain were injected with virus. the gfp method might prove more useful and clearer because there should be better accuracy when injecting virus into ca1 and not as much random diffusion as with the dye. the illumination of gfp appears as tiny dots rather than long axon fibers. we hypothesize that this is because the virus collected in the bouton at the end of the axons. to test this we will carry out antibody staining specific to the bouton. dii application in the dg of 400 micron thick transverse slices proved to be very successful and showed dii application in thick slices to label projections well. these projections ran from the dg to the ca3 field, showing the path of the well studied mossy fibers. application of the dii crystals to the sagittal slices has produced some positive results, but further investigation must be conducted. experimentation with the angle of sectioning will be performed to assess which plane is best for preserving the most connections from ca1 to the ec. the ultimate aim of this project is to develop methodology to carry out successful electrophysiological studies of acute slices, with a focus on the communication between the ec and the ca1 field of the hippocampus in order to further the investigation of memory storage. in the future, we will focus more on studying the connections between ca1 and the ec by stereotaxically injecting gfp virus. this method appears to be more promising as it labels more selectively than the dye, which somtimes diffuses too much and the causes � � � � � figure 8a figure 8b figure 8c figure 8d figure 8e figure 8 these are images from a whole mount cut horizontally pu�����tpjyvu�[opjr�zspjlz��� labeling is seen in the cent. -pn\yl��)�pz�h�thnupäjh[pvu� of 8a. figures 8d and 8e are thnupäjh[pvuz�vm��*� � � � � figure 9 figure 10������������������������;opz�pthnl�pz�vm�h�����tpjyvuz�[opjr�zhnp[[hs�zspjl���;opz�pthnl�kpzwsh`z�shilspun�vm�[ol�jvyw\z�jhssvz\t��hyyv^�wvpu[z�[v�[opz����;olzl�hyl�äilyz�[oh[� connect the two hemispheres on the brain. this is interesting because it means [oh[�äilyz�wyvqlj[pun�myvt�*(��hyl�nvpun�[oyv\no�[ol�jvyw\z�jvssvz\t�tlhupun� this is an image of a sagital slice of a gfp injected brain. this slice also zov^z�shilspun�pu�[ol�\wwly�*,u[�ylnpvu��jvuzpz[lu[�^p[o�+p0�shilspun�äukpunz� future directions columbia undergraduate science journal open-access publication | http://cusj.columbia.com 20 cusjvolume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu n eu ro sc ie n c e cusjcolumbia undergraduate science journalresearch articles 21cusj volume 5spring 2011 columbia undergraduate sci j http://cusj.columbia.edu m o lec u la r bio lo g y identifying serum-inducible sequence elements on the mouse zfp36 promoter max shutran1,*, ron prywes 2,* 1tufts university school of medicine, boston, ma 02110, 2department of biological sciences, columbia university, new york, ny 10027 apuj�äunly�wyv[lpu�����a-7����pz�hu�95(�ipukpun�wyv[lpu�[oh[�kv^uyln\sh[lz�[ol�[yhuzsh[pvu�vm�jly[hpu�nlul�wyvk\j[z�i`� inducing the degradation of mrna. zfp36 is an immediate-early gene: its transcription is activated directly following appropriate cellular stimuli, without any need for intervening protein synthesis. prominent mrna targets of zfp36 are tumor uljyvzpz�mhj[vy����;5-�����lwpklyths�nyv^[o�mhj[vy��,.-��huk�j�mvz��(z�klmlj[z�pu�[ol�j�mvz�huk�,.-�wh[o^h`z�oh]l�illu� consistently found in cancers, zfp36 is thought to be a tumor-suppressor. zfp36 also is responsible for the suppression vm�;5-����np]pun�p[�hu�hkkp[pvuhs�yvsl�pu�[ol�jvu[yvs�vm�[ol�puåhtth[vy`�ptt\ul�ylzwvuzl��;o\z��z[\k`pun�[ol�yln\sh[pvu� huk�hj[pvu�vm�a-7���th`�`plsk�puzpno[z�pu[v�tljohupztz�\uklys`pun�jhujly�huk�puåhtth[vy`�z`ukyvtlz��;v�l_htpul�[ol� regulation of zfp36, two conserved regions upstream of the start of transcription were assayed for their ability to confer serum induction. using a luciferase reporter gene system in hela cells, we show that neither region examined is induced by serum. key words: electron transfer, photosynthesis, quinones, uv-vis spectroscopy immediate-early genes are characterized by immediate transcriptional activation upon appropriate stimulation, including treatment of cells with mitogen-containing serum, without any need for intervening protein synthesis (johansen, 1995). as they are among the !rst genes to be activated in many signaling pathways, immediate-early genes provide a gateway to study the cellular response. zinc !nger protein 36 (zfp36) is an immediate-early gene product that regulates the expression of some genes by destabilizing their mrna. "e protein binds to au-rich sequences on mrna and catalyzes removal of the polyadenylated tail, reducing the half-life of the rna (blackshear, 2002). "e zinc !nger motif is common in dna and rna binding proteins, though zfp36 is unusual in that it has a cys3-his zinc-coordinating domain, as opposed to the more common cys2-his2 domain (tchen, 2004). "is slight di#erence in peptides might subtly impact the nucleotide binding dynamics of zfp36. notable genes regulated in part by zfp36 include tumor necrosis factor-ơ� (tnf-ơ), epidermal growth factor (egf) and c-fos, a transcription factor that promotes proliferation (amit, 2007) mice de!cient in zfp36 develop an excess of tnf-ơ, and the overabundance of this cytokine results in autoimmune complications such as arthritis (lai, 1998). as egf and c-fos are both implicated in proliferation, zfp36 is also thought to act as a tumor-suppressor (amit, 2007) "e zfp36 gene is regulated in several ways. tchen and colleagues et al. have shown that the zfp36 gene is mitogen-inducible via protein kinase p38 (tchen, 2004). "is result is consistent with qpcr data demonstrating that zfp36 is induced (upregulated) strongly by serum after one hour (figure 1). however, the upregulation seems to be independent of serum response factor (srf), the transcription factor that primarily mediates the serum response (prywes, unpublished data). lai and colleages showed that there are important regulatory elements downstream of the start of zfp36 transcription, and at least one of those regulatory elements is on an intron (lai, 1995; lai, 1998). given that zfp36 is induced by serum one hour after induction, we hypothesized that there may be additional serum-dependent regulatory elements upstream of the start of transcription. we identi!ed two conserved regions in the promoter region of mouse zfp36 and used a luciferase reporter gene assay to test the ability of each region to confer serum induction in hela cells. however, we ultimately ascertained that neither region was able to confer serum induction of zfp36 in hela cells. "is study and other studies of the regulation and action of zfp36 will provide vital information about this protein, and further research will play a signi!cant role in gaining a better understanding of the biological pathways for proliferation and in$ammation. introduction copyright: © 2009 the trustees of columbia university, columbia university libraries, some rights reserved, albert, et al. received dec. 29, 2009. accepted feb. 8, 2009. published april 5, 2009. *to whom correspondence should be addressed: columbia university, department of biology. 813b fairchild center. neew york, ny 10027 max.shutran@tufts.edu mrp6@columbia.edu the connections to appear unclear. by stereotaxically injecting virus we can also be more accurate and consistent with future injections into the mouse brain. different orientations of slicing must be tested in order to find the optimal orientation for preserving connections. once the connection from ca1 to the ec is clearly established, we also plan to examine the connections projecting from the ec to ca1. this anatomical data will contribute to the success of future electrophysiological studies. i would like to thank dr. siegelbaum for giving me the amazing opportunity to perform research in his lab and rebecca piskorowski for her helpful advice, constant guidance, and patience. i would also like to thank justine barry for her frequent and important support, guidance, and instruction, sebastian thuault for one of his mouse brains, and the rest of the siegelbaum lab. buzsaki, g. 1989. two-stage model of memory trace formation: a role for “noisy” brain states. neurosci 31: 551-570. de curtis, m., d. paré, and r.r. llinás. the electrophysiology of the olfactory-hippocampal circuit in the isolated and perfused adult mammalian brain in vitro. hippocampus. 1991 oct;1(4):341-54. eichenbaum, h., a.p. yonelinas, and c. ranganath. “the medial temporal lobe and recognition memory.” annu. rev. neurosci. 2007.30:123-152. eldridge, l.l., b.j. knowlton, c.s. furmanski, s.y. bookheimer, s.a. engel. nat neurosci. remembering episodes: a selective role for the hippocampus during retrieval. 2000 nov;3(11):1149-52. franklin, keith b. j., and george paxinos. the mouse brain in stereotaxic coordinates, third edition. new york: academic, 2007. godement, p., j. vaselow, s. thanos, and f. bonhoeffer. (1987) a study in developing visual systems with a new method of staining neurons and the processes in fixed tissue. development, 101: 697-713. heredia, m., m. santacana, and f. valverde. a method of using dii to study the connectivity of cortical transplants. journal of neurosci methods. 36 (1991) 17-25. honig, m.g. and r.i. hume (1985) interactions between sympathetic preganglionic neurons and sympathetic ganglion neurons in vitro. neurosci. abstr. 11:98. hyman, b.t., g.w. van hoesen, a.r. damasio, and c.l. barnes. alzheimer’s disease: cell-specific pathology isolates the hippocampal formation. science 225: 11681171. johnston, daniel, and david g. amaral. “hippocampus.” the synaptic organization of the brain. ed. gordon m. sheperd. 4th ed. new york: oxford up, 1998. print. naber, p. a., f. h. lopes da silva, m. p. witter. reciprocal connections between the entorhinal cortex and hippocampal fields ca1 and the subiculum are in register with the projections from ca1 to the subiculum. hippocampus. 11 (2001) 99-104. o’keefe, john. a review of the hippocampal place cells. prog neurobiol 13: 419–439, 1979. tamamaki, n. and y. nojyo. (2003) preservation of topography in the connections between the subiculum, field ca1, and the entorhinal cortex in rats. journal of comparative neurology, 353: 379-390. van groen, t., i. kadish, and j. m. wyss. species differences in the projections from the entorhinal cortex to the hippocampus. brain res bull. 2002 feb-mar 1;57(34):553-6. van groen, t., p. miettinen, and i. kadish. the entorhinal cortex of the mouse: organization of the projection to the hippocampal formation. hippocampus. 2003;13(1):133-49. vindal-sanz, m., m.p. villegas-pérez, g.m. bray, and a.j. aguayo. (1988) persistent retrograde labeling of adult rat retinal ganglion cells with carbocyanine dye dii. exp. neurol., 102:92-101. witter, m.p., floris g. wouterlood, pieterke a. naber, and theo van haeften. anatomical organization of the parahippocampal-hippocampal network. ann n y acad sci. 2000 jun;911:1-24. zola-morgan s., l.r. squire, and d. g. amaral. human amnesia and the medial temporal region: enduring memory impairment following a bilateral lesion limited to field ca1 of the hippocampus. j neurosci. 1986 oct;6(10):2950-67. references (jruv^slknltlu[z © 2023 lang, elvis. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 7 design of a genetically encoded tool to study the function of mitochondria–lysosome interactions yuliia kohut1, robert coukos1, dimitri krainc1 1department of neurology, northwestern university fienberg school of medicine, chicago, il, usa keywords: biosynthetic tools, membrane contact sites, mitochondria-lysosome contacts, abi-pyl domain, immunofluorescence, confocal microscopy, neurodegeneration abstract: membrane contact sites are regions of close proximity between organelles that mediate crucial organelle functions. however, the list of effective biological tools to study membrane contact sites function still needs expansion. mitochondria and lysosomes are organelles whose dysfunctions are closely associated with parkinson’s disease pathology, and some parkinson’s disease-associated genes have been shown to play a role in mitochondria-lysosome contact regulation. a novel biosynthetic system for reversible induction of interactions between mitochondria and lysosomes is presented, leveraging abscisic-acid-mediated heterodimerization of abi and pyl protein domains from arabidopsis thaliana. combinations of genetically encoded protein heterodimer constructs each containing the abi or pyl domain, a fluorescent protein tag (egfp, mcherry, irfp, or bfp), and a transmembrane sequence targeting the lysosomal membrane or outer mitochondrial membrane were engineered. using confocal microscopy to visualize organelle colocalization, the efficient and reversible induction of mitochondria-lysosome contacts is shown with the abscisic-acid-responsive system. findings indicate that mitochondrial constructs tagged with mcherry and irfp effectively mediate abscisic-acid-induced mitochondria-lysosome colocalization when paired with lysosomal constructs tagged with egfp/bfp or lacking a fluorescent tag. in live-cell imaging experiments, mitochondrialysosome contact durations are tunable with this system, increasing proportionally with abscisic-acid concentration used. overall, the system of abscisic-acid-induced reversible organelle interactions is positioned as a powerful biological tool for studying organelle dynamics and membrane contact sites functions, with relevance to therapeutic strategies in neurodegenerative disorders and organelle dysfunction. introduction membrane contact sites (mcs) are areas of close proximity between two membranebound organelles that fulfill a specific function and that are distinct from incidental membrane association or fusion events. the involvement of organelle contacts has been implicated in a variety of cellular functions, including organelle fission, trafficking, and cellular metabolism. importantly, misregulation of mcs is a feature of several neurodegenerative diseases, though the precise contribution of disrupted mcs to disease pathology remains an important question.1,2 while the scientific field realizes the significance of studying organelle contacts, developing effective molecular tools to probe their functions remains a limitation. in this paper, the design of a biosynthetic system that can be used to study mcs function via the reversible induction of mitochondria-lysosome contacts is presented. mitochondria are sites of essential cellular processes such as adenosine triphosphate production, oxidative phosphorylation, apoptosis, calcium homeostasis, and lipid storage, which makes them central sites of cellular metabolism.3 mitochondrial dysfunction and dysregulation are also columbia undergraduate science journal vol. 19, 2025 kohut et. al © 2025 kohut, yuliia. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 8 hallmarks of aging, neurodegeneration, cardiovascular problems, and even cancer.4 likewise, lysosomes are versatile organelles critical to metabolic regulation and cell homeostasis.5 they are prominently understood to function in the recycling of proteins and lipids, but an underappreciated functional corollary is their role in storing metabolites and signaling nutrient status to the cell in order to direct metabolic programming.6,7 importantly, impairment of normal lysosomal function is oftentimes associated with neurodegenerative diseases.8,9 mitochondria-lysosomal contacts are more transient than other organelle contacts, but they still share similar characteristics, such as metabolite transfer and dedicated tether proteins.10,11 recent research suggests that mitochondria-lysosomal contacts play an important role in the regulation of localized protein synthesis in neurons, which raises the importance of this type of organelle contact in the context of neurodegenerative disease and pathology.1 in particular, genes associated with parkinson’s disease (pd) appear to cluster around mitochondrial and lysosomal processes, and mitochondrialysosomal contacts are often disrupted by mutations in these genes, which suggests that the crosstalk between mitochondria and lysosomes may feature prominently in pd pathogenesis.12 many biological techniques have been used to characterize and study membrane contact sites. the most popular used in mcs studies are electron microscopy (em), proximity ligation assay (pla), live-cell confocal microscopy, and single molecule imaging involving fluorescence resonance energy transfer (fret), or structured illumination microscopy (sim).13,14 these techniques have unique advantages in the study of mcs, however, each technique faces limitations. for example, sim, fret, and live-cell confocal microscopy can be used on live samples but demand specialized equipment, while em and pla require tissue fixation and dehydration procedures that may disrupt the original architecture of the contact sites, which in turn prevents the study of dynamics and function in living cells. moreover, to examine the causal effect of contacts on other cellular functions, there is a need for a biological system that will allow us to directly manipulate the induction of mitochondria-lysosome contacts instead of passively observing them. this issue is addressed by proposing a biosynthetic inducible system of reversible organelle interactions, where the small molecule abscisic acid (aba) is used as a mediator of mitochondria-lysosome colocalization. aba is a plant stress hormone that mediates heterodimeric complexation of proteins containing abi (aba insensitive 1) and pyl (pyrabactin resistance-like) domains.15 to induce colocalization between mitochondria and lysosomes, protein constructs were engineered on the membranes of each organelle to include these compatible domains. fluorescent protein tags were included to visualize mitochondria-lysosome colocalization via confocal microscopy. interactions between organelles were induced by introducing aba to cells that express these domains on the organelle membranes. organelle contact was reversed via washout of aba from the cells to relax the association of heterodimer complex. different mitochondrial and lysosomal membrane construct variants are tested to identify more compatible pairs with additional biological assays. columbia undergraduate science journal vol. 19, 2025 kohut et. al © 2025 kohut, yuliia. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 9 methods plasmids the plasmids were designed and cloned manually. addgene plasmid #171013 was used as a backbone for designing each plasmid. plasmids were designed that contained the following coding sequences: 3xmyc-pyl-mcherry; 3xmyc-pyl-irfp; 3xmyc-pyl; tmem192-abi-egfp-3xha; tmem192-abi-bfp-3xha; and tmem192abi-3xha. cloning of constructs abi and pyl domain sequences were generated by synthetic gene fragment synthesis, featuring mammalian codon optimization of the domains present in addgene plasmid #38247. lentiviral plasmids containing the mitochondria and lysosomal targeting domains were linearized by restriction digestion of addgene plasmid 171013 in-frame with the targeting domains. polymerase chain reaction (pcr) was used to generate products containing the abi/pyl domains and products containing fluorophores and epitope tags. primers in these pcr reactions were designed to have overlapping homology regions. then, gibson ligation reactions were used to combine pcr reactions into the digested backbones. finally, plasmids were transformed into e. coli and individual bacterial clones were screened for successfully combined plasmids, which were then purified from bacterial dna preparations for preparation of lentivirus. transfection of lysosomal and mitochondrial plasmids lentivirus generation and a stable transduction of constructs into hela cells was performed. cell culture and passaging of hela cells cells were regularly passaged to maintain log-phase growth in fbs-supplemented dmem. when cells were nearly confluent, they were passaged with trypsin to lift and separate cells from the surface of the dish to which they were adhered, then the cells were pelleted in a centrifuge, resuspended in dmem, and appropriate dilutions of cells were added to a fresh plate. treatment with abscisic acid and mitochondrial stressors after aspirating old cell culture media, new media with either 1 µm, 500 nm, or 300 nm aba was added to the cells. hela cells were then incubated for 90 minutes in the course of one aba treatment. for treatment of hela cells with mitochondrial stressors, the following drug concentrations were used with their respective incubation times: 5 µm mitoparaquat for 60 minutes, 5 µm rotenone for 90 minutes, 5 µm oligomycin for 120 minutes, 5 µm cccp for 150 minutes, and 5 µm antimycin a for 180 minutes. immunoblotting proteins from cell lysates were prepared and resolved in 4-20% sds-page gradient gel at 140v for 1 hour and transferred on a pvdf membrane for 15 minutes via semi-dry transfer. membranes were blocked with 1% bsa-tbst for 1 hour. primary antibody stain was applied in 1% bsa-tbst and membranes were incubated overnight at 4c. the following primary antibodies and concentrations were used: 1:500 of ha (rb, cell sig. 3724s), and 1:1000 of myc (rb, cell sig. 2272s). membranes were then washed 4 times for 20 minutes total with tbst solution and incubated with 1:10,000 of goat antirabbit hrp (jackson 115-035-144) secondary antibody for 1 hour in 1% bsatbst. the membranes were washed again 4 times for 20 minutes total with tbst solution and stored in di water until they were imaged with horseradish peroxidase substrate. images were edited using fiji image analysis software. columbia undergraduate science journal vol. 19, 2025 kohut et. al © 2025 kohut, yuliia. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 10 immunofluorescence in preparation for immunofluorescence, cells were grown on number 1.5 thickness slide covers. after corresponding aba treatment, media was aspirated from all samples and formaldehyde was added to each sample. samples were then washed with pbs and transferred into the blocking solution for 30 minutes. afterwards, samples were treated with 1:500 of ha (rb, cells sig. 3724s) and 1:500 myc (mo, cell sig. 2276s) as primary antibodies and incubated at room temperature for 1 hour. next, samples were washed with pbs again, treated with 1:1000 of rb-488 (invitrogen a21206 or a11034) and 1:1000 of mo-568 (invitrogen a11031) secondary antibodies, and incubated at room temperature for 30 minutes. samples were washed with pbs again, inverted, and adhered to slides with liquid mountant. confocal microscopy microscopy was performed using a nikon axr confocal microscope with a 4-laser system (408, 488, 561, 640 nm), with corresponding emission filters and multispectral detectors. the system was equipped with perfectfocus system and co2 and a heat-controlled tokai hit chamber for live-cell imaging. nikon elements imaging software was used. image analysis was conducted using cell profiler and fiji imagej. anova p-test with post hoc tukey test were conducted to determine significance between signal reading before aba treatment, with aba treatment, and after washout of aba. live cell microscopy imaging parameters were 1024x1024 pixels, encompassing a space of 73.66 um by 73.66 um. this calculates to a resolution of 72 nm per pixel, though the true resolution limit for conventional light microscopy is ~200 nm. image series were taken at 3 second intervals over a span of 3 minutes, for a total of 61 images per times series. over the course of imaging, 5% co2 and 37°c were maintained. post-hoc, 1 nm radius rolling-ball method background subtraction was applied using fiji to facilitate analysis. blinded analysis was performed on the images. during image series analysis, any lysosomal fluorescence signal that completely overlapped or visually touched the surface of mitochondrial signal on the first image in the series was considered to be in contact with mitochondria. next, the proximity of such lysosomes with mitochondria was tracked with each consecutive image. when a tracked lysosome dissociated from mitochondrial signal and proceeded to move away for more than 2 images (6 seconds), the number of the last image that featured mitochondriallysosomal contact was recorded as the duration of contact in seconds. about 15 lysosomes per image series were tracked for contact duration in different locations inside a given cell. results in the initial design of the inducible system for mitochondria-lysosome interactions, lysosomal transmembrane protein 192 (tmem192) was modified with c-terminal fusion of an abi domain, a green fluorescent protein (egfp), and triplicate ha epitope tag (for antibody-based recognition). additionally, the transmembrane domain of the mitochondrial protein synj2bp was modified with n-terminal fusion of the fluorescent protein mcherry, the pyl domain, and a triplicate myc epitope tag (figure 1a). when hela cells expressing these constructs are treated with aba, aba links abi and pyl domains together, which forms a synthetic tether between the two organelles. to test if this interaction is enough to induce mitochondria-lysosome contacts in human cells, confocal microscopy and immunofluorescence were used to measure the colocalization between mitochondria and lysosomes. columbia undergraduate science journal vol. 19, 2025 kohut et. al © 2025 kohut, yuliia. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 11 figure 1. a system to reversibly induce mitochondria-lysosome contacts. a) aba mitochondrial and lysosomal heterodimer constructs. abi/pyl are the aba responsive domains; tmd – synj2bp transmembrane domain. b) fluorescence microscopy imaging of construct colocalization in hela cells with or without 1 µm aba. c) quantification of mitochondria-lysosome association in individual cells for each treatment. scale bars in (b) are 10 µm. aba treatment duration is 30 minutes, and the duration of the washout is 90 minutes. to test the system, hela cells expressing mitochondrial and lysosomal constructs were treated with dmso (null treatment) or 1 µm aba, followed by fixation of the samples. fluorescence microscopy protocols were implemented to image fluorescence signals from lysosomes and mitochondria. the percentage of lysosomal fluorescence intensity was quantified in pixels of overlapping mitochondrial and lysosomal signal to define the extent of mitochondria-lysosome colocalization. while organelles contacts are classically defined as distances (10-80 nm) below the resolution limit of conventional light microscopy (~200 nm), the protein heterodimerization mediated by aba would indeed tether the two organelles at distances within the classically defined range.2 thus, while not an absolute measure of contacts, changes in mitochondria-lysosome colocalization due to aba treatment will strongly correlate with changes in contacts (figure 1a). under baseline conditions, such as when hela cells are treated with dmso control, only some lysosomes colocalized with mitochondria. however, treatment of cells with 1µm aba was visually striking, with lysosomes colocalizing with the mitochondrial membrane to produce pixels of vivid, bright white overlap between fluorescent signals (figure 1b). when aba is washed out, the cells returned to baseline distribution of the organelles, with most lysosomes dissociated from mitochondria (figure 1b). after quantifying mitochondria-lysosome colocalization, only 35% of the lysosomal intensity colocalized with mitochondria in dmso (figure 1c). upon 1µm aba treatment, the extent of mitochondrialysosome colocalization significantly increased to 85%, consistent with qualitative observations. this substantial increase in colocalization suggests a strong inducible effect of the abi-pyl domain system. the colocalization percentage returned to 40% after washing out aba, indicating the reversibility of contact induction with aba. significant statistical differences were found between the colocalization signals in each group (figure 1b), which confirms that colocalization between two organelles was induced and reversed after aba wash-out. these results demonstrate that the abi-pyl heterodimer works well for inducing and reversing proximity between organelle membranes expressing these domains. after the system was seen for reversible organelle contacts working with egfp-tagged lysosomes and mcherry-tagged mitochondria, the next question was what other fluorescent protein pairs can effectively visualize columbia undergraduate science journal vol. 19, 2025 kohut et. al © 2025 kohut, yuliia. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 12 mitochondria-lysosome colocalization. egfp and mcherry are among the brightest and most stable fluorophores and are thus suitable choices for experiments in which only lysosomes and/or mitochondria are being imaged. however, other biological assays, such as those which measure mitochondrial reactive oxidant species (ros), mitochondrial membrane potential, or lysosomal ph, often feature red or green fluorescent probes. these probes usually have significant spectral overlaps with egfp and mcherry, requiring use of alternative fluorophores to visualize these probes simultaneously with the lysosomal and mitochondrial protein constructs. hence, further investigation focused on identifying fluorophore tags that would be effective for contact induction with the aba-induced system and compatible with assays that use green or red fluorescent indicators. mitochondrial mcherry construct performance was compared to the same construct lacking the mcherry fluorophore and the construct with infra-red fluorescent protein (irfp670 or “irfp”) instead of mcherry. figure 2. mitochondrial and lysosomal variants. a) constructs of pyl with mcherry/no-fp/irfp on mitochondrial tmd, paired to lysosomal construct with abi and egfp. b) constructs of abi with egfp/no-fp/bfp on lysosomal tmem192pyl, paired to mitochondrial construct with pyl and mcherry. an additional aim was to determine if there would be an advantage in removing egfp from the lysosomal membrane construct and if blue fluorescent protein bfp would provide a similarly effective signal to egfp. hence, six constructs with different fluorescent protein tags on mitochondrial and lysosomal membranes were design. the following combinations of fluorophore tags were tested: lysosomal-egfp against mitochondrialmcherry/no-fp (no-fluorescent protein)/irfp as well as mitochondrial-mcherry with lysosomal egfp/no-fp/bfp (figure 2a construct pairs 1m, 2m, 3m; figure 2b construct pairs 1l, 2l, 3l). figure 3. testing aba-induced lysosomemitochondria colocalization of mitochondrial construct variants. a) immunofluorescent imaging of construct colocalization in hela cells with or without 30minute treatment of 500 nm aba. b) quantification of mitochondria-lysosome association in individual cells. *p values for each variant combination provided by 1-way anova with post hoc tukey test between treatments. scale bars in (a) are 10 µm. columbia undergraduate science journal vol. 19, 2025 kohut et. al © 2025 kohut, yuliia. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 13 hela cells that expressed the different combinations of construct pairs were treated with 500 nm aba or dmso (null treatment) for 30 minutes before fixation or with aba treatment and 90 minutes of washout before fixation (figure 3a). lysosomal colocalization was quantified for each treatment (figure 3b). to keep quantification consistent across samples, immunofluorescence was performed against the ha and myc tags present in each construct, rather than taking the direct measurement of the fluorophores. constructs expressing mitochondrial mcherry and irfp (figure 3a, construct pairs 1m and 3m respectively) produced the same profile of induced and reversible contacts. on microscopy images, hela cells were seen expressing both 1m and 3m construct pairs showed a similar lysosomal colocalization to mitochondria upon aba treatment and the same relaxation of contacts after the washout (figure 3a, construct pairs 1m and 3m), which correlates with an increase in organelle colocalization from 40% to 70% (figure 3b, mito-mch and mito-irfp). similarly, both construct variants exhibited return of mitochondria-lysosome colocalization to the baseline of about 40% colocalization (figure 3b, mito-mch and mito-irfp). in contrast, it appears that cells expressing mitochondrial no-fp variant were unable to mediate mitochondria-lysosome contact induction, as there are no lysosomes colocalizing with the mitochondria upon aba treatment on microscopy (figure 3a, construct pair 2m). furthermore, there is no change in extent of colocalization qualitatively (figure 3b, mito no-fp). it is possible that removing the fluorophore from the mitochondrial construct variant situates the pyl domain too close to the mitochondrial membrane which sterically impedes the abi-pyl complex from forming. this leads to absence of contact induction upon aba treatment. figure 4. testing aba-induced lysosomemitochondria colocalization of lysosomal construct variants. a) immunofluorescent imaging of construct colocalization in hela cells with or without 30minute treatment of 500 nm aba. b) quantification of mitochondria-lysosome association in individual cells for each construct combination in each treatment. a similar approach was taken to test hela cells expressing varying lysosomal constructs. cells expressing all lysosomal egfp, no-fp, and bfp constructs produced visually the same contact induction profile with 70% mitochondria-lysosome colocalization (figure 4). after aba treatment, cells expressing the lysosomal no-fp construct variant showed increased mitochondria-lysosome colocalization, but most lysosomes remained in contact with mitochondria after the washout with about 55% colocalization post-wash compared to egfp baseline value of 30% (figure 4a construct pair 2l, figure 4b lyso no-fp, *p values for each variant combination provided by 1-way anova with post hoc tukey test between treatments. scale bars in (a) are 10 µm. columbia undergraduate science journal vol. 19, 2025 kohut et. al © 2025 kohut, yuliia. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 14 lyso egfp). as expected, the design of the the lysosomal no-fp construct did not change the distance between abi and the organelle membrane (figure 2a, construct pair 2l). thus, in this case, removing the fluorophore should not have had as strong of an impact on contact induction as in the case of the mitochondrial no-fp construct. interestingly, changes in lysosomal morphology were present during contact induction in cells expressing lysosomal nofp construct, where lysosomes clustered more in cells expressing lysosomal no-fp construct than in cells expressing other constructs (figure 4a, construct pair 2l). the lack of fluorophore in the lysosomal construct may allow for tighter packing of the heterodimer complex, which could explain this morphological change as well as the reduced recovery after washout. some lysosomes remained in contact with mitochondria post-washout in cells expressing lysosomal bfp (figure 4a, construct pair 3l), which correlated with a slightly diminished post-washout lysosomal colocalization of about 40% (figure 4b, lyso bfp). if a construct variant is expressed at a higher level, there may be increased binding opportunities at contact sites, leading to more stable tethers that make relaxation of contacts upon aba washout more difficult. thus, it was hypothesized that differences in expression levels of the heterodimer constructs might underlie small differences in the reduction of mitochondria-lysosome colocalization after washout. to address this question, immunoblotting on the whole-cell lysate of the hela cell lines expressing the construct variants was performed. toward this end, blotting was carried out for the lysosomal ha and mitochondrial myc tags, incorporated in the lysosomal and mitochondrial construct designs (figure 2, figure 5). cell lines with lysosomal-egfp and a mitochondrial variant had similar ha expression across construct variants. figure 5. immunoblotting of the whole-cell lysate to examine construct level expression. western blots against ha (a) and myc (b) in whole-cell lysate of hela cells expressing tested construct variants. proteolysis product is indicated with an asterisk * and the band in 2m line (b) is a full length and a lower weight construct variant product, but not a proteolysis product. this data is a result of a single immunoblotting experiment (n=1). however, myc expression levels were lower in cells with 2m construct, which correlates with the lack of contact induction in 2m cells observed during microscopy experiments (figure 3b, mito no fp). in cell lines with mitochondrial mcherry and a lysosomal variant, there were similar ha expression across all lysates. there was a higher myc expression in cells with 2l and 3l construct pair variants, which aligns with reduced washout efficiency in cells with these constructs during microscopy (figure 4b). these results suggest that the expression of the lysosomal construct is a limiting factor in the stability of the induced mitochondrialysosome contacts in the aba-directed system. as the immunoblotting data was derived from a single experiment, further replicates are required to draw definitive conclusions. columbia undergraduate science journal vol. 19, 2025 kohut et. al © 2025 kohut, yuliia. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 15 figure 6: quantification of mitochondrialysosome contact duration. a) average contact durations when hela cells are treated with 0 nm, 30 nm, 100 nm, and 300 nm aba. b) average contact durations when hela are treated with mitopq, cccp, oligomycin, antimycin a, and rotenone. after extensively testing aba as a tool to induce mitochondria-lysosome contacts, this system was used to assess whether the extent of contact induction could be modulated by varying the concentration of aba. in addition to quantifying lysosomemitochondria overlap, the contact duration is another important measure for defining the in-duction of bona fide organelle contacts.10 organelle contacts are highly dynamic processes that involve multiple native or artificial tethers, such as the ones induced with aba. thus, the duration of organelle contacts reflects the equilibrium of the number of active tethers, which are expected to increase with increasing aba. to test this hypothesis, hela cells were transduced with mito-mcherry/lyso-egfp constructs and treated with aba. there were statistically significant increases in average contact duration in cells treated with increasing aba concentrations (figure 6a). these results confirm that it is possible to tune the extent of contact induction in abainduced systems by manipulating abscisic acid concentrations. in addition to studying induced mitochondrialysosome contacts, the fluorophores on mitochondria and lysosome construct pairs can be used to track changes in contacts produced by other sources of cellular stress. live cell imaging and contact quantification were used explore the effect of mitochondrial stressors on the duration of mitochondrialysosome contacts. altered contacts induced by these stressors could be indicators of cellular responses involving lysosomes, such as calcium exchange, metabolite transfer, or quality control of damaged proteins or membranes. cells were treated with five drugs (mi-to-paraquat, cccp, oligomycin, antimycin a, and rotenone) that induce mitochondrial stress through various mechanisms. contact duration in each treatment averaged close to 100 seconds, but there was no statistically significant differences between null treatment and mitochondrial stressors with respect to contact duration (figure 6b). none of mitoparaquat, cccp, oligomycin, antimycin a, or rotenone have any effect on the duration of mitochondria-lysosome contact duration at the concentrations and treatment times. discussion the system for aba-mediated heterodimerization represents a novel tool for the reversible induction of organelle contacts. in immunofluorescence and live-cell confocal microscopy imaging, the system demonstrated tunable control of mitochondrialysosome contacts, representing a powerful biosynthetic tool for studying dynamic organelle interactions in human cells. to maximize the applicability of this tool, it was investigated whether it was compatible * p values are provided by 1-way anova with post hoc tukey test between treatments. *p values are provided by dunnett’s test (95% family-wise confidence level). "***" p < 0.001, "**" p < 0.01, "*" p < 0.05, "†" p < 0.1, columbia undergraduate science journal vol. 19, 2025 kohut et. al © 2025 kohut, yuliia. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 16 with various fluorescent protein tags. mcherry and irfp as mitochondrial tags were found to be equally effective for visualizing organelle colocalization in the aba-induced system. additionally, bfp in the lysosomal construct shows comparable results to egfp, suggesting that the fluorescent protein tag on each construct can be altered without sacrificing the efficacy of reversible organelle contact induction. while the lysosomal construct could also tolerate the removal of the fluorescent protein tag, excision of the mitochondrial fluorophore tag abolished aba-inducible contact formation. this may be due to steric effects from shortening the distance between the transmembrane and pyl dimerization domains. optimizing geometric parameters of constructs, such as lengths of linker regions between functional domains, is an important aspect of protein engineering. with additional design iteration, it should be possible to generate a mitochondrial construct without fluorophores that properly mediates organelle contacts in the presence of aba. the immunoblotting experiments highlighted that the extent of contact formation between organelles shows some dependence on expression levels of the protein constructs, especially regarding changes in the lysosomal construct organization. additionally, increasing aba doses highlighted the tunability of the system. knowing that contact levels and their duration can be directly manipulated by changing abscisic acid dosage can be useful in future experiments that aim to balance the effect size of other genetic tools or chemical perturbations to mitochondria-lysosome contacts. while the investigation into the effect of mitochondrial stress on contact duration did not yield positive results, there is no definitive conclusion that mitochondrial stress does not affect mitochondria lysosome contact induction. the drugs used in the experiment induced mitochondrial stress via disruption of the electron transport chain (etc) activity or by generating reactive oxidant species. additional types of mitochondrial stress, along with different dosages and treatment durations, should be investigated in the future to study their effect on mitochondria-lysosome contact induction. future studies could investigate how lysosomal stress influences the formation of mitochondria–lysosome contacts, as it remains unclear which organelle provides the primary signal initiating this interaction. conclusion in this study, aba-responsive heterodimerization was demonstrated to be an effective biosynthetic approach to induce reversible mitochondria-lysosome interactions. data indicates that the developed system is quantifiable and modular with respect to the choice of fluorophore tags. substantial significance is emphasized because the ability to manipulate organelle proximity through the abi-pyl system provides new ways to investigate the functional roles of mitochondria-lysosome contacts. developing effective molecular proximity tools for mcs detection is an ongoing challenge in the field and has limited the ability to examine the effects of these contacts on organelle dynamics and metabolic processes. by providing means to induce and reverse interactions dynamically, the system offers a valuable model to test hypotheses about how these organelles communicate in normal physiological function or in the context of disease. importantly, although only the induction of mitochondrialysosome contacts was demonstrated, the paradigm of abi-pyl membrane protein heterodimers should be applicable to studies of other organelle contacts. the impaired function of mitochondria and lysosomes have been studied separately from each other for a long time in the context of columbia undergraduate science journal vol. 19, 2025 kohut et. al © 2025 kohut, yuliia. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 17 neurodegenerative diseases like pd, charcot-marie-tooth disease, and others. however, it is becoming increasingly clear that processes that emerge from their dynamic association must be studied.1 thus, tools like the presented aba-induced system for reversible mitochondria-lysosome contacts provide new avenues for investigating the pathology of these disorders. with additional engineering, the design of aba-induced protein heterodimer formation can also be modified to study contact sites of other organelles. in the context of neurodegeneration, recent literature highlights the importance of exploring er-mitochondria, erendolysosome, er-plasma membrane, and er-lipid droplet interactions.14 more research will be performed to elucidate the applicability of the aba-induced system to these types of mcs. to study the function of mitochondrialysosome association in the context of pd, it would be helpful to apply this system in ipsc-derived dopaminergic neurons—a cell line more relevant to parkison's. to do this, crispr donor plasmids for each of the constructs are being designed. these plasmids will leverage the gene-editing technology of crispr to integrate the constructs into ipsc lines, including those of individuals with pd-causal mutations, allowing for the identification of which pdassociated genes are impacted by alterations to mitochondria-lysosome contacts. this integrated approach may shed light on cell type-specific molecular mechanisms of neurodegeneration and could lead to the development of new or targeted therapeutic strategies. author information corresponding author yuliia kohut – ykohu@illinois.edu author contributions yuliia kohut performed microscopy, immunoblotting, cloning experiments, and data analysis. she also wrote the manuscript and generated most of the figures. dr. robert coukos provided guidance and direction on conceiving and performing experiments during the project, including editing the manuscript. he also performed transfection of hela cells with lentivirus, cell culture work, and live-cell imaging. dr. dimitri krainc also provided direction and support during the project. abbreviations aba abscisic acid abi aba insensitive 1 bfp blue fluorescent protein crispr clustered regularly interspaced short palindromic repeats egfp extra green fluorescent protein em electron microscopy er endoplasmic reticulum fret fluorescence resonance energy transfer ipsc induced pluripotent stem cells irfp infra-red fluorescent protein mcs membrane contact sites no-fp no-fluorescent protein pcr polymerase chain reaction pla – proximity ligation assay pyl pyrabactin resistance-like sim structured illumination microscopy tmem192 transmembrane protein 192 tmd synj2bp transmembrane domain columbia undergraduate science journal vol. 19, 2025 kohut et. al © 2025 kohut, yuliia. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 18 references [1] kim s, coukos r, gao f, krainc d (2022) dysregulation of organelle membrane contact sites in neurological diseases. neuron 110(15):2386–2408. [2] scorrano l, et al. (2019) coming together to define membrane contact sites. nature communications 10(1). doi:10.1038/s41467-019-09253-3. [3] chen w, zhao h, li y (2023) mitochondrial dynamics in health and disease: mechanisms and potential targets. signal transduction and targeted therapy 8(1). doi:10.1038/s41392-023-01547-9. [4] li y, berliocchi l, li z, rasmussen lj (2023) interactions between mitochondrial dysfunction and other hallmarks of aging: paving a path toward interventions that promote healthy old age. aging cell 23(1). doi:10.1111/acel.13942. [5] trivedi pc, bartlett jj, pulinilkunnil t (2020) lysosomal biology and function: modern view of cellular debris bin. cells 9(5):1131. [6] bonam sr, wang f, muller s (2019) lysosomes as a therapeutic target. nature reviews drug discovery 18(12):923–948. [7] lamming dw, bar‐peled l (2018) lysosome: the metabolic signaling hub. traffic 20(1):27–38. [8] udayar v, chen y, sidransky e, jagasia r (2022) lysosomal dysfunction in neurodegeneration: emerging concepts and methods. trends in neurosciences 45(3):184– 199. [9] ysselstein d, shulman jm, krainc d (2019) emerging links between pediatric lysosomal storage diseases and adult parkinsonism. movement disorders 34(5):614–624. [10] wong yc, ysselstein d, krainc d (2018) mitochondria–lysosome contacts regulate mitochondrial fission via rab7 gtp hydrolysis. nature 554(7692):382–386. [11] [12] wong yc, ysselstein d, krainc d (2018) mitochondria–lysosome contacts regulate mitochondrial fission via rab7 gtp hydrolysis. nature 554(7692):382–386. [13] [14] 11. cantarero l, et al. (2020) mitochondria–lysosome membrane contacts are defective in gdap1related charcot–marie–tooth disease. human molecular genetics 29(22):3589–3605. [15] 12. coukos r, krainc d (2024) key genes and convergent pathogenic mechanisms in parkinson disease. nature reviews neuroscience 25(6):393–413. [16] 13. huang x, jiang c, yu l, yang a (2020) cur-rent and emerging approaches for studying inter-organelle membrane contact sites. frontiers in cell and developmental biology 8. doi:10.3389/fcell.2020.00195. [17] 14. gamuyao r, chang c-l (2024) imaging and proteomics toolkits for studying organelle con-tact sites. frontiers in cell and developmental biology 12. doi:10.3389/fcell.2024.1466915. [18] liang f-s, ho wq, crabtree gr (2011) engineering the aba plant stress pathway for regulation of induced proximity. science signaling 4(164). doi:10.1126/scisignal.2001449. introduction author information corresponding author yuliia kohut performed microscopy, immunoblotting, cloning experiments, and data analysis. she also wrote the manuscript and generated most of the figures. dr. robert coukos provided guidance and direction on conceiving and performing experiments duri... abbreviations aba abscisic acid abi aba insensitive 1 bfp blue fluorescent protein crispr clustered regularly interspaced short palindromic repeats egfp extra green fluorescent protein em electron microscopy er endoplasmic reticulum fret fluorescence resonance energy transfer ipsc induced pluripotent stem cells irfp infra-red fluorescent protein mcs membrane contact sites no-fp no-fluorescent protein pcr polymerase chain reaction pla – proximity ligation assay pyl pyrabactin resistance-like sim structured illumination microscopy tmem192 transmembrane protein 192 tmd synj2bp transmembrane domain references ©2024 manjila, nihal. this is an open-access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. recent progress in engineered oncolytic viruses: a thematic review nihal manjila1 1department of biology, case western reserve university, cleveland, oh, usa keywords: cancer therapy, oncolytic viruses, medicine, oncology abstract: this review examines recent research in engineered oncolytic viruses and applies a thematic analysis to identify trends in the field. the major themes identified are tumor microenvironment remodeling, enhancing virus efficiency, treating slow-growing tumors, tumor targeting specificity, and combination with chimeric antigen receptor-t cell therapy. the tumor microenvironment plays a significant role in cancer progression through factors like immunosuppression and hypoxia. engineered oncolytic viruses may be used to remodel aspects of the tumor microenvironment to favor and facilitate an immune response. another major consideration in the development of oncolytic viruses is their efficiency in inducing antitumor effects. viral vectors may be engineered with pro-apoptotic or pro-inflammatory signaling molecules to cause cancer cell death or mediate immune cell infiltration, respectively. slow-growing tumors present a challenge to oncolytic virotherapy since many viruses infiltrate tumors through infected daughter cells arising from cancer cell division. this challenge may be overcome by engineering viruses to maintain a high viral load in infected cancer cells and drawing a sustained immune response. a concern in the development of oncolytic viruses is the issue of viral tropism and infection specificity. although some viruses have limited tropism, oncolytic viruses may be engineered to specifically target tumors using cancer-specific receptor-ligand mechanisms. oncolytic viruses may also be used in conjunction with chimeric antigen receptor-t cell therapy to reduce immunosuppression and enhance chimeric antigen receptor-t cell infiltration of tumors. this poses a promising approach in oncology research and treatment. introduction cancer is a class of diseases characterized by the uncontrolled proliferation of cells in the body. affecting approximately 20 million people globally, cancer has one of the highest mortality rates with 10 million deaths per year, posing as a significant issue in the medical field. cancer incidence is exacerbated by the lack of non-invasive and side effect-free treatment options. while treatments like chemotherapy may be effective at combating cancer, the side effects take a toll on patients.1 there are various types of cancer, each with its own genetic causes and clinical manifestations, which makes developing treatments more difficult, as an approach that is effective against one type of cancer may not be effective against another type. for this reason, more therapeutics for cancer are vital. engineered oncolytic viruses (ovs) are viral vectors designed to damage and destroy cancer cells. ovs can be developed using a number of different theoretical approaches and experimental modifications. once ovs have been engineered and undergo basic testing, they can advance to animal models and clinical trials. in the past five years, there has been an increasing interest in ovs due to the treatment method’s novelty and comparative safety, which has led researchers to perform studies taking different approaches to advancing the knowledge and repertoire of effective ovs for clinical application.2 it is useful to review recent advances in the development of engineered ovs to determine contemporary research trends. this analysis can be used columbia undergraduate science journal vol. 18, 2024 manjila to make predictions about possible directions that ov research may take in the future. in this review, a total of 42 primary research articles were analyzed. the majority of analyzed articles involve the engineering and testing of ovs, but clinical trials and theory papers are also included. among the virus engineering articles, herpes simplex virus (hsv) was the most represented (table 1). this is due to the advantages that hsv possesses as a vector: it does not integrate into the host genome since it is a double-stranded dna virus and possesses a large genome that can serve as a stable base for genetic modifications. adenovirus and vaccinia virus were also well represented, with approximately half the number of articles as herpes simplex virus. several other viruses were also utilized in the analyzed studies, including poliovirus and bacteriophage t7. viral vector number of studies herpes simplex virus 15 adenovirus 9 vaccinia virus 7 vesicular stomatitis virus 3 newcastle disease virus 2 reovirus 1 measles virus 1 zika virus 1 maraba virus 1 poliovirus 1 myxovirus 1 bacteriophage t7 1 *a number of the viruses counted were one of multiple viruses used in a single study. table 1. viral vectors represented in analyzed studies five major themes or approaches were identified in the included studies. first, remodeling of the tumor microenvironment led to improvement in immune cell infiltration. this approach allowed the immune system to target and destroy cancer cells with its activated response. second, additional modification of engineered ovs further improved their antitumor effect. this strategy utilized the variety of ovs already designed by researchers and advanced their work. third, specialized ovs showed antitumor effects in the treatment of slow-growing tumors. viral replication or antitumor effects that relied on cancer cell division was ineffective in slowgrowing tumors. fourth, targeting of cancer cells by viruses or immune cells were improved using signaling molecules. increasing infection specificity improved the effectiveness of ovs, as well as reducing the severity of potential side effects. fifth, oncolytic virotherapy in conjunction with chimeric antigen receptor-t (car-t) cell therapy shows improved antitumor effectiveness. the theme of combinatorial therapy can be applied to other treatments as well, with the hope that the anti-tumor effect will be greater than either treatment alone. tumor microenvironment remodeling: when cancer cells accumulate and form tumors, they create a local environment that decreases the effectiveness of the immune system in infiltrating the tumor and destroying the cancer cells.3 one of the primary effects of this tumor microenvironment is the prevention of natural killer (nk) cells from infiltrating the tumor mass. nk cells are one of the immune system’s major cell types, and they can target and destroy cancer cells. in a healthy individual, nk cells will identify and eliminate any abnormally dividing cells. in an individual with cancer, nk cells are unable to overcome the cancer cells. other components of the immune activity may also become inhibited, allowing cancer cells to proliferate and cause metastasis. therefore, remodeling and modifying the tumor microenvironment to facilitate targeting of tumor cells by nk cells could be a design goal for engineered ovs. currently, there is a well-known engineered ov on the market known as talimogene laherparepvec (tvec), a clinical treatment that utilizes an engineered herpes simplex virus to combat cancer. it was thought that t-vec columbia undergraduate science journal vol. 18, 2024 manjila functioned through direct oncolysis, or the virus infecting cancer cells and causing them to die. ramelyte and colleagues showed that t-vec does not directly cause oncolysis after conducting a phase i clinical trial of cutaneous b cell lymphoma patients. instead, research found that the treatment alters the tumor microenvironment and allows the patient’s immune cells to infiltrate and attack the tumor. the researchers examined tumor appearance, sequenced cell genetics, and identified immune cells present in biopsy samples. they confirmed that the treatment reduced tumor scope, both in the tumor at the injection site and in distant tumors. they also found that multiple types of immune cells began to infiltrate the primary tumor as the treatment progressed. lastly, they found that the viral genetic script was present in both cancer cells and healthy cells, indicating a nonspecific infection mechanism.4 these findings provide key insights into the significance of the tumor microenvironment and viral design in engineering effective ovs. ramelyte and colleagues demonstrated that it was not necessary for an ov to directly destroy cancer cells. instead, the ov could modify the tumor microenvironment and allow the immune response to take effect. this provides a mechanism for engineering ovs that would not infect healthy cells to a significant degree. one of the major strengths of ovs and modification of the tumor microenvironment is the specificity and efficacy that can be achieved relative to more traditional treatment approaches. additionally, the tumor microenvironment contains many cell types and active pathways that can be investigated and targeted by future studies, potentially showing anti-tumor effectiveness that further improves on preexisting ovs or targets other cancer types. the microenvironment remodeling approach has been utilized by a number of research groups addressing various pathways and cell types present in the tumor microenvironment. cervera-carrascon and colleagues developed an oncolytic adenovirus that causes infected tumor cells to release two key cytokines into the tumor microenvironment.5 the cytokines signal local cytotoxic t cells to approach and target the tumor. the antitumor effectiveness of the adenovirus is enhanced by the addition of immune checkpoint inhibitor treatment that activates the immune response. lin and colleagues engineered an oncolytic herpes virus that releases damage-associated signals to facilitate antigen presentation by dendritic cells and incite an immune response.6 rather than counteracting immunosuppressive signals, this approach utilized antigen presentation to incite infiltration of the tumor by immune cells. gentile and colleagues created an oncolytic herpes virus that causes infected tumor cells to express adenosine deaminase, an enzyme that breaks down extracellular adenosine—an immunosuppressive signaling molecule.7 the breakdown of adenosine in the tumor microenvironment allows local immune cells to become activated and attack the tumor cells. keshavarz and colleagues developed an oncolytic newcastle disease virus transported by mesenchymal stem cells to deliver apoptosis signaling molecules to tumors.8 upon binding to the appropriate receptors in the tumor cells, the apoptosis signaling molecules will cause the cells to undergo cell death. each of these studies used biochemical pathways as a mechanism of action for ovs to modify the signaling molecule repertoire present in the tumor microenvironment. other researchers have built on the mechanism behind tvec’s function, as in the case of the oncolytic herpes virus engineered by haines and colleagues to not infect healthy cells but still trigger immune recognition of the tumor.9 this is a significant advance in ov engineering as it improves on a virus in clinical use by increasing infection specificity. the tumor microenvironment also includes several cell types that are normally found in a healthy body but may be present within a tumor and closely associated with cancer cells. kurisu and colleagues used an oncolytic reovirus to target and destroy columbia undergraduate science journal vol. 18, 2024 manjila tumor-associated fibroblast cells, which contribute to the immunosuppressive environment created by tumors.10 the researchers also found that the ov did not harm healthy fibroblast cells, indicating a level of infection specificity. infection specificity is important in this case since the ov directly causes fibroblasts, a cell type that also naturally exists in the body, to lyse and die. however, it poses potential risks and side effects since an ov with imprecise action could destroy healthy cell populations. another interesting method of approaching tumor microenvironment remodeling is to utilize a viral vector that would not normally infect human cells. bacteriophages are viruses that infect bacterial cells instead of animal or plant cells. like other classes of viruses, bacteriophages can also be genetically and structurally engineered. the challenge of using a bacteriophage as the vector for ov engineering is that it does not naturally infect cancer cells. instead, they must be engineered to target and infect cancer cells through signaling or receptor interactions. hwang and myung engineered bacteriophage t7 to express a peptide that contributes to tumor growth inhibition and immune cell infiltration of tumors.11 this result indicates that bacteriophages are viable bases for ov engineering and that additional studies could explore this class of viruses. remodeling of the tumor microenvironment holds massive potential for engineered ov research. t-vec has already been implemented as a clinical treatment and has shown success in the clinical setting. microenvironment remodeling also allows immune cells to infiltrate tumors and destroy cancer cells. this approach is generally safe since the direct cause of cancer cell death is the immune response. additionally, there are a few signaling pathways that are upregulated in cancer cells but not in healthy cells. upregulating or downregulating these pathways appropriately could disrupt the tumor microenvironment and reduce immune exclusion. enhancing virus efficiency engineered ovs are designed using one or more genetic and structural modifications to cause an antitumor effect. ovs with one or more such improvements can be further improved by making additional modifications that enhance efficacy without compromising safety. these modifications can take the form of increased levels of surface proteins or signaling molecules, as well as directly causing the death of cancer cells.12 enhancing modifications could also address different barriers to antitumor effectiveness, such as signaling from tumor cells or the immune system containing and destroying ov particles before they can take effect and destroy cancer cells. some cancer types, such as acute myeloid leukemia, show limited susceptibility to ovs and require more effective viral vectors and active agents. one molecule utilized by engineered ovs to cause an antitumor effect is tumor necrosis factor-related apoptosis-inducing ligand (trail). this immune factor can be expressed on the surface of ovs and binds to trail receptors on the surface of cancer cells to initiate a signaling pathway that causes cell death. wang and colleagues aimed to enhance an engineered ov with additional trail to cause a greater antitumor effect.13 the researchers had previously engineered an ov to express trail on its protein shell that showed antitumor activity. they recognized that additional trail could be bound to their ov through a structural attachment and experimentally achieved this through a zipper domain. this addition of trail improved the antitumor effectiveness of the ov against acute myeloid leukemia when compared to the first iteration of their trail-enhanced ov. certain types of cancer, including acute myeloid leukemia, may be resistant to the effects of trail signaling. the mechanism behind this resistance is a reduction of trail receptors in the cancer cells, leading to fewer opportunities for trail on the ov to bind and initiate the signaling pathway. due to the nature of this resistance columbia undergraduate science journal vol. 18, 2024 manjila mechanism, increasing the availability of trail is not sufficient to induce cell death in cancer cells. ginsenosides, compounds derived from ginseng, were previously shown to increase the availability of trail receptors in cancer cells.13 wang and colleagues investigated the role of the ginsenoside rh (steroid glycosides) in amplifying the antitumor effect of their trail-enhanced ov. they found that rh activity led to greater expression of trail receptors on the surface of cancer cells. in conjunction, the inclusion of rh in the treatment course increased cancer cell death as trail had more receptors to bind to. this demonstrates that ov enhancement can be further improved using synergistic treatment with other compounds, including natural compounds as in the case of ginsenosides. other research groups have taken the approach of enhancing viral efficiency by focusing efforts on different structural and genetic targets. menotti and colleagues modified an engineered herpes simplex virus to target cancer cell-specific receptors by inserting different single chain antibodies into the viral protein structure.14 this modification increased the number of cancer cells that can be targeted by the ov since different antibodies could be inserted, making that virus more useful across more cancer types. kim and colleagues increased viral expression of interleukin-12 and granulocyte-macrophage colony-stimulating factor to amplify immune response.15 more cancer cells will be destroyed if more immune cells are activated and localize to the tumor site. rivadeneira and colleagues engineered an oncolytic vaccinia virus to express the adipokine leptin in infected tumor cells.16 furthermore, leptin activates a signaling pathway that contributes to an increased t cell response and prolonged immune memory. these studies demonstrate the variety of methods that are available for enhancing ovs. a major barrier to achieving greater viral efficiency is the containment and destruction of ovs by the human immune system. if the immune response is activated by the presence of viral particles in the body, the ov will not have any significant antitumor effect since the virus will not reach the tumor in sufficient quantity. atasheva and colleagues addressed this issue through the development of an oncolytic adenovirus that avoids immune response and containment by mutating the region of the viral protein that host antibodies bind to.17 the resulting ov was able to cause a systemic antitumor effect without being hindered by the immune response since antibodies could no longer bind to the virus. this study highlights the need for ovs engineered to avoid triggering an immune response from the host organism, particularly as ovs become increasingly effective at destroying tumors. in a similar manner, bah and colleagues engineered an oncolytic measles virus to better avoid targeting by the immune system by replacing key surface glycoproteins with those from another virus.18 this is another method by which ovs can be engineered to demonstrate stealth against the immune system. while further laboratory studies are being performed, a few ovs have also moved into the clinical phase. zhang and colleagues performed phase i and ii clinical trials on the antitumor effectiveness of an oncolytic herpes simplex virus designed to express granulocyte-macrophage colony-stimulating factor, which enhances immune response.19 the researchers found that the ov treatment was effective in metastatic, treatment-resistant esophageal and rectal cancer. the addition of an antibody for a tumor cell receptor that suppresses immune activity enhanced the oncolytic effectiveness of the virus. side effects observed in the patients were relatively mild, and improved immune infiltration of the tumor was demonstrated. these results are significant because they show that ovs that are not effective against certain types of cancer may be enhanced to become effective. fares and colleagues found that an oncolytic adenovirus showed antitumor effect in a phase i clinical trial of glioblastoma patients.20 glioblastoma is a very aggressive cancer type, and results indicating antitumor activity could be indicative of the potential for clinical columbia undergraduate science journal vol. 18, 2024 manjila application after additional research is performed. enhancing preexisting ovs provides more effective therapeutics. due to the recent interest in oncolytic virotherapy, many different ovs have been engineered with a few entering clinical trials.12 these ovs that have already shown antitumor effectiveness could be made more effective through different engineering approaches and antitumor mechanisms. therefore, future research is not limited to the development of entirely novel ovs. it may also potentially be more efficient to push enhanced ovs into clinical trials if the parent strain was already in clinical trials. treating slow-growing tumors different cancer types have different pathophysiology and rates of metastasis. tumors grow at different rates depending on the type of cancer, anti-cancer immune response, and other host-related factors. the growth of tumors are directly the result of cancer cell proliferation and accumulation. engineered ovs are able to target tumors that grow quickly by spreading through tumors via daughter cells produced by cancer cell division.21 slower-growing tumors undergo cell division at a slower rate, and therefore ovs that primarily proliferate in a tumor through cell division are not as effective at combating the cancer cells. additionally, given that slow-growing tumors usually take an extended period of time before becoming significantly dangerous to patients, there is more time to administer treatments, and treatments that could be very effective in the long-term may have more utility than those of the short term. slow-growing tumors may also exhibit treatment resistance, which necessitates ov engineering that overcomes resistance mechanisms. ovs require additional engineering and investigation to move towards the development of novel ovs that utilize mechanisms of action that are effective against slow-growing cancers. fukuhara and colleagues sought to develop an ov that could be an effective anti-cancer agent in slow-growing tumors. they engineered an oncolytic herpes simplex virus known as thtert which retains the infection specificity of its parent strain while preventing protein synthesis shutdown in infected cancer cells, leading to virus proliferation that does not damage healthy cells. the researchers found that the ov was effective against slow-growing tumors by maintaining a high viral load despite lower cell division rates.22 additionally, they demonstrated that the virus was safe for application in brain tissue, which is quite susceptible to impairment due to tissue damage. these findings indicate that ovs utilizing similar mechanisms may be suitable for clinical use in the treatment of slow-growing tumors. ov effectiveness is shown to not be limited to fast-growing tumors and can be effective against slowgrowing tumors given the necessary modifications. the researchers also showed that the ov is safe for healthy tissue and may not pose a significant risk to the host organism. other researchers have investigated the possibility of ovs that function effectively in slow-growing tumors or have the potential to be useful towards the development of such ovs. cao and colleagues engineered an oncolytic vaccinia virus that draws macrophages towards infected tumor cells.23 the involvement of another immune cell type increases the number of possible ov mechanisms that could be utilized by researchers. liu and colleagues engineered an oncolytic herpes simplex virus to express cytosine deaminase for the treatment of uveal melanoma.24 this ov was designed to treat a rare and chemotherapy-resistant cancer type, demonstrating the need for ovs for uncommon cancers. khalique and colleagues created an oncolytic herpes virus that targets immunosuppressive tumors while simultaneously drawing an immune response to cancer cells.25 multi-function ovs could be effective against slow-growing tumors, particularly if the tumors grow at higher or lower rates over a period of time. leoni and colleagues developed a noncancer cell line for the development of ovs.26 this cell line could be a useful tool in the engineering of more ovs, particularly ovs designed for slow-growing cancers. columbia undergraduate science journal vol. 18, 2024 manjila some research groups have worked towards the development of anti-cancer vaccines which enhance long-term, antitumor immune activity through the activity of viral vectors. roy and colleagues furthered the approach of engineering ovs that do not rely on division to be effective. the researchers developed an anti-cancer vaccine utilizing a number of viral vectors with antigenic peptides. the separation of these peptide components allowed for anticancer immunity enhancement without requiring cell division.27 tian and colleagues also developed an anti-cancer vaccine using a herpes simplex virus to carry a monoclonal antibody.28 this anti-cancer vaccine also downregulates cell adhesion, reducing the ability of a tumor to exclude immune cells. priming the immune system to combat cancer cells has significant potential in reducing cancer incidence at an earlier stage of disease development. additionally, increasing the natural immune response to cancer in individuals with a predisposition to cancer development could improve patient outcomes. improving immune memory through increasing memory cell counts or extending memory cell survival also contributes to viral effectiveness against slow-growing tumors. as a tumor grows, the immune memory of that cancer type could trigger a response and draw immune cells to the tumor. thomas and colleagues developed an oncolytic herpes simplex virus that improves immune memory after responding to tumor cells.29 the researchers achieved this by identifying a virus that demonstrated antitumor activity and expressed a truncated envelope glycoprotein in the virus. they found that the antitumor activity of this ov leads to the activation of signaling molecules that cause immune memory to form via memory cells. slow-growing tumors have gained interest as their low division rate causes ovs that rely on cell division to not be as effective. while this is a significant engineering challenge, recent studies have shown that there are still methods of overcoming or circumventing the low division rate. ovs that create high viral loads without quickly lysing cancer cells can gain the most benefit from each cell division in a slow-growing tumor. progress has also been made on the development of research tools and earlystage mechanisms for treating slow-growing tumors. anti-cancer vaccines have the potential to develop into clinical treatments. enhancing immune memory via ov action also holds potential for treating slowgrowing tumors through a natural, long-term immune amplification. tumor targeting specificity as demonstrated by t-vec, there is a need to improve the targeting specificity of ovs to infect tumor cells primarily or almost exclusively. this could effectively reduce any possible systemic effect of ov therapy and improve patient safety. one approach that has shown significant potential in facilitating improvements in infection specificity is the utilization of biochemical signaling pathway components as markers for ovs. biochemical signaling pathways play a significant role in the regulation of tissue maintenance and cell division. any imbalance or dysregulation in these signaling pathways can lead to the development of cancer. ligand-receptor binding interactions also provide an opportunity for improving tumor targeting specificity by taking advantage of a signaling pathway that is dysregulated to facilitate localization by ovs.30 there are many pathways that have the potential to be utilized as the basis of a tumor-targeting system. one such ligand-receptor interaction that has been studied extensively is that between c-c chemokine ligand 5 (ccl5) and c-c chemokine receptor type 5 (ccr5). in many types of cancer, the ccl5/ccr5 ligand-receptor interaction functions improperly and contributes to tumor growth and metastasis.31 li and colleagues developed a pair of ovs that upregulate ccl5 in infected tumor cells and upregulate ccr5 expression in nk cells, respectively. each of these components individually increased antitumor effectiveness, and the combined treatment system was more effective than either columbia undergraduate science journal vol. 18, 2024 manjila component alone. the result of this bilateral oncolytic virotherapy was the establishment of a tumor-targeting system which led to an improved antitumor effect. nk cells were also more effectively able to target and destroy the corresponding cancer cells.32 additionally, the ov engineered to cause ccl5 expression did not infect nk cells in significant numbers. this is an important finding since nk cells with ccr5 expression would be drawn to nk cells with ccl5 expression, rather than the tumor cells. researchers have explored additional avenues to improve target specificity. vijayakumar and colleagues engineered an oncolytic newcastle disease virus to deliver immunotherapeutic molecules to tumors, eliciting a targeted antitumor response.33 zuo and colleagues developed an oncolytic vaccinia virus that encodes for a singlechain variable fragment that targets immune cells.34 wang and colleagues created an oncolytic vaccinia virus that inhibits pd-1, an immunosuppressive signaling molecule, and elicits a cancer-specific immune response.35 wu and colleagues engineered an oncolytic vesicular stomatitis virus that also inhibits pd-1, causing an antitumor effect.36 walton and colleagues developed an oncolytic poliovirus that proliferates rapidly, evades immune response, and is cytotoxic to infected cancer cells to treat glioblastoma, an aggressive brain cancer.37 these studies utilize different structural interactions and signaling pathways to effectively target tumors and combat cancer cells. triple-negative breast cancer is very aggressive and difficult to treat. ovs engineered to take advantage of signaling pathways specific to triple-negative breast cancer can demonstrate infection specificity while also being more effective. rodriguez stewart and colleagues developed an oncolytic reovirus that activated the caspase-mediated apoptosis signaling pathway in a novel manner.38 the caspase pathway involves a series of caspase signals that end with caspase-3, triggering apoptosis of the cell. the researchers found that the caspase pathway was activated but caspase-3 was not, indicating that some other signaling mechanism was responsible for apoptosis of tumor cells.38 this difference in signaling pathway activity may explain why the tumor cells were affected with specificity while also causing apoptosis. additionally, this study indicates that variations of previously understood signaling pathways may potentially facilitate the discovery or design of other ovs with specific tumor targeting and antitumor effectiveness. other types of cancer, such as pancreatic ductal adenocarcinoma (pdac), have also demonstrated resistance to treatment. seegers and colleagues engineered an oncolytic vesicular stomatitis virus that effectively infects treatment-resistant pdac cells due to enhanced viral attachment. the researchers used directed viral evolution to create mutations in the viral genes and identified two mutations that appeared in more than one evolved virus. ovs with these mutations were advanced to testing and demonstrated increased replication in treatment-resistant pdac cells.39 this finding indicates that directed evolution could be used to identify other beneficial mutations in ovs. additionally, other treatment-resistant cancers could potentially be treated more effectively by ovs which are designed to utilize targeting mechanisms that are not involved with the cancer’s resistance to treatment. an ov that relies on a different targeting mechanism could identify and infect cancer cells more effectively than an ov that relies on a mechanism which is made ineffective by a mutation in the cancer cells. significant improvements have been made in tumor targeting specificity by ovs and immune cells. this approach includes a number of different specific mechanisms that have been utilized to improve oncolytic virotherapy. ovs can infect cancer cells and cause the expression of different signaling molecules and other proteins on their surfaces that aid immune cells in identifying them. ovs can also be engineered to improve their infection specificity and more selectively infect cancer cells. engineering of ovs can also be specialized to different cancer types in order to take advantage of columbia undergraduate science journal vol. 18, 2024 manjila certain receptor-ligand interactions that may be upregulated in different cancers. this could lead to the development of an ov repertoire that demonstrates a significant degree of precision in its mechanism and cytotoxicity. oncolytic viruses and car-t cell therapy certain types of cancer, particularly blood cancers, can be treated using chimeric antigen receptor t (car-t) cell therapy. this treatment utilizes immune cells taken from the patient and engineered to carry receptors that better allow them to target cancer cells. this method is beneficial because it relies on the patient’s own cells and reduces the likelihood of severe side effects. the immunosuppressive nature of the tumor microenvironment decreases the efficiency of immune cell-based therapies in the same manner as it decreases the efficiency of natural immune responses. this means that car-t cell therapy could potentially be ineffective against tumor growths that express high levels of immunosuppression.40 combining car-t cell therapy with another form of treatment could potentially overcome the challenge of immunosuppression and improve the antitumor effectiveness of the therapeutic approach. oncolytic virotherapy can be utilized to counteract immunosuppression in the tumor microenvironment. after ovs have been implemented to cause the tumor region to become immunologically active, car-t cells can more easily infiltrate the tumor and cause cancer cell death. mckenna and colleagues developed a multi-component oncolytic therapeutic delivered by mesenchymal stromal cells combined with car-t cell therapy to achieve a greater antitumor effect. the first stage of this therapeutic strategy is the delivery of oncolytic viruses by the mesenchymal stromal cells. after the viruses have been delivered, they infect the tumor and reduce immunosuppression in the microenvironment. the final stage involves infiltration of the tumor by car-t cells and the death of cancer cells. this combined treatment was more effective than the engineered ovs alone.41 the researchers also found that t cells were identified in larger numbers within the tumor mass. this result indicates that the immune response is amplified in addition to the car-t cell activity. the concept of utilizing chimeric receptors and combination therapy for anti-cancer activity has been investigated by researchers developing ovs as well. froechlich and colleagues engineered an ov targeted to mesothelin, a molecule expressed by cancer cells.42 they achieved this by creating a chimeric receptor that included an immune component and the binding site for the cancer ligand target. huang and colleagues developed an oncolytic adenovirus that contains a gene circuit for the control of viral replication after infecting cancer cells.43 simulation of the population dynamics of the ov revealed that controlling immune signaling at the tumor site would be more effective than combination therapy using an ov with immune signaling. this finding is interesting but was not experimentally determined to a sufficient degree. combination therapy using engineered ovs and car-t cell therapy can also be utilized to overcome the challenge of homogenous expression of car targets. tumor cells and healthy cells both express certain car targets, which causes car-t cell activity to be nonspecific to tumors. park and colleagues developed an ov that causes infected tumor cells to express truncated cd19 on their surface before lysing due to viral activity. car-t cells engineered with cd19 were drawn to infected cancer cells due to the binding mechanism of cd19. native t cells were also signaled and drawn towards the tumor. these findings indicate that car-t cell activity can be made specific through the use of ovs to express chimeric antigen receptor targets in tumors.44 the researchers concluded that they had developed a method of improving the tumor targeting of car-t cells using ovs and that this interaction had potential therapeutic applications in the future. combination therapy utilizing oncolytic columbia undergraduate science journal vol. 18, 2024 manjila virotherapy and other cancer therapeutic strategies, such as car-t cell therapy, holds significant potential for improving patient outcomes. multiple therapeutic strategies that can function effectively could have additive antitumor effects and prolong life or destroy tumors. additionally, therapeutics that do not cause severe side effects could be implemented in the clinical setting without causing great discomfort or additional risks for patients. currently, oncolytic viruses are entering clinical use and car-t cell therapy has been in use for treating patients suffering from blood cancers. it is reasonable that two clinically approved treatment strategies could be utilized as a combined treatment in the future. conclusion significant advances have been made in research on engineered oncolytic viruses in the last five years. several studies have led to the development of novel ovs which have demonstrated antitumor effectiveness, including in vivo. researchers have also taken many different approaches to developing engineered oncolytic viruses. tumor microenvironment remodeling has been investigated as a method of improving immune cell infiltration of tumors. genetic and structural modifications have been utilized to enhance the antitumor effectiveness of ovs. specialized ovs have entered development for the treatment of slow-growing tumors and rare cancers. the targeting and infection of tumor cells has been improved through signaling and receptor upregulation. oncolytic virotherapy is effective in combined treatment with car-t cell therapy. these approaches address a variety of issues and challenges in the development and refinement of ovs for potential clinical treatment. outside the direct development of ovs with genetic and structural modifications, researchers have approached the question of engineered ov implementation in a number of ways. engineering experts have applied their expertise to the delivery of ovs, as in the application of control theory for the regulation of ov dosage.45 theoretical biologists have also approached the question of implementation through mathematical models, such as a voronoi cell-based model for viral spread in tumors.46 these types of studies contribute additional information towards the larger effort to optimize and implement ovs in the clinical setting. some researchers have also published the specific methods they used to engineer ovs and quantify them, allowing others to pursue their own virus engineering projects.47,48,49 clinical trials in veterinary medicine are also being conducted, with the same types of cancers being treated in animals as in humans.50 this variety of research approaches is additional evidence for the interest engineered ovs are gaining. in conclusion, ovs demonstrate significant potential for treating cancers with efficacy and safety. ovs can be engineered to deliver signaling molecules to cancer cells, facilitate immune cell infiltration of tumors, and upregulate immune response to tumor growth. there are many viruses and viral modifications that are suitable candidates for engineering into ovs. cancer treatment strategies such as chemotherapy often cause systemic symptoms and side effects that are very unpleasant for patients. these symptoms may be mitigated by ovs since the viruses can be engineered to selectively target tumors and utilize safe molecular components. as more ovs are engineered and move into clinical trials, there will likely be a greater number of ov therapeutics available for medical use in the next decade. the future of oncolytic virus research appears very promising as it provides hope for the development of meaningful cancer therapeutics. author information corresponding author nihal manjila nihal_manjila@hms.harvard.edu acknowledgements the author expresses his gratitude to professor phoebe stewart (department of pharmacology, case western reserve university school of medicine) for guidance and feedback in the preparation of the article. columbia undergraduate science journal vol. 18, 2024 manjila abbreviations ov: oncolytic virotherapy hsv: herpes simplex virus car-t: chimeric antigen receptor t nk: natural killer t-vec: talimogene laherparepvec trail: tumor necrosis factor-related apoptosis-inducing ligand htert: human telomerase reverse transcriptase ccl5: chemokine ligand 5 ccr5: chemokine receptor type 5 pd1: programmed cell death protein 1 pdac: pancreatic ductal adenocarcinoma cd19: b-lymphocyte antigen cd19 references [1] o. hemminki, j. m. dos santos, a. hemminki, oncolytic viruses for cancer immunotherapy. j hematol oncol 13, 84 (2020). [2] p. k. bommareddy, m. shettigar, h. l. kaufman, integrating oncolytic viruses in combination cancer immunotherapy. nat rev immunol 18, 498-513 (2018). [3] e. ylosmaki, v. cerullo, design and application of oncolytic viruses for cancer immunotherapy. curr opin biotechnol 65, 25-36 (2020). [4] e. ramelyte et al., oncolytic virotherapymediated anti-tumor response: a singlecell perspective. cancer cell 39, 394406 e394 (2021). [5] v. cervera-carrascon et al., tumor microenvironment remodeling by an engineered oncolytic adenovirus results in improved outcome from pd-l1 inhibition. oncoimmunology 9, 1761229 (2020). [6] c. lin et al., intratumoral delivery of a pd-1-blocking scfv encoded in oncolytic hsv-1 promotes antitumor immunity and synergizes with tigit blockade. cancer immunol res 8, 632647 (2020). [7] c. gentile et al., generation of a retargeted oncolytic herpes virus encoding adenosine deaminase for tumor adenosine clearance. int j mol sci 22, (2021). [8] m. keshavarz et al., oncolytic newcastle disease virus delivered by mesenchymal stem cells-engineered system enhances the therapeutic effects altering tumor microenvironment. virol j 17, 64 (2020). [9] b. b. haines et al., oncr-177, an oncolytic hsv-1 designed to potently activate systemic antitumor immunity. cancer immunol res 9, 291-308 (2021). [10] n. kurisu et al., oncolytic reovirusmediated killing of mouse cancerassociated fibroblasts. int j pharm 610, 121269 (2021). [11] y. j. hwang, h. myung, engineered bacteriophage t7 as a potent anticancer agent in vivo. front microbiol 11, 491001 (2020). [12] s. chaurasiya, y. fong, s. g. warner, oncolytic virotherapy for cancer: clinical experience. biomedicines 9, (2021). [13] z. wang et al., enhancing the antitumor activity of an engineered trail-coated oncolytic adenovirus for treating acute myeloid leukemia. signal transduct target ther 5, 40 (2020). [14] l. menotti et al., hsv as a platform for the generation of retargeted, armed, and reporter-expressing oncolytic viruses. viruses 10, (2018). [15] k. j. kim et al., antitumor effects of il12 and gm-csf co-expressed in an engineered oncolytic hsv-1. gene ther 28, 186-198 (2021). [16] d. b. rivadeneira et al., oncolytic viruses engineered to enforce leptin expression reprogram tumor-infiltrating t cell metabolism and promote tumor clearance. immunity 51, 548-560 e544 (2019). [17] s. atasheva et al., systemic cancer therapy with engineered adenovirus that evades innate immunity. sci transl med 12, (2020). columbia undergraduate science journal vol. 18, 2024 manjila [18] e. s. bah, r. a. nace, k. w. peng, m. a. munoz-alia, s. j. russell, retargeted and stealth-modified oncolytic measles viruses for systemic cancer therapy in measles immune patients. mol cancer ther 19, 2057-2067 (2020). [19] b. zhang et al., intratumoral oh2, an oncolytic herpes simplex virus 2, in patients with advanced solid tumors: a multicenter, phase i/ii clinical trial. j immunother cancer 9, (2021). [20] j. fares et al., neural stem cell delivery of an oncolytic adenovirus in newly diagnosed malignant glioma: a first-in-human, phase 1, dose-escalation trial. lancet oncol 22, 1103-1114 (2021). [21] m. mondal, j. guo, p. he, d. zhou, recent advances of oncolytic virus in cancer therapy. hum vaccin immunother 16, 2389-2402 (2020). [22] h. fukuhara, y. takeshima, t. todo, triple-mutated oncolytic herpes virus for treating both fastand slow-growing tumors. cancer sci 112, 3293-3301 (2021). [23] f. cao et al., engineering oncolytic vaccinia virus to redirect macrophages to tumor cells. adv cell gene ther 4, (2021). [24] s. liu et al., antitumor efficacy of oncolytic hsv-1 expressing cytosine deaminase is synergistically enhanced by dpd down-regulation and emt inhibition in uveal melanoma xenograft. cancer lett 495, 123-134 (2020). [25] h. khalique et al., oncolytic herpesvirus expressing pd-l1 bite for cancer therapy: exploiting tumor immune suppression as an opportunity for targeted immunotherapy. j immunother cancer 9, (2021). [26] v. leoni et al., a strategy for cultivation of retargeted oncolytic herpes simplex viruses in non-cancer cells. j virol 91, (2017). [27] d. g. roy et al., adjuvant oncolytic virotherapy for personalized anti-cancer vaccination. nat commun 12, 2626 (2021). [28] c. tian et al., enhanced anti-tumor response elicited by a novel oncolytic hsv-1 engineered with an anti-pd-1 antibody. cancer lett 518, 49-58 (2021). [29] s. thomas et al., development of a new fusion-enhanced oncolytic immunotherapy platform based on herpes simplex virus type 1. j immunother cancer 7, 214 (2019). [30] m. m. rahman, g. mcfadden, oncolytic viruses: newest frontier for cancer immunotherapy. cancers (basel) 13, (2021). [31] d. aldinucci, c. borghese, n. casagrande, the ccl5/ccr5 axis in cancer progression. cancers (basel) 12, (2020). [32] f. li et al., ccl5-armed oncolytic virus augments ccr5-engineered nk cell infiltration and antitumor efficiency. j immunother cancer 8, (2020). [33] g. vijayakumar, s. mccroskery, p. palese, engineering newcastle disease virus as an oncolytic vector for intratumoral delivery of immune checkpoint inhibitors and immunocytokines. j virol 94, (2020). [34] s. zuo et al., an engineered oncolytic vaccinia virus encoding a single-chain variable fragment against tigit induces effective antitumor immunity and synergizes with pd-1 or lag-3 blockade. j immunother cancer 9, (2021). [35] g. wang et al., an engineered oncolytic virus expressing pd-l1 inhibitors activates tumor neoantigenspecific t cell responses. nat commun 11, 1395 (2020). [36] c. wu, m. wu, m. liang, s. xiong, c. dong, a novel oncolytic virus engineered with pd-l1 scfv effectively inhibits columbia undergraduate science journal vol. 18, 2024 manjila tumor growth in a mouse model. cell mol immunol 16, 780-782 (2019). [37] r. w. walton, m. c. brown, m. t. sacco, m. gromeier, engineered oncolytic poliovirus pvsripo subverts mda5-dependent innate immune responses in cancer cells. j virol 92, (2018). [38] r. m. rodriguez stewart, v. raghuram, j. t. l. berry, g. n. joshi, b. a. mainou, noncanonical cell death induction by reassortant reovirus. j virol 94, (2020). [39] s. l. seegers, c. frasier, s. greene, i. v. nesmelova, v. z. grdzelishvili, experimental evolution generates novel oncolytic vesicular stomatitis viruses with improved replication in virusresistant pancreatic cancer cells. j virol 94, (2020). [40] r. c. sterner, r. m. sterner, car-t cell therapy: current limitations and potential strategies. blood cancer j 11, 69 (2021). [41] m. k. mckenna et al., mesenchymal stromal cell delivery of oncolytic immunotherapy improves car-t cell antitumor activity. mol ther 29, 18081820 (2021). [42] g. froechlich et al., generation of a novel mesothelin-targeted oncolytic herpes virus and implemented strategies for manufacturing. int j mol sci 22, (2021). [43] h. huang et al., oncolytic adenovirus programmed by synthetic gene circuit for cancer immunotherapy. nat commun 10, 4801 (2019). [44] a. k. park et al., effective combination immunotherapy using oncolytic viruses to deliver car targets to solid tumors. sci transl med 12, (2020). [45] a. j. n. anelone, m. f. villa-tamayo, p. s. rivadeneira, oncolytic virus therapy benefits from control theory. r soc open sci 7, 200473 (2020). [46] a. l. jenner, f. frascoli, a. c. f. coster, p. s. kim, enhancing oncolytic virotherapy: observations from a voronoi cell-based model. j theor biol 485, 110052 (2020). [47] f. hamdan et al., gamer-ad: a novel and rapid method for generating recombinant adenoviruses. mol ther methods clin dev 20, 625-634 (2021). [48] l. e. torres-dominguez, a. l. de matos, m. m. rahman, g. mcfadden, methods for the construction of recombinant oncolytic myxoma viruses. methods mol biol 2225, 63-75 (2021). [49] r. e. means, s. g. roy, s. g. katz, the propagation and quantification of two emerging oncolytic viruses: vesicular stomatitis (vsv) and zika (zikv). methods mol biol 2097, 253-263 (2020). [50] n. b. omar et al., safety and interim survival data after intracranial administration of m032, a genetically engineered oncolytic hsv-1 expressing il-12, in pet dogs with sporadic gliomas. neurosurg focus 50, e5 (2021). introduction corresponding author acknowledgements the author expresses his gratitude to professor phoebe stewart (department of pharmacology, case western reserve university school of medicine) for guidance and feedback in the preparation of the article. abbreviations ov: oncolytic virotherapy hsv: herpes simplex virus car-t: chimeric antigen receptor t nk: natural killer t-vec: talimogene laherparepvec trail: tumor necrosis factor-related apoptosis-inducing ligand htert: human telomerase reverse transcriptase ccl5: chemokine ligand 5 ccr5: chemokine receptor type 5 pd1: programmed cell death protein 1 pdac: pancreatic ductal adenocarcinoma cd19: b-lymphocyte antigen cd19 references microsoft word cusj part 2.pdf.doc 2024 article of the year ©2024 bhaumik, sattik. this is an open-access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. sephi of exoplanets kepler-504 b, kepler-315 b and kepler-315 c sattik bhaumik1 1minerva university, san francisco, california, usa keywords: habitability, sephi, exoplanets, life, liquid water, temperature, density, star, phi, esi abstract: the search for habitable exoplanets has improved with every passing year. new methods and advanced instrumentation with higher precision help find more habitable exoplanets and refine existing parameters of highly likely habitable exoplanets. this paper presents the statistical-likelihood exo-planetary habitability index (sephi) values for kepler-504 b (of star kepler-504), kepler-315 b, and kepler-315 c (both revolving around star kepler-315).1,2,3 sephi is based on the geometric mean of the likelihood gaussian estimation of four different comparison criteria with earth as the only place we know harboring life: telluricity, atmosphere and planet gravity, surface liquid water, and magnetic field.4,5,6 the seven physical characteristics of exoplanets have been used to calculate those four criteria: planetary mass, planetary radius, orbital period, stellar mass, stellar radius, and stellar effective temperature. this is a follow-up to previously calculated esi values for the three exoplanets mentioned above, with kepler-504 b and kepler-315 b having a high esi of 71.23% and 69.44%, respectively.7 it has been found that kepler-504 b (with a host m-type star (small red dwarf)) has a sephi value of 0, kepler-315 b (with a host g-type star) has a sephi value of 0, and kepler-315 c (with a host g-type star) has a sephi value of 0. thus, more than a combination of host star type, the orbital radius of exoplanet, and the final esi to determine probable habitability, a further in-depth analysis through sephi can help us confirm its actual habitability for earth-based life. introduction astronomers have been observing exoplanets for 32 years and studying their properties. in the last two decades, there has been immense progress in refining existing parameters through multiple missions by kepler space telescope (kst), transiting exoplanet survey satellite (tess), which have found 5500+ exoplanets, out of which 247 have an esi > 0.5. 69 exoplanets are potentially habitable till date as per the habitability worlds catalog.8 however, calculating earth similarity index, which compares the physical properties of the exoplanet to that of earth, (esi) is not enough to determine the likelihood to hold earth-based life as habitability—an independent criteria of esi.9 the esi values provide a general idea of possible habitability, but habitability can’t solely be concluded based on similar physical properties. hence to confirm habitability, an independent criterion of the esi, we need to conduct further in-depth analysis of the exoplanets by comparing more factors that contribute to life’s existence, sustenance, and proper environment for it to thrive. astronomers have used phi to determine habitability, but certain problems are associated with it. the phi is defined as the geometric mean of four sub-indexes: columbia undergraduate science journal vol. 18, 2024 bhaumik where s is substrate presence, e is available energy, c is the appropriate chemical composition, and l is the presence of a liquid solvent on the particular exoplanet. each of these variables is subdivided into different parameters (like atmospheric presence, magnetosphere, and redox chemistry) which are considered individually. the problem with phi is that every parameter is determined through a modulus function (ad-hoc classification) and often combines complementary criteria. for example, the energy received from the star and tidal flexing are both considered together for the determination of e, but these events do not occur simultaneously. furthermore, the phi value for jupiter or saturn is 0.40 and for mars, phi is 0.56. this implies that mars has a similar planetary habitability potential to that of jupiter or saturn, which is not true given the latter two are gas giants while mars still has a surface. hence, the sephi is used as a new form of phi calculation that utilizes statistical likelihood. these likelihood functions are determined by gaussian-like probability profiles (which is the exponential form). unlike phi which uses single variables, sephi uses comparison criteria (single variable/combination of them) and defines likelihood functions. additionally, sephi does not take into account any free parameter which means that all the comparison criteria in the analysis have equal weight. adopting sephi, with respect to esi, the results for sephi values and esi values for the exoplanets along with planets from our solar system have been added to give a reference base for understanding sephi’s relevance. methods for determining the four comparison criteria for the sephi, mozos et al. (2017) focused on those describing the basic conditions allowing life on earth and defined four sub-indexes ( ) for sephi. the final result was the geometric mean of these four sub-indexes, and it represented the statistical likelihood of a planet being potentially habitable with respect to earth from an astrobiological perspective.4 this definition was a statistic as the i were statistically independent: the following are the four comparison criteria: (i) telluricity ( ): to be a telluric planet, with part of its surface composed of solid silicates similar to the general composition of earth modeled as 17% fe and 83% mgsio3 (ii) atmosphere and planet gravity ( ): existence of a dense atmosphere that can retain the necessary gasses to sustain life, and the presence of a proper planetary gravity that is compatible with life. (iii) surface liquid water ( ): exoplanets having liquid water on its surface which is determined from the calculation of the habitable zone for the particular star system. (iv) magnetic field ( ): exoplanets having a strong magnetic field that will be able to protect life from solar radiation. participants/organisms: the data provided by nasa’s mast archive database was used along with the analysis of previous data on esi for the three exoplanets. the main data used were the previously calculated parameters that were cross-checked for any updates in the parameters from the nasa exoplanet archive.10,11 columbia undergraduate science journal vol. 18, 2024 bhaumik experimental design: quantitative data from the archive database was used in this paper. this data was observed and collected by kst as part of its legacy kepler and k2 missions. additional data was also used from tess. these data packets were already used in the previous esi calculations. the calculations conducted in this paper were built on the data used previously such as calculation for relative planetary mass (mp), relative planetary radius (rp), relative escape velocity (ve), stellar luminosity (ln), relative density (⍴p), stellar effective temperature (teff), which forms the basis for calculating the four composition criteria used to determine sephi values for each of the exoplanets.4 these particular criteria are essential to understand whether those basic conditions are compatible enough to allow earth-based life (carbon-based life) on that exoplanet. measurements/calculations: python programming language was used to calculate all four sub-indexes which were included in the sephi calculation for the exoplanets. data, such as habitable zone coefficients, and coefficients for magnetic field likelihood calculation were taken from previously published papers.12,13,14 each of the four criteria used has the following likelihood function: ℒ1: telluricity likelihood function from mozos et al. (2017), we know the following conditions for decay of the likelihood function: figure 1: relationship between a planet’s relative radius and composition parameters. the mean radius for a 100% mgsio3 composition is denoted as µ1, mp, and for a 50% mgsio3 and 50% h2o composition as µ2, mp . the standard deviation of the planet’s telluric likelihood function, σ1, mp, is calculated from these mean radii. where µ1, mp, is the relative radius (with respect to earth) of a planet containing 100% mgsio3 composition, µ2, mp, is the relative radius (with respect to earth) of a planet containing 50% mgsio3 and 50% h2o composition, and σ1, mp is the standard deviation σ1 equal to one-third of the difference of the relative radius of the planet with the above two compositions which limits ℒ1, mp between 0 and 1.4,15,16,17 the telluricity likelihood function for a given relative planet mass (mp) and relative radius (rp) is: figure 2: the likelihood function ℒ1, mp(rp) as gaussian-like profiles for telluricity based on relative radius rp with respect to earth. it also shows the pre-condition for each likelihood function under which a planet can be assumed as telluric, with three conditions, one for rp being less than or equal to µ1, mp, another one for a gaussian decay of exponential form if rp is between µ1, mp and µ2, mp , and lastly, if rp is greater than or equal to µ2, mp . ℒ2: atmosphere and planet gravity likelihood function from mozos et al. (2017), the likelihood function for planet gravity being compatible with earth life is: figure 3: the likelihood function ℒ2(ve) for calculating the ability of an exoplanet to maintain an atmosphere on the basis of relative columbia undergraduate science journal vol. 18, 2024 bhaumik escape velocity ve to that of earth. the functions are based on gaussian distribution profiles on two conditions of ve, with ve< 1 and ve >= 1. where ve is the relative planet gravity measured by √grp, σ21=1/3 and σ22=7.66/3. ℒ3: surface liquid water likelihood function the habitable zone boundaries dn depend on the following effective stellar flux (seff) with coefficients a, b, c and d depending on habitable zone and stellar effective temperature (teff), seff = seff, ☉+ a(teff 5780) + b(teff 5780)² + c(teff 5780)³ +d(teff 5780)⁴ and au, where ln is the stellar luminosity in solar units. the likelihood function is defined on the basis of orbital major axis a (4). the following are the defined zones with respect to a: hot zone: a < d1 inner-transition zone (itz): d1 ≤ a < d2 green zone: d2 ≤ a < d3 outer-transition zone (otz): d3 ≤ a < d4 cold zone: a > d4 the likelihood of holding liquid water if it is near (or within) the zone was calculated based on the planet’s location with respect to the corresponding habitable zone: figure 4. the likelihood function calculates the probability of the exoplanet holding water in liquid form on its surface as a gaussian-like profile distribution function on the basis of the exoplanet’s orbital semi-major axis a. the function returns a constant 1 if the semi-major orbital axis is in the green zone of d2 and d3. the other two conditions give us the probability of containing surface liquid water, one being the inner transition zone, where a < d2 and the other outer transition zone, where a > d3. where μ31= d2 , σ31= (d2 d1)/3 for the itz and μ32= d3 , σ32= (d4 d3)/3 for the otz (4). ℒ4: magnetic field the likelihood function is defined as: mn= αρ1/20n r10/30n f1/3n where r0n (normalized radius) and fn (normalized average convective buoyancy flux) are defined in terms of two corrections β1= rp/rr and β2= ⍴p/⍴r (related to the size and density of the reference planet) and α= 1.14 in case of kepler-504 b, a super earth, ρ0n= 1, r0n= β1, fn= β2 and in case of of kepler-315 b and kepler-315 c, two gas giants, ρ0n= 0.16, r0n= 16β1β2, fn= 100β1β2.18,19 the likelihood function of having a strong magnetic field to protect life from strong solar and cosmic radiation depends on mn: figure 5. the likelihood function ℒ4(mn) calculates the probability, using a gaussianlike distribution, as to whether an exoplanet has a strong dipolar magnetic field, which is required to protect life from cosmic radiation. the function calculates a gaussian decay value for a relative magnetic moment less than that of earth and returns a constant 1 for a relative magnetic moment greater than or equal to that of earth. after this, all the four likelihood function values are taken as a geometric mean to arrive at the sephi value = (ℒ1, ℒ2, ℒ3, ℒ4)1/4. data analysis: columbia undergraduate science journal vol. 18, 2024 bhaumik in table 1, the four criteria for the sephi along with the final sephi value for the exoplanets are shown. table 1: sephi values along with each of the likelihood function values of the comparison criteria for exoplanets kepler-504 b, kepler315 b and kepler-315 c. the values for ℒ show the probability of the criteria existing for that exoplanet where ℒ1 is the exoplanet’s telluricity, ℒ2, is the exoplanet’s relative gravity, ℒ3, is the probability to contain liquid surface water, ℒ4, is the probability to have a strong dipolar magnetic field. for comparison, the results of sephi values along with the corresponding esi values, host star type, habitable zone, and orbital radius are shown in table 2. for the calculation of values, python’s numpy library was used. table 2. comparison table for exoplanets kepler-504 b, kepler-315 b, and kepler-315 c where the esi, sephi along with the inner transition zone (itz), outer transition zone (otz), and the exoplanet’s orbital radii values are shown. these values indicate the exoplanet’s similarity to earth to hold carbonbased life across multiple parameters. results kepler-504 b, a super earth, has a telluricity likelihood of about 28.52%, while kepler-315 b and kepler-315 c being gas giants have 0 telluricity likelihood. for the exoplanet's relative gravity to that of earth, all 3 of them had comparable values of 0.011110 or 1.111% that of earth’s gravity. both kepler-504 b and kepler-315 b have 0 surface liquid water likelihood. kepler504 b having a surface does not harbor liquid water due to its high temperature of 114.84 celsius (k), 14.84 degrees above the boiling point of water, and this is also confirmed by the ℒ3 calculation. this high temperature is due to its close orbit with its host star as its orbital radius is at only 0.0646 au. it was interesting to see that kepler-315 c being a gas giant has a high liquid water likelihood of 82.1863%. its temperature is 51.31 celsius (k) which allows for water (if present) to exist in liquid form. all the 3 exoplanets were more massive than earth which means that they have a strong dipole magnetic field which is confirmed by their magnetic field likelihood value of 1 (or 100%) for all. the sephi value for super earth kepler-504 b is 0. both of the gas giants kepler-315 b and kepler-315 c also have 0 sephi values. getting even one value as zero among the four sub-indexes will lead to a 0 sephi since the geometric mean of these sub-indexes is taken. hence, a combination of stellar properties and esi value cannot be used to completely determine habitability. it is necessary to conduct more in-depth analysis on the particular exoplanet such as sephi, analyzing certain characteristics like telluricity, location in the habitable zone, surface temperature, surface liquid water, and strength of magnetic field, etc. discussion previous papers selected g-type and mtype host stars since most of the discovered habitable exoplanets revolve around those star types.7 the present data suggests that the chances of finding habitable exoplanets around these stars is extremely high and finding the esi value confirms that. however, by including more variables and comparison criteria, a more detailed analysis like sephi shows us that it is not always the case. habitability cannot be determined columbia undergraduate science journal vol. 18, 2024 bhaumik on the basis of stellar properties and esi values. further studies are required to determine habitability. the previous claim does give us a sense of general habitability, but, that is not a confirmation of life-supporting capacity for that exoplanet. for telluricity, the corresponding relative radius of a planet with 100% mgsio3 and 50% mgsio3 50% h2o were used given the exoplanet relative mass in terms of earth. this was a grid search where corresponding radii can be determined as a function of relative earth mass.15 for relative planet gravity calculation, 100% fe was used as the conservative reference to get the calculation within the estimation of the highest relative gravity that can be compatible for life to exist.20 the most updated equations for determining habitable zones were used.12,13,14 for magnetic field calculation α = 1 was used since the exoplanets studied were more massive than earth.19 the analysis used python as the core program for calculations. therefore, it includes rounding-off limitations floating point numbers. the code returns values only up to certain decimal points. the study must be extended to a larger group of exoplanets. sephi should be checked for exoplanets around other star types as well since sephi being characteristic and planet-specific might bring out some interesting observations and conclusions about other host-star type exoplanets. there are also parameter restrictions like telescope gathered data limitation, which introduces uncertainty in the data with regards to planetary properties like orbital radius, surface temperature, and telluricity. conclusion this research wanted to find whether stellar parameters and esi value is enough for determining habitability. however, upon finding the sephi values to be zero for all three exoplanets, it is determined that this is not always the case (even for exoplanets with host stars required for life to thrive, proper surface temperature but lack of surface soil telluricity). hence, habitability should not be declared once we find the conforming stellar types and high esi values. there is no doubt that a high esi value suggests a high probability of finding lifesupporting capacity. more comparison criteria must be checked for such that the exoplanet should fulfill for it to support carbon-based life. for future work, both esi as well as sephi values must be analyzed to completely determine habitability. future advanced telescopes like the present jwst will be more capable of observing the atmosphere which in turn will make calculations more precise and allow for the inclusion of more comparison criteria in the sephi and determine habitability more accurately. author information corresponding author sattik bhaumik sattik14bhaumik@gmail.com sattik@uni.minerva.edu notes the code used for calculating the sephi values and each of the comparison criteria can be found in this github repository: https://github.com/sattikbhaumik/analyzin g_exoplanets/blob/main/sephi.ipynb acknowledgments i would like to thank cusj for the constructive criticism and feedback to make the paper more accessible for others. i would like to thank my parents for their constant support. the author would like to thank the nasa mast astronomical archive database which gives open access to all researchers to conduct analysis and study the optical, ultraviolet, and near columbia undergraduate science journal vol. 18, 2024 bhaumik infrared. abbreviations esi: earth similarity index phi: planet habitability index sephi: statistical-likelihood exo-planetary habitability index kst: kepler space telescope tess: transiting exoplanet surveying satellite references [1] “exoplanet-catalog.” exoplanet exploration: planets beyond our solar system. https://exoplanets.nasa.gov/exoplanetcatalog/5699/kepler-504-b/. [2] “exoplanet-catalog.” exoplanet exploration: planets beyond our solar system. https://exoplanets.nasa.gov/exoplanetcatalog/5862/kepler-315-b/. [3] “exoplanet-catalog.” exoplanet exploration: planets beyond our solar system. https://exoplanets.nasa.gov/exoplanetcatalog/5863/kepler-315-c/. [4] mozos, j. m. rodríguez, and a. moya. “statistical-likelihood exo-planetary habitability index (sephi).” monthly notices of the royal astronomical society, vol. 471, no. 4, nov. 2017, pp. 4628–36. [5] banks, h. t., and h. t. tran. mathematical and experimental modeling of physical and biological processes. 0 ed., chapman and hall/crc, 2009. [6] grönholm, t., and a. annila. “natural distribution.” mathematical biosciences, vol. 210, no. 2, dec. 2007, pp. 659–67. [7] bhaumik, s., and g. sethi. “studying habitability of the exoplanets kepler504 b, kepler-315 b, and kepler-315 c.” journal of emerging investigators, 2022. [8] phl @ upr arecibo data. https://phl.upr.edu/hwc/data. [9] schulze-makuch, d., et al. "a twotiered approach to assessing the habitability of exoplanets." astrobiology, vol. 11, no. 10, dec. 2011, pp. 1041–52. = [10] kepler-504 | nasa exoplanet archive. [11] kepler-315 | nasa exoplanet archive. [12] kopparapu, r. k., et al. “habitable zones around main-sequence stars: new estimates.” the astrophysical journal, vol. 765, no. 2, feb. 2013, p. 131. [13] kopparapu, r. k., et al. “habitable zones around main-sequence stars: dependence on planetary mass.” the astrophysical journal, vol. 787, no. 2, may 2014, p. l29. [14] christensen, u. r., and j. aubert. “scaling properties of convectiondriven dynamos in rotating spherical shells and application to planetary magnetic fields.” geophysical journal international, vol. 166, no. 1, july 2006, pp. 97–114. [15] zeng, l., and d. sasselov. “a detailed model grid for solid planets from 0.1 through 100 earth masses.” publications of the astronomical society of the pacific, vol. 125, no. 925, mar. 2013, p. 227. [16] dressing, c. d., et al. “the mass of kepler-93b and the composition of terrestrial planets.” the astrophysical journal, vol. 800, no. 2, feb. 2015, p. 135. [17] dressing, c. d., and d. charbonneau. “the occurrence of potentially habitable planets orbiting m dwarfs estimated from the full kepler dataset and an empirical measurement of the detection sensitivity.” the astrophysical journal, vol. 807, no. 1, june 2015, p. 45. [18] lópez-morales, m., et al. “magnetic fields in earth-like exoplanets and implications for habitability around mdwarfs.” origins of life and evolution of columbia undergraduate science journal vol. 18, 2024 bhaumik biospheres, vol. 41, no. 6, dec. 2011, pp. 533–37. [19] cuartas-restrepo, p. "planetary magnetic fields and habitability in super earths." open astronomy, 27(1), 183-231 (2018). [20] marcus, r. a., et al. “collisional stripping and disruption of superearths.” the astrophysical journal, vol. 700, no. 2, aug. 2009, pp. l118–22. © 2025 lee, seonyu. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 19 unveiling dark matter in a galaxy: a case study of ngc 1068 seonyu lee1, heeju kim1 1department of physics and astronomy, seoul national university, seoul, republic of korea keywords: dark matter distribution, sed fitting, galaxy photometry abstract: the dark matter distribution in ngc 1068 is analyzed by determining its mass using dynamical mass from hi-based rotation curves and stellar mass from spectral energy distribution fitting of ugriz photometric data. high-precision photometry was achieved through gaussian fitting, star masking, 3-sigma clipping, and error propagation, enabling robust flux measurements for spectral energy distribution analysis with cigale. beyond 1.5 kpc, the dynamical mass begins to exceed the stellar mass. as the error bars do not overlap, the presence of dark matter is confirmed. the difference between the two masses is used to map the dark matter distribution from 1.5 kpc to 13 kpc. this study reproduces earlier findings using high-resolution data. furthermore, the method for constraining dark matter distributions in individual galaxies and provides detailed spectral energy distribution results with uncertainties is refined. introduction in the 1930s, astronomer zwicky introduced the concept of “dark matter”, a form of matter that interacts gravitationally but not with light, while studying the coma cluster.1 later in the 1970s, vera rubin confirmed the existence of dark matter by revealing the flatness of galactic rotation curves.2 despite accumulating evidence of dark matter’s existence, its nature remains one of the most significant unsolved mysteries in modern astrophysics. astrophysicists have studied dark matter’s distribution and interaction with baryonic matter within the cosmic environment, uncovering characteristics of dark matter particles. these studies have focused on large-scale structures such as galaxy clusters.3,4,5 however, investigating dark matter distribution within individual galaxies is equally important, as it provides valuable insights into galaxy formation and evolution and the interaction between dark and baryonic matter.6,7 additionally, galaxy simulations under various dark matter models can help elucidate its nature.8,9 this study presents a method for determining the dark matter distribution within a single galaxy, using ngc 1068 as a case study. as a seyfert type 2 galaxy, ngc 1068 hosts a heavily obscured active galactic nucleus (agn), making it a valuable target for studying the interplay between dark matter, stellar populations, and agn activity. its welldocumented structure and abundant multiwavelength data allow for a detailed investigation of both baryonic and nonbaryonic components. by comparing gravitational mass derived from rotation curves with stellar mass from spectral energy distribution (sed) fitting, this study reproduces historical findings providing the existence of dark matter. rotation curves probe the total gravitational potential of a columbia undergraduate science journal vol. 19, 2025 lee et. al © 2025 kohut, yuliia. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 20 galaxy, while sed fitting constrains the stellar content based on photometry. combining both enables us to isolate the dark matter component by comparing luminous and total mass distributions across different radii. furthermore, it visualizes the radial distribution of dark matter within the galaxy. for simplicity, the galaxy is modeled as a two-dimensional, axisymmetric plane. although this ignores asymmetries, it helps us understand overall trends. gravitational mass and its uncertainties are calculated from rotational curve data to map the gravitational mass distribution. radial photometry is performed using optical filter fits files, and photometric errors are calculated for sed fitting. the stellar mass and its uncertainty are derived from the best-fit parameters of the sed fitting process. these two methods collectively confirm the existence of dark matter, and the difference between the masses reveals the dark matter distribution, visualized as a colormap. methods data the observational data used in this study include fits files of ngc 1068, captured with the ugriz filters using the lsgt at siding spring observatory in australia (fig. 1). additionally, the rotational curve data were employed, sourced from brinks et al. (1997) [10]. this rotational curve consists of velocity measurements derived from hi emission lines et radii ranging from r = 20 arcsec to r = 180 arcsec (fig. 2). the inclination of the galaxy is already corrected in the data, making this dataset well-suited for calculating the gravitational mass distribution across a broad radial range. figure 1. ngc 1068 imaged using ugriz filters with the lsgt. figure 2. rotation curve of ngc 1068. rotation curve the rotation curve provides the distribution of velocities of stars and gas in a spiral galaxy as a function of radius, serving as a crucial tool for studying the galaxy’s mass distribution and dynamics. from the rotation curve of ngc 1068 presented by brinks et al. (1997), the velocities and their uncertainties were extracted at approximately 10 arcsecond intervals over the range r = 20 to 160 arcseconds.10 assuming a balance between centrifugal force and gravitational force according to newtonian dynamics, the gravitational mass and its uncertainties were calculated using the following equation: 𝑀 = 𝑣2𝑟 𝐺 , ∆𝑀 𝑀 = 2∆𝑣 𝑣 sed fitting sed fitting is a method used to optimize the flux emitted by galaxies across a broad range of wavelengths, from radio to x-ray. this technique provides insights into various physical properties of galaxies, such columbia undergraduate science journal vol. 19, 2025 lee et. al © 2025 kohut, yuliia. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 21 as star formation rate, star formation history (sfh), stellar mass, attenuation, dust mass and properties, metallicity, and active galactic nuclei (agn). as a critical tool in galaxy studies, sed fitting enables the estimation of galaxy evolution, activity, formation epochs, and the characteristics of stars and interstellar material within galaxies.11,12,13,14 the effectiveness of sed fitting can be enhanced through two primary approaches. the first is conducting observations across the widest possible range of wavelengths.15 different wavelength ranges predominantly trace distinct components of a galaxy. for instance, radio wavelengths trace agns and star formation, while infrared reveals emissions from gas, dust, and older stars. optical wavelengths capture stellar light, and x-rays highlight agns and hot gas.16,17,18,19 observing galaxies across a comprehensive wavelength range thus facilitates more accurate sed fitting and modeling. the second approach involves refining the models used for sed fitting.20 current sed fitting software incorporates models and modules that represent various components, such as sfh, single stellar population (ssp), nebular emission, dust attenuation, dust emission, and agns. by adjusting the types of modules and input parameters, the models can be tailored for greater accuracy. the key steps in sed fitting are as follows: 1. acquiring galaxy flux measurements through multi-wavelength photometry. 2. setting up models and input parameters. 3. performing fitting using statistical algorithms. 4. outputting the optimized sed and corresponding physical parameters. among the physical parameters provided by sed fitting programs are stellar mass and its bayesian uncertainties. in this study, we employed sed fitting to derive stellar mass and its associated errors from galaxy photometric data. cigale cigale, short for code investigating galaxy emission, is a software designed for performing sed fitting. as a python-based program, cigale offers lower technical complexity compared to other sed fitting tools and improves user accessibility with its automated features.21,22 it also allows high flexibility by enabling users to configure various modules and input parameters, as well as exclude unnecessary modules. despite limited observational data, cigale outputs the best-fit sed and parameters through sophisticated statistical analyses, such as chi-square minimization and bayesian inference. for these reasons, this study employed cigale for sed fitting.20 cigale includes a wide range of modules for sfh, ssp, nebular emission, dust attenuation, dust emission, agn, x-ray, radio, and redshift. however, since the observational data in this study is limited to the ugriz filters and optical wavelengths, only modules relevant to stellar emission were selected. the modules used in this study are listed below: 1. sfh: sfhdelayed.23 2. ssp: bc03.11 3. nebular emission: nebular.24 4. dust attenuation: dustatt_modified_cf00.25 modules related to dust emission, agn, x-ray, and radio were excluded due to the absence of corresponding observational data. the selected modules for sfh, ssp, nebular emission, and dust attenuation were chosen to yield the most reliable results given the columbia undergraduate science journal vol. 19, 2025 lee et. al © 2025 kohut, yuliia. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 22 limited dataset. modules with fewer input parameters and simpler structures were prioritized. input parameters for each module were configured by slightly expanding their default ranges. the following steps summarize the workflow of cigale: 1. pcigale init: create the config.ini file, which includes flux data for each filter, their measurement uncertainties, and module selections. 2. pcigale genconf: configure the parameters for the selected modules. 3. pcigale check: verify the parameter settings to ensure they are correctly configured before execution. 4. pcigale run: generate all possible sed models based on the modules and parameters specified in config.ini, compare the models to the observed fluxes, select the model with the lowest chi-square value, and output the optimal sed model along with physical parameters. 5. pcigale-plots sed, pdf, chi2: output graphs for the best-fit sed, parameter pdfs, and chi-square distributions. cigale evaluates model fit quality and determines the optimal model using statistical techniques like chi-square minimization and bayesian analysis. the chi-square computation incorporates a normalization factor for sfh, ensuring its integral equals one solar mass. this requires scaling the model values by a factor when calculating the chi-square. α = ∑ 𝑓𝑖 ×𝑚𝑖 σ𝑖 2𝑖 ∑ 𝑚𝑖 2 σ𝑖 2𝑖 + ∑ 𝑓𝑗 ×𝑚𝑗 σ𝑗 2𝑗 ∑ 𝑚𝑗 2 σ𝑗 2𝑗 χ2 =∑( 𝑓𝑖 − α ×𝑚𝑖 σ𝑖 ) 2 𝑖 +∑( 𝑓𝑗 − α ×𝑚𝑗 σ𝑗 ) 2 𝑗 +∑( 𝑓𝑘 −𝑚𝑘 σ𝑘 ) 2 𝑘 values with the subscript i represent the observed and model fluxes, while those with the subscript j denote observed and model parameters that depend on magnitude. values with the subscript k correspond to observed and model parameters that are independent of magnitude. 𝜎 represents the observational uncertainties. cigale identifies the model with the smallest chi-squared value as the best-fit model and stores its corresponding sed graph and physical parameters in the results.fits and results.txt files. while cigale identifies the optimal physical parameters by minimizing chi-square, galaxies with fundamentally different physical properties can exhibit similar seds. this issue becomes particularly challenging with limited observational data, as in this study.20 to address this, cigale goes beyond parameter extraction by employing bayesian analysis to derive probability distributions for each parameter. this process yields bayesian mean values, bayesian uncertainties, and confidence intervals for the parameters. the bayesian mean and confidence intervals are calculated using the following equations: 𝑃(θ|𝐷) = 𝑃(𝐷|θ) ⋅ 𝑃(θ) 𝑃(𝐷) ⟨θ⟩ = ∫θ ⋅ 𝑃(θ|𝐷) 𝑑θ ∫ 𝑃(θ|𝐷)  𝑑θ 𝑃(𝑎 ≤θ ≤ 𝑏) = ∫ 𝑃(θ|𝐷) 𝑏 𝑎  𝑑 here, d represents the observational data, and 𝜃 represents the parameter of interest. in cigale’s bayesian framework, the likelihood columbia undergraduate science journal vol. 19, 2025 lee et. al © 2025 kohut, yuliia. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 23 𝑃(θ|𝐷) is weighted by the chi-square statistic. 𝑃(𝐷|θ) ∝ 𝑒− 1 2χ 2 galaxy photometry after cigale input data required the redshift of ngc 1068, along with fluxes and their uncertainties in the ugriz filters at different radii. since the initial values significantly impact cigale’s results, accurate photometry and error estimation are essential. for ngc 1068, a redshift of z = 0.0038 was used. detecting the center of the galaxy to identify the center of the galaxy, the surrounding stars and their associated background light were removed. this was achieved by averaging the brightness over small predefined regions and normalizing them to a uniform value. subtracting the generated background from the original image resulted in a background-subtracted image. additionally, excessively bright points were excluded to ensure accurate background estimation. this process was carried out using the photutils and astropy packages in python. the pixel coordinates with the maximum brightness were assumed to be the center of the galaxy. the brightness distribution was found to be sufficiently circularly symmetric based on the visualization of the galaxy’s fits file. consequently, a 2d gaussian fit was applied to model the peak in the symmetric distribution. the equation used for the gaussian fitting is as follows: 𝑓(𝑥, 𝑦) = 𝐴 ⋅ exp (−[ (𝑥−𝑥0) 2 2σ𝑥 2 + (𝑦−𝑦0) 2 2σ𝑦 2 ]) + 𝐶 the residual map indicated that pixel values near the center are close to zero (fig. 3). these results verified the reliability of the gaussian fitting, and the extracted coordinates were set as the center of the galaxy. figure 3. residual map. masking star & radial photometry to perform accurate photometry of the galaxy, bright stars within the aperture photometry region needed to be masked. figure 4 shows the maximum photometry region used in this study and the masking of a bright star within that region. the center coordinates of the stars to be masked were identified, and the brightness of the detected star regions was replaced with pixel brightness corresponding to the background light. the photometry radius was set to match the radii extracted from the rotation curve graph. as the galaxy was assumed to be circularly symmetric in this study, circular photometry was performed instead of elliptical photometry (figure 5). the flux of the galaxy was calculated by summing the pixel brightness within the circular region and subtracting the background brightness times the area. to convert the galaxy flux obtained in this manner into observational magnitudes, the following equation was used: 𝑚 = 22.5 − 2.5 log10(𝐹) figure 4. stellar masking. columbia undergraduate science journal vol. 19, 2025 lee et. al © 2025 kohut, yuliia. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 24 figure 5. radial circular photometry. secondary calibration to derive the absolute flux from the observed data, standard stars within the dataset were identified. their relative magnitudes were then compared to known absolute magnitudes to perform photometric calibration. however, the data in this study did not include standard stars. thus, the right ascension (ra), declination (dec), and relative magnitudes of stars were obtained and matched with sdss data to determine absolute magnitudes (figures 6, 7).26 using the distribution of matched points, linear regression was applied to determine the values of k and zp. to improve accuracy, outliers were removed using 3sigma clipping (figure 8). in 3-sigma clipping, points that deviate by more than three times the standard deviation from the distribution of other points are excluded. the equation used for this secondary calibration is as follows: 𝑚std −𝑚inst = 𝑘 ⋅ 𝐶inst + zp 𝐶inst = 𝑚𝑖,inst −𝑚𝑧,inst figure 6. star detection. figure 7. matching errors in ra and dec. figure 8. before and after 3-sigma clipping. photometric error analysis cigale requires input data in the form of fluxes and photometric uncertainties for each filter. the following equation was used to calculate magnitude uncertainties. the snr represents the signal-to-noise ratio of the telescope, which was calculated assuming that the number of photons reaching the ccd follows a poisson distribution.27,28 snr = 𝑓𝑠 √𝑓𝑠 + 𝑓bkg , σ𝑚,obj = 2.5 ln(10) ⋅ snr the equation used to determine the uncertainty for secondary calibration is provided below. σ𝑚 = √(σ𝑘(𝐵 − 𝑉)) 2 + (𝑘σ𝐵−𝑉)2 + σ𝑐2 + σ𝑚,gal 2 lastly, the uncertainty arising from the process of converting absolute magnitudes into fluxes was calculated, with the value of 𝑓0 set to 3.631*103. 𝑓 = 𝑓0 ⋅ 10 −0.4𝑚, σflux = σ𝑚 ⋅ 0.4 ln(10) ⋅ 𝑓 the final total uncertainty was determined by summing the squares of these individual columbia undergraduate science journal vol. 19, 2025 lee et. al © 2025 kohut, yuliia. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 25 uncertainties and then taking the square root of the result. results gravitational mass the following graph shows the gravitational mass of ngc 1068 as a function of radius, calculated from its rotation curve. error bars represent the uncertainties (figure 9). to visualize this in two dimensions, a circular colormap was used. the boundary of ngc 1068 was extracted from an image taken by the hubble space telescope. the shape of the circular colormap was then adjusted to resemble the galaxy’s image, and the two were overlaid for representation (figure 10). figure 9. gravitational mass distribution of ngc 1068. figure 10. colormap of the gravitational mass. photometry from the results of linear regression for each filter, the values of k and zp were obtained (figure 11). figure 12 shows the flux as a function of radius for each ugriz filter, presented in order from top to bottom as u, g, r, i, and z. comparing the observed flux in jy units with data from the nasa extragalactic database (ned), it was found that the fluxes were smaller than those in ned: approximately 21% for the u-filter, 6% for the g-, r-, and ifilters, and 13% for the z-filter. figure 11. linear regression results for the ugriz filters. figure 12. flux as a function of radius for each ugriz filters (from top to bottom: u, g, r, i, z). columbia undergraduate science journal vol. 19, 2025 lee et. al © 2025 kohut, yuliia. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 26 stellar mass the following results were obtained from running cigale, showing the best sed model for r = 1.5 kpc, the chi-squared distribution of stellar mass, the probability density function (pdf), and the radial distribution of stellar mass (figures 13, 14, 15, 16, 17). figure 13. best sed r = 1.5 kpc. figure 14. chi-square distribution of stellar mass. figure 15: pdf of stellar mass r = 1.5 kpc. figure 16. stellar mass distribution of ngc 1068. figure 17. colormap of the stellar mass distribution. dark matter the difference between gravitational and stellar mass increases with radius, and their 1σ error bars do not overlap beyond 1.5 kpc (fig. 18), indicating a statistically significant discrepancy. this excess of gravitational mass relative to stellar mass is consistent with the presence of dark matter. the trend remains robust across the full radial range considered. this provides further evidence for the existence of dark matter, which interacts gravitationally but not electromagnetically, as revealed through dynamical analysis. figures 19 and 20 present the resulting dark matter mass distribution of ngc 1068. figure 18. gravitational mass vs stellar mass in ngc 1068. columbia undergraduate science journal vol. 19, 2025 lee et. al © 2025 kohut, yuliia. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 27 figure 19. dark matter mass distribution of ngc 1068. figure 20. dark matter colormap of ngc 1068. discussion results of cigale the residual map from the sed fitting shows that the error bars for the u-filter are abnormally large compared to other filters. this discrepancy likely stems from issues with the u-filter data or from fewer stars being detected during secondary calibration, compared to other filters. given that the lsgt has a relatively small apertures of 0.43m, the accuracy of u-filter observations may have been low, even with longer exposure times. furthermore, the flux of galaxies and stars in the u-filter was observed to be smaller than in other filters. this could have introduced significant errors during the linear regression or background light subtraction processes. using a high-performance telescope would allow for more refined and accurate analyses. these enhancements could address the limitations identified above. dark matter mass estimation dark matter mass is typically calculated as the difference between the total dynamical mass of a galaxy and the combined mass of its stars and gas. however, in this study, gas mass was not included due to the lack of observational data in the radio and infrared wavelengths, where gas emissions are primarily detected. estimating gas mass using optical data alone is unreliable and could lead to inaccurate results.29,30 additionally, the cigale modules for dust emission were omitted, further limiting the analysis. as a result, the dark matter mass estimation in this study considered only stellar mass, representing a key limitation. the graph comparing gravitational mass and stellar mass indicates that the two are nearly identical in the central region, suggesting that dark matter is absent within approximately r = 1.5 kpc. this observation is another limitation of this study. ngc 1068 is a galaxy with strong agn emissions at its center, but no agn-related modules were configured in cigale. as a result, agn emissions may have been misinterpreted as stellar emissions during the sed fitting process, leading to an overestimation of the stellar mass in the central region.31,32 this issue is further exacerbated by the lack of infrared and agnrelated data, which led to the exclusion of relevant cigale modules—such as agn, dust, and radio. consequently, the stellar mass is likely overestimated in agndominated regions, biasing the dark matter estimates within ~1.5 kpc. since agn emissions are predominantly observed in xray and radio wavelengths, which were not included in this study, likely contributing to the issue. securing observational data in the radio and x-ray wavelengths could address the limitations discussed above. sed fitting in previous studies typically incorporates observational data across a broad wavelength range, using columbia undergraduate science journal vol. 19, 2025 lee et. al © 2025 kohut, yuliia. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 28 measurements from approximately 10 to 20 filters. in contrast, this study relied on data from a narrow wavelength range and only five filters, which posed significant limitations. consequently, the reliability of the dark matter mass distribution results is likely affected. conclusion this study demonstrated the existence of dark matter and determined its distribution within a single galaxy, ngc 1068, using photometric data from the lsgt and its rotation curve. gravitational mass and its uncertainties were calculated from the rotation curve, and photometric data were generated to perform sed fitting with cigale. to ensure accurate photometric data, the study involved four key processes: identifying the galaxy’s center, masking bright surrounding stars, conducting secondary calibration, and calculating photometric uncertainties through signal-to-noise ratio analysis and error propagation. from the sed fitting results, the optimal stellar mass and bayesian errors were obtained. a comparison between the calculated stellar mass and gravitational mass confirmed the presence of dark matter, which cannot be observed through electromagnetic waves. radial photometry was performed, and sed fitting results were obtained for each radius, allowing the determination of the dark matter distribution. this study proposed a method for determining the dark matter mass distribution within a single galaxy, which can contribute to refining known dark matter density profiles (e.g. the nfw profile). additionally, the results provide valuable insights into galaxy formation and evolution. the significance of this study lies in its focus on a single galaxy. this offers a new direction for dark matter research, which has traditionally focused on larger-scale structures like galaxy clusters. author information corresponding author seonyu lee – julius712@snu.ac.kr author contributions seonyu lee led the research from conception to execution, including designing the study, analyzing results, and writing the manuscript. heeju kim contributed to the photometric data analysis and data processing. references [1] f. zwicky, die rotverschiebung von extragalaktischen nebeln. helv. phys. acta. 6, 110-127 (1933). https://doi.org/10.5169/seals-110267 [2] v. c. rubin, w. ford, jr. kent, rotation of the andromeda nebula from a spectroscopic survey of emission regions. apj 159, 379 (1970). https://doi.org/10.1086/150317 [3] m. markevitch, et al., direct constraints on the dark matter self-interaction cross. apj 606(2), 819-824 (2004). https://doi.org/10.1086/383178 [4] d. clowe, et al., a direct empirical proof of the existence of dark matter. apj 648(2), l109-l113 (2006). https://doi.org/10.1086/508162 [5] m. a. henson, d. j. barnes, s. t. kay, i. g. mccarthy, j. schaye, the impact of baryons on massive galaxy clusters: halo structure and cluster mass estimates. mnras, 465(3), 3361-3378 (2017). https://doi.org/10.1093/mnras/stw2899 [6] k. g. begeman, hi rotation curves of spiral galaxies. i. ngc 3198. a&a 223, 47-60 (1989). [7] a. a. dutton, a. v. maccio, cold dark matter haloes in the planck era: evolution of structural parameters for einasto and nfw profiles. mnras 441(4), 3359-3374 (2014). https://doi.org/10.1093/mnras/stu742 https://doi.org/10.5169/seals-110267 https://doi.org/10.1086/150317 https://doi.org/10.1086/383178 https://doi.org/10.1086/508162 https://doi.org/10.1093/mnras/stw2899 https://doi.org/10.1093/mnras/stu742 columbia undergraduate science journal vol. 19, 2025 lee et. al © 2025 kohut, yuliia. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 29 [8] v. springel, et al., simulations of the formation, evolution and clustering of galaxies and quasars. nature 435(7042), 629-636 (2005). https://doi.org/10.1038/nature03597 [9] j. s. bullock, m. boylan-kolchin, small-scale challenges to the λcdm paradigm. annu. rev. astron. astrophys. 55, 343-387 (2017). https://doi.org/10.1146/annurev-astro091916-055313 [10]e. brinks, e. d. skillman, r. j. terlevich, e. terlevich, hi observations of ngc 1068. astrophys. space. sci. 248, 23-31 (1997). [11]g. bruzual, s. charlot, stellar population synthesis at the resolution of 2003. mnras 344(4), 1000-1028 (2003). https://doi.org/10.1046/j.13658711.2003.06897.x [12]d. le borgne, et al., evolutionary synthesis of galaxies at high spectral resolution with the code pegasehr. a&a 425, 881-897 (2004). https://doi.org/10.1051/00046361:200400044 [13]c. conroy, modeling the panchromatic spectral energy distributions of galaxies. annu. rev. astron. astrophys. 51, 393-455 (2013). https://doi.org/10.1146/annurev-astro082812-141017 [14]c. pacifici, et al., the evolution of star formation histories of quiescent galaxies. apj 832(1), 79 (2016). https://doi.org/10.3847/0004637x/832/1/79 [15]j. walcher, b. groves, t. budavari, d. dale, fitting the integrated spectral energy distributions of galaxies. astrophys. space. sci. 331(1), 1-51 (2011). https://doi.org/10.1007/s10509010-0458-z [16]j. j. condon, radio emission from normal galaxies. annu. rev. astron. astrophys. 30, 575-611 (1992). https://doi.org/10.1146/annurev.aa.30.0 90192.003043 [17]r. c. kennicutt jr., star formation in galaxies along the hubble sequence. annu. rev. astron. astrophys. 36, 189232 (1998). https://doi.org/10.1146/annurev.astro.3 6.1.189 [18]k. nandra, p. m. o’neill, i. m. george, j. n. reeves, an xmm-newton survey of broad iron lines in seyfert galaxies. mnras 382(1), 194-228 (2007). https://doi.org/10.1111/j.13652966.2007.12331.x [19]s. bianchi, e. m. xilouris, the extent of dust in ngc 891 from herschel/spire images. a&a 531, l11 (2011). https://doi.org/10.1051/00046361/201116772 [20]m. boquien, et al., cigale: a python code investigating galaxy emission. a&a 622, a103 (2019). https://doi.org/10.1051/00046361/201834156 [21]e. da cunha, s. charlot, d. elbaz, a simple model to interpret the ultraviolet, optical and infrared emission from galaxies. mnras 388(4), 1595-1617 (2008). https://doi.org/10.1111/j.13652966.2008.13535.x https://doi.org/10.1038/nature03597 https://doi.org/10.1146/annurev-astro-091916-055313 https://doi.org/10.1146/annurev-astro-091916-055313 https://doi.org/10.1046/j.1365-8711.2003.06897.x https://doi.org/10.1046/j.1365-8711.2003.06897.x https://doi.org/10.1051/0004-6361:200400044 https://doi.org/10.1051/0004-6361:200400044 https://doi.org/10.1146/annurev-astro-082812-141017 https://doi.org/10.1146/annurev-astro-082812-141017 https://doi.org/10.3847/0004-637x/832/1/79 https://doi.org/10.3847/0004-637x/832/1/79 https://doi.org/10.1007/s10509-010-0458-z https://doi.org/10.1007/s10509-010-0458-z https://doi.org/10.1146/annurev.aa.30.090192.003043 https://doi.org/10.1146/annurev.aa.30.090192.003043 https://doi.org/10.1146/annurev.astro.36.1.189 https://doi.org/10.1146/annurev.astro.36.1.189 https://doi.org/10.1111/j.1365-2966.2007.12331.x https://doi.org/10.1111/j.1365-2966.2007.12331.x https://doi.org/10.1051/0004-6361/201116772 https://doi.org/10.1051/0004-6361/201116772 https://doi.org/10.1051/0004-6361/201834156 https://doi.org/10.1051/0004-6361/201834156 https://doi.org/10.1111/j.1365-2966.2008.13535.x https://doi.org/10.1111/j.1365-2966.2008.13535.x columbia undergraduate science journal vol. 19, 2025 lee et. al © 2025 kohut, yuliia. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 30 [22]c. conroy, j. e. gunn, m. white, the propagation of uncertainties in stellar population synthesis modeling. i. the relevance of uncertain aspects of stellar evolution and the initial mass function to the derived physical properties of galaxies. apj 699(1), 486-506 (2009). https://doi.org/10.1088/0004637x/699/1/486 [23]k. malek, et al., help: modelling the spectral energy distributions of herschel detected galaxies in the elais n1 filed. a&a 620, a50 (2018). https://doi.org/10.1051/00046361/201833131 [24]a. k. inoue, rest-frame ultraviolet-tooptical spectral characteristics of extremely metal-poor and metal-free galaxies. mnras 415(3), 2920-2931 (2011). https://doi.org/10.1111/j.13652966.2011.18906.x [25]s. charlot, s. m. fall, a simple model for the absorption of starlight by dust in galaxies. apj 539(2), 718-731 (2000). https://doi.org/10.1086/309250 [26]a. meiksin, colour corrections for high-redshift objects due to intergalactic attenuation. mnras 365(3), 807-812 (2006). https://doi.org/10.1111/j.13652966.2005.09756.x [27]b. m. tinsley, evolution of the stars and gas in galaxies. fundam. cosmic phys. 5, 287-388 (1980). https://doi.org/10.48550/arxiv.2203.0 2041 [28]n. gehrels, confidence limits for small numbers of events in astrophysical data. apj 303, 336 (1986). https://doi.org/10.1086/164079 [29]j. r. taylor, an introduction to error analysis: the study of uncertainties in physical measurements (university science books, 1997). [30]a. k. leroy, et al., the star formation efficiency in nearby galaxies: measuring where gas forms stars effectively. astron j 136(6), 2782-2845 (2008). https://doi.org/10.1088/00046256/136/6/2782 [31]k. iwasawa, a. c. fabian, a. j. young, h. inoue, c. matsumoto, variation of the broad x-ray iron line in mcg-6-3015 during a flare. mon. not. r. astron. soc. 306, l19-l24 (1999). [32]m. polleta, spectral energy distributions of hard x-ray selected active galactic nuclei in the xmmnewton medium deep survey. apj 663(1), 81-102 (2007). https://doi.org/10.1086/518113 https://doi.org/10.1088/0004-637x/699/1/486 https://doi.org/10.1088/0004-637x/699/1/486 https://doi.org/10.1051/0004-6361/201833131 https://doi.org/10.1051/0004-6361/201833131 https://doi.org/10.1111/j.1365-2966.2011.18906.x https://doi.org/10.1111/j.1365-2966.2011.18906.x https://doi.org/10.1086/309250 https://doi.org/10.1111/j.1365-2966.2005.09756.x https://doi.org/10.1111/j.1365-2966.2005.09756.x https://doi.org/10.48550/arxiv.2203.02041 https://doi.org/10.48550/arxiv.2203.02041 https://doi.org/10.1086/164079 https://doi.org/10.1088/0004-6256/136/6/2782 https://doi.org/10.1088/0004-6256/136/6/2782 https://doi.org/10.1086/518113 introduction author information corresponding author seonyu lee led the research from conception to execution, including designing the study, analyzing results, and writing the manuscript. heeju kim contributed to the photometric data analysis and data processing. references letter from the editor-in-chief dear readers it is with great pleasure that the cusj team presents you with the 17th volume of the columbia undergraduate science journal. since 2006, cusj has worked to elevate scientific journalism among undergraduates and provide a forum for showcasing their achievements. to honor our journal’s 18-year legacy, we took a step back to conduct a major refresh of editorial committee members, policies, and practices. we pride ourselves on following a strict code of ethics, a rigorous peer review system, and an inclusive author support program. maintaining these standards requires annual updates to our practices, allowing us to publish more works and provide our authors effective support. we appreciate your patience with us this year, and we look forward to offering our dear readers and student researchers a better journal each year. it is imperative that we put 100% of our best efforts into editing and publishing exceptional work, despite the committee improvements we spent extra time working on this year. therefore, we thought it was best to publish a short special edition for the 2023 columbia undergraduate science journal, solely featuring the strongest submissions from the year. on behalf of the entire editorial team, i’d like to thank every student who submitted their hard work. the columbia undergraduate science journal is a space to celebrate and promote budding scientists like all of you, and we’re looking forward to reading more of your work in the future. i’d also like to give a special congratulations to our published authors. your innovative ideas and thoughtful experiments are what keep our journal at the forefront of undergraduate scientific discourse, inspiring and informing both peers and professionals alike. i want to give a special shout-out to the cusj editorial committee. i onboarded a lot of new members this year, and they did a phenomenal job editing their first journal. i’d also like to thank the club's other editors-in-chief on our e-board for supporting me as i trained new members and worked on revamping our editing and publication process for the new year. i couldn’t have done it without your support. finally, i extend my gratitude to our faculty advisory board and the advisors at columbia libraries, whose assistance has been instrumental in enabling us to deliver research of the utmost quality. it was an honor serving as the editor-in-chief for the 2022-2023 edition of cusj, and i am looking forward to coming back stronger with the 2023-2024 edition. thank you for reading cusj! sincerely, kynnedy simone smith editor-in-chief, columbia undergraduate science journal 3 39 neural oscillations as predictors of variability in second language proficiency victoria ogunniyi1*, david abugaber2, irene finestrat2, alicia luque4, kara morgan-short2,3 1department of biological sciences, university of illinois at chicago, chicago, il ²department of hispanic linguistics, university of illinois at chicago, chicago, il ³department of psychology, university of illinois at chicago, chicago, il 4department of language and culture, uit the arctic university of norway, tromsø (norway) abstract: understanding what traits facilitate second language (l2) learning has been the focus of many psycholinguistic studies for the last thirty years. one source of insight comes from quantitative electroencephalography (qeeg), i.e., electrical brain activity recorded from the scalp. using qeeg, [1] found that functional brain connectivity is predictive of language learning ability. this study extends prat et al. in investigating the association of qeeg measures for two measures of l2 proficiency, namely: 1. a grammaticality judgment task, wherein participants read and identified spanish sentences as either correct or incorrect based on possible grammar violations, and 2. a standardized spanish proficiency test (dele). participants were low-intermediate l2 learners recruited from thirdand fourth-semester university spanish classes. spectral power and coherence within and across six different regions were analyzed for correlations with either scores on the grammaticality judgment task or on the dele. follow-up linear regression models based on significant qeeg correlates explained up to 11% of variance in dele scores but none of the variance in grammaticality judgment task performance. negative correlations were found between theta frequency coherence and the dele. because theta activity has been associated with episodic and working memory performance, these findings suggest that less proficient learners might utilize memory-based strategies more often to compensate for their lack of familiarity with the l2. keywords: second language acquisition, language proficiency, quantitative eeg, psycholinguistics, resting-state studies introduction understanding what characteristics underlie successful learning is not only pedagogically important but also intriguing from a cognitive standpoint. in the study of linguistics, this question has frequently been examined in regard to second language (l2) acquisition (for a review, see [2]). researchers have used a variety of approaches to investigate which cognitive abilities are correlated with higher l2 proficiency. going beyond behavioral measures for psycholinguistic and cognitive constructs © 2021 ogunniyi, abugaber, finestrat, luque, morgan-short. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. columbia undergraduate science journal vol. 15, 2021 ogunniyi et. al. 40 [2], neuroimaging techniques such as electroencephalography (eeg) allow us to directly quantify and connect brain activity to the processes involved in language processing [3]. this study addresses the question of whether eeg can explain variance on outcomes of l2 proficiency. we first review the literature on qeeg and its use as a neurocognitive measure in l2 studies, focusing on the studies that examine resting-state qeeg as a potential factor in language learning ability and language proficiency. then, in a conceptual replication and extension of [1], we describe the results of our study in which eeg measures of resting-state brain rhythms were explored for relationships with behavioral measures of l2 proficiency. the potential neurocognitive and pedagogical implications of these findings are expounded on in the discussion section. through our conceptual replication of [1], which focused on measures of l2 learning, we aimed to examine whether prior research on resting-state qeeg and l2 learning rate would extend to the construct of l2 proficiency, that is, whether a learner’s intrinsic pattern of brain rhythms is associated with their l2 abilities. background on qeeg eeg is an electrophysiological technique that utilizes electrodes placed on the scalp to measure changes in voltage between electrodes [4, 5]. these transient shifts in electric potential are caused by the electrical activity of neurons. due to their proximity to the scalp, pyramidal neurons, which project information to neurons in local regions, produce most of the postsynaptic potentials recorded by eeg [5]. although eeg data cannot attribute the electrical activity to specific brain regions, its temporal resolution allows researchers to track changes in brain activity to the millisecond. thus, since the 1960s, eeg has been a widely used tool in cognitive studies [6]. there are various methods that can be used to analyze eeg data. the analysis of raw eeg, or qeeg, data yields useful information via neural oscillations, that is, rhythmic or repetitive patterns of neural activity. in contrast to more common methods that analyze voltage amplitudes within given time windows tied to a stimulus, such as event related potentials, qeeg has the advantage of providing information about neural activity occurring before and after the onset of a stimulus, or even in the absence of any particular stimulus. in prior literature, qeeg has been used in disparate domains, such as serving as evidence of mental dysfunction in criminal cases [7], providing neurofeedback for therapy patients with adhd [8], and predicting learners’ aptitude for learning computer programming languages [9]. neural oscillations can be quantified through three measures: synchrony, amplitude, and coherence. the first measure, synchrony, describes whether neural oscillations are increasing or diminishing during a cognitive process [6]. by measuring the phase synchronization and desynchronization of the neural oscillations in this way, researchers can map larger interactions among the brain’s networks and demonstrate patterns of activation. this may provide insight into a possible solution to the binding problem, which asks how the brain integrates separate streams of information into one cohesive mental representation [5]. the second measure, amplitude, describes local changes in synchrony. amplitude within a certain frequency band is also often referred to as power. though an increase in power does not always reflect the presence of oscillations, sustained power increases within a narrow frequency range is usually a good indicator that oscillations are likely present at that frequency [6]. the third measure, coherence, describes the similarity in waveform properties and the stability of phase differences between two oscillations across brain regions [6]. though these three measures do not encompass all possible properties of oscillations, they can describe how oscillations represent activation and suppression of different neural networks, how wave amplitude is related columbia undergraduate science journal vol. 15, 2021 ogunniyi et. al. 41 to increased general activation, and how oscillations communicate over long distances. though their specific ranges vary between studies, there are five primary frequency bands that neural oscillations can be divided into [5, 6, 10]: delta (1-2 hz), theta (3-7 hz), alpha (8-12 hz), beta (13-30 hz), and gamma (30-200 hz). these bands have been implicated in a variety of cognitive processes. for instance, prior research has suggested a relationship between oscillation frequency and the range of neural network interactions: the lower frequencies, alpha and below, represent local interactions, whereas higher frequencies represent interactions between more distant brain regions [10]. in regard to the more general applications of qeeg, all of the frequency bands have been shown to play a role in memory. for instance, the delta band, which is the predominant frequency found in slow wave sleep, has been associated with memory consolidation [5]. in particular, delta oscillations facilitate the formation of declarative memory, the memory of one's experiences and explicit knowledge. similarly, theta oscillations have been implicated in both working memory and long-term memory retrieval [5, 11]. theta oscillations have primarily been observed in the cortex, further echoing these memory functions [10, 11]. alpha oscillations, which are the most prominent in the adult brain, are related to attention paid to external stimuli [5, 6]. more importantly, alpha plays a role in blocking irrelevant information in working memory. additionally, alpha desynchronization and reductions in alpha power result in more successful information encoding [10]. recent studies have shown that the beta band, which is mostly generated in the fronto-central region of the brain, also plays a role in regulating information stored in working memory [12]. the gamma band, which has been observed in the cortex [5], has been associated with shortand long-term memory maintenance [6]. additionally, increases in gamma activity is anti-correlated with beta activity levels [12]. additionally, the bands have been shown to play significant roles in stimulus-based language tasks. for instance, increased power in the theta band, which occurs during grammatical violations and sentence contexts that are difficult to interpret, reflects its involvement in lexical-semantic memory retrieval. similarly, the alpha band helps organize information stored in short-term memory during sentence comprehension [10]. regarding the higher frequency bands, gamma and beta have been associated with unifying related word meanings and similar grammatical forms, respectively. more specifically, the gamma band has been attributed to semantic unification [6, 10], which is supported by the observed decreases in gamma power in response to phrases with unclear meanings and idiomatic expressions [10]. conversely, the beta band is involved in syntactic unification [5, 6, 10]. beta oscillations sentence to lower processing regions [10]. by reflecting both domain-general and stimulus-specific cognitive functions, qeeg has proven to be a useful neurocognitive measure in psycholinguistic studies. the use of qeeg in l2 studies several l2 studies have tested the relationship between qeeg measures and l2 constructs. these studies have generally addressed two issues: l2 proficiency [13-16], which describes a learner’s language abilities at a given point in time, and l2 grammatical learning [17-19], which describes how learners better understand the rules of a language with increasing proficiency. regarding l2 proficiency studies, results have shown that highly proficient l2 learners differ in qeeg measures from less proficient l2 learners [14-16]. when comparing differently proficient participant groups, significant differences are found in the timing and location of oscillatory activity, especially regarding lateralization of the location of the oscillations between the right and left hemisphere. though the delta band is not frequently analyzed in these studies, one study found no significant group differences [13]. for grammatical learning studies, in both natural and arcolumbia undergraduate science journal vol. 15, 2021 ogunniyi et. al. 42 tificial grammar learning tasks, higher relative power and coherence in the higher frequency bands ( >8 hz) were associated with higher proficiency and increased in prevalence over time, whereas higher relative power and coherence in the lower frequency bands were associated with lower proficiency and decreased over time [16, 17, 19]. gauging learning by assessing oscillatory responses to grammatical violations, other research has also found that the higher frequency bands were elicited in both semantic and syntactic conditions [17]. a relatively new approach in l2 studies is to examine whether resting-state qeeg measures (i.e., that are taken when the brain is “idling” in the absence of any explicit task, as opposed to stimulus-related qeeg measures) are associated with the rate at which an individual acquires second language abilities, or l2 learning, and proficiency. as of now, very few studies have tested the correlation between neural activity occurring in the absence of a stimulus, or resting-state qeeg, and l2 learning rates [1, 20]. in these studies, native-english speakers learned a second language over the span of a few months. prior to this learning period, resting-state eeg was performed, and various behavioral measures were collected as additional outcome measures. in the first of these studies, [20] found that, when entered as predictors into a regression model, resting-state qeeg measures explained up to 60% of the variance observed in l2 learning rates, meaning that learning rate accounted for more than half of the variability in the data. though the most predictive frequency range was found to be the low-beta range (13-14.5 hz), power in the beta and gamma frequency bands recorded primarily over right hemisphere electrode regions were found to be the strongest predictors of l2 learning ability in general. the authors also found that alpha power over frontal and temporal electrodes and low-beta power over temporal regions were indicative of better language learning ability. as with other studies, greater activation in left hemisphere electrode sites was associated with lower l2 proficiency. in a later study conducted by [1] with a higher sample size, the authors tested whether resting-state qeeg measures were significant predictors of different l2 learning measures. similar to their earlier study, the results implicated the qeeg activity in the right hemisphere with greater l2 learning ability. simultaneous regression analyses were run on three outcome variables: l2 learning rate, total speech attempts, and performance on a declarative memory posttest. mean right posterior beta power was found to be a significant predictor of l2 learning rate and total speech attempts. frontotemporal to posterior coherence in the right hemisphere was found to be a significant predictor of performance on the declarative memory posttest, whereas mean within left posterior coherence across all frequencies was a significant predictor of total speech attempts. altogether, these results indicate that mean beta power over posterior electrode regions plays a significant role in l2 ability. in all, due to their potential advantages over stimulus-locked measures, qeeg measures have been increasingly used in l2 studies. though some frequency bands have been more frequently studied than others, all five of the classic bands have been implicated in various language functions in some way. across various research designs, qeeg has been used to illustrate large-scale patterns of brain activation during language processing. however, relatively little research has explored the potential of resting-state paradigms for qeeg in l2 psycholinguistics. for example, although [1, 20] have examined whether qeeg can predict individual differences in l2 learning rate, speech attempts during learning, and l2 declarative knowledge, no study has yet examined whether or how resting-state qeeg measures may be predictive of l2 proficiency, which is the desired final outcome of l2 learning. purpose of research the goal of this study is to investigate the columbia undergraduate science journal vol. 15, 2021 ogunniyi et. al. 43 potential association between resting-state qeeg and l2 proficiency. the current study expands on the research design of previous resting-state l2 studies. at present, resting-state qeeg studies have involved extensive training sessions with participants in the initial stages of l2 learning. however, the participant population utilized in this study comprised l2 learners at a low-intermediate stage who were recruited from thirdand fourth-semester university-level spanish courses. in connecting the qeeg measures with observed variability in language proficiency, we reasoned that second-year spanish learners with the qeeg profiles most conducive to effective processing (e.g., through memory functions and other related cognitive processes) would have gained the most spanish proficiency from their classes. additionally, this study has the advantage of analyzing a greater number of electrodes than prior research [1, 20], providing higher spatial resolution for measuring electrical activity on the scalp. as with prior resting-state studies, a variety of qeeg measures were analyzed, including spectral power and coherence. considering the issues above, this study specifically investigated two research questions: research question 1: is mean spectral power calculated from resting-state qeeg data associated with l2 proficiency, as assessed by two spanish proficiency tasks? research question 2: is mean withinand between-network coherence calculated from resting-state qeeg data associated with l2 proficiency, as assessed by two spanish proficiency tasks? following the results of [1], we predicted that the qeeg measures that were most likely to show a relationship with the two l2 proficiency tasks were mean beta power and frontotemporal-to-posterior coherence. in spite of these tentative predictions, we sought to replicate [1]’s methods as closely as possible in reproducing their two-step exploratory analysis strategy (i.e., pairwise correlations followed by multiple regression using significant correlates as predictors) on all frequency bands. as we intended our analysis itself to be strictly exploratory rather than confirmatory, we included all frequency bands in the analysis. in order to assess l2 proficiency, we decided to administer two assessments, one that reflects specific grammatical knowledge acquired through the learners’ class (the grammaticality judgment task), and one that is a more general and widely measure in the l2 literature (diplomas de español como lengua extranjera), see methods. given that prior research [21] has found a distinction between automatic and controlled language processing, using both a timed and untimed l2 proficiency measure would allow us to examine the application of l2 knowledge in two distinct ways. methods participants forty-nine participants (29 female; 20 male; mean age = 21.54; age sd = 4.14; range = 1838) were initially recruited for participation in a two-part eeg study in which they were tested in english and in spanish in separate testing sessions. participants were recruited at a large, public urban university in chicago from third and fourth-semester spanish language courses, which centered on developing communicative abilities and aimed to help students obtain low-intermediate proficiency by the end of the fourth-semester course. thirteen of these participants were recruited from the third-semester spanish course, while twenty-two of these participants were recruited from the fourth-semester spanish course; thirteen participants did not report their course level. the recruitment process involved advertising to these language courses, after which participants self-selected whether to participate. regarding the racial distribution of the participant population, nineteen participants identified as white/ caucasian, fourteen participants identified as asian, six participants identified as black/afcolumbia undergraduate science journal vol. 15, 2021 ogunniyi et. al. 44 rican-american, four participants identified as multiracial, and six participants’ races were unreported. all the participants reported having english as a native language (even if they were simultaneous bilinguals with early exposure to other languages) and having no spanish exposure growing up. additionally, per our eeg criteria, all participants were right-handed as assessed by the abridged version of the edinburgh handedness inventory [22], with normal or normal-to-corrected vision. this was done to ensure that our results were not confounded by uncorrected vision or differing brain activity in those who are left-handed. none of the participants reported having psychiatric, neurological, or learning disorders. all participants gave informed consent according to the standards of the university of illinois at chicago institutional review board and were financially compensated for their participation based on the number of hours spent participating in the study, receiving $5 for every hour spent on completing the proficiency measures, $15 for every hour spent during the eeg recording process, and a $45 bonus for completing both sessions. participants were only included in the analysis if they completed the resting state eeg recording along with at least one of our spanish proficiency measures. the final number of participants included in the current analysis is 47, with 44 of these participants having completed both spanish proficiency measures (see table i below). a similar eeg recording procedure was performed for an l1 english reading task (not reported here). the participant data included in this study was collected over the span of two years, from 2018 to 2020. grammaticality judgment task participants read various spanish sentences and were asked to determine whether they followed spanish grammatical rules. the grammaticality judgment task consisted of three experimental conditions: phrase structure, subject-verb agreement, and noun-phrase violations (see table ii for examples). the phrase structure condition, wherein word order violations were introduced into a sentence by presenting a noun instead of a verb or vice-versa, consisted of 60 correct and 60 violation sentable i. participant language characteristics m (sd) [range] number of native languages 1.29 (0.45) [1-2] number of l2s 1.44 (0.53) [1-3] age of acquisition spanish (years) 14.53 (4.88) [0-29] self-rated spanish listening proficiencya 4.88 (2.01) [1-9] self-rated spanish reading proficiency 5.42 (1.72) [1-8] note: aself-rated proficiency on 0 (‘none) to 10 (‘perfect) scale. table ii. examples of stimulus sentences item type example phrase structure ella tiene mucho dinero/*gastar que gastar/dinero en ropa. [she has a lot of money/*spend to spend/money on clothes.] subject-verb la mujer dibuja/*dibujan en su habitación. [the lady draws/*draw in her bedroom.] noun phrase el hombre prepara estas papas/ *papa para su esposa. [the man prepares these pota-toes/*potato for his wife.] note: * = violation word. italics indicate the critical correct/violation word in each sentence. columbia undergraduate science journal vol. 15, 2021 ogunniyi et. al. 45 tence frames, totaling 120 sentences overall. the subject-verb agreement condition, for which the verb ending did not agree with the plurality of the subject, consisted of 60 correct sentences with a singular subject, 60 correct sentences with a plural subject, and 120 violation sentences, totaling 240 sentences. the noun-phrase condition, for which either the singular/plural status or the grammatical gender of an article (e.g., los, “the [masc. plural]”; esta, “this [fem. singular]”) did not match the noun, consisted of 124 number violation frames, 124 gender violation frames, and 248 correct sentence frames, for a total of 496 sentences. in total, 856 sentences were used across all three conditions. the sentences ranged from 5 to 12 words in length. none of the sentences contained violations in initial or final sentence positions, so as to avoid sentence “start-up” and “wrap-up” effects in the eeg [23-24], and none of the critical words were repeated between frames. to ensure that the participants would be familiar with the vocabulary contained in the sentence frames, all the words for this task were taken from a spanish textbook used at the university at the time that data collection began [25]. sentences were presented one word at a time on the computer screen (see figure 1) using e-prime 2.0 software. instructions for the task were read orally to the participants by the experimenter. preceding each sentence, there was a screen that read, “rest your eyes.” after three seconds, the sentence was then visually presented one word at a time. each word was displayed in the center of the screen for 350 ms, with a 150 ms interval of blank screen before the onset of the subsequent word. once the entire sentence was presented, a screen followed that said, “good/bad?” in response to this prompt, participants pressed a keyboard button to categorize the sentence as either grammatical or ungrammatical. participants first completed a short practice block containing 8 sentences. the stimuli sentences were then presented over 4 experimental blocks. there were three 3-minute breaks during the experiment, one at the end of each block. another eeg recording was performed over the duration of this task but was not analyzed in the current study. participants’ responses were used to calculate a d-prime (dʹ) score, which is a metric for signal detection that accounts for response bias by comparing how often a participant correctly identifies a signal to their false-alarm rate [26]. figure 1: diagram of a typical trial in the grammaticality judgment task. diplomas de español como lengua extranjera (diploma of spanish as a foreign language): a modified version of the diplomas de español como lengua extranjera (dele, [27]) was used to assess spanish proficiency. the three-part test was completed on a computer in the laboratory through a qualtrics survey form. participants were asked to read through the dele questions and answer them at their own pace. in the first section, participants were required to read through a passage in spanish and answer 20 fill-in-the-blank questions. each question had 3 possible answer choices. in the second section, participants were given 10 sentences and asked to choose the answer choice that best defines the bolded word in the sentence. each sentence also had 3 possible answer choices. the third section consisted of 19 grammatical questions . participants were asked to select the answer choice that fit best in the context of each of the sentences. eight of these questions had 2 possible answer choices, and the remaining questions had 4 possible answer choices. in total, participants answered 49 questions. columbia undergraduate science journal vol. 15, 2021 ogunniyi et. al. 46 procedures prior to testing, all participants completed pre-testing questionnaires that verified their eligibility and provided more detailed information about their language history. this included a language background questionnaire, a test-session questionnaire, and a handedness questionnaire. the language background questionnaire assessed each participant’s demographic background and language history and experience (leap-q, [28]). the test-session questionnaire assessed how much sleep a participant had and whether they had taken any psychoactive substances that may affect their ability to perform the task. the handedness questionnaire was used to gauge left-/right-handedness by assessing hand preference during various activities, following the standard edinburgh handedness inventory [22]. the items were read to participants, who provided their answers verbally to the experimenter. answers were recorded in computer-based survey forms. in replicating [1]’s procedure, we collected five minutes of eyes-closed resting-state eeg following completion of the pre-testing surveys. participants sat in a chair inside of a sound-attenuating booth. after fitting the participants for an eeg cap and placing eye electrodes, an electrolyte solution was applied to the scalp electrodes to minimize electrical impedances. participants were then instructed to close their eyes and remain still and awake during the recording. while recording, the lights were turned off and the door of the sound booth was closed. the eeg data was recorded using asa™ software with an ant neuro waveguard™ elastic cap with 32 ag/agcl electrodes distributed in standard and extended 10-20 system locations. scalp impedances were lowered to 10 kω or below. scalp electrodes were referenced to the common average of all the electrodes. to detect artifacts caused by eye movements, electrodes were placed above and below the right eye and on the left and right outer canthi to record a vertical electrooculogram and a horizontal electrooculogram, respectively. using an ant neuro bioamplifier system (amp-trf40ab refa-8 amplifier), the eeg signal was amplified to 22 bits. the signal was also recorded in dc mode, digitized with a 512 hz sampling rate, and filtered online using a low-pass filter with a cutoff of 138.24 hz. following the resting-state eeg session, the grammaticality judgment task was implemented in the sound booth (see above), and after disassembly of eeg equipment and a short break the dele task was performed on a computer outside of the sound booth (see above). analyses the qeeg data was pre-processed using the ee-glab toolbox [29] for matlab [30]. to ensure that the resting-state recording was exactly five minutes, each recording was limited to 300 seconds. seven participants had recordings that were slightly less than 300 seconds (minimum = 283 seconds) but were still included in the analysis. each participant’s recording was divided into epochs of two-second duration, with 50% overlap across epochs. these epochs were then cleaned for artifacts (e.g., from muscle movements, eyeblinks, faulty electrodes, etc.) using the pop_autorej() function from eeglab. participant datasets with less than 75 seconds of epoch-free recording were omitted from the final analysis, which resulted in the loss of 6 participants (12% of the data). the mean number of samples per participant was 144.44 (s.d. = 39.73). the pre-processed data were subsequently analyzed using a modified version of the script used in the prat et al. study [1] (available at: https://github.com/uwccdl/qeeg) for the r scripting language [31]. six electrode networks were defined (figure 2): medial frontal (electrodes fp1, fpz, fp2, and fz), left hemisphere fronto-temporal (electrodes f7, fc5, t7, c3), right hemisphere frontotemporal (f8, fc6, t8), left hemisphere posterior (cp5, cp1, p7, p3, o1), right hemisphere posterior (cz, cp6, cp2, c4, p4, pz), and right hemisphere posterior occipital (oz, o2, poz, p8). columbia undergraduate science journal vol. 15, 2021 ogunniyi et. al. 47 in defining the electrode clusters, we aimed to replicate prat et al. (2019), in which qeeg channels were collapsed into networks based on phase synchrony results from an earlier l2 qeeg study. prat et al. 2019’s network definitions are technically slightly different from ours in that they used 14 electrodes rather than a 32-electrode cap, but our network definitions were aligned as closely with theirs as possible based on visual inspection of scalp maps (and in fact having a higher spatial resolution is a point in our favor in a sense). as data-driven results in favor of our network definitions (which we left out due to space limits), we replicated [1] (2019, table i) in that independent samples t tests found that all within-network qeeg coherence values in our networks were significantly greater than all between-network coherence values, with all independent samples t-tests at p < .001. we then extracted the qeeg measures of interest, which were power and withinand between-network coherence for each of the frequency bands: theta (4-7.5 hz), alpha (8-12.5 hz), beta (13-29.5 hz), and gamma (30-40 hz). figure 2: diagram of the six network regions analyzed finally, to address the research questions, we first conducted correlations between mean power and performance on the grammaticality judgment task and the dele, followed by correlations between mean coherence and performance on the grammaticality judgment task and the dele. here we report statistically significant correlations. these exploratory correlations were not corrected for multiple comparisons following the main analyses reported by [1]. for power and coherence measures that showed statistically significant correlations for either proficiency variable, we then entered them as predictors into two, separate regression analyses. results individual differences in indicators of l2 proficiency before examining the qeeg measures, descriptive statistics were examined for the two outcome measures of spanish proficiency (see figures 3 and 4 and table iii). with respect to the dele, the group mean was 19 out of 49, which illustrates that the participants were overall low proficiency speakers [27]. the most proficient participant scored twice as much as the least proficient participant. with respect to the grammaticality judgment task, participants were given two scores: mean accuracy and dʹ. the average accuracy on the grammaticality judgment task was 76%. the average dʹ was 0.91. a bootstrapped simulation of chance-level dʹ values on 244 trials with 10,000 iterations performed using the psycho package for r [32] found a 95% confidence interval of -0.26 to 0.26. this suggests that our participants’ dʹ values were above chance at α = 0.05. dele scores did seem to be above chance, as indicated by a mean accuracy of 39.5%. considering that most of the dele test items had 2, 3, or 4 answer choices, a minimum accuracy of 25% would at least reflect chance levels on the items with the most answer choices. dele scores and grammaticality judgment task dʹ scores were not significantly correlated with one another, r(44) = 0.27, p = 0.069. as per [21], our results suggest that the grammaticality judgment task and dele might capture somewhat different facets of l2 proficiency. columbia undergraduate science journal vol. 15, 2021 ogunniyi et. al. 48 table iii. performance on the two proficiency tasks performance measure m (sd) [range] dele score 19.3 (3.15) [0.13-0.26] grammaticality judgment task dʹ score 0.91 (0.74) [-0.27-2.88] grammaticality judgment task accuracy 0.76 (0.10) [0.51-0.98] figure 3: participant performance on the grammaticality judgment task. the maximum possible dʹscore on the grammaticality judgment task was effectively 4.9. figure 4: participant performance on the spanish proficiency test (dele). the maximum possible score on the dele was 49. relating individual differences in resting-state qeeg power to l2 proficiency variables in order to determine the relationship between resting-state qeeg power and performance on the two proficiency tests, we performed correlation analyses between either the grammaticality judgment task or dele scores (in separate analyses) and mean power across six electrode networks. the frequency bands of interest were theta (3-7 hz), alpha (8-12 hz), beta (13-30 hz), and gamma (30-200 hz). after conducting these analyses, none of the correlations were found to be significant. however, two positive correlations were approaching significance: medial frontal alpha power and dele scores, r(47) = 0.28, p = .052; and left hemisphere frontotemporal alpha power and dele scores, r(47) = 0.27, p = 0.057. relating individual differences in resting-state qeeg coherence to l2 proficiency variables in order to determine the relationship between resting-state qeeg coherence and performance on the two proficiency tests, we performed correlation analyses between the l2 proficiency variables and mean withinand between-coherence across six electrode networks. the frequency bands of interest were theta, alpha, beta, and gamma. for the between-coherence values, each of the networks were paired together and coherence across the four frequency bands was calculated for every pair. regarding the dele, two significant negative correlations were found: theta coherence within the right hemisphere posterior occipital network, r(47) = -0.31, p = 0.028; theta coherence between the me-dial frontal and right hemisphere posterior occipital networks, r(47) = -0.35, p = 0.012. regarding the grammaticality judgment task, there were no signif-icant correlations (all p > 0.05). simultaneous linear regression analyses when the two predictors of performance on the dele (theta coherence within the right hemicolumbia undergraduate science journal vol. 15, 2021 ogunniyi et. al. 49 sphere posterior occipital network and theta coherence between medial frontal and right hemisphere posterior occipital) were entered into a simultaneous regression analysis, the overall model was found to be statistically significant, f(2, 46) = 3.97, p = 0.026, explaining up to 11% of the observed variance. however, neither theta coherence within the right hemisphere posterior occipital network (β = -13.91, t = -1.07, p = 0.290) nor theta coherence between medial frontal and right hemisphere posterior occipital (β = -41.78, t = -1.61, p = 0.114) were found to independently predict performance on the dele (see figures 5 and 6). figure 5: regression line between the dele and theta coherence within right hemisphere posterior occipital networks. figure 6: regression line between the dele and theta coherence between the medial frontal and right hemisphere posterior occipital networks. discussion our results suggest that certain resting-state qeeg measures, particularly over the theta frequency band, are associated with l2 proficiency. regarding the first research question, none of the correlations run between resting-state mean power and l2 proficiency reached significance. however, the two correlations that did approach significance were related to alpha power: the positive correlation between medial alpha power and the dele, and the positive correlation between left hemisphere frontotemporal alpha power and the dele. regarding the second research question, two significant negative correlations were found between resting-state withinand between-network coherence and the dele, one within right hemisphere posterior networks and another between medial frontal and right hemisphere posterior networks. both significant correlations were found over the theta frequency band. after performing regressions on the significant qeeg predictors for dele performance, the model was found to explain up to 11% of the variance. none of the variance in grammaticality judgment task performance was explained by qeeg measures. in relating our results with those of previous studies, the correlations found between the theta frequency band and l2 proficiency were anticipated, although the direction of the relationship varied by study. in [1], theta coherence within frontal electrode regions was positively correlated with several outcome measures of l2 learning. in [16], highly proficient speakers experienced increased theta synchronization in right frontal regions during a grammar learning task. however, in the l2 proficiency study conducted by [13], lower theta coherence in frontal and occipital electrodes was observed among highly proficient speakers. in our study, the relationship between theta coherence measures and l2 proficiency was also negative. why do we see these contradictory patterns among studies? this may be explained by the differences in language learning and language proficiency. for instance, the participant columbia undergraduate science journal vol. 15, 2021 ogunniyi et. al. 50 population used for language learning studies consists of speakers who were just beginning to learn an l2, whereas proficiency studies involve participants who have already had experience learning the additional language. we understand activity in the theta frequency band to reflect several memory functions, including specifically shortand long-term memory maintenance and memory retrieval [5, 10, 33, 34]. additionally, the theta frequency is believed to originate from the cortex [5], which plays a role in the formation of new memories. generally, studies have found positive relationships between theta activity and learning rate in earlier stages of learning, negative relationships between theta and l2 ability in later stages [1, 16, 19]. the negative relationship found in this study may signify that more proficient participants are good at applying and retrieving grammar rules and no longer need to rely on working memory, which would result in a decreased prevalence of theta oscillations. altogether, as suggested by [16], the negative correlations found between theta coherence and dele performance suggest that less proficient l2 speakers have a greater reliance on memory-based strategies to compensate for their lack of familiarity with the language. even though the alpha frequency band was not significant, it was approaching significance, which reflects the inverse theta-alpha relationship expressed in the literature [5]. increases in the alpha frequency have been associated with diminished attention paid to a linguistic task [10]. interpreting the negative correlation between theta and proficiency to be the consequence of decreased working memory load, a positive correlation with alpha would suggest that more proficient participants were able to pay less attention to the task and still be successful. this may signify that the more automatic a language task is to a participant, the more likely they are to be more proficient. the results of this study need to be considered in light of its limitations. one limitation was the proficiency measures that were employed. as mentioned in the results section, our participants did not score statistically above chance on average on the dele. thus, it is somewhat surprising that significant results were found for the dele and not for the grammaticality judgment task. perhaps the more difficult dele test, which had been developed to test up to near-native speaker status, allowed us to detect a role related to which learners are more successful on a more challenging task. conversely, the grammaticality judgment task was designed to reflect specific grammatical structures taught in intermediate-level spanish courses, so we would expect participants to perform more successfully on this task overall, which they did, as evidenced by above-chance performance. more generally, because the dele and grammaticality judgment task both reflect performance accuracy on grammatical tasks, it is important to note that they are not holistic measures. as a multidimensional construct, language proficiency encompasses all the skills necessary to engage with the language in a real-life context [2]. thus, it would be beneficial for future research to include more time-pressured proficiency measures in future research, such as an oral elicited imitation (eit) task. unlike the dele and grammaticality judgment task, which are both primarily prescriptive grammar tasks, the eit can assess more implicit language knowledge by asking participants to listen to sentences and repeat them [35, 36]. this testing method has been found to engage long-term memory and require a higher level of language comprehension. it is also worth noting that, while we found significant correlations with theta coherence, the majority of the qeeg measures were not associated with l2 proficiency. though we interpreted theta to reflect the engagement of working memory, one difficulty in interpreting the results of this study, is in interpreting what cognitive processes may be reflected by the different frequency bands. although previous research does suggest associations between activity in the frequency bands and different columbia undergraduate science journal vol. 15, 2021 ogunniyi et. al. 51 cognitive processes, future research will need to strengthen the validity of these claims regarding theta. in regard to the lack of a relationship between l2 proficiency and the other frequency bands, null results can be difficult to explain, but the results could be at least partially due to our processing procedures where a certain amount of the data was not included. for future analyses, we plan on implementing independent component analyses performed to correct for eye and muscle artifacts in eeg data (using iclabel; [37]), which is expected to lead to lead to cleaner data, higher sample sizes, and improved model fits for both dele and grammaticality judgment task. we note three further limitations in our study that should be addressed in future research. first, future research might want to analyze data from a particular semester rather than data spanning participants from two spanish course levels, or the course level could be included as a covariate in analyses. second, regarding the analysis, a more precise way to define the frequency ranges of the specific frequency bands is to use individual alpha frequency (iaf) peaks. for each person, the iaf peaks at a different number, which affects the ranges of the other frequency bands [38]. finally, given the highly exploratory nature of this study, we did not correct for multiple correlation analyses, and we entered regression predictors based on significance from the correlational analyses. future research should conduct more conservative, confirmatory analyses on a dataset with higher statistical power to mitigate possible type i errors. indeed, post hoc analyses for our dataset showed that the significant correlations reported above did not survive correction for type i error inflation using the family-wise discovery rate, which further suggests that a confirmatory study would need to be conducted to validate any of the exploratory findings reported in this study. using this method in future research may further solidify the validity of our results or may lead to different findings. conclusion the purpose of the current study was to investigate whether mean qeeg power and coherence are significant predictors of l2 proficiency. based on our results, withinand between-network coherence over the theta frequency band is closely related to spanish l2 proficiency. because the theta frequency has been associated with memory retrieval and load, these results suggest that there is an inverse relationship between l2 proficiency and reliance on memory-based strategies for interpreting linguistic inputs. additionally, increased theta activity may be characteristic of individuals in earlier stages of language learning. more research is needed to further validate the significance of theta in l2 proficiency, as well as to determine the importance of the other frequency bands. ultimately, this study adds to growing literature of resting-state l2 qeeg studies, echoing the implication of intrinsic patterns of neural activity as sources of individual variation in linguistic ability. over time, qeeg may help to reveal individual neurophysiological variations among students within a classroom, enabling educators to develop language learning strategies that will be most conducive to successful l2 outcomes for them. in other words, from the conclusions of this body of research, we might be able to identify particular cognitive processes that are associated with l2 learning and proficiency. with such information, further research could then examine how to leverage these processes in instruction. author information corresponding author victoria ogunniyi *vogunn2@uic.edu author contributions victoria ogunniyi assisted with data collection and performed the analyses along with irene martinez and david abugaber. victoria ogunniyi wrote the manuscript. kara columbia undergraduate science journal vol. 15, 2021 ogunniyi et. al. 52 morgan-short, david abugaber, irene finestrat, and alicia luque reviewed the manuscript and devised the experimental plan. competing interests the authors declare no competing financial and non-financial interests. acknowledgments thank you to kinsey bice for her help with performing the qeeg analyses. thank you also to all undergraduate research assistants affiliated with the cognition of second language acquisition laboratory who contributed towards data collection for this project, including hannah chaddah, sarah hassan, gayatri chavan, will camacho, ana hernandez, anahi gaytan, krishna bhatia, amber lewis, mallory webber, evelyn delgado, daniel henreckson, yasiel lacalle, and nethaum mizyed. we would also like to thank sarah malone, christopher powell, crystal galvan, and allen bryson of the university of illinois at chicago’s summer research opportunities program for undergraduates for their support in the development of this manuscript. abbreviations delediplomas de español como lengua extranjera eeg – electroencephalography l2 – second language qeeg – quantitative electroencephalography references [1] c.s. prat, b.l. yamasaki, e.r. peterson, individual differences in resting-state brain rhythms uniquely predict second language learning rate and willingness to communicate in adults. j cogn neurosci 31(1), 78-94 (2019). [2] p. leclercq, a. edmonds, h. hilton, eds. measuring l2 proficiency: perspectives from sla (multilingual matters, 2014). [3] c. j. doughty, m.h. long, “the scope of inquiry and goals of sla” in the handbook of second language acquisition (john wiley & sons, 2003), pp. 3-16. [4] s. j. luck, an introduction to the event-related potential technique (mit press, 2014). [5] k. majumdar, a brief survey of quantitative eeg, 2nd ed. 2018 (crc press, 2018). [6] m. bastiaansen, a. mazaheri, o. jensen, “beyond erps: oscillatory neuronal dynamics” in the oxford handbook of event-related potential components, s. luck & e. kappenman, eds. (oxford university press, 2012), pp. 31-50. [7] g. gluck, qeeg accepted in death penalty trial in florida v. nelson. biofeedback 39(2), 74-77 (2011). [8] m. arns, w. drinkenburg, j. leon kenemans, the effects of qeeg-informed neurofeedback in adhd: an open-label pilot study. appl psycho-physiol biofeedback 37(3), 171180 (2012). [9] c.s. prat, t.m. madhyastha, m.j. mottarella, c.h. kuo, relating natural language aptitude to individual differences in learning programming languages. sci rep 10(1), 1-10 (2020). [10] y. prystauka, a.g. lewis, the power of neural oscillations to inform sentence comprehension: a linguistic perspective. lang linguist compass 13(9), e12347 (2019). [11] l. meyer, the neural oscillations of speech processing and language comprehension: state of the art and emerging mechanisms. eur j neurosci 48(7), 2609-2621 (2018). [12] m. lundqvist, p. herman, m.r. warden, columbia undergraduate science journal vol. 15, 2021 ogunniyi et. al. 53 s.l. brincat, e.k. miller, gamma and beta bursts during working memory readout suggest roles in its volitional control. nat commun 9(1), 1-12 (2018). [13] s. reiterer, c. hemmelmann, p. rappels-berger, m.l. berger, characteristic functional networks in highversus low-proficiency second language speakers detected also during native language processing: an explorative eeg coherence study in 6 frequency bands. brain res cogn brain res 25(2), 566578 (2005). [14] s. reiterer, e. pereda, j. bhattacharya, measuring second language proficiency with eeg synchronization: how functional cortical networks and hemispheric involvement differ as a function of proficiency level in second language speakers. second lang res 25(1), 77-106 (2009). [15] s. reiterer, e. pereda, j. bhattacharya, on a possible relationship between linguistic expertise and eeg gamma band phase synchrony. front psychol 2, 334 (2011). [16] o. kepinska, e. pereda, j. caspers, n.o. schiller, neural oscillatory mechanisms during novel grammar learning underlying language analytical abilities. brain lang 175, 99-110 (2017). [17] a. kielar, j.a. meltzer, s. moreno, c. alain, e. bialystok, oscillatory responses to semantic and syntactic violations. j cogn neurosci 26(12), 2840-2862 (2014). [18] a.g. lewis, k. lemhӧfer, j.m. schoffelen, h. schriefers, gender agreement violations modulate beta oscillatory dynamics during sentence comprehension: a comparison of second language learners and native speakers. neuropsychologia 89, 254-272 (2016). [19] z.r. cross, a. w. corcoran, m. schlesewsky, m.j. kohler, i. bornkessel-schlesewsky, oscillatory and aperiodic neural activity jointly predict grammar learning. biorxiv (2020). [20] c.s. prat, b.l. yamasaki, r.a. kluender, a. stocco, resting-state qeeg predicts rate of second language learning in adults. brain lang 157, 44-50 (2016). [21] a. godfroid, s. loewen, s. jung, j. h. park, s. gass, r. ellis, timed and untimed grammaticality judgments measure distinct types of knowledge: evidence from eye-movement patterns. studies in second language acquisition 37(2), 269-297 (2015). [22] r.c. oldfield, the assessment and analysis of handedness: the edinburgh inventory. neuropsychologia 9(1), 97-113 (1971). available at: https://www.psytoolkit.org/survey-library/handedness-ehi.html. [23] v. kuperman, m. dambacher, a. nuthmann, r. kliegl, the effect of word position on eye-movements in sentence and paragraph reading. q j exp psychol (hove) 63(9), 18381857 (2010). [24] l.a. stowe, e. kaan, l. sabourin, r.c. taylor, the sentence wrap-up dogma. cognition 176, 232-247 (2018). [25] j.a. blanco, p.r. donley, vistas: introduccion a la lengua espan̆ola, 5th ed. 2016 (vista higher learning, inc., 2016). [26] n.a. macmillan, c.d. creelman, detection theory: a user's guide (psychology press, 2004). [27] s. montrul, second language acquisition and first language loss in adult early bilinguals: exploring some differences and similarities. second lang res 21(3), 199-249 (2005). columbia undergraduate science journal vol. 15, 2021 ogunniyi et. al. 54 [28] v. marian, h.k. blumenfeld, m. kaushanskaya, the language experience and proficiency questionnaire (leap-q): assessing language profiles in bilinguals and multilinguals. j speech lang hear res 50(4), 940-967 (2007). available at: https://bilingualism.northwestern.edu/leapq/. [29] a. delorme, s. makeig, eeglab: an open source toolbox for analysis of single-trial eeg dynamics including independent component analysis. j neurosci methods 134(1), 9-21 (2004). available at: https://sccn.ucsd.edu/eeglab/index.php. [30] mathworks, inc., matla8b 8.0 and statistics toolbox 8.1, natick, massachusetts, united states (2020). available at: https:// www.mathworks.com/downloads/. [31] r core team, r: a language and environment for statistical computing. r foundation for statistical computing, vienna, austria (2020). available at: https://www.r-project.org/. [32] d. makowski, the psycho package: an efficient and publishing-oriented workflow for psychological science. journal of open source software 3(22), 470 (2018). available at: https://github.com/neuropsychology/psycho. [33] w. klimesch, eeg alpha and theta oscillations reflect cognitive and memory performance: a review and analysis. brain res brain res rev 29(2-3), 169-195 (1999). [34] n.a. herweg, e.a. solomon, m.j. kahana, theta oscillations in human memory. trends cogn sci 24(3), 208-227 (2020). [35] r. erlam, elicited imitation as a measure of l2 implicit knowledge: an empirical validation study. applied linguistics 27(3), 464–491 (2006). [36] n. spada, j.l.j. shiu, y. tomita, validating an elicited imitation task as a measure of implicit knowledge: comparisons with other validation studies. lang learn 65(3), 723-751 (2015). [37] l. pion-tonachini, k. kreutz-delgado, s. makeig, iclabel: an automated electroencephalographic independent component classifier, dataset, and website. neuroimage 198, 181-197 (2019). [38] o.m. bazanova, alpha eeg activity depends on the individual dominant rhythm frequency. j neurother 16(4), 270-284 (2012). © 2025 patel, krishay. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 31 cardiac regeneration: a promising future for tissue engineering and cardiac repair krishay patel1 university of california, san diego, la jolla, ca, usa keywords: cardiac regeneration, cardiomyocytes, cell culture, reprogramming, tissue engineering, cardiovascular disease, imaging techniques abstract: cardiovascular diseases are the leading cause of human death worldwide. these diseases have detrimental effects on the heart, damaging the tissue and disrupting healthy heart function. cardiac functioning begins to dissipate as a result of cardiac cell death and subsequent damage to the structures of the heart. cardiac regeneration is an evolving field in cardiac therapy that aims to rebuild damaged cardiac tissue that may have previously been beyond repair. understanding how cardiac cells function within the heart is crucial in studying tissue engineering and cardiac regeneration. with the loss of cardiac cells following cardiovascular disease, the ability to extract healthy cardiac cells for tissue engineering becomes greatly limited. thus, tissue cultures and cellular reprogramming become crucial methods for in vitro expansion of healthy cells. understanding regeneration in the heart, cell sources necessary for cardiac regeneration, and the existing models that have been implemented for cardiac regeneration are crucial in advancing its study and improving methods to take cardiac regeneration to clinical trials. imaging techniques to model patient-specific structures of the heart will ultimately help map damaged tissue, assisting with the entire regeneration process. investigating the limitations of these techniques is equally important to improving cardiac regeneration, with the possibility to save countless lives. introduction for many decades, cardiovascular-related diseases (cvds) have been the leading cause of death globally as well as a major contributor to disabilities.1 despite advancements in medical care that have resulted in a decline in age-standardized cvd deaths occurring globally, the total cvd global deaths have continued to increase over the last three decades, and previous progress in counteracting the mortality effects of cvds has become uneven and begun to stall.2 the risk of exposure to cvds has become increasing prevalent, with changes in the ecological and geological environment, modifications in lifestyle, and shifts in social policies contributing to cvd risk.3 as a result, there has been great concern about heart repair and treatment. cvds have been understood to irreversibly damage cardiomyocytes, the cells in the heart responsible for the contractile functions of the heart. this leads to the formation of scar tissue, ultimately damaging the structure and function of the heart.4 the adult human heart has limited regenerative capacity and so regenerative medicine therapies have shed new hopes in repairing or even replacing damaged hearts.5,6 the advancements in the field of tissue engineering, particularly concerning cell reprograming, the development of functional biomaterials, and the fabrication of biomimetic cardiac tissue, have gained traction in providing possible columbia undergraduate science journal vol. 19, 2025 patel © 2025 patel, krishay. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 32 avenues for tissue regeneration applications within the heart. biomaterial-based approaches have gained increasing attention for cardiac tissue engineering and regeneration after contributing to improved cardiac function, promoting angiogenesis, and diminishing severe immune responses following clinical trials on animals.7 nevertheless, creating fully effective functional cardiac tissue in vitro is still a challenge in the field of tissue engineering, with the regeneration of an entire human organ in vitro, specifically the heart, being difficult in terms of specific chamber orientation and electrical function.8 despite this, implementations of tissue engineering to reconstruct cardiac tissue have reported better contractile function from the heart, and cardiac progenitor cells, responsible for contributing to cardiac tissue regeneration, have been reported to give rise to cardiomyocytes in vitro and in vivo after transplantation and to enhance cardiac function post-infarction.9 this review will highlight the progress made in the field of cardiac tissue engineering regarding regenerative tissue applications on the heart, with a particular focus on cell reprogramming in the context of tissue engineering, growing these cells in various culture media, and the applications of these cells within the cardiac tissue engineering process. an introduction to the cardiac system, along with the current status on regeneration and cell sources necessary to engineer cardiac tissue, and the experimental applications of working with the heart, including imaging techniques and the fabrication of biomimetic cardiac tissue, will also be discussed. cardiac cells introduction to cells cells are fundamental building blocks within an organism’s body, present in all living organisms. they are responsible for carrying out all the vital functions, and they ultimately play a crucial role in organ function. in the process of regeneration, organisms have been shown to produce new cells when replacing an amputated, damaged structure.10 therefore, it remains crucial to produce and introduce new cells to the body when implementing regeneration after a particular body part or organ has faced severe damage. with cells being a crucial factor necessary for regeneration, it is crucial to produce the cells that predominantly make up the composition of the heart: cardiomyocytes.11 cardiomyocytes are the striated, branched cardiac muscles found in the heart that are responsible for contracting together synchronously to enable the heart to work as a pump.12 following cardiac injury, a substantial number of cardiomyocytes are lost, and the heart forms a fibrotic scar to serve as patch where the cardiac muscle cells previously resided. however, the scar tissue does not maintain the ability to effectively contract.13 in the event of a heart disease, one also observes cardiomyocyte cell death.14 therefore, combating the effects of cvds and enhancing the possibilities of cardiac tissue regeneration can be made a possibility by exploring areas of cell reprogramming concerning cardiomyocytes, understanding how to culture these cells so that they can be reapplied within the human heart, and uncovering other cell sources necessary for cardiac tissue engineering. columbia undergraduate science journal vol. 19, 2025 patel © 2025 patel, krishay. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 33 cell culture cells occupy a crucial role in the functioning of organs, and damage to organs, such as cvds, often results in the cells within that organ being damaged. as a result, the understanding and induction of cells to be introduced into the body becomes extremely important, particularly when discussing regeneration. however, to understand the cells within the body that are necessary for regenerative purposes and develop techniques to increase the number of required cells, these cells need to be extracted in vitro and grown outside the body. this is achieved through cell cultures. cell cultures are laboratory methods that enable the growth of eukaryotic or prokaryotic cells in physiological conditions. they enable the study of human health and diseases through the extraction of infected human cells, which can then be examined in a sterile, cultured environment. utilizing the practice of cell cultures working with the homogeneity of clonal cell populations, specific cell types, and well-defined culture systems removes any interfering genetic or environmental confounding variables, thereby allowing data generation with high reproducibility, which may not otherwise be consistent when examining and working with an entire organ.15 this then enables a possible avenue for inducing a high number of cardiomyocytes required for cardiac regeneration. among the various culture techniques available, two well-established models are 3d and 2d culture systems. 2d systems are the traditional monolayer cultures that are straightforward, cost-effective, and convenient for observation. however, 2d cultures fall short in simulating the in vivo morphology of the extra cellular matrix (ecm), and cells in these 2d cultures are compelled to adapt to their complex biological behaviors. these complications have resulted in a turn to 3d systems that better mimic the physiological context of the ecm.16 3d culture systems are more complex systems than 2d culture systems, as they have been developed to mimic in vivo conditions, and therefore provide a good alternative for the in vitro imitation of the human heart tissue. generally divided into scaffold-based and scaffold-free, 3d heart tissue models allow cells to organize themselves into a 3d structure resembling the human myocardial cell organization. a major advantage of 3d cultures is that their more accurate imitation of the internal human environment allows researchers to better replicate in vitro cardiac tissue morphological, biochemical, and mechanical features.17 in the cardiovascular field, 3d culture systems are further differentiated into two categories; they either contain a scaffold matrix, generally a hydrogel, mixed with and populated by cells to form a strip shaped micro-tissue between attachment sites, or they are smaller cellular spheroids formed by self-assembly without scaffold proteins. an increasing number of studies have used a mix or co-culture several cell types, such as rodent and human induced pluripotent stem cell-derived cardiomyocytes.18 in fact, larger tissue formats like multilayered cell sheets, recellularized hearts, or large biomaterial patches, which are too expensive and slow in manufacturing for drug testing, are developed for regenerative medicine, leading to advancements in issue and cardiac regeneration.19 columbia undergraduate science journal vol. 19, 2025 patel © 2025 patel, krishay. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 34 cell reprogramming cellular reprogramming is the process by which cells can be reliably changed from one tissue type to another, enabling novel approaches to more deeply investigate the fundamental basis of cell identity.20 the process of cellular reprogramming is not simply limited to a technique, but rather an application that can be used within the human body for regenerative purposes. the importance of maintaining high quantities of cardiomyocytes following cvds remains a problem, and the issues extends to the fact that cardiomyocytes have little capability of proliferation after birth.21 through the process of trans differentiation, the conversion of one cell to another, cells that are abundant in the body can be reprogrammed into desired cell phenotypes to restore tissue function in damaged areas.22 this process presents a possible avenue for further investigation into reprogramming the abundant cells in the body into the necessary cardiomyocytes, which may be scarce. several approaches for cellular reprogramming concerning the induction of cardiomyocytes have been investigated, and have proved successful in cardiomyocyte production, indicating a promising avenue towards cardiac tissue regeneration. firstly, through combinatorial expression of two cardiac transcription factors, gata4 and tbx5, and a cardiac-specific subunit baf chromatin-remodeling complex, baf60c, the direct differentiation and conversion of mouse mesoderm into beating cardiomyocytes can occur.23 secondly, following experimentation using a combinatorial strategy, it was identified that a combination of micrornas was capable of inducing direct cellular reprogramming of fibroblasts into cardiomyocyte-like cells in vitro.24 thirdly, by treating human fibroblasts with a combination of nine compounds, termed 9c, cardiomyocyte-like cells can be generated that uniformly contract and resemble human cardiomyocytes in their epigenetic properties.25 lastly, the introduction of three cardiogenic transcription factors: gata4, mef2c, and tbx5 (gmt) can induce the direct reprogramming of cardiac fibroblasts into cardiomyocytes, and this discovery of the direct cardiac reprogramming factor gmt has also enabled the induction of cardiomyocytes inside the body.26 the heart and its regenerative capabilities introducing the system of the heart the cardiovascular system is a crucial system within the body, composed of the heart, blood vessels, and blood. the cardiovascular system is designed to ensure the continuous survival of all cells within the body, which it achieves by maintaining the intermediate chemical environment of each cell in the body.27 the heart remains at the center of the cardiovascular system, responsible for pumping blood through the pulmonary and systemic systems of the body. the cardiovascular system, in turn, serves as a body-wide network of vessels that transports nutrients, respiratory gases, metabolic waste, and hormones, distributes and dissipates heat, and assists in defending the body against disease.28 the cardiovascular system, and ultimately the heart, plays a crucial role in the healthy function of the human body. adult hearts fail to regenerate following injury and cvds, and this failure to regenerate myocardium is a crucial contributor to heart failure and ultimately death worldwide.29 columbia undergraduate science journal vol. 19, 2025 patel © 2025 patel, krishay. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 35 therefore, complications that occur within the cardiovascular system and affect the heart ultimately have a detrimental effect on the rest of the body, making the applications of regenerative medicine surrounding cardiac tissue more pressing. regeneration within the heart currently, experimentation in heart regeneration has been restricted to mouse model hearts. in a particular experiment, fetal rodent hearts that had undergone a specific genetic ablation were able to restore around 50% of the lost cardiomyocyte mass indicating that the embryonic environment and the transcriptional state of embryonic cardiomyocytes facilitate cardiomyocyte cell cycle re-entry and repopulation of the heart, an important discovery in understanding the regenerative potential from the embryo and the change thereafter following the first few weeks of life.29 the field of heart regeneration has entered a period of extraordinary progress in the understanding of endogenous heart regeneration, stem cell differentiation for exogenous cell therapy, and cell-delivery methods. however, the regenerative capacity expressed by adult individuals, human in particular, is not entirely favorable to generate cardiac tissue necessary following cvds and injury. although the heart retains the ability to regenerate in some lower vertebrates, adult mammals do not retain the capacity to replenish the heart with enough cardiomyocytes to restore function following injury. therefore, the concept of heart and cardiac regeneration does not refer to a single area of study but rather encompasses the advancements and novel discoveries in fields including native cardiomyocyte renewal, cell therapy, direct reprogramming, and tissue engineering approaches to rebuild damaged heart tissue.30 multiple approaches have been explored to promote and possibly achieve regeneration within the heart. firstly, there are cell-free therapies for cardiac repair, where recombinant dna, rna-based, or protein therapeutics have been used in regenerative medicine. in myocardial infarction (mi) models, the injection of particular cardioactive growth factors, such as neuregulin1, has induced sustained improvement in myocardial function.31 furthermore, the utilization of the cardiomyogenic factor follistatin like 1 can stimulate the recovery of contractile function and limit fibrosis following mi injury.32 secondly, cell-based therapies have been proposed as a promising approach for treating advanced heart failure and repairing damaged myocardial tissue. thirdly, adult stem cell transplantations have been a possible avenue for cardiac regeneration. experimental evidence suggests that adult stem cells, such as bone marrow cells, bone marrow purified hematopoietic stem cells, and bone marrow purified mesenchymal stem cells, can differentiate into cardiomyocytes. lastly, pluripotent stem cell-based therapies have opened another path for cardiac regenerative possibilities. specifically, it has been understood that human embryonic stem cells can differentiate into multiple cell types and thus have great therapeutic potential in regenerative medicine.33 cell sources cell sources for cardiac tissue engineering and regeneration refer to the variety of cell types that can be used in cardiac tissue engineering constructs. these cell types include cardiomyocytes, often derived from columbia undergraduate science journal vol. 19, 2025 patel © 2025 patel, krishay. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 36 induced pluripotent stem cells or isolated rodent hearts, cardiovascular cells, fibroblasts, various progenitor cells, stem cells, and spheroids.34 embryonic stem cells, as a type of stem cell, can be collected from the inner mass cells of blastocytes and can differentiate into cell types of all germ layers.35 through experimentation, cardiointrusive cues were used to differentiate embryonic stem cells to cardiac progenitor cells and embed them in a fibrin patch, which in turn improved cardiac function. induced pluripotent stem cells are widely used in regenerative medicine, and they are often obtained from a patient’s somatic cells. these cells are similar to embryonic stem cells and have been used to regenerate the heart in a variety of animal studies. cardiac stem cells, which can be acquired from the heart, are extremely useful for cardiac regenerative purposes, as they have the potential to self-renew and can differentiate into different cell types, such as vascular smooth muscle cells, cardiomyocytes, and endothelial cells.36 lastly, cardiac progenitor cells are defined by having self-renewing, clonogenic properties, as well as multipotent differentiation capabilities both in vivo and in vitro, which are desirable for promoting cardiac regeneration.37 at the core of the heart, however, lies cardiomyocyte-rich tissue to allow for rhythmic pump contractions. these cells appear to be the most prevalent cardiac cell population in the heart, and the need to acquire these cells is extremely important for regenerative applications to succeed.6 discovering sources to acquire cardiomyocyte cells is crucial to engineer cardiac tissue, and although there are existing limitations around human tissue availability, human adult cardiomyocytes can be isolated during myocardial biopsies.38 however, this process needs to be done at quickly because following isolation, cardiomyocytes undergo a profound structural and functional remodeling, leading to cell differentiation and loss of viability.39 regenerative models and methods existing models there has been increasing demand for new therapeutic models, and such models are necessary to advance the understanding of the biological mechanisms of cardiac tissue regeneration, and test therapeutic approaches to regenerate tissue and restore cardiac function. in vitro model systems allow researchers to study biological systems while still maintaining a high level of control over the experimental parameters.40 such in vitro models include single cell models, 2d culture models, 3d culture models, coculture models, microfluidic system models, tissue models, whole heart models, and lastly cell sources as discussed in the previous section. examining the electromechanical properties of single cardiomyocytes has been crucial to the understanding of cardiac physiology. the electrophysiology of a single cardiomyocyte can be evaluated using the patch clamp technique, which can characterize the phenotype of cms differentiated from embryonic stem cells.41,42 cardiomyocytes grown in 2d culture systems can elucidate molecular signaling pathways, assess the cardiotoxicity of a particular drug, and evaluate gene therapy approaches.43,44,45 3d models have been used as scaffolds to mimic the natural extracellular matrix, making the models extremely close to human myocardium, serving as models of healthy columbia undergraduate science journal vol. 19, 2025 patel © 2025 patel, krishay. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 37 human heart physiology.40 cocultures systems enable a variety of growing cell types to be cultured together, which in turn stimulates the cellular communication that would normally occur between cells of different types.46 the coculture of human embryonic stem cells with human umbilical-vein endothelial cells and mouse fibroblasts resulted in the formation of vascular networks resembling capillaries, and the coculture enhanced cardiomyocyte proliferation.47 microfluidic techniques make use of soft lithography techniques and have been constructed to study the electrophysiology of single cardiomyocytes in microfabricated patchclamp devices or cell-to-cell signaling between pairs of cardiomyocytes.48,49 these systems also make use of microfabrication techniques that provide advanced structure from well-defined scaffolds, which can improve drug discovery and better mimic heart physiology. lastly, examining the whole heart of an organism outside its body allows for the evaluation of cardiac function ex vivo. in vivo model systems provide a more accurate understanding of the complex behaviors of the heart within a living organism. in vivo model systems include noncardiac regenerative models and cardiac regenerative models. noncardiac animal models, and noncardiac mammalian regeneration, can provided possible clues to cardiac regenerative biology using methods that are more practical because of lower cost and fewer technical and logistical challenges, and then can provide insight into potential mechanisms of cardiac regeneration. cardiac regenerative models, on the other hand, focus on examining small and large animal mammalian models that have been studied extensively with the intention of discovering therapeutic interventions to effectively regenerate the heart and cardiac tissue following injury and cvds.40 advanced optical imaging techniques optical imaging uses light and special properties of photons to obtain detailed images of organs, tissue, cells, and possibly even molecules.50 optical imaging is a noninvasive imaging technique that functions based on the absorption, scattering, and fluorescence properties of the incident particle. optical imaging provides depthresolved, high-resolution images of tissue microstructures in real time by measuring the interference between the light backscattered from the sample tissue and the light reflected from a reference. the use of optical imaging techniques can help understand a patient’s heart structure, which can help prepare for procedures by identifying structures to avoid and locating areas for treatment and repair. the imaging plays a crucial role in early diagnosis of cvds, as well as in monitoring and providing guidance for cardiac procedures. due to the high resolution, fast imaging speeds, and non-invasive procedure obtained from optical imaging, this technique has seen increased clinical use over the past few decades.51 molecular imaging, in particular optical imaging, has advanced in a way that allows for the characterization of biological processes at the cellular and subcellular levels, which could ultimately play a crucial role in the examination of cardiac tissue to understand their biological processes and whether tissue may be damaged or not.52 current methods of therapy for cardiac regeneration, including stem-cell therapy, columbia undergraduate science journal vol. 19, 2025 patel © 2025 patel, krishay. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 38 have been promising as discussed, but the challenges of such therapies are dependent on cell survival of the introduced cells, possible implantation difficulties due to adverse mechanical forces, and possible hypoxia however, the use of advanced optical imaging techniques has emerged as a novel approach to track engraftment, the formation of healthy blood cells following a transplant, as well as track the survival and rejection of transplanted tissue, which is crucial in understanding the efficacy of procedures conducted for cardiac regeneration, making this technique very valuable.53,54 emerging models to constructure cardiac tissue the majority of 3d cardiac models, until now, have been fabricated to be flat or small structures, which can limit volumetric expansion and functional maturation. the need to establish a model that could overcome these limitations of scale, maturity, and functionality of human cardiac tissues has been fabricated by integrating native heart-like cellular and extracellular components and dynamic flow. this new 3d human cardiac tissue model was successfully validated for various biomedical applications, including cardiotoxicity prediction, cardiac disease modeling, and regenerative therapy.55 together, this 3d tissue model is extremely beneficial in imaging the heart and its cardiac tissue with respect to identity and predicting the presence of cardiac disease, as well as applying this model for cardiac regeneration. with the current shortage of heart donors available and the urgent need for new sources of cardiac regeneration medicine, tissue engineering using bioactive materials and 3d bioprinting to create microscale cardiac tissue has shown great advantages and promise in the field of cardiac regenerative therapy. 3d bioprinting focuses on the tight fitting of a sufficient number of functional cardiomyocytes with a feasible scaffold on a small scale.56 3d bioprinting has ultimately emerged as an approach for in vitro generation of functional cardiac tissue for drug screening and cardiac regenerative therapy. the use of 3d bioprinted cardiac patches, fabricated by human coronary artery endothelial cells, collagen microprinting, and an alginate matrix, resulted in high levels of cell proliferation and differentiation, which are crucial for regeneration of cardiac tissue. 3d bioprinting is a layer-by-layer additive technology that precisely deposits biomaterials and active cells in accordance with a certain spatial pattern that has high resolution simulation of the heart.57 using a combination of different bioprinting approaches, a variety of cardiac constructs, including cartilage and a heart valve, have been successfully constructed, which is crucial to repair damaged tissue structures within the heart. experiments show that bioprinted cardiac tissue possess cardiomyocytes orientated in an organized fashion and contracting capabilities, both consist with native heart tissue characteristics.57,58 fabrication of biomimetic cardiac tissues inducing the maturation of cardiomyocytes remains a problem in the field of cardiac tissue engineering and cardiac regeneration. specifically, addressing this issue within the complexity of the in vivo conditions of the human heart remains an important barrier in cardiac tissue engineering. a variety of columbia undergraduate science journal vol. 19, 2025 patel © 2025 patel, krishay. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 39 methods have been tested to grow biomimetic tissue in the laboratory that can adjust to the complexities within the body. these methods include the hydrogel method, decellularized bio scaffolds, microfabrication, and bioprinting as discussed in the previous section. hydrogels are widely used for their ability to be molded into several geometries and their scalability. the cardiac tissue can be fabricated by joint gelatin of cells and a polymer or by having a casting mold to confine the cells in a particular organization, which can mimic cell organization found in native cardiac tissue by freeze-drying the hydrogel.6 decellularized bio scaffolds aim to generate scaffolds with a native extracellular matrix ultrastructure and composition while removing all cells and genetic material present in the native tissue. the most immediate application of decellularized myocardial sheets is scaffold recellularization; however, the recellularization of such sheets with cardiomyocytes, which can be successfully achieved, does not translate to the goal of recellularizing human hearts, which requires a vast number of cardiomyocytes for which there is no current sustainable production method.6 the use of biomaterials with built-in electroconductive characteristics are used to mimic the electrical conductivity of the native heart tissue, and scaffolds infused with electroconductive elements create a conductive network across the constructed tissue, which imitate electrical conductivity seen in the heart.59 gelatin-based scaffolds have also gained attention in recent years with their great bio-affinity that encourages the regeneration of tissue, and how by strengthening such gelatin-based scaffolds, the natural tissue can be accurately mimicked. gelatin-based scaffolds cansimplify the inclusion of cells and growth factors for optimal scaffold construction, and advanced production strategies for these scaffolds create highly configurable scaffold geometric shapes needed for specific medical necessities and to fabricate patient-specific implants.60 advances in microfabrication have allowed for the detailed engineering of material features to resemble the in vivo conditions of an organ, particularly the heart. simpler systems use microfabrication to pattern material substrates, to ease cardiomyocyte spreading and sarcomere axial alignment, and following experimentation, a substantial improvement in cardiomyocyte alignment was shown. microfabrication methodologies can ultimately be harnessed to create complex tissues like the native myocardium, which is crucial for regenerative cardiac tissue to adjust to the in vivo construct and conditions of the native cardiac tissue.6 limitations of imaging methods for the heart the rapid and complex changes that occur from heart morphology make imaging the heart during development a challenging task. confocal microscopy, an advancement of fluorescence microscopy, offers higher spatial resolution and the ability to visualize subcellular details; however, this imaging technique is relatively slow and harsh on the sample, potentially causing phototoxicity and bleaching.61,62 a major challenge in biomimetic cardiac tissue engineering is the scalability of the fabrication technique and the translation of these approaches to clinical columbia undergraduate science journal vol. 19, 2025 patel © 2025 patel, krishay. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 40 applications. the issues that remain around biomimetic cardiac tissue engineering include standardized production methods, optimizing functional and structural maturation, and ensuring long-term stability of synthetic cardiac tissue.59 these challenges, ultimately, require significant attention to move forward with advancements in this field. the advancements in 3d bioprinting have revolutionized potential avenues in cardiac treatment in the future, with the opportunity to fabricate microscale cardiac tissue for potential in vivo applications. however, limitations occur with the use of artificial prints, compared to the original hearts of large mammals, as these prints remain naïve in their generation of both the input and output mechanical strengths long-term. furthermore, because of the complexity of heart movement, whether these 3d bioengineered bodies are able to achieve or even get close to perfect cardiac replication physically and chemically remains unknown.56 there are a vast number of imaging techniques to examine the cardiac system, and although there are specific advantages for each imaging technique that make these applicable in certain situations, each of them has their own limitations. these imaging techniques include radionuclide imaging, mri, ultrasound, ct scan, optical imaging, and magnetic particle imaging (mpi). radionuclide imaging has the limitations of radioactive exposure, high cost of operation, poor spatial resolution, and a short imaging time. mris are also expensive and require heavy involvement to operate. ultrasounds have low tissue contrast, and they are dependent on an operator. ct scans have ionizing radiation which is extremely dangerous, and they offer limited soft tissue discrimination. optical imaging offers poor penetration depth, poor spatial resolution, and it is only used in preclinical trials. lastly, mpis are unable to discriminate live cells from dead cells. conclusion cvds have had a fatal impact on the human population. thus, studying heart and cardiac tissue repair has become a forefront issue to combat the effects of cvds. the effects of cvds result in damage of cardiac tissue and the death of cardiac cells. the heart soon experiences degraded function, triggering other dysfunctions within the body that ultimately result in death. therefore, exploring human regenerative capacity and the potential application of heart and cardiac regeneration has evolved as a new therapeutic to engineering functional cardiac tissue following cvds. the potential of cardiac regeneration is very promising for cardiac repair, and the progress made in the field through tissue engineering has shown the beneficial outcomes that can be achieved through cardiac regeneration. the focus on the production of cardiac cells following cvds is crucial in restoring appropriate cardiac function. cells form the basis of living tissue, making the induction of cells important for regenerative purposes, especially cardiomyocytes which once introduced into the body are important for cardiac function. through various applications of cell culture, specifically 2d and 3d cultures, these cardiac cells can be effectively understood and examined. growing these cells in an artificial environment is extremely important in the context of cardiac columbia undergraduate science journal vol. 19, 2025 patel © 2025 patel, krishay. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 41 regeneration, as more cardiac cells can be produced and introduced into the human body as a supplement to compensate for the loss of cells following cell death. cellular reprogramming also plays a crucial role in cardiac regeneration, by allowing cells to be changed to another cell type which could assist the process of cardiac regeneration by differentiating greater populated cells into cardiomyocytes which are less abundant in the body. the field of cardiac regeneration actively growing, with a variety of approaches being explored to help promote and achieve regeneration of the heart. cells sources have also been of extreme importance in engineering cardiac tissue to contribute towards regeneration of the heart. cells sources for cardiac tissue engineering refer to the variety of cell types that can be used in cardiac tissue engineering constructs, including cardiomyocytes, induced pluripotent stem cells or isolated rodent hearts, cardia vascular cells and fibroblasts, various progenitor and stem cells and spheroids. discovering and utilizing various cells sources to acquire cardiomyocyte cells is crucial to engineer cardiac tissue, as these cells appear to be the most prevalent cardiac cell population in the heart, and the need to acquire these cells is extremely important for regenerative applications to succeed.6 various existing models have helped propel the concept of cardiac regeneration to become a forefront technique for therapeutics in the future to treat damaged cardiac tissue. the use of a variety of imaging techniques has been crucial in allowing internal areas of the cardiac system and the heart, as well as problems with a particular person’s heart, to be examined and understood externally. this is crucial in cardiac regeneration, in particular, as particular constructs of cardiac tissue could then be induced after understanding a person’s current heart state and the areas that are damaged. the utilization of imaging techniques also showcases the efficacy of the regenerative tissue and its ability to work within the body, which is useful when evaluating the use of the construct and understanding how it contribute to regeneration through a non-invasive procedure. however, the use of imaging techniques, and the development of models relating to cardiac regeneration have their inherent limitations, serving as a setback for the progress of cardiac regeneration. the need to address these issues is crucial in developing therapeutics and constructs that can advance cardiac regeneration and better suit their applications for human use. possible directions for future myocardial regeneration include the consolidation of transplantation of cells with “true” regenerative potential, tissue engineering with various scaffolds and cell types, a stimulation of resident cell sources by cytokines or growth factors, and a direct reprogramming of scar tissue by delivery of various transcription factors or mirnas63. cardiac regeneration is shaping to be a promising field in the future to treat damaged cardiac tissue following cvds. the applications of cardiac regeneration are endless, and the hope generated by the possibility of cardiac regeneration is likely to be carried into the future. the process of cardiac regeneration could ultimately columbia undergraduate science journal vol. 19, 2025 patel © 2025 patel, krishay. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 42 revolutionize cardiac tissue repair and treatment, and the promise of cardiac regeneration could be the step needed to save countless lives. author information corresponding author krishay patel – krishay.in.sa@gmail.com author contributions krishay patel provided full contribution to the research and writing of the manuscript. the manuscript was drafted through the collegiate mentorship program with the help of dr rosalyn abbott (rabbott@andrew.cmu.edu) references [1] kauffman, d. (n.d.). cardiovascular disease burden, deaths are rising around the world. american college of cardiology. https://www.acc.org/aboutacc/pressreleases/2020/12/09/18/30/http%3a %2f%2fwww.acc.org%2faboutacc%2fpressreleases%2f2020%2f12%2f09%2f18 %2f30%2fcvd-burden-and-deathsrising-around-the-world [2] di cesare, m., bixby, h., gaziano, t., hadeed, l., kabudula, c., vaca mcghie, d., mwangi, j., pervan, b., perel, p., piñeiro, d., taylor, s., & pinto, f. (n.d.). world heart report 2023 confronting the worlds number one killer. world heart federation. https://world-heart-federation.org/wpcontent/uploads/world-heart-report2023.pdf [3] bhatnagar, a. (2017). environmental determinants of cardiovascular disease. circulation research, 121(2), 162–180. https://doi.org/10.1161/circresaha.1 17.306458 [4] hotkar, a. j., & balinsky, w. (2012). stem cells in the treatment of cardiovascular disease—an overview. stem cell reviews and reports, 8(2), 494–502. https://doi.org/10.1007/s12015-0119302-2 [5] bergmann, o. et al. evidence for cardiomyocyte renewal in humans. science 324, 98–102 (2009). [6] tenreiro, m. f., louro, a. f., alves, p. m., & serra, m. (2021). next generation of heart regenerative therapies: progress and promise of cardiac tissue engineering. npj regenerative medicine, 6(1), 1–17. https://doi.org/10.1038/s41536-02100140-4 [7] scafa udriște, a., niculescu, a.-g., iliuță, l., bajeu, t., georgescu, a., grumezescu, a. m., & bădilă, e. (2023). progress in biomaterials for cardiac tissue engineering and regeneration. polymers, 15(5), 1177. https://doi.org/10.3390/polym15051177 https://www.acc.org/about-acc/press-releases/2020/12/09/18/30/http%3a%2f%2fwww.acc.org%2fabout-acc%2fpress-releases%2f2020%2f12%2f09%2f18%2f30%2fcvd-burden-and-deaths-rising-around-the-world https://www.acc.org/about-acc/press-releases/2020/12/09/18/30/http%3a%2f%2fwww.acc.org%2fabout-acc%2fpress-releases%2f2020%2f12%2f09%2f18%2f30%2fcvd-burden-and-deaths-rising-around-the-world https://www.acc.org/about-acc/press-releases/2020/12/09/18/30/http%3a%2f%2fwww.acc.org%2fabout-acc%2fpress-releases%2f2020%2f12%2f09%2f18%2f30%2fcvd-burden-and-deaths-rising-around-the-world https://www.acc.org/about-acc/press-releases/2020/12/09/18/30/http%3a%2f%2fwww.acc.org%2fabout-acc%2fpress-releases%2f2020%2f12%2f09%2f18%2f30%2fcvd-burden-and-deaths-rising-around-the-world https://www.acc.org/about-acc/press-releases/2020/12/09/18/30/http%3a%2f%2fwww.acc.org%2fabout-acc%2fpress-releases%2f2020%2f12%2f09%2f18%2f30%2fcvd-burden-and-deaths-rising-around-the-world https://www.acc.org/about-acc/press-releases/2020/12/09/18/30/http%3a%2f%2fwww.acc.org%2fabout-acc%2fpress-releases%2f2020%2f12%2f09%2f18%2f30%2fcvd-burden-and-deaths-rising-around-the-world https://www.acc.org/about-acc/press-releases/2020/12/09/18/30/http%3a%2f%2fwww.acc.org%2fabout-acc%2fpress-releases%2f2020%2f12%2f09%2f18%2f30%2fcvd-burden-and-deaths-rising-around-the-world https://www.acc.org/about-acc/press-releases/2020/12/09/18/30/http%3a%2f%2fwww.acc.org%2fabout-acc%2fpress-releases%2f2020%2f12%2f09%2f18%2f30%2fcvd-burden-and-deaths-rising-around-the-world https://world-heart-federation.org/wp-content/uploads/world-heart-report-2023.pdf https://world-heart-federation.org/wp-content/uploads/world-heart-report-2023.pdf https://world-heart-federation.org/wp-content/uploads/world-heart-report-2023.pdf https://doi.org/10.1161/circresaha.117.306458 https://doi.org/10.1161/circresaha.117.306458 https://doi.org/10.1007/s12015-011-9302-2 https://doi.org/10.1007/s12015-011-9302-2 https://doi.org/10.1038/s41536-021-00140-4 https://doi.org/10.1038/s41536-021-00140-4 https://doi.org/10.3390/polym15051177 columbia undergraduate science journal vol. 19, 2025 patel © 2025 patel, krishay. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 43 [8] zhao, y., rafatian, n., wang, e. y., wu, q., lai, b. f. l., lu, r. x., savoji, h., & radisic, m. (2020). towards chamber specific heart-on-a-chip for drug testing applications. advanced drug delivery reviews, 165–166, 60. https://doi.org/10.1016/j.addr.2019.12.0 02 [9] laflamme, m. a., & murry, c. e. (2011). heart regeneration. nature, 473(7347), 326–335. https://doi.org/10.1038/nature10147 [10] king, r. s., & newmark, p. a. (2012). the cell biology of regeneration. the journal of cell biology, 196(5), 553– 562. https://doi.org/10.1083/jcb.201105099 [11] arackal, a., & alsayouri, k. (2024). histology, heart. in statpearls. statpearls publishing. http://www.ncbi.nlm.nih.gov/books/nbk 545143/ [12] ripa, r., george, t., shumway, k. r., & sattar, y. (2024). physiology, cardiac muscle. in statpearls. statpearls publishing. http://www.ncbi.nlm.nih.gov/books/nbk 572070/ [13] patel, p., & karch, j. (2020). chapter four—regulation of cell death in the cardiovascular system. in j. k. e. spetz & l. galluzzi (eds.), international review of cell and molecular biology (vol. 353, pp. 153–209). academic press. https://doi.org/10.1016/bs.ircmb.2019.1 1.005 [14] chiong, m., wang, z. v., pedrozo, z., cao, d. j., troncoso, r., ibacache, m., criollo, a., nemchenko, a., hill, j. a., & lavandero, s. (2011). cardiomyocyte death: mechanisms and translational implications. cell death & disease, 2(12), e244. https://doi.org/10.1038/cddis.2011.130 [15] segeritz, c.-p., & vallier, l. (2017). cell culture. basic science methods for clinical researchers, 151–172. https://doi.org/10.1016/b978-0-12803077-6.00009-6 [16] sun, s., liu, y., gao, h., guan, w., zhao, y., & li, g. (2024). cell culture on suspended fiber for tissue regeneration: a review. international journal of biological macromolecules, 268, 131827. https://doi.org/10.1016/j.ijbiomac.2024.1 31827 [17] gisone, i., cecchettini, a., ceccherini, e., persiani, e., morales, m. a., & vozzi, f. (2022). cardiac tissue engineering: multiple approaches and potential applications. frontiers in bioengineering and biotechnology, 10, 980393. https://doi.org/10.3389/fbioe.2022.9803 93 [18] zuppinger, c. (2019). 3d cardiac cell culture: a critical review of current technologies and applications. frontiers in cardiovascular medicine, 6, 87. https://doi.org/10.3389/fcvm.2019.0008 7 https://doi.org/10.1016/j.addr.2019.12.002 https://doi.org/10.1016/j.addr.2019.12.002 https://doi.org/10.1038/nature10147 https://doi.org/10.1083/jcb.201105099 http://www.ncbi.nlm.nih.gov/books/nbk545143/ http://www.ncbi.nlm.nih.gov/books/nbk545143/ http://www.ncbi.nlm.nih.gov/books/nbk572070/ http://www.ncbi.nlm.nih.gov/books/nbk572070/ https://doi.org/10.1016/bs.ircmb.2019.11.005 https://doi.org/10.1016/bs.ircmb.2019.11.005 https://doi.org/10.1038/cddis.2011.130 https://doi.org/10.1016/b978-0-12-803077-6.00009-6 https://doi.org/10.1016/b978-0-12-803077-6.00009-6 https://doi.org/10.1016/j.ijbiomac.2024.131827 https://doi.org/10.1016/j.ijbiomac.2024.131827 https://doi.org/10.3389/fbioe.2022.980393 https://doi.org/10.3389/fbioe.2022.980393 https://doi.org/10.3389/fcvm.2019.00087 https://doi.org/10.3389/fcvm.2019.00087 columbia undergraduate science journal vol. 19, 2025 patel © 2025 patel, krishay. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 44 [19] jackman, c. p., ganapathi, a. m., asfour, h., qian, y., allen, b. w., li, y., & bursac, n. (2018). engineered cardiac tissue patch maintains structural and electrical properties after epicardial implantation. biomaterials, 159, 48–58. https://doi.org/10.1016/j.biomaterials.2 018.01.002 [20] wilmut, i., sullivan, g., & chambers, i. (2011). the evolving biology of cell reprogramming. philosophical transactions of the royal society b: biological sciences, 366(1575), 2183– 2197. https://doi.org/10.1098/rstb.2011.0051 [21] kuang, j., huang, t., & pei, d. (2022). the art of reprogramming for regenerative medicine. frontiers in cell and developmental biology, 10, 927555. https://doi.org/10.3389/fcell.2022.9275 55 [22] grath, a., & dai, g. (2019). direct cell reprogramming for tissue engineering and regenerative medicine. journal of biological engineering, 13(1), 14. https://doi.org/10.1186/s13036-0190144-9 [23] takeuchi, j. k., & bruneau, b. g. (2009). directed transdifferentiation of mouse mesoderm to heart tissue by defined factors. nature, 459(7247), 708–711. https://doi.org/10.1038/nature08039 [24] jayawardena, t. m., egemnazarov, b., finch, e. a., zhang, l., payne, j. a., pandya, k., zhang, z., rosenberg, p., mirotsou, m., & dzau, v. j. (2012). microrna-mediated in vitro and in vivo direct reprogramming of cardiac fibroblasts to cardiomyocytes. circulation research, 110(11), 1465– 1473. https://doi.org/10.1161/circresaha.1 12.269035 [25] cao, n., huang, y., zheng, j., spencer, c. i., zhang, y., fu, j.-d., nie, b., xie, m., zhang, m., wang, h., ma, t., xu, t., shi, g., srivastava, d., & ding, s. (2016). conversion of human fibroblasts into functional cardiomyocytes by small molecules. science (new york, n.y.), 352(6290), 1216–1220. https://doi.org/10.1126/science.aaf1502 [26] yamakawa, h., & ieda, m. (2021). cardiac regeneration by direct reprogramming in this decade and beyond. inflammation and regeneration, 41, 20. https://doi.org/10.1186/s41232-02100168-5 [27] pittman, r. n. (2011). the circulatory system and oxygen transport. in regulation of tissue oxygenation. morgan & claypool life sciences. https://www.ncbi.nlm.nih.gov/books/nb k54112/ [28] stephenson, a., adams, j. w., & vaccarezza, m. (2017). the vertebrate heart: an evolutionary perspective. journal of anatomy, 231(6), 787–797. https://doi.org/10.1111/joa.12687 https://doi.org/10.1016/j.biomaterials.2018.01.002 https://doi.org/10.1016/j.biomaterials.2018.01.002 https://doi.org/10.1098/rstb.2011.0051 https://doi.org/10.3389/fcell.2022.927555 https://doi.org/10.3389/fcell.2022.927555 https://doi.org/10.1186/s13036-019-0144-9 https://doi.org/10.1186/s13036-019-0144-9 https://doi.org/10.1038/nature08039 https://doi.org/10.1161/circresaha.112.269035 https://doi.org/10.1161/circresaha.112.269035 https://doi.org/10.1126/science.aaf1502 https://doi.org/10.1186/s41232-021-00168-5 https://doi.org/10.1186/s41232-021-00168-5 https://www.ncbi.nlm.nih.gov/books/nbk54112/ https://www.ncbi.nlm.nih.gov/books/nbk54112/ https://doi.org/10.1111/joa.12687 columbia undergraduate science journal vol. 19, 2025 patel © 2025 patel, krishay. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 45 [29] uygur, a., & lee, r. t. (2016). mechanisms of cardiac regeneration. developmental cell, 36(4), 362–374. https://doi.org/10.1016/j.devcel.2016.0 1.018 [30] garbern, j. c., & lee, r. t. (2022). heart regeneration: 20 years of progress and renewed optimism. developmental cell, 57(4), 424–439. https://doi.org/10.1016/j.devcel.2022.0 1.012 [31] bersell, k., arab, s., haring, b., & kühn, b. (2009). neuregulin1/erbb4 signaling induces cardiomyocyte proliferation and repair of heart injury. cell, 138(2), 257–270. https://doi.org/10.1016/j.cell.2009.04.0 60 [32] wei, k., serpooshan, v., hurtado, c., diez-cuñado, m., zhao, m., maruyama, s., zhu, w., fajardo, g., noseda, m., nakamura, k., tian, x., liu, q., wang, a., matsuura, y., bushway, p., cai, w., savchenko, a., mahmoudi, m., schneider, m. d., … ruiz-lozano, p. (2015). epicardial fstl1 reconstitution regenerates the adult mammalian heart. nature, 525(7570), 479–485. https://doi.org/10.1038/nature15372 [33] wang, j., an, m., haubner, b. j., & penninger, j. m. (2023). cardiac regeneration: options for repairing the injured heart. frontiers in cardiovascular medicine, 9, 981982. https://doi.org/10.3389/fcvm.2022.9819 82 [34] cell sources for cardiac tissue engineering. a variety of cell types... (n.d.). researchgate https://www.researchgate.net/figure/cell -sources-for-cardiac-tissue-engineeringa-variety-of-cell-types-can-be-usedin_fig2_349634500 [35] müller, p., lemcke, h., & david, r. (2018). stem cell therapy in heart diseases—cell types, mechanisms and improvement strategies. cellular physiology and biochemistry: international journal of experimental cellular physiology, biochemistry, and pharmacology, 48(6), 2607–2655. https://doi.org/10.1159/000492704 [36] boroumand, s., haeri, a., nazeri, n., & rabbani, s. (2021). review insights in cardiac tissue engineering: cells, scaffolds, and pharmacological agents. iranian journal of pharmaceutical research : ijpr, 20(4), 467–496. https://doi.org/10.22037/ijpr.2021.114 730.150124 [37] mauretti, a., spaans, s., bax, n. a. m., sahlgren, c., & bouten, c. v. c. (2017). cardiac progenitor cells and the interplay with their microenvironment. stem cells international, 2017, 7471582. https://doi.org/10.1155/2017/7471582 [38] watson, s. a., scigliano, m., bardi, i., ascione, r., terracciano, c. m., & perbellini, f. (2017). preparation of viable adult ventricular myocardial slices from large and small mammals. nature protocols, 12(12), 2623–2639. https://doi.org/10.1038/nprot.2017.139 https://doi.org/10.1016/j.devcel.2016.01.018 https://doi.org/10.1016/j.devcel.2016.01.018 https://doi.org/10.1016/j.devcel.2022.01.012 https://doi.org/10.1016/j.devcel.2022.01.012 https://doi.org/10.1016/j.cell.2009.04.060 https://doi.org/10.1016/j.cell.2009.04.060 https://doi.org/10.1038/nature15372 https://doi.org/10.3389/fcvm.2022.981982 https://doi.org/10.3389/fcvm.2022.981982 https://www.researchgate.net/figure/cell-sources-for-cardiac-tissue-engineering-a-variety-of-cell-types-can-be-used-in_fig2_349634500 https://www.researchgate.net/figure/cell-sources-for-cardiac-tissue-engineering-a-variety-of-cell-types-can-be-used-in_fig2_349634500 https://www.researchgate.net/figure/cell-sources-for-cardiac-tissue-engineering-a-variety-of-cell-types-can-be-used-in_fig2_349634500 https://www.researchgate.net/figure/cell-sources-for-cardiac-tissue-engineering-a-variety-of-cell-types-can-be-used-in_fig2_349634500 https://doi.org/10.1159/000492704 https://doi.org/10.22037/ijpr.2021.114730.150124 https://doi.org/10.22037/ijpr.2021.114730.150124 https://doi.org/10.1155/2017/7471582 https://doi.org/10.1038/nprot.2017.139 columbia undergraduate science journal vol. 19, 2025 patel © 2025 patel, krishay. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 46 [39] banyasz, t., lozinskiy, i., payne, c. e., edelmann, s., norton, b., chen, b., chen-izu, y., izu, l. t., & balke, c. w. (2008). transformation of adult rat cardiac myocytes in primary culture. experimental physiology, 93(3), 370– 382. https://doi.org/10.1113/expphysiol.2007 .040659 [40] garbern, j. c., mummery, c. l., & lee, r. t. (2013). model systems for cardiovascular regenerative biology. cold spring harbor perspectives in medicine, 3(4), a014019. https://doi.org/10.1101/cshperspect.a0 14019 [41] cerbai, e., sartiani, l., de paoli, p., & mugelli, a. (2000). isolated cardiac cells for electropharmacological studies. pharmacological research, 42(1), 1–8. https://doi.org/10.1006/phrs.1999.0654 [42] maltsev, v. a., wobus, a. m., rohwedel, j., bader, m., & hescheler, j. (1994). cardiomyocytes differentiated in vitro from embryonic stem cells developmentally express cardiac-specific genes and ionic currents. circulation research, 75(2), 233–244. https://doi.org/10.1161/01.res.75.2.233 [43] von gise, a., lin, z., schlegelmilch, k., honor, l. b., pan, g. m., buck, j. n., ma, q., ishiwata, t., zhou, b., camargo, f. d., & pu, w. t. (2012). yap1, the nuclear target of hippo signaling, stimulates heart growth through cardiomyocyte proliferation but not hypertrophy. proceedings of the national academy of sciences of the united states of america, 109(7), 2394–2399. https://doi.org/10.1073/pnas.111613610 9 [44] estimating the risk of drug-induced proarrhythmia using human induced pluripotent stem cell-derived cardiomyocytes—pubmed. (n.d.). https://pubmed.ncbi.nlm.nih.gov/216934 36/ [45] improving cardiac conduction with a skeletal muscle sodium channel by gene and cell therapy—pmc. (n.d.). https://www.ncbi.nlm.nih.gov/pmc/articl es/pmc3392452/ [46] kirkpatrick, c. j., fuchs, s., & unger, r. e. (2011). co-culture systems for vascularization—learning from nature. advanced drug delivery reviews, 63(4–5), 291–299. https://doi.org/10.1016/j.addr.2011.01.0 09 https://doi.org/10.1113/expphysiol.2007.040659 https://doi.org/10.1113/expphysiol.2007.040659 https://doi.org/10.1101/cshperspect.a014019 https://doi.org/10.1101/cshperspect.a014019 https://doi.org/10.1006/phrs.1999.0654 https://doi.org/10.1161/01.res.75.2.233 https://doi.org/10.1073/pnas.1116136109 https://doi.org/10.1073/pnas.1116136109 https://pubmed.ncbi.nlm.nih.gov/21693436/ https://pubmed.ncbi.nlm.nih.gov/21693436/ https://www.ncbi.nlm.nih.gov/pmc/articles/pmc3392452/ https://www.ncbi.nlm.nih.gov/pmc/articles/pmc3392452/ https://doi.org/10.1016/j.addr.2011.01.009 https://doi.org/10.1016/j.addr.2011.01.009 columbia undergraduate science journal vol. 19, 2025 patel © 2025 patel, krishay. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 47 [47] stevens, k. r., kreutziger, k. l., dupras, s. k., korte, f. s., regnier, m., muskheli, v., nourse, m. b., bendixen, k., reinecke, h., & murry, c. e. (2009). physiological function and transplantation of scaffold-free and vascularized human cardiac muscle tissue. proceedings of the national academy of sciences of the united states of america, 106(39), 16568– 16573. https://doi.org/10.1073/pnas.09083811 06 [48] mammalian electrophysiology on a microfluidic platform—pmc. (n.d.). https://www.ncbi.nlm.nih.gov/pmc/articl es/pmc1166618/ [49] klauke, n., smith, g., & cooper, j. m. (2007). microfluidic systems to examine intercellular coupling of pairs of cardiac myocytes. lab on a chip, 7(6), 731–739. https://doi.org/10.1039/b706175g [50] optical imaging. (n.d.). national institute of biomedical imaging and bioengineering. https://www.nibib.nih.gov/scienceeducation/science-topics/opticalimaging [51] hendon, c. p., lye, t. h., yao, x., gan, y., & marboe, c. c. (2019). optical coherence tomography imaging of cardiac substrates. quantitative imaging in medicine and surgery, 9(5), 882–904. https://doi.org/10.21037/qims.2019.05. 09 [52] janib, s. m., moses, a. s., & mackay, j. a. (2010). imaging and drug delivery using theranostic nanoparticles. advanced drug delivery reviews, 62(11), 1052–1063. https://doi.org/10.1016/j.addr.2010.08.0 04 [53] roura, s., gálvez-montón, c., & bayesgenis, a. (2013). bioluminescence imaging: a shining future for cardiac regeneration. journal of cellular and molecular medicine, 17(6), 693–703. https://doi.org/10.1111/jcmm.12018 [54] cao, y.-a., bachmann, m. h., beilhack, a., yang, y., tanaka, m., swijnenburg, r.-j., reeves, r., taylor-edwards, c., schulz, s., doyle, t. c., fathman, c. g., robbins, r. c., herzenberg, l. a., negrin, r. s., & contag, c. h. (2005). molecular imaging using labeled donor tissues reveals patterns of engraftment, rejection, and survival in transplantation. transplantation, 80(1), 134–139. https://doi.org/10.1097/01.tp.000016434 7.50559.a3 [55] min, s., kim, s., sim, w.-s., choi, y. s., joo, h., park, j.-h., lee, s.-j., kim, h., lee, m. j., jeong, i., cui, b., jo, s.h., kim, j.-j., hong, s. b., choi, y.-j., ban, k., kim, y.-g., park, j.-u., lee, h.-a., … cho, s.-w. (2024). versatile human cardiac tissues engineered with perfusable heart extracellular microenvironment for biomedical applications. nature communications, 15(1), 2564. https://doi.org/10.1038/s41467-02446928-y https://doi.org/10.1073/pnas.0908381106 https://doi.org/10.1073/pnas.0908381106 https://www.ncbi.nlm.nih.gov/pmc/articles/pmc1166618/ https://www.ncbi.nlm.nih.gov/pmc/articles/pmc1166618/ https://doi.org/10.1039/b706175g https://www.nibib.nih.gov/science-education/science-topics/optical-imaging https://www.nibib.nih.gov/science-education/science-topics/optical-imaging https://www.nibib.nih.gov/science-education/science-topics/optical-imaging https://doi.org/10.21037/qims.2019.05.09 https://doi.org/10.21037/qims.2019.05.09 https://doi.org/10.1016/j.addr.2010.08.004 https://doi.org/10.1016/j.addr.2010.08.004 https://doi.org/10.1111/jcmm.12018 https://doi.org/10.1097/01.tp.0000164347.50559.a3 https://doi.org/10.1097/01.tp.0000164347.50559.a3 https://doi.org/10.1038/s41467-024-46928-y https://doi.org/10.1038/s41467-024-46928-y columbia undergraduate science journal vol. 19, 2025 patel © 2025 patel, krishay. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 48 [56] liu, n., ye, x., yao, b., zhao, m., wu, p., liu, g., zhuang, d., jiang, h., chen, x., he, y., huang, s., & zhu, p. (2020). advances in 3d bioprinting technology for cardiac tissue engineering and regeneration. bioactive materials, 6(5), 1388–1401. https://doi.org/10.1016/j.bioactmat.202 0.10.021 [57] frontiers | 3d bioprinting technology – one step closer towards cardiac tissue regeneration. (n.d.). https://www.frontiersin.org/articles/10.3 389/fmats.2021.804134/full [58] kato, b., wisser, g., agrawal, d. k., wood, t., & thankam, f. g. (2021). 3d bioprinting of cardiac tissue: current challenges and perspectives. journal of materials science. materials in medicine, 32(5). https://doi.org/10.1007/s10856-02106520-y [59] biomimetic approaches in cardiac tissue engineering: replicating the native heart microenvironment—pmc. (n.d.). https://www.ncbi.nlm.nih.gov/pmc/articl es/pmc10423641/ [60] rashid, a. b., showva, n.-n., & hoque, m. e. (2023). gelatin-based scaffolds: an intuitive support structure for regenerative therapy. current opinion in biomedical engineering, 26, 100452. https://doi.org/10.1016/j.cobme.2023.1 00452 [61] raiola, m., sendra, m., & torres, m. (2023). imaging approaches and the quantitative analysis of heart development. journal of cardiovascular development and disease, 10(4), 145. https://doi.org/10.3390/jcdd10040145 [62] fluorescence microscopy—an outline of hardware, biological handling, and fluorophore considerations—pmc. (n.d.). https://www.ncbi.nlm.nih.gov/pmc/articl es/pmc8750338/ [63] doppler, s. a., deutsch, m.-a., lange, r., & krane, m. (2013). cardiac regeneration: current therapies—future concepts. journal of thoracic disease, 5(5). https://doi.org/10.3978/j.issn.20721439.2013.08.71 https://doi.org/10.1016/j.bioactmat.2020.10.021 https://doi.org/10.1016/j.bioactmat.2020.10.021 https://www.frontiersin.org/articles/10.3389/fmats.2021.804134/full https://www.frontiersin.org/articles/10.3389/fmats.2021.804134/full https://doi.org/10.1007/s10856-021-06520-y https://doi.org/10.1007/s10856-021-06520-y https://www.ncbi.nlm.nih.gov/pmc/articles/pmc10423641/ https://www.ncbi.nlm.nih.gov/pmc/articles/pmc10423641/ https://doi.org/10.1016/j.cobme.2023.100452 https://doi.org/10.1016/j.cobme.2023.100452 https://doi.org/10.3390/jcdd10040145 https://www.ncbi.nlm.nih.gov/pmc/articles/pmc8750338/ https://www.ncbi.nlm.nih.gov/pmc/articles/pmc8750338/ https://doi.org/10.3978/j.issn.2072-1439.2013.08.71 https://doi.org/10.3978/j.issn.2072-1439.2013.08.71 columbia undergraduate science journal vol. 19, 2025 patel © 2025 patel, krishay. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. 49 introduction the heart and its regenerative capabilities conclusion corresponding author author contributions references template for electronic submission to acs journals 2023 article of the year 6 © 2023 lang, elvis. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. expression of single mrna constructs encoding both crispr-cas9 protein and guide rnas for future gene therapy applications elvis lang1, john tilton2, thomas sweet3 1,2,3case western reserve university school of medicine, cleveland, oh, usa keywords: crispr-cas9, gene editing, mrna abstract: the basis of many life-threatening diseases is disruption in key genes. in many cases, repairing these disruptions can prevent or reverse disease. the development of crispr-cas9 technology, which consists of cas9 nuclease directed to specific genomic locations by guide rna (grna), has significantly progressed in the past decade and has shown signs of promise for treating diseases such as alzheimer’s and cystic fibrosis. one integral issue of gene editing therapy is the method and effectiveness of delivery. current approaches such as lentiviral and adeno-associated virus vectors suffer from either stable, constant expression of crispr components that causes unintended gene editing or an inability to efficiently carry large cargoes such as two independent genes: cas9 and guide rna. to begin to bypass these cargo limitations, we created a crispr-cas9 mrna structure that encompasses all of the necessary components for gene editing on a single rna. these constructs consist of a promoter, followed by a cas9 open reading frame, a triplex region from malat1 that protects the cas9 open reading frame, and then either 1, 2, or 4 grnas that target specific reporters, with each grna between two self-cleaving ribozyme sequences. these constructs successfully drove cas9 editing of two distinct reporters in human cells and thus open the door for many more experiments such as incorporation into various delivery constructs to further develop this technology for gene editing therapy. introduction crispr-cas9 (clustered regularly interspaced short palindromic repeatscrispr associated protein 9) gene editing technology has been shown to be a promising solution to genetic diseases. cas9 proteins are naturally found in streptococcus thermophilus as a defense system against invading viruses, but now scientists use this system to perform gene editing in many organisms.1 in the crispr system, guide rnas bind to a specific region of dna as well as cas9 nuclease, guiding cas9 to the targeted dna region to cause doublestranded breaks in an organism’s genome. the cell’s own non-homologous end joining dna repair system is used to repair the break made by cas9 which induces changes to targeted sequences in the genome (figure 1).2 this system has been extensively researched and expanded upon for the past decade, and more systems have been developed that use cas9 as well such as base and prime editors. in these systems, various proteins have been fused to nuclease-deficient cas9 to modify the genome in several different ways. columbia undergraduate science journal vol. 17, 2023 lang et. al. 7 the advent of base editing technology grants the ability to perform specific types of point mutations to nucleotides. the first effective iteration of this technology, a cytosine base editor, arose in 2016 where komor et al. was able to fuse rat derived deaminase protein to the amino terminal end of an inactive cas9 protein.3 instead of causing double stranded breaks, this editor is able to deaminate cytosine to uracil. since dna replication machinery does not recognize uracil, the replication results in a c-g to t-a mutation.4 through further research, this editing system was able to become more specific and effective. an adenine base editor was also developed using a similar method to the cytosine base editor.4 like cas9, cytosine and adenine base editors can cause targeted mutations, however these base editors are more specific due to the ability to specify the point mutation. in a gene therapy context, adenine base editors are being used to mutate premature stop codons, which are thought to cause faulty gene expression in up to a third of genetic diseases.5 despite the incredible capabilities of base editors, they are unable to perform more than specific point mutations. however, another cas9 related system, prime editors, can perform any type of mutations that can be encoded in a grna.4 simply speaking, these systems are created by fusing a reverse transcriptase to an inactive cas9 protein. a prime editing guide rna then directs targeted transition and transversion mutations.4 although there is great potential for this system, it still requires much more research to improve its efficacy and specificity. other applications include grna targeted modulation of transcription or epigenetic marks by fusing transcriptional or epigenetic regulators to inactive cas9.6 these methods can be used to up or downregulate genes like those related to cancer, therefore showing a lot of promise for therapeutic uses.4 in several mouse models, researchers have been able to reduce the severity of neurodegenerative diseases by editing genes with crispr. 7 recent studies have demonstrated cas9 to be promising in decreasing the effects of alzheimer’s and parkinson’s disease through the mutation of the app and lrrk2 gene respectively.7 a study from yang et al. has shown that crispr technology can be used to ameliorate the symptoms of huntington's disease in mouse models through mutations in the htt gene.8 in addition, clinical trials for crispr related therapies have already begun to treat cystic fibrosis by mutating the cftr gene.9 as a result, crispr, while still being heavily researched and developed, shows extremely promising capabilities for mitigating genetic diseases. crispr-cas9 delivery figure 1: illustration of how crispr-cas9 performs gene editing. columbia undergraduate science journal vol. 17, 2023 lang et. al. 8 one major hurdle to fully realizing the potential of cas9-based approaches is delivery to tissues and cells. there are two primary ways currently used to deliver cas9 and grna: viral and non-viral delivery. with viral delivery, there are methods that use adeno-associated viruses (aavs) or lentiviral vectors to encapsulate both cas9 and grna genetic information and deliver it to cells through a mechanism similar to viral infection. on the other hand, non-viral methods utilize other macromolecules like lipid nanoparticles (lnps), protein polymers, and au nanoparticles to deliver grna and cas9 mrna, protein, or dna.10 delivering the protein form of crispr-cas9 along with grna is efficient and doesn’t result in much off-target editing, however it is very expensive and there is risk of endotoxin contamination.10 cas9 mrna + grna delivery has similar figure 2: sequence of pctg plasmids involved with cas9 gene editing. 1x refers to one copy of the grna between two self-cleaving ribozyme sequences. the pctg 2x plasmids have two ribozyme + grna + ribozyme sequences subsequently and the pctg 4x (not shown) plasmid has four. figure 3: schematic of rfp and gfp reporters used. gene editing of tlr reporter causes a frameshift which makes rfp. gfp reporter is activated through editing multiple sites to delete transcriptional stops. columbia undergraduate science journal vol. 17, 2023 lang et. al. 9 benefits to the protein form; however, it requires two components to get into particles and cells.10 alternatively, cas9 delivery can also be performed through complexing and conjugating with proteins. ramakrishna et al. were able to design a cell-penetrating protein (cpp) that was conjugated to cas9 and combined with a cpp complexed to a grna, resulting in efficient endogenous gene editing.11 this technology has been proven effective in hek293t cells, embryonic stem cells, embryonic carcinoma cells, and dermal fibroblasts.11 additionally, this system was able to produce less off-target editing events, however it is not perfect and it is difficult in vivo without a protective lipid nanoparticle.11 while there have been studies showing effective delivery of crispr-cas9 into tissues using lentiviral and adeno-associated virus vectors, these methods of long-term expression can be problematic if used as a therapeutic12,13. for example, lentiviruses involve integration into the genome and potentially permanent, constant expression of cas9 along with a grna14. despite the specificity of cas9, the technology is not guaranteed to always edit targeted gene sequences, and this can potentially lead to malignant or deleterious mutations. in many of the previous methods, the size of cargo is also a significant barrier for the efficient delivery of gene editing technology. for instance, aav packaging limits require separate viruses to be made for cas9 and grna. lentiviral packaging limitations lead to reduced efficacy when carrying large cas9 and grna genes together. several non-viral approaches depend on two different components (cas9 and grna) to get into the same particles and cells. therefore, to increase the efficiency of delivery, we engineered both the grna and cas9 segments to be expressed from a single piece of rna. with this construct design, cargo size would be significantly reduced and in the case of rna delivery methods, two molecules reduced to one. this configuration has the potential to open the door for more efficient and a greater variety of delivery methods. methods plasmids pcmv-vsv-g was a gift from bob weinberg (addgene plasmid # 8454 ; http://n2t.net/addgene:8454 ; rrid:addgene_8454).15 lenticrispr v2 was a gift from feng zhang (addgene plasmid # 52961 ; http://n2t.net/addgene:52961 ; rrid:addgene_52961).16 paavs1-tlr targeting vector was a gift from ralf kuehn (addgene plasmid # 64215 ; http://n2t.net/addgene:64215 ; rrid:addgene_64215).17 pcag-loxpstoploxp-zsgreen was a gift from pawel pelczar (addgene plasmid # 51269 ; http://n2t.net/addgene:51269 ; rrid:addgene_51269).18 cloning of constructs encoding cas9 and grna on a single rna a pcmv-cas9 plasmid was cloned by replacing the vsv-g open reading frame in pcmv-vsv-g with cas9 open reading frame from lenticrispr v2 using pcr followed by hifi assembly (new england biolabs). to clone in grnas, pcmv-cas9 was linearized downstream of the cas9 stop codon using stui, then malat1 triplex (from campa et al.)19 + hammerhead (hh) ribozyme dna and guide rna sequence + hepatitis delta virus (hdv) ribozyme dna were ordered from integrated dna technologies as gblocks with 20 bp overhangs between all three dna fragments for hifi assembly. xmai, nhei, and hindiii sites as well as homologous regions for hifi assembly were created in these dnas in order to allow for expansion of the grna cassette according to the approach of spakman et al. (figure 2).20 the result of these steps was pcmvcas9-malat1 triplex-hh ribozyme-grnahdv ribozyme, or pctg-1x, where the 1x columbia undergraduate science journal vol. 17, 2023 lang et. al. 10 denotes a single grna. in the text, grna targeting is added to the name: pctg-tlr1x encodes grna targeting traffic light reporter (ggtagcgggcgaagcactgc) and pctg-gfp-1x encodes grna targeting gfp reporter (sgtom from wei et al.).20,21 to create 2x and 4x grna constructs, we used the approach of spakman et al. 2020 but with different restriction enzymes. briefly, 1x constructs were linearized with nhei and separately the entire hh ribozymegrna-hdv ribozyme region was cut out with xmai and hindiii then gel purified. this creates overhangs that allow the entire hh ribozyme-grna-hdv ribozyme region to be cloned upstream of the other copy of the grna cassette with hifi assembly to create 2x grna constructs. the same expansion was repeated starting with the 2x construct to get the 4x construct encoding 4 identical grnas. plasmid sequences were confirmed by sanger sequencing. transfection of pctg-tlr and pctggfp plasmids the day before transfection, hek293t cells were seeded in dulbecco’s modified eagle medium supplemented with 10% fetal bovine serum at roughly 25% confluency in 6-well plates and incubated overnight at 37c and 5% co2. the next day, pctg-tlr 1x, 2x, 4x, or pctg-gfp 1x and 2x were transfected into hek293t cells with either paavs1-tlr or pcag-loxpstoploxp-zsgreen using jetoptimus transfection reagent. plates were then left to incubate for 48 hours at 37c and 5% co2. imaging of all constructed plasmids for gene editing activity all samples were imaged after 48 hours incubation using the lionheart fx automated microscope with the 4x objective. the plates transfected with paavs1-tlr reporter were imaged for rfp and the plates that received pcag-loxpstoploxp-zsgreen reporter were imaged for gfp. gen5 figure 4: pctg-tlr constructs efficiently edit tlr reporter. hek293t cells were transfected with indicated cas9-grna plasmid and traffic light reporter plasmid, then imaged for tagrfp after 48 hours. fluorescence imaging was done with the lionheart fx automated microscope using the 4x objective. each red particle shown represents one cell expressing tlr. each panel represents 1 of 16 fields imaged. brightfield images were taken of all wells and showed similar cellular confluency (data not shown). columbia undergraduate science journal vol. 17, 2023 lang et. al. 11 software was used to count rfp+ or gfp+ cells as well as calculate intensity of expression per cell. after imaging, a two-tailed, two sample equal variance, statistical t-test was performed to determine significance of rfp or gfp expression (n=2). figures 1, 2, and 3 were made with biorender software. results in order to test cas9 and grna single rna configurations, we created constructs that contain cas9 open reading frame followed by an rna region that forms a protective triplex.22 this sequence is then followed by grna sequence sandwiched between self-cleaving ribozymes (figure 2). in the cell, these self-cleaving ribozymes cleave out grna so that it can bind to cas9 and the targeted dna. the triplex is needed because these ribozyme cleavages would expose the 3’ end of the cas9 open reading frame to nucleases in the absence of the triplex.22 we created versions expressing 1, 2, or 4 grnas that target traffic light reporter (tlr), or 1 or 2 grnas targeting the stop region of pcagloxpstoploxp-zsgreen (figure 3). these reporters were targeted because both are well characterized and are also integrated into mouse lines, and thus are useful grnas for future in vivo experiments.23 to test targeted editing of tlr, we transfected each of these constructs into hek293t cells along with traffic light reporter (tlr). traffic light reporter encodes an upstream stop codon as well as an out-of-frame tagrfp. upon cas9 editing upstream of the early stop codon, a portion of editing events will cause the upstream stop codon to go outof-frame and bring tagrfp in frame, resulting in rfp expression (figure 3). cells expressing tlr grnas exhibit rfp expression, which indicates successful gene editing by cas9 that targeted the rfp reporter (figures 4a-c). the quantification of rfp+ cells also demonstrate a significant number of cells underwent gene editing (figure 6a). when comparing the difference between 1x, 2x, and 4x tlr grnas, there is a significant increase in figure 5: pctg-gfp-2x construct edits gfp reporter. hek293t cells were transfected with indicated cas9-grna plasmid and green fluorescent protein reporter plasmid, then imaged for gfp after 48 hours. fluorescence imaging was done with the lionheart fx automated microscope using the 4x objective. each bright green particle shown represents one cell expressing gfp. each panel represents 1 of 16 fields imaged. brightfield images were taken of all wells and showed similar cellular confluency (data not shown). columbia undergraduate science journal vol. 17, 2023 lang et. al. 12 the number of cells that displayed gene editing (figure 6a). additionally, the magnitude of rfp expression per cell increased corresponding to the number of guide rna sequences, meaning that 1x tlr grna showed the lowest magnitude, followed by 2x, and finally 4x demonstrated the largest magnitude of fluorescence (figures 4a-c and 6b). importantly, cells not expressing grna (figure 4f) or expressing constructs encoding gfp targeting grna (figure 4d-e) do not result in rfp expression, indicating that editing of tlr is selective. similarly, constructs expressing cas9 and grnas targeted to the stop region of pcag-loxpstoploxp-zsgreen were tested and demonstrated selective gene editing as well. pcag-loxpstoploxp-zsgreen encodes a zsgreen reporter downstream of 3 successive transcription termination sites. normally zsgreen expression would be low but targeting cas9 to regions between transcription termination sites with grna results in removal of termination sites and zsgreen expression (figure 3). cells expressing cas9 and tlr grnas along with reporter exhibited some gfp (figures 5a–c), but co-expression of cas9 and 2 gfp grnas resulted in increased gfp expression (figures 5e and 6c-d). when quantified, there was no statistical significance between the gfp reporter only negative control (figure 5f) and the pctg-gfp 2x version of the plasmid with figure 6: quantitative analysis of imaging from figures 4 and 5. corresponds with figure 4: tagrfp+ cell counts (a) obtained over 16 images for each sample were obtained using gen5 software. average rfp intensity per cell (b) was also obtained using gen5 software. corresponds with figure 5: gfp+ cell counts (c) obtained over 16 images for each sample were obtained using gen5 software. average gfp intensity per cell (d) was also obtained using gen5 software. these data are the result of two independent experiments (n=2). columbia undergraduate science journal vol. 17, 2023 lang et. al. 13 regards to both number of cells that express gfp and the magnitude of expression/gene editing (figure 6c-d). this reporter has a higher background signal in the absence of editing as well as a higher threshold for signal (multiple cas9 have to hit the same reporter dna). in addition, the reporter only control is a transfection with only one plasmid and neither cas9 nor grna, meaning that the reporter only control is not the best direct comparison. however, there was a significant increase in both number of gfp+ cells, as well as gfp intensity per cell when comparing pctg-gfp 2x to pctg-gfp 1x or any of the pctg-tlr constructs (figure 6c-d) indicating some level of selective gene editing. for future research, it will be valuable to create a pctg-gfp 4x plasmid and assess whether or not it demonstrates an increase in gene editing activity similar to the pctg-tlr plasmids. discussion the single rna encoding cas9 + guide rna constructs were successful in performing gene editing on two different reporters. the plasmids containing tlr guide rna were able to produce significant gene editing only on tlr and the duplication of guide rna sequences was shown to improve not only the number of cells that had gene editing but also increased the magnitude of gene editing per cell as well. with regards to the gfp guide rna containing plasmid, there was a significant increase in the number of cells with gene editing between the 2x version of pctg-gfp when compared to the 1x version or any of the pctg-tlr versions, demonstrating the increase in number of grnas is also able to increase the level of selective gene editing in cells. while this result was clearly weaker in magnitude than the tlr system, there are some limitations to using this gfp reporter that likely account for this. first, there are nonzero amounts of background green fluorescence for negative controls (figure 5). this higher background is likely worse in the reporter only control in which only one plasmid is being transfected as opposed to all other samples which get two plasmids. in addition, the gfp system requires multiple editing events per cell to get gfp (figure 3), making the signal weaker than the single editing event required for the tlr system. an article written by mccarty et al. reviews construct designs somewhat similar to ours, including one encoding cas12a.19,24 we selected cas9 for our configuration due to the wealth of knowledge about this system and cas9’s high activity relative to other cas proteins. in addition, the selected self-cleaving ribozymes that process grnas are wellcharacterized and regulated cleavage can be engineered.25,26 since we found that this kind of construct design works, we can potentially also test other designs, such as one where different grna sequences are encoded on one rna strand which would allow for simultaneous gene editing at multiple gene loci.24 this kind of structure could be promising for gene therapy in the future and is worth considering. conclusion these results are significant because after demonstrating that these cas9/grna single rna constructs function in hek293t cells, future experiments can now be designed to test whether this configuration improves delivery of cas9/grna using systems that have been shown to be limited by cargo size like aavs, lentiviruses, and mrna delivered by nano-lipid particles. aavs are known to have limited cargo capacity despite being very effective for delivery27. combining our configuration with smaller promoters, more compact cas enzymes, and regulatable ribozymes may even allow a single aav or more efficient lentiviruses to be produced.28,29 with regards to nonviral delivery, lipid nanoparticles are already known to produce efficient delivery of cargo, as well as target columbia undergraduate science journal vol. 17, 2023 lang et. al. 14 specific cells. while nonviral delivery systems are generally able to hold larger cargos, it is still valuable to test this single rna construct’s effectiveness in lipid nanoparticles and whether it demonstrates improvement in gene editing frequency and magnitude since current methods often rely on getting two separate molecules, grna and cas9, into particles. in the future, these delivery approaches could be directly used in animal models too as the guide rnas we have used in this study target commercially available tlr and ai9 mice. beyond this, testing for effective modification of disease associated genes using other grnas would provide valuable insight for the prospect of using our single rna structure for gene therapy applications. in short, this technology has the potential to improve both existing and emerging cas9/grna delivery methods. author information corresponding author *elvis lang – ecl67@case.edu author contributions elvis lang conceived experiments, performed the cloning, and performed data analysis. he also wrote the manuscript and generated all the figures used. dr. john tilton provided direction and guidance during this project. dr. thomas sweet also provided direction and including with the manuscript, conceiving experiments, and also did some of the plasmid design, cell culture work, and imaging for this project. funding sources experiments presented here were supported by nih 5ug3hl151544-03. abbreviations crispr: clustered regularly interspaced short palindromic repeats cas9: crispr associated protein 9 grna: guide rna aav: adeno associated virus rnp: ribonucleoprotein complex lnp: lipid nanoparticles cpp: cell penetrating protein hek293t: human embryonic kidney 293t tlr: traffic light reporter rfp: red fluoresce ent protein gfp: green fluorescent protein pctg: pcmv-triplex-grna cmv: cytomegalovirus references 1. deveau, h., garneau, j. e. & moineau, s. crispr/cas system and its role in phage-bacteria interactions. annu rev microbiol 64, 475–493 (2010). https://doi.org/10.1146/annurev.micro.112408.134123 2. ran, f., hsu, p., wright, j. et al. genome engineering using the crisprcas9 system. nat protoc 8, 2281– 2308 (2013 https://doi.org/10.1038/nprot.2013.143 3. komor, a. c., kim, y. b., packer, m. s., zuris, j. a. & liu, d. r. programmable editing of a target base in genomic dna without double-stranded dna cleavage. nature 533, 420–424 (2016) https://doi.org/10.1038/nature17946 4. kantor, a., mcclements, m. & maclaren, r. crispr-cas9 dna base-editing and prime-editing. int j mol sci 21, 6240 (2020). https://doi.org/10.3390/ijms21176240 5. karijolich, j. & yu, y.-t. therapeutic suppression of premature termination codons: mechanisms and clinical considerations (review). int j mol med 34, 355–362 (2014). https://doi.org/10.3892/ijmm.2014.180 9 6. zannas, a. s. & west, a. e. epigenetics and the regulation of stress vulnerability and resilience. neuroscience 264, 157–170 (2014). https://doi.org/10.1146/annurev.micro.112408.134123 https://doi.org/10.1146/annurev.micro.112408.134123 https://doi.org/10.1038/nprot.2013.143 https://doi.org/10.1038/nature17946 https://doi.org/10.3390/ijms21176240 https://doi.org/10.3892/ijmm.2014.1809 https://doi.org/10.3892/ijmm.2014.1809 columbia undergraduate science journal vol. 17, 2023 lang et. al. 15 https://doi.org/10.1016/j.neuroscience.2013.12.003 7. karimian, a. et al. crispr/cas9 novel therapeutic road for the treatment of neurodegenerative diseases. life sci 259, 118165 (2020). https://doi.org/10.1016/j.lfs.2020.11816 5 8. yang, s. et al. crispr/cas9mediated gene editing ameliorates neurotoxicity in mouse model of huntington’s disease. journal of clinical investigation 127, 2719–2724 (2017). https://doi.org/10.1172/jci92087 9. da silva sanchez, a., paunovska, k., cristian, a. & dahlman, j. e. treating cystic fibrosis with mrna and crispr. hum gene ther 31, 940– 955 (2020). https://doi.org/10.1089/hum.2020.137 10. li, l., hu, s. & chen, x. non-viral delivery systems for crispr/cas9based genome editing: challenges and opportunities. biomaterials 171, 207–218 (2018). https://doi.org/10.1016/j.biomaterials.2018.04.031 11. ramakrishna, s. et al. gene disruption by cell-penetrating peptide-mediated delivery of cas9 protein and guide rna. genome res 24, 1020–1027 (2014). https://doi.org/10.1101/gr.171264.113 12. escors, d. & breckpot, k. lentiviral vectors in gene therapy: their current status and future potential. arch immunol ther exp (warsz) 58, 107– 119 (2010). https://doi.org/10.1007/s00005-0100063-4 13. coura, r. dos s. & nardi, n. b. a role for adeno-associated viral vectors in gene therapy. genet mol biol 31, 1–11 (2008). https://doi.org/10.1590/s141547572008000100001 14. ciuffi, a. mechanisms governing lentivirus integration site selection. curr gene ther 8, 419–429 (2008). https://dx.doi.org/10.2174/1566523087 86848021 15. steward, s. a. et al. lentivirus-delivered stable gene silencing by rnai in primary cells. rna 9, 493–501 (2003). https://doi.org/10.1261/rna.2192803 16. sanjana, n. e., shalem, o. & zhang, f. improved vectors and genome-wide libraries for crispr screening. nat methods 11, 783–784 (2014). https://doi.org/10.1038/nmeth.3047 17. chu, v. t. et al. increasing the efficiency of homology-directed repair for crispr-cas9-induced precise gene editing in mammalian cells. nat biotechnol 33, 543–548 (2015). https://doi.org/10.1038/nbt.3198 18. hermann, m. et al. binary recombinase systems for high-resolution conditional mutagenesis. nucleic acids res 42, 3894–3907 (2014). https://doi.org/10.1093/nar/gkt1361 19. campa, c. c., weisbach, n. r., santinha, a. j., incarnato, d. & platt, r. j. multiplexed genome engineering by cas12a and crispr arrays encoded on single transcripts. nat methods 16, 887–893 (2019). https://doi.org/10.1038/s41592-0190508-6 20. spakman, d., king, g. a., peterman, e. j. g. & wuite, g. j. l. constructing arrays of nucleosome positioning sequences using gibson assembly for single-molecule studies. sci rep 10, 9903 (2020). https://doi.org/10.1038/s41598-02066259-4 21. wei, t., cheng, q., min, y.-l., olson, e. n. & siegwart, d. j. systemic nanoparticle delivery of crispr-cas9 ribonucleoproteins for effective tissue specific genome editing. nat commun 11, 3232 (2020). https://doi.org/10.1016/j.neuroscience.2013.12.003 https://doi.org/10.1016/j.neuroscience.2013.12.003 https://doi.org/10.1016/j.lfs.2020.118165 https://doi.org/10.1016/j.lfs.2020.118165 https://doi.org/10.1172/jci92087 https://doi.org/10.1089/hum.2020.137 https://doi.org/10.1016/j.biomaterials.2018.04.031 https://doi.org/10.1016/j.biomaterials.2018.04.031 https://doi.org/10.1101/gr.171264.113 https://doi.org/10.1007/s00005-010-0063-4 https://doi.org/10.1007/s00005-010-0063-4 https://doi.org/10.1590/s1415-47572008000100001 https://doi.org/10.1590/s1415-47572008000100001 https://dx.doi.org/10.2174/156652308786848021 https://dx.doi.org/10.2174/156652308786848021 https://doi.org/10.1261/rna.2192803 https://doi.org/10.1038/nmeth.3047 https://doi.org/10.1038/nbt.3198 https://doi.org/10.1093/nar/gkt1361 https://doi.org/10.1038/s41592-019-0508-6 https://doi.org/10.1038/s41592-019-0508-6 https://doi.org/10.1038/s41598-020-66259-4 https://doi.org/10.1038/s41598-020-66259-4 columbia undergraduate science journal vol. 17, 2023 lang et. al. 16 https://doi.org/10.1038/s41467-02017029-3 22. wilusz, j. e. et al. a triple helix stabilizes the 3′ ends of long noncoding rnas that lack poly(a) tails. genes dev 26, 2392–2407 (2012). https://doi.org/10.1101/gad.204438.11 2 23. bryda, e. c. et al. a novel conditional zsgreen-expressing transgenic reporter rat strain for validating cre recombinase expression. sci rep 9, 13330 (2019). https://doi.org/10.1038/s41598-01949783-w 24. mccarty, n. s., graham, a. e., studená, l. & ledesma-amaro, r. multiplexed crispr technologies for gene editing and transcriptional regulation. nat commun 11, 1281 (2020). https://doi.org/10.1038/s41467-02015053-x 25. beilstein, k., wittmann, a., grez, m. & suess, b. conditional control of mammalian gene expression by tetracycline-dependent hammerhead ribozymes. acs synth biol 4, 526– 534 (2015). https://doi.org/10.1021/sb500270h 26. komatsu, y. regulation of ribozyme activity with short oligonucleotides. biol pharm bull 27, 457–462 (2004). https://doi.org/10.1248/bpb.27.457 27. salman, a. et al. non-viral delivery of crispr/cas cargo to the retina using nanoparticles: current possibilities, challenges, and limitations. pharmaceutics 14, 1842 (2022). https://doi.org/10.3390/pharmaceutics14091842 28. frommer, j., appel, b. & müller, s. ribozymes that can be regulated by external stimuli. curr opin biotechnol 31, 35–41 (2015). https://doi.org/10.1016/j.copbio.2014.07.009 29. kuwabara, t. allosterically controllable ribozymes with biosensor functions. curr opin chem biol 4, 669–677 (2000). https://doi.org/10.1016/s13675931(00)00150-2 https://doi.org/10.1038/s41467-020-17029-3 https://doi.org/10.1038/s41467-020-17029-3 https://doi.org/10.1101/gad.204438.112 https://doi.org/10.1101/gad.204438.112 https://doi.org/10.1038/s41598-019-49783-w https://doi.org/10.1038/s41598-019-49783-w https://doi.org/10.1038/s41467-020-15053-x https://doi.org/10.1038/s41467-020-15053-x https://doi.org/10.1021/sb500270h https://doi.org/10.1248/bpb.27.457 https://doi.org/10.3390/pharmaceutics14091842 https://doi.org/10.3390/pharmaceutics14091842 https://doi.org/10.1016/j.copbio.2014.07.009 https://doi.org/10.1016/j.copbio.2014.07.009 https://doi.org/10.1016/s1367-5931(00)00150-2 https://doi.org/10.1016/s1367-5931(00)00150-2 30 effect of eccentricity in microwave imaging of multiple composite pipes yuki gao*1, noshin raisa1, and reza k. amineh1 1department of electrical and computer engineering, new york institute of technology, new york, ny, usa abstract: the use of non-metallic composites that are durable, low cost, and lightweight is growing fast in various industries. in the oil and gas industry, a commonly used form of these materials is in the shape of pipes. such pipes can be damaged due to material loss (defects and holes), erosions, and more which may cause major production failures or environmental mishaps. to prevent these issues, non-destructive testing (ndt) methods need to be employed for regular inspections of such components. since traditional ndt methods are mainly used for metallic pipes, microwave imaging has recently been proposed as a promising approach for examination of non-metallic pipes. while microwave imaging can be employed for inspection of multiple layers of pipes, the effect of undesired eccentricity of the pipes (undesired distance between the centers of multiple pipes which are supposed to be concentric) can impose additional imaging errors. in this paper, for the first time, we study the effect of eccentricity of the pipes on the images reconstructed using near-field holographic microwave imaging on double pipes through simulations. to have a realistic study, we add artificial noise to the simulated data. keywords: eccentricity effect, microwave imaging, non-destructive testing, non-metallic pipes © 2021 gao, raisa, amineh. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. introduction recently, non-metallic pipes and composite components such as fiber reinforced plastic (frp), glass reinforced epoxy resin (gre), high density polyethylene (hdpe), reinforced rubber expansion joints (rejs), carbon fiber reinforced plastics (cfrp), and polyvinyl chloride (pvc) are replacing metallic pipes throughout different industries due to advantages such as durability, low cost, light-weight, resistance to corrosion, etc. the growing demand for these materials necessitates the use of proper non-destructive testing (ndt) techniques for material integrity inspections. in general, ndt methods such as ultrasonic testing, radiography, eddy current, and magnetic flux leakage have been widely applied in different industries for inspection of metallic components. however, these ndt methods cannot fulfill the demand for testing certain materials and components such as non-metallic composite pipes. for example, due to the complex structure of composite materials such as frp/gre [1] and the nature of defects and failure morphology in hdpe thermal fusion joints, ultrasonic testing fails to perform ndt for these mediums [2, 3]. on the other hand, radiography relies on the use of x-ray [4] which requires extra safety measures. besides, it is incapable of detecting decolumbia undergraduate science journal vol. 15, 2021 gao, raisa et. al. 31 lamination and planar cracks for defects when the local density remains nearly the same. thus, to fulfill the growing demand for ndt of non-metallic materials, microwave measurement techniques have been proposed (e.g., see [5, 6]). the usage of microwave imaging helps detecting defects, cracks, holes, and more in such components. in particular, microwave holographic imaging is a fast and robust imaging technique that has been successfully applied in various applications such as the security screening of airport passengers [7], etc. originally, microwave holographic imaging techniques were developed based on synthetic aperture radar (sar) imaging techniques [7] which employ far-field assumptions, i.e., the imaging distance which is the distance between the measuring antennas and the imaged object is assumed to be much larger than the wavelength. wide-band sar imaging has been used to produce three-dimensional (3d) images of the vertical cracks/flaws in fat and curved hdpe pipes [8]. recently, sar-based imaging techniques have been extended to the near-field applications where the distance between the measuring antennas and the imaged object is small (e.g., see [9, 10]). thus, these techniques can be called near-field holographic imaging techniques where the information related to a specific imaging system is obtained a priori through the measurement of the so-called point-spread functions (psfs) [11]. this offers several advantages such as: the reduction of modeling errors (as the modeling of the antennas and the imaging setup is not required), the reduction of errors due to uncertainties in the material properties, and the reduction of errors due to the size of antennas (measuring the psfs directly instead of having assumption-based point-wise antennas). although analytical expressions for the psf (instead of direct measurement of them) can still be used in near-field holographic imaging, the material selection and the ignored near-field terms for the antennas may degrade the image reconstruction quality. it is common to use multiple pipes in concentric configuration, as illustrated in figure 1, to improve the efficiency and increase the lifetime of the wellbore production in oil and gas industry [12] or to separate the flow in the fluid transfer pipeline [13]. near-field holographic imaging has been extended to the application of multiple non-metallic pipe imaging in [14, 15] where the pipes are assumed to be perfectly concentric. in this paper, we study the performance of the near-field holographic imaging of double pipes with different eccentricity values, i.e., the centers of the two pipes are not perfectly aligned. although, in industry, normally centralizers are employed for making the multiple pipes concentric, the small misalignment of the centers, called eccentricity, can impose errors in image reconstruction when using techniques that have been developed based on the zero-eccentricity assumption. thus, here, we consider this important factor for the first time and we use a quantitative measure, called reconstruction error (re), to evaluate the degradation of the images of the defects on the inner and outer pipes of a double-pipe configuration due to various eccentricity values. it is worth noting that the effect of other important parameters for the considered microwave imaging setup such as thickness, radius, and permittivity of the pipes as well as angular separation of the antennas have been already studied in [15] and will be excluded here. although the study is performed through simulations, we add artificial noise to the simulated data to have realistic results. methods in this section, we review the near-field holographic imaging approach for imaging of multiple pipes using an array of receiver antennas and multiple frequency data. figure 1 illustrates the microwave imaging setup. it consists of a transmitter antenna to illuminate the pipes and an array of an receiver antennas measuring the scattered fields. the transmitter and receiver antennas scan a circular aperture with radius of ar . it is assumed that the defects and columbia undergraduate science journal vol. 15, 2021 gao, raisa et. al. 32 pipes are infinite along the longitudinal direction (z). the scattered field is recorded at nφ angles along the azimuthal direction φ (within [0,2π]). the complex-valued scattered field ( )sce φ is measured, at each sampling position, at nω frequencies within the band of 1ω to nω ω by each receiver. such scattered response is obtained from subtracting the response of the pipes without defects from the response of the same pipes with defects. the image reconstruction process then provides one-dimensional (1d) images of the pipes with radii ir , where 1, , ri n= � . please note that the imaging along the z direction can be implemented using similar concepts discussed here. the imaging system is assumed to be linear and space-invariant (lsi) which allows us to use the convolution theory. the convolution theory allows to write the response to an unknown input to the system as the convolution of the point-spread functions (psf) (also known as impulse responses) of the system with that unknown input function. for implementation of the near-field holographic imaging, first, the psfs of the lsi imaging system are acquired. these psfs are approximated by measuring small defects, called calibration defects (cds) placed on each pipe one at a time, representing impulse functions as the inputs to the imaging system. in other words, the psfs are measured by the same imaging system that will be later used for imaging test objects. these cds are the smallest defects that can be measured by the system. to provide more data for image reconstruction, measurements can be implemented at multiple frequencies, nω , n=1,…, nω and by multiple receivers, ma , m=1,…, an . we denote the measured psf function for the i-th pipe measured by the receiver antenna am at frequency nω by , , ( , ) m sc cd i a ne φ ω . we also denote the measured scattered field by the receiver antenna am at frequency nω by ( , ) m sc a ne φ ω . let’s first consider the spatially-sampled versions of ( , ) m sc a ne φ ω , , , ( , ) m sc cd i a ne φ ω , and ( )if φ denoted by ( , ) m sc a nnφ ωe , , , ( , ) m sc cd i a nnφ ωe , and ( )i nφf , 1,...,n nφ φ= , with the angular interval denoted by φ∆ . using the convolution theory and discrete fourier transforms (dft) along φ direction, it can be shown that the unknown shape functions of the defects on the pipes ( )if φ , 1, , ri n= � , can be found by solving the following system of equations at each spatial frequency kφ [15]: (1) where (2) and (3) where ( , ) m sc a nkφ ωe�� , , , ( , ) m sc cd i a nkφ ωe�� , and ( )i kφf�� denote dft along φ axis of ( , ) m sc a nnφ ωe , , , ( , ) m sc cd i a nnφ ωe , and ( )i nφf , respectively. these systems of equations are solved at each spatial frequency kφ to obtain the values for ( )i kφf�� , 1, , ri n= � . then, inverse dft along φ is applied to reconstruct images ( )i nφf over all the pipes with radii ir r= , 1, , ri n= � . at the end, the normalized modulus of ( )i nφf sc =e df� � �� � � 1 a sc sc sc n = e e e �� �� � �� 1 an = d d d �� �� � �� 1( ) ( ) rn k k φ φ f f = f �� �� � �� 1( , ) ( , ) m m m sc a sc a sc a n k k ω φ φ ω ω = e e e �� �� � �� , , 1, 1 , 1 , , 1, , ( , ) ( , ) ( , ) ( , ) m r m m m r m sc cd sc cd a n a a sc cd sc cd a n n a n k k k k ω ω φ φ φ φ ω ω ω ω = e e d e e � �� �� �� � � � � �� �� figure 1: illustration of the simulation setup in feko for the case that the defects are on the inner pipe. columbia undergraduate science journal vol. 15, 2021 gao, raisa et. al. 33 ( ) | /i n mφ| f , where m is the maximum of ( ) |i nφ| f for all ir , is plotted versus φ to obtain a 1d image of the defects on the i-th pipe. we call ( ) | /i n mφ| f the normalized image. in [15], inspired by standardized low-resolution brain electromagnetic tomography, the systems of equations in (1) are solved using standardization of the minimum norm. using this concept, the objective function to be minimized is constructed as: (4) where 0α ≥ is a regularization parameter. the detailed solution has been explained in [15]. as discussed earlier, the psf data is collected beforehand for the configuration of the pipes under inspection assuming that they are concentric. a non-zero eccentricity, however, affects the measured data for the inspected pipes leading to errors in the reconstructed images. to evaluate the quality of reconstructed images, we define a reconstruction error (re) parameter as: (5) where , ( )i ideal nφf is the ideal image for which the values are all 0 except being 1 at the true positions of the defects. results to study the effect of eccentricity on imaging of the multiple non-metallic composite pipes, we conduct a study using simulation data provided by altair’s feko software [16] which is a high frequency modeling software. the study was done by 1d scanning and image reconstructions along the azimuthal direction. in order to have a more realistic simulation study, white gaussian noise with signal-to-noise ratio (snr) of 20 db is added to the simulated responses by using the awgn command in matlab. figure 2 illustrates the configuration 2 scj α= − +e df f� � � �� � � � , 1 re ( ) | / ( ) rn i i ideal i n m nφ φ = = −∑ | f f of the imaging setup in feko. we study the performance of the system where the antenna array is placed on the outside of two concentric pipes. there are 13 resonant dipole antennas separated by 10aφ∆ = $ angles along the φ direction. thus, all the antennas are used as receivers except the center element which acts as both transmitter and receiver. the radii of the inner and outer pipes, namely, rout1 and rout2 are 20 mm and 40 mm, respectively, and the thickness of both pipes is d = 2 mm. the pipes have a relative permittivity εr of 2.25 and a tangent loss of 0.0004. the defects have semi-cylindrical shape and their parameters are ld = 1.5d and wd = 0.75d. in addition, the models are simulated with two identical defects on the pipes. the studied scenarios are: (1) both defects on the outer pipe only and (2) both defects on the inner pipe only. the eccentricity parameter, denoted by ecc, represents the distance between the centers of the inner and outer pipes (the outer pipe is assumed to be concentric with the circular path scanned by the antennas known as measurement aperture). for data acquisition, we perform scanning of a circular aperture to get the complex-valued transmission scattering parameters (in microwave, these are the parameters representing the coupling of the transmitter to figure 2: illustration of the simulation setup in feko for the case that the defects are on the inner pipe. columbia undergraduate science journal vol. 15, 2021 gao, raisa et. al. 34 the receiver which here takes into account the field scattered back from the objects as well) by rotating the antennas along the azimuth angle (φ ) from 0° to 360° every 2° (181 grid points) in feko. for each scenario, the simulated responses without the presence of the defects are subtracted from the simulated responses with the presence of the defects to acquire the scattered responses only due to the defects. also, white gaussian noise with snr of 20 db is added to each defect response to imitate real-world measurement data. first, we study the effect of eccentricity when ecc = 0.5 mm (along the x axis) and both defects are on the inner pipe or on the outer pipes at various azimuthal angles from 10φ = ± $ to 170φ = ± $ . after applying near-field holographic imaging as described in the previous section along with the psf data collected for concentric pipes, the values of res are computed for each scenario. figures 3a and 3b show the variations of the computed res versus the angle of defects for the cases that both defects are on the outer pipe and inner pipe, respectively. from both figures, it is observed that there is no clear correlation between the angles of the defects and the values of the res. in general, the error seems to be larger for the defect angles between 20φ = ± $ to 140φ = ± $ . furthermore, the values of res are larger when the defects are on the inner pipe indicating that the image of the inner pipe is more affected by the adverse effects of eccentricity. next, we study the effect of value of eccentricity on the quality of the reconstructed images when the defects are on the outer and inner pipes by visually comparing the quality of the reconstructed images to the ideal images. for this study, the eccentricity parameter ecc varies from 0.1 mm to 0.9 mm with steps of 0.1 mm and we choose constant angles of 170φ = ± $ for the defects one time when they are on the outer pipe and another time when the defects are on the inner pipes. figure 4 shows the reconstructed images of the two pipes when the eccentricity parameter is 0.1 mm. figures 4a and 4b show the images when the defects are on the outer and inner pipes respectively. in general, we notice that the image deteriorates is more for the inner pipe than the outer pipe. the reconstructed image in figure 4a for which the defects are on the outer pipe clearly shows the figure 3: variation of computed re when ecc = 0.5 mm and the angular positions of two identical defects are varying from 10φ = ± $ to 170φ = ± $ : (a) defects are on outer pipe (b) defects are on inner pipe. a b columbia undergraduate science journal vol. 15, 2021 gao, raisa et. al. 35 presence of the defects and it is close to the ideal image. however, the image for the inner pipe in figure 4a shows artifacts around 0.3 level. the reconstructed images in figure 4b in which the defects are on the inner pipe, still show the presence of the defects on the inner pipes but again contains large artifacts with maximum of 0.5 level. in this case, the image on the outer pipe also contains some small level (around 0.1 level) of artifacts with some shadows of the defects on the inner pipe. in the following, we demonstrate that as we continue to increase the value of eccentricity parameter in this study, the quality of the reconstruction images deteriorates significantly. figure 5 shows the reconstructed images when figure 4: reconstructed 1d images when the defects are at 170φ = ± $ and eccentricity = 0.1 mm for: (a) defects on the outer pipe (b) defects on the inner pipe. figure 5: reconstructed 1d images when the defects are at 170φ = ± $ and eccentricity = 0.5 mm for: (a) defects on the outer pipe (b) defects on the inner pipe. a a b b columbia undergraduate science journal vol. 15, 2021 gao, raisa et. al. 36 the value of ecc is increased to 0.5 mm. compared to figure 4, images in figure 5 are more distorted. as expected, the increase in the value of ecc leads to larger image reconstruction errors. next, we increase the value of ecc even further. figure 6 displays the high deterioration of the images when the value of ecc is 0.9 mm. from these figures, it can be easily deduced that the reconstructed image is far off from the ideal image due to high error caused by eccentricity. as a final step, we compute the variation of res as the value of ecc increases and when the defects are at on the outer pipe or on the inner pipe. figure 7a shows this variation when the defects are on the outer pipe. it is observed that the value of re increases sharply as ecc increases. a similar trend is observed in figure 7b when the defects are on the inner pipe. figure 6: reconstructed 1d images when the defects are at 170φ = ± $ and eccentricity = 0.9 mm for: (a) defects on the outer pipe (b) defects on the inner pipe. figure 7: variation of computed re vs. eccentricity when defects are at 170φ = ± $ and both defects are on (a) outer pipe (b) inner pipe. a b a b columbia undergraduate science journal vol. 15, 2021 gao, raisa et. al. 37 discussion & conclusion in this paper, we studied the effect of eccentricity of the pipes on the results of the holographic microwave imaging of multiple non-metallic pipes. microwave imaging is a non-contact method that can be used for inspection of multiple pipes and it is also safe due to the use of low-level microwave power. in general, the results indicate that the quality of the reconstructed images is highly sensitive to non-zero eccentricity values such that even an eccentricity of 0.5 mm imposes image quality deteriorations. besides, it was observed that the images of the defects on the inner pipes would be affected more seriously by non-zero eccentricity effects. although we employed simulated results from altair feko software, we applied additive white gaussian noise to the simulated responses to mimic real-world measurements and have a more realistic study. here, the study was conducted with two scenarios: (1) changing the angles of the defects while the eccentricity value is fixed and (2) changing the value of eccentricity while the positions of defects are fixed. it is worth noting that the effect of other important parameters for the considered microwave imaging setup such as thickness, radius, and permittivity of the pipes as well as angular separation of the antennas have been already studied in [15] and have been excluded here. due to serious adverse effects of eccentricity, we plan to develop a technique to estimate the value of unknown eccentricity parameter and then reduce the effect of that on the reconstructed images. ultimately, the goal is to develop a robust tool for ndt of non-metallic pipes that promote using these components in various industries. author information corresponding author *email: ygao21@nyit.edu author contributions yuki gao and noshin raisa have contributed in performing the parametric analysis of the eccentricity effect. reza k. amineh has provided the original feko simulation models and matlab codes for applying the nearfield holographic imaging. all authors have contributed in preparation of this manuscript. funding sources u.s. national science foundation (nsf) under award 1920098 and the new york institute of technology’s institutional support for research and creativity (isrc) grants competing interests the authors declare no competing financial and non-financial interests. acknowledgements the authors thank new york institute of technology’s undergraduate research and entrepreneurship program (urep). abbreviations cd – calibration defect cfrp – carbon fiber reinforced plastic dft – discrete fourier transform dtft – discrete time fourier transform frp – fiber reinforced plastic gre – glass reinforced epoxy resin hdpe – high density polyethylene ndt – non-destructive testing psf – point-spread function pvc – polyvinyl chloride re – reconstruction error rej –rubber expansion joint sar – synthetic aperture radar references [1] l. tong, a. p. mouritz, m. k. bannister, 3d fiber reinforced polymer composites. 1st ed. (2002). columbia undergraduate science journal vol. 15, 2021 gao, raisa et. al. 38 [2] k. murphy, d. lowe, evaluation of a novel microwave based ndt inspection method for polyethylene joints. in proc. asme pressure vessels piping conf., baltimore, md, u.s.a. 321-327 (2002). [3] x. w. zhu, j. p. pan, l. j. tan, microwave scan inspection of hdpe piping thermal fusion welds for lack of fusion defects. appl. mech. mater. 333-335, 1523-1528 (2013). [4] r. pincu, “digital radiography and its advantages in field ndt inspections today” 17th world conf. on nondestructive testing, shanghai, china (2008). [5] r. zoughi, microwave non-destructive testing and evaluation. (2000). [6] s. kharkovsky, r. zoughi, microwave and millimeter wave nondestructive testing and evaluation overview and recent advances. ieee instr. and meas. mag., 10(2), 26-38 (2007). [7] d. m. sheen, d. l. mcmakin, t. e. hall, three-dimensional millimeter-wave imaging for concealed weapon detection. ieee trans. microw. theory techn., 49(9), 1581-1592 (2001). [8] m. t. ghasr, k. ying, r. zoughi, 3d millimeter wave imaging of vertical cracks and its application for the inspection of hdpe pipes. aip conf. proc., 1581(1), 1531–1536 (2014). [9] r. k. amineh, n. k. nikolova, m. ravan, real-time three-dimensional imaging of dielectric bodies using microwave/millimeter wave holography. wiley & ieee press, isbn: 978-1-119-53886-8, (2019). [10] r. k. amineh, m. ravan, h. wu, a. kasturi, three-dimensional holographic imaging using data collected over cylindrical apertures. microwave and optical technology letters, 61(4), 907–911 (2019). [11] r. k. amineh, j. mccombe, a. khalatpour, n. k. nikolova, microwave holography using point-spread functions meas-ured with calibration objects. ieee trans. on inst. and meas., 64(2), 403−417 (2015). [12] l. e. s. martin, a. e. fouda, r. k. amineh, i. capoglu, b. donderici, s. s. roy, f. hill, new high-definition frequency tool for tubing and multiple casing corrosion detection. spe abu dhabi int. petroleum exhibition & conference, (2017). [13] y. renardy, d. d. joseph, couette flow of two fluids between concentric pipes. j. of fluid mechanics, 150, 381–394 (1985). [14] r. k. amineh, m. ravan, r. sharma, nondestructive testing of nonmetallic pipes using wideband microwave measurements. ieee trans. microwave theory and techniques, 65w(5), 1763-1772 (2020). [15] h. wu, m. ravan, r. sharma, j. patel, r. k. amineh, microwave holographic imaging of nonmetallic concentric pipes. ieee trans. instrumentation & measurement, 69(10), 75947605 (2020). [16] altair feko, https://www.altair.com/feko/ template for electronic submission to acs journals 17 © 2023 datta, neelabh. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. molecular mechanisms and clinical features of huntington disease: a fatal neurodegenerative disorder with autosomal dominant inheritance neelabh datta1 1department of biochemistry, asutosh college (affiliated to university of calcutta), 92, shyama prasad mukherjee rd, kolkata, west bengal, 700025, india keywords: huntington’s disease, neurogenetics, molecular mechanisms, pathophysiology, therapy abstract: huntington disease (hd) is a fatal genetic disorder that affects the movement and cognition of affected individuals. it is inherited in an autosomal dominant manner, meaning that each child of a parent with hd has a 50% chance of inheriting the mutated gene. the mutation involves an expansion of a trinucleotide repeat (cag) in the hd gene, which is located on the short arm of chromosome 4p16.3. the hd gene encodes a protein called huntingtin, which has an unknown function. the number of cag repeats determines the severity and onset of the disease. normal individuals have 26 or fewer repeats, while hd patients have 40 or more repeats. individuals with 27 to 35 repeats do not develop hd, but they can pass on the mutation to their offspring, especially if the mutation is inherited from the father. individuals with 36 to 39 repeats may or may not develop hd, depending on other factors. the more cag repeats, the earlier the symptoms appear. hd is the most extensively studied neurodegenerative disorder with a genetic cause. there are genetic tests available to diagnose hd and to predict the risk of developing hd in asymptomatic individuals. there are also prenatal and preimplantation tests to prevent the transmission of hd to the next generation. hd is characterized by involuntary movements called chorea, which affect all muscles and impair all psychomotor functions. hd patients also suffer from cognitive decline and psychiatric symptoms, such as mood disorders and social changes. these symptoms are chronic and progressive, leading to complete dependence and death. chorea can also be caused by other conditions, such as metabolic disorders or drug-induced side effects. neuroimaging techniques, such as mr imaging, fluorodeoxyglucose positron emission tomography (fdg-pet), mr spectroscopy, and diffusion tensor imaging, can help to diagnose hd and monitor its progression. the pathophysiology of hd involves the loss of neurons and the dysfunction of neurotransmitter systems, especially the dopaminergic system. there is no cure for hd, but there are treatments to manage the symptoms and to improve the quality of life of hd patients. these include pharmacological interventions, such as dopamine receptor antagonists or depleters, and nonpharmacological interventions, such as psychological and social support. hd is a devastating disease that poses many challenges for patients, families, and healthcare providers. there is hope that genetargeted therapies will be developed in the near future to stop or slow down the disease process. introduction huntington disease (hd) is a fatal neurodegenerative disorder that affects the cns and causes motor, cognitive, and behavioural problems in the patients [1]. the disease is inherited in an autosomal dominant manner and is caused by a mutation in the htt gene that results in an expansion of the polyq domain of the htt protein beyond 36 glutamines [3]. the brain cells of hd patients have misfolded polyq htt proteins that form columbia undergraduate science journal vol. 17, 2023 datta 18 clumps, unlike the normal individuals who have diffuse localization of htt [12]. the length of the polyq expansion determines how fast the aggregation occurs [13]. the normal function of the cell to recycle and degrade proteins and produce energy is disrupted by the misfolded polyq htt, which also binds to other proteins such as creb binding protein and depletes them from the cell [14]. this leads to a toxic effect of the mutant htt, but it also affects the normal function of the wildtype htt, which does not have the polyq expansion [2, 15]. therefore, any treatment for hd must consider both the gain-of-function and the loss-of-function effects of the mutant htt. the age of onset of hd depends on the length of the polyq expansion, with longer expansions causing earlier and more severe symptoms [4]. the polyq length explains 60-70% of the variation in the age of onset, while the rest is influenced by environmental and genetic factors [5]. the main feature of hd pathology is the death of the gabaergic msn in the striatum, which is a part of the brain that controls movement and cognition [2, 6]. the death of the msn is accompanied by inflammation and activation of glial cells, which are the support cells of the cns [2]. microglia, which are the immune cells of the cns, are activated in both early and late stages of hd and cause damage to the neurons in the striatum and cortex [7]. astrocytes and oligodendroglia, which are the cells that provide nutrients and insulation to the neurons, are also increased in hd brains, especially in the globus pallidus and the white matter surrounding it [8]. hd patients also show a significant loss of brain volume in different regions, such as the cerebral cortex, the telencephalic white matter, the putamen, and the caudate nucleus [2, 9, 11]. the loss of brain volume can be detected even before the symptoms of hd appear, indicating an early degeneration of the brain [10]. hd is a disease that affects the whole brain and causes progressive deterioration of the patients' functions and quality of life. there is no cure for hd, and the current treatments are only palliative and symptomatic. structure of the huntingtin protein the huntingtin (htt) protein is a molecule that is found in humans and other vertebrates, and it has a very similar structure and function among them [2,16]. it is a big protein, with a molecular weight of about 350 kda, and it has a shape that is long and flexible [17,18]. the protein is made up of many repeated units called heat (huntingtin, elongator factor3, pr65/a regulatory subunits of pp2a, and tor1), which help the protein to interact with other proteins and to form complex structures [2,24]. one of the special features of the htt protein is that it has a region near the beginning of the protein that contains many glutamine residues, which are amino acids that have a nitrogen atom in their side chain [2]. this region is called the polyq tract, and it starts from the 18th amino acid in the protein [2]. the number of glutamines in the polyq tract can vary a lot, and this can affect how the protein works and how easily it dissolves in water [27,30]. in most people, the polyq tract has around 20 glutamines, but in some people, it can have more than 40 glutamines, and this can cause a brain disease called huntington's disease (hd) [20,21,22]. the polyq tract is surrounded by another region that has many proline residues, which are amino acids that have a ring-shaped side chain [31]. this region is called the polyp tract, and it helps the protein to stay dissolved in water and to interact with proteins that are involved in moving things inside the cell [31,32]. the polyq tract is not present in all animals that have a similar protein to htt, but it is very important for the brain functions of animals that have it, columbia undergraduate science journal vol. 17, 2023 datta 19 as shown by experiments with mice that do not have the polyq tract [2,29]. the polyq tract can also form structures that look like zippers, and these structures can bind to other parts of the protein or to other proteins that have similar structures [2,28]. the htt protein is therefore a very complex and versatile protein that is involved in many different processes in the cell. the way that hd is inherited and how it affects different people is related to the way that the polyq tract can change in size and cause problems, which is caused by an increase of the number of repeats of three nucleic acids (c, a, and g) in the first exon of the hd gene, which is located on chromosome 4p16.32 [19]. htt has various features that allow it to move between the nucleus and the cytoplasm of the cell [2]. it has a nuclear export signal (nes) at the end of its c-terminal domain, which enables it to exit the nucleus [33]. it also has a domain at the beginning of its nterminal domain, which consists of 18 amino acids and interacts with tpr, a protein that is part of the nuclear pore complex and facilitates nuclear import [33]. this domain also forms a membrane-binding domain that has an amphipathic alpha helical structure and can reversibly bind to different types of vesicles, such as those derived from the endoplasmic reticulum (er), endosomes, and autophagosomes [34]. this domain is essential for the normal function of htt, as mutations or deletions in this region cause htt to accumulate in the nucleus and induce cellular toxicity [33]. htt is also subject to proteolytic cleavage by various enzymes, such as caspases and calpains, which are conserved among higher vertebrates [35]. these enzymes generate fragments of htt that are found in the nucleus, but their role is unclear. the proteolysis of htt is influenced by the cellular context, as it is increased in diseased brains and more selective for the fragments that have the nand c-terminal domains, especially in the striatum [2,36]. another enzyme that can produce n-terminal fragments of htt is cathepsin, which belongs to the lysosomal degradation pathway [37]. htt can also undergo different types of posttranslational modifications in its n-terminal region, such as ubiquitination, sumoylation, and phosphorylation by kinases like akt, erk1, and cdk5 [2,38]. the phosphorylation level of htt is regulated by s/t phosphatases pp1 and pp2a [39]. furthermore, htt can be palmitoylated in its n-terminal region through the interaction with huntingtin-interacting protein 14 (hip 14) [40]. palmitoylation is a mechanism that is used by several proteins that are involved in vesicle trafficking to maintain their proximity to the plasma membrane [2]. htt is highly expressed in the human brain and testes [41]. in the brain, it is present in both neurons and glial cells [42]. htt has a complex subcellular localization pattern, which may depend on its conformational state, as different antibodies that recognize different epitopes within the protein show different subcellular staining profiles [43]. htt is not only localized in the nucleus, er, and golgi complex, but also in the axons and synapses of neurons [41], where it is associated with microtubules, caveolae, and synaptosomes [41]. significance and function of the huntingtin protein the function and role of the huntingtin protein is still unclear, despite the fact that its genetic location was identified a long time ago. some studies have proposed that huntingtin is involved in regulating gene expression and transporting molecules inside the cell, based on its presence in both the cytoplasm and the nucleus and its frequent interactions with other proteins [44]. the huntingtin protein also plays a role in neuronal development, synaptic transmission, axonal transport, and autophagy [51]. the huntingtin columbia undergraduate science journal vol. 17, 2023 datta 20 gene has a segment of dna that repeats the sequence cag, which codes for the amino acid glutamine. in normal huntingtin proteins, this segment has less than 27 repeats, and people with 27 to 35 repeats (called transitional alleles) [45] do not develop the disease, but they may pass on longer repeats to their offspring. people with 36 or more repeats will develop the disease [46]. the number of cag repeats tends to increase in each generation, which means that the disease becomes more severe and appears earlier in life. this phenomenon is called "anticipation" [47]. the age of onset of the disease is inversely related to the number of cag repeats. therefore, people with juvenileand infantile-onset hd have more repeats than their parents and show symptoms earlier [48]. the huntingtin protein is expressed in many different types of cells, but it is especially abundant in the brain and the testes, and to a lesser extent, in the liver and the lungs [49]. it has a protective role against cell death, but when it is mutated or underexpressed, it causes early apoptosis and dysfunction. the mutation that causes hd is an expansion of the cag repeats, which results in a longer stretch of glutamine in the protein [46]. this leads to the formation of abnormal aggregates of the protein in the nucleus and the cytoplasm of the cell, which disrupts the normal balance of the cell and triggers apoptosis [50]. the aggregates also interfere with the normal function of huntingtin and its interacting partners, leading to neuronal degeneration, inflammation, oxidative stress, mitochondrial dysfunction, and impaired autophagy[52]. the most affected brain region is the striatum, which is involved in motor control and cognition, followed by the cortex, which is responsible for higher cognitive functions [53]. the symptoms of hd include chorea, dystonia, rigidity, bradykinesia, cognitive impairment, and psychiatric disturbances [54]. there is no cure for hd, but some treatments can help to manage the symptoms and improve the quality of life of the patients [55]. clinical indicators of huntington disease hd affects the different aspects of behaviour, cognition, and motor function of humans [51]. the disease has three subtypes, depending on the age of onset: adult-onset, juvenile-onset, and infantile-onset. the most common subtype is adult-onset hd, which usually manifests in the fourth or fifth decade of life. patients with adult-onset hd experience a range of behavioural symptoms, such as irritability, agitation, loss of inhibition, and aggression, which often precede the motor symptoms [46,52]. the motor symptoms include involuntary movements (chorea), which become less prominent as the disease progresses and are replaced by rigidity and abnormal muscle contractions (dyskinesia). patients also lose their ability to maintain a sustained voluntary muscle contraction and their fine and gross motor skills, leading to severe disability and dependence [46,52]. the cognitive symptoms include impairment of memory, executive function, language, and visuospatial skills, which worsen over time and result in dementia [51]. the disease course of adult-onset hd is typically 15–20 years from the onset of symptoms to death [51]. the other two subtypes, juvenile-onset and infantile-onset hd, are much rarer and account for about 10% of hd cases. they are characterized by an earlier onset of symptoms, usually before the age of 20, and a more rapid progression of the disease [46,53]. the main features of these subtypes are rigidity, dyskinesia, and cognitive decline, with little or no chorea [46,53]. these patients often show signs of motor deterioration and poor academic performance before they are diagnosed [46,53]. columbia undergraduate science journal vol. 17, 2023 datta 21 genetic modifiers of huntington’s disease the discovery of the hd defect [54] prompted many researchers to investigate the role of hd genes that were chosen based on their functional relevance. however, a new approach emerged in the early 21st century, which was based on two major advances in human genetics: firstly, the identification of common genetic variations across the genome, known as single nucleotide polymorphisms (snps), and secondly, the development of oligonucleotide array technology, which enabled the simultaneous genotyping of hundreds of thousands to millions of snps for unbiased genetic studies [55]. this approach, called genome-wide association analysis (gwa), allowed the researchers to scan the entire genome for genetic factors that influence the hd phenotype, without relying on prior knowledge of gene function. to apply this approach to hd with sufficient statistical power, three additional requirements had to be met: firstly, the availability of genomic dna from a large number of hd subjects for genotyping; secondly, the definition of a robust phenotype that accounted for the effects of the cag repeat size, which is the main determinant of the hd phenotype; and thirdly, the exclusion of any potential modifier factors that are linked to htt and act in cis to modify the effect of the mutation, as these factors would confound the genome-wide search [55]. statistically assessing the relationship between the length of inherited cag repeats and the age of motor onset provided a vigorous hd phenotype that accounted for the effect of cag repeats. due to the danger of including inexplicably influential outliers, the analysis was limited to cag repeat lengths typical of adult-onset (40 to 53–55) and sufficient representation of subjects to guarantee consistent results [56]. based on their inherited cag repeat lengths, >90% of hd subjects met these criteria, allowing a standard curve to be produced that relates cag repeat size to average age-at-onset. in the absence of the effects of the cag repeat size, an evaluation of this expected age-at-onset with the individual's observed age-at-onset provided the phenotype for analysis of genetic effects on onset [57]. it was fundamentally a matter of subtracting the expected age-at-onset from the observed age-at-onset to obtain the test phenotype, or residual ageat-onset, which was either a positive or negative number of years based on whether the subject's onset was later or earlier than anticipated. it was possible to test whether genetic variations at the htt locus other than the cag repeat size affect age-at-onset by using residual age-at-onset as a relevant hd phenotype and several thousand unrelated hd subjects [55]. in order to examine this, common single nucleotide polymorphisms (snps) were compiled across the gene and defined as haplotypes (i.e. the linear array of alleles at multiple snps along the chromosome, conducted as a physically linked set to progeny—basically a digital fingerprint for the htt region) [55]. in addition, expanded cag alleles associated with hd were found in multiple haplotypes, indicating that multiple independent ancestral hd cag expansion mutations contributed to the contemporary population of hd individuals [58]. the most common haplotypes, representing more than 83% of hd subjects, were not related with differences in onset age, which suggests that genetic factors usually act in transfer through genes detached from htt [58]. accordingly, hd is viewed as a prototypical autosomal dominant genetic disorder based on whether it is passed on to progeny (or not), but the timing of disease onset is actually polygenic, determined by the combination of cag repeats and other genetic factors. columbia undergraduate science journal vol. 17, 2023 datta 22 factors that play a role in the development of hd 1. mitochondrial disorder in early studies, functional abnormalities in mitochondria were discovered, indicating that mitochondrial dysfunction plays an essential role in the pathogenesis of hd and this is usually seen in the early disease process. the caudate and to a lesser extent the cortex of post-mortem hd brains have succinate dehydrogenase deficiency, a component of both the krebs cycle and the electron transport chain's complex ii. as compared to levels in matched control brains, hd brains demonstrated a significant decrease in complex ii activity in the caudate nucleus (roughly 50%) [67]. in addition to reductions in complex ii activity, complex iii activity in the caudate and putamen, as well as complex iv activity in the putamen, have decreased as well [67]. in spite of this, since most of these patients suffered from advanced neuropathy, including severe striatal atrophy (pathological grades 3 and 4 of hd), mutations in mitochondrial sources (i.e., glial, neuronal, etc.) may have occurred [59]. 2. oxidative stress due to ros reactive oxygen species (ros) are produced in excess in the body, which results in oxidative stress when the body is unable to detoxify them and repair the damage they cause. animal models with hd showed increased levels of malondialdehyde, 8hydroxydeoxyguanosine, 3-nitrotyrosine, and heme oxygenase oxidative damage products, and free radicals in the areas of degeneration in the brain of hd affected individuals which propose that oxidative stress is connected with the disease, either as a primary event or a secondary component of the cascade processes of cell death [67]. there is ample evidence that oxidative damage subsidizes significantly to the pathogenesis of neurodegenerative diseases such as hd [60]. 3. apoptosis there is a link between the pathogenic mechanism of apoptosis and chronic neurodegenerative diseases like hd [61]. caspases are cysteine-dependent, aspartatespecific proteases that initiate and execute apoptosis. a transcriptional up regulation of caspase-1, caspase-3, and caspase-9 in hd patients as well as in animal models of hd has been reported [62]. amyloidogenic mutant huntingtin (mhtt) has been confirmed to induce apoptosis in hd patients [63]. 4. neuroinflammation by discharging cell mediators that combat foreign substances and prevent infections, the inflammatory process safeguards our bodies from harm and disease. neuroinflammation does not directly correlate with hd progression, despite inflammatory processes being evidently confirmed in its pathophysiology. post-mortem studies of degenerating neurons in hd have discovered high levels of activated microglia and macrophages as well as elevated levels of il-6, il1, and tnfin the plasma and striatum of hd patients [67]. microglial cells may identify the pathogenic mhtt aggregates as foreign substances, resulting in neuro-inflammation [64]. 5. neurotoxicity disproportionate glutamate neurotransmission leads to excitotoxic neuronal death, which is supplemented by insistent intracellular calcium level elevation [65]. as nmda (n-methyl-d-aspartic acid) receptors are over activated by excited amino acids, free radicals are formed and mitochondrial percolumbia undergraduate science journal vol. 17, 2023 datta 23 meability transition pores are unlocked, both of which are lethal. the role of neurotoxicity has been documented as significant since the undeviating injection of acids such as qa and kainic acid causes neuro-degeneration of gabaergic msn in hd [66]. advances in huntington's disease therapy the htt gene mutations that cause hd are still not fully understood, but research on the molecular mechanisms behind them is very promising and could lead to a cure. currently, there are no neuroprotective therapies that can prevent or slow down the disease, and the only treatments available are symptomatic [67]. one of the main causes of hd is the toxicity of mhtt, the mutant form of the htt protein, which is produced by the mutated gene. therefore, reducing the expression of mhtt, either by lowering the levels of htt mrna or the protein itself, is a potential strategy to treat hd [68]. some studies have suggested that gene-silencing techniques that target the cag repeats in the htt gene, which are responsible for the mutation, could improve the functional, motor and cognitive outcomes of hd patients, but not their weight loss [67, 69]. these techniques involve using different types of dnabinding elements, such as zinc-finger proteins, nucleases, epigenetic modulators, or transcription factors, to either block, disrupt, or correct the mutant gene. for example, zinc-finger transcriptional repressors can bind to the dna and prevent its transcription, while zinc-finger nucleases can cut and edit the dna [67,70]. another example of a genome editing technique is crispr/cas9, which can also target and modify the htt gene. these approaches have the advantage of permanently correcting the cag expansion that causes hd. another way of reducing mhtt expression is by using antisense oligonucleotides (asos), which are synthetic molecules that bind to the htt mrna and trigger its degradation by an enzyme called rnase h1 [71]. asos can reach the central nervous system without needing a viral or lipid carrier, and they are easy to develop [67]. a clinical trial by tabrizi et al [72] used an aso called ionis-httrx to lower the levels of mhtt in the cerebrospinal fluid of 34 hd patients, who received doses ranging from 10 to 120 mg, and compared them with a placebo group. the results showed that the mhtt reduction was dose-dependent [72]. a similar method of post-transcriptional gene suppression is rna interference (rnai), which uses noncoding double-stranded rna sequences to silence specific genes. rnai can be achieved by using different types of rna molecules, such as sirnas, shrnas, or artificial mirnas, which have been shown to reduce the hd symptoms [67]. rnai is also a promising technique for many other diseases. one of the main approaches for treating hd is to reduce the synthesis of the mutant htt (mhtt) protein, which forms toxic aggregates in the brain. this can be achieved by using rna interference (rnai) techniques, which involve the use of small rna molecules that bind to the mhtt mrna and prevent its translation. several types of rna molecules have been used for rnai, such as short hairpin rna (shrna), small interfering rna (sirna), and microrna (mirna) [67]. these molecules have been delivered to the brain of hd animal models using viral vectors, such as adeno-associated virus (aav), which contain enhancers and promoters to drive the expression of the rna molecules. the first trials of rnai for hd were performed in rodents two decades ago. it is seen that shrna targeting mhtt reduced its synthesis and prevented the formation of inclusions, gait deficits, and rotarod dysfunction in mice, in recent studies [67]. similarly, columbia undergraduate science journal vol. 17, 2023 datta 24 sirna injected into the mouse striatum prolonged the survival of striatal neurons, reduced mhtt aggregates, and prevented motor dysfunction [73]. these results were replicated in multiple animal systems, such as rats, monkeys, and sheep [74]. a recent study used a single-stranded sirna (sssirna) for rnai and achieved a selective decrease of cag-expanded htt protein in various regions of the mouse brain [75]. another type of rna molecule that has been used for the suppression of mhtt is mirna, which is a natural regulator of gene expression. mirnas have been shown to have promising effects in genetically modified mice with hd; for example, one study reported that mirna-mediated knockdown of mhtt prevented regional cortical and striatal atrophy and reduced weight loss [76]. most of the rnai techniques do not completely eliminate the production of mhtt, but only reduce it to a certain extent. therefore, another possible therapeutic approach is to overexpress the wild-type htt, which may have a protective role against the toxic effects of mhtt. early trials of this strategy showed that inserting the wild-type htt into mammalian cells that expressed mhtt reduced cell death [77]. several natural and synthetic compounds have been suggested as potential candidates for the treatment of hd and other neurodegenerative disorders. one of the most widely studied compounds for hd is tetrabenazine (tbz), which is an inhibitor of the vesicular monoamine transporter 2 (vmat2) that blocks the uptake of dopamine into vesicles [67]. tbz has been shown to exert antichorea effects in patients with hd and was the first approved drug for the disease [78]. however, tbz has some limitations, such as low bioavailability and adverse effects. therefore, studies have been conducted to optimize the drug delivery and bioavailability of tbz using nanotechnology techniques, such as nanoparticles and nanocapsules [79]. another class of compounds that may have beneficial effects for hd are flavonoids, which are natural polyphenolic compounds found in plants. flavonoids have been shown to reduce cellular stress and exert anti-inflammatory and antiapoptotic effects in the cell [80]. some examples of flavonoids that have been tested for hd are resveratrol, curcumin, and quercetin. marine compounds have been proposed as potential sources of novel drugs for hd and other neurodegenerative diseases. marine compounds have diverse chemical structures and biological activities, such as antioxidant, anti-inflammatory, and anti-apoptotic properties. some examples of marine compounds that have been investigated for hd are fucoidan, xyloketal b, fucoxanthin, and cerebrosides [67, 81]. a recent study suggested that pridopidine, a dopamine stabilizer, may be a promising drug for hd symptoms. pridopidine acts on the sigma-1 receptor, which is involved in the regulation of calcium homeostasis, mitochondrial function, and neuroprotection. pridopidine has been shown to improve motor and cognitive functions in hd animal models and patients [82]. the stage of huntington's disease (hd) is related to the amount of dopamine in the central nervous system, since dopaminergic conduction disorders are the main cause of hd. pridopidine, a drug that protects nerve cells from degeneration, has shown promising results in animal models [83]. drug treatment for hd has the advantage of being based on well-studied active compounds that are effective and tolerable for other similar neurodegenerative diseases [67]. this makes it easier to personalize the medication according to the patient's diagnostics. another potential treatment for hd is cell replacement therapy using stem cells, which could reduce the symptoms of the disease [84]. moreover, exercise and physical activity columbia undergraduate science journal vol. 17, 2023 datta 25 have been reported to have positive effects on the motor function, gait speed, balance, and social well-being of hd patients [67, 85]. therefore, exercise could be a complementary therapy for hd. a novel approach to target the underlying cause of hd is the use of monoclonal antibodies that bind to the mutant huntingtin (mhtt) protein and lower its concentration in the cell. this could prevent the mhtt from spreading and causing damage to the brain [84]. gene therapy: a possible treatment of hd in the future research into gene therapy has led to the most exhilarating and favorable advances in hd research. a potential therapy for dominant genetic disorders, silencing mutant genes can offer major benefits. it is generally believed that gene therapy could have a twofold effect: (i) restoring function to non-dead, but dysfunctional neuronal circuits, and (ii) protecting against disease progression [86]. as a matter of fact, gene therapy will not be used to treat disease, but to prevent it to eliminate symptoms entirely. molecular gene therapy targets the transcription and translation processes of dna into mrna (messenger rna) by a process called transcription and the synthesis of proteins using the information in mrna (a process called translation) [87]. antisense oligonucleotides (asos), zinc finger proteins, and rna interference techniques are three common methods for gene-silencing [88]. zinc finger proteins suppress transcription, antisense oligonucleotides suppress translation of mhtt, and rna interference blocks protein translation [89] in large neuroimaging studies performed during prehd, measurable measures of brain regions such as the striatum have been found to be excellent biomarkers of disease progression and will be useful in future gene therapy trials [90]. conclusion research indicates that the pathology of huntington's disease (hd) may be significantly influenced by the loss of function in the normal htt protein, despite the disease primarily being attributed to a toxic gain of function caused by the expansion of polyglutamine (polyq) sequences. the htt protein is known to interact with various effector proteins and is involved in critical cellular processes such as transcription and intracellular trafficking. these interactions and processes are vital for the proper processing and localization of numerous proteins. consequently, a deficiency in htt function could potentially have a more extensive impact on cellular physiology than previously comprehended. given the variety of cellular disruptions caused by this deficiency, it appears that certain neurons may be more resilient than others. identifying which altered physiological processes most significantly contribute to the progression of hd will be imperative for the development of effective therapeutic interventions. as current therapeutic strategies under development aim to reduce htt levels, it is also essential to ascertain the degree to which cells can tolerate a reduction in normal htt expression. in the field of human disease research, utilizing human subjects is considered the benchmark for validating experimental findings related to disease pathogenesis. this approach is instrumental in elucidating the mechanisms underlying disease onset and in exploring viable treatment modalities for genetic disorders. take hd, for instance, a genetic condition precipitated by the incessant elongation of the cag trinucleotide repeat within the htt gene. the extent of this repeat sequence is a determinant of the age at which hd symptoms manifest, as the accumulation of toxicity from the expanded repeat sequence reaches a critical threshold, instigating cellular damage processes that culcolumbia undergraduate science journal vol. 17, 2023 datta 26 minate in neuronal impairment and the onset of the disease. however, the length of the cag repeat is not static; it is subject to somatic expansion influenced by both genetic and environmental factors. this variability leads to differential susceptibility, damage, and toxicity across various cell types, resulting in diverse phenotypic expressions among individuals. thus, the dna repair mechanisms that govern the length of the cag repeat are pivotal targets for therapeutic interventions aimed at delaying or averting the onset of hd and similar trinucleotide repeat disorders. nonetheless, these dna repair mechanisms are modulated by a plethora of genes that serve as modifiers of hd onset, complicating the prediction of disease trajectory for individual patients. furthermore, these modifier genes may influence not only dna repair pathways but also other processes that initiate cellular damage and toxicity in hd. another dimension of hd that remains elusive is the normal function of the htt protein, which is implicated in synaptic vesicle trafficking and endosomal signaling— processes that are crucial for neuronal development and functionality. the precise mechanisms by which htt facilitates these synaptic functions, and how they are disrupted by the mutant form of htt in hdafflicted neurons, are areas that warrant further investigation. a more thorough characterization of the normal roles of htt could unveil new therapeutic avenues and foster a more holistic understanding of hd. . author information corresponding author neelabh datta neelabhdatta@gmail.com author contributions text. references [1] myers, r. h., macdonald, m. e., koroshetz, w. j., duyao, m. p., ambrose, c. m., taylor, s. a., barnes, g., srinidhi, j., lin, c. s., & whaley, w. l. (1993). de novo expansion of a (cag)n repeat in sporadic huntington's disease. nature genetics, 5(2), 168–173. https://doi.org/10.1038/ng1093168. [2] schulte, j., & littleton, j. t. (2011). the biological function of the huntingtin protein and its relevance to huntington's disease pathology. current trends in neurology, 5, 65–78. [3] myers r. h. (2004). huntington's disease genetics. neurorx : the journal of the american society for experimental neurotherapeutics, 1(2), 255–262. https://doi.org/10.1602/neurorx.1.2.255. [4] squitieri, f., cannella, m., & simonelli, m. (2002). cag mutation effect on rate of progression in huntington's disease. neurological sciences : official journal of the italian neurological society and of the italian society of clinical neurophysiology, 23 suppl 2, s107–s108. https://doi.org/10.1007/s100720200092 [5] rubinsztein, d. c., leggo, j., chiano, m., korn, s., dodge, a., norbury, g., rosser, e., & craufurd, d. (1997). homozygotes and heterozygotes for ciliary neurotrophic factor null alleles do not show earlier onset of huntington's disease. neurology, 49(3), 890– 892. https://doi.org/10.1212/wnl.49.3.890. [6] albin, r. l., reiner, a., anderson, k. d., dure, l. s., 4th, handelin, b., balfour, r., whetsell, w. o., jr, penney, j. b., & young, a. b. (1992). preferential loss of striatoexternal pallidal projection neurons in presymptomatic huntington's disease. annals of neurology, 31(4), 425–430. https://doi.org/10.1002/ana.410310412. columbia undergraduate science journal vol. 17, 2023 datta 27 [7] sapp, e., kegel, k. b., aronin, n., hashikawa, t., uchiyama, y., tohyama, k., bhide, p. g., vonsattel, j. p., & difiglia, m. (2001). early and progressive accumulation of reactive microglia in the huntington disease brain. journal of neuropathology and experimental neurology, 60(2), 161–172. https://doi.org/10.1093/jnen/60.2.161. [8] vonsattel, j. p., myers, r. h., stevens, t. j., ferrante, r. j., bird, e. d., & richardson, e. p., jr (1985). neuropathological classification of huntington's disease. journal of neuropathology and experimental neurology, 44(6), 559–577. https://doi.org/10.1097/00005072198511000-00003. [9] nopoulos, p. c., aylward, e. h., ross, c. a., johnson, h. j., magnotta, v. a., juhl, a. r., pierson, r. k., mills, j., langbehn, d. r., paulsen, j. s., & predict-hd investigators coordinators of huntington study group (hsg) (2010). cerebral cortex structure in prodromal huntington disease. neurobiology of disease, 40(3), 544–554. https://doi.org/10.1016/j.nbd.2010.07.014. [10] squitieri, f., cannella, m., simonelli, m., sassone, j., martino, t., venditti, e., ciammola, a., colonnese, c., frati, l., & ciarmiello, a. (2009). distinct brain volume changes correlating with clinical stage, disease progression rate, mutation size, and age at onset prediction as early biomarkers of brain atrophy in huntington's disease. cns neuroscience & therapeutics, 15(1), 1–11. https://doi.org/10.1111/j.17555949.2008.00068.x. [11] de la monte, s. m., vonsattel, j. p., & richardson, e. p., jr (1988). morphometric demonstration of atrophic changes in the cerebral cortex, white matter, and neostriatum in huntington's disease. journal of neuropathology and experimental neurology, 47(5), 516–525. https://doi.org/10.1097/00005072198809000-00003. [12] hatters d. m. (2008). protein misfolding inside cells: the case of huntingtin and huntington's disease. iubmb life, 60(11), 724– 728. https://doi.org/10.1002/iub.111. [13] scherzinger, e., sittler, a., schweiger, k., heiser, v., lurz, r., hasenbank, r., bates, g. p., lehrach, h., & wanker, e. e. (1999). self-assembly of polyglutaminecontaining huntingtin fragments into amyloidlike fibrils: implications for huntington's disease pathology. proceedings of the national academy of sciences of the united states of america, 96(8), 4604–4609. https://doi.org/10.1073/pnas.96.8.4604. [14] nucifora , f. c., jr, sasaki, m., peters, m. f., huang, h., cooper, j. k., yamada, m., takahashi, h., tsuji, s., troncoso, j., dawson, v. l., dawson, t. m., & ross, c. a. (2001). interference by huntingtin and atrophin-1 with cbp-mediated transcription leading to cellular toxicity. science (new york, n.y.), 291(5512), 2423–2428. https://doi.org/10.1126/science.1056784. [15] imarisio, s., carmichael, j., korolchuk, v., chen, c. w., saiki, s., rose, c., krishna, g., davies, j. e., ttofi, e., underwood, b. r., & rubinsztein, d. c. (2008). huntington's disease: from pathology and genetics to potential therapies. the biochemical journal, 412(2), 191–209. https://doi.org/10.1042/bj20071619. [16] baxendale, s., abdulla, s., elgar, g., buck, d., berks, m., micklem, g., durbin, r., bates, g., brenner, s., & beck, s. (1995). comparative sequence analysis of the human and pufferfish huntington's disease genes. nature genetics, 10(1), 67–76. https://doi.org/10.1038/ng0595-67. columbia undergraduate science journal vol. 17, 2023 datta 28 [17] li, w., serpell, l. c., carter, w. j., rubinsztein, d. c., & huntington, j. a. (2006). expression and characterization of full-length human huntingtin, an elongated heat repeat protein. the journal of biological chemistry, 281(23), 15916–15922. https://doi.org/10.1074/jbc.m511007200. [18] macdonald m. e. (2003). huntingtin: alive and well and working in middle management. science's stke : signal transduction knowledge environment, 2003(207), pe48. https://doi.org/10.1126/stke.2003.207.pe48. [19] myers rh, marans k, macdonald me. huntington’s disease. in: genetic instabilities and hereditary neurological diseases (warren st, wells rt, eds), pp 301–323. new york: academic press, 1998. [20] gusella, j. f., wexler, n. s., conneally, p. m., naylor, s. l., anderson, m. a., tanzi, r. e., watkins, p. c., ottina, k., wallace, m. r., & sakaguchi, a. y. (1983). a polymorphic dna marker genetically linked to huntington's disease. nature, 306(5940), 234–238. https://doi.org/10.1038/306234a0. [21] myers r. h. (2004). huntington's disease genetics. neurorx : the journal of the american society for experimental neurotherapeutics, 1(2), 255–262. https://doi.org/10.1602/neurorx.1.2.255. [22] duyao, m., ambrose, c., myers, r., novelletto, a., persichetti, f., frontali, m., folstein, s., ross, c., franz, m., & abbott, m. (1993). trinucleotide repeat length instability and age of onset in huntington's disease. nature genetics, 4(4), 387–392. https://doi.org/10.1038/ng0893-387. [23] tartari, m., gissi, c., lo sardo, v., zuccato, c., picardi, e., pesole, g., & cattaneo, e. (2008). phylogenetic comparison of huntingtin homologues reveals the appearance of a primitive polyq in sea urchin. molecular biology and evolution, 25(2), 330– 338. https://doi.org/10.1093/molbev/msm258. [24] takano, h., & gusella, j. f. (2002). the predominantly heat-like motif structure of huntingtin and its association and coincident nuclear entry with dorsal, an nfkb/rel/dorsal family transcription factor. bmc neuroscience, 3, 15. https://doi.org/10.1186/1471-2202-3-15. [25] chook, y. m., & blobel, g. (1999). structure of the nuclear transport complex karyopherin-beta2-ran x gppnhp. nature, 399(6733), 230–237. https://doi.org/10.1038/20375. [26] grinthal, a., adamovic, i., weiner, b., karplus, m., & kleckner, n. (2010). pr65, the heat-repeat scaffold of phosphatase pp2a, is an elastic connector that links force and catalysis. proceedings of the national academy of sciences of the united states of america, 107(6), 2467–2472. https://doi.org/10.1073/pnas.0914073107. [27] tartari, m., gissi, c., lo sardo, v., zuccato, c., picardi, e., pesole, g., & cattaneo, e. (2008). phylogenetic comparison of huntingtin homologues reveals the appearance of a primitive polyq in sea urchin. molecular biology and evolution, 25(2), 330– 338. https://doi.org/10.1093/molbev/msm258. [28] li, s. h., & li, x. j. (2004). huntingtinprotein interactions and the pathogenesis of huntington's disease. trends in genetics : tig, 20(3), 146–154. https://doi.org/10.1016/j.tig.2004.01.008. [29] clabough, e. b., & zeitlin, s. o. (2006). deletion of the triplet repeat encoding polyglutamine within the mouse huntington's disease gene results in subtle behaviorcolumbia undergraduate science journal vol. 17, 2023 datta 29 al/motor phenotypes in vivo and elevated levels of atp with cellular senescence in vitro. human molecular genetics, 15(4), 607– 623. https://doi.org/10.1093/hmg/ddi477. [30] fiumara, f., fioriti, l., kandel, e. r., & hendrickson, w. a. (2010). essential role of coiled coils for aggregation and activity of q/n-rich prions and polyq proteins. cell, 143(7), 1121–1135. https://doi.org/10.1016/j.cell.2010.11.042. [31] ignatova, z., & gierasch, l. m. (2006). inhibition of protein aggregation in vitro and in vivo by a natural osmoprotectant. proceedings of the national academy of sciences of the united states of america, 103(36), 13357–13361. https://doi.org/10.1073/pnas.0603772103. [32] qin, z. h., wang, y., sapp, e., cuiffo, b., wanker, e., hayden, m. r., kegel, k. b., aronin, n., & difiglia, m. (2004). huntingtin bodies sequester vesicle-associated proteins by a polyproline-dependent interaction. the journal of neuroscience : the official journal of the society for neuroscience, 24(1), 269– 281. https://doi.org/10.1523/jneurosci.140903.2004. [33] atwal, r. s., xia, j., pinchev, d., taylor, j., epand, r. m., & truant, r. (2007). huntingtin has a membrane association signal that can modulate huntingtin aggregation, nuclear entry and toxicity. human molecular genetics, 16(21), 2600–2615. https://doi.org/10.1093/hmg/ddm217. [34] cornett, j., cao, f., wang, c. e., ross, c. a., bates, g. p., li, s. h., & li, x. j. (2005). polyglutamine expansion of huntingtin impairs its nuclear export. nature genetics, 37(2), 198–204. https://doi.org/10.1038/ng1503. [35] gafni, j., hermel, e., young, j. e., wellington, c. l., hayden, m. r., & ellerby, l. m. (2004). inhibition of calpain cleavage of huntingtin reduces toxicity: accumulation of calpain/caspase fragments in the nucleus. the journal of biological chemistry, 279(19), 20211–20220. https://doi.org/10.1074/jbc.m401267200. [36] mende-mueller, l. m., toneff, t., hwang, s. r., chesselet, m. f., & hook, v. y. (2001). tissue-specific proteolysis of huntingtin (htt) in human brain: evidence of enhanced levels of nand c-terminal htt fragments in huntington's disease striatum. the journal of neuroscience : the official journal of the society for neuroscience, 21(6), 1830–1837. https://doi.org/10.1523/jneurosci.21-0601830.2001. [37] kim, y. j., sapp, e., cuiffo, b. g., sobin, l., yoder, j., kegel, k. b., qin, z. h., detloff, p., aronin, n., & difiglia, m. (2006). lysosomal proteases are involved in generation of n-terminal huntingtin fragments. neurobiology of disease, 22(2), 346–356. https://doi.org/10.1016/j.nbd.2005.11.012. [38] humbert, s., bryson, e. a., cordelières, f. p., connors, n. c., datta, s. r., finkbeiner, s., greenberg, m. e., & saudou, f. (2002). the igf-1/akt pathway is neuroprotective in huntington's disease and involves huntingtin phosphorylation by akt. developmental cell, 2(6), 831–837. https://doi.org/10.1016/s15345807(02)00188-0. [39] metzler, m., gan, l., mazarei, g., graham, r. k., liu, l., bissada, n., lu, g., leavitt, b. r., & hayden, m. r. (2010). phosphorylation of huntingtin at ser421 in yac128 neurons is associated with protection of yac128 neurons from nmdamediated excitotoxicity and is modulated by columbia undergraduate science journal vol. 17, 2023 datta 30 pp1 and pp2a. the journal of neuroscience : the official journal of the society for neuroscience, 30(43), 14318–14329. https://doi.org/10.1523/jneurosci.158910.2010. [40] yanai, a., huang, k., kang, r., singaraja, r. r., arstikaitis, p., gan, l., orban, p. c., mullard, a., cowan, c. m., raymond, l. a., drisdel, r. c., green, w. n., ravikumar, b., rubinsztein, d. c., el-husseini, a., & hayden, m. r. (2006). palmitoylation of huntingtin by hip14 is essential for its trafficking and function. nature neuroscience, 9(6), 824–831. https://doi.org/10.1038/nn1702. [41] difiglia, m., sapp, e., chase, k., schwarz, c., meloni, a., young, c., martin, e., vonsattel, j. p., carraway, r., & reeves, s. a. (1995). huntingtin is a cytoplasmic protein associated with vesicles in human and rat brain neurons. neuron, 14(5), 1075– 1081. https://doi.org/10.1016/08966273(95)90346-1. [42] hebb, m. o., denovan-wright, e. m., & robertson, h. a. (1999). expression of the huntington's disease gene is regulated in astrocytes in the arcuate nucleus of the hypothalamus of postpartum rats. faseb journal : official publication of the federation of american societies for experimental biology, 13(9), 1099–1106. https://doi.org/10.1096/fasebj.13.9.1099. [43] ko, j., ou, s., & patterson, p. h. (2001). new anti-huntingtin monoclonal antibodies: implications for huntingtin conformation and its binding proteins. brain research bulletin, 56(3-4), 319–329. https://doi.org/10.1016/s03619230(01)00599-8. [44] borrell-pagès, m., zala, d., humbert, s., & saudou, f. (2006). huntington's disease: from huntingtin function and dysfunction to therapeutic strategies. cellular and molecular life sciences : cmls, 63(22), 2642–2660. https://doi.org/10.1007/s00018-006-6242-0. [45] semaka, a., creighton, s., warby, s., & hayden, m. r. (2006). predictive testing for huntington disease: interpretation and significance of intermediate alleles. clinical genetics, 70(4), 283–294. https://doi.org/10.1111/j.13990004.2006.00668.x. [46] s. mahalingam and l.m. levy (2014). genetics of huntington disease american journal of neuroradiology 35 (6) 1070-1072; doi: https://doi.org/10.3174/ajnr.a3772. [47] walker f. o. (2007). huntington's disease. lancet (london, england), 369(9557), 218–228. https://doi.org/10.1016/s01406736(07)60111-1. [48] tabrizi, s. j., langbehn, d. r., leavitt, b. r., roos, r. a., durr, a., craufurd, d., kennard, c., hicks, s. l., fox, n. c., scahill, r. i., borowsky, b., tobin, a. j., rosas, h. d., johnson, h., reilmann, r., landwehrmeyer, b., stout, j. c., & track-hd investigators (2009). biological and clinical manifestations of huntington's disease in the longitudinal track-hd study: crosssectional analysis of baseline data. the lancet. neurology, 8(9), 791–801. https://doi.org/10.1016/s14744422(09)70170-x. [49] trottier, y., devys, d., imbert, g., saudou, f., an, i., lutz, y., weber, c., agid, y., hirsch, e. c., & mandel, j. l. (1995). cellular localization of the huntington's disease protein and discrimination of the normal and mutated form. nature genetics, 10(1), 104– 110. https://doi.org/10.1038/ng0595-104. [50] bates g. (2003). huntingtin aggregation and toxicity in huntington's disease. lancet (london, england), 361(9369), 1642–1644. columbia undergraduate science journal vol. 17, 2023 datta 31 https://doi.org/10.1016/s01406736(03)13304-1. [51] ross, c. a., & tabrizi, s. j. (2011). huntington's disease: from molecular pathogenesis to clinical treatment. the lancet. neurology, 10(1), 83–98. https://doi.org/10.1016/s14744422(10)70245-3. [52] walker f. o. (2007). huntington's disease. lancet (london, england), 369(9557), 218–228. https://doi.org/10.1016/s01406736(07)60111-1. [53] koutsis g, karadima g, kladi a, et al. (2013). the challenge of juvenile huntington disease: to test or not to test. neurology; 80:990–96. [54] gusella, j. f., macdonald, m. e., & lee, j. m. (2014). genetic modifiers of huntington's disease. movement disorders : official journal of the movement disorder society, 29(11), 1359–1365. https://doi.org/10.1002/mds.26001. [55] hong, e. p., macdonald, m. e., wheeler, v. c., jones, l., holmans, p., orth, m., monckton, d. g., long, j. d., kwak, s., gusella, j. f., & lee, j. m. (2021). huntington's disease pathogenesis: two sequential components. journal of huntington's disease, 10(1), 35–51. https://doi.org/10.3233/jhd-200427. [56] lee, j. m., ramos, e. m., lee, j. h., gillis, t., mysore, j. s., hayden, m. r., warby, s. c., morrison, p., nance, m., ross, c. a., margolis, r. l., squitieri, f., orobello, s., di donato, s., gomez-tortosa, e., ayuso, c., suchowersky, o., trent, r. j., mccusker, e., novelletto, a., … cohort study of the hsg (2012). cag repeat expansion in huntington disease determines age at onset in a fully dominant fashion. neurology, 78(10), 690–695. https://doi.org/10.1212/wnl.0b013e318249f 683. [57] genetic modifiers of huntington’s disease (gem-hd) consortium (2015). identification of genetic factors that modify clinical onset of huntington's disease. cell, 162(3), 516–526. https://doi.org/10.1016/j.cell.2015.07.003. [58] lee, j. m., gillis, t., mysore, j. s., ramos, e. m., myers, r. h., hayden, m. r., morrison, p. j., nance, m., ross, c. a., margolis, r. l., squitieri, f., griguoli, a., di donato, s., gomez-tortosa, e., ayuso, c., suchowersky, o., trent, r. j., mccusker, e., novelletto, a., frontali, m., … gusella, j. f. (2012). common snp-based haplotype analysis of the 4p16.3 huntington disease gene region. american journal of human genetics, 90(3), 434–444. https://doi.org/10.1016/j.ajhg.2012.01.005. [59] quintanilla, r. a., & johnson, g. v. (2009). role of mitochondrial dysfunction in the pathogenesis of huntington's disease. brain research bulletin, 80(4-5), 242–247. https://doi.org/10.1016/j.brainresbull.2009.07 .010. [60] chen c. m. (2011). mitochondrial dysfunction, metabolic deficits, and increased oxidative stress in huntington's disease. chang gung medical journal, 34(2), 135– 152. [61] friedlander r. m. (2003). apoptosis and caspases in neurodegenerative diseases. the new england journal of medicine, 348(14), 1365–1375. https://doi.org/10.1056/nejmra022366. [62] leegwater-kim, j., & cha, j. h. (2004). the paradigm of huntington's disease: therapeutic opportunities in neurodegeneration. neurorx : the journal of the american society for experimental neurotherapeutics, 1(1), columbia undergraduate science journal vol. 17, 2023 datta 32 128–138. https://doi.org/10.1602/neurorx.1.1.128. [63] li, s. h., lam, s., cheng, a. l., & li, x. j. (2000). intranuclear huntingtin increases the expression of caspase-1 and induces apoptosis. human molecular genetics, 9(19), 2859–2867. https://doi.org/10.1093/hmg/9.19.2859. [64] kalonia, h., kumar, p., & kumar, a. (2011). attenuation of proinflammatory cytokines and apoptotic process by verapamil and diltiazem against quinolinic acid induced huntington like alterations in rats. brain research, 1372, 115–126. https://doi.org/10.1016/j.brainres.2010.11.06 0. [65] difiglia m. (1990). excitotoxic injury of the neostriatum: a model for huntington's disease. trends in neurosciences, 13(7), 286–289. https://doi.org/10.1016/01662236(90)90111-m. [66] jamwal, s., & kumar, p. (2015). antidepressants for neuroprotection in huntington's disease: a review. european journal of pharmacology, 769, 33–42. https://doi.org/10.1016/j.ejphar.2015.10.033. [67] palaiogeorgou, a.m., papakonstantinou, e., golfinopoulou, r., sigala, m., mitsis, t., papageorgiou, l. ... vlachakis, d. (2023). recent approaches on huntington's disease (review). biomedical reports, 18, 5. https://doi.org/10.3892/br.2022.1587 [68] barker, r. a., fujimaki, m., rogers, p., & rubinsztein, d. c. (2020). huntingtinlowering strategies for huntington's disease. expert opinion on investigational drugs, 29(10), 1125–1132. https://doi.org/10.1080/13543784.2020.1804 552 [69] aziz, n. a., van der burg, j. m. m., tabrizi, s. j., & landwehrmeyer, g. b. (2018). overlap between age-at-onset and diseaseprogression determinants in huntington disease. neurology, 90(24), e2099–e2106. https://doi.org/10.1212/wnl.0000000000005 690 [70] tabrizi, s. j., ghosh, r., & leavitt, b. r. (2019). huntingtin lowering strategies for disease modification in huntington's disease. neuron, 102(4), 899. https://doi.org/10.1016/j.neuron.2019.05.001 [71] lane, r. m., smith, a., baumann, t., gleichmann, m., norris, d., bennett, c. f., & kordasiewicz, h. (2018). translating antisense technology into a treatment for huntington's disease. methods in molecular biology (clifton, n.j.), 1780, 497–523. https://doi.org/10.1007/978-1-4939-78250_23 [72] tabrizi, s. j., leavitt, b. r., landwehrmeyer, g. b., wild, e. j., saft, c., barker, r. a., blair, n. f., craufurd, d., priller, j., rickards, h., rosser, a., kordasiewicz, h. b., czech, c., swayze, e. e., norris, d. a., baumann, t., gerlach, i., schobel, s. a., paz, e., smith, a. v., … phase 1–2a ionishttrx study site teams (2019). targeting huntingtin expression in patients with huntington's disease. the new england journal of medicine, 380(24), 2307–2316. https://doi.org/10.1056/nejmoa1900907 [73] keiser, m. s., kordasiewicz, h. b., & mcbride, j. l. (2016). gene suppression strategies for dominantly inherited neurodegenerative diseases: lessons from huntington's disease and spinocerebellar ataxia. human molecular genetics, 25(r1), r53– r64. https://doi.org/10.1093/hmg/ddv442 [74] difiglia, m., sena-esteves, m., chase, k., sapp, e., pfister, e., sass, m., yoder, j., reeves, p., pandey, r. k., rajeev, k. g., columbia undergraduate science journal vol. 17, 2023 datta 33 manoharan, m., sah, d. w., zamore, p. d., & aronin, n. (2007). therapeutic silencing of mutant huntingtin with sirna attenuates striatal and cortical neuropathology and behavioral deficits. proceedings of the national academy of sciences of the united states of america, 104(43), 17204–17209. https://doi.org/10.1073/pnas.0708285104 [75] yu, d., pendergraff, h., liu, j., kordasiewicz, h. b., cleveland, d. w., swayze, e. e., lima, w. f., crooke, s. t., prakash, t. p., & corey, d. r. (2012). single-stranded rnas use rnai to potently and alleleselectively inhibit mutant huntingtin expression. cell, 150(5), 895–908. https://doi.org/10.1016/j.cell.2012.08.002 [76] dufour, b. d., smith, c. a., clark, r. l., walker, t. r., & mcbride, j. l. (2014). intrajugular vein delivery of aav9-rnai prevents neuropathological changes and weight loss in huntington's disease mice. molecular therapy : the journal of the american society of gene therapy, 22(4), 797–810. https://doi.org/10.1038/mt.2013.289 [77] ho, l. w., brown, r., maxwell, m., wyttenbach, a., & rubinsztein, d. c. (2001). wild type huntingtin reduces the cellular toxicity of mutant huntingtin in mammalian cell models of huntington's disease. journal of medical genetics, 38(7), 450–452. https://doi.org/10.1136/jmg.38.7.450 [78] huntington study group (2006). tetrabenazine as antichorea therapy in huntington disease: a randomized controlled trial. neurology, 66(3), 366–372. https://doi.org/10.1212/01.wnl.0000198586.8 5250.13 [79] arora, a., kumar, s., ali, j., & baboota, s. (2020). intranasal delivery of tetrabenazine nanoemulsion via olfactory region for better treatment of hyperkinetic movement associated with huntington's disease: pharmacokinetic and brain delivery study. chemistry and physics of lipids, 230, 104917. https://doi.org/10.1016/j.chemphyslip.2020.1 04917 [80] devi, s., kumar, v., singh, s. k., dubey, a. k., & kim, j. j. (2021). flavonoids: potential candidates for the treatment of neurodegenerative disorders. biomedicines, 9(2), 99. https://doi.org/10.3390/biomedicines9020099 [81] catanesi, m., caioni, g., castelli, v., benedetti, e., d'angelo, m., & cimini, a. (2021). benefits under the sea: the role of marine compounds in neurodegenerative disorders. marine drugs, 19(1), 24. https://doi.org/10.3390/md19010024 [82] jabłońska, m., grzelakowska, k., wiśniewski, b., mazur, e., leis, k., & gałązka, p. (2020). pridopidine in the treatment of huntington's disease. reviews in the neurosciences, 31(4), 441–451. https://doi.org/10.1515/revneuro-2019-0085 [83] garcia-miralles, m., geva, m., tan, j. y., yusof, n. a. b. m., cha, y., kusko, r., tan, l. j., xu, x., grossman, i., orbach, a., hayden, m. r., & pouladi, m. a. (2017). early pridopidine treatment improves behavioral and transcriptional deficits in yac128 huntington disease mice. jci insight, 2(23), e95665. https://doi.org/10.1172/jci.insight.95665 [84] bartl, s., oueslati, a., southwell, a. l., siddu, a., parth, m., david, l. s., maxan, a., salhat, n., burkert, m., mairhofer, a., friedrich, t., pankevych, h., balazs, k., staffler, g., hayden, m. r., cicchetti, f., & smrzka, o. w. (2020). inhibiting cellular uptake of mutant huntingtin using a monoclonal antibody: implications for the treatment of huntington's disease. neurobiology of disease, 141, 104943. https://doi.org/10.1016/j.nbd.2020.104943 columbia undergraduate science journal vol. 17, 2023 datta 34 [85] fritz, n. e., rao, a. k., kegelmeyer, d., kloos, a., busse, m., hartel, l., carrier, j., & quinn, l. (2017). physical therapy and exercise interventions in huntington's disease: a mixed methods systematic review. journal of huntington's disease, 6(3), 217–235. https://doi.org/10.3233/jhd-170260 [86] nopoulos p. c. (2016). huntington disease: a single-gene degenerative disorder of the striatum. dialogues in clinical neuroscience, 18(1), 91–98. https://doi.org/10.31887/dcns.2016.18.1/pn opoulos. [87] shannon, k. m., & fraint, a. (2015). therapeutic advances in huntington's disease. movement disorders : official journal of the movement disorder society, 30(11), 1539–1546. https://doi.org/10.1002/mds.26331. [88] paulsen, j. s., long, j. d., johnson, h. j., aylward, e. h., ross, c. a., williams, j. k., nance, m. a., erwin, c. j., westervelt, h. j., harrington, d. l., bockholt, h. j., zhang, y., mccusker, e. a., chiu, e. m., panegyres, p. k., & predict-hd investigators and coordinators of the huntington study group (2014). clinical and biomarker changes in premanifest huntington disease show trial feasibility: a decade of the predict-hd study. frontiers in aging neuroscience, 6, 78. https://doi.org/10.3389/fnagi.2014.00078. [89] maxwell m. m. (2009). rnai applications in therapy development for neurodegenerative disease. current pharmaceutical design, 15(34), 3977–3991. https://doi.org/10.2174/13816120978964929 5. [90] beal, m. f., ferrante, r. j., swartz, k. j., & kowall, n. w. (1991). chronic quinolinic acid lesions in rats closely resemble huntington's disease. the journal of neuroscience : the official journal of the society for neuroscience, 11(6), 1649–1659. https://doi.org/10.1523/jneurosci.11-0601649.1991. [91] rozas, j. l., gómez-sánchez, l., tomás-zapico, c., lucas, j. j., & fernández-chacón, r. (2010). presynaptic dysfunction in huntington's disease. biochemical society transactions, 38(2), 488–492. https://doi.org/10.1042/bst0380488. [92] von zastrow, m., & sorkin, a. (2007). signaling on the endocytic pathway. current opinion in cell biology, 19(4), 436–445. https://doi.org/10.1016/j.ceb.2007.04.021. [93] hong, e. p., macdonald, m. e., wheeler, v. c., jones, l., holmans, p., orth, m., monckton, d. g., long, j. d., kwak, s., gusella, j. f., & lee, j. m. (2021). huntington's disease pathogenesis: two sequential components. journal of huntington's disease, 10(1), 35–51. https://doi.org/10.3233/jhd-200427. 26 introduction age-related macular degeneration (amd) is the leading cause of irreparable blindness in the developed world, affecting approximately 170 million people worldwide. over 11 million people in the united states suffer from amd, with that number projected to reach over 22 million over the coming decades [1]. people affected by amd experience a significant decrease in their quality of life due to impaired visual acuity at or near 20/200 (figure 1) [2]. while methods such as complement therapy and neuroprotection are currently being researched, there are no efficacious treatments for dry amd. amd must progress into the later and more severe wet amd stage for current treatments to be effective [3]. the prevailing therapies of photodynamic therapy and anti-vascular endothelial growth factor (vegf) therapy are inadequate, as their focus is not on curing amd and restoring sight, but on slowing down and preventing further vision loss [4]. cellular and biomaterial approaches for treating age-related macular degeneration ashrit challa*, srikar sama, selik morrishetty © 2022 challa, sama, morishetty. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. university of pennsylvania, philadelphia pa keywords: age-related macular degeneration, bioplastics, tissue engineering abstract: tissue engineering offers a novel, curative approach to treating age-related macular degeneration (amd), a disease characterized by excessive drusen deposition beneath the retinal surface and consequent vision loss. preclinical studies in rats have shown that transplanted retinal pigment epithelium (rpe) derived from human embryonic stem cells (hesc) have not only slowed amd but have also restored vision. there are two main methods of delivering rpe cells: direct injection and monolayer surgical insertion, the latter demonstrating long-term integration. biocompatible scaffolds allow for better delivery of rpe cells, induced pluripotent stem cells (ipsc), and retinal progenitor cells (rpc). unlike animal-derived extracellular matrix components, soft modulus biomaterials such as poly(lactic-co-glycolic acid) (plga) and poly(l-lactic acid) (plla) are ideal for amd cell transplants because of fast degradation times, high cellular attachment proliferation, and strong adherence to bruch’s membrane. these biomaterials can also be created at a 10-100μm thickness so that vision is not distorted. use of biomaterials could be improved by cross-linking them with anti-vascular endothelial growth factors (vegfs) like brolucizumab and retinal growth factors such as fibroblast growth factor (fgf). similarly, hesc and ipsc cells can be genetically modified to secrete anti-vegf factors. 27 as such, a tissue-engineered approach for treating amd has impressive implications, potentially allowing for the reversal of a disease previously thought to be irreversible. this article reviews the clinical features of amd, its current treatment options, stem cell and biomaterial tissue engineering therapeutic approaches, and pitfalls of and suggestions for such tissue-engineered approaches. clinical features of amd the clinical trademark of amd is the accumulation of fat and protein deposits, commonly known as drusen, in the macula, an area that is dense with photoreceptors responsible for high visual acuity [5]. drusen accumulates underneath the photoreceptors beneath the retinal pigment epithelium (rpe), which functions as a source of nutrients and growth factors, as well as a photoreceptor phagocytosis mechanism. the rpe attaches to bruch’s membrane (bm), which acts as a barrier between the retina and the choroid while regulating diffusion between the choroid and the rpe [6]. while the disease can be categorized into early, intermediate, and late stages based on the extent of drusen proliferation and vision loss, the most important distinction remains between dry and wet amd. dry, or non-exudative, amd occurs when excess drusen is deposited between the rpe and bm. this causes gradual rpe and photoreceptor cell death, as well as central macular atrophy and blind spots. wet, or neovascular/exudative, amd usually follows dry amd and is characterized by choroidal neovascularization (cnv), or the invasion of choroidal blood vessels into the rpe. bleeding and leaking from these vessels result in rpe cell death and rapid progression of blurriness and loss of visual acuity [7]. current treatments and medications for amd there are currently no effective curative treatments for amd; however, existing therapies aim to manage the disease and stop its progression. treatments for wet amd include laser photocoagulation and anti-vegf therapy, such as brolucizumab or ranibizumab. inhibiting vascularization in the eye prevents further progress of cnv, but it also contributes to chorioretinal atrophy due to less vascularization and potential narrowing of choroid capillaries. in fact, followups with anti-vegf treated eyes indicate an extremely high level (98%) of macular atrophy, particularly in the fovea [8]. photodynamic therapy (pdt) like vertreporfin aims to simply stop the progression of destructive vascularization through laser-activated medication. however, there is no clinically significant improvement of visual acuity because of this therapeutic method [9]. as such, monthly anti-vegf drug injections are currently prescribed and have had minimal to moderate success in restoring some vision to patients (about 30% of treated individuals), only maintaining the eye equivalent to its initial figure 1. as amd progresses, greater amounts of drusen disrupt the rpe-bm interface which leads to increased photoreceptor death. columbia undergraduate science journal vol. 16, 2022 28 state in which treatment first began. even when pdt and anti-vegf drugs are used in conjunction, a secondary approach when neither therapy is individually effective, there is a similar rate of visual acuity improvement (~11-13 letters) [10]. recent advancements in pharmacology have led to the development of better-performing injections, such as brolucizumab, which has a higher rate of visual acuity correction than ranibizumab. coupled with its small size, this compound allows for better vision improvement with fewer injections, but still holds the risk of injection-related complications and excessive drying [11]. currently, there are no current treatments for dry amd besides a surgical transplantation of a homologous donor retina. transplantation has previously been shown to be ineffective due to a failure in synapse formation between fully differentiated tissue and the host. surgical approaches of cleaning the debris near the retina and attempting to replace the degenerating retina with bolus injections have only offered temporary respite to the afflicted retina as it regresses back to its damaged form [12]. cell therapy approaches due to the difficulties of conventional therapies, recent years have witnessed much research and development in the use of stem cells and induced pluripotent cells as treatment or cures for amd. in 1987, the first significant study on the use of transplanted rpe cells for treatment was conducted by gouras et al, who placed transplanted rabbit rpe adjacent to the neural retina of a different rabbit [13]. another landmark occurred ten years later, when al-gvere et al. transplanted human fetal rpe patches into the subretinal space of human patients with wet or dry amd. the results of this grafting indicated that rpe transplants would not be rejected and further degrade vision, even without immunosuppression, and that dry amd had a lower graft rejection rate [14]. since these landmark studies of rpe transplants, iris pigment epithelium (ipe) cells and human embryonic stem cells (hesc) have been transplanted in many animal models, showing vision improvement. additionally, researchers have transplanted rpe cells, choroid-bruch’s-rpe explants, ipe cells and hesc in human patients with amd, aiming to evaluate transplantation safety (figure 2) [15]. preclinical studies conducted in rats have shown that using transplanted rpe cells derived from hesc have slowed retinal degeneration [16] and even improved visual acuity [17][18]. additionally, hesc-rpe has formed a polarized epithelial layer in vitro, secreting growth factors such as pigment epithelium described factor and vegf, all while expressing the barrier properties of normal adult human rpe cells [19]. currently the two preferred methods for delivering hesc-rpe into the subretinal space are either injections of cells suspended in a fluid, which are inexpensive and simple but carry the risk of rpe cell dedifferentiation, or creating monolayers of hescrpe which can be surgically placed subretinally, necessitating a biologically compatible substrate [20]. monolayers have shown higher rates of cell survival in comparison to injected figure 2: hesc-rpe cells injected into the subretinal space of human patients exhibit increased pigmentation and result in regeneration of the macula. columbia undergraduate science journal vol. 16, 2022 29 cell suspensions, as well as less clumping of cells [21]. recent advancements with hescrpe therapy include a completed three-year and an ongoing clinical trial conducted in part by the london project to cure blindness, with the former demonstrating the safety of long-term grafts while being the first to record longitudinal effects of hesc-rpe monolayer implantations. the completed study also showed lasting improvement in visual acuity of about 14-15 letters, as well as no adverse proliferative reactions such as teratoma formation, even after 37 months of observation. similarly, the phase i results of the ongoing clinical trial, with interventions provided to two human patients, indicate hesc-rpe integration while presenting improvement of visual acuity and reading speeds [22][23]. while the former study was conducted using bolus cell injections of hesc-rpes, the latter utilized a polyester sub-strate for monolayer insertion. furthermore, there are also other clinical trials using cell sheets, such as a trial using induced pluripotent stem cell (ipsc)-rpe cell sheet transplantation in patients with dry amd rather than wet amd, although this trial has only very recently begun (nct04339764). though the retinal space in the eye is immunologically privileged, various studies have used immunosuppression (such as tacrolimus and mycophenolate mofetil) during their trials of implanted stem cells, while others have relied purely on the immunological status of the eye [24]. both methods have had been successful in avoiding immunological rejection. ipscs have been studied for in vivo curative treatments and for the modeling of amd and other eye-related diseases. this is because there is an information deficit in the exact mechanisms for the progression of amd, specifically the non-exudative form. in response, the wide range of ipsc potential differentiation offers novel breakthrough methods in terms of replacement therapy and disease modeling [25]. to dedifferentiate human fibroblasts for use in ipsc-rpe procedures and other ipsc techniques, cells can be transfected with vectors such as the venezuelan equine encephalitis rna vector, inducing exogenous expression of pluripotency markers such as oct4, sox2, klf4, and glis1. with a dedifferentiation rate of >95%, newly formed ipsc cells are then suspended to form embryoid bodies, which are then placed in rpe medium. this allows for ipsc-rpe cells from different lineages to express high levels of rpe genes and proteins, such as rpe65 and mertk [26]. already, ipsc-rpe cells have been used to better understand the molecular etiology of amd, with one study identifying a single-nucleotide polymorphism near the vegfa gene in amd patients that decreases gene expression (figure 3). in conjunction, another study has been able to identify genes that are differently regulated in amd patients, alongside cell proliferation and localized immune response changes [27][28]. in this vein, ipsc-rpe research is exciting in its ability to better elucidate previously unknown mechanisms in amd development. figure 3. expression levels of vegfa for six ipsc-rpe samples, with the risk variant/ amd snp sample exhibiting significantly lower amounts of the gene columbia undergraduate science journal vol. 16, 2022 30 biomaterials approaches surgeons currently do not have complete control over where the rpe or retinal progenitor cells (rpcs) are placed subretinally, with random clumps of cells not being conducive to regeneration. additionally, without a way to ascertain the polarity and orientation of injected bolus or unoriented sheet cells, adherence to bruch’s membrane is drastically lowered. the use of biocompatible materials allows the parameters for delivery to be adjusted so that rpe cells, ipscs, and other rpcs can be delivered subretinally to regenerate the damaged rpe cell layer, attaching to photoreceptors apically and bm basally. thus, biomaterials must allow for cell attachment, proliferation, and correct orientation/polarization. to allow for rpe cell proliferation and bm attachment, the material must degrade by 2-3 weeks post-implant, and is ideally 10-100μm to allow for precise manipulation without retinal distortion and nutrient diffusion limitation [29]. some biomaterials that have been explored as substrates for rpe monolayer insertion are collagen, matrigel®️, fibronectin, laminin, vitronectin, and oligopeptides. however, these organic, animal-derived extracellular matrix components were found to discourage cell proliferation and have variable degradation times based on individual enzymatic digestion rates [30]. nanowires of poly(e-caprolactone) (pcl) may also be cast, but require precise construction in order to ensure porosity levels conducive to rpc polarization and attachment [31]. poly(dl-lactic-co-glycolic acid) (plga), poly(glycolic acid) (pga), poly(ethylene glycol) (peg), and poly(dl-lactic acid) (pla), are synthetic, thin, and degradable bio-materials. however, plga has an ideal degradation time of 2-3 weeks (whereas the others do not significantly degrade until about four weeks after initial cell seeding), can be 10-130μm thick, and has a feasible manufacturing process [32] [33]. once solvated in chloroform or hexafluoroisopropanol (hfip), these polymers are left to deposit on an even glass or scaffdex surface for 8 hours while the solvent evaporates. the thin sheets are then left to dry, potentially stored in nitrogenous atmospheres over desiccators like calcium sulfate. they may then be crosslinked and sterilized using uv light [34][35]. the rpe, rpc, and ipscs can thus be seeded. it has been found that there is 99-100% attachment of non-hesc human rpe cells to plga after 8 hours, and both plga and pga sheets allow for rpe cell metabolism and protein expression. these polymers are therefore viable for cell attachment and proliferation, allowing for apical microvilli and basilar diffusion and molecule excretion [36][37]. however, these rpe cells do not attach significantly to pla or peg. after 3-7 days, cells become confluent, forming a multilayer or monolayer (depending on the cell type) of polarized rpe or rpcs and are ready for implantation. plga sheets have been found to be the smoothest, thinnest, have the highest polarized rpe cell attachment, and most proliferation/ material area repopulation. this is due to a 50:50 high molecular weight pgla that contains an optimal ratio of lactic acid to glycolic acid. in comparison to previous substrates such as collagen, plga sheets are much smoother, thinner, and allow for the formation of an rpe monolayer with both correct orientation and polarity of cells. additionally, multiple plga ratios of lactic acid to glycolic acid, such as high molecular weight (high mw) 50:50 and 75:25 plga, allowed for significant levels of cell proliferation. however, due to its faster degradation time, a high mw 50:50 plga blend is thought to be optimal [38]. the porosity of this blend allows correct adherence to bm so that as the biomaterial is degraded, the rpe cells attach columbia undergraduate science journal vol. 16, 2022 31 and are integrated into the eye. studies also indicate that creating a polymer blend of plga and poly(l-lactic acid) (plla) may improve the porosity of the substrate, thus allowing for higher rates of cell proliferation. the modulus (hardness) of plga and poly(l-lactic acid) plla 50:50 is low enough to overcome the otherwise stiff composition of pure plga, with the flexibility provided by plla mitigating the risk of retinal damage. this 50:50 plga:plla was implanted in the rat model and the viability of the cells was monitored for 14 days [39]. unlike previous attempts to directly inject rpc cells into the retina, the plga:plla bound rpc and rpe cells were still viable and expressed gfp, unlike transplanted cells, which only had a 10% survival rate 14 days after implantation (figure 4). a more recent study shows that a 25:75 plla:plga blend might be more efficient, with a higher level of porosity and a lower elastic modulus (figure 5) [40]. soft modulus biomaterials of plga and plla are ideal for amd cell delivery because they degrade fast, have high cell attachment, rpe and rpc cell polarization, cell proliferation, adherence to bm, and high viability after implantation. shortcomings although cell therapies have evolved greatly over the past few decades, regarding both transplantations and surgical monolayer integration, there are still many unanswered questions. transplanted hesc-rpe cells demonstrated an increase in retinal pigmentation and visual acuity by about 14 letters in patients [41]. however, a 2008 study that surgically implanted fetal rpe into ten patients saw a four times improvement in visual acuity in one patient, from 20/800 to 20/200, which remained stable for five years [42]. this magnitude in visual improvement has not yet been seen from hesc or ipsc approaches, which are also marred by complications such as cataract formation and vitreous inflammation. and while ipsc-rpe clinical trial results are increasing in number, there is no published research on the longitudinal effects of ipsc-rpe cell transplantation in human patients with amd. it is known that ipsc-rpe cells also have a faster rejection time than hesc-derived cells, triggering macrophage-mediated phagocytosis such that almost no ipsc cells remain after 13 weeks [43]. given that donor retina scaffolds are difficult to procure, tissue engineering approaches provide alternative avenues. regardless, the figure 4. attachment levels of rpe cells for different cell substrates, with three different compositions of plga, the most proliferative substrate figure 5: predicted vs. actual elastic moduli for various plla: plga ratios, with the 25:75 exhibiting the lowest modulus columbia undergraduate science journal vol. 16, 2022 32 use of cutting-edge stem cell engineering techniques and substrates like plga and plla still pose problems. some of these problems include inflammation and injury due to the injection of cells, incomplete attachment of rpe cells to bm or a lack of subsequent proliferation, and the inability to replicate the true retinal environment. cells may also dedifferentiate once attached to bm, which would only lead to the presence of more harmful debris in the retina and subretinal space. both injected and surgically inserted cells have not yet shown complete restorations of vision and have only demonstrated peripheral and minor macular vision improvement. finally, the toxicity of the material used as the cell substrate must also be considered, as degradation of these scaffolds will inevitably lead to the presence of small subunits in the macular region. with the use of plga, a potentially toxic and immunogenic substance in the eye, careful construction of scaffolding must be used in accordance with shape and size restraints. as more data comes out regarding biomaterial interactions with the actual human eye, rather than approximating animal models, it will be important to modify scaffold compositions and morphologies accordingly [44]. improvements for current techniques based on current research, there are many hypothesized methods of improving the treatment of amd. cell therapies utilizing stem cells can be further improved with genetically modified rpe cells, able to impede or revert the progression of amd through genome and transcriptome modifications to counteract the changes of the disease. neovascular amd is due in part to an imbalance in growth factors such as vegf. rpe cells naturally express many anti-vegfs like pigment epithelium-derived factor (pedf), but damage to these cells allows for excess vascularization. to further treatment, transplanted cells should be genetically modified to upregulate the production of anti-vegfs, such as endostatin, pedf, basic fibroblast growth factor (bfgf), brain-derived neurotrophic factor (bdnf), or ciliary neurotrophic factor (cntf). to perform the necessary genetic modifications for the aforementioned suggestions, possible transfection methods include the sleeping beauty (sb100x)^ transposon system or the crispr/ cas9 system. during the later stages of amd when damage has occurred to both photoreceptors and rpe cells, another improvement would be to transplant autologous esc or ipsc which differentiate under a wider range of factors. this would allow for the reconstruction of the entire bm-rpe-photoreceptor complex, potentially leading to greater improvements in vision. in addition to therapeutic uses, it would also be helpful to use ipsc-rpe cells to model dry amd alongside wet amd in humans, as these cells could uncover more about dry amd while finding use as high throughput drug screens. gene therapy efficacies could also be tested in this way. this knowledge could lead to not just curative methods, but perhaps preventative interventions as well. in using biomaterials as a substrate for rpe cells, endogenous factors can perhaps be crosslinked to plga or plla. as protases break the crosslinks in plga scaffolds, anti-vegfs such as ranibizumab and retinal growth factors like fgf could be freed, allowing for retinal regeneration [45]. delivery of ranibizumab via nanoparticles has already been demonstrated, so it is now a matter of delivering these types of medications on plga scaffolds for amd patients (figure 6) [46]. columbia undergraduate science journal vol. 16, 2022 33 because rpe cells must be polarized, incorporating plga with this cell type alongside rpcs and ipscs may allow for differentiation into rpe and other retinal cells. in this manner, the eye can heal any damage done by the degradation or delivery of the substrate biomaterial and cells in the first place. similarly, the use of drugs to modulate calcium systems has shown down-regulating effects on proliferation, which could be used in the case of teratomacreating ipsc-rpe cells. if cross-linked in plga scaffolds, these inhibitors could be used as a form of cell control [47]. in regard to surgical placement, current methods require the rolling up of plga/plla scaffolds. these rolled-up scaffolds are then delivered to the subretinal space. an improvement would be to develop a method by which a higher modulus material can be delivered to the eye directly in between the bm and the photoreceptors, perhaps via a small incision. this higher modulus material could then be removed to leave softer biomaterials, which are less harmful to retinal health and vision. finally, cleaning up dedifferentiated cell and degraded scaffolding debris with lasers may also allow for better vision and healing. conclusion age-related macular degeneration is one of the leading causes of blindness in aging adults and is prevalent in today’s population. tissue engineering solutions for amd have improved upon conventional treatments, thanks to their ability to not just stop the progression of amd, but also result in visual improvement. the use of hescrpe and ipsc-rpe, coupled with a thin plga/ plla scaffold, may allow for effective and safe integration of rpe cells in the macula. with more research on polymer toxicity, molecular crosslinking, and methods of efficient insertion, these cellular engineering techniques may become the best way to reverse amd damage and fully regenerate the patient’s eye. author information corresponding author *ashrit challa, university of pennsylvania, department of biology; achalla@sas.upenn.edu, 937-750-1530 author contributions ashrit challa – text srikar sama – figures and proofreading selik morishetty – figures and proofreading abbreviations amd – age-related macular degeneration vegf – vascular endothelial growth factor rpe – retinal pigment epithelium bm – bruch’s membrane cnv – choroidal neovascularization pdt – photodynamic therapy ipe – iris pigment epithelium hesc – human embryonic stem cells ipsc – induced pluripotent stem cells rpc – retinal progenitor cell plga – poly(dl-lactic-co-glycolic acid) pga – poly(glycolic acid) figure 6: cumulative release percentage of ranibizumab by plga nanoparticle system; if modified into plga scaffolding, could be used to deliver anti-vegfs to integrating rpe cells columbia undergraduate science journal vol. 16, 2022 34 peg – poly(ethylene glycol) pla – poly(dl-lactic acid) hfip hexafluoroisopropanol plla – poly(l-lactic acid) pedf – pigment epithelium-derived factor bfgf – basic fibroblast growth factor bdnf – brain-derived neurotrophic factor cntf – ciliary neurotrophic factor references [1] pennington, k. l., & deangelis, m. m. (2016). epidemiology of age-related macular degeneration (amd): associations with cardio-vascular disease phenotypes and lipid factors. eye and vision (lon-don, england), 3, 34. https://doi. org/10.1186/s40662-016-0063-5 [2] fernández-robredo, p., sancho, a., johnen, s., recalde, s., gama, n., thumann, g., groll, j., & garcía-layana, a. (2014). current treatment limitations in age-related macular degeneration and future approaches based on cell therapy and tissue engineering. journal of ophthalmology, 2014, 510285. https://doi. org/10.1155/2014/510285 [3] ammar, m. j., hsu, j., chiang, a., ho, a. c., & regillo, c. d. (2020). age-related macular degeneration therapy: a review. current opinion in ophthalmology, 31(3), 215–221. https://doi. org/10.1097/icu.0000000000000657 [4] fernández-robredo, p., sancho, a., johnen, s., recalde, s., gama, n., thumann, g., groll, j., & garcía-layana, a. (2014). current treatment limitations in age-related macular degeneration and future approaches based on cell therapy and tissue engineering. journal of ophthalmology, 2014, 510285. https://doi. org/10.1155/2014/510285 [5] jager, r. d., mieler, w. f., & miller, j. w. (2008). age-related macular degeneration. the new england journal of medicine, 358(24), 2606–2617. https://doi.org/10.1056/nejmra0801537 [6] bhutto, i., & lutty, g. (2012). understanding age-related macular degeneration (amd): relationships between the photoreceptor/retinal pigment epithelium/bruch's membrane/choriocapillaris complex. molecular aspects of medicine, 33(4), 295–317. https://doi.org/10.1016/j. mam.2012.04.005 [7] mathenge w. (2014). age-related macular degeneration. communi-ty eye health, 27(87), 49–50. [8] fernández-robredo, p., sancho, a., johnen, s., recalde, s., gama, n., thumann, g., groll, j., & garcía-layana, a. (2014). current treatment limitations in age-related macular degeneration and future approaches based on cell therapy and tissue engineering. journal of ophthalmology, 2014, 510285. https://doi. org/10.1155/2014/510285 [9] brown, d. m., kaiser, p. k., michels, m., soubrane, g., heier, j. s., kim, r. y., sy, j. p., schneider, s., & anchor study group (2006). ranibizumab versus verteporfin for neovascular age-related macular degeneration. the new england journal of medicine, 355(14), 1432–1444. https://doi.org/10.1056/nejmoa062655 [10] fernández-robredo, p., sancho, a., johnen, s., recalde, s., ga-ma, n., thumann, g., groll, j., & garcía-layana, a. (2014). current treatment limitations in age-related macular degeneration and future approaches based on cell therapy and tissue engineering. journal of ophthalmology, 2014, 510285. https://doi. org/10.1155/2014/510285 columbia undergraduate science journal vol. 16, 2022 35 [11] yannuzzi, n. a., & freund, k. b. (2019). brolucizumab: evidence to date in the treatment of neovascular age-related macular degenera-tion. clinical ophthalmology (auckland, n.z.), 13, 1323–1329. https://doi.org/10.2147/ opth.s184706 [12] maclaren, r. e., bird, a. c., sathia, p. j., & aylward, g. w. (2005). long-term results of submacular surgery combined with macular translocation of the retinal pigment epithelium in neovascu-lar age-related macular degeneration. ophthalmology, 112(12), 2081–2087. https://doi. org/10.1016/j.ophtha.2005.06.029 [13] lopez, r., gouras, p., brittis, m., & kjeldbye, h. (1987). trans-plantation of cultured rabbit retinal epithelium to rabbit retina using a closed-eye method. investigative ophthalmology & visual science, 28(7), 1131–1137. [14] algvere, p. v., berglin, l., gouras, p., sheng, y., & kopp, e. d. (1997). transplantation of rpe in age-related macular degeneration: observations in disciform lesions and dry rpe atrophy. graefe's archive for clinical and experimental ophthalmology = albrecht von graefes archiv fur klinische und experimentelle ophthalmologie, 235(3), 149–158. https://doi. org/10.1007/bf00941722 [15] fernández-robredo, p., sancho, a., johnen, s., recalde, s., ga-ma, n., thumann, g., groll, j., & garcía-layana, a. (2014). current treatment limitations in age-related macular degeneration and future approaches based on cell therapy and tissue engineering. journal of ophthalmology, 2014, 510285. https://doi. org/10.1155/2014/510285 [16] idelson, m., alper, r., obolensky, a., benshushan, e., hemo, i., yachimovich-cohen, n., khaner, h., smith, y., wiser, o., gropp, m., cohen, m. a., even-ram, s., berman-zaken, y., matzrafi, l., rechavi, g., banin, e., & reubinoff, b. (2009). di-rected differentiation of human embryonic stem cells into functional retinal pigment epithelium cells. cell stem cell, 5(4), 396–408. https://doi.org/10.1016/j.stem.2009.07.002 [17] lu, b., malcuit, c., wang, s., girman, s., francis, p., lemieux, l., lanza, r., & lund, r. (2009). long-term safety and function of rpe from human embryonic stem cells in preclinical models of macular degeneration. stem cells (dayton, ohio), 27(9), 2126–2135. https://doi. org/10.1002/stem.149 [18] lund, r. d., wang, s., klimanskaya, i., holmes, t., ramos-kelsey, r., lu, b., girman, s., bischoff, n., sauvé, y., & lanza, r. (2006). human embryonic stem cell-derived cells rescue visual function in dystrophic rcs rats. cloning and stem cells, 8(3), 189–199. https:// doi.org/10.1089/clo.2006.8.189 [19] carr, a. j., smart, m. j., ramsden, c. m., powner, m. b., da cruz, l., & coffey, p. j. (2013). development of human embryonic stem cell therapies for age-related macular degeneration. trends in neuro-sciences, 36(7), 385–395. https://doi.org/10.1016/j.tins.2013.03.006 [20] carr, a. j., smart, m. j., ramsden, c. m., powner, m. b., da cruz, l., & coffey, p. j. (2013). development of human embryonic stem cell therapies for age-related macular degeneration. trends in neuro-sciences, 36(7), 385– 395. https://doi.org/10.1016/j.tins.2013.03.006 [21] diniz, b., thomas, p., thomas, b., ribeiro, r., hu, y., brant, r., ahuja, a., zhu, d., liu, l., koss, m., maia, m., chader, g., hinton, d. r., & humayun, m. s. (2013). subretinal implantation columbia undergraduate science journal vol. 16, 2022 36 of retinal pigment epithelial cells derived from human embryonic stem cells: improved survival when implanted as a monolayer. investigative ophthalmology & visual science, 54(7), 5087– 5096. https://doi.org/10.1167/iovs.12-11239 [22] schwartz, s. d., regillo, c. d., lam, b. l., eliott, d., rosenfeld, p. j., gregori, n. z., hubschman, j. p., davis, j. l., heilwell, g., spirn, m., maguire, j., gay, r., bateman, j., ostrick, r. m., morris, d., vincent, m., anglade, e., del priore, l. v., & lanza, r. (2015). human embryonic stem cell-derived retinal pigment epithelium in patients with age-related macular degeneration and stargardt's macu-lar dystrophy: follow-up of two open-label phase 1/2 studies. lancet (london, england), 385(9967), 509–516. https://doi.org/10.1016/s0140-6736(14)61376-3 [23] da cruz, l., fynes, k., georgiadis, o., kerby, j., luo, y. h., ahmado, a., vernon, a., daniels, j. t., nommiste, b., hasan, s. m., gooljar, s. b., carr, a. f., vugler, a., ramsden, c. m., bictash, m., fenster, m., steer, j., harbinson, t., wilbrey, a., tufail, a., … coffey, p. j. (2018). phase 1 clinical study of an embryonic stem cell-derived retinal pigment epithelium patch in age-related macular degeneration. nature biotechnology, 36(4), 328–337. https://doi. org/10.1038/nbt.4114 [24] lu, b., malcuit, c., wang, s., girman, s., francis, p., lemieux, l., lanza, r., & lund, r. (2009). long-term safety and function of rpe from human embryonic stem cells in preclinical models of macular degeneration. stem cells (dayton, ohio), 27(9), 2126–2135. https://doi. org/10.1002/stem.149 [25] melville, h., carpiniello, m., hollis, k., staffaroni, a., & goles-taneh, n. (2013). stem cells: a new paradigm for disease modeling and developing therapies for age-related macular degeneration. journal of translational medicine, 11, 53. https://doi.org/10.1186/1479-5876-11-53 [26] hazim, r. a., karumbayaram, s., jiang, m., dimashkie, a., lopes, v. s., li, d., burgess, b. l., vijayaraj, p., alva-ornelas, j. a., zack, j. a., kohn, d. b., gomperts, b. n., pyle, a. d., lowry, w. e., & williams, d. s. (2017). differentiation of rpe cells from integration-free ips cells and their cell biological characterization. stem cell research & therapy, 8(1), 217. https://doi. org/10.1186/s13287-017-0652-9 [27] smith, e. n., d'antonio-chronowska, a., greenwald, w. w., borja, v., aguiar, l. r., pogue, r., matsui, h., benaglio, p., bo-rooah, s., d'antonio, m., ayyagari, r., & frazer, k. a. (2019). human ipsc-derived retinal pigment epithelium: a model system for prioritizing and functionally characterizing causal variants at amd risk loci. stem cell reports, 12(6), 1342–1353. https://doi.org/10.1016/j.stemcr.2019.04.012 [28] gong, j., cai, h., nyscf global stem cell array team, noggle, s., paull, d., rizzolo, l. j., del priore, l. v., & fields, m. a. (2020). stem cell-derived retinal pigment epithelium from patients with age-related macular degeneration exhibit reduced metabolism and matrix interactions. stem cells translational medicine, 9(3), 364–376. https://doi.org/10.1002/sctm.19-0321 [29] lu, l., yaszemski, m. j., & mikos, a. g. (2001). retinal pigment epithelium engineering using synthetic biodegradable polymers. biomaterials, 22(24), 3345–3355. https://doi. org/10.1016/s0142-9612(01)00172-7 [30] lu, l., yaszemski, m. j., & mikos, a. g. (2001). retinal pigment epithelium engineercolumbia undergraduate science journal vol. 16, 2022 37 ing using synthetic biodegradable polymers. biomaterials, 22(24), 3345–3355. https://doi. org/10.1016/s0142-9612(01)00172-7 [31] mchugh, k. j., tao, s. l., & saint-geniez, m. (2014). porous poly(ε-caprolactone) scaffolds for retinal pigment epithelium trans-plantation. investigative ophthalmology & visual science, 55(3), 1754–1762. https://doi.org/10.1167/ iovs.13-12833 [32] reed, a. m., gilding, d. k. (1981). biodegradable polymers for use in surgery-poly(glycolic)/poly(lactic acid) homo and copoly-mers: 2. in vitro degradation. polymer, 22(4), 494-498. https://doi.org/10.1016/0032-3861(81)90168-3 [33] mccormick, r., pearce, i., kaye, s., & haneef, a. (2020). opti-misation of a novel bio-substrate as a treatment for atrophic age-related macular degeneration. frontiers in bioengineering and bio-technology, 8, 456. https://doi.org/10.3389/fbioe.2020.00456 [34] lu, l., nyalakonda, k., kam, l., bizios, r., göpferich, a., & mikos, a. g. (2001). retinal pigment epithelial cell adhesion on novel micropatterned surfaces fabricated from synthetic biode-gradable polymers. biomaterials, 22(3), 291–297. https://doi.org/10.1016/s01429612(00)00179-4 [35] mccormick, r., pearce, i., kaye, s., & haneef, a. (2020). opti-misation of a novel bio-substrate as a treatment for atrophic age-related macular degeneration. frontiers in bioengineering and bio-technology, 8, 456. https://doi.org/10.3389/fbioe.2020.00456 [36] lu, l., garcia, c. a., & mikos, a. g. (1998). retinal pigment epithelium cell culture on thin biodegradable poly(dl-lactic-co-glycolic acid) films. journal of biomaterials science. polymer edi-tion, 9(11), 1187–1205. https://doi. org/10.1163/156856298x00721 [37] mccormick, r., pearce, i., kaye, s., & haneef, a. (2020). opti-misation of a novel bio-substrate as a treatment for atrophic age-related macular degeneration. frontiers in bioengineering and bio-technology, 8, 456. https://doi.org/10.3389/fbioe.2020.00456 [38] giordano, g. g., thomson, r. c., ishaug, s. l., mikos, a. g., cumber, s., garcia, c. a., & lahiri-munir, d. (1997). retinal pig-ment epithelium cells cultured on synthetic biodegradable polymers. journal of biomedical materials research, 34(1), 87–93. https://doi.org/10.1002/ (sici)1097-4636(199701)34:1<87::aid-jbm12>3.0.co;2-m [39] lavik, e. b., klassen, h., warfvinge, k., langer, r., & young, m. j. (2005). fabrication of degradable polymer scaffolds to direct the integration and differentiation of retinal progenitors. biomaterials, 26(16), 3187–3196. https:// doi.org/10.1016/j.biomaterials.2004.08.022 [40] thomson, h. a., treharne, a. j., walker, p., grossel, m. c., & lotery, a. j. (2011). optimisation of polymer scaffolds for retinal pigment epithelium (rpe) cell transplantation. the british journal of ophthalmology, 95(4), 563–568. https://doi.org/10.1136/bjo.2009.166728 [41] schwartz, s. d., regillo, c. d., lam, b. l., eliott, d., rosenfeld, p. j., gregori, n. z., hubschman, j. p., davis, j. l., heilwell, g., spirn, m., maguire, j., gay, r., bateman, j., ostrick, r. m., morris, d., vincent, m., anglade, e., del priore, l. v., & lanza, r. (2015). human embryonic stem cell-derived retinal pigment epithelium in patients with age-related columbia undergraduate science journal vol. 16, 2022 38 macular degeneration and stargardt's macu-lar dystrophy: follow-up of two open-label phase 1/2 studies. lan-cet (london, england), 385(9967), 509–516. https://doi.org/10.1016/ s0140-6736(14)61376-3 [42] radtke, n. d., aramant, r. b., petry, h. m., green, p. t., pidwell, d. j., & seiler, m. j. (2008). vision improvement in retinal degener-ation patients by implantation of retina together with retinal pigment epithelium. american journal of ophthalmology, 146(2), 172–182. https://doi.org/10.1016/j.ajo.2008.04.009 [43] carr, a. j., smart, m. j., ramsden, c. m., powner, m. b., da cruz, l., & coffey, p. j. (2013). development of human embryonic stem cell therapies for age-related macular degeneration. trends in neuro-sciences, 36(7), 385–395. https://doi.org/10.1016/j.tins.2013.03.006 [44] thackaberry, e. a., farman, c., zhong, f., lorget, f., staflin, k., cercillieux, a., miller, p. e., schuetz, c., chang, d., famili, a., daugherty, a. l., rajagopal, k., & bantseev, v. (2017). evaluation of the toxicity of intravitreally injected plga microspheres and rods in monkeys and rabbits: effects of depot size on inflamma-tory response. investigative ophthalmology & visual science, 58(10), 4274–4285. https://doi. org/10.1167/iovs.16-21334 [45] sievers, j., hausmann, b., unsicker, k., berry, m. (1987). fibro-blast growth factors promote the survival of adult rat retinal ganglion cells after transection of the optic nerve. neuroscience letters, 76(2), 157-162. https://doi. org/10.1016/0304-3940(87)90708-7 [46] tanetsugu, y., tagami, t., terukina, t., ogawa, t., ohta, m., & ozeki, t. (2017). development of a sustainable release system for a ranibizumab biosimilar using poly(lactic-co-glycolic acid) biode-gradable polymer-based microparticles as a platform. biological & pharmaceutical bulletin, 40(2), 145–150. https://doi. org/10.1248/bpb.b16-00437 [47] wagner, m., benson, m. t., rennie, i. g., & macneil, s. (1995). effects of pharmacological modulation of intracellular signalling systems on retinal pigment epithelial cell attachment to extracellular matrix proteins. current eye research, 14(5), 373–384. https://doi. org/10.3109/02713689508999935 columbia undergraduate science journal vol. 16, 2022 39 introduction the poor health of rural communities is well-established in literature, vividly illustrated by consistently reduced scores in health metrics. one example is urban-rural discrepancies in life expectancy. from 2006 to 2009, residents of nonmetropolitan counties were expected to live 76.8 years compared to 78.8 years in metropolitan counties [1]. compounding this disparity is its stagnation; urban lifespans have steadily increased over time, while rural lives have remained at their current length for decades [2]. only half a year separated rural and urban lifespans from 1971 to 1974, but this gap quadrupled to 2.0 years from 2005 to 2009 [1]. rural populations also have a higher prevalence of obesity than metropolitan populations [3,4]. with close ties to chronic, severe illnesses such as type 2 diabetes, high obesity rates imply poor health [5,6]. challenges between and within rural minority demographics benjamin blue © 2022 blue. this is an open access article distributed under the terms of the creative commons attribution license, which permits the user to copy, distribute, and transmit the work provided that the original authors and source are credited. university of california, davis, davis, ca, usa keywords: rural communities, health disparities, access to care, minorities, social determinants abstract: rural minorities in america are in poorer health compared to other demographics, as illustrated by deficits in several health metrics. such deficits are great in magnitude and temporally persistent. furthermore, the unrecognized diversity of rural areas means that these deficits affect a larger proportion of rural areas than conventional stereotype dictates. this review analyzes existing research to characterize the health of rural minorities, focusing on specific challenges and elucidating recurring themes that arise within and between minorities. specific demographics were chosen for in-depth analysis based on prominence in nonmetropolitan areas, utility as case studies, and availability and quality of research. population-specific analyses delineate themes that apply across demographics as well as challenges specific to each group, revealing the complexity of rural minority health. analysis elucidates some key themes surrounding rural minority health. the invisibility of these demographics underlies many surface-level issues such as ineffective interventions and the underfunding of health organizations. social context also plays a great yet underappreciated role in health, lending complexity to this issue through unique demographic-specific challenges. above all, knowledge gaps are one of the most impactful and long-staying factors affecting rural minorities, impeding solutions and promoting disparity through ignorance on the part of both policymakers and providers. 40 rural mental health is challenged as well. in rural areas, all psychiatric disorders except for non-ptsd anxiety disorders are more prevalent [7]. furthermore, from 1999 to 2016 in rural areas, drug overdose deaths increased by 749.4% and the suicide rate for those aged 25 to 64 years increased by 38.3% [8]. these statistics led the nih to establish rural residents as a health-disparity population [9]. the healthcare struggles of rural areas are both distinct and difficult to overcome, rooted in the unique communities, cultures, and environments of a rural context. for instance, healthcare workers cite different obstacles in the workplace depending on their locality. compared to urban providers, rural providers report struggling more with provider scarcities, finding continuing medical education, and staying compliant with the health insurance and portability accountability act, a law that dictates acceptable usage and privacy protocols for medical records [10]. rural culture is distinct as well, emphasizing family, traditionalism, trust, and independence. cultural competence is paramount in rural health to gain patient trust, correctly and appropriately convey information, and foster accessibility [11]. however, proportionally few providers are recruited from rural backgrounds, promoting a cultural disconnect [12], facilitating the high patient-to-provider ratios that characterize rural healthcare [10]. providers that fail to understand these values encounter community distrust and struggle to meet patient needs [13]. a defining struggle in rural health is long geographic distances. compared to those in large metropolitan counties, five times as many rural households have no car and live further than a mile from a grocery store [1]. these vast distances spawn communication issues that hamper patient-provider contact, block cohesion between services, and damage efficiency, all of which limit healthcare access, making accessibility a prominent issue in rural health [14,15]. rural providers, already few in number, are spread thin over vast distances; lower service demand due to diffuse populations particularly affects specialists, who cluster in urbanized areas to support their practice [16]. high business expenses minimize profit margins for rural providers as well [17]. large metro counties have 263 specialists per 100,000 people whereas rural counties see only 30 per 100,000, contributing to reliance on primary care [16]. with greater dependence on primary care, rural areas also suffer from the nationwide shortage of general practitioners, where 25-30 providers serve 100,000 people in both urban and rural areas [16]. less well-understood, however, are the obstacles faced by minorities within rural communities. despite a common stereotype of ethnic homogeneity, rural areas are home to significant proportions of vulnerable demographics and host a unique demographic makeup. the distinctions between rural and urban healthcare are mirrored by the distinctions between rural minorities and the majority. furthermore, this diversity is growing due to the growth of low-wage industries. the number of minorities in rural areas increased by 20% between 2000 and 2010, while the number of rural whites remained largely the same [18]. just as rural communities have unique needs, cultures, and contexts compared to metropolitan areas, so do individual minority groups. issues that independently plague rural and minority populations overlap in the doubly-underserved rural minorities, damaging patient outcomes [19]. understanding these growing demographics is critical to serving their unique healthcare needs. medicine that understands and addresses the impact of rural minority standing columbia undergraduate science journal vol. 16, 2022 41 is better able to reach these vulnerable populations. this review will first identify several prominent demographics of rural areas: the disabled, native americans, veterans, and elderly. these are explored as illustrative case studies due to their high rural prevalence and availability of research. their distinguishing factors are highlighted and, as these separations are identified, the overarching themes that unite these demographics emerge. common ideas, observations, and challenges throughout rural minority health research are then explored. keynote research motifs include ongoing research gaps on both rural health and rural minorities, leading to unaddressed issues and uninformed, ineffective policy. this contributes to another recurring theme, the invisibility of rural minorities in both public policy and the research world; the enduring struggles of these demographics is testament to this. finally, the powerful negative effects of overlapping minority statuses continuously appears throughout research. concluding this review is an analysis of these research themes and current deficiencies in our knowledge of these complex demographics. overview of rural minority health the rural-urban dichotomy in geography, culture, and historical context has been well-described with demonstrated, quantifiable impacts on health. however, distinctions between rural demographics are less recognized. for example, ethnic disparities in rural communities remain largely overlooked and undiscussed in literature [20]. non-racial and non-ethnic demographic factors, such as the traits that define veteran, elderly, and disabled groups, are equally forgotten, rendering these groups invisible. small population sizes and lack of representation in healthcare further obscure rural minorities, and underrepresentation of the rural perspective among medical providers and researchers encourage stereotypes of homogeneity. this melange of factors occludes the recognition of specific needs and contextual factors both in literature and in medical practice. this theme of unacknowledged distinctiveness underpins rural minority health research. such lack of recognition conceals the desperate status of rural minority health. a strong recurring theme in literature is how intersecting geographic and minority disadvantages damage patient outcomes. as medically-underserved groups make up great amounts of rural populations, substantial amounts of people suffer from these intersections. for example, spatial isolation and low healthcare accessibility disproportionately impact rural minorities compared to their urban counterparts or the rural majority [20,21]. compounding these common challenges are the unique struggles of individual groups, such as the powerful yet cryptic effects of social and cultural contexts. apostolopoulos et al. see syndemics as the defining factor of rural minority health, describing the pathogenic interaction of biological and social factors [22]. the effects of syndemics appear frequently in literature; even cursory reviews of rural minority research underscore their obvious strength. for example, among black americans, several struggles – structural racism, inequitable social, education, labor, and health policies, and economic – are linked to trauma, stress, and syndemics [22]. rurally, research regarding the magnitude of hardship for minorities takes disproportionately little notice of these contextual effects. this invisibility amalgamates with other factors, such as high and increasing rates of rurally-concentrated poverty and high employment in dangerous industries like manufacturing, to injure rural minority health [23]. furthermore, for non-white rural residents, columbia undergraduate science journal vol. 16, 2022 42 the white-dominated rural setting means that specific cultural needs and experiences find little understanding. coethnic networks and communities are sparse due to diffuse populations, and culturally-intelligent resources are few and far between. a prime example is in the rural native american community, which struggles with fragmented healthcare services, low funding, and lack of community support [16]. unsurprisingly, its members experience remarkably poor health even compared to other rural underserved [2]. all of these factors combine to produce visible, quantifiable damage to health. rural minorities have lower life expectancies, see higher mortality rates, rate their healthcare as lower quality, receive fewer preventative services, and suffer higher rates of chronic disease compared to their urban counterparts [6,24]. table 1 summarizes health statistics in rural ethnic and racial demographics [20]. the complex aggregation of social, cultural, and geographic factors generates the current healthcare plight of rural minorities. “health outcomes are inexorably linked to zip code, are notably worse for the millions who live in rural and underserved communities and are further compounded by health disparities common among racial and ethnic groups.” [21] interventions to address social determinants of health are equally as important as secondary prevention and treatment efforts yet remain underutilized in policy due to poor recognition of these determinants. [25] research is limited for all rural minorities, but available literature reveals the distinctive qualities of each group as well as overarching patterns that unite them. understanding the interplay of unique challenges and broader factors will allow for more powerful interventions in these underserved communities. native americans native americans exemplify the struggles of rural minorities, experiencing some of the poorest health outcomes in america despite their intimate ties to rural land: in 2010, 29% of native americans lived in rural areas compared to 15% of the u.s. population [26]. furthermore, where other rural ethnicities have experienced gradual but measurable gains in healthcare access and quality, lingering historical biases and social factors continue to negatively affect native american health [27]. notably, rural native americans suffer one of the highest risks of maternal and infant mortality in the u.s [21,28], high rates of mental illness, stress [27], substance abuse [29], and suicide [24], and damaged mental wellbeing as well [4]. native americans also suffer higher mortality rates of preventable disease. per 100,000 people, pneumonia and influenza kill 26.6 native americans per year versus 15.1 for all races, and heart disease kills 194.7 native americans as opposed to 179.1 for all races [29]. unsurprisingly, the rural-urban disparity in life expectancy is exaggerated for native americans, where a gap of 11 years stands between metropolitan and nonmetropolitan indigenous lifespans [1]. economic obstacles hinder native healthcare, as well: 33% of rural native americans report recent problems accessing healthcare, and 28% recently experienced major problems paying medical bills. high costs do not indicate high quality, either, as 28% of rural native americans reported recent problems with quality of healthcare [30]. perhaps one of the greatest issues faced by this group is healthcare fragmentation. “services and support for health and social programmes are typically fragmented in indigenous populations… fragmentation results in the isolation of symptomatic issues— addiction, suicide, fetal alcohol syndrome, poor columbia undergraduate science journal vol. 16, 2022 43 housing, and unemployment—followed by the design of stand-alone programmes to try to manage each issue separately.” [31] fragmentation is bred by chronic underfunding, small populations, and long geographic distances [32], and as a population known for its small size and economic hardship, indigenous health is the paragon of such fractionation. fragmented healthcare services interrupt cohesion between services, leading to poor outcomes, inaccessibility, and inefficient resource use [33]. the indian health service (ihs), which only provides non-comprehensive health insurance and experiences chronic underfunding, suffers extensively from fractionation [29]. as many rural american indians/alaska natives receive healthcare through the ihs (approximately 1.5 million of 4.1 million native americans [34]), a great number of rural indigenous encounter fractionated, underfunded healthcare: 46% of american indian/alaskan natives receiving care through the ihs experienced funding shortfalls and subsequent reductions in healthcare [29]. more factors than just the ihs affect rural native american healthcare, such as stark socioeconomic status. compared to the total u.s. population, native americans working fulltime earned less and were more likely to live in poverty compared to non-hispanic whites [26]. there are also substantial differences between the experiences of rural native americans and whites not only in healthcare discrimination, but also in police treatment, racial violence, and housing [30]. health outcomes are not produced in a vacuum; social factors influence healthcare at all levels, from policymaking to patient-provider interactions, to generate palpable impacts [21,25]. rural black and hispanic americans suffer the same pattern, where structural racism influences health to damage outcomes [35]. from the poorer healthcare outcomes across all rural ethnic minorities, we see how minority standing in majority-white rural areas affects patient outcomes. interventions in rural indigenous healthcare struggle to address these. prominently, even the introduction of the ihs, specifically created to address native american healthcare, made little change in native american healthcare over time [29]. other interventions outside of the ihs are primarily targeted at individuals and communities [63]. while helpful, the greater systemic pressures on rural native americans remain, thus no substantial, national change in indigenous health has been accomplished. rural native americans’ twice-erased status in locality and ethnicity contributes to their current healthcare struggles. thus, the healthcare effects of such a unique context are forgotten. as indigeneity is an inherently social and cultural concept, poor indigenous health is rooted in cultural factors such as loss of community, lack of land connection, and feelings of spiritual, emotional, and mental disconnectedness [31]. this disconnection is a great mental burden: the rural indigenous experience the highest suicide rate out of all rural demographics, double that of rural whites and twelve times that of rural africanand asian-americans [24]. the fractionated state of rural native healthcare mirrors this disconnection. overall, the rural indigenous have not received the benefit of medical advancements, leading to stagnated improvement in health [21]. the pervasive stereotype of rural communities as homogenous oversimplifies them, rendering the diverse experiences of ethnic and racial constituents invisible. such invisibility affects healthcare at all levels, from public health initiatives and funding to patient-provider interactions. this invisibility is woven throughout rural health research, not just among the native american population. other columbia undergraduate science journal vol. 16, 2022 44 rural health researchers have commented on the difficulty of finding high-quality samples, trends, and pre-existing research. the issues dogging the unseen rural indigenous -economic, social, and cultural, not just in healthcare -have halted improvement in their health as others advance. disability rural populations have markedly high proportions of disability. compared to metropolitan adults, rural adults are 9% more likely to report having any disability, 24% more likely to report having more than one disability [36], and are more likely to have hearing or vision loss [4]. low workforce participation and economic constraints facilitate the high rates of poverty experienced by disabled households [37]. despite the substantial presence of disability in rural areas, the rural disabled are a rare topic in healthcare research [21]. accessibility frames the daily lives of disabled people. in a healthcare context, disability management and quality of life are contingent on accessible, high-quality care, as the disabled community requires more health services than abled persons [38]. thus, rural scarcity of both specialty and primary care, a necessity in the lives of many disabled people, disproportionately impacts this population. geographic inaccessibility defines rural areas, where vast distances between services make car travel essential. as driving requires well-functioning sight, hearing, cognitive ability, and mobility, many disabled people cannot drive long distances and struggle to access care. provider scarcities build only more obstacles, as many disabled people struggle to locate specialist care for disability-specific services [38]. the vacuum of nonurban specialty care impacts the health of the rural disabled, who need these services most. rural residents rely heavily on primary care providers as well, even to manage conditions largely overseen by specialists [12]. regardless of the fact that primary care cannot provide the same services as a specialist, the shortage of rural primary care inordinately affects the rural disabled, for whom consistent access to healthcare is critical. low socioeconomic status magnifies the effect of provider scarcity. many disabled adults (42.0%) depend on social security disability insurance or supplemental security income as well, including services such as medicare [38]. as these public programs pay around half of what private insurance pays practitioners [39], providers have less incentive to care for rural disabled, further limiting care options. financial barriers also impede access to care: working-age adults with disabilities are much more likely than those without to report inability to pay medical bills, problems paying medical bills, and not accessing care due to cost [38]. a recurring theme throughout rural minority healthcare is the numerous overlaps between marginalized populations. few demographics exemplify this better than rural disability. non-white populations [37] and veterans experience higher rates of disability and veterans have high rates of disability as well [40]. the elderly often encounter disability as they age, and healthcare obstacles in these separate populations frequently overlap, as in transportation struggles, high incidence of poverty, and dependence on public programs. overall, frequent overlaps with underserved groups results in rural disability generating more destitute outcomes. disability and rurality have a close relationship that directly affects healthcare and quality of life. several layers of inaccessibility hinder rural health, from financial to geographic, and the overlap of disability status with other individual factors lends further comcolumbia undergraduate science journal vol. 16, 2022 45 plexity to rural disabled healthcare. however, research into the rural disabled is noticeably lacking despite a critically underrepresented status. for example, the interplay of disability with other minority standings is known in only very broad, general trends. financial impacts remain understudied as well. further research to characterize rural disability is needed to fully understand its effects and its intersections with other groups. veterans rurality and veteranship are closely tied, with approximately 4.7 million veterans currently living in rural and highly rural areas [41]. unsurprisingly, the issues that plague rural healthcare merge with veteran-specific traits to damage rural veteran health, resulting in low health-related quality-of-life scores for rural veterans compared to their nonrural counterparts [42]. institutional underfunding and high intrinsic rates of disability are some of the prominent factors affecting the health of rural veterans. similar to the struggles of the ihs, the veterans health administration (vha) has seen regionalization and cutting of rural services due to small service populations and limited finances [7,43]. dilute service populations require the establishment of large referral regions and, as a result, vha tertiary care referral centers tend to be located in urban areas even if many clients are rurally-located [43]. this creates referral systems that are geographically separated from patients. for veterans, this disconnection exacerbates foundational issues in rural healthcare, generating long travel times and augmenting the lack of specialty care. as illustrated in the rural indigenous, fractionation and defunding of healthcare is common among the rural underserved and damages healthcare outcomes. as found above, disability frequently intersects with other minority statuses to impact health, exemplified by the high rates of disability among veterans. 29% of veterans are classified by the us government as having a service-connected disability [40] compared to the general disability rate of 4.3% [44]. service-connected disabilities include physical maladies as well as mental illness, with mental illness being a defining veteran’s issue. of veterans utilizing veterans affairs health services, 21.8% are diagnosed with post-traumatic stress disorder and 17.4% are diagnosed with depression, proportions much higher than in civilians [44]. high disability rates affect more than just health, contributing to, for example, high rates of household poverty. 13.19% of disabled veterans are in poverty compared to 5.51% of non-disabled veterans, damaging quality of life and introducing financial barriers to healthcare [37]. mental illness is a prominent veteran struggle due to traumatic experiences and military culture discouraging the use of behavioral healthcare. veteran suicide rates increased by 15% for men and 35% for women from 2001 to 2010 [45]. rural veterans struggle even more: physical and mental health scores are far lower for nonurban veterans compared to urban veterans, showing the larger burden of mental illness for rural veterans [7]. the lack of care sensitive to veterans’ particular needs and context, such as for mental health, fuels these struggles. if rural healthcare is sparse in general, there are less informed on the specific circumstances of veterans. military culture, with values of strength and independence, impacts how veterans approach healthcare; effective healthcare for veterans is sensitive to these cultural considerations and the distinct obstacles veterans face [40]. though veteranship holds unique contextual considerations for healthcare, recurring challenges across minorities impact columbia undergraduate science journal vol. 16, 2022 46 this population. demographic-specific healthcare organizations struggle to support diffuse rural populations, demonstrated in rural native americans with the ihs and in veterans with the vha. similarly, the lack of veteran-specific care echoes the cultural gaps that impact the rural indigenous. predominately white providers struggle with the specific issues surrounding the indigenous; metropolitan providers struggle to fully appreciate rural culture; the effects of military culture often go unrecognized by civilian providers. military values emphasize “the physical and mental well-being of soldiers, as well as their independence and self-reliance…” discouraging veterans from seeking healthcare [40]. nonmilitary providers unaware of these considerations thus struggle to reach veterans. coupled with rural obstacles to care, there is little incentive for veterans to pursue healthcare, facilitating health disparity. rural veterans, by virtue of both their locality and life experiences, represent a deeply underserved population. thinly-spread health services, lack of funding, high disability and poverty rates, poorly-defined referral areas, and care uninformed on military culture combine to generate uniquely poor health circumstances for rural veterans. this mixture of struggles incorporates wider issues in rural health and unique contextual obstacles. the poorer health of veterans, reflected in high suicide rates and reduced quality of life, is rooted in many different factors, difficult to untangle from each other but powerful in their combined effect. elders quiet, idyllic, and slow-paced, rural locales make popular retirement destinations. 25% of the rural population is age 65 or older, though the elderly constitute 12% of the u.s. population [48]. more americans are reaching retirement age and increasing this proportion, foreshadowing growth of current issues with provider shortages [49]. the effects of this shortage are compounded by the intensive healthcare needs of elderly patients. the medical intricacy of aging merges with issues in service access and provider shortages to establish the elderly as a medically vulnerable population. rural geography and overlaps with other minority statuses only exacerbate these issues. elderly patients benefit from specialty healthcare that is trained in complex and aged patient care [48,50]. however, a drastic shortage in geriatricians is developing, especially in rural areas [51]: for example, 56% of geriatric fellowship spots remained unfulfilled in the 2015 main residency match of the national resident matching program [52]. rural elders thus suffer not just from broad deficiencies in rural specialty care, such as cardiology for age-related heart disease, but for a demographic-specific specialty trained in their specific needs. mental health is also severely underemphasized in the elderly. social and spatial isolation plague the rural elderly [49], yet psychological services to cope with such loneliness are scarce [54]. suicide risk increases significantly after age 65, and 18% of completed suicides in 2019 were by older americans [55]. aside from specialty care, rural elders encounter pronounced challenges in primary care access, such as physical frailty and long geographic distances [5]. compounding this hardship is insufficient public transportation in rural areas [17]. care that would circumvent these physical barriers, such as home-based support services, is lacking. limited provider availability, inadequate transportation, poor telecommunications access (i.e. poor internet and technological literacy in elders), and low caregiver columbia undergraduate science journal vol. 16, 2022 47 recruitment and retention plague these services [14,17]. the same issues that haunt all of rural healthcare impact home health, damaging the health of rural elders who need these services most. socioeconomic factors also impact the elderly. high poverty rates in rural elders introduce financial barriers to care. compared to urban areas, rural elders have lower incomes and are more likely to be below the poverty line. although they are more likely to own their homes, those homes are more likely to be substandard [55]. greater reliance on medicaid and social security may also limit provider options [55]. long distances to work or support services only worsen financial issues. overall, vast distances may impose insurmountable barriers to healthcare, employment, prescriptions, or social services for rural elders [56]. as american lifespans stretch longer and more retirees seem to seek calm, idyllic towns and views, healthcare in the rural elderly will only become more important. these elders, however, will find only sparse healthcare services, whether in primary care, specialty care, or home health. frequent intersections with other marginalized statuses mar the health outcomes of rural elderly as well. for example, 56% of the 2.8 million rural veterans enrolled in the vha are over the age of 65 [57]. furthermore, the proportion of ethnic and racial minority groups among the american elderly is expected to rise sharply in coming years: the percentage of elderly non-hispanic whites is expected to drop from 78% in 2014 to 55% in 2060 [49]. throughout rural health, overlapping circumstances facilitate poor outcomes, and rural elders follow the same pattern [56]. little research has been completed about these overlaps, however, leaving a number of unanswered questions. ongoing knowledge gaps the experiences of rural minorities are important to consider in rural health due to the surprising heterogeneity of rural areas. factors that independently affect minority and rural populations overlap to damage healthcare quality and access. the intricacy and significant numbers of rural minorities make them critical aspects of rural areas, yet are often overlooked. current research into rural minority health highlights unique challenges of each demographic and implies universal challenges that unite them. some overarching patterns include the magnification of extant rural health issues for minorities, socioeconomic obstacles, and underfunded programs for specific demographics. of these patterns, persistent invisibility in their own communities and healthcare itself is the most consistent and impactful. such invisibility camouflages the impact of rural minority standing on health, causing it to be overlooked and understudied. large, ongoing gaps in our knowledge unify rural minority research, with rural native americans being a prime example of these gaps. the indigenous are a group already poorly-understood, with the rural indigenous only doubly so. small samples and inconsistencies in research methods, such as how tribes and ethnic groups are merged, make the indigenous difficult to research and contribute to the lack of literature surrounding them [33]. baldwin et al. report that, even for a basic measure such as mortality rate, “few studies have reported [american indian/alaska native] mortality nationally and stratified by residence location.” [58] other minorities suffer the same invisibility. for the disabled community, little is known about disability prevalence and types of disability by urbanization level [36]. for rural elders, demographic-specific issues such as elder mistreatment are known columbia undergraduate science journal vol. 16, 2022 48 to be understudied, and such issues affect rural elders even more due to geographic obstacles in knowledge and intervention efforts [49]. as for veterans, there is little information on current suicide rates, let alone historical trends, who make up great proportions of rural areas [45]. multiple-minority demographics are doubly affected by these gaps and remain chronically understudied. knowledge gaps are particularly wide in rural minority mental health. this is deeply concerning for a number of reasons. as both rural and minority populations experience separate shortages in mental health services, a poor understanding of these overlaps is deeply detrimental [15]. jensen et al. argue that rural areas require stronger interventions to overcome mental health deficits, as the combination of a distinct culture and strong, established barriers to behavioral health care damage accessibility and quality [59]. they echo king et al. who, in reference to rural native americans, argue that further research into social determinants of health will form better interventions [31]. across demographics and healthcare specialties, research agrees that a broad lack of research makes effective solutions difficult to form. more broadly, research into rural healthcare itself is lacking. many studies are limited to phone surveys or analyses of aggregated hospital data, a shaky choice of data collection considering the rapid closures of rural hospitals [60] and rural dependence on small clinics and private practices. general consensus holds that rural health is poorer than that in urban areas, but large-scale studies in specific health indicators are scarce. for example, even as the us has defined increasing life expectancy as a key goal [61], few studies analyze rural-urban life expectancy disparities and factors affecting lifespan. literature examining the plight of rural specialty care is also diffuse. cyr et al. find an “...insufficient understanding of differences in facilitators and barriers between u.s. urban versus rural specialty care. while conceptual frameworks exist to guide these efforts... none specifically focus on us urban versus rural specialty care.” [62] a lack of research leaves the details of rural health unexplored. the depth and persistence of these knowledge gaps may render interventions in rural minority health ineffective. for example, apostolopoulos et al. identify several common yet low-leverage interventions (e.g., self-management education) often employed in rural communities that generate underwhelming results [22]. our thready understanding of rural areas, frequently informed by stereotypes and assumptions, obscures more effective solutions. thus, the health of the rural underserved stagnates and declines. previous literature unanimously agrees that rural communities and their substituent minority groups are distinct from both urban populations and each other. each demographic has particular needs, values, and problems influenced by unique contextual and cultural factors; however, research has not fully studied the effect of intersecting minority statuses. such research gaps preclude a complete understanding of the unique context of minorities, impeding the implementation of solutions to uplift individual demographics and rural communities as a whole. a richer understanding of the rural minority experience -their unique, defining factors, their prevalence, and their historical context -will illuminate effective solutions and additional factors to bridge the gap between them and their urban counterparts. targeted research into rural minorities is a direct pathway to this deeper understanding. as shown in literature, there are several unanswered questions about the specific challenges rural minorities face, occluding effective solucolumbia undergraduate science journal vol. 16, 2022 49 tions. there is a wealth of information still to be gained about these highly complex, incredibly interesting groups. greater research efforts will increase public awareness and direct effective interventions, creating decisive changes to overcome these ongoing challenges. figures author information corresponding author benjamin blue (university of california, davis) amblue@ucdavis.edu author contributions benjamin blue, as primary author, performed all research and preparation of this manuscript. funding sources no external funding was used in the preparation of this manuscript. competing interests the author declares no competing financial and non-financial interests. references [1]. g. singh, m. siahpush, widening rural–urban disparities in life expectancy, u.s., 1969– 2009. am. j. of preventive medicine 46.2, e19-e29 (2014). [2]. j. saint onge, s. smith., demographics in rural populations. surg. clin. north am. 100.5, 823-833 (2020). [3]. k. matthews et al., health-related behaviors by urban-rural county classification — united states, 2013. mmwr surveill. summ. 66.5, 1-8 (2017). [4]. j. schiller, j. lucas, j. peregoy, summary health statistics for u.s. adults: national health interview survey, 2011. vital health stat. 10.256, 1-218 (2011). [5]. j. dixon, “the effect of obesity on health outcomes.” mol. cell. endocrinol. 316.2, 104108 (2010). [6]. america’s health insurance plans, “serving rural america: how health insurance providers break down barriers to ensure access to care.” (ahip reports, 2019). [7]. a. wallace, w. weeks, s. wang, a. lee, l. kazis, rural and urban disparities in health-related quality of life among veterans with psychiatric disorders. psych. serv. 57.6, 851-856 (2006). [8]. s. woolf, h. schoomaker, life expectancy and mortality rates in the united states, 19592017.” jama 322.20, 1996-2016 (2019). table 1. health metrics of rural demographics, reporting period 2012 2017 selected data concerning healthcare access and health of rural demographics. * indicates significant difference from non-hispanic white. adapted from james et al. columbia undergraduate science journal vol. 16, 2022 50 [9]. national institute of health, “nih health disparities strategic plan and budget: fiscal years 2009-2013.” (national institute of health, 2009). available from https://www.nimhd.nih. gov/docs/2009-2013nih_health_disparities_ strategic_plan_and_budget.pdf. [10]. l. hart, e. larson, d. lishner. rural definitions for health policy and research. am. j. public health 95.7, 1149-1155 (2005). [11]. t. thomas, r. diclemente, s. snell, overcoming the triad of rural health disparities: how local culture, lack of economic opportunity, and geographic location instigate health disparities. health ed. j. 73.3, 285-294 (2013). [12]. r. rosenblatt, f. chen, d. lishner, m. doescher, “the future of family medicine and implications for rural primary care physician supply.” final report #125. (wwami rural health research center, 2010). [13]. s. letvak, the importance of social support for rural mental health. issues ment. health nurs. 23.3, 249-261 (2002). [14]. k. gonzalez, m. shaughnessy, e. kabigting, d. west, j. robinson, q. chen, p. fahs, a systematic review of the health of vulnerable populations within u.s. rural societies. online j. rural nurs. health care 18.1, 112-147 (2018). [15]. d. morales, c. barksdale, a. beckel-mitchener, a call to action to address rural mental health disparities. j. of clin. transl. sci. 4.5, 1-20 (2020). [16]. m. meit et al., “the 2014 update of the rural-urban chartbook.” (rural health reform policy research center, 2014). [17]. d. siconolfi et al., rural-urban disparities in access to homeand community-based services and supports: stakeholder perspectives from 14 states. j. am. med. dir. assoc. 20.4, 503-508 (2019). [18]. d. lichter, immigration and the new racial diversity in rural america. rural sociol. 77.1, 3-35 (2012). [19]. k. kozhimannil, c. henning-smith, racism and health in rural america. j. health care poor underserved 29.1, 35-43 (2018). [20]. c. james, r. moonesinghe, s. wilson-frederick, j. hall, a. penman-aguilar, k. bouye, racial/ethnic health disparities among rural adults — united states, 2012–2015. surveillance summaries 66.23, 1-8 (2017). [21]. rural and underserved communities health task force, “left out: barriers to health equity for rural and underserved communities.” (u.s. house of representatives committee on ways and means, 2020). [22]. y. apostolopoulos, m. lemke, n. hosseinichimeh, i. harvey, k. lich, j. brown, embracing causal complexity in health disparities: metabolic syndemics and structural prevention in rural minority communities. prev. sci. 19.8, 1019–1029 (2018). [23]. t. farrigan, t. parker, the concentration of poverty is a growing rural problem. amber waves 4 (2012). [24]. web-based injury statistics query and reporting system, fatal injury reports. centers for disease control and prevention. available at https://www.cdc.gov/injury/wisqars/fatal. html. accessed 11/14/2021. columbia undergraduate science journal vol. 16, 2022 51 [25]. r. thornton, c. glover, c. cené, d. glik, j. henderson, d. williams, evaluating strategies for reducing health disparities by addressing the social determinants of health. health aff. 35.8, 1416–1423 (2015). [26]. s. ogunwole, “we the people: american indians and alaska natives in the united states.” (u.s. census, 2006). [27]. j. waldram, a. herring, t. young, aboriginal health in canada (university of toronto press, 2006). [28]. d. ely, a. driscoll, infant mortality in the united states, 2018: data from the period linked birth/infant death file. national vital statistics reports 69.7, 1-17 (2020). [29]. d. warne, l. frizzell, american indian health policy: historical trends and contemporary issues. am. j. public health 104.3, s263– s267 (2014). [30]. m. findling, r. blendon, j. benson, c. miller, the unseen picture: issues with health care, discrimination, police and safety, and housing experienced by native american populations in rural america. j. rural health 36, (2020). [31]. m. king, a. smith, m. gracey, indigenous health part 2: the underlying causes of the health gap. lancet 374.9683, 76-85 (2009). [32]. council on graduate medical education, “physician distribution and health care challenges in rural and inner-city areas.” (department of health and human services, 1998). [33]. l. trout, c. kramer, l. fischer, social medicine in practice: realizing the american indian and alaska native right to health. health hum. 20.2, 19-30 (2018). [34]. m. cahn, s. harvey, m. town, american indian and alaska native men's use of sexual health services, 2006-2010. perspectives on sexual and reproductive health 49.3, 181-189 (2017). [35]. d. nguyen, k. ho, j. williams, social determinants and health service use among racial and ethnic minorities: findings from a community sample. soc. work health care 50.5, 390405 (2011). [36]. g. zhao, c. okoro, j. hsia, w. garvin, m. town, prevalence of disability and disability types by urban–rural county classification—u.s., 2016. am. j. prev. med. 57.6. 749-756 (2019). [37]. a. london, c. heflin, j. wilmoth, work-related disability, veteran status, and poverty: implications for family well-being. j. poverty 15.3, 330-349 (2011). [38]. j. kennedy, e. wood, l. frieden, disparities in insurance coverage, health services use, and access following implementation of the affordable care act: a comparison of disabled and nondisabled working-age adults.” inq. 54, 1-10 (2017). [39]. e. lopez, t. neuman, l. levitt, g. jacobson, how much more than medicare do private insurers pay? a review of the literature. (kff, 2012). available at https://www.kff.org/ medicare/issue-brief/how-much-more-thanmedicare-do-private-insurers-pay-a-review-ofthe-literature/. [40]. c. griffin jr., m. stein, self-perception of columbia undergraduate science journal vol. 16, 2022 52 disability and prospects for employment among u.s. veterans. work 50.1, 49-58 (2014). [41]. veterans affairs office of rural health, rural veteran healthcare challenges. (veterans affairs, 2012). available at https://www.ruralhealth.va.gov/aboutus/ruralvets.asp [42]. a. wallace, r. lee, t. mackenzie, a. west, s. wright, b. booth, k. hawthorne, w. weeks, a longitudinal analysis of rural and urban veterans’ health-related quality of life. j. rural health 26.2, 153-163 (2010). [43]. r. tessler, r. rosenheck, g. gamache, declining access to alcohol and drug abuse services among veterans in the general population. military medicine 170, 234-238 (2005). u.s. census bureau, 2020 census acs 1-year data. (u.s. census bureau, 2020). [44]. k. seal, d. bertenthal, c. miner, s. sen, c. marmar, bringing the war back home: mental health disorders among 103,788 us veterans returning from iraq and afghanistan seen at department of veterans affairs facilities. arch. intern. med. 167.5, 476-482 (2007). [45]. j. mccarten, c. hoffmire, r. bossarte, changes in overall and firearm veteran suicide rates by gender, 2001–2010. am. j. prev. med. 48.3, 360–364 (2015). [46]. a. skoufalos, j. clarke, e. ellis, v. shepard, e. rula, rural aging in america: proceedings of the 2017 connectivity summit. popul. health manag. 20.s2, s1-s10 (2018). [47]. institute of medicine. retooling for an aging america: building the health care workforce (national academies press, 2008). [48]. c. dyer, c. murdock, j. hiner, j. halphen, j. burnett, elder mistreatment intervention: strategies for connecting with diverse and rural populations. generations 44.1, 91-97 (2020). [49]. a. hintenach, o. raphael, w. hung, training programs on geriatrics in rural areas: a review. curr. geriatr. rep. 8.2, 117-122 (2019). [50]. a. golden, m. silverman, m. mintzer, is geriatric medicine terminally ill? ann. intern. med. 156, 654–656 (2012). [51]. national resident matching program, “results and data: specialties matching service, 2015 appointment year.” (national resident matching program, 2015). available from http:// www.nrmp.org/wp-content/uploads/2015/02/ results-and-data-sms-2015.pdf. [52]. m. kaplan, n. huguet, b. mcfarland, j. mandle, factors associated with suicide by firearm among u.s. older adult men. psychol. men masc. 13.1, 65-74 (2012). [53]. r. hudson, lack of social connectedness and its consequences. public policy & aging report 27.4, 121–123 (2017). [54]. a. coburn, e. bolda, rural elders and long-term care. west. j. med. 174.3, 209-213 (2001). [55]. b. collins, t. borders, k. tebrink, k. xu, utilization of prescription medications and ancillary pharmacy services among rural elders in west texas: distance barriers and implications for telepharmacy. journal of health and human services administration 30.1, 75-97 (2007). [56]. department of veterans affairs, “fy2020 columbia undergraduate science journal vol. 16, 2022 53 funding and fy2021 advance appropriations, volume ii medical programs and information technology programs.” (department of veterans affairs, 2019). [57]. a. scharlach, e. fuller-thomson, b. kramer, “curriculum model on aging and ethnicity.” (university of california berkeley school of social welfare, 2002). [58]. l. baldwin, d. grossman, s. casey, w. hollow, j. sugarman, w. freeman, l. hart, perinatal and infant health among rural and urban american indians/alaska natives. am. j. public health 92.9, 1491–1497 (2002). [59]. e. jensen, e. wieling, t. mendenhall, a phenomenological study of clinicians’ perspectives on barriers to rural mental health care. j. rural ment. health 44.1, 51-61 (2020). [60]. b. kaufman, s. thomas, r. randolph, j. perry, k. thompson, g. holmes, g. pink, the rising rate of rural hospital closures. the journal of rural health 32.1, 35-43 (2016). [61]. healthy people 2020. healthy people 2020. (office of disease prevention and health promotion, 2014). available at https://www. healthypeople.gov/2020. [62]. m. cyr, a. etchin, b. guthrie et al., access to specialty healthcare in urban versus rural us populations: a systematic literature review. bmc health serv res 19, 974 (2019). [63]. v. blue bird jernigan, e. j. d’amico, b. duran, d. buchwald, multilevel and community-level interventions with native americans: challenges and opportunities. prevention science 21, 65–73 (2018). columbia undergraduate science journal vol. 16, 2022 volume 14 | 2020 | cusj.columbia.edu atomic materials analysis: correction of self-absorption distortion in xanes page 9 administration letter from the editor-in-chief…………………………………………..3 letter from the president………………………………………………...5 faculty advisory board and editorial board…………………………...7 articles characterization and correction of self absorption distortion in xanes aidan p. reddy, apurva mehta………………………………………….9 protein-polymer complex coacervates as synthetic membraneless organelles hansen tjo, nicholas zervoudis, allie c. obermeyer……………....21 i’m upset, not ashamed: an investigation into adolescent shame acknowledgement olivia putnam, keely lake……………………………………………..27 contents 1 2 dear readers, it is with great excitement that i am able to present to you the cusj volume 14 2020 edition. since our very first volume in 2006, it has been the mission of cusj to provide an accessible and professional, open access platform for undergraduates to communicate their research advances in various scientific fields. we cusj editors believe, as a community of young scientists and writers, that it is absolutely essential for researchers of all ages and levels of experience to have the opportunity to share their ideas and learn from their peers. creating a space to facilitate that discussion is exactly what cusj aims to achieve not only with each volume, but also with our outreach to the public and numerous research oriented events. covid-19 has posed an incredibly difficult and tragic challenge to communities all over the world, scientists and non-scientists alike. with the shutdowns of universities accompanied by sudden changes in employment status and housing, the perseverance and determination shown by both our authors and editing team has been a true testament to their character, work ethic, and resilience. the hard work evident in each article of this volume, from experimental design to data interpretation, is a display of the talent and strength of our authors, and their ability to push on through an unprecedented global emergency. these undergraduates have shown that even during the most trying of times, while physically isolated, we can all still come together and communicate through science. volume 14 of cusj is diverse in scientific fields represented, ranging from biochemistry and the development of enzymatic nanoreactors for industrial synthesis, to the social science of shame in teens. our cover features a graph from author aidan reddy’s cutting-edge investigation into x-ray absorption spectroscopy data processing, in which he develops a new method for spectra correction. his work should allow for more accurate analysis of various materials’ atomic properties, advancing the fields of both materials science and physics. it has been a pleasure to serve as the editor-in-chief of cusj for the 2019-2020 academic year. my sincere thank you to everyone involved in the creation of this volume for their passion and commitment. julia parsley editor-in-chief letter 3 4 dear readers, founded in 2006, the columbia undergraduate science journal was created by a team of graduate students who wanted to provide a collegiate platform for undergraduates to publish their scholarship and to educate students about the academic publication process. with guidance from advisors and faculty, the editorial board strove to produce a professional-level, open-access science journal that provides participants with a valuable experience in publication. since then our board has shifted to being solely composed of undergraduates. while still maintaining our core editorial standards, the journal has evolved to more explicitly try and foster a supportive student research community on our morningside campus. it is this notion of community that can help us cope and process the unfamiliar circumstances of a global pandemic. in quarantine, student researchers can reflect on the utility of science and its capacity to contain and eradicate this crisis. the collegiate scientists of today should recognize this moment for its affirmation in the power of collaboration and community. recall this moment as you go onto make the next breakthroughs in the natural and applied sciences. this year, the columbia undergraduate science journal formalized our intentions for community building, made inroads in connecting previous authors and alumni of our publication, and developed our editorial process further with open journal infrastructure. i was impressed by the quality of our submissions this year and i want to thank each of the authors for their cooperation in the compilation of our journal. i am incredibly proud of what we have accomplished this year. thank you all for your support in the growth of the student research community. the columbia undergraduate science journal editorial board is proud to present the 14th annual columbia undergraduate science journal. congratulations to the authors, thank you to our readers, and enjoy! sincerely, jason mohabir president cusj letter 5 6 faculty advisory board hugh ducklow…………………………….……….professor of earth and environmental sciences ivana hughes………..director of frontiers of science, senior lecturer in discipline chemistry marko jovanovic………………………………………………...….professor of biological sciences laura kaufman………………………………………………………….……...professor of chemistry patricia lindemann...lecturer in discipline and director of undergraduate studies psychology kyle mandli……………………………………………….………..professor of applied mathematics matthew palmer…………………...senior lecturer and director of undergraduate studies e3b gerard parkin……………………………………………………………...…...professor of chemistry ron prywes………………………………………………………….professor of biological sciences angela rasmussen…….…….associate research scientist center for infection and immunity editors julia parsley…………..….editor-in-chief, cusj kate johnson….……..…...editor-in-chief, cjsj jason mohabir…………....…………….president alex ying………....……………director of events arya rao…………......director of internal affairs ethan chen……...…………..director of finance maria trifas……..………….director of outreach neeraj sakhrani…..director of communications ilaria simeone………………....…………...editor vikas chelur……...…………....…………...editor huck jun hong……………......…………...editor noah krever………………....…...………...editor elin hu………………................…………...editor harrison zhang…………..…....…………...editor elena gribelyuk………….…....…………...editor brendon choy………………....…………...editor isabella leite…………..……....…………...editor jacy fang……………...……....…………...editor katie long…………..………....…………...editor people ashley rosenberg……...…....…………...editor sophia a ladyzhets…....…....…………...editor deena shefter……………......…………...editor kanav kalucha…………….....…………...editor sahith vadada……………......…………...editor lydia wu………………….......…………...editor arjun kudinoor……………….........……...editor mayeesa rahman……………....………...editor aswath suryanarayanan…....…………...editor chloe gong………………......…………...editor ellen ren…………...…….......…………...editor avi gupta…………...…….......…………...editor shivali verma…………...........…………...editor hibah vora…………….….......…………...editor unal yigit ozulku………….....……….…...editor michael wang……………….......………...editor caroline haoud……………….…………...editor shamara yearwood……….....…………...editor 7 8 &kdudfwhul]dwlrq &ruuhfwlrq ri 6hoi�$evruswlrq 'lvwruwlrq lq ;$1(6 $lgdq 3� 5hgg\ dqg $sxuyd 0hkwd 6/$& 1dwlrqdo $ffhohudwru /derudwru\� ���� 6dqg +loo 5g� 0hqor 3dun� &$ ����� 7klv lv qrw wkh dxwkru¶v suhvhqw dgguhvv ;$1(6� ;�5$< $%62537,21 63(&7526&23<� 1($5 ('*( 6758&785( $%675$&7� ;�ud\ devruswlrq vshfwurvfrs\ �;$6� lv dq h[shulphqwdo whfkqltxh xvhg wr sureh wkh dwrplf surshuwlhv ri pdwhuldov� ;$6 phdvxuhphqwv shuiruphg lq ioxruhvfhqfh prgh duh vxemhfw wr vr�fdoohg ³vhoi�devruswlrq´ glvwruwlrq� hvshfldoo\ zkhq wkh vdpsoh xqghu lqyhvwljdwlrq lv wklfn ru frqfhqwudwhg zlwk uhvshfw wr wkh dwrp ri lqwhuhvw� +huh zh lqyhvwljdwh wkh ehkdylru ri vhoi�devruswlrq dqg suhvhqw vhyhudo vxjjhvwlrqv wr dyrlg dqg pdqdjh vhoi�devruswlrq wkurxjk h[shulphqwdo ghvljq� 3duwlfxoduo\� zh ilqg wkdw lw lv hdvlhu wr fruuhfw iru yhu\ wklfn vdpsohv wkdq iru vdpsohv ri lqwhuphgldwh wklfnqhvv� 6hoi�devruswlrq glvwruwlrq qhfhvvlwdwhv d fruuhfwlrq ri wkh phdvxuhg vshfwuxp gxulqj gdwd surfhvvlqj� :h suhvhqw d whfkqltxh iru fruuhfwlqj vhoi�devruehg vshfwud lq wkh ;$1(6 uhjlrq wkdw lv hiihfwlyh iru doo vdpsoh wklfnqhvvhv dqg h[shulphqwdo jhrphwulhv� ,1752'8&7,21 ;�ud\ devruswlrq vshfwurvfrs\ �;$6� lv d srsxodu whfkqltxh xvhg wr lqyhvwljdwh wkh dwrplf surshuwlhv ri pdwhuldov� ;$6 vshfwud lqglfdwh wkh devruswlrq frhiil� flhqw� µe�(�� ri d sduwlfxodu hohphqw dqg hgjh dv d ixqf� wlrq ri wkh hqhuj\ ri wkh lqflghqw ;�ud\� 7kh vwuxfwxuh ri wkhvh vshfwud uhyhdo hohfwurqlf dqg jhrphwulf surshu� wlhv ri d sduwlfxodu hohphqw zlwklq d vdpsoh� ;$6 vshf� wud pd\ eh glylghg lqwr wzr sulpdu\ uhjlrqv� ;�ud\ de� vruswlrq qhdu�hgjh vwuxfwxuh �;$1(6�� zklfk surylghv sulpdulo\ hohfwurqlf lqirupdwlrq (h�j� r[lgdwlrq vwdwh)� dqg ;�ud\ devruswlrq ilqh vwuxfwxuh �;$)6�� zklfk sur� ylghv sulpdulo\ jhrphwulf lqirupdwlrq (h�j� frruglqd� wlrq qxpehu)� ;$6 phdvxuhphqwv duh iuhtxhqwo\ wdnhq lq wudqv� plvvlrq prgh� lq zklfk rqh frpsduhv wkh wudqvplwwhg lqwhqvlw\ dqg lqlwldo lqwhqvlw\ ri wkh ;�ud\ ehdp wr gh� ulyh wkh devruswlrq frhiilflhqw ri wkh vdpsoh ri lqwhuhvw µe�(�� dffruglqj wr wkh uhodwlrqvkls µt�(� !t oq( i0i )� zkhuh 7 lv wkh vdpsoh wklfnqhvv� i0 lv wkh lqlwldo ;�ud\ lqwhqvlw\� i lv wkh lqwhqvlw\ ri wkh ;�ud\ diwhu wudyholqj wkurxjk wkh pdwhuldo� dqg µt�(� lv wkh wrwdo devruswlrq frhiilflhqw ri wkh vdpsoh� 7kh edfnjurxqg frpsrqhqw ri µt�(�� µb(e)� zklfk lv wkh devruswlrq frhiilflhqw gxh wr hohphqwv dqg hgjhv rwkhu wkdq wkh rqh ri lqwhuhvw� lv vxewudfwhg iurp µt�(� wr \lhog µe�(�� µe�(� lv wkhq gl� ylghg e\ µe0+ "µe(e0+�� zkhuh e0+ lv dq hqhuj\ mxvw deryh wkh hgjh� wr rewdlq µn�(�� wkh qrupdol]hg yhuvlrq ri µe(e)� 7udqvplvvlrq prgh phdvxuhphqwv uhtxluh wkdw wkh vdpsoh devrue d vljqlilfdqw sruwlrq ri wkh lqwhqvlw\ ri wkh lqflghqw ;�ud\ dqg wkxv wkdw wkh vdpsoh eh vxiil� flhqwo\ wklfn (odujh 7) ru rswlfdoo\ frqfhqwudwhg (odujh µt(e)) >�@� $qrwkhu frpprq ;$6 whfkqltxh zklfk grhv qrw uhtxluh wkhvh frqglwlrqv lv ioxruhvfhqfh prgh� ,q ioxruhvfhqfh prgh� rqh phdvxuhv wkh ioxruhvfhqfh hplwwhg iurp wkh vdpsoh dv d uhvxow ri wkh devruswlrq ri lqflghqw ;�ud\ skrwrqv wr lqihu µt�(�� )oxruhvfhqfh phdvxuhphqwv� xqolnh wudqvplvvlrq phdvxuhphqwv� fdq eh wdnhq iurp pruh wklq dqg rswlfdoo\ gloxwh vdpsohv� ,q ioxruhvfhqfh prgh� wkhuh lv d olqhdu uhodwlrqvkls eh� wzhhq wkh udwlr ri wkh phdvxuhg ioxruhvfhqfh lqwhqvlw\ wr wkh lqflghqw ;�ud\ lqwhqvlw\ dqg wkh devruswlrq frhi� ilflhqw ri wkh wdujhw hohphqw dqg hgjh� (if/i0)(e) # µe(e)� +rzhyhu� wklv rqo\ krogv zkhq wkh vdpsoh xq� ghu lqyhvwljdwlrq lv vxiilflhqwo\ wklq ru gloxwh zlwk uh� vshfw wr wkh hohphqw ri lqwhuhvw� )ru wklfnhu dqg pruh frqfhqwudwhg vdpsohv� wkhuh lv d qrqolqhdu uhodwlrqvkls ehwzhhq wkh ioxruhvfhqfh if i0 (e) dqg devruswlrq µe(e) vshfwud� dqg wkhlu vwuxfwxuhv gliihu dv d uhvxow� 7kh ydul� dwlrq ehwzhhq wkh ioxruhvfhqfh dqg devruswlrq vshfwud wkdw rffxuv iru vxiilflhqwo\ wklfn vdpsohv lv uhihuuhg wr dv vhoi�devruswlrq� 2qh vkrxog qrw uhdg wrr pxfk lqwr wklv � 9 qdph� ,w lv zlgho\ uhjdughg dv d plvqrphu� exw qrqhwkh� ohvv d zhoo�hvwdeolvkhg whup� ,w lv srvvleoh wr dyrlg vhoi�devruswlrq wkurxjk h[� shulphqwdo ghvljq� prvw vwudljkwiruzdugo\ wkurxjk wkh fkrlfh ru suhsdudwlrq ri d vdpsoh wkdw lv vxiilflhqwo\ wklq ru gloxwh zlwk uhvshfw wr wkh hohphqw ri lqwhuhvw� +rz� hyhu� iru ydulrxv sudfwlfdo uhdvrqv� wklv lv qrw dozd\v ihd� vleoh� ,q wklv fdvh� lw lv qhfhvvdu\ wr fruuhfw wkh phdvxuhg ioxruhvfhqfh vshfwuxp gxulqj wkh gdwd�surfhvvlqj vwdjh wr uhiohfw wkh vwuxfwxuh ri wkh ghvluhg vshfwuxp� µn�(�� $ qxpehu ri vhoi�devruswlrq fruuhfwlrq dojrulwkpv kdyh ehhq suhvhqwhg lq wkh sdvw >�� �� �@� 9dulrxv fruuhf� wlrq dojrulwkpv dsso\ wr gliihuhqw uhjlrqv ri wkh vshf� wuxp �;$1(6 ru ;$)6�� dqg hdfk kdv lwv rzq olpl� wdwlrqv� )ru lqvwdqfh� d srsxodu fruuhfwlrq dojrulwkp iru wkh ;$1(6 uhjlrq� )/82� lqfoxghg lq wkh srsx� odu ;$6 gdwd surfhvvlqj surjudp $7+(1$� dvvxphv wkdw wkh uhohydqw vdpsoh lv dw ohdvw d fhuwdlq wklfnqhvv� zklfk zh zloo glvfxvv ixuwkhu� 7kxv� wkhuh uhpdlqv d qhhg iru lpsuryhg vhoi�devruswlrq fruuhfwlrq phwkrgv zlwk juhdwhu dffxudf\ dqg zlghu dssolfdelolw\� +huh zh zloo h[dplqh wkh ghshqghqfh ri vhoi� devruswlrq rq ydulrxv h[shulphqwdo sdudphwhuv dqg surylgh h[shulphqwdo ghvljq vxjjhvwlrqv iru dyrlglqj dqg frqwuroolqj vhoi�devruswlrq� $gglwlrqdoo\� zh zloo suhvhqw d fruuhfwlrq phwkrg iru wkh ;$1(6 uhjlrq zklfk pd\ eh dssolhg wr d udqjh ri vdpsoh wklfnqhvvhv dqg h[shulphqwdo jhrphwulhv� :h zloo glvfxvv lwv hiil� fdf\ dqg olplwdwlrqv� 7kh h[shulphqwdo gdwd vhw xvhg wkurxjkrxw lv ri dk! hplvvlrq olqh ri d frsshu vdpsoh dqg zdv wdnhq e\ $sxuyd 0hkwd dqg 5\dq 'dylv� )ljxuh �� 6fkhpdwlf 'ldjudp ri ([shulphqw� ,q vwdqgdug jhrphwu\� ! = ± = 45� (yhq lq qrqvwdqgdug jhrphwulhv� w\slfdoo\ ! + " = 90� 0(7+2'6 &kdudfwhul]dwlrq ri 6hoi�$evruswlrq :h ehjlq zlwk wkh ioxruhvfhqfh ixqfwlrqdo� d zhoo� hvwdeolvkhg htxdwlrq wkdw ghvfulehv wkh udwlr ri ioxruhv� fhqfh lqwhqvlw\ wr wkh lqwhqvlw\ ri wkh lqflghqw ;�ud\� (if/i0)(e)� rqh zrxog phdvxuh jlyhq dq h[shulphqwdo jhrphwu\ dqg fkhplfdo frpsrvlwlrq ri d vdpsoh >�� �@� 7kh ixoo ghulydwlrq ri wklv htxdwlrq lv jlyhq lq dsshqgl[ $� if i0 = ( ! 4# )$(e) µe(e) µt(e) + µf vlq(") vlq(#) ! 1! e!(µe(e)+µb(e)+µf vlq(!) vlq(") ) t vlq(!) " ��� +huh� µf "µt(ef ) lv wkh wrwdo olqhdu devruswlrq frhiil� flhqw dw wkh ioxruhvfhqfh hqhuj\ ri wkh wdujhw hgjh �ef �� *hqhudoo\� wkh olqhdu devruswlrq frhiilflhqw µ ri dq duel� wudu\ pdwhuldo lv jlyhq e\ µ = # i µim%i zkhuh µim lv wkh pdvv devruswlrq frhiilflhqw dqg %i lv wkh ghqvlw\ ri d vshflhv zh zlwklq wkh pdwhuldo� $oo rwkhu sdudphwhuv duh ghilqhg lq wkh lqwurgxfwlrq ru lq )lj� ,� -xvw dv µt(e) lv edfnjurxqg�vxewudfwhg dqg qrupdo� l]hg wr rewdlq µn(e)� if i0 (e) lv edfnjurxqg�vxewudfwhg dqg qrupdol]hg wr rewdlq (if/i0)n (e)� ,q sduwlfxodu� d olqh lv ilwwhg wr wkh suh�hgjh gdwd wr rewdlq (if/i0)b(e)� dqg wklv olqh lv vxewudfwhg iurp wkh hqwluh vshfwuxp� 7khq wkh uhvxowdqw vshfwuxp lv glylghg e\ (if/i0)0+ = (if/i0)(e0+) wr qrupdol]h� $v lv dssduhqw lq )lj� ,,�d�� wkh wklfnhu d vdpsoh lv� wkh pruh vhoi�devruehg lwv vshf� wuxp zloo eh� :h fdq ghilqh wklfnqhvv pruh suhflvho\ lq whupv ri dwwhqxdwlrq ohqjwk� ghilqhg dv latt " 1 µt � ru wkh ghswk zlwklq d vdpsoh dw zklfk wkh lqwhqvlw\ ri wkh lqflghqw ;�ud\ zloo uhdfk ��h ri lwv lqlwldo ydoxh� 7kh dwwhqxdwlrq ohqjwk ri frsshu dw lwvk hgjh lv$ 3.87µp� 4xdolwdwlyho\� vhoi�devruehg (if/i0)n (e) ;$1(6 � 10 vshfwud glvsod\ hqkdqfhg suh�hgjh ihdwxuhv dqg gdps� hqhg srvw�hgjh rvfloodwlrqv �)lj� ,,�d��� 7kh uhdvrq iru wklv ehfrphv dssduhqw jlyhq d sorw ri wkh gliihuhqfh eh� wzhhq d qrupdol]hg devruswlrq vshfwuxp dqg ioxruhv� fhqfh vshfwuxp �)lj� ,,�e��� :h vhh wkdw wkhuh lv dq doprvw vlqxvrlgdo udlvlqj ri ydoxhv ri wkh ioxruhvfhqfh vshfwuxp frpsduhg wr wkh devruswlrq vshfwuxp ehorz µn � (fruuhvsrqglqj wr wkh hgjh dqg suh�hgjh� dv zhoo dv wurxjkv ri wkh srvw�hgjh rvfloodwlrqv) dqg dv\ps� wrwlf orzhulqj ri ydoxhv deryh µn � �fruuhvsrqglqj wr shdnv ri wkh srvw�hgjh rvfloodwlrqv�� �d� �e� )ljxuh �� %hkdylru ri 6hoi�$evruswlrq 'lvwruwlrq iru 9dulrxv 7� d� 6lpxodwhg ioxruhvfhqfh vshfwud iru ydulrxv vdp� soh wklfnqhvvhv frpsduhg wr ruljlqdo� xq�vhoi�devruehg gdwd ri frsshu k hgjh� 6hoi�devruehg vshfwud vkrz lqfuhdvhg ydoxhv ehiruh dqg dorqj wkh hgjh� 7kh ghjuhh ri vhoi�devruswlrq lqfuhdvhv zlwk wklfnqhvv� krzhyhu� t = 10latt \lhogv d yluwxdoo\ lghqwlfdo vshfwuxp wr t = %� e� 'liihuhqfh ehwzhhq fruuhfw devruswlrq frhiilflhqw� �µn�� dqg vlpxodwhg� qrupdol]hg ioxruhvfhqfh vshfwuxp� if/i0n � dv d ixqfwlrq ri µn iru ydulrxv wklfnqhvvhv� 1rwlfh wkdw wkh ioxruhvfhqfh vshfwuxp lv doprvw vlqxvrlgdoo\ udlvhg ehorz wkh hgjh dqg dw srvw�hgjh wurxjkv� fruuhvsrqglqj wr µe(e) < µe0+� zlwk wkh juhdwhvw ghfuhdvh rffxuulqj durxqg ��� ri wkh zd\ xs wkh hgjh� $gglwlrqdoo\� lw lv dqg pruh�ru�ohvv olqhduo\ lqfuhdvhg iru srvw�hgjh shdnv� fruuhvsrqglqj wr µe(e) > µe0+� /dvwo\� qrwh wkdw� iru wklqqhu vdpsohv� vhoi�devruswlrq lv prvw vljqlilfdqw iru srvw�hgjh shdnv� dqg iru wklfnhu vdpsohv� vhoi�devruswlrq lv prvw vljqlilfdqw iru wkh suh�hgjh dqg srvw�hgjh wurxjkv� %\ shuiruplqj d iluvw�rughu 7d\oru vhulhv dssur[l� pdwlrq rq wkh h[srqhqwldo whup lq ���� zh ilqg wkdw if i0 # µe�(� zkhq wkh txdqwlw\ >(µt(e)+µf vlq(") vlq(#) ) t vlq(") @<< �� fruuhvsrqglqj sk\vlfdoo\ wr yhu\ wklq �vpdoo /� ru rs� wlfdoo\ gloxwh (vpdoo µt) vdpsohv� 1! e!(µt(e)+µf vlq(!) vlq(") ) t vlq(!) $ 1! (1! (µt(e) + µf vlq(!) vlq(") ) t vlq(!) ) = (µt(e) + µf vlq(!) vlq(") ) t vlq(!) ��� if i0 = ( ! 4# )$(e) µe(e) µt(e) + µf vlq(") vlq(#) ((µt(e)+µf vlq(!) vlq(") ) t vlq(!) ) = ( ! 4# )$(e) t vlq(!) µe(e) & if i0 # µe(e) ��� :khq wklv lv wkh fdvh� wkh qrupdol]hg ioxruhvfhqfh vshf� wuxp lv wkh vdph dv wkh qrupdol]hg devruswlrq vshfwuxp� wkhuh lv qr vhoi�devruswlrq� krzhyhu� zkhq wkh vdpsoh lv qrw yhu\ wklq ru gloxwh� wklv iluvw�rughu dssur[lpdwlrq qr orqjhu krogv� dqg� dv lv dssduhqw iurp ���� wkh uhodwlrq� vkls ehwzhhq µe dqg if/i0 ehfrphv pruh frpsolfdwhg� ,q rughu iru wklv iluvw�rughu dssur[lpdwlrq wr eh ydolg� wkh vdpsoh pxvw eh h[wuhpho\ wklq� $ vdpsoh hyhq ��� latt fdq h[klelw vljqlilfdqw vhoi�devruswlrq (vhh )lj� ,,)� ,q idfw� zh fdq suhglfw wkh ghjuhh ri vhoi�devruswlrq glv� � 11 wruwlrq wkdw d jlyhq h[shulphqwdo vshfwuxp zloo glvsod\� :h fdq ghilqh wkh ghjuhh ri vhoi�devruswlrq glvwruwlrq dv wkh gliihuhqfh ehwzhhq d qrupdol]hg� edfnjurxqg� vxewudfwhg� fruuhfw devruswlrq vshfwuxp� µn� dqg d qru� pdol]hg� edfnjurxqg vxewudfwhg� vhoi�devruehg ioxr� uhvfhqfh vshfwuxp� (if/i0)n � dqg fdoo wklv gliihuhqfh 6$'� ,q rughu wr gr vr� zh zloo wdnh dgydqwdjh ri uh� vxowv xvhg iru wkh vhoi�devruswlrq fruuhfwlrq glvfxvvhg odwhu rq �vhh dsshqgl[ %�� 6$' = µn[1! (µe0+ + &)[1! e!(µnµe0++$)l] (µnµe0+ + &)[1! e!(µe0++$)l] ] ��� ,q rughu iru 6$' wr eh ohvv wkdq ���� wkurxjkrxw wkh zkroh vshfwuxp ri wkh frsshu vdpsoh xvhg lq wklv zrun� fruuhvsrqglqj wr glvwruwlrq ri ohvv wkdq ��zkhuh wkh shufhqwdjh lv uhodwlyh wr � rq wkh qrupdol]hg vshf� wuxp� 7 zrxog qhhg wr eh ����µp� ru ohvv wkdq ���� ri dq dwwhqxdwlrq ohqjwk� 7klv h[dpsoh ghprqvwudwhv wkdw wkh lpsruwdqfh ri gloljhqw vdpsoh suhsdudwlrq wr dyrlg vhoi�devruswlrq fdqqrw eh xqghuvwdwhg� 8qohvv wkh h[� shulphqwhu wdnhv vshfldo fduh wr suhsduh dq h[fhswlrq� doo\ wklq vdpsoh� vljqlilfdqw vhoi�devruswlrq glvwruwlrq zloo rffxu� 7kh htxdwlrq iru 6$' ���� lv qrw lqyhuwleoh lq d forvhg dojheudlf irup iru / " t/ vlq(!)� krz� hyhu� rqh fdq hdvlo\ hydoxdwh wkh h[suhvvlrq qxphulfdoo\ wr ghulyh wkh pd[lpxp vdpsoh wklfnqhvv wkdw zloo nhhs 6$' xqghu d fhuwdlq ydoxh� vd\ ����� ,w lv jhqhudoo\ ehvw wr fdofxodwh 6$' iru kljkhvw wkh ydoxh ri wkh qrupdo� l]hg fruuhfw devruswlrq vshfwuxp� µn� rqh zrxog h[� shfw wr ilqg� zklfk fruuhvsrqgv wr wkh kljkhvw shdn lq wkh µn vshfwuxp� 7klv lv ehfdxvh� dv ghprqvwudwhg lq )lj� ,,�e�� vhoi�devruswlrq iru wklq vdpsohv lv juhdwhvw dw wkh kljkhvw ydoxh ri wkh vshfwuxp� 2i frxuvh� dq h[� dfw pd[lpxp ydoxh iru µn fdqqrw eh nqrzq zlwkrxw dq devruswlrq vshfwuxp wr uhihuhqfh� exw d uhdvrqdeoh jxhvv fdq xvxdoo\ eh pdgh� )ru prvw devruswlrq vshf� wud� µnmax < 1.5� vr fdofxodwlqj 6$' iru µn ��� zloo \lhog d ydoxh kljkhu wkdq rqh vkrxog dfwxdoo\ hqfrxqwhu� krzhyhu� iru vshfwud zlwk pruh h[djjhudwhg srvw�hgjh rvfloodwlrqv� fdofxodwlqj 6$' iru d odujhu ydoxh ri µn zloo eh qhfhvvdu\� )ru dq lqgh[ ri wkh ghjuhh ri vhoi� devruswlrq zklfk lv frpsdudeoh iurp h[shulphqw wr h[� shulphqw� lw pd\ eh xvhixo wr fdofxodwh 6$' iru d w\slfdo ydoxh ri µnmax� vd\ ���� �d� �e� )ljxuh �� %hkdylru ri 6$'� d� 6$' dv d ixqfwlrq ri µb0/µe0 iru 7 ��µe0 (lq rwkhu zrugv� rqh dwwhqxdwlrq ohqjwk iru wkh hohphqw ri lqwhuhvw)� µf �� dqg µn ���� 7klv ghprqvwudwhv wkdw� wkh odujhu wkh edfnjurxqg devruswlrq frhiilflhqw lq sursruwlrq wr wkdw ri wkh hohphqw dqg hgjh ri lqwhuhvw� wkh ohvv vljqlilfdqw vhoi�devruswlrq glvwruwlrq lv� e� 6$' dv d ixqfwlrq ri µe0 iru 7 �� µf �� dqg µn ���� 7klv ghprqvwudwhv wkdw d juhdwhu frqfhqwudwlrq ri wkh hohphqw ri lqwhuhvw uhvxowv lq pruh vljqlilfdqw vhoi�devruswlrq� 6$' surylghv dqrwkhu xvhixo lqvljkw� e\ lqfuhdvlqj µb(e)� zh fdq ghfuhdvh 6$' iru d jlyhq µe(e) dqg t � dv ghprqvwudwhg lq )lj� ,,,� ,q rwkhu zrugv� li zh gloxwh wkh dwrp ri lqwhuhvw lq d ghqvh pdwul[ ri khdylhu dwrpv �dqg pdlqwdlq vdpsoh wklfnqhvv�� zh fdq ghfuhdvh wkh dprxqw ri vhoi devruswlrq� 7klv hiihfw lv dovr dssduhqw lq ���� ,i µb�(� dqg µf duh odujh� wkh yduldwlrq lq µe zloo uhvxow lq uhodwlyho\ ohvv fkdqjh lq wkh ydoxh ri wkh h[� srqhqwldo whup� $gglwlrqdoo\� yduldwlrqv lq µe(e) zloo uhvxow lq juhdwhu yduldwlrqv ri wkh ydoxh ri wkh iudfwlrqdo whup dv d zkroh� ehfdxvh wkh ghqrplqdwru zloo ydu\ ohvv� 3k\vlfdoo\� dq lqfuhdvh lq µb(e) zloo fruuhvsrqg wr dq lqfuhdvh lq µf (e) ehfdxvh zloo eh pruh dqg�ru khdylhu dwrpv lq wkh pdwul[� zklfk zloo dovr devrue ioxruhvfhqfh skrwrqv� 7kxv� gloxwlqj wkh hohphqw ri lqwhuhvw lq d pd� wul[ ri khdylhu dwrpv zloo ghfuhdvh vhoi�devruswlrq hyhq � 12 pruh wkdq )lj� ,,,�d� lpsolhv� &rqyhuvho\� vhoi�devruswlrq glvwruwlrq zloo eh prvw vljqlilfdqw zkhq µb(e) dqg µf (e) duh plqlpl]hg ± wkdw lv� zkhq wkh vdpsoh lv frqfhqwudwhg zlwk wkh hohphqw ri lqwhuhvw� ,i zh dvvxph wkdw rxu vdpsoh lv yhu\ frqfhq� wudwhg zlwk wkh hohphqw ri lqwhuhvw dqg wkxv µb(e) = µf = 0� wkhq� lq rughu iru 6$' ' ���� iru µn ���� 7 pxvw eh ohvv wkdq ����� latt $ (1/12)latt� ,i µn ���� wkhq 7< 0.27latt� %hfdxvh wklv pd[lpxp ydoxh ri 7 uhtxluhg wr dyrlg vhoi�devruswlrq ydulhv vljqlilfdqwo\ iurp fdvh wr fdvh� wkhuh lv qr suhflvh uxoh ri wkxpe uh� jduglqj wkh pd[lpxp wklfnqhvv zlwk zklfk wr suhsduh \rxu vdpsoh� +rzhyhu� lw lv w\slfdoo\ rq wkh rughu ri kxqguhgwkv wr whqwkv ri dq devruswlrq ohqjwk� )ru wklfn vdpsohv� lw lv sduwlfxoduo\ dgylvdeoh wr dyrlg kljk frqfhqwudwlrq zlwk uhvshfw wr wkh hohphqw ri lqwhuhvw ehfdxvh vhoi�devruswlrq fdq ehfrph yhu\ vljqli� lfdqw� 6shflilfdoo\� µf = 0, µt(e) = µe(e) & if = i0( ! 4# )$(e)[1! e!µt(e) t vlq(!) ] ��� ,i wkh vdpsoh lv yhu\ wklfn� wkh h[srqhqwldo whup zloo dssurdfk ]hur dqg wkh ioxruhvfhqfh vshfwuxp zloo h[� klelw plqlpdo vwuxfwxuh� pdnlqj fruuhfwlrq txlwh gliil� fxow� 7kh orzhu µb(e) lv lq uhodwlrq wr µe(e)� wkh kljkhu wkh suhylrxvo\�glvfxvvhg fds rq vhoi�devruswlrq glvwru� wlrq lq wkh wklfn olplw ehfrphv� 7kh juhdwhu wkh wklfnqhvv ri rxu vdpsoh lv� wkh ohvv dffxudwh wkh iluvw�rughu dssur[lpdwlrq ��� ehfrphv� dqg wkxv vhoi�devruswlrq glvwruwlrq lqfuhdvhv� +rzhyhu� df� fruglqj wr wkh ioxruhvfhqfh htxdwlrq� wkhuh lv d pd[l� pxp dprxqw ri vhoi devruswlrq wkdw fdq rffxu gxh wr vdpsoh wklfnqhvv� 7klv lv looxvwudwhg lq )lj� ,,�e�� 7kh uhdvrq iru wklv lv wkdw� iru vxiilflhqwo\ odujh 7� zh pd\ dssur[lpdwh wkh ioxruhvfhqfh htxdwlrq wr vdwlvi\ wkh 7 ( % olplw� \lhoglqj d pxfk vlpsohu dqg dojheudlfdoo\� lqyhuwleoh htxdwlrq >�@� olp t"# $ if = i0( ! 4# )$(e) µe(e) µt(e) + µf vlq(") vlq(#) [1!e!(µt(e)+µf vlq(!) vlq(") ) t vlq(!) ] % = i0( ! 4# )$(e) µe(e) µt(e) + µf vlq(") vlq(#) ��� 7kh shufhqw huuru ri wkh t ( % dssur[lpdwlrq iru d sduwlfxodu h[shulphqw lv vlpso\ wkh pdjqlwxgh ri wkh h[srqhqwldo whup lq ���� 7hpsrudulo\ qhjohfwlqj µf dqg dvvxplqj vwdqgdug jhrphwu\ �! = " ����� lw zloo wdnh d vdpsoh ri dssur[lpdwho\ (! oq(0.01)/ vlq(45) $ 3.3)latt iru wkh t ( % wr eh dffxudwh zlwklq ��� +rz� hyhu� li zh dwwhpsw wr dsso\ d fruuhfwlrq wr wklv vshfwuxp dvvxplqj wkdw wkh t ( % olplw lv vdwlvilhg� rxu fru� uhfwhg vshfwuxp pd\ eh ³ryhufruuhfwhg´ pruh vhyhuho\ wkdq ��� 7kxv� dv d jhqhudo uxoh ri wkxpe iru vwdqgdug jhrphwu\� lw lv dgylvdeoh wr hqvxuh d vdpsoh wklfnqhvv ri ' (! oq(0.001)/ vlq(45) $ 4.9)latt ehiruh dvvxp� lqj wkh t ( % olplw� $gglwlrqdoo\� dv zh kdyh vhhq� vljqlilfdqw vhoi�devruswlrq pd\ rffxu hyhq iru d vdpsoh zlwk d wklfnqhvv rq wkh rughu ri plfurqv� 2qhpxvw kdyh pruh suhflvh nqrzohgjh ri d wklq vdpsoh¶v wklfnqhvv wr surshuo\ fruuhfw iru vhoi�devruswlrq² dqrwkhu uhdvrq wr lqwhqwlrqdoo\ vdwlvi\ wkh t ( % wkurxjk wkh fkrlfh ri d wklfn vdpsoh� 7klv surylghv d xvhixo lqvljkw iru h[shulphqwdo gh� vljq� lw lv hdvlhu wr fruuhfw d ioxruhvfhqfh vshfwuxp wdnhq iurp d vdpsoh ri vljqlilfdqw wklfnqhvv wkdq lw lv iru d vdpsoh ri lqwhuphgldwh wklfnqhvv� 2qh fdq xvh wklv htxdwlrq dv wkh edvlv iru d vwudljkwiruzdug fruuhf� wlrq dojrulwkp� 'dqlho +dvnho¶v )/82� d srsxodu vhoi� devruswlrq fruuhfwlrq dojrulwkp iru ;$1(6� grhv suh� flvho\ wkdw >�@� ,i d vdpsoh lv yhu\ wklfn� rqh fdq xvh )/82 ru wkh fruuhfwlrq dojrulwkp zh zloo suhvhqw wr fruuhfw ioxruhvfhqfh vshfwud zlwkrxw suhflvh nqrzohgjh ri wkh vdpsoh¶v wklfnqhvv� 7kxv� d yhu\ wklfn vdp� soh pd\ eh suhihudeoh wr rqh ri lqwhuphgldwh wklfnqhvv� zklfk zloo uhtxluh wkh h[shulphqwhu wr kdyh suhflvh nqrzohgjh ri lwv wklfnqhvv lq rughu wr fruuhfw wkh vshf� wuxp dffxudwho\� )ru wklfn vdpsohv� lw lv sduwlfxoduo\ dgylvdeoh wr dyrlg kljk frqfhqwudwlrq zlwk uhvshfw wr wkh hohphqw ri lqwhuhvw� ru� vlploduo\� gloxwlrq lq oljkwhu hohphqwv� ,q wklv fdvh� vhoi�devruswlrq fdq ehfrph yhu\ vljqlilfdqw� ,q sduwlfxodu� $ µf = µb(e) = 0 % & if i0 n = ( ! 4# )$(e)[1!e!µt(e) t vlq(!) ] ��� ,i wkh vdpsoh lv yhu\ wklfn� wkh h[srqhqwldo whup zloo ds� surdfk ]hur dqg wkh ioxruhvfhqfh vshfwuxp zloo h[klelw plqlpdo vwuxfwxuh� pdnlqj fruuhfwlrq txlwh gliilfxow� $gglwlrqdoo\� dv zh ghfuhdvh wkh lqflghqw dqjoh !� zh lqfuhdvh wkh dprxqw ri vhoi�devruswlrq� dv vkrzq lq )lj� 9,� ,qwxlwlyho\� zh fdq xqghuvwdqg d ghfuhdvh lq lqflghqw dqjoh dv dq lqfuhdvh lq wkh hiihfwlyh wklfnqhvv ri wkh vdpsoh� ,q wkh fdvh ri qrupdo lqflghqfh� jud]lqj h[lw jhrphwu\ �! ���� " �� zh qrwlfh wkdw if ehfrphv sursruwlrqdo wrµ�(�� dqg wkxv wkhuh lv qr vhoi devruswlrq� � 13 $vvxplqj ! + " ���� olp ""90 if i0 = ( ! 4# )$(e) µe(e) µe(e) + µb(e) + µf ( vlq(90) vlq(0) = %) & 1 ! e!(µe(e)+µb(e)+µf ( vlq(90) vlq(0) =#)) t vlq(0) ' & ( ! 4# )$(e) µe(e) µe(e) + µb(e) + µf ( vlq(90) vlq(0) = %) # µe ��� 7khuhiruh� lw lv dgylvdeoh wr shuirup ioxruhvfhqfh phd� vxuhphqwv lq wklv jhrphwu\ li dw doo srvvleoh� �d� �e� )ljxuh �� %hkdylru ri 6hoi�$evruswlrq 'lvwruwlrq )ru 9dulrxv !� d� 6lpxodwhg ioxruhvfhqfh vshfwud iru ydulrxv lqflghqw dqjohv frpsduhg wr ruljlqdo� xq�vhoi�devruehg gdwd ri frsshuk hgjh� zlwk wkh frqglwlrq wkdw !+" ���� 7kh ghjuhh ri vhoi devruswlrq lqfuhdvhv dv ! ghfuhdvhv� 1rwlfh wkdw iru ! = 89.9� wkhuh lv yluwxdoo\ qr vhoi�devruswlrq� e� 'liihuhqfh ehwzhhq fruuhfw dqg vlpxodwhg vshfwud� 6hoi�$evruswlrq &ruuhfwlrq 'hulydwlrq 7kh edvlv ri wklv fruuhfwlrq phwkrg lv wr xvh qxphu� lfdo fdofxodwlrq dqg wdexodwhg gdwd wr zrun durxqg wkh idfw wkdw wkh ghvluhg txdqwlw\� µe�(�� pd\ qrw eh dojh� eudlfdoo\ lvrodwhg lq ���� :h ehjlq zlwk d voljkwo\ uhiru� pxodwhg ioxruhvfhqfh htxdwlrq zlwkrxw frqvwdqw pxowl� solfdwlyh idfwruv� zklfk ehfrph xqlpsruwdqw xsrq qru� pdol]dwlrq� :h dovr pdnh wkh dssur[lpdwlrq wkdw $(e) lv frqvwdqw ryhu wkh ;$1(6 uhjlrq� zklfk lv dffxudwh ehfdxvh $(e) grhv qrw ydu\ vljqlilfdqwo\ dv d ixqfwlrq ri hqhuj\ iru d jlyhq fkhplfdo vshflhv� if i0 n # µe(e) µt(e) + µfg [1! e!(µt(e)+µfg)l] ��� +huh� l " t/ vlq(!)� g " vlq(!)/ vlq(")� 7dnlqj wdexodwhg gdwd iurp >�@� zh pd\ fdofxodwh� µf = ( i µm0i(e0+)%i ���� µb0 = ( i µm0i(epreedge)%i ���� µt0+ = ( i µm0i(e0+)%i ���� zkhuh zh lv dq hohphqw zlwklq wkh vdpsoh� µn µe�µe0+ lv wkh vshfwuxp wkdw wklv fruuhfwlrq zloo xowlpdwho\ ghulyh� 5hfdoo� lw lv wkh qrupdol]hg� edfnjurxqg�vxewudfwhg� dqg fruuhfwhg devruswlrq vshf� wuxp� 8qohvv wkhuh lv dqrwkhu dwrp lq wkh vdpsoh zlwk dq hgjh dw dq hqhuj\ lq wkh ;$1(6 hqhuj\ uhjlrq ri wkh hohphqw xqghu lqyhvwljdwlrq (d vfhqdulr zklfk vkrxog eh dyrlghg li dw doo srvvleoh)� µb�(� zloo eh qhduo\ frq� vwdqw ryhu wkh ;$1(6 uhjlrq� dqg wkxv zh fdq dssur[l� pdwh µb(e) $ µb0 = µb(ebeforeedge)� zh pd\ wkhq fdofxodwh µe0+ = µt0+ ! µb0� :h pd\ dovr ghilqh & " µb0++µfg� $gglwlrqdoo\� zhpd\ gurs wkh h[solflw �(� ghshqghqfh ri whupv zlwklq wkh h[suhvvlrq� *lyhq � 14 wkhvh dphqgphqwv� zh kdyh� if i0 n # µnµe0+ µnµe0+ + & [1! e!(µnµe0++$)l] ���� 7r qrupdol]h� zh hydoxdwh wkh uljkw kdqg vlgh ri wkh htxdwlrq dw µn = 1� ru� htxlydohqwo\� µe = µe0+� :h wkhq glylgh wklv ydoxh e\ � dqg fdoo lw ³1´ iru qrupdo� l]dwlrq idfwru� :h wkhq pxowlso\ wkh uljkw kdqg vlgh e\ 1� $w wklv srlqw� zh pd\ vhw wkh uljkw dqg ohiw kdqg vlghv ri wkh htxdwlrq wr eh htxdo ehfdxvh wkh\ duh erwk qru� pdol]hg� n " 1 µe0 µe0++$ [1! e!(µe0++$)l] ���� if i0 n = nµnµe0+ µn + & [1! e!(µnµe0++$)l] ���� 1h[w� zh fdofxodwh if i0 n (e) iru d odujh qxpehu ri ydoxhv ri wkh ilqdo� qrupdol]hg devruswlrq vshfwuxp µn ehwzhhq ��� dqg �� 7klv udqjh vkrxog vxiilfh iru prvw vshfwud rqh frxog uhdvrqdeo\ hqfrxqwhu� exw� ri frxuvh� wkh udqjh pd\ eh hdvlo\ dgmxvwhg iru dw\slfdo vshfwud� :h qrz kdyh d pds 6$&� if i0 n (e) )(µn(e)� zkhuh 6$& lv dq dfurq\p iru ³vhoi�devruswlrq fruuhfwlrq´� zklfk zh fdq xvh wr fruuhfw rxu vhoi�devruehg vshfwuxp� 2qhpd\ uhsodfh hdfk gdwd srlqw lq d vhoi�devruehg vshf� wuxp� )�(�� zlwk wkh ydoxh ri µn(e) zklfk fruuhvsrqgv wr wkh qhduhvw ydoxh iru )�(� lq wkh vhw ri ydoxhv lq wkh pds 6$& wkdw kdv ehhq fdofxodwhg� 7kh uhvxow lv wkh fru� uhfwhg vshfwuxp� µn�(�� 3xw vlpso\� wklv phwkrg zrunv durxqg wkh odfn ri forvhg�irup lqyhuwlelolw\ ri wkh iox� ruhvfhqfh htxdwlrq wr ilqg dq �doprvw� h[dfw lqyhuvh qxphulfdoo\ xvlqj wdexodwhg ydoxhv dqg dssur[lpdwlrqv zklfk duh w\slfdoo\ yhu\ vwurqj ryhu wkh ;$1(6 uhjlrq� 7r whvw wkh hiihfwlyhqhvv ri wklv fruuhfwlrq phwkrg� zh dssolhg lw wr gdwd wkdw zh vlpxodwhg xvlqj wkh ioxr� uhvfhqfh htxdwlrq dqg s\wkrq� ,q sduwlfxodu� zh wrrn if/i0(e) gdwd froohfwhg e\ 5\dq 'dylv dqg $sxuyd 0hkwd rq wkh k! hplvvlrq olqh ri d frsshu vdpsoh� dv� vxphg wr qrw eh vhoi�devruehg� :h ilwwhg wklv gdwd vhw wr wdexodwhg olqhdu devruswlrq frhiilflhqw gdwd iru frs� shu iurp 1,67 >�@� :h wkhq xvhg wkh uhvxowdqw vshf� wuxp dv µt(e) lq wkh vlpxodwlrqv� )ru wkh vlpxodwlrqv� zh rewdlqhg µb(e) wkurxjk d suh�hgjh olqhdu uhjuhvvlrq ilw wkh wkh suh�hgjh gdwd ri wkh µt(e) fxuyh� :h fdo� fxodwhg µe(e) xvlqj µe(e) = µt(e) ! µb(e)� :h vxewudfwhg wkh edfnjurxqg ri wkh vlpxodwhg e\ shuirup� lqj d olqhdu uhjuhvvlrq ilw rq suh�hgjh gdwd dqg vxewudfw� lqj wklv olqh iurp p\ hqwluh gdwd vhw� :h wkhq iodwwhqhg p\ gdwd e\ shuiruplqj d olqhdu uhjuhvvlrq ilw rq srvw� hgjh gdwd dqg vxewudfwlqj wklv olqh iurp gdwd diwhu wkh hgjh� $ sro\qrpldo ilw wr srvw�hgjh gdwd zrxog eh pruh vxlwdeoh iru vshfwud zlwk ohvv�olqhdu srvw�hgjh vwuxfwxuhv� /dvwo\� zh qrupdol]hg e\ glylglqj wkh hqwluh vshfwuxp e\ (if/i0)0+� wkh ydoxh dw zklfk wkh srvw�hgjh iodw� whqhg �exw qrw \hw qrupdol]hg� gdwd frqyhujhg� $oo gdwd surfhvvlqj� vlpxodwlrqv� dqg sorwwlqj lq wklv zrun zh uh grqh xvlqj 3\wkrq� $ gdwd�surfhvvlqj surjudp vxfk dv 6,;3$&. ru $7+(1$ frxog eh xvhg iru wkhvh lqlwldo vwhsv� $v h[klelwhg lq )lj� 9 dqg 9,� wkh fruuhfwlrq phwkrg zrunv xqlirupo\ zhoo dfurvv gliihuhqw ydoxhv ri t dqg !� $gglwlrqdoo\� lw lv jhqhudoo\ urexvw djdlqvw huuruv lq xvhu�ghwhuplqhg ydoxhv ri 7 dqg !� +rz� hyhu� dv ! lqfuhdvhv� ru 7 ghfuhdvhv� wkh fruuhfwlrq eh� frphv ohvv urexvw djdlqvw plvphdvxuhphqwv� dqg wkh xvhu pxvw nqrz wkhlu ydoxhv zlwk juhdwhu dffxudf\ wr fruuhfw zhoo� )lj� 9 dqg 9, dovr vkrz wkdw wkh gliihu� hqfh ehwzhhq ruljlqdo dqg fruuhfwhg vshfwud lv juhdwhvw dw wkh ulvlqj hgjh ihdwxuh ($ 8985ev ) dqg wkh iluvw shdn ($ 9000ev )� lqglfdwlqj wkdw wkh fruuhfwlrq lv ohdvw hi� ihfwlyh lq wkhvh uhjlrqv ri wkh vshfwuxp� :h qrwh wkdw wklv fruuhfwlrq lv qrw dssolfdeoh wr wkh ³jud]lqj�lqflghqfh´ uhjlph� zkhuhlq ! ' !c� zkhuh !c lv wkh vdpsohv fulwlfdo dqjoh iru wrwdo h[whuqdo uhiohfwlrq� ,q wklv uhjlph� ��� qr orqjhu dssolhv� � 15 �d� �e� �f� �g� �h� �i� )ljxuh �� %hkdylru ri fruuhfwlrq iru ydulrxv 7� d�f� dqg h ghprqvwudwh wkh fruuhfwlrq rq vlpxodwhg vshfwud ri ydulrxv wklfnqhvvhv� 7 1 10 latt� 7 �latt� dqg 7 ��latt uhvshfwlyho\� 7kh ruljlqdo� xq�vhoi�devruehg vshfwud duh frpsduhg wr wkh vlpxodwhg vshfwud iru ydulrxv 7 dqg wkh fruuhfwhg yhuvlrqv ri wkhvh vlpxodwhg vshfwud� e� g� dqg i �fruuhvsrqglqj wr d� f� dqg h� vkrz wkh gliihuhqfh ehwzhhq wkh ruljlqdo� xq�vhoi�devruehg vshfwud dqg wkh fruuhfwhg vshfwud �lq eoxh�� dv zhoo dv wkh gliihuhqfhv ehwzhhq wkh ruljlqdo xq�vhoi�devruehg vshfwud dqg wkh fruuhfwhg vshfwud� zkhuh wkh fruuhfwlrqv duh shuiruphg zlwk lqfruuhfw vdpsoh wklfnqhvv lqsxw� zlwk ���� lq juhhq dqg ���� lq uhg� 7klv ghprqvwudwhv wkh hiihfwlyhqhvv ri wkh fruuhfwlrq dqg lwv urexvwqhvv djdlqvw plvphdvxuhg vdpsoh wklfnqhvvhv �7�� � 16 �d� �e� �f� �g� �h� �i� )ljxuh �� %hkdylru ri fruuhfwlrq iru ydulrxv !� d�f� dqg h ghprqvwudwh wkh fruuhfwlrq rq vlpxodwhg vshfwud ri ydulrxv lqflghqw dqjohv� ! ��� ! ���� dqg ! ��� uhvshfwlyho\� 7kh ruljlqdo� xq�vhoi�devruehg vshfwud duh frpsduhg wr wkh vlpxodwhg vshfwud iru ydulrxv ! dqg wkh fruuhfwhg yhuvlrqv ri wkhvh vlpxodwhg vshfwud� e� g� dqg i �fruuhvsrqglqj wr d�f� dqg h� vkrz wkh gliihuhqfh ehwzhhq wkh ruljlqdo� xq�vhoi�devruehg vshfwud dqg wkh fruuhfwhg vshfwud lq eoxh� dv zhoo dv wkh gliihuhqfhv ehwzhhq wkh ruljlqdo� xq�vhoi�devruehg vshfwud dqg wkh fruuhfwhg vshfwud� zkhuh wkh fruuhfwlrqv duh shuiruphg zlwk lqfruuhfw !� zlwk ���� lq juhhq dqg ���� lq uhg� 7klv ghprqvwudwhv wkh dffxudf\ ri wkh fruuhfwlrq dqg lwv urexvwqhvv djdlqvw plvphdvxuhg !� � 17 ',6&866,21 &21&/86,21 :h kdyh lqyhvwljdwhg wkh ehkdylru ri vhoi�devruswlrq glvwruwlrq� $gglwlrqdoo\� zh kdyh jdlqhg d ihz lqvljkwv zklfk pd\ eh zruwk wkh frqvlghudwlrq lq ghvljqlqj d ioxruhvfhqfh�prgh ;$6 h[shulphqw� ,ghdoo\� d vdpsoh zloo eh vxiilflhqwo\ wklq vxfk wkdw vhoi�devruswlrq glvwru� wlrq pd\ eh dyrlghg dowrjhwkhu� :h kdyh ghprqvwudwhg wkdw wklv wklqqhvv lv w\slfdoo\ kxqguhgwkv wr whqwkv ri dq dwwhqxdwlrq ohqjwk� ,i rewdlqlqj wklv ghjuhh ri wklqqhvv lv qrw srvvleoh� lw lv suhihudeoh wr fkrrvh d vdpsoh ri > 4.9latt wkdq lw lv wr fkrrvh rqh ri lqwhuphgldwh wklfn� qhvv� 7klv lv ehfdxvh� lq rughu wr fruuhfw d ioxruhvfhqfh vshfwuxpphdvxuhg iurp d vdpsoh ri lqwhuphgldwh wklfn� qhvv� d pruh suhflvh nqrzohgjh ri 7 lv uhtxluhg wkdq lv uhtxluhg wr fruuhfw d vshfwuxp iurp d vdpsoh ri vlj� qlilfdqw wklfnqhvv� ,i lw lv nqrzq wkdw wkh vdpsoh 7 lv > 4.9latt� qr ixuwkhu nqrzohgjh ri 7 lv uhtxluhg wr fru� uhfw wkh vshfwuxp dffxudwho\� 7kh fruuhfwlrq zh kdyh suhvhqwhg lv edvhg rq wkh zhoo�hvwdeolvkhg ioxruhvfhqfh htxdwlrq ���� dqg vlplodu fruuhfwlrq phwkrgv kdyh ehhq suhvhqwhg ehiruh� prvw qrwdelolw\ e\ ,lgd dqg 1rpd >�@� krzhyhu� wklv fruuhfwlrqphwkrg lv xqltxh lq wkdw lw wdnhv lq dqg sxwv rxw qrupdol]hg� edfnjurxqg�vxewudfwhg gdwd� $gglwlrqdoo\� lw qrupdol]hv e\ µe0+� zkhuhdv ,lgd dqg 1rpd¶v fruuhfwlrq qrupdol]hv e\ µe0max >�@� 7kh fruuhfwlrq phwkrg zh suhvhqw lv vwurqj� exw wkhuh lv urrp wr lqfrusrudwh pruh sk\vlfdo ghwdlo iru lp� suryhphqw� )ru lqvwdqfh� lw grhv qrw h[solflwo\ dffrxqw ioxruhvfhqfh ghwhfwru jhrphwu\ dqg ydu\lqj dwwhqxdwlrq iurp sl[ho wr sl[ho ri wkh lqflghqw ;�ud\ lqwhqvlw\ rq wkh sdwk ehwzhhq wkh lqwhqvlw\ prqlwru dqg wkh vdpsoh dqg dv dwwhqxdwlrq ri wkh ioxruhvfhqfh ;�ud\ lqwhqvlw\ rq wkh sdwk iurp wkh vdpsoh wr wkh ioxruhvfhqfh ghwhf� wru� krzhyhu� wklv pd\ eh uhphglhg� wkurxjk wkh xvh ri wkh 6h$))ox; vriwzduh sdfndjh� zklfk fdq dgmxvw phd� vxuhg vshfwuxp wr dffrxqw iru wkhvh dwwhqxdwlrq idfwruv >�@� 7klv dgmxvwhg vshfwuxp fdq wkhq eh fruuhfwhg xv� lqj wkh phwkrg zh kdyh suhvhqwhg deryh� krzhyhu� dv %rrwk dqg %ulgjhv qrwhg lq >�@� wkh hiihfw ri lqwhjudwlrq ri ��� ryhu wkh vrolg dqjoh ri wkh ghwhfwru lv uhodwlyho\ lqvljqlilfdqw lq ! + " = 90 jhrphwulhv� $owkrxjk wklv fruuhfwlrq phwkrg zloo qrw eh shu� ihfw lq doo fdvhv� zh eholhyh wkdw wkhuh lv olwwoh urrp iru lpsuryhphqw eh\rqg d fruuhfwlrq zklfk wdnhv lqwr df� frxqw ioxruhvfhqfh ghwhfwru jhrphwu\ dqg ehdp dwwhq� xdwlrq iurp vrxufhv rwkhu wkdq wkh vdpsoh dv lq 6h$)� )/x;� dqg wkhq lqyhuwv ��� dv lq wkh fruuhfwlrq phwkrg suhvhqwhg khuh >�@� 0rylqj iruzdug� wkhuh uhpdlqv d qhhg iru d ;$1(6 fruuhfwlrq lq wkh jud]lqj lqflghqfh uhjlph� $87+25 ,1)250$7,21 &ruuhvsrqglqj $xwkru $lgdq 5hgg\� &roxpeld 8qlyhuvlw\ lq wkh &lw\ ri 1hz <run� 1<� 86$� dlgdq�uhgg\#froxpeld�hgx )xqglqj 6rxufhv 7klv zrun zdv vxssruwhg lq sduw e\ wkh 8�6� 'hsduwphqw ri (qhuj\� 2iilfh ri 6flhqfh� 2iilfh ri :runirufh 'h� yhorsphqw iru 7hdfkhuv dqg 6flhqwlvwv �:'76� xqghu wkh 6flhqfh 8qghujudgxdwh /derudwru\ ,qwhuqvklsv 3ur� judp �68/,�� $&.12:/('*(0(176 :h wkdqn $sxuyd 0hkwd dqg 5\dq 'dylv iru wkhlu vxssruw dqg jxlgdqfh� 5()(5(1&(6 >�@ $� ,lgd� 7� 1rpd� &ruuhfwlrq ri wkh 6hoi�$evruswlrq (iihfw lq )oxruhvfhqfh ;�5d\ $evruswlrq )lqh 6wuxfwxuh� -sq� $sso� 3k\v� ��� ��������� ������� >�@ &�+� %rrwk� )� %ulgjhv� ,psuryhg 6hoi�$evruswlrq &ruuhfwlrq iru )oxruhvfhqfh 0hdvxuhphqwv ri ([� whqghg ;�5d\ $evruswlrq )lqh 6wuxfwxuh� 3k\vlfd 6fulswd� 7��� ���±��� ������� >�@ '� +dvnho� )/82� &ruuhfwlqj ;$1(6 iru 6hoi� $evruswlrq lq )oxruhvfhqfh 0hdvxuhphqwv� ������� >�@ -� *rxorq� &� *rxorq�*lqhw� 5� &ruwhv� -�0� 'xerlv� 2q h[shulphqwdo dwwhqxdwlrq idfwruv ri wkh dpsolwxgh ri wkh (;$)6 rvfloodwlrqv lq devruswlrq� uhiohfwlylw\ dqg oxplqhvfhqfh phdvxuhphqwv� -rxu� qdo gh 3k\vltxh� �� ���� ������� ������� >�@ -� -dnohylf� -�$� .lue\� 0�3� .ohlq dqg $�6� 5rehuw� vrq� *�6� %urzq dqg 3� (lvhqehujhu�)oxruhvfhqfh 'hwhfwlrq ri (;$)6� 6hqvlwlylw\ (qkdqfhphqw iru 'loxwh 6shflhv dqg 7klq )lopv� 6rolg 6wdwh &rppx� qlfdwlrqv ��� ������� ������� >�@ 1dwlrqdo ,qvwlwxwh iru 6wdqgdug dqg 7hfk� qrorj\� 3k\vlfdo 0hdv� /derudwru\� 'dwd iurp ´;�5d\ )rup )dfwru� $wwhqxd� wlrq� dqg 6fdwwhulqj 7deohv� $ydlodeoh dw kwwsv���sk\vlfv�qlvw�jry�3k\v5hi'dwd�))dvw�kwpo�irup�kwpo� >�@ 5� 0� 7uhyrudk� &� 7� &kdqwohu� 0� -� 6fkdonhq� 6roylqj 6hoi�$evruswlrq lq )oxruhvfhqfh� ,8&u�� ������� ������� �� 18 >�@ :�+� 0f0dvwhu� 1� .huu 'ho *udqgh� -�+� 0do� ohww� -�+� +xeeho� &rpslodwlrq ri ;�5d\ &urvv 6hf� wlrqv� $wrplf 'dwd dqg 1xfohdu 7deohv� �� ������� ������� �� 19 $33(1',; $� '(5,9$7,21 2) )/825(6&(1&( (48$7,21 :h ehjlq zlwk %hhu¶v /dz� zklfk ghvfulehv wkh h[sr� qhqwldo ghfd\ ri wkh lqwhqvlw\ ri wkh lqflghqw ;�ud\ dv lw wudyhov wkurxjk d vdpsoh dv d uhvxow ri devruswlrq� 6hh )lj� , iru jhrphwu\� di(x) dx = !i(x)µt(e) ���� ) x sin(!) 0 1 i(x) di(x) = ) x sin(!) 0 !µt(e)dx ���� i(x) = i0e !µt(e)x sin(!) ���� +huh� ,�[� lv wkh lqwhqvlw\ ri wkh ;�ud\ dw d jlyhq ghswk� µ lv wkh olqhdu devruswlrq frhiilflhqw� htxlydohqw wr# i µi = # i µmi%i� zkhuh zh lv dq hohphqw zlwklq wkh pdwhuldo� µmi lv lwv pdvv devruswlrq frhiilflhqw dqg %i lv wkh ghqvlw\ ri d wkh hohphqw zlwklq d pdwhuldo� dqg i0 lv wkh lqwhqvlw\ ri wkh ;�ud\ sulru wr hqwhulqj wkh pdwh� uldo� 7kh devruswlrq ri d skrwrq grhv qrw qhfhvvdulo\ uhvxow lq wkh uhohdvh ri d ioxruhvfhqfh skrwrq� )ru lq� vwdqfh� lw pd\ uhvxow lq wkh uhohdvh ri dq $xjhu hohfwurq lqvwhdg� 7kh sruwlrq ri devruswlrq hyhqwv uhvxowlqj lq ioxruhvfhqfh pd\ eh ghilqhg dv $(e)� d ixqfwlrq ri hq� huj\� ,qwxlwlyho\� wkh wrwdo ioxruhvfhqfh� iftot(x) pxvw eh htxdo wr iabs(x)� wkh wrwdo devruehg lqwhqvlw\ pxowl� solhg e\ $(e)� iftot(x) = $(e)iabs = $(e)(i0 ! i(x)) = $(e)i0(1! e! µt(e)x sin(!) ) & diftot(x) dx = $(e)i0 µt(e) sin(!) e! µt(e)x sin(!) ���� :khq d skrwrq lv uhohdvhg iurp dq dwrp� lw pxvw wudyho wkurxjk wkh exon ri wkh pdwhuldo ehiruh duulylqj dw iox� ruhvfhqfh ghwhfwru� 2q wklv mrxuqh\� ioxruhvfhqfh ;� ud\v dwwhqxdwh dffruglqj wr %hhu¶v /dz� $gglwlrqdoo\� d jlyhq hgjh ri d jlyhq dwrp zloo ioxruhvfh skrwrqv ri d fkdudfwhulvwlf hqhuj\�ef ex �eb� zkhuhex lv wkh hq� huj\ ri wkh lqfrplqj ;�ud\ skrwrq dqgeb lv wkh elqglqj hqhuj\ ri wkh devruelqj hohfwurq� )oxruhvfhqfh skrwrqv duh hplwwhg lvrwurslfdoo\� $ ioxruhvfhqfh ghwhfwru grhv qrw hqwluho\ vxuurxqg d pdwhuldo� vr lw grhv qrw ghwhfw doo ri wkh ioxruhvfhqfh hplwwhg iurp wkh vdpsoh� ,qvwhdg� lw ghwhfwv d sruwlrq ri ! 4% ri doo wkh ioxruhvfhqfh skrwrqv hplwwhg� zkhuh ! lv wkh vrolg dqjoh lq vwhudgldqv durxqg wkh vdpsoh zklfk wkh ioxruhvfhqfh ghwhfwru fryhuv� 7dn� lqj wklv lqwr dffrxqw� zh ilqg� dif (x) dx = ! 4# $(e)i0 µt(e) sin(!) e! µt(e)x vlq(!) e! µfx vlq(") ���� :khuh if lv wkh phdvxuhg ioxruhvfhqfh lqwhqvlw\ dqg µf µt(ef ) lv wkh wrwdo olqhdu devruswlrq frhiilflhqw dw wkh hqhuj\ ri wkh ioxruhvfhqfh skrwrqv� $ jlyhq hgjh ri d jlyhq dwrp zloo ioxruhvfh skrwrqv ri d fkdudfwhulvwlf hqhuj\� ef ex � eb� zkhuh ex lv wkh hqhuj\ ri wkh lqfrplqj ;�ud\ skrwrq dqg eb lv wkh elqglqj hqhuj\ ri wkh devruelqj hohfwurq� :h pxvw qrz lqwhjudwh ryhu wkh hqwluh ghswk ri wkh pdwhuldo� )lqdoo\� zh lqwhjudwh ryhu wkh hqwluh ghswk ri wkh pdwhuldo� ) t 0 dif (x) = ) t 0 ! 4# $(e)i0 µt(e) vlq(!) e!(µt(e) vlq(!) + µf vlq(") )xdx ���� if (t )! if (0) = if = ! 4# $(e)i0 µt(e) vlq(!) 1 !(µt(e) vlq(") + µf vlq(#) ) [e!(µt(e) vlq(!) + µf vlq(") )t!1] & if = ! 4# $(e)i0 µt(e) µt(e) + µf vlq(") vlq(#) [1!e!(µt(e)+µf vlq(!) vlq(") ) t vlq(!) ] ���� $33(1',; %� '(5,9$7,21 2) 6$' 6$' " µn ! if i0 n ���� :h wkhq frqvlghu ���� dqg ����� \lhoglqj� if i0 n = µnµe0+ µnµe0++$ [1! e!(µnµe0++$)l] µe0+ µe0++$ [1! e!(µe0++$)l] = (µe0+ + &)[1! e!(µnµe0++$)l] (µnµe0+ + &)[1! e!(µe0++$)l] & 6$' = µn[1! (µe0+ + &)[1! e!(µnµe0++$)l] (µnµe0+ + &)[1! e!(µe0++$)l] ] ���� �� 20 protein-polymer complex coacervates as synthetic membrane-less organelles hansen tjo1*, nicholas zervoudis2, allie c. obermeyer2 1department of chemical engineering, university of massachusetts amherst 2department of chemical engineering, columbia university in the city of new york complex coacervation, protein engineering, soft matter abstract: in solution, oppositely-charged macromolecules undergo charge-mediated liquid-liquid phase separation into a complex coacervate phase – a dense, macromolecule-rich liquid. in nature, the basis for intracellular compartmentalization in the formation of membrane-less organelles has been shown to follow similar complexation principles, where charged proteins represent the ionic species. we seek to capture the spatiotemporal tunability properties of such organelles for enzymatic reactions in vitro. however, the in vitro formation and deformation of protein-based coacervate microenvironments as a nanoreactor is a limiting factor. here, we prescribe high-precision turbidimetry coupled with optical microscopy, to characterize the phase behavior of binary protein-polymer complexes between the weak anionic enzyme glucose oxidase (gox) and four different synthetic polycations as functions of composition and ionic strength. establishment of conditions ideal for coacervate formation in each protein-polymer system informed ph titration experiments on characterizing self-assembly regulation. the results from this study will help inform the design of novel coacervate microenvironments for industrial enzyme cascades and elucidate the role of associative phase separation in cellular evolution. introduction selectivity and specificity are an enzyme’s key characteristics [1, 3, 4]. as biological catalysts, enzymes increase rates of reaction such that greater amounts of biological product may be obtained under less time. they also have been used to improve chemical processes in industries from food, agriculture, and petroleum in addition to reducing energy costs and operation time [4]. it is difficult to synthetically match the efficacy and specificity of enzymes as biochemical systems have had millions of years to evolve [4, 16]. while advents in directed evolution and protein engineering attempt to circumvent the time required for natural evolution, amongst other approaches to improve enzyme activity, there is also particular interest to achieve similar goals by optimizing an enzyme’s surroundings [1, 4, 16]. this strategy is inspired by metabolic reactions, such as those that make up cellular respiration, where the spatiotemporal efficacy of enzymes is enhanced by its surrounding biological environment [1, 4]. with recent reports highlighting the complex functions of biological condensates in vivo (signaling, reaction networks etc.), we took a biomimetic approach in constructing a stable and responsive enzymatic compartment [1-5]. using de novo liquid-liquid phase separated synthetic organelles, these microenvironments may be the bridge to advance how enzymatic power is harnessed industrially. complex coacervation, an example of associative liquid-liquid phase separation, describes how oppositely charged polyelectrolytes phase separate into a coacervate phase – a dense, polyelectrolyte-rich phase with potential applications in biomolecular encapsulation – and a dilute phase, the supernatant [1]. because the coacervate phase compartmentalizes both enzymes and substrates within the same microenvironment, enzymes can perform their catalytic functions with greater spatiotemporal efficacy [1-4]. these de novo systems are especially advantageous due to their tuneability by a variety of parameters: ph, charge stoichiometry, ionic strength, mixing order, and others [1, 2]. the use of proteins or other charged biomacromolecules as coacervating macro-ions allows further structural modulation through ionic tagging and supercharging [6, 7, 9]. complex coacervate systems typically involve binary mixtures of oppositely charged components. to simplify coacervate formation, a net charged enzyme is chosen such that it constitutes one of the electrostatic components for complexation [10, 11]. in particular, the weakly anionic enzyme glucose oxidase (gox) was used in conjunction with four structurally distinct polycations: poly(4-vinyl nmethyl pyridinium iodide) (qp4vp), poly(allylamine hydrochloride) (pah), poly(ethyleneimine) (pei), and poly(1-vinyl imidazole methyl iodide) (pvi). in sum, we sought to elucidate the effects of mixing order, salt concentration, and finally ph on the phase behavior of four different gox-polycation systems: gox-qp4vp, gox-pah, gox-pei, gox-pvi. 21 methods the phase behavior of four sets of binary gox-polycation systems were studied. the four sets of polymer-protein mixtures consisted of poly(4-vinyl n-methyl pyridinium iodide) (qp4vp), poly(allylamine hydrochloride) (pah), polyethylenimine (pei), and poly(1-vinylimidazole methyl iodide) (pvi) as the polycations, with glucose oxidase (gox) as the anionic charged protein. additionally, we investigated the effects of salt on coacervate formation via the addition of sodium chloride (nacl). finally, we utilized ph titrations to explore how complex coacervation can be regulated by solution ph. sample preparation: glucose oxidase from aspergillus niger was purchased from sigma aldrich (g2133). a stock solution of 2 mg/ml gox was prepared in 10 mm tris at ph 7.4. application of beer’s law was used to determine true concentrations of gox via its absorbance at 280 nm in a 4 ml quartz cuvette. polymer solutions were diluted from liquid stocks of 5 mg/ml; relevant stoichiometric calculations using the molar equivalency equation were used to determine the requisite volumes of 10 mm tris needed to dilute appropriated samples to a mass concentration of 2 mg/ml. each polycation solution was subsequently adjusted to a ph of 7.4. mixing ratios: four mixing ratios of gox/polymer were investigated: 88% gox/12% polymer, 84% gox/16% polymer, 80% gox/20% polymer, 76% gox/24% polymer. such values were determined from preliminary data indicating an optimum mixing ratio range for gox at roughly 80%. data results for variations in mixing ratios were obtained through turbidimetry analysis and optical microscopy. to exclude external ionic strength contributions in studying mixing ratio effects, salt species were absent in all mixtures. salt effects: the effects of added salt on system phase behavior were examined via turbidimetry analyses and optical microscopy. sodium chloride (nacl) concentration was varied from 25 mm to 50 mm on all polymer/protein systems. this salt range was predicted to be conducive for liquid-liquid phase separation in gox-polymer systems based on preliminary work. for the gox-pei system, no form of phase separation was observed at all salt concentrations. thus, further planned investigations on the gox-pei system with salts were abandoned. ph titration: the effects of ph on phase behavior reversibility was examined via turbidimetry analysis, based on absorbance readings from an uv-vis spectrophotometer, and a ph probe at constant ambient temperature (25 ℃). 1 m hydrochloric acid (hcl) was used as the titrant; a 1 cm stirbar at roughly 500 rpm was used to ensure consistent solution mixing throughout the procedure. each gox-polycation system was set at its experimentally-determined optimum mixing ratio and salt concentration: 88% mixing ratio, and 50 mm nacl for all systems with the exception of goxpah, which was set at 175 mm nacl. turbidimetry analyses: turbidimetry analyses were done to investigate the effects of mixing ratios and salt concentration. each sample was prepared in triplicate in tissue culturetreated polystyrene 96-well half-area plates (corning), followed by incubation at room temperature for 3 h. using a plate reader (tecan infinite m200 pro), the absorbance of the mixture was taken at a wavelength (λ = 600 nm) to monitor scattering of the phase separated mixture. each sample had an invariant volume of 50 µl allowing for appropriate absorbance measurements and physical mixing by a tecan infinite m200 pro plate reader (10 s of orbital shaking). finally, the absorbance was converted into turbidity using the following set of relationships: 𝜏 = 100 −%𝑇 (1) %𝑇 = 10(./0) (2) where: • 𝜏 is the turbidity of the solution and an indicator of the extent of phase separation present within the sample. • 𝑇 is the transmittance of the solution as a function of mixture absorbance. • 𝐴 is the measured absorbance of the mixture. optical microscopy: all four protein-polymer samples were prepared in triplicate, followed by individual well examination with optical microscopy using an evos fl auto 2 inverted fluorescence microscope (invitrogen). each sample, controlled at a volume of 50 μl, were formulated in an optically clear 384-well plate (nunc) and then underwent 3 h incubation period at room temperature to maximize the degree of liquid-liquid phase separation taking place in the wells (preliminary data indicated that samples tended to favor figure 1. chemical structures of the four cationic polymers used in this work. structures were drawn using the chemdraw prime software. 22 precipitation following initial mixing before eventual transition into a coacervate phase). all optical microscopy images were taken under 20x objectives with transmitted light. results and discussion initial experiments investigated the effects of mixing ratio of protein to polymer using gox and four different polycations (qp4vp, pah, pei, pvi). preliminary data suggested each binary systems’ tendency to phase separate at all four preselected mixing ratios; however, certain mixing ratios resulted in greater extents of phase separation as indicated by their relative turbidity magnitudes. nevertheless, this initial assumption did not hold as no phase separation was observed with pei at all mixing ratios. however, we hypothesized that additions of salt may facilitate phase separation through external charge compensations and increasing the total possible conformations of electrostatic associations given the presence of non-polyelectrolyte ions [1, 2, 6]. thus, pei was kept for subsequent experiments on the effects of salt addition. turbidity data as a function of mixing ratio was plotted in such a way that optimum mixing ratios for each gox-polycation system can be determined. although ranging in value from 0 – 100%, turbidity is typically employed as a qualitative indicator in characterizing phase behavior: for example, turbidity values above 20% usually suggest (but do not guarantee) the presence of phase separation whereas values nearing 0% imply the absence of phase separation. however, the presence and exact nature of phase separation, be it liquid-liquid or liquid-solid phase separation, can only be verified through optical microscopy. nevertheless, turbidity is useful as a continuous measure for when a system exhibits the same morphology under different conditions, and therefore can indicate the conditions most conducive to the desired phase behavior. in this paper, for example, turbidity is used to determine mixing ratios that best drives phase separation. out of the four mixing ratios investigated, the two mixing ratios most favoring liquid-liquid phase separation will be selected for subsequent investigations involving salt and ph. it was expected that maximum complexation, as determined by peaks in relative turbidity, for a given gox-polycation system would lie around mixing ratios of 84% and 88% (although turbidity values were generally higher at a mixing ratio of 92%, optical microscopy showed greater tendencies for precipitation). indeed, all gox-polycation systems except for gox-pei exhibited coacervate formation at such mixing ratios as determined via optical microscopy (data not shown). phase separation absence in gox-pei also explains the low turbidity magnitudes (below 20% at all mixing ratios). we predicted the absence of phase separation in the gox-pei system to be due to the lack of entropic gains from bound counter-ion release in electrostatic interactions within the system as a result of pei being too weakly charged. the larger relative sample standard deviations with gox-pvi and gox-pei suggests possible systematic inaccuracies: e.g., bubble formation interference at select mixing ratios, but the consistency of turbidity maximums coupled with optical microscopy images (data not shown) at mixing ratios of 84% and 88% supported the selection of these particular mixing ratios for subsequent experiments. while the quantitative limitations of turbidity must be considered, this may also suggest the dominance of (charged) protein-polymers system’s electrostatic interactions driving macro-phase separation on turbidity readings as opposed to the strengths of individual components. salt effects based on the two selected mixing ratios (88% and 84%) from the previous experimental section, the effects of salt (nacl) concentration on gox-polycation phase behavior were investigated (fig. 3). fig. 3c illustrates the characteristic binodal curve phase diagram in phase separating polyelectrolyte systems [1, 8, 12, 14]. in this investigation, however, we are more concerned with determining salt concentrations that promoted liquid-liquid phase separation for the two mixing ratios: 84% and 88%. turbidimetry analysis was used to provide a relative measure for the degree of phase separation across all gox-polycation systems and facilitate comparison across the two different salt concentrations. the two chosen salt concentrations were 25 mm and 50 mm based on preliminary data suggesting that this resulted in liquid-liquid phase separation in the gox-polycation systems being studied. at both mixing ratios of 84% and 88%, all gox-polycations that exhibited liquid-liquid phase separation in the absence of salt (fig. 2) continued to do so with increasing salt concentration. although gox-pvi and gox-qp4vp systems underwent a decrease in turbidity with the addition of salt from 25 mm to 50 mm at constant mixing ratio of 84% (fig. 3b), optical microscopy still showed liquid-like morphologies. given that there were still no indicators of phase separation in gox-pei with the addition of salt, as suggested by low turbidity magnitudes below 20% (figs. 3a, figure 2. complex coacervation of glucose oxidase (gox) enzyme with a palette of synthetic polycations. mixtures were prepared across a select range of macromolecule mixing ratios informed from a more thorough investigation of the protein’s phase behavior. error bars describe the standard deviation of each triplicated data point (n = 3). as a control, all turbidity values shown have had 10 mm tris turbidity reference values subtracted. 23 3b), and confirmed by optical microscopy, the system was excluded from subsequent experiments on the effects of ph on coacervate formation. as an exception amongst the phase-separating systems, pah did not demonstrate the desired formation of biomolecular condensates at both salt concentrations of 25 mm and 50 mm based on optical microscopy (data not shown). instead, its turbidity values consistently above 20% for 25 mm and 50 mm salt concentrations (figs. 3a, 3b) were shown to be a result of precipitate formation, which furthers evidence for the gox-pah system’s greater relative propensity for liquid-solid phase separation at the current salt concentration range. for gox-pah, low concentrations of salt ions may not provide sufficient charge compensations to favor liquidliquid phase separation, so much as it is strengthening electrostatic interactions via increasing possible coulombicdriven conformations. conversely, at higher salt concentrations, its charge screening effects would work towards dampening such electrostatic interactions to reduce the entropic gains from bound counter-ion release and drive coacervation [2, 6, 12]. an alternative explanation as to the goxpah system’s propensity for liquid-solid phase separation at low salt concentrations could be due to kinetic trapping effects as is prevalent in solid phases [2, 8, 12]. next, experiments with the gox-pah system involving higher salt concentrations were necessary to identify a “minimum” salt concentration at which the system undergoes complex coacervation instead of precipitation. thus, we conducted an additional salt titration exploring gox-pah phase behavior at higher salt concentrations: 150 mm to 200 mm nacl at 25 mm intervals. samples were analyzed via turbidimetry (data not shown) and optical microscopy (fig. 3d). while the possibility to utilize a different salt as per the hofmeister series e.g. kbr was considered, ultimately liquid-liquid phase separation was observed at a nacl concentration of 150 mm, with the optimum salt concentration for gox-pah determined to be 175 mm (fig. 3d), thereby negating the need to use a different salt species and maintaining consistency across all gox-polycation systems. ph effects building on preceding work determining ideal mixing ratio and salt compositions conducive to coacervate formation, the goal of this section is to effectively explore coacervate figure 3. the phase behaviors of multiple gox-polycation mixtures as functions of salt (nacl) concentration. goxpolycation mixtures were prepared using the two optimum mixing ratios for complex coacervation from fig. 2 (84% and 88%). both turbidity (λ = 600 nm) and optical microscopy were used to confirm liquid-liquid phase separation. error bar values represent sample standard deviation; n = 3 for all data points. a, turbidity versus salt concentration at constant mixing ratio of 88%. b, turbidity versus salt concentration at constant mixing ratio of 84%. all turbidity values shown in figs. 3a and 3b have had 10 mm tris turbidity reference values subtracted; results from fig. 2 were also included as no-salt controls. c, binodal phase boundary of complex coacervate systems where both charged polymers and charged proteins facilitate phase separation; φ denotes ‘phase’. arrow points to increasing two-phase region with increasing macromolecular charge density or patterning, demonstrating phasic tuneability with charge-associated parameters. d, optical microscopy images of gox-pah phase behavior (88% mixing ratio) provide sufficient qualitative evidence of liquid-liquid phase separation at salt concentration ranges beyond those in figs. 3a and 3b. 24 self-regulation by ph. all gox-polycation systems were set at an initial ph of ~9.5 in addition to their optimum compositions (specified in the experimental section), and a fixed volume of acid was titrated into the solution to slowly decrease the solution ph (fig. 4b). reductions in turbidity as ph decreases marks the dissolution of the coacervate microenvironment. since gox is characterized as a weak polyelectrolyte and has an isoelectric point (pi) of roughly ~4.2, protonation of gox below its pi reduces its net negative charge. thus, the coacervate microenvironment formed by gox and a polycation would deform as liquid-liquid phase separation dissipates from weakening electrostatic interactions. this is evident from the turbidity of each system being lowest at ph values below 4.2, the isoelectric point of gox (fig. 4b). however, the periodicity of the trends, notably with the gox-pah and gox-pvi systems, was unexpected (fig. 4b). we hypothesize possible induced charging effects on prolonging liquid-liquid phase separation given the uneven anisotropy and charge patchiness of gox [5, 8]. it is possible the coacervate microenvironment may persist at some of the lower ph ranges due to such effects, which may explain why each system undergoes a transition towards a turbidity maximum at the same ph value of 6 (indicating gox as the limiting factor) following a local minimum at ph 8~ due to the initial decrease in ph. ultimately, we hope to demonstrate ph-mediated regulation of our coacervate microenvironments. throughout the experimental section, we have shown how coacervate microcompartments can be formed with polymer-enzyme complexes. further, we can tune the interaction strength and morphology via mixing ratio, salt, and ph. due to many enzymatic reactions being affected by system ph, demonstrating the microenvironment’s ability to form and dissolve reversibly in response to ph represents the primary goal of our work on protein-polymer synthetic nanoreactors [4, 16]. it is important to stress that these results are building blocks towards this objective. conclusion an enzyme-polymer complex coacervate system was investigated on the basis of its ability to capture the complexities seen in biological condensates. the phase behaviors of multiple gox-polycation systems as functions of mixing ratios and ionic strength were investigated. we were interested in each gox-polycation system’s propensity to undergo liquidliquid phase separation, and the ionic stability of their formed coacervate microenvironments. gox phase separated upon mixing with the polycations qp4vp, pah, and pvi; but did not phase separate with pei. for the gox-pah system, the addition of salt was needed to screen existing charge such that the entropic gains favoring liquid-solid phase separation may be suppressed. the gox-pei system did not undergo any type of phase separation both with and without the addition of salt. it is also worth remarking that the same species of salt, nacl, may be used to induce liquidliquid phase separation for all phase separating gox-polycation systems, suggesting the salt species’ versatility for driving complexation. ph titrations were used to investigate the dynamics of the formed coacervate microenvironments for each gox-polycation system; turbidimetry suggests that all gox-polycation systems that phase separated at a ph of ~9.5 no longer phase separated upon reaching ph < pi of gox. we suggested the possibility of induced-charging effects as an explanation for the periodicity in phase behavior of gox-pvi as ph decreases, where the anisotropy of gox and presence of charge patches prolonged phase separation despite non-ideal ph [5, 10, 11]. nevertheless, while the exact phase behavior of gox-polycation systems at low ph were not determined as with optical microscopy, due to methodology limitations, the global minimums in turbidity at low ph compared to relative maximums at high ph strongly suggested dissolution of coacervates and dissipation of liquid-liquid phase separation. this crucial finding will inform future work on demonstrating the reversibility of coacervate formation in gox-polycation systems such that these microenvironments can be made smart and self-regulated to mimic the complexity of condensates found in cells. the establishment of ideal parameters at which various gox-polycation systems undergo liquid-liquid phase separation provides a foundation for coacervate microenvironment formation for use in advanced synthetic nanoreactor design and elucidation of cellular compartmentalization phenomena. figure 4. the effects of ph on complex coacervation in multiple gox-polycation systems. each gox-polycation mixture was set at ph of roughly 9.5, followed by titration of hydrochloric acid (hcl) until dissolution of complexation. turbidity (λ = 600 nm) was used to indicate absence of phase separation upon convergence towards a minimum value after multiple ph titrates. a, schematic depicting ph-induced dissolution of complex coacervation via protonation of gox, which reduces its net negative charge. b, measured turbidity of gox-polycation systems as a function of ph. 5th order polynomials were plotted to visualize turbidity trends with ph and to guide the eye (a 5th order polynomial represented the lowest-degree polynomial containing the local extremas that track the data) 25 author information corresponding author *htjo@umass.edu author contributions hansen tjo performed the experiments, analyzed data, and produced the figures and the manuscript. nick zervoudis devised the experimental plan and reviewed the manuscript. dr. allie obermeyer reviewed the manuscript. funding sources this work was supported by the fu foundation school of engineering and applied sciences at columbia university and funding from the columbia chemical engineering reu program. acknowledgments the authors would like to acknowledge members of the obermeyer group for helpful discussions. abbreviations gox – glucose oxidase hcl – hydrochloric acid nacl – sodium chloride pah – poly(allylamine hydrocholoride) pei – poly(ethylenimine) pvi – poly(1-vinyl imidazole methyl iodide) pi – isoelectric point qp4vp – poly(4-vinyl n-methyl pyridinium iodide) references [1] l.p. bergeron-sandoval, n. safaee, s.w. michnick, mechanisms and consequences of macromolecular phase separation. cell 165(5),1067–1079 (2016). [2] w.c. blocher, s.l. perry, complex coacervate-based materials for biomedicine. wiley interdisciplinary reviews: nanomedicine and nanobiotechnology 9(4), doi:10.1002/wnan.1442 (2016). [3] m. castellana, m. z. wilson, y. xu, p. joshi, i. m. cristea, j. d. rabinowitz, z. gitai, n. s. wingreen, enzyme clustering accelerates processing of intermediates through metabolic channeling. nature biotechnology 32(10), 1011–1018 (2014). [4] j.-m. choi, s.-s. han, h.-s. kim, industrial applications of enzyme biocatalysis: current status and future aspects. biotechnology advances 33(7),1443–1454 (2015). [5] f. comert, p.l. dubin, liquid-liquid and liquid-solid phase separation in protein-polyelectrolyte systems. advances in colloid and interface science 239, 213– 217 (2017). [6] c.s. cummings, a.c. obermeyer, phase separation behavior of supercharged proteins and polyelectrolytes. biochemistry 57(3), 314–323 (2017). [7] j. horn, r. a. kapelner, a.c. obermeyer, macroand microphase separated protein-polyelectrolyte complexes: design parameters and current progress. polymers 11(4), 578 (2019). [8] p. jha, p. desai, j. li, r. larson, ph and salt effects on the associative phase separation of oppositely charged polyelectrolytes. polymers 6(5):1414–1436 (2014). [9] r. a. kapelner, a.c. obermeyer, ionic polypeptide tags for protein phase separation. chemical science 10(9), 2700–2707 (2019). [10] l. li, s. srivastava, m. andreev, a.b. marciel, j. j. d. pablo, m. tirrell, phase behavior and salt partitioning in polyelectrolyte complex coacervates. macromolecules 51(8), 2988–2995 (2018). [11] a.c. obermeyer, c.e. mills, x.-h. dong, r. j. flores, b.d. olsen, complex coacervation of supercharged proteins with polyelectrolytes. soft matter 12(15), 3570–3581 (2016). [12] s.l. perry, phase separation: bridging polymer physics and biology. current opinion in colloid & interface science 39, 86–97 (2019). [13] s.l. perry, y. li, d. priftis, l. leon, m. tirrell, the effect of salt on the complex coacervation of vinyl polyelectrolytes. polymers 6(6), 1756–1772 (2014). [14] d. priftis, m. tirrell, phase behaviour and complex coacervation of aqueous polypeptide solutions. soft matter 8(36), 9396–9405 (2012). [15] d. priftis, n. laugel, m. tirrell, thermodynamic characterization of polypeptide complex coacervation. langmuir 28(45), 15947–15957 (2012). [16] e. ricca, b. brucher, j.h. schrittwieser, multi-enzymatic cascade reactions: overview and perspectives. advanced synthesis & catalysis 353(13), 2239–2262 (2011). 26 i’m upset, not ashamed: an investigation into adolescent shame acknowledgement olivia putnam1 , keely lake2 1department of psychology, northwestern university, evanston, il 2wayland academy, beaver dam, wi shame, emotion, psychology, negative affect, mood disorder, self-conscious emotion abstract: shame is a powerful and acutely painful “master emotion” that is strongly correlated with maladaptive behaviors and a host of psychological symptoms. concerningly, the affect remains under-researched and difficult to identify or address in a clinical setting. this may be caused, at least in part, by shame’s intrinsically hidden nature, which drives people to deny the emotion and express it through other means. this study aimed to understand the degree to which people fail to acknowledge their own shame and the psychological and behavioral implications of this shame. participants completed both a self-report measure of shame and an empirical assessment of internalized shame, as well as measures of shame coping methods and emotional regulation. as expected, results showed no significant correlation between participants’ self-rated shame and measured shame. we also saw a significant correlation between assessed internalized shame and use of shamecoping methods as well as difficulty in emotion regulation – specifically, difficulty with clarity of emotion, acceptance of emotions, and strategies for coping with emotions. these findings indicate that people struggle to acknowledge their own shame and also speak to the maladaptive, dysregulated ways people manage their shame. recognizing shame as a powerful emotion with implications in psychiatric disorders and understanding the factors that prevent people from acknowledging their own shame may help improve treatment for those who struggle with the emotion and reduce the likelihood that they will engage in maladaptive coping behaviors. introduction shame is one of the most under-researched emotions. whereas emotions like sadness, anger, and nervousness have undergone decades of psychological research and have come to be seen as the underlying feelings behind widespread clinical diagnoses like anxiety or depression, shame research within the field of psychology was virtually nonexistent until the last two decades and remains remarkably limited today. this lack of research is particularly troubling when it comes to shame because the affect is highly maladaptive. unsurprisingly, as a result, it is highly correlated with addiction, depression, violence, aggression, bullying, suicide, and eating disorders. in recent years, it has also become an emerging component of ptsd (taylor, 2015). a brief history of shame conceptualization one especially influential early premise for defining shame, and differentiating it from guilt, is early anthropologists’ focus on public vs. private transgressions (e.g., benedict, 1946). anthropologists commonly distinguished shame based on the situations they believed elicited it. more specifically, shame was conceived as a "public" emotion, arising from public exposure and disapproval of some transgression in societal rules and norms. guilt, on the other hand, was described as a more "private" experience arising from self-inflicted criticism and regret. however, more recent empirical research has failed to support this public/private distinction (tangney, marschall, rosenberg, barlow & wagner, 1994; tangney, miller, flicker & barlow, 1996). one such example is a study conducted in 1992 asking participants to describe three guilt-inducing events and three shameinducing events. a systematic analysis of the social context of these events found that shame and guilt are equally likely to be experienced in the presence of others (tangney, et al., 1992). "solitary" shame experiences were equally as common as "solitary" guilt experiences. even more to the point, “the frequency with which others were aware of the respondents' behavior did not vary as a function of shame and guilt” (tracy, 2011). this led to a new conceptualization of shame which remains today: shame as holistic negative self-concept. helen block lewis, renowned psychologist and pioneer of this shame-understanding, asserted that while guilt involves a negative evaluation of a specific behavior, shame involves a negative evaluation of the global self ("i did something bad" vs. "i am bad") (1971). though the distinction may appear inconsequential, this contrasting emphasis on the self “sets the stage for very different emotional experiences and very different patterns of motivation `and subsequent behavior” (tracy, 2011). the two emotions, for instance, produce distinct * 27 “action tendencies.” shame is commonly accompanied by attempts to deny, hide from, or escape the experiences that elicit shame, while guilt typically leads to “reparative action” – confessing, apologizing, undoing. this difference in internal conceptualization of the self and subsequent “action tendencies” is part of what ultimately makes shame maladaptive. while guilt can be painful and overwhelming, it is generally limited to the guilt-inducing action or experience. shame consumes the entire self, leaving experiencers with a globally-negative self-conception (“i am a terrible person”). this negative self-concept is not only painful and distressing, but it also feels irreparable. a person can correct a behavior, but one’s fundamental essence seems permanent. this sense of futility drives much of shame’s maladaptivity. rather than embracing adaptive behaviors like apologies or changes in behavior, which increase psychosocial success, shame-experiencers tend to recede and hide from the shame-inducing event. often, this leads people to isolate themselves socially, withdraw from activities that potentially remind them of the shame, and engage in anhedonic-behaviors. in other cases, this avoidance manifests as anger or hostility, as experiencers attempt to “turn the tables” on others to avoid their own shame, or as risky behavior (i.e. substance abuse) which many use a distraction from their shame (ellison, 2006). in instances where shame-experiencers have in fact committed some wrongdoing (the determination of which is of course subjective) the failure to take the expected “reparative action” can lead to social conflict (tangney, stuewig & mashek, 2007). for those who experience shame about events for which reparative action is not typically expected (i.e. being the victim of sexual assault, mental illness, minor mistakes or failures), shame can cause dissociation, debilitate people from talking about their experience, and limit much-needed processing of their own emotions and/or trauma (taylor, 2015). either way, these maladaptive responses to shame led researchers to “consistently report a positive relationship between proneness to shame and a host of psychological symptoms, including depression, generalized anxiety and social anxiety, low self-esteem, ptsd, eating disorder symptoms, cluster c personality disorders, suicidal behavior and self-injurious behavior, and substance abuse” (tracy, 2011). shame acknowledgement in the context of shame’s maladaptive consequences, the lack of research into the affect becomes dangerous. without comprehensive research into shame and its implications in psychological disorders, we cannot develop evidence-based treatments for shame-related disorders or, more importantly, adjust treatments for preexisting mental health disorders in which shame plays a more important role than previously-realized. one roadblock that commonly hinders the development of these treatments or the confrontation of shame within a therapy setting is the fact that shame often goes unacknowledged by the experiencer (mcgonigal, 2016). as terry f. taylor ph.d. writes in a review article of peritraumatic shame, “shame...is a virtually invisible, ubiquitous part of everyday life. because the experience of shame is often considered to be painful and disempowering, and because recognition of shame in itself can be felt as shameful… shame remains unacknowledged and is expressed as avoidant behavior” (2015). this instinct to hide one’s shame “makes it difficult to recognize internally when it happens,” let alone acknowledge out loud (luoma, 2012). concerningly, this tendency among people not to acknowledge their own shame also makes it difficult to study the affect, as it renders self-report measures unreliable. present study the consensus that shame characteristically goes unrecognized has never been scientifically reviewed. further, researchers have not studied whether the degree to which people report their own shame correlates with their mental health in other capacities. our research attempts to fill this gap. like most shame-studies before it, we utilize assessments intended to empirically measure participants’ levels of shame. in addition to these assessments, however, we also use an assessment of emotional affect that asks participants to self-rate the frequency with which they experience different emotions, “ashamed” being one them. comparisons of participants’ scores on the shame assessments and their self-reported level of shame will act as quantified measurements of how well they acknowledge their own shame. this research will not only test the assumption of shame’s unidentified nature but may also provide some insight into the prevalence of unacknowledged shame and how it affects people’s mental health. methods participants and procedure the assessments were administered to a sample of 54 adolescent students between the ages of 14 and 17 (58% female, 42% male, m = 16.3 years old, sd = 1.3, 55% american). participants were recruited from the student population at wayland academy, a small boarding high school with international students. as an incentive for participation, students were offered a small amount of extra credit in their science and math classes. participants were each given a battery of psychological assessments including the positive and negative affect scale (panas), the internalized shame scale (iss), the compass of shame scale (css), and the difficulty in emotion regulation scale (ders). participants were assured that their responses would be anonymous and confidential. assessments were given in a quiet, distraction free room. assessments positive and negative affect scale: (panas; watson and clarke, 1998) the most commonly used measure of affect in scholarly research, the panas is comprised of 10 negative affects (afraid, upset, distressed, jittery, nervous, ashamed, guilty, irritable, hostile) and 10 positive affects (enthusiastic, interested, determined, excited, in28 spired, alert, active, strong, proud, attentive). participants use a 0-4 likert scale to rate the frequency with which they tend to experience each affect. for the purposes of this research, the “ashamed” item was used to measure self-rated shame because, unlike empirical assessments of shame, it requires participants to explicitly endorse the word “ashamed.” internalized shame scale: (iss; rosario and white, 2006) the internalized shame scale, a 30-item questionnaire, is the most widely used empirical measure of shame across psychology and sociology research. the assessment has two subscales that are intended to be reported separately: a 24-item shame scale and a 6-item self-esteem scale. the shame scale attempts to tease apart different experiences of shame to create a holistic measure of the affect. importantly, the iss does not actually use the word shame, because shame can itself be a shameful thing to admit. there are four identified cutoffs: a score of 50 or higher indicates problematic levels of shame, a score of 60 or higher indicates possible depression and/or other emotional or behavioral problems, a score of 70 or higher indicates a high probability of depression and/or other emotional or behavioral problems. compass of shame scale: (css; ellison, 2006) because shame is an emotion that commonly goes unacknowledged both internally and outwardly, it is often expressed through other emotions or behaviors. the compass of shame scale recognizes this tendency and assesses the maladaptive ways people cope with shame. the four coping methods that it identifies are “attack self,” “attack others,” “withdraw,” and “avoid.” this scale is of particular importance because it acknowledges shame as a fundamental source for many other maladaptive, unhealthy behaviors and is the first of its kind to assess and quantify these shame-based behaviors. additionally, the test questions are situational rather than experiential. participants cannot always recognize feelings as shame, but they can often identify situations that produce those shame feelings. difficulty in emotional regulation scale: (gratz, 2004) this assessment represents one of the most popular, comprehensive and well-established measures of emotion regulation and is widely used in both clinical and nonclinical settings. the questionnaire assesses five primary components of emotional regulation: emotional awareness, emotional clarity, emotional acceptance, impulse control, ability to engage in goal-directed behavior while experiencing negative emotions, and ability to use situationally appropriate emotion regulation strategies flexibly to modulate emotional responses as desired. results self-rated shame vs. measured shame to investigate the relationship between self-rated shame and measured shame, pearson correlations were computed between ratings from the panas item “ashamed” and scores from the internalized shame scale. as expected, there was no significant correlation between self-rated shame and assessed “true” shame. the correlation and significance are displayed in figure 1. figure 1. pearson correlation between panas “ashamed” score and internalized shame score. this graph shows the pearson correlation between panas “ashamed” rating and iss scores, r = 0.0806, p > 0.05 to visualize how shame acknowledgment relates to shame’s maladaptivity, an anova test for the significance of differences in mean panas “ashamed” ratings among internalized shame cutoff groups – shame w/in normal limits, problematic levels of shame, possible indicator of mental health disorders, and likely indicator of depression/mental health disorder -was performed and is displayed in figure 2. an anova test for the differences in panas “ashamed” ratings between iss cutoff groups was computed and is displayed in figure 2. figure 2. analysis of variance test of panas “ashamed” score by iss cutoff groups. this graph shows the mean panas ashamed ratings reported by individuals in different iss cutoff groups, f ratio = 0.903, p > 0.05 as expected, and in line with our other findings, there was no significant difference in self-rated shame (panas “ashamed”) between among iss cutoff groupings (i.e. all participants reported similar levels of sham, no matter how much shame an individual was actually experiencing). 29 identifying shame as negative affect in order to explore whether participants may identify their internalized shame as other negative emotions, pearson correlations were computed between panas negative affect item scores and internalized shame scale scores. both are displayed in table 1. table 1. pearson correlations between panas negative affect subscores and internalized shame scale score. this table shows the results of pearson correlations between various panas negative affect subscores and iss score. there was also a significant positive correlation between panas negative affect subscale total score and internalized shame score (r = 0.534, p < 0.01). implications of shame to investigate shame’s relationship with other emotional and behavioral problems, pearson correlations were computed between iss scores and scores on the individual compass of shame scales and the difficulty in emotion regulation scale (figure 3). strong correlations between the iss and the withdrawal and attack self scales were expected and obtained. both correlations were significantly stronger than the iss correlations with the avoidance and attack other scales. all correlations were significant. category differences male/female, age, and continent of origin differences were assessed for both the internalized shame scale and the panas “ashamed” rating. women tended to have higher internalized shame scores than men (t = 1.76, p < 0.05). there was no significant difference, however, between men and women for the panas “ashamed” rating. there was also no significant correlation between age and internalized shame score or the panas “ashamed” rating. continent of origin did not produce any significant differences in iss score or panas “ashamed.” figure 3. pearson correlations between css subscores and iss scores. a, pearson correlation between css attack self subscore and iss score, r = 0 .75, p < 0.001 b, pearson correlation between css withdraw subscore and iss score, r = 0 .72, p < 0.001 c, pearson correlation between css avoidance subscore and iss score, r = 0.41, p < 0.01 d, pearson correlation between css attack others subscore and iss score, r = 0.27, p < 0.05 30 discussion the lack of correlation between iss score and panas “ashamed” rating, as displayed in figure 1, indicates that there is no relationship between self-rated shame and empirically-measured “true” shame. this is further demonstrated by the analysis of variance displayed in figure 2, which shows that there is no significant difference in selfrated shame between cutoff categories of “true” iss shame. essentially, even those who experience shame at an intensity high enough to indicate depression or other mental health disorders tended to describe their experience of shame as “rare.” a paired t-test comparing average iss shame and panas shame affirmed this underreporting phenomenon in individuals, with nearly 74 percent of participants reporting lower self-rated panas shame than “true” iss shame. importantly, there was a significant positive correlation between iss score and panas negative affect subscale score, suggesting that while people struggle to accurately identify their shame, they may describe it broadly as negative affect. this point is reiterated by the fact that, with the exception of “distressed,” the individual panas negative affect items had weak or insignificant correlations with iss score and none had stronger correlations than the negative affect subscale score. this indicates that people are not calling shame by another name, but instead use negative umbrella terms, or a variety of different emotion words, to imprecisely describe the feeling. these results both confirm a common understanding that shame often goes unacknowledged and also underline the difficulty of identifying and properly addressing shame. this pattern becomes especially meaningful in the context of internalized shame’s negative implications. iss score was found to have positive correlations with all css scales: attack self, withdraw, avoid, and attack others. given their naturally internalized nature, the “withdraw” and “attack self” coping mechanisms had a stronger relationship with internalized shame than “avoid” or “attack other.” internalized shame also showed a strong positive relationship with ders score and with the lackof-clarity, non-acceptance, and strategies subscales specifically. this fits well with the findings about unacknowledged shame, as clarity and acceptance are both components of emotional acknowledgment, and are necessary for strategic management of one’s emotions. the male/female differences on the iss reflect those of previous studies, with women tending to experience more internalized shame than men. also similar to adult studies, age was not a significant factor in iss score, suggesting that experience of shame does not change significantly during adolescent development. given that this comes from a cross-sectional review rather than a longitudinal one, however, the accuracy of this conclusion is limited. continent of origin, which has never before been studied in relationship to internalized shame, produced no significant difference. the accuracy of this conclusion, however, may be compromised by the small and varying number of participants within each continent group. conclusion the comparisons of self-rated shame and empirically measured “real” shame confirm both our hypothesis and a larger long-held public understanding that people hesitate to acknowledge their shame. this finding is important not only because it is the first of its kind to quantitatively validate that informal understanding, but, more significantly, because it speaks to the extent of the disparity. more than one third of participants experienced internalized-shame with a frequency associated with depression and other clinical disorders, yet the great majority of these participants rated their own experience of shame as rare. this is problematic because, if people cannot acknowledge shame as a component of their emotional distress or mental illness, then these problems become much more difficult to address within a clinical or intrapersonal context. this lack of acknowledgment becomes especially concerning if the shame centers around a specific, potentially-traumatic event (i.e. sexual assault). if shame prevents a person from speaking up about and working through such an experience, then symptoms can worsen dramatically. as time goes on, this can also become a self-fulfilling prophecy of sorts because the longer one avoids their shame the more internalized it becomes and, consequently, the more inhibiting it becomes. given that shame is a powerful emotion, however, it cannot be entirely ignored or suppressed. in fact, this study’s findings indicate that people may be able to recognize the emotion broadly as emotional distress. however, without specific identification and management, shame is often expressed through maladaptive coping mechanisms. the correlation between internalized shame and each pole of the shame-coping scale suggests that, rather than addressing shame head on, shameexperiencers tend to avoid the emotion through problematic behaviors such as: risk-taking and distraction, which can develop into substance abuse and have been shown to be severely maladaptive; withdrawal from social interaction, often a symptom of depression; excessive selfcriticism, also related to mental illness; or attacking others, an instinct associated with aggression and potential violence. though they vary in commonality, each pole is maladaptive in its own right—an attempt to ignore, hide from, wallow in, or push back one’s shame without ever truly acknowledging it. it is also important to note that these individuals coping mechanisms are not orthogonal and in fact tend to converge. this tendency to cope with shame is similarly reflected by the significant correlation between internalized shame and difficulty in emotional regulation. even more to the point, the ders subscales that had the strongest relationships with shame were lack-of-clarity, nonacceptance, and strategies. this indicates that people with high levels of internalized shame have significant trouble identifying their emotions, acknowledging their emotions without guilt or embarrassment, and coping with their emotions effectively. this not only speaks to the troubling ways people manage their shame but also to the dysregulatory nature of shame itself, which can make it 31 more difficult for a person to manage any of their emotions. the implication of these findings is three-fold: tools for shame recognition must be better integrated into both clinical and intrapersonal settings, shame must be better accounted for within diagnostic criteria, and more research into the affect must be conducted. it is imperative that mental health practitioners recognize that patients are unlikely to forthrightly acknowledge or report their shame and that these practitioners are trained in how to identify this underlying shame and address it without causing the patient to shut down or react with anger. it is also an unfortunate truth that, because the perception of shame in others “can also evoke a discomforting emotion in the observer,” it may fail to be addressed in therapy, with the therapist remaining in an “unconscious collusion with the patient” to ignore the shame (taylor, 2015). additional shame-specific training and wider acknowledgement of the emotion’s role in psychopathology may help mediate this problem an important step in helping clinicians (and, in fact, any individual) better identify and work with shame is better accounting for the emotion in diagnostic criteria. despite the fact that the emotion is strongly correlated with many different mental illnesses and social-emotional problems, shame is rarely listed as a symptom in the dsm iv and is generally relegated to the “associated features” of a disorder. not only does this lack of representation reinforce shame’s hidden nature, but it also fails to account for the way shame’s role in a disorder can change the way it must be treated. perhaps most importantly, more scientific research must be conducted into shame. since shame’s psychological conceptualization as holistic self-blame was established, only a handful of scientific studies into the affect have been conducted and, even fewer yet have investigated shame’s role in psychiatric disorders. if we are to help people recognize and cope with their internalized shame, we must first understand what prevents that acknowledgment and which tools are the most effective in fostering it. author information department of psychology, northwestern university evanston, il, usa references [1] h. l. bash, a. papa, shame and ptsd symptoms. psychological trauma: theory, research, practice, and policy. 6(2), 159–166 (2014). [2] r. benedict, the chrysanthemum and the sword, (houghton mifflin harcourt, 1946). [3] p. m. del rosario, r. m. white, the internalized shame scale: temporal stability, internal consistency, and principal components analysis. personality and individual differences. 41(1), 95–103 (2006). [4] j. elison, r. lennon, s. pulos, investigating the compass of shame: the development of the compass of shame scale. social behavior and personality: an international journal. 34(3), 221–238 (2006). [5] k. l. gratz, l. roemer, multidimensional assessment of emotion regulation and dysregulation: development, factor structure, and initial validation of the difficulties in emotion regulation scale. journal of psychopathology and behavioral assessment. 26(1), 41–54 (2004). [6] h. b. lewis, shame and guilt in neurosis, (international universities press, 1974). [7] j. luoma, j. lejeune, association for contextual behavioral science. (2005). available at: https://contextualscience.org/acbs [8] k. mcgonnigal, r. d. siegel, breaking the power of shame by engaging it. (2016). available at: https://nextlevelpractitioner.nicabm.co/week40/ [9] l. a. medrano, m. trogolo, construct validity of the difficulties in emotion regulation scale: further evidence using confirmatory factor analytic approach. abnormal and behavioural psychology. 2(2), 1000117 (2016). [10] p. michl, et al, neurobiological underpinnings of shame and guilt: a pilot fmri study. social cognitive and affective neuroscience. 9(2), 150–157 (2012). [11] d. sznycer, et al, cross-cultural differences and similarities in proneness to shame: an adaptationist and ecological approach. evolutionary psychology. 10(2), 147470491201000 (2012). [12] k. tangey, w. fischer, self-conscious emotions: shame, guilt, embarrassment, and pride, (guilford press, 1995). [13] j. p. tangney, recent advances in the empirical study of shame and guilt. american behavioral scientist. 38(8), 1132–1145 (1995). [14] j. p. tangey, "shame and guilt in interpersonal relationships" in self-conscious emotions: the psychology of shame, guilt, embarrassment, and pride, j. p. tangey, k. w. fisher, eds. (guilford press, 1995), pp 114-139. [15] j. p. tangney, conceptual and methodological issues in the assessment of shame and guilt. behavior research and therapy. 34(9), 741-754 (1996). [16] j. p. tangney, s. a. burggraf, p. e. wagner, "shame-proneness, guilt-proneness, and psychological symptoms" in self-conscious emotions: the psychology of shame, guilt, embarrassment, and 32 pride, j. p. tangey, k. w. fisher, eds. (guilford press, 1995), pp. 343-367. [17] g. taylor, pride, shame and guilt: emotions of selfassessment. philosophical books. 28(1), 40–41 (1987). [18] t. f. taylor, the influence of shame on posttrauma disorders: have we failed to see the obvious? european journal of psychotraumatology. 6(1), 28847 (2015). [19] j. tracy, j. p. tangney, m. r. leary, handbook of self and identity, (guilford press, 2012). [20] d. watson, l. a. clark, a. tellegen, development and validation of brief measures of positive and negative affect: the panas scales. journal of personality and social psychology. 54(6), 1063–1070 (1988). [21] m. r. zaslav, shame-related states of mind in psychotherapy. j psychother prac res. 7(2), 154–166 (1998). 33