81 Screening of Multiple Potential Control Genes for use in Caste and Body Region Comparisons Using RT-qPCR in Coptotermes formosanus by Matthew R. Tarver1, Christopher P. Mattison2, Christopher B. Florane1, Doug J. Hinchliffe3, Dunhua Zhang1 & Alan R. Lax1 ABsTRACT Formosan subterranean termites, Coptotermes formosanus, are a significant worldwide pest. Molecular gene expression is an important tool for understand- ing the physiolog y of organisms. The recent advancement of molecular tools for Coptotermes formosanus is leading to the advancement of the understand- ing of termite physiolog y. One of the first steps in analyzing gene expression is the normalization to constant reference genes. stable reference genes that have constant expression across multiple treatments are important for accu- rately comparing target genes' expression. The objective of this investigation was to analyze and validate a set of potential reference genes including 18S rRNA; Glyceraldehyde 3-phosphate dehydrogenase (Gadphd); ribosomal protein L7 (RPL); β-actin (BA1); α-tubulin (Atube); α-actin (Aactin); and elongation factor (Elong) as standards for analysis of transcriptional changes in the termite Coptotermes formosanus, across two phenotypic castes, body regions, and colonies. We also compared the expression of hexamerin-1 and 2 using stable and unstable reference genes to demonstrate the importance of consistent control genes. Our results demonstrate that 18S and RPL can serve as reliable expression standards when comparing these different castes and body regions, and we show that C. formosanus Hex-1 and Hex-2 have expression patterns similar to that previously described in R. flavipes. INTRODuCTION Formosan subterranean termites (Coptotermes formosanus) are an ex- tremely destructive urban pest and are expanding their range throughout united states Department of Agriculture, Agricultural Research service, southern Regional Research Center, 1100 Robert E. Lee Blvd., New Orleans, LA 70124 1Formosan subterranean Termite Research unit 2Food Processing sensory Quality Research unit 3Cotton Chemistry and utilization Research unit 82 sociobiolog y Vol. 59, No. 1, 2012 the world. Following introduction of the species into the united states, they have successfully spread through the southern gulf coast states (su & scheffrahn 2000, Lax & Osbrink 2003). Research focused on the control of this species has advanced through the years but there is an increased need for species-specific control methods. The implementation of molecular biolog y resources has added to the knowledge of the basic biolog y of this serious pest. understanding the molecular processes underlying termite growth, differen- tiation, and response to stress will lead to the identification of novel target sites for termite-specific control. Gene expression profiles will be beneficial in focusing the search for termite control targets. One established method for measuring gene expression is through reverse transcription quantitative real-time polymerase chain reaction (RT-qPCR). The RT-qPCR technique is an extremely sensitive method for nucleic acid detection, genotyping, snp analysis and the quantification of specific messenger RNA (mRNA) levels. several factors, including quality of template RNA, choice of enzyme and other reagents, selection of control(s) (or reference) genes, and method of analysis can have a significant effect on results. slight variations in the transcript abundances of chosen reference gene(s) thought to be temporally and/or spatially static can artificially amplify or lessen ap- parent changes in target transcript quantification. Application of insufficient reference standards can cause inaccurate normalization of experimental data leading to increased error or incorrect conclusions. Rather than using a single reference gene, the use of several internal reference standards can be applied to more accurately measure changes in gene expression (Vandersompele et al., 2002). Careful consideration must be taken in the choice of these standards for each system and experimental condition analyzed to ensure adequate normalization for accurate analyses of results. In previous studies characterizing gene expression in termites, several dif- ferent control genes were utilized (scharf et al. 2003, 2005a, 2005b, Zhou et al. 2006a, 2006b, 2006c, 2007a, 2007b, 2008, Wheeler et al. 2010, Korb et al. 2009, Wiel et al. 2009, Hojo et al. 2009). However, it is unclear from these studies which control genes provide reliable high-quality standards for gene expression studies in C. formosanus termites. These types of analyses would benefit from a well characterized and validated set of control genes that provide a stable standard for analysis across various experimental conditions. 83 Tarver, M.R. et al. —screening of Multiple Control Genes in C. formosanus The objective of this investigation was to analyze and validate a set of potential reference genes including 18S rRNA; Glyceraldehyde 3-phos- phate dehydrogenase (Gadphd); ribosomal protein L7 (RPL); β-actin (BA1); α-tubulin (Atube); α-actin (Aactin); and elongation factor (Elong) as standards for analysis of transcriptional changes in the termite Coptotermes formosanus, across two phenotypic castes, body regions, and colonies. In ad- dition we incorporated one exogenous (Alien) RNA control into our study. A wide range of Ct (threshold cycle) variation was observed among this set of candidate control genes, and to test their utility we applied these control genes as standards for the analysis of two target genes Hexamerin 1(Hex-1) and Hexamerin 2 (Hex-2). When we applied the most stable set of refer- ence genes, we found that Hex-1 and Hex-2 expression in C. formosanus was similar to Reticulitermes flavipes, with an overall up-regulation in workers and increased expression in the carcass. Predictably, target gene comparison between variable and non-variable control genes resulted in statistically dif- ferent results, underscoring the idea that careful selection of control genes is essential for accurate gene expression analysis. MATERIALs AND METHODs Termites Formosan subterranean termites were collected from cardboard rolls placed in bucket traps at various locations in New Orleans City Park (New Orleans, LA.). The termites were removed from the collection material, separated, and placed into plastic containers (17.1 x 12.2 x 6.0cm Pioneer Plastics) contain- ing slats of moist spruce (Picea sp.) wood. The containers were then placed into an incubator maintained at 25°C, 70 % relative humidity and constant darkness. Individuals collected from two independent colonies were used in the experiments. 16s mitochondrial rDNA sequencing was used to verify their identities as Coptotermes formosanus (szalanski et al. 2003). Dissection A total of four sets of 25 workers and soldiers from a colony designated 1732 were dissected to obtain the tissue from each individual’s head, gut and carcass for extraction of RNA. Termites were dissected using an Olym- pus sZX-12 dissecting microscope (Olympus America Inc., Center Valley, 84 sociobiolog y Vol. 59, No. 1, 2012 PA). Each individual was decapitated and the entire gut removed with fine forceps by gently squeezing the hind end and pulling until the entire gut was withdrawn from the carcass. The gut was placed into a 1.5 ml eppendorf tube containing 100ul phosphate buffered saline (PBs). The remaining head and carcass were placed into separate 1.5 ml eppendorf tubes. This procedure was repeated 25 times to provide isolated head, gut, and carcass samples for each individual termite in separate tubes. The tubes were then placed into a -80°C freezer prior to extraction of RNA. This process was repeated to obtain 4 biological replicates of 25 heads, guts and carcasses from worker termites. The same procedures were followed to obtain 4 biological replicates of heads, guts, and carcasses from soldier termites as well. This was then repeated to obtain 4 biological replicates of 25 from workers and soldiers from a second colony designated 1314. Extraction of RNA Total RNA was isolated from each set of the 25 worker or soldier fro- zen body regions using the sV Total RNA Isolation kit (Promega Corp., Madison, WI) according to manufacturer’s instructions. RNA was similarly extracted from the extirpated guts using the same kit. The extracted RNA was quantitated using a NanoDrop instrument (NanoDrop Technologies Inc., Wilmington, DE). Design of primers Gene targets and primers were chosen based upon homolog y to past termite control genes and control genes from other species (Hojo et al. 2009, Tarver et al. 2010. Zhou et al. 2006a,b,c; 2007a,b). specific gene sequences were identified from an in-house EsT library (D.Z. usDA-ARs-sRRC, unpub- lished results). Eight control and two target genes were chosen. Primers were designed using Beacon Designer 7.61 (Premier Biosoft, Palo Alto, CA) for sybr qRT-PCR. Hexamerin-1 and -2 were chosen as test target genes based on previous research (Zhou et al. 2006a, 2007b). All primer sequences are listed in Table 1. Primers were purchased from Integrated DNA Technologies (IDT www.idtdna.com). Reverse Transcription Quantitative Real-Time PCR Reverse transcription reactions for cDNA synthesis were performed using a QuantiTect Reverse Transcription Kit (Qiagen Inc., Valencia, CA) and the 85 Tarver, M.R. et al. —screening of Multiple Control Genes in C. formosanus Alien Reference RNA qRT-PCR Detection Kit (Agilent Technologies Inc., santa Clara, CA). A 1 μl aliquot of the Alien RNA transcript (3x1010 copies / μl) was diluted with 59 μl of RNAse free water to a final concentration of 5x108 copies / μl. From the initial 1:60 dilution, four 10-fold serial dilutions were made to achieve a final concentration of 5x104 copies / μl. For the initial reaction, 2 μl of 7X gDNA Wipeout Buffer (Qiagen Inc.) and a 2 μl aliquot of the 5x104 concentration Alien RNA template were added. The amount of template RNA used from each sample was 130 ng and 240 ng for Colonies 1732 and 1314 respectively. Additional RNase free water was then added as needed until the final volume of each reaction was 14 μl. samples were then incubated at 42ºC for 2 minutes and immediately placed on ice. To each reaction, 1 μl of Quantiscript Reverse Transcriptase, 4 μl of 5X Quantiscript Reverse Transcriptase Buffer and 1 μl Reverse Transcriptase Primer Mix was added to bring the total volume of each sample to 20 μl. samples were then incubated at 42ºC for 15 minutes followed by a 3 minute incubation at 95ºC to inactivate the Quantiscript Reverse Transcriptase. All above reagents were from Quiagen (Valencia, CA) The qPCR reactions were performed in a Bio-Rad CFX96 real time PCR detection system (Bio-Rad Laboratories, Hercules, CA) using the fast protocol Table 1. Primer pair sequences designed for RT-qPCR. RT-qPCR Primer Pairs ( 5’ - 3’ ) Blastx sequence Description Forward Reverse Ribosomal protein L7 (RPL) attgacaacagagttcgtaggc cgttatgcaccagtaccttcc 18s ribosomal rRNA (18S) cgagattgagcaataacaggtc acgtaatcaacgcgagcttatg β-Actin (BA1) ctatgtcgccctggactt cgttgccaatggtgatga Glyceraldehyde 3-phosphate dehydrogenase (GADPHD) ggtagttaatggtcacaagatt gcagaagccttatcaatgg α-tubulin (Atube) cttaccacgagcagttgagt cgaggatcacacttcaccat α-actin (Aactin) tcttggatgattacgagtgtga tctcaatcctctcagcaattct Elongation factor (Elong) cttggatggtggaaaggagtaact atcttggattggaaggcgaagt Hexamerin 1 (Hex-1) gaatctgctgctacccaagg aatcccagtggcttgtcatc Hexamerin 2 (Hex-2) actctgctccgaaaggtgaa tcgaaatcatgttcgctctg 86 sociobiolog y Vol. 59, No. 1, 2012 with iTaq™ Fast sYBR® Green supermix (Bio-Rad Laboratories). All primer sets were first run at a temperature gradient (53-63°C) to obtain the optimal annealing temperature. Optimal annealing temperature was determined to be 60°C. All samples (worker head, gut, carcass; soldier head, gut, carcass; Colony 1 and Colony 2) had four biological replicates per treatment (n=48). Three technical replicates were run for each biological sample. A dissociation curve was generated and used to validate that a single amplicon was present for each RT-qPCR reaction. Data Analysis Averaged biological C t values of the reference genes were analyzed using BestKeeper software (Pfaffl et al., 2004) to obtain a ranking for control gene stability. Two reference genes (18S, RPL) and the Alien exogenous control gene were used to calculate the relative expression of the two target genes Hex-1 and Hex-2 according to the ABI Guide to Performing Relative Quan- titation of Gene Expression using Real-Time Quantitative PCR (Applied Biosystems, Foster City, CA) with the following modification: the geometric mean of the reference genes and Alien exogenous control gene was used to calculate the ∆C t values for the individual target genes (Vandersompele et al., 2002). The ∆C t values were normalized to the treatment that had the lowest overall expression to calculate the ΔΔC t . The values were then graphed using the sigmaPlot 11.0 software (systat software, Inc., san Jose, CA). statistical differences were calculated by analysis of variance (ANOVA) and student’s t-test using JMP® 8.0 (sAs Institute Inc., Cary, NC, usA). statistical dif- ferences were not calculated using the ΔΔC t values because normalization using the ΔΔC t leads to incorporation of values from one of the treatments and violates the ANOVA rules. To validate our findings and establish a set of stable control genes, the target gene Hex-1 was analyzed using the two most consistent (18S and RPL) reference genes, and the least consistent (Atube) reference gene independently for normalization. Results / Discussion The RT-qPCR technique provides a precise method for characterizing changes in gene transcription. Inclusion of an appropriate normalization strateg y is essential to the integrity of the data and its analysis. To establish a set of reliable control genes for Coptotermes formosanus we analyzed the expression pattern of several potential standards. 87 Tarver, M.R. et al. —screening of Multiple Control Genes in C. formosanus Gene stability and ranking We first compared the Ct values for each gene across all treatments to visualize the expression variability (Fig. 1). Data in Fig. 1 indicated that the 18S and Alien controls provided the most consistent C t values. Both the RPL and Gadph C t values varied slightly more, while the β-actin, α-actin, α-tubulin, Elong, Hexamerin-1, and Hexamerin-2 C t values varied widely from the various body parts. C t values were also entered into the BestKeeper analysis program and the results summarized in Table 2. This analysis provided a ranking of the expression of control genes based upon their standard deviation. The ranking based upon CV value was very similar with the Alien and 18S genes providing the most stable reference standard. The α-actin and Elong were not included in the BestKeeper analysis because of the increased number of reactions for which no product was detected (α-actin 26/48 rx, Elong 41/48 rx ). undetectable reactions resulted in a lack of measurable product, which caused excessive variation in C t value. However, none of the samples lacked RNA as evidenced by the expression levels of the other genes, indicating that transcript abundances of α-actin and Elong were below the limit of detection in these samples. The transcript abundances of valid reference genes should be present at sufficient levels to reliably generate detectable amplification products, and so α-actin and Elong may be invalid for use as reference genes due to their low expression levels. Target gene analysis (single versus multiple gene normalization) To validate a set of reference genes that can provide accurate normalization, we chose to normalize Hex-1 and Hex-2 expression profiles using 18S and Table 2. RT-qPCR C t values obtained from BestKeeper software RPL 18S BA1 Gadpdh Atube Alien n 48 48 48 48 48 48 geo Mean [C t ] 23.4 13.69 21.75 23.27 28.04 27.97 am Mean [C t ] 23.47 13.73 21.88 23.34 28.72 27.99 min [C t ] 21.3 11.17 17.93 20.77 19.27 26.45 max [C t ] 29.07 17.33 27.44 28.45 37.56 32.68 std dev [± C t ] 1.45 0.79 1.85 1.38 5.21 1 CV [% C t ] 6.19 5.77 8.47 5.9 18.13 3.59 88 sociobiolog y Vol. 59, No. 1, 2012 Fig. 1. Control and target gene Ct values across all treatments. 89 Tarver, M.R. et al. —screening of Multiple Control Genes in C. formosanus RPL, two internal control genes, and one exogenous control gene “Alien”. The results with this set of three reference standards indicate that both Hex-1 and Hex-2 were more highly expressed in workers compared to soldiers (Fig. 2 A,B). Hex-1 and 2 were also more expressed in the worker and soldier carcass compared to the head and gut. These findings are similar to results obtained with R. flavipes (Zhou et al. 2006c) that demonstrated up-regulation of Hex-1 and -2 in the termite fat body. Our results clearly demonstrate an up- regulation in the carcass which contained the fat body, therefore our results are consistent with the findings of Zhou et al. 2006c. To demonstrate the potential pitfalls associated with application of a single non-validated reference gene for RT-qPCR normalization, we varied the application of our set of reference genes in the normalization of one target gene, Hex-1. The expression levels of Hex-1 across soldier body regions were compared using individual control genes (18S, RPL, Alien, and Αtube). Our findings indicated a similar expression pattern between two consistent con- trol genes (18S, RPL), and demonstrated a higher Hex-1 expression in the carcass compared to the head and gut (Fig. 3A and B). Even between these two “stable” control genes there is a significant difference in Hex-1 expres- sion in the gut. We find that with the application of 18S alone as a reference, Hex-1 expression in the gut is relatively higher compared to normalization with RPL alone (Fig. 3A and B). Analysis of Hex-1 expression using the most variable control gene (Atube) significantly alters the results. Applying Atube as a normalization control, Hex-1 expression is lowest in the gut compared to the head and carcass (Fig. 3C). These results are dramatically different from the results using the two previous normalizers (18S, RPL) and clearly demonstrate that using only Atube as a control gene standard could result in misleading data analysis. This examination is consistent with similar types of analyses of multiple reference genes from other model systems and clearly supports the use of multiple standards (Vandesomepele et al. 2002). When performing gene expression analysis it is important for researchers to characterize and validate a set of control genes specific for the type of study being performed. Relatively few studies have been conducted on termite gene expression using RT-qPCR. The majority of the past experiments have used only one gene as a reference, and in most cases this was the β-actin gene (scharf et al. 90 sociobiolog y Vol. 59, No. 1, 2012 Fig. 2. Hexamerin 1 (A) and Hexamerin 2 (B) relative expression in worker and soldier termites in different body locations. 18S, RPL and Alien were used as control genes with the “soldier head” being the control treatment. Hex-2 relative expression in worker carcass (B) was too high to fit on to the graph (5663.5 ± 981 sE). Different letters represent significant differences between treatments (P<0.05). 91 Tarver, M.R. et al. —screening of Multiple Control Genes in C. formosanus Fig. 3. Relative Hexamerin 1 expression in soldier termites after application of different reference standards (A) 18S, (B) RPL, and (C) Atube. soldier head was used as the normalizing treatment. Different letters represent significant differences (P<0.05). 92 sociobiolog y Vol. 59, No. 1, 2012 2003, 2005a, 2005b, Wheeler et al. 2010). Zhou et al. (2006b) used only NADH-dh to compare different regions of the termite gut, while Korb et al. (2009) used only 18S to compare gene expression in workers with different social situations. Weil et al. (2009) also used only 18S and to compare gene expression in Cryptotermes cynocephalus and Cryptotermes scundus female neotenics. A number of researchers utilized the BestKeeper software to compare three control genes. Zhou et al. (2006a; 2006b) used the BestKeeper software to compare β-actin, HSP-70 and NADH-dh and determined that β-actin was the most stable when measuring hexamerin expression levels in worker termites. In another set of experiments Zhou et al. (2006b) used the BestKeeper software to compare β-actin, HSP-70 and NADH-dh and determined that NADH-dh was the most stable when comparing cytochrome p450 expression levels in termite workers and hexamerin body location. Weil et al. (2007) used the BestKeeper software to compare 18S, β-actin, and hexamerin and determined that, similar to our findings, 18S was the most stable when comparing castes in C. secundus. Hojo et al. (2009) also used the BestKeeper software and compared 18S, β-actin, Eelong, and NADH-dh when measuring gene expression to compare soldier and minor workers in a higher termite. This study demonstrated that β-actin was the most stable and compared the stability of each potential control gene with each other. Zhou et al. (2008) again used the BestKeeper software to compare β-actin, hsp-70 and NADH-dh, but this time normalized to all three control genes when documenting RNAi expression knockdown of hexamerin expression. To date, the most extensive control gene identification experiment was performed in a study by Tarver et al. (2010) that analyzed 49 different genes across multiple treatments, days and colonies in R flavipes. This study com- pared the standard deviations of the C t values across all genes and treatments and selected three genes with low variation to use as control genes (Stero-1, LIM and Mev-1). Taken together, these different RT-qPCR gene expression studies indicate a need for control genes capable of performing as reference standards inde- pendent of termite colony, body-region, and experimental condition. Our results suggest that 18S and RPL may serve as stable control genes when comparing the various samples that we investigated (caste, body location, 93 Tarver, M.R. et al. —screening of Multiple Control Genes in C. formosanus and colony) in our termite species, C. formosanus. Testing other treatments or species of termites may require a different set of control genes and a suite of control genes needs to be evaluated for variability. Predefined suites of control genes that can be applied across multiple species, time points, and treatments are highly desirable and would provide ease of use if it is possible to define them. Finally, the use of the Alien Reference RNA (stratagene) has a number of advantages. By “spiking” in an exogenous “alien RNA” before cDNA synthesis and then measuring Alien gene expression, a valid control is created to com- pensate for systematic error associated with pipetting and multiple reactions. Increased use of the Alien Reference RNA in future RT-qPCR studies will assist in the reduction of erroneous observations. Care must be taken not to rely solely on the Alien Reference RNA, but the addition of this reference with appropriate internal control genes should provide an additional reliable standard to more accurately quantify gene expression. Conclusion The RT-qPCR technique provides a precise method for characterizing gene transcription changes. Inclusion of an appropriate normalization strat- eg y is essential to the accuracy of the data and its analysis. While it is labor and reagent intensive, the use of several well characterized reference genes ensures that small variations in target genes are more likely to be detected and this translates into more accurate results. Here we have characterized six potential control genes and applied the most stable (18S, RPL, Alien) to measure target gene expression across body region (head, gut, and carcass) and between castes (worker and soldier). Our results demonstrate that 18S and RPL can serve as reliable expression standards when comparing these different castes and body regions, and we show that C. formosanus Hex-1 and Hex-2 have expression patterns similar to that previously described in R.. flavipes. Further use of these reference genes when comparing expression levels among various potential termite control treatments will allow more accurate assessment of effects on target gene expression. ACKNOWLEDGMENTs We would like to thank Drs. Maureen Wright and Michael santiago Cin- trón for manuscript review and helpful comments. 94 sociobiolog y Vol. 59, No. 1, 2012 REFERENCEs Hojo, M., K. Toga, I. Itai, & K. Maekawa 2009. Reference genes for real-time quantitative reverse transcriptase-PCR in the higher termite Nasutitermes takasagoensis (Isoptera: Termitidae) comparing soldiers with minor workers. sociobiolog y 54: 509-520. Korb, J., T. Weil, K. Hoffmann, K.R. Foster & M. Rehli 2009. A gene necessary for reproductive suppression in termites. science 324: 758. Pfaffl, M.W., A. Tichopád, C. Prgomet & T.P. Neuvians 2004. Determination of stable housekeeping genes, differentially regulated target genes and sample integrity: BestKeeper – Excel-based tool using pair-wise correlations. Biotechnolog y Letters 26:509-515. Lax, A.R., & W. LA. Osbrink 2003. united states Department of Agriculture – Agriculture Research service research on targeted management of the Formosan subterranean termite Coptotermes formosanus shiraki (Isoptera: Rhinotermitidae). Pest Management science 59:788-800. scharf, M.E, D. Wu-scharf, B.R. Pittendrigh & G.W. Bennett 2003. Caste- and development- associated gene expression in a lower termite. Genome Biolog y 4, R62. scharf, M.E., C.R. Ratliff D. Wu-scharf, X. Zhou & B.R. Pittendrigh 2005a. Effects of juvenile hormone III on Reticulitermes flavipes: changes in hemolymph protein composition and gene expression. Insect Biochemistry and Molecular Biolog y 35: 207-215. scharf, M.E., Wu-scharf D., Pittendrigh B.R. & G.W. Bennett 2005b. Gene expression profiles among immature and adult reproductive castes of the termite Reticulitermes flavipes. Insect Molecular Biolog y 14: 31-44. su, N., & R.H. scheffrahn 2000. Termites as pests of buildings. Termites: Evolution, Sociality, Symbiosis, Ecolog y (eds. T. Abe, D.E. Bigness & M. Higashi), pp.437-454. Dordrecht, Kluwer Acedemic Publishers. szalanski, A.L., Austin J.W. & C.B. Owens 2003. Identification of Reticulitermes spp. (Isoptera: Reticulitermatidae) from south central united states by PCR-RFLP. Journal of Economic Entomolog y 96: 1514-1519. Tarver, M.R., Zhou X., & M.E. scharf 2010. socio-environmental and endocrine influences on developmental and caste-regulatory gene expression in the eusocial termite Reticulitermes flavipes. BMC Molecular Biolog y 11: 28. Vandesomepele, J., DePreter K., Pattyn F., Poppe B., Van Roy N., DePaepe A., & F. speleman 2002. Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes. Genome Biolog y 3: 34. Weil, T., Korb J. & M. Rehli 2009. Comparison of queen-specific gene expression in related lower termite species. Molecular Biolog y and Evolution 26: 1841-1850. Weil, T., Rehli M. & J. Korb 2007. Molecular basis for the reproductive division of labour in a lower termite. BMC Genomics 8: 198.1-198.9. Wheeler, M.M., Tarver M.R., Coy M.R. & M.E. scharf 2010. Characterization of four esterase genes and esterase activity from the gut of the termite Reticulitermes flavipes. Archives of Insect Biochemistry and Physiolog y 73: 30-48. 95 Tarver, M.R. et al. —screening of Multiple Control Genes in C. formosanus Zhou, X., Oi F.M. & M.E. scharf 2006a. social exploitation of hexamerin: RNAi reveals a major caste-regulatory factor in termites. Proceedings of the National Academy of sciences 103: 4499-4504. Zhou, X., song C., Grzymala T.L., Oi F.M. & M.E. scharf 2006b. Juvenile hormone and colony conditions differentially influence cytochrome P450 gene expression in the termite Reticulitermes flavipes. Insect Molecular Biolog y 15: 749-761. Zhou, X., Tarver M.R., Bennett G.W., Oi, F.M. & M.E. scharf 2006c. Two hexamerin genes from the termite Reticulitermes flavipes: sequence, expression, and proposed functions in caste regulation. Gene 376: 47-58. Zhou, X., smith, J.A., Koehler, P.G., Oi, F.M., Bennett, G.W. & M.E. scharf 2007a. Correlation of cellulase gene expression and cellulolytic activity throughout the gut of the termite R. flavipes. Gene 395: 29-39. Zhou, X, Tarver M.R. & M.E. scharf 2007b. Hexamerin-based regulation of juvenile hormone-dependent gene expression underlies phenotypic plasticity in a social insect. Development 134: 601-610. Zhou, X, Wheeler M.M., Oi F.M. & M.E. scharf 2008. RNA interference in the termite Reticulitermes flavipes through ingestion of double-stranded RNA. Insect Biochemistry and Molecular Biolog y 38: 805-815.