981Vol. 10 | No. 3 | Summer 2013 |U R O LO G Y J O U R N A L 1Molecular Genetics Department, Faculty of Biological Science, Tarbiat Modares University, Tehran, Iran 2Urology and Nephrology Research Center, Labbafi-Nejad Medical Center, Shahid Beheshti University of Medical Science, Tehran, Iran 3Cellular and Molecular Research Center, Ahvaz Jundishapour Univer- sity of Medical Science, Ahvaz, Iran 4ParsGenome Company, Tehran, Iran 5Genetic Research Center, Baqiyat- allah University of Medical Sciences, Tehran, Iran Hamideh Monfared,1 Seyed Amir Mohsen Ziaee,2 Mahmoud Hashemitabar,3 Hamid Khayatzadeh,1 Vahid Kheyrollahi,1,4 Mahmood Tavallaei,5 Seyed Javad Mowla1 Co-Regulated Expression of TGF-β Variants and miR-21 In Bladder Cancer Corresponding Author: Seyed Javad Mowla, PhD Molecular Genetics Department, Faculty of Biological Science, Tarbiat Modares University, Tehran, Iran Tel: +98 21 82883464 Fax: +98 21 82884717 E-mail: sjmowla@modares.ac.ir Received May 2012 Accepted June 2013 Purpose: To investigate a potential alteration in the expression levels of transforming growth factor β (TGF-β) and miR-21 in bladder cancer tissues. Material and Methods: Using real-time polymerase chain reaction (PCR) method, we examined a potential correlated expression of miR-21 and TGF-β variants in 30 bladder tumors and their marginal/non-tumor biopsy specimens obtained from the same patients. Results: Our data revealed a significant down-regulation of TGF-β variants (P = .03) along with a non-significant alteration in the expression of miR-21 in tumor vs. non-tumor sam- ples. However, in contrast to low-grade tumors, the expression of miR-21 was upregulated in high-grade ones, and the expression level can efficiently discriminate low-grade tumors from high-grade ones (P = .03). Conclusion: In accordance to the observed similarity between TGF-β variants and miR-21 gene expression alterations in bladder tumors, treating 5637 bladder cancer cell line with TGF-β recombinant protein caused a significant upregulation of miR-21. The later finding further confirmed a correlated expression of TGF-β and miR-21 in bladder tumors. Keywords: biological markers; analysis; urinary bladder neoplasms; miR-21; TGF-β; gene expression profiling CELLULAR AND MOLECULAR UROLOGY 982 | INTRODUCTION Recently, several high-throughput studies have fo-cused on delineating genomic changes and gene expression alterations in bladder cancer, in hope to find novel markers correlated to different grades and stages of the tumors.(1) The recent studies have also documented a link between the expression of microRNAs (miRNAs or miRs) and cancer pathogenesis. MiRNAs are a class of non- coding RNA molecules with 19–24 nucleotides in length. They contribute to cancer initiation and progression, and are differentially expressed in normal and tumor tissues.(2) MiRNAs exert their effects by silencing the expression of their target mRNAs via a perfect and/or an imperfect com- plementary base-pairing, causing either mRNA degradation or translational inhibition.(3) Almost 50% of human miRNA genes are located in fragile chromosomal regions, which are frequently amplified, deleted, or translocated in various cancers.(4) Based on accumulating data, some miRNAs can function as either tumor suppressors or oncogenes.(5,6) Deregulations of miRNA expression have been already re- ported for several human cancers. For example, elevated expression of miR-21 has been implicated in the acquisition of invasive and metastatic properties of human cancer cells. MiR-21 functions as an oncomir, by targeting multiple tu- mor suppressor genes such as: PTEN, PDCD4, TPM1, and MASPIN.(7,8) Likewise, overexpression of miR-21 is linked to higher stages and lymph node metastasis in human breast cancer.(9) MiRNA's biogenesis begins with the transcription of its gene, followed with the maturation of the primary transcript (pri- miRNA) into a hair-pin structure (pre-miRNA), and then processing of a ~20-24 nucleotides mature form. Based on a recent report, miR-21 was rapidly induced by the growth fac- tors bone morphogenetic protein-4 (BMP-4) and transform- ing growth factor-β (TGF-β) in vascular smooth muscle cells. (10) These factors induced a rapid elevation of mature miR-21, by enhancing the processing of primary transcripts of miR- 21 (pri-miR-21) into its precursor form (pre-miR-21).(10) The later finding suggests that the autocrine TGF-β signaling can contribute to an altered expression of miR-21 in cancer cells. TGF-β, a family of multifunctional homodimeric poly- peptides, has three different isoforms in mammalian cells: TGF-β1, TGF-β2, and TGF-β3. The 25-kDa mature forms of all isoforms are structurally and functionally similar.(11) This family of cytokines can regulate cell growth, differentiation, inflammatory responses, apoptosis and extracellular matrix (ECM) production.(12) Because of its potent anti-proliferative effects on many cell types, such as endothelial and epithe- lial cells, TGF-β1 is regarded as a tumor suppressor gene.(11) However, an unexpected overexpression of TGF-β has been described in several tumor tissues, which demonstrated to be associated with the tumor progression and metastasis.(13-16) In the present study, we employed a quantitative real-time polymerase chain reaction (PCR) method to investigate whether the expression level of TGF-β variants is altered in bladder tumor tissues. We also measured the expression level of mature miR-21, a direct target of TGF-β, in tumor and non-tumor specimens of bladder. We then investigated the possibility of any correlation between TGF-β and miR-21 expression levels in bladder cancer tissues. MATERIAL AND METHODS Table 1. Clinicopathological characteristics of patients with bladder cancer Variables Value. Number of patients 30 Age groups, years 30-60 10 61-90 20 Mean age at diagnosis, year 63.8 Gender Male 28 Female 2 Number of specimens Tumor 30 Non-Tumor 30 Tumor stage at diagnosis PT a -Pt 1 26 PT 2 -PT 3 3 CIS 1 Tumor grade at diagnosis High grade 10 Low grade 20 Cellular and Molecular Urology 983Vol. 10 | No. 3 | Summer 2013 |U R O LO G Y J O U R N A L Patients and clinical samples Thirty pairs of human tumor and non-tumor (apparently normal tissue biopsy from the same patients) specimens of bladder were obtained from Labbafi-Nejad hospital in Teh- ran, Iran. The samples had been immediately snap-frozen in liquid nitrogen and had been stored in -70ºC, until being used for RNA extraction. For each patient, the clinicopathologi- cal information including: gender, age, grade, and stage of tumors were gathered (Table 1). The experimental procedure was approved by the ethical committee of Tarbiat Modares University. Cell culture The 5637 bladder cancer cell line was obtained from the na- tional cell bank of Iran (Pasteur institute of Iran, Tehran) and was cultured in RPMI-1640 (Gibco, USA) medium, supple- mented with penicillin/streptomycin (100 U/ml and 100 µg/ ml, respectively) and 10% fetal bovine serum (FBS), at 37°C in a humidified atmosphere of 5% CO2. After reaching to a confluency of 70-80%, cells were treated with 400 µM/ml of TGF-β recombinant protein (Peprotech, USA) for different time-points. The cells were then lysed for RNA extraction and performance of real-time PCR. RNA extraction and real-time PCR Total RNA was isolated from the homogenized cell culture or tissue specimens; using Trizol solution (Invitrogen, USA), according to the manufacturer’s instructions. The 260/280 absorbance ratio was determined by the Nano Drop ND-100 spectrometer (Nano Drop Technologies, Wilmington, DE). RNase-free DNase (TAKARA, Japan) treatment of the total RNA was performed to eliminate any potential contamina- tion with genomic DNA. cDNA synthesis was carried out by using two commercial kits: “Universal cDNA synthesis kit” (Exiqon, Denmark) for miR-21 cDNA synthesis, and “Pri- meScriptTM 1st strand cDNA Synthesis Kit” (TAKARA, Japan) for TGF-β cDNA synthesis. For miR-21 amplification, the real-time PCR reactions were performed using 8µL of diluted cDNA(20X) products, miR- 21 LNA™ primers (Exiqon, Denmark), and SYBR Premix Ex Taq II (Perfect Real Time) (TAKARA, Japan), accord- ing to the manufacturer’s protocol. The PCR reactions were conducted at 95˚C for 30 seconds, followed by 40 cycles of 95˚C for 10 seconds, and 60˚C for 1 minute, in an ABI 7500 real-time quantitative PCR system (Applied Biosystems, USA). U6 snRNA gene was used as a housekeeping internal control. All real-time PCR reactions were done in duplicates. To minimize the data variation in separate runs, paired tumor and non-tumor samples from the same patient were always examined on the same runs. To amplify TGF-β variants, the real-time PCR reactions were performed using 2µL of cDNA products, specific primers for each variant (table 2), SYBR Premix Ex Taq II (Perfect Real Time) (TAKARA, Japan), and the ABI 7500 real-time quan- titative PCR instrument. The specific primers for TGF-β vari- ants as well as GAPDH (as an internal control) were designed by Gene Runner software, version 3.5 (Hastings Software, New York, USA), and synthesized by TAG company (Co- penhagen A/S, Denmark), as high purified salt-free grade. To make sure about the uniqueness of all PCR products, all de- signed primers were blasted against human genome. Statistical analysis: Table 2. The sequences of the designed primers and the PCR product sizes for each primer pairs Gene Bank access number Forward primer sequence Reverse primer sequence Amplicon size TGF-β1 NM_000660 5´-TGGCGATACCTCAGCAAC-3´ 5´-ACCCGTTGATGTCCACTTG-3´ 181-bp TGF-β2 NM_001135599 (variant 1) NM_003238 (variant 2) 169 bp (variant 2) 5´-AGGAGCGACGAAGAGTACTAC-3´ 5´-ACTCTGCTTTCACCAAATTG-3´ 169 bp (variant 1) TGF-β3 NM_003239 5´-TGTCCATGTCACACCTTTCAG-3´ 5´-TGTGGTGATCCTTCTGCTTC-3´ 145 bp GAPDH NM_002046 5´-GTGAACCATGAGAAGTATGACAAC-3´ 5´-CATGAGTCCTTCCACGATACC-3´ 123 bp TGF-β Variants and miR-21 in Bladder Cancer | Monfared Et Al 984 | All real-time PCR data were analyzed with Representation- al State Transfer (REST) 2008 (Corbett research Pty Ltd, Australia) software, and Statistical Program for Social Sci- ences (SPSS) software version 13.0 (SPSS Inc., Chicago, USA). REST software was used to analyze and normalize the real-time PCR data. Kolmogorov–Smirnof normality test (KS-test) was used to examine the normal distribution of the samples. Statistical differences between tumor and non- tumor samples were determined by paired t-test (if KS-test was passed) or Mann-Whitney non-parametric test (if KS- test wasn’t passed). The probability of a statistically signifi- cant difference between the clinicopathological parameters and the gene of interest expression fold change was tested by Mann-Whitney non-parametric test. Receiver operating characteristics (ROC) curve was plotted to determine the suitability of the miR-21 expression level as a potential tu- mor marker for bladder cancer. All tests were performed as two-tailed, and a P-value of <.05 was considered as statisti- cally significant. Cellular and Molecular Urology Figure 1. (A) A representative acryl-amid gel electrophoresis re- sult demonstrating a single amplified product for each TGF-β vari- ant, as well as the internal house-keeping control gene, GAPDH. The sizes of all PCR products appeared to be identical to the ones expected by primer designing. Representative melt curve graphs of real-time PCR products of TGF-β variants and GAPDH (B), as well as miR-21 and the internal control U6 (C) confirmed the au- thenticity of amplified products. Figure 2. The real-time RT-PCR quantification of the expression levels of TGF-β1 (A), TGF-β2 (B), and TGF-β3 (C) variants is com- pared between tumor vs. non–tumor (total samples), and in high-grade vs. low-grade tumors of bladder. The expression of each variant is normalized to that of GAPDH (the internal control). Histograms show the mean values of TGF-β’s relative expression, with standard deviation as error bar. Note that the expression of all TGF-β variants is significantly down-regulated in tumor sam- ples, compared to their non-tumor counterparts. 985Vol. 10 | No. 3 | Summer 2013 |U R O LO G Y J O U R N A L RESULTS Overall, tissue specimens from 30 patients were collected (30 tumor and 30 non-tumors) and examined. To categorize tumor samples based on their stages and grades, the clinico- pathological characteristic of all patients were obtained and further confirmed by an expert urologist (Table1). Optimization of real-time PCR experiments After RNA extraction with TRIZOL reagent, to assure that equal amounts of RNA were used for each PCR, we used GAPDH and U6 as internal controls (reference genes), and normalized the relative TGF-β and miR-21 expression of each sample with its own reference expression. The specific- ity of all primers was determined by examining the melting curves and the sizes of PCR products (Figure1). Statistical analysis of primers efficiencies was done by LinReg PCR software. Direct sequencing of the PCR products was fur- ther confirmed the accuracy and authenticity of each PCR product. Altered expression of TGF-β variants and miR-21 in blad- der tumors Real-time PCR was performed as duplicates for all samples, under similar conditions. The obtained data revealed a signifi- cant down-regulation of TGF-β variants in tumor samples, in comparison to non-tumor ones (Figure 2). The relative (nor- malized to that of internal control, GAPDH) down-regulation for each variant of TGF-β in tumor vs. non-tumor samples was as follow: TGF-β1 (0.35 times, P = .029), TGF-β2 (0.13 times, P = .001), and TGF-β3 (0.17 times, P = .000). The down-regulation of TGF-β variants was more significant in low grade samples: TGF-β1 (0.33 times, P = .041), TGF-β2 (0.08 times, P = .006), and TGF-β3 (0.11 times, P = .000). However, the down-regulation of all variants was much less in high grade specimens, where there was no statistically sig- nificant difference between the expression of all variants in high-grade tumors and their non-tumor counterparts (Figure 2). We observed a similar down-regulation of miR-21 in total bladder tumors (0.33 times) as well as in low-grade tumors (0.17 times), compared to their non-tumor counterparts (Fig- ure 3A). interestingly, miR-21 expression was much higher in high-grade tumors, compared to their non-tumor counter- parts (2.2 times); nevertheless, the difference was not statisti- cally significant. But, a comparison between high-grade and low-grade tumor samples revealed that miR-21 is significant- Figure 3. (A) The relative expression of miR-21 in tumor vs. non- tumor tissue samples and high-grade vs. low-grade tumors, normalized to the expression level of U6, as the internal control. Note that the altered expression between tumor and non-tumor samples is not statistically significant. However, in contrast to low- grade tumors, miR-21 is upregulated in high-grade ones, in a way that it could significantly discriminate between high-grade and low-grade tumor samples (B). C and D) ROC curve analysis was performed to determine the sensitivity and specificity of miR-21 expression level to discriminate between tumor and non-tumor (C), as well as low-grade and high-grade states of the samples (D). Note that the area under curve (AUC) is much bigger in (D), sug- gesting that miR-21 could discriminate high-grade tumors from low-grade ones much better than tumor from non-tumor states of the samples. Figure 4. miR-21 expression level is significantly elevated after treatment of 5637 bladder cancer cell line with TGF-β protein, at different time points. TGF-β Variants and miR-21 in Bladder Cancer | Monfared Et Al 986 | Cellular and Molecular Urology ly upregulated (9.8 times, P = .031) in high grade specimens (Figure 3B). To determine the suitability of miR-21 expres- sion level in discriminating tumor and non-tumor samples of bladder, their ROC curve plots were created and analyzed. As shown in Figure 3C, the area under the curve (AUC) for miR- 21 expression level in tumor vs. non-tumor samples was 61% (Larger AUC value means better overall performance of the medical test to correctly discriminate two groups of samples). The finding suggests that miR-21 expression level is not a good tumor marker in discriminating tumors from non-tumor samples (P = .1548). In contrast, the AUC was much bigger (0.79) and statistically significant (P = .0251), when miR-21 expression level were used to discriminate high-grade from low-grade tumors (Fig 3D). However, despite this significant result, we couldn’t define a valid and applicable cut-off range to determine the specificity, sensitivity, and predictive value for expression level of miR-21 as a grading marker. Co-regulation of TGF-β and miR-21 expression The similar expression pattern of TGF-β and miR-21 sug- gested a potential co-expression regulation of the genes. To examine this possibility, we treatedthe 5637 bladder cancer cell line with recombinant TGF-β protein. The addition of TGF-β significantly elevated the expression of miR-21 in treated cells. As it is evident in figure 4, following a slight decline at the beginning of the experiment, miR-21 was rap- idly upregulated at 2, 3, and 4 hours after treatment of the cells. DISCUSSION In this study, we examined a potential alteration in the ex- pression of TGF-β and miR-21 in bladder tumors. Our data revealed that all variants of TGF-β were down-regulated in bladder tumors. The finding is in accordance with the com- mon knowledge about TGF-β function as a potent inhibitor of normal epithelial cell proliferation.(17) TGF-β is probably one of the regulatory factors that are genetically or epigenetically altered during bladder tumorigenesis.(11) An altered expres- sion of TGF-β, its receptors (TGF-βR1 and TGF-βR2), and its signaling components (the smad family) are already re- ported for several tumors.(13,14,18-26) Our finding is in conflict with a previous report,(16) which claimed a significant eleva- tion of serum TGF-β1 levels in patients with invasive bladder cancer. The inconsistency between the two reports could be due to examining different samples, as the same group failed to show similar finding in the urine of the same patients.(16) Moreover, the tumor cells are not the sole source of TGF-β1 release within the serum. Consistence with our finding, and in a follow-up research, the same group reported a down- regulation of TGF-β variants in bladder tumor tissues.(27) In this study, we discriminate the expression of each variant of TGF-β in our samples. The data revealed that the variants of the gene had a similar pattern of expression in tumor vs. non-tumor samples. The later finding implies that all variants of TGF-β are under the same regulatory elements inside the cells, despite being located on different chromosomes. The most significant down-regulation of TGF-β was observed in low-grade tumors. A logical interpretation of this observation is that the alteration of the gene is a vital step in tumor initia- tion, but not that critical in tumor progression. We also in- vestigated a potential alteration in the expression of miR-21, which its biogenesis is directly regulated by TGF-β. Despite previous reports on upregulation of miR-21 in most cancer types and cancer-derived cell lines,(28,29) the expression level of miR-21 failed to show any statistically significant dif- ference between tumor and non-tumor samples of bladder. However, while miR-21 showed a slight down-regulation in low-grade tumors, its expression was noticeably elevated in high-grade ones. Indeed, the expression level of miR- 21 could strongly discriminate high-grade from low-grade tumors. In contrast to the conclusion we made for TGF-β, miR-21's upregulation seems to be a vital step for tumor pro- gression, but not for tumor initiation. Our data demonstrated a significant correlation between expression levels of miR- 21 and TGF-β variants. In low-grade tumor samples which TGF-β expression was significantly down-regulated, miR-21 expression was also down-regulated. And in high grade sam- ples, where the TGF-β expression level was increased, the expression level of miR-21 was accordingly elevated. CONCLUSION Our data suggest a significant down-regulation of all TGF-β variants in bladder cancer samples, mostly in low-grade tu- mors. We also observed a non-significant down-regulation 987Vol. 10 | No. 3 | Summer 2013 |U R O LO G Y J O U R N A L TGF-β Variants and miR-21 in Bladder Cancer | Monfared Et Al of miR-21 in low-grade bladder tumors. In contrast, miR- 21 showed upregulation in high-grade bladder tumors, in a way that the expression level of miR-21 could discriminate low-grade tumors from high-grade ones. Similar pattern of gene expression for TGF-β variants and miR-21 suggests a co-regulated expression for these genes. ACKNOWLEDGMENTS This work was supported partially by a research grant from the Urology and Nephrology Research Center and the re- search deputy of Ahvaz Jundishapur University of Medical Science. CONFLICT OF INTEREST None declared. REFERENCES 1. Dyrskjot L, Zieger K, Orntoft TF. Recent advances in high-through- put molecular marker identification for superficial and invasive bladder cancers. Front Biosci. 2007;12:2063-73. 2. Croce CM, Calin GA. MiRNAs, cancer, and stem cell division. Cell. 2005;122:6-7. 3. Bartel DP. MicroRNAs: genomics, biogenesis, mechanism, and func- tion. Cell. 2004;116:281-297. 4. Calin GA, Sevignani C, Dumitru CD et al. Human microRNA genes are frequently located at fragile sites and genomic region sinvolved in cancers. Proc Natl Acad Sci USA. 2004;101:2999-3004. 5. Esquela-Kerscher A, Slack FJ. Oncomirs - microRNAs with a role in cancer. Nat Rev Cancer. 2006;6:259-69. 6. Baffa R, Fassan M, Volinia S et al. MicroRNA expression profiling of human metastatic cancers identifies cancer gene targets. J Patholo. 2009;219:214-221. 7. Zhu S, Wu H, Wu Fet al. MicroRNA-21 targets tumor suppressor genes in invasion and metastasis. Cell Res. 2008;18:350-359. 8. Asangani IA, Rasheed SAK, Nikolova DA,et al. MicroRNA-21 (miR- 21) post-transcriptionally downregulates tumor suppressor Pdcd4 and stimulates invasion, intravasion and metastasis in colorectal cancer. Oncogene. 2008;27:2128-2136. 9. Yan LX, Huang XF, Shao Q, et al. MicroRNA miR-21 overexpres- sion in human breast cancer is associated with advanced clinical stage, lymph node metastasis and patient poor prognosis. RNA. 2008;14:2348-2360. 10. Davis BN, Hilyard AC, Lagna G, Hata A. SMAD proteins control DRO- SHA-mediated microRNA maturation. Nature. 2008;454:56-61. 11. Massague J. TGF-β in Cancer. J Cell. 2008;134:215-230. 12. Keski-Oja1 J, Koli K, von Melchner H. TGF-β activation by traction? J Cell Biology. 2004;14:657-659. 13. Gorsch SM, Memoli VA, Stukel TA, et al. Immunohistochemical stain- ing for TGF-β1 associates with disease progression in human breast cancer. Cancer Res. 1992;52:6949-6952. 14. Thompson TC, Truong LD, Timme TL, et al. Transforming growth factor β1as a biomarker for prostate cancer. J Cell Biochem. 1992; 16H:54-61 15. Weidner N, Semple JP, Welch WR, Folkman J. Tumor angiogenesis and metastasis - correlation in invasive breast carcinoma. N Engl J Med. 1991;324:1-8. 16. Eder IE, Stenzl A, Hobisch A, et al. Transforming growth factors-beta 1 and beta 2 in serum and urine from patients with bladder carci- noma. J Urol. 1996;156:953-7. 17. Huang SS, Huang JS. TGF-beta control of cell proliferation. J Cell Bio- chem. 2005; 96(3):447-62. 18. Coombes RC, Barrett-Lee P, Luqmani Y. Growth factor expression in breast tissue. J Steroid Biochem Mol Biol. 1990;37:833-6. 19. Jonson T, Albrechtsson E, Axelson J, et al. Altered expression of TGFB receptors and mitogenic effects of TGFB in pancreatic carci- nomas. Int J Oncol. 2001;19:71-81. 20. Ryu B, Kern SE. The essential similarity of TGF beta and activin re- ceptor transcriptional responses in cancer cells. Cancer Biol Ther. 2003;2:164-70. 21. Johnson MD, Shaw AK, O'Connell MJ et al. Analysis of transforming growth factor β receptor expression and signaling in higher grade meningiomas. J Neurooncol. 2011;103:277-85. 22. Kim SH, Lee SH, Choi YL et al. Extensive alteration in the expres- sion profiles of TGFB pathway signaling components and TP53 is observed along the gastric dysplasia-carcinoma sequence. Histol Histopathol. 2008; 23:1439-52. 23. BertucciF, Finetti P, Cervera N et al. How different are luminal A and basal breast cancers? Int J Cancer. 2009;124:1338-48. 24. Wang JC, Su CC, Xu JB et al. Novel microdeletion in the transform- ing growth factor beta type II receptor gene is associated with gi- ant and large cell variants of non-small cell lung carcinoma. Genes Chromosomes Cancer. 2007;46:192-201. 25. Guzinska-Ustymowicz K, Kemona A. Transforming growth factor beta can be a parameter of aggressiveness of pT1 colorectal cancer. World J Gastroenterol. 2005;11:1193-5. 26. Barlow J, Yandell D, Weaver D et al. Higher stromal expression of transforming growth factor-beta type II receptors is associated with poorer prognosis breast tumors. Breast Cancer Res Treat. 2003;79:149-59. 27. Eder IE, Stenzl A, Hobisch A et al. Expression of transforming growth factors beta-1, beta 2 and beta 3 in human bladder carcinomas. Br J Cancer. 1997;75:1753-60. 28. Selcuklu SD, Donoghue MT, Spillane C. miR-21 as a key regulator of oncogenic processes. Biochem Soc Trans. 2009;37:918-25. 29. Pan X, Wang ZX, Wang R. MicroRNA-21: a novel therapeutic target in human cancer. Cancer Biol Ther. 2011;10:1224-32.