Cross-talk between miR-29c and transforming growth factor-β3 is mediated by an epigenetic mechanism in leiomyoma.

Using paired leiomyoma and matched myometrium (N=42) the co-expression of miR-29c and TGF-β3 in the same specimens was determined. The analysis indicated that miR-29c ( Fig. 1A ) expression is reduced while TGF-β3 expression at both mRNA ( Fig. 1A ) and protein ( Fig. 1B and 1C ) levels is elevated in leiomyomas as compared to paired myometrium. The analysis further indicated no significant racial/ethnic differences in expression of miR-29c and TGF-β3 ( Fig. 1D and 1E ). Collectively, this data confirmed previous results from our laboratory and others with regard to miR-29c and TGF-β3 expression in leiomyomas demonstrating an inverse pattern of expression ( 9 , 21 , 23 ). Since TGF-β3 is a predicted target of miR-29c based on Targetscan, we assessed such regulatory function of miR-29c on TGF-β3 expression in isolated LSMC. As shown in Fig. 2A and 2B , luciferase reporter assay revealed that miR-29c directly interacts with the 3’ UTR of TGF-β3 thereby regulating its expression in LSMC. Transfection of LSMC with pre-miR-29c downregulated the expression of TGF-β3, whereas knockdown of miR-29c with anti-miR-29c, which mimics the expression pattern of miR-29c in leiomyoma, increased the expression of TGF-β3 at mRNA ( Fig. 2C ) and protein ( Fig. 2D and 2E ) levels. We also demonstrated that treatment of LSMC with TGF-β3 (5 ng/ml) resulted in inhibition of miR-29c expression ( Fig 3A ). Since miR-29c expression in LSMC is under epigenetic control ( 15 , 37 , 39 ), we determined whether TGF-β3 mediated inhibition of miR-29c expression is mediated through an epigenetic mechanism. As such we determined the effect of TGF-β3 on epigenetic enzymes namely, DNMT1 and DNMT3A which catalyze DNA methylation and EZH2 which is involved in histone methylation. The analysis indicated that treatment of LSMC with TGF-β3 had no significant effect on DNMT3A and EZH2 expression (data not shown); however, TGF-β3 induced the expression of DNMT1 protein levels ( Fig. 3B and 3C ). To further establish the role of DNMT1 in mediating the effects of TGF-β3 on miR-29c expression, LSMC were treated with DNMT1 siRNA which effectively inhibited DNMT1 ( Fig. 3D and 3E ). Inhibition of DNMT1 in LSMC resulted in suppression of TGF-β3 at both mRNA and protein levels ( Fig. 3D , 3E and 3F ), while it increased the expression of miR-29c ( Fig. 3G ). Furthermore, treatment of LSMC with DNMT1 siRNA attenuated the inhibitory effect of TGF-β3 on miR-29c expression indicating that the effect of TGF-β3 on miR-29c is mediated by DNMT1 ( Fig. 3G ). To provide additional support that TGF-β3-induced inhibition of miR-29c is epigenetically mediated, LSMC were treated with TGF-β3 and the methylation status of miR-29c promoter was determined by methylation specific PCR. As shown in Fig. 4A and 4B , TGF-β3 increased the methylation level of miR-29c promoter, thus highlighting the importance of an epigenetic mechanism for TGF-β3-induced inhibition of miR-29c.

Leiomyomas and paired myometrium were obtained from patients (N=42) not on hormonal treatments for at least 3 months prior to surgery at Harbor-UCLA Medical Center with prior approval obtained from LA BioMed at Harbor-UCLA Medical Center Institutional Review Board (#036247). Informed consent was obtained from all the patients participating in the study before surgery. The paired tissues were obtained from white Hispanics (n=14), African Americans (n=14) and Caucasians (n=14) aged 35–56 years (mean 45 ± 5.1 years). Among the Caucasians group 6 pairs were kindly provided by Dr. Al-Hendy (University of Chicago). The tissues were either snap frozen and stored in liquid nitrogen for further analysis, or used for isolation of LSMC as previously described ( 29 ). Briefly, LSMC were cultured in DMEM supplemented with 10% fetal bovine serum until reaching confluence with a change of media every 2–3 days. Cells at passages p1 to p3 were used for all experiments. Cell culture experiments were performed at least three times using LSMC obtained from different patients. All supplies for isolation and cell culture were purchased from Sigma-Aldrich (St. Louis, MO), Invitrogen (Carlsbad, CA) and Fisher Scientific (Atlanta, GA). LSMC were seeded in 6-well plates and at sub-confluence transfected with 50 nM of pre-miR-29c, anti-miR-29c, pre-miR negative (NC) or anti-miR negative control (aNC) (Applied Biosystems, Foster city, CA) for 72 to 96 hours using PureFection transfection reagent (System Biosciences, Inc., Mountain View, CA) according to the manufacturer’s protocol. LSMC were cultured as above and at sub-confluence transfected with 50 nM of siRNA negative control (siNC) or siRNA against DNMT1 (siDNMT1; Santa Cruz Biotechnology, Dallas, Texas) for 72–96 hours using PureFection transfection reagent (System Biosciences, Inc., Mountain View, CA) according to the manufacturer’s protocol. Total RNA was extracted from LSMC using Trizol (Thermo Fisher Scientific, Waltham, MA) and their quantity and quality was determined (ND-1000 Spectrophotometer, NanoDrop Technologies, Wilmington, DE) as previously described ( 30 – 32 ). Subsequently, RNA sample of 1 μg each was reverse transcribed using random primers for TGF-β3. MiR-29c primer design and PCR conditions have been described previously ( 33 ). Quantitative RT-PCR was carried out using SYBR gene expression master mixes (Applied Biosystems, Carlsbad, CA). Reactions were incubated for 10 min at 95°C followed by 40 cycles for 15 seconds at 95°C and 1 min at 60°C. Levels of mRNA and miRNA were quantified using the Invitrogen StepOne System and normalized to FBXW2 ( 34 ) and RNU6B, respectively. All reactions were run in triplicate and relative expression was determined using the comparative cycle threshold method (2–ΔΔCT), as recommended by the supplier (Applied Biosystems). Abundance values were expressed as fold changes compared to the corresponding control group. The primer sequences used were as follows: TGF-β3 (sense, 5’-CGGGCTTTGGACACCAATTA-3’; antisense, 5’-GGGCGCACACAGCAGTTC-3’); and FBXW2 (sense, 5’-CCTCGTCTCTAAACAGTGGAATAA-3’; antisense, 5’-GCGTCCTGAACAGAATCATCTA-3’). miR-29c (sense, 5’-GCAGTAGCACCATTTGAAATC-3’; antisense, 5’-GGTCCAGTTTTTTTTTTTTTTTAACC-3’); and RNU6B (sense, 5’-ATTGGAACGATACAGAGAAGATTAG-3’; antisense, 5’-AATATGGAACGCTTCACGAAT-3’). Total protein isolated from paired tissue samples and LSMC following treatment conditions was subjected to immunoblotting as previously described ( 10 , 35 ). Briefly, samples were suspended in RIPA buffer containing 1 mM EDTA and EGTA (Boston BioProducts, Ashland, MA) supplemented with 1 mM PMSF and a complete protease inhibitor mixture (Roche Diagnostics, Indianapolis, IN), sonicated, and centrifuged at 4°C for 10 min at 14,000 rpm. The concentration of protein was determined using the BCA™ Protein Assay Kit (Thermo Scientific Pierce, Rockford, IL). Equal aliquots (Fifty micrograms) of total protein for each sample were denatured with SDS-PAGE sample buffer, and separated by electrophoresis on an SDS polyacrylamide gel. After transferring the samples to a nitrocellulose membrane, the membrane was blocked with TBS-Tween + 5% milk, and probed with the following primary antibodies: TGF-β3 and DNMT1 (Santa Cruz Biotechnology). The membranes were washed with TBS containing 0.1%Tween-20 wash buffer after each antibody incubation cycle. SuperSignal West Pico Chemiluminescent Substrate™ (Thermo Scientific Pierce, Rockford, IL) was used for detection, and photographic emulsion was used to identify the protein bands, which were subsequently quantified by densitometry. The membranes were also stripped and probed with glyceraldehyde 3-phosphate dehydrogenase (GAPDH) antibody (Santa Cruz Biotechnology) serving as the loading control. The densities of the specific protein bands were quantified with a scanning densitometer (Bio-Rad GS-800, Hercules, CA), and the results were expressed as means ± SEM normalized to GAPDH. LSMC were seeded in six-well plates and at 70–80% sub-confluence were transfected with 50 nM pre-miR-29c oligonucleotides or a negative control (NC) using the PureFection transfection reagent as previously described ( 36 ). At the same time, the cells were co-transfected with a luciferase reporter plasmid (1 μg/well) containing a 3’ untranslated region (UTR) sequence of TGF-β3 (GeneCopoeia, Rockville, MD). After 48 hrs of transfection Firefly and Renilla luciferase activities were measured using the Dual-Luciferase Reporter Assay System (Promega, Madison, WI). Firefly luciferase activity was normalized to Renilla luciferase activity. The level of induction was reported as mean ± SEM of three experiments performed in triplicates and compared with a ratio in cells transfected with negative control independently set as 1. LSMC at sub-confluence were treated with TGF-β3 (5 ng/ml) for 48 hours and genomic DNA was extracted and used for bisulfite conversion (EZ DNA Methylation Kit, Zymo Research Corporation, Irvine, CA) as previously described ( 37 ). Since methylated cytosine residues remained unchanged after bisulfite conversion this approach allows distinguishing DNA sequences which are methylated or un-methylated in specific genomic regions using sequence-specific PCR primers ( 38 ). The pair of primers used to detect the methylated (Me) sequence of miR-29c promoter were 5’-GATGGGATTAAATTTTGGAATATTC-3’ and 5’-AAATCCTAAAACCCGTCGAA-3’ and for the un-methylated (Un) sequence of miR-29c promoter were 5’-GGGATTAAATTTTGGAATATTTGG-3’ and 5’-ATCAAATCCTAAAACCCATCAAA-3’. DNA amplification was carried out following manufacturer’s protocol (HotStarTaq Plus PCR reagent, Qiagen, Valencia, CA). The PCR condition was 5 min at 95°C, followed by 40 cycles of amplification at 94°C for 30 second s, 30 seconds at 55°C, and 1 min at 72°C. PCR amplified products were electrophoresed on 2% agarose gel and visualized under ultraviolet illumination. Power calculation indicated that the sample size estimates for studies have been based on a power of 80% to detect 30% changes between the two groups (Myometrium vs Leiomyoma) (assuming an expected standard deviation of 20% of mean values). This analysis results in a requirement for 10 patients in each group. Throughout the text, all data are presented as mean ± SEM and analyzed by PRISM software (Graph-Pad, San Diego, CA). Dataset normality was determined by the Kolmogorov-Smirnov test. Comparisons involving two groups were analyzed using paired or un-paired Student’s t-tests as appropriate. One-way ANOVA was used for comparisons involving multiple groups. Statistical significance was established at P<0.05.

In the present study we provide evidence for a novel feed forward crosstalk between miR-29c and TGF-β3. We demonstrate an inverse pattern of expression between miR-29c and TGF-β3 in fibroids which is race independent. We confirm that TGF-β3 is a direct target of miR-29c in LMSC and knockdown of miR-29c results in overexpression of TGF-β3 whereas its overexpression has the opposite effect in LMSC. TGF-β3 in turn influences miR-29c expression through an epigenetic mechanism by increasing the expression of DNMT1 resulting in increased methylation of miR-29c promoter. Blockade of DNMT1 by siRNA transfection of LMSC blocked the inhibitory effect of TGFβ3 on miR-29c expression, thus establishing the importance of this epigenetic mechanism. A schematic demonstrating this cross talk between miR-29c and TGF-β3 is shown in Fig. 4C . Leiomyomas are considered a fibrotic disorder characterized by increased cell proliferation, inflammation, angiogenesis and excess accumulation of ECM where elevated expression of TGF-β family plays a central role ( 2 , 3 , 6 – 8 ). Both driver mutations [MED12 (mediator complex subunit 12), HMGA2 (high mobility group AT-hook 2), FH (fumarate hydratase), COL4A5 (collagen type IV, alpha 5)] and chromosomal alteration in addition to widespread alteration in DNA methylation in multiple genes critical to fibroid growth have been reported ( 40 , 41 ). The miR-29 family consists of miR-29a, miR-29b, and miR-29c which are encoded by two gene clusters, and are predicted to target many genes functionally associated with cellular transformation, inflammatory responses and ECM turnover ( 30 , 42 – 47 ). It is clear that a low level of anti-fibrotic miR-29c and elevated level of pro-fibrotic TGF-β3 in leiomyoma provides an environment necessary to promote and maintain tumor fibrotic characteristic. MiR-29c directly interacted with the 3’UTR of TGF-β3 regulating TGF-β3 at both transcriptional and translational levels which we further confirmed using gain-or loss-of-function of miR-29c. Although our result is the first to confirm the regulatory function of miR-29c on TGF-β3 expression in leiomyoma, in a recent study using unilateral ureteral obstruction (UUO) model for renal fibrosis, intramuscular injection of exosome-encapsulated miR-29 was shown to target TGF-β3 ( 48 ), thus providing support for our observation. Furthermore, we found that TGF-β3 through an epigenetic mechanism involving DNMT1 expression inhibited miR-29c expression. The regulatory action of TGF-β3-induced DNMT1 occurred by increased methylation of miR-29c promoter, thus further support the importance of epigenetic regulatory mechanism in leiomyoma ( 49 ). A similar crosstalk involving miR-29b and TGF-β1 and TGF-β2 (transforming growth factor-β2) has been previously reported in trabecular meshwork cells ( 50 ), in renal tubular cells where miR-29b was downregulated by TGF-β1 via Smad3 ( 51 , 52 ), and in cardiac fibroblast cells where TGF-β1 inhibited the expression of miR-29 family ( 46 ). However, to our knowledge our data is the first evidence demonstrating a feedback interaction between miR-29c and TGF-β3. Our data indicated that TGF-β3 induced DNMT1 expression in LSMC which in turn resulted in methylation of miR-29c promoter and lowering of miR-29c. Interestingly, although DNMT1 is not a direct target of miR-29 recent studies have demonstrated that miR-29b by targeting the expression of transcription factor SP1 regulates DNMT1 expression in acute myeloid leukemia cell lines ( 53 , 54 ). In addition, our laboratory and others have reported that leiomyoma express significantly higher DNMT1 as compared to matched myometrium ( 15 , 55 , 56 ). In line with our findings showing a significant decrease in the expression of TGF-β3 after DNMT1 knockdown in LSMC, zebrafish with a mutation in DNMT1 were reported to have 50% decrease in expression of TGF-β3 in the lens ( 57 ). Several studies in other tissues and cells have demonstrated an effect of the TGF-β1 on DNA methylation, including global changes in DNA methylation of genes critical to epithelial to mesenchymal transition in ovarian cancer cell line through induction of DNMT1 and DNMT3A (DNA methyltransferase 3A) and DNMT3B (DNA methyltransferase 3B) ( 58 ). In cardiac fibroblast TGF-β1 reduced the expression of DNMT1 and DNMT3A and induced COL1A1 expression ( 59 ). Collectively, these studies indicate the tissue/cell specific effect of TGFβ isoforms on DNA methylation. Our data presented here further support the regulatory function of miRNAs and epigenetic modification of genes that drive tumorigenesis and tissue fibrosis ( 60 – 63 ), and leiomyoma growth and progression ( 56 , 64 – 67 ). miRNAs have been implicated in the pathogenesis of a number of gynecologic and obstetrics conditions including endometriosis ( 68 ), polycystic ovary syndrome ( 69 ), fetal growth restriction ( 70 ) and pre-eclampsia ( 71 ), and as such have been proposed to serve as biomarkers or targets of therapeutics. Our profiling data indicating the inverse pattern of expression between miR-29c and TGF-β3 expression with anti-fibrotic and pro-fibrotic functions respectively, implicate their importance in leiomyoma pathogenesis. Additionally, the regulatory interaction between miR-29c and TGF-β3 provides a feed forward mechanism which could explain leiomyoma growth through ECM accumulation and an ideal therapeutic target to interrupt fibroid growth. This aim could be achieved by overexpression of miR-29c or inhibition of TGF-β3 to disrupt the fibrotic crosstalk between TGF-β3 and miR-29c. The strength of our study is the in vitro demonstration of the link between TGF-β3 known to play a key role in fibrosis associated with leiomyoma and the anti-fibrotic miRNA miR-29c and the limitation of the study is that it is an in vitro study and unknown if it applies to the in vivo setting. In summary, our data established a novel interaction between miR-29c and TGF-β3 in leiomyoma cells. In this scheme ( Fig. 4C ) TGF-β3 is a direct target of miR-29c. Elevated levels of TGF-β3 in fibroids induce the expression of DNMT1 which in turn increases the methylation of miR-29c promoter and its reduced expression. This feed forward axis would result in increased accumulation of the ECM and progression of fibroid growth.

Leiomyomas are benign fibrotic tumors afflicting a significant number of women being the most common indication for all hysterectomies performed ( 1 ). These tumors whose growth is dependent on ovarian steroids are characterized by excess accumulation of extracellular matrix (ECM), inflammation, and increased angiogenesis ( 2 , 3 ). Additionally, in leiomyomas altered expression of a number of non-protein coding genes, including microRNA (miRNAs), which target the expression of protein coding genes, has been documented ( 4 , 5 ). The pathogenic mechanisms for leiomyoma initiation and progression have been under intense investigation ( 6 – 8 ). Our laboratory has focused on the mechanism underlying ECM accumulation and we have identified miR-29c, a member of small non-coding RNAs as being pivotal to the fibrosis associated with these tumors ( 9 ). The levels of miR-29c are suppressed in fibroids compared with myometrium and are under ovarian steroid control, transcription factors SP1 (specificity protein 1), NF-κB (nuclear factor kappa-light-chain-enhancer of activated B cells) and epigenetic regulation ( 9 ). miR-29c has a number of targets including collagens ( 9 – 12 ), elastin (ELN) ( 10 , 11 , 13 ) and matrix metalloproteinases (MMPs) ( 11 , 14 ) many of which have been demonstrated to play a role in fibroid pathogenesis ( 2 ), and previously published reports demonstrated an increased expression of miR-29c targets in leiomyomas including collagen I (COL-1), collagen III (COL-3), MMPs and epigenetic enzymes DNMTs (DNA methyltransferases) ( 9 , 15 , 16 ). Other groups have reported other members of the miR-29 family namely miR-29a and miR-29b to also be lower in fibroids compared with myometrium ( 16 , 17 ). Furthermore, overexpression of miR-29b in an animal model for fibroid resulted in shrinkage of tumors ( 17 ). Collectively, these studies point to the highly significant role of miR-29 family in fibroid pathogenesis. TGF-β3 (transforming growth factor-β3) is a pro-fibrotic cytokine which interacts with type 1 and 2 receptors propagating its effect through activation of signaling pathways such as MAPK/ERK (mitogen activated protein kinase/extracellular-signal-regulated kinase) ( 18 ), Smad2/3 (SMAD family member 2/3) ( 19 ) and PI3K/Akt (phosphatidylinositol 3-kinases/protein kinase B) ( 20 ), the end result of which is increased synthesis of collagen I, CTGF (connective tissue growth factor), fibronectin (FN1) and versican V0 (VCAN V0) all of which are critical to leiomyoma progression ( 18 – 22 ). Furthermore, TGF-β3 inhibits the expression of MMPs thus resulting in ECM accumulation ( 22 ). Treatment of leiomyoma smooth muscle cell (LSMC) with TGF-β3 but not TGF-β1 (transforming growth factor-β1) stimulated DNA synthesis ( 23 ). Others have shown an increase in cell proliferation in response to TGF-β1 ( 24 ). A number of factors have been shown to reduce the levels of TGF-β3 and fibrotic genes regulated by TGF-β3 in in vitro studies, including tranilast ( 15 ), vitamin D3 ( 18 ) and selective progesterone receptor modulators ( 25 – 27 ), whereas activation of β-catenin (Catenin beta-1) which is critical in fibroid pathogenesis induces TGF-β3 ( 28 ). The ovarian steroid dependence of TGF-β family and their influence on smooth muscle cell proliferation underscore the significance of TGF-β and its receptors in fibroid pathophysiology. Since TGF-β3 is a predicted target of miR-29c (Targetscan, http://www.targetscan.org ) and both miR-29c and TGF-β3 are critical in regulation of fibrosis which is the hallmark of fibroids, we hypothesized that miR-29c might exert its effects on the ECM through modulation of TGF-β3 levels, and TGFβ3 in turn through a feedback loop could influence miR-29c transcription. This hypothesis was tested in an in vitro primary cell culture obtained from fibroids removed at the time of hysterectomy from women on no hormonal medications three months prior to surgery.