Introduction
Polycystic ovary syndrome (PCOS), a multi-factorial endocrine condition, affects up to 5-10% of women in reproductive age and accounts for 75% of anovulatory infertility [
1‐
3]. PCOS is a heterogeneous syndrome with complex pathologies with follicle growth arrest at the small antral stage, minimal granulosa cell proliferation, hyperthecosis and hyperandrogenemia, and chronic anovulation [
1]. It is associated not only with infertility, but also with increased risk of metabolic disorders, such as insulin resistance, diabetes, obesity, and cardiovascular diseases [
4]. Both genetic and environmental factors are known to contribute to the pathogenesis of PCOS; however, its etiology is unclear.
Differential gene expression is evidenced in human PCOS ovaries when compared to the normal ones [
5]. Genomic studies in PCOS have revealed a variety of altered genes that fall into many functional categories, including cell division, apoptosis and regulation of gene expression and metabolism. While genes involved in steroidogenesis, including those related to retinoic acid biosynthesis and luteinizing hormone-responsive gene, are generally up-regulated in theca cells of PCOS subjects [
6], those associated with Wnt signalling and ovarian folliculogenesis appear to be down-regulated [
7,
8]. Although ovarian apoptotic genes are altered in PCOS, the net effect of these changes is unclear [
7]. In addition, significant differences in mRNA abundance in oocytes from PCOS patients have also been reported and cluster analysis of 374 genes indicated an association with chromosomal alignment and segregation in mitosis and/or meiosis [
9].
Despite ample evidences on gene dysregulation in PCOS ovary as reported above, precisely how these genes are transcriptional and post-transcriptional regulated is poorly understood. New advances in post-transcriptional gene regulation have resulted from the discovery of hundreds of miRNAs in different mammalian species. Diverse expression pattern of miRNAs and the high abundance of potential targets in the ovary have led to the notion of their importance in the regulation of ovarian function. A large number of miRNAs has now been identified in the ovary [
10,
11] and found to be regulated by gonadotropins during follicular development [
12,
13]. Recent studies indicate that miRNAs are involved in the regulation of steriodogenesis, cell proliferation and apoptosis in human granulosa cells [
14,
15]. Differential expression of miR-23a, miR-23b,miR-542–3p, miR-211, and miR-17–5p in granulosa/cumulus cells from women undergoing assisted reproduction suggests aberrant miRNA expression may be an underlying etiology in female infertility [
16,
17]. Furthermore, miRNAs have been implicated to play a role in obesity and metabolic syndrome, including insulin resistance [
18], conditions often associated with PCOS. However, whether miRNAs contribute to the follicle arrest, dysregulated steroid production and insulin resistance in PCOS is not known. Understanding how miRNAs are regulated in the ovary and the identification of their specific targets and functions may offer novel insights into the etiology of PCOS and the development of target-specific gene regulation for its treatment.
Hyperandrogenism is one key feature in human PCOS patients and circulating levels of androgens, such as testosterone, androstenedione and dihydrotestosterone (DHT), are elevated in these patients [
19,
20]. A number of animal models have been developed that allow one to focus on different aspects of PCOS pathology, including ovarian follicular and metabolic features [
21‐
23]. A recent report on a rat PCOS model involving chronic DHT implant mimics the hyperandrogenic status and exhibits both polycystic ovaries and metabolic abnormalities similar to those in human PCOS [
23], allowing one to examine not only ovarian characteristics but also metabolic features of PCOS.
We hypothesize that PCOS is associated with dysregulation of ovarian miRNAs which may be involved in post-transcriptional regulation of genes involved in the ovarian pathology. Taking advantage of DHT-induced rat PCOS model, we tested this hypothesis by comparing the expression of 349 miRNAs between DHT-treated vs untreated rat ovaries. Moreover, selected candidate miRNAs were investigated for their localization in follicular cells. Results of the present study revealed that PCOS is associated with differential expression of regulatory non-coding miRNAs in rat ovaries.
Materials and methods
DHT-filled silastic capsules preparation
The DHT-filled capsules were prepared as previously described [
24]. Silastic tubing (I.D. 1.98 mm x O.D. 3.18 mm; Dow Corning, Cat. No. 2415577, Midland, MI), cut to an appropriate length to achieve a desired surface area of 299.70 mm
2, was filled with DHT (Sigma) from 1 ml syringe. The tubing was closed at each end (3 mm) with a sealant (Silicone Type A, Dow Corning Cat. No. 891, Midland, MI), ensuring that adhesive is in contact with tubing walls and that there are no air bubbles. Control capsules were empty with only sealant on both ends. After overnight dry, capsules were rinsed for 2 days in 3% bovine serum albumin (BSA) in phosphate buffer saline (PBS) with 0.1% NaN
3 solution, thoroughly washed with PBS and sterilized by dipping briefly in 70% ethanol before use.
Animal preparation
Sprague Dawley rats (Charles River, Montreal, Canada) were maintained on 12-h light, 12-h dark cycles and given food and water ad libitum. All procedures were carried out in accordance with the Guidelines for the Care and Use of Laboratory Animals, Canadian Council on Animal Care, and were approved by University of Ottawa and Ottawa Hospital Research Institute Animal Care Committee.
Animal surgery, DHT capsule implantation and animal care
Thirty-six female rats at 21 days of age were randomly divided into two groups (control vs. DHT group) and were implanted with a DHT-filled silicone capsule continuous-releasing (83 μg per day; empty capsule for controls) for 12 weeks to mimic the hyperandrogenic state in women with PCOS, whose plasma DHT levels are approximately 1.7-fold higher than those of healthy control [
23]. Animals were weighed weekly to monitor body weight gain and euthanized at 12 wks post-implantation. Ovaries were collected for analyses.
Determination of estrous cycle phases of rat
The estrous cycle of rats was monitored by microscopic analysis of vaginal smear obtained every morning for 12 days from 10–11 wks post-implantation. Predominant cells associated with each phase of the estrus cycle were: epithelial cells (proestrous; P), enucleated cornified cells (estrous; E), equal mix of leukocytes and cornified epithelial cells (metestrous; M) and leukocytes (diestrous; D) [
25].
Determination of insulin sensitivity
Insulin sensitivity test (IST) was performed after 12 h overnight fasting at 10 wks of surgery. Human insulin (0.2 U/100 g body weight intravenously, Novo Nordisk Canada Inc. Mississauga, ON, Canada) was administered to control or DHT-treated rats through tail vein and blood samples were obtained from the tail at 0, 5, 10, 20, 40, 80, 160 min. after insulin injection. Glucose levels were determined by glucose test strips (ACCU-CHEK, Roche). The insulin sensitivity index
K
ITT
was calculated from the slope of a linear portion of the regression line of natural logarithm of glucose versus time. Raw data were transformed by Log
K
ITT
for analysis.
K
ITT
= (0.693/t
1/2) × 100, t
1/2 represents the half-life of glucose decay after insulin injection [
26].
Ovarian tissue collection and morphological evaluation
Pairs of ovaries were excised, and weighed. For morphological evaluation, whole ovaries were embedded in paraffin block and longitudinal sections (4 μm thick) were stained with hematoxylin and eosin. Since the considerable discrepancy in estimating follicle numbers exists between studies due to different section thickness and correction factors adopted [
27], the percentage of number of cystic follicles (FC) and corpus lutea (CL) over total follicular structure (including all kinds of follicles and CL) per ovary were determined. A FC was defined as an collapsed anovulatory follicle exhibiting the absence of oocyte, reduced granulosa cell layer, but intact theca layer. The maximal longitudinal ovarian sections were investigated to count the numbers of FC and CL, mean of the percentage of FC and CL from 3–5 sections after sampling every fifth section were defined as the relative values per ovary from 5 individual animals. In an additional experiment, three rats from DHT group at each time point were administered with 20 IU of equine chorionic gonadotropin (eCG; Sigma; PBS as control) for designated time (0–30 h) to test their ovarian responsiveness to gonadotropin in vivo.
miRNA expression profiling
Ovarian cortex tissues were snap-frozen in liquid nitrogen and stored at −80°C until analysis. Total RNAs including small RNAs from 3 independent DHT (DHT1-3) treated and 3 non-treated control (CTL1-3) ovaries have been isolated using miRNeasy mini kit (Qiagen, Hilden, Germany) following manufacturer’s protocol. Immediately after isolation, contaminating DNA and divalent cations from the RNA samples were removed using TURBO DNA-free™ kit (Ambion, Foster City, CA) according to instruction of the manufacturer. The quality and the concentration of the RNA samples were assessed by NanoDrop 8000 spectrophotometer (NanoDrop, Wilmington, Delaware, USA). Subsequently, a total of 250 ng purified RNA from each samples were reverse transcribed for qPCR analysis, using the QuantiMir™ RT systems (System Biosciences, Mountain View, CA).
MiRNA expression analysis was performed using miRNA PCR array platform containing 349 well characterized Rat miRNAs including 3 endogenous controls, namely Rat U6 snRNA, RNU43 snoRNA & U1 snRNA (System Biosciences, Mountain View, CA, Cat # RA680A-1) according to the user manual of miRNome microRNA Profilers QuantiMir™ (System Biosciences, Mountain View, CA) with some modifications. Briefly, 20 μl of cDNA from each sample mixed with 1300 μl of Maxima 2x SYBR Green, (Fermentas Gmbh, Leon-Rot, Germany, Cat # K0222), 50 μl of universal reverse primer and 800 μl of nuclease-free water. Five microlitres of mixed cDNA cocktail were dispersed into each well of 384 well PCR plate containing 1 μl of pre-aliquoted miRNA specific primer. Multichannel laboratory automation workstation was used for pipetting the mix and mixing the reaction into 384 well plates (Biomek® NXP, Beckman Coulter, Krefeld, Germany). The assays were performed in ABI 7900 HT real time PCR system (Applied Biosystems, Foster City, CA, USA) with thermal cycling programme at 52°C for 2 min, 95°C for 10 min, 40 cycles of (95°C for 15 sec, 60°C for 1 min). Dissociation stage was included at the end of the programme to assess the Tm of the PCR amplicons and to verify the specificity of the amplification reactions. In addition, amplification was verified for randomly selected reactions through 2.5% agarose gel electrophoresis. Data were analysed by ΔΔCt method and normalized by the geometric mean of the three endogenous controls through System Bioscience’s free Software developed to analyse PCR Array data for Rat miRNome (System Biosciences, Mountain View, CA), equipped with t test. A fold regulation of ≥ 2. P ≤ 0.05 was considered as significant in expression differences.
Localization of selected miRNAs in ovarian cryo-section
Ovaries were fixed in 4% paraformaldehyde (pH 7.3; 4°C, overnight) followed by incubation in 0.5M Sucrose in DEPC-treated 1X PBS (4°C, overnight). They were embedded in Tissue-Tek Oct reagent (Shandon Cryomatrix, Thermo Cat# 6769006) in a tissue path base mold, underwent a slow freezing process (−1°C/min rate) in a freezing container (NALGENE™ Cryo, Cat# 5100–0001). Serial ovarian sections (10 μm) were cut at −23°C using rapid sectioning cryostat (Leica microsystem Nussloch GmbH, Heidelberger, Germany) and mounted on SuperFrost® Plus slides (Menzel-Glaeser, Germany) at −80°C until hybridization began. Sections were checked for quality and required cell content at a regular interval by staining with TB in a light microscope. Post-fixation, acetylation and proteinase K treatment was done as described previously [
28]. Two hours of pre-hybridization was performed at 52°C in hybridization solution (50% formamide, 5x sodium chloride/sodium citrate [SSC; pH 6.0], 0.1% Tween-20, 50 μg/ml heparin, and 500 mg/ml yeast tRNA). Ovarian sections were incubated overnight at 52°C in a hybridization buffer containing 3'-Digoxigenin (DIG) labeled LNA-modified oligonucleotide probes (1pM) for miRNA, U6 RNA and scrambled miRNA (Exiqon, Vedbaek, Denmark) in a humidified chamber. Blocking, incubation with anti-DIG-AP antibody, washing and color development with alkaline phosphatase substrates solution [5-Bromo-4-chloro-3-indolyl phosphate/Nitro blue tetrazolium (BCIP/NBT)] were performed as described previously [
28]. Vector® Methyl Green (Vector laboratories, Burlingame, CA) was used as nuclear counter-stain.
Statistical analyses
All data were analyzed using Prism 5.0 statistical software (GraphPad, San Diego, CA). The results were presented as means ± SEM as detailed in figure legends. Differences in body weight gains, follicle cysts and corpus luteum percentage were analyzed by two-way ANOVA, ovarian weight between groups were analyzed by one-way ANOVA and followed by Bonferroni post-hoc test. Insulin sensitivity in PCOS and control groups was compared by Student t-test. Statistical significant difference was defined at P < 0.05.
Discussion
PCOS is a complex and heterogeneous endocrine condition characterized by hyperandrogenemia, hyperinsulinemia, insulin resistance, and chronic anovulation. Using a rodent model for PCOS which recapitulates the reproductive and metabolic phenotypes of the human conditions [
23], we have begun to examine molecular and cellular mechanisms underlying the dysregulated reproductive and metabolic functions of PCOS. We have demonstrated that chronically androgenised rats, exhibited increased body weight, disrupted estrus cyclicity, decreased insulin sensitivity and decreased ovarian weight. In the present study we have for the first time that these phenotypes are associated with dysregulation of gene regulatory noncoding miRNA.
Ovarian growth is the result of a series of complex and coordinated processes, which include morphological and functional changes in different follicular cells and their interactions. Among the small non-coding RNAs associated to various biological and physiological disorders (sncRNAs), miRNAs are the best characterized. Studies on the bovine and mouse ovary have demonstated the presence of miRNAs and suggested their potential role in ovarian functions [
10,
29]. Several intra-ovarian regulators and their receptors have been predicted as targets of this ovarian miRNAs, which mediate important intracellular events necessary for normal follicular development and functions.
Although miRNAs are expressed in the human, mouse and cow ovaries [
10,
14,
29‐
34], little is known of their role and dysregulation in PCOS. The present study represents the first report on the dysregulation of miRNAs in a rat ovary chronically exposed to DHT, a condition mimicking the hyperandrogenic condition in human PCOS. Among the miRNAs examined, 79 miRNAs (24%) responded to the hyperandrogenic condition and interestingly, 80% of which were upregulated compared to the control group supporting the notion that hyperandrogenic condition down-regulates androgen receptors in the granulosa cells [
35] which could be mediated by these upregulated miRNAs (rno-miR-379*, rno-let-7d, rno-miR-24, rno-miR-673, rno-miR-26b, rno-miR-335, rno-miR-382*, rno-miR-412, rno-miR-99a*, rno-miR-543, rno-miR-674-3p, rno-miR-409-3p). It has been reported that the expression of miR-143, let-7a, miR-15b is under negative control of follicle stimulating hormone (FSH) during follicular development [
36] and may be involved in FSH-induced rat granulosa cell progesterone production [
37]. Thus, it is possible that the down-regulation of miRNAs (rno-miR-770, rno-miR-466c, rno-miR-31, rno-miR-183, rno-miR-96, rno-miR-132, rno-miR-182, rno-miR-384-3p and rno-miR-184) observed in this study could be associated with promoted thecal hyperandrogenesis [
37,
38]. Similarly, stepwise artificial neural networks analysis revealed predictive miRNA signatures (miR-342, miR-299, miR-217, miR-190, miR-135b, miR-218) corresponding to oestrogen receptor status in breast cancer [
39]. These findings raises the possibility that the modulated miRNA expression by exogenous DHT treatment activates estrogen receptors (ERs) to exert indirect steroidal actions on the follicular cells or directly on androgen receptor pathway for the regulation of androgen metabolism [
40].
Recent evidences suggest that miRNAs are involved in the regulation of steroid hormone receptors (SHRs) in ovarian follicle growth and hormone-responsive cancers. Follicular development is the result of tightly regulated activities of FSH, estrogen, androgen and luteinizing hormone receptor (LHR) in granulosa and theca cells, as well as oocyte in response to hormonal stimulation and actions of intra-ovarian regulators [
15]. MicroRNAs expression profile and analysis of their target genes revealed that microRNAs exhibits characteristics important in regulating downstream LHR signaling in ovarian cancer cells [
41]. MiR-132, 5-fold down-regulation in DHT-treated group compared to control in this study, has previously been shown to be related to luteinizing hormones [
12], raising the possibility that elevated LH in PCOS could involve the expression and action of miR-132.
Androgen Receptors (ARs) mRNA is predominantly expressed in rat granulosa cells and its expression decreases with the stage of follicle development [
35]. Hyperandrogenemia induced by DHT is characterized by down-regulation of granulosa cell AR in rat preantral which could be restored by FSH administration [
35]. Ovarian follicle expression of miR-221 and miR-222 has been reported to be repressed by androgens [
42‐
44] which regulates cell proliferation by targeting p27/kip1 [
44,
45]. Here we observed higher expression of miR-222 in the DHT-induced PCOS rats, a response most evidenced in both theca and granulosa of early stage follicles. Interestingly, its expression pattern changes with increased follicular growth and antrum formation. In the cystic follicle, expression of miR-222 becomes restricted to theca only, suggesting that miR-222 modulates AR expression and hereby paracrine regulation.
Clinical studies have shown that follicular ERα and ERβ expression is significant altered in PCOS [
46]. In breast cancer cells, over-expression of miR-221, miR-222, let-7 and miR-20b is associated with reduced of ERα protein content, signaling and expression of ERα target genes [
47‐
49]. Our present study also showed higher expression of these miRNAs in the induced PCOS, consistent with a similar mechanism found in the breast cancer cells and their possible role in the regulation of ERα signaling.
In situ localization shows specific expression of miR-222 in the theca cells (Figure
5) where it may be involved in regulation of the ERα. In addition to miR-221/222, several studies also highlighted the differential regulation of let-7d, let-7f, miR-25 and miR-26b in prostate and breast cancer, as well as in leukemia by the estrogen receptor pathways and that their expression was up-regulated in ERα-positive cells [
50‐
52]. Expression analysis of these miRNAs revealed their higher expression in PCOS, raising the possibility that estrogen-mediated regulation of these miRNAs might be involved in the etiology of PCOS
. Unlike other miRNAs in DHT-treated PCOS rats, miR-183 was found to be highly expressed in granulosa cells of control ovaries but was down-regulated by 10-fold in the PCOS counterpart. Interestingly, this miRNA was found to be correlated with estrogen receptor in breast cancer [
53].
Hyperandrogenism in PCOS is one of the main causes for early follicle excess, which induces nuclear forkhead box transcription factors 3a (FOXO3a) exclusion and follicular arrest [
54]. FOXO3a, a target of miR-96, promotes cancer cell proliferation [
55]. Accordingly, expression of miR-96 in PCOS shows a 6-fold down-regulation compared to control ovary and is localized in the theca and cumulus-granulosa of cystic follicles. Thus, the cell proliferation in the preantral follicles in hyperandrogenic condition could be induced through a mechanism utilizing miR-96-mediated regulation.
PCOS is a complex syndrome with both reproductive and metabolic disorders. Although the rat PCOS model (DHT-treatment) simulates the hyperandrogenic state found in human PCOS and recapitulates some of the phenotypes of the human syndrome (obesity, disruption of menstrual cycle, insulin resistance), an inherent problem of any animal model for the complex syndrome is that it does not address all the phenotypes in PCOS. While the rat PCOS model may provide some mechanistic insight in the pathophysiology of the human syndrome, the significant findings from the animal studies need to be validated in the human context. In addition, although miRNAs are known to regulate multiple gene targets in different system [
56,
57], they can also be regulated by other paracrine/autocrine factors [
44,
58]. The precise target (s) for the reported dysregulated miRNA and their immediate down-stream processes need to be defined.
In summary, using an androgenized rat model which recapitulates many of the reproductive and metabolic phenotypes of human PCOS, we have shown that miRNAs are differentially regulated in a follicular cell-specific manner. Since one given miRNA could target multiple mRNAs of different genes and a given target may be regulated by multiple miRNAs [
59], whether these miRNAs involve in the dysregulation of the hormone receptors (AR, FSHR and ER) and intra-ovarian regulators of ovarian follicle growth and function in PCOS is unknown. The expression of predicted target genes and their regulation by these miRNAs needs to be clarified by further experiments. Our ongoing studies not only include target gene expression analysis but also their correlation with the miRNAs, and functional analysis. Taken together, the present findings support the notion that the present DHT-treated rat model is useful for investigating the role and regulation of miRNAs in the molecular and cellular mechanism of PCOS and may offer new insights for the treatment of reproductive and metabolic disorders associated to PCOS.
Competing interests
The authors declare that they have no competing interests
Authors’ contributions
MMH, MJC, QW and JYK performed the experiments, prepared the data and drafted the manuscript. KS, DT and BKT are co-mentors, provided input of studies and edited the manuscript. All authors read and approved the final manuscript.