DNA Methylation and Histone Modifications Are Associated with Repression of the Inhibin α Promoter in the Rat Corpus Luteum

Abstract

The transition from follicle to corpus luteum after ovulation is associated with profound morphological and functional changes and is accompanied by corresponding changes in gene expression. The gene encoding the α subunit of the dimeric reproductive hormone inhibin is maximally expressed in the granulosa cells of the preovulatory follicle, is rapidly repressed by the ovulatory LH surge, and is expressed at only very low levels in the corpus luteum. Although previous studies have identified transient repressors of inhibin α gene transcription, little is known about how this repression is maintained in the corpus luteum. This study examines the role of epigenetic changes, including DNA methylation and histone modification, in silencing of inhibin α gene expression. Bisulfite sequencing reveals that methylation of the inhibin α proximal promoter is low in preovulatory and ovulatory follicles but is elevated in the corpus luteum. Increased methylation during luteinization is observed within the cAMP response element in the promoter, and EMSA demonstrate that methylation of this site inhibits cAMP response element binding protein binding in vitro. Chromatin immunoprecipitation reveals that repressive histone marks H3K9 and H3K27 trimethylation are increased on the inhibin α promoter in primary luteal cells, whereas the activation mark H3K4 trimethylation is decreased. The changes in histone modification precede the alterations in DNA methylation, suggesting that they facilitate the recruitment of DNA methyltransferases. We show that the DNA methyltransferase DNMT3a is present in the ovary and in luteal cells when the inhibin α promoter becomes methylated and observe recruitment of DNMT3a to the inhibin promoter during luteinization.

Hormones are powerful regulators of gene expression in both space and time. The selective effects of hormones on particular cell types or specific genes are specified by receptors and second messengers, transcription factors and cofactors, and chromatin-modifying and -remodeling complexes downstream of the hormonal signal. In the ovary, the pituitary gonadotropin hormones FSH and LH control the dynamic gene expression changes that occur during the reproductive cycles and are necessary for proper growth and maturation of ovarian follicles, release of the oocyte during ovulation, and formation of the corpus luteum. Under the influence of FSH, granulosa cells proliferate, support the maturation of the oocyte, and secrete hormones such as estrogen. The preovulatory LH surge triggers ovulation and initiates terminal differentiation of granulosa cells into luteal cells, which produce hormones such as progesterone, which is necessary for proper implantation of the embryo. The differing functions of preovulatory and luteinized granulosa cells requires that many genes that were highly expressed in the follicle to be turned off in the corpus luteum and, conversely, other genes repressed in the follicle to be induced in the corpus luteum. Gene expression profiling has identified hundreds of genes in the ovary that are differentially expressed before and after the LH surge or human (h) chorionic gonadotropin (CG) administration (1–4).

The inhibin α subunit gene is highly expressed in granulosa cells of the preovulatory follicle and then rapidly and robustly down-regulated at the time of the LH surge, remaining repressed in the corpus luteum (5, 6). The inhibin α subunit dimerizes with the β_A or β_B subunit to form inhibin A and B, respectively (7–10). Dimerization of the β subunits results in the formation of a related hormone, activin (11, 12). The main function of ovarian inhibin is to negatively regulate FSH production and secretion from the anterior pituitary (13, 14), and it antagonizes the activity of activin (11, 12, 15). In addition to these endocrine roles, both hormones have autocrine and paracrine effects within the ovary. Because the α subunit is unique to inhibin and is also the most abundant subunit in the granulosa cells (6), its regulation is often the focus of studies with respect to understanding inhibin expression in the ovary.

The down-regulation of inhibin at the time of the LH surge plays an important role in the estrous cycle and normal reproduction. Repression of inhibin on the evening of proestrus allows for the secondary FSH surge on estrous morning, which contributes to the recruitment of new follicles for the subsequent cycles (14). Its chronic expression in mice carrying an α-inhibin transgene results in altered levels of FSH and LH, reduced ovulation, ovarian cysts, multioocytic follicles, and subfertility (16, 17). Therefore, understanding the mechanisms by which this gene is negatively regulated and by which repression is maintained is important with respect to reproductive health.

In rats (6, 18), sheep (19), cows, and pigs (20), inhibin α expression remains repressed in the corpus luteum. In the rat, this repression is maintained despite the continued expression of transcription factors that have been shown to activate the inhibin α gene in granulosa cells, such as liver receptor homolog-1 (21, 22), steroidogenic factor 1 (SF-1) (21, 22), and cAMP response element binding protein (CREB) (23, 24). Although inducible cAMP early repressor (ICER) (25, 26), CCAAT/enhancer binding protein β (C/EBPβ) (27), and the NR4A orphan nuclear receptors (our unpublished data) have been identified as transcription factors involved in the down-regulation of inhibin α in response to the LH surge, these factors are only transiently expressed and cannot account for the sustained repression in the corpus luteum.

The epigenetic environment of the gene, including histone modifications, chromatin organization, and DNA methylation, is a critical determinant of gene transcription. DNA methylation is often associated with more stable gene expression changes that occur in cancer (28), early development (29), or genomic imprinting (30). However, there is growing evidence that dynamically regulated genes can also be methylated and demethylated to attenuate promoter activity in response to hormonal (31) or other stimuli (32, 33).

Epigenetic modifications on the inhibin α promoter have been observed in several tissues. Increased DNA methylation of the inhibin α subunit gene has been shown to occur in prostate and gastric cancer tumors and cell lines in which inhibin α is not expressed (34–36). Increased histone H3 and H4 acetylation and H3 phosphorylation of the proximal inhibin α promoter have been associated with active transcription in granulosa cells (21, 22, 37). In this study, we hypothesized that repressive epigenetic changes, such as DNA methylation and specific histone modifications, would be associated with the silencing of inhibin α gene expression in the rat corpus luteum.

Materials and Methods

Animal and hormone treatments

Immature 23- to 25-d-old Sprague Dawley rats (Charles River Laboratories, Inc., Lexington, MA) were maintained on a 14-h light, 10-h dark cycle. Rats were injected sc with 10 IU pregnant mare serum gonadotropin (PMSG) (Sigma Aldrich, St. Louis, MO). After 48 h, rats were either killed or given an ip injection of 15 IU hCG (Sigma Aldrich) and killed at various time points thereafter. All animal protocols were approved by the Animal Care and Use Committee at Northwestern University (Evanston, IL). Animals were maintained in accordance with the NIH Guide for Care and Use of Laboratory Animals.

Tissue and cell preparations

To isolate ovarian cells and tissue, dissected ovaries were placed in 4F medium [DMEM-Ham's F12, 1:1 supplemented with 15 mm HEPES (pH 7.4), transferrin (5 μg/ml), insulin (2 μg/ml), hydrocortisone (40 ng/ml), 10% fetal bovine serum, and antibiotics] and extraneous tissue was removed. Needles were used to dissociate follicles and corpora lutea, and they were immediately frozen on dry ice until use. For cells, ovaries were placed in 4F containing 0.5 m sucrose and 10 mm EGTA for 30 min at 37 C and then washed twice with 4F. Using 23- to 26-gauge needles, the cells were mechanically dispersed from the ovaries and collected by centrifugation.

Bisulfite sequencing

Genomic DNA was isolated either by the Purelink Genomic DNA Mini Kit (Invitrogen, Carlsbad, CA) or digestion buffer supplemented with proteinase K overnight, followed by Rnase treatment, phenol-chloroform extraction, and ethanol precipitation. DNA (1 μg) was used in the bisulfite conversion reaction performed with the CpGenome Fast DNA Modification Kit (Millipore Corp., Billerica, MA) as per the manufacturer's instructions.

Heminested PCR was performed on 5 μl of bisulfite-treated DNA. The following parameters were used for the first PCR: 95 C for 5 min followed by five cycles of 95 C for 1 min, 54 C for 2 min, and 72 C for 3 min; 25 cycles of 95 C for 1 min, 54 C for 2 min, and 72 C for 2 min; and a final extention of 72 C for 10 min. Two microliters of the first PCR were used for the second PCR. Four reactions of the second PCR were pooled per sample. The second PCR was performed as follows: 95 C for 5 min; 20 cycles of 95 C for 30 sec, 58 C for 1 min, and 72 C for 3 min; and a final extention of 10 min at 72 C. The primers are listed below and also contained the 5′-overhang for ligation-independent cloning into pet30Xa/LIC vector (Novagen, San Diego, CA).

The PCR products were gel purified using the Qiaquick Gel Extraction Kit (QIAGEN, Chatsworth, CA). Samples were treated with T4 DNA polymerase to create the 5′-overhangs and annealed into the pet30 Xa/LIC vector as per the manufacturer's instructions. The annealed plasmid was transformed into dh5α cells. Plasmid DNA was purified using the DNA Wizard Miniprep Kit (Promega Corp., Madison, WI). DNA was sequenced at the Northwestern University Genomics Core Sequencing Facility using the T7 terminator primer.

−461 rat inhα 5′: 5′-AGAGTAGGTAGGATTATTTGTTTTTTATTTAG-3′

−393 rat inhα 5′: 5′-AGTAGTTGTTAYGGTTGAAAAGAGTTTTAG-3′

+153 rat inhα 3′: 5′-ACAAATTCTAACCCCTAACAACTATCC-3′

EMSA

Double-stranded nucleotide probes (36 bp) containing the inhibin α SF-1 binding site (SBS) and either the wild-type or symmetrically methylated cAMP response element (CRE) were end labeled with [γ-32^P]ATP by T4 polynucleotide kinase (Promega). Probes were purified with Illustra Microspin G-50 columns (GE Healthcare, Waukesha, WI). Recombinant CREB bZip or recombinant SF-1 DNA-binding domain (DBD) protein were incubated in binding buffer [10 mm Tris (pH 7.5), 1 mm MgCl₂, 1 mm dithiothreitol, 200 ng/μl polydeoxy(inosinic-cytidylic)acid] and 50 μm ZnCl₂ (for SF-1 DBD only) and probe (40,000 cpm) for 20 min at room temperature. The protein-DNA complexes were separated on a 5% polyacrylamide gel at 4 C. The gel was dried and exposed to film at −80 C. For quantification, gels were placed on phosphor screens and scanned by a Storm 860 phosphoimager (GE Healthcare) at Northwestern University's Keck Facility. Densitometry was calculated using ImageQuant software (GE Healthcare).

Chromatin immunoprecipitation (ChIP) assay

MagnaChIP beads (Millipore) were washed three times with blocking buffer (0.5% BSA/PBS) and resuspended in blocking buffer. Antibodies [α-CREB (Millipore), 17-600; α-H3K27me3, ab6002 (Abcam, Cambridge, MA); α-H3K4me3, ab8580, or α-H3K9me3, ab8898; α-DNMT3a, ab2850, (Abcam) or normal rabbit IgG, (Invitrogen), as a negative control] were added to the beads and rotated for 8 h at 4 C. The beads were washed and resuspended in blocking buffer.

Cells were washed twice with PBS and cross-linked for 10 min with 1% formaldehyde. Glycine was added to a final concentration of 1.5 mm for 5 min. The cells were washed three times with PBS and protease inhibitors (1 μg/ml each of antipain, leupeptin, aprotinin, and pepstatin A, and 0.5 mm phenylmethylsulfonyl fluoride). The cells were resuspended in lysis buffer [1% sodium dodecyl sulfate (SDS), 50 mm Tris, 10 mm EDTA, protease inhibitors] and incubated for 5 min on ice. Lysates were snap frozen on dry ice and stored at −80 C until all samples were collected. Chromatin was sonicated by 8 times 20 one-second pulses at 30% output at 4 C. Chromatin was removed from the cell debris and divided, saving 20 μl for use as inputs. Dilution buffer [0.1% SDS, 1.1% Triton, 1.2 mm EDTA, 17.6 mm Tris-Cl (pH 8.0), 1.67 mm NaCl and protease inhibitors] was added to each sample and incubated with the antibody-coupled beads overnight at 4 C with rotation. The beads were washed two times each with low-salt wash buffer [0.1% SDS, 1% Triton, 2 mm EDTA, 20 mm Tris-Cl (pH 8.0), 150 mm NaCl], high-salt wash buffer [0.1% SDS, 1% Triton, 2 mm EDTA, 20 mm Tris-Cl (pH 8.0), 150 mm NaCl], LiCl₂ wash (250 mm LiCl₂, 1% sodium deoxycholate, 1% Igepal), and 1×TE/50 mm NaCl. Immunocomplexes were eluted [50 mm Tris-Cl (pH 8.0), 20 mm EDTA, and 1% SDS] at 65 C for 15 min. The beads were removed by centrifugation. All samples and inputs were placed at 65 C to reverse cross-links. Samples were treated with Rnase followed by proteinase K. The DNA was extracted by phenol-chloroform and ethanol precipitation and resuspended in 35 μl of 10 mm Tris-Cl, pH 8.0. Five microliters were used for quantitative PCR (qPCR).

Quantitative RT-PCR (qRT-PCR) and PCR

cDNA was generated by reverse transcriptase with avian myeloblastosis virus reverse transcriptase at 42 C for 1 h 15 min and 95 C for 5 min. qPCR was performed on the cDNA using intron-spanning primers specific to each target gene and Power SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA). The reactions were performed in duplicate on a 7300 Real-time PCR System (Applied Biosystems). Ribosomal protein L19 (RPL19) was used as a control for normalization. Melting curves were generated for each reaction to confirm a single product was generated by each primer set. The primers are listed below.

rInhα 5′: 5′-AACTCTGAACCAGAGGAGGA-3′

rInhα 3′: 5′-GGCCGGAATACATAAGTGAA-3′

DNMT3a 5′: 5′-ATGCCAAGACTCACCTTCC-3′

DNTM3a 3′: 5′-ACTGCAATCACCTTGGCTT-3′

rRPL19 5′: 5′-AAACCAACGAAATCGCCAAT-3′

rRPL19 3′: 5′-GGCAGTACCCTTCCTCTTCC-3′

For the ChIP assays, qPCR was performed as described above, using primers within the inhibin α promoter or first exon. The proximal promoter of RPL19 and distal inhibin α promoter served as negative controls. The primers are listed below. Data were first normalized to the input for each sample and then plotted as the fold change from the preovulatory cells; n = 3–6 independent experiments per time point.

rInha prox 5′: 5′-CCTGAATGGGTCAGGTCACT-3′

rInhα prox 3′: 5′-ACCCTTCTTCCCCAGTCTC-3′

rInhα exon 1 5′: 5′-GTCTCTGCTGCTCCTTTTGC-3′

rInhα exon 1 3′: 5′-CCCCAAGGCATCTAGGAATAG-3′

rRPL19 prox 5′: 5′-TGATCACGGTCTTTTAGGAAGAA-3′

rRPL19 prox 3′: 5′-CGGGCTCCTCCCATTAACTA-3′

rInhα distal 5′: 5′-TGTCTCGATTTGCTCCTGTGT-3′

rInhα distal 3′: 5′-AGTGCAGACCCTTGCAGAAT-3′

Immunohistochemistry

Ovaries were fixed in 4% paraformaldehyde for 24 h, stored in 70% ethanol at 4 C, and embedded in paraffin. Sections (5 μm) were cut. Slides were deparaffinized and rehydrated through alcohol and incubated with avidin-biotin (Vector Laboratories, Inc., Burlingame, CA) for 30 min at room temperature. Slides were incubated with 10% goat serum (Vector Labs) and 1% BSA for 1 h at room temperature, and incubated with primary Dnmt3a antibody (1:200 dilution) or inhibin α antibody (1:200 dilution) overnight at 4 C. After a PBS wash, sections were incubated with biotinylated antirabbit IgG (Vector Labs) for 30 min at room temperature. The color was developed with 3,3′-diamino-benzidine (Vector Labs) for 3 min, and slides were counterstained with hematoxylin for 1 min.

Statistical analysis

Statistical analysis was performed using Graphpad Prism 5.0 for Macintosh (GraphPad Software, La Jolla, CA).

Results

Tissue-specific expression of inhibin α is inversely correlated with promoter methylation

To determine whether promoter methylation is associated with tissue-specific inhibin expression in the rat, we used bisulfite DNA sequencing to compare proximal promoter methylation in the liver, which has very low inhibin α expression, and in primary granulosa cells, which express inhibin α at a much higher level, as shown through qRT-PCR in Fig. 1A. The region of the proximal promoter and first exon analyzed contains five CpG dinucleotides, as shown in Fig. 1B, including those contained within the cAMP response element (CRE, −120), a putative specificity factor 1 (Sp1)-binding site (−158), and an activator protein 2 (AP2)/Sp1 binding site (−66). There was an increased level of methylation in the liver compared with the granulosa cells, where the promoter was largely unmethylated (Fig. 1C).

Fig. 1.

Tissue-specific expression of inhibin α is inversely related to proximal promoter methylation. A, qRT-PCR was used to quantify the amount of inhibin α mRNA in liver and ovarian granulosa cells from PMSG-primed immature female rats. Relative abundance of inhibin α mRNA in granulosa cells and liver are shown relative to RPL19 mRNA; n = 3 rats for granulosa cells and liver. B, Schematic of the inhibin α proximal promoter amplified in the bisulfite sequencing analysis. Circles indicate CpG dinucleotides, and the numbers below indicate position relative to the transcription start site. Relative transcription factor-binding sites are indicated; those in italics have not been characterized. C, Bisulfite sequencing was performed on genomic DNA isolated from liver and granulosa cells from PMSG-primed immature rats, and the average number of clones that exhibited at least one methylated CpG site is shown. Granulosa cells were pooled from six rats; n = 3 for liver; 30–45 clones analyzed per tissue. Different letters represent statistical significance as determined by unpaired two-tailed t test (P < 0.05). Error bars represent sem.

Repression of inhibin α expression in the rat corpus luteum is associated with increased proximal promoter DNA methylation

Previous work from our laboratory has shown through multiple methods that inhibin α mRNA is highly expressed in rat preovulatory follicles but absent in corpora lutea (6), and this down-regulation of inhibin α mRNA begins within 1 h after hCG administration to immature rats primed with PMSG (25). Because tissue-specific inhibin α expression is inversely associated with promoter methylation, we postulated that the changing levels of inhibin α gene expression within the ovary might also be associated with different levels of promoter methylation. We decided to examine methylation of the promoter in follicles and corpora lutea where inhibin α levels are expected to be the most different.

To quantitatively compare inhibin α mRNA levels, RNA was prepared from preovulatory follicles isolated 48 h after PMSG injection, ovulatory follicles isolated 18–20 h after hCG injection of PMSG-primed rats, and corpora lutea isolated 6 d after hCG injection. Livers from PMSG and PMSG/hCG-stimulated animals were used as a negative control. As shown in Fig. 2A, qRT-PCR results demonstrate that inhibin α mRNA is highly expressed in the preovulatory follicles and dramatically reduced in the corpora lutea. Robust repression is also achieved 18 h after hCG injection in ovulatory follicles before luteinization. Levels of inhibin α mRNA in the livers of PMSG- and PMSG/hCG-treated animals were very low (Fig. 2A).

Fig. 2.

Inhibin α gene repression in the rat corpus luteum and liver is associated with increased promoter methylation. A, qRT-PCR for inhibin α was performed on RNA isolated from preovulatory follicles (black bar), ovulatory follicles (gray bar), and corpora lutea (white bar) obtained from gonadotropin-stimulated immature female rats at time points PMSG 48 h, PMSG 48 h/hCG 18 h, and PMSG 48 h/hCG 6 d, respectively; n = 3 rats per time point. The liver mRNA level (hatched bar) is the average from three PMSG- and three PMSG/hCG 6 d-stimulated animals. There was no significant difference in inhibin α mRNA levels in the liver between the PMSG 48-h and PMSG/hCG 6-d treatments (data not shown). B. Bisulfite sequencing was performed on genomic DNA from rat preovulatory follicles, ovulatory follicles, corpora lutea, and liver (n = 6 rats for each tissue; 9–10 clones per sample). The liver data are the average of three PMSG 48 h- and PMSG 48 h/hCG 6 d-stimulated animals; there was no significant difference in methylation in the liver between the two hormone treatment groups (data not shown). Different letters signify statistical significance compared with “a” (P < 0.05) whereas same letters represent statistical similarity as determined by unpaired two-tailed t test. Error bars represent sem.

To determine whether the inhibin α repression coincides with promoter methylation, bisulfite sequencing was performed on genomic DNA isolated from rat preovulatory follicles, ovulatory follicles, and corpora lutea. Genomic DNA from livers obtained from PMSG- and PMSG/hCG-treated animals were also used for bisulfite sequencing to confirm that the hormone regimens did not nonspecifically change the methylation status in other tissues. As shown in Fig. 2B, approximately 37% of the analyzed clones obtained from corpora lutea were methylated, a significant increase compared with the preovulatory follicles, in which only 5% of the clones were methylated. Interestingly, the promoter was also largely unmethylated in the ovulatory follicles, despite the gene being significantly repressed at this time. These results suggest the acute repression in the ovulatory follicles does not involve DNA methylation, whereas the longer-term repression in the corpus luteum is associated with increased promoter methylation, which could play a role in gene silencing. The average methylation levels in liver were approximately 86%, again confirming that increased promoter methylation is associated with decreased inhibin α expression.

Representative individual methylation patterns obtained from the bisulfite sequencing of clones are shown in Fig. 3. There was an increase in the diversity of methylation patterns obtained from the corpora lutea and liver compared with the follicles. The increased diversity suggests that the clones analyzed are derived from distinct cellular DNA molecules rather than resulting from any bias in the PCR amplification process.

Fig. 3.

Diversity of inhibin promoter methylation patterns is increased in rat corpora lutea and liver. Individual methylation patterns obtained from bisulfite sequencing of clones in the ovary and liver. Each circle represents a CpG dinucleotide (schematic with position number at the top of each column) with open circles representing unmethylated sites and black circles representing methylated sites. Each horizontal line is one clone pattern with the number in parentheses at the end of each line indicative of the number of clones with that specific pattern (n = 6 rats for each tissue sample).

Methylation of the inhibin α CRE inhibits CREB binding

Four of the five CpG in the proximal promoter were significantly hypermethylated in corpora lutea compared with follicles, as shown in Fig. 4. Interestingly, with the exception of the −66 site, the methylation levels in the corpora lutea were similar to the levels seen in the liver. As stated previously, three of the five CpG within the proximal inhibin α promoter are contained within transcription factor binding sites. The CpG at −120 is an atypical CRE that binds CREB and is critical to transcriptional activation (23, 24). The CRE also binds ICER and C/EBPβ when they are induced after the LH surge (25–27).

Fig. 4.

Relative hypermethylation of transcription factor binding sites occurs on the inhibin α promoter in corpora lutea. Quantification of site-specific methylation levels from preovulatory follicles, ovulatory follicles, corpora lutea, and liver (average of PMSG 48 h- and PMSG 48 h/hCG 6d-stimulated livers). Statistical significance (P < 0.05) at each CpG is represented by different letters, whereas same letters represent statistically similar samples as determined by unpaired two-tailed t test. Error bars represent sem.

Because CREB expression is high in luteinized granulosa cells (38, 39), where inhibin α expression is low, CREB must be prevented from binding to the CRE. We hypothesized that the observed CRE methylation in the corpus luteum could block CREB from binding as a functional consequence of promoter methylation. EMSA were performed using oligonucleotide probes spanning the inhibin α CRE and SBS containing a methylated or unmethylated cytosine within the CpG at position −120. As shown in Fig. 5A, CRE methylation inhibited recombinant CREB bZIP protein from binding. Binding of the SF-1 DBD protein was used to normalize for the total amount of DNA probe added, because it binds a separate site within the probe. CREB binding to the unmethylated probe was approximately 7-fold higher than to the methylated probe (Fig. 5B). Similar results were obtained with full-length recombinant CREB and endogenous CREB from granulosa cells (data not shown). These data suggest that methylation of CpG-containing transcription factor binding sites alone can hinder such factors from binding as one possible mechanism contributing to repression.

Fig. 5.

Methylation of the inhibin α CRE inhibits CREB binding. A, EMSA was performed with 36-bp oligonucleotide probes spanning the inhibin α CRE and SBS. The methylated probe contained a symmetrically methylated cytosine at the central CpG within the CRE. Radiolabeled probes were incubated with 0.6 μm recombinant CREB bZip or 3 μm SF1-DBD (negative control) and size separated on a 5% polyacrylamide gel. Similar results were obtained with higher concentrations of CREB bZip and full-length recombinant CREB (data not shown). B, Binding of CREB (0.6 μm and 3 μm) was normalized to SF-1 DBD binding (3 μm and 6 μm) by densitometry on scanned phosphor screens (n = 2). Lanes 1 and 5, probe alone; lanes 2 and 6, probe and CREB bZip; and lanes 3 and 7, probe and SF-1 DBD. C, ChIP for CREB was performed in primary cells isolated from gonadotropin-stimulated immature rats at PMSG 48 h (preovulatory cells) and PMSG 48 h/hCG 6d (luteal cells). The proximal promoter, including the CRE, was amplified with the primers (black arrows) shown in the schematic above the graph.

To assess whether CREB binding is reduced in vivo when the CpG within the proximal promoter are methylated, we performed ChIP for CREB on ovarian cells isolated from rats treated with PMSG for 48 h (“preovulatory cells”) or PMSG for 48 h plus hCG for 6 d (“luteal cells”). The proximal promoter that was amplified in the ChIP assays is shown in Fig. 5C. CREB binding was found to be significantly reduced in the luteal cells, at a time when the CpG within the CREB-binding site shows enhanced methylation.

Inhibin α promoter histone modifications change during luteinization

Modifications on the N-terminal tails of histones within nucleosomes can change the chromatin environment by altering the compactness of the DNA and through the recruitment of proteins that modify the accessibility of DNA to transcription factors and the RNA polymerase machinery. Specific modifications of particular residues on the histone tails are often preferentially associated with either transcriptionally active or repressed genes. To characterize the histone modifications on the inhibin α promoter, ChIP for active and repressive histone marks was performed using primary cells isolated from rat ovaries at time points corresponding to high inhibin α expression (PMSG 48 h, preovulatory cells), acute repression (PMSG 48 h /hCG 18 h, ovulatory cells), and longer-term repression (PMSG 48 h/hCG 6 d, luteal cells).

Figure 6A shows the relative inhibin α mRNA levels at these time points, confirming the down-regulation of inhibin α mRNA after hCG treatment in vivo. The reduced extent of repression seen in the ovulating and luteal cells compared with the ovulatory follicles and corpora lutea (Fig. 2A) is likely due to heterogeneity in the cell populations, which are mechanically dispersed from the intact ovary. A schematic of the primer locations used in the ChIP assays is shown in Fig. 6B. Primers within the first exon were also used, because some of the histone modifications are preferentially localized within the first exons and coding regions of other genes (40).

Fig. 6.

Histone modifications within the inhibin α proximal promoter and first exon change as gene expression is repressed. A, qRT-PCR for inhibin α was performed on cells isolated from ovaries from gonadotropin-stimulated immature female rats at PMSG 48 h (preovulatory cells), PMSG 48 h/hCG 18 h (ovulatory cells), and PMSG 48 h/hCG 6 d (luteal cells); (n = 3 animals per time point). B, Schematic of the regions of the inhibin α promoter and exon 1 amplified in the ChIP assay qPCR; black arrows indicate primer position. C, ChIP was performed in ovarian cells using an antibody against H3K27me3. D, ChIP was performed for H3K9me3 in preovulatory, ovulatory, and luteal cells using primers specific to the inhibin α proximal promoter. E, ChIP was performed for H3K4me3 in each cell type, and primers specific to the first exon of inhibin α were used in the qPCR. F, ChIP was performed for all three histone modifications in liver tissue, which has low inhibin α expression. Primers to the proximal promoter were used to amplify immunoprecipitated DNA from H3K27me3 and H3K9me3 antibodies, whereas primers to the first exon were used to amplify DNA immunoprecipitated with H3K4me3 antibody.

Trimethylation of histone H3 lysine 27 (H3K27me3) and histone H3 lysine 9 (H3K9me3) have been associated with repressed genes. To determine the enrichment of H3K27me3 and H3K9me3 on the inhibin α promoter in the three cell types, ChIP assays were performed and revealed an increase of H3K27me3 and H3K9me3 on the proximal promoter in ovulatory cells, with further enrichment in luteal cells compared with the preovulatory cells (Fig. 6, C and D). Conversely, H3K4 trimethylation (H3K4me3) is generally linked to active genes and is often present in the proximal promoter and coding regions of these genes (40). ChIP showed a decrease in H3K4me3 levels within the first exon of inhibin α in the ovulatory and luteal cells compared with the preovulatory cells (Fig. 6E). Importantly, we observed an increase in repressive marks and a decrease in the activating mark in ovulatory cells, indicating that these changes in histone modification precede the changes in DNA methylation. Consistent with these histone modifications reflecting levels of inhibin α gene expression, the repressive histone marks were highly enriched and the active mark was suppressed in the liver, where inhibin α mRNA is not expressed (Fig. 6F).

The de novo methyltransferase DNMT3a is expressed throughout the peri- and postovulatory period and is recruited to the inhibin α gene

It has recently been shown that histone H3 tails, particularly those deficient in K4 methylation, can directly bind and activate the de novo methyltransferase DNMT3a to subsequently methylate target DNA (41). Because we observed a decrease in inhibin α H3K4me3 in ovulatory and luteal cells, this might be expected to be an environment that would recruit DNMT3a to methylate the promoter. We examined the expression of DNMT3a in preovulatory follicles, ovulatory follicles, and corpora lutea. DNMT3a mRNA was expressed in all three tissues without significant differences (Fig. 7A). To determine the expression and localization of DNMT3a protein, immunohistochemistry was performed on whole ovarian sections at PMSG 48 h, hCG 18 h, and hCG 6 d for inhibin α and DNMT3a. Inhibin α protein was abundantly expressed in the granulosa cells of preovulatory follicle, reduced in the ovulatory follicle, and absent in the corpus luteum (Fig. 7B). Although DNMT3a protein was also present in preovulatory follicles, protein expression persisted in ovulatory follicles and in the corpus luteum. Thus, DNMT3a is expressed in luteinizing cells at times corresponding to decreased inhibin α mRNA and protein expression and increased inhibin α promoter methylation.

Fig. 7.

De novo methyltransferase DNMT3a is expressed throughout the peri- and postovulatory period in the ovary and is recruited to the inhibin α promoter during luteinization. A, qRT-PCR was performed for DNMT3a mRNA using RNA from preovulatory follicles, ovulatory follicles, and corpora lutea (n = 3 animals per time point). B, IHC was performed on sections obtained from fixed ovaries isolated from animals at PMSG 48 h, PMSG 48 h/hCG 18 h and PMSG 48 h/hCG 6 d for DNMT3a and inhibin α. The panels shown are adjacent sections from the same ovary. Preovulatory follicles, ovulatory follicles, and corpora lutea are indicated on each image by POF, OF, and CL, respectively. C, ChIP assays were performed in primary cells isolated from ovaries at PMSG 48 h, PMSG 48 h/hCG 18 h, PMSG 48 h/hCG 24 h, and PMSG 48 h/hCG 48 h, and qPCR was used to amplify the immunoprecipitated DNA with primers flanking the inhibin α first exon.

To determine whether DNMT3a is recruited to the inhibin promoter before the DNA methylation changes observed in the corpus luteum, ChIP was performed using cells isolated from rat ovaries at PMSG 48 h, hCG 18 h, hCG 24 h, and hCG 48 h. As shown in Fig. 7C, there was a modest increase in DNMT3a binding at hCG 18 h and hCG 24 h within the first inhibin exon, which was reduced after 48 h of hCG in vivo. Therefore, the DNMT3a recruitment occurs at approximately the same time that H3K4me3 depletion occurs in ovulatory cells, and this precedes the CpG DNA methylation that occurs on the inhibin promoter in the corpus luteum.

Discussion

Inhibin α is a key ovarian hormone that must be precisely regulated to maintain normal ovarian cyclicity and function. Transgenic mice overexpressing inhibin α develop ovarian pathologies (16, 17). Likewise, inha^−/− mice also have abnormal ovarian phenotypes. They exhibit precocious follicular growth and development (42) and develop sex cord stromal tumors at puberty (43). Therefore, defining the mechanisms by which inhibin α is regulated is key to understanding how ovarian development and female fertility are achieved. In this study, we found that the rat inhibin α proximal promoter is subject to changes in DNA methylation and histone modifications that correlate with the differing gene expression levels observed in the preovulatory follicle and corpus luteum.

A previous study showed that there is a decrease in methylation of the inhibin α promoter that accompanies increased gene expression in the mouse testis from 3–6 wk of age (44). Studies have also associated inhibin α proximal promoter methylation with decreased inhibin α expression in prostate (34, 35) and gastric cancers (36). Treatment of prostate and gastric cancer cell lines with 5-aza-deoxycytidine reactivated inhibin α expression, suggesting the DNA methylation functionally contributes to the observed repression (34, 36). Based on the precedence that the inhibin α promoter can be differentially methylated in other tissues, we examined promoter methylation in three different structures within the ovary that display varying levels of inhibin α expression: the preovulatory follicle, in which inhibin α gene expression is high; the ovulating follicle, in which inhibin α is undergoing acute repression in response to the LH surge; and the corpus luteum, in which long-term repression or silencing has been established.

We found that repression of inhibin α gene expression in the corpus luteum is associated with increased proximal promoter DNA methylation compared with preovulatory follicles in which inhibin gene expression is higher. Interestingly, the promoter was largely unmethylated in rat ovulating follicles where inhibin α expression is already significantly reduced compared with the preovulatory state. Therefore, the acute repression in response to the LH surge that occurs in the ovulating follicle is likely achieved by mechanisms that do not involve DNA methylation. This observation is reminiscent of the repression of aromatase in the bovine ovary (45). Aromatase, an enzyme that catalyzes the rate-limiting step in the conversion of androgens to estrogen, has a proximal promoter containing several regulatory elements common to inhibin α, including a CpG-containing CRE (46). Hypermethylation of the aromatase PII promoter occurs in bovine luteinized granulosa cells in which the gene is silenced, but not in late preovulatory follicles during the initial repression phase in response to the LH surge (45, 47). Hypermethylation of the aromatase PII promoter has also been observed in the corpus luteum of the buffalo (48), and DNA methylation has also been attributed to controlling tissue-specific aromatase expression in other tissues and cell types (49–51).

There is a growing body of literature describing promoter DNA methylation as a normal regulatory mechanism of genes in response to hormones and other signals (52). The cytochrome P450 27B1 promoter becomes increasingly methylated after vitamin D-induced repression and then demethylated as the gene is reactivated by PTH (31). The brain-derived neurotrophic factor proximal promoter is also demethylated in mouse cortical neurons upon depolarization, resulting in reactivation of the gene (32). These and other studies demonstrate that DNA methylation and demethylation can occur on relatively short time scales to attenuate gene expression (53). Therefore, it is clear that in some cases DNA methylation changes are associated with dynamically regulated genes and may contribute to tissue- and cell-specific gene expression (54–59).

The inhibin α CRE was a commonly methylated site in the promoter, and we demonstrated that methylation of this site can inhibit CREB from binding in vitro and in vivo. Methylation of CRE has been documented for other promoters, and this methylation corresponded with a diminishment of CREB binding (32, 50, 60–62). Although CREB does not interact directly with the cytosine that becomes methylated within the CRE, the methyl groups have been shown to create a stiffness of the DNA by reducing the conformational space and causing the DNA to fold around the hydrophobic methyl groups (63). Although our data suggest that methylation of the CpG within the CRE could keep inhibin α repressed in the corpus luteum by blocking CREB binding, it is also possible that altered localization of phospho-CREB in luteal cells contributes to preventing reactivation of inhibin α expression. Although CREB is localized to the nuclei of luteal cells, CREB phosphorylated at serine 133 has been shown to be mostly cytoplasmic. It is unclear whether phosphorylation of nuclear CREB occurs at other residues in luteal cells or the phosphoserine 133 epitope was masked by protein interactions in the luteal cell nuclei (38).

A recent study showed that DNMT3a was highly up-regulated by PR-B overexpression and R-5020 treatment in primary mouse granulosa cells, suggesting it is controlled by PR signaling (64). Because PR mRNA is highly up-regulated in response to the LH surge (65), DNMT3a is a good candidate for methylating the inhibin α gene during the periovulatory period. Additionally, we have shown H3K4me3 is decreased when inhibin α is repressed before DNA methylation occurs, and it is possible this reduction of H3K4 methylation could play a role in the recruitment of DNMT3a (41) to methylate the promoter in the corpus luteum. Our results reveal a modest recruitment of DNMT3a to the first exon at 18–24 h after hCG administration, times at which H3K4me3 is decreased. Although the recruitment was not robust, DNMT3a binding is likely transient and/or dynamic, and thus may not be fully captured by ChIP at the time points analyzed.

Many genes, including inhibin β_A and β_B (6), the FSH and LH receptors (66), steroidogenic acute regulatory protein (67), and cytochrome P450 side chain cleavage (CYP11A1) (68), undergo a dramatic switch in expression during ovulation and luteinization, which could also be controlled by epigenetic changes. There is emerging evidence consistent with this idea. Histone H3K9 dimethylation has been shown to decrease on the steroidogenic acute regulatory protein promoter in primary mouse granulosa cells after hCG administration, concomitant with an increase in gene expression (69). DNA methylation and histone modifications have been shown to contribute to LH receptor expression in nonovarian cell lines, but it remains to be seen whether these events contribute to gene expression changes in primary ovarian cells (70, 71). The use of primary cells and tissue in our study was aimed at avoiding epigenetic events that could be consequences of in vitro culture and immortalization (72, 73).

Based on work from our laboratory and others, we propose the following model for the down-regulation of inhibin α gene expression in the rat ovary. The LH surge initiates the rapid induction of transiently expressed factors, including ICER (25, 26) and C/EBPβ (27), which block CREB from binding to the CRE, resulting in acute repression of inhibin α transcription within a few hours. C/EBPβ can interact with histone-modifying enzymes (74) and chromatin-remodeling complexes (75); therefore, it is likely that C/EBPβ, other repressive transcription factors, and corepressors recruit histone-modifying enzymes that result in increased promoter H3K27me3 and H3K9me3 and decreased H3K4me3 by the time ovulation occurs. These histone modifications allow for the recruitment of DNMT3a to methylate the promoter as luteinization progresses. After luteinization, increased DNA methylation maintains inhibin α repression, in part, by limiting access of transcription factors such as CREB to the promoter. The promoter is further enriched with H3K27me3 and H3K9me3 at this time, whereas H3K4me3 remains reduced. Thus, the inhibin α promoter undergoes a series of complex chromatin-related changes before, during, and after ovulation to appropriately control inhibin expression both temporally and spatially.