ERβ- and Prostaglandin E2-Regulated Pathways Integrate Cell Proliferation via Ras-like and Estrogen-Regulated Growth Inhibitor in Endometriosis

In endometriosis, stromal and epithelial cells from the endometrium form extrauterine lesions and persist in response to estrogen (E2) and prostaglandin E2 (PGE2). Stromal cells produce excessive quantities of estrogen and PGE2 in a feed-forward manner. However, it is unknown how estrogen stimulates cell proliferation and survival for the establishment and persistence of disease. Previous studies suggest that estrogen receptor-β (ERβ) is strikingly overexpressed in endometriotic stromal cells. Thus, we integrated genome-wide ERβ binding data from previously published studies in breast cells and gene expression profiles in human endometriosis and endometrial tissues (total sample number = 81) and identified Ras-like, estrogen-regulated, growth inhibitor (RERG) as an ERβ target. Estradiol potently induced RERG mRNA and protein levels in primary endometriotic stromal cells. Chromatin immunoprecipitation demonstrated E2-induced enrichment of ERβ at the RERG promoter region. PGE2 via protein kinase A phosphorylated RERG and enhanced the nuclear translocation of RERG. RERG induced the proliferation of primary endometriotic cells. Overall, we demonstrated that E2/ERβ and PGE2 integrate at RERG, leading to increased endometriotic cell proliferation and represents a novel candidate for therapeutic intervention.

The estrogen-dependent persistence of extrauterine lesions in endometriosis causes chronic pelvic pain, reduced fertility, and decreased quality of life in women (1, 2). Endometriosis affects 6%–10% of women of reproductive age and poses a heavy health care economic burden, with treatment, surgical interventions, and hospitalizations associated with the disease totaling $22 billion annually in the United States (3). Surgical intervention is rarely curative and multiple surgeries are often necessary due to recurrence of disease and to address chronic pain (4, 5). Currently medical therapies for endometriosis are solely temporizing and focus on either minimizing the effects of estrogen or ameliorating the accompanying inflammation (6). Thus, identifying the cellular, genetic, and epigenetic basis for estrogen action in endometriosis is important for the development of targeted and effective therapies for this disease.

In endometriosis, the importance of estrogen produced during ovulatory cycles is iterated by the risk factors associated with the disease, which consist of prolonged exposure to estrogen either through early menarche or late menopause (7). At the molecular level, estrogen activates genes that increase cell proliferation as well as signaling pathways that result in cell survival, such as the phosphatidylinositol 3-kinase/Akt and serum- and-glucocorticoid-regulated kinase-3 pathways (8–10). In endometriosis, estrogen action is directly linked to inflammation, which is demonstrated by the estrogen-mediated induction of cytokine expression in endometriotic stromal cells (11). Additionally, the inflammatory milieu of the disease, characterized by elevated prostaglandin E2 (PGE2), directly activates estradiol (E2) synthesis in endometriotic cells via steroidogenic factor 1/CYP19A1 activation (12–14). PGE2 also activates signaling pathways that increase endometriotic cell survival (15). Thus, proinflammatory factors aimed at eliminating endometriotic lesions instead exacerbate the disease by inducing hormone synthesis and stimulating prosurvival signaling pathways, resulting in a feed-forward mechanism that promotes endometriotic lesion survival. Accordingly, inflammatory signals via PGE2 and E2 signaling are major regulators of disease in endometriosis. It remains unknown, however, how E2-dependent proinflammatory signals are integrated in endometriosis, but we predict that altered estrogen receptor function underlies part of this mechanism.

Despite extensive research, it is still unclear how the estrogen receptors jointly mediate estrogen's effects in endometriosis. Previous studies in stromal cells derived from ovarian endometriosis identified that a hypomethylated estrogen receptor (ER)-β promoter region resulted in remarkably elevated ERβ mRNA and protein expression relative to the normal endometrium (16). In addition, the eutopic endometrium of women with endometriosis have elevated ERβ expression when compared with the endometrium of healthy women (17, 18), suggesting that high levels of ERβ in the endometrium may predispose women to endometriosis. Further studies demonstrate that in endometriotic stromal cells, ERβ transcriptionally represses ERα (19), indicating that elevated ERβ confers a unique estrogen response mechanism that may contribute to disease progression. Mechanistically, this is supported by the observation the two estrogen receptors possess highly conserved DNA-binding domains (95% identity), but each has two independent and highly unique activation function (AF) domains. The AF-1 (20% identity) and AF-2 (30% identity) domains are involved in coregulator recruitment and dictate transcriptional activation or repression (8).

Genome-wide binding studies demonstrated that although ERα and ERβ share a large number of transcriptional targets, they also bind independently to other DNA regions, resulting in distinct binding patterns (20–22). On average, ERβ binds more closely to transcription start sites (TSS) and to GC-rich regions, whereas ERα binds more distally from TSSs and is enriched in AT-rich regions (21). The difference between the ERα/ERβ binding patterns prompted us to ask whether unique ERβ transcriptional targets in endometriotic tissues mediate the pathogenesis/pathoprogression of the disease. We hypothesized that unique ERβ-regulated genes in endometriotic stromal cells are responsible for the E2-driven mechanisms characteristic of endometriosis and would likely represent novel targets for development of new therapeutics. We interrogated the interactions between gene expression profiles from endometriotic stromal cells with previously published genome-wide ERβ binding data sets (20, 22) to identify Ras-like and estrogen regulated growth inhibitor (RERG) as a novel ERβ-regulated gene in endometriosis. We then uncovered a novel mechanism showing how RERG integrates estrogenic and inflammatory signaling pathways to promote proliferation.

Materials and Methods

Subjects, tissue sample processing, and cell culture

Tissues were obtained with informed consent from the Northwestern Memorial Prentice Women's Hospital under a protocol approved by the Institutional Review Board at Northwestern University (Reproductive Tissue Registry STU00018080). A total of 53 endometriosis specimens (listed in Supplemental Table 1) were obtained from women with a mean age of 32.9 years. Most of the tissues were obtained from women in the proliferative phase of the menstrual cycle (n = 34), with another 12 samples from the secretory phase and seven samples with an unknown phase. For the normal endometrium tissues, a total of 28 samples from human subjects were used, with a mean age of 40.2 years; the majority were from the proliferative phase (n = 20), and seven samples were from the secretory phase, and one was unknown.

All experiments were performed in freshly isolated tissues or stromal cells isolated from the fresh tissues and used in experiments within three passages after the initial culturing. For all experiments, we compared (ectopic) endometriotic tissues meticulously dissected from ovarian endometriomas by a pathologist with the (eutopic) endometrium obtained from women without clinical or pathological diagnosis of endometriosis.

Endometriotic tissue was obtained after surgery from women undergoing surgical removal of endometriosis, and normal endometrial tissue was obtained from women undergoing hysterectomies due to benign conditions and not endometriosis. For the ectopic endometriosis tissues, pathological confirmation of endometriosis was obtained from clinical diagnosis and by pathology examination. For experiments conducted in stromal cells, primary cultures were established by isolating stromal cells from normal endometrium (NoEM) and diseased endometriotic tissue (E-Osis) using 0.2 mg/mL deoxyribonuclease and 5 mg/mL collagenase digestion followed by a second digestion in 0.2 mg/mL deoxyribonuclease, 5 mg/mL collagenase, 1 mg/mL pronase, and 2 mg/mL hyaluronidase digestion in Hanks' buffered saline solution. E-Osis cells were cultured in DMEM/F-12 (Life Technologies) supplemented with 10% fetal bovine serum (FBS) and antibiotic/antimycotic solution (VWR). Human embryonic kidney-293 (HEK-293)-T cells were cultured in DMEM (Life Technologies) supplemented with 10% FBS and antibiotic/antimycotic solution.

Other compounds and reagents

Estradiol, diarylpropionitrile (DPN), and propylpyrazoletriol (PPT) (resuspended in ethanol) were purchased from Tocris and used at a final 100 nM concentration for 1- to 6-hour treatments. PGE2 was purchased from Cayman, resuspended in ethanol and used at a final concentration of 50–100 nM. 8-Bromoadenosine-3′,5′-cyclic monophosphate, was purchased from Axxora, Inc (resuspended in 1× PBS) and cells were treated at a concentration of 100 mM. Recombinant TNF was purchased from Sigma (resuspended in 1× PBS) and a final concentration of 10 ng/mL was used for cell culture treatments. Phosphatase inhibitor cocktail (catalog number P5726) and protease inhibitor cocktails (catalog number P8340) were purchased from Sigma and diluted 1:100 in the appropriate lysis buffer. Actinomycin D was purchased from Tocris and resuspended in dimethylsulfoxide (DMSO). All antibodies are listed in Supplemental Table 2.

Microarray analysis

Stromal cells from six normal endometrial tissues and six ovarian endometriosis tissues were obtained from the proliferative (n = 3) or secretory (n = 3) phases of the menstrual cycle and digested as described above. The cells were cultured for one passage and then harvested, followed by RNA isolation and quantification. RNA was then transcribed to cDNA, converted to biotinylated cRNA, fragmented, and hybridized to HumanHT-12 v4 expression BeadChip arrays from Illumina, which provides coverage of approximately 48 000 genes and expressed sequence tags (Illumina, Inc). Raw signal intensities of each probe were obtained using data analysis software (Beadstudio; Illumina) and imported to the lumi package of Bioconductor for data analysis. Before transformation and normalization (23–25), A/P call detection was performed based on the detection P value. A total of 16 771 of 48 802 probes with less than P = .01 were considered as valid signals. Differentially expressed genes were identified using an ANOVA model with empirical Bayesian variance estimation (26). Initially, genes were identified as being differentially expressed on the basis of a statistically significant (raw value of P < .01), and 1.5-fold change (up or down) in expression level in endometriosis RNA samples compared with normal endometrial samples (Gene Expression Omnibus accession number GSE58178).

Genome-wide ERβ binding data sets

The ChIP-Seq and Chip-on-Chip peaks from Zhao et al (22) and Grober et al (20) were imported by GenomicFeatures and rtracklayer packages of Bioconductor (27) to find overlapping differentially expressed genes from the microarray analysis. Functional analysis of the identified overlapped genes were performed by the GeneAnswers package of Bioconductor (28) to discover the potential involved functions and pathways between endometriosis and normal endometrial samples. One thousand four hundred fifty-seven peaks were obtained from Zhao el al (22). ERβ ChIP-on-ChIP, of which 112 mapped within 5 kb upstream or 1 kb downstream of an ERβ binding site. Nine thousand seven hundred two ERβ binding sites were obtained from the study by Grober et al (20), of which 1684 had a gene mapping −5 kb to +1 kb of the binding site. All binding data, gene expression profiles, and overlapping ERβ binding sites are included as Supplemental Tables.

Real-time quantitative PCR

RNA was prepared from the samples according to the RNeasy kit (QIAGEN). One microgram of RNA was used to reverse transcribe into cDNA using qScript cDNA Supermix (Quanta; 101414-106). SYBR Green (Life Technologies; 4368708) and primers were used to amplify genes of interest. Gene expression data were normalized to the GAPDH or TBP genes. All primer sequences or catalog numbers are listed in Supplemental Table 2. Samples were processed in the 7900HT Fast real-time PCR system and data were collected with SDS 2.3 software (Applied Biosystems).

Small interfering RNA (siRNA) knockdown

Reverse transfections were used for all E-Osis and NoEM transfections. Specifically, 40 nM ON-TARGET Plus nontargeting siRNA number 1 (Thermo Scientific; D-001810–01-05), ON-TARGET PLUS human RERG (85004) siRNA (Thermo Scientific; J-008204–09-0010), or 100 nM Silencer Select siESR2 (Ambion; s4827) or Silencer Select nontargeting siRNA number 1 (Ambion; 4390843) were diluted in Opti-MEM reduced serum media with RNAiMAX (Life Technologies) in a 1:5 (siRNA to RNAiMAX ratio). Complex formation was carried out for 15 minutes at room temperature and then added to serum-free/antibiotic-free DMEM/F-12 media overnight. Media were changed the following day to complete FBS- and antibiotic-containing media. The cells were collected for mRNA or protein 48–96 hours after transfection.

pLenti-ERβ and pLenti-RERG-FLAG overexpression in E-Osis and NoEM cells

Constructs were generated according to the methods published by Campeau et al (29). The RERG sequence-verified cDNA construct was purchased from Open Biosystems (MHS6278–202827994) and amplified with high-fidelity Pfx polymerase (Life Technologies) using primers with Kpn1 and Not1-restriction enzyme sequences and ligated in to the pENTR4-FLAG construct (Addgene number 17423). To generate the DsRED-FLAG lentiviral construct, the DsRed gene was amplified with primers with Not1 and BamHI restriction enzyme sequences and ligated into the pENTR4-FLAG construct. After sequence verification, the constructs were recombined using LR Clonase II (Life Technologies; number 11791) into pLenti-CMV Puromycin destination vector (Addgene; number 17452), transformed into Stbl3 competent cells (Life Technologies), and plated onto ampicillin-selective LB-agar plates. The pLenti-EF1α-ERβ construct was generated in a similar manner; briefly, ESR2 cDNA construct was purchased from Origene (119216) digested and ligated into pEF1α-ENTR1A (Addgene; number 17427). After sequence verification, the construct was recombined into the pLentiCMV-GFP destination vector (Addgene; number 19732) and propagated and purified as described above.

To generate the virus, lentiviral envelope constructs were purchased from Addgene, psPAX (number 12260) pMD2.G (number 12259) and transfected along with the lentiviral overexpression construct into HEK-293T/17 cells. Briefly, 4 μg of lentiviral construct, 3 μg psPAX, and 1 μg pMD2.G were added to 850 μL OptiMEM with FuGene (1:3 DNA to FuGene ratio) and incubated for 5–15 minutes. The complex was then added to freshly trypsinized HEK-293T/17 (110⁷ cells) into a 10-cm plate in DMEM/10% FBS without antibiotics. The media were changed to complete antibiotic-containing media the following day, and viral supernatants were collected 48 and 72 hour after the transfection. Viral titers were obtained using the quantitative PCR (qPCR) lentivirus titration assay (ABM; number LV900), and cells were transduced with virus using a multiplicity of infection 10 and supplemented with 8 μg/mL Polybrene (Sigma; number H9268).

Chromatin immunoprecipitation

We performed ERβ ChIP in E-Osis stromal cells following the protocol published by Lee et al (30) and using the conditions for ERβ ChIP used by Charn et al (31). Briefly, cells were grown until 70%–80% confluency and serum starved overnight, followed by a 45-minute treatment with 100 nM E2. The cells were then fixed with one tenth the volume of formaldehyde solution (11% formaldehyde; 50 mM HEPES-KOH, pH 7.5; 100 mM NaCl; 1 mM EDA; 0.5 mM EGTA) for 10 minutes and quenched with 0.125 M glycine. Cells were lysed and chromatin was sheared by sonication for 20 cycles (10 sec on, 50 sec rest). Protein lysates were incubated with 4 μg each of an ERβ antibody mixture (CWK-F12 (31), Pierce PA1–311, GeneTex GTX70182, Calbiochem GR-40) or IgG (Sigma) overnight at 4°C. Dynal Beads (Life Technologies) were used to capture the protein/DNA/antibody complexes for 4 hours on a rocker at 4°C. Dynal Beads were then washed with radioimmunoprecipitation assay buffer (50 mM HEPES-KOH, pH 7.5; 500 mM LiCl; 1 mM EDTA; 1% Nonidet P-40 (NP-40); 0.7% Na-deoxycholate) and protein/DNA complexes were eluted (50 mM Tris-HCl, pH 8; 10 mM EDTA; 1% sodium dodecyl sulfate) followed by crosslink reversal overnight at 65°C. After DNA purification, quantitative RT-PCR was performed on INPUT and ChIP'ed DNA using primers spanning 2.5 kb upstream and downstream of the RERG TSS. The regions used to validate ERβ binding were chosen based on the published binding sites for RERG and used to design primers upstream and downstream of the identified binding site. Data were quantified as fold enrichment relative to IgG and normalized to vehicle treatment. Experiments were replicated in E-Osis stromal cells from at least three subjects.

Immunoblot analysis

Protein from stromal cells was collected using M-PER (Pierce) lysis buffer and quantified using the bicinchoninic assay protein assay (Pierce) as indicated by the manufacturer's instructions. All antibodies are listed in Supplemental Table 2. At least 20 μg of protein was diluted with reducing 4× lithium dodecyl sulfate sample buffer (Life Technologies), electrophoresed on 4%–12% Novex Bis-Tris polyacrylamide precast gels (Life Technologies), and transferred onto polyvinyl difluoride or nitrocellulose membranes. The membranes were blocked with 5% milk in Tris-buffered saline and 0.2% Tween 20 and probed for each specific antibody shaking overnight at 4°C. Horseradish peroxidase-conjugated secondary antibodies (Cell Signaling) were diluted in 5% milk at 1:5000 and incubated for 1 hour shaking at room temperature. The membranes were then washed four times in Tris-buffered saline (TBS) and 0.2% Tween 20 and once in TBS for 10 minutes each time followed by incubation with chemiluminescence reagent for 5 minutes (Femto from Pierce or Luminata Crescendo from Millipore). Film was developed in a Konika Minolta developer.

Immunofluorescence

A total of 20 000 E-Osis cells were plated on glass coverslips in 24-well plates and treated with vehicle or 100 nM E2 and 50 nM PGE2. After treatment, cells were fixed with 4% paraformaldehyde in 1× PBS for 10 minutes, followed by permeabilization with 0.2% Triton X-100. Blocking and antibody dilutions were performed in 1% milk + 1% normal donkey serum (Jackson ImmunoResearch). Primary antibody incubation was performed in a humidified chamber at room temperature for 1 hour followed by three 5-minute washes in TBS. Secondary antibody (1:250) was added in the dark for 1 hour followed by incubation with 4′,6-diamidino-2-phenylindole (DAPI) in TBS for 5 minutes, and two 5-minute washes in TBS. Coverslips were mounted on slides with ProLong Gold antifade media (Life Technologies) and imaged on a Leica CTR 5000 microscope.

Immunohistochemistry

Frozen or paraffin-embedded sections were obtained of endometriotic tissue from women with pathologically confirmed ovarian endometriosis and of normal endometrium tissue from women without endometriosis. Paraffin-embedded sections were deparaffinized and rehydrated with serial washes in xylene and 100%, 90%, 80%, and 60% ethanol followed by washing in deionized water. Sections underwent antigen retrieval, blocking in 5% normal donkey serum and incubation in primary antibody for 1 hour at room temperature in a humidified chamber. Secondary antibody was added for 1 hour at room temperature in the dark, followed by incubation with DAPI in 1× TBS (5 ng/mL) and two washes in TBS. Tissue sections were mounted on coverslips using ProLong Gold antifade mounting media from Life Technologies. Frozen sections were stained using the same protocol except that fixation was performed in 4% paraformaldehyde in 1× PBS followed by permeabilization in 0.2% Triton X-100, and blocking in 5% normal donkey serum.

Cell proliferation assays

To quantify cell proliferation after RERG siRNA knockdown, we used Trypan Blue exclusion and a 5-bromo-2′-deoxyuridine (BrdU) ELISA assay (Roche; number 11647229001). For the cell counting experiments, 50 000 cells were plated in triplicate in a six-well plate at the time of transfection and were harvested and counted 24, 48, and 72 hours after RERG knockdown. For the BrdU ELISA, 1000 cells were plated in a 96-well plate at the time of transfection. BrdU incorporation was then allowed to proceed for 4 hours at 37°C, followed by DNA denaturation, BrdU antibody labeling, and detection at 492 nm/370 nm absorbance. BrdU incorporation was then quantified relative to the cells transfected with a nontargeting siRNA.

Immunoprecipitation

E-Osis stromal cells were lysed in 0.5% NP-40 lysis buffer (50 mM Tris-HCl, 250 mM NaCl, 5 mM EDTA, 50 nM NaF, 1 mM Na₃VO₄, 0.5% NP-40, 0.0.2% NaNa₃) supplemented with protease and phosphatase inhibitors. FLAG-conjugated magnetic beads were used to immunoprecipitate RERG-FLAG and DsRed-FLAG constructs. One milligram of protein was used in each immunoprecipitation, beads were washed five times with TBS wash buffer, and RERG-FLAG- or DsRed-FLAG-associated proteins were eluted with 100 μg/mL FLAG peptide. For all other immunoprecipitations, antibodies or IgG was covalently conjugated to agarose beads and incubated with 1 mg of protein overnight at 4°C in a rocker as indicated by the manufacturer's instructions (direct immunoprecipitation kit; Pierce). Elution was performed in acidic glycine (50 mM glycine, pH 2.8) or under reducing conditions with lithium dodecyl sulfate sample buffer with 10% β-mercaptoethanol.

Statistical analysis

All experiments were performed in three or more subjects and analyzed using Student's t tests or ANOVA analyses with Tukey's multiple comparison posttests using GraphPad versions 5–6.

Results

Identification of ERβ-regulated genes in endometriotic stromal cells

A total of 624 differentially expressed genes were identified by comparing the mRNA profiles of the normal endometrial stromal cells isolated from six healthy women (NoEM) with stromal cells isolated from the endometriomas from six women (E-Osis) (Figure 1A and Supplemental Table 3). These 624 differentially expressed genes were then correlated with the previously published ERβ ChIP-on-ChIP (22) and ChIP-Seq data sets (20) obtained using MCF7 breast cancer cells. Using this strategy, we identified a total of 70 genes that were differentially regulated in NoEM and E-Osis and that also had an ERβ binding site (Figure 1B and Supplemental Table 4). Gene ontology analysis (Supplemental Table 5) of these 70 genes revealed that regulators of protein phosphorylation (59.4%, P = .0003) and apoptosis (13%, P = .01) were overrepresented in this group. We selected 12 of these genes based on previous studies showing their involvement in estrogen-dependent processes or in the regulation of cell proliferation or apoptosis (32–37) and used real time PCR to quantify their mRNA levels using NoEM (n = 8) and E-Osis (n = 8) tissues from a separate set of subjects and verified the results of the microarray (Supplemental Table 6).

Of the ERβ gene targets identified, we were particularly interested in RERG because of its striking overexpression in E-Osis compared with NoEM (31-fold, P = .01; Figure 1C). Although the microarray indicated that RERG was increased by 2.3-fold in E-Osis compared with NoEM (Supplemental Table 4), real-time PCR validation in a separate set of patient samples indicated a striking overexpression of 31-fold, P = .01 (Figure 1C). The discrepancy between the microarray and real-time PCR validation can be the result of the different set of patient samples used or may reflect the higher sensitivity and accuracy of real-time PCR compared with microarray. In Figure 1, D and E, we verified that the elevated expression of RERG was also observed at the protein level. Densitometric analysis of the immunoblots showed that the levels of RERG were 13.17-fold higher (P = .007) in stromal cells isolated from E-Osis compared with NoEM (Figure 1E). We also analyzed the expression of RERG in whole NoEM and E-Osis tissues using immunohistochemistry (Figure 1F and Supplemental Figure 1). Confirming our previous results, the immunoreactive RERG (Figure 1F) was readily detected in E-Osis but absent in NoEM tissues.

E2 regulates expression of RERG in E-Osis stromal cells

To verify that E2 regulates RERG in endometriosis, we treated E-Osis cells with 10⁻⁷ M E2 and quantified RERG mRNA and protein expression. E2 treatment modestly increased mRNA levels of RERG (1.55-fold, P = 0.032) within 2 hours (Figure 2A). Although the induction was weak and observed only after 2 hours of E2 treatment, we did detect a strong and significant increase of RERG protein after the E2 treatment. Significantly increased RERG protein levels were observed within 6 hours for RERG (1.54-fold, P=0.05) (Figure 2, B and C). Incubation of E-Osis cells with the RNA polymerase II inhibitor actinomycin D (ActD) prior to E2 treatment blocked E2-stimulated RERG expression (Figure 2D), suggesting that E2 regulates the transcription of RERG. The inhibition of the E2 induction of RERG protein by the RNA polymerase II inhibitor actinomycin D strongly suggests that E2-dependent stimulation of RERG expression is transcriptionally regulated.

Because the increased ratio of ERβ to ERα in E-Osis is thought to drive the distinct estrogen-mediated effects in endometriosis (16, 18, 38–40), we predicted that the ERβ-selective agonist DPN would strongly induce the expression of RERG in E-Osis. Compared with vehicle-treated cells, after a 2-hour treatment, DPN robustly induced RERG expression (Figure 2E). Although not statistically significant, there was a tendency for DPN to more strongly induce the expression of RERG than E2 or the ERα-selective agonist PPT in E-Osis stromal cells (Figure 2, E and F). Conversely, DPN treatment in NoEM did not significantly induce RERG expression (Figure 2, G and H). Our results indicate that the estrogen receptor agonists DPN and PPT up-regulate RERG expression in in E-Osis but not in NoEM.

ERβ regulates the expression of RERG in E-Osis stromal cells

We performed ERβ ChIP-qPCR in E2-treated E-Osis stromal cells to verify the enrichment of ERβ at the RERG promoter region (Figure 3A). Transcription factor binding element analysis was performed to identify the RERG promoter region most likely to contain estrogen receptor element (ERE) binding motifs. We identified a palindromic ERE with 91% homology to a consensus ERE in the region 1 kb downstream of the RERG TSS (Supplemental Figure 2). ERβ-ChIP was initially performed in E-Osis stromal cells 45 minutes, 3 hours, and 24 hours after the 10⁻⁷ M E2 treatment (not shown); the maximal ERβ binding at the RERG promoter region was observed after 45 minutes of E2 treatment. ERβ-ChIP was performed in E-Osis stromal cells from additional subjects treated with or without 10⁻⁷ M E2 for 45 minutes, and enrichment was measured via qPCR using tiled primers spanning a 5-kb region that included the predicted binding site by ChIP-on-ChIP and centered on its TSS. After E2 treatment, we observed significant ERβ enrichment (2.74-fold, P < .001) at the predicted ERE located 1 kb downstream of the RERG TSS (Figure 3A).

We used siRNAs to silence the expression of ERβ in E-Osis cells to further demonstrate ERβ-mediated regulation of RERG in E-Osis. After siRNA-mediated knockdown of ERβ, a corresponding decrease was observed in RERG protein expression (Figure 3B). Given that the ERβ knockdown in Figure 3B was performed in the absence of a ligand, these results suggest that ERβ can regulate RERG in a ligand-independent manner. Densitometric analysis was performed to verify the response to ERβ knockdown across different subjects (n = 3; Figure 3, C and D); we confirmed a significant decrease in ERβ (0.57-fold, P = .01) and RERG (0.48-fold, P = .002). We also overexpressed ERβ in NoEM stromal cells with a lentiviral system that drives ERβ expression from the EF1α promoter (Figure 3, E and F) (29). Lentiviral ERβ overexpression in NoEM stromal cells (0.26 vs 0.39 A.U., P = .022, Figure 3F) resulted in a small but significant increase in RERG protein expression when compared with control cells overexpressing green fluorescent protein (GFP) (0.10 vs 0. 22 A.U., P = .028, Figure 3G). These results demonstrate that ERβ binds the RERG promoter regulatory region in E-Osis to induce its gene expression.

PGE2 and E2 regulate RERG expression and nuclear localization

We sought understand the effects of the proinflammatory molecule, PGE2, on the regulation of RERG in E-Osis stromal cells. We observed a dose-dependent increase in RERG protein expression in response to PGE2 treatment (Supplemental Figure 3A). We used 50 nM PGE2, the lowest concentration that can induce RERG expression for all subsequent experiments. To assess whether PGE2 regulation of RERG was controlled by transcription, we incubated E-Osis cells with the RNA polymerase II inhibitor ActD prior to PGE2 treatment and measured RERG protein levels. Figure 4, A and B, shows that ActD did not abolish PGE2-induced RERG protein levels, suggesting that PGE2 regulates RERG by a posttranscriptional process. In fact, preincubation of HEK-293T cells overexpressing FLAG-RERG with the protein kinase A (PKA) inhibitor H89 decreased the immunodetection of RERG-FLAG by the phospho-PKA substrate antibody (Figure 4C). Thus, RERG is phosphorylated in response to PGE2 via a PKA-activated pathway.

Small GTPases such as RERG are dynamically regulated by various posttranslational modifications as well as by their association with effector proteins that facilitate GTP/GDP exchange (41). SCANSITE analysis (Massachusetts Institute of Technology) identified putative RERG phosphorylation sites on three serine residues (Ser19, Ser181, and Ser 182) and on one threonine residue (Thr 183) (Supplemental Figure 3B). We also identified a nuclear localization signal in the C-terminal domain of RERG that overlaps with three of the putative phosphorylation residues (S181/S182/T183) (Supplemental Figure 3B). We first determined that although individual E2 and PGE2 treatments increased RERG expression, combined E2+PGE2 did not synergize RERG expression in E-Osis stromal cells (Figure 4D). To answer whether the E2- or PGE2-mediated expression of RERG also led to its nucleocytplasmic translocation, we performed confocal microscopy of E-Osis stromal cells after treatment. Figure 4E demonstrates that compared with vehicle, E2, PGE2, and E2+PGE2 treatments led to pronounced nuclear translocation of RERG. These results demonstrate the interplay of estrogenic and proinflammatory pathways on RERG expression.

RERG regulates proliferation of E-Osis stromal cells

We next characterized the contribution of RERG to proliferation and apoptosis in E-Osis. Immunoblotting confirmed that RERG knockdown was achieved at 48 and 72 hours after the transfection (Figure 5A). Depleting RERG in E-Osis significantly decreased total cell number relative to the cells treated with control siRNA after 48 hours and 72 hours (Figure 5B). Because the greatest effect on cell count was observed 72 hours after the transfection, we performed a BrdU incorporation assay using this time point to gauge proliferation. We observed a significant decrease in BrdU incorporation 72 hours after siRNA knockdown of RERG when compared with the nontargeting control (0.88-fold decrease, P = .0002; Figure 5C). We also observed a robust decrease in the expression of another marker of cell proliferation, proliferating cell nuclear antigen (PCNA), 72 hours after RERG knockdown in E-Osis stromal cells (Figure 5D). Densitometry analysis (Figure 5, E and F) of immunoblots of E-Osis cells from four additional subjects confirmed the statistically significant changes in PCNA (0.71-fold, P = .03) and RERG (0.40-fold, P = .002) protein expression after the RERG knockdown. To assess the role of RERG in apoptosis, we performed a new set of experiments and measured cleaved poly(ADP-ribose) polymerase expression after the RERG knockdown. We detected a modest and transient increase in cleaved poly(ADP-ribose) polymerase expression at 48 hours but not 72 or 96 hours after the transfection (Supplemental Figure 4), indicating that RERG is not a key regulator of apoptosis in E-Osis.

In summary, we identified that ERβ overexpression and proinflammatory signaling via PGE2 in endometriosis led to the overexpression of a key gene, RERG (Figure 6). We found that RERG is transcriptionally regulated in response to estrogen and posttranscriptionally modified by the proinflammatory factor, PGE2. Functionally, we found that ERβ, via RERG, contributes to the pathogenesis of endometriosis by promoting endometriotic stromal cell proliferation.

Discussion

We uncovered a novel signaling mechanism in endometriosis downstream of ERβ and PGE2 that involves the regulation of RERG. By verifying that RERG is a transcriptional target of estrogen and its receptor ERβ, we determined that RERG contributes to endometriotic cell proliferation. In our system, hormonal signaling via E2/ERβ complemented the proinflammatory actions of PGE2 and promoted the phosphorylation of RERG in human endometriotic cells. We summarize our proposed model of E2/ERβ and PGE2 regulation of RERG in Figure 6.

ERβ is up-regulated in human endometriotic tissues due to promoter hypomethylation (16), and elevated ERβ expression is observed in the eutopic endometrium of women with endometriosis, suggesting a role for this receptor in establishing the disease (17, 18). To characterize ERβ-regulated genes in endometriosis, we took advantage of the publicly available genome-wide ERβ binding sites identified in MCF-7 cells and correlated them with the differentially expressed mRNAs in E-Osis stromal cells. Although this strategy may pose difficulty for interpreting pathway analyses across the different cell types, our experimental approach demonstrated that a discovery-based strategy can successfully identify transcriptional targets in different cell types. Although our study focused on ovarian endometriotic and normal endometrium from endometriosis-free women, additional studies using tightly matched eutopic endometrium from women with endometriosis will provide a more accurate control for these experiments.

Our strategy identified 70 genes that had altered gene expression in E-Osis as well an ERβ binding site, for which we characterized one that regulated cell proliferation. Analysis of the additional 69 ERβ targets will further establish the role of ERβ in endometriosis. Interesting candidates among the remaining ERβ targets include the serum- and glucocorticoid-regulated kinase (SGK1), which regulates endometrial cell survival in response to oxidative stress, thioredoxin (TXNRD1), which encodes an enzyme involved in the regulation of intracellular reactive oxygen species, or cytochrome P450 1B1 (CYP1B1), an estradiol-metabolizing enzyme that is associated with endometriosis and increased exposure to dioxins (33, 36, 42, 43). Future studies using ERβ-ChIP-Seq in E-Osis are warranted to obtain a comprehensive data set of ERβ transcriptional targets in E-Osis.

We found that treatment with the ERβ- or ERα-specific agonists, DPN and PPT, had opposing effects on the NoEM and E-Osis cells. These results were expected, given that the major receptor in E-Osis is ERβ, whereas ERα is the predominant estrogen receptor in NoEM. Future studies using ERα- and ERβ-selective antagonists will clarify the suppressive effects that ERβ has on ERα in E-Osis. Because little is known about the molecular function of RERG, almost any information presented here is novel with respect to its biological roles in any mammalian cell type. Intriguingly, we showed that RERG is phosphorylated in response to PGE2 via a cAMP-dependent signaling pathway that involves PKA action. We also showed that RERG has a bipartite nuclear localization signal (NLS) that overlaps with the predicted phosphorylated residues. A bipartite NLS is characterized by two clusters of basic amino acids separated by a 10- to 12-amino acid linker region (44), and in certain proteins, phosphorylation within or around the NLS is a key regulatory step in nucleocytoplasmic transport (45, 46). Indeed, E2 and PGE2 increased nuclear localization of RERG, suggesting that phosphorylation/activation of RERG modulates its nucleocytoplasmic shuttling. Additional experiments, such as site-directed mutagenesis of the putative phosphorylation residues in the RERG NLS, are necessary to completely characterize this process.

RERG was initially characterized in the epithelial-derived MCF7 breast cancer cell line, in which RERG overexpression resulted in decreased cell proliferation (47). It was also shown to be a transcriptional target of ERβ in HEK-293T cells overexpressing the receptor (48). In our system using primary human stromal E-Osis cells, knockdown of endogenous RERG resulted in decreased cell proliferation. It is possible that RERG action differs in epithelial and stromal cells and that although estrogen drives proliferation and expression of RERG in E-Osis stromal cells, the effects of RERG may be different in the epithelial MCF7 cells. It is also possible that we observed different effects on cell proliferation due to the technique used to study RERG function: we observed a decrease in cell proliferation after siRNA-mediated knockdown, whereas Finlin et al (47) observed growth inhibition after overexpressing it in MCF cells using a plasmid.

The contribution of estrogen to the proinflammatory state of endometriosis has been previously demonstrated (11). In primary human cell cultures derived from endometriotic lesions, estrogen induces the expression of several cytokines and chemokines. For example, in endometriotic stromal cells estrogen increases the mRNA and protein expression of regulated on activation, normal T cell expressed and secreted (RANTES; also known as CCL5) (49) and macrophage chemoattractant protein (MCP1; also known as CCL2) (50). RANTES and MCP1 then recruit monocytes and macrophages to endometriotic lesions and further stimulate these immune cells, resulting in a feed-forward cycle that amplifies the inflammatory response. Similarly, the proinflammatory factor PGE2 stimulates the expression of aromatase in endometriotic stromal cells, increasing local estradiol synthesis and fueling endometriotic cell survival and growth (12, 51). More recently, the interplay between hormone action and inflammation was demonstrated to contribute to endometriotic cell survival. Specifically, the presence of a cleaved steroid receptor coactivator isoform induced endometriotic cell survival in response to TNF (18). Our findings establish that the link between estrogen signaling and inflammation in endometriosis occur via the ERβ- and PGE2-dependent regulation of RERG.

Overall, our results demonstrate that estrogen, via ERβ, and PGE2, via cAMP/PKA, coordinately regulates RERG in endometriosis. These results highlight the interplay between E2 and proinflammatory signaling in endometriosis and demonstrate how they converge to modulate the expression of RERG to ultimately control cell proliferation. Furthermore, our findings identify RERG as a new promising target for therapeutic intervention in endometriosis.

Acknowledgments

Author contributions included the following: D.M., S.E.B., M.T.D., P.Y., J.J.W., E.S., and M.E.P. designed the project, provided intellectual guidance, and reviewed the manuscript. D.M. wrote the manuscript and performed the experiments. M.T.D., P.Y., A.N., S.S.M., M.O., J.S.C., and C.M.E. provided intellectual guidance, performed the experiments, and reviewed the manuscript.

This work was supported by National Institutes of Health Grant R37HD038691.

Current address for C.M.E.: Department of Obstetrics and Gynecology, Gulhane Military Medical Faculty, Etlik, Ankara 06018, Turkey.

Disclosure Summary: The authors have nothing to disclose.

Abbreviations

 
• ActD

  actinomycin D

 
• AF

  activation function

 
• BrdU

  5-bromo-2′-deoxyuridine

 
• DAPI

  4′,6-diamidino-2-phenylindole

 
• DMSO

  dimethylsulfoxide

 
• DPN

  diarylpropionitrile

 
• E2

  estradiol

 
• E-Osis

  diseased endometriotic tissue

 
• ER

  estrogen receptor

 
• ERE

  ER element

 
• FBS

  fetal bovine serum

 
• GFP

  green fluorescent protein

 
• HEK-293

  human embryonic kidney-293

 
• NLS

  nuclear localization signal

 
• NoEM

  normal endometrium

 
• NP-40

  Nonidet P-40

 
• PCNA

  proliferating cell nuclear antigen

 
• PGE2

  prostaglandin E2

 
• PKA

  protein kinase A

 
• PPT

  propylpyrazoletriol

 
• qPCR

  quantitative PCR

 
• RERG

  Ras-like, estrogen-regulated, growth inhibitor

 
• siRNA

  small interfering RNA

 
• TBS

  Tris-buffered saline

 
• TSS

  transcription start site.

References