Decreased Expression of EZH2 in Granulosa Cells Contributes to Endometriosis-Associated Infertility by Targeting IL-1R2

Reactive oxygen species (ROS) accumulation or an imbalance between ROS and antioxidants induces oxidative damage of ovarian GCs, leading to anovulatory disorders such as endometriosis, polycystic ovary syndrome, and premature ovarian failure ( 49 ). Oxidative stress in the follicular microenvironment has been speculated to negatively affect folliculogenesis, oocyte maturation, ovulation, and embryogenesis ( 20 , 50 ). To explore the effect of excessive oxidative stress on histone methylation in GCs, we examined expression of the histone methyltransferase EZH2 using immunofluorescence. EZH2 was mainly located in the nuclei of GCs (COV434, SVOG, and mGCs) with faint staining in the cytoplasm; however, after 2 hours of incubation with H 2 O 2 , EZH2 expression in the cytoplasm increased ( Fig. 1A and Supplementary Fig. S1A [ 39 ]). Chromatin fractionation assay results show that H 2 O 2 -induced ROS resulted in a time-dependent cytoplasmic accumulation of EZH2 in GCs ( Fig. 1B ). Owing to the rapid degradation of EZH2 protein and our previous results, we found that 2-hour H 2 O 2 exposure was feasible to observe cytoplasmic induction of EZH2 in GCs, similar to results in other cell types ( 51 , 52 ). Western blot analysis ( Fig. 1C ) and immunofluorescence ( Fig. 1D ) confirmed decreases of EZH2 and H3K27Me3 protein expression following excessive oxidative stress in COV434, SVOG, and mGCs, whereas protein expression of phospho-EZH2 (Thr311) was significantly increased after 2-hour H 2 O 2 treatment. Phosphorylation of Thr311 (p-EZH2) disrupts the interaction between EZH2 and SUZ12, leading to attenuation of EZH2 histone methyltransferase activity ( 53 ). These results are consistent with studies implicating ROS in the regulation of histone methylation through effects on the nuclear-cytoplasmic distribution of EZH2 ( 51 ). Decreased EZH2 and H3K27Me3 expression in endometriosis granulosa cells (GCs) correlated with decreased numbers of retrieved and mature oocytes. A, Representative fluorescence microscopy images of COV434 (left panel) and SVOG cells (right panel) after exposure to 100-µM H 2 O 2 for 2 hours. EZH2 (red); DAPI staining of nuclei (blue). Scale bar = 50 µm. Magnification, 100×. B, An anti-EZH2 antibody was used to detect EZH2 expression in cells subjected to cellular fractionation. GAPDH and LaminB1 served as cytoplasmic and nuclear indicators, respectively. C, Western blot assay of COV434, SVOG, and mouse GCs (mGCs) showing time-dependent decreases of EZH2 and H3K27Me3 protein expression under 100 µM H 2 O 2 treatment, and increased p-EZH2 expression after exposure to 100 µM H 2 O 2 for 2 hours. D, Immunofluorescence assay of mGCs showing reduced expression of Ezh2 (red) after exposure to 100 µM H 2 O 2 for 8 hours compared with 0-hour exposure. DAPI (blue) was used to stain cellular nuclei. Scale bar = 50 µm. Original magnification, 100×. E, qRT-PCR assay of EZH2 mRNA expression in human granulosa cells (hGCs; n = 35 for Con-hGC collected from women with tubal infertility and EM-hGC collected from women with endometriosis-associated infertility); * P less than .05, unpaired t test. F, ELISA of H3K27Me3 in hGCs (n = 18 for Con-hGC, n = 18 for EM-hGC); * P less than .05, unpaired t test. G, Western blot analysis of key histone methylation-associated proteins in hGCs (n = 6 for Con-hGC, n = 7 for EM-hGC). H, Scatter diagram showing significant Pearson correlation between numbers of retrieved and mature oocytes with EZH2 mRNA expression in endometriosis GCs based on qRT-PCR results (n = 33 for EM-hGC). Left panel shows the correlation between oocyte retrieval number and EZH2 mRNA expression, and right panel displays the correlation between mature oocyte number and EZH2 mRNA expression; both P less than .01. I, Left scatter diagram shows the correlation between oocyte retrieval number and H3K27Me3 protein expression in endometriosis GCs (n = 18 for EM-hGC), and right scatter diagram displays the correlation between high-quality embryo rate and H3K27Me3 protein expression (n = 18 for EM-hGC); both P less than .05. To examine the histone methylation status of endometriosis GCs during ovulation, EZH2 expression was evaluated by qRT-PCR. We found significantly decreased EZH2 mRNA expression in endometriosis GCs compared with that in GCs from women with tubal infertility ( Fig. 1E ; P < .05). ELISA results showed significantly reduced H3K27Me3 protein levels in endometriosis GCs ( Fig. 1F ; P < .05) with no differences in H3K27 monomethylation or dimethylation, or levels of monomethylation, dimethylation, and trimethylation of H3K4, H3K9, H3K36, or H3K79 (Supplementary Fig. S1C-S1G) ( 39 ). Decreased EZH2 and H3K27Me3 levels in endometriosis GCs were confirmed by Western blot analysis ( Fig. 1G and Supplementary Fig. S1B [ 39 ]). H3K4Me3, H3K9Me3, and H3K27Me3 are the most important histone marks regulated by EZH2. A reduction in H3K9Me3 protein expression and increase in H3K4Me3 protein expression were observed (see Fig. 1G ), but quantitative analysis revealed no statistical significance (see Supplementary Fig. S1B) ( 39 ). Moreover, no statistical differences were found in H3K4 trimethylation or H3K9 trimethylation levels by ELISA after an increase of GC sample size (see Supplementary Fig. S1D-S1E) ( 39 ). On the basis of our PCR, Western blot, and ELISA results, we chose H3K27Me3 for further research. Linear regression analysis showed positive correlations between EZH2 mRNA expression in endometriosis GCs and numbers of oocytes retrieved and mature oocytes in patients with endometriosis ( Fig. 1H , R = 0.454 for oocyte retrieval number and R = 0.530 for mature oocyte number; both P < .01). Positive correlations were also found between H3K27Me3 levels in endometriosis GCs and numbers of retrieved oocytes and rates of high-quality embryos in patients with endometriosis ( Fig. 1I , R = 0.482 for oocyte retrieval number and R = 0.579 for high-quality embryo rate; both P < .05). In addition, a positive correlation was found between H3K27Me3 levels in endometriosis GCs and high-quality embryo rates in patients with endometriosis, indicating that histone methylation modification may profoundly affect embryo quality. However, more research is needed to clarify the detailed mechanism. Epigenetic modifications change dramatically during ovulation. The LH surge induces rapid alterations of steroidogenesis-related genes by changing DNA methylation, histone acetylation, and/or histone methylation levels in promoters for StAR ( 54 , 55 ), Cyp19a1 ( 24 ), Cyp11a1 ( 35 ), and inhibin α ( 56 ). To explore the influence of EZH2 deficiency on ovulation, IHC was performed to evaluate expression patterns of Ezh2 and H3K27Me3 in mGCs. Exogenous HCG induced phosphorylation of ERK1/2 (p-ERK1/2) in mGCs, with maximum levels observed at 4 hours ( Fig. 2A and 2B ). In addition, the downstream effector cyclooxygenase 2 (Cox-2) was elevated at 2 hours and peaked at 4 hours ( Fig. 2C ). Sharply increased p-ERK1/2 and Cox-2 protein expression in mGCs indicated the occurrence of ovulation. Conversely, Ezh2 mRNA and Ezh2 protein expression gradually decreased after HCG injection and reached the lowest levels after 12 hours (Supplementary Fig. S2A [ 39 ] and Fig. 2D , all compared with PMSG for 48 hours). IHC showed that Ezh2 and H3K27Me3 were moderately expressed in the nuclei of mGCs from secondary, preantral, and antral follicles compared with those in the nuclei of ovarian stroma cells, theca cells, and luteinizing GCs ( Fig. 2A ). Additionally, H3K27Me3 levels rapidly decreased after exogenous HCG injection, with the lowest level observed at 4 hours and remaining stable until 12 hours ( Fig. 2E ). These results indicate that the ovulation process in mGCs was accompanied by decreased expression of Ezh2 and H3K27Me3. Decreased Ezh2 and H3K27Me3 levels in mouse granulosa cells (GCs) during ovulation. A, Immunohistochemistry of indicated proteins in mouse ovary GCs from C57BL/6 mice (n = 5 for each time point). Staining was developed by diaminobenzidine and nuclei were stained with hematoxylin. Scale bars = 50 µm. The right curve graph reflects dynamic changes of ovulation marker proteins (p-ERK1/2 and Cox-2), the histone methyltransferase Ezh2, and histone mark H3K27Me3 at different time points during ovulation. B, Average H-scores of p-ERK1/2 after treatment with PMSG for 48 hours (P48), HCG for 2 hours (H2), HCG for 4 hours (H4), HCG for 8 hours (H8), HCG for 12 hours (H12), HCG for 16 hours (H16), and HCG for 20 hours (H20). C, Average H-scores of Cox-2 at P48, H2, H4, H8, H12, H16, and H20. D, Average H-scores of Ezzh2 at P48, H2, H4, H8, H12, H16, and H20. E, Average H-scores of H3K27Me3 at P48, H2, H4, H8, H12, H16, and H20. To further explore the influence of Ezh2 deficiency on infertility, we generated GC-specific Ezh2 KO mice ( Ezh2 fl/fl ; Cyp19-Cre ; Fig. 3A ). Successful disruption of Ezh2 was confirmed by marked decreases of Ezh2 and H3K27Me3 levels in GCs from 3w- to 6w-old mice after stimulation with PMSG for 48 hours ( Fig. 3B ). IHC of unstimulated 3w-old mice demonstrated selective deficiency of Ezh2 in GCs of growing and antral follicles ( Fig. 3C ). Fertility testing demonstrated the subfertility of KO females, although no significant difference in mouse body or ovary weights were found between 6w-old KO and WT females (Supplementary Fig. S2B and S2C) ( 39 ). Selective Ezh2 depletion in ovary GCs causes female mouse subfertility by disturbing ovulation. A, Flowchart showing generation of ovary GC-specific Ezh2 -knockout (KO) mice Ezh2 fl/fl ; Cyp19-Cre (KO) mice, which were obtained by mating Ezh2 fl/+ ; Cyp19-Cre mice (wild-type [WT] mouse) with Ezh2 fl/fl mice. B, Western blot analysis of Ezh2 and H3K27Me3 levels in fresh extracted mGCs from 3-week-old to 6-week-old WT and KO mice following treatment with PMSG for 48 hours (n = 6 for WT, n = 6 for KO; each group contained 3 3-week-old mice and 3 6-week-old mice). C, IHC results showing classical Ezh2 staining in preovulatory mGCs of 3-week-old WT mice and paired 3-week-old KO mice (n = 3 for KO mice, n = 3 for WT mice, without superovulation). D, The development of mouse fertility within 6 months of mating. The horizontal axis shows time (day) from mating, while the vertical axis indicates average numbers of pups for each delivery (pups per delivery). The first generation of WT mice were born an average of 21 days after mating. Ten WT mice and 10 KO mice were used to plot fertility curves. E, qRT-PCR results for relative mRNA expression of the indicated ovulation-related genes in fresh extracted mGCs after PMSG injection for 48 hours, followed by HCG treatment for 4 hours (n = 4 for WT, n = 4 for KO; only 3-week-old immature mice were used for superovulation); unpaired t test. F, Antral follicles constitute a specialized niche that satisfies the needs of ovulation. The number of antral follicles per ovary at P48 were counted by 2 technicians simultaneously (n = 10 for KO mice, n = 10 for WT mice); unpaired t test. G, Cumulus–oocyte complexes (COCs) in the oviducts of mice at HCG 16 (n = 10 for KO mice, n = 10 for WT mice); unpaired t test. H, In vitro COC expansion assay results. The defect of COC expansion in KO mice could not be rescued by FSH (100 ng/mL) addition. Images shown in the left panel were acquired 12 to 14 hours after culture of COCs with 100 ng/mL FSH. Scale bars = 100 µm. Magnification, 100×. Right panel shows the statistical analysis of average COC expression rates from 10 paired WT or KO mice; unpaired t test. I, Hematoxylin and eosin staining showing typical folliculogenesis and ovulation images of 3-week-old WT and KO ovaries at P48, H4, and H16 (n = 30 for KO mice, n = 30 for WT mice; each point reflects 10 mice). Scale bars = 100 µm. Magnification, 100×. J, Results of western blot analysis of Ezh2, H3K27Me3, and ovulation-associated protein (p-ERK1/2, C/EBPβ, Smad4, Smad2/3, and Cox-2) expression in fresh extracted mGCs from 3-week-old to 6-week-old WT and KO mice after a superovulatory regimen of PMSG and hCG for 4 hours (n = 8 for WT, n = 8 for KO; each group reflects 4 mice). Fig. 3D shows the development of mouse fertility within 6 months of mating. The horizontal axis shows time (days) from mating, while the vertical axis indicates average numbers of pups for each delivery (pups/delivery). A total of 10 WT mice with Ezh2 fl/fl ; Cyp19-Cre genotype and 10 KO mice with Ezh2 fl/+ ; Cyp19-Cre genotype were used for this fertility test. Vaginal plugs were checked to confirm mating the day after mating, with the mating day defined as day 0 (beginning of the horizontal axis). Births were recorded for all 10 female mice from WT and KO groups. Specifically, the number of pups produced by each female mouse at every delivery was recorded, and the average number of pups produced by all 10 female mice in each group at every delivery was calculated. Each node on the fertility curve represents the average number of pups for all 10 female mice in each group at the corresponding delivery. Mice in the KO group had 7 deliveries within 6 months, while mice in the WT group had 8 (see Fig. 3D , each node on the fertility curve refers to a delivery). Mice in the WT group took an average of 21 days to produce their first pups, while mice in the KO group took an average of 32 days to produce their first pups (see Fig. 3D , first node of the WT fertility curve is located at day 21, while the first node of the KO fertility curve is located at day 32). However, the interval time between 2 adjacent nodes in both groups was 21 days, indicating that no significant delay of production time occurred after the first delivery. Collectively, these results indicate that first litters were severely compromised in KO mice. The fecundity of female mice undulates, first rising and then falling at later reproductive ages. Female mice first mated with fertile male mice at age 6 weeks and stopped mating at age 30 weeks, yielding a normal fertility curve that rises first and falls later (see Fig. 3D ). Although fertility curves for WT and KO mice showed the same tendency, the average number of pups per delivery decreased in mice with Ezh2 -depleted GCs (see Fig. 3D ), and the average number of offspring over a 6-month period was significantly decreased in KO mice compared with WT mice (Supplementary Fig. S2D; P < .001) ( 39 ). Induction of epidermal growth factor–like factors (amphiregulin, epiregulin, and betacellulin) and other ovulation-related genes ( Ptgs2 , Tnfaip6 , and Sult1e1 ) by LH is required for cumulus cell expansion and follicle rupture in vivo ( 57 , 58 ). We performed qRT-PCR to quantify mRNA levels of these genes. Prostaglandin synthase 2 ( Ptgs2 ), amphiregulin ( Areg ), epiregulin ( Ereg ), and Sult1e1 were induced in mGCs of WT mice but upregulated to a lesser degree in mGCs of KO mice ( Fig. 3E ). No differences were observed in primordial, primary, secondary, or antral follicles of 8w-old WT and KO mice (Supplementary Fig. S2E) ( 39 ), and no significant difference in antral follicle counts per ovary was observed after PMSG treatment for 48 hours ( Fig. 3F ), indicating comparative follicular development between WT and KO mice. WT mice had ovulated many COCs by 16 hours after hCG injection, but only a few COCs were observed in the oviducts of KO mice ( Fig. 3G ). Consistent with ovulation defects, the average number of corpora lutea was reduced in KO mice at 16 hours post HCG injection compared with that in WT mice ( Fig. 3I and Supplementary Fig. S2F [ 39 ]). Our IHC results confirmed inhibition of ovulation in the KO strain, although follicular development patterns in the 2 groups were similar (see Fig. 3I ). Specifically, COC expansion was impaired in preovulatory follicles of KO mice even following treatment with 100-ng/mL FSH for 12 hours ( Fig. 3H ). Mammalian ovulation is a multistep physiological process. In addition to canonical signaling molecules triggered by gonadotropins, such as CCAAT/enhancer binding protein β (C/EBPβ), p-ERK1/2, and COX-2, the transforming growth factor β signaling pathway plays crucial roles. Previous studies showed that GC-specific depletion of Smad4 caused premature luteinization followed by ovulation failure ( 42 ), while double KO of Smad2 and Smad3 in GCs reduced female fecundity by disrupting COC expansion ( 59 ). Western blot analysis confirmed decreased expression of ovulation-related proteins (p-ERK1/2, C/EBPβ, Smad4, Smad2/3, and Cox-2) in KO-GCs following hCG treatment for 4 hours ( Fig. 3J ), indicating an ovulation defect in GC-specific KO mice. Furthermore, serum progesterone concentrations were increased in 6w-old KO mice following hCG treatment for 4 hours (Supplementary Fig. S2G) ( 39 ), revealing complicated developmental stage-specific functions of Ezh2 in ovulation and luteinization. To clarify the influence of Ezh2 deletion on the estrous cycle, we observed estrous cycle patterns of WT and KO mice for over 20 days, which were similar (Supplementary Fig. S2H) ( 39 ). Therefore, follicular development was unaffected in KO mice, but GC-specific Ezh2 deficiency interfered with ovulation by disrupting paracrine signals. RNA-seq was performed to investigate the effects of Ezh2 depletion in GCs. Normalized RNA-seq data from 4 WT-GCs and 3 KO-GCs were subjected to PCA. PC1 and PC2 visually separated WT-GCs and KO-GCs as 2 distinctive clusters ( Fig. 4A ). We found that 1420 genes were significantly upregulated and 827 genes were significantly downregulated in mGCs of KO mice compared with WT-GCs following treatment with hCG for 4 hours (fold change > 2; corrected P value < .05). KEGG analysis results indicated enrichment of the “mitogen-activated protein kinase (MAPK) signaling pathway,” “cytokine–cytokine receptor interaction,” and other inflammation-associated pathways in KO-GCs, while decreased enrichment of the “cyclic adenosine monophosphate signaling pathway” and “ovarian steroidogenesis” were observed in KO-GCs (Supplementary Fig. S2I) ( 39 ). GSEA results also indicated abundant enrichment of functional genes from the “MAPK signaling pathway” (normalized enrichment score [NES] = 1.373, false discovery rate [FDR] q  value = 0.051; P ≤ .01), “cytokine–cytokine receptor interaction” (NES = 1.373, FDR q-value = 0.055; P ≤ .01), and “transcriptional misregulation” (NES = 1.384, FDR q-value = 0.067; P ≤ .01) in KO-GCs ( Fig. 4B ). To identify target genes transcriptionally misregulated by Ezh2 KO, we performed Venn analysis. The suppressive receptor IL-1R2 was the common differentially expressed gene in the 3 aforementioned signaling pathways ( Fig. 4C ). Ezh2 knockout (KO) in granulosa cells (GCs) upregulated il-1R2 expression by decreasing H3K27Me3 methylation on the il-1R2 promoter. A, Principal component analysis (PCA) of mRNA data from 4 wild-type (WT)-GCs and 3 KO-GCs. B, GSEA revealed enrichment of functional genes from the “MAPK signaling pathway,” “cytokine–cytokine receptor interaction,” and “transcriptional misregulation” gene set in KO-GCs. NES, normalized enrichment score; false discovery rate (FDR) of all sets were less than 25%; all P less than .01. C, Venn analysis of abundant differentially expressed genes between WT-GCs and KO-GCs based on ovulation, inflammation, and transcriptional regulation data sets. D, ChIP-PCR results showing fold change of immunoprecipitated Il-1r2 mRNA in fresh mGCs extracted from WT and KO mice at H4 (n = 20 for WT, n = 20 for KO). ChIP-RCR data are shown as mean ± SD of 3 independent experiments; *** P less than .001, unpaired t test. E, qRT-PCR results of indicated genes in fresh extracted mGCs from WT and KO mice at H4 (n = 3 for WT, n = 3 for KO); *** P less than .001, * P less than .05, unpaired t test. F, mGCs from C57BL/6 mice were infected by LV-Sh- Ezh2 or LV-Sh-NC, and mRNA expression of indicated genes were detected via qRT-PCR; *** P less than .001, ** P less than .01, paired t test. G, Results of Western blot analysis of fresh mGCs extracted from WT and KO mice at H4 (n = 4 for WT, n = 4 for KO). H, Western blot analysis of indicated protein expression in mGCs infected by LV-Sh-NC or LV-Sh- Ezh2 with or without P/FSK stimulation. To clarify the specific targets disrupted by Ezh2 KO, ChIP sequencing of mGCs from WT or KO mice following a superovulatory regimen of PMSG and hCG 4 hours was performed. Differentially expressed pathways enriched in KO mice are summarized in Supplementary Fig. S3A ( 39 ). The 5′-upstream region of mRNA is critical for the regulation of gene expression. H3K27Me3 levels in distal regions of the Il1r2 promoter were markedly lower in KO-GCs than WT-GCs, but no differences were observed in the proximal promoter region (Supplementary Fig. S3B) ( 39 ). ChIP-PCR results show that the binding of an H3K27Me3 antibody to the Il1r2 promoter region was decreased after Ezh2 depletion at hCG-4 hours (H4) ( Fig. 4D ). qRT-PCR results of mGCs obtained from WT or KO mice treated with PMSG, followed by HCG for 4 hours, indicated reduced Ezh2 mRNA expression and increased Il1r2 mRNA expression in KO-GCs, but no changes in Il1r1 or Il1rn mRNA ( Fig. 4E ). Importantly, Ezh2 knockdown in vitro induced Il1r2 expression but had no influence on ovulation-related molecules ( Ptgs2 , Areg , Ereg , Sult1e1 , Cox-2, Smad2/3, Smad4, C/EBPβ, or p-ERK1/2) in the absence of PMA and FSK (P/FSK; Fig. 4F and 4H ). mGCs from 3w-old KO mice showed decreased Cox-2, Smad2/3, Smad4, C/EBPβ, and p-ERK1/2 expression after treatment with HCG for 4 hours, whereas IL-1R2 protein expression was increased ( Figs. 3J and 4G ). Consistent with in vivo results, mGCs from C57BL/6 mice with Ezh2 knockdown showed increased IL-1R2 expression and decreased Cox-2, Smad2/3, Smad4, C/EBPβ, and p-ERK1/2 protein expression compared with controls following exposure to P/FSK (see Fig. 4H ). P/FSK are compounds used to induce luteinization. Our Western blot analysis shows reduced Ezh2 and H3K27Me3 expression in cells treated with P/FSK for 4 hours, while expression of Cox-2, Smad2/3, Smad4, C/EBPβ, and p-ERK1/2 was increased compared with levels in untreated cells (P/FSK treatment for 0 hours; see Fig. 4H ). Although the LH surge from the pituitary gland plays a central role in the initiation of ovulation, the function of other paracrine factors in FF should not be ignored. IL-1β concentrations are elevated in endometriosis FF and thought to aggravate GC senescence by acting as a senescence-associated secretory phenotype factor ( 20 ). qRT-PCR showed that the addition of IL-1β significantly activated ovulation-related gene expression, even when co-cultured with P/FSK, and had a superposition effect on ovulation ( Fig. 5A ; P < .05). IL-1β function depends on the receptor it binds (IL-1R1, IL-1R2, or IL-1RN) ( 21-23 ). Expression of activated receptor Il1r1 was upregulated after hCG treatment for 4 hours, while expression of the suppressive receptor Il1r2 was downregulated; no change was observed in Il1rn ( Fig. 5B , left panel). Similar trends were observed in vitro, with elevated Il1r1 expression and decreased Il1r2 expression after P/FSK exposure for 4 hours (see Fig. 5B , right panel). Interleukin (IL)-1β markedly amplified ovulation signals but had no effect on Ezh2, H3K27Me3, or IL-1R2 expression levels. A, qRT-PCR results showing Ptgs2 , Areg , Ereg , and Sult1e1 mRNA expression after treatment with 10-ng/mL IL-1β for 0, 15, and 30 minutes; * P less than .05, 10 ng/mL IL-1β for 15 minutes compared with 10 ng/mL IL-1β for 0 minutes, unpaired t test; # P less than .05, 10 ng/mL IL-1β for 30 minutes compared with 10 ng/mL IL-1β for 0 minutes, unpaired t test. B, Left panel shows qRT-PCR results of IL-1β receptor mRNA ( Il1r1 , Il1r2 , and Il1rn ) in fresh mGCs extracted from C57BL/6 mice at P48 and H4 (P48 or PMSG 48 hours refers to PMSG injection for 48 hours, H4 or HCG 4 hours refers to PMSG injection for 48 hours followed by HCG treatment for 4 hours); * P less than .05, unpaired t  test. Right panel shows qRT-PCR results of IL-1β receptor mRNA in mGCs treated with P/FSK for 0 hours or 4 hours; * P less than .05, paired t test. C, Western blot analysis of ovulation-associated protein expression after treatment with 10 ng/mL IL-1β for 0, 15, and 30 minutes; P/FSK treatment was used as positive control. D, Western blot analysis showing the effect of IL-1β (10 ng/mL), P/FSK, and IL-1β with P/FSK co-culture on Ezh2, H3K27Me3, IL-1R2, Cox-2, and p-ERK1/2 protein expression. E, Relative mRNA expression of Ezh2 , Il-1r2 , Ptgs2 , Areg , Ereg , and Sult1e1 after infection with LV-Sh-NC or LV-Sh- Ezh2 , followed by P/FSK treatment for 4 hours; * P less than .05, ** P less than .01, paired t test. F, Relative mRNA expression of Ezh2 , Il-1r2 , Ptgs2 , Areg , Ereg , and Sult1e1 after infection with LV-Sh-NC or LV-Sh- Ezh2, followed by P/FSK and 10 ng/mL IL-1β co-culture; * P  less than .05, ** P less than .01, *** P less than .001, paired t test. From our previous data, we chose 15 minutes as the IL-1β treatment period to detect mRNA expression changes. G, Western blot analysis of indicated protein expression in cells infected with LV-Sh-NC or LV-Sh- Ezh2 , followed by treatment with P/FSK for 4 hours, 10 ng/mL IL-1β for 0.5 hours, or both. To detect changes in protein expression, the treatment time of IL-1β was 30 minutes. P/FSK and IL-1β co-culture refers to P/FSK treatment for the first 3.5 hours, followed by the addition of 10 ng/mL IL-1β for the last 30 minutes. To clarify the function of IL-1β, we examined ovulation signaling pathway molecules after the addition of 10-ng/mL IL-1β for 0 minutes, 15 minutes, and 30 minutes; P/FSK was used as a positive control. Western blot analysis showed that expression of Cox-2, Smad2/3, C/EBPβ, and p-ERK1/2 were increased after treatment with IL-1β for 30 minutes ( Fig. 5C ), although Smad4 protein expression was unaffected. Moreover, the addition of IL-1β did not influence levels of Ezh2, H3K27Me3, or IL-1R2. We further explored the combined effect of IL-1β and P/FSK on ovulation. After incubation of cells with 10-ng/mL IL-1β for 0.5 hours, p-ERK1/2 and Cox-2 levels were increased. The addition of IL-1β aggravated P/FSK-induced Cox-2 and p-ERK1/2 protein expression, but IL-1β did not influence levels of Ezh2, H3K27Me3, or IL-1R2 ( Fig. 5D ). Additionally, in cells with Ezh2 knockdown (LV-Sh- Ezh2 ), expression of Il1r2 was increased, but P/FSK-induced increases of Ptgs2 , Areg , Ereg , and Sult1e1 were compromised compared with levels in the LV-Sh-NC group (Fig.  5E ). These results not only indicate the vital role of Ezh2 on ovulation, they suggest that the LH surge could stimulate IL-1β secretion. Importantly, the decrease of ovulation-related gene expression induced by Ezh2 knockdown was not rescued by P/FSK and IL-1β co-culture ( Fig. 5F ). IL-1β addition also failed to reverse the inhibition of ovulation-related protein expression by Ezh2 knockdown ( Fig. 5G ). To further clarify the role of IL-1R2 in ovulation, shRNA was used to inhibit IL-1R2 (si- IL-1R2 ) expression after knockdown of Ezh2. qRT-PCR and Western blot analyses show that the inhibition of IL-1R2 reversed the decreases of Cox-2, Smad2/3, Smad4, C/EBPβ, and p-ERK1/2 expression caused by Ezh2 knockdown under IL-1β incubation ( Fig. 6A and 6B ). Additionally, inhibition of IL-1R2 partly recovered changes in ovulation-related gene expression induced by P/FSK ( Fig. 6C ) and partly reversed the suppression of Cox-2, Smad2/3, Smad4, C/EBPβ, and p-ERK1/2 levels induced by Ezh2 knockdown ( Fig. 6D ). These results further suggest that the LH surge stimulated IL-1β secretion. Our IHC results show decreased Ezh2, H3K27Me3, p-ERK1/2, and Cox-2 protein expression in KO-GCs at H4, while IL-1R2 expression in KO-GCs was significantly increased at H4 ( Fig. 6E and 6F ; P < .05). Collectively, these results indicate that EZH2- and H3K7Me3-based histone methylation modification was reduced during normal ovulation when IL-1R2 protein expression was decreased in GCs, while selective Ezh2 depletion in GCs caused ovulatory dysfunction by upregulating IL-1R2 expression and consequently suppressing IL-1β-mediated ovulation reactions. Interleukin (IL)-1R2 inhibition partly recovered the decrease of ovulation-related genes induced by Ezh2 knockdown under incubation with p/FSK or IL-1β. A, Relative messenger RNA (mRNA) expression of indicated genes after transfection with LV-Sh- Ezh2 or LV-Sh- Ezh2 combined with Si- IL1R2 , followed by 10 ng/mL IL-1β for 15 minutes. si-NC or si- IL-1R2 (100 nM) was added to the medium after infection with LV-Sh- Ezh2 , and then IL-1β was added. ** P less than .01, *** P less than .001, paired t  test. B, Western blot analysis of indicated proteins after treatment with LV-Sh-NC, LV-Sh- Ezh2, or LV-Sh- Ezh2 combined with Si- IL1R2 , followed by IL-1β treatment for 0.5 hours. C, Relative mRNA expression of indicated genes after treatment with LV-Sh- Ezh2 or LV-Sh- Ezh2 combined with Si- IL1R2 , followed by P/FSK treatment for 4 hours. si-NC or si- IL-1R2 (100 nM) was added to the medium after infection with LV-Sh- Ezh2, and then P/FSK was added to induce expression of LH target genes. ** P less than .01, *** P less than .001, paired t test. D, Western blot analysis of indicated proteins after treatment with LV-Sh-NC, LV-Sh- Ezh2 , or LV-Sh- Ezh2 combined with si -IL1R2, followed by P/FSK treatment for 4 hours. E, IHC images showing decreased Ezh2, H3K27Me3, p-ERK1/2, and Cox-2 protein expression in knockout (KO)-GCs at H4, as well as increased IL-1R2 protein expression in KO-GCs at H4. Scale bar = 100 µm. Magnification, 200×. F, Immunohistochemistry H scores of Ezh2, H3K27Me3, p-ERK1/2, Cox-2, and IL-1R2 expression in mGCs of wild-type (WT) and KO mice (n = 10 for WT, n = 10 for KO); * P less than .05, ** P less than .01, *** P less than .001, unpaired t test. To explore ovulation in vivo, we established an endometriosis mouse model using C57BL/6 mice, as previously described (Supplementary Fig. S3C) ( 39 ). To improve the implantation of endometrial tissue to ectopic locations ( 60 ), we administered 200-µg/kg estrogen every other day to mice. Because the effect of estrogen is believed to last less than 14 days ( 60 ), a 2-week washout period was administered before induction of ovulation to reduce the effects of hormones on follicular development. Two weeks after endometrium transplantation surgery, 4 mice were randomly selected to be examined for the formation of exogenous cysts. Vibrant exogenous cysts were observed in the abdominopelvic cavity of the endometriosis mouse. qRT-PCR results revealed decreased Ezh2 , Ptgs2 , Areg , Ereg , and Sult1e1 mRNA expression in GCs from endometriosis mice following treatment with HCG for 4 hours ( Fig. 7A ), while Il1r2 mRNA expression was increased. Western blot analyses indicated decreased Cox-2, p-ERK1/2, C/EBPβ, Smad2/3, Smad4, H3K27Me3, and Ezh2 levels in EM-GCs compared with the levels in Con-GCs at H4, while IL-1R2 protein expression was abundant in EM-GCs at H4 ( Fig. 7B ), consistent with the phenotype of KO mice. Notably, IHC results of EM-GCs and Con-GCs showed the same changes ( Fig. 7C ). Ezh2 and H3K27Me3 expression were found in the nuclei of Con-GCs, while Cox-2 and p-ERK1/2 expression were significantly decreased in EM-GCs with increased IL-1R2 staining in the cytoplasm at H4 (see Fig. 7C and 7D ; P < .05). The H-scores of these proteins in mGCs are summarized in Fig. 7D . Granulosa cells (GCs) from endometriotic mice show decreased histone methylation modifications and ovulation gene expression, and increased interleukin (IL)-1R2 expression. A, Relative messenger RNA (mRNA) expression of indicated genes in fresh extracted mouse GCs (mGCs) at H4 from control mice (Con-HCG 4 hours) or EM mice (EM-HCG 4 hours); n = 3 for Con-HCG 4 hours, n = 3 for EM-HCG 4 hours; * P less than .05, ** P less than .01, *** P less than .001, unpaired t  test. B, Western blot analysis of indicated proteins in fresh extracted mGCs from Con or EM mice at P48 or H4 (each time point reflects 3 mice). C, Representative IHC images showing decreased Ezh2, H3K27Me3, p-ERK1/2, and Cox-2 protein expression in EM-GCs at H4, as well as increased IL-1R2 protein expression in EM-GCs at H4. Some images were derived from serial sections of mouse ovary. D, Immunohistochemistry H scores of Ezh2, H3K27Me3, IL-1R2, p-ERK1/2, and Cox-2 expression in mGCs from Con or EM mice (n = 12 for Con, n = 10 for EM); all compared with the Con group, unpaired t  test. We next examined expression of IL-1R1 , IL-1R2 , and IL-1RN in hGCs from preovulatory follicles. Our results show significantly elevated expression of IL-1R2 in endometriosis GCs compared with that in control GCs ( Fig. 8A ; P < .05). However, IL-1R1 and IL-1RN expressions were similar between endometriosis GCs and control GCs (see Fig. 8A and Supplementary Fig. S3D [ 39 ]). We initially focused on the pathophysiology of deep infiltrating endometriosis, but in-depth studies found no significant differences of Ezh2 , H3K27Me3, IL1R2 , or ovulation-related gene expression between ovarian endometriosis and deep infiltrating endometriosis. We considered that the pathophysiology of these 2 types of endometriosis was not identical, although both showed abnormal ovulatory phenotypes. We reanalyzed our published RNA-seq data ( 20 ) and performed GSEA, the results of which indicate enrichment of functional genes from the “IL-1 signaling pathway” ( Fig. 8B with NES = 1.481, FDR q value = 0.054, and P value < .05) and “IL-1R pathway” in EM-hGCs (see Fig. 8C with NES = 1.380, FDR q value = 0.208, and P value = .098), confirming the importance of IL-1R2 in endometriosis GCs. Increased IL-1R2 expression and reduced ovulation gene expression in granulosa cells (GCs) contributes to endometriosis-associated infertility. A, qRT-PCR assay of IL-1R1 and IL-1R2 mRNA expression in human GCs (hGCs; n = 32 for Con-hGC, n = 28 for EM-hGC); * P less than .05, unpaired t test. B and C, GSEA revealed enrichment of functional genes from the “IL 1 signaling pathway” and “IL-1R pathway” in GCs of patients with endometriosis (n = 4 for Con-hGC, n = 5 for EM-hGC; data previously detected but not published). D, qRT-PCR assay of PTGS2 mRNA expression in hGCs (n = 38 for Con-hGC, n = 34 for EM-hGC); *** P less than .001, unpaired t test. E, Thirty-eight paired control FF samples and 34 paired endometriosis FF samples were collected to detect concentrations of PGE2, PGF2α, and progesterone by ELISA; *** P less than .001, * P less than .05, unpaired t test. F, qRT-PCR assay of AREG , EREG , BTC , and PTX3 mRNA expression in hGCs (n = 20 for Con-hGC, n = 22 for EM-hGC; these samples overlapped with samples used to examine IL-1R1 and IL-1R2 mRNA expression; Btc, betacellulin), ** P less than .01, * P less than .05, unpaired t test. G, Western blot analysis of IL-1R2 and Cox-2 expression in hGCs (n = 6 for Con-hGC, n = 7 for EM-hGC). H, Typical images of dominant follicles from a control patient and luteinized unruptured follicles from a patient with endometriosis. I, ELISA results of progesterone concentrations in 38 paired control FF samples and 34 paired endometriosis FF samples; * P less than .05, unpaired t test. J, Scatter diagram showing the correlation between EZH2 mRNA expression in endometriosis GCs and PEG2 expression in paired endometriosis follicular fluid (n = 33 for patients with endometriosis; P < .01). K, Scatter diagram displaying the correlation between PGE2 concentrations and mature oocyte numbers in patients with endometriosis (n = 33 for EM-FF; P < .05). Next, PTGS2 mRNA expression was detected in 38 EM-GC samples and 34 Con-GC samples. qRT-PCR results show reduced PTGS2 mRNA expression in patients with endometriosis ( Fig. 8D ). Western blot results show decreased COX-2 protein expression in EM-hGCs compared with that in Con-hGCs ( Fig. 8G ). To better elucidate the microenvironment of follicles in endometriosis, paired human samples for 38 control FF (Con-hFF) and 34 endometriosis FF (EM-hFF) were examined for concentrations of PGE2, PGF2α, and progesterone. As displayed in Fig. 8E , PGE2, a pivotal stimulator of the ovulatory cascade ( 61 ), was decreased in EM-hFF compared with that in Con-hFF. However, PGF2α, a major luteolytic prostaglandin, was elevated in EM-hFF compared with that in Con-hFF (see Fig. 8E , right). In addition, mRNA expression of AREG , EREG , BTC , and PTX3 was decreased in EM-hGCs compared with that in Con-hGCs ( Fig. 8F ). In contrast, IL-1R2 protein expression was increased in EM-hGCs compared with that in Con-hGCs ( Fig. 8G ). The concentration of progesterone in preovulatory EM-hFF was also notably elevated ( Fig. 8I ), indicating intensification of luteinization in patients with endometriosis. Trapped oocytes that did not ovulate are more commonly observed in patients with endometriosis. Fig. 8H shows a typical dominant follicle from a control patient and luteinized unruptured follicle from a patient with endometriosis. Linear regression analysis showed positive correlations between EZH2 mRNA levels in endometriosis GCs and PGE2 concentrations in 33 paired EM-hFF samples ( Fig. 8J , R = 0.467; P < .01). Consistent with most published studies, a positive correlation between PGE2 concentration and mature oocyte number was also identified by linear regression analysis ( Fig. 8K , R = 0.437; P < .05). These data demonstrate that GCs from patients with endometriosis have an abnormal ovulatory portfolio that eventually contributes to endometriosis-associated infertility. We also explored the direct effects of excessive oxidative stress on ovulation via Western blot. Our results show that excessive oxidative stress suppressed both EZH2 and H3K27Me3 protein expression, and inhibited expression of ovulation-related molecules following stimulation with P/FSK (Supplementary Fig. S3E) ( 39 ).

Ovulatory dysfunction (anovulation, oligo-ovulation, and LUFS) has been intensively studied in endometriosis research ( 27 , 62 , 63 ). Reduced ovulation ostia on corpora lutea has been observed in patients with moderate and severe endometriosis ( 64 ). The incidence of ovulatory dysfunction in patients with endometriosis and infertility is approximately 27.2% ( 65 , 66 ). LUFS, the most common type of ovulatory dysfunction, involves the absence of follicular rupture and ovum release after the LH peak, and/or disrupted follicular maturity, follicle growth maintenance, or in situ luteinization, accompanied by elevated progesterone levels and secretory changes in the endometrium. LUFS occurs in 11% to 23% of fertile women ( 67 , 68 ) and up to 75% of patients with endometriosis ( 69 ). The incidence of LUFS increases with the severity of endometriosis ( 70 , 71 ). With rapid development of assisted reproductive technology and broadened applications of ultrasound in monitoring follicle growth, less attention has been paid to LUFS in the last decade. LUFS, an inherent disorder of endometriosis, results in secondary phenomena of reduced oocyte retrieval number and compromised oocyte quality during IVF. Elucidating the pathogenesis of LUFS will improve understanding of endometriosis-associated infertility and treatment outcomes of IVF for patients with endometriosis. Inflammation, which affects both ovulation and hormone secretion, is involved in endometriosis-associated infertility ( 72 ). A previous study showed significant increases of the proinflammatory factor IL-1β in endometriosis FF, and a tendency toward endometriosis severity-dependence ( 73 ). IL-1β expression was also elevated in serum, peritoneal fluid, and endometriotic lesions of patients with endometriosis compared with controls, and may contribute to the progression of endometriosis by facilitating angiogenesis and neuroangiogenesis ( 74-77 ). The effects of elevated FF IL-1β levels on ovulation in patients with endometriosis is unclear. We speculate that LUFS may be associated with the inflammatory niche that preovulatory follicles are exposed to during endometriosis, but the detailed mechanism requires further research. The function of IL-1β depends chiefly on the receptor it binds, namely IL-1R1 and IL-1R2. IL-1R1 forms a functional heterotrimeric signaling complex with the IL-1RAcP accessory protein to complement IL-1–dependent cell activation ( 21 , 22 ). In contrast to IL-1R1, IL-1R2 more efficiently binds IL-1β ( 22 ). IL-1R2, which has a short cytoplasmic tail with no TIR domain, negatively regulates IL-1β activity by at least 3 different mechanisms ( 23 ). By binding IL-1β precursor with high affinity, IL-1R2 blocks its cleavage by caspase 1 ( 78 ). IL-1R2 also blocks signal transmission by acting as a direct molecular trap for IL-1β ( 79 , 80 ). By forming the classic IL-1β/IL-1R2/IL-1RAcP complex, IL-1R2 efficiently inhibits the amplification of IL-1β signaling ( 80 ). On the basis of the biological functions of IL-1R2, we assume that increased IL-1β in endometriosis FF may participate in endometriosis-associated infertility through its high-affinity ligand-binding receptor IL-1R2. However, no study has reported the expression profiles of IL-1 receptors in endometriosis GCs. The aim of this study was to elucidate the causes of reduced oocyte retrieval numbers in patients with endometriosis in the context of increased IL-1β expression in the follicular microenvironment. On the basis of multiple previous studies of oxidative stress ( 20 ), we found an unexpected suppressive effect of excessive oxidative stress on histone modification, as well as a marked decrease of EZH2 histone methyltransferase activity in EM-GCs, although we cannot completely rule out regulation of EZH2 by oxidative stress–related proteins. Our results reveal spatial and temporal differences in EZH2 expression during ovulation, verifying the importance of histone methylation modification in ovulation. In cells incubated with P/FSK for 4 hours, protein expression of EZH2, H3K27Me3, and IL-1R2 were decreased, indicating a strong regulatory effect of LH surge on histone methylation modifications and the IL-1β receptor. However, neither EZH2, H3K27Me3, nor IL-1R2 expression was affected by IL-1β, even when coincubated with P/FSK, indicating that the reduction of histone methylation modifications in endometriosis GCs was not caused by elevated IL-1β. Notably, co-culture of cells with IL-1β and P/FSK increased protein expression of COX-2, p-ERK1/2, and C/EBPβ compared with culture of cells with P/FSK alone (see Fig. 5D ), demonstrating the superimposed effects of IL-1β on LH surge-induced ovulation. However, IL-1β does not directly affect histone methylation modifications. Although GC-specific Ezh2 -KO mice cannot completely represent the pathologic changes of human endometriosis, this model effectively reflects the effect of EZH2 on ovulation through targeting of IL-1R2. Moderate reductions in H3K27Me3 levels during ovulation resulted in the activation of some genes, while extreme reductions in H3K27Me3 activated the expression of other important genes. Moreover, in vivo and in vitro experiments showed that Il1r1 mRNA expression was elevated during ovulation, while expression of Il1r2 was decreased after LH stimulation and Il1rn expression remained unchanged (see Fig. 5B ). The contrasting expression prolife of IL-1β receptors suggests that IL-1R1–mediated activation of IL-1β signaling is predominant during normal ovulation, and an unimpeded ovulation process requires the suppressive receptor IL-1R2 to remain at low activity. Our results indicate that IL-1R2–based inhibition of IL-1β signaling leads to reduced ovulation-related gene expression, thereby contributing to endometriosis-associated infertility ( Fig. 9 ). The decreases in histone methylation modifications derived from reduced EZH2 and H3K27Me3 expression in endometriosis GCs are considered the initiating factor for endometriosis-associated infertility (see Fig. 9 ). Moreover, as a high-affinity ligand-binding receptor of IL-1β, the disrupting function of elevated IL-1R2 on ovulation is conceivable. Interestingly, our in vitro results also suggest that the LH surge induces IL-1β expression because Ezh2 knockdown suppressed ovulation-related gene expression with P/FSK treatment only. However, whether the LH surge enhances IL-1β expression in vivo is unknown and cannot be addressed because of ethics issues. Decreased histone methyltransferase activity of EZH2 in ovarian granulosa cells suppresses ovulatory signals of endometriosis-associated infertility patients by targeting high-affinity ligand-binding suppressive receptor interleukin (IL)-1R2. Folliculogenesis, ovulation, and luteinization are successive and complex processes. Untimely decreases in EZH2 histone methyltransferase activity in GCs stimulates IL-1R2 expression and interferes with the ovulation cascade by blocking IL-1β signal transmission, ultimately contributing to endometriosis-associated infertility. Although the LH surge plays a fundamental role in induction of ovulation, we confirmed that histone methylation modification-activated upregulation of IL-1R2 also interferes with the ovulation process. Taken together, our study took a comprehensive view of the follicular microenvironment to elucidate a novel mechanism by which reduced histone methylation leads to ovulation dysfunction and endometriosis-associated infertility. Specifically, reduced histone methylation upregulated the high-affinity suppressive receptor IL-1R2 and inhibited IL-1β signaling to disrupt the ovulation cascade, ultimately reducing ovulation-related gene expression (see Fig. 9 ). We suggest that administration of nonsteroidal anti-inflammatory drugs, especially PGE2 synthesis inhibitors, may be reasonable in clinical practice for treatment of women with endometriosis.

This study was initiated November 1, 2016 and terminated January 23, 2022. The ethics committee of Sir Run Shaw Hospital at Zhejiang University approved and monitored this study. Informed written consent was obtained from every patient before sample collection. We enrolled 99 patients with both laparoscopic and histological diagnosis of endometriosis (ovarian endometriosis, deep infiltrating endometriosis, or both) and 104 controls with tubal infertility at the Sir Run Shaw Hospital of Zhejiang University School of Medicine. Detailed inclusion and exclusion criteria are described in the supplementary material ( 39 ). FF was collected at the time of oocyte retrieval only from leading follicles with a diameter greater than 17 mm. Human GCs were collected by gently cutting the cumulus layer of each leading oocyte and then washing it twice in phosphate-buffered saline, followed by centrifugation (800 ×g for 5 min at 4 °C). Detailed procedures are presented in the supplementary material ( 39 ). Demographics, clinical characteristics, and assisted reproductive therapy outcomes of all recruited patients are listed in Table 1 . Clinical characteristics and outcomes of all patients Values in italic bold indicate statistically significant difference. Data are presented as means ± SD. Abbreviations: AMH, antimüllerian hormone; ART, assisted reproductive technology; BMI, body mass index; E2, 17β-estradiol; EM, endometriosis; FSH, follicle-stimulating hormone; hCG, human chorionic gonadotropin; ICSI, intracytoplasmic sperm injection; IVF, in vitro fertilization; LN, luteinizing hormone; Pg, progestin; PN, pronuclear; rAFS, revised American Fertility Society Score. Estimated by unpaired t test with Welch correction. Estimated by nonparametric test with Mann-Whitney U test. Estimated by Pearson chi-square. Human GC lines COV434 and SVOG were purchased from American Type Culture Collection. COV434 and SVOG were cultured in Dulbecco’s modified Eagle’s medium (DMEM)/F-12 (Jinuo Biomedical Technology Company) supplemented with 10% fetal bovine serum (FBS) and incubated at 37 °C in 5% CO 2 . Mouse GCs (mGCs) were cultured in DMEM/F-12 supplemented with 5% FBS and incubated at 37 °C in 5% CO 2 . The human GC lines COV434 and SVOG were previously validated ( 20 ). Murine GCs were harvested from pregnant mare serum gonadotropin (PMSG)-primed mice as previously described ( 40 ). Briefly, mGCs were released from antral follicles by puncturing with a 26.5-gauge needle. All animals were housed under a 12:12-hour light:dark schedule and provided food and water ad libitum. To avoid the complexity of ovarian functions associated with estrous cycles and endogenous surges of gonadotropins, female mice at postnatal day 21 were used to study ovarian responses to exogenous gonadotropins. Wild-type (WT) C57BL/6 mice were purchased from Hangzhou Ziyuan Laboratory Animal Science and Technology. To precisely disrupt Ezh2 in follicle maturation and luteinization, we used Cyp19-Cre mice ( 41 ). Cyp19-Cre transgenic mice are a Cre-expressing mouse model widely used in reproduction research and highly specific for ovary GCs ( 40-43 ). Mice with GC-specific knockout of Ezh2 (KO mice) were generated by crossing Cyp19-Cre mice with Ezh2 flox/flox mice. The recommended Cre-lox breeding program is described at https://www.jax.org/news-and-insights/jax-blog/2011/september/cre-lox-breeding-for-dummies . Cyp19-Cre mice (No. 027038) and Ezh2 flox/flox mice (No. 022616) were purchased from The Jackson Laboratory. Female mice (aged 3- or 6-week-old, abbreviated as 3- or 6-week-old, respectively) were used in experiments. Three-week-old (postnatal day 21) immature female mice were used for superovulation to reduce mGC differentiation. Physiological ovulation changes were explored in mature 6- or 8-week-old females without superovulation. Immature female mice were intraperitoneally injected with 5 IU of PMSG (Ningbo San Sheng Biotech) for 48 hours to stimulate follicle development, followed by injection with 5 IU human chorionic gonadotropin (hCG; Ningbo San Sheng Biotech) to induce ovulation. All mice used in this study were bred in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. The Committee of Experimental Animal Ethics of Zhejiang University approved the experiments. Our previous research found that excessive oxidative stress in endometriosis induces GC senescence by activating endoplasmic reticulum stress, contributing to endometriosis-associated infertility ( 20 ). To explore the effect of excessive oxidative stress on histone modification, 100 µM H 2 O 2 was used to induce oxidative stress in human GC lines COV434 and SVOG. An immunofluorescence assay was performed as described in the supplementary material ( 39 ) in cells treated with H 2 O 2 for 2 hours. Cell fractionation was performed using NE-PER Nuclear and Cytoplasmic Extraction Reagents (78833; Thermo Scientific) according to the manufacturer's protocol. After 2 washes in in phosphate-buffered saline, 8 × 10 6 COV434 or SVOG cells were lysed with cytoplasmic extraction reagent containing protease inhibitor cocktail for cytoplasmic extraction, suspended in ice-cold nuclear extraction reagent, and then centrifuged for 10 minutes for nuclear extraction. Cytoplasmic and nuclear fractions were collected and incubated in 5% bovine serum albumin with a glyceraldehyde-3-phosphate dehydrogenase (GAPDH) antibody (60004-1-Ig; Proteintech) or LaminB1 antibody (13435; Cell Signaling Technology) as cytoplasmic or nuclear indicators for quantification, respectively. Western blot analysis was performed according to standard protocols in our laboratory using primary antibodies listed in Table 2 . Image J (National Institutes of Health) was used to analyze the signal intensity of protein bands. Antibodies used in immunofluorescence, Western blot, immunohistochemistry, and chromatin immunoprecipitation Abbreviations: ChIP, chromatin immunoprecipitation; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; IF, immunofluorescence; IHC, immunohistochemistry; WB, Western blot; USA, United States of America. RNA isolation and quantitative reverse-transcriptase polymerase chain reaction (qRT-PCR) were performed with an RNA-Quick Purification Kit (RN001; Shanghai Yishan Biotechnology) according to the manufacturer's instructions. A total of 38 control GC samples and 34 endometriosis GC samples were collected to detect PTGS2 mRNA expression. Because of sample loss, 35 control GC samples and 33 endometriosis GC samples were collected to detect EZH2 mRNA expression. Another 32 control GCs samples and 28 endometriosis GC samples were collected to detect IL-1R1 , IL-1R2 , and IL-1RN mRNA expression, and the remaining 20 control GC samples and 22 endometriosis GC samples were used to evaluate expression of ovulation-related genes. Primer sequences, including those for ovulation-related genes Ptgs2 , Areg , Ereg , Sult1e1 , Btc , and Ptx3 , are listed in Table 3 . List of primers used in quantitative reverse-transcription polymerase chain reaction (PCR) , chromatin immunoprecipitation–PCR, and genome identification Abbreviations: BTC , betacellulin; ChIP, chromatin immunoprecipitation; EREG , epiregulin; Gapdh , glyceraldehyde-3-phosphate dehydrogenase; IL , interleukin; PCR, polymerase chain reaction; WT, wild-type. We used human samples of 38 control FF and 34 endometriosis FF to detect concentrations of prostaglandin E2 (PGE2; EK8103/2-96; Multi Sciences Biotech), prostaglandin F2α (PGF2α; ab133041, Abcam), and progesterone (25-0175, ET Healthcare) according to the manufacturers' protocols. We performed enzyme-linked immunosorbent assays (ELISAs) for histone modification using the following kits from Epigentek: P-3044-96, P-3028-96, P-3036-96, P-3052-96, P-3060-96, and an EpiQuik Total Histone Extraction Kit (catalog No. OP-0006). A total of 18 control GCs samples and 18 endometriosis GC samples were collected to detect H3K27 monomethylation, dimethylation, and trimethylation using kits from Epigentek according to the manufacturer's instructions. The remaining 12 control GC samples and 12 endometriosis GC samples were used to evaluate monomethylation, dimethylation, and trimethylation levels of H3K4, H3K9, H3K36, and H3K79. Detailed procedures are described in the supplementary material ( 39 ). Immunohistochemistry (IHC) was performed as previously described ( 20 ). A semiquantitative grading system (H-score) was used to evaluate the staining intensity and percentage of protein expression. Detailed procedures are provided in the supplementary material ( 39 ) and primary antibodies are listed in Table 2 . COCs were collected from ovaries of PMSG-treated (48 hours) immature WT and KO mice. COCs were plated in 50-µL defined COC medium ( 44 ) under the cover of mineral oil and treated with 100 ng/mL follicle-stimulating hormone (FSH) for 12 hours. The expansion status was observed by microscopy. Six-week-old WT (n = 10) and KO mice (n = 10) in estrus were housed with 8w-old fertile male mice (2:1; defined as day 0). The following day, vaginal plugs were checked to confirm mating (defined as day 1) and all female mice were used for recording births. Pup sizes were determined at 21 days after birth. The fertility test was conducted over 6 months, as previously described ( 20 ). High-throughput sequencing and bioinformatics analyses were conducted at Biomarker Technologies (BMKCloud). To compare global gene expression profiled between WT-GCs (n = 4) and KO-GCs (n = 3), WT and KO mice were injected with 5 IU PMSG, and 48 hours later 5 IU HCG was injected to induce ovulation. No less than 2 × 10 6 mGCs were collected from WT and KO mice ovaries, and these cells were further used to perform RNA-seq. According to the results of quality control measurements by Biomarker Technologies (BMKCloud), 2.5 μg RNA from WT-GCs and 2.5 μg RNA from KO-GCs were ultimately used for transcriptome analysis. The process of RNA-seq mainly includes sample detection, library construction, quality control, and sequencing on an high-throughput sequencing platform (Illumina). Approximately 24 008 genes were successfully annotated. Principal component analysis (PCA) and GSEA were conducted to identify differentially expressed genes between WT and KO mice. Using the Kyoto Encyclopedia of Genes and Genomes (KEGG) biological pathway database ( http://www.genome.jp/kegg/ ) and Gene Ontology Consortium database ( http://www.geneontology.org/ ), we performed GSEA (using GSEA 3.0, http://www.broadinstitute.org/gsea/ ) to explore differentially expressed gene sets in paired WT-GCs and KO-GCs. A chromatin immunoprecipitation (ChIP) assay was performed using a Simple ChIP Enzymatic Chromatin IP Kit (9003, Cell Signaling Technology) in accordance with the manufacturer's instructions. ChIP-grade anti-H3K27Me3 (9733, Cell Signaling Technology) and negative control Normal Rabbit immunoglobulin G (IgG) (2729, 1 µg per IP sample; Cell Signaling Technology) were used for IP. ChIP-enriched DNA and input DNA were subjected to ChIP-Seq. ChIP fold enrichment was calculated using the comparative Ct method. ChIP-PCR was performed to verify the results of ChIP-Seq. Primer sequences for ChIP-PCR are listed in Table 3 . Undifferentiated mGCs were harvested from PMSG-primed immature mice as previously reported ( 44 , 45 ). Briefly, 1 × 10 6 mGCs were cultured in DMEM/F12 (Invitrogen) containing 5% FBS (Invitrogen) in 6-well culture dishes. Cells were cultured in serum-free medium for at least 2 hours before treatments ( 43 ). Next, 10-µM forskolin (FSK; HY-15371; MedChemExpress) plus 20-µM phorbol 12-myristate 13-acetate (PMA; HY-18739, MedChemExpress) were applied for 4 hours to induce expression of the LH target. Lentiviral vectors expressing short hairpin RNA (shRNA) targeting Ezh2 (LV-Sh- Ezh2 ) and corresponding negative control lentiviruses (LV-Sh-NC) were purchased from GeneChem. Cells were seeded in 6-well plates and infected with lentivirus at 10 multiplicity of infection for 48 hours before induction of luteinization. The medium was changed after 24 hours of culture, and GCs were observed by fluorescence microscopy to confirm the infection efficiency. IL-1R2 shRNA (small interfering [si]- IL-1R2 ) and a nonsense mutation negative control (si-NC) were purchased from RiboBio. Our preliminary experiments performed according to the manufacturer instructions confirmed that 100-nM si- IL-1R2 for 48 hours effectively knocked down IL-1R2 expression in mGCs. To explore the function of IL-1R2 on ovulation, 100 nM si- IL-1R2 was added to suppress IL-1R2 expression after infection of cells with LV-Sh- Ezh2 . A total of 25 endometriosis model mice were established by mouse-mouse intraperitoneal implantation, as previously described ( 20 , 46-48 ). Two mice died of intestinal obstruction. Detailed information for establishing the endometriosis mouse model is described in the supplementary material, and a concise procedure is presented in Supplementary Fig. S3C ( 39 ). All experiments were repeated at least 3 times. SPSS version 19.0 (SPSS) and GraphPad Prism 5 (GraphPad Software) were used for statistical analysis. Statistical comparison between 2 groups was performed using the unpaired t test or Mann-Whitney U test after the test of data normality. Comparison of continuous variables among groups was conducted by one-way analysis of variance followed by least significant difference tests. Pearson correlation analysis was used to estimate the correlation between different clinical outcomes and independent variables of EZH2 mRNA expression in GCs, H3K27Me3 expression in GCs, PTGS2 mRNA expression in GCs, and progesterone in FF.