Introduction

Follicles are functional units of the mammalian ovary, and each follicle consists of an oocyte surrounded by one or more layers of somatic cells. A neonatal ovary is endowed with several million primordial follicles, consisting of an oocyte surrounded by a single layer of flattened pregranulosa cells [1]. Of those dormant primordial follicles that survive and are recruited into the growing follicle population, very few are destined to ovulate; the majority will degenerate at some point along this lengthy developmental continuum (duration of weeks in rodents to months in humans) [28].

Progression of follicle development requires bidirectional communication between an oocyte and granulosa cells (GCs). Later stages of follicle development, including the process of follicle selection and ovulation, are also dependent on appropriately timed endocrine signals, notably, pituitary gonadotropins and a myriad of locally produced factors operating in an autocrine or paracrine manner to coordinate and control ovarian function [11]. Three growth factors, growth differentiation factor 9 (GDF9), bone morphogenetic protein 15 (BMP15), and fibroblast growth factor 8 (FGF8), are expressed in an oocyte-specific manner and play key roles in promoting follicle growth beyond the primary stage [10, 29, 40]. Furthermore, they determine the differentiation of cumulus cells and their responses to ovulation signals in preovulatory follicles [20, 39]. Therefore, they are also called oocyte-derived cumulus expansion-enabling factors [12, 42].

On the other hand, the accumulation and resetting of epigenetic marks during oocyte development enables the oocyte genome to establish the competence to generate a new organism [16, 46, 48]. Despite extensive mapping of oocyte chromatin patterns, the extent to which histone modifications form an instructive component of ovarian follicle development remains unclear. A hallmark of active chromatin is trimethylated H3 at lysine-4 (H3K4me3), which is enriched in active promoters and correlates with transcriptional activity [3, 6, 35]. Recent reports revealed deposition of broad domains of H3K4me3 in mature oocytes, predominantly in partially methylated regions [25, 46]. Mixed-lineage leukemia 2 (MLL2) and SET domain-containing 1 (SETD1)-Cxxc finger protein 1 (CFP1) histone methyltransferases are reported to confer this unusual distribution of H3K4me3 as well as the developmental competence onto mouse oocytes [2, 44]. Nevertheless, whether H3K4me3 contributes to oocytes’ capacity for facilitating follicle recruitment and ovulation in estrus cycles remains unexplored.

To address these outstanding questions, we took advantage of an oocyte-specific Cxxc1 knockout mouse model where H3K4 trimethylation is selectively ablated in oocytes of developing follicles [44]. Employing multiple experimental approaches, we demonstrated that the CFP1 deletion in growing oocytes impaired both follicle growth and ovulation in response to gonadotropin stimulations. CFP1-mediated histone H3 lysine-4 trimethylation in an oocyte was necessary to maintain the normal expression of key paracrine factors which facilitate the communication between an oocyte and the surrounding GCs. Therefore, the posttranslational histone modification of the oocyte genome not only gives developmental competence to a zygote after fertilization as demonstrated in previous studies, but also maintains the appropriate responses of ovarian somatic cells to gonadotropin in a cell-nonautonomous manner.

Results

Cxxc1 deletion in mouse oocytes decreases transcription activity in growing follicles

We have reported selective deletion of Cxxc1 in growing oocytes via crossing of Cxxc1fl/fl mice with Zp3-Cre mice [44]. In this study, we knocked out Cxxc1 in oocytes as early as the primordial follicle stage, using the well-characterized Gdf9-Cre (Cxxc1fl/fl;Gdf9-Cre, later termed as Cxxc1oo−/−). The strategy of developmental stage-specific Cxxc1 knockout is illustrated in Fig. 1a. As previously reported, both Cxxc1fl/fl;Gdf9-Cre and Cxxc1fl/fl;Zp3-Cre females were completely infertile [34, 44]. Efficient deletion of CFP1 protein in the oocyte of Cxxc1oo−/− mice at postnatal day (PD) 21 was confirmed by western blotting (Fig. 1b) and immunofluorescent staining on ovarian sections (Fig. 1c, arrows indicate oocytes). Nonetheless, the expression level of CFP1 in the surrounding granulosa cells was not affected by Cxxc1-deletion in the oocytes (Fig. 1c). It was shown that knockout of Mll2, another H3K4 methyltransferase, in mouse oocytes impairs H3K4me3 accumulation and caused defects of oocyte development [17]. Therefore, we compared the levels of mRNAs encoding Setd1a, Setd1b, Cxxc1, and Mll2, by quantitative RT-PCR in fully grown oocytes. The amplification efficiencies of each pair of PCR primers were normalized. The results demonstrated that the expression levels of Cxxc1 transcripts were more abundant than that of Mll2 transcripts in oocytes (Fig. 1d). Importantly, the expression of Setd1a, Setd1b, and Mll2 mRNAs was not significantly affected in Cxxc1-deleted oocytes (Fig. 1e). In addition, we detected an increase of MLL2 protein level in Cxxc1 null oocytes, probably due to a feedback mechanism trying to compensate the loss of CFP1 (Fig. 1b).

Fig. 1
figure 1

Expression of SETD1-CFP1 and MLL2 in Cxxc1-deleted oocytes. a The strategy of developmental stage-specific Cxxc1 knockout in mouse oocytes. b Western blotting results showing the protein levels of MLL2, CFP1, and H3K4me3 in fully grown GV oocytes collected from WT and Cxxc1oo−/− mice. Endogenous DDB1 was used as a loading control. Total proteins from 100 oocytes were loaded in each lane. c Immunofluorescence results showing the levels of CFP1 in the oocytes and GCs of WT and Cxxc1oo−/− mice at postnatal day (PD) 21. Arrows indicate oocytes. Scale bar 50 μm. Six ovaries from three females for each genotype were analyzed. d Quantitative RT-PCR results showing expression levels of mRNAs encoding Setd1a, Setd1b, Mll2, and Cxxc1 in fully grown GV oocytes. The expression level of Cxxc1 was set as 1, and levels of other transcripts were normalized accordingly. e Quantitative RT-PCR results showing expression levels of mRNAs encoding Setd1a, Setd1b, Mll2, and Cxxc1 in fully grown GV oocytes of WT and Cxxc1oo−/− mice. The expression level of each transcript in WT oocytes was set as 1. Error bars: SEM. ***P < 0.001 according to two-tailed Student’s t test; n.s. nonsignificant

Full size image

In the ovaries of wild-type (WT) mice on PD21, the H3K4me3 level in oocytes of primordial follicles (indicated by arrows) was comparable to that in the surrounding somatic cells, but gradually increased in oocytes during follicle growth, particularly at secondary and antral follicle stages (Fig. 2a, c). In comparison, H3K4me3 failed to accumulate in Cxxc1-null oocytes of growing follicles. The H3K4me3 level in GCs remained stable during follicle development and was not affected by the Cxxc1 knockout in oocytes (Fig. 2a). In WT oocytes, the phosphorylation level of RNA polymerase II (pS2, which reflects global transcriptional activity) was high in oocytes at the secondary follicle stage, but decreased at the antral follicle stage. On the other hand, the pS2 level was significantly lower than normal in Cxxc1-null oocytes at these stages (Fig. 2b, d).

Fig. 2
figure 2

Changes of gene expression and histone H3 trimethylation in Cxxc1 knockout oocytes. Immunofluorescence results showing the levels of H3K4me3 (a) and phosphorylated RNA polymerase II CTD repeat YSPTSPS (pS2) (b) in the oocytes and GCs of WT and Cxxc1oo−/− mice. The stages of follicles are specified. Arrows indicate oocytes within primordial follicles. Scale bar 50 μm. Six ovaries from three females for each genotype were analyzed. c, d Quantification of H3K4me3 and pS2 signals in a, b, respectively. Error bars: SEM. ***P < 0.001 according to two-tailed Student’s t test; n.s. nonsignificant. Relative expression of selected genes extracted from RNA-seq data (e) and determined by qRT-PCR (f). Error bars: SEM. ***P < 0.001, **P < 0.01, and *P < 0.05 according to two-tailed Student’s t test. g ChIP-qPCR analysis of H3K4me3 levels on the promoter regions of indicated genes. All enrichment values are relative to input values and then normalized with enrichment values of WT samples. Error bars: SEM. ***P < 0.001, **P < 0.01, and *P < 0.05 according to two-tailed Student’s t test

Full size image

Both RNA sequencing and quantitative RT-PCR results showed that genes encoding oocyte-specific paracrine factors, including Gdf9, Bmp15, Fgf8, c-Kit, and Oosp1, were significantly downregulated in Cxxc1-null oocytes at PD21 (Fig. 2e, f). So was Gja4, which encodes a connexin-37 subunit crucial for oocyte–GC communication [33, 37]. To further verify whether oocyte-specific paracrine factors are regulated by H3K4me3 levels, we performed chromatin immunoprecipitation (ChIP)-qPCR assays using 300 GV oocytes to investigate the levels of H3K4me3 on the promoter regions of Gdf9, Bmp15, Fgf8, c-Kit, Oosp1, and Gja4. The results showed that the H3K4me3 levels on the promoters of these genes in Cxxc1-null oocytes were significantly lower than those in WT oocytes (Fig. 2g). These results indicate that the downregulation of these genes was caused by the effect of the SETD1–CFP1 complex on transcription-coupled H3K4 trimethylation at the promoter regions.

Follicle growth is impaired by the Cxxc1 knockout in oocytes

At PD28, which is the stage where female mice undergo the transition from puberty to sexual maturation, the ovaries of Cxxc1oo−/− mice were already smaller than those of their control litter mates (Fig. 3a, b). Six-month-old adult Cxxc1oo−/− females also had smaller ovaries than did WT mice of the same age (Fig. 3a, b). Because our previous study has already proved that the Cxxc1 knockout does not affect the survival of primordial follicles [44], we hypothesized that CFP1 in oocytes might be necessary to support follicle development beyond this stage.

Fig. 3
figure 3

Follicle development in the ovaries of Cxxc1oo−/− mice. Size (a) and weight (b) comparisons between WT and Cxxc1oo−/− ovaries at 4 weeks and 6 months; n: the number of ovaries analyzed for each genotype. Error bars: SEM; 0.01 < *P < 0.05 according to two-tailed Student’s t test. Scale bar 500 μm. c, e H&E staining results showing ovarian histology of 4-week-old and 6-month-old WT and Cxxc1oo−/− mice. Scale bar 100 μm. Serial follicle counts in ovaries of 4-week-old (d) and 6-month-old (f) WT and Cxxc1oo−/− mice. Six ovaries from different females for each genotype were analyzed. Error bars: SEM. ***P < 0.001, **P < 0.01, and *P < 0.05 according to two-tailed Student’s t test; n.s. nonsignificant

Full size image

Ovarian histological analyses at PD28 indicated that the numbers of primordial and primary follicles were similar between WT and Cxxc1oo−/− mice, suggesting that the primordial follicle formation and activation processes were not affected by the Cxxc1 deletion in oocytes (Fig. 3c, d). Nevertheless, there were much fewer early secondary follicles (which contain 2–3 layers of GCs) and antral follicles (characterized by the presence of an antral cavity) in the Cxxc1oo−/− ovaries than in normal ones (Fig. 3c, d). The numbers of late secondary follicles that contain more than three layers of GCs also moderately decreased in Cxxc1oo−/− ovaries but not as significantly as in the early secondary follicles and antral follicles. A likely explanation for this phenomenon is that the transition from the secondary to antral follicle stage was impaired in Cxxc1oo−/− ovaries and, therefore, caused relative accumulation of late secondary follicles. Follicles and corpus lutea (CL) at various developmental stages were also present on ovarian sections of 6-month-old Cxxc1oo−/− mice (Fig. 3e). Similar to the results of follicle counts at 4 weeks old, the number of secondary follicles, but not primordial and primary follicles, decreased in the ovary of 6-month-old Cxxc1oo−/− mice, but the numbers of primary follicles were not affected (Fig. 3f). These results further indicated that Cxxc1 deletion in oocytes caused follicle growth retardation instead of premature ovarian failure.

GC proliferation and apoptosis were affected by the Cxxc1 knockout in oocytes

At PD28, increased GC apoptosis and follicle atresia were detected by immunohistochemical staining of cleaved caspase 3 (CC3) and a TdT-mediated dUTP Nick-End Labeling (TUNEL) assay in the ovaries of Cxxc1oo−/− mice in contrast to the ovaries of control mice (Fig. 4a). Quantifying atretic follicles versus total follicles on ovarian sections indicated that the ovaries of Cxxc1oo−/− mice had greater numbers of atretic follicles (Fig. 4b). On the contrary, fewer dividing GCs labeled with the mitotic cell marker phosphorylated histone H3 at serine-10 were detected in the growing follicles of Cxxc1oo−/− mice than in those of control mice (Fig. 4c, d).

Fig. 4
figure 4

Impaired signaling pathways and gene expression patterns in the developing follicles of Cxxc1oo−/− mice. a Immunohistochemical data on cleaved caspase 3 and TUNEL assay show an increase in GC apoptosis and follicle atresia in 4-week-old Cxxc1oo−/− ovaries as compared with WT. Scale bar 100 μm. Asterisks indicate atretic follicles. b Quantification of TUNEL-positive atretic follicles in 4-week-old WT and Cxxc1oo−/− ovaries. Error bars: SEM. **P < 0.01 according to two-tailed Student’s t test. Immunohistochemical (c) and quantification (d) results for phospho-histone H3 (pH3-Ser10) indicating GC proliferation in 4-week-old WT and Cxxc1oo−/− ovaries. Scale bar 100 μm. Error bars: SEM. *P < 0.05 according to two-tailed Student’s t test. e Quantitative RT-PCR results for the expression levels of FSH receptor (Fshr) mRNA in GCs and COCs isolated from 4-week-old WT and Cxxc1oo−/− ovaries. Error bars: SEM; n.s. nonsignificant. f Immunofluorescence results on p-AKT(Ser 473) in 4-week-old WT and Cxxc1oo−/− ovaries. Arrows indicate oocytes within activated follicles. Scale bar 100 μm. Arrows indicate growing follicles. g Immunohistochemical results on p-RPS6 (Ser235/236) in 4-week-old WT and Cxxc1oo−/− ovaries. Scale bar 100 μm. h Western blot analysis of PI3K pathway components, including FOXO1, p-AKT (Ser 473), and p-RPS6 (Ser235/236) in 4-week-old WT and Cxxc1oo−/− ovaries. ERK1 and -2 served as a loading control. Total-protein samples were extracted from four ovaries in each group, and 30 μg of total protein was loaded per lane; each experiment was repeated at least three times. i Quantitative RT-PCR results on mRNA levels of genes specifically expressed in COCs (Npr2, Slc38a3, and Amh) and genes more expressed in GCs (Nppc, Cyp19a1, and Cyp11a1) from WT and Cxxc1oo−/− ovaries at 44 h after PMSG injection. Error bars: SEM. **P < 0.01 and *P < 0.05 according to two-tailed Student’s t test; n.s. nonsignificant

Full size image

FSH receptor (encoded by Fshr gene) expressed by GCs is essential for follicle growth beyond the early antral follicle stage [9]. Expression of Fshr was not statistically affected by the Cxxc1 knockout in oocytes (Fig. 4e). Because FSH stimulates follicle growth and prevents atresia by elevating phosphoinositide 3-kinase (PI3K) pathway activity [13, 23], we evaluated the amounts of phosphorylated AKT and ribosome protein S6 (pRPS6) in the ovaries of WT and Cxxc1oo−/− mice by immunofluorescent and immunohistochemical analyses, respectively. AKT is a key component of the PI3K pathway and a multifunctional kinase. Its general function is to inhibit apoptosis and to stimulate cell proliferation [5]. Besides, activated AKT increases the cellular protein synthesis rate by inducing sequential phosphorylation of its downstream proteins P70 ribosome protein S6 kinase (P70RSK) and ribosome protein S6 (RPS6) [32]. The growing follicles of Cxxc1oo−/− mice had significantly lower p-AKT and pRPS6 levels compared with those in control mice (Fig. 4f, g). The reduced activity of the PI3K signaling pathway was next verified by western blot analyses of GCs isolated from the ovaries of 4-week-old WT and Cxxc1oo−/− mice (Fig. 4h). Amounts of phosphorylated AKT, P70RSK, and RPS6 all decreased in GCs from Cxxc1oo−/− mice. In addition, transcriptional repressor forkhead box O1 (FOXO1) is highly expressed in GCs of slowly growing follicles. It represses GC proliferation and steroidogenesis and facilitates follicle atresia. Normally, an activated PI3K pathway releases follicles from this inhibitory state by triggering phosphorylation and degradation of FOXO1 [27]. Our western blot results indicated that the protein levels of FOXO1 were significantly higher in GCs of Cxxc1oo−/− ovaries (Fig. 4h). Collectively, these results suggested that the Cxxc1 deletion in oocytes indirectly impaired PI3K signaling in the surrounding GCs and facilitated follicle atresia.

CFP1 in an oocyte is required for maintaining a specific gene expression pattern in cumulus cells

Cumulus cells, which are closely associated with an oocyte, promote oocyte growth and developmental competence through complex bidirectional interactions with the oocyte. Influenced by concentration gradients of oocyte-secreted factors, mural and cumulus cells each express a subset of different transcripts [8]. The C-type natriuretic peptide–natriuretic peptide receptor 2 (NPR2) signaling pathway is essential for maintaining high cGMP levels in cumulus–oocyte complexes (COCs) and for preventing precocious germinal vesicle breakdown (GVBD) [47]. Nppc is a gene that encodes secreted factor C-type natriuretic peptide and is highly expressed in mural GCs. In contrast, the gene of its receptor, Npr2, is more prominently expressed in cumulus cells than in mural GCs (Fig. 4i). This result is in agreed with the reported observation [47]. In the COCs of PMSG-primed Cxxc1oo−/− mice, the mRNA levels of Npr2 and Nppc were dramatically lower and higher as compared with WT controls, respectively (Fig. 4i). Similarly, other genes that are preferentially expressed by COCs rather than GCs, e.g., Slc38a3 and Amh, were downregulated after the Cxxc1 knockout in oocytes. By contrast, the genes that normally have a lower expression level in COCs than in GCs, including Cyp19a1 and Cyp11a1, were upregulated in COCs of Cxxc1oo−/− mice. Overall, the expression levels of these genes in GCs were not significantly affected by the Cxxc1 knockout in oocytes. Collectively, these results indicated that the distinct gene expression patterns in cumulus cells within preovulatory follicles were disrupted by Cxxc1 deletion in oocytes.

The in vitro growth defect of Cxxc1oo−/− follicles can be partially rescuing by the PI3K activator 740Y-P

It has been reported that artificial activation of PI3K pathway in the mouse and human ovaries was able to stimulate follicle growth [41]. Therefore, we cultured secondary follicles isolated from Cxxc1oo−/− mice in medium-containing a PI3K activator 740Y-P (100 µg/ml). This treatment had a rescuing effect and increased the rate of Cxxc1oo−/− follicles that developed to the antral stage (Fig. 5a, b). We also investigated if PI3K pathway in the GCs of Cxxc1oo−/− follicles was indeed activated by 740Y-P treatment. Western blot results showed that PI3K pathway components including AKT, P70S6K and RPS6 were all phosphorylated after 740Y-P treatment, in GCs from both WT and Cxxc1oo−/− mice (Fig. 5c). Furthermore, we isolated COCs from cultured WT and Cxxc1oo−/− follicles (with or without 740Y-P treatment) and further cultured with or without EGF, a stimulator of cumulus cell expansion. Compact COCs with multiple layers of cumulus cells could be isolated from WT follicles after culture, and these COCs underwent expansion after EGF treatment (Fig. 5d). However, COCs isolated from Cxxc1oo−/− follicles contained less cumulus cells which frequently dissociated from the oocytes after culture. The presence of 740Y-P in medium improved the morphology of the COCs as well as their in vitro expansion responses to EGF (Fig. 5d).

Fig. 5
figure 5

Rescuing effect of the PI3K activator 740Y-P on cultured Cxxc1oo−/− follicles. a Representative images of secondary follicles isolated from WT and Cxxc1oo−/− ovaries and the same follicle at 12 days after culture with or without adding the PI3K activator 740Y-P. Scale bars 100 μm. b Rates of secondary follicles that developed to the antral stage after 12 days’ culture. c Western blot analysis of PI3K pathway components in cultured GCs before and after 740Y-P (100 µg/ml) treatment for 1 h. Total AKT was blotted as a loading control. d COCs isolated from WT and Cxxc1oo−/− follicles (with or without 740Y-P treatment) were further cultured in medium with or without adding 100 ng/ml EGF for 16 h before being imaged. Scale bar 150 μm

Full size image

CFP1 in oocytes regulates GC function during ovulation by a cell-nonautonomous mechanism

Although the follicles containing Cxxc1-null oocytes have a low growth rate under natural conditions, the first wave of follicle development to the preantral stage was normal in the ovaries of 3-week-old Cxxc1oo−/− mice. PMSG injection at PD21–23 stimulated the formation of comparable numbers of large antral stage follicles in the ovaries of Cxxc1oo−/− mice (Fig. 6a, b). Nonetheless, subsequent hCG injection failed to induce effective meiotic resumption (characterized by GVBD, Fig. 6c) and ovulation (Fig. 6d) in these mice. At 8 h after hCG injection, ovaries of Cxxc1oo−/− mice were much smaller than the control ovaries, indicating a poor response to gonadotropins (Fig. 6e). Cumulus expansion was significantly impaired in the preovulatory follicles of Cxxc1oo−/− mice when compared with that in WT mice (Fig. 6f). Subsequently, preovulatory follicles did not rupture at 16 h after hCG injection (Fig. 6g). In an in vitro cumulus expansion assay, EGF mimicked the function of EGF-like factors and induced expansion of cultured COCs isolated from WT mice. In contrast, COCs collected from Cxxc1oo−/− mice were nonresponsive to the EGF treatment (Fig. 6h).

Fig. 6
figure 6

The impact of oocyte-specific Cxxc1 deletion on ovulation. a H&E staining results showing ovarian histology of 3-week-old WT and Cxxc1oo−/− mice 44 h after PMSG priming. Scale bar 100 μm. b Antral follicle counts of 3-week-old WT and Cxxc1oo−/− ovaries after PMSG priming for 44 h. The numbers of ovaries used for each genotype (n) are indicated. Error bars: SEM; n.s. nonsignificant. c In vivo GVBD rates in antral follicles of 3-week-old WT and Cxxc1oo−/− mice at 8 h after hCG injection. Four females of each genotype (n) were used in this experiment. Error bars: SEM. ***P < 0.001 according to two-tailed Student’s t test. d Numbers of oocytes ovulated in 3-week-old WT and Cxxc1oo−/− mice at 16 h after hCG injection. Four females of each genotype (n) were employed in this experiment. Error bars: SEM. ***P < 0.001 according to two-tailed Student’s t test. e Images of ovaries of 3-week-old WT and Cxxc1oo−/− mice at 8 h after hCG injection. Scale bar 500 μm. f H&E staining results showing ovarian histology of 3-week-old WT and Cxxc1oo−/− mice at 8 h after hCG injection. Scale bar 100 μm. g H&E staining results showing ovarian histology of 3-week-old WT and Cxxc1oo−/− mice at 16 h after hCG injection. Scale bar 100 μm. h COC (collected from 4-week-old WT and Cxxc1oo−/− mice) expansion induced by EGF at 16 h of culturing. Scale bar 100 μm. i Immunofluorescence assay results showing the FOXO1 expression in GCs of WT and Cxxc1oo−/− ovaries after PMSG (44 h) and hCG injection (8 h). Scale bar 100 μm. j Serum progesterone (P4) concentrations of WT and Cxxc1oo−/− mice at 8 h after hCG injection. Four females of each genotype (n) were used in this experiment. Error bars: SEM; 0.01 < *P < 0.05 according to two-tailed Student’s t test

Full size image

Some studies have shown that transcription factor FOXO1 is highly expressed in growing follicles and is a negative regulator of steroidogenesis [18, 26, 27]. LH or hCG induces degradation of FOXO1 before ovulation and stimulates steroidogenesis and terminal differentiation of GCs [27]. We found that FOXO1 expression remained high in Cxxc1oo−/− ovaries, even after hCG treatment (Fig. 6i), indicating that GC differentiation was impaired after the Cxxc1 deletion in oocytes. This phenomenon was indeed associated with a lower serum progesterone level in hCG-treated Cxxc1oo−/− mice than that in control mice (Fig. 6j). Because the CFP1 protein is deleted only in oocytes (not in mural granulosa or cumulus cells), these results that CFP1 in oocytes was required for ovulation in a cell-nonautonomous manner.

Ovulation-stimulating signals were compromised in granulosa and cumulus cells of Cxxc1oo−/− ovaries

In the previous studies, we have shown that the preovulatory LH surge induces transient activation of extracellular signal regulated kinases 1 and -2 (ERK1 and 2) in granulosa and cumulus cells, and that ERK1 and -2 activities are indispensable for LH target gene expression and for ovulation [15]. Thus, we determined protein levels of phosphorylated and activated ERK1 and -2 and their target genes in the GCs of Cxxc1oo−/− mice by immunofluorescent staining, western blot, and/or qRT-PCR. These results revealed that ERK1 and -2 phosphorylation remarkably decreased in the granulosa and cumulus cells of Cxxc1oo−/− mice at 2 h after hCG injection (Fig. 7a, b). The LH downstream genes—that are regulated by ERK1 and -2 and are essential for cumulus expansion (Ptgs2, Ptx3, Has2, and Tnfaip6) and for ovarian endocrine functions (Sult1e1 and Star) during ovulation [15]—had significantly lower expression levels in both COCs and GCs of Cxxc1oo−/− mice, when compared with those of control mice, at 4 h after hCG injection (Fig. 7c). Notably, in agreement with our observation that CFP1 is required for COCs to maintain their unique gene expression pattern, we noted derepression of LH receptor expression in COCs of Cxxc1oo−/− mice (Fig. 7c). Furthermore, CEBPβ (a transcription factor regulated by ERK1 and 2), which is crucial for ovulation [14], was not effectively induced in the ovaries of Cxxc1oo−/− mice after hCG injections (Fig. 7b, d).

Fig. 7
figure 7

Detection of ovulation-related signaling events in the ovaries of WT and Cxxc1oo−/− mice. a Immunofluorescence assay results for phosphorylated ERK1 and -2, SMAD2 and -3, and SMAD1, -5, and -8 in the ovaries of WT and Cxxc1oo−/− mice at 2 h after hCG injection. Scale bar 100 μm. b Western blot results showing levels of phosphorylated ERK1/-2, SMAD1/5/8, and CEBPβ in the ovaries of WT and Cxxc1oo−/− mice before and after hCG injection (0, 2, and 4 h). Total ERK1/2 was blotted as a loading control. Total proteins were extracted from GCs of four ovaries in each group, and 30 μg proteins were loaded per lane; each experiment was repeated for at least three times with the similar results. c qRT-PCR results on the expression levels of LH target genes in GCs and COCs isolated from WT and Cxxc1oo−/− ovaries (0 and 4 h after hCG injection). Error bars: SEM. ***P < 0.001, **P < 0.01, and *P < 0.05 according to two-tailed Student’s t test. d Immunofluorescence assay results showing the expression of transcription factor CEBPβ in the ovaries of WT and Cxxc1oo−/− mice before and after hCG injection (2 and 8 h). Scale bar 100 μm. e Immunofluorescence assay results showing the expression of PTGS2 and PTX3 in the ovaries of WT and Cxxc1oo−/− mice after hCG injection (8 h). Scale bar 100 μm

Full size image

Some studies have also shown that oocyte-secreted GDF9 and BMP15 induce phosphorylation and activation of their downstream signaling molecules SMAD2 and -3 and SMAD1, -5, and -8, respectively, in adjacent cumulus cells, and that this signaling pathway is a prerequisite for successful cumulus expansion and ovulation [24, 31, 45]. Because the expression of Gdf9 and Bmp15 decreased in Cxxc1-null oocytes, the TGF-β signaling between a CFP1-deficient oocyte and its surrounding GCs might be impaired. As presented in Fig. 7a, pSMAD1, -5, and -8 and pSMAD2 and -3 levels were significantly lower in cumulus cells that surrounded Cxxc1-null oocytes when compared with those that surrounded WT oocytes. The decrease of pSMAD1/5/8 levels in GCs of Cxxc1oo−/− mice when compared with WT mice was also confirmed by western blot (Fig. 7b). Similarly, the protein expression of oocyte factor (GDF9, BMP15, and FGF8)-regulated, ovulation-related genes in cumulus cells, including prostaglandin synthase 2 (PTGS2) and pantraxin-3 (PTX3), decreased in the preovulatory follicles of Cxxc1oo−/− mice (Fig. 7e). This result is consistent with the qRT-PCR data in Fig. 7c.

We also checked the mRNA expression of genes coding for EGF-like paracrine factors (Areg, Ereg, and Btc) in granulosa and cumulus cells. These three factors are intrafollicular mediators of LH actions, and are known to be essential for ovulation [36]. Although the mRNA level of Btc was lower in granulosa and cumulus cells of Cxxc1oo−/− mice after hCG injection, mRNA levels of Areg and Ereg were not significantly affected by the Cxxc1 knockout in oocytes (Fig. 8a). Nor was the LH-induced expression of transcription factor GATA-binding protein 4 (GATA4) [30] affected in the ovaries of hCG-treated Cxxc1oo−/− mice (Fig. 8b). These results indicated that the Cxxc1 knockout in oocytes impaired the expression of only a subset of LH-regulated ovulation-related genes in granulosa and cumulus cells.

Fig. 8
figure 8

Expression of EGF-like factors and GATA4 in the ovaries of WT and Cxxc1oo−/− mice. a qRT-PCR results on the mRNA expression of three EGF-like factors (Areg, Ereg, and Btc) in GCs and COCs isolated from WT and Cxxc1oo−/− ovaries (0 and 4 h after hCG injection). Error bars: SEM. **P < 0.01 and *P < 0.05 according to two-tailed Student’s t test; n.s.: nonsignificant. b Immunofluorescence assay results showing expression of GATA4 in the ovaries of WT and Cxxc1oo−/− mice before and after hCG injection (2 and 8 h). Scale bar 100 μm. c A diagram showing functions of CFP1 in mouse oocytes of growing and preovulatory follicles. Accompanied by FSH stimulation, CFP1-SETD1-mediated H3K4 trimethylation in oocytes facilitates the survival of growing follicles and the transition from the secondary to antral follicle stage. In fully grown mouse oocytes of preovulatory follicles, CFP1 is required for oocyte expression of the cumulus expansion-enabling factors, which activate ovulation signaling pathways in the surrounding cumulus cells and maintain COC responsiveness to the ovulation signal triggered by LH

Full size image

Discussion

In this study, we identified CFP1 in oocytes as a key regulator of ovarian functions, particularly follicle development and ovulation, as summarized in Fig. 7c. The H3K4me3 level was low in growing oocytes but high in fully grown oocytes, indicating that H3K4me3 is a marker of epigenetic maturation. Defective H3K4 trimethylation in growing oocytes under the influence of the Cxxc1 knockout impaired the responses of GCs to gonadotropin signals, implying that balanced H3 K4 methylation of the oocyte genome not only is essential for maintaining transcription within the oocyte but also determines the developmental fate of the follicles.

Paracrine factors secreted by oocytes have profound effects on follicle growth and development both before and after the LH surge. For example, the absence of oocyte-derived GDF9 results in a failure to progress beyond the early preantral stages of follicle development [10]. A lack of BMP15, another oocyte-secreted protein, impairs cumulus cell differentiation and cumulus expansion [43]. In particular, oocyte-secreted paracrine factors are necessary to enable upregulation of Has2, Ptgs2, Ptx3, and Tnfaip6 transcripts during cumulus expansion. Each of these transcripts is absolutely required for cumulus expansion, because the phenotype of null mutations in Ptgs2, Ptx3, or Tnfaip6 or inhibition of HAS2 activity severely compromises cumulus expansion.

The CFP1 deletion affected the expression of a wide range of maternal genes, regardless of their original expression magnitude [44]. In addition, deletion of the Mll2 gene, which encodes another histone H3 lysine-4 trimethylase, in mouse oocytes impairs H3K4me3 accumulation and causes an ovulation defect [2]. Therefore, the obvious question is how those two distinct H3K4 methyltransferases split the labor in terms of H3K4me3 induction. Although the oocyte-specific Mll2 mice have ovulation defects, the expression levels of Gdf9, Bmp15, and c-Kit in oocytes were not affected by Mll2 KO [2]. In contrast, these genes were downregulated in Cxxc1 KO oocytes. Therefore, CFP1 might be more important than MLL2 in maintaining expression of these oocyte-derived paracrine factors and the normal communications between oocytes and surrounding granulosa/cumulus cells. Oocyte-specific knockout of murine Setd1b also caused a reduction in ovary size, an increase in atretic follicles, and ovulation defects. These phenotypes are comparable with those observed after CFP1 ablation in oocytes [4]. Because CFP1 is part of the complex that recruits SETD1B to chromatin, the overlap in the phenotypes of the knock-outs further strengthened the notion that the SETD1–CFP1 complex in oocyte is required to maintain normal ovarian functions.

Although GCs express many mRNA transcripts and proteins in common, the mural and cumulus cells each express a subset of different transcripts, with markedly different steady-state levels. The cumulus cells surrounding the oocyte are specialized to service the development of the oocyte, and steroidogenesis is the principal role of mural GCs that line the follicle wall [8, 38]. The findings in this report show that Cxxc1 deletion in mouse oocytes decreases mRNA transcript levels of cumulus markers. In addition, TGF-β signaling is positively involved in cumulus expansion, and upregulation of Ptx3, Ptgs2, and Has2 mRNAs was compromised in follicles containing the Cxxc1-null oocytes. By contrast, expression of mural GC transcripts increased in cumulus cells of Cxxc1oo−/− ovaries. Thus, without the CFP1-mediated H3K4me3 accumulation, oocytes failed to establish opposing gradients of influence in the follicle toward the pituitary-secreted FSH. Only the oocytes with appropriate H3K4me3 levels are capable of specifying the mural and cumulus GC phenotypes that are pivotal for follicle development and ovulation.

In recent years, the features and establishment of epigenetic modifications in developing oocytes were extensively studied. Nonetheless, the potential involvement of these epigenetic events in the regulation of ovarian function via cell-nonautonomous mechanisms has not been assessed. In the ovaries of Cxxc1oo−/− mice, the primordial follicle pool is intact, but awakened follicles have low efficiency of development into antral follicles. Instead, they have a higher-than-normal rate of becoming atretic. This phenomenon is most likely due to the lack of PI3K signaling activity in growing follicles. It has been recognized that oocytes coordinate their own survival, GC proliferation, and follicle growth by secreting local signaling molecules and communicate with GCs through gap junctions. KIT and KIT ligand (KITL) have well-documented functions in the proliferation, migration, and differentiation of ovarian follicle cells, primarily by activating the PI3K signaling pathway [19]. KIT is expressed on the primordial oocyte membrane, while KITL is produced by granulosa cells. Therefore, the significant decrease of c-Kit expression in Cxxc1-null oocytes might be a major reason for the weakened AKT activity in the surrounding GCs. Consequently, the compromised AKT activity led to the increase of FOXO1 protein level and the high apoptosis rate of GCs. Our results also indicate that the expression of Gja4, which encodes oocyte-specific gap junction protein connexin-37 [37], is low in Cxxc1-null oocytes. Recently, small molecules that modulate the PI3K pathway have been used for treating the ovarian tissues of human patients with primary ovarian insufficiency for assisted reproduction purposes [22, 41]. On the other hand, it has been unclear whether CFP1 function and H3K4me3 accumulation are also stimulated in oocytes alone with follicle awakening. Our observations in this novel mouse model suggest that the key role of CFP1 in oocytes of growing follicles should be considered when using these new drugs in clinical practice.

FSH supports follicle growth beyond the early antral stage [9]. In the current study, we provide evidence that depletion of Cxxc1 impairs the physiological effect of FSH on follicle growth. Thus, we hypothesize that in WT mice, CFP1 together with SETD1 histone methyltransferase promotes accumulation of H3K4me3 in oocytes of growing follicles, thereby increasing the secretion of oocyte-derived paracrine factors as well as the communication between oocytes and cumulus cells and leading to increased follicular responses to gonadotropins (Fig. 8c). This positive feedback loop may potentially perform a key function in follicle selection during each estrous cycle.

Materials and methods

Animals

Mice were maintained in a controlled environment of 20–22 °C, with a 12 h/12 h light/dark cycle, 50–70% humidity, and food and water provided ad libitum. All mice used were handled according to the Animal Research Committee guidelines of Zhejiang University. Cxxc1fl/fl mouse strain has been engineered and described previously [7]. They were bred over the Gdf9-Cre transgenic mice, which allow CRE-mediated recombination specifically in the oocytes at the primordial follicle stage [21]. All mice have a C57BL/6J genetic background.

Superovulation

To study ovarian responses to exogenous gonadotropins, female mice at postnatal day (PD) 23 were analyzed to avoid the complexity of ovarian functions associated with estrous cycles and endogenous surges of gonadotropins. Immature mice were injected intraperitoneally with 5 IU PMSG (pregnant mare serum gonadotropin; Calbiochem) to stimulate preovulatory follicle development followed 48 h later with 5 IU hCG (human chorionic gonadotropin, American Pharmaceutical Partners, Schaumburg, IL, USA) to stimulate ovulation.

In vitro COC expansion assay

Fully grown, non-expanded COCs were collected from ovaries of PMSG-primed immature mice. COCs (~ 30) were plated in 100 µl of defined COC medium (MEM w/ES, 25 mM HEPES, 0.25 mM sodium pyruvate, 3 mM l-glutamine, 1 mg/ml BSA, 100 U/mL penicillin, and 100 μg/ml streptomycin) with 1% FBS under the cover of mineral oil and treated with EGF (100 ng/ml) or FSH (500 nM). Expansion status was evaluated by microscopic examination after overnight culture.

Histological analysis

Ovaries were collected and fixed in formalin overnight, processed, and embedded in paraffin using standard protocols. IHC was performed using standard protocols. For immunofluorescence, sections or cells were sequentially probed with primary antibodies as indicated in Results and Alexa Fluor 594- or 488-conjugated secondary antibodies (Molecular Probes). Slides were mounted using VectaShield with 4′,6-diamidino-2-phenylindole (DAPI, Vector Laboratories). Digital images were acquired using an epifluorescence microscope (Nikon Eclipse 80i) with 4–100× objectives. Semi-quantitative analysis of the fluorescence signals was conducted using the NIH Image analysis program ImageJ. The relative H3K4me3 and pS2 levels were quantified by comparing the Alexa Fluor 594 or 488 fluorescence intensities with the intensity of DAPI staining within the same oocyte. The blue channel of DAPI staining was not presented in the original figure due to the space limitation. The antibodies used are listed in Supplementary Table 1.

Follicle counts

Ovaries were serially sectioned at 5 μm and stained with hematoxylin and eosine (H&E). Numbers of follicles were determined by recording the numbers of oocyte nuclei on all sections. These criteria were used to determine the follicle stages: primordial follicles, small oocytes surrounded by a single layer of flat granulosa cells; primary follicles, middle sized oocytes surrounded by a single layer of cubical granulosa cells; secondary follicles, large oocytes surrounded by more than two layers of granulosa cells; antral follicles and presence of follicle cavity.

TUNEL assay

TUNEL assays were performed on 10% formalin-fixed paraffin-embedded sections using the ApopTag Plus Peroxidase In Situ Apoptosis Detection Kit (Serologicals Corporation, Norcross, GA, USA) according to the manufacturer’s instructions.

Hormone-level assays

Mice were anesthetized and blood was collected by cardiac puncture. Serum and blood cells were separated by centrifugation. Serum hormone levels were determined by Di’an Medical Diagnostics Limited Corporation (Hangzhou, China).

Western blot

Ovaries were lysed in 2-mercaptoethanol containing loading buffer and heated at 95 °C for 5 min. SDS-PAGE and immunoblots were performed following the standard procedures using a Mini-PROTEAN Tetra Cell System (Bio-Rad, Hercules, CA, USA). The antibodies used are listed in Supplementary Table 1.

Quantitative RT-PCR

Total RNA was extracted from oocytes using the RNeasy Mini kit (Qiagen) according to the manufacturer’s instruction, followed by reverse transcription (RT) using Superscript RT kit (Bio-Rad). Quantitative RT-PCR was performed using a Power SYBR Green PCR Master Mix (Applied Biosystems, Life technologies) with ABI 7500 Real-Time PCR system (Applied Biosystems) using primers listed in Supplementary Table 2. For each indicated gene, the relative transcript level of the control sample (left-hand bar of each graph) was set as 1. The relative transcript levels of other samples were compared to the control, and the fold-changes shown in the graph. For each experiment, qPCR reactions were done in triplicate.

ChIP-qPCR

For ChIP-qPCR, 300 oocytes were used per reaction. The procedure of ULI-NChIP was carried out as previously described [46]. 1 μg of histone H3K4me3 antibody (Cell Signaling Technology, catalog no.9727) was used for each immunoprecipitation reaction. Then, we used 3 ng/μl DNA obtained from the ChIP experiments and 10 ng/μl input DNA for ChIP-qPCR analysis. The primers used in the qPCR experiment are listed in Supplementary Table 2.

Statistical analysis

The experiments were randomized and were performed with blinding to the conditions of the experiments. No statistical method was used to predetermine sample size. Results are given as mean ± SEM. Each experiment included at least three independent samples and was repeated at least two times. Results for two experimental groups were compared by two-tailed unpaired Student’s t tests. Statistically significant values of P < 0.05, P < 0.01, and P < 0.001 are indicated by asterisks (*), (**), and (***), respectively.

Publicly available data

Raw sequencing reads were obtained from Gene Expression Omnibus and processed with the parameters detailed below: Cxxc1-null oocytes (GSE85019).