GSK-3β displayed decreased activity in fetal oocytes in mice
To explore the physiological function of GSK-3β during early oogenesis, we first examined the specific location of GSK-3β in mouse ovaries. Immunofluorescence detection showed that GSK-3β (green) was extensively expressed in mouse ovaries from 13.5 dpc to 1 dpp (days postparturition) and was primarily located within the cytoplasm of both somatic (arrowhead) and germ cells (arrow), which were marked by DEAD-Box Helicase 4 (DDX4) (red). Hoechst (blue) was used to mark the cell nucleus. On 1 dpp, when the primordial follicles began to form, GSK-3β was expressed in the cytoplasm of both oocytes and the surrounding pre-granulose cells of the primordial follicle (dashed line) (Fig. 1a). Meanwhile, we detected the expression of p-GSK-3β (Ser9), which is the inactive form of GSK-3β, in fetal and neonatal ovaries (Fig. 1b). Different from the consistent expression pattern of total GSK-3β, p-GSK-3β was almost undetectable in fetal oocytes before 17.5 dpc. Thereafter, p-GSK-3β appeared in the cytoplasm of the part of the oocytes on 17.5 dpc (arrowhead) and displayed apparent expression in most of the oocytes in 1 dpp ovaries (arrowhead). This progressively increasing expression of p-GSK-3β in fetal oocytes indicated decreasing GSK-3β activity correlating with meiotic prophase progression.
Furthermore, to determine the expression level of GSK-3β in fetal oocytes, ovarian somatic and germ cell components were separated before examination (Additional file 1: Figure S1). Western blotting results from the fetal oocyte component demonstrated that total GSK-3β protein was constantly but invariantly expressed from 13.5 dpc to 1 dpp, while p-GSK-3β expression increased significantly from 17.5 dpc onward (Additional file 1: Figure S1). The increased protein level of p-GSK-3β confirmed the downregulation of GSK-3β activity in fetal oocytes, which implied a potential functional role of GSK-3β during early meiotic prophase I in mice.
Inhibition of GSK-3β led to dramatic fetal oocyte loss during meiotic prophase I
To thoroughly study the developmental stage-dependent role of GSK-3β in fetal ovarian development, the GSK-3β-specific inhibitor BIO (6-bromoindirubin-3′-oxime) was applied to block its activity in an in vitro culture system [30].
First, the role of GSK-3β in PGC proliferation and meiosis initiation was studied. Because the majority of PGCs switched from mitosis to meiosis at approximately 13.5 dpc in mice, 12.5 dpc ovaries were cultured for 2 days in vitro (equaling 14.5 dpc) with 1 μM dimethylsulfoxide (DMSO, as a control) or BIO. The 5-bromo-2-deoxyuridine (BrdU) labeling assay showed that proliferating PGCs (co-stained with both DDX4 antibody and BrdU antibody) were comparable to those of the control following GSK-3β inhibition (Additional file 2: Figure S2A), as was confirmed by the statistical analysis (58.00 ± 11.26 for BIO versus 55.50 ± 12.63 for the control per section; P > 0.05) (Additional file 2: Figure S2B). Similar results were obtained when the cultured ovaries were stained with another cellular proliferation marker, proliferating cell nuclear antigen (PCNA) (144.40 ± 33.43 for BIO versus 133.67 ± 30.73 for the control per section; P > 0.05) (Additional file 2: Figure S2C-D). Moreover, to examine the meiotic initiation in PGCs following GSK-3β inhibition, cultured ovaries were stained with an antibody against synaptonemal complex protein 3 (SYCP3) to mark the germ cells that had entered meiosis. The results showed that SYCP3 showed a weak appearance on 13.5 dpc but became intensively expressed in the germ cell nuclei from the leptotene stage onward (Additional file 2: Figure S2E) [31]. As shown in Additional file 2: Figure S2E, the majority of the germ cells from both the control and GSK-3β-inhibited group entered meiosis normally. Taken together, inhibition of GSK-3β had no significant impact on PGC proliferation or meiosis initiation in the fetal ovaries in mice.
Next, to explore the function of GSK-3β during meiotic prophase I, 14.5 dpc ovaries were cultured for 2, 3, and 4 days (equaling 16.5 dpc, 17.5 dpc, and 18.5 dpc, respectively) with BIO. Immunofluorescence examination revealed that the oocyte quantity declined dramatically in GSK-3β-inhibited ovaries compared with that in the control after 4 days of culture (Fig. 2a). The statistical analysis revealed that the oocyte number decreased in a time-dependent manner after GSK-3β inhibition and that GSK-3β inhibition resulted in the ovaries containing approximately 50% fewer oocytes (4370.83 ± 790.14) than the control ovaries (7580.00 ± 964.40) (P < 0.001) after 4 days of treatment (Fig. 2b). To confirm the role of GSK-3β in fetal oocyte conservation in the mouse ovary, another GSK-3β-specific inhibitor, CHIR99021 (5 μM), was used. The results of the immunofluorescence (Additional file 3: Figure S3A) and statistical analysis (Additional file 3: Figure S3B) verified consistent oocyte loss (4183.75 ± 667.85 for CHIR99021 versus 7805.00 ± 961.99 for the control per ovary; P < 0.001) following 4 days of GSK-3β inhibition. Moreover, after 14.5 dpc ovaries were cultured with BIO for 3 days (equaling 17.5 dpc), we found apparently severe oocyte apoptosis according to the immunofluorescence co-staining of DDX4 and active Caspase-3, which has been proven to be involved in programmed oocyte apoptosis in the fetal ovary [32] (Fig. 2c). Statistical analysis confirmed the significantly increased oocyte apoptosis after GSK-3β inhibition in the fetal ovary (31.00 ± 12.68 for BIO versus 3.50 ± 3.11 for the control per section; P < 0.01) (Fig. 2d). In addition, western blotting results showed that the protein level of Caspase-3 in the fetal ovaries increased significantly following GSK-3β inhibition (Additional file 3: Figure S3C). These results implied that GSK-3β was indispensable for fetal oocyte survival during meiotic prophase I, as inhibition of GSK-3β in the fetal ovaries led to massive oocyte apoptosis.
Finally, to determine the function of GSK-3β during germline cyst breakdown and primordial follicle formation, 17.5 dpc ovaries were cultured with BIO for 4 days (equaling 2 dpp). Immunofluorescence results showed that primordial follicle assembly (arrowhead) was intact following GSK-3β inhibition (Additional file 3: Figure S3D). The numbers of total oocytes and the established primordial follicle showed insignificant differences between the control and GSK-3β inhibition groups (Additional file 3: Figure S3E). In addition, after 14.5 dpc ovaries were cultured with BIO for 7 days (equaling 3 dpp), the treated ovaries contained significantly fewer primordial follicle compared to control (Additional file 3: Figure S3F-G), which implied that inhibition of GSK-3β not only influenced fetal oocyte survival but also impaired further folliculogenesis.
Inhibition of GSK-3β impeded meiotic progression and resulted in meiotic defects
Based on previous results that GSK-3β was essential for maintaining fetal oocyte survival during meiotic prophase I, we next assessed meiotic progression and meiotic events following GSK-3β inhibition. Histological immunofluorescence with antibodies against Y box protein 2 (MSY2), which is present exclusively from the diplotene stage and afterward in oocytes [33, 34], demonstrated that when 14.5 dpc ovaries were cultured for 4 days (equaling 18.5 dpc) with BIO, MSY2-positive oocytes were obviously reduced (dashed line; Fig. 3a). The statistic analysis verified that the vast majority (95.72% ± 4.02% per section) of the oocytes were MSY2+ in the control group, whereas significantly less (68.84% ± 4.9% per section; P < 0.001) oocytes from the BIO-treated group were MSY2+ (Fig. 3b), which implied that fetal oocytes were impeded in reaching the diplotene stage following GSK-3β inhibition.
Since partial fetal oocytes in the GSK-3β-inhibited ovaries failed to reach the diplotene stage, when programmed DSB repair was completed and synapsis was achieved, we next assessed whether GSK-3β inhibition impacted synaptic events in oocytes. Phosphorylated histone H2AX at Ser139 (referred to as γ-H2AX), which marked DSBs on meiotic chromatin, appeared in the oocytes from the leptotene stage and disappeared markedly in the pachytene stage (Additional file 4: Figure S4A) [35]. However, the chromosome spreads showed that a substantial number of fetal oocytes exhibited distinct γ-H2AX signals in the late pachytene stage following GSK-3β inhibition, indicating unrepaired DSBs sustained on the chromosomes; in contrast, normal fetal oocytes that reached the late pachytene stage were devoid of γ-H2AX signals on the chromosomes (Fig. 3c). Statistical analysis revealed that the percentage of oocytes with unrepaired DSBs in the pachytene stage was increased significantly following GSK-3β inhibition (18.00% ± 3.61% for BIO versus 6.00% ± 2.65% for the control per slide; P < 0.01) (Fig. 3d), which might be the reason for the failure to progress to the diplotene stage after GSK-3β inhibition. Moreover, immunofluorescence co-staining for RAD51 recombinase (a RecA homolog; a key factor in homologous recombination repair) and SYCP3 demonstrated that partial fetal oocytes showed ectopic RAD51 foci on the pachytene chromosome following GSK-3β inhibition, which implied incomplete DSB repair in the oocytes in GSK-3β-inhibited ovaries (Fig. 3e). Similarly, statistical analysis demonstrated that the percentage of abnormal RAD51-persistent oocytes in the pachytene stage in GSK-3β-inhibited ovaries (15.50% ± 4.65% per slide) increased significantly than that in the control ovaries (5.80% ± 3.83% per slide; P < 0.05) (Fig 3f). In summary, GSK-3β ensured the normal process of meiotic prophase I in fetal oocytes, whereas inhibition of GSK-3β resulted in abnormal meiotic DSB repair and meiotic progression errors.
Premature TAp63 upregulation was evident during meiotic prophase I following GSK-3β inhibition
Since there was a deficiency in DSB repair, cell cycle arrest, and increased apoptosis in fetal oocytes following GSK-3β inhibition, DNA damage checkpoint signaling was presumed to be impaired in these fetal oocytes.
To investigate the validity of the DNA damage checkpoint signaling during meiotic prophase I, we first examined the expression pattern of key components of the signaling in fetal and neonatal ovaries in vivo. Histological sections and immunofluorescence staining showed that γ-H2AX signaling, which marked unprocessed DSBs on chromosomes, emerged intensively in the oocyte nucleus from 15.5 to 17.5 dpc and disappeared afterward (Fig. 4a). This expression pattern was correlated with meiotic prophase progression, as most DSBs were induced in the leptotene stage and accomplished recombinational repair in the late pachytene stage [36]. Accordingly, p-ATM (phosphorylated ataxia telangiectasia mutated kinase) showed notable expression from 15.5 to 17.5 dpc and was primarily located within the oocyte nucleus (Additional file 4: Figure S4B). Consistently, p-CHK2 (phosphorylated checkpoint kinase 2) showed similar expression peaks to γ-H2AX in the fetal ovary (Additional file 4: Figure S4C). Intriguingly, as the major downstream effecter that is required for culling the oocytes bearing unrepaired DSBs, TAp63 was completely absent in fetal oocytes; instead, it showed a peak in the perinatal oocyte nucleus. TAp63 displayed strong expression within the oocyte nucleus until 1 dpp, when most oocytes had reached the diplotene stage (Fig. 4b). Together, these results showed that fetal oocytes were devoid of TAp63 expression until DSB repair completion around birth in vivo.
However, when 14.5 dpc ovaries were cultured for 2 days (equaling 16.5 dpc) with BIO, both the mRNA and protein level of TAp63 increased significantly, which suggested an unexpected enhancement of TAp63 expression during meiotic prophase I (Fig. 4c, d). Correspondingly, TAp63 was found to be prematurely expressed within the nucleus of partial oocytes in the fetal ovary following GSK-3β inhibition (Fig. 4e). These findings indicated that inhibition of GSK-3β resulted in the disrupted expression pattern of TAp63 in fetal oocytes in mouse ovaries.
Furthermore, we detected the expression of downstream apoptotic inducers, p21 (cyclin-dependent kinase inhibitor), Bad, Noxa, and Puma, which are BH3-only proapoptotic Bcl-2 family members and are essential mediators of p63-dependent apoptosis pathways [37]. Qualitative reverse transcription polymerase chain reaction (qRT-PCR) results showed that the relative mRNA expression levels of p21, Bad, and Noxa were significantly increased in GSK-3β-inhibited ovaries (Fig. 4f), which indicates p63-dependent apoptotic activation. Puma was rarely detectable in both the control and treatment ovaries (data not shown). In summary, inhibition of GSK-3β resulted in premature TAp63 expression and triggered transcriptions of proapoptotic genes, which might induce fetal oocyte attrition.
GSK-3β regulated cytoplasmic-nuclear translocation of β-catenin in fetal ovaries
Since GSK-3β is pivotal for the survival of fetal oocytes, which in turn influences further folliculogenesis, it is necessary to explore the mediators downstream of GSK-3β in oocyte fate determination. According to previous reports, GSK-3β negatively regulates the canonical WNT signaling pathway via modulating the stabilization of β-catenin, an active co-transcriptional factor in the cell nucleus. Thus, the expression of β-catenin following GSK-3β inhibition was studied in fetal ovaries.
First, western blotting results revealed that following GSK-3β inhibition, the phospho-β-catenin (Ser37/41/Thr49) expression level decreased significantly in the fetal ovary, which implied attenuated β-catenin phosphorylation mediated by GSK-3β activity (Fig. 5a). Histological sections and immunofluorescence assays provided additional evidence that inhibition of GSK-3β significantly promoted both the cytoplasmic and nuclear staining of β-catenin in oocytes (arrowhead) compared with that in the control (Fig. 5b). Furthermore, to clarify whether the cytoplasmic accumulation of β-catenin resulted in its nuclear translocation and transcriptional activation, several canonical target genes of β-catenin were examined by qRT-PCR (Fig. 5c). The results showed that the genes were significantly upregulated following GSK-3β inhibition, which confirmed the nuclear importation of β-catenin as a co-transcriptional factor to initiate target gene transcription. In summary, inhibition of GSK-3β resulted in aberrant cytoplasmic accumulation and subsequent nuclear translocation of β-catenin in fetal oocytes.
Aberrant nuclear translocation of β-catenin induced fetal oocyte attrition
Since GSK-3β inhibition in fetal ovaries led to dramatic fetal oocyte loss and aberrant nuclear translocation of β-catenin, accumulation and nuclear translocation of β-catenin may be responsible for the fetal oocyte attrition following GSK-3β inhibition. To evaluate this assumption, ICG-001, a specific antagonist of β-catenin-mediated transcription [38], was used to examine the presumptive role of β-catenin-mediated transcriptional activation following GSK-3β inhibition.
As shown in Fig. 6a, GSK-3β inhibition significantly increased the mRNA expression levels of canonical β-catenin target genes; however, co-treatment with ICG-001 (0.5 μM) efficiently reduced the expression levels of these genes accordingly, which is indicative of the antagonizing effect of ICG-001 on β-catenin transcription activity. Importantly, the upregulated TAp63 level following BIO treatment was effectively attenuated by BIO plus ICG-001 treatment. As shown in Fig. 6b, c, when 14.5 dpc ovaries were cultured for 2 days (equaling 16.5 dpc), both the mRNA and protein levels of TAp63 increased significantly after GSK-3β inhibition but were reduced after the simultaneous block of β-catenin transcriptional activity. Meanwhile, immunofluorescence assays proved that fetal oocytes displayed aberrant nuclear TAp63 expression after GSK-3β inhibition but were devoid of nuclear TAp63 expression significantly following BIO plus ICG-001 co-treatment (Fig. 6d). These results demonstrated that by modulating the nuclear translocation of β-catenin, GSK-3β was responsible for the fine-tuning of TAp63 expression in the fetal ovary.
Next, the effect of aberrant β-catenin transcriptional activation on fetal oocyte survival was examined. Interestingly, the DSB repair deficiency in fetal oocytes following GSK-3β inhibition could be efficiently reduced when the transcription activity of β-catenin was blocked simultaneously. The percentage of the pachytene-stage oocytes with γ-H2AX signals, which indicated abnormal meiotic DSB repair, increased significantly following GSK-3β inhibition (15.04% ± 2.35% BIO versus 3.76% ± 0.93% for the control per slide; P < 0.05), but reduced significantly in the ovaries treated with BIO plus ICG-001 (7.86% ± 3.62% for BIO plus ICG-001 versus 15.04% ± 2.35% for BIO per slide; P < 0.05) (Fig. 6d).
Moreover, massive oocyte loss induced by GSK-3β inhibition was partially rescued in BIO plus ICG-001-treated ovaries (Fig. 6f). Statistical analysis confirmed a sharp decrease in oocyte quantity following GSK-3β inhibition (4468.33 ± 781.62 for BIO versus 9691.67 ± 2116.50 for the control per ovary; P < 0.05), whereas oocyte quantity was significantly rescued following co-treatment with BIO plus ICG-001 (6181.25 ± 1221.00 for BIO plus ICG-001 versus 4468.33 ± 781.62 for BIO per ovary; P < 0.05) (Fig. 6g). Accordingly, immunofluorescence co-staining of DDX4 and active Caspase-3 demonstrated that fetal oocyte apoptosis was significantly alleviated following blockage of β-catenin transcription activity (Fig. 6h). The statistical analysis revealed that the number of active Caspase-3-positive oocytes significantly increased from 4.33 ± 4.51 to 33.00 ± 9.54 per section (P < 0.05) following GSK-3β inhibition but decreased to 6.00 ± 4.00 per section (P < 0.05) when BIO plus ICG-001 treatment was applied, which was insignificantly different from the control group (P > 0.05) (Fig. 6i). Collectively, these results implied that following GSK-3β inhibition, nuclear translocation and transcriptional activity of β-catenin were detrimental to oocytes in the fetal ovary due to the transcriptional activation of TAp63 expression.
β-catenin regulated TAp63 transcription in mouse ovaries in vivo
The regulatory relationship between β-catenin and TAp63 was then assessed in the mouse ovary in vivo. Previous studies reported that β-catenin acts as a protein with dual functions: as an intracellular adhesion on the cytomembrane and as a co-transcriptional factor in the cell nucleus [39]. According to the immunofluorescence results (Fig. 7a), β-catenin was primarily expressed on the cytomembrane of fetal oocytes before 15.5 dpc; from 17.5 dpc onward, β-catenin displayed obvious accumulation in the cytoplasm and nucleus of oocytes (arrowhead), which was correlated with the contemporaneously attenuated phosphorylation activity of GSK-3β according to our results. In addition, we detected the expression of active β-catenin, which is unphosphorylated on Ser37 or Thr41 but functionally active, in fetal and neonatal ovaries. Before 15.5 dpc, few fetal oocytes showed active β-catenin expression; however, the oocyte nucleus began to display intensive active β-catenin staining from 17.5 dpc onward (Fig. 7b). Collectively, β-catenin showed increasing nuclear expression along with meiotic prophase I progression, which was consistent with the upregulated expression of TAp63 in fetal ovaries (Additional file 4: Figure S4D). These results implied a transcriptional regulatory role of β-catenin on TAp63.
Furthermore, chromatin immunoprecipitation (ChIP) assays were applied to verify whether β-catenin induced TAp63 transcription activation directly in fetal ovaries. Mouse ovaries of 17.5 dpc were collected, and the DNA fragment that immunoprecipitated with the anti-β-catenin antibody was examined with primers designed within 2000 bp of the TAp63 promoter sequence region. The qRT-PCR results indicated that β-catenin enriched the − 963 to − 793 region of the TAp63 promoter sequence (Fig. 7c). In summary, these findings demonstrated that β-catenin translocated to the nucleus of fetal oocytes and activated TAp63 transcription by directly binding to the promoter region of TAp63 in vivo.
Conditional deletion of Gsk-3β in germline cells caused oocyte loss in mice
To identify the physiological role of GSK-3β in developing fetal ovaries in vivo, a germ cell-specific deletion of Gsk-3β mouse model was produced. In brief, by crossing Gsk-3βflox/flox mice with Ddx4-Cre mice, Gsk-3βflox/−;Ddx4-Cre mice (which exhibited Cre-mediated recombination confined to the germline cells beginning at approximately 15.5 dpc [40]) were produced and are referred to as Gsk-3β cKO mice. The wild-type littermates were generally used as the control. Immunofluorescence results verified a significant decrease in GSK-3β expression in the cytoplasm of oocytes in cKO mice compared with that in the control. The conditional deletion strategy did not affect GSK-3β expression in somatic cells, which proved the efficiency and accuracy of Cre-mediated deletion (Fig. 8a).
Next, the effect of germ cell-specific GSK-3β deletion on oogenesis and early folliculogenesis in mice was assessed. Histological sections and immunofluorescence revealed insignificant differences between Gsk-3β cKO and the control ovary at 15.5 dpc, while on 1 dpp, when germ cell cyst breakdown and primordial follicle formation was progressing, fewer oocytes and primordial follicles were available in the Gsk-3β cKO ovaries compared with those in the control (Fig. 8b). On 7 dpp, the Gsk-3β cKO ovary showed a significantly lower follicle reserve than that in the control, and the statistical analysis proved that the number of primordial follicles in the cKO mice was significantly less than that in the control (5338.33 ± 1727.99 in the cKO ovary versus 11,662.50 ± 2459.24 in the control per ovary; P < 0.01). However, the numbers of growing follicles in the ovaries showed insignificant differences between the cKO and control ovaries (1130.00 ± 568.11 for the cKO ovary versus 1315.00 ± 528.50 for the control per ovary; P > 0.05) (Fig. 8c). To clarify whether fetal oocyte loss following GSK-3β deletion was due to increased apoptosis, the TUNEL (TdT-mediated dUTP Nick-End Labeling) assay was performed on 1 dpp ovaries, which showed that the TUNEL-positive cells in Gsk-3β cKO ovaries were obviously more than that in the control, indicating excessive cell apoptosis in the ovaries of Gsk-3β cKO mice (Fig. 8d). These observations demonstrate that GSK-3β is essential for fetal oocyte survival during meiotic prophase I in the fetal mouse ovary.
Moreover, the meiotic DSB repair following germ cell-specific Gsk-3β deletion was assessed. When 1 dpp ovaries were stained with γ-H2AX, more γ-H2AX-positive oocytes were observed in Gsk-3β cKO ovaries compared with those in the control, which further confirmed the sustained DSBs in the oocytes after Gsk-3β deletion (Fig. 8e). Similarly, more RAD51 signal-positive oocytes were found in the Gsk-3β cKO ovaries than those in the control, which demonstrated incomplete DSB repair after Gsk-3β deletion (Fig. 8f). Additionally, we examined the expression of β-catenin in Gsk-3β-deleted ovaries. According to the histological section and immunofluorescence results, cytoplasmic accumulation and nuclear translocation of β-catenin in the oocytes of cKO ovaries were detected (arrowhead) (Fig. 8g). Statistical analysis demonstrated that the percentage of oocytes showing β-catenin accumulation per section increased significantly in cKO ovaries (P < 0.001) (Fig. 8h). In summary, the results from the Gsk-3β cKO mice were in accordance with the in vitro GSK-3β inhibition findings, which indicated that GSK-3β is a prerequisite for fetal oocyte survival in mice.