ABSTRACT
A successful mitosis-to-meiosis transition in germ cells is essential for fertility in sexually reproducing organisms. In mice and humans, it has been established that expression of STRA8 is crucial for meiotic onset in both sexes. Here, we show that BMP signalling is also essential, not for STRA8 induction but for correct meiotic progression in female mouse fetal germ cells. Largely in agreement with evidence from primordial germ cell-like cells (PGCLCs) in vitro, germ cell-specific deletion of BMP receptor 1A (BMPR1A; ALK3) caused aberrant retention of pluripotency marker OCT4 and meiotic progression was compromised; however, the timely onset of Stra8 and STRA8 expression was unaffected. Comparing the transcriptomes of Bmpr1a-cKO and Stra8-null models, we reveal interplay between the effects of BMP signalling and STRA8 function. Our results verify a role for BMP signalling in instructing germ cell meiosis in female mice in vivo, and shed light on the regulatory mechanisms underlying fetal germ cell development.
INTRODUCTION
In most animals, the sole means by which genetic material is transferred from one generation to the next is via haploid gametes, and their production is crucially dependent on the process of meiosis. Although the chromosomal events of meiosis are relatively conserved from yeasts to humans, the molecular regulators of the decision to switch from mitosis to meiosis are less conserved; perhaps for this reason, meiotic initiation has been under-studied (Kimble, 2011).
Primordial germ cells (PGCs) are specified in the mouse epiblast at embryonic (E) day ∼7.25 through the coordinated signalling activities of different factors, including BMP2, BMP4, BMP8b, and WNT3a (Lawson et al., 1999; Ohinata et al., 2009; Saitou et al., 2002; Ying et al., 2000, 2001). Nascent PGCs maintain or upregulate pluripotency marker expression [e.g. Oct4 (Pou5f1) and Sox2], repress somatic genes and begin to express germ cell specific genes. PGCs migrate to and colonise the embryonic gonads from ∼E10.5, at which time genome-wide epigenetic reprogramming occurs (Hajkova, 2010; Hajkova et al., 2002). Once in the gonads, PGC sexual fate is determined not by their genetic sex but by the signals received from the gonadal soma (McLaren, 1995, 2003). Gonadal somatic cues direct ovarian germ cells to enter meiosis during fetal life, but testicular germ cells avoid meiotic entry until after birth. Despite this knowledge, we still lack a comprehensive understanding of soma-to-germ cell signalling during fetal gonad development.
Evidence exists that the signalling molecule retinoic acid (RA) directly induces mouse fetal ovarian germ cells to initiate Stra8 (stimulated by retinoic acid gene 8) expression (Bowles et al., 2006, 2016; Feng et al., 2021; Koubova et al., 2006; Soh et al., 2015). Stra8 encodes a transcription factor that is crucial for the initiation of meiotic S (synthesis) phase and for meiotic progression in both sexes. In the Stra8-null mouse, crucial features of meiosis, including chromosome condensation, cohesion, synapsis, and recombination, are absent (Baltus et al., 2006). Because RA levels are highest at the anterior end of the developing ovary, Stra8 transcript and STRA8 protein are induced in an anterior-to-posterior ‘wave’ (Bowles et al., 2006; Menke et al., 2003; Feng et al. 2021). In the fetal testis, RA is cleared by the RA-degrading enzyme CYP26B1 (Bowles et al., 2006; Dokshin et al., 2013; MacLean et al., 2007; Soh et al., 2015) and, therefore, testicular germ cells are spared from meiotic entry during fetal life. Besides Stra8, other studies have shown that RA directly induces other key meiotic genes, including Rec8 (Koubova et al., 2014) and Meiosin (Ishiguro et al., 2020).
Recent findings suggest that, at least in vitro, another signalling molecule, bone morphogenetic protein (BMP), is necessary for XX germ cell development (Miyauchi et al., 2017). BMPs are TGFβ superfamily signalling ligands; they transduce signal via binding to type I and type II transmembrane serine/threonine kinase receptors (Mueller and Nickel, 2012). BMPs have numerous roles in germ cell specification (BMP2, BMP4 and BMP8b) (Lawson et al., 1999; Ying et al., 2000, 2001), migration (BMP4) (Dudley et al., 2007), and postnatal oocyte development (GDF9 and BMP15) (Dong et al., 1996; Yan et al., 2001). In a landmark study, primordial germ cell-like cells (PGCLCs), generated in vitro from mouse embryonic stem cells and cultured without somatic cells (Ohta et al., 2017), were studied to delineate the signalling activity necessary to drive mitosis-to-meiosis transition, meiotic progression, and oogenesis (Miyauchi et al., 2017). It was concluded that RA and BMP work synergistically to promote oogenic fate (Miyauchi et al., 2017) and the transcription factor ZGLP1 was subsequently identified as a key effector of BMP signalling in PGCLCs (Nagaoka et al., 2020). Despite this progress, it remains unclear whether BMP signalling plays a crucial role in vivo and, if so, at which step(s) of ovarian germ cell development it is required.
Investigating whether BMPs act directly on gonadal germ cells to instruct female-specific development is complicated because: (1) BMP signalling is required for earlier PGC specification and proliferation (Lawson et al., 1999; Ross et al., 2007; Ying et al., 2000, 2001); (2) several BMP ligands (BMP2, BMP4 and BMP5) are expected to be present in the E11.5 fetal ovary (Jameson et al., 2012; Ross et al., 2007; Yao et al., 2004); (3) BMP2 likely also supports correct fetal ovarian somatic cell development (Kashimada et al., 2011; Yao et al., 2004); and (4) BMPs signal through receptors shared with other TGFβ family members. Given these considerations, we investigated the role of BMP in control of ovarian germ cell fate using two approaches: broad chemical inhibition of BMP signalling in ex vivo culture and germ cell-specific deletion of the gene encoding the BMP receptor 1A (BMPR1A) in vivo.
RESULTS
Antagonising canonical BMP signalling impacts expression of Sycp3 and Oct4 in ex vivo UGR cultures
To investigate the potential role of BMP signalling in meiotic onset in fetal ovarian germ cells, we inhibited BMP signalling in urogenital ridges (UGRs) cultured ex vivo. LDN193189, a small molecule inhibitor of BMPR1A (ALK3) and ACVR1 (ALK2) (Cuny et al., 2008), specifically inhibits canonical BMP signalling by blocking SMAD1/5/8 phosphorylation. UGRs were dissected from E11.5 C57BL/6 embryos and individually cultured ex vivo for 24, 48 or 72 h in hanging drops with or without LDN193189 (500 nM) (Fig. S1A). Antagonism of BMPR1A and ACVR1 led to a significant upregulation of Bmp2 expression in the 24 h group (Fig. S1B), possibly reflecting a feedback mechanism by the gonadal somatic cells to compensate for reduced BMP signalling.
Treatment with LDN193189 did not adversely affect the onset of Stra8 expression; rather, Stra8 was elevated in 48- and 72-h treatment groups (Fig. S1C). This finding could suggest that disruption to BMP signalling enhances Stra8 expression. It seems more likely, however, that this result reflects delayed germ cell progression through meiosis, because STRA8 downregulates its own expression as meiosis proceeds (Soh et al., 2015). Consistent with the latter possibility, Sycp3 expression, a marker of meiotic progression, was significantly lower in 72 h treated samples compared to controls (Fig. S1C). Meiotic entry coincides with downregulation of pluripotency marker Oct4 (Bullejos and Koopman, 2004). We found that blocking of BMP signalling in UGRs did not affect Oct4 expression in 24 or 48 h groups, though expression was abnormally maintained at 72 h (Fig. S1C). These ex vivo culture studies suggest that BMP signalling is likely required for efficient meiotic progression and not necessarily for meiotic onset as perturbations to Sycp3 and Oct4 only occurred after 72 h.
Germ cell-specific knockout of Bmpr1a is effective and specific in mouse fetal ovaries
Next, we used a genetic deletion approach to investigate the in vivo role for BMP signalling and to test whether its effects are direct or indirect on the germ cells. BMP2 is a likely candidate for directing fetal ovarian germ cell development due to its female-specific gene expression (Yao et al., 2004). The Bmp2-null mutant is embryonic lethal at E10.5 (Zhang and Bradley, 1996), and BMP2 likely plays important roles in ovarian soma development (Bayne et al., 2016; Kashimada et al., 2011; Yao et al., 2004). Furthermore, Bmp4 and Bmp5 expression may also contribute to signalling in the fetal ovary (Jameson et al., 2012). To circumvent these problems, we deleted the type I BMP receptor BMPR1A (used by BMP2, BMP4 and BMP5) specifically in gonadal germ cells (Bmpr1aΔPGC) using Bmpr1atm2.1Bhr (Bmpr1afl/fl) (Mishina et al., 2002) and Oct4-CreERT2 (Greder et al., 2012) mouse lines.
Bmpr1afl/fl females were time-mated with Bmpr1afl/fl;Oct4-CreCre/WT males, and injected with 4-hydroxytamoxifen at 9.5, 10.5, and 11.5 days post coitum (dpc) to induce CRE-dependent deletion of Bmpr1a in embryonic germ cells. Bmpr1aΔPGC and Bmpr1afl/fl control embryos were harvested from E12.5 to E16.5 (Fig. 1A). Successful deletion was confirmed by the loss of BMPR1A immunostaining in MVH(DDX4)+ germ cells, but not the surrounding gonadal somatic cells, in E13.5 ovaries (Fig. 1B). Additionally, we confirmed loss of Zglp1 expression, a BMP target in germ cells, in MACS-enriched germ cell populations (Fig. S2).
Germ cell-specific knockout of Bmpr1a substantially reduced expression of downstream effector Zglp1 but did not affect germ cell survival. (A) Bmpr1afl/fl females were time-mated with Bmpr1afl/fl;Oct4-CreCre/WT males; CRE-dependent recombination of Bmpr1a in the embryos was induced by intraperitoneal administration of 4-hyroxytamoxifen (4-OHT) and progesterone (P) to the dams at 9.5, 10.5, and 11.5 dpc. Embryos were harvested at E12.5-E16.5 for gene and protein expression analysis. (B) Immunostaining for BMPR1A and MVH in E13.5 embryos confirmed successful depletion of BMPR1A in germ cells. (C,D) qRT-PCR showed significant downregulation of (C) Bmpr1a and (D) Zglp1 in Bmpr1aΔPGC ovaries compared with the control across all timepoints analysed. (E) Mvh was significantly higher in the mutants at E13.5 only. (F) Bmpr1aΔPGC germ cells are less round compared to control germ cells. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001 (n≥4, unpaired t-test; data are mean±s.e.m. for qRT-PCR and mean±s.d. for cell roundness). Cytoplasmic MVH (magenta) marks germ cells. Scale bars: 20 μm.
Germ cell-specific knockout of Bmpr1a substantially reduced expression of downstream effector Zglp1 but did not affect germ cell survival. (A) Bmpr1afl/fl females were time-mated with Bmpr1afl/fl;Oct4-CreCre/WT males; CRE-dependent recombination of Bmpr1a in the embryos was induced by intraperitoneal administration of 4-hyroxytamoxifen (4-OHT) and progesterone (P) to the dams at 9.5, 10.5, and 11.5 dpc. Embryos were harvested at E12.5-E16.5 for gene and protein expression analysis. (B) Immunostaining for BMPR1A and MVH in E13.5 embryos confirmed successful depletion of BMPR1A in germ cells. (C,D) qRT-PCR showed significant downregulation of (C) Bmpr1a and (D) Zglp1 in Bmpr1aΔPGC ovaries compared with the control across all timepoints analysed. (E) Mvh was significantly higher in the mutants at E13.5 only. (F) Bmpr1aΔPGC germ cells are less round compared to control germ cells. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001 (n≥4, unpaired t-test; data are mean±s.e.m. for qRT-PCR and mean±s.d. for cell roundness). Cytoplasmic MVH (magenta) marks germ cells. Scale bars: 20 μm.
In Bmpr1aΔPGC fetal ovaries, expression of Bmpr1a was significantly reduced compared to the control at all timepoints analysed (Fig. 1C). Similarly, Zglp1 expression was significantly reduced compared to the control from E12.5-E15.5 (Fig. 1D), confirming Zglp1 as a downstream target of BMPR1A-mediated BMP signalling in mouse ovarian germ cells in vivo. Contrary to observations in PGCLCs, and after ubiquitous inhibition of BMP signalling in pregnant dams (Miyauchi et al., 2017), Mvh expression was not diminished in our Bmpr1a-cKO model. Mvh transcription did not differ between mutant and control samples, except at E13.5, when significantly higher Mvh expression was found in the mutants (Fig. 1E). As Mvh expression (relative to Tbp expression) can be considered a proxy for germ cell number, this might indicate that mutant germ cells continued dividing mitotically when compared with control germ cells. We noted that germ cells deficient for BMPR1A were more irregular in shape than control germ cells (Fig. 1F); the implication of this is unknown as the survival of the cells, at least to E16.5, appeared normal.
Germ cell-specific knockout of Bmpr1a did not affect Stra8 initiation, but subcellular localisation of STRA8 protein was affected from E15.5
We first investigated the initial steps of germ cell differentiation. Analysis by qRT-PCR showed that loss of BMPR1A did not affect the initiation of Stra8 expression in E12.5-E14.5 fetal ovaries (Fig. 2A). To test whether the anterior-to-posterior ‘wave’ of Stra8 upregulation is affected (Menke et al., 2003), E12.5 fetal ovaries were bisected to compare Stra8 expression between the anterior and posterior halves. Comparable expression between the mutants and controls indicates that loss of Bmpr1a did not compromise the anterior-to-posterior ‘wave’ of Stra8 onset (Fig. 2B), suggesting that this is primarily determined by RA availability, as postulated previously (Bowles et al., 2010, 2006). At E15.5, significantly higher Stra8 expression was observed (∼5.8-fold difference) in Bmpr1aΔPGC ovaries (Fig. 2A), corroborating our ex vivo gonad culture observations (Fig. S1C), and possibly reflecting a lack of meiotic progression.
Loss of BMP signalling did not affect temporal onset or anterior-to-posterior initiation of Stra8 expression but resulted in increased cytoplasmic STRA8. (A) qRT-PCR showed no difference in the onset of Stra8 expression between control and Bmpr1aΔPGC fetal ovaries at E12.5-E14.5, but expression level at E15.5 was substantially higher (∼5-fold) in the mutants. (B) Quantification of Stra8 expression from anterior and posterior halves of E12.5 fetal ovaries showed the highest Stra8 expression in the anterior half in both control and Bmpr1aΔPGC mutant gonads. (C-E) The number of STRA8+ germ cells was comparable between E13.5 and E15.5 (green, STRA8; magenta, germ cell marker MVH). (E) At E15.5, most MVH+ germ cells expressed STRA8. Consistent across E15.5 and E16.5 control ovaries, punctate STRA8 immunostaining was found predominantly localised to the nucleus, or to both the nucleus and cytoplasm (arrowheads) of germ cells. Very few cells exhibited stronger cytoplasmic than nuclear STRA8 immunosignal (double-headed arrow). In the mutants, most Bmpr1aΔPGC germ cells expressed both nuclear and cytoplasmic STRA8, with most cells exhibiting stronger cytoplasmic STRA8 immunosignal. MVH (magenta) and TRA98 (red) are cytoplasmic and nuclear markers for germ cells, respectively. (F) E15.5 Bmpr1aΔPGC ovaries had significantly more germ cells with cytoplasmic-accumulated STRA8 than the control. (G) Stra8 transcript splice junctions visualised by sashimi plot showed that both control and Bmpr1aΔPGC ovaries expressed Stra8 isoforms with or without the nuclear localisation sequence (NLS, represented in red within exon 3), or nuclear export sequence (NES, represented in blue across exons 6 and 7) (ENSMUST00000185102.1, ENSMUST00000114999.7, or ENSMUST00000114997.2) in similar proportions. *P<0.05, ***P<0.0001 (n≥4, unpaired t-test; data are mean±s.e.m.). Scale bars: 50 μm.
Loss of BMP signalling did not affect temporal onset or anterior-to-posterior initiation of Stra8 expression but resulted in increased cytoplasmic STRA8. (A) qRT-PCR showed no difference in the onset of Stra8 expression between control and Bmpr1aΔPGC fetal ovaries at E12.5-E14.5, but expression level at E15.5 was substantially higher (∼5-fold) in the mutants. (B) Quantification of Stra8 expression from anterior and posterior halves of E12.5 fetal ovaries showed the highest Stra8 expression in the anterior half in both control and Bmpr1aΔPGC mutant gonads. (C-E) The number of STRA8+ germ cells was comparable between E13.5 and E15.5 (green, STRA8; magenta, germ cell marker MVH). (E) At E15.5, most MVH+ germ cells expressed STRA8. Consistent across E15.5 and E16.5 control ovaries, punctate STRA8 immunostaining was found predominantly localised to the nucleus, or to both the nucleus and cytoplasm (arrowheads) of germ cells. Very few cells exhibited stronger cytoplasmic than nuclear STRA8 immunosignal (double-headed arrow). In the mutants, most Bmpr1aΔPGC germ cells expressed both nuclear and cytoplasmic STRA8, with most cells exhibiting stronger cytoplasmic STRA8 immunosignal. MVH (magenta) and TRA98 (red) are cytoplasmic and nuclear markers for germ cells, respectively. (F) E15.5 Bmpr1aΔPGC ovaries had significantly more germ cells with cytoplasmic-accumulated STRA8 than the control. (G) Stra8 transcript splice junctions visualised by sashimi plot showed that both control and Bmpr1aΔPGC ovaries expressed Stra8 isoforms with or without the nuclear localisation sequence (NLS, represented in red within exon 3), or nuclear export sequence (NES, represented in blue across exons 6 and 7) (ENSMUST00000185102.1, ENSMUST00000114999.7, or ENSMUST00000114997.2) in similar proportions. *P<0.05, ***P<0.0001 (n≥4, unpaired t-test; data are mean±s.e.m.). Scale bars: 50 μm.
Consistent with the evidence that Stra8 mRNA is initiated correctly, immunofluorescence staining on sectioned tissue showed the presence of STRA8+ germ cells at E13.5 and E14.5, with no obvious difference between mutants and controls (Fig. 2C,D, Fig. S3). At E15.5 and E16.5, however, close examination revealed an increase in cytoplasmic STRA8 in Bmpr1aΔPGC germ cells compared to controls (Fig. 2E). STRA8 is known to shuttle between the nucleus and cytoplasm, which is important for its transcriptional activity (Tedesco et al., 2009). In control germ cells, STRA8 was found predominantly in a punctate pattern in the nucleus, with some cells being STRA8+ in both the nucleus and cytoplasm. In Bmpr1aΔPGC germ cells, nuclear signal for STRA8 was observed but this was frequently accompanied by a stronger cytoplasmic signal. E15.5 mutant ovaries possessed 10 times more germ cells with a stronger STRA8 signal in the cytoplasm than nucleus (cyt>nuc) compared to the control (mutant: 17.7%, 95/937 STRA8+ germ cells; control: 1.8%, 17/944 STRA8+ germ cells) (Fig. 2F). RNA-Seq analysis (see later sections) showed that Stra8 isoforms, with or without the N-terminal nuclear localisation signal (NLS, amino acids 28-33, depicted in red) and nuclear export sequence (amino acids 174-348, depicted in blue), were expressed in comparable ratios between E14.5 control and Bmpr1aΔPGC ovaries (Fig. 2G), suggesting the difference in subcellular localisation is not a result of differential isoform expression. These results indicate direct BMP signalling in ovarian germ cells is dispensable for initiation of Stra8 expression, but is necessary for the extinction of Stra8 expression and nuclear localisation of STRA8 at later timepoints (E15.5-E16.5).
Germ cell-specific knockout of Bmpr1a delayed meiotic progression in vivo
Despite normal onset of Stra8, expression of meiotic marker Sycp3 was significantly reduced in the Bmpr1aΔPGC mutants at E13.5 and E15.5, as was the expression of leptotene/zygotene marker Spo11 at E15.5 (Fig. 3A). At E14.5, 77.6% of germ cells in control ovaries were SYCP3+, whereas only 29.1% mutant germ cells were SYCP3+ (Fig. 3Bi-C). By E15.5, most, if not all, control germ cells were SYCP3+, but the mutant ovaries still contained fewer SYCP3-expressing germ cells, consistent with the lower Sycp3 transcription described above (Fig. 3Bi-C). By E16.5, more mutant germ cells expressed SYCP3, but the protein did not decorate the lengths of the chromosomes, as in control germ cells. Instead, we observed SYCP3 protein in a pre-leptotene-like pattern (Prieto et al., 2004) in the nucleus, as seen in controls at earlier timepoints, E14.5 and E15.5 (Fig. 3Bi,D). At E15.5, ∼80% of control but <40% of mutant germ cells were positive for the DNA double-stranded break marker γH2AX (Fig. 3Bii,E). These results suggest that germ cells in the mutant ovaries were able to initiate meiosis (express STRA8) but, in a considerable portion of them, progression into meiotic prophase, indicated by the expression of SYCP3, was delayed.
Delayed meiotic progression in Bmpr1aΔPGC germ cells. (A) Expression of meiotic progression markers Sycp3 and Spo11 was significantly reduced in the Bmpr1aΔPGC mutants at E13.5 and E15.5 (Sycp3) and E15.5 (Spo11). (Bi,C) Quantification of immunofluorescence revealed reduced expression of SYCP3 in mutants at E14.5-E16.5, compared to control germ cells. (Bi,D) At E16.5, most control germ cells expressed SYCP3, which decorated the length of the chromosomes; fewer mutant germ cells expressed SYCP3, which, when present, displayed a punctate nuclear pattern similar to that observed at earlier timepoints in the control. (Bii, E) At E15.5, fewer Bmpr1aΔPGC germ cells were positive for the DNA double-stranded breaks marker γH2AX. (Fi) Immunostaining for SYCP1 (green) and SYCP3 (magenta) in meiotic chromosome spreads showed a similar distribution pattern of SYCP3+ cells at different stages of meiotic prophase I at E15.5 but a different distribution pattern at E16.5 between control and Bmpr1aΔPGC cells. (Fii) At E15.5, both control and Bmpr1aΔPGC SYCP3+ cells were mostly in leptotene and zygotene stages of meiotic prophase. At E16.5, >90% of SYCP3+ cells in the control ovaries had reached pachytene, some had reached late pachytene, and none were found in the leptotene stage. On the contrary, very few Bmpr1aΔPGC SYCP3+ cells had reached pachytene, no late-pachytene cells were found, and 85% of the SYCP3+ cells remained in leptotene or zygotene. *P<0.05, **P<0.01 (n≥4, unpaired t-test; data are mean±s.e.m.). Cytoplasmic MVH (magenta) marks germ cells. Scale bars: 50 μm.
Delayed meiotic progression in Bmpr1aΔPGC germ cells. (A) Expression of meiotic progression markers Sycp3 and Spo11 was significantly reduced in the Bmpr1aΔPGC mutants at E13.5 and E15.5 (Sycp3) and E15.5 (Spo11). (Bi,C) Quantification of immunofluorescence revealed reduced expression of SYCP3 in mutants at E14.5-E16.5, compared to control germ cells. (Bi,D) At E16.5, most control germ cells expressed SYCP3, which decorated the length of the chromosomes; fewer mutant germ cells expressed SYCP3, which, when present, displayed a punctate nuclear pattern similar to that observed at earlier timepoints in the control. (Bii, E) At E15.5, fewer Bmpr1aΔPGC germ cells were positive for the DNA double-stranded breaks marker γH2AX. (Fi) Immunostaining for SYCP1 (green) and SYCP3 (magenta) in meiotic chromosome spreads showed a similar distribution pattern of SYCP3+ cells at different stages of meiotic prophase I at E15.5 but a different distribution pattern at E16.5 between control and Bmpr1aΔPGC cells. (Fii) At E15.5, both control and Bmpr1aΔPGC SYCP3+ cells were mostly in leptotene and zygotene stages of meiotic prophase. At E16.5, >90% of SYCP3+ cells in the control ovaries had reached pachytene, some had reached late pachytene, and none were found in the leptotene stage. On the contrary, very few Bmpr1aΔPGC SYCP3+ cells had reached pachytene, no late-pachytene cells were found, and 85% of the SYCP3+ cells remained in leptotene or zygotene. *P<0.05, **P<0.01 (n≥4, unpaired t-test; data are mean±s.e.m.). Cytoplasmic MVH (magenta) marks germ cells. Scale bars: 50 μm.
To interrogate the progression through meiotic prophase I, meiotic chromatin spreads were prepared for E15.5 and E16.5 control and mutant germ cells. Meiotic prophase sub-stages were determined by immunostaining patterns of SYCP1 and SYCP3 (Fig. 3F). At E15.5, meiotic progression was similar in control and mutant SYCP3+ germ cells, with most in leptotene and zygotene stages, and a small proportion reaching early pachytene. However, by E16.5, >90% of control SYCP3+ cells were in mid- to late-pachytene stage, while the majority of Bmpr1aΔPGC SYCP3+ cells (∼85%) remained in leptotene or zygotene stages (Fig. 3F). Thus, distribution of SYCP3+ meiotic germ cells in various stages of prophase was similar between the control and Bmpr1aΔPGC mutants at E15.5, but vastly different at E16.5 (Fig. 3F), suggesting a potential defect in meiotic progression in Bmpr1aΔPGC germ cells, possibly immediately before pachytene stage.
Bmpr1aΔPGC germ cells abnormally retained pluripotency marker OCT4 and were slow to abandon the mitotic cell cycle
Entry of fetal germ cells into meiotic prophase I is marked by the loss of pluripotency-associated Oct4/OCT4 (Bullejos and Koopman, 2004; Menke et al., 2003; Pesce et al., 1998; Western et al., 2010). Interestingly, we found that Oct4 expression was lower in the mutant ovaries compared to the control at E12.5, though no change was seen at later timepoints (Fig. 4A). The result was clearer when we investigated OCT4 protein. OCT4 immunostaining showed a gradual loss of the protein from E13.5 to E14.5 in the control germ cells, and an almost complete absence by E15.5, as expected (Bullejos and Koopman, 2004). In the mutants, however, OCT4 immunosignal persisted through to E15.5 with little evidence of reduction (>70% remained OCT+ at E15.5; Fig. 4B,C). This suggests that post-transcriptional regulation of Oct4/OCT4 may require BMP signalling.
Bmpr1aΔPGC germ cells abnormally retained OCT4 and mitotic exit was delayed. (A) Oct4 expression in Bmpr1aΔPGC mutant ovaries was lower than that in the control at E12.5, but not significantly different at later timepoints. (B-Di) Control germ cells lost OCT4 gradually from E13.5 to E15.5, but Bmpr1aΔPGC germ cells abnormally retained OCT4 expression (B,C), and maintained expression of the mitotic G2/M marker phospho-histone H3 at E15.5 (Di). (Dii,iii) To monitor S phase, embryos were exposed to BrdU for 2 h. Fewer germ cells (MVH+) in E15.5 control ovaries were positive for BrdU immunosignal compared to Bmpr1aΔPGC germ cells. Of the cells quantified, 110/226 (48.6%, control) and 134/301 (44.5%, mutant) were located at the anterior end of the ovaries. **P<0.01 (n≥4, unpaired t-test; data are mean±s.e.m.). Cytoplasmic MVH (magenta) marks germ cells. Scale bars: 50 μm.
Bmpr1aΔPGC germ cells abnormally retained OCT4 and mitotic exit was delayed. (A) Oct4 expression in Bmpr1aΔPGC mutant ovaries was lower than that in the control at E12.5, but not significantly different at later timepoints. (B-Di) Control germ cells lost OCT4 gradually from E13.5 to E15.5, but Bmpr1aΔPGC germ cells abnormally retained OCT4 expression (B,C), and maintained expression of the mitotic G2/M marker phospho-histone H3 at E15.5 (Di). (Dii,iii) To monitor S phase, embryos were exposed to BrdU for 2 h. Fewer germ cells (MVH+) in E15.5 control ovaries were positive for BrdU immunosignal compared to Bmpr1aΔPGC germ cells. Of the cells quantified, 110/226 (48.6%, control) and 134/301 (44.5%, mutant) were located at the anterior end of the ovaries. **P<0.01 (n≥4, unpaired t-test; data are mean±s.e.m.). Cytoplasmic MVH (magenta) marks germ cells. Scale bars: 50 μm.
To investigate the exit from mitosis, we performed immunofluorescence staining for the mitotic G2/M marker phospho-histone H3 (PHH3). At E15.5, control ovaries were completely devoid of PHH3+ germ cells but some germ cells in the Bmpr1aΔPGC ovary were PHH3 positive (Fig. 4Di). Punctate mitotic prophase-like pattern and bouquet-like metaphase/telophase pattern of PHH3 (Medani et al., 2021) could be seen in some Bmpr1a-null germ cells, suggesting they were still proliferating mitotically. We also assessed cell cycle progression by injecting BrdU into pregnant females 2 h before embryo collection. At E15.5, BrdU signal was detected in 29.2% of MVH+ germ cells in the control ovaries, and 42.5% of MVH+ cells in the mutants (Fig. 4Dii,iii), indicating that some mutant germ cells fail to abandon the mitotic cell cycle on schedule.
RNA-Seq analysis revealed transcriptional impacts of BMP signalling in fetal germ cells
Following the candidate gene/protein assessment, we then sought to evaluate the regulatory role of BMP signalling more systematically. Bulk RNA-Seq was conducted to compare the transcriptomes of E14.5 mutant and control ovaries. Differentially expressed genes (DEGs) were defined as those with |Log2-fold change (Log2FC)|>0.5 (i.e. upregulated or downregulated by at least 1.4 fold) and a false discovery rate (FDR)<0.05. Of the 14,244 genes mapped, 1440 were differentially expressed (600 downregulated and 840 upregulated) in Bmpr1aΔPGC ovaries compared to the control (Fig. 5; Fig. S4; Table S1).
RNA-Seq revealed aberrant gene expression in E14.5 Bmpr1aΔPGC mutant ovaries. RNA-Seq was conducted on E14.5 Bmpr1aΔPGC mutant and control ovaries (n=4, pools of two ovary pairs each), and differentially expressed genes (DEGs) were defined as those with a |Log2FC|>0.5 and FDR<0.05 (downregulated, 600; upregulated, 840; see Table S1 for full list of DEGs). (A) Consistent with the in vivo phenotype observed, Stra8 was not differentially expressed but some other meiosis-associated genes were downregulated, while pluripotency-associated genes were upregulated in the Bmpr1a-cKO. Expression levels of selected genes are visualised in the heatmap. (B) Functional annotation clustering identified an enrichment for genes associated with mitosis and stem cell population maintenance among the upregulated clusters, and an enrichment for genes associated with meiosis and germ cell development among the downregulated clusters. Full lists of enriched biological processes GO terms and the associated genes among the top 5 enriched annotation clusters are listed in Table S2 (downregulated in Bmpr1aΔPGC) and Table S3 (upregulated in Bmpr1aΔPGC).
RNA-Seq revealed aberrant gene expression in E14.5 Bmpr1aΔPGC mutant ovaries. RNA-Seq was conducted on E14.5 Bmpr1aΔPGC mutant and control ovaries (n=4, pools of two ovary pairs each), and differentially expressed genes (DEGs) were defined as those with a |Log2FC|>0.5 and FDR<0.05 (downregulated, 600; upregulated, 840; see Table S1 for full list of DEGs). (A) Consistent with the in vivo phenotype observed, Stra8 was not differentially expressed but some other meiosis-associated genes were downregulated, while pluripotency-associated genes were upregulated in the Bmpr1a-cKO. Expression levels of selected genes are visualised in the heatmap. (B) Functional annotation clustering identified an enrichment for genes associated with mitosis and stem cell population maintenance among the upregulated clusters, and an enrichment for genes associated with meiosis and germ cell development among the downregulated clusters. Full lists of enriched biological processes GO terms and the associated genes among the top 5 enriched annotation clusters are listed in Table S2 (downregulated in Bmpr1aΔPGC) and Table S3 (upregulated in Bmpr1aΔPGC).
As expected, BMP signalling target Zglp1 (Nagaoka et al., 2020) was downregulated in the Bmpr1a-cKO (down 2.44-fold), as were canonical BMP target genes Msx1 (down 2.31-fold) and Msx2 (down 3.53-fold) (Fig. 5A). Consistent with the results reported above, functional annotation of the downregulated genes found an enrichment for those associated with ‘meiotic cell cycle’, ‘reproductive process’, ‘germ cell development’, ‘cell differentiation’, and ‘double-strand break repair involved in meiotic recombination’ (Fig. 5B; Table S2). Male germ cell-related GO terms such as ‘spermatogenesis (GO:0007283)’ and ‘spermatid development (GO:0007286)’ were among the over-represented GO terms. However, we noted that many genes associated with these ‘male’ processes (e.g. Majin, Ythdc2, Sycp3, Meioc and Zglp1) are also known meiotic factors in the female germline, suggesting this enrichment is related to the meiotic process during spermatogenesis rather than to the male program per se.
DEGs upregulated in Bmpr1aΔPGC ovaries were enriched for those associated with mitosis, cell proliferation, and stem cell population maintenance (Fig. 5B; Table S3), consistent with our observation that Bmpr1a-null germ cells exhibited a delay in early meiotic progression and the abandonment of mitosis. As observed in our candidate approach, genes that are not dependent on BMP signalling for their expression included Mvh and Stra8 (Fig. 5A). Induction of Dazl, a marker of germ cell ‘licencing’ (Gill et al., 2011) was unaffected by the loss of Bmpr1a, but, rather, upregulated in the mutant (up 1.54-fold). In line with our observation of abnormal OCT4 protein retention, we found that loss of BMP signalling was associated with upregulation of pluripotency-associated genes (e.g. Oct4, Sox2, Nanog, Cdh1, Prdm14, Tcl1, Dppa2, Esrrb and Zic3). Moreover, some genes that are normally highly expressed by fetal testicular but not ovarian germ cells (neither pre-meiotic nor meiotic) were aberrantly upregulated in the Bmpr1a-cKO [e.g. Lefty1, Hesx1, Asb9, Upp1 and Pramef12 (Pramel13); Jameson et al., 2012; Spiller et al., 2012]. This suggests BMP signalling is required to prevent some degree of germ cell masculinisation.
Some key genes expressed during mitosis-to-meiosis transition are BMP dependent, STRA8 dependent or co-dependent
Recent studies of PGCLCs (Miyauchi et al., 2017; Nagaoka et al., 2020) proposed that RA and BMP signalling synergistically promote oogenic germ cell fate, but the interplay between BMPs and RA/STRA8 has not been studied in vivo. To better understand meiotic onset in fetal ovarian germ cells, we analysed our Bmpr1a-cKO data alongside Stra8-null data (both RNA-Seq datasets derived from E14.5 whole ovaries, with DEGs defined as above). Appreciating the caveats that these single E14.5 timepoint datasets would not only identify direct downstream BMP/STRA8 targets, but would also capture differences due to delayed germ cell progression, we classified genes that were broadly (1) regulated only by BMP signalling, (2) regulated only by STRA8, and (3) regulated to some extent by both (Fig. 6; Table S4).
Key genes expressed during germ cell mitosis-to-meiosis transition are BMP dependent, STRA8 dependent, or co-dependent. DEGs in E14.5 Bmpr1aΔPGC and Stra8null ovaries were compared to identify genes that are regulated by BMP signalling, by STRA8, or by both. (A) BMP signalling and STRA8 both independently and cooperatively regulate expression of meiotic genes, early PGC markers, pluripotency-associated genes, and cell cycle regulators. Expression of selected genes are visualised in the heatmap. (B,C) Venn diagrams represent genes that are (B) induced by BMP signalling, STRA8, or both (downregulated in mutants), or (C) repressed by BMP signalling, STRA8, or both (upregulated in mutants). See Table S4 for the full gene list.
Key genes expressed during germ cell mitosis-to-meiosis transition are BMP dependent, STRA8 dependent, or co-dependent. DEGs in E14.5 Bmpr1aΔPGC and Stra8null ovaries were compared to identify genes that are regulated by BMP signalling, by STRA8, or by both. (A) BMP signalling and STRA8 both independently and cooperatively regulate expression of meiotic genes, early PGC markers, pluripotency-associated genes, and cell cycle regulators. Expression of selected genes are visualised in the heatmap. (B,C) Venn diagrams represent genes that are (B) induced by BMP signalling, STRA8, or both (downregulated in mutants), or (C) repressed by BMP signalling, STRA8, or both (upregulated in mutants). See Table S4 for the full gene list.
Comparing Stra8null versus Stra8WT returned 367 differentially downregulated and 77 differentially upregulated genes (Table S5). Interestingly, we found far fewer DEGs than the 2361 identified in a study where high-STRA8 and Stra8-null preleptotene cells from postnatal testis were similarly compared (Kojima et al., 2019). A large proportion of identified DEGs were also differentially expressed in the Bmpr1aΔPGC ovaries [52.3% (192 out of 367) of downregulated genes; 66.2% (51 out of 77) of upregulated genes] (Fig. 6). Despite the considerable overlap, 408 genes were induced (downregulated in Bmpr1a-cKO) and 789 genes were repressed (upregulated in Bmpr1a-cKO) because of BMP signalling, independently of STRA8. On the other hand, we found that STRA8 induces 175 genes (downregulated in Stra8null) and represses 26 genes (upregulated in Stra8null) independently of BMP signalling. Whether these expression changes are the direct or indirect consequences of SMAD1/5/8 activation (for BMP signalling) or direct STRA8 transcriptional activity cannot be ascertained from our datasets.
Meiotic onset and progression genes
Our results suggest most of the known meiotic and synaptonemal complex-related genes are regulated by both BMP signalling and STRA8 (downregulated in Bmpr1aΔPGC and Stra8null samples) (Fig. 6B; Table 1; Table S4). This group includes Mei4, Cntd1, Stag3, Syce1, and Sycp3. Notably, expression of Meiosin (Gm4969, encoding the interacting partner of STRA8, which is crucial for meiotic initiation; Ishiguro et al., 2020) was dependent on both STRA8 (down 6.5-fold) and BMP signalling (down 3.59-fold) (Fig. 6A; Table 1). In postnatal male germ cells, expression of Stra8 and Meiosin was reported as being mutually independent, with both genes relying on RA for their expression (Ishiguro et al., 2020); thus, our results suggest a sexual dimorphic regulation of Meiosin, consistent with a recent report that RA does not directly regulate Meiosin expression in female germ cells (Shimada et al., 2023). Direct transcriptional regulation of Meiosin by STRA8 is plausible, as STRA8 binding sites are present on the Meiosin promoter region (Ishiguro et al., 2020) and expression of Stra8 precedes that of Meiosin (Shimada et al., 2023). Consistently, target genes of MEIOSIN/STRA8 (e.g. Dmc1, Prdm9, Meiob, Msh5, Mcmdc2 and Brme1; Ishiguro et al., 2020; Takemoto et al., 2020) were significantly downregulated in both Bmpr1aΔPGC and Stra8null ovaries. Others found that Meioc upregulation at the onset of meiosis does not require RA or STRA8 (Abby et al., 2016); however, we found that Meioc expression is substantially affected by loss of STRA8 (down 4.07-fold) and by loss of BMP signalling (down 3.31-fold). This is in line with evidence that Meioc expression is dependent on STRA8 (Soh et al., 2015), and that its promoter region is bound by MEIOSIN and STRA8 (Ishiguro et al., 2020).
Expression of meiotic genes in Bmpr1aΔPGC and Stra8null mutants
Gene . | Function in the context of meiosis . | Female KO reproductive phenotype . | Stra8 null* . | Bmpr1a cKO* . | Reference . |
---|---|---|---|---|---|
Ankrd31 | Double-stranded break (DSB) formation | Premature sterility | n.s. | 0.37 | Papanikos et al. (2019) |
Cntd1 | Homologous recombination | Sterile | 0.32 | 0.59 | Holloway et al. (2014) |
Dmc1 | DSB repair | Sterile | 0.19 | 0.47 | Pittman et al. (1998); Yoshida et al. (1998) |
Hfm1 | Crossover formation and complete synapsis | Sterile | 0.32 | 0.31 | Guiraldelli et al. (2013) |
Figla | Transcription factor for oocyte-specific gene expression | Lack of ovarian follicle formation | 0.26‡ | 0.24 | Soyal et al. (2000) |
Hormad1 | Promotes homolog alignment and synaptonemal complex (SC) formation | Sterile | 0.44 | n.s. | Shin et al. (2010) |
Hormad2 | Removal of asynaptic oocytes | Fertile | 0.40 | 0.74 | Kogo et al. (2012); Wojtasz et al. (2012) |
Iho1 (Ccdc36) | DSB formation | Sterile | 0.12 | 0.33 | Stanzione et al. (2016) |
Kash5 (Ccdc155) | Chromosome synapsis | Sterile | 0.18 | 0.35 | Horn et al. (2013) |
M1ap | Crossover formation | Fertile | 0.16 | 0.34 | Li et al. (2023) |
Majin | Meiotic telomere complex protein | Sterile | 0.06 | 0.38 | Shibuya et al. (2015) |
Mcmdc2 | Meiotic recombination | Sterile | 0.27 | 0.38 | Finsterbusch et al. (2016) |
Mei1 | Chromosome synapsis | Sterile | 0.04 | 0.13 | Libby et al. (2003) |
Mei4 | DSB formation | Nearly devoid of primordial and primary follicles | 0.32 | 0.58 | Kumar et al. (2010) |
Meiob | Meiotic recombination | Sterile | 0.16 | 0.17 | Souquet et al. (2013) |
Meioc (Gm1564) | Post-transcriptional regulator | Sterile | 0.25 | 0.30 | Abby et al. (2016); Soh et al. (2015, 2017) |
Meiosin (Gm4969) | STRA8 co-factor | Sterile | 0.15 | 0.28 | Ishiguro et al. (2020) |
Mlh1 | DNA mismatch repair and crossing over | Sterile | n.s. | n.s. | Baker et al. (1996) |
Msh4 | Chromosome pairing | Sterile | 0.48§ | 0.48 | Kneitz et al. (2000) |
Msh5 | DSB repair | Sterile | 0.23 | 0.43 | de Vries et al. (1999) |
Prdm9 | Epigenetic events in meiotic prophase; determines sites of DSB | Sterile | 0.07 | 0.23 | Baudat et al. (2010); Hayashi et al. (2005) |
Rad21l | Cohesin to link homologous chromosomes | Pre-mature sterility | 0.05 | 0.18 | Herrán et al. (2011); Ishiguro et al. (2011); Lee and Hirano (2011) |
Rec8 | Meiotic-specific cohesin | Sterile | n.s. | 0.71 | Bannister et al. (2004); Xu et al. (2005) |
Rnf212 | Homologous recombination | Sterile | n.s. | 0.36 | Reynolds et al. (2013) |
Slc25a31 (Ant4) | Adenine nucleotide translocase | Fertile with smaller litter size | 0.40 | n.s. | Lim et al. (2015) |
Smc1b | Meiotic-specific cohesin | Sterile | 0.33 | 0.48 | Revenkova et al. (2004) |
Sohlh1 | Oocyte differentiation-related transcription factor | Sterile | n.s. | n.s | Pangas et al. (2006); Shin et al. (2017) |
Spata22 | Meiotic prophase progression | Sterile | 0.16¶ | 0.20 | La Salle et al. (2012) |
Spo11 | Catalyse DSBs | Sterile (premature ovarian failure) | n.s. | n.s. | Baudat et al. (2000); Keeney et al. (1997) |
Stag3 | Meiotic-specific cohesin | Sterile | 0.39 | 0.57 | Hopkins et al. (2014) |
Stra8 | Meiosis-specific transcription factor | Sterile | KO | n.s. | Baltus et al. (2006) |
Sun1 | Homologous pairing and synapsis formation | Sterile | n.s. | n.s. | Ding et al. (2007) |
Syce1 | SC structural protein | Sterile | 0.30 | 0.55 | Bolcun-Filas et al. (2009) |
Syce3 | SC structural protein | Sterile | n.s. | 0.32 | Schramm et al. (2011) |
Sycp1 | SC structural protein | Sterile | n.s. | 0.72 | de Vries et al. (2005) |
Sycp2 | SC structural protein | Subfertile | 0.24 | 0.49 | Yang et al. (2006) |
Sycp3 | SC structural protein | Subfertile | 0.26 | 0.39 | Yuan et al. (2002) |
Taf4b | Component of the Transcription factor IID (TFIID) complex | Sterile | 0.52 | 0.69 | Grive et al. (2014) |
Taf7l | Component of the TFIID complex | Fertile | 0.33§§ | 0.49 | Cheng et al. (2007) |
Terb1 | Meiotic telomere complex protein | Sterile | 0.09 | 0.26 | Shibuya et al. (2014) |
Terb2 | Meiotic telomere complex protein | Sterile | 0.27** | 0.48 | Shibuya et al. (2015) |
Tex11 | DSB repair and regulation of crossing over | Subfertile | 0.30 | n.s. | Yang et al. (2008) |
Tex12 | SC structural protein | Sterile | 0.20 | 0.39 | Hamer et al. (2008) |
Top6bl (Gm960) | DSB formation; forms a complex with SPO11 | Few primordial and primary follicles | 0.17 | 0.49 | Robert et al. (2016) |
Ythdc2 | Post-transcriptional regulator | Sterile | 0.68‡‡ | 0.71 | Bailey et al. (2017) |
Zglp1 | Oogenic program activator | Sterile | n.s. | 0.41 | Nagaoka et al. (2020) |
Gene . | Function in the context of meiosis . | Female KO reproductive phenotype . | Stra8 null* . | Bmpr1a cKO* . | Reference . |
---|---|---|---|---|---|
Ankrd31 | Double-stranded break (DSB) formation | Premature sterility | n.s. | 0.37 | Papanikos et al. (2019) |
Cntd1 | Homologous recombination | Sterile | 0.32 | 0.59 | Holloway et al. (2014) |
Dmc1 | DSB repair | Sterile | 0.19 | 0.47 | Pittman et al. (1998); Yoshida et al. (1998) |
Hfm1 | Crossover formation and complete synapsis | Sterile | 0.32 | 0.31 | Guiraldelli et al. (2013) |
Figla | Transcription factor for oocyte-specific gene expression | Lack of ovarian follicle formation | 0.26‡ | 0.24 | Soyal et al. (2000) |
Hormad1 | Promotes homolog alignment and synaptonemal complex (SC) formation | Sterile | 0.44 | n.s. | Shin et al. (2010) |
Hormad2 | Removal of asynaptic oocytes | Fertile | 0.40 | 0.74 | Kogo et al. (2012); Wojtasz et al. (2012) |
Iho1 (Ccdc36) | DSB formation | Sterile | 0.12 | 0.33 | Stanzione et al. (2016) |
Kash5 (Ccdc155) | Chromosome synapsis | Sterile | 0.18 | 0.35 | Horn et al. (2013) |
M1ap | Crossover formation | Fertile | 0.16 | 0.34 | Li et al. (2023) |
Majin | Meiotic telomere complex protein | Sterile | 0.06 | 0.38 | Shibuya et al. (2015) |
Mcmdc2 | Meiotic recombination | Sterile | 0.27 | 0.38 | Finsterbusch et al. (2016) |
Mei1 | Chromosome synapsis | Sterile | 0.04 | 0.13 | Libby et al. (2003) |
Mei4 | DSB formation | Nearly devoid of primordial and primary follicles | 0.32 | 0.58 | Kumar et al. (2010) |
Meiob | Meiotic recombination | Sterile | 0.16 | 0.17 | Souquet et al. (2013) |
Meioc (Gm1564) | Post-transcriptional regulator | Sterile | 0.25 | 0.30 | Abby et al. (2016); Soh et al. (2015, 2017) |
Meiosin (Gm4969) | STRA8 co-factor | Sterile | 0.15 | 0.28 | Ishiguro et al. (2020) |
Mlh1 | DNA mismatch repair and crossing over | Sterile | n.s. | n.s. | Baker et al. (1996) |
Msh4 | Chromosome pairing | Sterile | 0.48§ | 0.48 | Kneitz et al. (2000) |
Msh5 | DSB repair | Sterile | 0.23 | 0.43 | de Vries et al. (1999) |
Prdm9 | Epigenetic events in meiotic prophase; determines sites of DSB | Sterile | 0.07 | 0.23 | Baudat et al. (2010); Hayashi et al. (2005) |
Rad21l | Cohesin to link homologous chromosomes | Pre-mature sterility | 0.05 | 0.18 | Herrán et al. (2011); Ishiguro et al. (2011); Lee and Hirano (2011) |
Rec8 | Meiotic-specific cohesin | Sterile | n.s. | 0.71 | Bannister et al. (2004); Xu et al. (2005) |
Rnf212 | Homologous recombination | Sterile | n.s. | 0.36 | Reynolds et al. (2013) |
Slc25a31 (Ant4) | Adenine nucleotide translocase | Fertile with smaller litter size | 0.40 | n.s. | Lim et al. (2015) |
Smc1b | Meiotic-specific cohesin | Sterile | 0.33 | 0.48 | Revenkova et al. (2004) |
Sohlh1 | Oocyte differentiation-related transcription factor | Sterile | n.s. | n.s | Pangas et al. (2006); Shin et al. (2017) |
Spata22 | Meiotic prophase progression | Sterile | 0.16¶ | 0.20 | La Salle et al. (2012) |
Spo11 | Catalyse DSBs | Sterile (premature ovarian failure) | n.s. | n.s. | Baudat et al. (2000); Keeney et al. (1997) |
Stag3 | Meiotic-specific cohesin | Sterile | 0.39 | 0.57 | Hopkins et al. (2014) |
Stra8 | Meiosis-specific transcription factor | Sterile | KO | n.s. | Baltus et al. (2006) |
Sun1 | Homologous pairing and synapsis formation | Sterile | n.s. | n.s. | Ding et al. (2007) |
Syce1 | SC structural protein | Sterile | 0.30 | 0.55 | Bolcun-Filas et al. (2009) |
Syce3 | SC structural protein | Sterile | n.s. | 0.32 | Schramm et al. (2011) |
Sycp1 | SC structural protein | Sterile | n.s. | 0.72 | de Vries et al. (2005) |
Sycp2 | SC structural protein | Subfertile | 0.24 | 0.49 | Yang et al. (2006) |
Sycp3 | SC structural protein | Subfertile | 0.26 | 0.39 | Yuan et al. (2002) |
Taf4b | Component of the Transcription factor IID (TFIID) complex | Sterile | 0.52 | 0.69 | Grive et al. (2014) |
Taf7l | Component of the TFIID complex | Fertile | 0.33§§ | 0.49 | Cheng et al. (2007) |
Terb1 | Meiotic telomere complex protein | Sterile | 0.09 | 0.26 | Shibuya et al. (2014) |
Terb2 | Meiotic telomere complex protein | Sterile | 0.27** | 0.48 | Shibuya et al. (2015) |
Tex11 | DSB repair and regulation of crossing over | Subfertile | 0.30 | n.s. | Yang et al. (2008) |
Tex12 | SC structural protein | Sterile | 0.20 | 0.39 | Hamer et al. (2008) |
Top6bl (Gm960) | DSB formation; forms a complex with SPO11 | Few primordial and primary follicles | 0.17 | 0.49 | Robert et al. (2016) |
Ythdc2 | Post-transcriptional regulator | Sterile | 0.68‡‡ | 0.71 | Bailey et al. (2017) |
Zglp1 | Oogenic program activator | Sterile | n.s. | 0.41 | Nagaoka et al. (2020) |
*Expression levels relative to the control.
‡P=0.064, §P=0.060, ¶P=0.051, **P=0.063, ‡‡P=0.068, §§P=0.1447.
n.s., not significant.
Although a large subset of meiotic genes is regulated by both BMP signalling and STRA8, some genes are independently regulated by either BMP or STRA8 (Fig. 6; Table 1; Table S4). Genes downregulated only in Bmpr1aΔPGC ovaries include Zglp1, Figla, Msx1, Msx2, Syce3, Terb2, Ankrd31, Rnf212, and Spata22. Genes that are expressed independently of BMP signalling but are STRA8 dependent include Hormad1, Hormad2, and Tex11.
Interestingly, some genes known to be upregulated at meiotic onset were expressed normally in the absence of either BMP signalling or STRA8 protein, suggesting their induction might depend on RA, RA-responsive factors other than STRA8, unknown signalling factors present in the fetal gonad, or synergistic actions of BMP signalling and STRA8 (Table 1). As previously reported, Rec8 expression was fully independent of STRA8 (Soh et al., 2015), and its expression was reduced only marginally in the Bmpr1a-cKO (down 1.41-fold). This group also includes Sycp1, Spo11, and Sohlh1.
Markers of early PGCs and pluripotency
Diminished expression of early PGC markers and pluripotency-related genes upon meiotic entry could be regulated by STRA8, BMP signalling, or both (Fig. 6C; Table S4). Those with expression apparently repressed by STRA8 (upregulated in Stra8null) but not BMP signalling include Tfap2c, Prdm1, and Klf5, whilst downregulation of Dmrt1, Etv4, and Lncenc1 was dependent on BMP signalling but not STRA8. Genes that appear to be downregulated by both (upregulated in both Stra8null and Bmpr1aΔPGC ovaries) include Dppa2, Nanog, Oct4, Sox2, Prdm14, Trib3, and Utf1.
Genes potentially associated with the mitosis-to-meiosis switch
Consistent with findings that cell cycle genes are dynamically regulated during meiosis initiation (Zhao et al., 2020), such genes were aberrantly expressed in both Bmpr1aΔPGC and Stra8null ovaries (Fig. 6B,C; Table S4). We show that BMP signalling represses expression of Ccnb1 and Ccna2 but seems responsible for upregulating Ythdc2; elevation of YTHDC2 is necessary for downregulating Ccna2 during the mitotic-to-meiotic transition (Bailey et al., 2017). Elevated expression of Ccnd1 at meiotic onset also relies on intact BMP signalling. Although it is generally accepted that cyclin-dependent kinases are constitutively present during the cell cycle while levels of cyclins fluctuate cyclically, we find evidence that BMP signalling downregulates Cdk1 expression. This could be functionally relevant, as decreased expression of Cdk genes is also observed in postnatal testicular germ cells as they progress into meiosis (Diederichs et al., 2005). In agreement with findings from a recent study (Shimada et al., 2023), our data indicate that STRA8 is required to upregulate Cdkn2a, and further clarifies that it is only the p19ARF-encoding isoform of Cdkn2a that is upregulated when STRA8 is present.
Our data support the theory that BMP signalling and STRA8 act together to influence the mitosis-to-meiosis switch. Ccne2, which is upregulated in early meiotic germ cells and encodes a cyclin required in meiotic prophase I, at least in males (Martinerie et al., 2014), was significantly downregulated in both Stra8null and Bmpr1aΔPGC E14.5 ovaries. Expression of the meiosis-specific Ccnb3 also requires both STRA8 and BMP signalling. Another gene regulated by both BMP signalling and STRA8 is Inca1, encoding a novel Cyclin A1/CDK2 inhibitor expressed in male and female germ cells as they enter meiotic prophase (Diederichs et al., 2005; Jameson et al., 2012). Overall, our results support a multi-faceted role for BMP signalling in instructing germ cell meiosis in female mice in vivo, and demonstrate the importance of both BMP signalling and STRA8 in ensuring a proper mitosis-to-meiosis transition.
DISCUSSION
Successful mitosis-to-meiosis transition is essential for gametogenesis. In mammalian females, this process occurs during embryogenesis, giving rise to the non-renewable ovarian reserve. Evidence indicates that ovarian germ cells respond to endogenous RA by expressing the transcription factor STRA8, which drives meiotic initiation (Bowles et al., 2006; Feng et al., 2021; Koubova et al., 2006; Soh et al., 2015). Recent findings from mouse PGCLC studies suggest that BMP signalling is also required, at least in vitro, for adoption of the oogenic fate (Miyauchi et al., 2017). Such in vitro studies are powerful and allow a relatively simple analysis of germ cell sex determination ‘in a reductive and a constructive fashion’ (Miyauchi et al., 2017). Nonetheless, it remains uncertain as to whether and how BMP signalling impacts ovarian germ cell development in vivo. In this study, we provide in vivo evidence that expression of Stra8 is independent of BMP signalling but expression of Meiosin, which encodes the STRA8-interacting protein, is partially dependent on BMP signalling, in the mouse fetal ovary. Combining transcriptome analyses of Bmpr1aΔPGC and Stra8null fetal ovaries, we demonstrate that BMP signalling and STRA8 have independent but also common targets for ensuring proper expression of meiotic factors and downregulation of pluripotency (Fig. 7).
Both BMP signalling and STRA8 are required for fetal ovarian germ cell meiosis in vivo. This article demonstrates an essential role for BMP signalling in fetal germ cell meiosis. We conclude that BMP signalling and STRA8 both directly or indirectly contribute to downregulating pluripotency in differentiating germ cells, and cooperatively ensure a proper mitotic-to-meiotic switch and activate factors required for meiotic chromosome activity. Through ZGLP1, BMP signalling in germ cells also promotes a meiotic transcription program. BMP signalling and STRA8 also independently promote germ cell meiosis by downregulating the spermatogenic program and repressing proliferation, respectively.
Both BMP signalling and STRA8 are required for fetal ovarian germ cell meiosis in vivo. This article demonstrates an essential role for BMP signalling in fetal germ cell meiosis. We conclude that BMP signalling and STRA8 both directly or indirectly contribute to downregulating pluripotency in differentiating germ cells, and cooperatively ensure a proper mitotic-to-meiotic switch and activate factors required for meiotic chromosome activity. Through ZGLP1, BMP signalling in germ cells also promotes a meiotic transcription program. BMP signalling and STRA8 also independently promote germ cell meiosis by downregulating the spermatogenic program and repressing proliferation, respectively.
BMP signalling is important in many aspects of fetal germ cell development; here, we aimed to investigate the function of BMPs in post-migratory germ cells with a view to confirming a role in sexual fate commitment. Noting that BMP2 and BMP5 are likely to be present in the fetal ovary, but not the fetal testis (Jameson et al., 2012; Yao et al., 2004), we chose to genetically delete Bmpr1a, encoding the BMP receptor BMPR1A, specifically in the germ cells after gonadal colonisation. Both BMP2 and BMP5 signal through BMPR1A and BMPR1B (ALK6), and BMP5 additionally uses ACVR1 (Mueller and Nickel, 2012). Bmpr1a expression is high, and Bmpr1b expression is negligible in fetal germ cells, with low expression detected for Acvr1 (Jameson et al., 2012). Whilst we cannot exclude the possibility that BMPs could still signal through ACVR1 in our Bmpr1a-deleted model, we expect that BMPR1A-mediated signalling from soma to germ cells is largely abrogated. Our near-complete loss of BMPR1A immunostaining along with the BMP signalling effector Zglp1 confirmed that the model was appropriate for our in vivo investigation.
BMP signalling is not required for initiation of Stra8 expression but is essential for meiotic entry and progression
The requirement for STRA8 in female germ cell meiosis is unequivocal (Baltus et al., 2006); however, the necessity for RA in this context has been a long-standing debate (Chassot et al., 2020; Kumar et al., 2011; Vernet et al., 2020). Although we acknowledge the controversy, here we are not focusing on the question of whether RA is required to induce Stra8 expression. Rather, we set out to determine whether BMP signalling is crucial, alongside STRA8, to ensure correct transition from mitosis to meiosis in vivo.
Using our germ cell-specific Bmpr1a-cKO model, we showed that the loss of BMP signalling does not abrogate the timely onset, level of expression, or the anterior-to-posterior induction pattern of Stra8 in fetal ovaries. These results support the hypothesis that Stra8 induction is driven by RA, as this signalling factor is present at higher levels at the anterior end of the ovary (Bowles et al., 2006). It is, of course, formally possible that non-RA and non-BMP signalling factors, which are present particularly at the anterior end of the gonad, are involved. This result is consistent with findings in PGCLCs, where provision of RA alone was just as successful in inducing full activation of Stra8 expression as was RA together with BMP2 (Miyauchi et al., 2017).
Despite normal onset of Stra8 expression, its downregulation is perturbed, and there is elevated cytoplasmic accumulation of STRA8 protein in Bmpr1a-null germ cells after E14.5. It is possible that elevated Stra8 expression results in increased STRA8 protein production; if this was to saturate its active nuclear import (Tedesco et al., 2009), then STRA8 might accumulate in the cytoplasm. Another more-specific mechanism could relate to the reduced expression of Meiosin in germ cells in the absence of BMP signalling (Fig. 5A). In mouse spermatogenic cells, loss of Meiosin led to cytoplasmic accumulation of STRA8 (Ishiguro et al., 2020), a phenotype similar to that observed in our mutant ovaries. Whether MEIOSIN regulates the nuclear import of STRA8 in fetal ovaries requires further investigation. With respect to aberrant maintenance of Stra8 expression in Bmpr1a-null germ cells, increased expression of Dmrt1 (Fig. 5A) could enhance and possibly maintain expression of Stra8 through the binding of DMRT1 to its proximal promoter (Feng et al., 2021; Krentz et al., 2011). More generally, it is possible that Stra8 expression is maintained in the absence of BMP signalling because meiosis is not progressing or, more specifically, because there is insufficient nuclear STRA8 to downregulate its own expression (Soh et al., 2015). Future assessment of the expression, (co)localisation, and modification of STRA8 and MEIOSIN would help shed light on these possibilities.
Although Stra8 expression is initiated normally, we find that loss of BMP signalling causes a delay in meiotic prophase entry and progression. In Bmpr1aΔPGC ovaries, transcription of key meiotic genes was significantly reduced, fewer germ cells were positive for SYCP3, and those that did express SYCP3 were temporally delayed in meiotic progression. However, despite very low Zglp1 expression, our phenotype did not recapitulate the near complete absence of SYCP3 observed in the ubiquitous Zglp1−/− model (Nagaoka et al., 2020), possibly due to the specificity of our knockout to post migratory germ cells: in the Zglp1−/− model, ZGLP1 would have been absent throughout development in germ cells and somatic cells. We were unable to analyse postnatal phenotypes because treatment with 4-hydroxytamoxifen precludes parturition (Savery et al., 2020). Future studies with non-inducible CRE lines that are active in early gonadal germ cells (Burnet et al., 2023) will be required to assess the adult phenotype.
Both BMP signalling and STRA8 are required for proper mitosis-to-meiosis transition in vivo
Transcriptome profiling of Bmpr1aΔPGC ovaries allowed identification of direct or indirect targets of BMP signalling in fetal ovarian germ cells. Comparison of Bmpr1a-cKO and Stra8null datasets further confirmed that both factors are indispensable for fetal germ cell meiosis. Regulatory targets of STRA8 have been reported for males (Kojima et al., 2019) but not yet for females. Further studies to identify and verify factors immediately downstream of BMP signalling and STRA8 in the context of germ cell mitosis-to-meiosis transition would be valuable for elucidating this complicated mechanism.
Our combined transcriptome analyses revealed that a plethora of cell cycle genes are independent or shared targets of BMP signalling and/or STRA8. For example, we find that expression of Ccnd1 is dependent on BMP signalling. Despite Cyclin D1 being considered a proliferation-associated G1 mitotic marker, Ccnd1 transcript expression increases sharply in early meiotic germ cells, and is downregulated during later stages of meiosis in an anterior-to-posterior wave (Heaney et al., 2012; Zhao et al., 2020). Thus, Cyclin D1 potentially has a role in the mitosis-to-meiosis transition and its expression is induced by BMP. We find that STRA8 is required for Cdkn2a transcription, as has recently been highlighted (Shimada et al., 2023). Cdkn2a encodes alternative isoforms p16INK4a and p19ARF, both of which can reduce CDK activity, thereby directly (p16INK4a) or indirectly (p19ARF) impacting cell cycle progression (Chin et al., 1998). We extend what is known regarding these cell cycle regulators as we find that STRA8 upregulates only the isoform encoding p19ARF, and that BMP signalling is not required for Cdkn2a induction. In the mouse testis, the p19ARF variant is transiently expressed in spermatogonia (Gromley et al., 2009) and prevents spermatocytes from undergoing p53-dependent apoptosis, thereby supporting meiotic progression (Churchman et al., 2011). In the females, however, loss of Arf caused no discernible reproductive defects (Churchman et al., 2011). The functional importance of p19ARF expression at the mitosis-to-meiosis transition in ovarian germ cells remains uncertain and warrants further investigation.
BMP signalling likely also affects germ cell development prior to sexual fate determination
Whilst our findings on the role of BMP signalling in female germ cell fate specification are mostly in line with the observations from PGCLCs (Miyauchi et al., 2017; Nagaoka et al., 2020), we noted a few discrepancies between the behaviour of gonadal germ cells and PGCLCs. In our Bmpr1a-cKO model, loss of BMP signalling did not affect the expression of germ cell marker Mvh and ‘licensing factor’ Dazl; this is in contrast to the report from Miyauchi et al. (2017), where inhibition of the BMP signalling pathway by LDN193189 injection (to pregnant dams) impaired Mvh and Dazl expression in fetal ovarian germ cells. Since LDN193189 treatment inhibits BMP signalling in both germ cell and gonadal somatic cells, and, further, because induction of MVH in germ cells seems to require an intercellular interaction with gonadal somatic cells (Toyooka et al., 2000), these disparate results raise the interesting possibility that normal Mvh induction in ovarian germ cells requires BMP signalling, but not direct signalling to germ cells. Moreover, LDN193189 treatment in pregnant dams resulted in reduced Mvh expression in XX but not XY germ cells (Miyauchi et al., 2017): it is possible that upregulation of Mvh, upon gonadal colonisation (Seligman and Page, 1998; Toyooka et al., 2000), is driven by different mechanisms in fetal ovary and fetal testis, and that BMP plays a role in driving ovarian somatic cell fate rather than directly acting on germ cells in this context. As PGCLCs transcriptionally resemble migratory PGCs rather than gonadal PGCs (Ohta et al., 2017), it seems likely that one function of BMP in the in vitro system is to mature PGCLCs to a state that is responsive to fate-determining signals provided by the gonadal environment.
Conclusion
A role for BMPR1A-mediated signalling in ensuring correct fetal ovarian germ cell meiosis is now unequivocal. Our in vivo data support most of the conclusions reached by Miyauchi et al. (2017) and Nagaoka et al. (2020) in their studies of PGCLCs. Importantly, our findings support the contention that the PGCLC system recapitulates, to a great extent, germ cell development in vivo. Our work helps clarify the in vivo mechanisms that underlie commitment of germ cells to the female fate – information that may help in the quest to increase efficiency and derive safe and effectives oocytes, in vitro, for eventual therapeutic use. Our data also contribute to the knowledgebase with respect to the question of how germ cells lose pluripotency and differentiate – information that is relevant to stem cell biology more broadly.
MATERIALS AND METHODS
Animals
All procedures involving animals and their care were carried out in accordance with Australian national, state, and institutional guidelines. Breeding of mice and animal experiments were approved by The University of Queensland Animal Ethics Committee. The Bmpr1atm2.1Bhr (referred to as Bmpr1afl/fl in this work), Oct4-MerCreMer (referred to as Oct4-CreERT2 in this work) and Stra8Δ173 (referred to as Stra8null in this work) mouse lines have been described by Mishina et al. (2002), Greder et al. (2012), and Feng et al. (2021), respectively. All mouse lines were maintained on a pure C57BL/6 background; imported Bmpr1afl/fl animals were received on a C57BL/6×129SV background, and subsequently backcrossed with wild-type C57BL/6 for more than nine generations to obtain a largely C57BL/6 background. Oct4-CreERT2 animals were intercrossed with Bmpr1afl/fl animals to produce Bmpr1afl/fl;Oct4-CreCre/WT male studs for timed matings with Bmpr1afl/fl females. All mice were housed at the University of Queensland Biological Resources, Queensland Bioscience Precinct, Research Animal Facility with a 12 h light/dark cycle. Genotyping of the mice were performed by PCR as described previously (Chuma and Nakatsuji, 2001; Greder et al., 2012). Primer sequences are listed in Table S6.
Timed matings and tissue collection
For embryo collections, timed matings were set up and noon of the day on which a vaginal plug was observed was designated as 0.5 days post coitum (dpc)/embryonic day (E) 0.5. Wild-type C57BL/6 males and females were housed together for timed matings for ex vivo gonad cultures. Embryos were collected at E11.5, and urogenital ridges (UGRs; gonad plus adjacent mesonephros) were dissected out for ex vivo gonad culture. For in vivo experiments, Bmpr1afl/fl;Oct4-CreCre/WT studs and Bmpr1afl/fl females were time-mated. Pregnant females were injected intraperitoneally with 1 mg 4-hydroxytamoxifen (4-OHT) (H6278, Sigma-Aldrich) and 0.5 mg progesterone (P0130, Sigma-Aldrich) at 9.5, 10.5 and 11.5 dpc to induce germ cell-specific CRE-dependent recombination of Bmpr1a in the embryos. Pregnant females were euthanised, and Bmpr1afl/fl;Oct4-CreCre/WT (referred to as Bmpr1aΔPGC or Bmpr1a-cKO in this article) embryos and Bmpr1afl/fl control littermates were collected (at timepoints ranging from E12.5 to E16.5) for subsequent analysis.
Embryo sex was determined by visual inspection of dissected gonads (for embryos aged E12.5 or older) while Ube1 genotyping using tail tissue (Chuma and Nakatsuji, 2001) was performed to confirm the sex of samples collected before E12.5 or for whole embryo collections (all ages). For gene expression and immunofluorescence analyses, E12.5 – E15.5 right gonads were dissected with the mesonephroi removed and stored in RNAlater (R0901, Merck) at 4°C until RNA extraction. Left gonads were left in the embryos and fixed in 4% paraformaldehyde in PBS (PFA/PBS), dehydrated through an ethanol series (25%, 50% and 75% v/v in water), processed (Leica ASP300 S Tissue Processor), and embedded in paraffin for sectioning and histological staining.
Organ culture
Freshly dissected C57BL/6 E11.5 UGRs were cultured in 24-well plates in hanging drops (McClelland and Bowles, 2016) of 45 μl StemPro-34 SFM (10639011, Gibco) for 24, 48, or 72 h at 37°C in 5% CO2, replacing with fresh medium (including treatments where relevant) every 24 h. UGRs were treated with LDN193189 (500 nM in DMSO, SML0559, Sigma-Aldrich) or DMSO (D2650, Sigma-Aldrich). For each embryo, one UGR was used in the control group and one in the treatment group. Cultured samples were collected in pools of 2-4 and stored in RNAlater (AM7021, Thermo Fisher Scientific) at 4°C until RNA extraction.
Nucleic acid extraction
Genomic DNA samples for genotyping was extracted from ear notches, toe clippings, or embryonic tail using QuickExtract Solution (QE0905, Lucigen) according to the manufacturer's protocol. Total RNA was extracted from cultured UGRs, individual gonads, or gonad pairs using the RNeasy Micro Kit (74004, Qiagen), including on-column DNase treatment, according to the manufacturer's protocol.
Quantitative reverse transcription PCR (qRT-PCR)
Reverse transcription was performed immediately after RNA extraction to synthesise cDNA using the High-Capacity cDNA Reverse Transcription Kit (4368813, Applied Biosystems). TaqMan Gene Expression Assays (Table S7) and Universal TaqMan Master Mix (4318157, Applied Biosystems) were used for gene expression quantification. Realtime-PCR runs were performed on a QuantStudio 7 Flex Real-Time PCR System (Applied Biosystems). Expression of germ cell-specific transcripts was normalised against the expression level of mouse vasa homolog (Mvh, also known as Ddx4), a marker of germ cells, to control for germ cell number. Expression levels of other genes were normalised against the house-keeping gene Tbp.
Histological sectioning and staining
Whole embryos with right gonad excised (retained for RNA extraction) were collected for histological analyses and fixed in 4% paraformaldehyde in PBS (PFA/PBS), dehydrated in ethanol series (25%, 50%, 75% and 100% v/v in water), and embedded in paraffin wax for sectioning. Whole embryos were serial-sectioned at 5 µm on a Leica RM Rotary Microtome, and dewaxed in xylene, followed by rehydration through an ethanol series from 100% to 35% v/v in water. Immunofluorescence was performed as previously described (Bowles et al., 2010). Primary and secondary antibodies used are listed in Table S8. Images were obtained with a DP70 colour camera (Olympus) on a BX-51 upright florescence microscope (Olympus), or a LSM900 Fast AiryScan2 Confocal microscope (Zeiss). All immunofluorescence images are representative images of n≥3; analyses were conducted in FIJI on single images. For measuring cell roundness, cell boundaries were manually defined based on MVH immunosignal, ‘Roundness’ was measured using the ‘Measure’ function. For quantification of STRA8+, SYCP3+, OCT4+, or BrdU+ germ cells, the ‘Cell Counter’ plug-in was used for cell counting from at least two serial sections (every tenth) of each biological replicate, except for BrdU, which was counted from image sections from anterior or posterior ends. For STRA8 subcellular localisation, cytoplasmic and nuclear signal max. intensity was compared using the ‘Plot Profile’ function.
Meiotic chromatin spreads
Meiotic chromatin spreads were prepared from E15.5 and E16.5 ovaries as per Hwang et al. (2018), with the following modifications. Briefly, ovaries were placed in 0.5 ml freshly made Hypotonic Extraction Buffer [HEB (pH 8.2-8.4); 30 mM Tris-HCl (pH 7.2), 50 mM sucrose, 17 mM trisodium citrate dihydrate, 5 mM EDTA, 0.5 mM DTT and 0.1 mM PMSF] in a 12-well plate and incubated on ice for 15-30 min. After incubation, each ovary was placed in 25 µl of 100 mM sucrose on a microscopic slide, within a square pre-drawn using hydrophobic barrier PAP pen (00-8877, Invitrogen). Cells from the ovaries were released by teasing open the ovary using 27 G needles, and carefully dispersed by pipetting. Cells were fixed in the squares with 40 µl 1% paraformaldehyde/0.15% Triton X (pH 9.2), and the slides were incubated overnight in a humid chamber at room temperature. Slides were air-dried on day 2 and stored at −80°C until immunofluorescence staining.
Immunofluorescence staining reagents were prepared as follows: antibody dilution buffer (ADB) – 3 g BSA (BSAS 0.05, Bovogen), 10 ml horse serum, 250 µl 20% Triton X and 10 ml 10×PBS made up to 100 ml with water; ADB/PBS solution – 10% ADB solution in PBS; PBTX solution – 0.1% Triton X in PBS; Photoflo/PBS – 0.4% Kodak Photo-Flo 200 Solution in PBS; Photoflo/H2O – 0.4% Kodak Photo-Flo 200 Solution in water; anti-fade mounting media – 0.932 g DABCO, 3.2 ml water, 800 µl 1 M Tris-HCl (pH 8.0), 36 ml glycerol and 40 µl DAPI (0.2 mg/ml).
Chromatin spreads slides were blocked for 10 min each in Photoflo/PBS, PBTX, and ADB/PBS, and incubated overnight at room temperature with primary antibodies: rabbit polyclonal anti-SYCP1 (ab15090, Abcam) and mouse monoclonal anti-SYCP3 (ab97672, Abcam) diluted at 1:200 in ADB. The blocking steps were repeated prior to incubation with secondary antibodies at 37°C for 1 h. Secondary antibodies used were: goat anti-rabbit IgG (H+L) Alexa Fluor 488 (A11034, Invitrogen) and goat anti-mouse IgG (H+L) Alexa Fluor 594 (A11032, Invitrogen), both of them at a concentration of 1:2000 diluted in ADB. Slides were washed three times in Photoflo/PBS, once in Photoflo/H2O, and then air-dried in the dark before mounting with 60 µl antifade mounting media. Chromatin spreads were visualised and imaged at 100× with DP70 colour camera (Olympus) on a BX-51 upright florescence microscope (Olympus). Meiotic prophase I substages for SYCP3+ cells were classified based on the synaptonemal complex structure visualised by anti-SYCP1 and anti-SYCP3 as follows: leptotene – synaptonemal complex starts to accumulate on chromosome, fragmented SYCP1 and SYCP3 signals; zygotene – chromosomes show signs of condensation, axes of the chromosomes marked by long thread-like SYCP3 signals with SYCP1 signals colocalised in some regions undergoing synapsis; pachytene – chromatin fully condensed and synapsed, SYCP1 and SYCP3 signals colocalised and appearing shorter and thicker than previous stages; late pachytene – chromatin shows early signs of SYCP1 dissociation from the lateral elements.
Magnetic-activated cell sorting
Gonads collected at E13.5, were dissociated using Cell Dissociation Buffer (13151014, Gibco) for 15 min at 37°C followed by pipetting up and down until a single cell suspension was formed. Germ cells were enriched from the cell suspension using anti-SSEA-1 (CD15) microbeads (130-094-530, Miltenyi Biotec) on MS columns as per the manufacturer's protocol.
Bulk RNA sequencing and data analysis
Fetal ovaries were collected from eight E14.5 Bmpr1aΔPGC embryos and eight Bmpr1afl/fl littermate controls across six litters; gonad pairs were pooled into eight samples (n=4 each of mutant and control, two ovary pairs per sample) and total RNA was freshly isolated as described above. cDNA libraries were constructed from 500 ng total RNA using NEBNext Ultra II Directional RNA Library Prep Kit for Illumina (E7760S, NEB), optimised for an approximate final library size of 520 bp, following manufacturer's protocols. Strand-specific paired-end libraries were sequenced on Illumina NovaSeq with a read length of 150 bp at the Australian Genome Research Facility (AGRF), Melbourne.
Reads were cleaned using Trimmomatic (Bolger et al., 2014) and mapped to the Mus musculus genome (mm10) using STAR 2.5.2 (Dobin et al., 2013), and read counts per gene were quantified using the same tool. Differential gene expression analysis was conducted using edgeR (Robinson et al., 2010) with the GLM approach.
Differential gene expression data were visualised using R, sashimi plots were generated with the tool ggsashimi (Garrido-Martín et al., 2018), and Gene ontology (GO) analysis was performed using DAVID (Huang et al., 2009; Sherman et al., 2022), using all expressed genes in the samples as the background.
Statistical analyses
Statistical analyses were conducted using Prism 9, n≥3 unless otherwise stated, graphs were plotted as mean±s.e.m. for TaqMan gene expression analyses and protein localisation analyses, and as mean±s.d. for cell roundness. Paired t-test (two-tailed) was used to calculate statistical significance in ex vivo gonad culture experiments; unpaired t-test (two-tailed) was used to calculate statistical significance in in vivo experiments.
Acknowledgements
Confocal microscopy, qRT-PCR, and some histological procedures were carried out at the Core Facilities at the School of Biomedical Sciences, University of Queensland. We thank all the staff at the University of Queensland Biological Resources animal facility for their assistance with mouse husbandry. We acknowledge the University of Queensland Research Computing Centre (RCC) for its support in this research.
Footnotes
Author contributions
Conceptualization: C.M.S., J.B.; Formal analysis: F.K.M.C.; Funding acquisition: C.M.S., J.B.; Investigation: F.K.M.C., C.-W.A.F., C.C.; Methodology: F.K.M.C., C.M.S., J.B.; Resources: Y.M., J.B.; Supervision: C.-W.A.F., C.M.S., J.B.; Writing – original draft: F.K.M.C., J.B.; Writing – review & editing: F.K.M.C., C.-W.A.F., C.C., Y.M., C.M.S., J.B.
Funding
This research was supported by the Australian Research Council (DP200102896 to J.B. and C.M.S). F.K.M.C was supported by the University of Queensland Research Training Program Scholarship and the Lalor Foundation Fellowship. Open Access funding provided by the University of Queensland. Deposited in PMC for immediate release.
Data availability
Bulk RNA-Seq data for E14.5 Bmpr1aΔPGC and control ovaries are have been deposited in GEO under accession number GSE268565.
References
Competing interests
The authors declare no competing or financial interests.