In mammalian sex determination, SRY directly upregulates the expression of SOX9, the master regulatory transcription factor in Sertoli cell differentiation, leading to testis formation. Without SRY action, the bipotential gonadal cells become pre-granulosa cells, which results in ovarian follicle development. When, where and how pre-granulosa cells are determined to differentiate into developing ovaries, however, remains unclear. By monitoring SRY-dependent SOX9 inducibility (SDSI) in an Sry-inducible mouse system, we were able to identify spatiotemporal changes in the sexual bipotentiality/plasticity of ovarian somatic cells throughout life. The early pre-granulosa cells maintain the SDSI until 11.5 d.p.c., after which most pre-granulosa cells rapidly lose this ability by 12.0 d.p.c. Unexpectedly, we found a subpopulation of the pre-granulosa cells near the mesonephric tissue that continuously retains SDSI throughout fetal and early postnatal stages. After birth, these SDSI-positive pre-granulosa cells contribute to the initial round of folliculogenesis by the secondary follicle stage. In experimental sex reversal of 13.5-d.p.c. ovaries grafted into adult male nude mice, the differentiated granulosa cells re-acquire the SDSI before other signs of masculinization. Our data provide direct evidence of an unexpectedly high sexual heterogeneity of granulosa cells in developing mouse ovaries in a stage- and region-specific manner. Discovery of such sexually bipotential granulosa cells provides a novel entry point to the understanding of masculinization in various cases of XX disorders of sexual development in mammalian ovaries.
In mammalian gonadogenesis, both testicular Sertoli cells and ovarian granulosa cells develop from a common supporting cell precursor (Albrecht and Eicher, 2001) which arises from the coelomic epithelium (Karl and Capel, 1998; Schmahl et al., 2000; Schmahl and Capel, 2003; Sekido et al., 2004). In XY males, Sry is transiently [11.0–12.0 days post coitum (d.p.c.)] activated in a center-to-pole wave-like pattern along the anteroposterior (AP) axis of the indifferent XY gonads (Harikae et al., 2013). In these bipotential supporting cells, SRY directly upregulates another, autosomal Sry-box gene, Sox9 (Sekido and Lovell-Badge, 2008), and Sox9 induces Fgf9 expression in a similar center-to-pole pattern (Kim et al., 2006; Hiramatsu et al., 2010). FGF9 signals, in turn, upregulate SOX9 expression, resulting in the maintenance of high-level SOX9 expression in Sertoli cells.
In contrast, without Sry expression, the bipotential supporting cells become pre-granulosa cells, leading to the first follicle formation soon after birth (Albrecht and Eicher, 2001; Mork et al., 2012). Similar to SRY/SOX9 expression in Sertoli cell differentiation, ovarian differentiation is marked by early female-specific expression of several genes including Wnt4 (Vainio et al., 1999), Rspo1 (Parma et al., 2006) and FoxL2 (Uda et al., 2004; Ottolenghi et al., 2005). The enforced expression of either FOXL2 or active β-catenin disrupts testis development in XY gonads (Ottolenghi et al., 2007; Maatouk et al., 2008; Garcia-Ortiz et al., 2009). In contrast, ovarian development in XX gonads lacking a single gene among them is not severely affected (Vainio et al., 1999; Uda et al., 2004; Chassot et al., 2008; Tomizuka et al., 2008; Liu et al., 2009). Even in Wnt4/Foxl2 double-null embryos, female-to-male sex reversal starts to occur at the perinatal period (Ottolenghi et al., 2007). Therefore, it remains unclear how many other ovarian genes are involved in the switch from the sexually bipotential state to an ovarian fate (Nef et al., 2005; Beverdam and Koopman, 2006; Garcia-Ortiz et al., 2009; Munger et al., 2009; Chen et al., 2012; Jameson et al., 2012). It is also unclear when, where and how granulosa cell fate is determined in developing fetal ovaries.
In mammalian folliculogenesis, it is well known that ovarian follicles can be classified into two different populations: (1) the first follicles in the initial round of folliculogenesis soon after birth; and (2) the primordial follicle pool in the cortex region of the adult ovaries (Hirshfield, 1992; Hirshfield and DeSanti, 1995; Mork et al., 2012). Recently, Mork and colleagues demonstrated that, in developing XX gonads, the coelomic epithelia continuously contribute to the pre-granulosa cell population in ovarian parenchyma even after birth (Mork et al., 2012). Soon after birth, the pre-granulosa cells located in the ovarian parenchyma at the fetal stage mainly contribute to the initial round of folliculogenesis in the ovarian medullary region. In contrast, the granulosa cells in the resting primordial follicles in the cortex region appear to be newly recruited from the coelomic epithelia at the perinatal period (Hirshfield, 1992; Mork et al., 2012). However, in these resting primordial follicles, it remains unknown how pre-granulosa cell differentiation occurs in the ovarian cortex region during the perinatal periods. It is also unclear to what extent each granulosa cell population located in the cortex/medullary region has the heterogeneity in their properties including the sexual bipotentiality and plasticity in developing ovaries.
Previously, we have demonstrated that in XX Sry transgenic gonads, forced ubiquitous SRY expression in the entire gonadal area at the stages prior to normal Sry induction resulted in neither any advance in timing nor ectopic activation of Sox9 expression in developing gonads (Kidokoro et al., 2005). This finding indicates that SRY-dependent Sox9 inducibility is tightly regulated in gonadal precursor cells, and that only supporting cells which correspond to Sry-expressing pre-Sertoli cells in XY gonads achieve a competent state to respond to SRY action in XX gonads. Moreover, we have developed a novel Sry-inducible transgenic (Hsp-Sry Tg) system to induce XX testis sex reversal (Hiramatsu et al., 2009). In this Hsp-Sry Tg system (which we refer to as ‘Tg’ in this paper), heat shock (HS) treatment induces ectopic Sry expression, leading to transient Sox9 induction in XX Tg supporting cells. When the Hsp-Sry transgene was induced during the critical time window of 11.0–11.25 d.p.c. [i.e. 12–14 tail somites (ts)], Sox9 expression was not only transient but maintained at high levels, resulting in testis formation of XX Tg gonads. After this critical time period, ectopic Sry induction initially induced SOX9 expression in most supporting cells by 11.5 d.p.c., however, Sox9 expression was not maintained, resulting in ovarian differentiation (Hiramatsu et al., 2009). In fetal ovaries at 12.0–12.5 d.p.c., HS-dependent Sry expression did not induce ectopic SOX9 initiation in most ovarian cells, suggesting that the loss of SOX9 inducibility is one of the earliest key events in pre-granulosa cell differentiation. Hence, we hypothesized that the loss and re-acquisition of the potency to initiate SRY-dependent Sox9 activation can be used to monitor the sexually bipotential states in ovarian development under normal and experimental conditions.
By using our Hsp-Sry Tg system, this study is the first to visualize spatiotemporal changes in SRY-dependent SOX9 inducibility (referred to as ‘SDSI’ in this paper) in developing granulosa cells of normal and masculinized ovaries. Here, we demonstrate the sexual heterogeneity of ovarian granulosa cells in a stage- and region-specific manner.
XX gonadal somatic cells rapidly lose SRY-dependent SOX9 inducibility in an anterior-to-posterior wave-like manner during 11.5 to 12.0 d.p.c.
First, we examined the onset of the loss of SDSI in developing XX gonads. In brief, XX Hsp-Sry Tg (Tg) gonads were isolated at various tail somite stages, subjected to heat shock (HS) treatment (43°C, 10 minutes), and then cultured for 9 hours to allow immunohistochemical detection of SOX9-positive ovarian cells (Fig. 1A). SOX9 immunostaining detected no positive signals in any non-HS-treated XX Tg explants (‘-HS’ in Fig. 1B). In 12–30 tail somite stage (ts) XX Tg gonads, HS treatment induced ubiquitous SRY expression at 3 hours after HS (supplementary material Fig. S1) (Hiramatsu et al., 2009). In 12–18 ts XX Tg gonads, HS treatment initiates SOX9 expression in most gonadal supporting cells ∼9 hours after HS (upper panel in ‘15ts’ of Fig. 1B), indicating the SDSI-positive states in most XX supporting cells at 12–18 ts. Interestingly, in 19–21 ts XX Tg explants, the number of SDSI-positive (i.e. SOX9-positive) cells was reduced in the anterior pole region, as compared to the center and posterior regions (‘19ts’ and ‘21ts’ in Fig. 1B; see Fig. 2D). Subsequently, the number of SDSI-positive ovarian cells in HS-treated explants was rapidly reduced in most somatic cells in XX Tg gonads by 24–25 ts (Fig. 1B).
Two populations of ovarian somatic cells retain SDSI in fetal mouse ovaries
Although there is a rapid loss of SDSI in most XX gonadal cells by 12.0 d.p.c., interestingly, we have noticed the presence of two populations of somatic cells that sustain the SDSI in fetal ovaries at 30 ts (12.5 d.p.c.; Fig. 1B–D). The first population was located within and immediately beneath the coelomic epithelium (solid arrowheads in Fig. 1B,C), while the second population is the gonadal cells in the presumptive ovarian medullary region adjacent to the mesonephros (open arrowheads in Fig. 1B,D). Whole-mount in situ hybridization of HS-treated XX Tg explants (12.5 d.p.c.; 6-hour culture) also confirmed that the Sox9-positive signals were detected in the cells within the ovarian surface and the gonadal somatic cells surrounding the DDX4-positive germ cells near the mesonephric tissues (Fig. 1E,F).
In the coelomic epithelial and subepithelial regions, ectopic SOX9 activation was frequently found in HS-treated XX Tg explants isolated at 18 to 30 ts (with a peak between 19 and 24 ts; Fig. 1B), whereas only a few SOX9-positive cells were found within the coelomic epithelia of HS-treated ovaries at 13.0 d.p.c., and no SDSI-positive cells were seen before 18 ts (‘15ts’ in Fig. 1B) and after 13.5 d.p.c. (not shown). By whole-mount anti-SOX9/anti-ZO1 (a marker for tight junctions between coelomic epithelial cells) double-immunostaining of HS-treated ovaries at 12.5 d.p.c. (28–30 ts), the SDSI-positive cells (a single cell or two to three clustered cells) were located throughout the gonadal surface along the anteroposterior (AP) axis at the level of average cell number 16.9±1.5 per gonad (n = 8; Fig. 1G). These SDSI-positive coelomic epithelial cells appeared to be positive for SF1 (a marker for gonadal somatic cells), but all coelomic epithelial/subepithelial cells were negative for FOXL2 [an early pre-granulosa cell marker (Uda et al., 2004; Ottolenghi et al., 2005)] (supplementary material Fig. S2A).
The second population that sustained the SDSI in 12.5 d.p.c. ovaries appeared to be a subpopulation of the supporting cells that is located in the presumptive ovarian medullary region adjacent to the mesonephros (open arrowheads and arrows in Fig. 1B,D,F). In HS-treated XX Tg explants isolated at 30 ts, the SDSI-positive cells were closely connected with germ cells (Fig. 1D,F) at the bottom of the ovarian parenchyma corresponding to the region with only weak signals for anti-heparin sulfate proteoglycans (HSPG) staining (Fig. 1H). Most interestingly, most of SDSI-positive ovarian cells near the mesonephric tissue were FOXL2 positive at 12.5 d.p.c. (Fig. 1I). Moreover, some of these cells appeared to be positive for SPRR2d, another early pre-granulosa cell marker (Lee et al., 2009; Bouma et al., 2010) (supplementary material Fig. S2B). These data indicate that the SDSI-positive population in the ovarian medulla is likely to be a subpopulation of FOXL2-positive pre-granulosa cells.
Neither WNT4 nor FOXL2 is responsible for the loss of SDSI in XX gonads
The mechanisms underlying the loss of SDSI in the major population of XX supporting cells are still unclear. It was previously shown that fetal ovarian differentiation is promoted by the two major ovarian factors, FOXL2 transcription factor and WNT4 signaling molecule (Vainio et al., 1999; Uda et al., 2004; Ottolenghi et al., 2005; Ottolenghi et al., 2007; Maatouk et al., 2008; Garcia-Ortiz et al., 2009; Liu et al., 2009).
To identify a possible link of the loss of SDSI with WNT4 signaling, we crossed the inducible Sry Tg mouse line onto the Wnt4-heterozygous or -homozygous mutant background and examined the transient SOX9 expression in the HS-treated XX Tg explants in wild-type, Wnt4-heterozygous and Wnt4-homozygous backgrounds. No significant differences could be detected in the spatiotemporal patterns of the loss of SDSI between Wnt4-heterozygous and wild-type XX Tg explants (Fig. 2A,B,D). In both XX Tg explants initiated at 21 ts, the SDSI-positive gonadal cells in the anterior region were significantly lower in number than those in the posterior region (asterisks on the bars in Fig. 2D). In Wnt4-homozygous XX Tg explants, the loss of SDSI properly occurred in the anterior region at 20–21 ts (upper panels in Fig. 2C; D), albeit a 2–3 ts delay of the loss of SDSI and persistence of a small number of SOX9-positive cells at 30 ts.
The female-specific FOXL2 expression started in the centro-medullary region near the adjacent mesonephros from around 24–26 ts in developing XX gonads in vivo (supplementary material Fig. S3). The timing of the onset of the reduced SDSI in the anterior region (19–20 ts) was ∼8 hours earlier than the activation of FOXL2 in developing XX gonads in vivo. Interestingly, in HS-treated Wnt4-homozygous XX Tg explants, female-specific FOXL2 expression appears to be delayed (lower panels in Fig. 2A,C), showing no appreciable FOXL2-positive cells in these Wnt4-homozygous XX Tg gonads at 21–26 ts (i.e. a similar situation to Wnt4/Foxl2 double null XX gonads). Even in these XX Wnt4-homozygous explants that lacked detectable FOXL2 (n = 4), the loss of SDSI properly occurs in most of the gonadal cells by 26 ts (upper panels in Fig. 2C; right lower graph in Fig. 2D).
Next, we examined the potential contribution of retinoic acid (RA) signaling to the loss of SDSI in developing XX gonads (supplementary material Fig. S4). This is because, in developing ovaries, female differentiation (i.e. Stra8 expression and meiotic induction) of germ cells is initiated in an anterior-to-posterior wave-like manner at 12.5–13.5 d.p.c. (Bowles and Koopman, 2007). We isolated XX Tg genital ridge at 14–15 ts, and cultured them in the presence or absence of RA or its receptor antagonist, AGN193109 until the stage corresponding to 30 ts. However, no appreciable changes in the loss of SDSI were detected in XX Tg explants treated with either RA or AGN193109 (n = 4 for each; supplementary material Fig. S4).
We tested a possible contribution of the adjacent mesonephric tissue to the loss of SDSI in developing XX gonads (Fig. 3A–C). We isolated XX Tg genital ridge with or without adjacent mesonephros at 14–15 ts, cultured them for 12 or 30 hours until the stage corresponding to 21 or 30 ts, respectively. We finally examined the SDSI-positive patterns in these explants. Removal of the adjacent mesonephros from XX Tg gonads at 14–15 ts did not have any appreciable defects in the anterior-to-posterior loss of SDSI in developing fetal ovaries (Fig. 3B,C). Both explants showed anterior-to-posterior wave-like reduced pattern even in the absence of mesonephros, which were similar to the patterns observed in XX Tg gonads isolated at 21 ts (Fig. 3A) and 30 ts (not shown). Finally, we tested a possible contribution of the anterior gonadal region to the loss of SDSI in the middle and posterior regions of developing XX gonads (Fig. 3D,E). We separated the XX Tg genital ridges into three equal (anterior, middle or posterior) segments at 14–15 ts and cultured for 12 or 30 hours until the stage corresponding to 21 or 30 ts, respectively. Separation of three segments from the XX Tg genital ridge did not have any appreciable influences in the anterior-to-posterior loss of SDSI in developing ovarian segments (Fig. 3D,E).
All these data suggest that the intrinsic factor(s) other than WNT4, FOXL2 and RA lead(s) to the spatiotemporal loss of SDSI in developing XX gonads.
Trans-differentiation of SDSI-positive pre-granulosa cells into Sertoli-like cells
It was previously reported that a high level of SOX9 expression is maintained by FGF9 signals, resulting in the establishment of Sertoli cells in XY gonads (Kim et al., 2006). Moreover, this action of FGF9 is antagonized by female-specific WNT4 signaling in developing ovaries (Kim et al., 2006). In order to trace the potential cell fate of SOX9-expressing ovarian cells, XX Tg wild-type and Wnt4-heterozygous ovaries at 13.0 d.p.c. were treated with HS and cultured in the presence or absence of FGF9 (100 ng/ml) for 12–72 hours (Fig. 4A). As anticipated, either exogenous FGF9 or reduced WNT4 activity resulted in the maintenance of high-level SOX9 expression in a certain ovarian medulla subpopulation, albeit a small one, of pre-granulosa cells even after 72 hours of culture (8/9 explants in the presence of FGF9, 10/12 explants in Wnt4-heterozygous background; Fig. 4B). In particular, in FGF9-treated ovarian explants, some SOX9-positive cells in the ovarian medulla showed negative for FOXL2 and positive for anti-Müllerian hormone (AMH), secreted from differentiated Sertoli cells (5/6; Fig. 4C,D). These findings indicate that this ovarian medulla population of pre-granulosa cells still maintains the sexual bipotentiality to become Sertoli cells by forced SRY expression.
The SDSI-positive pre-granulosa cell population contributes to the initial round of folliculogenesis after birth
In order to map the cell fate of the SDSI-positive subpopulation of pre-granulosa cells in the ovarian medulla, we examined HS-dependent SOX9 expression patterns in the developing ovaries at the fetal and postnatal stages both in vitro (Fig. 5) and in vivo (supplementary material Fig. S5).
In developing ovaries after 12.5 d.p.c., SDSI was maintained in a restricted population of pre-granulosa cells in the presumptive ovarian medullary region adjacent to the mesonephros throughout the fetal stages (Fig. 5A). Moreover, PCNA (a cell proliferation marker)/SOX9-double immunostaining (supplementary material Fig. S6A–D,G) revealed poor proliferative activities in the SDSI-positive ovarian medulla population in fetal ovaries at 12.5 d.p.c. [PCNA-positive ratio (PCNA index) in the SDSI-positive ovarian medulla population: 14.2±6.2% (n = 4)] and 17.5 d.p.c. [5.6±3.2% (n = 6)], as opposed to a high PCNA-positive rate in the SDSI-positive coelomic epithelial population at 12.5 d.p.c. [91.7±8.3% (n = 4)]. Analyzes of BrdU incorporation before HS treatment also confirmed a similar pattern to those of PCNA index: lower BrdU-positive rates in the ovarian medulla population than those in the coelomic epithelial population at 12.5 d.p.c. (supplementary material Fig. S7). After birth, the SDSI-positive granulosa cells were positive for anti-PCNA staining [PCNA index: 64.0±7.2% at 3 days post-partum (d.p.p.; n = 4) and 60.9±11.4% at 14 d.p.p. (n = 4), respectively; supplementary material Fig. S6C,D,G], and they were found in considerable numbers in developing ovarian follicles during the initial round of folliculogenesis at 3–14 d.p.p. (Fig. 5A,B; also see Fig. 7A). This finding is consistent with the recent cell-lineage tracing data showing a large contribution of Foxl2-positive pre-granulosa cells at 12.5 d.p.c. to the first follicular formation soon after birth (Mork et al., 2012). At 3–14 d.p.p., only a small population of FOXL2-positive granulosa cells maintained the SDSI in primary and secondary follicles located in the centro-medullary region (Fig. 5A–C). During the transition from the primordial to secondary follicular stages, SDSI-positive granulosa cells did not appear to be increased in number and restricted to only a small population within the basal layer of stratified granulosa cells in the secondary follicles (also see Fig. 7C). In ovarian surface epithelia and cortex regions, no SDSI-positive granulosa cells were detected in the developing primordial and primary follicles (three lower right panels in Fig. 5B). Any SDSI-positive granulosa cells were not found in the antral follicles at 14 d.p.p. Similarly, in adult ovaries, no SDSI-positive granulosa cells were detected in both in vitro and in vivo HS treatment experiments (data not shown). Therefore, it is likely that certain pre-granulosa cells in the spatially restricted domain of the ovarian medulla continuously retain the SDSI throughout fetal stages. After birth, almost all of this population appear to contribute to the initial round of folliculogenesis, consequently leading to the complete loss of SDSI during the transition from the secondary to antral follicle stages. In addition, we examined the distribution patterns of TUNEL-positive cells by using the serial sections adjacent to the SOX9-immunostained sections (supplementary material Fig. S8), in order to examine the influence of HS treatment and 9-hour culture on the dynamics of SDSI-positive granulosa cells. We confirmed no appreciable overlapping patterns between the SDSI-positive follicles and TUNEL-positive apoptotic cell death in the XX Tg ovarian explants, in which only few TUNEL-positive signals appear to be found predominantly in the degenerating oocytes in the ovarian cortex region in contrast to the SDSI-positive granulosa cells located in the centro-medullary region.
Re-acquisition of SDSI in a partial sex-reversal model of the fetal ovaries grafted into adult male mice
It is well known that fetal ovaries grafted under the kidney capsule undergo partial sex reversal, by which a subset of granulosa cells transdifferentiate into SOX9-positive Sertoli-like cells around 2–3 weeks after transplantation (Taketo et al., 1984; Taketo and Merchant-Larios, 1986; Morais da Silva et al., 1996). First, in order to examine the contribution of the SDSI-positive granulosa cells in the ovarian medulla to this partial masculinization, fetal ovaries were isolated from wild-type and Tg embryos at 13.5 d.p.c. and transplanted under the kidney capsule of male nude mice (7 weeks old). The grafted ovaries were retrieved from the host kidney at 4–20 days after transplantation, treated with HS, and cultured for 9 hours to evaluate the SDSI (Fig. 6A; supplementary material Fig. S9). In non-HS-treated control transplants of both wild-type and Tg ovaries, some of the developing primordial follicles started to transform by day 7, leading to the partial formation of a testis-cord-like structure by day 10 post-transplantation (arrowheads in supplementary material Fig. S9). At around 12–14 days after transplantation, most of the testis cord-like structures had become positive for DMRT1, an ancient, conserved factor of the vertebrate testis differentiation pathway (Matson et al., 2011) (Fig. 6B; supplementary material Fig. S10). At 15–20 days after transplantation, SOX9-positive/FOXL2-negative Sertoli-like cells were first detected in testis-cord like structures, and they were restricted to the ovarian medullary region (Fig. 6B–D), which is confirmed by DiI labeling of fetal ovaries before transplantation (Fig. 6E).
During this process of granulosa cell transdifferentiation, we further analyzed the SDSI in grafted ovarian explants at early stages of masculinization (i.e. days 7 and 10 post-transplantation). At day 10 post-transplantation, no SOX9-positive signals were detected in any non-treated XX Tg (n = 5; ‘-HS’ in Fig. 6F) or HS-treated wild-type explants (n = 11; figure not shown). In HS-treated Tg ovarian transplants at 7–10 days after transplantation, SDSI was maintained in a subpopulation of granulosa cells, which was restricted to the presumptive ovarian medullary area (large open arrows in Fig. 6F). Surprisingly, we found that, at days 7 and 10 post-transplantation, a considerable number of ovarian granulosa cells in not only the ovarian medullary region but also throughout the ovarian parenchyma gradually re-acquired the SDSI (arrowheads in Fig. 6F). Anti-PCNA/SOX9 double immunostaining revealed high proliferative activities in these SDSI-positive granulosa cells of the follicular structures [PCNA index: 91.0±2.2% in primary and secondary follicular structures (n = 4); supplementary material Fig. S6E,F]. Moreover, these SDSI-positive cells are FOXL2-positive/DMRT1-negative granulosa cells, which are located in both follicular and testis-cord-like structures (‘DMRT1’ in Fig. 6H). Quantitative real-time RT-PCR analysis also confirmed the re-acquisition of SDSI in the XX Tg explants at days 7 and 10 post-transplantation (*P<0.05, **P<0.01; Fig. 6I).
Moreover, we calculated both the relative numbers of the SDSI-positive follicle per total number of all follicles and the relative numbers of the SDSI-positive granulosa cell per total granulosa cells (in each SDSI-positive follicle) in the grafted ovaries at day 10 post-transplantation, and then compared them with those in the postnatal ovaries at P3∼21 (Fig. 7). In the grafted ovaries at day 10 post-transplantation, the relative number of the SDSI-positive follicle (gray bar in Fig. 7B) showed 52.8±10.3% [38.4±6.3% in primary follicles plus 14.6±6.3% in secondary follicles (orange bar and green bar in Fig. 7B, respectively)]. Each value in the grafted ovaries (Fig. 7B) showed significantly higher than each corresponding value in the postnatal ovaries at any stages of P3∼P21 (Fig. 7A). Moreover, in both primary and secondary follicle stages, the relative number of SDSI-positive granulosa cells per follicle in the grafted ovaries (Fig. 7D) was significantly higher than each corresponding value in the postnatal ovaries at any stages of P3∼P21 (Fig. 7C). These findings show that most of the SDSI-negative pre-granulosa cells in 13.5 d.p.c. ovaries re-acquire the SDSI within 10 days after transplantation into the adult male mice. All these data also suggest that the SDSI can be used as a functional marker for one of the earliest masculinization signs of granulosa cells.
By monitoring the SRY-dependent SOX9 inducibility (SDSI), this study is the first to visualize dynamic changes in sexual bipotentiality in developing XX gonads, showing an unexpected wide range of bipotential states of granulosa cell lineage, from the precursor state in the coelomic epithelium/bipotential state of gonadal supporting cells in early XX gonads to the high level of sexual plasticity in granulosa cells of postnatal ovaries. First, the present study demonstrated that, in the majority of XX supporting cells, the loss of SDSI occurs in an anterior-to-posterior wave-like pattern from 18 to 24 ts (around 11.5∼12.0 d.p.c.). Moreover, this loss was reversible in most of these cells of the grafted ovaries, showing re-acquisition of the SDSI before the onset of a partial sex reversal (Fig. 6F–I). FOXL2 is well known to be crucial for the maintenance of ovarian phenotype in the granulosa cells at the adult stage (Uhlenhaut et al., 2009). However, it was shown that either loss or re-acquisition of the SDSI in these cells occurs independently of FOXL2 expression in developing ovaries (Fig. 1I; Fig. 5C; Fig. 6G,H). These findings suggest that the Hsp-SRY transgene can induce transient SOX9 expression in some of FOXL2-positive granulosa cells in vitro, although this SOX9 expression cannot be maintained in these cells after the disappearance of Hsp-SRY expression (i.e. transient co-existence of SOX9 and FOXL2 for only 3–4 hours). In Wnt4-mutant embryos, it was shown that reduced Wnt4 activity does not alter any reduced patterns of SDSI in developing XX gonads (Fig. 2). Moreover, the present data demonstrated that proper reduction of the SDSI occurred even in the Wnt4-homozygous XX gonads at 21–26 ts (before the onset of FOXL2 expression) where immunoreactive FOXL2 activity was almost abolished, similar to the situation seen in XX gonads of the Wnt4/Foxl2-double-null embryos (Ottolenghi et al., 2007) (Fig. 2C). Taken together, these data suggest that neither WNT4 nor FOXL2 is responsible for the loss of SDSI in the majority of XX supporting cells. It was also shown that exogenous addition of either RA or its receptor antagonist had no appreciable influence on the loss of SDSI in developing XX gonads in vitro (supplementary material Fig. S4). Moreover, the loss of SDSI properly occurs in the culture of the XX genital ridges without either mesonephric tissue or anterior gonadal region that is the potential major source of retinoic acid (Bowles et al., 2006) and WNT4 (Mizusaki et al., 2003). Therefore, these data raise the possibility of the existence of certain intrinsic factor(s) which spatiotemporally regulate(s) the SDSI in supporting cell lineage along the AP axis of developing XX gonads.
In contrast to a rapid loss of SDSI in most XX supporting cells, interestingly, here we have discovered two populations of supporting cell lineage that sustain the SOX9 inducibility in fetal ovaries at 12.5 d.p.c. One is a subpopulation of the coelomic epithelial and subepithelial cells that cover the surface area of ovarian parenchyma from 11.5 to 12.5 d.p.c. Since testis-specific recruitment of Sertoli cell precursor occurs in the proliferating coelomic epithelial cells from 14 to 18 ts (Karl and Capel, 1998; Schmahl et al., 2000, Colvin et al., 2001; Schmahl and Capel, 2003), this population in the ovarian coelomic epithelium may be homologous to the reserved pool of a part of supporting cell precursors, which are fated to become SRY-positive pre-Sertoli cells in developing testis. Interestingly, it was recently shown that the coelomic epithelial cells continuously contribute to the pre-granulosa cell population in ovarian parenchyma even after birth (Mork et al., 2012). Therefore, these data raise the possibility that this SDSI-positive coelomic epithelial population at 12.5 d.p.c. may contribute to the pre-granulosa cells of fetal and postnatal ovaries at later stages. However, the present study showed no detectable SDSI-positive signals within and beneath the coelomic epithelia after 13.5 d.p.c. (Fig. 5A). In the ovarian cortex region, the presumptive pre-granulosa cells (i.e. the somatic cells near germ cells) and the primordial follicles near the ovarian surface (coelomic) epithelia showed no SDSI-positive signals at the late fetal and perinatal stages (lower right panels in Fig. 5B). Since the primordial follicle pool was shown to be newly recruited from the coelomic epithelia at the perinatal period (Mork et al., 2012), this finding suggests that, at the perinatal stage, pre-granulosa cell recruitment from the coelomic epithelia and its subsequent differentiation may not require the SDSI-positive (sexually bipotential) state during the primordial follicle formation in the ovarian cortex region.
Most interestingly, another SDSI-positive population is identical in a subpopulation of differentiated granulosa cells that are restrictedly localized in the ovarian medullary region adjacent to the mesonephros. The present data revealed that this ovarian medulla population is mitotically silent throughout the fetal life, and then soon after birth, contributes to the first wave of folliculogenesis at prepubertal stages. These data provide direct evidence showing the unexpected heterogeneity in the sexual bipotentiality/plasticity of granulosa cells in the first follicles of mouse postnatal ovaries. It was previously shown that the first follicles in the ovarian medullary region have several different characteristics from the resting primordial follicle pool in the cortex region; such as their tendency to undergo atresia at later stages without the stimulation from pituitary-derived FSH, and their temporal differences in pre-granulosa cell recruitment from the coelomic epithelia, etc. (Mork et al., 2012). Moreover, the co-existence of some granulosa cells with sexual bipotentiality in the first follicles is consistent with the ovotestis-like phenotype in various partial XX sex reversal models such as the Esr1/Esr2 (estrogen receptor α/β) (Couse et al., 1999; Dupont et al., 2003), Foxl2/Wnt4 (Schmidt et al., 2004; Uda et al., 2004; Ottolenghi et al., 2007) and Rspo1-null ovaries (Chassot et al., 2008). For example, in ovaries of Foxl2/Wnt4-double-null mice, it was reported that the testis-cord-like tubules in the medullary region harbored well-differentiated spermatogonia, in contrast to the ovarian structures including oocytes in the cortex region (Ottolenghi et al., 2007). In Esr1/Esr2-double-null prepubertal ovaries, it was also shown that the initial transdifferentiation into the seminiferous-like tubules with SOX9-positive Sertoli-like cells appears to be found in the first wave of folliculogenesis in the centro-medullary region (Dupont et al., 2003). The present data using a partial sex reversal model of the grafted ovaries also demonstrated that, in the wild-type ovarian transplants, spontaneous transdifferentiation of pre-granulosa cells into SOX9-positive Sertoli-like cells is repeatedly found in ovarian medullary region (Fig. 6C–E). Taken together, these data therefore raise an interesting possibility that such SDSI-positive pre-granulosa cells in the medullary region cause or contribute to masculinization in various cases of XX disorders of sexual development in mammals.
Finally, what is the regulator(s) for the loss and re-acquisition of the SDSI in gonadal supporting cell linage? One strong candidate is the ovarian factor(s) that can repress the SRY action in SOX9 activation in a FOXL2/WNT4/RA-independent manner. Garcia-Ortiz et al. identified several candidate genes that potentially act in developing ovaries independent of FOXL2 and WNT4 [e.g. Irx3 (Jorgensen and Gao, 2005; Garcia-Ortiz et al., 2009; Kim et al., 2011), Lzts1 and Zbtb3c (Garcia-Ortiz et al., 2009)]. Moreover, Nr0b1/Dax1 was previously shown to act as anti-SRY gene that can repress SOX9 expression (Swain et al., 1998; Ludbrook et al., 2012). Since the expression levels of these ovarian genes (i.e. Irx3, Lzts1, Zbtb3c and Nr0b1/DAX1) were downregulated in the ovarian grafts within 10 days after transplantation into the male mice (K.H. and Y.K., unpublished observation), they may be the candidate repressors for SDSI in developing XX supporting cells. On the other hand, we cannot exclude a possible involvement of SF1/Nr5a1-like co-factor(s) which synergistically act(s) with SRY to initiate SOX9 expression in supporting cell lineage (Sekido and Lovell-Badge, 2008). Interestingly, spatiotemporal dynamics of the SRY-expressing supporting cells in XY gonads (Bullejos and Koopman, 2001; Kidokoro et al., 2005) roughly coincides with that of the SDSI-positive supporting cells in XX gonads, except for the two populations that retain the SDSI beyond 12.5 d.p.c. (Fig. 1). These data give rise a possibility that the loss and re-acquisition of SDSI in XX supporting cells are caused by some SRY co-factors that are transiently (i.e. 11.0∼12.0 d.p.c.) expressed in supporting cell lineage of both sexes. The identification of these positive/negative regulators for SDSI and their characterization must be required to resolve the molecular mechanisms of the initial differentiation of pre-granulosa cells at around 11.5–12.0 d.p.c. and the biological significance of the SDSI-positive (sexually bipotential) state in some granulosa cells of the first follicles soon after birth.
Materials and Methods
All animal experiments in this study were carried out in strict accordance with the Guidelines for Animal Use and Experimentation as set out by the University of Tokyo. The procedures were approved by the Institutional Animal Care and Use Committee of the Graduate School of Agricultural and Life Sciences in the University of Tokyo (approval ID: P11-502). The inducible Sry transgenic line #44 [with the HSP-Sry (Hsp70.3 promoter-driven murine Sry) transgene; ICR/C57BL6[B6]-mixed background (Hiramatsu et al., 2009)] and Wnt4 mutant line [129svJ/ICR-mixed background (Vainio et al., 1999)] were used in this study (Kidokoro et al., 2005; Mizusaki et al., 2003). Male nude mice (8 weeks old; BALB/c, nu/nu, SLC Japan) were used as host mice for the transplantation of fetal ovaries.
Transplantation of fetal mouse ovaries
Transplantation of fetal ovaries was carried out by the previously reported procedure (Taketo et al., 1984; Taketo and Merchant-Larios, 1986). In brief, fetal ovaries (without mesonephroi) were isolated from XX Tg and wild-type embryos at 13.5 d.p.c. and transplanted beneath the kidney capsule of male nude mice. In some experiments, the gonad/mesonephros border of the 13.5 d.p.c. ovaries was labeled with DiI (0.84 µg/µl, Invitrogen) using a micropipette before transplantation. Transplants were dissected from host animals on days 4∼20 post-transplantation.
Heat shock treatment in vitro and in vivo
For in vitro heat shock (HS) treatment, embryos were collected from pregnant female mice at various stages (11.0–18.5 d.p.c.). From 10.5 to 12.5 d.p.c., the tail somites of each embryo were counted for accurate staging. Using tail somite stages, 11.0 d.p.c. corresponds to ∼12 ts, 11.5 d.p.c. to 18 ts, 12.0 d.p.c. to 24 ts, and 12.5 d.p.c. to 30 ts. Genital ridges and ovaries were isolated from the embryos, pups (1, 3, 7, 14 and 21 d.p.p.) and adult females (7 weeks old) in cold Dulbecco's modified Eagle's medium (DMEM; Sigma). One genital ridge of each pair was subjected to HS treatment (43°C for 10 minutes) in a 0.2 ml thin-wall PCR tube as described previously (Hiramatsu et al., 2009). In most cases, the other genital ridge was used as a non-HS control. In some experiments of adult ovaries, the ovaries were fragmented into small pieces before HS treatment. Fetal ovary transplants and fragments of adult ovaries were HS treated under the same conditions (43°C for 10 minutes). All samples were cultured for 9 hours before SOX9 expression was detected immunohistochemically.
For in vivo HS treatment, the postnatal and adult ovaries were surgically separated from the mesovarium under anesthesia. Then, the left ovaries with oviducts and uterine horns were gently extracted from the abdominal cavity and immersed in pre-warmed DMEM (43°C) in a 0.5 ml tube for 15 minutes (right ovaries were used as a non-treated control).
All ovarian samples were cultured on ISOPORE membrane filters (Millipore) in DMEM containing 10% horse serum at 37°C for appropriate periods (2 hours to 3 days). Some genital ridges were cultured in 10% horse-serum–DMEM supplemented with FGF9 (Sigma; 100 ng/ml) (Hiramatsu et al., 2010), all-trans retinoic acid (Sigma; 2 µM) or retinoic acid receptor antagonist, AGN193109 (Toronto Research Chemicals. Inc.; 7.5 µM) (Bowles et al., 2006). The segment culture assay using anterior, middle and posterior segments of the genital ridge (Hiramatsu et al., 2003) and the gonad culture without adjacent mesonephros (Matoba et al., 2005; Matoba et al., 2008) were also initiated from the XX Tg gonads at 14–15 ts. In addition, under the present culture conditions of XX and XY genital ridges initiated at 11.5 d.p.c., the gonadal explants at the 1st, 3rd and 5th day after culture roughly correspond to the testes/ovaries at 12.0–12.5, 13.5–14.5 and 15.5 d.p.c., respectively (Byskov, 1978; Taketo and Koide, 1981; Kanai et al., 1991; Mizukami et al., 2008; Hiramatsu et al., 2003; Hiramatsu et al., 2009; Hiramatsu et al., 2010; Wu et al., 2013).
Histology and Immunohistochemistry
The samples were fixed in 4% PFA–PBS at 4°C for 4 hours, dehydrated, and then embedded in paraffin. Serial sagittal sections (∼5 µm in thickness) were cut in the middle portion including the largest area and then every 10th section of the samples was principally analyzed by anti-SOX9 immunostaining, while adjacent section was used for immunostaining for other markers as described below. The sections were incubated with anti-AMH (1∶200 dilution; Santa Cruz), anti-laminin (1∶400 dilution; ICN Pharmaceuticals), anti-BrdU (1∶100 dilution; Dako Cytomation), anti-DMRT1 (1∶100 dilution; Santa Cruz), anti-FOXL2 [1∶200 dilution (Uda et al., 2004); 1∶200 dilution (goat antibodies for double-staining with SOX9), Abcam], anti-heparan sulfate (1∶200 dilution; 10E4 epitope; Seikagaku Corporation, Japan), anti-MVH/DDX4 (1∶5,000 dilution) (Toyooka et al., 2000), anti-PCNA (1∶1,000 dilution; PC10; DAKO), anti-SF1/Ad4Bp [1∶1,000 (Ikeda et al., 2001)], anti-SOX9 [1∶250 dilution (Kidokoro et al., 2005; Kent et al., 1996; Polanco et al., 2010)], anti-SPRR2d (1∶300 dilution; Alexis Biochemicals), or anti-SRY [1∶10 dilution (Wilhelm et al., 2005)] at 4°C for 12 hours. The reaction was visualized with biotin-conjugated secondary antibody in combination with Elite ABC kit (Vector Laboratories) or by Alexa-Fluor-488/594-conjugated secondary antibodies (Invitrogen).
For whole-mount immunohistochemistry, the PFA-fixed samples were treated with 0.5% Triton X-100 for 15 minutes at 4°C, and then incubated with rabbit anti-SOX9 and mouse anti-ZO1 (1∶400 dilution; Invitrogen) antibodies at 4°C for 12 hours. The reaction was visualized with Alexa-Fluor-488/594-conjugated secondary antibodies. After the mesonephric tissue was removed, they were mounted mesonephric side down (coelomic epithelium side up) using an antiphotobleaching medium. Finally, we calculated the total number of the SOX9/ZO1-double-positive cells located within the coelomic epithelium (total 8 explants).
For quantitative analysis of SDSI-positive cells, we calculated the number of SOX9-positive cells located in regions I–V (see right lower panel in Fig. 2D) of three to five longitudinal sagittal sections per explant (the section containing the largest area in the middle position of the gonad, and two sections before and behind it at an interval of ∼50 µm). The area of each region was also measured using the ImageJ program (Ver. 1.44). Finally, the cell number per area (mm2) was separately estimated in each region of XX gonadal explants.
For the number of SDSI-positive cells in postnatal ovaries and grafted ovarian transplants, we separately counted the numbers of each primary, primordial or secondary follicle in three to five longitudinal sagittal sections per ovary. We also counted the numbers of SOX9-positive and -negative granulosa cells per each follicle. After counting the numbers in each section, the total number of oocytes was counted in each section image. Finally, relative numbers of SOX9-positive follicle relative to total follicle number (i.e. total oocytes number) and relative SOX9-positive granulosa cells in each follicle were estimated in postnatal ovaries at P3∼P21 and in grafted ovaries at day 10 post transplant.
For the cell proliferation assays, the PCNA- and BrdU-positive indices of the SDSI-positive/-negative cell population were estimated in HS-treated XX Tg ovaries. The BrdU solution (BD Biosciences; 0.1 mg/g body weight) was intraperitoneally injected into the pregnant females and pups at 2 hours prior to the tissue sampling and subsequent in vitro HS treatment. The PCNA-/BrdU-positive cells were calculated separately in the SOX9-positive (coelomic epithelial/gonadal somatic) and the SOX9-negative (that are located near the SOX9-positive cells) cell population using three sagittal sections per explant (PCNA staining: n = 4, 6, 4 and 4 ovaries at 12.5 d.p.c., 17.5 d.p.c., 3 d.p.p., and 14 d.p.p., respectively; BrdU staining: n = 8 and 6 ovaries at 12.5 d.p.c. and 3 d.p.p., respectively). TUNEL assays were also performed by an Apoptotic Detection Kit (TaKaRa, Japan).
Whole-mount in situ hybridization
Whole-mount in situ hybridization was performed using 4% PFA-fixed explants (n = 6) as described previously (Hiramatsu et al., 2003).
Total RNA was reverse transcribed using random primer with a Superscript-III cDNA synthesis kit (Invitrogen). Specific primers and fluorogenic probes for Sox9 (Mm00442795_m1) and Gapdh (Taqman control reagents) were purchased from Applied Biosystems. PCR was performed using an Applied Biosystems Step One Real-Time PCR System. The expression levels represented the relative expression levels of each marker gene per Gapdh amplicon ratio (mean ± standard error).
Quantitative data (i.e. relative number of SOX9-positive cells, real-time PCR data, and PCNA/BrdU-positive index) were analyzed by Student's t-test. One-way analysis of variance (ANOVA) was also conducted to analyze significance of the regional difference of the SOX9-positive cell number among wild-type, Wnt4-heterozygous and Wnt4-homozygous XX Tg gonads.
The authors thank Dr Dagmar Wilhelm for her comments on and critical reading of the manuscript, Drs Hitomi Suzuki and Miyuri Kawasumi for their technical advice, and Ms Itsuko Yagihashi and Taeko Nagano for their secretarial assistance. The authors also thank Drs Dagmar Wilhelm, Peter Koopman, David Schlessinger and Toshiaki Noce for kindly providing the antibodies. K.H. is DC1 JSPS Research Fellow.
K.H. carried out most of the experiments; K.M., M.S., S.M., R.H. and N.T. partially contributed to the experiments; M.K-A., M.K. and K-i.M. evaluated the experiments and the final manuscript; K.H. and Y.K. designed the project and wrote the manuscript.
This work was supported by Grants-in-Aid for Scientific Research (KAKENHI) [grant number 23132504, 24228005, and 25132703 to Y.K.; 10J01100 to K.H.] from the Ministry of Education, Culture, Sports, Science and Technology (MEXT) and the Japan Society for the Promotion of Science (JSPS).