The conditional deletion of steroidogenic factor 1 (Nr5a1) in Sox9-Cre mice compromises testis differentiation

Steroidogenic factor 1 (NR5A1) is essential for gonadal development. To study the importance of NR5A1 during early gonadal sex differentiation, we generated Sox9-Cre-Nr5a1 conditional knockout (cKO) mice: Sox9-Cre;Nr5a1flox/flox and Sox9-Cre;Nr5a1flox/− mice. Double-immunostaining for NR5A1 and AMH revealed silenced NR5A1 in Sertoli cells and reduced AMH+ cells in the gonads of XY Sox9-Cre-Nr5a1 cKO mice between embryonic days 12.5 (E12.5) and E14.5. Double-immunostaining for SOX9 and FOXL2 further indicated an early block in Sertoli cells and ectopic granulosa cell differentiation. The number of cells expressing the Leydig cell marker 3βHSD obviously reduced in the gonads of XY Sox9-Cre;Nr5a1flox/− but not Sox9-Cre;Nr5a1flox/flox mice at E15.5. The presence of STRA8+ cells indicated that germ cells entered meiosis in the gonads of XY Sox9-Cre-Nr5a1 cKO mice. The results of qRT-PCR revealed remarkably reduced and elevated levels of testis and ovary markers, respectively, in the gonads of XY Sox9-Cre-Nr5a1 cKO mice at E12.5‒E13.5. These data suggested that the loss of Nr5a1 abrogates the testicular pathway and induces the ectopic ovarian pathway, resulting in postnatal partial/complete male-to-female gonadal sex reversal. Our findings provide evidence for the critical role of NR5A1 in murine gonadal sex determination in vivo.

www.nature.com/scientificreports/ an important role in Sertoli cell maturation and spermatogenesis during postnatal testis development 16 . Anti-Müllerian hormone (Amh, also known as Müllerian inhibiting substance, Mis)-Cre-mediated Nr5a1 cKO mice have also been generated 17 . The gonads of XY Amh-Cre-Nr5a1 cKO mice are strikingly hypoplastic, and the loss of Sertoli and germ cells along with dysgenic testis cords from E15.5 suggested a role for NR5A1 in Sertoli cell survival and proliferation during embryonic testis differentiation after sex determination. However, despite Nr5a1 upregulation in the testis as early as E12.5, Nr5a1 was inactivated at E14.5 in XY Amh-Cre-Nr5a1 cKO mouse gonads. Thus, the role of NR5A1 in testis differentiation at E12.5-E14.5 has remained unclear.
To study the role of NR5A1 during early testis differentiation, we generated a different Nr5a1 cKO using the Sox9-Cre transgene, which is active in the gonads, osteo-chondrogenic tissues, intestine, spinal cord, and pancreas, during mouse embryogenesis 18 . Because the expression of Sox9 is specifically elevated in Sertolilineage cells from the beginning of testis differentiation 4,5,10 , Sox9-Cre should induce a deletion of the Nr5a1 allele before Amh-Cre, in which the Cre recombinase is expressed at E14.5 17 . Here, we generated genotypically different Sox9-Cre;Nr5a1 flox/flox cKO (Sox9-Cre;Nr5a1f. /f ) and Sox9-Cre;Nr5a1 flox/− cKO (Sox9-Cre;Nr5a1f. /− ) mice that respectively exhibited postnatal partial and complete XY sex reversal. We analysed the expression of cell type-specific gonadal markers between E12.5 and E15.5 to determine the onset and progression of sex reversal in XY Sox9-Cre-Nr5a1 cKO mice.
Distinct postnatal phenotypes of genetically different Sox9-Cre-Nr5a1 cKO mice. Gross observation of XY Sox9-Cre;Nr5a1f. /f mice revealed ambiguous external genitalia and two internal reproductive tracts with very small gonads, located at the level of the bladder two weeks after birth ( Supplementary Fig. 3). On the other hand, external genitalia were indistinguishable between XY Sox9-Cre;Nr5a1f. /− and XX control female mice ( Supplementary Fig. 4). An internal uterus was identified, but gross observation did not reveal any gonads. We then histologically analysed the reproductive tissues. Two reproductive tracts were defined as a uterus and vas deferens in XY Sox9-Cre;Nr5a1f. /f mice (Fig. 1b,c), and their gonads were markedly smaller than those of XY control mice (Fig. 1a). The gonads (Fig. 1b) comprise a large interstitial area comprising connective tissue and a few irregularly shaped tubule-like structures (asterisks in Fig. 1b′) that were similar to seminiferous tubules ( Fig. 1a′) in size and enclosed Sertoli-like cells but not male gametes. Other sections of the same gonad (Fig. 1c) contained a few small vesicular structures (arrows in Fig. 1c′). We also used IHC to detect SOX9 and Forkhead box protein L2 (FOXL2), markers of Sertoli cells (Fig. 1f) and granulosa cells (Fig. 1j) in the testes and the ovaries, respectively. Tubule-like and vesicular structures contained SOX9 + (Fig. 1g) and FOXL2 + cells (Fig. 1h), respectively, indicating that the gonads of XY Sox9-Cre;Nr5a1f. /f mice were "ovotestes", with a testicular portion comprising seminiferous-like tubules and an ovarian portion with follicle-like structures, and both portions were atypical. We examined six gonads from XY Sox9-Cre;Nr5a1f. /f mice and did not identify any apparent variability in histological and IHC results among them. Histological staining showed that gonads of XY Sox9-Cre;Nr5a1f. /− mice were < 1 mm in length and were located below the kidneys, where ovaries are usually positioned (Fig. 1d,e). All 10 histologically stained gonads from XY Sox9-Cre;Nr5a1f. /− mice were ovaries with primary and secondary follicles (Fig. 1d′,e′) comprising FOXL2 + granulosa cells (Fig. 1i), similar to those in XX control ovaries (Fig. 1j).

Differential perturbation of Leydig cells in XY Sox9-Cre-Nr5a1 cKO gonads.
We assessed the expression of 3 beta-hydroxysteroid dehydrogenase (3βHSD) using IHC to determine Leydig cell differentiation (Fig. 4). The expression of 3βHSD was detected in the interstitial region of XY control gonads at E13.5-E15.5 ( Fig. 4a-c). 3βHSD + cells were fewer in number in the gonads from the two XY Sox9-Cre-Nr5a1 cKO models and from XY Nr5a1 +/− mice compared with those in the gonads of XY control animals at E13.5 ( Fig. 4d,g,j,m). The number of 3βHSD + cells in gonads from XY Sox9-Cre;Nr5a1f. /f and Nr5a1 +/− animals returned to levels that  www.nature.com/scientificreports/ Ala-Asp] box polypeptide 4 (DDX4, also known as VASA), a marker of germ cells (Fig. 5). DDX4 + cells were enclosed within the testis cord in XY control gonads (Fig. 5a-c) and were evenly distributed within the gonads of XX controls (Fig. 5j-l). The distribution of DDX4 + cells in XY Sox9-Cre-Nr5a1 cKO gonads appeared irregular ( Fig. 5d-i). To determine germ cell differentiation, we examined the expression of stimulated by retinoic acid gene 8 (STRA8), a marker of meiotic germ cells, also via IHC (Fig. 6). We found STRA8 + cells in XX (Fig. 6g,h) but not in XY (Fig. 6a,b) control gonads at E14.5 and E15.5. On the other hand, STRA8 + cells were detected in www.nature.com/scientificreports/ XY Sox9-Cre-Nr5a1 cKO gonads (Fig. 6c-f), but appeared less abundant than those in XX controls and tended to localise to the pole and dorsal sides.

Impairment of testis cord organisation in XY
Sox9-Cre-Nr5a1 cKO gonads. Foetal testes are morphologically characterised by the formation of testis cords that enclose Sertoli cells. Because the basement membrane of testis cords can be outlined using laminin immunostaining, we investigated testis cord formation by double staining for laminin and AMH (Fig. 7). Normal testicular cords enclosing AMH + cells were detected in XY (Fig. 7a-c) but not in XX (Fig. 7j-l) control gonads at E12.5-E14.5, as expected. The testis cords in XY Sox9-Cre;Nr5a1f. /f gonads were morphologically irregular and narrower than those in the XY controls at E12.5, and were only partially detected at E14.5 (Fig. 7d-f). The cords in XY Sox9-Cre;Nr5a1f. /− mice appeared more disorganised than those in XY Sox9-Cre;Nr5a1f. /f gonads at E12.5 and E13.5, and the structure outlined by laminin resembled that of control XX ovaries at E14.5 (Fig. 7g-i). The testis cords in XY Nr5a1 +/− mice also appeared irregular at E12.5 and E13.5, but were indistinguishable from those of XY controls at E14.5 (Fig. 7m-o).

Discussion
XY Amh-Cre-Nr5a1 cKO mice lacking Nr5a1 in Sertoli cells at E14.5 showed a reduction of Sertoli cell numbers and a loss of proliferating Sertoli cells at E15.5-E18.5, suggesting a role for NR5A1 in the regulation of Sertoli cell proliferation 17 . We used the Sox9-Cre transgene to investigate the importance of NR5A1 in the early stages of testes differentiation in vivo, because Sox9 is specifically elevated in Sertoli cells from the beginning of testis differentiation. The phenotypes of postnatal XY Sox9-Cre-Nr5a1 cKO gonads with partial and complete maleto-female sex reversal were obviously more severe than those of Amh-Cre-Nr5a1 cKO mice. The results of IHC for AMH and SOX9 indicated that both XY Sox9-Cre-Nr5a1 cKO models had conspicuously less Sertoli cells at E12.5-E14.5. The spatial and temporal IHC findings of gonadal protein markers during the early phase of testis morphogenesis revealed novel and important information about the cellular consequences of Nr5a1 loss. Sox9-Cre should have inactivated Nr5a1 along with elevated Sox9 expression in Sertoli cells. However, undifferentiated somatic cells in indifferent gonads of both sexes express low levels of Sox9 at E11.5-E12.5 10 .   The results of IHC double staining to detect SOX9 and FOXL2 showed that the decrease in the numbers of SOX9 + Sertoli cells was accompanied by a gradual increase in that of FOXL2 + cells in both XY Sox9-Cre-Nr5a1 cKO models at E13.5-E14.5. Sertoli cell differentiation spatially spreads in a centre-to-pole pattern in XY gonads 19 , while FOXL2 expression begins at the anterior-dorsal side in XX gonads 19 . We detected SOX9 + and FOXL2 + cells in the central and pole-to-dorsal regions, respectively, in XY Sox9-Cre-Nr5a1 cKO gonads, suggesting a blockage of Sertoli cell differentiation and the initiation of granulosa cell differentiation within the gonads. Testis cords were concomitantly irregular in XY Sox9-Cre-Nr5a1 cKO gonads from E12.5. Since the formation and maintenance of testis cords require Sertoli cells 1,20 , the different severity of cord disruption between the two Sox9-Cre-Nr5a1 cKO genotypes probably reflects differences in the sizes of Sertoli cell populations.
Although most AMH + cells in XY control gonads were positive for NR5A1 (NR5A1 + /AMH + ), the gonads of the two XY Sox9-Cre-Nr5a1 cKO models contained NR5A1 − /AMH + cells, indicating that AMH + cells can differentiate in the absence of NR5A1. Despite disrupted NR5A1 binding to the Amh gene promoter, Amh transcription initiated in foetal testes produced Amh mRNA at levels sufficient for the induction of Müllerian duct regression, suggesting that NR5A1 is a quantitative regulator of Amh transcripts 21 . However, mutations of the SOX9-binding site in that study prevented Amh transcription, suggesting that SOX9 acts as an essential initiator of Amh transcription 21 . Furthermore, although Gata-binding protein 4 (GATA4) is also considered as an essential regulator of Amh transcription, Amh can be induced in Sertoli cells without GATA4 binding to the Amh promoter, thus leading to Müllerian duct regression 22 . Therefore, appropriate transcription of the Amh gene requires the cooperative action of several transcription factors, including NR5A1, SOX9, GATA4, Wilms tumour suppressor gene 1, and NR0B1 (also called Dax1) during testis development. Although the disruption of any one of these factors might alter interactions among them, some Amh might still be generated.
The development of the reproductive organs of both sexes is dependent on hormones secreted from the testis during a critical period of embryonic development 23,24 . In males, AMH induces the regression of the Müllerian ducts, and testosterone acts on the Wolffian ducts that develop into the epididymides, vas deferens, and seminal vesicles. In females that do not have AMH, the Müllerian ducts develop into the oviducts, uterus, and upper vagina, and the Wolffian ducts regress. The numbers of AMH + cells and levels of Amh were obviously reduced at E12.5-E14.5 in both XY Sox9-Cre-Nr5a1 cKO models in the present study. Müllerian ducts had histologically ambiguous epithelial mesenchymal borders with TUNEL-positive apoptotic cells in XY control mice, whereas those in both XY Sox9-Cre-Nr5a1 cKO models retained clear tubular structures without apoptotic cells, similar to those in XX control mice at E14.5 ( Supplementary Fig. 6). These results indicate that insufficient AMH/Amh during the critical period induced persistent Müllerian ducts and the subsequent development of the uterus in XY Sox9-Cre-Nr5a1 cKO mice.
The vas deferens and seminal vesicles were present in XY Sox9-Cre;Nr5a1f. /f mice but not in XY Sox9-Cre;Nr5a1f. /− mice. We did not measure testosterone levels, but they were probably the main determinant of Wolffian duct development in XY Sox9-Cre;Nr5a1f. /f and XY Sox9-Cre;Nr5a1f. /− gonads. Since foetal Leydig cells lack 17 beta-hydroxysteroid dehydrogenase (17βHSD), which converts androstenedione to testosterone during steroidogenesis, they co-operate with Sertoli cells that possess 17βHSD to produce testosterone 25 . The different testosterone levels might be due to the different numbers of Sertoli cells between the two cKO models.
Foetal Leydig cells also secrete insulin-like peptide 3 (INSL3), which influences testis descent 26 . XY Sox9-Cre;Nr5a1f. /f gonads were located at the level of the bladder within the abdominal cavity, indicating insufficient INSL3. www.nature.com/scientificreports/ Significantly fewer cells expressed the Leydig cell marker 3βHSD in the cKO models than in XY controls at E13.5, and qRT-PCR findings showed significantly lower expression of the Leydig cell marker Cyp11a1, in XY Sox9-Cre-Nr5a1 cKO, than in XY control gonads at E12.5-E13.5. These results indicated that Leydig cell differentiation is also affected. As described above, Nr5a1 could become ablated in cells expressing Nr5a1 including Leydig-lineage cells at indifferent stages. Since foetal Leydig cell differentiation is regulated by paracrine factors secreted from Sertoli cells, including AMH and DHH 27 , foetal Leydig cell differentiation might be indirectly affected by the reduced numbers of Sertoli cells. However, subsequent Leydig cell fates greatly differed between gonads of the two XY Sox9-Cre-Nr5a1 cKO models. The numbers of 3βHSD + cells in XY Sox9-Cre;Nr5a1f. /f recovered to levels equivalent to those in XY controls by E14.5, whereas those in XY Sox9-Cre;Nr5a1f. /− became undetectable by E15.5. Further studies are needed to understand the mechanistic differences underlying these phenotypes.
The presence of DDX4 + cells and normal Ddx4 levels in both XY Sox9-Cre-Nr5a1 cKO models suggested that germ cells entered the gonads normally; therefore, their irregular distribution is probably due to testis cord disruption. The presence of STRA8 + cells in the gonads of both XY Sox9-Cre-Nr5a1 cKO models indicates meiotic entry of germ cells. The expression of STRA8 in females is notably induced by retinoic acid (RA), which is synthesised in the mesonephros, whereas germ cells in males are protected from RA exposure by cytochrome P450 family 26 subfamily B member 1 (CYP26B1), an RA metabolising enzyme, which is produced by Sertoli cells and regulated by NR5A1 and SOX9 28,29 . The meiotic entry confirmed herein might have been caused by insufficient CYP26B1 expression due to the reduced numbers of Sertoli cells. Since the meiotic entry of germ cells spatially proceeds from the anteroposterior regions to the dorsoventral regions of the gonads along the diffusion flow of RA from the mesonephros into the gonad 19 , localisation of STRA8 + cells to the pole-to-dorsal regions indicated that male-to-female sex reversal occurred in somatic and germ cells at these regions in XY Sox9-Cre-Nr5a1 cKO gonads.
In summary, our results indicated that both male and female programmes in XY Sox9-Cre;Nr5a1f. /f gonads were ongoing at least until E15.5, whereas the male pathway was abolished by E15.5 in XY Sox9-Cre;Nr5a1f. /− gonads, suggesting that the dissimilar developmental profiles resulted in different reproductive phenotypes between the two cKOs. To the best of our knowledge, we showed that the early loss of Nr5a1 in XY mice can induce gonadal sex reversal, providing the first evidence supporting a critical role of NR5A1 in gonadal sex determination in mice in vivo.

Methods
All methods were carried out in accordance with ARRIVE guidelines.
Animals. All procedures were conducted in accordance with the National Institutes of Health Guide for the Sections from at least three gonads in each genotype group at each time point were processed in parallel to reproducibly compare immunostaining on at least two separate occasions. Negative control tissue sections were incubated without primary antibodies. Digital images were acquired using a Keyence BIOREVO microscope (BZ-9000; Keyence, Osaka, Japan) and transferred to Photoshop CS (Adobe Systems, Mountain View, CA, USA) to generate figures. Contrast, brightness, and cropping in all images were adjusted. Investigators who were blinded to the assignments of gonads quantified IHC data derived from at least three gonads from three mice per genotype.