Homozygous Inactivation of Sox9 Causes Complete XY Sex Reversal in Mice¹

Abstract

In the presence of the Y-chromosomal gene Sry, the bipotential mouse gonads develop as testes rather than as ovaries. The autosomal gene Sox9, a likely and possibly direct Sry target, can induce testis development in the absence of Sry. Sox9 is thus sufficient but not necessarily essential for testis induction. Mutational inactivation of one allele of SOX9/Sox9 causes sex reversal in humans but not in mice. Because Sox9^–/– embryos die around Embryonic Day 11.5 (E11.5) at the onset of testicular morphogenesis, differentiation of the mutant XY gonad can be analyzed only ex vivo in organ culture. We have therefore conditionally inactivated both Sox9 alleles in the gonadal anlagen using the CRE/loxP recombination system, whereby CRE recombinase is under control of the cytokeratin 19 promoter. Analysis of resulting Sox9^–/– XY gonads up to E15.5 reveals immediate, complete sex reversal, as shown by expression of the early ovary-specific markers Wnt4 and Foxl2 and by lack of testis cord and Leydig cell formation. Sry expression in mutant XY gonads indicates that downregulation of Wnt4 and Foxl2 is dependent on Sox9 rather than on Sry. Our results provide in vivo proof that, in contrast to the situation in humans, complete XY sex reversal in mice requires inactivation of both Sox9 alleles and that Sox9 is essential for testogenesis in mice.

Introduction

Mammalian testis determination is triggered by the Y-chromosomal testis-determining factor SRY. In its presence, the bipotential gonadal anlagen differentiate into testes, while in its absence, ovaries develop. Whereas the testis-inducing function of SRY has been unambiguously demonstrated [1], its direct target gene or genes still await identification. As one likely SRY target, the Sry-related gene Sox9 has emerged. In humans, heterozygous SOX9 mutations cause partial or complete XY sex reversal in the context of the skeletal malformation syndrome campomelic dysplasia [2, 3], and duplication of the chromosomal region 17q23.1-q24.3 encompassing the SOX9 locus was present in an XX individual with female-to-male sex reversal [4]. During gonadogenesis of the mouse, Sox9 is initially expressed in both sexes, with expression decreasing in the developing ovary and strongly increasing in the developing testis, concomitant with the peak of Sry expression at Embryonic Day 11.5 (E11.5) [5, 6]. Furthermore, SOX9 has been shown to colocalize with SRY in the nucleus of Sertoli cell precursors as early as E11.5 [7], consistent with the hypothesis that Sox9 is a direct, and possibly the only, target of SRY [8]. Sox9 can also substitute for Sry as a testis-determining factor, as ectopic expression of a Sox9 transgene [9] and mutational upregulation of Sox9 expression [10] cause testis development in XX mice. These latter data show that Sox9 is sufficient for testis induction but do not prove that it is essential for this process.

In contrast to the situation in humans, heterozygous Sox9 mutations do not cause XY sex reversal in mice [11]. As mouse embryos homozygously mutant for Sox9 die at E11.5 at the onset of testicular morphogenesis, the fate of the mutant XY gonad could be studied only ex vivo in organ culture, revealing no signs of testis cord formation after 3 days in culture [12]. To follow the development of Sox9^–/– XY gonads in vivo and during the entire phase of gonadogenesis, a conditional, gonad-specific knockout of Sox9 is needed. A Cre transgene under control of an Sf1 (steroidogenic factor 1, also known as Nr5a1) regulatory element has been used for this purpose, but because of inefficient and/or late Cre-mediated deletion of the Sox9^flox allele, mutant gonads always showed some sex cord formation [12]. By using the Ck19:Cre mouse line, where the CRE recombinase is under control of the cytokeratin 19 promoter [13], we have achieved complete, homozygous inactivation of Sox9 at the initial stage of XY gonadal development, providing in vivo evidence for an essential role of Sox9 in testogenesis.

Materials and Methods

Mouse Crosses

Sox9^flox/+ mice, originally on a 129P2/OlaHsd × C57BL/6 mixed genetic background [14], have been backcrossed to C57BL/6 for three generations. These N3 mice were made Sox9^flox/flox by sister-brother matings and crossed with the Ck19:Cre transgenic mouse line [13]. Resultant Cre/+;Sox9^flox/+ offspring were backcrossed to Sox9^flox/flox mice to obtain Cre/+;Sox9^flox/flox mice. Embryos at E11.5 were staged by counting the number of tail somites posterior to the hind-limb bud. Older embryos were staged according to the day of plug formation. PCR was used on tail tip or yolk sac DNA for genotyping. Primers and PCR conditions for the Ck19:Cre allele [15], for the Sox9 and Sox9^flox alleles [14], and for Sry [16] were used as previously described.

The experiments with animals were performed at the Animal Facility of the Institute of Human Genetics and Anthropology, Freiburg, Germany. The animals were housed under a 12L:12D cycle with free access to standard mouse chow and tap water. All experimental procedures complied with the rules of the German Animal Welfare Law and were licensed by the local authorities. This is in accordance with the International Guiding Principles for Biomedical Research Involving Animals.

Histology, Immunohistochemistry, and In Situ Hybridization

For histology, embryos from E11.5 to E15.5 were collected in PBS, fixed in Serra (ethanol: 37% formaldehyde:acetic acid, 6:3:1), embedded in paraffin, and sectioned at 7 μm. Staining of sections with hematoxylin and eosin followed standard techniques.

For immunohistochemistry, sections were deparaffinized in xylene and hydrated through descending ethanols, and antigens were retrieved in sodium citrate buffer. Sections were blocked in 5% serum from the species where the secondary antibody was raised for 2 h and incubated with primary antibody at 4°C overnight. Rabbit anti-SOX9 (1:200; kind gift of M. Wegner), goat anti-DMC1 (Santa-Cruz; 1:200), and rabbit anti-SF1 (1:1000; kind gift of K. Morohashi) were used as primary antibodies. After washing with PBS + 0.1% Tween20, sections were incubated in secondary biotinylated antibodies (Vector Laboratories) for 1 h at room temperature. After washing, sections were incubated with avidin-fluorescein (Vector Laboratories) for 30 min at room temperature, counterstained with DAPI, and mounted with Vectashield mounting medium (Vector Laboratories). For double immunostaining, following SOX9 immunohistochemistry, sections were treated with an Avidin/Biotin Blocking Kit (Vector Laboratories), and immunostained using goat anti-AMH antibody (1:200; Santa Cruz) and Avidin-Texas Red (Vector Laboratories). Section in situ hybridization with a probe for Wnt4 [17] was performed as described [18], with some modifications.

For histology and immunohistochemistry, a minimum of four and up to 10 mutant gonads were analyzed for each time point and marker; for in situ hybridization, four mutant gonads were analyzed for each time point.

Quantitative RT-PCR Analysis

For real-time RT-PCR analysis, total RNA from single pairs of urogenital ridges (mesonephros + gonad) or from individual urogenital ridges was isolated using the Absolutely RNA Microprep Kit (Stratagene), including a DNase I treatment, and eluted in 30 μl. Subsequently, an aliquot of 10 μl RNA was reverse transcribed with SuperScript II RNase H^– Reverse Transcriptase (Invitrogen) and oligo-dT primers in a total volume of 20 μl. To verify the absence of genomic DNA contamination, an aliquot of each RNA sample was used for PCR without reverse transcription, as Sry and Foxl2 are intronless genes. Primer pairs for all other genes covered at least one intron. One microliter of cDNA was used for PCR analysis employing the QuantiTect SYBR green real-time PCR kit (Qiagen) on a Biorad iCycler in a 96-well format. Because of the limited amount of material and to ensure that each sample could be subjected to at least two independent experiments, all samples were measured as duplicates. A standard curve was generated for each gene using serial dilutions of a cDNA pool from several male embryos with 13–23 tail somites. Expression levels were determined in one plate for all samples simultaneously and normalized to the corresponding amounts of Tbp (TATA box binding protein) cDNA. Control experiments using additional housekeeping genes (beta-actin, glyceraldehyde-3-phosphate dehydrogenase, and hypoxanthine-guanine-phosphoribosyl transferase) had shown that Tbp expression was not affected by Sox9 inactivation. Analysis was done on a total of five mutant samples of E11.5 together with six male and four female controls. For stage E12.5, two independent experiments using two (experiment 1) and four (experiment 2) samples of each genotype (mutant and control males and females) were performed. Since the two experiments were done at different times, the results could not be pooled.

The following primers were used: for Tbp, GGC CTC TCA GAA GCA TCA CTA and GCC AAG CCC TGA GCA TAA (see http://medgen31.rug.ac.be/primerdatabase/; for Sox9, CGG AGG AAG TCG GTG AAG A and GTC GGT TTT GGG AGT GGT G; for Sry, GCA AAC AGC TTT GTG GTC AA and GGA AAA GGG GAT GAA ATG GT; for Amh, ACC CTT CAA CCA AGC AGA GA and CCT CAG GCT CCA GGG ACA; and for Foxl2, GCA GAA GCC CCC GTA CTC and ATG CTA TTC TGC CAG CCC TTC. All primers (except for Tbp) were designed using the Primer3 program (http://fokker.wi.mit.edu/cgi-bin/primer3/primer3_www.cgi).

Results

Ck19:Cre;Sox9^flox/flox XY Embryos Show Absence of Testis Differentiation and of Müllerian Duct Regression

We used the CRE/loxP system to generate a homozygous knockout of Sox9 during early gonadal development. For this, we crossed our Sox9^flox mouse line, where exons 2 and 3 of Sox9 are flanked by loxP sites [14], with the Ck19:Cre mouse line [13]. This line expresses the CRE recombinase throughout the early postimplantation mouse embryo, including the embryonic ectoderm, mesoderm, and definitive endoderm [19]. Relevant to the present work, intense lacZ staining of E10.5 and of E11.5 urogenital ridges in the CRE reporter line R26R [20] reveals that the Ck19:Cre transgene must be efficiently expressed in this tissue already before the upregulation of Sox9 expression in XY embryos (see Supplemental Fig. 1 [available online at http://www.biolreprod.org]). Cre/+;Sox9^flox/+ animals survive, are fertile, and can be backcrossed to Sox9^flox/flox mice to obtain Cre/+;Sox9^flox/flox offspring. As revealed by SOX9 immunohistochemistry at E8.5 to E11.5, embryos with the latter genotype show moderate to severe reduction of SOX9-positive cells in the otic placode/vesicle, premigratory and early migratory cranial neural crest cells (CNCCs), migrating sklerotomal cells, and heart endocardial cushions and thus display reduction in CNCC-derived tissue and in all cartilage anlagen and heart hypoplasia (F. Barrionuevo, unpublished). As a consequence of these various defects, Cre/+;Sox9^flox/flox embryos show significant early embryonic lethality, with only about one eighth of the expected number of embryos recovered after E11.5. Whereas Cre/+;Sox9^flox/flox XX and Cre/+;Sox9^flox/+ XY embryos show normal ovary and testis development, respectively (not shown), Cre/+;Sox9^flox/flox XY embryos (hereafter termed mutants) show complete sex reversal. The failure of testis differentiation is already apparent at E12.5 when testis cords first become visible in the control male gonad (Fig. 1A), while no signs of testis differentiation are observed in the mutant gonad, which has the morphology of a control female gonad (Fig. 1, B and C). This difference is more pronounced at later stages when control male gonads show distinct testis cords, while mutant gonads remain in the differentiation stage typical of female gonads up to E15.5, the latest stage analyzed (Fig. 1, D–I).

Fig. 1

Histology of control and mutant gonads and sexual ducts. Starting at E12.5, testis cords (arrows) are visible in the control male gonad (A); no signs of differentiation can be observed in the control female (B) and mutant gonad (C). At E13.5, control male gonads present with clearly visible testis cords (D), whereas control female (E) and mutant (F) gonads show no sign of differentiation. This can also be observed at E15.5 (G–I). At E15.5, in the control male, the Müllerian duct (md) has started to degenerate (J), while the epithelium of the Wolffian duct (wd) is well developed. In the control female (K), the Müllerian duct presents no signs of degeneration, whereas the Wolffian is regressing. Mutant ducts develop like the control female ducts (L). Transverse sections; bar = 80 μm for A–C and 50 μm for D–L

The absence of testicular function in the mutant gonad is also apparent from the differentiation status of the sexual ducts. At E15.5, in the control male, secretion of AMH (anti-Müllerian hormone) from testicular Sertoli cells has initiated the degeneration of the Müllerian duct, while testosterone produced from testicular Leydig cells has stimulated differentiation of the Wolffian duct, which shows a well-developed epithelium (Fig. 1J). In the control female as well as in the mutant, the Müllerian duct displays no signs of degeneration and presents with an open lumen, while the Wolffian duct has started to degenerate (Fig. 1, K and L).

Immunohistochemistry Reveals Presence of Female-Specific Markers in XY Mutant Gonads

To characterize the mutant gonads further, immunohistochemistry was performed. Double staining for SOX9 and AMH, an early Sertoli cell marker regulated by SOX9 [21, 22], gives signals for both markers within the testicular cords in E13.5 control male gonads (Fig. 2A) but not in control female or mutant gonads (Fig. 2, B and C). Staining with an antibody against SF1, a key regulator of steroidogenesis that is expressed in the gonads of both sexes from E9.5 on, gradually increasing its expression until E14.5 and subsequently downregulated in ovaries but not in testes [23], reveals strong expression in interstitial Leydig cells surrounding the testis cords and weak expression in Sertoli cells in E13.5 control male gonads (Fig. 2D). The different SF1 expression pattern characteristic of E13.5 control female gonads [24] (Fig. 2E) is also shown by the mutant gonad (Fig. 2F). If E15.5 gonadal sections are stained with an antibody against DMC1, a marker for female germ cells at the stage of meiotic arrest [25], the control male gonad gives no signal (Fig. 2G), while the mutant gonad gives a strong signal similar to that seen in the control female gonad (Fig. 2, H and I).

Fig. 2

Immunohistochemistry of control and mutant gonads. A–C) Double immunohistochemistry with antibodies against SOX9 (green-yellow signal, nuclear) and AMH (red signal, cytoplasmatic) at E13.5. In the control male (A), fluorescent signals are observed for both antibodies within the testis cords but no signals in the control female (B) and the mutant gonad (C). D–F) Immunohistochemistry with antibody against SF1 at E13.5. SF1 is strongly expressed in Leydig cells surrounding testis cords in the control male gonad (D), displaying a distinctly different staining pattern in the control female and mutant gonads (E and F, respectively). G–I) Immunohistochemistry with antibody against DMC1, a marker for meiotic oocytes, at E15.5. Control male gonad shows no signal (G), whereas both control female (H) and mutant gonads (I) show clear signals. Transverse sections; bar = 100 μm for A–I

One E13.5 and one E15.5 mutant embryo, out of 11 mutant embryos analyzed histologically, had a completely sex-reversed gonad on one side and an ovotestis on the contralateral side, revealing areas with and areas without testis cords (Fig. 3, A and D). These ovotestes most likely resulted from mosaic expression of the Ck19:Cre transgene, a phenomenon observed previously [13, 19, 26]. Immunohistochemical analysis shows that the Sertoli cell markers SOX9 and AMH are expressed only in regions of the ovotestes that contain regularly formed testis cords and in nearby clusters of cells that may constitute incompletely formed testis cords (Fig. 3, B and E). Likewise, the Leydig cell marker SF1 is expressed only in these regions (Fig. 3, C and F), while the region devoid of testis cords in the E15.5 ovotestis stains positive for the oocyte marker DMC1 (Fig. 3G). This confirms that Sox9 is needed for Sertoli cell formation [9] and that Leydig cells can differentiate only in the neighborhood of testis cords.

Fig. 3

Ovotestes in XY Cre/+;Sox9^flox/flox mice. A–C) E13.5 ovotestis. Hematoxylin-eosin(HE)-stained section shows a small area with testis cords and a larger area devoid of testis cords (A). Immunohistochemistry shows that testis cords express male markers SOX9 and AMH (B) and that surrounding regions express SF1 in Leydig cells (C). D–G) E15.5 ovotestis. HE staining shows regions with and without testis cords (D). E–G) Regions corresponding to the area boxed in D from parallel sections. Testicular portion stains positive for SOX9 + AMH (E) and SF1 (F), ovarian portion stains positive for DMC1 (G). Arrowheads point to testis cords. Transverse sections; bar in C = 100 μm for A–C; in D, 50 μm; bar in G = 100 μm for E–G

Gene Expression Analyses Corroborate Complete Sex Reversal of Ck19:Cre;Sox9^flox/flox XY Gonads

To quantify the expression levels of selected sex-determination or -differentiation genes, we performed real-time quantitative RT-PCR using RNA isolated from mutant and from control gonads. At around E11.5, mutant gonads show low Sox9 expression levels in the range of XX control gonads that is drastically below the level seen in XY control gonads (Fig. 4A, left). Also, mutant gonads display no measurable Amh expression at stages when control male gonads begin to express Amh (Fig. 4A, right). Figure 4A also indicates that a threshold of Sox9 has to be reached to initiate Amh expression, in accordance with the known role of Sox9 in this process [21, 22]. At E12.5, mutant embryos display even lower Sox9 expression levels when compared to female controls (Fig. 4B, left). The ovary-specific marker Foxl2 that is required for granulosa cell differentiation starts to be expressed in female gonads around E12.5 [27]. Accordingly, we detect significant E12.5 Foxl2 expression in the mutant and control female but only very weak expression in the control male gonads (Fig. 4B, middle), whereas the mirror-image expression pattern is seen for Amh in the same gonads (Fig. 4B, right). Wild-type Sry expression starts at E10.5, peaks at E11.5, and disappears around E12.5 [28]. In our analysis, control and mutant XY gonads show similar Sry levels at E11.5 (Fig. 4C, left). Thus, Sry expression is independent of Sox9 dosage, illustrating that Sry acts upstream of Sox9. At E12.5, Sry is downregulated in control male gonads but remains at least 5-fold higher in the mutant XY gonads (Fig. 4C, right). Similarly, at E13.5, 4.5-fold higher Sry levels were detected in Sox9 mutant embryos compared to male controls (data not shown).

Fig. 4

Gene expression analysis of control and mutant gonads by real-time PCR. A) Analysis of Sox9 (left) and Amh (right) expression in control and mutant gonads at E11.5. B) Analysis of Sox9 (left), the female-specific gene Foxl2 (middle), and the male-specific gene Amh (right) in control and mutant gonads at E12.5. C) Determination of Sry expression levels at E11.5 and E12.5. In case of E12.5 (in B and C), the left bars correspond to experiment 1, the right bars to experiment 2, performed with two and four samples of each genotype, respectively. All experiments were performed at least twice, yielding virtually identical results. In each case, the result of one measurement is shown. Numbers on the y-axis refer to relative expression levels. Note that to define a point of reference, the mean values for male controls are set as 1. Error bars denote standard deviation

Wnt4 Is Expressed in XY Mutant Gonads

Wnt4 is an early ovary-specific marker that is expressed at E11.0 in the genital ridges of both sexes, becoming downregulated in the male and maintained in the female gonad [17], where it inhibits endothelial and steroidogenic cell migration into the developing ovary [29]. Because Wnt4 is also expressed in the mesonephros that was not removed from the urogenital ridges used for real-time PCR, Wnt4 expression was analyzed by RNA in situ hybridization. At E11.5, the Wnt4 signal is observed only in the mesonephros of the male urogenital ridge (Fig. 5A), while in control female and mutant gonads, a clear signal is also visible in the gonadal compartment of the urogenital ridge (Fig. 5, B and C). At later stages, the control male gonad shows only weak background signals for Wnt4, while control female and mutant gonads continue to show significant Wnt4 expression (Fig. 5, D–I). At these later stages, continuous Wnt4 expression in the mesonephroi and expression in the developing kidney is observed in both sexes, as described [17].

Fig. 5

Wnt4 is expressed in mutant gonads from E11.5 onward. A–I) RNA in situ hybridization for Wnt4, a marker for early female gonadal differentiation. A–C) Transverse sections of E11.5 urogenital ridges. Control male shows signals only in the mesonephros (A), while control female (B) and mutant (C) show signals also in the genital ridge. D–I) Transverse sections at E13.5 (D–F) and E15.5 (G–I). Control male gonads show no (D) or weak (G) background signals, while control female (E and H) and mutant gonads (F and I) show significant Wnt4 expression. Gonad, g; mesonephros, ms; kidney, k. Bar = 50 μm for A–C and 80 μm for D–I

Discussion

We show here that by use of the Ck19:Cre line, complete, homozygous inactivation of a conditional Sox9 allele could be achieved at the earliest stages of gonadal development. In contrast to constitutively inactivated Sox9^–/– embryos, which die at E11.5 at the onset of testicular morphogenesis, some of the Ck19:Cre;Sox9^flox/flox embryos survive up to at least E15.5. This allowed us to follow the fate of Sox9^–/– XY gonads in vivo throughout this entire phase of testicular development and circumvented the necessity for organ culture as in the case of constitutively inactivated Sox9^–/– XY gonads [12]. Both our in vivo study and the ex vivo organ culture study by Chaboissier et al. [12] show that, in contrast to the situation in humans, complete XY sex reversal in mice requires inactivation of both Sox9 alleles and that Sox9 is essential for testis induction to occur.

A detailed study of the spatial and temporal activity of the CRE recombinase expressed from the Ck19 locus revealed that CRE reaches functional levels already between E4.5 and E5.5 throughout the epiblast and that, by E7.5, β-galactosidase-positive cells of the R26R reporter line used were distributed throughout the entire embryo proper, including the embryonic ectoderm, mesoderm, and definitive endoderm [19]. Correspondingly, we observed Ck19:Cre-mediated inactivation of Sox9 not only in gonads but also in many other tissues, causing significant early embryonic lethality, as mentioned in Results. It may be argued that Sox9 inactivation in these additional tissues would have an indirect effect on gonad development. However, the fact that Sry expression in the mutant gonads initiates normally as shown by quantitative RT-PCR strongly suggests that the gonadal anlagen form properly and that development along the ovarian pathway is based solely on Sox9 inactivation in the gonad itself.

Another aspect of the Ck19:Cre allele is that it can cause recombination of a loxP-flanked allele in a mosaic pattern [13, 19, 26]. In the detailed study by Means et al. [19], the extent of CRE-mediated recombination of the R26R lacZ reporter allele was shown to vary from embryo to embryo, with about two thirds of the embryos exhibiting β-galactosidase activity in 75%–100% of cells and about one third of embryos in 25%–75% of cells; however, the tissue distribution was consistent. We observed the formation of a unilateral ovotestis in only two out of 11 mutant embryos (22 gonads) analyzed histologically, indicating that the degree of gonadal mosaicism in our cross was rather low.

Efficient CRE-mediated excision of the Sox9 alleles in mutant embryos is also demonstrated by our real time RT-PCR analysis. At E11.5, Sox9 expression in the mutant gonads was not completely extinguished and was similar to the low levels detected in female controls, whereas at E12.5, Sox9 expression in the mutants was even lower than in female embryos [5, 6]. This indicates that CRE-mediated inactivation of Sox9 was not complete but was efficient enough to reduce Sox9 expression to a level that causes complete, immediate XY sex reversal.

Expression analyses by real time RT-PCR have been performed using five mutant samples of E11.5 and two and four mutant samples of E12.5, respectively. As exemplified by Sry, gene expression during gonad development and differentiation is a very dynamic and stage-dependent process. In our analysis this can be observed in case of the Sox9 and Amh expression in male E11.5 controls, which shows a linear correlation with developmental stage, indicated by the number of tail somites. This dynamic gene expression might also be an explanation for the variability between the two experiments for E12.5, in which case the embryos were staged according to plug formation. Lower Sox9 and Foxl2 and higher Sry levels in the mutant gonads of the second experiment might be due to a slightly earlier developmental stage of the respective embryos.

We found that, in contrast to wild-type male gonads, Sry expression is not turned off at E12.5 (and E13.5) but remains high in mutant XY gonads. These findings are in line with the observation by Chaboissier and coworkers [12] of persistent Sry expression in E13.5 XY gonads of Sf1:Cre;Sox9^flox/flox mice with low levels of Sox9. It appears that one function of SOX9 is to downregulate Sry expression, directly or indirectly, after it has itself been upregulated by SRY.

The fact that the mutant XY gonads analyzed by us express Wnt4 and Foxl2 in a female-specific manner even though Sry expression persists beyond E11.5 furthermore indicates that downregulation of these early ovary-specific markers in wild-type XY gonads is dependent not on SRY but rather on SOX9 or on a downstream target of SOX9, or on SOX9 and on a downstream target of SRY with which SOX9 may interact in this process. Our results also indicate that Sox9 may be involved in the rapid downregulation, between E11.0 and E11.5, of Wnt4, the earliest known female-specific gonadal marker. A recent report by Qin and Bishop [30] arrives at the same conclusion. In humans, an additional copy of WNT4 has been implicated in dosage-dependent XY sex reversal [31]. Based on our data, one could speculate that this sex reversal is caused by the extra dosage of WNT4 overriding the repressive action of SOX9, which normally downregulates WNT4 below a certain threshold level to allow testis development to proceed. Under this hypothesis, the XY sex reversal resulting from SOX9 haploinsufficiency in humans may, at least in part, be caused by the inability of the reduced amount of SOX9 protein to downregulate WNT4 below the critical threshold level. On the other hand, it has been reported that Sertoli cell differentiation is compromised in Wnt4 mutant testes and that Wnt4 acts downstream of Sry and upstream of Sox9 [32]. Together with the data presented here, it seems possible that both Sry and Wnt4 are required for Sox9 upregulation and that for normal testis development to proceed, Wnt4 subsequently has to be downregulated. This effect appears to be mediated by SOX9, either directly or indirectly.

In conclusion, by crossing our mouse line carrying a conditional Sox9 allele with the Ck19:Cre line, complete, homozygous inactivation of Sox9 was achieved at the earliest stages of gonadal development, resulting in XY sex reversal. Thus, even though the Ck19:Cre line expresses the CRE recombinase in many embryonic tissues in addition to the gonadal anlagen, this line may nevertheless prove to be a valuable tool for the effective, early conditional inactivation of other genes in the sex determination pathway.

Acknowledgments

We thank Kenichiro Morohashi and Michael Wegner for generous gifts of antibodies, Katrin Wieland and Jürgen Zimmer for dedicated technical assistance, and Ulrike Dohrmann for breeding Sox9^flox mice to homozygozity and for comments on the manuscript.

Author notes

¹

Supported by the Plan de Perfeccionamiento de Doctores de la Junta de Andalucia to F.B. and by grants from the Deutsche Forschungsgemeinschaft to C.E. (En 280/6–1) and G.S. (Sche 194/15–1+2).

³

Current address: Prince Henry’s Institute of Medical Research, Monash Medical Centre, 246 Clayton Rd., Melbourne, Victoria 3168, Australia.

⁴

Current address: Leibniz Institute for Age Research-Fritz Lipmann Institute e.V. (FLI), Beutenbergstr. 11, D-07745 Jena, Germany.