Sexually dimorphic development of the gonads is controlled by positive and negative regulators produced by somatic cells. Many Wnt ligands, including ones that signal via the canonical β-catenin pathway, are expressed in fetal gonads. β-catenin, a key transcriptional regulator of the canonical Wnt pathway and an element of the cell adhesion complex, is essential for various aspects of embryogenesis. To study the involvement of β-catenin in sex determination, we ablated β-catenin specifically in the SF1-positive population of somatic cells. Although β-catenin was present in gonads of both sexes, it was necessary only for ovarian differentiation but dispensable for testis development. Loss of β-catenin in fetal testes did not affect Sertoli cell differentiation, testis morphogenesis or masculinization of the embryos. However, we observed molecular and morphological defects in ovaries lacking β-catenin, including formation of testis-specific coelomic vessel, appearance of androgen-producing adrenal-like cells and loss of female germ cells. These phenotypes were strikingly similar to those found in the R-spondin1 ( Rspo1 ) and Wnt4 knockout ovaries. In the absence of β-catenin, expression of Wnt4 was down-regulated while that of Rspo1 was not affected, placing β-catenin as a component in between Rspo1 and Wnt4 . Our results demonstrate that β-catenin is responsible for transducing sex-specific signals in the SF1-positive somatic cell population during mouse gonadal development.
The gonadal primordium uniquely has the potential to differentiate into two distinct organs, an ovary or a testis. In most mammals, the sex of the gonad is determined by the Y-linked gene, SRY (sex-determining region of the Y chromosome) ( 1 ). Sry serves as a genetic switch to induce expression of Sry -related HMG-box gene 9 ( Sox9 ), which in turn activates a cascade of events leading to testis morphogenesis ( 2–7 ). The transformation of the gonadal primordium into an ovary is considered to occur by ‘default’ in the absence of SRY and SOX9 . However, the finding of testis formation in XX individuals lacking SRY has led to an alternative proposal, the ‘Z’ hypothesis ( 8–12 ). According to this model, an ovary-specific Z factor(s) acts as an inhibitor of the testis morphogenesis pathway, therefore, allowing the emergence of the ovary. SRY, instead of being a testis-inducer, antagonizes the Z factor and thereby permits the progression of the testis program. Much evidence indicates that the level of antagonism is not restricted to the level of transcription. The testis and ovary both produce unique signaling molecules with opposing functions such as fibroblast growth factor 9 ( Fgf9 ) and Wnt4 , respectively ( 13 ). The outcome of the sex determination apparently hinges on the balance between the antagonizing pathways induced by these signaling molecules.
WNT4, a paracrine and/or autocrine factor involved in the antagonizing pathways in sex determination, belongs to a highly conserved family of secreted glycoproteins. Wnt4 is expressed in mouse gonads in both sexes at embryonic day 9.5 (E9.5) and becomes ovary-specific at the time of sex determination around E11.5 ( 14 ). In the absence of Wnt4 , partial female-to-male sex reversal occurred in mice with the appearance of testis vasculature, ectopic androgen-producing cells and the androgen-dependent Wolffian duct, the precursor of the male reproductive tract ( 14–16 ). Various degrees of female-to-male sex reversal were also found in human patients with mutations in the WNT4 gene ( 10 , 17 ). Rspo1 , another ovary-specific gene, is thought to be a component of the Wnt4 pathway based on the fact that similar ovarian phenotypes developed in both Rspo1 and Wnt4 knockouts ( 18 , 29 ). WNT4 and RSPO1 have been shown to signal via β-catenin in certain tissues; moreover, in the Rspo1 knockout ovaries, genes involved in β-catenin signaling were down-regulated ( 18 ). However, there is no direct genetic proof that connects Wnt4 , Rspo1 and β-catenin in regulating ovarian development.
In addition to Wnt4 , sex-specific expression was also found for other Wnt genes such as Wnt5a , Wnt6 and Wnt9a in the ovary ( 19 , 20 ) and WNT1, WNT3 and WNT7A in the testis ( 20–26 ). The overlapping expression of multiple Wnt genes in the gonads suggests that these WNTs could act redundantly or synergistically via common signaling pathways such as β-catenin. In this study, we investigated the role of β-catenin by inactivating it specifically in the somatic cell population of the gonadal primordium. Our results revealed that although β-catenin was present in gonads in both sexes, it was only required for maintaining ovarian identity but dispensable for testis formation.
Removal of β-catenin in somatic cells in gonads using the SF1/cre transgenic mice
To investigate the role of β-catenin in gonadal development, we generated a conditional knockout model in which β-catenin was ablated in somatic cells of fetal gonads by SF1/cre recombinase ( 27 ). The activity of the SF1/cre was detected in the somatic cells of gonads of both sexes as early as E11.5 ( 27 ). We first confirmed that β-catenin was removed specifically from the somatic cells of gonads using immunohistochemistry with a β-catenin antibody that has been shown to detect β-catenin in fetal mouse testis ( 28 ). In the control testes ( SF1/cre;Ctnnb1+/− or Ctnnb1f/− ), β-catenin was detected on the membrane of anti-Müllerian hormone (AMH)-positive Sertoli cells and germ cells starting at E15.5 (Fig. 1 A and C), consistent with previous findings ( 28 ). In contrast, in the β-catenin conditional knockout mice (or SF1/cre;Ctnnb1f/− , Fig. 1 B and D), the staining for β-catenin in Sertoli cells was abolished, whereas it was maintained in germ cells. No β-catenin staining was detected in Leydig cells, which also express SF1/cre (data not shown).
In control fetal ovaries as early as E12.5 (Fig. 1 E and G), membrane β-catenin was detected in most somatic cells (PECAM negative) and germ cells (PECAM positive). In the SF1/cre;Ctnnb1f/− ovaries (Fig. 1 F and H), membrane staining of β-catenin was unaffected in PECAM-positive germ cells but disappeared in the PECAM-negative somatic cells. These results demonstrated that Cre recombinase directed by the SF1/cre transgene inactivated β-catenin in a somatic cell-specific manner.
Normal testis development in the absence of β-catenin
We first examined whether testes form properly in the absence of β-catenin in the somatic cells. SOX9 and AMH, two molecular markers for fetal Sertoli cells, were expressed in a manner indistinguishable between control and SF1/cre;Ctnnb1f/− testes at E12.5 (Fig. 2 A and B). Testis cords and testis-specific vasculature also developed normally (Fig. 2 A–D). At birth, the entire male reproductive system was intact and testis morphology was normal in the SF1/cre;Ctnnb1f/− testes (Fig. 2 I and J). This was further confirmed by molecular analysis of Sertoli cell marker AMH (Fig. 2 E and F) and Leydig cell marker CYP17 (Fig. 2 G–H). These results indicated that loss of β-catenin in SF1-positive somatic cells did not affect the development of fetal testes.
Changes of the ovarian program and appearance of the testis-specific vasculature in the SF1/cre;Ctnnb1f/− ovaries
Next, we investigated whether the development of the ovaries was affected. We first performed whole-mount in situ hybridization for Rspo1 , Wnt4 and Fst , three genes that normally are expressed in an ovary-specific manner (Fig. 3 A–B, D–E and G–H). In the SF1/cre;Ctnnb1f/− ovaries, Rspo1 expression was maintained (Fig. 3 C), whereas expression of Wnt4 and Fst was down-regulated (Fig. 3 F and I) compared with the control ( SF1/cre;Ctnnb1+/− or Ctnnb1f/− , only SF1/cre;Ctnnb1+/− is shown here; Fig. 3 B, E and H). This result indicated that β-catenin in the SF1-positive cells is required for expression of Wnt4 and Fst but not Rspo1 .
When either Rspo1 , Wnt4 or Fst was lost, fetal ovaries form the testis-specific coelomic vessel, an artery normally found only in the testis ( 15 , 16 , 29 ). This evidence supports the hypothesis that Rspo1 , Wnt4 and Fst form a linear signaling cascade that suppresses the appearance of testicular vasculature in the ovary. As shown in Figure 3 L, loss of β-catenin in the ovary also resulted in ectopic formation of the coelomic vessel, which was demarcated by staining for PECAM. The coelomic vessel did not form in the control ovary but was present in the control testis (Fig. 3 J and K). This result implicates β-catenin as a component of the Rspo1 / Wnt4 / Fst pathway that inhibits testicular vasculature formation in fetal ovaries.
The appearance of the testis-specific coelomic vessel in the SF1/cre;Ctnnb1f/− ovaries raised the possibility that the absence of β-catenin allowed activation of the testis pathway, which in turns caused formation of the vessel. Based on the observation that Sox9 and Fgf9 were transiently expressed in the Wnt4 knockout ovaries, it was proposed that Wnt4 antagonizes the expression of Sox9 and Fgf9 , two factors critical for specifying the testis fate ( 13 ). We therefore examined whether Sox9 and Fgf9 were up-regulated and thus could be responsible for coelomic vessel formation in the SF1/cre;Ctnnb1f/− ovaries. Sox9 and Fgf9 mRNA were expressed at a high level in the control testes as expected (Fig. 3 M and S). However, we found no evidence of up-regulation of Sox9 and Fgf9 mRNA in the control or SF1/cre;Ctnnb1f/− ovaries at E11.5 and 12.5 (Fig. 3 N, O, T and U). Immunohistochemical analysis of SOX9 protein confirmed the mRNA observation (Fig. 3 P–R). These data revealed that loss of β-catenin did not lead to activation of the testis pathway, indicating that disruption of the ovarian program and the appearance of the coelomic vessel were direct effects of loss of β-catenin.
Maintenance of the Wolffian duct derivatives and appearance of adrenal-like cells in the SF1/cre;Ctnnb1f/− female
Development of the reproductive tract in the SF1/cre;Ctnnb1f/− female grossly resembled that of the control female (Fig. 4 A and B). However, examination under higher magnification revealed that the SF1/cre;Ctnnb1f/− female developed epididymal structure and vas deferens, both derivatives of the Wolffian duct (Fig. 4 D). Histological analysis using H&E staining further confirmed the identity of these structures as similar to those found in the control male (Fig. 4 F and G). As expected, these Wolffian duct-derived structures were absent in the control female (Fig. 4 C and E). The Müllerian duct derivatives (e.g. oviduct and uterus) were still intact in the SF1/cre;Ctnnb1f/− female (Fig. 4 C and D). Although some of the Wolffian duct derivatives were maintained, the external genitalia of the SF1/cre;Ctnnb1f/− female was indistinguishable from that of the control female (data not shown).
Maintenance of the Wolffian ducts requires androgens produced by fetal Leydig cells. Thus, the epididymal structures and vas deferens observed in the SF1/cre;Ctnnbf/− female could result from the ectopic appearance of androgen-producing fetal Leydig cells in a manner similar to Rspo1 and Wnt4 knockout mice ( 14 , 29 ). To test this possibility, we performed immunohistochemistry for CYP17, a key enzyme in the pathway of androgen production, and INSL3, a Leydig cell-specific factor that mediates testis descent ( 30 , 31 ). CYP17 was expressed in clusters of cells in the SF1/cre;Ctnnb1f/− ovaries but was absent in the control ovaries (Fig. 5 A and B). However, INSL3 was not detected in either the control or SF1/cre;Ctnnb1f/− ovaries (Fig. 5 C and D) but was present in control testis (inset in Fig. 5 D). These results indicate that the CYP17-positive cells were probably not fetal Leydig cells. In the SF1/cre;Ctnnb1f/− female, the ovaries remained attached to the posterior part of the kidney and did not descend to the position adjacent to the bladder (Fig. 4 B), further confirming the absence of INSL3.
We speculated that the ectopic CYP17-positive cells in the SF1/cre;Ctnnb1f/− ovaries might be of adrenal origin. As noted earlier, adrenal-like cells expressing both Cyp17 and Cyp21 were found in the Wnt4 knockout ovaries. We therefore examined expression of Cyp21 mRNA and found that clusters of Cyp21 -positive cells were present exclusively in the SF1/cre;Ctnnb1f/− ovaries but not in the control (Fig. 5 E and F). These data demonstrated that the pathway induced by β-catenin suppresses the appearance of the androgen-producing adrenal-like cells in fetal ovaries.
Loss of germ cells in the SF1/cre;Ctnnb1f/− ovaries
Differentiation and survival of female germ cells require a unique somatic environment in the fetal ovary, such that disturbance of the ovarian program (Fig. 3 ) and/or other phenotypes in the SF1/cre;Ctnnb1f/− ovaries could affect germ cell development. We performed immunohistochemistry for two germ cell markers, PECAM and TRA98 (only TRA98 is shown here) ( 32 ), and found that the number of germ cells was similar between the control and the SF1/cre;Ctnnb1f/− ovaries at E15.5 (Fig. 6 A and B). A significant decrease in germ cell numbers became apparent at E16.5 (Fig. 6 C and D) and female germ cells in the SF1/cre;Ctnnb1f/− ovaries were almost completely lost at birth (Fig. 6 E and F). H&E staining further revealed the absence of female germ cells in the SF1/cre;Ctnnb1f/− ovaries (Fig. 6 G and H), indicating that the absence of TRA98 staining was due to germ cell loss instead of decreased TRA98 expression. To assess whether the germ cells enter meiosis properly, we examined the expression of phosphorylated gamma histone 2AX (γH2AX), a meiotic maker. We found similar expression patterns between the SF1/cre;Ctnnb1f/− and control ovaries (Fig. 6 I and J), indicating the β-catenin is not required for the initiation of meiosis but is essential for the survival of meiotic germ cells.
β-Catenin is not only a key secondary effector downstream of the canonical Wnt pathway but also involved in cell–cell adhesion via cadherins, particularly in epithelial cell types. Conditional deletion of β-catenin in the SF1-positive somatic cells could affect the ability of these cells to respond to Wnt signaling as well as their interaction with neighboring cells via cell adhesion. Previous studies have shown that E- and N-Cadherins were not present in the somatic cells of fetal gonads and that loss of P-cadherin did not affect fertility ( 33 , 34 ). Furthermore, conditional deletion of β-catenin only induced phenotypes in ovaries, suggesting that loss of β-catenin did not affect general cell–cell adhesion in SF1-positive somatic cells that are the common precursors of gonads in both sexes. Although the antibody used in this study detected only membrane-bound β-catenin in the fetal gonads, this result does not exclude important nuclear functions of β-catenin. Using the same antibody, others observed nuclear β-catenin in fetal gonads only when a mutant form of β-catenin resistant to degradation was over-expressed ( 28 ). This observation suggests that under normal condition, nuclear shuttling of β-catenin is a transient and dynamic process that is difficult to monitor using immunohistochemistry. Based on the loss of membrane-bound β-catenin staining and occurrence of unique phenotypes in the β-catenin conditional knockout gonads, we are confident that we have generated a valid functional model to study the role of β-catenin in the SF1-positive somatic cells.
β-Catenin is dispensable for fetal testis formation
When β-catenin was inactivated in the SF1-positive somatic cells population at E11.5, the time when the sex determination occurs, the ovarian program but not the testis pathway was affected. Molecular markers for Sertoli and Leydig cell differentiation were expressed properly and testicular structures and germ cells differentiated normally. At the time of birth, the β-catenin conditional knockout male had virilized reproductive organs indistinguishable from the control male. Using Amh-cre line to remove β-catenin specifically in Sertoli cells starting at E15.5, others found no defects in later testis development ( 28 ). Our results extend this finding by showing that β-catenin also is not required in the SF1-positive Leydig cell population during early testis morphogenesis and later maintenance. Along with the finding that activation of stabilized β-catenin in the somatic cell of fetal testis disrupted the testicular program and caused male-to-female sex reversal ( 35 ), our results show that β-catenin is dispensable for normal testis development. It was previously reported that in the Wnt4 knockout testis, an early defect in Sertoli cell differentiation and disorganized testis cords was found ( 36 ). However, these defects were not observed in our model. It is possible that WNT4 does not signal via β-catenin in the fetal testis. In our SF1/Cre conditional knockout model, β-catenin protein became undetectable starting between E11.5 and E12.5. Therefore, we cannot exclude the possibility that WNT4 regulates testis development via β-catenin before E11.5.
β-Catenin is a key component of the RSPO1/WNT4 pathway in fetal ovarian development
When abolishing the function of β-catenin in the SF1-positive ovarian cells, unique changes were observed in three ovary-specific genes, Rspo1 , Wnt4 and Fst . Expression of Wnt4 and Fst , but not Rspo1 , was down-regulated in fetal ovaries in the absence of β-catenin, placing β-catenin upstream of Wnt4 and Fst . Interestingly, in the Rspo1 knockout ovaries, Wnt4 and Fst expression was also down-regulated ( 29 ). We also found that loss of β-catenin in the SF1-positive somatic cells produced phenotypes similar to those in Rspo1 and Wnt4 knockout female. RSPO family members, including RSPO1, have been shown to exert their functions via β-catenin. Using the Axin-LacZ reporter line, Chassot et al . showed that the pathway downstream of β-catenin was affected in the Rspo1 knockout ovaries, suggesting a connection between RSPO1 and β-catenin in the fetal ovary ( 18 ). These data together support the model that β-catenin is an intracellular mediator of Rspo1 that stimulates Wnt4 expression in the SF1-positive somatic cells in fetal ovary. Although we envision that β-catenin might act downstream of Rspo1 , our results do not exclude the possibility that β-catenin could also be a downstream mediator of Wnt4 as found in other model systems ( 15 , 37–39 ).
The ovarian pathway, induced by β-catenin, suppresses formation of the coelomic vessel without activating the testis-determining pathway
The appearance of the testis-specific coelomic vessel in the β-catenin conditional knockout, as well as in Rspo1 , Wnt4 and Fst knockout ovaries, indicated that these molecules are components of a common pathway that antagonizes the formation of testis vasculature. When these ovarian genes are inactivated, the testis pathway could become active, leading to appearance of the testis characteristics. However, we found no evidence of activation of the testis pathway based on the absence of expression of the testis markers SOX9, Fgf9 and AMH in the β-catenin conditional knockout ovaries; these observations are similar to what others have found in Rspo1 , Wnt4 and Fst knockout ovaries ( 14 , 15 , 18 , 29 ). Transient expression of SOX9 and Fgf9 were reported in the Wnt4 knockout ovaries at E11.5, but the expression disappeared by E12.5 and no other Sertoli cell markers (e.g. Amh and Desert hedgehog) were present ( 13 ). In our conditional knockout model, β-catenin may be inactivated too late to allow the transient expression of Sox9 and Fgf9 . However, if this is the case, we should not observe the appearance of the coelomic vessel, which is thought to be controlled by Sox9 and Fgf9 . We therefore propose that some aspects of the testis-determining mechanisms—at least the formation of the coelomic vessel—may not require Sox9 and its downstream components.
β-Catenin inhibits the appearance of adrenal-like cells in the fetal ovary
Ectopic cells that express Cyp17 were first observed in the Wnt4 knockout ovary ( 14 ). These Cyp17 -positive androgen-producing cells were initially thought to be fetal Leydig cells; subsequent studies indicated that this is not the case ( 14 , 40 ), as these cells expressed adrenocortical markers ( 40 ). We also observed the androgen-producing cells in the β-catenin conditional knockout ovaries, as evidenced by their expression of CYP17. The adrenal identity of these cells was further confirmed by their expression of the adrenal marker Cyp21 . The lack of expression of the Leydig cell-specific INSL3 also argues that these androgen-producing cells are not fetal Leydig cells. These CYP17-positive cells in the β-catenin conditional knockout female embryos apparently produced androgens in amounts sufficient to sustain the androgen-dependent Wolffian duct derivatives such as epididymis and vas deferens. However, the amount of androgen produced by these adrenal-like cells was not sufficient to virilize the external genitalia, probably due to their small number. Similar phenomena were also observed in Rspo1 and Wnt4 knockout females, indicating that Rspo1 , β-catenin and Wnt4 together suppress the appearance of these adrenal-like cells in fetal ovaries ( 14 , 15 , 18 , 29 ). Gonadal and adrenal cells derived from a common adrenogonadal primordium and are both positive for SF1. In our conditional knockout model, β-catenin was removed in both gonads and adrenals at approximately the time of their separation. Therefore, it is possible that one function of the Rspo1 /β-catenin/ Wnt4 pathway is to ensure proper separation of adrenal and gonadal precursors during organogenesis. It is interesting to note that the appearance of CYP17-positive adrenal-like cells and maintenance of Wolffian duct derivatives were not found in Fst knockout ( 15 ). This observation indicates that although Fst plays a role in suppressing coelomic vessel formation, it is not involved in inhibiting the appearance of adrenal-like cells in the fetal ovary.
β-Catenin in SF1-positive somatic cells is essential for female germ cell survival
Our study reveals that the removal of β-catenin from the SF1-positive somatic cells of fetal ovaries caused loss of female germ cells. Similar oocytes depletion was also found in the Rspo1 , Wnt4 and Fst knockout ovaries ( 14 , 15 , 18 , 29 ). The loss of female germ cells could be due to improper differentiation of somatic cells in the absence of β-catenin, appearance of the coelomic vessel, ectopic production of androgens and/or adrenal hormones from the adrenal-like cells, or a combination of these effects. Based on the decreased expression of the meiotic marker Stra8 , it was proposed that female germ cells in the Rspo1 knockout ovaries failed to enter meiosis properly ( 41 , 42 ). However, we observed that the female germ cells entered early meiosis properly in the conditional β-catenin knockout ovary. In addition, the female germ cells also enter meiosis normally in Wnt4 and Fst knockout ovaries ( 15 ). Further experiments are required to establish a conclusive link between Rspo1 and other components of the ovarian pathway in maintaining germ cell survival. It is interesting to note that in the β-catenin conditional knockout testis, male germ cells remained undisturbed, further supporting that β-catenin is only required for development of the ovary.
It is illustrative to consider the different effects of β-catenin disruption mediated by the SF1/cre transgene in somatic cells of the ovaries, testes and adrenal gland. All three organs are believed to emerge from a common pool of adrenogonadal precursors that express SF1 and potentially can assume the adrenal or gonadal fates. Thereafter, the ovaries and testes arise from a bipotential gonad as directed by signals discussed ealier, while the adrenocortical precursors form an adrenal primordium that is not sexually dimorphic. Despite their common origin and conserved functions as primary steroidogenic organs, the effects of β-catenin inactivation differ markedly. At one extreme, β-catenin inactivation causes severe defects in the adrenal primordium such that the organ is completely lacking in newborn mice ( 43 ). In our study, we also observed the severe adrenal phenotypes with a significant decrease of the cortex where SF1 is expressed (data not shown). In contrast, the testes in newborn mice do not exhibit any structural or functional defects (this paper), although the absence of adrenal glands precludes analysis of their post-natal function. Finally, the ovaries are affected to an intermediate degree; they form along the normal pattern, but then exhibit abnormalities that apparently reflect impairments in mechanisms that normally inhibit certain aspects of normal testis (i.e. the testes artery) or adrenal (i.e. expression of Cyp21) development. Based on these emerging insights into tissue-specific differentiation of the common adrenogonadal precursor pool, it will be of considerable interest to explore further the in vivo effects of proposed modulators of this differentiation. For example, is ectopic expression of Rspo1 (or Wnt4 ) in the adrenal progenitors sufficient to repress the normal steroidogenic phenotype of these cells? Experiments such as these should provide novels insights into the developmental mechanisms that mold the common precursors into three distinct steroidogenic organs.
In summary, we provide genetic evidence to demonstrate a sexually dimorphic role of β-catenin in the SF1-positive somatic cell population during early gonad development. β-catenin is dispensable for testis formation and maintenance; however, it serves as a critical mediator that maintains ovarian characteristics. We propose that β-catenin is an intracellular component in response to Rspo1 in the SF1-positive somatic cells that acts to trigger expression of Wnt4 ; it thereby suppressing the appearance of testis-specific vasculature and androgen-producing adrenal-like cells while maintaining the survival of female germ cells. In addition to β-catenin, transcription factors such as Foxl2 have been linked to Wnt4 in maintaining ovarian cell fate ( 44 ). It remains to be determined whether FOXL2 also play roles in the Rspo1/Wnt4 pathway by interacting with β-catenin.
MATERIALS AND METHODS
Transgenic mice ( SF1/cre ) carrying Cre recombinase under the control of the Steroidogenic factor 1 ( Sf1 ) promoter and regulatory elements ( 27 ) and floxed β-catenin mice ( Ctnnb1floxed/floxed or B6.129- Ctnnb1tm2Kem /KnwJ, Jax#004152) ( 45 ) were maintained on a B6 background. To increase the efficiency of production of homozygous null alleles and decease the incidence of mosaic deletion, we generated SF1/cre;Ctnnb1floxed/− embryos. Female and male mice were paired together and checked for the presence of a vaginal plug the next morning. The day when the vaginal plug was detected was considered 0.5 days of gestation, or E0.5. Samples were collected at E12.5, E13.5, E15.5 and birth (P0). For genotyping, tail bits were digested in 50 m m sodium hydroxide and subjected to standard PCR using gene-specific primers: 5′-GAGTGAACGAACCTGGTCGAAATCAGTGCG-3′ and 5′-GCATTACCGGTCGATGCAACGAGTGATGAG-3′ for SF1/Cre genotyping, 5′-AATCACAGGGACTTCCATACCAG-3′ and 5′-GCCCAGCCTTAGCCCAACT-3′ for Ctnnb null allele and 5′-AAGGTAGAGTG ATGAAAGTTGTT-3′ and 5′-CACCATTGTCCTCTGTCTATTC -3′ for Ctnnb1 wild-type and floxed alleles. For sex genotyping, primers 5′-TGAAGCTTTTGGCTTTGAG-3′ and 5′-CCGCTGCCAAATTCTTTGG-3′ were used to detect the presence of Sry gene. All procedures described were reviewed and approved by the Institutional Animal Care and Use Committee at University of Illinois and were performed in accordance with the Guiding Principles for the Care and Use of Laboratory Animals. All experiments were performed on at least three animals for each genotype.
Tissues were fixed in 4% paraformaldehyde in PBS overnight at 4°C, processed through graded ethanol series and Histo-cleanII (International Diagnostics), and eventually embedded in paraffin. Five micrometer sections were cut and rehydrated for Hematoxylin and Eosin staining (2 min for Hematoxylin and 30 s for Eosin). Slides were mounted using Permount.
Testes and ovaries were isolated and fixed in 4% paraformaldehyde overnight at 4°C. On the next day, samples were rinsed three times in PBS for 5 min each. Then samples were put through graded sucrose series (10, 15 and 20%) and incubated in 1:1 20% sucrose and OCT-freezing media (Tissue-Tek) overnight at 4°C. The samples were embedded in 1:3 20% sucrose and OCT mix and cut to 8 µm frozen sections. Sections were washed in PBS for 10 min and blocked in blocking solution (5% heat-inactivated donkey serum and 0.1% Triton X-100 in PBS) for 1 h at room temperature. Then sections were incubated with primary antibodies (see later) at 4°C overnight. The next day sections were washed three times for 10 min each with washing solution (1% heat-inactivated donkey serum and 0.1% Triton X-100 in PBS) followed by incubation in the corresponding secondary antibodies. After secondary antibody treatment, sections were washed three times for 10 min each in washing solution and mounted with DAPI. The dilutions and sources of primary antibodies used were: rabbit polyclonal antibody against β-catenin (1:500, Sigma), rabbit polyclonal antibody against AMH (1:200, Santa Cruz), rat monoclonal antibody against germ cell nuclear fraction Tra98 (1:1000, provided by H. Tanaka) ( 32 ), rabbit polyclonal antibody against CYP17 (1:100, a gift from Buck Hales), rat polyclonal antibody against PECAM1 (1:500, Pharmingen), rabbit polyclonal antibody against INSL3 (1:100, a gift from Stefan Hartung) ( 46 ), rabbit polyclonal antibody against Laminin (1:200, Sigma), rabbit polyclonal antibody against phosphorylated histone 2AX (1:100, Upstate) and rabbit polyclonal antibody against SOX9 (1:1000, provided by Ken Morohashi) ( 47 ). All of the secondary antibodies (1:200) were fluorescein-conjugated and purchased from Jackson Immunochemical.
Whole-mount in situ hybridization
Tissues were fixed overnight in 4% paraformaldehyde in PBS at 4°C and dehydrated through a methanol gradient (25, 50, 70 and 100%) in PTW (0.1% Tween20 in DEPC-PBS). Samples were stored in 100% methanol at −20°C. In situ hybridization was processed according to the standard non-radioisotopic procedure using digoxigenin-labeled RNA probes. In brief, samples were dehydrated through a methanol gradient and then treated with proteinase K (10 mg/ml) at 37°C for 12 min followed by fixation in 4% paraformaldehyde/0.1% glutaraldehyde immediately at room temperature for 20 min. Samples were pre-hybridized in hybridization buffer (5× SSC pH 5.0, 50% formamide, 0.1% CHAPS, 0.1% Tween20, 1 mg/ml Yeast tRNA, 50 µg/ml Heparin and 5 m m EDTA pH 8.0) at 65°C for 1 h. Then digoxigenin-labeled RNA probe was added into the solution and samples were rotated in an oven at 65°C overnight (12–16 h). Samples were incubated in 20% sheep serum in MABTL blocking solution at room temperature for 2 h followed by incubating in alkaline phosphatase-conjugated anti-digoxigenin at 4°C on shaker overnight. On the third day, after washed in MABTL three times for 1 h each, samples were incubated in alkaline phosphates substrate in NTMTL (0.1 m NaCl, 0.01 m Tris–HCl pH 9.5, 0.05 m MgCl 2 , 1% Tween20, 0.05% Levamisole) for color development The optimal hybridization temperature and sources of RNA probes are: Rspo1 (65°C; Blanche Capel), Wnt4 (65°C; Blanche Capel) ( 13 ), Fst (65°C; Martin Matzuk) ( 15 ), Cyp21 (65°C; Keith Parker), Sox9 (65°C; Blanche Capel) ( 13 ) and Fgf9 (60°C; Blanche Capel) ( 13 ).
This work was supported by National Institutes of Health (HD46861 to H.H.Y, HD046743 and DK54480 to K.L.P.) and March of Dimes Birth Defect Foundation.
We would like to thank Dr. Buck Hales for the CYP17 antibody, Dr. Stefan Hartung for the INSL3 antibody, Dr. Ken Morohashi for the Sox9 antibody, Dr. H. Tanaka for the TRA98 antibody and Dr. Blanche Capel for the plasmids for in situ hybridization. We also appreciate Dr. Ana Vieira for maintaining the mouse colony and all of the Yao lab members for their assistance and support.
Conflict of Interest statement . None declared.