We have previously reported a dominant mouse mutant, Odd sex (Ods), in which XX Ods/+ mice on the FVB/N background show complete sex reversal, associated with expression of Sox9 in the fetal gonads. Remarkably, when crossed to the A/J strain ∼95% of the (AXFVB) F1 XX Ods/+ mice developed as fully fertile, phenotypic females, the remainder developing as males or hermaphrodites. Using a (AXFVB) F2 population, we conducted a genome-wide linkage scan to identify the number and chromosomal location of potential Ods modifier genes. A single major locus termed Odsm1 was mapped to chromosome 18, tightly linked to D18Mit189 and D18Mit210. Segregation at this locus could account for the presence of sex reversal in 100% of XX Ods/+ mice which develop as males, for the absence of sex reversal in ∼92% of XX Ods/+ mice which develop as females, and for the mixed sexual phenotype in ∼72% of XX Ods/+ mice that develop with ambiguous genitalia. We propose that homozygosity for the FVB-derived allele strongly favors Ods sex reversal, whereas homozygosity for the A/J-derived allele inhibits it. In mice heterozygous at Odsm1, the phenotypic outcome, male, female or hermaphrodite, is determined by a complex interaction of several minor modifying loci. The close proximity of Smad2, Smad7 and Smad4 to D18Mit189/210 provides a potential mechanism through which Odsm1 might act.
In the mouse, male development is initiated by the transient expression of a single dominant gene on the Y chromosome, Sry (1). During embryogenesis, Sry is thought to act cell-autonomously in the bipotential gonad by specifying the Sertoli cell lineage which in turn directs testis differentiation (2–4). In the mouse, if Sry is deleted (5,6), or in the human if mutations are present in distinct regions of the SRY gene, the gonads develop into ovaries and XY females result (7,8). Similarly, mice transgenic for Sry develop as XX (Sry+) males (1), and humans containing SRY translocated into the XX genome develop as XX (SRY+) sex-reversed males (9). To date, potential Sry targets are those genes which are upregulated during early testicular development such as Sox9, Sf1 and Dmrt1 (2,10,11). Sox9 is particularly interesting in this respect because at 10.5 days post coitum (dpc) it is expressed at a low level in the genital ridges of male and female embryos. Normal XX fetal gonads downregulate Sox9 expression by 11.5 dpc, whereas XY fetal gonads upregulate and maintain Sox9 expression (12,13). This pattern of expression is consistent with Sox9 upregulation in the gonad being initiated by Sry action, either directly or through as yet unidentified intermediate genes.
We have recently described an insertional mouse mutant, Odd sex (Ods), in which two copies of a tyrosinase minigene have integrated ∼1 Mb upstream of Sox9 (14). XX Ods/+ mice show complete female to male sex reversal despite the absence of Sry. In addition, the pattern of Sox9 expression in developing XX Ods/+ genital ridges from 10.5 dpc onwards is indistinguishable from that of their XY +/+ or XY Ods/+ littermates (1) and (Y.Qin, unpublished data). Based on these data, we proposed that the transgenic insertion in Odd Sex mice had compromised the control of Sox9 expression such that Sox9 was no longer repressed in XX Ods/+ genital ridges. Its specific upregulation in the gonad was responsible for triggering Sertoli cell differentiation and subsequent male development. This is consistent with recent data showing that Sox9 can induce testis development in transgenic XX mice, essentially substituting for Sry in the gonad-determining pathway (15).
The initial Ods data were all generated on the inbred FVB/N genetic background. More recently, we have crossed FVB/N XY Ods/+ carrier males to the inbred strains CAST/Ei, C3H/He/J, C57BL/6/J, 129/Sv and A/J. With the first four strains, F1 XX Ods/+ mice developed as typical sex-reversed males. Remarkably, when crossed to the A/J strain ∼95% of the F1 (AXFVB) XX Ods/+ mice developed as fully fertile females, the rest developing as males or displaying ambiguous external genitalia. This suggested the existence of modifiers of Ods sex-reversal which could be detected using the A/J background. Such autosomal modifiers, termed tda1-3, have been shown to exist in the XYPos male to female sex reversal model of Eicher and colleagues (16). In the human, modifiers of sex determination have been postulated to exist to explain the variability of phenotype in familial cases of gonadal dysgenesis (17).
Using the Odd Sex mouse mutant, we report here the identification of a major locus, Odsm1mapping to chromosome (Chr) 18, tightly linked to D18Mit189/Mit210, Smad2, 4 and 7. Segregation at this locus could account for the presence of sex reversal in 100% of XX Ods/+ mice which develop as males, for the absence of sex reversal in ∼92% of XX Ods/+ mice which develop as females, and for the mixed sexual phenotype in ∼72% of XX Ods/+ mice that develop with ambigious genitalia. We favor the hypothesis that Odsm1 may be a component of a repressor complex responsible for extinguishing Sox9 expression in the normal embryonic female gonad.
RESULTS AND DISCUSSION
In order to map the modifiers of Ods sex reversal, fertile (AXFVB) F1 XX Ods/+ females were mated to (AXFVB) F1 XY Ods/+ carrier males to produce a segregating F2 population. At 3–4 weeks of age the external genitalia were carefully examined and mice were placed into one of three categories: male, female or ambigious. The appearance of the internal reproductive tract and the histology of the gonads was also examined after sacrifice. Genotyping for Ods and the sex chromosomes was performed by PCR as previously described (14). The genotype and phenotype of 619 F2s is presented in Figure 1. Genotypes were found in the expected mendelian frequency, indicating that all classes were equally viable. In the XX Ods/+ class, approximately equal numbers of males (72) and females (64) were found, together with 26 mice presenting with ambiguous external genitalia. Significantly, all (75) homozygous XX Ods/Ods mice were clear sex-reversed males. These data suggest that, whatever the genetic background, expression of Sox9 primarily controls sex determination in XX Odd Sex mice. In combination with the A/J background, however, expression of the sex-reversed phenotype is inhibited in some way, providing a powerful ‘sensitized cross’ with which to analyze the effect of genetic modifiers of sex determination.
Histological examination of the gonads of XX Ods/+ F2 mice in the male category showed them to be relatively homogenous (Fig. 2A–C). All had two small well-descended testis devoid of germ cells, with varying degrees of interstitial hyperplasia. In ∼10% of mice, some sections contained small areas of disorganized tubules or ovarian tissue (Fig. 2C), which was not previously observed on the pure FVB/N background. All mice in the females category had two gonads in the typical ovarian position, inferolateral to the kidneys. Histologically, the majority (∼80%) resembled normal ovaries (Fig. 2D), but were often associated with epididymal tissue (Fig. 2E). The rest ranged in appearance from streak gonads to ovotestis (Fig. 2F). Gonads from mice with ambiguous external genitalia were the most heterogeneous in appearance. Approximately 50% had an ovary on one side and a testis on the other (Fig. 2G and H), characteristic of true hermaphrodites. However, there did not appear to be a preference for any particular side. Frequently, the epididymis and uterus co-existed (arrowed in Fig. 2H). Histologically, there was considerable variability from grossly normal ovaries and testis, streak gonads with little recognizable structure, to ovotestis with developing oocytes in testicular tubules (Fig. 2I–K).
A complete genome scanning using 88 polymorphic microsatellite markers (average spacing ∼15 cM) was performed on a sample of 30 females and 32 males (AXFVB) F2 XX Ods/+. For each marker, the two genotype distributions (females versus males) were compared using a classical chi-square test. For each autosome and the X chromosome the lowest P-value is indicated (Fig. 3A, open bars). Significant distortion between the two distributions was observed for three chromosomes with P-values of 1.56%, 0.59% and 5.27×10−5 at D4Mit214, D11Mit242 and D18Mit184, respectively. The samples sizes were then increased to 50 males and 50 females, and more polymorphic markers, flanking the three STSs previously identified, were analyzed in order to better resolve the location of Ods modifier loci. In addition, 25 mice from the ambiguous category were included in the analysis. The distortion between the three genotype distributions (females, males and ambiguous) on mouse chromosome 4, shown by solid bars on Figure 3A, remained barely significant after full analysis, P=1.49%, ruling out the presence of a strong modifier on Chr 4.
After full analysis, D11Mit242 continued to show significant distortion on Chr 11 with a P-value of 2.4×10−4. In the XX Ods/+ male group 20 AA, 25 AF and five FF mice were observed compared with four AA, 28 AF and 18 FF XX Ods/+ female mice (where AA indicates homozygosity for the A/J-derived allele; FF homozygosity for the FVB/N-derived allele and AF heterozygosity). Thus, AA appears over-represented in sex-reversed males, and the FF allele over-represented in females. This direction of distortion on Chr 11 is difficult to explain in terms of a modifier of Ods, since data from the F1 (AXFVB) Ods/+ mice, which are >90% female, indicates that A/J-derived alleles, should inhibit the sex reversion. A confounding factor may be that all F2 mice are by definition, heterozygous for the Ods locus on distal Chr 11, which may influence the segregation at D11Mit214 in some way. In addition, sex-specific transmission ratio distortion (TRD) affecting mouse Chr 11 has previously been reported in two different crosses (18,19). The unexpected sense of action of this Chr 11 allele, its strength at ∼100 times less than the chromosme 18 Odsm1 locus (see below), and the known TRD on Chr 11, raised the possibility that the distortion was not specific to the sex reversion. Further genetic experiments need to be performed to confirm the relevance of the Chr 11 locus to Ods-mediated sex reversal.
Chi-square comparison of the Chr 18 allele distribution in the three classes (Fig. 3) clearly shows that they are significantly different, and a major modifier of Ods could be identified closely linked to D18Mit210/189 (P=3.4×10−6, LOD 9.4). A closer examination of the segregation in this region shows there to be a marked deficit of FF homozygotes in Ods/+ females, no AA homozygotes in Ods/+ males and a deficit of both homozygous classes in Ods/+ hermaphrodites. These data suggest that homozygosity at the FVB allele (FF) promotes Ods sex reversal, whereas homozygosity at the A/J-derived allele (AA) inhibits it. Heterozygosity at Odsm1 (FA) can give rise to males and females in approximately equal numbers, or to hermaphrodites.
As shown in Figure 4, placing Odsm1 in the D18Mit184–D18Mit189/210 interval would lead to discordance between the genotype and the phenotype of two FF females in recombinant classes 1 and 9 (circled), no discordant males and five non-AF hermaphrodites (classes 1, 3, 9 and 12). Placing Odsm1 in the telomeric segment would lead to three discordant females (classes 1 and 4), no discordant males (classes 3, 7 and 8) and six discordant hermaphrodites (classes 1, 3, 4, 7 and 8). Thus segregation at D18Mit189/210 can account for the presence of sex reversal in 100% of XX Ods/+ males (50/50), for the absence of sex reversal in >92% of XX Ods/+ females (46/50) and in ∼72% (18/25) of hermaphrodites.
Based on these data we propose that Odsm1 represents a major modifier locus of sex determination in the Ods mouse. In the F2 cross, being homozygous for the FVB-derived alleles at Odsm1 will strongly favor males. Similarly, being homozygous for A/J-derived alleles at Odsm1 will strongly favor females. In mice heterozygous at Odsm1, neither allele dominates. Although we cannot formally rule out non-classical mechanisms such as random or imprinted monoallelic expression at Odsm1 (20–22), the simplest explanation for the presence of males, females and hermaphrodites in the Odsm1 heterozygotes is that the phenotypic outcome is determined by a complex interaction of multiple minor modifiers. The existence of such genes has been demonstrated in mice (16), and postulated to exist in human (17,23). As yet we cannot determine whether Odsm1 acts through a direct effect on Sox9 expression/timing, or whether it works through a parallel pathway rendering the gonad more or less vunerable to sex reversal. In the event that its action is direct, the ODSM1 protein could act during embryonic development by mediating the repression of SOX9 in normal female gonads (2,12,14). In homozygous Ods/Ods mice it can no longer do this efficiently as expression of both alleles of Sox9 is under control of the transgenic insertion (14). In the case of the Ods/+ heterozygote, certain ‘strong’ or early acting alleles of Odsm1, such as are contained in the A/J background, would still be able to sufficiently repress Sox9 to produce females. ‘Weak’ or late acting alleles from FVB/N and other inbred strains would still be incapable of repressing Sox9, leading to males.
An examination of the gene map around D18Mit210/189, which currently define Odsm1 (Fig. 4) for potential Odsm1 candidates, reveals the presence of Smad2 (15 kb from Mit189) Smad7 (506 kb from Mit210) and Smad4 (2.2 Mb from Mit210). This gene family of co-activators and repressors are known to be downstream mediators of the TGF-β beta signaling pathway (24). The involvement of such important pathway regulators would provide an attractive model for the mechanism of Odsm1 function in sex determination.
MATERIAL AND METHODS
Mice and breeding
FVB/N mice and A/J mice were purchased from Charles River Laboratory and Harlan Sprague–Dawley, respectively. The Ods mutation is maintained in our laboratory on an inbred FVB/N background. An outcross–intercross strategy was used for mapping potential modifiers of Ods sex reversal. An FVB XY Ods/+ male was first crossed to an A/J female. Fertile (AXFVB) F1 XX Ods/+ females were then intercrossed to (AXFVB) F1 XY Ods/+ males. F2 mice were classified as male, female or hermaphrodites based solely on the appearance of external genitalia. After sacrifice, gonads were fixed in Bouin's solution, processed for routine histological analysis and stained with hemotoxylin/eosin according to standard procedures.
Genotyping and data analysis
DNA was extracted from the spleens or livers of 3–4-week-old mice using a DNA isolation kit (Promega Inc.) according to the manufacturer's instructions. The presence or absence of the Y chromosome and tyrosinase minigene transgene was determined by PCR as previously described (14). Homozygous Ods/Ods mice were identified visually by their dark coat color, and confirmed by PCR using primers Contig19F (gctgtgtttccagggtgaat) and Contig19R (caaatttccctggcagtgtt). These primers map within the DNA segment deleted in Ods mice and amplify a 200 bp product from +/+ or Ods/+ DNA but no product from Ods/Ods homozygotes. Whole genome screening was carried out in the Kleberg Genotyping Center (Baylor College of Medicine) using 88 fluorescently labeled microsatellite primers, polymorphic between A/J and FVB/N (available upon request), obtained from Research Genetics (Murine MapPairsTM). The primers chosen were labeled with FAM, TET or HEX and cover the 19 autosomes as well as the X chromosome with an average spacing of ∼15 cM. PCR products were detected using an ABI PRISM 377 sequencer. Products from additional informative Chr 18 markers were analyzed on standard 4% agarose gels after amplifying 100 ng of genomic DNA with unlabeled primer pairs.
Genotyping data was imported into the MapManager software package (25) for ease of management and analyzed using standard statistical methods. For each marker the genotype distributions among male, female and hermaphrodite populations were compared using a standard chi-square test for contingency table. For the LOD calculation on MMU18, the following model was used after observing the raw data: among (AXFVB) F2 XX Ods/+ offspring, males should not be homozygous AA at Odsm1, females should not be homozygous FF and hermaphrodites should not be either homozygous AA or FF. The male, female and hermaphrodite populations were split into two classes each: homozygous AA and non-homozygous AA for the males; homozygous FF and non-homozygous FF for the females; and homozygous and heterozygous for the hermaphrodites.
We would like to thank Ms Edna Wright and Helen Martinez for excellent technical assistance and Dr Jan Rohozinski for critical reading of the manuscript. This work was funded by grants from the NIH and the March of Dimes Birth Defects Foundation (to C.E.B.).
To whom correspondence should be addressed: Department of Obstetrics and Gynecology, Baylor College of Medicine, 6550 Fannin Street, Houston, TX 77030, USA. Tel: +1 7137988221; Fax: +1 7137985074(1); Email: firstname.lastname@example.org