Introduction

Gonadal sex determination is a fundamental process in vertebrates because the phenotypic sex depends on whether the gonad develops into a testis or an ovary. In most mammals, a single sex-determining gene on the Y chromosome, SRY is sufficient to induce the undifferentiated gonad to develop as a testis (Gubbay et al., 1990; Sinclair et al., 1990; Koopman et al., 1991; Capel et al., 1993). SRY encodes a high mobility group domain transcription factor, and its downstream target is the Sox9 gene (Sekido and Lovell-Badge, 2008). Many other key downstream genes have been identified as having roles in the sex-differentiation pathway in human and mouse, although their molecular functions and the regulatory networks among these genes are unknown (Wilhelm et al., 2007).

A second vertebrate sex-determining gene was identified in the medaka fish, Oryzias latipes. This species also has an XX/XY sex-determination system (Aida, 1921), but initiation of the male-determination pathway begins with the expression of another Y chromosome-specific gene, Dmy (Matsuda et al., 2002). Dmy is the Y-specific paralog of the autosomal Dmrt1 gene which appears to be involved in male sexual development in vertebrates (Nanda et al., 2002). Mutations with impaired Dmy function cause XY sex reversal (Shinomiya et al., 2004; Otake et al., 2006, 2008) and the presence of an exogenous Dmy gene induces XX sex reversal (Matsuda et al., 2007; Otake et al., 2009), suggesting that this gene is necessary and sufficient to induce testicular development. Dmy contains a DNA-binding motif (the DM domain), and is thus thought to encode a putative transcription factor that activates or represses downstream target genes to regulate sexual differentiation of the bipotential gonad. However, it remains unclear how Dmy controls the male-determination pathway and whether its downstream genes are conserved among vertebrates.

A straightforward approach for identifying genes involved in the sex-determination pathway is the genetic analyses of sex-reversal conditions. Our previous study demonstrated an intriguing sex reversal in the interspecific hybridization between O. latipes and O. curvinotus (Shinomiya et al., 2006), despite these two species having a common sex-determining pathway with the Dmy gene on the homologous Y chromosome (Matsuda et al., 2003). In XYlatipes hybrids between O. curvinotus females and O. latipes males, Hd-rR strain males produced XY females in the hybrids (110 males and 30 females), whereas HNI males produced no XY females (70 males and 0 females). This demonstrates that there is a strain difference in the incidence of XYlatipes females (Shinomiya et al., 2006; Kato et al., 2010). A congenic approach successfully demonstrated that a small region on the Y chromosome, which included Dmy, was responsible for the strain-specific difference in this XYlatipes sex reversal (Kato et al., 2010).

In the opposite cross between O. latipes females (Hd-rR strain) and O. curvinotus males, all XYcurvinotus hybrids developed as females. This suggests that Dmy derived from O. curvinotus cannot determine maleness in the hybrid background (Shinomiya et al., 2006). A previous study reported that O. latipes females from some wild strains produced both males and females in the interspecific hybrids (Sakaizumi et al., 1992), implying that there were genetic background differences in the incidence of the XYcurvinotus sex reversal. However, these strain differences have not been confirmed because there were no sex chromosome-linked markers available in the hybrid fish until recently.

In the present study, we compared the ability to induce maleness in the XYcurvinotus hybrids among three medaka inbred strains, Hd-rR, HNI and HO4C, using the Dmy gene as a Y chromosomal marker. Because these strains showed different incidences of XY sex reversal in the interspecific hybrids, we mapped the chromosomal region responsible for the sex reversal. Linkage analysis using the Hd-rR and HNI strains identified a single major locus on linkage group (LG) 17, which was associated with the strain differences in the XY sex reversal. This region also contributed to differences in the frequency of hybrid sex reversal between the Hd-rR and HO4C inbred strains, suggesting that the hybrid XY sex reversal may be caused by an incompatibility between a single autosomal locus derived from O. latipes and the Dmy allele of O. curvinotus. Further analysis will reveal the molecules that interact with Dmy, and allow us to understand the molecular mechanisms of sex determination that begin with Dmy.

Materials and methods

Fish

We used a laboratory stock of O. curvinotus, as well as three inbred strains (HNI, Hd-rR and HO4C) of O. latipes. All fish were supplied by a sub-center (Niigata University, Japan) of the National Bioresource Project (Medaka) (http://www.shigen.nig.ac.jp/medaka/). The wild stock of O. curvinotus was originally collected in Hong Kong in 1986 (see Takehana et al., 2005). HNI was established from the Northern Japanese Population, whereas Hd-rR and HO4C were from the Southern Japanese Population (Hyodo-Taguchi, 1996; Kinoshita et al., 2009). The fish were maintained in aquaria under an artificial 14-h light/10-h dark photoperiod at 27±2 °C.

Mating and sexing

Hybrid fish were obtained from matings between O. latipes females and O. curvinotus males (XY or YY males) by pair mating. Naturally spawned eggs were collected and incubated under the same conditions as the adult fish. Hatched hybrid fish were reared until maturation for 2–3 months, and examined for their phenotypic and genotypic sexes.

Phenotypic sex was judged based on secondary sex characteristics, namely, the shapes of the dorsal and anal fins. Genotypic sex (XY or XX) was determined by the presence or absence of the Dmy gene of O. curvinotus evaluated by PCR amplification of caudal fin clip genomic DNA extracted from adult fish. PCR amplification was performed with the primers PG17.eS (5′-CGCCTTGAGGAGGCAGCAGG-3′) and PG17.20U (5′-GCATCTGCTGGTACTGCTGGTAGTTG-3′) with an annealing temperature of 65 °C (Shinomiya et al., 2006). Because the PCR primers amplify both Dmy and its autosomal paralog Dmrt1, individuals with only Dmrt1 fragments were judged to have an XX genotype, and those with both Dmy and Dmrt1 fragments were judged to have an XY genotype.

Production of YY males

Fertilized eggs of O. curvinotus were incubated in water containing estradiol-17β (Sigma, St Louis, MO, USA) at 0.2 μg ml−1 until hatching. Hatched fry were transferred to normal tap water and fed on a commercial pet-food diet until sexual maturation. Sex-reversed XY females were identified by PCR genotyping of Dmy and subsequently mated with normal XY males. YY males were selected by PCR genotyping of Dmy and Casp6 (Kondo et al., 2001; Matsuda et al., 2003).

Genetic mapping

Sixty expressed sequence tag (EST) markers covering twenty-four chromosomes were used to map the locus responsible for strain-specific differences in hybrid sex reversal. The primer sequences used in this study are listed in Supplementary Table S1, and the mapped positions of the EST markers are available in M Base (http://earth.lab.nig.ac.jp/~mbase/medaka_top.html). In addition, we designed sequence-tagged site (STS) primers based on genome sequence data (UTGB medaka genome browser: http://medaka.utgenome.org/ or Ensembl genome browser: http://www.ensembl.org), and used these STS markers for fine mapping. Primer sequences and chromosomal locations of these markers are listed in Supplementary Table S2. Individual PCR conditions were optimized for each primer pair, and products were digested with restriction enzymes if necessary. Separation of PCR products was performed by conventional polyacrylamide gel electrophoresis (Kimura et al., 2004).

RNA extraction and RT-PCR

Total RNA was extracted from fry at 8 days post fertilization (dpf) using an RNeasy Mini Kit (Qiagen, Tokyo, Japan), and subjected to reverse transcriptase PCR (RT-PCR) using a OneStep RT-PCR Kit (Qiagen). Aliquots (20 ng) of the total RNA were used as templates in 25 μl reaction volumes. The PCR amplification conditions were as follows: 30 min at 55 °C; 15 min at 95 °C; cycles of 30 s at 96 °C, 30 s at the annealing temperature and 60 s at 72 °C; and 5 min at 72 °C. The number of PCR cycles for Dmy and β-actin was adjusted to be 36 cycles and 22 cycles, and the annealing temperature at 63 °C and 55 °C, respectively. The primers for Dmy (cY331, 5′-AGG CTT CGT CCG GCC CTG AA-3′ and cY4-5U, 5′-GAG GCT CCT GGT GCA GAA CG-3′) amplified a 449-bp DNA fragment. The primers for β-actin (3b, 5′-CMG TCA GGA TCT TCA TSA GG-3′ and 4, 5′-CAC ACC TTC TAC AAT GAG CTG A-3′) amplified a 322-bp DNA fragment (Otake et al., 2006). Aliquots of 8 and 2 μl of the Dmy and β-actin RT–PCR products, respectively, were electrophoresed in a 2% agarose gel and stained with ethidium bromide.

Results

The HNI strain produces XY males in the hybrids

Our previous study indicated that interspecific hybridization between O. latipes females and O. curvinotus males caused complete XY sex reversal (Shinomiya et al., 2006). We used Hd-rR inbred strain females for matings with O. curvinotus males (Hd-rR-curvinotus), and found that all 569 hybrids developed as females. Among 99 genotyped individuals, 55 individuals had the XX genotype and 44 had XY, demonstrating that all XY individuals were sex reversed (Table 1). In the present study, we used another inbred strain, HNI, for matings with O. curvinotus males (HNI-curvinotus). All 69 XX hybrids were female, whereas the 60 XY hybrids consisted of 46 females and 14 males (23%), indicating that the HNI strain could produce XY males in the hybrids. This means that the two inbred strains HNI and Hd-rR differ in their incidences of XY sex reversal in the interspecific hybrids. Next, we used (Hd-rR × HNI) F1 females and (F1 × Hd-rR) BC1 females for matings with O. curvinotus males. The F1 females produced 147 females and 38 males (21%) in the XY hybrids, and four out of five BC1 females produced males in the XY hybrids wheras the one remaining female did not (Table 1).

Table 1 Genotypic and phenotypic sexes of hybrids between Oryzias latipes females and O. curvinotus males
Full size table

A single major locus controls the sex reversal ratio in the XY hybrids

To map the chromosomal region(s) responsible for the strain differences in the XY sex reversal ratio, an initial linkage analysis was performed using EST markers established in O. latipes. In the cross between (Hd-rR × HNI) F1 females and O. curvinotus males, the hybrids have either an Hd-rR- or an HNI-derived allele at each locus (Figure 1). Because XY males were found only in the HNI-curvinotus hybrids, we assumed that the chromosomal regions heterozygous for HNI and O. curvinotus were responsible for producing XY males. Among 38 hybrid males obtained from the cross, we selected 24 males and used them for genome-wide genotyping with 60 EST markers covering 24 chromosomes. Then we searched for loci that showed a significant deviation from the expected 1:1 ratio, using a classical chi-squared test. The tests revealed significantly high chi-squared values at four loci, OLa21.11f, OLc30.09h, AU168385 and MF01SSA074C05 (P<0.01), which are all located on LG 17 (Table 2). At these loci, HNI/curvinotus heterozygotes appeared at a significantly higher frequency than the 1:1 ratio, but other loci did not show a significant deviation from the expected 1:1 segregation. Furthermore, genotyping for the BC1 females used in progeny tests (see Table 1) revealed that BC1 females producing XY males had the HNI/Hd-rR genotype whereas those producing only XY females had the Hd-rR/Hd-rR genotype at the EST markers on LG 17 (data not shown). These results suggested that a single major locus, which we named Hybrid maleless (Hml), was associated with the strain differences in the hybrid sex reversal ratio.

Figure 1
figure 1

Mating scheme for linkage analysis. F1 females between Hd-rR and HNI strains were crossed with O. curvinotus males, and the resulting XY males in the hybrids were used for genetic mapping. These XY males had an O. curvinotus-derived Y chromosome and various combinations of HNI chromosomal segments in the autosomes and the X chromosome. We searched for chromosomal regions heterozygous for HNI and O. curvinotus alleles, which could produce XY males. Solid, HNI-derived chromosome; open, Hd-rR-derived chromosome; dotted, O. curvinotus-derived chromosome. The position of the sex-determining gene Dmy is shown.

Full size image
Table 2 EST markers associated with the strain difference in the XY sex reversal
Full size table

Genetic mapping using all 38 XY hybrid males showed that the Hml locus was located on LG 17 flanked by the two molecular markers AU171199 and LG17-29 (Figure 2a). To map the Hml locus precisely, we obtained an additional 198 XY hybrid males from matings between (Hd-rR × HNI) F1 females and O. curvinotus YY males, and genotyped new STS markers. This analysis narrowed the chromosomal location of Hml to a 1.2 cM interval between the two markers gs20859 and LG17–29, with strong linkage to an additional four markers LG17–54, sca3316, LG17–49 and LG17–51 (0 recombination events in 236 meioses; Figure 2b). On the basis of the latest genome assembly (version 1.0), three scaffolds (354, 435 and 1423) were mapped to this region, although there remain large gaps (approximately 120 and 140 kb) between the scaffolds (Figure 2c). To find candidate genes in this region, we surveyed mapped genes on these scaffolds using the UCSC genome browser (http://genome.ucsc.edu). Among 12 genes found in this region (chr17: 10.8–11.6 Mb), seven encoded zinc finger proteins, suggesting a clustered organization of these genes on the Hml region.

Figure 2
figure 2

Genetic and physical map around the Hml locus. (a) A sparse genetic map around the Hml locus on linkage group 17. (b) A high-resolution recombination map around the Hml locus. The Hml gene was mapped to a 1.2 cM interval based on the genotyping of 236 XY hybrid males. The numbers under each map indicate the number of recombinants between Hml and the adjacent marker. (c) A physical map around the Hml locus based on the available genome sequence data (version 1.0; http://medaka.utgenome.org/). Three scaffolds (354, 435 and 1423) are mapped to this region but there remain gaps between the scaffolds.

Full size image

The HO4C strain did not cause XY sex reversal in the hybrids

For further fine mapping, we used another inbred strain, HO4C, for matings with YY males of O. curvinotus (HO4C-curvinotus). In this cross, all 34 XY hybrids developed as males with no sex reversal (Table 3), demonstrating that Dmycurvinotus sufficiently induced maleness on this hybrid background. Then, we used (Hd-rR × HO4C) F1 females for the matings, and found a 1:1 Mendelian ratio of males to females in the XY hybrids (135 and 121 progeny, respectively) (Table 3). Furthermore, two STS markers, gs20859 and LG17-10, showed strong linkage to the phenotypic sex (Table 4). For these loci, XY females had the Hd-rR/curvinotus genotype whereas XY males had the HO4C/curvinotus genotype. These results strongly suggest that the Hml locus is involved in the different sex reversal ratios between Hd-rR-curvinotus and HO4C-curvinotus hybrids.

Table 3 Genotypic and phenotypic sexes of hybrids between Oryzias latipes females and O. curvinotus YY males
Full size table
Table 4 Linkage analysis between phenotypic sex and STS markers on linkage group 17 in XY hybrids between (Hd-rR × HO4C)F1 females and O. curvinotus males
Full size table

Reduced Dmy expression in Hd-rR-curvinotus XY hybrids

Reduced expression levels of the sex-determining gene Dmy have been observed in some XY sex-reversal mutants of O. latipes (Matsuda et al., 2002; Otake et al., 2006). To examine the relationship between the hybrid sex reversal and the Dmycurvinotus expression level, we performed an RT-PCR analysis of fry at hatching (8 dpf) in the Hd-rR-curvinotus hybrids, the HO4C-curvinotus hybrids and the male parental strain of O. curvinotus (Figure 3). The Dmy mRNA level was slightly lower in XY fry of the HO4C-curvinotus hybrids than in that of O. curvinotus, although all the XY hybrids developed as males in adulthood. On the other hand, Dmy expression was not detected in XY fry of the Hd-rR-curvinotus hybrids, which developed as all females in adulthood, suggesting that reduced expression of Dmycurvinotus was associated with the XY sex reversal in the Hd-rR-curvinotus hybrids.

Figure 3
figure 3

Expression of Dmy in the medaka hybrids. Dmycurvinotus mRNA expression in fry at hatching (8 dpf) was analysed by RT-PCR. β-actin expression was determined for calibration. Depressed or eliminated expression of Dmy transcripts was observed in XY fry of the Hd-rR-curvinotus hybrids.

Full size image

Discussion

Our mating experiments clearly demonstrated that the incidence of XY females was different among inbred strains of O. latipes. All XY individuals (44/44) in Hd-rR-curvinotus hybrids were sex-reversed females, whereas 23% (14/60) of XY HNI-curvinotus hybrids developed as males. Furthermore, all XY (34/34) HO4C-curvinotus hybrids developed as males, indicating that there are strain differences in the ability to induce maleness in the XY hybrids. Among these hybrids, the Y chromosome and one set of autosomes are derived from O. curvinotus whereas another set of autosomes and the X chromosome are different, suggesting that the presence or absence of XY females in the hybrids can be attributed to the differences in the maternal genome. Notably, male fish appeared only in XY individuals, indicating that Dmycurvinotus is necessary for the male determination of the hybrids. Taken together, these findings suggest that the XY sex reversal in the hybrids results from incompatibility between the Dmycurvinotus allele and the O. latipes alleles of autosomal and/or X chromosomal loci.

The sex reversal condition observed in the medaka hybrids is similar to XY sex reversal in B6-YPOS mice. This situation occurs when the Y chromosome derived from Mus poschiavinus (YPOS) is transferred onto the C57BL/6J (B6) inbred strain that normally carries a M. musculus Y chromosome (Eicher et al., 1982). Genetic background differences have been characterized in the inbred mouse strains. The B6 background was particularly sensitive to XY sex reversal, whereas other strains, including DBA/2J and 129S1/SvImJ, were found to be completely resistant to YPOS-associated sex reversal (Nikolova et al., 2008). Quantitative trait loci mapping using the B6-YPOS and DBA/2J strains identified multiple loci that conferred some but not all of the observed sensitivity to XY sex reversal in B6 (Eicher et al., 1996). More recently, a congenic approach using the B6.129S1/SvImJ-YPOS strain identified that a chromosome 11 region derived from the 129S1/SvImJ strain provided partial protection from sex reversal in XYPOS mice (Nikolova et al., 2008). These findings implicate the combined effects of many loci (rather than a single gene) in conferring the sensitivity to sex reversal in B6-YPOS; this has been confirmed by a recent expression quantitative trait loci approach (Munger et al., 2009).

By contrast, a single locus is implicated in sex reversal in the medaka hybrids. (Hd-rR × HNI) F1 females produced XY males in the interspecific hybrids, and four out of five (F1 × Hd-rR) BC1 females produced XY males, whereas the remaining BC1 female did not, implying that a small number of genes contributed to the strain differences in the hybrid XY sex reversal. Our linkage analysis using XY hybrid males obtained from (Hd-rR × HNI) F1 females successfully identified the single major locus Hml on LG 17 that confers sensitivity to the XY sex reversal in the hybrids. In addition, genotyping for XY hybrids obtained from (Hd-rR × HO4C) F1 females revealed strong linkage to STS markers around the Hml locus, suggesting that the same Hml locus contributed to the sex reversal in the different XY hybrids. These results suggest that a single gene, rather than a disrupted global network, causes XY sex reversal in the medaka hybrids. Thus, it is likely that allelic differences at the single Hml locus affect Dmycurvinotus function in the male-determining process in the hybrid fish.

Dmy transcripts first appear just before hatching exclusively in XY individuals of O. latipes and O. curvinotus (Matsuda et al., 2003; Kobayashi et al., 2004; Shinomiya et al., 2006). Reduced expression of the Dmy gene causes male-to-female sex reversal in O. latipes, suggesting that a threshold level of Dmy expression is required for male development (Matsuda et al., 2002; Otake et al., 2006, 2008). Our RT-PCR analysis of the medaka hybrids also revealed that Dmy expression levels in XY fry differed between Hd-rR-curvinotus and HO4C-curvinotus hybrids, in agreement with their sex reversal ratio. Hd-rR-curvinotus hybrids showed a loss of Dmycurvinotus expression in the XY fry, which developed as all phenotypic females, suggesting that the Hml gene is an upstream regulator of Dmy. Thus, the XY sex reversal in the hybrids may result from an incompatibility between the cis-regulatory region of the Dmycurvinotus allele and some Hml alleles of O. latipes as a trans-acting factor.

In this study, only 23% of XY individuals developed as males in the HNI-curvinotus hybrids. XY hybrids harboring identical genotypes developed as males or females, suggesting an incomplete penetrance for this phenotype. In the medaka mutants, a low level of Dmy expression below a threshold can induce sex reversal in a subset of XY individuals having the same chromosomal condition (Otake et al., 2006). Thus, a certain threshold level of Dmy expression may be required for male determination also in the hybrid backgrounds. Similar to the HNI-curvinotus hybrids, we obtained male XY hybrids from (Hd-rR × HNI) F1 females. However, the incidence of XY males (21%) was higher than expected. In this cross, we estimated that approximately 11% of XY individuals would become males, because half of the XY hybrids would have the Hd-rR allele at the Hml locus, and develop as females, whereas the other half of the XY hybrids would have the HNI allele, and consist of males (23%) and females (77%). The observed high incidence of XY hybrid males is probably due to a background effect when the HNI allele at the Hml locus is in the hybrid background containing a part of the Hd-rR genome. Therefore, the Hd-rR background in the hybrids seems to affect the Dmy expression level itself or the threshold for Dmy expression, and thus cause the higher frequency of the XY males. However, further expression analysis of Dmy is necessary to test these hypotheses.

In both O. latipes and O. curvinotus, the first appearance of morphological sex differentiation was a difference in the number of germ cells between XX and XY embryos, and a subset of germ cells in XX entered meiosis around hatching (Shinomiya et al., 2006). Later, morphological sex differences in somatic cells were observed. When Dmy function is impaired, germ cells in XY embryos start to proliferate and then enter meiosis just like XX embryos (Otake et al., 2006, 2008). These findings suggest that Dmy is involved in the regulation of germ cell proliferation at the early sex-determining stage and the formation of the testicular architecture. Previous histological observation of developing gonads in Hd-rR–curvinotus hybrids demonstrated that active proliferation of germ cells and oogenesis occurred in early gonadal development in the XY hybrids, identical to the early ovarian development in the XX hybrids (Shinomiya et al., 2006). Therefore, it is likely that the mismatching of the Hd-rR allele at the Hml locus with the regulatory region of Dmycurvinotus in the hybrids causes reduced expression of Dmy and disruption of the initial testis-determination steps including suppression of germ cell proliferation, thus resulting in XY sex reversal.

Detailed understanding of the molecular mechanisms underlying the sex reversal in the XY hybrids requires isolation and molecular characterization of the Hml locus. Although we found 12 candidate genes including 7 zinc finger proteins within the Hml region, other genes should be potentially located on the gap regions. To find them, we performed genomic synteny analysis using the Genomicus genome browser (version 60.01; http://www.dyogen.ens.fr/genomicus/). However, we could not find other candidate genes by this approach, because the genes around the Hml region were mapped to several chromosomes and/or scaffolds in other fish species including stickleback and Tetraodon. Furthermore, the gene order was not conserved among these fish species. Therefore, we have taken a positional approach to isolate the Hml gene, and started a chromosome walking using a bacterial artificial chromosome clones. So far, we have isolated two bacterial artificial chromosome clones that fill the gaps, and estimated the 1.2 cM region of interest to be 670–720 kb (with the gap region as 200–250 kb in total) which is similar to that in the genome data. To identify the Hml gene, we will determine the complete genomic sequence of these bacterial artificial chromosome clones, identify the functional genes in this genomic region and perform functional analyses of the candidate genes. Isolation of the Hml gene will help to define not only the processes and mechanisms underlying the hybrid XY sex reversal but also the transcriptional regulatory mechanism of Dmy in normal development of O. latipes and O. curvinotus.