Change of the Heterogametic Sex From Male to Female in the Frog
M. Ogata, H. Ohtani, T. Igarashi, Y. Hasegawa, Y. Ichikawa, I. Miura


Two different types of sex chromosomes, XX/XY and ZZ/ZW, exist in the Japanese frog Rana rugosa. They are separated in two local forms that share a common origin in hybridization between the other two forms (West Japan and Kanto) with male heterogametic sex determination and homomorphic sex chromosomes. In this study, to find out how the different types of sex chromosomes differentiated, particularly the evolutionary reason for the heterogametic sex change from male to female, we performed artificial crossings between the West Japan and Kanto forms and mitochondrial 12S rRNA gene sequence analysis. The crossing results showed male bias using mother frogs with West Japan cytoplasm and female bias using those with Kanto cytoplasm. The mitochondrial genes of ZZ/ZW and XX/XY forms, respectively, were similar in sequence to those of the West Japan and Kanto forms. These results suggest that in the primary ZZ/ZW form, the West Japan strain was maternal and thus male bias was caused by the introgression of the Kanto strain while in the primary XX/XY form and vice versa. We therefore hypothesize that sex ratio bias according to the maternal origin of the hybrid population was a trigger for the sex chromosome differentiation and the change of heterogametic sex.

THE most common mechanisms of genetic sex determination are male heterogamety as designated XX female/XY male and female heterogamety as designated ZZ male/ZW female (Bull 1983). In vertebrates, the heterogametic sex is male in mammals whereas in birds it is female. The other lower vertebrates such as reptiles, amphibians, and fishes have both types; the type may differ between species or any larger taxonomic groups. In Amphibia, female heterogamety is assumed to have evolved first, because the morphologically primitive species are most commonly heterogametic in females. Male heterogamety is thought to have appeared later at certain evolutionary branching points and quite rarely to have reversed back again to females (Hillis and Green 1990). Here, the following questions concerning sex determination arise: Why do the two types of heterogamety exist? How is the one type changed to the other? These are the basic questions to be solved to understand the genetic mechanisms of sex determination and their evolution.

Concerning the evolution of the sex-determining mechanisms, the Japanese frog Rana rugosa is quite unique (Figure 1). Two different types of sex chromosomes exist in two separate local forms (XX/XY and ZZ/ZW forms), which are assumed to share a common origin in hybridization between the other West Japan and Kanto forms, both having homomorphic sex chromosomes and male heterogametic sex determination (Miuraet al. 1998). Coexistence of the two types of sex-determining mechanisms in the same species is unusual and, apart from R. rugosa, has been reported only in the platy fish, housefly, and midge (Bellamy 1922; Gordon 1927, 1944; Thompson 1971; Macdonald 1978). Such a species would be an excellent animal model to study the evolution of one heterogametic mechanism to another. Because the ancestral-type forms of R. rugosa still remain and are separated geographically, it is possible to trace sequentially the evolution of male heterogamety to female heterogamety. In this study, to find out how the two different types of sex chromosomes differentiated in this species, we artificially performed crossings between the two ancestral-type forms of West Japan and Kanto according to the hybrid origin hypothesis, paying special attention to the sex ratio of the offspring. Also, we examined the maternal (cytoplasmic) origins of the XX/XY and ZZ/ZW forms by analyzing the mitochondrial 12S rRNA gene sequence. On the basis of the results, we infer the evolutionary reason why the two different types of sex chromosomes differentiated and discuss the evolution of male heterogamety to female heterogamety.


Frogs: The frogs used for crossing were prepared from the strains of Hiroshima (West Japan form) and Isehara (Kanto form) that had been reared through inbreeding at the Institute for Amphibian Biology (Higashihiroshima, Japan). Ovulation was accelerated by injection of pituitary gland solution prepared from R. nigromaculata (Ohtaniet al. 1997). The gynogenetic diploids ZZ and WW were produced according to the method of Ohtani et al. (1997). The localities from which the frogs used for mitochondrial gene analysis were collected are shown in Figure 1. One male and one female per population were used for the analysis.

Figure 1.

—The four geographic forms of R. rugosa in Japan. The West Japan and Kanto forms have a male heterogametic sex determination with homomorphic sex chromosomes. The other two forms with ZZ/ZW and XX/XY sex chromosomes are designated, respectively, ZZ/ZW and XX/XY forms. The sex chromosomes, together with the homologous no. 7 chromosomes of the West Japan and Kanto forms, are diagrammatically shown. The numbers on the map denote the localities from which the frogs were collected for mitochondrial gene analysis. 1, Takigawa city of Hokkaido; 2, Namioka-machi of Aomori Prefecture; 3, Niigata city of Niigata Prefecture; 4, Kanazawa city of Ishikawa Prefecture; 5, Isehara city of Kanagawa Prefecture; 6, Kamogawa city of Chiba Prefecture; 7, Fujieda city of Shizuoka Prefecture; 8, Hamakita city of Shizuoka Prefecture; 9, Hiroshima city of Hiroshima Prefecture; 10, Hakata-cho of Ehime Prefecture; 11, Kaimon-cho of Kagoshima Prefecture.

Mitochondrial 12S rRNA gene analysis: Mitochondrial DNA was isolated from blood or liver cells together with nuclear genomic DNA. The 409 bp of 12S rRNA was amplified using 20 μm sense and antisense primers (Sumidaet al. 1998) in a 50-μl reaction solution containing 5 μl of 10× buffer, 4 μl of 2.5 mm dNTP, 1 μl of Taq polymerase (Takara, Berkeley, CA), and distilled water, 30 cycles at 95° for 1 min, 50° for 1 min, and 72° for 1 min. The products were purified with MicroSpin TMS-300HR columns (Pharmacia Biotech, Piscataway, NJ) and sequenced by PCR direct sequencing method with an ABI PRISM 310 genetic analyzer. To construct the gene tree, a distance matrix based on Kimura's two-parameter method (Kimura 1980) was calculated and clustered by the neighbor-joining method (Saitou and Nei 1987) using PHYLIP version 3.572 software (Felsenstein 1996).

cDNA subtraction and RT-PCR: Total RNA was prepared from the gonads and mesonephroi of ZZ and WW tadpoles at day 20 after fertilization according to the manufacturer's instruction [Promega (Madison, WI) SV total RNA isolation system]. cDNA was synthesized and subjected to subtraction between ZZ and WW tadpoles according to the manual of the CLONTECH (Palo Alto, CA) PCR-select cDNA subtraction kit. The cDNA sources of WW tadpoles subtracted with that of ZZ ones were ligated into PCR vector (Invitrogen, San Diego). Out of 64 clones picked up, 24 distinct clones were identified and 11 of them showed higher expression in female tadpoles than in male tadpoles at day 20. Out of the 11 clones, only 1 showed much higher expression in females and almost none in the ZZ males or in the tadpoles from the Hiroshima (West Japan) and Isehara (Kanto) forms at the same stage. It was designated W13.

Expression of W13 at day 19, 21, and 23 after fertilization was examined by reverse transcription polymerase chain reaction (RT-PCR). Total RNA was isolated and purified from the gonads plus mesonephroi of tadpoles at the days indicated above. First-strand cDNA was synthesized using 1 μg of the total RNA as the template for 1 hr at 42° in a 20-μl reaction solution containing 4 μl of 5× buffer, 4 μl of 2.5 mm dNTP, 2 μl of 0.1 m dithiothreitol, 2 μm of dT24 oligomer, 1 μl of Superscript (BRL), and distilled water. One microliter of the cDNA solution was amplified in a 50-μl reaction solution containing 5 μl of 10× buffer, 0.3 μl of Ex Taq (Takara), 4 μl of 2.5 mm dNTP, and 1 μl of 12.5 μm each of W13 sense and antisense primers, 35 cycles at 94° for 40 sec, 62° for 40 sec, 72° for 50 sec, ending with 72° for 2 min. Amplification of the 304-bp EF1α fragment of R. rugosa was according to the method of Nakajima et al. (2000). ZZ male and ZW female tadpoles were sexed by polymerase chain reaction-restriction fragment length polymorphism of the sex-linked gene ADP/ATP translocase (Sakisakaet al. 2000).

PCR for genomic DNA: Amplification of a genomic DNA fragment of W13 was carried out using 0.5 μg of ZZ or WW DNA under the same conditions as those for RT-PCR of total RNA.


Artificial hybridization: According to the hypothesis on hybrid origin of the XX/XY and ZZ/ZW forms (Miuraet al. 1998), we chose one population from each of the ancestral-type forms of West Japan and Kanto and hybridized and backcrossed them with each of the two forms.

Developmental ability: The following designations were chosen: H represents the haploid nuclear genome of the Hiroshima frog belonging to the West Japan form, I represents that of the Isehara frog belonging to the Kanto form, and HI denotes a hybrid of HH female × II male. The reciprocal F1 hybrids were all as viable as the controls during development from embryos to adults (Table 1). The backcrossed offspring, on the contrary, showed external deformities and developmental arrest in the two crossings in which maternal cytotype is paired with a predominantly nuclear genome of another form. In the two crossings of IH females × HH male, 39.4 and 40.4% of tadpoles suffered from microencephalon after feeding and died before completion of metamorphosis (Figure 2). In another two crossings of HI females × II male, 80.8 and 82.0% suffered from microencephalon and/or microphthalmia after feeding and died before complete metamorphosis (Figure 2). The morphological abnormalities and developmental arrest seem to be due to the incompatibility of the nuclear genome with the cytoplasm. In the other crossings, the dead embryos during development and until metamorphosis were all underdeveloped with no specific external deformity; thus the death may be due to egg immaturity.

View this table:

Development of the hybrid and backcrossed offspring

Sex ratio: Sex of the offspring was examined according to the external morphology of the gonads just after metamorphosis and/or in a year. The results are summarized in Table 2. First, a sex bias was observed in the F1 generation. In the HH female × II male crosses, the sex ratios were skewed toward males in 1-year-old offspring although the ratios were ∼1:1 just after metamorphosis. Conversely, in the II female × HH male crosses, sex ratios were skewed toward females both after metamorphosis and in a year. Likewise, in the backcrossings, the sex ratios were skewed toward males using female parents with H cytoplasm and toward females using those with I cytoplasm (Table 2; Figure 3). Exceptions were the following two kinds of crossings in which maternal cytotype is paired with a predominantly nuclear genome of the same form. In one crossing, each of HI × HH and IH × II, the sex ratios were ∼1:1. This suggests that maternal factors derived from two kinds of nuclear genomes (Isehara and Hiroshima) in the hybrid female eggs restored the sex ratio to 1:1 in the backcrossings. In the remaining one cross of HI × HH, the sex ratio was skewed toward males, as in other crosses with H cytotype. Conversely, in the remaining one cross of IH × II, the sex ratio was skewed toward males, which was opposite to the female-biased sex ratio in other crosses with I cytotype. In this case, the IH mother frog no. 2 was found to be genetically XY because its gynogenetic offspring showed ∼1:1 in the sex ratio (XX:YY; Table 2). In a cross of XY female with XY male, the sex ratio is expected to be 3:1 (75% males). The actual result was only 64.3% males, showing occurrence of sex reversal from male to female and thus a female bias.

View this table:

Sex of the hybrid and backcrossed offspring

Figure 2.

—Graphs showing development of the backcrossed offspring. (A) IH female. ▴, IH1 female × HH4 male; ▪, IH2 female × HH4 male; ▵, IH1 female × II4 male; □, IH2 female × II4 male. (B) HI female. Χ with shaded background, HI1 × HH4; •, HI2 × HH4; Χ, HI1 × II4; ○, HI2 × II4.

Here, as for the mechanism to induce sex bias, the following three candidates can be given: sex reversal, maternal control by meiotic drive, and gender-biased lethality. Sex reversal is the most plausible in this case, because hermaphrodites were actually found in some hybrids and backcrossed offspring, particularly with the IH2 female parent being proved to be a genetic male of XY. In contrast, the maternal control is impossible in this case because all the females are homogametic sex XX. Also, gender-biased lethality is not likely because no evident lethality with external deformity occurred before metamorphosis except in the two crossings HI × II and IH × HH, and sex bias was evident at the metamorphosis. However, we could not exclude its occurrence during the 1 year after metamorphosis in the two crossings HH4 × HI1 and HI1 × HH4, because 20 and 18.8% frogs, respectively, were lost and the sex ratios of adult frogs were much more skewed than those of the 1-month-old frogs.

Consequently, artificial crosses between the two forms showed a sex bias (Figure 3): The sex ratio was skewed toward males using the mother frogs with H (West Japan) cytoplasm, whereas it was skewed toward females using those with I (Kanto) cytoplasm.

Mitochondrial 12S rRNA gene: In the crosses between the West Japan and Kanto forms, sex ratios of the offspring were skewed according to the cytoplasmic origin of the female parent. The next question is whether or not the cytotypes of the XX/XY and ZZ/ZW forms are actually different from each other. To approach this, 409 bp of the mitochondrial 12S rRNA gene was sequenced from 22 frogs of 11 populations belonging to the four different forms and the results were compared (Figure 4). The substitution rate per site was 0.0074 ∼ 0.1168 and the 22 sequences were classified into nine distinct haplotypes. The gene tree constructed using the neighbor-joining method (Saitoh and Nei 1987) is shown in Figure 5. Two major clusters were formed: One was composed of the populations belonging to the West Japan and ZZ/ZW forms, and the other was composed of those belonging to the Kanto and XX/XY forms. A tree using the maximum-likelihood method (Felsenstein 1981) gave the same topology. These results suggest that cytoplasm of the XX/XY form originates from the Kanto form and that of the ZZ/ZW form originates from the West Japan form.

Figure 3.

—Sex ratios of the hybrids and backcrossed offspring from the Hiroshima population of the West Japan form and the Isehara population of the Kanto form. The data are based on Table 2 and are shown as postmetamorphosis sex ratios (1 year plus 1 month postmetamorphosis). In some crosses, the sex ratio of either 1-month or 1-year frogs is indicated by m (in the case of 1 month) or y (in the case of 1 year) on the right of each bar. Solid bar, male percentage; open bar, female percentage. The female parents are shown above the bar and the male parents on the left. The crossing labeled with an asterisk was found to be an XY × XY crossing because of the 1:1 (XX:YY) sex ratio of the gynogenetic offspring from the same IH female no. 2.


Hybrid origin: The hypothesis that the ZZ/ZW and XX/XY forms share a common origin in hybridization between the other West Japan and Kanto forms is based on the following findings. The first concerns the sex chromosome morphology (Nishioka et al. 1993a, 1994; Miura et al. 1996, 1997; Ohtaniet al. 2000). The Y and Z sex chromosomes are subtelocentric, as is no. 7 of the West Japan form in morphology, while the X and W are metacentric and identical in shape to each other (Figure 1). Chromosome banding patterns, lampbrush chromosome morphology, and pairing configurations also indicate the similarity of Z and Y to West Japan no. 7 and the similarity of X to W. The second concerns the sequence of a sex-linked gene ADP/ATP translocase, studies on which added to the above similarities between the sex chromosomes and revealed that the X and W sex chromosomes were the most closely related to the more subtelocentric no. 7 of the Kanto form (Miuraet al. 1998). On the basis of these results, it is assumed that the Y and Z chromosomes originated from no. 7 chromosome of the West Japan form, whereas the X and W chromosomes originated from that of the Kanto form. For this assumption, hybridization between the West Japan and Kanto forms is prerequisite.

Figure 4.

—Variable sites in the DNA sequences of the PCR fragments of mitochondrial 12S ribosomal RNA derived from the frog R. rugosa from 11 populations in Japan. A dot indicates that the nucleotide is the same as that shown on the top line. Nine haplotypes are classified: ZW1, Takigawa male and female, Namioka male and female, Niigata male and female; ZW2, Kanazawa male and female; Kanto, Isehara male and female, Kamogawa male and female; XY1, Fujieda male; XY2, Fujieda female, Hamakita male and female; West1, Hiroshima male; West2, Hiroshima female and Hakata-cho female; West3, Hakata-chomale; West4, Kagoshima male and female. Since the sequence of R. catesbeiana (Yoneyama 1987) used as an outgroup with a specific one-nucleotide insert is not included in the alignment, the nucleotide number ends at 408.

The four forms of XX/XY, ZZ/ZW, West Japan, and Kanto were shown by electrophoretic analyses of isozymes to be genetically distinct, forming two groups (Nishiokaet al. 1993b): One contains the West Japan form only and the other is composed of the remaining three. This means that the nuclear genomes of XX/XY and ZZ/ZW forms are genetically close to that of the Kanto form. However, among the three forms of XX/XY, ZZ/ZW, and Kanto, the XX/XY form shares the most isozyme alleles with the West Japan form. The next question is: What about their cytoplasmic similarities? The answer was obtained in this study from the mitochondrial 12S rRNA gene sequence. The 12S rRNA genes of the XX/XY and ZZ/ZW forms, respectively, were similar to those of the Kanto and West Japan forms in sequence (Figure 4). Thus, it can be concluded that the XX/XY and ZZ/ZW forms have different combinations of nuclear genome and cytotype: a nuclear genome similar to Kanto with West Japan cytotype for the ZZ/ZW form and a nuclear genome less similar to Kanto, but with Kanto cytotype for the XX/XY form.

Figure 5.

—The 12S rRNA gene tree of the nine haplotypes of R. rugosa. The numbers shown at the nodes of the tree are the percentages of 1000 bootstrap replicates.

Sex bias: The experimental hybridization of the West Japan and Kanto forms caused opposite types of sex bias in the offspring: a male bias when choosing the West Japan cytoplasm and a female bias when choosing the Kanto cytoplasm. Together with the information about nuclear and cytoplasmic (mitochondrial) genotypes of the XX/XY and ZZ/ZW forms, the following is assumed: In the primary XX/XY population, the Kanto strain was maternal, being subject to a slight introgression of the West Japan nuclear genome and thus suffering female bias. On the contrary, in the primary ZZ/ZW population, the West Japan strain was maternal, being subject to repeated introgression of the Kanto nuclear genome and thus suffering male bias. Here, because sex bias would threaten survival of the populations, there would have been a strong positive selective pressure for the minor sex-favoring gene to restore a 1:1 sex ratio in the circumstance in which one-way sex reversal frequently occurs. Actually, such genes are acquired. The present Y chromosome is dominant in male determination and stronger than the Z chromosome: XXY triploids are all males but XXZ are mostly males with some females (Ohtaniet al. 2000). The female determination of the present W chromosome is dominant and stronger than that of the X chromosome: The WY hybrids produced from ZW females × XY males become females (Nishioka and Hanada 1994). We infer that the new sex-determining genes on the Y and W chromosomes may have become dominant after a hybridization event by chromosomal or genomic rearrangements causing their recruitment or elevation of their expression.

In summary, in the primary ZZ/ZW form in which a male bias would have occurred, a female-determination-related gene on the W chromosome (metacentric no. 7) would have become newly dominant and thus have been selected, whereas either its homolog or another male-determination-related gene on the Y chromosome (subtelocentric no. 7) would have acted dominantly as a male determiner and thus have been selected in the primary XX/XY form in which a female bias would have occurred. Consequently, we hypothesize that sex ratio bias according to the maternal origin of the hybrid population was a trigger for the sex chromosome differentiation and change of the heterogametic sex (Figure 6).

Figure 6.

—Assumptive history of establishment of the two types of sex chromosomes with female and male heterogametic sex determination in R. rugosa.

Change of heterogametic sex: Bull and Charnov (1977) hypothesized that a new-born sex-determining gene can be rapidly extended and fixed in the population, if it is linked to a gene with high adaptive value, and finally cause a change of the heterogametic sex. However, as seen in this study on R. rugosa, a minor sex-favoring gene can also be easily fixed in a sex-biased population and cause a change of the heterogametic sex.

Wilkins (1995) has proposed an intriguing hypothesis on the evolution of a genetic sex-determination pathway in the nematode Caenorhabditis elegans. Its pathway is a negative regulatory cascade, in which all the regulatory steps involve negative control. Wilkins proposed that the pathway evolved in reverse order—from the final step up to the first; the pathway arose in steps, driven by frequency-dependent selection for the minority sex at each step and involving the sequential acquisition of dominant negative genetic switches. A sex ratio bias is a prerequisite for each step. Wilkins (1995) also implied that the hypothesis is applicable to other animals. It is indeed interesting when considering the sex-determining genes of vertebrates. The mammalian SRY, a male determiner, is considered a dominant negative regulator acting as a repressor of a repressor of Sox 9, inducing testis differentiation (Graves 1998; Huanget al. 1999; Bishopet al. 2000). In birds, a candidate gene for female determination, Wpkci or ASW, has been isolated recently (Horiet al. 2000; O'Neillet al. 2000). The Wpkci (ASW) is an altered form of PKC inhibitor with no HIT domain and is amplified ∼40 times on the W chromosome. A single copied homolog is located on the Z chromosome. The Wpkci's expression level in females is about seven times higher than that of pkci of ZZ males in the gonads plus mesonephroi at stage 29, suggesting that the much higher production of Wpkci is involved in ovary determination by interfering with PKCI function of the Z chromosome. This function is also a dominant negative regulation. In the platyfish, which has a male heterogametic sex determination, the newly appeared female determiner W is dominant against the male determiner Y (Gordon 1944), which is also likely to be a dominant negative regulator.

Again, the Wilkins hypothesis is applicable to the evolution of heterogamety: If a male dominant determiner appears and becomes fixed against a female bias, the sequential selective process ends as male heterogamety, and if a new female determiner becomes fixed against a male bias, it ends as female heterogamety. Therefore, the change of heterogamety may be recognized as a step in the process of creating the sex-determination pathway. Wilkins (1995) added that the sequential selective process ends with the gene switch that becomes fixed in the evolution of the sex chromosome. The case of R. rugosa seems to clearly fit this model.

Figure 7.

—W-linkage of W13 (A) and its expression in the gonads plus mesonephroi around the beginning of morphological gonad differentiation (B). The geographical forms of the tadpoles examined are shown at the top. Black- and-white images are reversed. EF1α (elongation factor 1α) is a control amplified product.

A new female determiner: We have attempted to identify the gene that was acquired as a new female dominant determiner on the W chromosome in the evolution of the ZZ/ZW mechanism in R. rugosa. Since the morphological differentiation of gonads starts around day 22 after fertilization, subtraction was carried out in mRNA expressed in the gonads and mesonephroi of WW females at day 20 with that of the ZZ males. One mRNA species, the sequence of which does not appear on any database, has been isolated; it shows a W-link and an intriguing expression pattern (Figure 7). The mRNA clone designated W13 was expressed from the W chromosome and thus only in ZW female gonads plus mesonephroi but not in ZZ male gonads or in either sex of tadpoles from the ancestral-type forms of Kanto and West Japan at the same stages. This probably means that the W13 was recruited in the W chromosome or acquired the higher expression in females after the hybridization event in the process of W chromosome differentiation through an inversion. Work is now underway to confirm the functional relationship of W13 with female determination and to analyze its homolog on the Y chromosome to uncover the mechanism of heterogametic sex change at the molecular level.


We thank J. N. Raybould for his critical reading of the manuscript.


  • Communicating editor: K. Golic

  • Sequence data from this article have been deposited with the DDBJ Data Library under accession nos. AB075890–AB075898.

  • Received February 5, 2003.
  • Accepted February 25, 2003.


View Abstract