• A sex-independent transmission ratio distortion (siTRD) system detected in the interspecific cross in rice was analyzed in order to understand its significance in reproductive barriers. The S1 gene, derived from African rice Oryza glaberrima, induced preferential abortion of both male and female gametes possessing its allelic alternative (), from Asian rice O. sativa, only in the heterozygote.
• The siTRD was characterized by resolving it into mTRD and fTRD occurring through male and female gametes, respectively, cytological analysis of gametophyte development, and mapping of the S1 locus using near-isogenic lines. The allelic distribution of the S1 locus in Asian and African rice species complexes was also analyzed.
• The siTRD system involved at least two components affecting male and female gametogeneses, respectively, including a modifier(s) that enhances fTRD. The chromosomal location of the major component causing the mTRD was delimited within an approx. 40 kb region. The S1 locus induced hybrid sterility in any pairwise combination between Asian and African rice species complexes.
• The allelic state of the S1 locus has diverged between Asian and African rice species complexes, suggesting that the TRD system has a significant role in the reproductive barriers in rice.
Transmission ratio distortion (TRD) refers to the situation in which one of a pair of alleles is preferentially recovered in the progeny of a heterozygote. TRD can be caused by various genetic factors operating on chromosome segregation during meiosis (Pardo-Manuel de Villena & Sapienza, 2001; Birchler et al., 2003; Fishman & Willis, 2005) as well as those operating at the pre- and postzygotic phases of reproduction (Zamir & Tadmor, 1986; Oka, 1988; Xu et al., 1997; Harushima et al., 2001). Well-known TRD systems include the mouse t-haplotype (Silver, 1993) and segregation distorter (SD) of Drosophila (Temin et al., 1991), in which the postmeiotic dysfunction of spores or gametes is involved. SD and t-haplotype have been termed as meiotic drive (MD) in a broad sense, although MD was originally defined as preferential transmission of chromosomes during meiosis in the female (Pardo-Manuel de Villena & Sapienza, 2001; Birchler et al., 2003; Fishman & Willis, 2005). Genes exhibiting preferential transmission by promoting the elimination of its allelic alternative without giving any fitness advantage to the individual are considered to be selfish genetic elements (Crow, 1988). Such a pattern of selfish behavior will drastically alter the frequencies of alleles in a population, thereby affecting the genome and the species (Sandler & Novitski, 1957; Crow, 1988; Hurst & Werren, 2001). Recently, TRD (or MD in a broad sense) systems were suggested to play a role in the formation of reproductive barriers and speciation in animal species (Frank, 1991; Hurst & Pomiankowski, 1991; Orr et al., 2006). The genes that underlie the reduction in fitness of hybrids are considered to be involved in reproductive barriers, which might be important in driving a nascent species to become an independent genetic entity (Wu & Ting, 2004). Current interest in the study of speciation is focused on the identification and characterization of genes for reproductive barriers. However, it is unclear whether TRD systems are involved in reproductive barriers between species in plants.
Transmission ratio distortion associated with gametic dysfunction has been frequently detected in inter- and intraspecific hybrids of plants (Crow, 1991; Lyttle, 1991; Morishima et al., 1992). Preferential dysfunction of gametes occurs in either male (Cameron & Moav, 1957; Loegering & Sears, 1963; Sano, 1983) or female gametes (Maguire, 1963; Scoles & Kibirge-Sebunya, 1983). In this study, the former is termed male TRD (mTRD) and the latter female TRD (fTRD). On the other hand, sex-independent TRD (siTRD) results from preferential dysfunction in both male and female gametes (Rick, 1966; Endo & Tsunewaki, 1975; Sano et al., 1979; Finch et al., 1984). Genes for TRD detected in interspecific crosses can cause hybrid sterility between plant species; they have no deleterious effects in the homozygote but cause the dysfunction of gametes only in the heterozygote. However, it remains unclear, in most cases, whether such a phenomenon is observed only in the specific pair of strains or in any pairwise combinations of the individuals between species. During the process of speciation, the potential of a given locus to act as a reproductive barrier that prevents interspecific gene flow depends on its allelic distribution within the two diverging species. Although recent genetic studies on speciation have revealed the genetic architecture of reproductive barriers and characterized the genes responsible at the molecular level (Wu & Ting, 2004; Mallet, 2006; Orr et al., 2006; Rieseberg & Willis, 2007), there have only been a few studies that directly examined the patterns of genetic variation at the loci responsible for hybrid incompatibility or sterility between species (Christie & Macnair, 1987; Wu & Palopoli, 1994; Sweigart et al., 2007).
Among three TRD systems (mTRD, fTRD, and siTRD), siTRD has the severest effects on the degree of hybrid sterility as well as that of TRD in the progeny. Genes for siTRD have not been reported in animal species so far (Lyttle, 1991; Úbeda & Haig, 2005). In plants, genes for siTRD were found in tomato (Rick, 1966, 1971) and rice (Sano et al., 1979, 1994; Sano, 1992; Ren et al., 2005), and in wheat, alien chromosomes introduced from Aegilops species by intergeneric crosses also caused siTRD (Endo & Tsunewaki, 1975; Finch et al., 1984). The Gamete eliminator (Ge) gene in tomato was the first gene for siTRD reported in plants (Rick, 1966, 1971). The Ge locus causes a preferential elimination of gametes only in GecGep heterozygotes, and the Gep allele induces abortion of both male and female gametes possessing the Gec allele. Intraspecific allelic differentiation at the Ge locus was observed in cultivated and wild tomato species (Rick, 1971), indicating the occurrence of TRD in intraspecific hybrids as observed in t-haplotype and SD systems (Lyttle, 1991; Temin et al., 1991; Silver, 1993).
In contrast to siTRD caused by intraspecific crosses, we previously found a gene for siTRD (the S1 gene) in interspecific hybrids between the Asian (Oryza sativa) and African (Oryza glaberrima) cultivated rice species (Sano et al., 1979; Sano, 1990). The S1 gene was previously the most fully characterized gene for siTRD in plants. The two species are reproductively isolated by sterility barriers, although hybrids exhibit no aberration in the course of meiosis (Chu et al., 1969). The S1 locus was found to be responsible for the hybrid sterility in the cross between O. glaberrima and O. sativa ssp. indica (Sano et al., 1979). When a chromosomal region containing the S1 locus of a strain of O. glaberrima was introduced into a strain of O. sativa ssp. indica by backcrossing, dysfunction of both male- and female-gametes possessing the allele of O. sativa () was induced only in the heterozygote (), resulting in siTRD (Sano et al., 1979). The homozygotes with either the S1 or the alleles did not exhibit any reduction in fertility of male and female gametes. When the same chromosomal region was transferred to a strain of O. sativa ssp. japonica by backcrosses, the manner of transmission in the female gamete was altered. Furthermore, the transmission of the allele through the female gamete was increased when the chromosomal segment of O. glaberrima containing the S1 locus became smaller by recombination during backcrossing, suggesting the compound nature of the S1 locus (Sano, 1990).
In the present study, we characterized the siTRD system caused by the S1 locus by means of cytological examination and analysis using molecular markers. We firstly confirmed that the TRD in female gametes was altered depending on the genetic background of O. sativa. Genetic dissection of the S1 locus revealed the presence of a linked modifier gene(s) in addition to the primary gene(s) for TRD. We also carried out fine mapping of the major component of the S1 locus causing TRD in male gametes. In addition, we surveyed the distribution of the mTRD component at the S1 locus in Asian (O. sativa–O. rufipogon complex) and African (O. glaberrima–O. barthii complex) rice species complexes. The present results provide evidence for species-specific allelic differentiation at the S1 locus, and suggest that the siTRD system could actually play a significant role in reproductive barriers between rice species.
Materials and Methods
Study system and genetic stocks
In this study, we focused on two rice species, Asian (Oryza sativa) and African (O. glaberrima) cultivated rice. O. rufipogon and O. barthii are considered to be the wild progenitors of Asian and African cultivated rice, respectively. O. sativa comprises two subspecies, ssp. japonica and ssp. indica, which are well distinguished from each other by molecular markers as well as morphological and physiological characteristics (Oka, 1988; Garris et al., 2005). An annual type of Asian wild rice that is sometimes called O. nivara was classified as O. rufipogon in this study. Two cultivated rice species (O. sativa and O. glaberrima) and their five wild relatives (O. rufipogon, O. barthii, O. glumaepatura, O. meridionalis, and O. longistaminata) share the same genome A. According to the classification of the gene pool, Asian cultivated rice (O. sativa) and its wild progenitor (O. rufipogon) belong to the same biological species forming a primary gene pool (O. sativa–O. rufipogon complex), whereas African cultivated rice (O. glaberrima) and its wild progenitor (O. barthii) belong to another biological species (O. glaberrima–O. barthii complex) (Harlan, 1975; Oka, 1988).
Taichung 65 (O. sativa ssp. japonica, denoted T65) and Pehkuh (O. sativa ssp. indica, denoted Acc108), both of which carry the allele at the S1 locus, and four near-isogenic lines (NILs: T65wx, T65S1, T65wxS1, and Acc108S1) were used. T65wx is a NIL carrying the wx (waxy) gene from Kinoshita-mochi (Oka, 1974, derived from BC12). The three NILs T65S1, T65wxS1, and Acc108S1 harbor the S1 allele introduced from a strain of O. glaberrima (W025 from Guinea) by backcrosses (Sano et al., 1979; Sano, 1990). T65S1and T65wxS1 harbor the Wx and wx alleles at the wx locus, respectively. The wx locus encodes a granule-bound starch synthase that plays a role in amylose synthesis, and the Wx and wx alleles are functional and nonfunctional alleles, respectively (Sano, 1984). The Wx gene expresses in pollen grains as well as in the endosperm. Since the wx locus is tightly linked with the S1 locus, the phenotypic difference between Wx and wx alleles in pollen grains was used as a marker to detect TRD caused by the S1 locus (see later discussion). For the examination of the allelic distribution at the S1 locus, 27 strains of Asian (O. sativa–O. rufipogon) and African (O. glaberrima–O. barthii) rice species complexes were used.
Detection of TRD
The degree of TRD is usually measured in terms of a k value, where k is defined as the proportion of progeny that received the allele exhibiting the preferential transmission from the heterozygote (Lyttle, 1991). The value varies from 0.5 (Mendelian segregation) to 1.0 (complete elimination of its allelic alternative). Since the male and female gametes can be affected differently, the two parameters km and kf were defined in order to distinguish the three TRD systems based on the model of Úbeda & Haig (2005), where km and kf stand for the proportion of progeny that received the allele exhibiting the preferential transmission through the male and female gametes, respectively. Therefore, the two parameters distinguish mTRD (km = 1 and kf = 0.5), fTRD (km = 0.5 and kf = 1), and siTRD (km = kf = 1).
With regard to the S1 locus, km and kf could be estimated as shown in Fig. 1. km is estimated from backcrossing data using the heterozygote as the pollen parent. When the homozygote was pollinated with the heterozygote (), km was estimated from Ns/(Nf+Ns) in the backcrossed progeny where Nf and Ns stand for the number of pollen-fertile () and pollen-semi-sterile () plants, respectively (Fig. 1a). In the present case, no S1a allele transmitted from the heterozygote () to the progeny through male gametes (see later discussion). When km = 1, kf is estimated from Nf/(Nf+ Ns) in the selfed progeny where Nf and Ns stand for the numbers of pollen-fertile (S1S1) and pollen-semi-sterile () plants, respectively (Fig. 1b).
The Wx/wx phenotypes in pollen grains can be used to detect the degree of mTRD since the wx locus is located close to the S1 locus. By a treatment with potassium iodine solution (I2-KI), Wx and wx pollen grains are stained blue and reddish brown, respectively, while aborted pollen grains are not stained. The difference in the allelic state at the wx locus (Wx or wx) has no effects on pollen fertility (e.g. both T65 and T65wx were fully pollen-fertile). In the heterozygotes at both the S1 and wx loci (S1-wx/-Wx or S1-Wx/-wx), about half of the pollen grains were not stained (aborted), resulting in deviation of the Wx : wx ratio from 1 : 1 in stainable (fertile) pollen grains, which confirms that the preferential transmission is caused by gamete abortion. In the fertile pollen grains, the recombination value between the S1 and wx loci was computed based on the frequency of Wx pollen grains in S1-wx/-Wx or wx pollen grains in S1-Wx/-wx since all fertile pollen grains are expected to carry the S1allele (see later discussion).
Survey for the allelic distribution at the S1 locus
To examine the allelic distribution at the S1 locus, 25 strains of Asian rice species (O. sativa and O. rufipogon) were crossed with the two NILs carrying or S1S1. The allelic state was determined based on pollen and seed fertilities in the F1 plants and distorted segregation of the Wx phenotype in pollen grains. On the other hand, the allelic state of two strains of African rice species (C7639 of O. glaberrima and B19 of O. barthii) was examined by introducing the region containing the S1 locus into T65wx by backcrosses because interspecific F1 hybrids between the African rice species and O. sativa exhibited nearly complete male sterility. After confirming the introgression of the S1 region using molecular markers, test crosses were carried out to examine their allelic states.
Genetic analysis using molecular markers
Genomic DNA was isolated from 2-month-old plants by the CTAB method for RFLP analysis according to the method of Murray & Thompson (1980). RFLP markers (RZ398 and S1520) were provided by Dr S. R. McCouch (Cornell University, USA) and Dr T. Sasaki, RGP (Rice Genome Research Program), National Institute of Agrobiological Resources, Tsukuba, Japan. The detection of polymorphisms was carried out according to Matsubara et al. (2003) and Saitoh et al. (2004).
For the analysis using PCR-based markers, DNA was isolated from a small piece of frozen leaf tissue according to the method of Monna et al. (2002), with slight modifications. Two microsatellite markers (RM589 and RM204) were selected from the public database (http://www.gramene.org). To detect the polymorphism in the wx locus, a primer pair was designed to amplify the 10th intron (Table 1), in which the presence of a retroposon (p-SINE1-r2) is polymorphic (Hirano et al., 1994). In addition, seven cleaved amplified polymorphic sequence (CAPS) markers (R2291, C1496, G8008, E0605P, E1316, E1920, and P18.22) and three single-nucleotide polymorphism (SNP) markers (E0102, E0506, and E21) were designed based on sequences in the public database (accession numbers D24638, D15906, and AP000399). The primers for PCR amplification were listed in Table 1. To detect the polymorphisms in the CAPS markers (R2291, C1496, G8008, E0605P, E1316, E1920, and P18.22), the amplified products were digested with EcoRI, AccI, HinfI, DraI, EcoRI, DraI, and PvuII, respectively. For the SNP markers, the amplified products were sequenced using a Big Dye Terminator Cycle Sequencing Kit (Applied Biosystems, Inc., Foster City, CA, USA) on an ABI 310 automatic sequencer (Applied Biosystems) and SNPs were identified for genotyping. Recombination values were calculated by the maximum-likelihood method (Allard, 1956) and converted to cM using the Kosambi function (Kosambi, 1944).
Table 1. Primers used in this study
Forward primer (5′–3′)
Reverse primer (5′–3′)
Cytological analysis of gametophyte development
Spikelets were collected from panicles before heading. Samples were fixed in FAA (formalin : glacial acetic acid : 70% ethanol = 1 : 1 : 18) and stored in 70% ethanol until use. The percentages of stainable pollen grains were examined at the mature stage before flowering with a potassium iodine solution (I2-KI) and with a solution of 4′,6-diamidine-2′ phenylindole dihydrochloride (1 µg ml−1 DAPI, 0.05 k Tris-HCl pH 7.0, and 0.5% Triton X-100) according to the method of Hihara et al. (1996). Ovaries were dehydrated in a graded ethanol-butanol series, embedded in Paraplast Plus (Oxford Labware, St Louis MO, USA), and then sectioned at 10 µm. Sections were stained with safranin and fast green (Sylvester & Ruzin, 1993) and observed using a light microscope (BH-2, Olympus, Tokyo, Japan).
Effects of the S1 locus in the different backgrounds
A previous study (Sano, 1990) suggested that the formation of fertile male and female gametes was affected differentially by the S1 locus, depending on the genetic background. This prompted us to resolve the siTRD system into the two components: the male and female TRD. For this purpose, siTRD caused by the S1 locus was characterized in terms of km and kf values. NILs carrying a chromosomal segment around the S1 locus derived from W025 in the genetic backgrounds of a ssp. indica strain Acc108 (Acc108S1) or a ssp. japonica strain T65 (T65S1) were used for the analysis. Characterization using molecular markers revealed that Acc108S1and T65S1harbored the segments introgressed from W025 covering regions at least between markers RM589 and R2291 and between RM589 and RZ398, respectively (Fig. 2a).
To analyze siTRD, the two NILs were crossed with their recurrent parents (Acc108 and T65wx). As expected, the seed and pollen fertilities of the F1 hybrids between Acc108 and Acc108S1 were 44.5 and 42.3%, respectively, and those of the F1 hybrids between T65wx and T65S1 were 61.8 and 52.6%, respectively. The recurrent parents (Acc108 and T65wx; genotype, ) were pollinated with these F1 plants to estimate km (Fig. 1a). No pollen-fertile plants () were observed in both crosses (0/185 and 0/213, respectively), indicating that km = 1.0. However, in the F2 populations, the frequency of pollen-semi-sterile () plants differed between the two crosses (Table 2). Assuming km = 1.0, kf was estimated to be 0.97 in Acc108 × Acc108S1 and 0.81 in T65wx× T65S1, based on the frequency of pollen-semi-sterile plants (Table 2, Fig. 1b). This indicated that only the kf value was altered, depending on the genetic background, and the allele was transmitted from female gametes more frequently into the progeny under the background of T65. The difference in the kf value between the two crosses might be the result of the difference in the length of segments introgressed from W025 between Acc108S1 and T65S1. The results of a previous study (Sano, 1990) suggested that the kf value increased when the chromosomal segment of W025 containing the S1 locus became smaller by recombination. However, although Acc108S1 had a shorter chromosomal segment of W025 than T65S1, the kf value was higher in the Acc108 background than in the T65 background (Fig. 1a, Table 2). Therefore, it is likely that the difference in the kf value is the result of difference(s) in factor(s) present in the other regions of the genome between Acc108 and T65, rather than the difference in the length of the introgressed segment.
Table 2. Segregation patterns for pollen fertility and two markers (wx and R2291) in the F2 populations derived from Acc108 × Acc108S1 and T65wx × T65S1
Pollen fertility was examined by I2-KI treatment. Only stainable pollen grains were produced in fertile plants, while about a half of pollen grains were not stained in semi-sterile plants (see text). Two loci (wx and R2291) were genotyped using molecular markers.
G and S indicate homozygotes for the alleles derived from Oryza glaberrima and O. sativa (Acc108 and T65wx), respectively, and H indicates heterozygotes.
Assuming km= 1.0, kf was estimated by no. of pollen-fertile plants/total no. of plants observed in the progeny (see Fig. 1b and text). The recombination values of wx-S1 and S1-R2291 were estimated to be 0.024 and 0.015, respectively, in the Acc108 background, and 0.017 and 0.020, respectively, in the T65 background.
Corresponding to the recombinant T65S1(r) shown in Fig. 2.
By assuming km and kf values, the recombination values between the S1 gene and other markers could be calculated based on the phenotype of pollen fertility in the two F2 populations of T65 or Acc108 genetic backgrounds. Linkage analyses revealed that the S1 gene is located between the wx locus and R2291. By analyzing two additional markers in the cross between Acc108 and Acc108S1, the S1 gene was shown to be present between markers C1496 and G8008 (Fig. 2b).
Cytological analysis of gametophyte development in the S1S1a heterozygote
To examine the effects of the S1 locus on gametophyte development, cytological examination of male and female gametophytes was carried out in heterozygotes. About half of the pollen grains (male gametophytes) were not stained by I2-KI treatment, indicating that these pollen grains were aborted (Fig. 3a). Staining of pollen grains with DAPI revealed that most aborted pollen grains had less than three nuclei, whereas fully developed pollen grains had a vegetative nucleus and two sperm cells (Fig. 3b), suggesting that pollen grains carrying the allele were arrested before the completion of the second mitotic division in microspores. The Wx gene, which is tightly linked to the S1 locus, is expressed in pollen grains and was used as a marker for analyzing the segregation of fertile/aborted pollens. Assuming that no TRD occurs, the Wxwx heterozygote is expected to produce blue (Wx) and reddish brown (wx) pollen grains at a ratio of 1 : 1 when stained with I2-KI. The heterozygote (S1-wx/-Wx) of the F1 hybrid from T65 × T65wxS1 showed a distorted segregation for Wx/wx phenotypes (Fig. 3a). Most of the stained pollen grains showed a reddish brown color, while blue pollen grains appeared at a low frequency, suggesting that pollen grains carrying the Wx allele linked tightly with the allele were frequently aborted. The frequency of Wx pollen grains (0.027) among the fertile pollen grains was consistent with the recombination value (0.024 and 0.017 in Acc108 × Acc108S1 and T65wx× T65S1, respectively) between the wx and S1 loci based on the segregation data in the F2 population (Table 2), which suggested that Wx pollen grains possessed the S1 allele by recombination between them. Thus, it is not likely that the fertile Wx pollen grains resulted from the incomplete penetrance of the S1 allele, which supported the assumption that abortion of male gametophytes possessing the allele is almost complete, showing km = 1.0.
To examine why the allele was more frequently transmitted to the progeny from the heterozygotes () obtained by T65wx× T65S1 than those obtained by Acc108 × Acc108S1, the formation of female gametophytes was compared in the two genetic backgrounds. If kf = 1.0, half of the female gametophytes are also expected to degenerate in the heterozygote (). The developmental pattern of the female gametophyte is referred to as Polygonum-type in rice (Lopez-dee et al., 1999). In the F1 of Acc108 × Acc108S1, out of the 35 ovules examined, 16 had a mature seven-celled structure similar to that found in the parental lines (Fig. 3c), whereas the remaining 19 had ovules without any embryo sac structure (Fig. 3d,e). On the other hand, in the F1 of T65wx × T65S1, out of the 43 ovules examined, 28 had a normally developed embryo sac, whereas the remaining 15 had an arrested embryo sac during megagametogenesis. Most of the arrested embryo sacs contained more than four nuclei (Fig. 3f) or a failure in the polarization of the egg apparatus (Fig. 3g). As shown in Table 3, the frequency of arrested embryo sacs was 35% (15/43) in the F1 of T65wx× T65S1, which deviated significantly from the ratio of 1 : 1 (kf = 1) but fitted to the expectation from kf = 0.81 calculated from the selfed progeny (Table 2). These results indicate that the female gametophytes carrying the allele were partially rescued in the heteozygotes obtained from T65wx× T65S1 and suggest that this accounts for the higher value of kf in the Acc108 background than in the T65 background.
Table 3. Frequencies of normal and abnormal embryo sacs in the heterozygotes () in the Acc108 (Acc108 × Acc108S1) and T65 (T65wx× T65S1) backgrounds
To examine the compound nature of the S1 locus, the region near the S1 locus was dissected using molecular markers. In the F2 population derived from T65wx × T65S1, a recombinant, which was heterozygous at the S1 locus (exhibiting semi-sterility in pollen) and had a segment recombined between the S1 locus and marker R2291, was selected and named T65S1(r) (Table 2). Although T65S1(r) showed semi-sterility in pollen grains, the seed fertility (c. 75%) was markedly higher than expected. The distorted segregation of Wx/wx phenotypes was observed in a manner similar to that shown in Fig. 3a, indicating that km is 1.0 in T65S1(r). However, the selfed progeny of T65S1(r) revealed that pollen-semi-sterile plants () occurred frequently (50/109), suggesting a high rate of transmission of the allele. This was confirmed by the genotypes of the two markers adjacent to the S1 locus (Table 4). Assuming km = 1.0, kf was estimated to be 0.54, which was consistent with a 1 : 1 ratio of S1S1 : (Table 4). These results indicated that the S1 gene in T65S1(r) caused only a preferential transmission through pollen (mTRD), suggesting that the S1 locus comprises at least two elements controlling male and female fertility. Surveys of the flanking markers in T65S1(r) showed that the segment from W025 was shortened by recombination in the region between the markers G8008 and R2291 (Fig. 2a). This implies that an element (or elements) inducing the abortion of female gametes is located in the region between the markers G8008 and S1520 in which the segment from W025 was present in T65S1 but not in T65S1(r).
Table 4. The segregation pattern for pollen fertility and two markers (C1496 and G8008) in the selfed progeny of the recombinant T65S1(r)
G and S indicate homozygotes for the alleles derived from Oryza glaberrima and O. sativa, respectively, and H indicates heterozygotes.
Assuming km= 1.0, kf was estimated by no. of fertile plants/total no. of plants observed in the selfed progeny (see Fig. 1b and text).
Fine mapping of the component of the S1 locus responsible for mTRD
For fine mapping of the major component of the S1 locus that causes mTRD, 33 recombinants between C1496 and G8008 were identified in 2450 segregating plants obtained through self-pollination of the heterozygotes (Fig. 2c). The recombinants were further characterized using seven additional CAPS and SNP markers, which revealed the presence of progressively shortened segments introduced from W025. Four recombinants were selected to delimit the S1 locus (Fig. 2c). When the homozygous recombinants were obtained by selfing, all the plants (A-6-34, A-3-94, T-8-60, and A-1-2-16) were pollen-fertile, and their genotypes were identified as S1S1 based on the pollen-semi-sterility in the F1 plants obtained by crossing these plants with Acc108 or T65wx. These results allowed the delimitation of the major component of the S1locus causing mTRD to a region of approx. 40 kb between the markers E0506 and E1920 (Fig. 2c). In this region, eight open-reading frames (ORFs) that correspond to full-length cDNAs or ESTs comprising four repeated early nodulin genes, a ribosome biogenesis regulatory protein homolog, and three ORFs coding for unknown proteins are present (http://rapdb.dna.affrc.go.jp).
Distribution of the S1 and S1a alleles in Asian and African rice species complexes
To examine the allelic distribution at the S1 locus, 27 strains of Asian (O. sativa–O. rufipogon complex) and African (O. glaberrima–O. barthii complex) rice species complexes were crossed with the tester NILs carrying the S1 or alleles. All the F1 hybrids between the strains of O. sativa or O. rufipogon and the S1 carrier exhibited lower pollen and seed fertilities than the F1 hybrids between these strains and the carrier (Table 5). The extent of the reduction of seed fertility varied depending on the strains used for the test crosses. On the other hand, a marked deviation from 50% production of Wx (or wx) pollen grains was observed in all the F1 hybrids between the strains of the O. sativa or O. rufipogon and the S1 carrier. These results indicate that all the strains of O. sativa–O. rufipogon complex examined carried the allele at the S1 locus, at least for the major component causing mTRD.
Table 5. Allelic distribution at the S1 locus estimated from test crosses between near-isogenic lines (NILs) for the S1 or alleles and 28 strains of Asian (Oryza sativa–O. rufipogon) and African (O. glaberrima–O. barthii) rice species complexes
The strains of African rice species (O. glaberrima and O. barthii) could not be used for the test crosses because most interspecific F1 hybrids between African rice species and O. sativa exhibited nearly complete male sterility (Chu et al., 1969). To determine the allelic state of the African rice species, the Wx–S1region was introduced from C7639 (O. glaberrima) and B19 (O. barthii) into T65wx by backcrosses, and the resultant plants (T65Wx (C7639) and T65Wx (B19)) were used for test crosses. Both the very high frequency of Wx pollen grains and semi-sterility of pollen and seeds in the F1 hybrids between these plants and the carrier indicated that the two strains (C7639 and B19) carried the S1 allele (Table 5). These results suggested that the allelic state at the S1 locus has diverged between the Asian (O. sativa–O. rufipogon complex) and African (O. glaberrima–O. barthii complex) rice species complexes.
Multi-component systems of siTRD involving the S1 locus
Cytological observations confirmed that the siTRD caused by the S1 locus is the result of the aberrant development of both female and male gametophytes carrying the allele in the heterozygote (). Characterization of this siTRD using the two genetic backgrounds revealed that the degree of TRD in female gametes (kf) was altered depending on the genetic background, while the abortion of male gametes carrying the S1a allele is always complete (km= 1). To explain the difference in the kf value depending on the genetic background, one possible assumption is that an unlinked gene(s) modifying the effect of only fTRD exists in the genome and its presence and/or function might be polymorphic within O. sativa. Recent studies on gametogenesis have revealed that a large number of genes are required during the haploid gametophytic phase in plants (Vollbrecht & Hake, 1995; Yang & Sundaresan, 2000; Drews & Yadegari, 2002). Gametic mutations are classified into three groups: male gametophyte-specific, female gametophyte-specific, and general gametophytic mutations (Drews & Yadegari, 2002), suggesting that the development of the female and male gametophytes could be differentially regulated by a variety of haploid-expressed genes. In this study, only fTRD is affected by the genetic background, suggesting that the siTRD system consists of at least two genetic components controlling mTRD and fTRD.
Genetic dissection of the S1 locus using molecular markers also demonstrated that the siTRD system is a multi-component system, in which the fTRD was affected not only by the genetic background of O. sativa but also by a genetic element(s) of O. glaberrima within the gene complex of the S1 locus. Furthermore, a recombinant carrying the chromosomal segment of the presumed region of the genetic element(s) affecting fTRD derived from W025 (G8008 to S1520, Fig. 2a) exhibited full seed fertility (data not shown). This strongly suggested that the S1 locus consisted of the gene(s) that causes mTRD and its modifier(s), which enhances the fTRD only in the heterozygote of the former. In the case of SD of Drosophila and the mouse t-haplotype, TRD is caused by the interaction between the linked genes, the trans-acting distorter and the cis-acting responder, and the modifiers or enhancers are also linked with each other, resulting in the formation of a gene complex on the same chromosome for selective advantages in their transmission (Temin et al., 1991; Silver, 1993). The present results demonstrated that the siTRD system involving the S1 locus is a multi-component system including the interaction of linked genes, which resembles the TRD systems in SD and t-haplotype. This suggests that the siTRD system might not have evolved as a single mutation but by a combination of multiple genetic changes. The results of fine mapping revealed that the major component of the S1 locus causing TRD in male gametes was delimited within a region of c. 40 kb where eight ORFs are annotated. It remains to be determined whether the mTRD is caused by a single gene or interaction of multiple genes similar to the distorter and responder observed in SD and t-haplotype.
TRD system as a reproductive barrier between species
Although whether hybrid sterility is induced by a given locus depends on its allelic state, the pattern of allelic differentiation has been examined in only a few genes (Wu & Palopoli, 1994; Sweigart et al., 2007). In two species of yellow monkeyflower, Mimulus guttatus and M. nastus, a pair of complementally acting loci (hms1 and hms2) was found to cause hybrid male sterility (Sweigart et al., 2006). The survey of intraspecific allelic variation revealed that geographic distribution of the allele causing sterility at one of the two loci (hms1) was extremely restricted, indicating that the potential of the locus for making a reproductive barrier between the two species is limited (Sweigart et al., 2007). By contrast, no intraspecific polymorphisms were observed at the Odysseus locus, which is responsible for hybrid male sterility between Drosophila mauritiana and D. simulans, suggesting that it could act as a reproductive barrier in any pairwise combination between species (Wu & Palopoli, 1994).
The TRD near the S1 locus (near the wx locus) was not observed in crosses between O. sativa and O. glumaepatura (Brondani et al., 2001) or O. longistaminata (Causse et al., 1994; Hu et al., 2003), despite the fact that they are reproductively isolated by sterility barriers. By contrast, strong TRD was frequently reported in the S1 region in the cross combination between O. sativa and O. glaberrima (Yabuno, 1990; Doi et al., 1998; Lorieux et al., 2000). In a survey for allelic distribution at the S1 locus, at least for the major component of the locus causing mTRD, all strains of Asian rice species examined had the allele, while the S1 allele was distributed in three strains (W025, C7639, and B19) of African rice species. No F1 pollen semi-sterility was observed when W025 (O. glaberrima) was crossed with 41 strains of O. glaberrima and 28 strains of O. barthii (formerly O. breviligulata) (Chu et al., 1969), suggesting that there was no allele in the African rice species complex. These results suggest that the S1 and alleles are specific to the Asian (O. sativa–O. rufipogon complex) and African (O. glaberrima–O. barthii complex) rice species complexes, respectively. Thus, the S1 locus potentially causes F1 hybrid sterility in any pairwise combinations between these two species complexes, which leads to a reproductive barrier between rice species.
Inferring the evolutionary origin of the siTRD system
A question arises as to how the genes for siTRD, which are responsible for reproductive barriers, could be maintained in a species despite the fact that they have deleterious effects that reduce the fitness of individuals. One possible explanation is that stepwise mutations allow the development of reproductive barriers without reducing the fitness of individuals (Nei et al., 1983). Rick (1966, 1971) reported that there were three alleles (Gep, Gec, and Gen) at the Ge locus in tomato, and suggested that the Gep allele, which causes the abortion of male and female gametes carrying the Gec allele in the heterozygote, might have arisen from the neutral allele Gen, which induced no abortion in the heterozygotes with either of the other two alleles. This resembles the siTRD system caused by the S6 locus detected in a hybrid between Asian cultivated rice and its wild ancestor, which also consists of three alleles, including a neutral allele (our unpublished data). In the case of a siTRD system in wheat, the presence of a suppressor that alleviates gametic abortion was reported (Tsujimoto & Tsunewaki, 1985). However, no suppressor or neutral allele () was detected in the rice species examined, regarding the siTRD caused by the S1 locus.
The alternative assumption is that the selfish genetic elements might play an important role in the origin of reproductive barriers (Hurst & Werren, 2001; Coyne & Orr, 2004). Frank (1991) and Hurst & Pomiankowski (1991) indicated that sterility barriers could evolve by a mutual imbalance between TRD (MD in a broad sense) systems. According to their scenario, mutations causing the TRD, which are more likely to evolve on sex chromosomes than autosomes, can increase in frequency because of their selfish nature by killing the gametes carrying their homologous alleles, but the effect of the mutation should be often suppressed because of a fertility cost on their bearers and the deleterious effect of a biased sex-ratio. As a result, two populations might evolve different TRD systems and an unmasking of normally suppressed TRD might yield sterility in hybrids. Recent empirical work in Drosophila supported the hypothesis for the contribution of the TRD system (sex-ratio drive) to the rise of the reproductive barriers (Dermitzakis et al., 2000; Tao et al., 2001; Juiter et al., 2004; Orr et al., 2006).
In plants, in which there are no sex chromosomes in most species, it has not been known how selfish genetic elements causing TRD are maintained in populations and whether such elements are involved in the formation of reproductive barriers. In this study, dissection of the S1 locus revealed that the major component of the locus causes mTRD, which made us assume that the siTRD system has arisen from mTRD. In rice, male sterility appears more frequently than female (or embryo sac) sterility in F1 hybrids of a number of intra- and interspecific cross combinations (Chu et al., 1969). The fact that the male gamete is frequently affected in TRD systems has been explained by the notion that the wastage of male gametes has a smaller effect on individual fecundity (Lyttle, 1991). In plants, genes that cause mTRD or male semi-sterility could not be eliminated quickly from populations, especially from those comprising perennials, since partial male sterility only has a small effect on fecundity (Stebbins, 1977; Sano, 1983). As a result, a vast number of haploid genomes can be screened by selection acting on male gametes in heterozygotes, and genetic changes are achieved at a cost relatively lower than that required for the selection of individuals (Mulcahy, 1979; Walbot & Evans, 2003).
The maintenance of TRD system in a population may also be affected by mating system of plants. Unlike most animal species, plant species sometimes have a self-pollination system, which makes it readily possible to yield homozygotes. This allows the gene responsible for mTRD to be maintained in the population even in the absence of the suppressor gene(s), which reduces heterozygote disadvantages. Since both the Asian and African rice species complexes consist predominantly of self-pollinating plants (Oka, 1988), the homozygotes can frequently appear in the populations. The distribution of the component of the S1 locus responsible for mTRD showed that all strains of Asian rice species examined had the allele and the S1 allele may also predominate in African rice species. Although it remains to be studied what evolutionary forces have caused the fixation of the and S1 alleles in Asian and African rice species complexes, respectively, the present results suggest that the TRD system caused by the S1 locus plays a significant role in the reproductive barriers between rice species.
We thank Drs S. R. McCouch and T. Sasaki for providing the molecular markers, and Y. Kishima, H. Nagano, N. Sawamura, and R. Suzuki for their suggestions and assistance. This work was supported in part by grants from the Ministry of Education, Culture, Sports, Science and Technology and the Ministry of Agriculture, Forestry and Fisheries of Japan (integrated research project for plant, insect and animal using genome technology GD-2001).