Genetic Mapping of the Gamete Eliminator Locus, S 2 , Causing Hybrid Sterility and Transmission Ratio Distortion Found between Oryza sativa and Oryza glaberrima Cross Combination

: Hybrid sterility is a reproductive barrier that prevents gene ﬂow between species. In Oryza species, some hybrid sterility loci, which are classiﬁed as gamete eliminators, cause pollen and seed sterility and sex-independent transmission ratio distortion ( si TRD) in hybrids. However, the molecular basis of si TRD has not been fully characterized because of lacking information on causative genes. Here, we analyze one of the hybrid sterility loci, S 2 , which was reported more than forty years ago but has not been located on rice chromosomes. Hybrids between African rice ( Oryza glaberrima ) and a near-isogenic line that possesses introgressed chromosomal segments from Asian rice ( Oryza sativa ) showed sterility and si TRD, which conﬁrms the presence of the S 2 locus. Genome-wide SNP marker survey revealed that the near-isogenic line has an introgression on chromosome 4. Further substitution mapping located the S 2 locus between 22.60 Mb and 23.54 Mb on this chromosome. Signiﬁcant TRD in this chromosomal region was also observed in a calli population derived from cultured anther in hybrids of another cross combination of African and Asian rice species. This indicates that the pollen abortion caused by the S 2 locus occurs before callus induction in anther culture. It also suggests the wide existence of the S 2 -mediated si TRD in this interspeciﬁc cross combination. Chromosomal location of the S 2 locus will be valuable for identifying causative genes and for understanding of the molecular basis of si TRD. class these results, we two F 3 families self-pollination of the two recombinant classes. The F 3 plants were further classiﬁed by genotype of a marker linking to the S 2 locus 2C). If the S 2 locus is located between markers RM3643 and S2_1, low pollen and seed fertilities are expected in plants which are heterozygous in this chromosomal region. The result showed that all F 3 plants had high pollen and seed fertilities compared to the heterozygotes. These results conﬁrmed that the S 2 locus is located between S2_1 and S2_10.


Introduction
African cultivated rice species, O. glaberrima Steud, are a useful genetic resource for improving Asian cultivated rice species (O. sativa L.). However, severe seed and pollen sterility is observed in F 1 hybrids derived from these species [1][2][3]. Such a severe hybrid sterility prevents the transfer of useful genes in O. glaberrima to O. sativa during breeding.
In Oryza species, numerous loci for hybrid sterility that fit in the "single locus sporogametophytic interaction model" have been reported [4][5][6][7]. In this model, loci are classified in to three classes, namely pollen-killer, egg-killer, and gamete-eliminator, based on sexspecificity in their actions [7]. Pollen-killers and egg-killers abort male and female gametes with one of the two alleles, respectively. However, gamete eliminators abort both male and female gametes with one of the two alleles, which causes sex-independent transmission ratio distortion (siTRD) in progenies of hybrids. Although the existence of pollen-killers or egg-killers is frequently reported, reports on gamete-eliminators are relatively rare.
In the interspecific cross between O. glaberrima and O. sativa, more than 11 loci for hybrid sterility have been reported [6,8]. Among them, only three loci, S 1 , S 2 , and S37(t) are reported as gamete eliminators [5,[9][10][11]. Recent studies identified the causative genes for a gamete eliminator, the S 1 locus [12][13][14]. Xie et al. [12] and Koide et al. [13] reported the involvement of peptidase-domain containing protein coding genes in S 1 -locus-mediated hybrid sterility. Additionally, Xie et al. [14] recently reported the involvement of an additional gene that constitutes a tripartite system in S 1 locus-mediated hybrid sterility. However, causative genes remain unclear for the other gamete eliminator loci. Therefore, overall mechanisms causing a severe sterility in hybrids between O. glaberrima and O. sativa remain unknown.
Here, we identify the chromosomal location of the gamete eliminator locus, S 2 . Sano et al. [5] suggested the existence of genetic factor(s) causing both seed and pollen sterility in the F 1 derived from the cross between W025 (a strain of O. glaberrima) and a nearisogenic line (NIL) developed by successive backcrossing with W025 to Acc108 (a strain of O. sativa). They termed the genetic factor inducing hybrid sterility in this cross as the S 2 locus. The two alleles in the S 2 locus, S 2 g , derived from O. glaberrima and S 2 s , derived from O. sativa, were assumed (S 2 g and S 2 s were previously denoted S 2 and S 2 a , respectively [5]. To reduce the confusion among the locus name and the allele name, we renamed them as S 2 g and S 2 s .). In heterozygotes (S 2 g /S 2 s ), male and female gametes carrying the S 2 g allele are preferentially aborted, causing about 50% sterility in pollens and seeds. In addition, only gametes with the S 2 s allele are transmitted to the next generation, which causes siTRD in later generations. Although Sano et al. [5] suggested the presence of the S 2 locus by NILs, chromosomal location of the S 2 locus is unknown, which limits molecular studies on this locus. The present study confirms results obtained by Sano et al. [5] and identifies the chromosomal location of the S 2 locus using genome-wide SNP typing and DNA marker surveys.

Genetic Stocks
The W025 African rice (Oryza glaberrima Steud) strain and the Pehkuh (denoted as Acc108) Asian rice (O. sativa L.) strain were used. A NIL, developed by Sano et al. [5], W025S2s, was also used. W025 and Acc108 harbor S 2 g and S 2 s alleles at the S 2 locus, respectively. W025S2s harbors the S 2 s allele at the S 2 locus introduced from Acc108 in the genetic background of W025. For genetic mapping, a total of 150 F 2 plants and two F 3 families derived from the W025 and W025S2s cross were used. To examine the genotype of plants derived from the anther culture, we used 27 calli derived from the anther culture of hybrids developed by crossing Nipponbare (a strain of O. sativa) and IRGC 104038 (denoted as WK21, a strain of O. glaberrima).

DNA Marker Survey
For the SNP survey, 1468 KASP markers designed to detect polymorphism between African and Asian rice species were used [15]. The amplification and inflorescence detection were conducted by Kbioscience/LGC. A total of 1182 KASP markers, which can determine the genotypes of SNP in both W025 and W025S2s, were used for analysis. Among the 1182 KASP markers, 26 showed polymorphism between W025 and W025S2s. If two neighboring SNPs are both polymorphic, we considered the region between them to be an SNP cluster region. To confirm the detailed chromosomal region in which the NIL has an introgression from the donor, two SSR markers and seven InDel markers in the SNP cluster region were used (Supplementary Table S1).

Genetic Mapping Using a Segregating Population
To genetically map the S 2 locus, a total of 150 F 2 plants were used. These F 2 plants were developed by self-pollination of the F 1 plants derived from the W025 and W025S2s cross. The F 2 plants were grown in a greenhouse in the Research Faculty of Agriculture, Hokkaido University, Sapporo, Japan, in 2018 and 2020. The plants were grown in a shortday field (10 h light and 14 h dark) after sowing to prevent delay of heading because of the Agriculture 2021, 11, 268 3 of 10 plants' photosensitivity. Genomic DNA was isolated from 2-week-old seedlings using a simple method [16]. Genotypes of markers in the SNP cluster region of each F 2 plant were determined using two SSR markers and seven InDel markers described above. To confirm the location of the S 2 locus, two F 3 families were developed using self-pollination of two F 2 plants that have a segment recombined between markers. The genotypes of F 3 families were also determined by markers in the SNP cluster region.
To investigate whether the effect of the S 2 was also observed in calli derived from anther culture, we determined the genotype of anther-culture derived calli. For genotyping, a marker S2_4, which links to the S 2 locus, was used. The procedure of callus induction is based on that of Kanaoka et al. [8].

Seed and Pollen Fertility
The panicle fertility of each F 2 and F 3 plant was determined by counting fertile and sterile seeds in a panicle. The panicle-fertilities of two or three panicles in a plant were examined and the plant's maximum panicle-fertility was used for the plant's seed fertility. To examine the pollen fertility, spikelets were collected from panicles before flowering. The percentages of stainable and normal size pollen grains were examined with a potassium iodine solution (I 2 -KI). Non-stainable or small size pollen grains were classified as sterile pollen. For a plant, pollen fertility of at least two spikelets was examined and their maximum values were used as the plant's pollen fertility.

The S 2 Locus Causes Pollen and Seed Sterility in Heterozygotes
To confirm the genetic action of the S 2 locus, we examined pollen and seed fertilities of heterozygous plants (S 2 g /S 2 s ). Two parents, W025 and the NIL W025S2s showed 85.6% and 91.2% pollen fertility, respectively. However, the F 1 plants derived from the cross between the parents showed significantly lower pollen fertility (60.2%) than parents ( Figure 1). W025 and W025S2s showed 87.8% and 80.7% seed fertility, respectively, while their F 1 plants showed significantly lower seed fertility (60.8%) than parents ( Figure 1). These results indicated that genetic factors reduce pollen and seed fertilities in F 1 plants derived from the W025 and W025S2s cross. This result was consistent with one expected from the genetic action of gamete-eliminator. Because we used the same materials as Sano et al. [5], we considered this genetic factor to be the S 2 locus.

Chromosomal Regions Introgressed into W025S2s
Because W025 has fertile pollens and seeds in self-pollination, the reduction of pollen and seed fertilities in F 1 plants was caused by the S 2 locus located in the chromosomal segment introgressed into the W025S2s. Therefore, we conducted genome-wide SNP marker survey to locate introgressed segments of a donor variety, Acc108, in the genetic background of W025. Genotypes of 1182 SNP loci in W025 and W025S2s were determined using the KASP marker system [15]. Among the 1182 SNP loci, 26 loci showed polymorphism between W025 and W025S2s, which indicates the existence of introgressed segments (Table S2). Seven of the 26 polymorphic SNP loci are located between 21.45 Mb and 22.73 Mb on chromosome 4 as neighboring SNPs ( Figure 2A). Additionally, an SNP locus (id4006835) located closely to these (21.12 Mb on chromosome 4) also showed polymorphism between W025 and W025S2s ( Figure 2A). Therefore, we considered the chromosomal region (from 21.12 Mb to 22.73 Mb on chromosome 4) to be an SNP cluster. Because there was only one SNP cluster identified by a genome-wide survey, we focused this region for further investigation.

Chromosomal Regions Introgressed into W025S2s
Because W025 has fertile pollens and seeds in self-pollination, the reduction of pollen and seed fertilities in F1 plants was caused by the S2 locus located in the chromosomal segment introgressed into the W025S2s. Therefore, we conducted genome-wide SNP marker survey to locate introgressed segments of a donor variety, Acc108, in the genetic background of W025. Genotypes of 1182 SNP loci in W025 and W025S2s were determined using the KASP marker system [15]. Among the 1182 SNP loci, 26 loci showed polymorphism between W025 and W025S2s, which indicates the existence of introgressed segments (Table S2). Seven of the 26 polymorphic SNP loci are located between 21.45 Mb and 22.73 Mb on chromosome 4 as neighboring SNPs ( Figure 2A). Additionally, an SNP locus (id4006835) located closely to these (21.12 Mb on chromosome 4) also showed polymorphism between W025 and W025S2s ( Figure 2A). Therefore, we considered the chromosomal region (from 21.12 Mb to 22.73 Mb on chromosome 4) to be an SNP cluster. Because there was only one SNP cluster identified by a genome-wide survey, we focused this region for further investigation.  To confirm and delimit the introgressed segment on chromosome 4 in the W025S2s, we used two SSR markers and nine Indel markers around the SNP cluster. Five of the 11 markers that are located in the region between 21.28 Mb and 23.50 Mb on chromosome 4 showed polymorphism between W025 and W025S2s (Supplementary Figure S1). In these five markers, polymorphisms were not observed between W025S2s and Acc108 ( Figure  S1). These results indicate that a segment of Acc108 was introgressed into W025S2s in this chromosomal region. However, the other six markers that are located on 20.10 Mb or in the region between 23.54 Mb and 26.99 Mb on chromosome 4 did not show polymorphism between W025 and W025S2s (Supplementary Figure S1). Because there were polymorphisms between W025 and Acc108 in these six markers, these results indicated that W025S2s does not have introgressed segments in these regions. Therefore, we concluded that W025S2s possesses the introgressed segment from Acc108 in the region at most between 20.10 Mb and 23.54 Mb on chromosome 4. To confirm and delimit the introgressed segment on chromosome 4 in the W025S2s, we used two SSR markers and nine Indel markers around the SNP cluster. Five of the 11 markers that are located in the region between 21.28 Mb and 23.50 Mb on chromosome 4 showed polymorphism between W025 and W025S2s (Supplementary Figure S1). In these five markers, polymorphisms were not observed between W025S2s and Acc108 ( Figure S1). These results indicate that a segment of Acc108 was introgressed into W025S2s in this chromosomal region. However, the other six markers that are located on 20.10 Mb or in the region between 23.54 Mb and 26.99 Mb on chromosome 4 did not show polymorphism between W025 and W025S2s (Supplementary Figure S1). Because there were polymorphisms between W025 and Acc108 in these six markers, these results indicated that W025S2s does not have introgressed segments in these regions. Therefore, we concluded that W025S2s possesses the introgressed segment from Acc108 in the region at most between 20.10 Mb and 23.54 Mb on chromosome 4.

Transmission Ratio Distortion Observed in the F2 Population
To examine whether the S 2 locus is in the introgressed segment on chromosome 4, we developed the F 2 population and determined the genotype of the individual plant in the population ( Figure 2B). The S 2 locus is assumed to be the factor for hybrid male and female sterility induced via preferential abortion of male and female gametes with the S 2 g allele in heterozygotes [5]. Therefore, if hybrid sterility induced by the S 2 locus occurs, TRD in the flanking region of the S 2 locus is expected in progenies of heterozygotes.
We observed a significant TRD in the F 2 population derived from the W025 and W025S2s cross in 2018 and 2020 (Table 1). In the marker, RM16991, the ratio of homozygotes of the W025S2s-derived allele:heterozygotes:homozygotes of the W025-derived allele was 134:16:0, which shows significant segregation distortion (p < 0.01) towards the excess of W025S2s allele. Stronger distortion was also observed in the marker S2_2, S2_4 and S2_5 (Table 1). These results indicated that a genetic factor for TRD exists in the introgressed segment in W025S2s. We also analyzed pollen and seed fertility of the 95 F 2 plants grown in 2020 ( Figure 3). All 90 plants homozygotes of the W025S2s-derived allele in the marker S2_4 showed higher pollen and seed fertility than any heterozygote (Figure 3). This result indicated that the factor for hybrid sterility is also located in this region. Because the S 2 locus is considered to be a factor for hybrid sterility causing TRD, these results strongly indicate that the S 2 locus is in this chromosomal region (between 20.10 Mb and 23.54 Mb on chromosome 4).

Substitution Mapping of the S 2 Locus
Using four individuals that harbor recombination points between markers RM16991 and S2_5, we further narrowed down the candidate region of the S 2 locus ( Figure 2B). High pollen and seed fertilities (92.4% and 89.3%, respectively) were observed in individuals homozygous for W025S2s-derived allele for all five markers (F2_s in Figure 2B). Conversely, low pollen and seed fertilities (53.1% and 48.2%, respectively) were observed in the heterozygote (F2_h in Figure 2B). These results show that hybrid sterility occurred in the heterozygote. In two recombinant classes (recombinant class 1 and 2), pollen and seed fertilities were higher than those of heterozygotes (94.4% and 90.2% for pollen fertility and 86.4% and 89.5% for seed fertility, respectively), though only one individual was obtained for the recombinant class 2. For further confirmation of these results, we developed two F 3 families derived from the self-pollination of the two recombinant classes. The F 3 plants were further classified by genotype of a marker linking to the S 2 locus ( Figure 2C). If the S 2 locus is located between markers RM3643 and S2_1, low pollen and seed fertilities are expected in plants which are heterozygous in this chromosomal region. The result showed that all F 3 plants had high pollen and seed fertilities compared to the heterozygotes. These results confirmed that the S 2 locus is located between S2_1 and S2_10.
plants grown in 2020 ( Figure 3). All 90 plants homozygotes of the W025S2s-derived allele in the marker S2_4 showed higher pollen and seed fertility than any heterozygote ( Figure  3). This result indicated that the factor for hybrid sterility is also located in this region. Because the S2 locus is considered to be a factor for hybrid sterility causing TRD, these results strongly indicate that the S2 locus is in this chromosomal region (between 20.10 Mb and 23.54 Mb on chromosome 4).

Transmission Ratio Distortion Observed in Anther-Culture Derived Calli
Kanaoka et al. [8] showed that, in calli derived from the anther culture, TRD is not observed for some loci for hybrid sterility, though it is observed in the F 2 population. This phenomenon was assumed to occur if callus induction begins before pollen abortion; genes for hybrid sterility are unlikely to function in microspores after the initiation of callus induction. Although Kanaoka et al. [8] examined the effect of eleven hybrid sterility loci in calli derived from anther culture in hybrids between O. sativa and O. glaberrima, the S 2 locus was not included in the analysis. To investigate whether TRD caused by the S 2 locus in the F 2 population is also observed in calli derived from anther culture, we genotyped 27 calli derived from the cultured anther in hybrids between Nipponbare and WK21. Of these, 24 and 3 plants are O. sativa homozygote and O. glaberrima homozygote for the marker, S2_4, respectively (Table 2). Because a significant distortion was observed in the marker linked to the S 2 locus, we concluded that TRD was induced by the S 2 locus also in the anther-culture derived calli.

Discussion
The S 2 locus was first reported by Sano et al. [5], who found that NILs with O. glaberrima genetic background showed pollen and seed semi-sterility when they are crossed with a strain of O. glaberrima. They also reported restoration of fertility in the progenies of the above cross (Sano et al. [5] described them as BnF2 plants). Based on these observations, they hypothesized that sterility observed in F 1 plants between the NIL and O. glaberrima was caused by the locus acting in the mode of "single locus sporo-gametophytic interaction model". They assumed this locus as the S 2 locus [5].
In this study, we also used the same materials described in Sano et al. [5] and confirmed that pollen and seed sterility occur in the F 1 derived from the NIL (W025S2s) and W025 cross (Figure 1). This result indicated that hybrid sterility observed in this study is caused by the S 2 locus. Genome-wide SNP marker survey and genetic mapping showed that the The present study showed that TRD also occurred in the population derived from anther culture. Kanaoka et al. [8] suggested that if callus induction begins before pollen abortion induced by the gene for hybrid sterility, TRD does not occur in the population derived from anther culture. The present result that a significant distortion was observed in the anther-culture-derived population suggested that pollen abortion induced by the S 2 locus begins before callus induction. Because the timing of callus induction is from uninucleate to two nuclei stages in the development of pollen, these results indicated that the S 2 locus induces pollen abortion before/in the two nuclei stage during pollen development. To confirm the timing of pollen abortion, detailed cytological analysis is necessary. For the anther culture, we used hybrids derived from Nipponbare and WK21. This result suggested that the S 2 locus induces hybrid sterility not only in the specific pairwise-cross combination between Asian and African rice varieties, but also in the broader pairwise-cross combinations between these two species. To confirm this, a survey of allelic distribution is necessary.
Mechanisms causing/maintaining a strong reproductive barrier between species are a long-standing interest in evolutionary biology. In the genus Oryza, it is plausible that severe sterility arises from accumulation of hybrid sterility loci. Although the overall effect on sterility would increase with the increasing number of hybrid sterility loci, it is also affected by the phase and strength of linkage between hybrid sterility loci ( Figure 4). If preferentially-transmitting alleles in two different hybrid sterility loci (that is, their alternative alleles do not transmit because of the gametes' selective abortion) link in the coupling phase, the overall effect on hybrid sterility decreases with their linkage strength ( Figure 4A). Conversely, if these two alleles link in the repulsion phase, the overall effect on hybrid sterility increases with the loci's linkage strength ( Figure 4B). Therefore, a difference in the direction of gametic selection in hybrid sterility loci might have a role for arising/maintaining a severe sterility barrier between species. However, a bias in the use of the genetic background species may cause an asymmetric detection of hybrid sterility loci with the specific direction of gametic selection.  Generally, it is common to use the NILs with the genetic background of O. sativa, an Asian cultivated rice species, because this species is easier to handle and maintain than other related species of Oryza. In the case of S2, preferential abortion of gametes carrying the S2 g allele derived from O. glaberrima occurs. Therefore, in the progeny of heterozygotes (S2 g /S2 s ), almost all plants are homozygotes for O. sativa-derived allele (S2 s ) at the S2 locus (Table 1; Figure 3). In such a case, it is difficult to develop NILs containing the S2 g allele at the S2 locus by backcrossing O. sativa to heterozygotes, just as in general cases. The NIL used in the present study was developed by backcrossing O. glaberrima to heterozygotes to introduce the S2 s allele at the S2 locus in the genetic background of O. glaberrima. Such NILs with an "alternative" genetic background enabled detailed genetic analysis of the S2 locus. Currently, four loci for hybrid sterility in Oryza have been reported as gamete eliminator type [7,11,17]. In these four loci, all loci except for the S2 locus cause preferential abortion of the gametes with the allele derived from O. sativa. Further development of NILs with genetic background of species other than O. sativa may help us to detect unknown loci for gamete eliminator that potentially contribute to severe sterility barriers between species.
The present study confirmed the existence of the S2 locus causing the abortion of male and female gametes with the allele derived from O. glaberrima. In addition to the S2 locus, the presence of the S1 locus causing preferential abortion of the gametes with the allele derived from O. sativa has been confirmed by using the same materials [5,9,13,18]. Koide et al. [18] showed that a gamete eliminator locus S1 is a compound locus of pollen-killer and its modifier affecting the abortion of female gametes, though such a compound effect might depend on the genetic background [18]. These studies have raised the question of how such a (compound) locus with a strong sterility effect appears during/after speciation in Oryza. The information of causal genes of the S2 locus might also shed light on the evolutionary process of gamete eliminators in Oryza. Generally, it is common to use the NILs with the genetic background of O. sativa, an Asian cultivated rice species, because this species is easier to handle and maintain than other related species of Oryza. In the case of S 2 , preferential abortion of gametes carrying the S 2 g allele derived from O. glaberrima occurs. Therefore, in the progeny of heterozygotes (S 2 g /S 2 s ), almost all plants are homozygotes for O. sativa-derived allele (S 2 s ) at the S 2 locus (Table 1; Figure 3). In such a case, it is difficult to develop NILs containing the S 2 g allele at the S 2 locus by backcrossing O. sativa to heterozygotes, just as in general cases. The NIL used in the present study was developed by backcrossing O. glaberrima to heterozygotes to introduce the S 2 s allele at the S 2 locus in the genetic background of O. glaberrima. Such NILs with an "alternative" genetic background enabled detailed genetic analysis of the S 2 locus. Currently, four loci for hybrid sterility in Oryza have been reported as gamete eliminator type [7,11,17]. In these four loci, all loci except for the S 2 locus cause preferential abortion of the gametes with the allele derived from O. sativa. Further development of NILs with genetic background of species other than O. sativa may help us to detect unknown loci for gamete eliminator that potentially contribute to severe sterility barriers between species.
The present study confirmed the existence of the S 2 locus causing the abortion of male and female gametes with the allele derived from O. glaberrima. In addition to the S 2 locus, the presence of the S 1 locus causing preferential abortion of the gametes with the allele derived from O. sativa has been confirmed by using the same materials [5,9,13,18]. Koide et al. [18] showed that a gamete eliminator locus S 1 is a compound locus of pollenkiller and its modifier affecting the abortion of female gametes, though such a compound effect might depend on the genetic background [18]. These studies have raised the question of how such a (compound) locus with a strong sterility effect appears during/after speciation in Oryza. The information of causal genes of the S 2 locus might also shed light on the evolutionary process of gamete eliminators in Oryza.

Conclusions
The S 2 locus induced both pollen and seed sterility in hybrids of O. sativa and O. glaberrima. We have mapped this locus between 22.60 Mb and 23.54 Mb on rice chromosome 4 using the segregating population. Although the causative gene(s) remains unidentified,