Evaluation of the fertility restoration capability of C418
To examine the ability of C418 to restore fertility in WA-type japonica CMS lines, we determined the pollen grain and spikelet fertility levels of individual plants of two WA-CMS lines and the testcross F1 hybrids obtained from the crosses between two WA-type CMS lines and C418. WA-NipA and WA-LqxA exhibited amorphous aborted pollen grains, and the natural spikelet levels of these two CMS lines were 0, indicating that the WA-type japonica CMS lines were completely sterile (Fig. 1a, b). The pollen stainability of the testcross F1 plants was > 95%, and the natural spikelet fertility levels of these testcross F1 plants ranged from 10.86–64.63%, indicating that plants carrying the WA-type CMS cytoplasm produced morphologically normal pollen, although partially stained pollen grains may lack the ability to germinate (Fig. 1). Also, the average natural spikelet fertilities of the WA-LqxA/C418 F1 and WA-NipA/C418 F1 plants were statistically different (24.63% ± 7.70% vs. 54.06% ± 8.47%, P = 4.24E-6). These results indicate that the restorer line C418 is able to partially restore fertility to the WA-type japonica CMS lines tested in this study, and that fertility restoration was influenced by the subtypes of the CMS line nuclear backgrounds.
Genetic analysis of fertility restoration and gene mapping
In order to explore the mechanism underlying fertility restoration, the pollen grains and natural spikelet fertility levels of 160 plants in the WA-NipA//Nip/C418 population were investigated. Based on the pollen grains, the plants in the WA-NipA//Nip/C418 population could be divided into two categories: (1) the plants resembled the CMS lines with degenerated anthers and shrunken pollen grains and were sterile, and (2) the plants were similar to the testcross F1 hybrids with normal anthers and dark-stained pollen grains and were partially fertile. The segregation ratio of sterile plants to partially fertile plants was 1:1 (χ2 = 0.21, which was < χ20.05 = 3.84), indicating that fertility restoration is conditioned by one dominant restorer gene (Table 1). The natural spikelet fertilities of the sterile plants were < 3.88%, and the natural spikelet fertilities of the partially fertile plants ranged from 0 to 61.92% and showed a continuous variation. These results indicate that there might be one or more minor-effect Rf genes in C418 that affect fertility restoration in WA-type CMS lines.
Table 1
Fertility segregation in the F2 populations and three-cross F1 progeny plants
Population | Total plants | Partially fertile plants | Sterile plants | Observed ratio |
WA-NipA//Nip/C418 | 160 | 88 | 72 | 1.22:1 |
WA-NipA/C418 F2 | 350 | 343 | 7 | 49:1 |
WA-LqxA/C418 F2 | 254 | 251 | 3 | 83.67:1 |
Based on the breeding pedigrees of BT-type japonica hybrids in China, most BT-type restorer lines should theoretically carry the major restorer gene Rf4 (Yang et al. 2016; Kazama and Toriyama, 2014). In order to map the target Rf gene in C418, we first needed to determine whether Rf4 is present in C418. We sequenced the Rf4 allele from C418, and the nucleotide sequence of Rf4 in the C418 was identical to that from IR24, demonstrating that it carried the functional Rf4 allele. Subsequently, we tested whether Rf4 is related to the fertility restoration of WA-type CMS by genetic linkage analysis. A previous study identified two InDel marker loci, STS10-27 and STS10-16, that flank Rf1/Rf4 on Chromosome10, and these markers show polymorphisms between C418 and Nip. We used these two markers to genotype all plants in the WA-NipA//Nip/C418 population. All 88 partially fertile plants showed heterozygous genotypes at these two loci, while all 72 sterile plants were homozygous for the NIP alleles at the two marker loci. These results indicate that the Rf gene in C418 that is responsible for the fertility restoration of WA-type CMS is most probably Rf4.
The Rf4 pollen grains are favored in fertilization in the testcross F1 plants
A total of 350 and 254 plants in the WA-NipA/C418 F2 and WA-LqxA/C418 F2 populations, respectively, were grown to maturity, and the pollen grains and natural spikelet fertilities of all F2 plants were quantified. Based on the fertility criteria (Materials and methods), there were 343 and 251 partially fertile plants, and seven and three sterile plants in the WA-NipA/C418 F2 and WA-LqxA/C418 F2 populations, respectively (Table 1). These results indicate that the fertility restoration in WA-type CMS japonica rice is sporophytic, but the segregation ratio of partially-fertile plants to sterile plants did not fit a one-gene (3:1) sporophytic model. To explore the reasons behind the aberrant segregation of fertility restoration, we used the markers STS10-27 and STS10-16 to genotype all of the F2 plants. In the WA-NipA/C418 F2 population, there were 192 plants with the Rf4Rf4 genotype and 151 plants with the Rf4rf4 genotype, all of which were partially fertile, while the seven sterile plants had the rf4rf4 genotype. Similarly, 251 partially fertile plants in the WA-LqxA/C418 F2 population were Rf4-containing plants, including 137 plants with the Rf4rf4 genotype and 114 plants with the Rf4Rf4 genotype, and three sterile plants carried the rf4rf4 genotype. Obviously, the genotypes at the Rf4 locus deviated from the expected 1:2:1 segregation ratio in the two F2 populations for a common single gene model. However, plant phenotypes were completely consistent with their genotypes, indicating that Rf4 is indeed correlated with the fertility restoration of WA-type CMS.
In considering the aberrant fertility and genotypic segregation ratios, we hypothesized that both Rf4 and rf4 pollen grains in the testcross F1 plants are able to restore fertility in WA-type japonica CMS lines, but that the rf4 pollen grains are less competitive than the grains carrying Rf4. To test this hypothesis, we further constructed the WA-NipA//WA-NipA/C418 population that consisted of 37 plants, and the pollen grains and genotypes of these plants were investigated. We found that 35 plants had > 90% stained pollen grains and the Rf4rf4 genotype, while two plants showed > 95% shrunken pollens and had the rf4rf4 genotype. This distorted segregation indicated that the majority of male gametes in the WA-NipA/C418 F1 plants involved in pollination should carry the Rf4 gene. Taken together, these results demonstrate that the pollen grains carrying Rf4 in the testcross F1 plants have different viability compared to grains that do not carry Rf4, and that Rf4 pollen grains are preferentially selected during fertilization.
Development of NIL Rf4 in the ‘Nipponbare’ japonica genetic background
In an attempt to identify the effects of Rf4 on the fertility restoration of WA-type CMS in japonica lines, we developed a near-isogenic line homozygous for Rf4 in the Nip genetic background using marker-assisted selection (Fig. 2a). In 2017, a BC5F3 line was obtained that was very similar to Nip in terms of the main biological characteristics (Fig. 2b). In addition, the genetic background of the BC5F3 line was analyzed using the 40K rice SNP-array, which showed that the recurrent parent genome recovery was 98.45% in this line, and the C418 chromosome segment containing Rf4 had been successfully transferred into Nip. Thus, we successfully developed an Rf4 NIL, which we named NILRf4, and it was suitable for further study.
The effects of Rf4 on the fertility restoration of WA-type japonica CMS lines
We next examined the pollen grains following I2-KI staining and determined the natural spikelet fertility levels of WA-NipA/NILRf4 F1 and WA-LqxA/NILRf4 F1 plants. All F1 plants had dark-staining pollen grains, and there were no differences between the different crosses (data not shown). The natural spikelet fertility levels of the testcross F1 hybrids from WA-NipA and NILRf4, and WA-NipA and NILRf4 were all < 1%. These results indicate that Rf4 has minor effects on the fertility restoration of WA-type japonica CMS lines. In the WA-NipA/NILRf4//NILRf4 population, there were 14 plants with the Rf4rf4 genotype and 19 plants with the Rf4Rf4 genotype, and the average natural spikelet fertility levels of the plants were 4.40% and 31.32%, respectively (Fig. 3c). These results imply that Rf4 exerts a dosage effect on the fertility restoration of WA-type japonica CMS lines.