Genetic analysis of male sterility obtained from a rice cultivar Lebed backcrossed with Taichung 65

Male sterility is a useful agronomic trait for breeding of self-pollinating crops and is often observed in the progenies of hybrids of distantly related species, for example, Oryza sativa L. subsp. indica and O. sativa L. subsp. japonica. To explore new male sterile lines in rice, we performed successive backcrosses using a japonica cultivar Taichung 65 (T65) as a recurrent pollen parent and various indica cultivars as the initial female parents. We observed male sterile plants in the backcross progeny from an indica cultivar, Lebed. Both fertile and sterile plants were present in the BC4F1 generation. The sterile plants segregated for fertile and sterile plants when backcrossed with T65 in BC5F1, BC6F1 and BC7F1 with a ratio of 1:1. Conversely, all the backcross progenies from the fertile BC4F1 were consistently fertile. Anthers of the male sterile line were stunted and did not shed pollen; cross-sectional observations revealed defects in sporophytic cells. The male sterility appears to be caused by heterozygous alleles derived from T65 and Lebed. A male sterility gene was mapped between two INDEL markers on the long arm of chromosome 10, which corresponded to a 407 kb region in the Nipponbare genome. Since the heterozygous Lebed allele acts as dominant sporophytic pollen killer, it would be useful for recurrent selection breeding of japonica rice.


Findings
Male sterility is a useful agronomic trait for the production of F 1 hybrids in self-pollinating crops, eliminating laborious emasculation (Huang et al. 2014). Male sterility is often observed in the progenies of hybrids of distantly related species, for example, hybrids of indica group rice (Oryza sativa L. ssp. indica) and japonica group rice (O. sativa L. ssp. japonica). Such intersubspecific hybrid male sterility is thought to be controlled by interaction between indica alleles and japonica alleles of several loci such as Sa, Sb, Sc, Sd, and Se (Ouyang et al. 2010). Male sterility can also be caused by incompatibility between mitochondrial and nuclear genomes in nuclear/cytoplasm substitution lines (Huang et al. 2014), referred to as cytoplasmic male sterility.
To explore new cytoplasmic male sterile lines in rice, we produced many cytoplasm substitution lines of a japonica cultivar Taichung 65 (T65) by successive backcross, with various rice cultivars as the initial female parent and T65 as a recurrent pollen parent. During this process, we found a male sterile line in the backcross progeny from an indica cultivar, Lebed, from the Philippines, which was provided by the National Institute of Agrobiological Sciences Genebank (Tsukuba, Japan) as WRC23 (Kojima et al. 2005).
An F 1 hybrid between Lebed and T65 was backcrossed four times with T65. The resulting BC 4 F 1 generation contained two sterile plants and four fertile plants. The progeny of a fertile plant backcrossed with T65 were all fertile (Additional file 1: Table S1). In contrast, segregation of fertile and sterile plants was observed in the BC 5 F 1 generation obtained from a sterile plant backcrossed with T65. Seed setting rates of each plant indicated that four plants were fertile, while six plants were sterile (Additional file 1: Table S2). When the sterile line was backcrossed with T65, sterile and fertile plants were always segregated in BC 5 F 1, BC 6 F 1, and BC 7 F 1 generations, with an overall ratio of 18:14, corresponding to a theoretical 1:1 ratio (χ 2 = 0.5, p > 0.05). This indicates that the male sterility was caused by heterozygous alleles derived from T65 and Lebed in a single locus. We named the male sterility gene LEBED-TAICHUNG 65 MALE STERILITY (LTMS), and designated the sterile plant as the LTMS line.
A day before flowering, anthers were whitish yellow and stunted in the LTMS line, while those of T65 were yellow and engorged (Fig. 1a). Fig. 1b shows anthers protruding from florets just after flowering. No pollen grains were shed from anthers of the LTMS line after flowering. In contrast, empty anthers, following the shedding of pollen grains, were evident in T65 (Fig. 1b). These phenotypes of anthers, indehiscence and dehiscence, were used to identify sterile and fertile plants, respectively, for the mapping of the LTMS gene. Microscopic observation of anthers and pollen grains revealed that pollen grains were not stained with 0.1% potassium iodide in the LTMS line, while they were darkly stained in T65, indicating no starch accumulation in the pollen grains of the LTMS line ( Fig. 1c and d). The anthers of F 1 hybrids between Lebed and T65 contained both starch-engorged and empty pollen grains, showing a seed setting rate of 36.2%, as against a 51.3% seed setting rate in Lebed (Additional file 2: Figure S1). Relatively higher seed setting rate and pollen stainability of the F 1 plant (Additional file 2: Figure S1), compared with those of the LTMS line, suggests that Lebed and the F 1 plant might carry an epistatic restorer gene to suppress the function of the LTMS gene. It is interesting to note that the LTMS phenotype was only manifested in progenies backcrossed with T65.
Anthers at the meiotic stage, the early and late uninucleate microspore stages, the bicellular pollen stage, and the tricellular pollen stage were fixed with a formaldehyde-acetic acid-ethanol solution, and embedded in Technovit 7100 resin (Kluzer, Wehrheim, Germany) as described by Zhou et al. (2011). Transverse sections (7 μm) were stained with 0.05% toluidine blue (Fig. 2). In anthers of T65, microspores developed into pollen grains full of starch. In contrast, pollen grains were empty in the LTMS lines. Morphological differences between T65 and the LTMS line were not evident until the early uninucleate microspore stage, and development ceased at the late uninucleate microspore stage in the LTMS line. Observation of anther walls revealed that the tapetum began to degenerate in T65 at the late uninucleate microspore stage, whereas tapetal, middle layer, and endocecium cells were clearly observed in the LTMS line. The middle layer and endocecium cells of the LTMS line were enlarged at the tricellular pollen stage, whereas they were degraded in T65. These results indicate that defects in sporophytic cells cause pollen abortion.
The genome of the LTMS line in the BC 5 F 1 generation was expected to consist predominantly of the T65 Table 1 Chromosomal position of markers, and genotypes, anther phenotypes, and seed setting rates in BC 6 F 1 recombinant plants  Table S5). These results indicate that the LTMS gene is located between KNJ8-indel702 (IRGSP-1.   Table S6 showed variable seed setting rates in panicles. We also noticed that some panicles of the LTMS lines occasionally set a few seeds (Additional file 1: Table S2). We do not know the reason of this occasional recovery of seed setting. Six recombinant plants were further investigated using additional INDEL markers. Four plants carrying heterozygous Lebed /T65 genotypes at KNJ8-indel707 and KNJ8-indel710 were shown to be male sterile (Table 1). We conclude that the LTMS gene is located between KNJ8-indel704 (IRGSP-1.0 position: 11,269 kb) and C5-indel8296 (IRGSP-1.0 position: 11,676 kb) in chromosome 10, corresponding to a 407 kb-region in the Nipponbare genome.
Based on the integrated rice science database, Oryzabase, only a single gene is listed as a hybrid sterility gene on chromosome 10-F1 POLLEN STERILITY 18 (S18). The S18 locus has been reported in backcross progenies of japonica (Oryza sativa L. 'T65') and African rice (O. glaberrima Steud., Acc. IRG104038) as a new locus affecting F 1 pollen sterility, and shown to be linked to RFLP markers G1084 and R1629 (Doi et al. 1998). The map position corresponds to a Nipponbare genomic region between IRGSP-1.0 position 10,637 kb and 12,757 kb, which includes the candidate region of the LTMS gene. It is possible that the LTMS gene is identical to S18, although the originated species are different; the LTMS gene from indica rice (O. sativa) and the S18 gene from African rice (O. glaberrima). Efforts are now in progress for molecular cloning of the LTMS gene.
Several hybrid sterility genes/QTL have been reported in hybrids between O. sativa subsp. japonica and subsp. indica. For F 1 pollen sterility; for example, Sa in chromosome 1, Sb (S24) in chromosome 5, Sc in chromosome 3, Sd (S35) in chromosome 1, and Se (S25) in chromosome 12 were identified through a series of allelic testcrosses (Guo et al. 2016). Pollen grains carrying japonica alleles of Sa, Sb, Sc, Sd or Se are known to cause selective pollen abortion in F 1 heterozygous plants, resulting in the predominant transmission of indica alleles to their progeny. Molecular cloning has been reported for the Sa and Sc genes (Long et al. 2008, Shen et al. 2017. The Sa locus comprises two adjacent genes, SaM and SaF, encoding a small ubiquitin-like modifier E3 ligase-like protein and an F-box protein, respectively (Long et al. 2008). Male semisterility occurred only when heterozygous SaM alleles (SaM + /SaM − ) and at least one SaF allele (SaF + /Saf − or SaF + /SaF + ) were present, causing pollen abortion only in pollen grains carrying the SaM − allele. It is possible that the LTMS locus is also composed of several adjacent genes. Shen et al. (2017) reported that the japonica allele of the Sc gene (Sc-j) corresponds to Os03g0247300, which encodes a 446 amino-acid protein with a DUF1618 domain. Sc-j has been shown to be essential for pollen development. It has been reported that Sc-i contains two or three tandem-duplicated Sc-j-homologs with distinct promoters, and the high expression of Sc-i in Sc-j/Sc-i hybrids causes suppression of Sc-j expression in pollen and selective abortion of Sc-j-pollen (Shen et al. 2017). The gene dosage-dependent allelic suppression has been proposed as a mechanism of hybrid sterility.
Hybrid pollen sterility is caused by the reciprocal gene loss of duplicated genes has also been reported. In the F 1 hybrid between a japonica cultivar, Nipponbare, and an indica cultivar, Kasalath, pollen sterility is caused by interaction between DOPPELGANGER1 (DPL1; LOC_ Os01g15448) and DOPPELGANGER2 (DPL2; LOC_ Os06g08510). The Kasalath allele of DPL2 and the Nipponbare allele of DPL1 have been reported to be nonfunctional, and pollen carrying the two defective DPL alleles fails to germinate (Mizuta et al. 2010). Molecular cloning of another hybrid sterility gene has been reported in F 1 hybrids between O. sativa ssp. japonica and O. glumaepatula. Hybrid sterility is governed by interaction of the S28 gene on chromosome 4 and the S27 gene on chromosome 8. The reciprocal loss of the duplicated genes encoding mitochondrial ribosomal protein L27 has been reported to cause pollen sterility (Yamagata et al. 2010). Notably, the above-mentioned pollen sterility genes act as gametophytic pollen killers, whilst the LTMS gene identified in our study acts as a sporophytic pollen killer.
Although most genic male sterility is caused by lossof-function alleles of genes essential for anther and pollen development (see Wang et al. 2013 for reviews), two dominant genic male sterile mutants have also been reported in rice. The Pingxiang dominant male-sterile gene was designated as Ms-p and mapped on the SSR markers RM311 and RM3152 on chromosome 10 (Huang et al. 2007). We noticed that the mapping position of Msp corresponds to a Nipponbare genomic region between IRGSP-1.0 position 19,120 kb to 19,416 kb, which is near, but different from, the position of the LTMS gene in our study (Table 1). Molecular cloning of the Ms-p gene has not yet been reported. The gene for Sanming dominant