Findings

A wild abortive (WA)-type CMS has been almost exclusively used for breeding three-line hybrid rice and contributes to 10% of the total rice cultivated area worldwide (Li et al. [2007]; Barclay [2010]). Because of the great impact of WA-type CMS on agriculture, many studies have attempted to elucidate the CMS-causing gene in WA mitochondria and to determine the fertility restorer genes for WA-CMS. In 2013, a mitochondrial orf352 (WA352) gene that confers WA-CMS was discovered; this gene encodes 352-amino-acids protein (Luo et al. [2013]). We also found a sequence variant of orf352 in an RT102 CMS line derived from O. rufipogon (Okazaki et al. 2013). However, the Rf genes in nuclear genome have not yet been cloned, although two major Rf genes, Rf3 and Rf4, have been mapped on the chromosomes 1 and 10, respectively (Lu et al. [1997]; Yao et al. [1997]; Tan et al. [2008]; Jing et al. [2001]; Ahmadikhah and Karlov [2006]; Ngangkham et al. [2010]; Suresh et al. [2012]). In this study, we report the cloning of Rf4. We also showed that the cloned Rf4 reduced the orf352-containing transcripts and restored pollen fertility.

To identify the Rf4 gene for WA-CMS, we performed map-based cloning of Rf4 using a cultivar IR24 (Additional file 1: "Methods"), because this cultivar is known to be a strong restorer line for WA-CMS (Jing et al. [2001]). We delimited the Rf4 candidate region between two SSR markers, SSR1045 and AT801, that corresponded to a 213-kb region of Nipponbare genome (Additional file 2: Table S1). While we were conducting the fine mapping, Ngangkham et al. ([2010]) reported that Rf4 was located in a region between RM6737 and RM6100 with a distance of 104 kb in the Nipponbare genome, which further narrowed down the candidate region. We isolated a bacterial artificial chromosome (BAC) clones covering the corresponding region from IR24 genomic libraries (Figure 1) and determined their nucleotide sequences (Additional file 1: "Methods"). Because the presence of Rf4 reduced the orf352-containing transcripts in the mitochondria, we identified the candidate gene according to the following two criteria, (i) it would encode a protein whose function is related to RNA metabolism, and (ii) it would encode a mitochondrial-targeting protein. The candidate genes that fulfilled the criteria were four new pentatricopeptide repeat (PPR)-encoding genes, PPR454, PPR782a, PPR782b, and PPR458 (Figure 1). These genes were named on the number of amino acids that they encoded. PPR proteins are, in general, known to be involved in RNA regulation in the mitochondria and plastids (Schmitz-Linneweber and Small [2008]). We obtained genomic fragments containing PPR454, PPR782a, PPR782b, or PPR458 from the IR24 BAC clone (Additional file 1: "Methods"; Additional file 3: Figure S1). We introduced each genomic fragment into a WA-CMS line, WAA, which carries a Taichung 65 nuclear background, and obtained at least twelve transgenic lines (Additional file 4: Figure S2).

Figure 1
figure 1

Candidate genes present in the IR24 BAC clone. Candidate genes and SSR markers in the IR24 BAC clone are indicated.

Full size image

The anthers of WA-CMS are milky white, slender, and stunted (Additional file 4: Figure S2) and contain shrunken pollen grains lacking starch accumulation ability, because of which they are not stained with 1% potassium iodide (Figure 2a). In contrast, anthers of fertile Taichung 65 are yellow and engorged and contain darkly stained pollen grains. In four of the thirteen transgenic plants with PPR782a, restorations of anther and pollen development and starch accumulation in pollen grains were observed (Figure 2a; Additional file 4: Figure S2). The percentage of stainable pollen grains (pollen stainability) was 48% for plant Nos. 9, 10, and 12, and 79% for plant No. 13 (Figure 2b). Especially, one of the plants (No. 13) showed 47% seed setting rate (Figure 2c). Restorations of anther and pollen development were not observed in the remaining nine plants. Introduction of three other genomic fragments, which contained PPR782b, PPR454, or PPR458, did not recover the anther and pollen morphology (Additional file 4: Figure S2), and resulted in 0% of pollen stainability and no seed setting. These data indicate that Rf4 is the PPR782a gene.

Figure 2
figure 2

Restoration of fertility in transgenic plants with PPR782a . (a) Pollen grains stained with iodine-potassium iodide. (b) Pollen stainability of each transgenic plant. Error bars indicate the standard deviation. (c) Seed setting of the transgenic plant, No. 13.

Full size image

T1 seeds were set by self-pollination. The progeny segregated into 8 fertile plants, all of which carried the introduced gene, and 7 male-sterile plants, all of which did not carry the introduced gene (Additional file 5: Figure S3). The appearance of null segregants indicated that PPR782a controlled the fertility restoration sporophytically. The pollen fertility was not completely restored by the introduced Rf4 gene. Other fertility restorer genes, such as the Rf3 gene on chromosome 1, might be necessary for fully restoration of WA-CMS (Suresh et al. [2012]).

The predicted amino acid sequence of PPR782A carries 18 repeats of PPR motif (Additional file 6: Figure S4) and is highly similar to RF1A for BT-CMS showing 86% identity (Kazama and Toriyama [2003]; Akagi et al. [2004]; Komori et al. [2004]; Kazama et al. [2008]). Amino acid identity between PPR782A and PPR782B was 94%, with a completely identical region in the N- and C-terminal ends. A non-restorer line, Nipponbare, contained a putative allele of PPR782a, which is encoded by the Rice Annotation Project (RAP) locus ID Os10g0495200. The amino acid sequence of PPR782A_Nipponbare shows 95% identity to that of PPR782A_IR24 (Additional file 6: Figure S4). Some amino acid substitutions might be crucial for the function of the PPR protein, as was reported for other PPR-type Rf genes in petunia (Bentolila et al. [2002]) and radish (Brown et al. [2003]; Desloire et al. [2003]; Koizuka et al. [2003]).

Expression of Rf4 was assessed using RT-PCR and gene-specific primers (Additional file 1: "Methods"). Rf4 was expressed weakly in leaf blades. Expression was increased in the course of anther development, with the highest levels noted in the anthers at the tri-cellular pollen stage (Figure 3). Rf4 has been proposed to express highly in the tapetum, because the fertility restoration was sporophytically controlled. Our results of expression analysis indicated that Rf4 might play a gametophytic role in the maturation of pollen grains after tapetum degradation, as well as a sporophytic role.

Figure 3
figure 3

PPR782a is expressed in various tissues with the highest expression in mature anthers. RT-PCR analysis of PPR782a in anthers at the meiotic stage (Mei), uni-nucleate microspore stage (Uni), bi-cellular pollen stage (Bi), and tri-cellular pollen stage (Tri), and in the leaf blades. PCR results obtained using genomic DNA is also shown.

Full size image

In a recent report on the WA-CMS-associated mitochondrial gene, orf352 (WA352), RF4 was proposed to function post-transcriptionally, leading to the degradation of orf352-containing transcripts, half of which were co-transcribed with the upstream gene rpl5 encoding the ribosomal protein large subunit 5 (rpl5-orf352 transcripts; Figure 4a), whereas RF3 acted post-translationally and suppressed translation of orf352 (Luo et al. [2013]). To determine whether the cloned Rf4 reduced rpl5-orf352 transcripts, we performed northern blot analysis by using RNA extracted from the leaf blades (Additional file 1: "Methods"). In a WA-CMS plant, WAA, we detected two strong signals of 4.5- and 2.7-kb bands corresponding to the rpl5-orf352 and orf352 transcripts, respectively (Figure 4b), as reported (Luo et al. [2013]). In a fertility restorer line, WAR, the signals of both the bands almost disappeared (Figure 4b). In four transgenic plants (Nos. 9, 10, 12, and 13) showing recovery of anther morphology and pollen stainability, the amount of the two transcripts decreased (lanes 9, 10, 12, and 13; Figure 4b). These four plants showed higher expression of the introduced PPR782a gene as detected by RT-PCR (Figure 4c). Northern blot analyses performed using cox1 and atp9 probes revealed that the introduction of PPR782a did not affect the accumulation of these transcripts (Additional file 7: Figure S5). These results suggested that Rf4 was involved in the reduction of orf352-containing transcripts.

Figure 4
figure 4

Reduction of orf352 -containing RNA in transgenic plants with PPR782a . (a) Genomic structure around orf352 and orf352-containing transcripts. orf352 is located between rpl5 and nad5 _exon4 _exon5. orf 352 is transcribed as both 4.7-kb and 2.5-kb RNA. Horizontal lines represent both transcripts containing orf352. (b) Northern blot analysis of RNA isolated from leaf blades of a cytoplasmic male sterility (CMS) line (WAA), a fertility restorer line (WAR), and transgenic WAA plants with PPR782a. Staining of ribosomal RNA is shown as a loading control. (c) Expression of the introduced PPR782a and tubulin detected by RT-PCR.

Full size image

RF4 is considered to recognize and bind orf352-containing transcripts and promote the degradation of orf352-containing transcripts to avoid ORF352-mediated premature programmed cell death and consequent male sterility (Luo et al. [2013]). Finding a nuclear Rf factor provide novel insights into reconciliation between mitochondria and nuclei in agronomically important crops, and has practical implications for production of hybrid rice.

During preparation of this manuscript, Tang et al. ([2014]) have reported a short letter entitled as "The rice restorer Rf4 for wild-abortive cytoplasmic male sterility encodes a mitochondrial-localized PPR protein that functions in reduction of WA352 transcripts". Their study turned to be essentially the same to ours. They used a PCR-amplified genomic clone of cv. Minghui 63 for a complementation test, but did not investigate the self-pollinated progeny. In contrast, our study used the genomic clones isolated from BAC libraries of IR24 and examined the segregation in the T1 plants, showing that the fertility restoration was controlled sporophytically (Additional file 5: Figure S3). The nucleotide sequence of the RF4 allele of IR24 in our study is completely identical to that presented in Supplementary Figure of Tang et al. ([2014]). An adjacent gene, PPR782b, identified in our study has not been reported in their study. Thus our study is not just a confirmation of Tang et al. ([2014]), but further provides more information.

Accession codes

The nucleotide sequences of PPR454, PPR782a, PPR782b, and PPR458 have been deposited at the DDBJ under accession numbers [AB900791 to AB900794].

Authors' contributions

TK performed the experiments and drafted the manuscript. KT designed and supervised the study and revised the manuscript. Both authors read and approved the final draft of the manuscript.

Additional files