A rice nuclear gene, Rf-1, restores the pollen fertility disturbed by the BT-type male sterile cytoplasm, and is widely used for commercial seed production of japonica hybrid varieties. Genomic fragments carrying Rf-1 were identified by conducting chromosome walking and a series of complementation tests. Isolation and analysis of cDNA clones corresponding to the fragments demonstrated that Rf-1 encodes a mitochondrially targeted protein containing 16 repeats of the 35-aa pentatricopeptide repeat (PPR) motif. Sequence analysis revealed that the recessive allele, rf-1, lacks one nucleotide in the putative coding region, presumably resulting in encoding a truncated protein because of a frame shift. Rice Rf-1 is the first restorer gene isolated from cereal crops that has the property of reducing the expression of the cytoplasmic male sterility (CMS)-associated mitochondrial gene like many other restorer genes. The present findings may facilitate not only elucidating the mechanisms of male sterility by the BT cytoplasm and its restoration by Rf-1 but also isolating other restorer genes from cereal crops, especially rice.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Cytoplasmic male sterility (CMS) is a maternally inherited phenotype characterized by the inability of a plant to produce functional pollen. It has been observed in more than 150 plant species including agriculturally important crops (Laser and Lersten, 1972). The CMS phenotype is thought to be caused by aberrant mitochondrial genes that are often chimeric in structure (Hanson, 1991).
Nuclear genes termed restorers of fertility (Rf) function to suppress the CMS phenotype. They often alter the expression of CMS-associated genes, and thus presumably reduce or remove the deleterious effects caused by CMS-associated genes. Systems of CMS and Rf have been utilized in hybrid seed production because they eliminate the need for laborious hand emasculation. Furthermore, the systems provide excellent models for the study of nuclear–mitochondrial interactions in multicellular organisms. However, limited information is available on the molecular mechanisms of sterilization and restoration. In order to understand the mechanisms, three restorer genes, Rf2 of maize (Cui et al., 1996), Rf of Petunia (Bentolila et al., 2002) and Rfk1 (Rfo) of radish (Brown et al., 2003; Desloire et al., 2003; Koizuka et al., 2003), have been cloned so far. Maize Rf2 was proved to encode an aldehyde dehydrogenase (Liu et al., 2001). On the other hand, Petunia Rf and radish Rfk1 (Rfo) were demonstrated to encode a protein composed of 14 repeats and 16 repeats of the 35-aa pentatricopeptide repeat (PPR) motif, respectively. The fact that the three restorer genes encode two different kinds of protein is consistent with previous observations regarding their restoration mechanisms; maize Rf2 does not reduce the accumulation of the CMS-associated protein URF13 (Dewey et al., 1987), while Petunia Rf and radish Rfk1 (Rfo) decrease the CMS-associated PCF protein (Nivison and Hanson, 1989) and the CMS-associated ORF125 protein (Koizuka et al., 2003), respectively.
In rice, the cytoplasm of an indica variety, Chinsurah Boro II, caused male sterility under the nuclear genome of a japonica variety, Taichung 65, and was named BT cytoplasm or cms-bo (Shinjyo, 1975). The BT cytoplasm in combination with the dominant and gametophytic restorer gene, Rf-1, has been widely used for seed production of japonica hybrids. It has been proposed that Rf-1 is involved in processing of the CMS-associated transcripts (Akagi et al., 1994; Iwabuchi et al., 1993; Kadowaki et al., 1990).
For better understanding of fertility restoration, we decided to isolate rice Rf-1 by means of map-based cloning. Our previous study (Komori et al., 2003) revealed that the Rf-1 locus was located between S12564 Tsp509I and C1361 MwoI, and that the genetic distance from S12564 Tsp509I to Rf-1 was 0.1 cM. Considering that the average physical length per centimorgan is 280 kbp in rice (Harushima et al., 1998), we decided to isolate Rf-1 by using the S12564 Tsp509I locus as a start point of the map-based cloning.
In this study, we conducted further fine-scale mapping of the Rf-1 locus, constructed an overlapping contig spanning the Rf-1 candidate region with lambda phage clones, and identified genomic fragments carrying Rf-1 by a series of complementation tests. Furthermore, we isolated cDNA clones corresponding to the genomic fragments and identified the coding region of the Rf-1 gene. The deduced amino acid sequence was compared with those of maize Rf2 (Cui et al., 1996), Petunia Rf (Bentolila et al., 2002) and radish Rfk1 (Rfo; Brown et al., 2003; Desloire et al., 2003; Koizuka et al., 2003).
Screening of recombinants
Using two near-isogenic lines (MS-Koshihikari and FR-Koshihikari) with their recurrent parent (Koshihikari), a three-way cross, MS-Koshihikari (BT, rf-1/rf-1)//FR-Koshihikari (BT, Rf-1/Rf-1)/Koshihikari (Normal, rf-1/rf-1), was conducted as described in our previous report (Komori et al., 2003). FR-Koshihikari inherits the Rf-1 region including S12564 Tsp509I and C1361 MwoI from the Rf-1 donor (IR8, indica), and Rf-1 is located between the two loci (Komori et al., 2003). To screen recombinants carrying a recombination point near the Rf-1 locus, the 4103 individuals obtained by the three-way cross were subjected to genotyping for S12564 Tsp509I and C1361 MwoI following the method previously described by Komori et al. (2003). Among the 4103 plants, one and six plants were homozygous for Koshihikari allele at S12564 Tsp509I and C1361 MwoI, respectively. The other plants were heterozygous at both of the two loci. As the F1 plants (FR-Koshihikari/Koshihikari) have the BT cytoplasm, only pollens carrying Rf-1 can develop normally in the F1 plants, and the 4103 plants are expected to be heterozygous at the Rf-1 locus. Therefore, the results described above mean that the one and six plants were recombinants carrying a recombination point between Rf-1 and S12564 Tsp509I, and between Rf-1 and C1361 MwoI, respectively. In our previous study (Komori et al., 2003), three recombinants between S12564 Tsp509I and C1361 MwoI had been found. Thus, a total of 10 recombinants were obtained and subjected to the subsequent fine-scale mapping of the Rf-1 locus.
Chromosome walking and fine-scale mapping
As summarized in Figure 1(a), chromosome walking was conducted with genomic clones of Asominori (japonica, rf-1/rf-1) and IR24 (indica, Rf-1/Rf-1), which inherits Rf-1 from IR8 as well as FR-Koshihikari. Clones named WSA, WSE, WSF, and WSG were screened from the genomic library of Asominori with Probe A, Probe E, Probe F and Probe G, respectively. In a similar way, clones named XSE, XSF, XSG, and XSH were screened from the genomic library of IR24 with Probe E, Probe F, Probe G, and Probe H, respectively. Nucleotide sequences of the walked regions were determined and utilized to develop eight new PCR-based markers (Table 1). The 10 recombinants were subjected to genotyping for the new marker loci in order to monitor the genetic distance from the Rf-1 locus. Results of the walking and fine-scale mapping are illustrated in Figure 1(b). Based on the idea that the 10 plants were Rf-1/rf-1 as mentioned above, a plant homozygous for Koshihikari allele at a certain marker locus was considered to have a recombination point between Rf-1 and the marker locus. The number of plant(s) with a recombination point between Rf-1 and each of the marker loci was two at the beginning of the walking (S12564 Tsp509I and M1239 MboI), decreased to zero as the walking proceeded (M6227 BslI, M20680 MboI, M45461 TaqI, M49609 BstUI and M57629 BsaJI), and increased to one after further walking (M66267 XbaI and M69331 BstZ17I). These genetic results implied that the Rf-1 locus was between M1239 MboI and M66267 XbaI. As shown in Figure 1(a), the contig constructed with genomic clones of IR24 fully covered the Rf-1 locus, and each overlapping region was more than 4.7 kbp long. Therefore, at least one clone within the contig seemed to include the Rf-1 gene in full length.
Table 1. Summary of PCR-based markers developed in this study
Forward primer (5′–3′)
Reverse primer (5′–3′)
After pre-heating for 2 min at 94°C, 30 PCR cycles (1 min at 94°C, 1 min at 58°C, and 1 min at 72°C), followed by 2 min at 72°C.
After pre-heating for 2 min at 94°C, 35 PCR cycles (0.5 min at 94°C, 0.5 min at 58°C, and 0.5 min at 72°C), followed by 2 min at 72°C.
After pre-heating for 2 min at 94°C, 30 PCR cycles (1 min at 94°C, 1 min at 61°C, and 1 min at 72°C), followed by 2 min at 72°C.
To identify which clone in the IR24 contig contains the entire Rf-1 gene, a series of complementation tests was conducted. Ten restriction fragments (ctE1, ctE7, ctF4, ctF18-del, ctF18, ctG22, ctG16, ctG8, ctG8H18, and ctH18; Figure 1c) derived from the nine clones constituting the IR24 contig were introduced to MS-Koshihikari, according to the subcloning strategy summarized in Table 2, in order to test for their ability to restore fertility. More than 40 transgenic plants (T0) for each construct were grown in a greenhouse and judged with respect to their seed fertility after the maturity period. Some of the transgenic plants carrying ctG16, a 15.6-kbp fragment derived from XSG16, were fertile (Figure 2a). On the other hand, all of the transgenic plants of the other nine constructs remained sterile as MS-Koshihikari. As regards the transgenic plants carrying ctG16, the numbers of filled seeds and empty seeds were counted for two standard panicles per plant, and the seed set percentage was calculated for each plant. As shown in Figure 2(b), the seed set percentage of 47 plants ranged from 0 to 87%. Although the degree of fertility restoration differed among plants possibly because of incomplete introduction of ctG16 in some plants and/or the transgenic process such as in vitro culture, 39 plants out of 47 became fertile to some degree (>10% seed set), and 3 plants showed a similar seed set percentage (>80%) to F1 plants (85%) obtained by a cross, MS-Koshihikari/FR-Koshihikari, and FR-Koshihikari (91%). Hence, it was experimentally proved that ctG16 includes the Rf-1 gene in a functional structure.
Table 2. Strategy of vector construction for functional complementation tests
Among the fertile transgenic plants (T0) carrying ctG16, some were proved to carry a single copy of the introduced fragment. One of such T0 plants was self-pollinated, and the T1 progeny (12 plants) was grown and subjected to the pollen fertility analysis. The rate of fertile pollens that became dark blue upon iodine staining was close to 100% like FR-Koshihikari in five plants and around 50% like F1 of MS-Koshihikari/FR-Koshihikari in seven plants (Figure 2c). The seed set percentage was more than 80% in all of the T1 plants (data not shown). Considering that the male sterility of the BT cytoplasm is restored by Rf-1 in a gametophytic manner (Shinjyo, 1969), the above results suggested that the introduced fragment was stably inherited to the T1 generation and restored the BT-CMS just like the endogenous Rf-1 gene.
In order to identify the genomic region of the Rf-1 gene more precisely, additional complementation tests were conducted for two restriction fragments, ctG16-sub6.8 and ctG16-sub11.4 within ctG16 (Figure 1c; Table 2). Results revealed that both of the fragments had the ability to restore fertility as well as ctG16 (data not shown).
Isolation of cDNA clones
To identify expressed sequences in ctG16, a cDNA library of FR-Koshihikari was screened using three genomic fragments as probes (Probe X, Probe Y, and Probe Z; Figure 3). When screened with Probe X and Probe Y, eight clones were detected with both of the two probes, and their partial nucleotide sequences were determined. By comparing their sequences with the sequence of ctG16, six clones were proved to be cDNA clones corresponding to ctG16 and were named XY clones. When screened with Probe X and Probe Z, 12 clones were detected with both of the two probes. Sequence analysis revealed that six clones were cDNA clones corresponding to ctG16 and were named XZ clones. Full sequences of XY and XZ clones revealed the structure of the Rf-1 gene (Figure 3). Although a great diversity was observed with respect to splicing patterns and polyadenylation sites, it generated no variance at the deduced amino acid level because of the existence of a stop codon in the first exon. To determine whether the 3′ untranslated region could play an important role in expressing the gene function of Rf-1, a 4.2-kbp fragment supposed to contain the promoter region and the coding region of the Rf-1 gene (Figure 3) was fused to the Nos-terminator and introduced to MS-Koshihikari. More than 80% of seed set percentage was observed in some transgenic plants (T0), and the frequency distribution of seed set percentage was similar to that for the 15.6-kbp fragment of ctG16. This result suggested that adequate expression of the predicted coding region is sufficient to restore fertility.
Sequence analysis of the Rf-1-encoded protein
The Rf-1 gene was predicted to encode a 791-aa protein containing 16 PPR motifs, 14 of which are in tandem from Ser192 to Gly681 (Figure 4a), and was therefore designated as PPR791. The consensus motif derived from the 16 PPRs is similar to that derived from 1303 PPRs (Small and Peeters, 2000; Figure 4b), indicating that the Rf-1-encoded protein, PPR791, is a typical member of the PPR proteins. Furthermore, the consensus motif shows a quite high level of homology with that derived from the deduced proteins encoded by Petunia Rf (Bentolila et al., 2002) and radish Rfk1 (Rfo; Brown et al., 2003; Desloire et al., 2003; Koizuka et al., 2003; Figure 4b). PPR791 was identical to rice PPR8-1, which was regarded as a candidate of the Rf-1-encoded protein by a recent report (Kazama and Toriyama, 2003).
mitoprot analysis revealed that PPR791 has a mitochondrial targeting sequence whose putative cleavage site is Gly27. Mitochondrial targeting sequences have been predicted in the deduced proteins encoded by Petunia Rf (Bentolila et al., 2002), radish Rfk1 (Rfo; Brown et al., 2003; Desloire et al., 2003; Koizuka et al., 2003) and maize rf2 (Liu et al., 2001). Comparison of the targeting sequences among restorer gene products revealed that the rice Rf-1-encoded protein shares the N-terminal seven amino acids with the maize rf2-encoded protein (Figure 4c).
Analysis of the genomic region including the Rf-1 locus
The genomic sequences of the contigs shown in Figure 1(a), 76 kbp long for IR24 and 45 kbp long for Asominori, were subjected to similarity search against the sequence of XY4 (Figure 3), one of the isolated cDNA clones for the Rf-1 gene. The results revealed the genomic structures of the region including the Rf-1 locus as illustrated in Figure 5. First, the Asominori allele at the Rf-1 locus or PPR791 had 1-bp and 574-bp deletions as compared to the corresponding region of IR24. The 1-bp deletion causes a frame shift and generates a stop codon in front of the 574-bp deletion, resulting in a truncated 266-aa protein instead of the 791-aa protein encoded by Rf-1. Second, IR24 harbored three genes highly homologous to Rf-1 within 25 kbp upstream of Rf-1. They were predicted to encode 794-aa protein with 17 PPRs, 683-aa protein with 16 PPRs, and 762-aa protein with 17 PPRs, and were therefore named PPR794, PPR683, and PPR762, respectively. According to mitoprot analysis, all of the putative PPR proteins contained a mitochondrial targeting sequence. The cDNA clones for the PPR794 gene and the PPR762 gene, whose deduced proteins are 89.2 and 84.2% identical with PPR791, respectively, were unintentionally isolated when the cDNA library of FR-Koshihikari was screened with Probe X and Probe Y in search of cDNA clones for the Rf-1 gene, indicating that the two Rf-1 homologous genes are expressed in young panicles of FR-Koshihikari. Analysis of the Asominori sequence flanking the Rf-1 locus revealed that Asominori has a 7947-bp deletion and has no allele at the PPR762 locus. In the Asominori allele at the PPR683 locus, two deletions and one insertion are present, and as a result, a 401-aa protein is predicted. In contrast, the Asominori allele at the PPR794 locus is likely to encode a 794-aa protein that is 99.8% identical with the predicted gene product of the IR24 allele.
Our study demonstrated that the rice Rf-1 gene encoded a 791-aa protein containing a mitochondrial target signal and 16 PPR motifs. Among three restorer genes, which have been cloned so far, maize Rf2 (Cui et al., 1996), Petunia Rf (Bentolila et al., 2002), and radish Rfk1 (Rfo; Brown et al., 2003; Desloire et al., 2003; Koizuka et al., 2003), two of them (Petunia Rf and radish Rfk1 or Rfo) encode proteins which have PPR motifs, and rice Rf-1 belongs to this group. The restorer proteins encoded by those four genes contain a putative mitochondria targeting sequence in the N-terminal region. The putative mitochondrial targeting sequence of rice Rf-1 has the same N-terminal seven amino acids as that of maize Rf2, while no homology is found between the target sequence of rice Rf-1 and those of Petunia Rf and radish Rfk1 (Rfo; Figure 4b). Mitochondrial targeting sequences encoded by restorer genes seem to be conserved depending on the degree of evolutional relationship of species rather than on the similarity of the restoration mechanism. As CMS is caused by aberrant mitochondrial genes in general, it should be universal that nuclear restorer genes encode proteins with a mitochondrial targeting sequence regardless of the restoration mechanisms. The restorer genes coding for a PPR protein have been demonstrated to alter the expression of CMS-associated genes (Akagi et al., 1994; Bentolila et al., 2002; Iwabuchi et al., 1993; Kadowaki et al., 1990; Koizuka et al., 2003). In contrast, maize Rf2 encodes aldehyde dehydrogenase (Liu et al., 2001), and does not affect the accumulation of the CMS-associated proteins URF13 (Dewey et al., 1987). Moreover, maize rf2 is deleterious to lower, but not upper, florets in normal cytoplasm plants (Liu et al., 2001), leading to an argument whether Rf2 is a restorer gene of CMS-T (Schnable, 2002; Touzet, 2002). Therefore, rice Rf-1 is the first restorer gene isolated from cereal crops that influences the CMS-associated transcript and encodes a PPR protein. It is likely that, in addition to Petunia Rf, radish Rfk1 (Rfo), and rice Rf-1, some of other restorers influencing the expression profile of the CMS-associated genes encode a PPR protein. Thus, isolation of rice Rf-1 may facilitate isolation of other restorer genes from agriculturally important cereal crops such as rice and maize.
A previous study on the CMS mechanism of the BT cytoplasm (Kadowaki et al., 1990) identified a chimeric gene containing the 5′ portion of ATPase subunit 6 (atp6) together with a normal atp6 gene in the BT cytoplasm. The chimeric gene, named unidentified reading frame of rice mitochondria associated with CMS (urf-rmc), was probably generated by an intramolecular homologous recombination between atp6 and a mitochondrial sequence. As urf-rms was found only in CMS lines, and as introduction of Rf-1 to a CMS line changed the transcript size of urf-rmc from 2.7 kbp in a CMS line (BT cytoplasm, rf-1/rf-1) to 2.8 kbp in a restorer line (BT cytoplasm, Rf-1/Rf-1), it was presumed that the increase in the length of the urf-rmc transcript under the influence of Rf-1, possibly because of alteration in RNA synthesis or processing, would lead to an immature or untranslatable transcript, and might thereby prevent the expression of the CMS phenotype (Kadowaki et al., 1990). Another study (Iwabuchi et al., 1993) reported that CMS mitochondria contained a second atp6 gene, B-atp6, in addition to a normal atp6 gene, N-atp6, and that the transcript of B-atp6 was correlated with the CMS phenotype. B-atp6 was transcribed into 2.0 kbp in the absence of Rf-1, whereas two RNAs, 1.5 and 0.45 kbp, were detected in the presence of Rf-1. Together with the results of the editing pattern of the atp6 RNAs, it was proposed that RNA processing of the first transcript of B-atp6 by Rf-1 would influence the sequential post-transcriptional editing and be involved in controlling CMS expression (Iwabuchi et al., 1993). Thus, in two studies, the genes or transcripts involved in the CMS phenotype are different, but both suggested that a chimeric gene containing the 5′ portion of atp6 was involved in the CMS phenotype and that the gene product of Rf-1 would modify the transcript profile of the chimeric gene. Two features of the protein encoded by Rf-1 are consistent with the previous interpretation on the Rf-1 function. First, the protein is predicted to move into mitochondria, and second, the protein contains 16 repeats of the PPR motif. As repeated PPR motifs are thought to be involved in RNA binding (Bentolila et al., 2002; Small and Peeters, 2000), it is likely that Rf-1 modifies the expression of mitochondrial gene(s) by interacting with specific RNA species. In fact, a recent report (Kazama and Toriyama, 2003) demonstrated that a 4.7-kbp genomic fragment carrying PPP8-1, which was proved to be identical to PPR791 (Rf-1), affected the expression of B-atp6 in callus. Probably, the B-atp6 expression is similarly affected by Rf-1 in panicles, which may lead to the efficient editing of the B-atp6 RNAs and the production of normal polypeptide as previously proposed by Iwabuchi et al. (1993), resulting in the restoration of pollen fertility. For better understanding of the mechanisms of CMS by the BT cytoplasm and its restoration by Rf-1, it should be a crucial step to identify the transcript(s) to which the Rf-1-encoded protein would bind. Isolation of Rf-1 made this approach possible.
The genomic sequence of IR24 around Rf-1 uncovered the presence of three genes that encoded mitochondrially targeted PPR proteins within 25 kbp upstream of Rf-1. The results of a series of complementation tests revealed that the three genes did not function as a restorer of the BT cytoplasm. However, two of these, PPR794 and PPR762, were proved to be expressed at least in young panicles of a restorer line, and their deduced proteins showed high homology with the Rf-1-encoded protein. As the Asominori allele at the PPR794 locus encodes a protein almost the same as the predicted gene product from the corresponding allele of IR24, it is unlikely that this locus plays an important role in expressing a phenotypic difference between IR24 and Asominori. On the other hand, Asominori has no allele at the PPR762 locus, and a similarity search using the sequence of a Nipponbare bacterial artificial chromosome (BAC), AC068923, revealed no nucleotide polymorphism at the locus between Asominori and Nipponbare (data not shown). This leads to an assumption that common japonica varieties lack the allele, and that the PPR762 locus is involved in a potential difference between MS-Koshihikari and FR-Koshihikari in an expression profile of mitochondrial gene(s), urf-rmc and/or B-atp6, for example. Future study on the gene functions of these PPR genes may provide a clue to elucidate generation and differentiation of restorer genes.
Comparison of genomic sequences at the Rf-1 locus between IR24 and Asominori revealed that the recessive allele (rf-1) encodes a truncated protein because of a 1-bp deletion in the putative coding region. In contrast, it was assumed that the recessive allele at the Petunia Rf locus was generated by a recombination between two PPR genes, one of which is the functional restorer, Rf-PPR592, and that a 530-bp deletion in the promoter region in the rf allele would alter the expression pattern, resulting in a loss of the gene function (Bentolila et al., 2002). Interestingly, such a recombination actually occurred during in vitro cloning in our study; a 4645-bp region sandwiched by a perfect direct repeat of 356 bp was lost during isolation of a lambda phage clone, XSF18, and the resulting clone was named XSF18-del (Figures 1a and 5). Perhaps, also in plants, this kind of recombination may cause generation and/or loss of PPR genes around restorer genes, and be involved in conferring and/or removing a function as a restorer gene. The fact that both in Petunia (Bentolila et al., 2002) and in radish (Brown et al., 2003; Desloire et al., 2003), the isolated restorer gene was also adjacent to other PPR gene(s) possibly support the hypothesis.
Provided that restorer genes often encode a PPR protein, isolation of other restorer genes may be facilitated by conducting a search of the PPR motif based on the map information. The consensus PPR motifs were proved to be highly conserved among restorer gene products (Figure 4c), and the conserved consensus motif has some amino acid substitutions compared to the consensus motif derived from 1303 PPR motifs (Small and Peeters, 2000). If this change from common PPRs is conserved among restorer gene products, the change may be applied to identify candidate PPRs involved in CMS restoration among numerous PPRs. Rice is an adequate plant to perform this type of approach because its genome sequencing has been almost finished and information of predicted gene products is available.
Transcripts of rice Rf-1 were highly varied with respect to splicing patterns and polyadenylation sites. Alternative splicing is a common mechanism for regulating gene expression in higher eukaryotes (Adams et al., 1996). In plants, a report (Reddy, 2001) described 29 pre-mRNAs that undergo alternative splicing, many of which are known to encode proteins with different functions and/or different localization patterns. In the case of rice Rf-1, the complex splicing patterns generated no functional difference at the deduced amino acid level because of the existence of a stop codon in the first exon. The complementation test for the 4.2-kbp fragment (Figure 3) actually indicated no apparent effect of the additional exons at least on the function of restoring CMS by the BT cytoplasm, although it is too early to conclude that the observed diversity of transcripts has nothing to do with controlling gene expression.
Two near-isogenic lines, named MS-Koshihikari and FR-Koshihikari, were used with their recurrent parent, Koshihikari. The cytoplasm and the nuclear genome of MS-Koshihikari (a CMS line) were derived from Chinsurah Boro II and Koshihikari, respectively. The Rf-1 gene derived from IR8 was introduced to MS-Koshihikari by recurrent backcross to breed FR-Koshihikari (a restorer line). As a result, FR-Koshihikari has the same cytoplasmic and nuclear genomes as MS-Koshihikari, except for the Rf-1 region.
Genomic libraries of Asominori and IR24 were constructed with the Lambda DASH II/BamHI Vector Kit (Stratagene, La Jolla, CA, USA), as described previously by Komori et al. (2003).
The walking was started by screening the Asominori library with a genomic region corresponding to a part of a cDNA clone, S12564, as a probe (Probe A, Figure 1a). Table 3 lists sequences of PCR primers used in this study to amplify genomic regions for preparation of probes. 32P-labeling of the probes, plaque hybridization, and isolation of phage DNA were performed as described by Komori et al. (2003).
For the second walking, the Asominori library and the IR24 library were screened with a partial fragment of a clone obtained by the first walking as a probe (Probe E, Figure 1a). In a similar way, the third and the fourth walkings were conducted with Probe F and Probe G (Figure 1a), respectively.
After the fourth walking, we found that a Nipponbare (japonica, rf-1/rf-1) BAC clone, AC068923, including the S12564 locus, was accessible through a web site of TIGR (The Institute of Genome Research). Therefore, the probe for the fifth walking (Probe H, Figure 1a) was prepared based on the BAC sequence, and was used to screen the IR24 library.
The sequences of the contigs constructed by piling the isolated clones were determined by primer walking.
Development of new PCR-based markers flanking Rf-1
To develop additional PCR-based markers flanking the Rf-1 locus, genomic sequences of the flanking region were compared between varieties with the genotype of rf-1/rf-1 and varieties with the genotype of Rf-1/Rf-1. Primers were designed for cleaved amplified polymorphic sequence (CAPS; Konieczny and Ausubel, 1993) or derived CAPS (dCAPS; Michaels and Amasino, 1998; Neff et al., 1998) markers.
Ten restriction fragments (ctE1, ctE7, ctF4, ctF18-del, ctF18, ctG22, ctG16, ctG8, ctH18, and ctG8H18; Figure 1c) derived from nine lambda phage clones harboring genomic fragments of IR24 were ligated to an adequate site of pSB200, a vector carrying a hygromycin-resistant gene for plant selection. Furthermore, two restriction fragments, ctG16-sub6.8 and ctG16-sub11.4, within ctG16 were ligated to pSB11 (Komari et al., 1996). Recombinants were cloned into Escherichia coli DH5a, and then mobilized into Agrobacterium tumefaciens LBA4404 (pSB1; Komari et al., 1996) by triparental mating using E. coli HB101 (pRK2013; Ditta et al., 1980) as a helper strain, as described by Hiei et al. (1994). Immature embryos of MS-Koshihikari were transformed with the Agrobacterium prepared above and regenerated as described by Hiei et al. (1994).
Iodine staining of pollen
Spikelets before flowering were harvested just after heading, and fixed in 70% (v/v) ethanol. Pollens were stained with a solution containing 0.18% (w/v) iodine and 1% (w/v) iodine potassium.
cDNA library construction and screening
Total RNA was extracted from young panicles (5–10 cm long) of FR-Koshihikari by the standard SDS-phenol method (Watanabe and Price, 1982), and then poly(A)+ RNA was purified with the QuickPrep mRNA Purification Kit (Amersham Pharmacia Biotech, Piscataway, NJ, USA). A cDNA library was constructed with the poly(A)+ RNA using the ZAP-cDNA Synthesis Kit (Stratagene) according to the manufacturer's instructions.
Approximately 1 × 106 phage clones from the library were screened with Probe X and Probe Y (Figure 3; Table 3). Membranes for hybridization were made in duplicate from each plate, and plaque hybridization was performed as described previously by Komori et al. (2003). Phage clones detected with both of the two probes were isolated and converted to pBluescript SK(−) phagemid clones by in vivo excision following the supplier's instructions (Stratagene). Screening was also conducted with a probe combination of Probe X and Probe Z (Figure 3; Table 3).
Multiple blocks in the Rf-1-flanking regions of IR24 and Asominori were determined by aligning the genomic sequences against the nucleotide sequence of the Rf-1 gene with genetyx sv/rc ver. 6.1 (Software Development Co. Ltd, Tokyo, Japan). Similarity search and translation analysis were also conducted with the software. The presence of a transit peptide in the deduced proteins was predicted with mitoprot (http://mips.gsf.de/cgi-bin/proj/medgen/mitofilter; Claros and Vincens, 1996).