BackgroundPlant mitochondrial genome has unique features such as large size, frequent recombination and incorporation of foreign DNA. Cytoplasmic male sterility (CMS) is caused by rearrangement of the mitochondrial genome, and a novel chimeric open reading frame (ORF) created by shuffling of endogenous sequences is often responsible for CMS. The Ogura-type male-sterile cytoplasm is one of the most extensively studied cytoplasms in Brassicaceae. Although the gene orf138 has been isolated as a determinant of Ogura-type CMS, no homologous sequence to orf138 has been found in public databases. Therefore, how orf138 sequence was created is a mystery. In this study, we determined the complete nucleotide sequence of two radish mitochondrial genomes, namely, Ogura- and normal-type genomes, and analyzed them to reveal the origin of the gene orf138.ResultsOgura- and normal-type mitochondrial genomes were assembled to 258,426-bp and 244,036-bp circular sequences, respectively. Normal-type mitochondrial genome contained 33 protein-coding and three rRNA genes, which are well conserved with the reported mitochondrial genome of rapeseed. Ogura-type genomes contained same genes and additional atp9. As for tRNA, normal-type contained 17 tRNAs, while Ogura-type contained 17 tRNAs and one additional trnfM. The gene orf138 was specific to Ogura-type mitochondrial genome, and no sequence homologous to it was found in normal-type genome. Comparative analysis of the two genomes revealed that radish mitochondrial genome consists of 11 syntenic regions (length >3 kb, similarity >99.9%). It was shown that short repeats and overlapped repeats present in the edge of syntenic regions were involved in recombination events during evolution to interconvert two types of mitochondrial genome. Ogura-type mitochondrial genome has four unique regions (2,803 bp, 1,601 bp, 451 bp and 15,255 bp in size) that are non-syntenic to normal-type genome, and the gene orf138 was found to be located at the edge of the largest unique region. Blast analysis performed to assign the unique regions showed that about 80% of the region was covered by short homologous sequences to the mitochondrial sequences of normal-type radish or other reported Brassicaceae species, although no homology was found for the remaining 20% of sequences.ConclusionsOgura-type mitochondrial genome was highly rearranged compared with the normal-type genome by recombination through one large repeat and multiple short repeats. The rearrangement has produced four unique regions in Ogura-type mitochondrial genome, and most of the unique regions are composed of known Brassicaceae mitochondrial sequences. This suggests that the regions unique to the Ogura-type genome were generated by integration and shuffling of pre-existing mitochondrial sequences during the evolution of Brassicaceae, and novel genes such as orf138 could have been created by the shuffling process of mitochondrial genome.

To reveal the molecular and genetic mechanism of fertility restoration in Ogura male sterility in Japanese wild radish (Raphanus sativus var. hortensis f. raphanistroides), we investigated fertility restoration of a plant that lacks the dominant type of orf687, a previously identified fertility restorer gene. A total of 100 F2 plants were made from the cross between a male-sterile strain with the Ogura cytoplasm, 'MS-Gensuke', and a Japanese wild radish plant. Segregation of pollen fertility in the F2 plants led us to assume that 2 dominant complementary genes controlled the fertility restoration of the plants. However, the fertility of 27 of 59 male-fertile plants was not completely restored, resulting in a group of plants with partial male fertility. Northern blot analysis of the CMS-associated gene orf138 indicated that one restorer allele (termed Rft) was involved in the processing of orf138 RNA. Rapid amplification of cDNA ends (RACE) and subsequent Northern blot analysis confirmed that the orf138 transcript lost a 5' part of the coding region of the orf138 gene in the restored plants. The accumulation of ORF138 protein was significantly reduced by Rft, but trace amounts of the protein were recognized in both partially male-fertile and male-sterile plants with Rft. The relationship of pollen fertility and segregation of co-dominant sequence tagged site (STS) markers in the F2 generation suggested that the penetrance of Rft was so low that Rft needs suitable conditions to function sufficiently for the complete restoration of fertility.

The complete mitochondrial genome sequences of Brassica species have provided insight into inter- and intraspecific variation of plant mitochondrial genomes. However, the size of mitochondrial genome sequenced for Brassica oleracea hitherto does not match to its physical mapping data. This fact led us to investigate B. oleracea mitochondrial genome in detail. Here we report novel B. oleracea mitochondrial genome, derived from var. capitata, a cabbage cultivar ''Fujiwase''. The genome was assembled into a 219,952-bp circular sequence that is comparable to the mitochondrial genomes of other Brassica species (ca. 220-232 kb). This genome contained 34 protein-coding genes, 3 rRNA genes and 17 tRNA genes. Due to absence of a large repeat (140 kb), the mitochondrial genome of ''Fujiwase'' is clearly smaller than the previously reported mitochondrial genome of B. oleracea accession ''08C717'' (360 kb). In both mitotypes, all genes were identical, except cox2-2, which was present only in the Fujiwase type. At least two rearrangement events via large and small repeat sequences have contributed to the structural differences between the two mitotypes. PCR-based marker analysis revealed that the Fujiwase type is predominant, whereas the 08C717 type coexists at low frequency in all B. oleracea cultivars examined. Intraspecific variations in the mitochondrial genome in B. oleracea may occur because of heteroplasmy, coexistence of different mitotypes within an individual, and substoichiometric shifting. Our data indicate that the Fujiwase-type genome should be used as the representative genome of B. oleracea.

In order to determine the molecular basis of cytoplasmic male sterility (CMS) in alloplasmic lines of eggplant, the genomic structures and transcription patterns of mitochondrial ATP synthase subunit (atp) and cytochrome oxidase subunit (cox) genes were studied for wild and cultivated eggplants. Alloplasmic eggplant lines with cytoplasms of wild Solanum species showing either anther indehiscent type of CMS or non-pollen production type of CMS were studied with the cultivated eggplant Solanum melongena, used as a control. Southern hybridization of the mitochondrial genes indicated the difference between the two types of CMS and showed complete identity within each type. The cytoplasmic patterns of all wild species differed from that of the cultivated eggplant. Thus, the cytoplasm of the six wild eggplants and the one cultivated eggplant was classified into three groups. Male sterile plants of both types of CMS showed novel transcription patterns of atp1, whereas a different transcription pattern of cox2 was observed only in the anther indehiscent type. Based on these differences, we determined the DNA sequences of about a 4 kbp segment in the atp1 region. Although the coding and 3' flanking regions were almost identical among the cytoplasms, the 5' flanking region was completely different and novel open reading frames (orfs) were found for each of the CMS types and the cultivated eggplant. The cytoplasm of Solanum kurzii inducing the anther indehiscent type CMS had orf312, and those of Solanum aethiopicum and Solanum grandifolium of non-pollen production type CMS had orf218. The correspondence between the transcription patterns of these orfs and phenotypic expression of male sterility strongly suggests that these orfs are causal genes for each type of CMS.

In radish, the mitochondrial gene orf138 found in Ogura cytoplasm is responsible for male sterility, whereas the nuclear restorer gene (Rf gene) inhibits the expression of orf138. To investigate the distribution of the Rf gene for Ogura male sterile cytoplasm in Japanese wild radish, 226 plants collected from 15 regions in Japan and two additional sites of Korea were used as materials. PCR analysis indicated that 42% of the wild radishes possessed orf138. On the other hand, 207 plants (91.6%) were judged to have an Rf gene by the observation of pollen fertility in wild radishes themselves and in hybrids obtained by crosses with a male sterile variety having Ogura cytoplasm. For plants having restorer function, we analyzed the nucleotide sequence of the Rf gene, orf687, identified to date. Among the 207 plants having the Rf gene, 199 amplified the orf687 fragment of the expected size by PCR. PCR products were further analyzed by mismatch-specific endonuclease digestion and PCR-RFLP analysis. As a result, plants that showed an identical RFLP pattern to the known Rf type of orf687 were restricted to 30 plants (14.5% of the 207 plants) collected mainly in populations of southern region of Japan. It was also found that most of the plants (148 plants; 71.5% of 207 plants) possessed the sequence of the orf687 corresponding to the rf genotype lacking the restorer function. We conclude that the Rf gene widely distributed in Japanese wild radish is not orf687.