The mitochondrial genome of an asymmetrically cell-fused rapeseed, Brassica napus , containing a radish-derived cytoplasmic male sterility-associated gene

male sterility (CMS) is an important trait caus-ative genes are located in the mitochondrial genome. A CMS rapeseed, Brassica napus ‘SW18’, was made 25 years ago by an asymmetric (or “donor-recipient”) cell fusion between B. napus ‘Westar’ and a CMS radish ( Raphanus sativus ‘Kosena’), in order to transfer the radish CMS-associated gene without disturbing the rapeseed features. Here, we determined the nucleotide sequences of the mitochondrial genomes of Kosena radish and SW18. SW18 has a recombinant mitochondrial genome, which includes the whole 222-kb genome of Westar (54 genes) and a total of 23 kb insertions of four fragments from Kosena radish (three genes: orf125 , trnfM and atp1 ). All of the Kosena radish-derived fragments in the SW18 mitochondrial genome had sequences at their ends (ranging from 63 bp to 628 bp) that are identical to the sequences at the sites of insertion on the Westar rapeseed-derived mitochondrial genome. This suggests that these insertions were mediated by homologous recombination. These results conﬁrm at the nucleotide level that a desired CMS-associated gene ( orf125 ) along with a few extraneous genes from radish were successfully transferred.


INTRODUCTION
Cytoplasmic male sterility (CMS) is frequently used in F1 hybrid breeding for efficient F1 seed production, because it causes bisexual plants to become female, resulting in secure outcrossing (Chen and Liu, 2014). Many Brassica vegetables show strong hybrid vigor, but good CMS is scarce. Therefore, Ogura-type CMS in the radish, Raphanus sativus L. (Brassicaceae) (Ogura, 1968), is now widely transferred to Brassica vegetables (Yamagishi and Bhat, 2014). Early attempts to transfer Ogura cytoplasm to Brassica vegetables by hybridization between radish and rapeseed and backcrossing to the rapeseed failed to produce a good product, because genes other than CMS genes that were acquired from radish mitochondrial or plastid genomes caused inappropriate phenotypes (yellowish and slow-growth) (Pelletier et al., 1983; The mitochondrial genome of an asymmetrically cell-fused rapeseed, Brassica napus, containing a radish-derived cytoplasmic male sterility-associated gene Shin-ichi Arimura 1 * , Shungo Yanase 1 , Nobuhiro Tsutsumi 1 and Nobuya Koizuka 2 1 Graduate School of Agricultural and Life Sciences, The University of Tokyo, 1-1-1, Yayoi, Bunkyo-ku, Tokyo 113-8657, Japan 2 Faculty of Agriculture, Tamagawa University, 6-1-1, Tamagawa Gakuen, Machida, Tokyo 194-8610, Japan (Received 20 February 2018, accepted 30 May 2018J-STAGE Advance published date: 23 August 2018) Cytoplasmic male sterility (CMS) is an agronomically important trait whose causative genes are located in the mitochondrial genome. A CMS rapeseed, Brassica napus 'SW18', was made 25 years ago by an asymmetric (or "donor-recipient") cell fusion between B. napus 'Westar' and a CMS radish (Raphanus sativus 'Kosena'), in order to transfer the radish CMS-associated gene without disturbing the rapeseed features. Here, we determined the nucleotide sequences of the mitochondrial genomes of Kosena radish and SW18. SW18 has a recombinant mitochondrial genome, which includes the whole 222-kb genome of Westar (54 genes) and a total of 23 kb insertions of four fragments from Kosena radish (three genes: orf125, trnfM and atp1). All of the Kosena radish-derived fragments in the SW18 mitochondrial genome had sequences at their ends (ranging from 63 bp to 628 bp) that are identical to the sequences at the sites of insertion on the Westar rapeseedderived mitochondrial genome. This suggests that these insertions were mediated by homologous recombination. These results confirm at the nucleotide level that a desired CMS-associated gene (orf125) along with a few extraneous genes from radish were successfully transferred.
Key words: asymmetric cell fusion, donor-recipient cell fusion, plant mitochondrial genome and Bhat, 2014). Cell fusion is another strategy for introducing genes from distinct species. Plants regenerated from fused cells and their progeny are known to have only one of the parental plastid genomes while the mitochondrial genome is recombined from the parental ones (Belliard et al., 1979;Pelletier et al., 1983;Wang et al., 2012;Sanchez-Puerta et al., 2015). Sakai and Imamura (1992) tried to produce a CMS rapeseed harboring the CMS-associated gene orf125 from R. sativus 'Kosena', by using asymmetric cell fusion (donor-recipient cell fusion) between Brassica napus 'Westar' and an X-ray-irradiated Kosena radish ( Fig. 1) (Sakai and Imamura, 1992). The resultant line, B. napus 'SW18', having 38 nuclear chromosomes (the same as rapeseed, 2n = 38, but not radish, 2n = 18) was consecutively backcrossed (more than six times) and showed growth phenotypes identical to those of Westar rapeseed except for CMS (Sakai and Imamura, 1992;Iwabuchi et al., 1999).
Here, we compared the mitochondrial genome of SW18 rapeseed, the asymmetrically cell-fused line, with the mitochondrial genomes of the two parent lines, Westar rapeseed (Handa, 2003) and Kosena radish, the CMS donor line, to see which genes (and which DNA fragments) were transferred to, or recombined in, the SW18 mitochondrial genome.

MATERIALS AND METHODS
Plant materials and growth conditions Seeds (B. napus 'Westar' and 'SW18', and R. sativus 'Kosena') were obtained from Dr. Jun Imamura (Sakai and Imamura, 1992). These plants were grown on Jiffy pellets (Jiffy Products) at 22 °C under 16 h of diurnal light or continuous light (100 μmol m − 2 sec − 1 ). Total DNA was isolated by Plant DNeasy Mini Kit (Qiagen) from leaves of 14-dayold plants.
DNA sequencing, assembly and analyses Pairedend (350-bp) libraries of genomic DNAs were constructed and sequenced on Illumina HiSeq 2500 and HiSeq10X platforms by BGI. The Kosena radish sequencing generated 97,692,126 reads with a total length of 14,751,511,026 bp, while the SW18 rapeseed sequencing generated 97,191,262 reads with a total length of 14,675,880,562 bp. The reads were mapped to the reference sequences of R. sativus 'MS-Gensuke' (NC_018551) and Westar rapeseed (NC_008285), respectively. Reads mapping to the reference sequences of the mitochondrial genomes (530,362 reads of Kosena and 455,699 reads of SW18) were collected and re-used for generating de novo contigs with the software Geneious (Biomatters). Reads of the nuclear and chloroplast genomes that were erroneously mapped to the mitochondrial genomes could be easily detected by differences in the depths of their read coverage (about ten times less or ten times more, respectively, than those of the genuine reads for the mitochondrial genomes) and they were manually removed or ignored. The generated de novo contigs were manually connected by the sequences at their ends and the connections were confirmed by long PCRs. The sequences determined in this study were deposited in DDBJ as accession numbers AP018472 (Kosena radish mitochondrial genome), AP018473, AP018474 and AP018475 (subgenome 1, 2 and 3, respectively, of the SW18 rapeseed mitochondrial genome).
PCR analyses Long PCR experiments to confirm the mitochondrial genome structures mapped by Illumina short reads were done using Prime Star GXL DNA polymerase (TaKaRa) according to the manufacturer's instruction.

RESULTS & DISCUSSION
Kosena radish mitochondrial genome The mitochondrial genome of Westar rapeseed, the recipient line of asymmetric cell fusion, has a length of 221,853 bp, 34 protein-coding genes, three rRNAs and 17 tRNAs (Handa, 2003) Fig. 1. SW18 rapeseed, made by an asymmetric cell fusion and consecutive backcrosses. Protoplasts of Raphanus sativus 'Kosena' were irradiated with X-rays to disrupt DNA. Protoplasts of Brassica napus 'Westar' were treated with iodoacetamide (IOA) to prevent protein functions. Protoplasts of the two species were then fused. These procedures were done previously by Sakai and Imamura (1992). a length of 258,424 bp, 34 protein-coding genes, three rRNAs and 18 tRNAs ( Fig. 2 and Table 1). The Kosena mitochondrial genome displays 99.959% identity with the previously sequenced mitochondrial genome of an Oguratype CMS radish, MS-Gensuke (258,426 bp, 34 proteincoding genes, three rRNAs and 18 tRNAs, Tanaka et al., 2012). The two radish mitochondrial genomes have the same order of genes as each other except for a single 90-kb inversion. The inverted region has a pair of 442bp inverted repeats at its ends (100% identity between the inverted repeat sequences), which are the secondlargest pair of repeats in the genome (the largest repeat is 9,731 bp). Park et al. (2013) reported that an Ogura-type CMS radish has multiple master-circle genome structures with a difference of a 79,976-bp inversion with a pair of 311-bp inverted repeats at the ends; however, this inversion is distinct from that reported here.
Mitochondrial genome of the asymmetrically cellfused SW18 rapeseed The mitochondrial genome of SW18 appears to consist of three circular subgenomes (Fig. 3A, 227,181 bp, 156,632 bp and 14,497 bp). They have long identical sequences (region W2-C, 4,136 bp) between them, so that they can be drawn as a master circle with a total length of 398,310 bp (Fig. 3B). The nucleotide sequence and the structure of the first circular    (1), (2) and (3) refer to the regions from subgenomes 1 to 3. Subgenome 3, which is inserted into subgenome 1 in this map, could be inserted into subgenome 2 as another configuration.
subgenome are similar to those of the master circular mitochondrial genome sequence of Westar rapeseed (Handa, 2003) but with an 11,612-bp inversion of the rapeseed sequence (Region W3) and a 6,211-bp insertion from the Kosena radish mitochondrial genome (Region K1) (Fig.  3A). The inversion has a pair of 147-bp inverted repeats at its ends. Moreover, the Kosena inserted sequence, Region K1, has 281-bp and 628-bp sequences at its ends that are common to the sequences of the connected ends of Regions W1 and W2, respectively. These common sequences suggest that the inversion and insertion were generated via homologous recombination events. The second circular subgenome (subgenome 2; 156,632 bp) has the same sequence and structure as subgenome 1, except that it has a 5,651-bp Kosena-derived insertion (Region K2) instead of the middle of Region W4, 75,675 bp. In addition, the connected ends of Regions W4-A and K2 and those of Regions K2 and W4-B have common sequences (462 bp and 63 bp) (Fig. 3B). The third small 14,497-bp circular subgenome consists of two Kosena fragments of 8,250 bp and 2,755 bp (Regions K3 and K4), and another 4,136-bp Westar mitochondrial DNA sequence (Region W2-C). These connected sites also have three common sequences: 139 bp (Regions W2-C and K3), 23 bp (Regions K3 and K4) and 482 bp (Regions K4 and W2-C). All of the new connected sequences found in the SW18 mitochondrial genome have common identical sequences at both of the connected ends, suggesting that homologous recombination events were responsible for making the new genome. These three subgenomes have similar mean coverages of Illumina short reads (553, 390 and 322), suggesting that the stoichiometry of these three molecules is also similar. A comparison of the genomes (Table 1) shows that SW18 has the full set of 54 Westar rapeseed genes and only three Kosena radish mitochondrial genes, namely orf125 (the CMS-associated gene), atp1 and trnfM.
Mitochondrial genomes of plants derived from normal cell fusion typically have more than 20 recombinations with almost equal amounts of parental genome fragments (Belliard et al., 1979;Wang et al., 2012;Sanchez-Puerta et al., 2015). The SW18 mitochondrial genome has the whole circular genome of Westar with about 60% redundancy, and four partial fragments of Kosena radish (a total of 23 kb, accounting for 8.8% of the genome). Such asymmetric contributions seem to be consistent with the achievement of the previous trial (Sakai and Imamura, 1992) to transfer a CMS-associated gene to another species without greatly disturbing other phenotypes. Because methods for stably transforming plant mitochondrial genomes are currently unavailable, asymmetric cell fusion, a relatively old technique, is still a valuable method to modify the mitochondrial genome.