Mitochondrial genome recombination in somatic hybrids of Solanum commersonii and S. tuberosum

Interspecific somatic hybridization has been performed in potato breeding experiments to increase plant resistance against biotic and abiotic stress conditions. We analyzed the mitochondrial and plastid genomes and 45S nuclear ribosomal DNA (45S rDNA) for the cultivated potato (S. tuberosum, St), wild potato (S. commersonii, Sc), and their somatic hybrid (StSc). Complex genome components and structure, such as the hybrid form of 45S rDNA in StSc, unique plastome in Sc, and recombinant mitogenome were identified. However, the mitogenome exhibited dynamic multipartite structures in both species as well as in the somatic hybrid. In St, the mitogenome is 756,058 bp and is composed of five subgenomes ranging from 297,014 to 49,171 bp. In Sc, it is 552,103 bp long and is composed of two sub-genomes of 338,427 and 213,676 bp length. StSc has 447,645 bp long mitogenome with two subgenomes of length 398,439 and 49,206 bp. The mitogenome structure exhibited dynamic recombination mediated by tandem repeats; however, it contained highly conserved genes in the three species. Among the 35 protein-coding genes of the StSc mitogenome, 21 were identical for all the three species, and 12 and 2 were unique in Sc and St, respectively. The recombinant mitogenome might be derived from homologous recombination between both species during somatic hybrid development.


Scientific Reports
| (2022) 12:8659 | https://doi.org/10.1038/s41598-022-12661-z www.nature.com/scientificreports/ research on the genetic and evolutionary nature of mitogenomes [14][15][16] . Mitogenome sequences were released for crop species in Solanaceae including pepper (Capsicum annum), potato (S. tuberosum), and tomato (S. lycopersicum) [17][18][19] , and mitogenome structures and phylogenetic relationships among these species have been revealed. Previously, we reported the complete mitogenome sequences for St and Sc; however, a detailed genome analysis was not conducted 18,20 . Somatic hybridization mediated by protoplast fusion can be a useful breeding tool for introducing valuable traits from related species [21][22][23] . Favorable traits such as disease resistance 24,25 and drought tolerance 26 were transmitted to the cultivated potato from wild species. Similar approaches have been applied to develop potato somatic hybrids with enhanced disease resistance using wild potato species [27][28][29][30][31] . The organelle genome structures in somatic hybrids have been investigated in these studies. Conserved intact plastomes were delivered to the somatic hybrid from either of the two parents. Novel mitogenomes have been identified via non-random rearrangement in somatic hybrids 21,29,32 . Previous studies have found putative rearrangement hotspots and specific regions preferentially inherited or eliminated. In addition, Lossl et al. 33 speculated that non-random mitochondrial rearrangement determining mitogenome types in somatic hybrids might be related to yield components in potatoes. However, these studies did not provide sequence-level evidence for mitogenome recombination events in somatic hybrids.
In this study, we characterized plastome, mitogenomes, and nuclear ribosomal DNA (nrDNA) of tetraploid potato (S. tuberosum), wild diploid potato (S. commersonii), and their somatic hybrid. The comparative analysis revealed their inheritance pattern and exhibited dynamic multipartite structural variation mediated by recombination events that share unique genes from both species in the somatic hybrid mitogenome.

Results
Complete mitochondrial genome assembly. The mitogenomes of St, Sc, and StSc were assembled into five to two subgenomes through de novo assembly using 5.3 to 6.6 Gb PE reads. Each assembly was validated by conducting PCR analysis and sequencing (Tables S1 and S2, Fig. S1). The St mitogenome size was 756,058 bp, and it was composed of five circular subgenomes of lengths 49 Fig. S2C).
Two large repeats (more than 1 kb) were identified in the St subgenome 1. R1 was 11,916 bp, and R2 was 7500 bp. In contrast, St subgenome 2 had only R1, and subgenome 3 had only 1589 bp of R3. Similarly, the R1 sequence co-existed in St subgenomes 1 and 2. The R2 repeat is shared between subgenomes 1 and 4 (  Fig. 2). Therefore, it is likely that the majority of the somatic hybrid mitogenomes originated from Sc (Fig. 2). GISH data using Sc genome probes revealed strong signals in 24 chromosomes but weak signals in the other 24 chromosomes in the StSc somatic hybrid (Fig. 3A). We also assembled and compared 45S nrDNA cistron sequence of three species. For example, multiple aligned position at 191 bp represents 'T' genotype in St and 'C' genotype in Sc. However, in StSc, it was identified that 75.6% of 'T' and 24.4% of 'C' were present. In conclusion, the overall 45S nrDNA sequences of StSc revealed both genotypes with average about 70 and 30 ratio for Sc and St, respectively (Fig. 3B). www.nature.com/scientificreports/ In summary, St, a dihaploid of tetraploid cultivated potato, has five mitogenomes. Sc, a diploid wild potato, has two mitogenomes. Somatic hybrids developed via protoplast fusion of these two diploids contain the Sc-unique plastome 37 but recombined mitogenomes and nuclear genomes derived from both St and Sc genomes (Fig. 4).  (Table S8, Fig. 5A). A Ka/Ks value of more than 2 was observed due to the extremely low Ks value.

Mutation rate of mitochondrial genes in Solanaceae.
Although the Ka and Ks values were generally low, ccmFc and mttB exhibited high Ka/Ks values of more than 1, indicating that these genes were positively selected during evolution (Fig. 5A). Considering that atp6 showed a high mutation rate above 0.1. Ka and Ks values relative to the other genes, the amino-acid sequences corresponding to atp6 were compared among Solanaceae species, which revealed that amino acid sequences were variable at the N-terminus but conserved at the C-terminus (Fig. 5B).
Phylogenetic trees were constructed using various programs, including RAxML, MEGA7, PhyML, and BEAST to examine the topology of the species. Trees treated with RAxML, PhyML, and BEAST displayed the same topology, while those treated with MEGA7 exhibited slightly different topologies (Fig. S3). In trees generated using RAxML representing an optimized topology (Figs. 6 and S3), Solanaceae species were divided into two subfamilies, Solanoideae and Nicotianoideae, and the somatic hybrid exhibited a moderate branch between St and Sc. During the evolution of Solanaceae mitogenome, first, rps1 and rps19 were present in Solanaceae, however, these were omitted completely in Oleaceae. Next, rps7 was confirmed to be completely deleted in Solanaceae compared to Oleaceae. Lastly, ycf14 in all Nicotianoideae species was pseudogenized in the divergence period between Solanoideae and Nicotianoideae (Fig. 6).

Discussion
Diverse mitogenome structures in potato. The potato (S. tuberosum) nuclear genome was reported by the Potato Genome Sequencing Consortium 40 , and mitogenomes were reported by our group for S. tuberosum (MF989953-MF989957) and S. commersonii (MF989960-MF989961) 20 . In this study, we sequenced a novel mitogenome for somatic hybrids (MF989958-MF989959) of St and Sc and conducted a comparative genome analysis. These assemblies play an important role in the unique genetic inheritance patterns of somatic hybrid mitogenomes. Plant mitogenomes exhibit dynamic recombinant structures mediated by HR 41,42 . The large repeats shared in St, Sc, and StSc mitogenomes might facilitate HR events in St, Sc, and StSc.
The HR of mitogenomes mediated by dispersed repeats was suggested to consist of a multivariate configuration and subgenomes 43 . Similar to the five sub-mitogenomes of S. tuberosum (accession no. PT56: St), another mitogenome was reported using the cultivar Cicero and Désirée (hereafter referred to as Cicero) based on PacBio sequencing 44 . Meanwhile, our assembled and Cicero mitogenomes exhibited noticeable differences. The largest mitogenome of Cicero is the fused form of mitogenome subgenomes 1, 2, and 4 (Fig. S5). Our mitogenome subgenome-1 was assembled in a circular form; in contrast, the Cicero's largest mitogenome was observed in Table 2. Common and unique mitogenome genes of a somatic hybrid and its parent species.     (Figs. 1 and 4), suggesting that various sub-mitogenomes can exist in different individuals or tissues. Furthermore, atp6 in St sub-mitogenome 1 was prematurely terminated as 321 amino acids (aa) in length, whereas the Cicero mitogenome had a full structure gene of 389 aa. The discrepancy of atp6 was examined thoroughly, and a linear fragment with a complete structure of 389 aa was found by mapping to atp6 sequence of Cicero mitogenome.

Group of genes
Not only master circle, but also other forms due to homologous recombination of mitochondrial subgenomes exist in the different cells 45 . This suggests that there might be a different form or an unrevealed linear form or short fragment could also contain essential genes in the mitogenome, such as atp6, even if not in the circular form.   hybrids were randomly selected and delivered from parents, and the nuclei or mitochondria are known to fuse 13 . In previous study, the StSc plastome was transferred only from S. commersonii 37 . In addition, StSc mitogenomes were randomly rearranged between St and Sc in most regions. The clustering analysis of CDSs revealed that the majority of the genes were derived from Sc; however, certain genes were derived from St (Fig. 2). Previous studies have suggested that the somatic hybrids harbor recombinant mitogenomes that share both mitogenome types based on fingerprinting patterns 13,24,32,33 . In this study, we have displayed sequence-level recombination events that share unique genes from each of the parental species of somatic hybrids. Although this phenomenon may have occurred entirely randomly, fundamental mitochondrial genes in the Solanaceae family were preserved. Although the recombination mechanism is unclear, we assume that the smaller mitogenome might be competitive during somatic hybridization. Low evolutionary rate of mitochondrial genes in Solanaceae. To date, missing or misnamed genes have been examined in the Solanaceae mitogenome, and it has been confirmed that 35 PCGs are commonly preserved (Table S7). These genes were found to have few mutations, and even if mutations existed, most of them were identified as synonymous substitutions. This result is consistent with the fact that mitogenomes have few gene variations 38,46 . Exceptionally, atp6 exhibited large length variation. Moreover, atp6 was present in an intact form with a short length; however, the Solanaceae mitogenome evolutionary process confirmed that the sequence continued to accumulate in the front of the conserved motif (Fig. 5). It was considered that atp6 could continuously promote mutations, suggesting that it can be useful for Solanaceae DNA barcoding. Novel ORFs that might be created during somatic hybridization have been identified; however, their function is unknown.
Based on the NCBI land plant organelle database (https:// www. ncbi. nlm. nih. gov/ genome/ organ elle/), 5295 cp genomes have been published, whereas only 279 mitogenomes have been released (July 2021). This can be attributed to the difficulty in assembling the mitogenome compared to the plastome. Therefore, our phylogenetic study will play an important role in identifying the relationship during mitochondrial evolution. Our phylogenetic trees were slightly different from those of traditional plastome-based trees 20 . The Solanum, Capsicum, and Nicotiana genera were assigned to the same lineage. However, in all other phylogenetic trees, Capsicum was grouped with S. lycopersicum and S. pennellii. This was due to the low mutation rate of the mitogenome. De novo assembly and validation of mitochondrial genomes and 45S nuclear ribosomal DNA. Raw PE data of approximately 5.8 Gb for S. tuberosum, 6.6 Gb for S. commersonii, and 5.3 Gb for somatic hybrid were generated and used for assembly (Table S1). De novo mitogenome assembly was performed using PE data according to a previous study 48 . Briefly, high-quality read sequences (Phred score > 20) were obtained, and de novo assembly was conducted using the clc_novo_assemble tool in the CLC Assembly Cell package (ver. 4.2.1, CLC Inc., Denmark). Contigs derived from raw data of mitogenomes were selected based on similarity with mitogenome sequences of other Solanaceae species, such as Capsicum annuum (GenBank acc. no. KJ865409), and Nicotiana tabacum (GenBank accession No. no. KR780036), and then extended, gap-filled, and merged through a series of read mapping to generate a draft circular mitogenome sequence.

Materials and methods
The draft mitogenome sequences were validated using bioinformatics and experimental methods. For validation based on PE read mapping, the high-quality PE reads were mapped again on the draft mitogenome sequences; subsequently, the consistency and connectivity of the mapped reads on draft mitogenomes and on junctions between contigs were confirmed. In read mapping, the coexistence of chloroplast or nuclear genomederived reads was manually confirmed and removed based on extremely high or low depth.
For validation based on PCR amplification and nucleotide sequencing, specific primers for each subgenome were designed using the NCBI Primer-BLAST tool (https:// www. ncbi. nlm. nih. gov/ tools/ primer-blast/ index. cgi) and used for genomic DNA PCR amplification and nucleotide sequencing (Table S2) www.nature.com/scientificreports/ Mitochondrial genome annotation. Mitogenomes were initially annotated using the GeSeq program (https:// chlor obox. mpimp-golm. mpg. de/ geseq-app. html) 49 , and the genes were further predicted by comparison with mitogenomes of other Solanaceae species such as Capsicum annuum (GenBank acc. no. KJ865409) and Nicotiana tabacum (GenBank accession No. no. KR780036). Ambiguous gene positions were manually corrected using NCBI BLASTN-based search analysis and the Artemis annotation tool 50 . A linear and circular map of the mitogenome with annotation information was drawn using the OGDRAW program (https:// chlor obox. mpimpgolm. mpg. de/ OGDraw. html) 51 . Repetitive sequences such as direct and palindrome repeats in the mitogenome were searched using the Vmatch program (http:// www. vmatch. de) 52 integrated with the REPuter program with a minimum repeat length of 20 bp and then selected using a length cut-off value of 100 bp. Tandem repeats were searched using the Tandem Repeats Finder program with parameters such as match 2, mismatch 7, indels 7, minimum alignment score 80, maximum period size 10, and maximum TR size 50 (https:// tandem. bu. edu/ trf/ trf. html) 53 .

Data availability
The complete mitogenome have been deposited in the National Center for Biotechnology