Intrageneric structural variation in organelle genomes from the genus Dystaenia (Apiaceae): genome rearrangement and mitochondrion-to-plastid DNA transfer

Introduction During plant evolution, intracellular DNA transfer (IDT) occurs not only from organelles to the nucleus but also between organelles. To further comprehend these events, both organelle genomes and transcriptomes are needed. Methods In this study, we constructed organelle genomes and transcriptomes for two Dystaenia species and described their dynamic IDTs between their nuclear and mitochondrial genomes, or plastid and mitochondrial genomes (plastome and mitogenome). Results and Discussion We identified the putative functional transfers of the mitochondrial genes 5′ rpl2, rps10, rps14, rps19, and sdh3 to the nucleus in both Dystaenia species and detected two transcripts for the rpl2 and sdh3 genes. Additional transcriptomes from the Apicaceae species also provided evidence for the transfers and duplications of these mitochondrial genes, showing lineage-specific patterns. Intrageneric variations of the IDT were found between the Dystaenia organelle genomes. Recurrent plastid-to-mitochondrion DNA transfer events were only identified in the D. takeshimana mitogenome, and a pair of mitochondrial DNAs of plastid origin (MIPTs) may generate minor alternative isoforms. We only found a mitochondrion-to-plastid DNA transfer event in the D. ibukiensis plastome. This event may be linked to inverted repeat boundary shifts in its plastome. We inferred that the insertion region involved an MIPT that had already acquired a plastid sequence in its mitogenome via IDT. We propose that the MIPT acts as a homologous region pairing between the donor and recipient sequences. Our results provide insight into the evolution of organelle genomes across the family Apiaceae.


Introduction
Dystaenia Kitag. is a genus of perennial herbs in the family Apiaceae (also known as Umbelliferae), which is endemic to Japan and Korea.It has two species: D. ibukiensis (Y.Yabe) Kitag.and D. takeshimana (Nakai) Kitag.(Kitagawa, 1937).This genus has attracted the attention of biologists because of its evolutionary patterns and processes in oceanic islands (Pfosser et al., 2005).This is because D. takeshimana is native to Ulleungdo, an oceanic island in Korea, and likely originated via anagenetic speciation from the Japanese species D. ibukiensis (Pfosser et al., 2005).Oceanic islands exhibit two modes of speciation: anagenesis and cladogenesis (Takayama et al., 2015).Among the oceanic islands, Ulleungdo has a higher level of endemism derived via anagenetic (88%) than cladogenetic speciation (Stuessy et al., 2006).Molecular phylogenetic studies based on nuclear internal transcribed spacer (ITS) and trnL-F regions have shown that Dystaenia is a monophyletic group, but its sister group remains unknown (Choi et al., 1998;Pfosser et al., 2005).Multilocus datasets from three genomic sequences are required to better understand the evolutionary history of Dystaenia.
Recently, next-generation sequencing platforms have generated deep coverage enabling the assembly of plastid and mitochondrial genomes.Comparative genomics of organelles helps shed new light on evolutionary events across the tree of life and provides valuable sources for phylogenetic studies.The plastid genome (plastome) of angiosperms generally has a conserved quadripartite structure with a pair of inverted repeat (IR), large single-copy (LSC), and small single-copy (SSC) regions (Ruhlman and Jansen, 2021).Angiosperm plastomes range from 120 kb to 170 kb in length and contain 79 proteins, 30 transfer RNA (tRNA), and four ribosomal RNA (rRNA) genes.In contrast, the mitochondrial genome (mitogenome) of angiosperms exhibits a dynamic structure with circular, linear, and branched molecules (Sloan, 2013).Angiosperm mitogenomes range from 222 kb to 983 kb in length and contain 41 proteins, 14 tRNA, and three rRNA genes (Mower et al., 2012).Furthermore, organellar phylogenomic analysis based on genome-scale data provides new insights into the origins of species that have undergone complex evolutionary histories with incomplete lineage sorting and hybridization (Liu et al., 2019;Park and Park, 2020;Wu et al., 2022).
Chloroplast capture resulting from hybridization or introgression has been suggested as a possible reason for phylogenetic incongruence (Rieseberg and Soltis, 1991;Stegemann et al., 2012).Whole-genome duplication and hybridization are involved in the evolutionary history of this family (Ren et al., 2018;Jia et al., 2023).Numerous nuclear single-copy genes from transcriptome data have been utilized to conduct a phylogenetic analysis of Apioideae, showing that this family has undergone a rapid evolutionary divergence and incomplete lineage sorting (Wen et al., 2020).To date, the complete plastomes of 439 species from 83 genera have been sequenced [National Center for Biotechnology Information (NCBI), accessed 26 June 2023], 421 species from the subfamily Apioideae, 16 species from the subfamily Saniculoideae, one species from the subfamily Azorelloideae, and one species from the subfamily Mackinlayoideae.The published plastomes of Apiaceae range from 141 kb to 179 kb and contain a full set of protein-coding genes.The plastomes have a conserved quadripartite structure, but the IR boundaries have shifted during Apiaceae genome evolution (Wang et al., 2021a;Wen et al., 2021;Yuan et al., 2021;Samigullin et al., 2022).The most notable feature among Apiaceae plastomes is the plastid DNA of mitochondrial origin (PLMT).Mitochondrionto-plastid DNA transfers have been reported in five lineages of this family: Crithmum and Petroselinum (Downie and Jansen, 2014;Downie and Jansen, 2015) and Caucalis, Cuminum, and Daucus (Iorizzo et al., 2012;Spooner et al., 2017).In contrast to plastomes, there are thus far only five species from the subfamily Apioideae with sequenced mitogenomes: Apium graveolens (Cheng et al., 2021), Bupleurum chinense (Qiao et al., 2022), Coriandrum sativum (Wang et al., 2021b), Daucus carota (Iorizzo et al., 2012), and Foeniculum vulgare (Palumbo et al., 2020).These mitogenomes range from 281 kb to 435 kb in size, with circular molecules.Losses of six mitochondria-encoded genes (rpl2, rps10, rps14, rps19, sdh3, and sdh4) were observed in all five mitogenomes.
As part of our ongoing research on the evolution of the genus Dystaenia, we determined the plastid and mitochondrial genomes of two species: D. ibukiensis and D. takeshimana.A comparison of the two-organelle genomes revealed intrageneric variations in size, structure, and gene content in Dystaenia.We demonstrated the functional replacement of rpl2, rps10, rps14, rps19, and sdh3 from the mitochondria to the nucleus and also compared intracellular DNA transfer (IDT) between the plastid and mitochondrial genomes.Interestingly, we identified a mitochondrion-to-plastid DNA transfer event in the D. ibukiensis plastome and discussed the evolutionary history of PLMT.Our results provide new insights into the evolution of Apiaceae organelle genomes, including intercompartmental transfers.

DNA/RNA extraction and sequencing
Fresh leaves of D. takeshimana and D. ibukiensis were collected from Ulleungdo, Geongbuk, South Korea, and Mt.Ibuki, Shiga, Japan, respectively.Total genomic DNAs were isolated using the Exgene Plant SV Mini Kit (GenAll, Seoul, South Korea), following the manufacturer's protocol.Total RNA was isolated, as described by Breitler et al. (2016).After gel electrophoresis and qualitative assessment, DNA and RNA samples were sequenced using the Illumina HiSeq2500 platform (Illumina, San Diego, CA, USA).Approximately 43 Gb and 20 Gb of 150-bp paired-end (PE) reads were generated from 550-bp insert libraries from the DNAs of D. takeshimana and D. ibukiensis, respectively.The DNAs from D. takeshimana and D. ibukiensis were used for long reads generated from four and two flow cells on the Oxford Nanopore Technologies (ONT) GridION platform (ONT, Oxford, UK), and 68 Gb and 12 Gb of ONT reads were produced, respectively.RNAs from D. takeshimana and D. ibukiensis were sequenced by the Illumina platform, generating 12 Gb and 8 Gb of 150-bp PE reads, respectively.

Organelle genome assembly, finishing, and annotation
Multiple assemblies for both species were generated using Canu v2.2 (Koren et al., 2017), MaSuRCA v4.0.5 (Zimin et al., 2017), SPAdes v3.15.3 (Antipov et al., 2016), and Velvet v1.2.10 (Zerbino and Birney, 2008) based on long-and short-read data.For example, single-type platform assemblies have been created using Canu for ONT reads and Velvet for Illumina reads.The default parameters were used for Canu.We used pairwise combinations of k-mers (99-145) and expected coverage values (50, 100, 150, 200, 500, and 1,000) without the scaffolding option for Velvet assemblies.Hybrid assemblies were generated using MaSuRCA and SPAdes, combining Illumina and ONT reads.The default parameters were used for MaSuRCA.For the SPAde assemblies, independent runs were executed with multiple coverage cutoffs (10, 50, 100, 200, and 500) using the "careful" option.All de novo assemblies were performed on a 64-core Linux workstation with 2,048 GB of memory.Plastid and mitochondrial contigs were identified using a BL A S T -l i k e a l g o r i t h m i n G e n e i o u s P r i m e 2 0 2 2 . 2 (www.geneious.com)with Liriodendron tulipifera plastome and mitogenome sequences (NC_008326 and NC_021152) as queries.The identified organellar contigs were manually aligned, and a consensus genome sequence was generated for each by tracking and end-inspecting the organellar contigs.The coverage depth of the whole plastome and mitogenome sequences was evaluated by mapping Illumina PE and ONT reads using Bowtie v2.4.2 (Langmead and Salzberg, 2012) and BWA v0.7.17 (Li, 2013).To predict all tRNA genes in the organelle genomes, ARAGORN v1.2.38 (Laslett and Canback, 2004) and tRNAscan-SE v2.0.9 (Chan et al., 2021) were used.Circular or linear organellar genomes were generated using OGDRAW v1.3.1 (Greiner et al., 2019).The newly sequenced genomes were deposited in GenBank with accession numbers OR231235-OR231238.

Comparative organellar analyses
Repetitive sequences in Dystaenia organelle genomes were identified using ROUSFinder.py(Wynn and Christensen, 2019).The two plastomes and two mitogenomes were aligned using the "progressive Mauve algorithm in Mauve v2.3.182(Darling et al., 2004) in Geneious Prime.Organellar protein-coding genes were collected from five species, each with available genomes.Individual gene alignments were generated using the "Translation Align" approach with MAFFT v7.49 in Geneious Prime.A maximum likelihood (ML) tree was constructed from a concatenated alignment of the 24-gene dataset using IQ-TREE2 v2.2.03, with a best-fit model (-m TEST) and 1,000 ultra-fast bootstrap replicates (-B 1000).

Identification of intracellular DNA transfer
To investigate PLMTs and mitochondrial DNAs of plastid origin (MIPTs), we performed a reciprocal "BLASTN" searches between the plastid genome and its mitochondrial counterpart with an e-value cutoff of 1 × 10 −6 , requiring at least 80% sequence identity and a minimum length of 50 bp.In addition, the CENSOR web server (Kohany et al., 2006) was used to search the mitochondrial genomes for putative nuclear transposable elements (TEs) with default parameters and "green plants" as a reference sequence source.To identify functional intracellular gene transfer (IGT) to the nucleus, candidate transcripts were verified by BLASTN searches for organellar genes against their transcriptomes.The conserved domain of the predicted open reading frame (ORF) was annotated by CD searches against the Conserved Domain Database (CDD) v3.19 (Lu et al., 2019). LOCALIZER v1.0.4 (Sperschneider et al., 2017) and TargetP v2.0 (Almagro Armenteros et al., 2019) were used to predict the presence of Nterminal presequences [chloroplast transit peptide (cTP) and mitochondrial targeting peptide (mTP)] and their potential cleavage sites.The transcript sequences were used as queries in "BLASTN" against the de novo genome sequence of two Dystaenia species.The sequenced genes were deposited in GenBank (Supplementary Table 1).Phylogenetic trees were constructed using ML methods as described above.

Migration of organelle genes into the nuclear genomes of Dystaenia
Although losses of the mitochondrial-encoded 5′ rpl2, rps10, rps19, and sdh3 genes were only identified in the Dystaenia organelle genomes, we performed BLASTN searches using 41 mitochondrial and 79 plastid gene sequences against each transcriptome dataset.BUSCO assessment of the transcriptomes resulted in 83.6% (D. takeshimana) and 62.1% (D. ibukiensis) (Supplementary Figure 3).We identified multiple transcripts with high sequence identities to the four genes containing the targeting sequences and conserved domain (Table 2, Supplementary Figure 4).Moreover, we identified a nuclear transcript for mitochondrial rps14, which is present in the mitogenome (Table 2, Supplementary Figure 4).In particular, two nuclear transcripts for the mitochondrial 5′ rpl2 and sdh3 were identified in both Dystaenia transcriptomes (Table 2, Supplementary Figure 4).To test the potential gene split transfer of truncated sdh4, we performed BLASTN searches using Amborella mitochondrial sdh4 as the query.One and two transcripts were identified in the D. takeshimana and D. ibukiensis transcriptome, respectively (Supplementary Figure 4).These transcripts contained a portion of the conserved peptidylprolyl isomerase (PPIase) domain upstream of the sdh4 domain.However, TargetP failed to predict the mitochondrial target sequences.No nuclear transcript was identified for the truncated mitochondrial rpl5 gene from the D. ibukienesis mitogenome.
Examining the draft D. takeshimana de novo genome assembly identified the generic structure of the nuclear genes (Figure 4A).This analysis revealed several exons and introns in the five nuclear genes.The nuclear-encoded RPS14 is intronless, whereas the 5′ RPL2, RPS10, and RPS19 contain two exons, whereas SDH3 contains four exons (Figure 4A).Variations in the intron length of each 5′ RPL2 and SDH3 gene were identified (Figure 4A).In the case of the nuclear-encoded RPS14, we found extra copies with 70.9% and 65.3% nucleotide sequence identity on the other D. takeshimana and D. ibukiensis nuclear genome scaffolds, respectively.The D. takeshimana nuclear genome contains an intact ORF; the D. ibukiensis nuclear genome has multiple internal stop codons.We constructed a phylogenetic tree using 24 mitochondrial genes from seven available Apiaceae mitogenomes (Figure 4B).The phylogenetic distribution of gene loss showed that all analyzed species shared the loss of mitochondrial rpl2, rps19, and sdh3 genes.species, multiple transcripts of mitochondrial sdh4, which also contains a portion of the conserved peptidylprolyl isomerase (PPIase) and sdh4 domains without mitochondrial targeting sequences, were found in three transcriptomes (A.graveolens, B. chinense, and D. carota subsp.Sativus; Supplementary Figure 5).However, transcripts from C. sativum and S. divaricate transcriptomes included a 5′ extension of 113 bp and 126 bp, respectively, with an incomplete ORF at the N-terminus.The first 32 and 42 amino acids of these ORFs (when "ATA" is translated as a start codon) were predicted as an mTP by LOCALIZER (0.974 and 0.889).We further generated a phylogenetic tree of each gene (5′ RPL2, RPS10, RPS14, RPS19, and SDH3) (Figure 5).The phylograms based on the 5′ RPL2 and RPS10 matrix conflicted with the 24-gene data matrix of the phylogenetic tree.However, the phylogenetic analyses of RPS19, SDH3, and RPS14 were consistent with the 24-gene data matrix of the phylogenetic tree, although A. graveolens and C. sativum were conflicting.Although it was difficult to infer the evolutionary history of the gene duplication events for the 5′ RPL2, it was possible for the SDH3 and RPS14 based on robust relationships with high bootstrap values.For example, phylogenetic analysis of the nuclear-encoded SDH3 copies suggested that duplication events occurred in the common ancestors of Dystaenia, A. graveolens, C. sativum, D. carota subsp.sativus, and S. divaricate, and the loss of one copy occurred independently in D. carota subsp.sativus (Figures 4, 5).Phylogenetic analysis of the nuclear-encoded RPS14 copies suggested that the IGT occurred in the common ancestor of the two Dystaenia species, A. graveolens, C. sativum, D. carota subsp.sativus, and S. divaricate, and then stochastic losses occurred after the duplication events (Figures 4, 5).

Migration of mitochondrial DNA into the plastome of Dystaenia
The regions that Mauve did not align indicated that the sequence lacked detectable homologous regions between the two plastomes (Supplementary Figure 2A).According to the possible scenario for IR boundary shifts, the region is associated with the end of the boundary (Figures 1, 6).Nucleotide sequence alignment of the trnH-psbB intergenic spacer from the D. ibukiensis plastome with two intergenic spacers (clpP-psbB and trnH-ycf1) from the D. takeshimana plastome revealed the insertion of a 2,411-bp fragment.The GC content of the insertion region was 46.5%, whereas the remaining LSC regions had a GC content of 36.2%.Illumina reads from D. ibukiensis mapped to its plastome confirmed the insertion of trnH and psbB resulting from uniform coverage.The primers designed to amplify this region yielded PCR products of the expected size (Supplementary Figure 6).
To identify possible mitochondrion-to-plastid DNA transfer [plastid DNA of mitochondrial origin (PLMT)], the inserted region sequences of the D. ibukiensis plastome were used to query the D. ibukiensis mitogenome using "BLASTN" searches.BLAST (> 30 bp Phylogenetic trees for the nuclear-encoded genes from seven Apiaceae species.A species phylogeny and copy number variation in each nuclearencoded genes (right).Maximum-likelihood trees based on each nuclear gene.The numbers "1" and "2" after each species in the phylograms represent paralogs.Bootstrap support values > 50% are shown at nodes.MT, mitochondrial gene; asterisk, pseudogene.
in length) returned two hits to the two non-coding regions of the mitogenome (Figure 6A, Supplementary Figure 7).A large fragment (2,294 bp) was located upstream from the nad6 gene in a small mitochondrial circle with 99.3% nucleotide identity.Compared with our MIPT analysis, a 579-bp sequence adjacent to the large fragment revealed MIPT.A small fragment (49 bp) was located upstream from the atp4 gene in a large mitochondrial circle with 100% nucleotide identity.The short sequences (10 bp, "AGAAAGGCCC") at the end of the small fragment were identical to the sequences downstream of the trnH gene in the D. takeshimana plastome, indicating that the short sequences were D. ibukiensis plastid DNA.Three hits to TEs (ATLANTYS_LC_I, Gypsy-2_LC_I, and Gypsy-54_GM-I) were detected in the mitogenome, whereas the plastome contained only one TE (Figure 6A).PLMT had three repetitive sequences of "CCTACGTATGCCTATG"; however, the mitogenome contained four copies (Figure 6A).No PLMT hits were identified from reciprocal BLAST searches between the D. takeshimana plastid and mitochondrial genomes, indicating that the mitochondrion-top l a s t i d D N A t r a n s f e r e v e n t w a s u n i q u e t o t h e D .ibukiensis plastome.

Discussion
Plastids and mitochondria originate from cyanobacterial and alpha-proteobacterial endosymbiont ancestors within plant cells, respectively (Lang et al., 1999;Keeling, 2010).The angiosperm organelle genomes are typically assembled into circular maps (Mower et al., 2012;Smith and Keeling, 2015;Ruhlman and Jansen, 2021).However, mitogenomes often exhibit subcircular, linear, and branched chromosomes (Bendich, 2007;Sloan, 2013;Smith and Keeling, 2015).The presence of recombination activities with repeats (particularly large repeats) contribute to variations in the mitogenome structural organization (Kühn and Gualberto, 2012).Complete plastid and mitochondrial genomic sequences are required to understand organelle genome evolution better.Here, we generated the complete plastid and mitochondrial genomes of two Dystaenia species and showed contrasting patterns of genomic structure within organellar genome evolution.For example, compared with other Apiaceae plastomes, the D. takeshimana plastome was highly conserved, whereas D. ibukiensis underwent relocation and inversion events of a large region (17 kb, a part of ycf2 through psbB genes) (Figure 1).We Frontiers in Plant Science frontiersin.orginferred that the structural changes in the D. ibukiensis plastome were likely caused by a series of IR expansions and contractions (Figure 1): first, an expansion at the IR B /LSC boundary to the clpP occurred, followed by a contraction at the IR A /LSC boundary, from the clpP to the middle of the ycf2 gene.In the plastomes of Apiaceae, the inversion of only a small fragment (490 bp, trnY-trnD-trnE) has been documented in Peucedanum (Liu et al., 2022) and Angelica (Wang et al., 2021a).It is well known that the expansion and contraction of the IR play a key role in the evolution of Apiaceae plastomes.In D. ibukiensis, IR expansion and contraction resulted in the relocation of a large gene block (Figure 1).The occurrence of large events in the D. ibukiensis plastome is a unique feature of the family Apiaceae.In particular, the D. ibukiensis plastome harbored foreign DNA transferred from its mitochondrial counterpart (Figure 6).Incorporating foreign DNA into the D. ibukiensis plastome is a valuable case study for the evolution of plastomes in Apiaceae.The D. takeshimana mitogenome mapped a circular molecule similar to those of Apiaceae mitogenomes (A.graveolens, B. chinense, C. sativum, D. carota subsp.sativus, and F. vulgare), whereas the D. ibukiensis mitogenome mapped two molecules with substoichiometric linear (Figure 2).Similar to D. ibukiensis, a recent study demonstrated that the Aragoa cleefii mitogenome from Plantaginaceae exhibits a high level of complexity, characterized by the presence of two circular maps connected by a substoichiometric linear (Mower et al., 2021).Large recombinogenic repeats are commonly found in plant mitogenomes.Although the D. takeshimana mitogenome (20) had more large repeat pairs (>1 kb) than the D. ibukiensis mitogenome (13), all D. ibukiensis repeat pairs were >2 kb.The presence of large repeats (>1 kb) in both Dystaenia mitogenomes indicates the existence of multiple major or minor alternative isoforms, as confirmed by contigs generated from hybrid assemblies of short and long reads.Therefore, additional chromosomal configuration may be present in the two Dystaenia mitochondria.Interestingly, we observed the same fragments of plastid-derived sequences in the D. takeshimana mitogenome (Figure 3), identifying a pair of repeats.The three MIPTs can be mediators of homologous recombination in the D. takeshimana mitogenome.It is likely that the fragments moved independently from the plastome and were inserted into the mitogenome or that two large fragments of plastid sequences were inserted into the mitogenome, one of which was split by genomic recombination.
Our analyses showed that gene loss and transfer to the nucleus during the evolution of Apicaeae organelles occurred only in mitogenomes.In angiosperm mitogenomes, a high frequency of 15 ribosomal proteins and two sdh gene losses have been documented (Adams et al., 2002).In the common ancestor of eudicots, two protein-coding genes, rps2 and rps11, were lost (Adams et al., 2002).The successful transfers of 10 mitochondrial genes (5′ rpl2, rpl5, rps4, rps7, rps10, rps12, rps14, rps19, sdh3, and sdh4) into the nucleus have been reported in multiple lineages (Adams et al., 2001;Park et al., 2014;Park et al., 2015;Park and Park, 2020).Comparative analyses of Dystaenia and related species indicated that three mitochondrial genes (rpl2, rps19, and sdh3) were transferred to the nucleus in the common ancestor of all analyzed Apiaceae species; one (rps10) was unique to the specific lineages, Dystaenia/Sposhnikovia clades and D. carota subsp.sativus; and rps14 was transferred to the nucleus in the common ancestor of all analyzed Apiaceae species except for Bupleurum (Figures 4, 5).The split transfer of the 3′ end of the mitochondrial rpl2 gene to the nucleus occurred in the common ancestor of core eudicots (Adams et al., 2001), in which the 5′ portion of rpl2 was present in the mitogenome.Subsequently, multiple transfers of 5′ rpl2 to the nucleus were documented in three lineages: Medicago Soybean (Adams et al., 2001) and Geranium (Park et al., 2015).We also found the 5′ RPL2 transcripts in multiple transcriptomes from Apiaceae, resulting in the complete transfer of the mitochondrial rpl2 to the nucleus during Apiaceae evolution.In particular, the two Dystaenia species and S. divaricate contained two copies of 5′ RPL2 genes, and the phylogenetic analysis indicated that its origin and evolutionary history remain unclear.In the case of RPS19, all analyzed species contained one copy, and the topology was consistent with the ML tree based on the 24 mitochondrial gene sets, although some support values at the nodes were weak.
The phylogenetic relationships of SDH3 among the analyzed Apiaceae lineages provide a good example of how mitochondrial genes are transferred to the nucleus and duplicated (Figure 5).The phylogenetic analysis of nuclear-encoded SDH3 copies suggested that a single transfer to the nucleus occurred in the common ancestor of the seven species.Duplication events occurred in the common ancestors of Dystaenia, A. graveolens, C. sativum, D. carota subsp.sativus, and S. divaricate, and, subsequently, the loss of one copy occurred independently in D. carota subsp.sativus.The nuclear-encoded RPS14 gene was found in all transcriptomes except Bupleurum, suggesting that functional replacement occurred in the common ancestors of Dystaenia, A. graveolens, C. sativum, D. carota subsp.sativus, and S. divaricate, although only the loss of mitochondrial rps14 occurred independently in A. graveolens and D. carota subsp.sativus.The coexistence of mitochondrial and nuclear rps14 homologs within Dystaenia, C. sativum, and S. divaricate indicates that successful functional replacement of the nucleus was necessary before the original mitochondrial copy was lost.In the case of Dystaenia rps14, two copies were detected in the nucleus.Phylogenetic analysis suggests that two independent transfers of mitochondrial rps14 to the nucleus occurred in the common ancestor of Dystaenia and that the copies acquired a targeting sequence.However, the second copy from both species may fail to function as a replacement because the targeting sequence from the second copy of D. takeshimana has a weak signal (Table 2), and the second copy of D. ibukiensis loses the targeting sequence and has multiple internal stop codons.Similar to Dystaenia, the Rhazya stricta genome contains two nuclear copies of rps14 and one mitochondrial copy; one of the nuclear copies is a pseudogene (Park et al., 2014).However, phylogenetic analysis supports a single transfer of mitochondrial rps14 to the Rhazya nuclear genome (Park et al., 2014).
Although multiple transcripts for mitochondrial sdh4 contained parts of the PPIase and sdh4 domains, our analyses did not find sufficient evidence for mitochondrial target peptides to shuttle the product back to the mitochondria.However, the Arabidopsis thaliana mitogenome also contains a truncated sdh4 gene at the C-terminus, previously annotated as a pseudogene.The nuclearencoded SDH4 gene presence, which contains an mTP and a partial domain of sdh4, suggests the possibility of split gene transfer and fission of sdh4 in the mitogenome.MIPT was located upstream of the truncated rpl5 gene, suggesting that the transfer event disrupted the ORF of rpl5, resulting in a loss of functionality.However, our analysis found no evidence for IGT of the truncated mitochondrial rpl5 in the D. ibukiensis mitogenome.The evolutionary fate of mitochondrial rpl5 in D. ibukiensis and the fate of mitochondrial sdh4 among the analyzed genera were unclear from the present data.Additional deep genome sequencing will be required to address this question.
In addition to IGT to the nucleus, inter-compartmental transfers between plant organelle genomes have been documented; plastid-to-mitochondrial DNA transfers are common, but transfers in the opposite direction are rare (Mower et al., 2012).Within angiosperm plastomes, mitochondrion-toplastid DNA transfers have been documented in five families: Anacardium in Anacardiaceae (Rabah et al., 2017); Caucalis, Crithmum, Cuminum, Daucus, and Petroselinum in Apiaceae (Iorizzo et al., 2012;Downie and Jansen, 2014;Downie and Jansen, 2015;Spooner et al., 2017), the tribe Asclepiadeae of Apocynaceae (Straub et al., 2013); Convallaria in Asparagaceae (Raman et al., 2019); and Paspalum and Pariana in Poaceae (Ma et al., 2015;Burke et al., 2016).Most fragments of mitochondrial DNA were inserted into their plastome-IR region, except for Crithmum, Petroselinum, and tribe Asclepiadeae.The foreign DNA of Crithmum and Petroselinum was found between the IR A and LSC gene trnH-GUG, and the tribe Asclepiadeae plastomes contained PLMT sequences between rps2 and rpoC2 genes in LSC.We discovered a mitochondrial insertion in the D. ibukiensis plastome trnH-psbB intergenic spacer in the LSC and inferred that the insertion sequences were associated with IR boundary shifts (expansion and contraction) (Figure 6).The complete D. ibukiensis mitogenome provided strong evidence for the occurrence of DNA transfer into the plastid counterpart, showing that the PLMT was fragmented into two in the mitogenome (Figure 6).However, the mitochondrial DNA was likely to transfer intact as plant mitogenomes have a high frequency of genomic rearrangements between recombinationally active repeats.Only D. carota and Asclepias syriaca had complete mitogenomes, which confirmed that the PLMTs comprise two or three pieces in their mitogenomes, also suggesting a single DNA transfer event (Iorizzo et al., 2012;Straub et al., 2013).
The integration of mitochondrial DNA into the D. ibukiensis plastome can be comprehensively explained by considering the hypothesis of invasion and DSB (Figure 6).We propose two hypotheses to explain how the mitochondrial DNA fragment was integrated into the D. ibukiensis plastome (Figure 6B).In both hypotheses, plastid-to-mitochondrial DNA transfer occurred first, followed by mitochondrion-to-plastid DNA transfer.In this case, the transfer portion of the mitochondrial DNA could be a circular or linear molecule containing MIPT, which acts as a homologous region pairing between the donor and recipient sequences.After mitochondrial DNA is imported into the plastid, it can be integrated into the plastome by two different processes: 1) a displacement loop (D-loop) associated with IR boundary shifts facilitates pairing with a homologous anchor at an MIPT site and strand invasion to initiate the recombination repair process; and 2) fusion between the donor and recipient DNA occurs within or at the end of the interacting microhomologous sequences (illegitimate recombination), followed by the deletion of a segment of the recipient DNA.Finally, the integration of mitochondrial sequences is completed by DNA replication, and a double-strand break (DSB) occurs before the interaction with the plastome.Next, integration occurs as a single-stranded DNA molecule is annealed at an MIPT site to a complementary single-stranded overhang present at a DSB site in the plastome.Finally, integration is completed by second-strand synthesis and ligation via nonhomologous end-joining.
The double D-loop strategy has been widely recognized as a fundamental aspect of plastid DNA replication (Heinhorst and Cannon, 1993).One potential mechanism that could contribute to IR expansion in the plastome was also associated with a DSB, which is subsequently followed by strand invasion, expansion, and recombination within the IR region (Goulding et al., 1996).Homology-facilitated illegitimate recombination between short regions has been observed in plastomes (Ogihara et al., 1988;Marećhal et al., 2009).Previous studies have shown that illegitimate recombination sites have GC-rich microhomologies of 3 bp to 10 bp (Prudhomme et al., 2002;De Vries et al., 2004).We present two hypotheses to explain how mitochondrial DNA can integrate into the plastome (Figure 6).In both hypotheses, identifying MIPT sequences adjacent to the transferred sequences in the D. ibukiensis mitogenome strongly indicated that homologous recombination plays a crucial role in facilitating the integration of its plastome.Both invasion and DSB are possible scenarios.The formation of a D-loop bubble is the predominant mechanism in plastid DNA and can facilitate homologous DNA pairing.The PLMT region in the D. ibukiensis plastome is associated with IR expansion, and DSB repair can occur via nonhomologous end-joining.Furthermore, identifying microhomologous sequences at the ends of the transferred sequences in the D. ibukiensis plastome suggested the possibility of illegitimate recombination.TE can be mobilized within a genome (Huang et al., 2012), and LTR retrotransposons can target tRNA genes.In the D. ibukiensis mitogenome, three TEs were detected around the transferred DNA, one within the transferred DNA, and two in either of the flanking regions (Figure 6), which were categorized as gypsy LTR retrotransposons.PLMT in the D. ibukiensis plastome is located upstream of the trnH-GUG.Thus, mitochondria-located PLMT may move from the mitogenome to the plastome.Two distinct mechanisms have been postulated to underlie these events: mitochondrial integration into the D. carota and A. syriaca plastomes were inferred by the likely mechanism of non-LTR retrotransposons and repair of a DSB by homologous recombination, respectively (Iorizzo et al., 2012;Straub et al., 2013).A physical connection between chloroplasts and mitochondria may facilitate the exchange of materials between these organelles.Stromules are dynamic tubular structures extending from the chloroplast surface of plant cells (Köhler et al., 1997;Hanson and Sattarzadeh, 2008).Stromules have been observed to interact with other organelles, including mitochondria, peroxisomes, and the endoplasmic reticulum (ER) (Hanson and Conklin, 2020), suggesting a potential role in inter-organelle communication and the exchange of molecules.The ER is a prevalent mediator in facilitating interactions among the other organelles (Mathur et al., 2022), suggesting that the delivered DNA may be able to be imported into the chloroplast through the ER.A previous study has shown that circular or linear foreign DNA can enter the chloroplast envelope under stressful environmental conditions (Cerutti and Jagendorf, 1995).

1
FIGURE 1 Plastid genome maps of Dystaenia.(A) D. takeshimana.(B) D. ibukiensis.Genes on the inside and outside of the map are transcribed in clockwise and counterclockwise directions, respectively.The hypothetical model of plastome rearrangement by IR expansion and contraction in the genus Dystaenia (right).
The loss of sdh4 occurred independently in D. carota subsp.sativus and B. chinense, and loss of rps10 occurred independently in D. carota subsp.sativus and in the two Dystaenia species/S.divaricate clade.The loss of rps14 occurred independently in A. graveolens and D. carota subsp.sativus.The truncation of sdh4 occurred in the common ancestor of A.

2
FIGURE 2 Mitochondrial genome maps of Dystaenia.(A) D. takeshimana.(B) D. ibukiensis.Genes on the inside and outside of the map are transcribed in clockwise and counterclockwise directions, respectively. 4

FIGURE 3
FIGURE 3 Distribution of repetitive DNA in two Dystaenia mitogenomes.(A) D. takeshimana.(B) D. ibukiensis.Black lines within circular maps indicate the positions of the pairs of repeats, with crossed connecting lines denoting reverse repeats.Black boxes on the inner and outer circle indicate the positions of mitochondrial genes.Green and red lines inside of the circular maps indicate mitochondrial DNAs of plastid origin and transposable elements, respectively.
(A) Schematic diagram of mitochondrial gene transfers from mitochondrial genome to the D. takeshimana nuclear genome.(B) Phylogenetic distribution of gene content among seven Apiaceae species.Park and Park 10.3389/fpls.2023.1283292

FIGURE 5
FIGURE 5 FIGURE 6 Intracellular DNA transfer events to the plastids in Dystaenia.(A) Schematic diagram of genomic regions surrounding the plastid DNA of mitochondrial origin (PLMT) and mitochondrial DNA of plastid origin (MIPT) from D. ibukiensis organelle genomes.Genome maps are based on the results of Figures 1, 2. The IR A and IR B indicate inverted repeats in the plastome.(B) Models of intracellular DNA transfer events in D. ibukiensis organelle genomes.The hypothetical models illustrate the potential integration of PLMT from the mitogenome into the plastome through invasion or double-stranded break (DSB) mechanism after plastid-to-mitochondrion DNA transfer.

TABLE 2
Transit peptide prediction of nuclear-encoded ORFs.