The order Alismatales is a hotspot for evolution of plant mitochondrial genomes characterized by remarkable differences in genome size, substitution rates, RNA editing, retrotranscription, gene loss and intron loss. Here we have sequenced the complete mitogenomes of Zostera marina and Stratiotes aloides, which together with previously sequenced mitogenomes from Butomus and Spirodela, provide new evolutionary evidence of genome size reduction, gene loss and transfer to the nucleus. The Zostera mitogenome includes a large portion of DNA transferred from the plastome, yet it is the smallest known mitogenome from a non-parasitic plant. Using a broad sample of the Alismatales, the evolutionary history of ribosomal protein gene loss is analyzed. In Zostera almost all ribosomal protein genes are lost from the mitogenome, but only some can be found in the nucleus.
Citation: Petersen G, Cuenca A, Zervas A, Ross GT, Graham SW, Barrett CF, et al. (2017) Mitochondrial genome evolution in Alismatales: Size reduction and extensive loss of ribosomal protein genes. PLoS ONE 12(5): e0177606. https://doi.org/10.1371/journal.pone.0177606
Editor: Dan Mishmar, Ben-Gurion University of the Negev, ISRAEL
Received: November 23, 2016; Accepted: April 28, 2017; Published: May 17, 2017
Copyright: © 2017 Petersen et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All sequences from this study are available from the GenBank data base (accession numbers KX808392, KX808393, KX808234-KX808391).
Funding: This work was funded by the Danish Council for Independent Research | Natural Sciences (grant no. 12-126713 to OS and GP), NSERC (Natural Sciences and Engineering Research Council of Canada) Discovery Grant to SWG, and US National Science Foundation (grants DEB 0830036 to SWG and DEB 0830020 to JID). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exists.
Compared to the mitochondrial genome (mitogenome) of most other eukaryotes, the structure of the mitogenome of land plants is astonishingly dynamic. More than a hundred mitogenomes covering most major clades of land plants have now been completely sequenced, and particularly the more than 75 mitogenomes from angiosperms reveal an intriguing evolutionary history: changes in size, structure, gene and intron content, horizontal gene transfer (HGT) and transfer of DNA between genomic compartments, extensive RNA editing, extreme differences in substitution rates and frequent incorporation of retrotranscribed sequences (reviewed in [1, 2]).
Plant mitogenomes are usually shown as single, circular structures, but increasing evidence suggests that they can be multichromosomal with the individual structures taking diverse shapes, i.e. being circular, linear or branched . Among angiosperms most mitogenomes have been depicted as 200–800 kb circular structures, however, the smallest being only 66 kb and the largest >11 Mb and assembled into more than 100 individual fragments [4, 5]. The ancestral angiosperm mitogenome is assumed to have included 41 protein coding genes: 20 encoding subunits of the respiratory chain complexes, 15 encoding proteins of the small and large ribosomal subunits, four involved in cytochrome c maturation, one encoding a translocase subunit and one encoding an intron maturase [1, 2]. Whereas the two sdh genes involved in the respiratory chain complex II and many of the ribosomal protein genes are recurrently lost, the remaining genes have been found in almost all angiosperms [1, 2, 6, 7]. Loss of protein coding genes from the mitogenome may occur following functional transfer of a gene to the nucleus . In addition to protein coding genes, the mitogenome includes three ribosomal RNA genes universally present (possibly with the exception of Viscum ) and approximately 20 tRNA genes, most of which are frequently lost.
Import of foreign DNA is one of the mechanisms responsible for mitogenome size expansion. The most common source of foreign DNA is the plastome . DNA of plastid origin has been found in all angiosperm mitogenomes investigated so far with various quantities ranging from 2–3 kb in Arabidopsis, Silene and Vigna to 113 kb in Cucurbita and 138 kb in Amborella [2, 10, 11]. Import of nuclear sequences to the mitogenome may occur but is difficult to detect. The availability of nuclear genomes is limited, and direction of potential transfers is not trivial to determine . The most frequent nuclear sequences reported from angiosperm mitogenomes are transposable elements, though rarely in monocots, including the Alismatales [2, 12, 13]. Acquisition of foreign DNA through HGT is another means of size increase. HGT has been described primarily in parasitic plants, but the most extraordinary case is non-parasitic Amborella, where the mitogenome, through fusion with entire foreign mitogenomes, has experienced a roughly six-fold size increase [11, 14].
To address questions related to mitogenome evolution appropriate taxon sampling is needed (e.g., [4, 15–17]). In this study we focus on a clade of monocots, Alismatales, which several studies have shown to be a hotspot of mitochondrial anomalies, including pronounced intron loss, substitution rate variation, extremely different levels of RNA editing and potential abundant incorporation of retrotranscribed gene sequences [18–22]. The Alismatales includes three major clades: the Araceae, the Tofieldiaceae and a clade of 10 families here referred to as the core alismatids [22, 23]. The known mitochondrial anomalies are confined to the latter clade. By adding complete mitogenome sequences from two core alismatids, Stratiotes aloides (Hydrocharitaceae) and Zostera marina (Zosteraceae), to data already available [12, 13] we widen previous comparative studies of Alismatales mitogenomes . Intergenomic transfer of sequences are explored using the complete nuclear genome sequence of Zostera marina  and the large amount of plastome gene data from Alismatales provided by Ross et al. . Additionally, we provide gene sequence data from Triantha (Tofieldiaceae) and 27 species representing all families of the core alismatids. We will use these data to study the evolution of ribosomal protein genes in phylogenetic framework.
Material and methods
Mitochondrial genome sequencing, assembly and annotation
Specimens of Zostera marina L. (specimen voucher: A. Cuenca C2544, Denmark, Humlebæk Strand 6 Sept. 2009 (C)) and Stratiotes aloides L. (specimen voucher: O. Seberg et al. C2459, Denmark, Køge Bugt 10 Sept. 2008 (C)) were collected in the wild, and no specific permits were required for the collection of the material. The species are not protected by Danish law and they are collected in public areas where no permits are needed.
The complete mitogenomes were assembled using a combination of 454 and Illumina sequencing data. The 454 data generated from mitochondrial enriched DNA extractions, were re-used from Cuenca et al. . For llumina sequencing, DNA was extracted from silica gel preserved material using a standard CTAB method . Short-insert, paired-end (PE) libraries with average insert size of 500 bp were constructed and run in 1/16 of a lane on an Illumina HiSeq 2000 (Illumina, San Diego, CA). Libraries and sequencing data were produced at the Danish National High-Throughput DNA Sequencing Centre. Raw reads were trimmed for quality, adapters and unidentified nucleotides using AdapterRemoval .
454 reads were initially assembled in Newbler v. 2.3 (454 Life Sciences Corp, CT, USA) using default settings. Contigs were ordered using the program bb.454contignet (http://www.vcru.wisc.edu/simonlab/sdata/software/) that reads the assembling information from Newbler and depicts the contigs as a web, indicating contig connections . Contigs were then extended and assembled as described by Cuenca et al. . To identify reads of adjacent contigs and to determine the borders of duplications, contigs were extended by blasting approximately the last 75 nt of each contig border against a database of all raw 454 sequence reads. These BLAST analyses were done using the BLASTN program  in stand-alone BLAST ver. 2.2.21 (ftp://ftp.ncbi.nlm.nih.gov/blast/executables/blast+/LATEST/). Consensus sequences of each contig were used as seed sequences and extended using both 454 reads and Illumina reads in the Short Sequence Assembly by K-mer search and 3' read Extension program, SSAKE ver. 3.5 , with parameters -m 15 -o 2 -r 0.6 -p 0 -t 0 -v 1. Finally, all 454 and Illumina reads where mapped to the assembled mitogenomes using Geneious ver. 7.1 (Biomatters Ltd.) to verify the assemblies, evaluate coverage and correct for potential homopolymer length errors in particular attributable to the 454 reads.
Sequences of protein coding genes and rRNA genes were identified by BLASTN using a local database of extracted gene sequences from the 27 angiosperm species including Butomus umbellatus (NC021399 ) and Spirodela polyrhiza (NC0178840 ) from the Alismatales (S1 Table). In addition to complete protein coding genes with intact reading frames, fragments of known genes >100 bp were annotated. A sequence was recognized as a pseudogene if it had a length comparable to known functional genes, but could not be translated into an amino acid sequence even following potential RNA editing. The tRNA genes were identified using tRNAscan-SE 1.21 . Annotation was performed manually in Geneious vers. 7.1–9.0 (Biomatters Ltd.). The assembled and annotated genomes are deposited in GenBank under accession numbers KX808393 (Stratiotes aloides) and KX808392 (Zostera marina).
To verify deviating gene sequences in the mitogenome of Zostera marina, we produced Illumina sequences for another species of Zostera, Z. noltii Hornem. (specimen voucher: Seberg et al. C2453, Denmark, N of Munkholmbroen 10 Sept. 2008 (C)). DNA extraction and sequencing were performed as above, except that the library was run on an Illumina HiSeq 2500. Complete assembly of this mitogenome was not attempted, but the genes except the tRNA genes, were extracted after mapping the Illumina reads to the Zostera marina mitogenome using Geneious ver. 8.0 (Biomatters Ltd.). Reads mapping to identified, complete or partial genes were used for de novo assembly of individual loci, and if necessary in order to obtain complete gene sequences one or more rounds of Map to Reference as implemented in Geneious ver. 8.0 (Biomatters Ltd.) were used to extend the sequence of individual loci. A local database of genes from other completely assembled mitogenomes (S1 Table) was used as reference sequences to search for genes not present in Z. marina. Identified genes and flanking sequence of Z. noltii are deposited in GenBank under accession numbers KX808258-KX808296.
To identify regions of potential plastid origin in the complete mitogenomes of Zostera and Stratiotes we performed a BLASTN search against a database of 23 angiosperm plastid genomes, including genomes from the genera Elodea (Hydrocharitaceae), Lemna and Spirodela (Araceae) from the Alismatales (S1 Table). Plastid genome sequences of Stratiotes and Zostera are not available. Only mitochondrial sequences larger than 100 bp and with a similarity score higher than 80% were considered. If the hits included protein coding gene sequence we used the mitochondrial sequence matching the plastid protein coding gene to perform a new BLASTN search against all sequences in GenBank and against local databases of plastid gene sequences created from the data provided by Ross et al.  available at figshare.com (DOI: 10.6084/m9.figshare.1407422.v1). We also included the protein coding sequences in phylogenetic analyses using the matrices from Ross et al. . We realigned the sequences after inclusion of the new sequence copies found in the mitogenomes using the MUSCLE  plugin in Geneious ver. 8.1. Subsequent phylogenetic analyses were performed using RAxML ver. 7.2.8  with 100 replicates of rapid bootstrapping and a GTR+GAMMA+I model as implemented in Geneious ver. 8.1.
To identify regions of potential nuclear origin in the mitogenomes of Zostera and Stratiotes we searched for repetitive elements using the Repbase Update repetitive element database . For Zostera we also performed a BLASTN search (maximum E-value = 1e-50) of the complete mitogenome against a local database created in Geneious ver. 9.0 including all the contigs from the Zostera marina nuclear genome (GenBank no. LFYR00000000 ). BLASTN results of sequences longer than 250 bp and a pairwise similarity >80% were inspected for sequence features, which could indicate the direction of potential transfers. E.g., finding a sequence in the mitogenome including (part of) a gene normally located in the nuclear genome would indicate transfer from the nuclear genome to the mitogenome.
We searched for dispersed repeated sequences in the complete mitogenomes of Zostera and Stratiotes by blasting the sequences against themselves using BLASTN. The results were filtered to retain only matches of sequences longer than 100 bp and pairwise similarity >80%.
To evaluate the conservation of gene order in the Alismatales, we searched for clusters of genes shared between the complete mitogenomes of Spirodela, Butomus, Zostera and Stratiotes. Gene clusters are defined as two or more adjacent genes in the same direction shared by at least two taxa. This was done by simple visual inspection of the annotated mitogenome sequences.
To get a rough estimate of mitogenome DNA similarity between pairs of Alismatales species we made pairwise BLASTN analyses with the four complete mitogenome sequences. Due to the heuristic nature of BLAST, blasting a sequence A against B, does not necessarily give the same result as blasting B against A. Accordingly, all analyses were done in both directions and average values were calculated as in Guo et al. .
Ribosomal protein genes in Alismatales
Illumina sequencing data from Triantha and 26 core alismatids provided by Ross et al.  were used to retrieve ribosomal protein coding genes as we previously retrieved other protein coding genes from the same data . The sequences are deposited under accession numbers KX808234-KX808257 and KX808297-KX808391. Assuming that genes can be lost, but not gained (see below), we mapped loss of ribosomal protein genes on a phylogenetic tree of the Alismatales using Dollo parsimony. We used a tree modified from the result of a total evidence analysis based on morphology, plastid and mitochondrial data (Figure 1 in ). The tree was modified by reducing the taxon content to match the present sampling and by assuming that the relative positions of Spirodela and Triantha are the same as for the representatives of Araceae and Tofieldiaceae, respectively, as used by Petersen et al. .
The nuclear genome of Zostera  includes several NUMTs (nuclear mitochondrial sequences) that contain mitochondrial ribosomal protein genes. We compared the ribosomal protein NUMTs by BLASTN to a local database including the Alismatales mitochondrial gene sequences plus sequences from up to 50 other angiosperms (S1 Table), and to assess the origin of two NUMTs we performed phylogenetic analyses using the same sequences as included in the local databases. The sequences were aligned using the default settings in MUSCLE  as implemented in Geneious ver. 9.0. Subsequent phylogenetic analyses were performed using RAxML ver. 7.2.8  with 100 replicates of rapid bootstrapping and a GTR+GAMMA+I model as implemented in Geneious ver. 9.0.
Mitogenome size and gene content of Stratiotes and Zostera
The Stratiotes aloides mitogenome could be assembled into a 349,058 bp circular structure whereas the mitogenome of Zostera marina was assembled into a shorter linear structure of just 191,481 bp (Fig 1, Table 1). Details of read numbers and coverage are listed in S2 Table. The large number of repeats in Zostera makes numerous submolecular structures possible, but we failed to map reads into a master circle unless accepting a duplication of more than 68 kb, which is not supported by read coverage. Although larger duplications have been reported , we prefer to draft a linear genome.
For Zostera upper genes are transcribed forward; lower genes are transcribed reverse. For Stratiotes inner genes are transcribed clockwise; outer genes are transcribed counter-clockwise. Genes are color coded as indicated in the lower left corner. The names of trans-spliced genes are followed by exon numbers. Only genes assumed to be functional are shown. The figure was created using OGDRAW .
Mitogenome characteristics are summarized in Table 1, and a complete list of genes including length information is provided in S3 Table. With regard to gene content, the two new complete genomes differ in their content of ribosomal protein genes and tRNA genes, only. Stratiotes has four ribosomal protein genes (rpl10, rps1, rps3, rps12) and two pseudogene sequences (rpl16, rps7), whereas Zostera marina only has one complete gene (rps7) and a fragments of another (rps3 exon2). In Zostera noltii two rps3 fragments were detected (S3 Table). Stratiotes also has more tRNA genes than Zostera marina. A total of 11 tRNA genes normally found in plant mitogenomes were detected in Stratiotes (S3 Table), however three of them are located in regions assumed to be recently transferred from the plastome (see below). In Zostera marina we found 8 tRNA genes, two of which are of recent plastid origin (see below).
The protein coding genes of Stratiotes and the two species of Zostera differ slightly with regard to introns (Tables 1 and 2). Due to the presence of rps3 (which includes a cis-spliced intron) in Stratiotes but not in Zostera, the total number of introns differs (21 vs. 22), and in Stratiotes nad1i728 is cis-spliced but trans-spliced in Zostera. For two of the protein coding genes (nad1 and nad4L) the start codon is created by RNA editing in both Stratiotes and Zostera. In Zostera editing is confirmed by comparison to the EST library data from Zostera marina (GenBank, SRR057351). In Zostera an alternative start codon of atp9 and cob may be used, possibly GTG in atp9 and ATA in cob.
Dispersed repeats in Stratiotes and Zostera
The two new, complete mitogenomes differ considerably both in number and pattern of repeated sequences (Table 1, S4 Table). Stratiotes has only 12 repeated sequences longer than 100 bp, and the longest is just 382 bp, thus accounting for only approximately 0.6% of the entire mitogenome. In Zostera 209 repeats were detected, the longest more than 10 kb. However, a number of the suggested repeats in Zostera are overlapping, representing different estimates of the repeats with just slightly different lenghts. Thus, the true number of repeats is lower than 209. Due to this ambiguity, the proportion of repeated sequences in the genome is also difficult to assess precisely, but we estimate that repeats account for approximately 20% of the genome.
Insert of foreign DNA in the mitogenomes of Stratiotes and Zostera
The mitogenomes of both Stratiotes and Zostera include a substantial amount of plastid DNA. In Stratiotes 11 fragments ranging in size from 641–12,253 bp with a total length of 49,392 bp (corresponding to 14.2% of the mitogenome) were detected, and in Zostera 27 fragments in the range of 140–6,614 bp, totaling 39,563 bp (corresponding to 20.5% of the mitogenome), were found (Table 1, S5 Table). Disregarding sequences that are repeated in the mitogenomes, the inserts in Zostera and Stratiotes correspond to 31,678 bp and 50,261 bp of the complete plastome of Elodea canadensis (NC018541), respectively. For the previously published mitogenome of Butomus, we calculated that the 6,928 bp of plastid inserts correspond to 7,235 bp of the Elodea plastome (S5 Table).
Results from blast searches and phylogenetic analyses of inserted plastid sequences are listed in S6 Table. The gene trees are not shown, but the placement and bootstrap support values for the relevant clade including the transferred plastid sequence are listed. For both Stratiotes and Zostera marina almost all inserted gene sequences, and in particular those included in long inserts, are most similar to sequences from Stratiotes and Zostera muelleri (the data set from Ross et al.  used for similarity searches and phylogenetic analyses includes Z. muelleri instead of Z. marina), respectively, and the sequences are grouped together in the gene trees (not shown). Noteworthy exceptions are two inserts in Stratiotes: a 931 bp insert including a partial sequence of ndhB that is most similar to Potamogeton (Potamogetonaceae) and placed in the gene tree as sister to a clade including Ruppiaceae, Cymodoceaceae, Zosteraceae and Potamogetonaceae, and a 569 bp insert including a partial ndhK sequence that is most similar (70.3%) to Illicium (Schisandraceae) and placed together with a transferred gene sequence from Butomus within early diverging angiosperm lineages in a gene tree. In contrast to the plastid inserts in Zostera and Stratiotes, only the longest insert (4,897 bp) in Butomus seems to stem from intraspecific, intergenomic transfer. The remaining inserts, including mostly very short gene fragments, show similarity and weakly supported or unsupported relationships to different groups of angiosperms (S6 Table).
Insertion of nuclear DNA could be most precisely investigated in Zostera marina where the complete nuclear genome has been sequenced . Blasting the complete mitogenome sequence of Zostera against the nuclear genome resulted in 329 matches fulfilling the search criteria. A total of 57 matches included sequences >250 bp long and with a similarity >80% (S7 Table). Of these, 17 matches involve plastid sequences, 25 involve sequences with part of or rarely a complete mitochondrial gene, and the remaining do not include sequences with any particular features. Eight matches between highly similar sequences involve an approximately 38 kb long sequence of the mitogenome and approximately 22 kb of the nuclear scaffold 54 (LFYR01001529). The 38 kb mitogenome sequence includes several complete genes (trnM, rrn5, rrn18, ccmFn, rps7, cox2, mttB), a large fraction of nad4 and five regions of plastid sequence, but the nuclear scaffold has some long assembly gaps, e.g., where plastid regions, rrn18 and cox2, should be located according to the mitogenome sequence.
A search for nuclear repetitive elements in both the Zostera and Stratiotes mitogenome retrieved no convincing positive results. Though the search criteria used here suggested similarity (82–96%) of three short sequences (121–310 bp) from each of Zostera and Stratiotes to repetitive elements, the sequences are all more similar to part of a chloroplast insert (92–100%) or to a mitochondrial rRNA gene (100%), and thus do not likely originate from repetitive elements.
Mitogenome similarity and gene clusters in Alismatales
Mitochondrial DNA similarity between pairs of species of Alismatales is given in Table 3. Butomus and Stratiotes share the greatest amount of highly similar DNA (112 kb) whereas the least is found in a comparison between Spirodela and Zostera (58 kb).
Gene order is conserved very poorly in Alismatales. Only 11 colinear gene clusters including two or three genes were found (Table 4). A single cluster (rrn18-rrn5) is shared by all four taxa, six are shared by three taxa, but only one of them (nad3-cox3) by the core alismatids.
Loss and transfer of ribosomal protein genes in Zostera and other core alismatids
A survey of ribosomal protein genes in 31 representatives of Alismatales revealed a substantial loss of genes from the mitogenome in all core alismatids (Fig 2). Triantha has an almost full complement with 14 ribosomal genes, but only one (in Zostera) to six genes (in Scheuzcheria) were found in the core alismatids. An evolutionary hypothesis about gene loss and pseudogenization is created by mapping the events on a phylogeny of the Alismatales assuming that genes may be lost but not gained, and that pseudogenes do not revert to functional copies (Fig 3). Nine genes (rpl2, rpl5, rpl16, rps2, rps4, rps10, rps13, rps14, rps19) are lost only once either completely or first functionally by pseudogene formation. The remaining six genes have been lost at least twice.
Blank boxes indicate absence of a gene from the mitogenome. Dark grey boxes indicate presence of a gene assumed to be functional. Light grey boxes and Ψ indicates a pseudogene or fragment of gene in the mitogenome. N indicates presence of a gene annotated as of mitochondrial origin in the nuclear genome of Zostera marina. (n) indicates likely presence of a gene in the nuclear genome of Zostera marina, though not annotated as of mitochondrial origin. *Zostera noltii has the same mitochondrial genes, but the potential nuclear location of genes is unknown.
Bars represent events of loss or pseudogenization. Black bars mark unique events of gene loss from the mitogenome. Red (pseudogenization) and blue (loss) bars mark homoplasious gene changes.
In the nuclear genome of Zostera marina five genes (rps2, rps4, rps7, rps12 and rps14; NCBI; LFYR00000000) are annotated as ribosomal protein genes of mitochondrial origin (NUMTs). Three of them (rps2, rps4, rps14) are missing from the mitogenome of all core alismatids, and the similarity of the NUMTs in Zostera to the mitochondrial genes in Triantha and Spirodela is small (approximately 40–60%). The rps12 gene is missing from the mitogenome only in Zostera and Ranalisma (Alismataceae). In most Alismatales the coding sequence of rps12 includes one reading frame of 378 bp, whereas the nuclear Zostera gene is split in two exons of 133 bp and 377 bp (scaffold 89, LFYR01002109), the second corresponding to the mitochondrial gene. Similarity between the nuclear rps12 exon2 and mitochondrial rps12 sequences is close to 75% for all Alismatales, and the phylogenetic analysis place the NUMT with very weak support (30% bootstrap) as the sister to a clade including Maundia, Triglochin, Halodule, Amphibolis, Potamogeton and Zannichellia. The mitochondrial rps7 gene is missing or pseudogenized in some core alismatids, but present in Zostera. The rps7 NUMT, which is located in scaffold 54 (LFYR01001529), is included in the 22 kb region described above and the gene sequence is identical to the sequence of the mitochondrial gene copy. Transcriptome data from Z. marina available in GenBank (SRX021650) indicate that rps7 transcripts are C to U edited, hence stemming from a mitochondrial gene.
A BLASTN search for the remaining mitochondrial genes both missing from the Zostera mitogenome and not annotated in the nuclear genome was performed with genes from other alismatids as templates. This gave a match to a full length coding sequence (480 bp) 81% similar to the rps1 gene in Potamogeton. The potential NUMT is included in a larger nuclear region annotated as a hypothetical protein (LFYR01001306, ZOSMA_440G00060). A phylogenetic analysis of the NUMT places it with very weak support (44% bootstrap) as sister to Triglochin (Juncaginaceae), however, the group is strongly supported within a clade also including the Cymodoceaceae (98% bootstrap). Another BLAST match (85.2% similarity) was found between the mitochondrial rpl10 gene (279 bp) from Potamogeton and a short nuclear sequence (169 bp) in Zostera. The nuclear sequence is part of a larger annotated hypothetical protein gene (LFYR01001739, ZOSMA_65G01050), but beyond the region with BLAST match the sequence has very limited similarity to rpl10 genes, even at the amino acid level.
Mitogenome size evolution
In general, mitogenomes of Alismatales are relatively small in overall size (bp) compared to other angiosperms (see e.g., ). The genome of Stratiotes (349,058 bp) lies in between the previously sequenced Butomus (450,826 bp) and Spirodela (228,493 bp) genomes whereas the mitogenome of Zostera (191,481 bp) is the smallest of all autotrophic angiosperms sequenced so far. Until now a smaller plant mitogenome (65,873 bp) has only been described in the hemiparasite Viscum scurruloideum , but despite Viscum mitogenomes being considerably and consistently reduced in gene content, they are not consistently reduced in size. The mitogenome of V. album has a rather standard size of 565,432 bp . With only four completely sequenced mitogenomes from Alismatales it is impossible to provide a comprehensive picture of size evolution, however, both expansions and reductions have occurred. Considering the major differences observed in mitogenome size even between other closely related taxa, e.g., in the genus Silene and the family Cucurbitaceae [4, 10, 39], a general trend is not to be expected, rather, mitogenome size tends to be idiosyncratic across the land plants. Christensen [40, 41] proposed that different mechanisms of DNA repair in coding and non-coding regions of plant mitogenomes could explain the co-occurrence of low mutation rate in the coding regions and high variability, including size expansion, of non-coding regions. However, the mechanism based on a model of inaccurate double-strand breaks proposed for non-coding regions seems insufficient to explain size reductions, which most likely have occurred also in other angiosperm lineages, e.g., the Brassicales .
Consistent with a lack of a general pattern of size evolution, DNA similarity between Alismatales mitogenomes is limited (Table 3). Though the two most closely related genera, Butomus and Stratiotes, do share the greatest amount of highly similar DNA, this corresponds to only a fourth or a third of the complete mitogenomes. The approximately 112 kb of shared, similar DNA is about twice as much as the amount of genic sequence (incl. cis-spliced introns) in the two genomes (approximately 59 kb and 63 kb), thus conserved regions also involve non-coding sequences. In contrast, the lowest amount of shared, similar DNA between any pair of species, viz. Zostera and Spirodela (approximately 58 kb), corresponds very closely to the amount of gene sequence in Zostera (approximately 56 kb), showing that virtually all non-gene DNA has either been replaced or shuffled beyond conventional limits of similarity recognition.
Mitogenome size can be positively affected by the occurrence of duplicated sequences as most dramatically seen in Silene conica where more than 40% (approximately 4,600 kb) of the mitogenome is composed of repetitive DNA . The three core alismatid mitogenomes do differ considerably in amount and pattern of repeated sequences (Table 1, S4 Table), and in the two closest relatives, Butomus and Stratiotes, repeated sequences account for approximately 8.3% and 0.6% of the mitogenomes, respectively. Thus, slightly more than 1/3 of the approximately 100 kb size difference between Butomus and Stratiotes, (i.e., 37.8 kb according to Cuenca et al. ), can be explained by sequence duplications in Butomus. Surprisingly, the smallest Alismatales mitogenome, the approximately 191 kb Zostera genome, has both the highest number of repeated sequences and the largest total amount of repeated DNA, accounting for approximately 20% of the total size. Thus, as previously observed, e.g., in the asterid lineage and in Silene [4, 42], there appears to be no clear correlation between mitogenome size and content of repeated sequences.
Mitogenome size may also be affected by the incorporation of foreign DNA either by intracellular transfer from the plastid or nuclear genome or from horizontal transfer of mitochondrial DNA. The mitogenomes of the four Alismatales all include plastid DNA within the range observed in other angiosperms , however, with the smallest (approximately 7 kb) and largest (approximately 49 kb) amounts found in the closest relatives, Butomus and Stratiotes (S5 Table). The relative amount of plastid DNA in the mitogenome of Butomus (1.5% ) is among the lowest yet observed in any angiosperm, where smaller fractions (0.8% and 1.1%) have only been reported in species of Gossypium . Thus, more surprising is the discovery of almost 40 kb of plastid DNA in the mitogenome of Zostera, accounting for more than 20% of its total size. This relative fraction by far exceeds the previously reported highest relative amounts of approximately 10–12% in Boea and Cucurbita [10, 43].
The data for plastid inserts suggest a correlation between quantify of inserted DNA, length of inserts and the relative timing of insert events. The large amount of plastid DNA in Zostera and Stratiotes is coupled with the presence of several long sequences, which are transferred recently. Recent transfer is suggested both on the basis of raw sequence similarity to congeneric plastid sequences and by phylogenetic analyses (S6 Table). However, without a denser taxonomic sampling of comparable plastome sequences any absolute timing of transfer events cannot be estimated meaningfully, so we just conclude that these transfers have taken place after divergence of each of the two genera from their respective sister groups. In contrast, Butomus includes only a single long insert of such recent origin. A single insert sequence (including the ndhK gene) is found in both Stratiotes and Butomus and our data suggest that it may have an origin among early angiosperms (S6 Table). Inserts including the ndhK gene have been found in several other angiosperms including Amborella, Liriodendron and many eudicots and monocots e.g., [11, 15, 43, 44]. This, however, does not support a scenario of one ancient transfer event, as almost all the inserted sequences are more similar to their conspecific or congeneric plastome gene copies than to each other (data not shown). Thus, repeated transfer events seem more likely, unless conversion between plastome and mitogenome sequences as suggested by Clifton et al.  also has occurred repeatedly. In both Amborella and Liriodendron the inserts including ndhK are longer (>5 kb and >1 kb, respectively) [11, 15] than those we find in Stratiotes and Butomus, thus it remains a possibility that the latter sequences are remnants of a longer, ancient transfer.
Although both Zostera and Stratiotes include large fractions of plastid DNA of recent origin there are some differences. In Zostera, the total amount of plastid DNA is less than in Stratiotes and the sizes of the inserts are generally smaller, too. However, the number of inserts is larger, suggesting that transfer events either occur more frequently or that the original inserts have been fragmented subsequently (S5 Table). The high number of short non-overlapping plastid inserts, including three non-overlapping inserts of the rpoB gene, suggests that fragmentation and rearrangements have shaped the Zostera mitogenome particularly rapidly compared to both Stratiotes and Butomus.
Based on the data from Butomus, Cuenca et al.  previously suggested that the transfer of protein coding sequences from the plastid to the mitochondrial genome possibly took place through reverse transcription. The current data from Zostera and Stratiotes, which include long inserts with several genes that are not likely to be co-transcribed, do not support that hypothesis. A scenario of frequent intracellular transfer events of longer pieces of DNA followed by rapid fragmentation and rearrangement as previously inferred in other plants [44–46] seems more likely.
Intracellular transfer of DNA between the nuclear and mitochondrial genomes may also shape both genomes. However, due to lack of features in larger parts of both the nuclear genomes and mitogenomes, the direction of a potential transfer can be difficult to assign even in the few cases where the nuclear genome has been sequenced . With the complete nuclear genome of Zostera marina being available  a full comparison of the mitogenome and the nuclear genome was possible, and some intracellular transfer has clearly taken place. Some similar or even identical sequences include mitochondrial genes, thus most likely they are the result of transfer from the mitogenome to the nuclear genome. Most prominent among these is a transfer of an approximately 38 kb region of mitochondrial DNA including several complete genes (S7 Table). Based on the highly conserved sequence, the transfer most likely occurred very recently—possibly after divergence of Z. marina from its sister species. The mitochondrial sequence contains inserts of plastid origin, but the NUMT has annotation gaps at the exact location of the plastid inserts. The most simple explanation is that the plastid sequences were transferred to and incorporated into the mitogenome first, and subsequently the entire region was transferred to the nucleus. Either due to assembly ambiguity caused by sequence similarity between genomic fractions or because plastid reads may have been filtered out, the transferred plastid sequences were not correctly incorporated in the nuclear genome sequence.
The mitogenomes of Stratiotes and Zostera are very similar with regard to GC content, 45.2% and 45.1%, respectively (Table 1). While these values are completely normal compared to Spirodela (45.7%) and other angiosperms, which usually have a GC content in the range of 43–45% [2, 13], they are in contrast to the increased GC level previously found in Butomus (49.1%) . The GC content of a mitogenome may be affected by recent transfer of DNA from the plastome, which is generally characterized by a low GC content [43, 46]. Both Stratiotes and Zostera have rather large amounts of plastid DNA in their mitogenomes (see above), which bias the total GC content as the plastid inserts in both mitogenomes have a GC content of only 39.3% (Table 1), very close to the 39.0% GC content estimated by Sloan and Wu  for ancestral plastid sequences transferred to a diverse range of angiosperm mitogenomes. The GC contents of the remaining part of these two core alismatid mitogenomes are 46.2% and 46.6%, respectively, thus in the high end of the usual spectrum, but still not as high as in Butomus, where the total GC content is unaffected by the small amount of plastid DNA. The GC content of the few plastid inserts in Butomus is even slightly higher (50.2%) than that of the remaining mitogenome, consistent with the inserts primarily being older and thus having been brought into balance with the rest of the mitogenome [46, 47].
Gene content and gene features
The gene contents of Stratiotes and Zostera are normal compared to other Alismatales and to angiosperms in general with regard to ribosomal RNA genes and protein coding genes, with the exception of the ribosomal protein genes (Table 1, S3 Table). Excluding the variation in ribosomal protein gene content, the only protein coding gene that is not consistently present is sdh4, which is absent from the mitogenomes of the three core alismatids. Together with sdh3, which is also lost in all Alismatales, the sdh4 gene is frequently missing in angiosperms [7, 48]. The number of tRNA genes in the core alismatids is relatively low compared to Spirodela and most other angiosperms (see e.g., [4, 13]), however, plant mitochondria are capable of importing cytosolic tRNAs encoded in the nucleus [49, 50] and all genes appear to be dispensable.
The Alismatales is characterized by extreme variation in the occurrence of introns in the protein coding genes, with only six cis-spliced introns retained in some members of the Hydrocharitaceae . However, the species investigated in the present study are among those with a quite normal array of introns compared to other angiosperms , and only minor differences exist among them (Table 2). A few differences in intron splicing are also observed. In Butomus and Stratiotes nad1i728 is cis-spliced, but it is trans-spliced in Spirodela and Zostera. In light of the phylogeny, the simplest evolutionary explanation is a change from trans to cis-splicing, however, all evidence from other plant mitochondria suggests that evolution proceeds only from cis- to trans-splicing [1, 52]. The nad1i728 intron has changed from cis- to trans-splicing several times during angiosperm evolution [1, 53], and could have done so in the Alismatales, too. However, a second intron, nad2i542 in Zostera marina, may suggest that a change from trans- to cis-splicing is possible. This intron is cis-spliced in Isoetes and Selaginella, but trans-spliced in all seed plants . In Zostera marina we find the two flanking exons just 2,764 bp apart, which is a normal length for a group II intron. The region includes both another protein coding gene, atp8, and a tRNA gene, and the entire configuration may be the result of random events of rearrangement. Thus, it is entirely possible that nad2i542 is still trans-spliced, but the proximity of the nad2 exons poses the obvious question, whether it is possible to regain a functional cis-spliced intron.
Richardson et al.  describe the distribution of 29 colinear gene clusters among angiosperms, 17 of which were assumed to be ancestral in the angiosperms and an additional three are supposed to be ancestral in the monocotyledons. Richardson et al.  did not include the intron splicing maturase gene, matR, which is located within nad1i728 in many angiosperms as well as other vascular plants, mosses and hornworts . Fifteen of these clusters are present in Spirodela, but a number of the clusters include genes which are not present in the core alismatids (ribosomal genes and tRNA genes mainly), thus only six of the gene clusters are present in Butomus and Stratiotes and only the universally occurring rrn18-rrn5 gene cluster is also present in Zostera. Among the Alismatales five new gene clusters not present in Spirodela are shared between Butomus and Stratiotes, only one of them also shared with Zostera. Though gene order is highly variable even among closely related species or within a species [17, 37], the quantity of rearrangements having shaped the Zostera mitogenome is unique.
Mitochondrial ribosomal protein genes in the Alismatales
A survey of mitochondrial ribosomal protein genes in 280 angiosperms identified the Alismatales as a hot spot for gene loss , and the present data confirms this (Fig 2). Whereas Triantha (Tofieldiaceae), representing the sister group to Spirodela (Araceae) and the core alismatids, has an almost full complement of ribosomal protein genes, lacking only rps19, and Spirodela has retained 10 genes, the core alismatids have retained from six to just one of these genes. Among the taxa surveyed here the two species of Zostera are the only ones that have retained just one ribosomal gene (rps7). A few genes (rps1, rps3, rps12) occur in most of the taxa, but not in all, suggesting that all the ribosomal protein genes are dispensable. These data are comparable to those from the genus Silene, where some species have only one gene left [4, 55, 56], from Ajuga reptans, which retains only three genes , and from the genus Viscum, where zero to six genes are retained [5, 9].
Mapping the loss of genes from the mitogenome on a phylogeny of the Alismatales reveals that nine genes are lost only once either completely or functionally by pseudogene formation (Fig 3). Three genes are lost twice (rpl10, rps7, rps12), but the history of gene loss of the remaining three ribosomal genes within the core alismatids becomes increasingly complicated with the genes being lost five to six times. Given the phylogeny five recurrent losses of rps11 must be postulated (Fig 3). However, in this case ambiguities concerning resolution of the Alismatales phylogeny should be considered. An alternative resolution provided by plastid data only  would reduce the number of losses to three, and another based exclusively on mitochondrial data  would reduce it to two. The possibility that only one loss has occurred in Alismatales and that the occurrence of rps11 in Scheuchzeria could be attributable to horizontal gene transfer, is not likely though it cannot be excluded. The rps11 gene in Scheuchzeria is more similar to the rps11 in Triantha than to any other available sequence, and thus consistent with vertical rather than horizontal transfer. The rps11 gene is lost repeatedly in the monocots and in all core eudicots , thus repeated loss in Alismatales may not be surprising. The phylogeny also implies five recurrent losses of rps1. All are mapped to terminal branches and none of these postulated losses are affected by the choice among the recently published phylogenies [21–23]. The greatest number of complete or functional losses of ribosomal protein genes is detected in rps3, which is lost six times from terminal branches in the phylogeny. Even in this case, the number of invoked losses is unaffected by the choice of phylogeny. Whereas Adams et al.  reported multiple losses of rps1 in the angiosperms they found only few losses of rps3, thus the recurrent losses in the core alismatids are surprising.
Whether genes missing from the mitogenome are completely lost or transferred to the nuclear genome is rarely known in detail, but in Zostera marina the complete nuclear genome sequence  provides evidence that at least four of the ribosomal protein coding genes have been transferred to the nucleus. Three of these genes (rps2, rps4, rps14) are missing from the mitogenome of all core alismatids. These NUMTs in Zostera are all very divergent compared to the mitochondrial genes in Triantha and Spirodela (approximately 40–60% similarity), suggesting ancient transfer events. Thus, rps14 could have been functionally transferred to the nucleus in a common ancestor of the core alismatids and Araceae, and rps2 and rps4 subsequently in the common ancestor of the core alismatids (Fig 3). Each of the transfer events could potentially have occurred previously and include a period of time with co-existence of nuclear and mitochondrial gene copies as found in some other plants (e.g., ). The rps2 gene has possibly been lost from the mitogenome of all eudicots and functionally transferred to the nucleus , but even if we now have evidence that it is also functionally transferred in Zostera and thus possible so in all core alismatids, we cannot be certain whether this involves a second transfer event or whether some groups (e.g., Araceae and Tofieldiaceae) have retained the mitochondrial gene copy. Functional transfers of rps14 have occurred in both eudicots and members of the Poales (e.g., [42, 57, 58]), and a functional transfer of rps4 is found in Geraniaceae , thus for these two genes the transfers observed in core alismatids are independent events.
The rps12 gene, which is only rarely lost from angiosperm mitogenomes , is present in most members of the Alismatales except Zostera and Ranalisma, suggesting that it was either lost/transferred twice, recently, or that a copy was transferred in a common ancestor. In most Alismatales the coding sequence of rps12 includes one reading frame, whereas the nuclear gene in Zostera is split into two exons (scaffold 89, LFYR01002109), the second corresponding to the mitochondrial gene. Expansion of the coding sequence, insertion of an intron and moderate sequence similarity between the nuclear rps12 second exon and mitochondrial rps12 sequences (approximately 75% for all Alismatales), suggest that the transfer is not recent. Weak phylogenetic evidence suggests that transfer took place ancestrally in this clade including Maundiaceae, Juncaginaceae, Cymodoceaceae, Potamogetonaceae and Zosteraceae (see Fig 3 for the Alismatales phylogeny).
Although rps7 is the most frequently lost ribosomal protein gene, according to Adams et al. , it appears only rarely to be functionally transferred to the nucleus [58, 59]. The complete gene has been transferred to the nuclear genome of Zostera (scaffold 54, LFYR01001529), but it is not likely to be functional. The nuclear rps7 gene sequence is identical to the gene sequence remaining in the mitogenome and it is included in a >38 kb NUMT that is almost identical to the mitogenome sequence, which strongly suggests a very recent transfer of DNA from the mitogenome to the nucleus.
In addition to the genes annotated as mitochondrial in the Zostera nuclear genome, BLASTN searches for the remaining ribosomal protein genes missing from the Zostera mitogenome further suggest that rps1 may also have been transferred. Absence of a mitochondrial rps1 gene in Maundia and a presence of a mitochondrial pseudogene in Zannichellia coupled with phylogenetic evidence suggest that a transfer occurred in the common ancestor of the group including Maundiaceae, Cymodoceaceae, Potamogetonaceae and Zosteraceae. Functional transfers of rps1 are known from several eudicots (e.g., [42, 59]).
Although a BLASTN search for rpl10 reveals a short sequence to be present in nuclear genome of Zostera we do not consider it a likely replacement of the mitochondrial rpl10 gene. It is possible that the mitochondrial rpl10 gene has been functionally replaced by a duplicated nuclear encoded chloroplast counterpart as described in other monocots and Brassicaceae . The nuclear genome of Zostera does include an rpl10 gene annotated as encoding a “putative 50S ribosomal protein L10” (LFYR01000749), which appear to have a plastid origin. Other putative ribosomal genes in the Zostera nuclear genome appear to have only remote similarity to seed plant mitochondrial genes, thus functional replacement may possibly account for the remaining ribosomal protein genes apparently missing in both the nuclear and mitochondrial genome of Zostera.
S1 Table. Reference mitogenome and plastome sequences from GenBank.
= indicates that the same species was used.
S2 Table. Sequence reads and coverage.
Total number of reads from 454 and Illumina sequencing, number of reads mapped to the mitogenomes and mean coverage of reads.
S3 Table. Gene content in mitogenomes of five representatives of Alismatales.
The length of coding sequences in kb. Z = Zostera. Ψ = pseudogene or fragment. CP = tRNA gene located in region of plastid origin.
S4 Table. Repeated sequence in Stratiotes aloides and Zostera marina.
S5 Table. Inserts of plastid origin in the mitogenomes of Stratiotes aloides, Butomus umbellatus and Zostera marina.
S6 Table. Similarity and phylogenetic relationship of plastid gene fragments located in mitochondrial genomes of Butomus umbellatus, Zostera marina and Stratiotes aloides.
BLASTN and phylogenetic analyses are done using matrices from Ross et al.  including ca. 150 taxa mostly from the Alismatales.
We thank Jeanine Olsen (University of Groningen) for preliminary access to the Zostera genome data and Charlotte Hansen for skillful assistance with lab work.
- Conceptualization: GP OS AC.
- Data curation: AC GP GTR CFB.
- Formal analysis: AC GP.
- Funding acquisition: OS GP JID SWG.
- Investigation: AC AZ GTR CFB GP.
- Project administration: OS GP.
- Resources: JID SWG OS GP.
- Writing – original draft: GP.
- Writing – review & editing: GP OS JID SWG CFB GTR AC AZ.
- 1. Knoop V: Seed plant mitochondrial genomes: complexity evolving. In: Genomics of chlooroplasts annd mitochondria. Dordrecht: Springer; 2012: 175–200.
- 2. Mower JP, Sloan DB, Alverson AJ: Plant mitochondrial genome diversity: the genomics revolution. In: Plant Genome Diversity: Plant genomes, their residents, and their evolutionary dynamics. Edited by Wendel JF, Greilhuber J, Dolezel J, Leitch AR, vol. 1. Vienna: Springer; 2012: 123–144.
- 3. Sloan DB: One ring to rule them all? Genome sequencing provides new insights into the 'master circle' model of plant mitochondrial DNA structure. New Phytol 2013, 200:978–985. pmid:24712049
- 4. Sloan DB, Alverson AJ, Chuckalovcak JP, Wu M, McCauley DE, Palmer JD, et al: Rapid Evolution of Enormous, Multichromosomal Genomes in Flowering Plant Mitochondria with Exceptionally High Mutation Rates. PLoS Biology 2012, 10:e1001241. pmid:22272183
- 5. Skippington E, Barkman TJ, Rice DW, Palmer JD: Miniaturized mitogenome of the parasitic plant Viscum scurruloideum is extremely divergent and dynamic and has lost all nad genes. Proc Nat Acad Sci USA 2015, 112:E3515–E3524. pmid:26100885
- 6. Adams KL, Rosenblueth M, Qiu YL, Palmer JD: Multiple losses and transfers to the nucleus of two mitochondrial succinate dehydrogenase genes during angiosperm evolution. Genetics 2001, 158:1289–1300. pmid:11454775
- 7. Adams KL, Qiu YL, Stoutemyer M, Palmer JD: Punctuated evolution of mitochondrial gene content: High and variable rates of mitochondrial gene loss and transfer to the nucleus during angiosperm evolution. Proc Nat Acad Sci USA 2002, 99:9905–9912. pmid:12119382
- 8. Adams KL, Song KM, Roessler PG, Nugent JM, Doyle JL, Doyle JJ, et al: Intracellular gene transfer in action: Dual transcription and multiple silencings of nuclear and mitochondrial cox2 genes in legumes. Proc Nat Acad Sci USA 1999, 96:13863–13868. pmid:10570164
- 9. Petersen G, Cuenca A, Moller IM, Seberg O: Massive gene loss in mistletoe (Viscum, Viscaceae) mitochondria. Sci Rep 2015, 5: 17588. pmid:26625950
- 10. Alverson AJ, Wei XX, Rice DW, Stern DB, Barry K, Palmer JD: Insights into the Evolution of Mitochondrial Genome Size from Complete Sequences of Citrullus lanatus and Cucurbita pepo (Cucurbitaceae). Mol Biol Evol 2010, 27:1436–1448. pmid:20118192
- 11. Rice DW, Alverson AJ, Richardson AO, Young GJ, Sanchez-Puerta MV, Munzinger J et al.: Horizontal Transfer of Entire Genomes via Mitochondrial Fusion in the Angiosperm Amborella. Science 2013, 342:1468–1473. pmid:24357311
- 12. Wang WQ, Wu YR, Messing J: The Mitochondrial Genome of an Aquatic Plant, Spirodela polyrhiza. PLoS One 2012, 7: e46747. pmid:23056432
- 13. Cuenca A, Petersen G, Seberg O: The Complete Sequence of the Mitochondrial Genome of Butomus umbellatus—A Member of an Early Branching Lineage of Monocotyledons. PLoS One 2013, 8: e61552. pmid:23637852
- 14. Barkman TJ, McNeal JR, Lim SH, Coat G, Croom HB, Young ND, et al: Mitochondrial DNA suggests at least 11 origins of parasitism in angiosperms and reveals genomic chimerism in parasitic plants. BMC Evol Biol 2007, 7:248. pmid:18154671
- 15. Richardson AO, Rice DW, Young GJ, Alverson AJ, Palmer JD: The "fossilized" mitochondrial genome of Liriodendron tulipifera: ancestral gene content and order, ancestral editing sites, and extraordinarily low mutation rate. BMC Biol 2013, 11: 29. pmid:23587068
- 16. Grewe F, Edger PP, Keren I, Sultan L, Pires JC, Ostersetzer-Biran O, et al: Comparative analysis of 11 Brassicales mitochondrial genomes and the mitochondrial transcriptome of Brassica oleracea. Mitochondrion 2014, 19:135–143. pmid:24907441
- 17. Tang MY, Chen ZW, Grover CE, Wang YM, Li SS, Liu GZ et al.: Rapid evolutionary divergence of Gossypium barbadense and G. hirsutum mitochondrial genomes. BMC Genomics 2015, 16: 770. pmid:26459858
- 18. Petersen G, Seberg O, Davis JI, Stevenson DW: RNA editing and phylogenetic reconstruction in two monocot mitochondrial genes. Taxon 2006, 55:871–886.
- 19. Cuenca A, Petersen G, Seberg O, Davis JI, Stevenson DW: Are substitution rates and RNA editing correlated? BMC Evol Biol 2010, 10: 349. pmid:21070620
- 20. Cuenca A, Petersen G, Seberg O, Jahren AH: Genes and Processed Paralogs Co-exist in Plant Mitochondria. J Mol Evol 2012, 74:158–169. pmid:22484699
- 21. Cuenca A, Ross TG, Graham SW, Barrett CF, Davis JI, Seberg O, et al: Localized Retroprocessing as a Model of Intron Loss in the Plant Mitochondrial Genome. Genome Biol Evol 2016, 8:2176–2189. pmid:27435795
- 22. Petersen G, Seberg O, Cuenca A, Stevenson DW, Thadeo M, Davis JI et al.: Phylogeny of the Alismatales (Monocotyledons) and the relationship of Acorus (Acorales?). Cladistics 2016, 32:141–159.
- 23. Ross TG, Barrett CF, Soto Gomez M, Lam VKY, Henriquez CL, Les DH et al.: Plastid phylogenomics and molecular evolution of Alismatales. Cladistics 2016, 32:160–178.
- 24. Olsen JL, Rouze P, Verhelst B, Lin YC, Bayer T, Collen J et al.: The genome of the seagrass Zostera marina reveals angiosperm adaptation to the sea. Nature 2016, 530:331–335. pmid:26814964
- 25. Doyle JJ, Doyle JF: A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochem Bull 1987, 19:11–15.
- 26. Lindgreen S: AdapterRemoval: easy cleaning of next-generation sequencing reads. BMC Res Notes 2012, 5:337. pmid:22748135
- 27. Iorizzo M, Senalik DA, Szklarczyk M, Grzebelus D, Spooner D, Simon P: De novo assembly of the carrot mitochondrial genome using next generation sequencing of whole genomic DNA provides first evidence of DNA transfer into an angiosperm plastid genome. BMC Plant Biol 2012, 12:61. pmid:22548759
- 28. Altschul SF, Gish W, Myers EW, Lipman DJ: Basic local aligment search tool. J Mol Biol 1990, 215:403–410. pmid:2231712
- 29. Warren RL, Sutton GG, Jones SJM, Holt RA: Assembling millions of short DNA sequences using SSAKE. Bioinformatics 2007, 23:500–501. pmid:17158514
- 30. Schattner P, Brooks AN, Lowe TM: The tRNAscan-SE, snoscan and snoGPS web servers for the detection of tRNAs and snoRNAs. Nucleic Acids Res 2005, 33:W686–W689. pmid:15980563
- 31. Edgar RC: MUSCLE: a multiple sequence alignment method with reduced time and space complexity. BMC Bioinformatics 2004, 5:1–19.
- 32. Stamatakis A: RAxML-VI-HPC: Maximum likelihood-based phylogenetic analyses with thousands of taxa and mixed models. Bioinformatics 2006, 22:2688–2690. pmid:16928733
- 33. Jurka J, Kapitonov VV, Pavlicek A, Klonowski P, Kohany O, Walichiewicz J: Repbase update, a database of eukaryotic repetitive elements. Cytogenet Genome Res 2005, 110:462–467. pmid:16093699
- 34. Guo WH, Grewe F, Fan WS, Young GJ, Knoop V, Palmer JD, et al: Ginkgo and Welwitschia Mitogenomes Reveal Extreme Contrasts in Gymnosperm Mitochondrial Evolution. Mol Biol Evol 2016, 33:1448–1460. pmid:26831941
- 35. Edgar RC: MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Res 2004, 32:1792–1797. pmid:15034147
- 36. Stamatakis A, Hoover P, Rougemont J: A Rapid Bootstrap Algorithm for the RAxML Web Servers. Syst Biol 2008, 57:758–771. pmid:18853362
- 37. Satoh M, Kubo T, Nishizawa S, Estiati A, Itchoda N, Mikami T: The cytoplasmic male-sterile type and normal type mitochondrial genomes of sugar beet share the same complement of genes of known function but differ in the content of expressed ORFs. Mol Genet Genomics 2004, 272:247–256. pmid:15340837
- 38. Lohse M, Drechsel O, Bock R: OrganellarGenomeDRAW (OGDRAW): a tool for the easy generation of high-quality custom graphical maps of plastid and mitochondrial genomes. Curr Genet 2007, 52:267–274. pmid:17957369
- 39. Alverson AJ, Rice DW, Dickinson S, Barry K, Palmer JD: Origins and Recombination of the Bacterial-Sized Multichromosomal Mitochondrial Genome of Cucumber. Plant Cell 2011, 23:2499–2513. pmid:21742987
- 40. Christensen AC: Plant Mitochondrial Genome Evolution can be Explained by DNA Repair Mechanisms. Genome Biol Evol 2013, 5:1079–1086. pmid:23645599
- 41. Christensen AC: Genes and Junk in Plant Mitochondria-Repair Mechanisms and Selection. Genome Biol Evol 2014, 6:1448–1453. pmid:24904012
- 42. Park S, Ruhlman TA, Sabir JSM, Mutwakil MHZ, Baeshen MN, Sabir MJ et al.: Complete sequences of organelle genomes from the medicinal plant Rhazya stricta (Apocynaceae) and contrasting patterns of mitochondrial genome evolution across asterids. BMC Genomics 2014, 15:405. pmid:24884625
- 43. Zhang TW, Fang YJ, Wang XM, Deng X, Zhang XW, Hu SN, et al: The Complete Chloroplast and Mitochondrial Genome Sequences of Boea hygrometrica: Insights into the Evolution of Plant Organellar Genomes. PLoS One 2012, 7: e30531. pmid:22291979
- 44. Clifton SW, Minx P, Fauron CMR, Gibson M, Allen JO, Sun H et al.: Sequence and comparative analysis of the maize NB mitochondrial genome. Plant Physiol 2004, 136:3486–3503. pmid:15542500
- 45. Allen JO, Fauron CM, Minx P, Roark L, Oddiraju S, Lin GN et al.: Comparisons among two fertile and three male-sterile mitochondrial genomes of maize. Genetics 2007, 177:1173–1192. pmid:17660568
- 46. Sloan DB, Wu ZQ: History of Plastid DNA Insertions Reveals Weak Deletion and AT Mutation Biases in Angiosperm Mitochondrial Genomes. Genome Biol Evol 2014, 6:3210–3221. pmid:25416619
- 47. Lawrence JG, Ochman H: Amelioration of bacterial genomes: Rates of change and exchange. J Mol Evol 1997, 44:383–397. pmid:9089078
- 48. Adams KL, Ong HC, Palmer JD: Mitochondrial gene transfer in pieces: Fission of the ribosomal protein gene rpl2 and partial or complete gene transfer to the nucleus. Mol Biol Evol 2001, 18:2289–2297. pmid:11719578
- 49. Hecht J, Grewe F, Knoop V: Extreme RNA Editing in Coding Islands and Abundant Microsatellites in Repeat Sequences of Selaginella moellendorffii Mitochondria: The Root of Frequent Plant mtDNA Recombination in Early Tracheophytes. Genome Biol Evol 2011, 3:344–358. pmid:21436122
- 50. Maréchal-Drouard L, Weil JH, Guillemaut P: Import of several transfer-RNAs from cytoplasm into the mitochondia in bean, Phaseolus vulgaris. Nucleic Acids Res 1988, 16:4777–4788. pmid:3387208
- 51. Mower JP, Case AL, Floro ER, Willis JH: Evidence against Equimolarity of Large Repeat Arrangements and a Predominant Master Circle Structure of the Mitochondrial Genome from a Monkeyflower (Mimulus guttatus) Lineage with Cryptic CMS. Genome Biol Evol 2012, 4:670–686. pmid:22534162
- 52. Bonen L: Cis- and trans-splicing of group II introns in plant mitochondria. Mitochondrion 2008, 8:26–34. pmid:18006386
- 53. Qiu YL, Palmer JD: Many independent origins of trans splicing of a plant mitochondrial group II intron. J Mol Evol 2004, 59:722–724.
- 54. Guo WH, Mower JP: Evolution of Plant Mitochondrial Intron-Encoded Maturases: Frequent Lineage-Specific Loss and Recurrent Intracellular Transfer to the Nucleus. J Mol Evol 2013, 77:43–54. pmid:23979261
- 55. Sloan DB, Alverson AJ, Storchova H, Palmer JD, Taylor DR: Extensive loss of translational genes in the structurally dynamic mitochondrial genome of the angiosperm Silene latifolia. BMC Evol Biol 2010, 10: 274. pmid:20831793
- 56. Zhu AD, Guo WH, Jain K, Mower JP: Unprecedented Heterogeneity in the Synonymous Substitution Rate within a Plant Genome. Mol Biol Evol 2014, 31:1228–1236. pmid:24557444
- 57. Ong HC, Palmer JD: Pervasive survival of expressed mitochondrial rps14 pseudogenes in grasses and their relatives for 80 million years following three functional transfers to the nucleus. BMC Evol Biol 2006, 6: 55. pmid:16842621
- 58. Park S, Grewe F, Zhu AD, Ruhlman TA, Sabir J, Mower JP, et al: Dynamic evolution of Geranium mitochondrial genomes through multiple horizontal and intracellular gene transfers. New Phytol 2015, 208:570–583. pmid:25989702
- 59. Liu SL, Zhuang Y, Zhang P, Adams KL: Comparative Analysis of Structural Diversity and Sequence Evolution in Plant Mitochondrial Genes Transferred to the Nucleus. Mol Biol Evol 2009, 26:875–891. pmid:19168566
- 60. Kubo N, Arimura S: Discovery of the rpl10 Gene in Diverse Plant Mitochondrial Genomes and Its Probable Replacement by the Nuclear Gene for Chloroplast RPL10 in Two Lineages of Angiosperms. DNA Res 2010, 17:1–9. pmid:19934175