lable at ScienceDirect

Biochimie 100 (2014) 107e120

Contents lists avai

Biochimie

journal homepage: www.elsevier .com/locate/biochi

Review

The plant mitochondrial genome: Dynamics and maintenance

José M. Gualberto, Daria Mileshina, Clémentine Wallet, Adnan Khan Niazi,Frédérique Weber-Lotfi, André Dietrich*

Institut de Biologie Moléculaire des Plantes, CNRS and Université de Strasbourg, 12 rue du Général Zimmer, 67084 Strasbourg, France

a r t i c l e i n f o

Article history:Received 22 July 2013Accepted 17 September 2013Available online 26 September 2013

Keywords:MitochondriamtDNAReplicationRecombinationRepair

* Corresponding author. Tel.: þ33 3 67 15 53 73; faE-mail addresses: jose.gualberto@ibmp-cnrs.u

Daria.Mileshina@ibmp-cnrs.unistra.fr (D. Mileshinacnrs.unistra.fr (C. Wallet), Adnan-Khan.Niazi@ibmplotfif@unistra.fr (F. Weber-Lotfi), andre.dietrich@ibmp

0300-9084/$ e see front matter � 2013 Elsevier Mashttp://dx.doi.org/10.1016/j.biochi.2013.09.016

a b s t r a c t

Plant mitochondria have a complex and peculiar genetic system. They have the largest genomes, ascompared to organelles from other eukaryotic organisms. These can expand tremendously in somespecies, reaching the megabase range. Nevertheless, whichever the size, the gene content remainsmodest and restricted to a few polypeptides required for the biogenesis of the oxidative phosphorylationchain complexes, ribosomal proteins, transfer RNAs and ribosomal RNAs. The presence of autonomousplasmids of essentially unknown function further enhances the level of complexity. The physical orga-nization of the plant mitochondrial DNA includes a set of sub-genomic forms resulting from homologousrecombination between repeats, with a mixture of linear, circular and branched structures. This materialis compacted into membrane-bound nucleoids, which are the inheritance units but also the centers ofgenome maintenance and expression. Recombination appears to be an essential characteristic of plantmitochondrial genetic processes, both in shaping and maintaining the genome. Under nuclear surveil-lance, recombination is also the basis for the generation of new mitotypes and is involved in the evo-lution of the mitochondrial DNA. In line with, or as a consequence of its complex physical organization,replication of the plant mitochondrial DNA is likely to occur through multiple mechanisms, potentiallyinvolving recombination processes. We give here a synthetic view of these aspects.

� 2013 Elsevier Masson SAS. All rights reserved.

1. Introduction

Mitochondria are key players in plant development, fitness andreproduction. Their contribution to energy production, metabolismand cell homeostasis relies on the performance of their own geneticsystem that makes them semi-autonomous. As in other organisms,the mitochondrial genome in plants encodes a series of essentialpolypeptides that build up the complexes of the oxidative phos-phorylation chain, together with nuclear-encoded subunits. Butplant mitochondrial DNAs (mtDNAs) have remarkable features thatdistinguish them from their animal and fungal counterparts. Inparticular, higher plants harbor large mtDNAs that are highly var-iable in size and structural organization. On top of that, in manyplant species, mitochondria contain various forms of plasmids thatreplicate independently from the main chromosome. In most plantspecies, the mtDNA gene sequences evolve very slowly, ascompared to animal mtDNA sequences, and point mutations arerare. It is believed that this is because plant mitochondria contain

x: þ33 3 88 61 44 42.nistra.fr (J.M. Gualberto),), clementine.wallet@ibmp--cnrs.unistra.fr (A.K. Niazi),-cnrs.unistra.fr (A. Dietrich).

son SAS. All rights reserved.

an active DNA recombination system that allows copy correction ofmutations. Indeed, numerous studies have shown that plantmitochondrial genomes undergo extensive and high frequencyhomologous recombination (HR). Such processes make plantmtDNA prone to rearrangements. When not lethal, mtDNA muta-tions/rearrangements can generate severe phenotypes or causecytoplasmic male sterility (CMS). Hence the requirement for effi-cient DNA repair and maintenance pathways in a context whereoxidative pressure, replication defaults or environmental hazardgenerate base modifications and strand breaks. Finally, mitochon-drial genome dynamics generates heteroplasmy, with alternativemtDNA configurations coexisting with the main mtDNA. Segrega-tion of alternative mitotypes significantly contributes to the rapidevolution of the plant mtDNA structure. The present review ad-dresses these issues, with special emphasis on the specificities ofthe plant mitochondrial genetic system.

2. Plant mitochondrial DNA structure and organization

2.1. Genome size and content

The structure of higher plant mitochondrial genomes has anumber of unique features. Whereas most animals possess circularmtDNAs of 15e17 kb in size, the mitochondrial genomes of plants

Fig. 1. Circular and linear models of the mtDNA structure in plants. The dynamics of the mtDNA of plants allow several models. a) A circular genome model representing thedifferent organisations of the mtDNA of A. thaliana ecotype C24. The two pairs of large, frequently recombining repeats (A and B) and their orientations are represented by blue andred arrows. The different parts of the genome are colored in a range of gray scale. Intramolecular recombination events are represented with dashed lines, and in orange,recombination leading to subgenomic molecules. Dotted line shows possible intermolecular recombination. b) The mtDNA has been observed as predominantly constituted by acomplex array of linear molecules. Branched forms of the mtDNA could be intermediates of recombination-dependent processes: 30 single-stranded sequences resulting from therecession of double-strand breaks can invade homologous double-stranded DNA, forming a D-loop and leading to the establishment of replication forks.

J.M. Gualberto et al. / Biochimie 100 (2014) 107e120108

are much larger and differ greatly in size, even between very closespecies or within species [1,2]. They commonly range between 200and 750 kb in angiosperms [1], but with a tremendous furtherextension in some lineages. As an example, proliferation ofdispersed repeats, expansion of existing introns and acquisition ofsequences (including sequences of nuclear, plastid, viral and bac-terial origin) account for the expansion of the cucumber (Cucumissativus) mitochondrial genome into a tri-chromosome structurewith components of 1556, 84, and 45 kb in size [3]. Sequencing ofthe mtDNA from two Silene species with exceptionally high mu-tation rate revealed enormousmitochondrial genomes of 6.7 (Silenenoctiflora) and 11.3 (Silene conica) megabase-pairs resulting frommassive proliferation of non-coding content [4]. Moreover, thesegenomes are distributed into a large number of circular chromo-somes: 59 for S. noctiflora and over 128 for S. conica. These chro-mosomes range themselves in size from 44 to 192 kb. As acomparison, the more typical, slowly evolving mitochondrialgenome of Silene latifolia accounts for a single chromosome of253 kb [4]. Remarkably, plant mitochondrial genomes are large buttheir ploidy appears to be low. Whereas thousands of mtDNAcopies can be present in a mammalian cell, much lower mtDNAlevels were detected in plants. Preuten et al. [5] analyzed the ploidyof several individual mitochondrial genes in various samples ofArabidopsis thaliana, Nicotiana tabacum and Hordeum vulgare, usingreal time quantitative PCR. The copy numbers of the investigatedgenes both differed from each other and varied greatly betweenorgans or during development, with the highest estimates around280 copies per cell for the atp1 gene in mature leaves. This wasactually less than the mean number of mitochondria per cell, whichwas around 450 in protoplasts derived from mature leaves. Thus,individual mitochondria in plants may contain only part of thegenome or potentially no DNA [5].

Despite their large sizes, land plant mtDNAs do not contain asignificantly greater number of genes than mitochondrial genomes

of other lineages. The known genes usually range between 50 and60, while multiple cis- or trans-spliced introns and large intergenicregions complete the genomes. Protein genes in plant mtDNAsencode subunits of the oxidative phosphorylation chain complexesbut also proteins involved in the biogenesis of these complexes, aswell as several ribosomal proteins. First to be entirely sequenced,the 367 kb A. thaliana mtDNA (ecotype C24) codes for 32 proteins,3 ribosomal RNAs (5S, 18S and 26S rRNAs) and 22 tRNAs [6], whilethe 16.5 kb human mtDNA codes for 13 proteins, 2 rRNAs (12S and16S) and 22 tRNAs [7]. With a total of 57 genes versus 37, thismakes a 1.5 fold difference in gene content for a 22 fold differencein size. Identified genes represent a minor part of the plant mtDNAcontent (10% in A. thaliana), whereas the major part (60% inA. thaliana) has no recognizable origin and function [6]. Intronsand duplications expand (8% and 7% in A. thaliana, respectively), aswell as integrated nuclear and plastid sequences (5% as a whole inA. thaliana). Further putative open reading frames of significantlength can be detected, which also add up to 10% of the genome inA. thaliana [6].

Differences in gene content among plant species are also notlinked to genome size. The large cucumber mtDNA carries just 4more protein genes than the A. thaliana genome within 1.3 mega-base of extra sequence [3]. On the other hand, it can be noted thatthe cucumber mitochondrial genome has integrated 7 more tRNAgenes of plastid origin, whereas 2 tRNA genes of authentic mito-chondrial origin present in other species are missing. Similarly, the253 kb mtDNA of S. latifolia and the 11.3 megabase mtDNA ofS. conica carry the same reduced set of 25 protein genes and, whilethe S. latifolia mtDNA contains 9 tRNA genes, that of S. conica hasonly 2 [4]. The intergenic sequences with no recognizable origin orfunction thus represent 91.5% (10.3 megabase) of the S. conicamitochondrial genome. With the emergence of second-generationsequencing technologies, the number of completed plant mito-chondrial genome sequences rapidly increased in the last few years,

J.M. Gualberto et al. / Biochimie 100 (2014) 107e120 109

with 77 deposited in the GenBank database (http://www.ncbi.nlm.nih.gov/genomes/) by June 2013.

2.2. Physical organization

The physical structure of plant mitochondrial genomes is usu-ally represented by a circle of double-stranded DNA, called “masterchromosome”, housing the complete set of mitochondrial genes.However, such a structure is derived from mapping and sequenceassembly and has not been directly observed. Thus, although itmight exist, the master circle must be rare. On the other hand, the620 kb mitochondrial insertion characterized in A. thaliana nuclearchromosome 2 [8] seems to derive from a concatemer of mito-chondrial genomes, suggesting that such concatemers may existin vivo. PlantmtDNAs usually contain repeated sequences dispersedthroughout the genome, which can be in direct or inverted orien-tation. They can pair up and recombine, redistributing sequencesand generating sub-genomic circular DNA molecules (Fig. 1a).Indeed, numerous studies have shown that plant mitochondrialgenomes undergo extensive and high frequency homologousrecombination. Especially, recombination between a few large re-peats (several kb in size) is frequent and reciprocal, leading tomultiple, interconverting configurations of the genome. Theserecombinationally active long repeats are not homologous betweenspecies, but the process is conserved. In A. thaliana, two large re-peats of 6.5 and 4.2 kbwere shown to be involved in recombination[9]. In theN. tabacummtDNA, there are three pairs of large repeatedsequences, with lengths of 18 kb, 6.9 kb and 4.7 kb [10]. But thenumber of recombinogenic repeats can be much larger, as forexample in wheat where there are 10 pairs of large repeats [11]. Asa consequence, the mtDNA of plants cannot be physically repre-sented as a unique, organized structure that could be used todetermine phylogenetic relationships between species. Further-more, besides sub-genomic circles, early observations highlightedlinear forms [12]. In the case of the maize (Zea mays) CMS-S cyto-plasm, it was proposed that the mitochondrial genome would existmainly as multiple linear molecules. These would result fromrecombination of the mitochondrial chromosome with a linearplasmid (see Section 4 below) specific to CMS-S. The sequencingdata seemed to be consistent with this premise [2]. Also in earlystudies, themorphology of carefully isolatedmtDNAwas consistentwith a predominance of linear molecules [13]. Altogether, it ap-pears that the entity of a plant mitochondrial genome is an array ofinter-convertible linear and circular DNAmolecules that most likelyreflect recombination intermediates between repeated sequences,yielding a mixture of various subgenomic DNA molecules withdifferent structures and contents (Fig. 1). The concept of a circularmaster chromosome remains because the question of how multi-partite and branched DNA molecules are properly transmitted tothe next generation is unsolved. Finally, a significant portion of themtDNA (2e8%) was reported to be in a single-stranded form inChenopodium album suspension culture or whole plants [14].

Homologous recombination between the numerous interme-diate size repeats (ISRs, usually ranging from 50 to 600 bp in size)also present in the genome is generally infrequent but, whenoccurring, leads to further complex rearrangements. It has beenproposed that most of the recombination processes involving ISRsresult from Break-Induced Replication (BIR) pathways [15]. Butother pathways, such as Single-Strand Annealing (SSA), might alsobe involved [16]. These low frequency recombination activities arenon-allelic, and the process is asymmetrical, resulting in accumu-lation of only one of the expected reciprocal recombination prod-ucts and leading to duplication or deletion of genomic sequences. Inaddition, illegitimate recombination processes involving very shortsequence homologies of a few nucleotides (microhomologies) have

also been described [17], which can lead to gene chimeras. Both lowfrequency recombination between ISRs and error-prone repairprocesses mediated by microhomologies (see below) yield low-copy number alternative configurations (or “mitotypes”) of themitochondrial genome that contribute to the natural heteroplasmyof the plant mtDNA. Finally, additional processes might be at work,depending on the species. In particular, the large size of the mito-chondrial genomes in the genera Cucumis and Silene have beenlinked with the accumulation of short (30e53 bp) repeated se-quences [3,4,18].

Contrary to a prevailing idea, the mtDNA is not naked but ispacked into nucleoprotein particles called nucleoids [19]. In thesemembrane-anchored particles, the DNA is associated with a num-ber of factors involved in its compaction andmetabolism. Nucleoidsare considered as the heritable units of mtDNA and their structureinfluences mtDNA segregation and transcriptional regulation [20].The actual set of nucleoid-associated proteins is still discussed andseems to vary between the different organisms, suggesting that thecomposition of mitochondrial nucleoproteins has been reinventedseveral times along with mtDNA evolution [19]. Mitochondrialnucleoids contain a set of core proteins involved in DNA mainte-nance and expression, as well as peripheral factors that are com-ponents of signaling pathways [21]. Plant mitochondrial nucleoidswere characterized as dynamic entities, with chromatin-like orfibril-like structures, depending on the developmental stage ortissue [22,23]. Chromatin-like structures were associated with avesicle component.

Whereas a nucleoid-enriched proteome has been establishedfor chloroplasts [24], information on the composition of theirmitochondrial counterparts remains scarce. Plant nucleoidsappeared to share a qualitatively similar phospholipid compositionwith the whole organelle and to contain polypeptides of both innerand outer membrane origin [22]. The association of the identifiedproteins with isolated nucleoids was a function of the detergentconcentration in the isolation procedure, so that the structural orfunctional significance has to be further evaluated. Little is knownabout the proteins responsible for nucleoid organization in plants[25]. The presence of nucleosome-like structures was suggested,depending on the developmental stage [22,26]. It was reported thathistone H3 and histone deacetylase might be targeted to mito-chondria in cauliflower [27,28]. However, such a view was contra-dicted by studies in tobacco cells [29]. Actin was persistentlyrecovered in plant nucleoids, where it seemed to associate with themtDNA [22]. Import of actin to the inside of mitochondria wasindeed reported and its potential function in the organelles wasdiscussed [30]. It would bind to the mtDNA in complexes alsocontaining porin and the ADP/ATP carrier. Plant mitochondrialnucleoids likely have a heterogeneous genomic organization. Opencircles, supercoils, complex forms, and linear molecules withinterspersed sigma-shaped structures and/or loops representedabout 70% of themtDNAmolecules present inmung bean nucleoids[23]. As in fungi or mammals, mitochondrial nucleoids in plants arenot only compaction and segregation units but they support theentire DNA metabolism. Isolated mung bean nucleoids were re-ported to run DNA synthesis and were capable of directing regu-lated RNA synthesis [22]. Membrane-association of the plantmitochondrial base excision repair (BER) components (see Section6.2) also supports the assumption that mtDNA maintenance andrepair pathways take place in the membrane-bound nucleoids[31,32].

3. Mitochondrial DNA replication

Replication of plant mitochondrial genomes is far from beingunderstood. Several factors involved in organellar genome

J.M. Gualberto et al. / Biochimie 100 (2014) 107e120110

replication have been identified by sequence homology and fromproteomic data. Among those, plant organellar DNA polymeraseswere identified by their similarities to the known mitochondrialDNA polymerase of animals and yeast. In A. thaliana, there are twoorganellar DNA polymerases, Pol1A and Pol1B, which are dual-targeted to both mitochondria and plastids [33e35] and whichare apparently redundant in their functions [36,37]. Individually,each of them is dispensable, because the mutants show no visiblephenotypes, apart from a small reduction in mtDNA and chloro-plast DNA (cpDNA) copy numbers. Conversely, the doublemutant isnot viable, showing that the two polymerases are redundant fororganellar genome replication [37]. However, the situation mightnot be the same in other species. While single mutants of each ofthe A. thaliana organellar DNA polymerases have a very mild effecton mtDNA and cpDNA copy numbers, in Z. mays, deficiency in aplastid-localized DNA polymerase results in albino plants becauseof a drastic reduction in cpDNA copy number [38]. This suggeststhat in maize the organellar DNA polymerases are not redundantand might be specific for each type of organelle.

Additional mtDNA replication factors have been identified,which are, in most cases, also dual-targeted to chloroplasts andmitochondria. A helicase similar to the Twinkle helicase of animalmitochondria is present in all plant genomes sequenced, and istargeted to both types of organelles in A. thaliana. Like its T7 phagehomolog, it combines helicase and primase activities in a singleprotein [35,39]. Genes coding for organellar Type I and Type IItopoisomerases have also been identified [24,35,40] and there arehomologs of bacterial single-stranded DNA-binding (SSB) proteinsin plant mitochondria [41].

Regarding the mechanisms of replication, they are still open todebate, mainly because the large size and multipartite complexityof the plant mtDNA render this process difficult to analyze. In asearch for replication origins as they had been described in meta-zoan mitochondrial genomes, two putative fragments (oriA andoriB) were identified from the genome of Petunia hybrida. Thesefragments were able to initiate DNA synthesis in an in vitro DNAsynthesizing system derived from purified P. hybrida mitochondria[42]. DNA synthesis performed with both the A and B origins in thesame plasmid, in complementary strands, initiated first in the A-origin and proceeded in the direction of the B-origin, after whichreplication was also initiated in the B-origin. Based on these ob-servations, the authors proposed a model of mtDNA replication inplants structurally comparable to the one of mammalian mtDNAheavy and light strands [43]. Similarly, a search for sequences ofPapaver somniferum (Opium poppy) mtDNA able to promoteautonomous plasmid replication in yeast revealed a 917 bp frag-ment with high A/T content [44]. However, these experimentsstrongly relied on the premises that most of the plant mtDNA existsas genome-size circles replicating according to a theta model ofcircular DNA replication. But, as commented on in Section 2.2,while data arising from restriction mapping can imply circularmtDNA genomes, direct observation of plant mtDNA failed tosupport that assumption. Early experiments aiming to characterizethe topology of the mtDNA of plants mainly revealed linear andsmall circular molecules, and the inability to purify full-size circulargenomes was attributed to artifactual shearing of large molecules(reviewed in Ref. [12]). Structural analysis of plant mtDNA usingpulsed-field gel electrophoresis and moving pictures of DNAmigration in agarose revealed that most of the genome is containedin large linear molecules and complex branched molecular struc-tures larger in size than the predicted genome size [13,45]. Analysisof the mtDNA of the liverwort Marchantia polymorpha by digestionwith restriction enzymes with one recognition site per genome andpulsed-field gel electrophoresis showed that the genome isconstituted of linear and head-to-tail concatemers [46]. These

concatemeric molecules could be generated both by rolling circlereplication and by recombination-mediated replication, as in phageT4. The latter model could also account for the complex branchedmtDNA molecules (Fig. 1b).

Rolling-circle and recombination-dependent replication havealso been inferred from electron microscopy observation of theC. album mitochondrial genome and plasmids (see Section 4below). In these studies, the mtDNA preparations containedmostly linear molecules of variable size, but also subgenomic cir-cular DNA molecules, circular molecules with tails (sigma-likestructures), as well as more complex rosette-like and catenate-likemolecules [47]. The distribution of single-stranded mtDNA in thesigma structures and the detection of entirely single-strandedmolecules indicated a rolling-circle type of replication of sub-genomic circles [48], while analysis of replicating mtDNA in cellcultures also revealed characteristic recombination intermediatescompatible with a phage T4-like mechanism of recombination-mediated replication [49]. Dramatic changes in the absolute andrelative amounts of the different replicative structures occurredduring cell growth, unveiling the plasticity of the mtDNA structure.Taken together, the pulsed-field gel electrophoresis, movies ofmigrating DNA and electron microscope observations suggest thecoexistence of a recombination-dependent mode of replicationinitiation with a rolling-circle mode of replication. In the yeastCandida albicans, a comprehensive topological analysis of themtDNA also supported a model of recombination-driven replica-tion initiation [50], and inverted repeats in the genome seem to bepredominantly involved in replication initiation via homologousrecombination. As in C. albicans, replication of plant mtDNA mightinitiate by recombination involving the large repeats that areusually present in plant genomes.

However, the present models of recombination-dependentreplication do not satisfactorily explain how the complex pop-ulations of mtDNA molecules are transmitted in a near homo-plasmic fashion from cell to cell and from one generation toanother. At present, it cannot be excluded that replication involvesdifferent processes and factors according to the plant tissue and thedevelopmental stage. These questions have to be addressed in or-der to understand how the mtDNA is faithfully and completelytransmitted, and how its copy number is regulated, probably incoordination with the replication of the plastid and nucleargenomes.

4. Mitochondrial plasmids

In addition to a large and dynamic main genome, mitochondriaof many higher plant species contain a variety of smaller DNAmolecules whose size can range from 0.7 to over 20 kb [51e54].These can be regarded as extrachromosomal replicons or plasmidsthat can be autonomously replicated in mitochondria because theyare usually present at a high stoichiometry relative to the maingenome [54]. Multimeric forms of these mitochondrial plasmidshave been observed that likely result from homologous recombi-nation processes [55]. The pattern of mitochondrial plasmids isspecies-specific and can be different among varieties of the sameplant species [56]. In Silene species, plasmids further contribute tothe complexity of mitochondrial genetics [57]. Although remnantsof putative integration into the mtDNA were sometimes reported[58], most plant mitochondrial plasmid DNAs have essentially nosequence similarity with the main chromosome and cannot beconsidered as subgenomic forms. In that respect, they resemblemitochondrial plasmids found in several fungus species, such asNeurospora crassa or Neurospora intermedia. Their presence orabsence has no correlation with species peculiarities but theiramplification changes along the development. Additional plasmid

J.M. Gualberto et al. / Biochimie 100 (2014) 107e120 111

variants can arise from recombination events [59]. The nuclearcontext participates in the control of mitochondrial plasmidappearance and copy number [59].

Mitochondrial plasmids in plants were studiedmore extensivelyin the 1980’s, when they were expected to provide vectors formitochondrial transformation. The field subsequently became lessactive. They can in turn be classified into two types, circular andlinear plasmids. Circular plasmids have been found in a diversegroup of plant species. Their size is often in the range of 1e2.5 kb,although it can extend to 9 kb or more [51e53]. They can shareshort regions of homology or contain sequences homologous to thenuclear genome (reviewed in Ref. [54]). In early studies, replicationof the mtp2 (1.7 kb) and mtp3 (1.5 kb) circular mitochondrialplasmids of Vicia fabawas reported to start at a specific origin, closeto sequences that can fold into hairpin structures, and to proceed inthe same, single direction around the molecules [60]. An asym-metric rolling circle replication mechanism with unique featureswas documented for the C. album mp1 circular DNA plasmid(1.3 kb) [61]. This plasmid would contain a putative replicationorigin sharing homology with double-stranded rolling circle origin(dso) or transfer origin (oriT) nicking motifs of bacterial plasmids[62]. Typical rolling circle intermediates and concatemers werecharacterized [63]. Sigma-like molecules with long double-stranded tails were identified and their structure suggested theoccurrence of strand switching during rolling circle replication, asfor bacterial replicons [62]. A/T-rich regions containing tandemsequences that resemble the yeast ARS (autonomously replicatingsequence)-type replication origin were highlighted in the Z. mays1.4 and 1.9 kb circular plasmids [64,65]. Within their limited size,plant circular mitochondrial plasmids mostly carry open readingframes of modest length, although some reach 1 kb. Nevertheless,these orfs are transcribed, as established in a number of species[64e69]. The function of such transcripts is not understood.Notably, differences were reported in the set of circular plasmidsand transcripts derived thereof between male fertile and CMS linesof Beta vulgaris [70].

Linear mitochondrial plasmids have been characterized in Beta,Brassica, Daucus, Sorghum and Zea species, but they seem to be lessfrequent than circular forms [54]. Linear plasmids are generallylarger and commonly range from 2 to 12 kb [54]. They usuallycontain terminal inverted repeats (TIRs) and their 50-termini werereported to carry specific covalently-bound proteins (reviewed inRef. [54]). Such “invertron” features are shared with differentclasses of fungal mitochondrial plasmids, virus or phage DNAs andtransposable elements [71]. The terminal proteins are presumablyinvolved in the initiation of DNA replication at both 50 termini. Fivelinear plant mitochondrial plasmids have been studied moreclosely: the Z. mays S1 [72], S2 [73] and 2.3 kb [74] plasmids, theB. vulgaris 10.4 kb plasmid (GenBank accession Y10854), and theBrassica napus 11.6 kb plasmid [75]. The terminal inverted repeatsat their ends range from 170 bp (Z. mays 2.3 kb plasmid) to 407 bp(B. vulgaris 10.4 kb plasmid). TIR sequences are unrelated, except forthe S1 and S2 plasmids, which share identical terminal repeats.Apart from the Z. mays 2.3 kb plasmid, linear mitochondrial plas-mids contain several open reading frames (up to 6 in the 11.6 kbB. napus plasmid) that are larger than those present in circularplasmids. Several of these orfs are considered to encode DNA orRNA polymerases, while the others have no recognizable function.Like the nuclear-encoded RNA polymerases that transcribe themain mitochondrial genome, linear plasmid-encoded RNA poly-merases are of phage-type. But the two classes of enzymes aredifferent. Plasmid-encoded DNA polymerases have similaritieswith group B DNA polymerases characteristic of phages and viruses.The presence of DNA and RNA polymerase genes in linear mito-chondrial plasmids supports the idea that these plasmids are

autonomously replicated and transcribed. It has been proposed thatgene conversion involving the TIRs (considered as palindromictelomeres) can shape the architecture of mitochondrial linear ge-netic elements, including plasmids [76]. Transcription and trans-lation of putative genes carried by linear plasmids have beendetected, especially of those coding for RNA or DNA polymeraseslikely to be required for expression and replication [77,78].Expression of orfs with unknown function was established inparticular for the 11.6 kb linear plasmid of B. napus, demonstratingthat these are functional genes [75]. Similarly, the protein encodedby orf1 of the Z. mays 2.3 kb plasmid could be detected [79]. Thispolypeptide is similar to a domain of the putative RNA polymeraseencoded by the Z. mays S2 linear plasmid, but it was hypothesizedthat it is rather the protein that covalently binds to the 50 ends ofthe 2.3 kb plasmid [79]. Interestingly, transcription appeared tostart inside or just at the end of the TIR sequence in both the 11.6 kbB. napus plasmid and the 2.3 kb Z. mays plasmid [75,79]. The lattercarries the only functional tRNATrp gene of Z. mays mitochondriaand is therefore the only plant mitochondrial episome with a clearfunction, explaining its ubiquitous presence in Z. mays lines [74].The presence of the S1 and S2 plasmids of Z. mays and their ex-changes with themainmtDNAwere reported to be involved in CMSor non-chromosomal stripe (NCS) phenotypes, although a strictcorrelation remained difficult to establish (discussed in Ref. [54]).Conversely, mtDNA rearrangements involving short repeats sharedwith the S2 plasmid were reported to be responsible for both CMS-S reversion to fertility and NCS mutation [80,81]. However, CMS-Sreversion is not necessarily accompanied by a loss of the S plas-mids [82]. Similarly, association of the 11.6 kb linear plasmid withCMS in Brassica was found not to be exclusive [83]. In Lolium per-enne, a linear plasmid-like element carrying the characteristic DNAand RNA polymerase orfs was found to be integrated into the mainmtDNA in a CMS line, while in fertile lines it is present as a low-copy number extrachromosomal replicon [84].

Mitochondrial plasmids are transmitted uniparentally from thematernal plant to the progeny, together with the main genome.However, deviations from a strict maternal inheritance wereobserved for the Beta maritima 10.4 kb linear plasmid [85]. Thepossibility of paternal inheritance through the pollen was estab-lished and extensively studied for the 11.6 kb Brassica linearplasmid [86,87].

The origin of mitochondrial plasmids is mostly unknown buttheir presence in many plant species indicates that their acquisitionis not a rare event. Short similarities were highlighted betweenseveral circular plasmids from different species, raising the idea of acommon ancestor [88]. Mitochondrial plasmids often contain se-quences that have nuclear counterparts, which in turn suggestsintercompartment exchange. Conversely, the invertron structure oflinear plasmids [71] rather points out to a common ancestralinvertron of viral origin [54]. However, the orf sequences and thestructural organization differ among plant mitochondrial linearplasmids, in agreement with independent origins. Finally, it hasbeen speculated that horizontal DNA transfer, in particular fromfungi, might have played a role in the acquisition of linear plasmidsby plant mitochondria [54].

5. Recombination control

5.1. Recombination factors

At the core of the organellar homologous recombination (HR)activities, there are eubacterial-type RecA proteins inherited fromthe corresponding prokaryotic endosymbiont ancestors. RecA cat-alyzes DNA strand-exchange during homologous recombination.Initially, RecA polymerizes on single-stranded DNA to form a long

J.M. Gualberto et al. / Biochimie 100 (2014) 107e120112

presynaptic DNA-protein filament. In the presence of homologousdouble-stranded DNA, RecA catalyzes strand exchange, forming aD-loop that is a common intermediate in homologous recombina-tion pathways. The process can be modulated by many factors.These include SSB proteins, which have very high affinity for single-stranded DNA that they sequester from RecA, and recombinationmediators, which can displace SSBs and promote DNA annealing.After strand invasion, several outcomes are possible, according tothe recombination pathway that is engaged (reviewed in Ref. [89])and the available recombination factors. The processes that areactive in plant mitochondria are still not identified.

In A. thaliana and all flowering plants, there are two mito-chondrial RecA-like proteins, RECA2 and RECA3 (Table 1). WhileRECA3 is strictly mitochondrial, RECA2 was described as dual-targeted [90], but loss of RECA2 only affects mitochondrial func-tions [16]. By heterologous expression in Escherichia coli, it wasshown that RECA2 and RECA3 are functional RecAs, with over-lapping but also specific activities [16]. Accordingly, the A. thalianarecA2 and recA3 mutants displayed different phenotypes: recA2plants are unable to develop past the seedling stage, while recA3plants are phenotypically normal [16,90].

The essential functions of RECA2 are still unknown, but theymight be related to C-terminal sequences that it shares with mostknown RecA proteins and that are not present in RECA3. The recA2and recA3 mutants have normal mtDNA copy numbers, indicatingthat RecA functions are not essential for replication. But both recA2and recA3 mutants display molecular phenotypes of increasedrecombination between ISRs [16]. The RecA-independent homol-ogous recombination activity is exacerbated in recA2 mutants,where accumulation of crossover products is an order of magnitudehigher than in recA3 mutants. As discussed below, RECA2 andRECA3, as well as additional genes, are apparently involved insurveillance pathways that restrict mtDNA recombination. Thereshuffling of the mitochondrial genome by recombination isprobably the cause of the observed recA2 lethal phenotypes.

Table 1Nuclear-encoded proteins involved or suspected to be involved in organelle DNA replica

Category Name AGI

DNA polymerase Pol1A At1g50840Pol1B At3g20540

Helicase/primase Twinkle At1g30680Topoisomerase GyrIIA At3g10690

GyrIIB1 At3g10270GyrIIB2 At5g04130Topo I At4g31210

Recombinase RECA2 At2g19490RECA3 At3g10140

ssDNA-binding/recombination mediator SSB1 At4g11060SSB2 At3g18580OSB1 At3g18580OSB2 At4g20010OSB3 At5g44785OSB4 At1g31010WHY1 At1g14410WHY2 At1g71260WHY3 At2g02740ODB1 At1g71310ODB2 At5g47870

MutS MSH1 At3g24320DNA glycosylase UNG At3g18630

FPG At1g52500OGG1 At1g21710

AP endonuclease ARP At2g41460APE1L At3g48425APE2 At4g36050

DNA ligase LIG1 At1g08130

ssDNA: single-stranded DNA; Mt: mitochondrion; Cp: chloroplast; Nuc: nucleus.a Predicted (SUBA3, http://suba.plantenergy.uwa.edu.au).

As RECA2 and RECA3, additional core elements of the plantmtDNA maintenance machinery were inherited from the bacterialancestors of plant organelles. That is the case of the eubacterial-type single-stranded DNA-binding proteins SSB. In bacteria, SSB isan abundant essential protein that cooperatively binds single-stranded DNA, forming protein filaments that saturate longstretches of the DNA. Binding of SSB protects the exposed single-stranded DNA from nucleases, but also melts secondary struc-tures that inhibit elongation of RecA-single-stranded DNA fila-ments. However, the high-affinity binding of SSB also limits RecAassembly onto single-stranded DNA [91]. To overcome recombi-nation inhibition by SSB, recombination mediators are required.Mediators can be other single-stranded DNA-binding proteins thatpromote the displacement of SSB and the assembly of RecA [92]. InA. thaliana and in most flowering plants, there are two SSB genes,SSB1 and SSB2 (Table 1) [33,41]. Both proteins are targeted tomitochondria ([41] and our unpublished results). SSB1 was shownto have similar properties as E. coli SSB: it specifically binds single-stranded DNA but not double-stranded DNA and stimulates RecA-dependent strand invasion [41]. Essential functions for SSB2 wereconfirmed in A. thaliana, where ssb2mutants are unable to developpast the seedling stage (our unpublished results).

In addition to eubacterial-type SSBs, plant mitochondria havespecific single-stranded DNA-binding proteins. The precise mo-lecular functions of these proteins are not well characterized, butthey might have multiple and possibly redundant roles in mtDNAreplication, recombination and repair. Among them there are theorganellar single-stranded DNA-binding proteins OSB (Table 1),which in A. thaliana form a small family of four proteins targeted toeither mitochondria or plastids, or dual-targeted to both organelles[93]. OSB proteins are characterized by the presence of a centraldomain similar to the OB-fold domain followed by one, two or threeC-terminal PDF domains. The PDF domain is a well-conserveddomain of 50 amino acids only found in OSB proteins. It isrequired for proteineDNA interaction [93]. In A. thaliana, the

tion, recombination and repair in A. thaliana.

Targeting Function References

Mt þ Cp Replication [34e37,144]Mt þ Cp Replication, RepairMt þ Cp Replication [35,39,145]Mt þ Cp Replication? [40]CpMtMt þ Cp Replication? [35]Mt þ Cp Recombination surveillance [16,90]Mt Recombination surveillance, RepairMta Replication, Repair? [33,41]MtMt Recombination surveillance [93]Cp n.d.Mt þ Cp n.d.Mt Recombination surveillance UnpublishedCp Repair [95,146]Mt [17,146]Cp [95,146]Mt þ Nuc? Repair [96,97,147]Cp þ Nuc? [96,97]Mt þ Cp Recombination surveillance [106]Mt Repair [31,148]Nuca [123,124]Nuc, Cpa [121,122]Cp Repair [118]Nuca þ Cp [127]Nuca þ Mta

Mt þ Nuc Replication, repair [130]

J.M. Gualberto et al. / Biochimie 100 (2014) 107e120 113

molecular phenotype of the mutants for the mitochondrion-targeted OSB1 and OSB4 proteins is the same as that of the recA2and recA3 mutants: mutant plants accumulate aberrant mtDNAsequences resulting from ectopic recombination between certainISRs ([93] and our unpublished results). This suggests that OSBproteins are also involved in the surveillance of mtDNA recombi-nation. The specific expression of OSB1 in gametophytic cellsfurther suggests that recombination surveillance by OSB proteins isrequired to prevent transmission of aberrant mtDNA configurationsgenerated by ectopic recombination [93].

Whirly (WHY) proteins are further single-stranded DNA-bind-ing proteins present in plant organelles (reviewed in Refs. [89,94]).In plastids, WHY proteins have multiple DNA- and RNA-relatedroles [89,94]. In A. thaliana mitochondria there is a single WHYprotein (WHY2) (Table 1), whose roles are not yet completely un-derstood. Over-expression of WHY2 affected the transcription ofseveral mitochondrial genes but this could be a non-specific effectdue to masking of the mtDNA by WHY2. Rather, like chloroplastWHY1 and WHY3, mitochondrial WHY2 seems to be involved inthe surveillance of illegitimate recombination, preventing error-prone pathways of double-strand break (DSB) repair throughsequence microhomologies [17,95]. It is possible that WHYproteinsfunction as recombination mediators, promoting RecA-dependenthomologous recombination repair pathways to the detriment ofalternative illegitimate recombination processes.

Proteins with sequence and structural similarities to eukaryoticRAD52, a recombinationmediator, are present in plants [96,97]. Thetwo RAD52-like genes identified in the A. thaliana genome werereported to generate four orfs through differential splicing, with therespective isoforms being distributed between the nucleus, mito-chondria and chloroplasts [96]. However, a specific antibody onlydetected the mitochondrial and chloroplast isoforms [97]. LikeWHY2, the RAD52-like mitochondrial protein ODB1 (organellarDNA-binding) is possibly an additional recombination mediator(Table 1). ODB1 co-purified with WHY2 from cauliflower mito-chondrial extracts and both proteins were shown to co-immunoprecipitate [97]. ODB1 has a high affinity for single-stranded DNA, with no apparent sequence specificity, but is alsoable to bind double-stranded DNA, although with much less af-finity. On the other hand, ODB1 is able to promote annealing ofcomplementary sequences [97]. In this respect, ODB1 seems to be afunctional homolog of yeast mitochondrial Mgm101p. Alsodistantly related to RAD52, the latter is involved in mtDNA main-tenance and has recombination mediator properties similar tothose of RAD52 [98e100]. In agreement with this assumption,A. thaliana knockout mutants of ODB1 are impaired in homologousrecombination-dependent repair of mtDNA breaks generated bygenotoxic treatment [97].

An important component of the mitochondrial recombinationmachinery is the MSH1 gene, which encodes a MutS-like protein(Table 1). MSH1 has essential functions in the surveillance ofmtDNA recombination. The gene was identified in A. thaliana asresponsible for the cytoplasmic inherited CHM (chloroplast muta-tor) phenotype. The msh1 mutants display phenotypes of varie-gated and distorted leaves that correlated with importantrearrangements of the mtDNA by recombination [15,90,101e103].In additional species where MSH1 was knocked down by RNAsilencing, mtDNA recombination was also increased, accompaniedby phenotypic defects that included male sterility [104]. InA. thaliana, MSH1 is targeted to bothmitochondria and chloroplasts[105] and recently it was shown that it is also active in the main-tenance of the plastidial genome, explaining the variegationphenotype of msh1 plants [106]. In bacteria, MutS is a protein thatparticipates in mismatch repair and in the suppression of ectopicrecombination. Together with MutL, it recruits the endonuclease

MutH to mismatches that arise during replication and homologousrecombination [107]. There is no known MutH-like gene inA. thaliana, but plant mitochondrial MSH1 contains a C-terminalGIY-YIG-type homing endonuclease domain. It is therefore possiblethat MSH1 combines mismatch recognition and endonucleasefunctions. In recombination reactions, MSH1 probably recognizesheteroduplexes in strand-exchange complexes when there is littlesequence homology. MSH1 would then promote their rejection[105,108].

Additional putative components of the plant mitochondrialrecombination machinery have been identified by their similarityto known factors from other species, including DNA polymerases.As discussed in Section 3, in A. thaliana there are two organellarDNA polymerases, Pol1A and Pol1B (Table 1), which are targeted toboth mitochondria and plastids and have redundant roles in DNAreplication [33e35,37]. However, Pol1B, but not Pol1A, apparentlyhas specific functions in DNA repair, because pol1B mutants, andnot pol1A mutants, are hypersensitive to genotoxic agents thatspecifically induce DNA breaks in organellar genomes [37].

Finally, although many factors involved in mtDNA recombina-tion have been identified either by their DNA-binding activities orfrom sequence comparisons, many additional ones are predicted tobe required that remain to be identified. These might not be ofeubacterial origin, because corresponding genes could not be foundin the A. thaliana genome. They might be of viral origin or mighthave been recruited from the eukaryotic host. For example, aninitial fundamental step in homologous recombination of DSBs isthe recognition of free DNA extremities and their resection by 50e30

exonucleases, to form long 30-overhangs that are substrates forRecA and other mediators. DNA helicases should also have impor-tant roles in recombination, for instance during recession, to un-wind branched duplex DNA and help to process Holliday-junctionsby catalyzing branch migration. Several candidates can be pin-pointed, from genes coding for proteins with corresponding func-tional motifs and predicted organellar targeting sequences.However, the putative mitochondrial functions of these proteinsremain to be validated.

5.2. Recombination specificity

Homologous recombination involves the exchange between theinvading single strand of the donor DNA and the double-strandedrecipient DNA. The length of sequence homology required forefficient strand-exchange is characteristic of the recombinase en-zymes that catalyze the reaction, which in plant mitochondria areeubacterial-type RecAs [16,90,109]. The minimal pairing sequenceof bacterial RecA can be as small as 25 bp, but efficiency greatlyincreases with sizes of sequence homology greater than 200 bp[110]. Repeated sequences are the main sites of intramolecularrecombination in plant mtDNAs and repeats of diverse size andnumber have been identified in the plant mitochondrial genomesso far sequenced [3,111]. Accordingly, the efficiencies of recombi-nation observed in plant mitochondria correlate with the size andhomology of the repeats. Large repeats of sizes larger than 1 kbmediate high frequency, reciprocal homologous recombination.This type of recombination involving large repeats is responsiblefor the well described multipartite structure of plant mitochondrialgenomes. On the other hand, ISRs recombine sporadically, andoften the products of their recombination are not detectable byconventional map-based sequencing approaches. But such prod-ucts can be visualized on overexposed Southern blots and by PCR.The infrequent ectopic recombination mediated by these repeatsleads to many possible alternative configurations of themtDNA andhas been associated with intraspecific variation of the mtDNA[15,103]. But this type of recombination can also lead to deleterious

RecA-independent Recombination

RECA3, RECA2, MSH1, OSB1, OSB4

Break-Induced Replication (BIR)

DSB in mtDNA

RecA-dependent D-loop formation

RECA3, RECA2 ODB1, WHY2

POL1B

Increased recombination involving ISRs

Single-Strand Annealing (SSA)

MicroHomology-Mediated Recombination (MHMR)

Synthesis-Dependent Strand Annealing (SDSA)

Gene conversion without crossovers

Fig. 2. Alternative recombination-dependent repair pathways that might exist in plant mitochondria. Double strand breaks (DSBs) of the mtDNA can be repaired by RecA-dependent and independent homologous recombination (HR) pathways. RecA-independent repair of mtDNA DSBs that occur in somatic tissues can lead to increased recombi-nation involving intermediate-size repeats (ISRs), probably by single-strand annealing (SSA). These recombination events are under recombination surveillance, and are de-repressed in certain mutant backgrounds (mutants deficient in RECA3, RECA2, MSH1, OSB1 or OSB4). DSBs that can be induced by genotoxic treatments are repaired by break-induced replication (BIR), which requires RecA activity for initial D-loop formation. Mutants deficient in RecA-dependent mtDNA repair show reduced recombination followinggenotoxic treatment and can accumulate products of illegitimate microhomology-mediated recombination (MHMR).

J.M. Gualberto et al. / Biochimie 100 (2014) 107e120114

changes in gene expression, and is under tight control. Thus, mu-tants of several genes that apparently modulate mtDNA recombi-nation can have dramatically higher frequencies of ectopicrecombination between ISRs, correlating with phenotypes of dis-torted and variegated leaves. Among these, mutants affected in theexpression of MSH1 have been extensively characterized in severalplant species [90,106,108]. In A. thaliana, in addition to MSH1, othergenes have been identified whose mutants have the same molec-ular phenotypes of increased ectopic recombination across ISRs.These are the genes coding for the two mitochondrion-targetedRecA proteins RECA2 and RECA3 [16] and the plant-specificorganellar single-stranded DNA-binding proteins OSB1 and OSB4([93] and our unpublished results). The mtDNA rearrangementsinduced by the loss of any of these surveillance genes concern ISRsranging from about 50 bp up to 556 bp in A. thaliana [15,16,103].These can be identical in sequence or imperfect repeated se-quences, and the efficiency of sequence exchange appears toloosely correlate with repeat size and sequence homology. How farapart the repeats are located in the genome and how they areoriented might also influence the recombination efficiency and thedifferential accumulation of the corresponding crossover products[15,16].

The similar molecular and visible phenotypes of these mutantssuggest that the corresponding genes are involved in recombina-tion surveillance processes that restrict recombination whensequence homology between repeats is below a certain size. In thecase of MSH1, like for other MutS factors, when there is littlesequence homology the protein might recognize heteroduplexes instrand-exchange complexes, and promote their rejection. The C-terminal GIY-YIG-type homing endonuclease domain of MSH1could be responsible for cleaving the heteroduplexes and startingover the process of sequence homology scanning [105]. However,different recombination pathways might be affected in other mu-tants. In the case of the RECA2 and RECA3 mutants, the observedincrease in ectopic recombination suggests that it is the loss ofRecA-dependent recombination that favors RecA-independentpathways under more relaxed control. These could involve theSingle-Strand Annealing recombination pathway (SSA) that canoccur when complementary single-stranded DNA sequences areexposed [16]. The OSB1 and OSB4 proteinsmight also be involved in

modulating RecA-dependent recombination, for instance by pro-moting or suppressing binding of RecA to single-stranded DNA.

In addition to ectopic recombination that is suppressed undernormal plant conditions, it was observed that genotoxic treatmentsalso result in increased ectopic recombination across ISRs. This canbe explained by the mobilization of recombination factors for therepair of DNA breaks by the BIR pathway. This homologousrecombination-dependent repair pathway depends on RecA-likeactivities and accordingly it was found that genotoxin-inducedmtDNA recombination is suppressed in RECA3 mutants [16].Other genes were also found to be involved in this genotoxin-induced repair process: the organellar DNA polymerases Pol1B[37], the Rad52-like protein ODB1 [97] and the plant-specific sin-gle-stranded DNA-binding protein WHY2 [17]. The correspondingmutants are not affected in recombination surveillance undernormal growth conditions, but under genotoxic treatment theyhave reduced recombination, as compared to wild-type plants.Concomitant with the loss of the RecA-dependent repair activity,these mutants also display increased illegitimate recombinationinvolving microhomologies of just a few nucleotides [17,37,97].Such sequence chimeras are of very lowabundance, and can only bedetected by PCR-based approaches. But collectively they likelyconstitute a major source of mtDNA heteroplasmy. On an evolu-tionary time scale, they are probably responsible for the importantchanges in the mtDNA structure observed between plant species.

6. Mitochondrial DNA repair

6.1. Recombination-dependent repair

Repair of DSBs is essential for genome stability because they canlead to collapsed replication forks, gene loss and genome insta-bility. The sequence of plant mtDNAs revealed numerous examplesof gene chimeras or fusions of unrelated sequences likely emergingfrom repair by illegitimate recombination, involving none or just afew nucleotides. Several of these gene chimeras have been associ-ated with CMS phenotypes.

In chloroplasts there is extensive evidence for the participationof plastid-targeted RecA in DNA repair [112e114]. In Physcomitrellapatens, it was shown that disruption of amitochondrial RecA results

Fig. 3. Short-patch base excision repair in plant mitochondria. The first step involveseither a monofunctional DNA-glycosylase (1) or a bifunctional DNA glycosylase (2).UNG, uracil-DNA glycosylase; OGG1, 8-oxoguanine-DNA glycosylase; FPG:formamidopyrimidine-DNA glycosylase; APE2 and APEL1, homologs of AP endonu-cleases from yeast and bacteria. For enzymes followed by question marks, theinvolvement is not established.

J.M. Gualberto et al. / Biochimie 100 (2014) 107e120 115

in reduced recovery of the mtDNA following genotoxic damage[115]. In the mitochondria of flowering plants, it was demonstratedthat RECA3 has specific functions in mtDNA repair [16]. In wild-type A. thaliana plants, treatments that induce DNA repair activ-ities result in the accumulation of specific ISR-derived crossoverproducts. This induced recombination activity is lost in the recA3background, showing that RECA3 is part of an HR-dependent repairactivity induced by genotoxic stress [16]. The possible roles ofRECA2 could not be assessed, given the seedling-lethal phenotypeof the recA2 mutant.

The patterns of recombination that are triggered by RECA3-dependent repair activities are specific and distinct from RECA-independent ones that spontaneously occur in recA3, recA2 andthe other mutants involved in recombination surveillance (Fig. 2).In particular, crossover events that result from recombinationbased on a repeated sequence spanning the 30-end of the atp9 geneand part of the downstream intergenic region towards the cox3gene are diagnostic of each one of these pathways (i.e. RECA-dependent repair induced by genotoxic treatment and RECA-independent recombination relaxed in surveillance mutants). Thiscould be tested by qPCR analysis of plants grown in the presenceand absence of genotoxic drugs, such as the antibiotic ciprofloxacinthat specifically inhibits the organellar-targeted gyrase, thus pro-voking DSBs in the organellar genomes but not in the nuclear DNA.The studies showed that RECA3 and the RAD52-like protein ODB1are involved in homologous recombination-dependent repair ofthe mtDNA [16,97]. Similar experiments confirmed the specificinvolvement in repair of the single-stranded DNA-binding proteinWHY2 and of the organellar DNA polymerase POL1B, but not ofPOL1A (our unpublished results).

As a consequence of deficient homologous recombination-dependent mtDNA repair, alternative pathways of illegitimaterecombination repair are apparently activated, in particular non-homologous end joining (NHEJ) and microhomology-mediatedrecombination (MHMR) (Fig. 2). This was shown in why2, pol1Band odb1mutants. Increased repair by NHEJ andMHMR-dependentpathways can explain the chimeric sequences often observed inplant mitochondrial genomes.

Importantly, mutants deficient in mitochondrial recombination-dependent repair can be hypersensitive to genotoxic drugs thatinduce DNA breaks [16,37], showing that repair of organellargenome lesions significantly contributes to the survival of plantsunder stress conditions. It is apparent that many studies dealingwith plant responses under stress often overlook the contributionof the organellar genomes to these processes.

6.2. Base excision repair

A major consequence of mtDNA oxidation caused by respirationbyproducts or environmental conditions is the formation ofabnormal bases. Oxidized bases are preferentially replaced throughbase excision repair (BER). The BER pathway [116] is initiated by aspecific DNA glycosylase that cleaves the N-glycosidic bond be-tween the abnormal base and the deoxyribose, thus generating anabasic (AP) site. Two ways are then possible for the process toproceed, short-patch BER (spBER) and long-patch BER (lpBER). Inthe case of spBER, if the cleavage has been carried out by a mono-functional DNA glycosylase, an AP endonuclease recognizes the APsite and cleaves the phosphodiester chain on the downstream side,leaving a 50-deoxyribose phosphate (dRp) end. DNA polymerasethen extends the upstream 30-hydroxyl end, adds the missingnucleotide and eliminates the 50dRp. Finally, the gap is sealed by aligase. When the DNA glycosylase is bi-functional and possesses alyase activity, it not only cleaves the N-glycosidic bond but alsomakes a single-stranded incision on the 30 side of the AP site. On the

other side, the AP endonuclease subsequently generates a 30-hy-droxyl end that is used by DNA polymerase. One nucleotide isinserted in the case of spBER, while for lpBER 2e6 nucleotides areadded from the 30-hydroxyl end, thus displacing the nucleotidesdownstream of the gap on the same DNA strand. Plant mitochon-dria possess an spBER pathway (Fig. 3) but did not seem to runlpBER [31], contrary to mammalian mitochondria [117].

The first step of the BER pathway involves a DNA glycosylase. InA. thaliana, DNA glycosylases were found in mitochondria [31] andin chloroplasts [118]. Several DNA glycosylase activities exist inplantmitochondria (Table 1). Uracil-DNA glycosylase (UDG) activitywas detected by in vitro assays with mitochondrial fractions fromZ. mays seedlings [119] and by in vitro and in organello assays withA. thaliana and Solanum tuberosum mitochondria [31]. The enzymedid not seem to have a lyase activity and was in part membrane-associated. In vivo targeting and in vitro import of GFP fusion pro-teins showed that the enzyme encoded by the single putativeA. thaliana UDG gene is indeed targeted into mitochondria [31]. For8-oxoguanine-DNA glycosylase (OGG1) and formamidopyrimidine-DNA glycosylase (FPG), which are functional analogs, the situationis less clear. A DNA glycosylase activity recognizing 8-oxoguanine inA. thaliana and S. tuberosummitochondria was characterized, usingin vitro and in organello assays (Boesch P., Paulus F. and Dietrich A.,unpublished data). Plants are the only organisms where the pres-ence of both OGG1 and FPG has been reported [120]. There are thustwo candidates to account for the 8-oxoguanine-cleaving activity.A. thaliana OGG1 is a bi-functional enzyme that was shown tolocalize to the nucleus by GFP fusion experiments [121]. Based onbioinformatic analysis, Macovei et al. [122] suggested the presenceof a putative mitochondrial targeting sequence in the N-terminalpart of theMedicago truncatula, A. thaliana, Populus nigra and Oryzasativa OGG1. However, when considering all relevant bioinformatictools, prediction of OGG1 localization in plants is either

J.M. Gualberto et al. / Biochimie 100 (2014) 107e120116

mitochondrial or plastidic, depending on the program used, withan overall preference for plastids. FPG, also a bi-functional enzyme,was found in M. truncatula and A. thaliana [122,123]. In both spe-cies, FPG contains a nuclear targeting sequence. Remarkably, inA. thaliana there is a single copy of the FPG gene that comprises tenexons and codes potentially for seven FPG forms, due to alternativesplicing of the mRNA [124]. The different forms of FPG could bespecific for different oxidized substrates but it can also be hy-pothesized that they would be localized in different compartmentsof the cell. As a whole, one might still speculate that OGG1 or FPGwould be relocalized to the organelles when the mtDNA or thecpDNA is oxidized. This was shown to be the case for Saccharomycescerevisiae N-glycosylase1 (Ntg1). Ntg1 possesses both a nuclear anda mitochondrial targeting sequence and is sent to the compartmentwhere the DNA is oxidized [125,126]. Sumoylation, a post-translational modification, seems to be involved in Ntg1 nuclearrelocalisation [125].

After base excision, an AP endonuclease is necessary for the BERpathway to proceed. An AP endonuclease activity was character-ized in A. thaliana and S. tuberosum mitochondria [31]. A. thalianapossesses three genes, named ARP, APE1L and APE2 (Table 1), thatencode homologs of AP endonucleases from human, yeast andbacteria. Transient expression of GFP fusions established that theARP protein is targeted to chloroplasts [118]. APE1L was found in aplastid proteome [127]. Alternative splicing yields two forms ofAPE2, the longest showing organelle targeting predictions. Also inB. napus, the putative AP endonuclease is predicted to be targetedto the organelles. Upon elimination of the damaged nucleotide, aDNA polymerase is needed to fill the gap. As discussed in previoussections, in A. thaliana, two organellar g-type DNA polymerases,POL1A and POL1B, are dual-targeted to chloroplasts and mito-chondria (Table 1) [33,34,128] and both function in mtDNA repli-cation [36]. Whether one or both of these enzymes are involved inBER remains to be established. At the final step, a ligase is needed toreconnect the two parts of the DNA strand. Cordoba-Canero et al.[129] showed that DNA ligase1 is required for both short-patch andlong-patch BER in the A. thaliana nucleus. In this species, tran-scription of the DNA ligase1 gene yields one major and two minormRNA variants differing in the length of the 50 leader preceding acommon orf. According to in planta targeting of GFP fusions, thelonger protein resulting from translation starting at the first inframe AUG codon is imported into mitochondria (Table 1), whereasthe other forms are not [130]. The mitochondrial isoform actuallypossesses both an internal nuclear localization signal and amitochondrial-targeting signal. Altogether, a number of factorsinvolved in plant mitochondrial BER are still to be confirmed oridentified. As in yeast and in animals (reviewed in Ref. [32]), sharingrepair proteins and modulating factors with other compartmentsmight turn out to be a major characteristic of plant mitochondrialDNA repair pathways. As mentioned above, it was suggested that inyeast BER factors re-localize either to the nucleus or to mitochon-dria upon genotoxic stress in one or the other compartment[125,126]. One can further speculate about a possible contributionof such mechanisms to mtDNA maintenance in plants, in that casewith the chloroplast as a third player.

7. Mitochondrial genome segregation and evolution

An important consequence of mtDNA heteroplasmy is that, froma heteroplasmic parent, individuals can segregate in which certainsubstoichiometric mitotypes were amplified and became the pre-dominant mtDNA. This process is called substoichiometric shifting(SSS) [131,132]. SSS can result in the activation or silencing ofmitochondrial sequences, including the expression of gene chi-meras, thus altering mitochondrial gene expression with

potentially deleterious consequences to the plant. But SSS is alsoresponsible for the rapid evolution of the mtDNA structure and tocytoplasm-nucleus conflicts, which can result in CMS [133]. Thiswas clearly revealed by intra-specific sequence comparisonsshowing that intergenic regions change rapidly through sequenceduplications, inversions, deletions and insertions resulting fromrecombination [2]. Through the production of specific stoichio-metric changes, the activity of the MSH1 recombination factorappeared to account for the polymorphisms distinguishingA. thaliana ecotypes [103]. Mutation or down regulation of genescoding for key recombination factors like MSH1, OSB1 or RECA3triggers extensive mitochondrial genome reorganization inA. thaliana or crop species, leading to marked phenotypes[90,93,103,104]. In OSB1-mutated lines, subgenomic levels of ho-mologous recombination products accumulate, one of thembecoming predominant in subsequent generations [93]. Remark-ably, the mtDNA structural changes triggered by recombinationfactor mutation or down regulation are irreversible after SSS, and itis possible to revert to the wild type nuclear genome by back-crossing, while maintaining the modified mtDNA genome that ismaternally inherited [93,104]. In that way, down regulating MSH1allowed the production of stable, non-transgenic, male sterile lines[104].

In the case of non-allelic recombination involving ISRs, a modelbased on BIR was proposed that could account for the eventsleading to stoichiometric changes [15,90]. However, it is difficult toreconcile this model with the accumulation of recombined mtDNAin mutants deficient for RecA [16,90,134], which should be requiredfor the BIR pathway. Most probably, both RecA-dependent BIR andadditional RecA-independent homologous recombination pro-cesses co-exist in plant mitochondria and contribute to the pool ofrecombined mtDNA sequences [16]. But while BIR and additionalhomologous and illegitimate recombination processes can accountfor the accumulation of re-shuffled mtDNA sequences in somatictissues, these pathways cannot explain the maintenance andsegregation of very low copy number sequences that cannot be re-created de novo by recombination [135]. It is therefore possible thatselective replication of alternative mtDNA sequences is alsoinvolved in the SSS process.

How these molecules segregate and can be amplified during SSSis not well understood. The segregation of alternative mitotypeslikely occurs during gametogenesis, and it is therefore significantthat several genes described as involved in the suppression ofectopic recombination involving ISRs are expressed in the femalegamete [16,90,93]. This suggests that their activities are importantto repress the transmission of recombined mtDNA molecules. As ithas been speculated in animals, the segregation of mitotypes couldinvolve a genetic bottleneck during oogenesis that would reducethe copy number of transmitted mtDNA molecules. However, thereis no evidence that such a process exists in plants, because themtDNA copy number in egg cells seems to be equivalent to the onein mesophyll cells, at least in A. thaliana [136]. Since the mtDNAcopy numbers in plants are relatively small, as compared to animals[5], the existence of a bottleneck is probably not necessary.

Models have also been proposed to account for the evolution ofmitochondrial genome structures and explain the combination oflow mutation rates and genome expansion. A “mutation burden”hypothesis put forward by Lynch et al. [137] proposes that thefundamental features of genome evolution are largely defined bythe relative power of random genetic drift and mutation, i.e. pri-mary non-adaptive forces. Especially, high mutation rates would bea barrier to organelle genome evolution. In this respect, that plantand animal mitochondrial genomes evolved to opposite architec-tural complexity would be a consequence of a two order ofmagnitude difference in their mutation rates. Such a model would

J.M. Gualberto et al. / Biochimie 100 (2014) 107e120 117

stand for plant species with low mtDNA mutation rates and highmitochondrial genome expansion, but it remains difficult to applyto species like S. noctiflora or S. conica, which show both excep-tionally high mutation rates in genes and highly expanded ge-nomes. Christensen proposed a model based on DNA repair andrecombination mechanisms [138] that might account both for lowmutation rates in genes combined with high expansion, rear-rangement and mutation rates in the rest of the mitochondrialgenome, and for the conundrum posed by species with huge ge-nomes and high mutation rates in genes [4,139,140]. The modelpostulates different repair mechanisms in transcribed and non-transcribed regions [138]. Accurate template-directed DSB repairby synthesis-dependent strand annealing (SDSA) with enhancedsecond strand capture, gene conversion and BER would keep mu-tation rates low in coding regions. Conversely, BIR, NHEJ andfurther error-prone repair pathways, such as microhomology-mediated BIR, would explain expansion, rearrangements andrapid divergence in non-coding regions. In species with both highgene mutation rates and dramatic expansion of non-coding re-gions, lack of some BER or mismatch repair mechanisms wouldpromote the accumulation of damaged and unpaired bases incoding regions, the fixation of mutations, the enhancement of DSBformation and the frequency of BIR [138]. This attractive hypothesisimplies that transcription-directed repair processes exist in plantmitochondria, although the mechanisms and factors involvedremain to be identified.

8. Conclusion

In many aspects, the genetics of plant mitochondria is morecomplex than that of most other organisms. The capacity to inte-grate and/or expand intergenic non-coding sequences in theorganellar genome is remarkable, especially versus the strictlycompact mammalian mtDNA. One might speculate that gain ofsequences is related to the competence of plant mitochondria forDNA import [141]. Notably, both the acquisition of plasmids and theimport competence are shared with fungal mitochondria [142].However, mammalian organelles are competent as well [143] andtheir genome size is stable. Also, many years after their discovery,the significance of mitochondrial plasmids and their cross-talk withthe main mitochondrial genome are still a mystery. Not only thesequence content but also the physical organization of the plantmtDNA is a complex mix that either results from or supports avariety of replication and maintenance mechanisms. While rare inmammalian organelles, recombination is a major player in shapingplant mitochondrial genomes, redistributing sequences, generatingpolymorphism, driving evolution and at the same time preservingthe genetic information. In the end, the challenge remains to un-derstand how the high plasticity of the plant mtDNA can combinewith gene conservation, intercompartment genome coordinationand appropriate functional efficiency of the organelles. In thisrespect, it can be noted that, apart from reported mutants, mtDNAmaps and restriction patterns have been reproduced many timesfor many species and generations, which indicates that, althoughdynamic, plant mitochondrial genomes remain stable on anobservable time scale.

Acknowledgments

Our projects are funded by the French Centre National de laRecherche Scientifique (CNRS, UPR2357), the Université de Stras-bourg (UdS), the Agence Nationale de la Recherche (ANR-06-MRAR-037-02, ANR-09-BLAN-0240-01) and the Ministère de laRecherche et de l’Enseignement Supérieur (Investissementsd’Avenir/Laboratoire d’Excellence MitoCross).

References

[1] T. Kubo, K.J. Newton, Angiosperm mitochondrial genomes and mutations,Mitochondrion 8 (2008) 5e14.

[2] J.O. Allen, C.M. Fauron, P. Minx, L. Roark, S. Oddiraju, G.N. Lin, L. Meyer,H. Sun, K. Kim, C. Wang, F. Du, D. Xu, M. Gibson, J. Cifrese, S.W. Clifton,K.J. Newton, Comparisons among two fertile and three male-sterile mito-chondrial genomes of maize, Genetics 177 (2007) 1173e1192.

[3] A.J. Alverson, D.W. Rice, S. Dickinson, K. Barry, J.D. Palmer, Origins andrecombination of the bacterial-sized multichromosomal mitochondrialgenome of cucumber, Plant Cell 23 (2011) 2499e2513.

[4] D.B. Sloan, A.J. Alverson, J.P. Chuckalovcak, M. Wu, D.E. McCauley, J.D. Palmer,D.R. Taylor, Rapid evolution of enormous, multichromosomal genomes inflowering plant mitochondria with exceptionally high mutation rates, PLoSBiol. 10 (2012) e1001241.

[5] T. Preuten, E. Cincu, J. Fuchs, R. Zoschke, K. Liere, T. Borner, Fewer genes thanorganelles: extremely low and variable gene copy numbers in mitochondriaof somatic plant cells, Plant J. 64 (2010) 948e959.

[6] M. Unseld, J.R. Marienfeld, P. Brandt, A. Brennicke, The mitochondrialgenome of Arabidopsis thaliana contains 57 genes in 366,924 nucleotides,Nat. Genet. 15 (1997) 57e61.

[7] S. Anderson, A.T. Bankier, B.G. Barrell, M.H. de Bruijn, A.R. Coulson, J. Drouin,I.C. Eperon, D.P. Nierlich, B.A. Roe, F. Sanger, P.H. Schreier, A.J. Smith,R. Staden, I.G. Young, Sequence and organization of the human mitochon-drial genome, Nature 290 (1981) 457e465.

[8] R.M. Stupar, J.W. Lilly, C.D. Town, Z. Cheng, S. Kaul, C.R. Buell, J. Jiang,Complex mtDNA constitutes an approximate 620-kb insertion on Arabidopsisthaliana chromosome 2: implication of potential sequencing errors caused bylarge-unit repeats, Proc. Natl. Acad. Sci. U. S. A. 98 (2001) 5099e5103.

[9] M. Klein, U. Eckert-Ossenkopp, I. Schmiedeberg, P. Brandt, M. Unseld,A. Brennicke, W. Schuster, Physical mapping of the mitochondrial genome ofArabidopsis thaliana by cosmid and YAC clones, Plant J. 6 (1994) 447e455.

[10] Y. Sugiyama, Y. Watase, M. Nagase, N. Makita, S. Yagura, A. Hirai, M. Sugiura,The complete nucleotide sequence and multipartite organization of the to-bacco mitochondrial genome: comparative analysis of mitochondrial ge-nomes in higher plants, Mol. Genet. Genom. 272 (2005) 603e615.

[11] Y. Ogihara, Y. Yamazaki, K. Murai, A. Kanno, T. Terachi, T. Shiina, N. Miyashita,S. Nasuda, C. Nakamura, N. Mori, S. Takumi, M. Murata, S. Futo, K. Tsunewaki,Structural dynamics of cereal mitochondrial genomes as revealed by com-plete nucleotide sequencing of the wheat mitochondrial genome, NucleicAcids Res. 33 (2005) 6235e6250.

[12] A.J. Bendich, Reaching for the ring: the study of mitochondrial genomestructure, Curr. Genet. 24 (1993) 564e588.

[13] D.J. Oldenburg, A.J. Bendich, Size and structure of replicating mitochondrialDNA in cultured tobacco cells, Plant Cell 8 (1996) 447e461.

[14] S. Backert, R. Lurz, O.A. Oyarzabal, T. Borner, High content, size and distri-bution of single-stranded DNA in the mitochondria of Chenopodium album(L.), Plant Mol. Biol. 33 (1997) 1037e1050.

[15] J.I. Davila, M.P. Arrieta-Montiel, Y. Wamboldt, J. Cao, J. Hagmann, V. Shedge,Y.Z. Xu, D. Weigel, S.A. Mackenzie, Double-strand break repair processesdrive evolution of the mitochondrial genome in Arabidopsis, BMC Biol. 9(2011) 64.

[16] M. Miller-Messmer, K. Kuhn, M. Bichara, M. Le Ret, P. Imbault, J.M. Gualberto,RecA-dependent DNA repair results in increased heteroplasmy of the Ara-bidopsis mitochondrial genome, Plant Physiol. 159 (2012) 211e226.

[17] L. Cappadocia, A. Marechal, J.S. Parent, E. Lepage, J. Sygusch, N. Brisson,Crystal structures of DNAeWhirly complexes and their role in Arabidopsisorganelle genome repair, Plant Cell 22 (2010) 1849e1867.

[18] J.W. Lilly, M.J. Havey, Small, repetitive DNAs contribute significantly to theexpanded mitochondrial genome of cucumber, Genetics 159 (2001) 317e328.

[19] M. Kucej, R.A. Butow, Evolutionary tinkering with mitochondrial nucleoids,Trends Cell Biol. 17 (2007) 586e592.

[20] A.P. Rebelo, L.M. Dillon, C.T. Moraes, Mitochondrial DNA transcriptionregulation and nucleoid organization, J. Inherit. Metab. Dis. 34 (2011) 941e951.

[21] R. Gilkerson, L. Bravo, I. Garcia, N. Gaytan, A. Herrera, A. Maldonado,B. Quintanilla, The mitochondrial nucleoid: integrating mitochondrial DNAinto cellular homeostasis, Cold Spring Harb. Perspect. Biol. 5 (2013) a011080.

[22] H. Dai, Y.S. Lo, A. Litvinchuk, Y.T. Wang, W.N. Jane, L.J. Hsiao, K.S. Chiang,Structural and functional characterizations of mung bean mitochondrialnucleoids, Nucleic Acids Res. 33 (2005) 4725e4739.

[23] Y.-S. Lo, L.-J. Hsiao, N. Cheng, A. Litvinchuk, H. Dai, Characterization of thestructure and DNA complexity of mung bean mitochondrial nucleoids, Mol.Cells 31 (2011) 217e224.

[24] W. Majeran, G. Friso, Y. Asakura, X. Qu, M. Huang, L. Ponnala, K.P. Watkins,A. Barkan, K.J. van Wijk, Nucleoid-enriched proteomes in developing plastidsand chloroplasts from maize leaves: a new conceptual framework fornucleoid functions, Plant Physiol. 158 (2012) 156e189.

[25] A. Sakai, H. Takano, T. Kuroiwa, Organelle nuclei in higher plants: structure,composition, function, and evolution, Int. Rev. Cytol. 238 (2004) 59e118.

[26] M. Takusagawa, T. Hayashi, H. Takano, A. Sakai, Organization ofmitochondrial-nucleoids in BY-2 cultured tobacco cells, Cytologia 74 (2009)329e341.

J.M. Gualberto et al. / Biochimie 100 (2014) 107e120118

[27] B.Z.M. Katherine, J.M. Donohue, B.A. Everitt, Evidence that core histone H3 istargeted to the mitochondria in Brassica oleracea, Cell Biol. Int. 34 (2010)997e1003.

[28] M.B. Scher, A. Vaquero, D. Reinberg, SirT3 is a nuclear NADþ-dependenthistone deacetylase that translocates to the mitochondria upon cellularstress, Genes Dev. 21 (2007) 920e928.

[29] M. Takusagawa, S. Tamotsu, A. Sakai, Histone H3 is absent from organellenucleoids in BY-2 cultured tobacco cells, Cell Biol. Int. 37 (2013) 748e754.

[30] Y.S. Lo, N. Cheng, L.J. Hsiao, A. Annamalai, G.Y. Jauh, T.N. Wen, H. Dai,K.S. Chiang, Actin in mung bean mitochondria and implications for itsfunction, Plant Cell 23 (2011) 3727e3744.

[31] P. Boesch, N. Ibrahim, F. Paulus, A. Cosset, V. Tarasenko, A. Dietrich, Plantmitochondria possess a short-patch base excision DNA repair pathway,Nucleic Acids Res. 37 (2009) 5690e5700.

[32] P. Boesch, F. Weber-Lotfi, N. Ibrahim, V. Tarasenko, A. Cosset, F. Paulus,R.N. Lightowlers, A. Dietrich, DNA repair in organelles: pathways, organiza-tion, regulation, relevance in disease and aging, Biochim. Biophys. Acta 1813(2011) 186e200.

[33] A. Elo, A. Lyznik, D.O. Gonzalez, S.D. Kachman, S.A. Mackenzie, Nuclear genesthat encode mitochondrial proteins for DNA and RNA metabolism are clus-tered in the Arabidopsis genome, Plant Cell 15 (2003) 1619e1631.

[34] A.C. Christensen, A. Lyznik, S. Mohammed, C.G. Elowsky, A. Elo, R. Yule,S.A. Mackenzie, Dual-domain, dual-targeting organellar protein pre-sequences in Arabidopsis can use non-AUG start codons, Plant Cell 17 (2005)2805e2816.

[35] C. Carrie, K. Kuhn, M.W. Murcha, O. Duncan, I.D. Small, N. O’Toole, J. Whelan,Approaches to defining dual-targeted proteins in Arabidopsis, Plant J. 57(2009) 1128e1139.

[36] J.D. Cupp, B.L. Nielsen, Arabidopsis thaliana organellar DNA polymerase IBmutants exhibit reduced mtDNA levels with a decrease in mitochondrialarea density, Physiol. Plant. 149 (2013) 91e103.

[37] J.S. Parent, E. Lepage, N. Brisson, Divergent roles for the two PolI-likeorganelle DNA polymerases of Arabidopsis, Plant Physiol. 156 (2011) 254e262.

[38] D.B. Udy, S. Belcher, R. Williams-Carrier, J.M. Gualberto, A. Barkan, Effects ofreduced chloroplast gene copy number on chloroplast gene expression inmaize, Plant Physiol. 160 (2012) 1420e1431.

[39] J. Diray-Arce, B. Liu, J.D. Cupp, T. Hunt, B.L. Nielsen, The ArabidopsisAt1g30680 gene encodes a homologue to the phage T7 gp4 protein that hasboth DNA primase and DNA helicase activities, BMC Plant Biol. 13 (2013) 36.

[40] M.K. Wall, L.A. Mitchenall, A. Maxwell, Arabidopsis thaliana DNA gyrase istargeted to chloroplasts and mitochondria, Proc. Natl. Acad. Sci. U. S. A. 101(2004) 7821e7826.

[41] A.C. Edmondson, D. Song, L.A. Alvarez, M.K. Wall, D. Almond, D.A. McClellan,A. Maxwell, B.L. Nielsen, Characterization of a mitochondrially targetedsingle-stranded DNA-binding protein in Arabidopsis thaliana, Mol. Genet.Genom. 273 (2005) 115e122.

[42] J.M. de Haas, J. Hille, F. Kors, B. van der Meer, A.J. Kool, O. Folkerts,H.J. Nijkamp, Two potential Petunia hybrida mitochondrial DNA replicationorigins show structural and in vitro functional homology with the animalmitochondrial DNA heavy and light strand replication origins, Curr. Genet.20 (1991) 503e513.

[43] D.L. Robberson, D.A. Clayton, Replication of mitochondrial DNA in mouse Lcells and their thymidine kinase e derivatives: displacement replication on acovalently-closed circular template, Proc. Natl. Acad. Sci. U. S. A. 69 (1972)3810e3814.

[44] J. Farkasovska, Sequence analysis of a Papaver somniferum L. mitochondrialDNA fragment promoting autonomous plasmid replication in Saccharomycescerevisiae and Kluyveromyces lactis, Curr. Genet. 24 (1993) 366e367.

[45] S. Backert, P. Dorfel, T. Borner, Investigation of plant organellar DNAs bypulsed-field gel electrophoresis, Curr. Genet. 28 (1995) 390e399.

[46] D.J. Oldenburg, A.J. Bendich, Mitochondrial DNA from the liverwort March-antia polymorpha: circularly permuted linear molecules, head-to-tail con-catemers, and a 50 protein, J. Mol. Biol. 310 (2001) 549e562.

[47] S. Backert, R. Lurz, T. Borner, Electron microscopic investigation of mito-chondrial DNA from Chenopodium album (L.), Curr. Genet. 29 (1996) 427e436.

[48] S. Backert, P. Dorfel, R. Lurz, T. Borner, Rolling-circle replication of mito-chondrial DNA in the higher plant Chenopodium album (L.), Mol. Cell. Biol. 16(1996) 6285e6294.

[49] S. Backert, T. Borner, Phage T4-like intermediates of DNA replication andrecombination in the mitochondria of the higher plant Chenopodium album(L.), Curr. Genet. 37 (2000) 304e314.

[50] J.M. Gerhold, A. Aun, T. Sedman, P. Joers, J. Sedman, Strand invasion struc-tures in the inverted repeat of Candida albicans mitochondrial DNA reveal arole for homologous recombination in replication, Mol. Cell 39 (2010) 851e861.

[51] D.M. Lonsdale, J.M. Grienenberger, The mitochondrial genome of plants, in:R.G. Hermann (Ed.), Cell Organelles, Springer-Verlag, New York, 1992, pp.181e218.

[52] R.H. Brown, M. Zhang, Mitochondrial plasmids: DNA and RNA, in:C.S. Levings III, K. VasiI (Eds.), The Molecular Biology of Plant Mitochondria,Kluver Academic Publishers, Dordrecht, The Netherlands, 1995, pp. 61e91.

[53] B.O. Bengtsson, H. Andersson, The population genetics of plant mitochondrialplasmids, J. Theor. Biol. 188 (1997) 163e176.

[54] H. Handa, Linear plasmids in plant mitochondria: peaceful coexistences ormalicious invasions? Mitochondrion 8 (2008) 15e25.

[55] C. Hallden, C. Lind, T. Bryngelsson, Minicircle variation in Beta mitochondrialDNA, Theor. Appl. Genet. 77 (1989) 337e342.

[56] M.R. Hanson, M.F. Conde, Functioning and variation of cytoplasmic genomes:lesson from cytoplasmic-nuclear interactions affecting male fertility inplants, Intern. Rev. Cytol. 94 (1985) 213e267.

[57] H. Andersson-Ceplitis, Evolutionary dynamics of mitochondrial plasmids innatural populations of Silene vulgaris, Evolution 56 (2002) 1592e1598.

[58] M.M. Robison, D.J. Wolyn, A mitochondrial plasmid and plasmid-like RNAand DNA polymerases encoded within the mitochondrial genome of carrot(Daucus carota L.), Curr. Genet. 47 (2005) 57e66.

[59] M.C. Flamand, G. Duc, J.P. Goblet, L. Hong, O. Louis, M. Briquet, M. Boutry,Variant mitochondrial plasmids of broad bean arose by recombination andare controlled by the nuclear genome, Nucleic Acids Res. 21 (1993) 5468e5473.

[60] J.A. Wahleithner, D.R. Wolstenholme, Origin and direction of replication inmitochondrial plasmid DNAs of broad bean, Vicia faba, Curr. Genet. 14 (1988)163e170.

[61] S. Backert, K. Meissner, T. Borner, Unique features of the mitochondrialrolling circle-plasmid mp1 from the higher plant Chenopodium album (L.),Nucleic Acids Res. 25 (1997) 582e589.

[62] S. Backert, M. Kunnimalaiyaan, T. Borner, B.L. Nielsen, In vitro replication ofmitochondrial plasmid mp1 from the higher plant Chenopodium album (L.): aremnant of bacterial rolling circle and conjugative plasmids? J. Mol. Biol. 284(1998) 1005e1015.

[63] S. Backert, Strand switching during rolling circle replication of plasmid-likeDNA circles in the mitochondria of the higher plant Chenopodium album(L.), Plasmid 43 (2000) 166e170.

[64] S.R. Ludwig, R.F. Pohlman, J. Vieira, A.G. Smith, J. Messing, The nucleotidesequence of a mitochondrial replicon from maize, Gene 38 (1985) 131e138.

[65] A.G. Smith, D.R. Pring, Nucleotide sequence and molecular characterizationof a maize mitochondrial plasmid-like DNA, Curr. Genet. 12 (1987) 617e623.

[66] J.A. Wahleithner, D.R. Wolstenholme, Mitochondrial plasmid DNAs of broadbean: nucleotide sequences, complex secondary structures, and transcrip-tion, Curr. Genet. 12 (1987) 55e67.

[67] L. de la Canal, D. Crouzillat, M.C. Flamand, A. Perrault, M. Boutry, G. Ledoigt,Nucleotide sequence and transcriptional analysis of a mitochondrial plasmidfrom a cytoplasmic male-sterile line of sunflower, Theor. Appl. Genet. 81(1991) 812e818.

[68] M.C. Flamand, J.P. Goblet, G. Duc, M. Briquet, M. Boutry, Sequence andtranscription analysis of mitochondrial plasmids isolated from cytoplasmicmale-sterile lines of Vicia faba, Plant Mol. Biol. 19 (1992) 913e923.

[69] C.M. Thomas, Sugarbeet minicircular mitochondrial DNAs: high-resolutiontranscript mapping, transcript abundance and copy number determination,Mol. Gen. Genet. 234 (1992) 457e465.

[70] C.M. Thomas, The nucleotide sequence and transcription of minicircularmitochondrial DNA’s associated with male-fertile and cytoplasmic male-sterile lines of sugarbeet, Nucleic Acids Res. 14 (1986) 9353e9370.

[71] K. Sakaguchi, Invertrons, a class of structurally and functionally related ge-netic elements that includes linear DNA plasmids, transposable elements,and genomes of adeno-type viruses, Microbiol. Rev. 54 (1990) 66e74.

[72] M. Paillard, R.R. Sederoff, C.S. Levings, Nucleotide sequence of the S-1mitochondrial DNA from the S cytoplasm of maize, EMBO J. 4 (1985) 1125e1128.

[73] C.S. Levings, R.R. Sederoff, Nucleotide sequence of the S-2 mitochondrial DNAfrom the S cytoplasm of maize, Proc. Natl. Acad. Sci. U. S. A. 80 (1983) 4055e4059.

[74] P. Leon, V. Walbot, P. Bedinger, Molecular analysis of the linear 2.3 kbplasmid of maize mitochondria: apparent capture of tRNA genes, NucleicAcids Res. 17 (1989) 4089e4099.

[75] H. Handa, K. Itani, H. Sato, Structural features and expression analysis of alinear mitochondrial plasmid in rapeseed (Brassica napus L.), Mol. Genet.Genom. 267 (2002) 797e805.

[76] D.R. Smith, P.J. Keeling, Gene conversion shapes linear mitochondrialgenome architecture, Genome Biol. Evol. 5 (2013) 905e912.

[77] C. O’Brien, G. Zabala, V. Walbot, Integrated R2 sequence in mitochondria offertile B37N maize encodes and expresses a 130 kD polypeptide similar tothat encoded by the S2 episome of S-type male sterile plants, Nucleic AcidsRes. 17 (1989) 405e422.

[78] G. Zabala, V. Walbot, An S1 episomal gene of maize mitochondria isexpressed in male sterile and fertile plants of the S-type cytoplasm, Mol. Gen.Genet. 211 (1988) 386e392.

[79] P. Leon, C. O’Brien-Vedder, V. Walbot, Expression of ORF1 of the linear 2.3 kbplasmid of maize mitochondria: product localization and similarities to the130 kDa protein encoded by the S2 episome, Curr. Genet. 22 (1992) 61e67.

[80] C.L. Schardl, D.R. Pring, D.M. Lonsdale, Mitochondrial DNA rearrangementsassociated with fertile revertants of S-type male-sterile maize, Cell 43 (1985)361e368.

[81] K.J. Newton, J.M. Mariano, C.M. Gibson, E. Kuzmin, S. Gabay-Laughnan,Involvement of S2 episomal sequences in the generation of NCS4 deletionmutation in maize mitochondria, Dev. Genet. 19 (1996) 277e286.

[82] J.T. Matera, J. Monroe, W. Smelser, S. Gabay-Laughnan, K.J. Newton, Uniquechanges in mitochondrial genomes associated with reversions of S-typecytoplasmic male sterility in maizemar, PLoS One 6 (2011) e23405.

J.M. Gualberto et al. / Biochimie 100 (2014) 107e120 119

[83] R.J. Kemble, J.E. Carlson, L.R. Erickson, J.L. Sernyk, D.J. Thompson, The Brassicamitochondrial DNA plasmid and large RNAs are not exclusively associatedwith cytoplasmic male sterility, Mol. Gen. Genet. 205 (1986) 183e185.

[84] P. McDermott, V. Connolly, T.A. Kavanagh, The mitochondrial genome of acytoplasmic male sterile line of perennial ryegrass (Lolium perenne L.) con-tains an integrated linear plasmid-like element, Theor. Appl. Genet. 117(2008) 459e470.

[85] P. Saumitou-Laprade, G. Pannebecker, F. Maggouta, R. Jean, G. Michaelis,A linear 10.4 kb plasmid in the mitochondria of Beta maritima, Curr. Genet.16 (1989) 181e186.

[86] L. Erickson, R. Kemble, E. Swanson, The Brassica mitochondrial plasmid canbe sexually transmitted. Pollen transfer of a cytoplasmic genetic element,Mol. Gen. Genet. 218 (1989) 419e422.

[87] M. Oshima, H. Handa, The identification of quantitative trait loci that controlthe paternal inheritance of a mitochondrial plasmid in rapeseed (Brassicanapus L.), Genes Genet. Syst. 87 (2012) 19e27.

[88] T. Shikanai, Z.Q. Yang, Y. Yamada, Nucleotide sequence and molecularcharacterization of plasmid-like DNAs from mitochondria of cytoplasmicmale-sterile rice, Curr. Genet. 15 (1989) 349e354.

[89] A. Maréchal, N. Brisson, Recombination and the maintenance of plantorganelle genome stability, New Phytol. 186 (2010) 299e317.

[90] V. Shedge, M. Arrieta-Montiel, A.C. Christensen, S.A. Mackenzie, Plant mito-chondrial recombination surveillance requires unusual RecA and MutS ho-mologs, Plant Cell 19 (2007) 1251e1264.

[91] R.D. Shereda, A.G. Kozlov, T.M. Lohman, M.M. Cox, J.L. Keck, SSB as anorganizer/mobilizer of genome maintenance complexes, Crit. Rev. Biochem.Mol. Biol. 43 (2008) 289e318.

[92] S.L. Gasior, H. Olivares, U. Ear, D.M. Hari, R. Weichselbaum, D.K. Bishop,Assembly of RecA-like recombinases: distinct roles for mediator proteins inmitosis and meiosis, Proc. Natl. Acad. Sci. U. S. A. 98 (2001) 8411e8418.

[93] V. Zaegel, B. Guermann, M. Le Ret, C. Andres, D. Meyer, M. Erhardt,J. Canaday, J.M. Gualberto, P. Imbault, The plant-specific ssDNA bindingprotein OSB1 is involved in the stoichiometric transmission of mitochondrialDNA in Arabidopsis, Plant Cell 18 (2006) 3548e3563.

[94] D. Desveaux, A. Marechal, N. Brisson, Whirly transcription factors: defensegene regulation and beyond, Trends Plant Sci. 10 (2005) 95e102.

[95] A. Marechal, J.S. Parent, F. Veronneau-Lafortune, A. Joyeux, B.F. Lang,N. Brisson, Whirly proteins maintain plastid genome stability in Arabidopsis,Proc. Natl. Acad. Sci. U. S. A. 106 (2009) 14693e14698.

[96] A. Samach, C. Melamed-Bessudo, N. Avivi-Ragolski, S. Pietrokovski, A.A. Levy,Identification of plant RAD52 homologs and characterization of the Arabi-dopsis thaliana RAD52-like genes, Plant Cell 23 (2011) 4266e4279.

[97] S. Janicka, K. Kuhn, M. Le Ret, G. Bonnard, P. Imbault, H. Augustyniak,J.M. Gualberto, A RAD52-like single-stranded DNA binding protein affectsmitochondrial DNA repair by recombination, Plant J. 72 (2012) 423e435.

[98] X.J. Chen, M.X. Guan, G.D. Clark-Walker, MGM101, a nuclear gene involved inmaintenance of the mitochondrial genome in Saccharomyces cerevisiae,Nucleic Acids Res. 21 (1993) 3473e3477.

[99] S. Meeusen, Q. Tieu, E. Wong, E. Weiss, D. Schieltz, J.R. Yates, J. Nunnari,Mgm101p is a novel component of the mitochondrial nucleoid that bindsDNA and is required for the repair of oxidatively damaged mitochondrialDNA, J. Cell Biol. 145 (1999) 291e304.

[100] M. Mbantenkhu, X. Wang, J.D. Nardozzi, S. Wilkens, E. Hoffman, A. Patel,M.S. Cosgrove, X.J. Chen, Mgm101 is a Rad52-related protein required formitochondrial DNA recombination, J. Biol. Chem. 286 (2011) 42360e42370.

[101] J.M. Martinez-Zapater, P. Gil, J. Capel, C.R. Somerville, Mutations at theArabidopsis CHM locus promote rearrangements of the mitochondrialgenome, Plant Cell 4 (1992) 889e899.

[102] W. Sakamoto, H. Kondo, M. Murata, F. Motoyoshi, Altered mitochondrialgene expression in a maternal distorted leaf mutant of Arabidopsis inducedby chloroplast mutator, Plant Cell 8 (1996) 1377e1390.

[103] M.P. Arrieta-Montiel, V. Shedge, J. Davila, A.C. Christensen, S.A. Mackenzie,Diversity of the Arabidopsis mitochondrial genome occurs via nuclear-controlled recombination activity, Genetics 183 (2009) 1261e1268.

[104] A.P. Sandhu, R.V. Abdelnoor, S.A. Mackenzie, Transgenic induction of mito-chondrial rearrangements for cytoplasmic male sterility in crop plants, Proc.Natl. Acad. Sci. U. S. A. 104 (2007) 1766e1770.

[105] R.V. Abdelnoor, A.C. Christensen, S. Mohammed, B. Munoz-Castillo,H. Moriyama, S.A. Mackenzie, Mitochondrial genome dynamics in plants andanimals: convergent gene fusions of a MutS homologue, J. Mol. Evol. 63(2006) 165e173.

[106] Y.Z. Xu, M.P. Arrieta-Montiel, K.S. Virdi, W.B. de Paula, J.R. Widhalm,G.J. Basset, J.I. Davila, T.E. Elthon, C.G. Elowsky, S.J. Sato, T.E. Clemente,S.A. Mackenzie, MutS HOMOLOG1 is a nucleoid protein that alters mito-chondrial and plastid properties and plant response to high light, Plant Cell23 (2011) 3428e3441.

[107] M.J. Schofield, P. Hsieh, DNA mismatch repair: molecular mechanisms andbiological function, Annu. Rev. Microbiol. 57 (2003) 579e608.

[108] R.V. Abdelnoor, R. Yule, A. Elo, A.C. Christensen, G. Meyer-Gauen,S.A. Mackenzie, Substoichiometric shifting in the plant mitochondrialgenome is influenced by a gene homologous to MutS, Proc. Natl. Acad. Sci. U.S. A. 100 (2003) 5968e5973.

[109] Z. Lin, H. Kong, M. Nei, H. Ma, Origins and evolution of the recA/RAD51 genefamily: evidence for ancient gene duplication and endosymbiotic genetransfer, Proc. Natl. Acad. Sci. U. S. A. 103 (2006) 10328e10333.

[110] S.T. Lovett, R.L. Hurley, V.A. Sutera Jr., R.H. Aubuchon, M.A. Lebedeva,Crossing over between regions of limited homology in Escherichia coli. RecA-dependent and RecA-independent pathways, Genetics 160 (2002) 851e859.

[111] A.J. Alverson, S. Zhuo, D.W. Rice, D.B. Sloan, J.D. Palmer, The mitochondrialgenome of the legume Vigna radiata and the analysis of recombinationacross short mitochondrial repeats, PLoS One 6 (2011) e16404.

[112] H. Cerutti, A.M. Johnson, J.E. Boynton, N.W. Gillham, Inhibition of chloroplastDNA recombination and repair by dominant negative mutants of Escherichiacoli RecA, Mol. Cell. Biol. 15 (1995) 3003e3011.

[113] T. Kwon, E. Huq, D.L. Herrin, Microhomology-mediated and nonhomologousrepair of a double-strand break in the chloroplast genome of Arabidopsis,Proc. Natl. Acad. Sci. U. S. A. 107 (2010) 13954e13959.

[114] B.A. Rowan, D.J. Oldenburg, A.J. Bendich, RecA maintains the integrity ofchloroplast DNA molecules in Arabidopsis, J. Exp. Bot. 61 (2010) 2575e2588.

[115] M. Odahara, T. Inouye, T. Fujita, M. Hasebe, Y. Sekine, Involvement ofmitochondrial-targeted RecA in the repair of mitochondrial DNA in the moss,Physcomitrella patens, Genes Genet. Syst. 82 (2007) 43e51.

[116] J.S. Sung, B. Demple, Roles of base excision repair subpathways in correctingoxidized abasic sites in DNA, FEBS J. 273 (2006) 1620e1629.

[117] P. Boesch, N. Ibrahim, A. Dietrich, R.N. Lightowlers, Membrane association ofmitochondrial DNA facilitates base excision repair in mammalian mito-chondria, Nucleic Acids Res. 38 (2010) 1478e1488.

[118] B.L. Gutman, K.K. Niyogi, Evidence for base excision repair of oxidative DNAdamage in chloroplasts of Arabidopsis thaliana, J. Biol. Chem. 284 (2009)17006e17012.

[119] R.J. Bensen, H.R. Warner, The partial purification and characterization ofnuclear and mitochondrial uracil-DNA glycosylase activities from Zea maysseedlings, Plant Physiol. 83 (1987) 149e154.

[120] T. Morales-Ruiz, M. Birincioglu, P. Jaruga, H. Rodriguez, T. Roldan-Arjona,M. Dizdaroglu, Arabidopsis thaliana Ogg1 protein excises 8-hydroxyguanineand 2,6-diamino-4-hydroxy-5-formamidopyrimidine from oxidativelydamagedDNAcontainingmultiple lesions,Biochemistry42 (2003)3089e3095.

[121] H. Chen, P. Chu, Y. Zhou, Y. Li, J. Liu, Y. Ding, E.W. Tsang, L. Jiang, K. Wu,S. Huang, Overexpression of AtOGG1, a DNA glycosylase/AP lyase, enhancesseed longevity and abiotic stress tolerance in Arabidopsis, J. Exp. Bot. 63(2012) 4107e4121.

[122] A. Macovei, A. Balestrazzi, M. Confalonieri, M. Fae, D. Carbonera, New in-sights on the barrel medic MtOGG1 and MtFPG functions in relation tooxidative stress response in planta and during seed imbibition, Plant Physiol.Biochem. 49 (2011) 1040e1050.

[123] T. Ohtsubo, O. Matsuda, K. Iba, I. Terashima, M. Sekiguchi, Y. Nakabeppu,Molecular cloning of AtMMH, an Arabidopsis thaliana ortholog of theEscherichia colimutM gene, and analysis of functional domains of its product,Mol. Gen. Genet. 259 (1998) 577e590.

[124] T.M. Murphy, M.J. Gao, Multiple forms of formamidopyrimidine-DNA gly-cosylase produced by alternative splicing in Arabidopsis thaliana,J. Photochem. Photobiol. B 61 (2001) 87e93.

[125] L.M. Griffiths, D. Swartzlander, K.L. Meadows, K.D. Wilkinson, A.H. Corbett,P.W. Doetsch, Dynamic compartmentalization of base excision repair pro-teins in response to nuclear and mitochondrial oxidative stress, Mol. Cell.Biol. 29 (2009) 794e807.

[126] D.B. Swartzlander, L.M. Griffiths, J. Lee, N.P. Degtyareva, P.W. Doetsch,A.H. Corbett, Regulation of base excision repair: Ntg1 nuclear and mito-chondrial dynamic localization in response to genotoxic stress, Nucleic AcidsRes. 38 (2010) 3963e3974.

[127] T. Kleffmann, D. Russenberger, A. von Zychlinski, W. Christopher,K. Sjolander, W. Gruissem, S. Baginsky, The Arabidopsis thaliana chloroplastproteome reveals pathway abundance and novel protein functions, Curr.Biol. 14 (2004) 354e362.

[128] Y. Mori, S. Kimura, A. Saotome, N. Kasai, N. Sakaguchi, Y. Uchiyama,T. Ishibashi, T. Yamamoto, H. Chiku, K. Sakaguchi, Plastid DNA polymerasesfrom higher plants, Arabidopsis thaliana, Biochem. Biophys. Res. Commun.334 (2005) 43e50.

[129] D. Cordoba-Canero, T. Roldan-Arjona, R.R. Ariza, Arabidopsis ARP endonu-clease functions in a branched base excision DNA repair pathway completedby LIG1, Plant J. 68 (2011) 693e702.

[130] P.A. Sunderland, C.E. West, W.M. Waterworth, C.M. Bray, An evolutionarilyconserved translation initiation mechanism regulates nuclear or mitochon-drial targeting of DNA ligase 1 in Arabidopsis thaliana, Plant J. 47 (2006) 356e367.

[131] I. Small, R. Suffolk, C.J. Leaver, Evolution of plant mitochondrial genomes viasubstoichiometric intermediates, Cell 58 (1989) 69e76.

[132] O.A. Kajander, A.T. Rovio, K. Majamaa, J. Poulton, J.N. Spelbrink, I.J. Holt,P.J. Karhunen, H.T. Jacobs, Human mtDNA sublimons resemble rearrangedmitochondrial genomes found in pathological states, Hum. Mol. Genet. 9(2000) 2821e2835.

[133] F. Budar, P. Touzet, R. De Paepe, The nucleo-mitochondrial conflict in cyto-plasmic male sterilities revisited, Genetica 117 (2003) 3e16.

[134] M. Odahara, H. Kuroiwa, T. Kuroiwa, Y. Sekine, Suppression of repeat-mediated gross mitochondrial genome rearrangements by RecA in themoss Physcomitrella patens, Plant Cell 21 (2009) 1182e1194.

[135] M. Woloszynska, D. Trojanowski, Counting mtDNA molecules in Phaseolusvulgaris: sublimons are constantly produced by recombination via shortrepeats and undergo rigorous selection during substoichiometric shifting,Plant Mol. Biol. 70 (2009) 511e521.

J.M. Gualberto et al. / Biochimie 100 (2014) 107e120120

[136] D.Y. Wang, Q. Zhang, Y. Liu, Z.F. Lin, S.X. Zhang, M.X. Sun, Sodmergen, Thelevels of male gametic mitochondrial DNA are highly regulated in angio-sperms with regard to mitochondrial inheritance, Plant Cell 22 (2010) 2402e2416.

[137] M. Lynch, B. Koskella, S. Schaack, Mutation pressure and the evolution oforganelle genomic architecture, Science 311 (2006) 1727e1730.

[138] A.C. Christensen, Plant mitochondrial genome evolution can be explained byDNA repair mechanisms, Genome Biol. Evol. 5 (2013) 1079e1086.

[139] C.L. Parkinson, J.P. Mower, Y.L. Qiu, A.J. Shirk, K. Song, N.D. Young,C.W. dePamphilis, J.D. Palmer, Multiple major increases and decreases inmitochondrial substitution rates in the plant family Geraniaceae, BMC Evol.Biol. 5 (2005) 73.

[140] J.P. Mower, P. Touzet, J.S. Gummow, L.F. Delph, J.D. Palmer, Extensive vari-ation in synonymous substitution rates in mitochondrial genes of seedplants, BMC Evol. Biol. 7 (2007) 135.

[141] M. Koulintchenko, Y. Konstantinov, A. Dietrich, Plant mitochondria activelyimport DNA via the permeability transition pore complex, EMBO J. 22 (2003)1245e1254.

[142] F. Weber-Lotfi, N. Ibrahim, P. Boesch, A. Cosset, Y. Konstantinov,R.N. Lightowlers, A. Dietrich, Developing a genetic approach to investigatethe mechanism of mitochondrial competence for DNA import, Biochim.Biophys. Acta 1787 (2009) 320e327.

[143] M. Koulintchenko, R.J. Temperley, P.A. Mason, A. Dietrich, R.N. Lightowlers,Natural competence of mammalian mitochondria allows the molecularinvestigation of mitochondrial gene expression, Hum. Mol. Genet. 15 (2006)143e154.

[144] Y. Ono, A. Sakai, K. Takechi, S. Takio, M. Takusagawa, H. Takano, NtPolI-like1and NtPolI-like2, bacterial DNA polymerase I homologs isolated from BY-2cultured tobacco cells, encode DNA polymerases engaged in DNA replica-tion in both plastids and mitochondria, Plant Cell Physiol. 48 (2007) 1679e1692.

[145] T.E. Shutt, M.W. Gray, Twinkle, the mitochondrial replicative DNA helicase, iswidespread in the eukaryotic radiation and may also be the mitochondrialDNA primase in most eukaryotes, J. Mol. Evol. 62 (2006) 588e599.

[146] K. Krause, I. Kilbienski, M. Mulisch, A. Rodiger, A. Schafer, K. Krupinska, DNA-binding proteins of the Whirly family in Arabidopsis thaliana are targeted tothe organelles, FEBS Lett. 579 (2005) 3707e3712.

[147] M. Vermel, B. Guermann, L. Delage, J.M. Grienenberger, L. Maréchal-Drouard,J.M. Gualberto, A family of RRM-type RNA-binding proteins specific to plantmitochondria, Proc. Natl. Acad. Sci. U. S. A. 99 (2002) 5866e5871.

[148] D. Cordoba-Canero, E. Dubois, R.R. Ariza, M.P. Doutriaux, T. Roldan-Arjona,Arabidopsis uracil DNA glycosylase (UNG) is required for base excision repairof uracil and increases plant sensitivity to 5-fluorouracil, J. Biol. Chem. 285(2010) 7475e7483.