Copyright � 2009 by the Genetics Society of AmericaDOI: 10.1534/genetics.109.108514

Diversity of the Arabidopsis Mitochondrial Genome Occurs viaNuclear-Controlled Recombination Activity

Maria P. Arrieta-Montiel,* Vikas Shedge,* Jaime Davila,* Alan C. Christensen† andSally A. Mackenzie*,†,1

*Center for Plant Science Innovation, †School of Biological Sciences, University of Nebraska, Lincoln, Nebraska 68588

Manuscript received August 11, 2009Accepted for publication October 1, 2009

ABSTRACT

The plant mitochondrial genome is recombinogenic, with DNA exchange activity controlled to a largeextent by nuclear gene products. One nuclear gene, MSH1, appears to participate in suppressingrecombination in Arabidopsis at every repeated sequence ranging in size from 108 to 556 bp. Present in awide range of plant species, these mitochondrial repeats display evidence of successful asymmetric DNAexchange in Arabidopsis when MSH1 is disrupted. Recombination frequency appears to be influenced byrepeat sequence homology and size, with larger size repeats corresponding to increased DNA exchangeactivity. The extensive mitochondrial genomic reorganization of the msh1 mutant produced alteredmitochondrial transcription patterns. Comparison of mitochondrial genomes from the Arabidopsisecotypes C24, Col-0, and Ler suggests that MSH1 activity accounts for most or all of the polymorphismsdistinguishing these genomes, producing ecotype-specific stoichiometric changes in each line. Our ob-servations suggest that MSH1 participates in mitochondrial genome evolution by influencing the lineage-specific pattern of mitochondrial genetic variation in higher plants.

THE plant mitochondrial genome is characterizedby several unusual features in its organization and

structure. These features include the presence of se-quence chimeras (Schnable and Wise 1998), foreignDNA insertions (Xiong et al. 2008), and unusually highlevels of recombination (Mackenzie 2007). Recombi-nation in the mitochondrial genome can involve large-size (.1 kb) repeats that undergo high-frequencyreciprocal DNA exchange to subdivide the genome, aswell as intermediate-size repeats that mediate low-frequency, asymmetric DNA exchange to produce onlyone of the predicted recombination products. This low-frequency recombination activity is associated withrapid stoichiometric changes in genome configuration(Shedge et al. 2007), referred to as substoichiometricshifting (Small et al. 1987).

Three nuclear genes have been shown to influencerecombination within the plant mitochondrial genome:MSH1 (Abdelnoor et al. 2003), RECA3 (Shedge et al.2007), and OSB1 (Zaegel et al. 2006). Of these, MSH1appears to have the most profound and immediateeffect on the genome and on plant phenotype (Sandhu

et al. 2007; Shedge et al. 2007). To better understand theinfluence of MSH1-regulated recombination on ge-nome structure, we investigated mitochondrial sites in

Arabidopsis at which MSH1 appears to function. Weshow that at least 33 sites within the genome simulta-neously become active with MSH1 disruption. Repeat-mediated DNA exchange activity results in extensivechanges in mitochondrial genome organization andgene expression patterns by permitting differentialmodulation of transcription. The pattern of genomicreorganization is highly reproducible, and Arabidopsiscross-ecotype comparisons suggest that these rearrange-ment processes participate directly in higher plantmitochondrial genome evolution.

MATERIALS AND METHODS

Arabidopsis thaliana growth and mutants: Arabidopsis plantswere grown by cold treating (4�) and then sowing seedsdirectly in potting mix (Metro Mix 360). Plants were grownat an 8-hr-day length at 24� for 8 weeks and then transferred toa 16-hr-day length. Two MSH1 mutants were used for thestudy: msh1-1(Abdelnoor et al. 2003) and Salk_041951. TheRECA3 mutant line Sail_252_C06 was also used for geneticanalyses (The Arabidopsis Information Resource, http://www.Arabidopsis.org).

DNA gel blot and PCR assays: Total genomic DNA wasextracted from above-ground tissues of flowering plants usingthe DNeasy plant mini kit (Qiagen). Total RNA was isolatedusing TRIzol reagent (Invitrogen) according to the manufac-turer’s instructions and purified using the RNeasy mini kit(Qiagen). DNA gel blot and hybridizations were as describedpreviously (Janska and Mackenzie 1993). Primers used toPCR amplify repeats are listed in the supporting information(Table S1).

Supporting information is available online at http://www.genetics.org/cgi/content/full/genetics.109.108514/DC1.

1Corresponding author: N305 Beadle Center, University of Nebraska,Lincoln, NE 68588-0660. E-mail: smackenzie2@unl.edu

Genetics 183: 1261–1268 (December 2009)

Quantitative PCR: Equal amounts of DNA were used forquantitative PCR using the SYBR GreenER kit for iCycler(Invitrogen). Quantitative PCR data collection and analysiswere conducted using iCycler iQ software (version 3.1; Bio-Rad). Experiments were repeated, each sample was run intriplicate, and the results were averaged. Primers used for real-time analysis of regions present in molecules B and D andmolecules A and C, respectively, were RealBDF (59-ATTCCATCCACTCCGGCTTAGCTT-39) and RealBDR (59-TCGCTGTGAAAGG TGGAATCCGTT-39) and RealACF (59-ATGTAGAGCCAACTGGAGAGCA-39) and RealACR (59-CGGAAAGCCCAAATTCTCCTGCAT-39).

Bioinformatics analyses: BLAST and blast2seq (http://www.ncbi.nlm.nih.gov/blast/bl2seq/wblast2.cgi) were usedto identify repeated sequences within the C24 mitochondrialgenome with a word size of 50 nucleotides. This processidentified 104 different repeats, repeats A–Z and AA–ZZ, indescending order of BLAST score. Each repeat copy wasnumbered (e.g., repeat A-1 and repeat A-2) and distinguishedby the flanking sequences. The 33 largest of the intermediateclass of repeats, from 108 to 556 bp, are indicated in Figure 1Aand Table S2. Most repeats were present in two copies only,with the exceptions of H, V, FF, NN, OO, and RR, present inthree copies and BB present in four copies.

To identify repeats in the mitochondrial sequences ofsorghum, tobacco, and maize, the software REPuter (Kurtz

et al. 2001) was used and its results were processed with a Perlscript designed to filter close appearances of the repeats. Tomap ecotype mitochondrial genomes, a script written in Perlgenerated a network whose nodes are the regions between therepeats and whose arcs indicate evidence of linkages on thesame molecule using information gathered from DNA gel blotanalysis from the repeated regions for each ecotype. A circuitin this network corresponds to a map of the mitochondrialgenome, and circular maps of these circuits were generatedfor each ecotype. Arabidopsis Genome Initiative locus identi-fiers are MSH1 (At3g24320) and RECA3 (At3g10140).

RESULTS

Mitochondrial recombination is prolific in theabsence of MSH1: In Arabidopsis msh1 mutants, suc-cessful DNA exchange activity increased markedly atrepeats ranging in size from 108 bp to 556 bp (Figure1A, Table S2) relative to wild- type Columbia-0 (Col-0).There exist 33 near-perfect (.90% sequence identity)

Figure 1.—Sites of re-combination in theArabidopsis mitochondrialgenome. (A) Computer-generated map of the 33identified small mitochon-drial repeats active in Ara-bidopsis msh1. Coloredrepeats represent thosetested by gel blot hybrid-ization, with correspond-ing colors designatinginteracting repeat pairs.Numbers in parenthesesindicate lengths of the re-peats (bp). (B) DNA gelblots (BamHI) displayingevidence of ectopic recom-bination at repeats D andF. A and B designate paren-tal forms; R designates therecombinant product. Re-peats D and F are proximalto genes CoxII and Atp8.Arrows indicate secondaryrecombination products.(C) Recombination is notevident at an imperfect re-peat of 342 bp of 80% se-quence homology. A andB designate the two re-peats (BamHI), with ar-rows predicting the sizesof recombinant products.Probes used are the re-peats.

1262 M. P. Arrieta-Montiel et al.

repeats within this size range in Arabidopsis, named inalphabetical order by BLAST score, and 23 wereconfirmed to be active in msh1 mutants (Figure 1B;Figure S1). The remaining 10 are assumed to be active,but definitive confirmation was precluded either bytheir proximity to other active repeats or by theirrelative product sizes. No repeats are detected in thesize range between 556 bp, the upper limit for in-termediate repeats, and 4.3 kbp, the lower limit of high-frequency large repeats in the genome.

A selected imperfect (80% sequence identity) repeat(IR-1/IR-2) within the intermediate-size range (342 bp)did not show evidence of recombination (Figure 1C,Table S2), suggesting that a near-perfect sequencehomology of a given size range may be essential forsuccessful DNA exchange. This observation appearsconsistent with what has been reported in yeast mito-chondria (Phadnis et al. 2005). Similarly, two repeatsbelow the indicated size range tested negative forrecombination activity (Table S2). As reported pre-viously (Shedge et al. 2007), msh1-regulated recombi-nation at each repeat was asymmetric, with only oneproduct retained in the genome. Similar intermediate-size repeated sequences could be identified in themitochondrial genomes of other plant species, al-though size range varied slightly, with no evidence of

sequence conservation among repeats from differentspecies (Table 1). We postulate that these repeats areactivated with the disruption of MSH1 as well (Sandhu

et al. 2007).The extent of the observed mitochondrial rearrange-

ment in the msh1 mutant (formerly designated chm1) issurprising in light of previous reports suggesting alimited, localized point of mitochondrial rearrange-ment (Sakamoto et al. 1996; Shedge et al. 2007). Theseeming disparity in results derives from differingexperimental approaches. Evidence of recombinationis considerably more difficult to detect when probingwith large, cosmid-size genomic segments than withprobes designed to specifically target the repeats due tothe complex hybridization patterns from multiple frag-ments and because most recombinant products arepresent in low stoichiometry in the first generationfollowing MSH1 disruption.

Behavior of msh1 mutants: Comparison can be madeof ‘‘early’’ vs. ‘‘advanced’’ msh1 mutants by comparinglines derived from Col-0 3 msh1 for F2 msh1/msh1progeny (early) with the original chm1-1 mutant re-ported by Redei (1973) and maintained by recurrentself-pollination (advanced). Differences exist betweenthese two genomes in relative DNA stoichiometries(Figure 2), suggesting that mitochondrial rearrange-

TABLE 1

Sample plant species containing intermediate-sizemitochondrial repeats

SpeciesNo. of repeats of

100–200 bpNo. of repeats

.200 bpMaximum-size

repeat

Zea mays 26 6 830Sorghum bicolor 15 5 303Solanum tabacum 25 1 405

Figure 2.—Early (first-generation) andadvanced-generation msh1 mutants differ in mito-chondrial genome configuration. Evaluationswere made with three different repeats—B, D,and F—to demonstrate recombination (A andB, parental molecules; C, recombinant mole-cules). Changes from early to advanced genera-tion include altered stoichiometries, novelpolymorphisms, and loss or reduction of parentalforms. Of 20 repeats studied in more detail, 6showed no further changes in the advanced mu-tant. For the remaining 12, later-occurring differ-ences usually involved the loss of one of theparental forms. Data shown are from DNA gelblot hybridizations with repeats as probes.

Plant Mitochondrial Genome Recombination 1263

ment continues at some level indefinitely in the absenceof MSH1. Analysis of early generation msh1 mutantsindicates that all repeats are active in recombinationsimultaneously, with differences in rate appearing tocorrespond to repeat size (Figure 3A). The seeminglycontinuous recombination activity may be a conse-quence of recombination products serving as substratefor additional DNA exchange activity (Figure 3B), butthe process does not appear to be stochastic on the basisof comparisons of the advanced msh1 mutants. Plant-to-plant comparisons of advanced msh1 mutants show verysimilar mitochondrial genome configuration, with mi-nor differences accounted for largely by secondaryrecombinations. However, in association with activerecombination, a process of cytoplasmic sorting occurs,giving rise to mitochondrial genome variation amongindividual F3 progeny (Col-0 3 msh1) (Figure 3C).

The msh1 recA3 double mutant (Shedge et al. 2007)undergoes more extensive rearrangement of the mito-chondrial genome than is observed in the advancedmsh1 (chm1-1) mutants. The nature of recombination isaltered in the double mutant, so that reciprocal re-combination products accumulate for some of therepeats (Figure 4). The extensive degree of genomicrearrangement that occurs in msh1 and the msh1 recA3double mutant produces alterations in mitochondrialgene expression. RNA gel blot analysis showed thatseveral transcripts were altered in either size or abun-dance as a consequence of recombination (Figure 5).However, transcript changes were not observed for allgenes. For example, three additional mitochondrialgenes tested, Atp9, Atp6, and orf 111b, showed nodetectable change in transcript level or size (data notshown).

Figure 3.—Features of msh1-regulated recombination. (A) A gradient of recombination based on repeat size may exist. Forexample, molecules A and B recombine at repeat G (335 bp) to give product C, and molecules B and D (which do not hybridizewith this probe) recombine at repeat L (249 bp) to give product E. The stoichiometry of recombinant product C is consistentlyhigher than recombinant molecule E. Parental molecule A shifts to substoichiometric levels in the msh1 mutant. Two early gen-eration (F3) progeny show that there is initial plant-to-plant variability for the rate at which A is reduced in concentration. F3 plantswere derived from different F2 progeny. (B) In the msh1 mutant, recombination products can serve as substrate for subsequentrecombination events. Recombinant molecule C is substrate for recombination at repeat V, giving rise to recombinant molecule E.Molecule D does not hybridize to the probe used in this experiment. (C) Genetic variation is observed among F3 progeny fol-lowing recombination in Col-0 3 msh1 populations. This variation is presumably the consequence of cytoplasmic sorting and oftenincludes loss of one of the parental A or B forms (example indicated by arrow; also see panel A). All results shown are from DNAgel blot hybridization experiments.

1264 M. P. Arrieta-Montiel et al.

Mitochondrial genome comparisons in Arabidopsis:Arabidopsis C24, Col-0, and Ler mitochondrial genomesare readily distinguished by numerous DNA polymor-phisms, some of which have been characterized indetail. Previous studies by another group showed thatat least one novel DNA insertion exists in Col-0 but isabsent from C24 (Forner et al. 2005). This insertion wassuggested to involve a 9-bp repeat-mediated integration.The sequence in question represents a recombination-ally active region (Zaegel et al. 2006; Shedge et al.2007). Similarly, mitochondrial genomic environmentspresent in Col-0 have seemingly disappeared from theLer genome (Figure 6). Remarkably, introduction of themsh1 mutation to C24, Col-0, and Ler ecotypes producedgenomic changes such that each of the three mitochon-drial genomes show evidence of recombination in allregions assayed (Figure 6). Following this observation,we employed PCR-based methods to detect an ex-tremely low-level presence of regions undetectable bygel blot hybridization in the wild type (Table 2). In somecases, the presence of regions could be predicted on thebasis of recombination patterns in the msh1 mutant, butin such a small number of cells as to go completelyundetected.

These results indicate that several features distin-guishing the three mitochondrial genomes are largelythe consequence of differential substoichiometric shift-ing across ecotypes under the influence of MSH1.The region reported by Forner et al. (2005) to be pre-sent in Col-0 but absent from C24 is present at near-undetectable levels in C24, with its recombinationactivity evident only in the C24 msh1 mutant (Table 2).

Similarly, genomic regions present in Col-0 but absentin Ler were detectable within the Ler msh1 mutant(Figure 6). These observations suggest that much of themitochondrial variation reported in Arabidopsis ecotypesis the consequence of MSH1-regulated recombination.

While the recombinationally active regions of theArabidopsis mitochondrial genome were dispersedthroughout the genome, some repeats tended to cluster(Figure 1A and Figure 7A). Several regions show repeatsorganized densely, even overlapping. In fact, the re-combinational region suggested by Forner et al. (2005)to be absent from C24 actually encompasses fourdistinct and active repeats within a 1.8-kb segment(Figure 7B). Given the observed association of recom-bination frequency with repeat size, these clusterregions are presumed to represent regions of highestDNA exchange activity within the genome.

Mitochondrial genome comparisons of C24, Col-0,Ler, and their msh1 counterparts allowed us to model theorganization of these genomes (Figure 8). From aprocess of cross-ecotype diagnostic gel blot hybridiza-tions with repeat probes, we postulate that the threegenomes, while differing dramatically in their organ-izations in wild-type lines, appear much more similar intheir msh1 counterparts. This similarity emerges fromsubstoichiometric forms present in all three ecotypes

Figure 4.—Reciprocal recombinants can accumulate inthe msh1 recA3 double mutant. Experiments using repeatsD, K, and I as probes show that parental forms A and B arepredominant in Col-0. In the msh1 mutant, only recombinantmolecule C accumulates. In the double mutant, the recipro-cal recombinant product D also accumulates.

Figure 5.—Mitochondrial transcription is influenced byDNA rearrangement activity. RNA gel blot analysis of totalRNA preparations probed with the mitochondrial coding se-quence of orf452, Atp8, and CoxII to demonstrate examples ofenhanced transcript levels (orf452) and novel transcriptsemerging in response to mitochondrial rearrangement(Atp8 and CoxII). Ubiquitin is shown as a loading control.

Plant Mitochondrial Genome Recombination 1265

that become amplified with the disruption of MSH1 andenhanced ectopic recombination.

DISCUSSION

Loss of MSH1 in Arabidopsis results in extensivechanges in mitochondrial genome configuration andaltered gene expression: The impact of MSH1 on DNAexchange activity within the Arabidopsis mitochondrialgenome is wide-ranging, influencing the activity of everyidentifiable repeat in the genome within a size range of�100–600 bp. The pervasive nature of mitochondrialrearrangement in the msh1 mutant suggests that thegene participates in defining the mitochondrial ge-nome organization that is inherited within a lineage.This influence by MSH1 was demonstrated in cross-ecotype comparisons within Arabidopsis. Loss of MSH1activity in C24, Col-0, and Ler leads to stoichiometricchanges that essentially obviate polymorphisms be-tween the genomes. From these results we postulatethat a significant proportion of the within-speciesmitochondrial DNA polymorphisms observed in plantsmay be the consequence of low-frequency recombina-tion and substoichiometric changes within genomesthat are otherwise quite similar or even identical insequence.

The influence of MSH1 in establishing genomicstoichiometric relationships also affected mitochon-drial transcription. Transcriptional modulation occursas the apparent consequence of altered gene copynumber or local environment during genomic shifting.In fact, cross-ecotype variation in mitochondrial tran-scription has already been shown genetically to be theconsequence, at least in part, of mitochondrial genomeconfiguration (Forner et al. 2008). Thus, MSH1 ap-pears to play a significant role in plant mitochondrialgenome diversity by controlling low-frequency, asym-metric recombination and, hence, genetic variationthat is observed within plant mitochondrial genomes(Small et al. 1989).

A model for MSH1 action: Results presented hereand in a previous study (Shedge et al. 2007) suggest thatMSH1 influences the fate of double-strand breaks andstrand invasion within near-perfect repeats of a givensize range. MSH1 apparently limits successful DNAexchange, so that recombination at intermediate re-peats is extremely low but detectable in wild type, andunsuccessful events presumably result in gene conver-sion events that would account for the continuedmaintenance of sequence identity in the repeats.Disruption of MSH1 permits high-frequency DNAexchange and massive reorganization of the genome.Remarkably, this reorganization in C24, Col-0, and Lerleads to very similar genome organizations in the threeecotypes, perhaps eventually producing a ‘‘default’’organization with final DNA configurations approach-ing more uniform levels. The process by which thesethree mitochondrial genomes assume distinct, ecotype-specific, heritable genome configurations with theparticipation of MSH1 is a detail for future study.

The observations that we report in Arabidopsis arestrikingly similar to previous studies of the commonbean (Phaseolus vulgaris L), where the accession lineG08063 was shown to derive directly from the maternalparent line POP, although their mitochondrial ge-nomes appear remarkably different in configuration

Figure 6.—Substoichiometric shifting accounts for DNA polymorphisms distinguishing Arabidopsis ecotypes. Recombinationdata are presented for two different repeats, B and L. In both panels, molecules A and B are predominant forms in Col-0. MoleculeC is the product of A/B recombination and accumulates in the Col-0 msh1 mutant. Reciprocal recombinant D, while not visible inany ecotype in either panel, is present in very low stoichiometry and serves as substrate for recombination with molecule C in theLer msh1 mutant to produce A and B in both panels (dashed line indicates the expected size of molecule D). Arrowheads indicatethe ‘‘environments’’ not present in the parental ecotypes and emerging after disruption of the MSH1 locus.

TABLE 2

Evidence for substoichiometric repeat forms inArabidopsis ecotypes

Ecotype Molecules B/D Molecules A/C

Col-0 12.3279088 99.76922C-24 0.01085013 100Ler 0.01246353 82.93195

Data from real-time quantitative PCR experiments show es-timates of copy number for a unique (nonrepeat) segmentpresent in molecules B and D (see Figure 6) relative to anAtp9 segment in molecules A and C.

1266 M. P. Arrieta-Montiel et al.

(Janska et al. 1998). In bean, as in Arabidopsis, thesemitochondrial genomes were apparently interconvert-ible via substoichiometric shifting processes, as deducedby more sensitive mapping methods.

For such genomic interconversions to occur, onemust assume that successful DNA exchange involvesone parental molecule that is substoichiometric andpresent only in some small fraction of the plant’s cells.In Arabidopsis, recombination occurs at a repeat pre-sent at nearly undetectable levels in C24 (Forner et al.2005). Similar substoichiometric recombination hasbeen described in the common bean (Woloszynska

and Trojanowski 2009) and pearl millet (Feng et al.2009) mitochondria. This can be explained by hypoth-esizing that these events occur only in tissue(s) wherethe repeats are present in normal copy number. We havespeculated that this is the meristem (Arrieta-Montiel

et al. 2001). Whether loss of MSH1 can produce amitochondrial genome configuration in the vegetativetissues of a plant that resembles that of the meristemcells, the elusive ‘‘master’’ chromosome (Lonsdale et al.1988), is an interesting possibility that merits furthertesting.

MSH1-associated processes likely account for othermitochondrial rearrangement activity observed inplants: Massive mitochondrial genome rearrangementin plants has been reported in the past. Many of thesereports involved cell suspension cultures (Kemble et al.1982; Kemble and Shepard 1984; Schmidt et al. 1996).Suspension-culture-associated mitochondrial rearrange-ments have also been associated with repeat-mediatedrecombination (Belliard et al. 1979) and likely substoi-chiometric shifting (Ozias-Akins et al. 1988; Forner et al.2005). These results, while not conclusive, would be

Figure 8.—Model representing themitochondrial genomes of Arabidopsisecotypes based on diagnostic gel blot hy-bridization experiments including inter-mediate repeat regions and an assemblyscript. The predominant form is repre-sented as the larger forms, with smallercircles representing the more abundantof many substoichiometric forms. Differ-ent orientations for the same regions arerepresented by a different texture. Thelarge repeat I is represented by red arrowsand the large repeat II is represented byblack arrows. A fragment absent in theC-24 published sequence, but present inthe predominant form of Col-0 and as a

subfragment in Ler, is denoted by a green triangle. The dominant form for Ler has lost one copy of large repeat I and four ofthe regions present in C-24. For simplicity of the depicted model, not all possible substoichiometric configurations are shown.Similarly, the various configurations arising from recombination between large repeats are omitted.

Figure 7.—Intermediaterepeats within the Arabi-dopsis mitochondrial ge-nome show evidence ofclustering. (A) Position ofthe repeat clusters on themitochondrial genomemap. The y-axis representsthe map position with acluster represented by asymbol. Repeats presentwithin each cluster are des-ignated. (B) Four activerepeats exist within an�1.8-kb segment that differ-entiates ecotype C24 fromCol-0 (Forner et al. 2005).The schematic depictsdense organization of re-peats E, H, L, and K. In thiscase, several partial (p)gene segments exist withinthe region that is locatedupstream to an intact copyof CoxIII.

Plant Mitochondrial Genome Recombination 1267

consistent with a model for enhanced mitochondrialrecombination under conditions where MSH1 might bereduced in its expression or activity. There are probablyother environmental conditions under which modulationof MSH1 could influence mitochondrial genome organi-zation within a plant. Perhaps such conditions permit low-frequency induction of cytoplasmic male sterility (Sandhu

et al. 2007) in natural plant populations or, perhaps, thespontaneous reversion to pollen fertility ( Janska et al.1998)? Interestingly, a report made several years ago onmaize following crosses to the mutant iojap suggests thatthese crossing experiments resulted in the emergence of acytoplasmic male sterile cytoplasm type from a normalfertile cytoplasm (Lemke et al. 1985). These reported obser-vations possibly could be accounted for by a substoichio-metric shifting process identical to that reported here.

We thank Hardik Kundariya for excellent technical assistance. Thiswork was funded by a grant from the National Science Foundation(NSF) (MCB 0744104) to S.A.M. and M.P.A.-M. This material wasbased on work supported by the National Science Foundation whileA.C.C. was working at the NSF. Any opinion, finding, conclusions, orrecommendations expressed in this material are those of the authorsand do not necessarily reflect the views of the NSF.

LITERATURE CITED

Abdelnoor, R. V., R. Yule, A. Elo, A. C. Christensen, G. Meyer-Gauen et al., 2003 Substoichiometric shifting in the plant mito-chondrial genome is influenced by a gene homologous to MutS.Proc. Natl. Acad. Sci. USA 100: 5968–5973.

Arrieta-Montiel, M., A. Lyznik, M. Woloszynska, H. Janska, J.Tohme et al., 2001 Tracing evolutionary and developmentalimplications of mitochondrial stoichiometric shifting in thecommon bean. Genetics 158: 851–864.

Belliard, G., F. Vedel and G. Pelletier, 1979 Mitochondrial re-combination in cytoplasmic hybrids of Nicotiana tabacum by pro-toplast fusion. Nature 281: 401–403.

Feng, X., A. P. Kaur, S. A. Mackenzie and I. M. Dweikat,2009 Substoichiometric shifting in the fertility reversion of cy-toplasmic male sterile pearl millet. Theor. Appl. Genet. 118:1361–1370.

Forner, J., B. Weber, C. Wietholter, R. C. Meyer and S. Binder,2005 Distant sequences determine 59 end formation of cox3transcripts in Arabidopsis thaliana ecotype C24. Nucleic AcidsRes. 33: 4673–4682.

Forner, J., A. Holzle, C. Jonietz, S. Thuss, M. Schwarzlander

et al., 2008 Mitochondrial mRNA polymorphisms in differentArabidopsis accessions. Plant Physiol. 148: 1106–1116.

Janska, H., and S. Mackenzie, 1993 Unusual mitochondrial ge-nome organization in cytoplasmic male sterile common beanand the nature of cytoplasmic reversion to fertility. Genetics135: 869–879.

Janska, H., R. Sarria, M. Woloszynska, M. Arrieta-Montiel andS. A. Mackenzie, 1998 Stoichiometric shifts in the commonbean mitochondrial genome leading to male sterility and spon-taneous reversion to fertility. Plant Cell 10: 1163–1180.

Kemble, R. J., and J. F. Shepard, 1984 Cytoplasmic DNA variation in apotato protoclonal population. Theor. Appl. Genet. 69: 211–216.

Kemble, R. J., R. I. S. Brettell and R. B. Flavell, 1982 Mito-chondrial DNA analyses of fertile and sterile maize plants fromtissue culture with the Texas male sterile cytoplasm. Theor. Appl.Genet. 62: 213–217.

Kurtz, S., J. V. Choudhuri, E. Ohlebusch, C. Schleiermacher,J. Stoye et al., 2001 REPuter: the manifold applications of re-peat analysis on a genomic scale. Nucleic Acids Res. 29: 4633–4642.

Lemke, C. A., V. E. Gracen and H. L. Everett, 1985 A new sourceof cytoplasmic male sterility in maize induced by the nucleargene, iojap. Theor. Appl. Genet. 71: 481–485.

Lonsdale, D. M., T. Brears, T. P. Hodge, S. E. Melville and W. H.Rottmann, 1988 The plant mitochondrial genome: homo-logous recombination as a mechanism for generating hetero-geneity. Philos. Trans. R. Soc. Lond. B 319: 149–163.

Mackenzie, S. A., 2007 The unique biology of mitochondrial ge-nome instability in plants, pp. 36–46 in Plant Mitochondria, editedby D. Logan. Blackwell Publishing, Oxford.

Ozias-Akins, P., Z. Tabaeizadeh, D. R. Pring and I. K. Vasil,1988 Preferential amplification of mitochondrial DNA frag-ments in somatic hybrids of the Graminae. Curr. Genet. 13:241–245.

Phadnis, N., R. A. Sia and E. A. Sia, 2005 Analysis of repeat-mediated deletions in the mitochondrial genome of Saccharomycescerevisiae. Genetics 171: 1549–1559.

Redei, G. P., 1973 Extra-chromosomal mutability determined by anuclear gene locus in Arabidopsis. Mutat. Res. 18: 149–162.

Sakamoto, W., H. Kondo, M. Murata and F. Motoyoshi,1996 Altered mitochondrial gene expression in a maternal dis-torted leaf mutant of Arabidopsis induced by chloroplast muta-tor. Plant Cell 8: 1377–1390.

Sandhu, A. P., R. V. Abdelnoor and S. A. Mackenzie,2007 Transgenic induction of mitochondrial rearrangementsfor cytoplasmic male sterility in crop plants. Proc. Natl. Acad.Sci. USA 104: 1766–1770.

Schmidt, M., C. Walz and C.-U. Hesemann, 1996 Somaclonalvariation of the mitochondrial ATPase subunit 6 gene regionin regenerated triticale shoots and full-grown plants. Theor.Appl. Genet. 93: 355–360.

Schnable, P. S., and R. P. Wise, 1998 The molecular basis of cyto-plasmic male sterility. Trends Plant Sci. 3: 175–180.

Shedge, V., M. Arrieta-Montiel, A. C. Christensen and S. AMackenzie, 2007 Plant mitochondrial recombination surveil-lance requires unusual RecA and MutS homologs. Plant Cell19: 1251–1264.

Small, I. D., P. G. Isaac and C. J. Leaver, 1987 Stoichiometric dif-ferences in DNA molecules containing the atpA gene suggestmechanisms for the generation of mitochondrial genome diver-sity in maize. EMBO J. 6: 865–869.

Small, I., R. Suffolk and C. J. Leaver, 1989 Evolution of plant mi-tochondrial genomes via substoichiometric intermediates. Cell58: 69–76.

Woloszynska, M., and D. Trojanowski, 2009 Counting mtDNAmolecules in Phaseolus vulgaris: sublimons are constantly pro-duced by recombination via short repeats and undergo rigorousselection during substoichiometric shifting. Plant Mol. Biol. 70:511–521.

Xiong, A. S., R. H. Peng, J. Zhuang, F. Gao, B. Zhu et al.,2008 Gene duplication and transfer events in plant mitochon-drial genome. Biochem. Biophys. Res. Commun. 376: 1–4.

Zaegel, V., B. Guermann, M. Le Ret, C. Andres, D. Meyer, et al.,2006 The plant-specific ssDNA binding protein OSB1 is in-volved in the stoichiometric transmission of mitochondrialDNA in Arabidopsis. Plant Cell 18: 3548–3563.

Communicating editor: J. A. Birchler

1268 M. P. Arrieta-Montiel et al.

Supporting Information http://www.genetics.org/cgi/content/full/genetics.109.108514/DC1

Diversity of the Arabidopsis Mitochondrial Genome Occurs via Nuclear-Controlled Recombination Activity

Maria P. Arrieta-Montiel, Vikas Shedge, Jaime Davila, Alan C. Christensen and Sally A. Mackenzie

Copyright © 2009 by the Genetics Society of America DOI: 10.1534/genetics.109.108514

M. P. Arrieta-Montiel et al. 2 SI

FIGURE S1.—Evidence of extensive repeat-mediated mitochondrial DNA recombination in Arabidopsis. DNA gel blot

hybridization experiments with BamH1-digested total genomic DNA from wildtype Columbia-0 and the msh1 mutant, testing for evidence of recombination at 23 different mitochondrial repeats. Results from eleven repeats are shown, with repeat size indicated. Repeat identifiers are indicated by R (repeat) followed by the alphabetical letter designator shown in Table S2. A and B indicate the two parental forms of the repeats, with R indicating recombinant product. DNA exchange is asymmetric, so only one recombination product accumulates.

M. P. Arrieta-Montiel et al. 3 SI

TABLE S1

Primers used for PCR amplification of mitochondrial repeats in Arabidopsis

Repeat

Primer Sequence (5’-3’)

Repeat_B1-f1 TCAACTCAATTCTCTCGTTC

Repeat_B1-r1 ATCTGCCTCCTGCACTATAC

Repeat_D1-f1 AGTGATCTGTTCATCTAACTCA

Repeat_D1-r1 TACTACTACCTCGTCCATTG

Repeat_F1-f1 CACGAGGAATGGAAAGAAACAT

Repeat_F1-r1 GCGCACAAACCACTCTAAAG

Repeat_G1-f1 CCCAAAATGGCATAACCAAATGATTG

Repeat_G1-r1 AAAATCAATAGGTGCCGGAGCTGCT

Repeat_H1-f1 AAAGGAATTCCATCCACTCCGGCTT

Repeat_H1-r1 CCGTAAGAGGCATTTCACTGTGCTC

Repeat_K1-f1 ACAACTTTTAGTGCCATATCTTCCAAAACCC

Repeat_K1-r1 TCAGTGAGGGCGACCACGAAAGTAT

Repeat_L1-r1 AAAATCAATAGGTGCCGGAGCTGCT

Repeat_L1-f1 GAAGGTTTCCATTCAGTCTCATAAAGCAAG

Repeat_G1-r1 CCCAAAATGGCATAACCAAATGATTG

Repeat_I1-f1 GTGATACAGGAAGCTGTTTC

Repeat_I1-r1 CTCACTCAATCAATAGTCGG

Repeat_N1-f1

Repeat_N1-r1

AGTGTGGAGAGGAACCCTGC

ACTCGGGCCATCATAAAGAG

CACATTAGGAATCCTCAATC Repeat_Q1-f1

Repeat_Q1-r1 TATGGTTCCTAGACCTGTAC

Repeat_V1-f1 GCTCATCTTCTTCTAACGAA

Repeat_V1-r1 CATAACCAGAAGAATTGTGA

Repeat_YY1-f1 TTCGATCTTAGTCGTAAGCT

Repeat_YY1-r1

Imperfect repeat (IR)-f1

Imperfect repeat (IR)-r1

TCTATCGCCTCTACGCTATC

AAATGACAGTCGGGCTCCCA

CATTTGTCCAATCCACCGAA

M. P. Arrieta-Montiel et al. 4 SI

TABLE S2

Repeats within the Arabidopsis mitochondrial genome identified by size and location

Repeat Length (bp)

Start End

Neighboring

Repeat

Length

(bp)

Start

End

Neighboring

1 6590 252404 258993 U-1 187 217220 217406 nad6 5'

1 6590 118239 124828 U-2 187 308860 309046

2 4197 184955 189151 V-1 180 135820 135999 ca 5' atp8

2 5198 5000 10197 V-2 180 251827 252006 ca 5' atp6-1

A-1 556 19679 20234 tRNA 1 part of rps3 V-3 170 117599 117768

A-2 556 347125 347680 tRNA 13 W-1 184 100391 100574

B-1 533 244062 244594 repeat 2, tRNAs 5, 6 W-2 183 279398 279580

B-2 533 322365 322897 repeat 5, tRNAs 15, 14 X-1 204 289252 289455

C-1 457 36349 36805 X-2 206 307898 308103

C-2 456 238966 239421 Y-1 175 117785 117959

D-1 452 6121 6572 cox 2 5' Y-2 175 119395 119569

D-2 452 96210 96661 Y-3 175 253560 253734

E-1 441 269735 270175 orf 262 Z-1 142 11399 11540

E-2 441 333117 333557 orf262 portion Z-2 142 92672 92813

F-1 350 136378 136727 atp8 3' AA-1 209 117570 117778

F-2 350 176902 177251 AA-2 223 135787 136009

G-1 335 30927 31261 orf 315 5' BB-1 145 1960 2104

G-2 335 270932 271266 atp9 3' BB-2 147 36460 36606

H-1 342 31714 32055 orf 315 3' BB-3 147 239076 239222

H-2 342 333111 333452 orf 262, subset E-2 BB-4 145 187048 187192

H-3 336 269735 270070 orf 262 subset E-1 CC-1 148 101381 101528

I-1 281 30431 30711 CC-2 149 149654 149802

I-2 281 128373 128653 DD-1 128 66960 67087

J-1 285 45840 46124 DD-2 128 276363 276490

J-2 281 125568 125848 EE-1 127 107016 107142

K-1 251 23248 23498 rpl16 EE-2 127 268131 268257

K-2 251 332511 332761 rpl16 portion FF-1 118 119256 119373

L-1 249 270646 270894 atp9 FF-2 117 362998 363114

L-2 249 332807 333055 3' atp9 FF-3 118 253421 253538

M-1 259 149013 149271 GG-1 108 176014 176121

M-2 258 270444 270701 GG-2 108 217415 217522 nad6 5'

N-1 293 70157 70449 SS-1* 87 30692 30778

N-2 286 251484 251769 ca 5' atp6-1 SS-2 86 109794 109879

O-1 296 333590 333885 orf275 portion YY-1* 53 288389 288441

O-2 311 359898 360208 YY-2 53 36035 360401

M. P. Arrieta-Montiel et al. 5 SI

P-1 465 41386 41850 IR-1 342 182596 182938

P-2 430 124907 125336 IR-1 342 348500 348842

Q-1 206 113689 113894 ca 3' atp-1

Q-2 206 269441 269646 ca 3' orf262

R-1 206 175812 176017

R-2 206 200416 200621 cob 5'

S-1 230 45567 45796

S-2 221 125322 125542

T-1 185 111907 112091 ca 5' atp1

T-2 185 361780 361964

Repeats in bold were tested for recombination activity in the msh1 mutant. Asterisks designate those putative repeats at which no recombination was detected.