Utility of in vitro culture to the study of plant mitochondrial genome configuration and its dynamic features

Recombination activity plays an important role in the heteroplasmic and stoichiometric variation of plant mitochondrial genomes. Recent studies show that the nuclear gene MSH1 functions to suppress asymmetric recombination at 47 repeat pairs within the Arabidopsis mitochondrial genome. Two additional nuclear genes, RECA3 and OSB1, have also been shown to participate in the control of mitochondrial DNA exchange in Arabidopsis. Here, we demonstrate that repeat-mediated de novo recombination is enhanced in Arabidopsis and tobacco mitochondrial genomes following passage through tissue culture, which conditions the MSH1 and RECA3 suppressions. The mitochondrial DNA changes arising through in vitro culture in tobacco were reversible by plant regeneration, with correspondingly restored MSH1 transcript levels. For a growing number of plant species, mitochondrial genome sequence assembly has been complicated by insufficient information about recombinationally active repeat content. Our data suggest that passage through cell culture provides a rapid and effective means to decipher the dynamic features of a mitochondrial genome by comparative analysis of passaged and non-passaged mitochondrial DNA samples following next-generation sequencing and assembly.


Introduction
The plant mitochondrial genome is organized into an unusual multipartite structure derived from high and low frequency DNA recombination between repeated sequences in the genome (Fauron et al. 1995). Large-sized ([1,000 bp) repeats participate in high-frequency reciprocal DNA exchange to subdivide the genome and facilitate inter-conversions between DNA molecules (Mackenzie and McIntosh 1999). Intermediate-sized (50-550 bp) repeats mediate low frequency asymmetric DNA exchange that results in accumulation of only one of the expected recombinant products (Shedge et al. 2007). Frequency of DNA exchange at these intermediate repeats appears to control the relative copy number of the recombinant forms within the genome.
In plants, the mitochondrial DNA population in vegetative tissues is organized into predominant and substoichiometric DNA configurations. Changes in relative abundance of these mitochondrial genomic forms, often occurring within a single plant generation, are referred to as substoichiometric shifting (SSS) (Arrieta-Montiel and Mackenzie 2011), a phenomenon first reported in maize (Small et al. 1987). The SSS process participates in the rapid generation of mitochondrial genome variation within a plant species (Davila et al. 2011), and appears to underlie reversible phenotypic transitions between cytoplasmic male sterile and male fertile plants within a population (Mackenzie 2011). Asymmetric recombination at intermediate-size repeats in mitochondria accounts for SSS activity (Arrieta-Montiel et al. 2009).
Three nuclear genes have been cloned and reported to participate in the control of plant mitochondrial recombination in Arabidopsis: MSH1, RECA3 and OSB1 (Abdelnoor et al. 2003;Shedge et al. 2007;Zaegel et al. 2006). Among these nuclear genes, MSH1 appears to have the most profound effect on plant mitochondrial recombination surveillance and inherited plant phenotypic effects (Shedge et al. 2007;Sandhu et al. 2007). In Arabidopsis, mutation of MSH1 elevates the mitochondrial DNA exchange activity at 47 repeat pairs ranging from 50 to 550 bp in size (Arrieta-Montiel et al. 2009;Davila et al. 2011). Although mitochondrial recombination and SSS activity can occur spontaneously at low frequency, disruption of nuclear genes MSH1, RECA3 or OSB1 enhances SSS frequency.
Earlier evidence of mitochondrial DNA polymorphisms arising in cultured cells is plentiful (Rode et al. 1987;Shirzadegan et al. 1989;Vitart et al. 1992;Hartmann et al. 2000). Most provocative among these reports is the observation of recombination events in tobacco that are reversible with plant regeneration, implying SSS activity under culture conditions (Kanazawa et al. 1994). We investigated the nature of mitochondrial genome changes under tissue culture conditions to assess the feasibility of capitalizing on this process for mitochondrial genome mapping. Here, we present evidence that tissue culture results in reduced expression of both MSH1 and RECA3, together with enhanced recombination at intermediate repeats. This effect is reversible with plant regeneration, providing a useful system for the assembly of mitochondrial genome sequence into a more accurate physical map of the genome.
In order to induce regeneration of tobacco shoots, 4-week-old tobacco callus was cultured on basic Murashige and Skoog inorganics and Gamborg's B5 vitamins, supplemented with 1 mg L -1 benzylaminopurine (BAP), 0.1 mg L -1 1-Naphthyl acetic acid (NAA) and 3 % sucrose. After formation of shoots, regenerated young seedlings were transferred to tobacco rooting medium with basic Murashige and Skoog inorganics, Gamborg's B5 vitamins, 0.1 mg L -1 1-Naphthyl acetic acid (NAA) and 3% sucrose. All media were solidified with 0.8% Grade A agar (Sigma) with pH 5.7 ± 0.1. Shooting and rooting occurred in a growth chamber at 24°C and 18-h day length provided by cool light fluorescent tubes, producing a photon flux density of 92-97 lmol m -2 s -1.
Total genomic DNA isolation, gel blot and PCR assays 1-to 4-week-old Arabidopsis and tobacco calli were sampled. Young leaves from Arabidopsis Col-0 and msh1 mutant plants (Abdelnoor et al. 2003) and tobacco wildtype (Xanthi) plants served as control samples, which were grown in the growth chamber under the same condition as the growth chamber where seed plates were maintained. Total genomic DNA was isolated from calli and plants, according to the Li and Chory (1998) protocol. Total genomic DNA was digested with BamHI (Arabidopsis) and ClaI (tobacco), and analyzed by DNA gel blot hybridization (Hybond-N, Amersham). Mitochondrial DNA repeats Total genomic DNAs from tobacco regenerants at three different growth stages were assayed by three-primer competitive PCR. Primers (Table 1) were designed to assay each environment flanking Repeat A. Actin was used as internal control.

RNA isolation and real-time quantitative PCR analysis
Total RNA was extracted from Arabidopsis and tobacco callus, and wildtype and regenerated plants, with TRIzol (Invitrogen) and purified with RNeasy (Qiagen). Purified RNA was used to synthesize first-strand cDNA (Super-Script III First-Strand Synthesis SuperMix for qRT-PCR; Invitrogen). Equal amounts of cDNA were used for quantitative PCR with SYBR GreenER for iCycler (Invitrogen). Quantitative PCR analysis used iCycler iQ software (version 3.1, Bio-Rad). The ubiquitin gene was used as internal standard in gene expression analysis (Arrieta-Montiel et al. 2009). Fold change for MSH1, RECA3 or OSB1 expression in each sample, compared to wildtype, was calculated as 2 ÀDDC t (DDC t = DC tS -DC tWT , DC tS = C tMSH1 -C tUbq , DC tWT = C tMSH1 -C tUbq ). Primers for RT-PCR analysis of ubiquitin, MSH1, RECA3 and OSB1 in Arabidopsis and ubiquitin and MSH1 in tobacco are included in Table 1.

Mitochondrial recombination increased markedly at very early stages of Arabidopsis and tobacco callus culture
Initiation of callus formation could be detected 1 week following placement of Arabidopsis and tobacco leaf segments on callus-inducing medium, with mature friable callus formed after 4 weeks (Fig. 1). Total genomic DNA extracted from 1-to 4-week-old calli was digested with BamHI (Arabidopsis) or ClaI (tobacco) for mitochondrial genome analysis. In Arabidopsis, repeat-mediated recombination was detected in 1-week-old cultures, with the level of recombination increasing over time. The 4-week-old cultured tissues displayed the same 4.1-kb recombinant form at Repeat F that is observed in the msh1 mutant (Arrieta-Montiel et al. 2009), while both parental forms were also retained (Fig. 2a). Recombination at Repeat D gave the predicted 2.2-kb recombinant molecule in callus culture, again increasing with time (Fig. 2a). In tobacco callus, recombination at Repeat A was evident in the predicted recombinant 6-kb form (Fig. 2c), which is observed only at very low levels in the wildtype plant.  wildtype (Col-0 and Xanthi) plant samples (Fig. 3). In both Arabidopsis and tobacco, the decrease in expression was more pronounced with age of callus. RECA3 expression was also down-regulated in Arabidopsis, but OSB1 transcript levels did not change under tissue culture conditions. We suggest that the mitochondrial genomic rearrangements observed are the likely consequence of altered MSH1 and RECA3 expression.

MSH1 expression is correlated with SSS activity during tobacco plant regeneration
During tobacco regeneration from 4-week-old callus, three stages were investigated (Fig. 4a): Stage 1, 1 month following transfer of 4-week-old callus to shooting medium; Stage 2, 2 months following callus transfer, when roots emerge in rooting medium; and Stage 3, 1 week following transfer of the young seedling to potting mix. Genomic DNA was extracted from the different growth stages, and three-primer competitive PCR analysis (Fig. 4b) allowed resolution of changes in relative stoichiometries for parental and recombinant forms during the regeneration process. Recombinant forms were predominant in callus, and Stage 1 reversed the trend back toward substoichiometric levels in the subsequent regeneration stages. Recombinant and parental forms were confirmed by DNA sequencing. Similarly, down-regulation of MSH1 expression in Stage 1 was gradually reversed during regeneration, reaching normal MSH1 transcript levels in the regenerated plant (Fig. 4c).

Discussion
Over the past 20 years, mitochondrial genome rearrangements were often reported to occur in plant cells grown in vitro (Cloutier and Landry 1994). It was generally thought that these rearrangements were the consequence of extended culture periods. Evidence presented here of mitochondrial SSS activity within the first week of callus These earlier studies showed that an ''early'' generation msh1 mutant produces mitochondrial genome rearrangements at lower frequency that are reversible following re-introduction of the MSH1 gene (Davila et al. 2011). The ''advanced'' generation msh1 mutants produced more extensive rearrangements that were less readily reversible and could become fixed in the mitochondrial population.
Combining the msh1 and recA3 mutations resulted in the most extensive mitochondrial genome rearrangements (Shedge et al. 2007). In this study, we have shown that down-regulation of both MSH1 and RECA3 occurs under cell culture conditions, suggesting that conditions are appropriate for rapid and extensive mitochondrial genome changes.
A question raised by this study is whether plant regeneration would be facilitated if the mitochondrial genome were stably maintained during in vitro culture. Tobacco, where plants can be readily regenerated from callus, has been reported to display a highly reversible mitochondrial rearrangement process (Kanazawa et al. 1994). We also observed reversibility of mitochondrial genome rearrangement in tobacco upon regeneration from callus. Might this reversibility be a factor in the plant's amenability to regeneration? There appear to be several plant developmental implications of the msh1 recA3 double mutation (Shedge et al. 2007(Shedge et al. , 2010, suggesting that this type of gene expression change can have dramatic implications for plant growth.  Because the plant mitochondrial genome is characterized by numerous recombinational repeats, assembly of an intact genome sequence, particularly using deep sequence reads of relatively small size, can be extremely difficult. Sequence assembly without the availability of information from substoichiometric forms provides an incomplete picture of the genome, so that intra-specific mitochondrial comparative studies can imply far more extensive genomic variation than is actually present. Information about substoichiometric forms can allow one to deduce the interconvertibility of related mitochondrial configurations. In many plant species, particularly those being investigated for ecological studies, little mitochondrial genome information is currently available and details of intra-specific mitochondrial relationships may be of crucial importance. In these cases, implementing callus culture for mitochondrial genome analysis may be valuable. Comparative assembly of the callus culture-derived mitochondrial genome sequence and intact plant-derived form would reveal recombinational repeats and substoichiometric forms. This type of information is valuable for subsequent ecotype comparisons and for developing an understanding of evolutionary trends within a species (Davila et al. 2011).