Introduction

The evolution of the plant mitochondrial genome (mtDNA) is thought to be mainly driven by extensive structural reorganization, recurrent incorporation of foreign DNA material, and frequent gene transfer to the nucleus, while the roles of substitutions and small indel changes are believed to be rather minor, when compared to the animal or fungal mtDNAs (e.g., Palmer et al. 2000; Barr et al. 2005). However, recent reports of high mutation rates within specific lineages (Cho et al. 2004; Mower et al. 2007; Ran et al. 2010) suggest that this type of changes also play a role in the evolution of plant mtDNA. Surprisingly, the majority of these examples involve large indels up to 3,000 bp long, instead of the short-scale length variation (i.e., indels less than 100 bp) that is most commonly reported in plant population studies (e.g., Bastien et al. 2003; Jaramillo-Correa et al. 2003; Godbout et al. 2005, 2008).

In eukaryote genomes, small indels are often associated with repetitive sequences, such as microsatellites or mobile elements (e.g., Buschiazzo and Gemmell 2006). In the mtDNA of some plants, these repetitive sequences are so abundant that they have led to important size increases and structural rearrangements (e.g., Palmer et al. 2000; Lilly and Harvey 2001; Hecht et al. 2011). This profusion raises the question of whether such repetitive elements tend to accumulate in specific regions of the plant mtDNA, and if these regions have higher mutation rates than other parts of the genome. Such higher rates might imply that repeated sequences do not only play a key role as promoters of recombination (e.g., Allen et al. 2007; Hecht et al. 2011), but also as mutagenic hotspots (e.g., Amos et al. 2008; McDonald et al. 2011). Indeed, mutation rate differences between repetitive and non-repetitive regions have been reported for the nuclear and chloroplast genomes of plants (e.g., Thuiller et al. 2002; Jackobson et al. 2007), and also for the animal mtDNA (e.g., Broughton and Dowling 1997; Wilkinson et al. 1997).

Sequence-tandem repeats (STRs), which include both micro and minisatellites, are formed by iterations of DNA-sequence motifs. In the mitochondrial genome of different taxa, they have been associated with diverse processes including recombination, gene regulation, and some disorders (e.g., Vergnaud and Denoeud 2000; Gemayel et al. 2010). In general, STRs can arise spontaneously from unique sequences or can be brought in by mobile elements; they grow by replication slippage and/or unequal cross-over during recombination, until reaching a contraction–expansion equilibrium, and finally become interrupted by the proliferation of random mutations that break up the repeat pattern (reviewed by Buschiazzo and Gemmell 2006). Some studies (e.g., Mouhamadou et al. 2007; Morse et al. 2009) have reported that STRs tend to accumulate in specific regions of the genome. Such a pattern has been interpreted as the “renaissance” of previously interrupted motifs and/or as the recurrent insertion of repetitive elements in particular hotspots (Buschiazzo and Gemmell 2006). Nevertheless, only a few STR-rich regions have been investigated in plants, most of them in the nuclear genome (e.g., Karhu et al. 2000; Chen et al. 2002).

In firs (Abies Miller, Pinaceae), a predominantly temperate group of conifers, the fourth intron of the mitochondrial gene nad5 (nad5-4) represents such a STR hotspot. It contains a complex array of up to four different repeated motifs (Fig. 1a), most of them shared by distantly related taxa from Asia, the Mediterranean, and Mesoamerica (e.g., Ziegenhagen et al. 2005; Jaramillo-Correa et al. 2008; Peng et al. 2012). The concentration of that amount of STRs in a region that is only ~800 bp long, and the availability of population data to assess its length variation in different species (Liepelt et al. 2002, 2010; Ziegenhagen et al. 2005; Jaramillo-Correa et al. 2008, 2011; Jiang et al. 2011; Wang et al. 2011; Aguirre-Planter et al. 2012; Peng et al. 2012), provides a unique opportunity for surveying the potential role of small indels in the evolution of the plant mitochondrial genome.

Fig. 1
figure 1

Schematic organization of the sequence tandem repeats (STRs) found in the fourth intron of the mitochondrial gene nad5 of the genus Abies. The sequence and position of each STR is indicated for the two most common haplotypes (12 and 18), and the sites where substitutions are observed in other haplotypes are indicated by arrows. Note the presence of STR-2 interrupting the first array on haplotype 12 and its absence in haplotype 18

Full size image

In this study, we investigated the organization and evolution of this STR-rich mitochondrial intron in the genus Abies and its related taxa within the Pinaceae. First, we mapped its indel and substitution variation onto the phylogeny of this family, and estimated and compared its mutation rates with that of other mtDNA genes. Then, we correlated the available population data with the sequence variability (i.e., number of STRs, and repeats and substitutions per STR), and with some species distribution-traits (i.e., size and type of natural range, a proxy for species size), to examine if these variables are related to mtSTR diversity as reported for nuclear microsatellites in other species (e.g., Zhu et al. 2000; Nybom 2004). Finally, we searched the DNA sequences of some of the complete plant mtDNA genomes available to investigate how frequent STR hotspots are across taxa.

Materials and Methods

Sampling and Sequencing of mtDNA Genes

A total of 31 species and two varieties from the approximately 50 taxa currently recognized in the genus Abies (Liu 1971; Rushforth 1989; Farjon and Rushforth 1989) were considered together with one species of some of its related genera within the Pinaceae, namely: Cedrus atlantica, Keteleeria evelyniana, Larix decidua, Picea abies, Pinus pseudostrobus, Pseudotsuga menziesii, and Tsuga canadensis (Table S1). These taxa were selected in order to represent (i) as broad a phylogeographic and phylogenetic range as cross-amplification would allow, (ii) to sample the widest possible range of nad5-4 allele sizes, and (iii) to include both monomorphic and polymorphic taxa.

DNA was extracted with a CTAB mini-prep protocol (Vázquez-Lobo 1996) or a DNeasy Plant Mini Kit (Qiagen) from foliage collected in natural populations, arboreta, and botanical gardens (Table S1). PCR amplification was carried out in a PTC-225 thermal cycler (MJ Research) with the universal and internal primers and the conditions described elsewhere (Dumolin-Lapègue et al. 1997; Jaramillo-Correa et al. 2008). PCR products were examined by agarose gel electrophoresis (2 % in TAE) in order to verify that only one single band was amplified. Both DNA strands were then directly sequenced in an Applied Biosystems 3130xl DNA Genetic Analyser by using the appropriate primers, a Sequenase GC-rich kit (Applied Biosystems) and a dideoxynucleotide chain termination procedure. Ten additional mitochondrial regions (ccmC; cox1, cox3, matR, nad1 b/c, nad3-rps12, nad4-3, nad5-1, nad7-1, and SSU-V1 region) were amplified and sequenced, for comparative purposes, following the same methods as above and using previously reported primers (see Jaramillo-Correa et al. 2003, 2008, and references therein). Sequences available in Genbank were downloaded to complete some of the datasets above (i.e., nad1 b/c, nad3-rps12, nad4-3, and nad5-4) and to build an additional one for the gene atp8 (Polezhaeva et al. 2010; Semerikova et al. 2011).

Sequence and Phylogenetic Analyses

Phylogenetic relationships were retrieved from a combined data set taken from previous studies representing the most recent calibrated phylogenies of the Pinaceae and the genus Abies (Gernandt et al. 2008; Xiang et al. 2009; Lin et al. 2010; Aguirre-Planter et al. 2012). DNA sequences from the mtDNA genes (including the regions flanking the nad5-4 STR-rich zone) were excluded from this analysis because of their low levels of polymorphism. In brief, within the genus Abies, only 14 parsimony informative sites were obtained for the eleven mtDNA regions sequenced (representing more than 10,000 bp), which resulted in poorly resolved phylogenetic trees (data not shown).

DNA sequences of the gene nad5-4 were aligned with BioEdit (Hall 1999) and the SSRs were color-coded according to sequence resemblance (Fig. S1, see Barros et al. (2008) for an example in human SSRs). The different motifs observed were then traced onto the phylogenetic tree described above, and their putative ancestral states determined with the parsimony reconstruction method available in Mesquite (Maddison and Maddison 2011; see also Primer and Ellegren 1998 and Zhu et al. 2000). When more than one result was produced, the most parsimonious reconstructions were averaged. The order and age of each STR insertion, expansion, contraction, and interruption (i.e., substitutions) were then inferred by comparisons with previously reported divergence dates (Gernandt et al. 2008; Lin et al. 2010; Aguirre-Planter et al. 2012).

For each mtDNA region in Abies, the following indices were estimated: the mean number of pairwise nucleotide differences (π; Tajima 1983), the mutation-scaled effective population size derived from the number of segregating sites (θ W; Watterson 1975), the insertion/deletion (I) and substitution (K o) rates per site, and the average number of pairwise substitutions between species (D xy). Calculations were made with DnaSP v. 5 (Librado and Rozas 2009) following Laroche et al. (1997), and using sequences from the Pinoideae subfamily (Larix, Picea, Pinus, and Pseudotsuga) as outgroups. For the particular case of nad5-4, such indices were also determined for each independent STR and both flanking regions.

Population mtDNA Variation and Comparative Analysis

An extensive literature survey was made in the ISI Web of Science and Scopus to gather information on the variation of nad5-4 in as many Abies taxa as possible (last update in July 2012). Only those studies that sequenced and published the observed variants were retained. Then, for each taxa, the mean number of mitotypes (nh) and the mean mtDNA diversity (H E) was noted from the original publications, while the number of STRs, their length, and number of repeats were retrieved from the available DNA sequences. Previous works on nuclear SSRs (e.g., Zhu et al. 2000; Nybom 2004) showed that some species life-traits, such as the size and degree of fragmentation of the natural range, were related to the length and number of repeats, and variability of microsatellites. Thus, in order to test if such relationships could be also observed at the mtDNA level, the natural distribution of each retained taxon was classified as wide or restricted, and continuous or fragmented. Then, differences between categories were determined with two-independent Student’s t tests, while the relationships between the number of repeats, and mean nh and H E per species were tested with Pearson’s correlation coefficients (Zar 2010; see results).

Database Search for Other STR-Rich mtDNA Regions

A second literature search was performed for identifying studies reporting additional polymorphic plant mtSTRs and investigating if STR hotspots were common in other taxa. Again, only the cases with sequenced and published variants were retained, but in this case, only the type of STR (micro- or minisatellite, perfect or imperfect, and simple or compound) and the region of occurrence were noted. This data set was complemented by an intensive search of repeated motifs in 33 of the complete plant mitochondrial genomes available in GenBank (see Table 1), which was performed with the tandem repeats finder (Benson 1999) by using the default parameters. Only those sequence blocks with more than 80 % similarity and repeated at least five times where considered as STRs. When more than one sequence was available for a given genus, only the longest one was analysed (e.g. Zea mays spp. mays genotype CMS-C and Oryza rufipogon). The regions of occurrence for each STRs detected were determined from the original notation of each genome, complemented with blast searches, and compared across taxa.

Table 1 Diversity indices (nh, π, θ w), indel and substitution rates (I, K o), and average pairwise divergence (D xy) for 12 mitochondrial DNA regions across species in the genus Abies
Full size table

Results

Description of the nad5-4 mtSTR Hotspot

Depending on the individual and species surveyed, the whole intron spanned between 832 and 1012 base pairs (bp), and contained a STR-rich region covering between 156 and 336 bp. No heteroplasmy was observed by electrophoresis, cloning or sequencing of PCR products. All the newly generated sequences for Abies and the other Pinaceae are available in Genbank (Accession nos. KC578686–KC578818).

Within the genus Abies, the extensive sequencing of 135 individuals revealed 24 different haplotypes that harbored between two and four STRs (Fig. 1). The first of these repeats was an imperfect array with two to eight repetitions with the basic motif CTATAT. The fifth of these repeats extended into a small and monomorphic AT succession, while one G/T and four A/C substitutions were also observed in different repeat copy members across haplotypes (Fig. S1). The second STR was an imperfect stretch of eight to twelve iterations of the basic sequence GATA. It was exclusive to the haplotypes harboring more than three copies of the first microsatellite (haplotypes 1–16), which was interrupted by these motifs (Fig. 1). The first three copy members of this stretch extended into small and monomorphic AT successions with some T/G substitutions in three particular haplotypes (5–7). The third and fourth STRs were both imperfect AT repeats (Fig. 1) that were located 30 and 100 bp downstream the end of the first STR, respectively. The former was present in all but three haplotypes (14, 15, and 21) and harbored one TTT and two AAA motifs that interrupted an array of six to nine repeats. The later was observed in all 21 haplotypes and was formed by 12 or 13 repetitions that were disrupted by one TT and one AA motif, and one A/C substitution (Figs. 1, S1).

Most of the intron, especially the sequences flanking the STRs, appeared to be well conserved within the Pinaceae (Fig. S1). No within-species polymorphism was observed outside the genus Abies, in spite of an intensive screening of the natural distributions of three additional taxa (Pinus pseudostrobus, Picea abies, Tsuga canadensis; data not shown). The basic sequence pattern of the first (1–3 repeats), third (9–13 repeats), and fourth (12–13 repeats) STRs were present in all species surveyed except Picea abies, while a proto-STR (2 repeats) of the second motif was observed in Tsuga. Blast searches and visual comparisons with the homologous intron of other plants revealed that the regions flanking the STRs were also well conserved in the gymnosperm Cycas taitungensis and in many angiosperms. However, although a proto-STR (1 repeat) of the first array, together with four imperfect AT-tandem repeats, were detected in Cycas, the zone containing the tandem stretches was highly variable outside the Pinaceae and was virtually impossible to align (data not shown). Interestingly, one of the tandem arrays found during the analyses of the complete mtDNA genomes of Zea and Tripsacum was located in this same region, nad5-4 (see below). Nevertheless, no sequence similarities were observed between this array and the conifer STRs (Table S2), thus suggesting independent origins for these motifs.

Evolution of the nad5-4 STR-Rich Region

Tracing the variable STR characters onto the Pinaceae phylogeny (Fig. 2) revealed that the insertion and proliferation of the first, third, and fourth-repeated motifs preceded the divergence of the subfamilies Abietoideae (i.e. Abies, Cedrus, Keteleeria, and Tsuga) and Pinoideae (i.e. Larix, Picea, Pinus, and Pseudotsuga) of the Pinaceae. Interestingly, most of the substitutions interrupting these arrays were observed in the branches leading to the Pinoideae subfamily and the genus Cedrus. The insertion of the second repetitive motif apparently occurred after the differentiation of Cedrus from the other Abietoideae, while its proliferation took place exclusively within the genus Abies. This genus seemingly had two ancestral haplotypes (12 and 18), which are currently distributed in species from Eurasia and North America, and which, respectively, showed the presence (haplotype 12 and derived forms) and absence (haplotype 18 and derived forms) of the second STR (Figs. 1, 2, 3). The different polymorphisms that characterized the remaining haplotypes in Abies were restricted to a few species or populations within species, and generated a significant phylogeographic structure (Fig. 3; see Liepelt et al. 2002, 2010 and Jaramillo-Correa et al. 2008 for more details), suggesting a more recent origin.

Fig. 2
figure 2

Tandem-repeats found in the fourth intron of the mitochondrial gene nad5 mapped onto the phylogeny of the Pinaceae, with emphasis on the genus Abies (adapted from Gernandt et al. (2008), Lin et al. (2010), and Aguirre-Planter et al. 2012). STRs are represented by squares with different fillings. Only the mutations leading to new motifs on each STR are mapped on the phylogenetic tree. Ancestral states were determined by parsimony reconstruction using Mesquite (Maddison and Maddison 2011). Branches exhibiting more than one haplotype indicate putative ancestral polymorphisms. Thick horizontal lines represent branches supported by bootstrap values over 95, thinner lines have values between 50 and 95, and dashed lines have no statistical support (i.e., bootstrap below 50) and are shown only to ease the visualisation of particular mutations

Full size image
Fig. 3
figure 3

Minimum spanning network of the haplotypes found in the fourth intron of the mitochondrial gene nad5 of the genus Abies. Each mtDNA type is represented by a circle whose filling indicates its geographical location. Lines connecting mitotypes denote mutations that are represented by three different symbols (see “mutation codes”). The mtDNA types observed in the six outgroups are shown as dashed circles, and the missing mitotypes are represented by small white dots

Full size image

Sequence and Population Diversity Across Abies Taxa

As a whole, the nad5-4 intron had higher diversity and mutation rates in the genus Abies than other mitochondrial genes (Table 1). In particular, both the indel and substitution rates were an order of magnitude higher than those of other introns, while all the coding regions surveyed (atp8, ccmC, cox1, cox3, matR), excepting SSU rRNA V1, were monomorphic across species. However, when analysing each nad5-4 region independently, it was observed that this increase of the mutation rates (for both indels and substitutions) actually occurred around the first two STRs, while the remaining of the intron, including the regions with the last two-repeated stretches, were far less variable across Abies taxa. Nevertheless, it is noteworthy that these last two regions were highly divergent (D xy values in Table 1) when compared to the outgroups (i.e., the taxa in the Pinoideae subfamily), while the zones of the intron devoid of STRs had similar D xy estimates than the observed for other mtDNA genes (Table 1). Pairwise divergence could not be estimated for the first two stretches given their short size (STR-1) and complete absence (STR-2) in the outgroups, respectively.

On the other hand, and after collecting data from eight population studies implicating 24 Abies taxa (i.e., Liepelt et al. 2002, 2010; Ziegenhagen et al. 2005; Jaramillo-Correa et al. 2008, 2011; Jiang et al. 2011; Wang et al. 2011; Aguirre-Planter et al. 2012; Peng et al. 2012), no significant differences were detected for the number of STRs within nad5-4, their length, or number of repeats, across species distributed in wide, restricted, continuous, or fragmented ranges (Fig. 4a). Similar results were obtained for the number of mitotypes (nh) and average mtDNA diversity (H E) (Fig. 4b), which suggests that, contrary to nuclear STRs, species effective sizes appear to be independent of the mtSTRs length and diversity for this particular locus.

Fig. 4
figure 4

a Length (left y axis) and number of repeats (right y axis), and b number of mitotypes (left y axis) and mtDNA diversity (right y axis) observed for the fourth intron of the mitochondrial gene nad5 in population studies of Abies species distributed in wide or restricted (left), and continuous or fragmented (right) natural ranges. Boxes, lines within boxes, and black dots in a represent 25–75th percentiles, and mean and outlier values, respectively. Error bars in a and b denote 95 % confidence intervals

Full size image

STR Mining in Other Plant Mitochondrial Regions

The initial literature review resulted in 45 plant mtSTRs, most of which (i.e., 21) were reported for wheat (Triticum aestivum) and its wild relatives (Ishii et al. 2006). Six minisatellites have also been reported in Brassica napus, five in Oryza sativa and its relatives (Honma et al. 2011), four in Beta vulgaris (Nishizawa et al. 2000), and nine tandemly repeated stretches in various conifers (Tables 2, S2). However, it must be noted that half of the STRs reported for Beta, and most of those described for Brassica and Oryza did not meet our selection criteria (80 % homology and at least five-repeated stretches) and were excluded, thus reducing the final set to 33 STRs (Table 2).

Table 2 Summary of the number and location of plant mitochondrial sequence tandem repeats (mtDNA STRs) found and meeting selection criteria (see below) with two different search methods
Full size table

The in silico search yielded 309 additional STRs for taxa as varied as liverworts, mosses, cycads or grasses, whose complete mtDNA genomes ranged between ~100 and ~1,000 kb long (Table 2, S3). The method used was able to recover all repeated stretches previously reported in the literature for those plants with a complete mtDNA genome (e.g., Beta vulgaris, Brassica napus, or Oryza). However, only those tandem repeats meeting the pre-established selection criteria were retained. The taxa exhibiting the highest number of STRs were Selaginella moellendorfii (Selaginellaceae), Tripsacum dactyloides, Zea mays (Poaceae), Cucurbita pepo (Cucurbitaceae), and Cycas taitungensis (Cycadaceae). A positive but non-significant correlation was observed between the genome length and the number of STRs (r 2 = 0.249; t test P = 0.17), which became significant once the atypical S. moellendorfii was removed from the dataset (r 2 = 0.475; t test P < 0.01; see “Discussion” section).

Sequence annotations, complemented by further blast analyses, revealed that most of these STRs (94 %) were located in mitochondrial intergenic regions, followed by group-II introns (6 %) (Tables 2, S3). Only two STRs were observed in coding regions, one in Brassica napus (orf101a) and the other in Raphanus sativus (orf138). The first STR was conserved in other species of the genus Brassica, while the second one was exclusive to a male-sterile variety of radish (data not shown). Other than the fourth intron of the gene nad5, for which STRs were observed in Abies, Zea, and Tripsacum, repeated stretches were also retrieved across taxa in the following introns: nad1 b/c (six Eurasian Picea, Pinus ponderosae complex, Triticum–Aegilops complex, Selaginella), nad4-1 (most taxa in the Poaceae), and nad4-2 (Cycas, Phoenix and Selaginella). However, compound STRs were found only in the first case, in specific taxa of the genera Picea and Pinus (Mitton et al. 2000; Sperisen et al. 2001). Within species, 17 intergenic regions bore more than one set of repeated motifs (Table 2); some examples include: rps2B2–rrn26-2 in Z. mays with six STRs, nad1–ycf2 in C. pepo with five stretches, and cob–atp1 in S. moellendorfii with four different motifs. Nevertheless, these STRs were dispersed in regions spanning more than 3 kb, except for the last case where the repeated stretches were restricted to a segment of only 330 bp (see Hecht et al. 2011).

Discussion

In this study, we showed that STRs can accumulate and be maintained over long periods of time (i.e., between 180 and 130 Ma; Aguirre-Planter et al. 2012) in short regions of the plant mitochondrial genome, and that this accumulation can translate in increased mutation rates (both indels and substitutions). The population data for the case study region (intron nad5-4) further suggested that mtSTR variability in Abies is quiet independent of the species’ effective sizes, thus indicating that its variability is the result of the differential fixation of ancestral polymorphisms, instead of the accumulation of more recent mutations following speciation. Also, searches in complete plant mtDNA genomes revealed that although STRs can accumulate in all sorts of plants, proliferation of STR hotspots appears to be taxa specific.

Recurrent STR Insertions in the Plant Mitochondrial Intron nad5-4

The co-location of various STRs in short-genomic regions (i.e., less than 1 kb) has been interpreted as previously interrupted motifs that resurge, by recurrent mutations, as linked sets of tandem repeats and/or as a the continual insertion of repeated stretches in mutational hotspots (e.g., Mouhamadou et al. 2007; Morse et al. 2009). The basic repeat sequence of the STRs found in the Pinaceae nad5-4 was very different in three out of four cases, which suggests at least three-independent insertions during the evolution of this family. A putative fourth insertion could be at the origin of the last-repeated stretch, although the separate evolution of a previously interrupted motif cannot be ruled out, given that the third and fourth STRs share the same basic sequence (AT; see Figs. 1 and S1).

Further evidence for recurrent STR insertion in this intron derives from the analysis of complete plant mtDNA genomes, which revealed another microsatellite in the homologous intron of Zea and Tripsacum (Table S3). Our blast analyses showed a proto-minisatellite formed by two repeats of the basic motif (CAATCC) in Sorghum and other Poaceae (data not shown), which indicates that the proliferation of this STR occurred after the divergence of the lineage leading to Zea and Tripsacum, some 12 Ma (Gaut et al. 2000; Swigoňová et al. 2004).

Like many other plant mitochondrial introns, nad5-4 is a group-II intron that has apparently lost its self-splicing ability (Kelchner 2002). This type of introns share a basic secondary structure, with a relatively well conserved core region and six helical domains with different degrees of divergence, and within which indel variation and insertions of mobile elements are rather common (e.g., Laroche and Bousquet 1999; reviewed by Bonen 2008). Indeed, polymorphic STRs have been observed within plant group-II introns of mitochondrial (e.g., nad1 b/c in Picea; Sperisen et al. 2001, nad7-1 in Pinus; Godbout et al. 2005, 2008), and chloroplast genes (petD, Löhne and Borsch 2005), while 6 % of the STRs recovered herein from the analyses of complete mtDNA genomes were located in this type of introns (see Table S2). Explanations to account for the accumulation of STRs in group-II introns include relaxed evolutionary constrains, as the self-splicing functions were taken over by external elements (Carrillo et al. 2001), and intragenomic recombination among short-direct repeats (e.g. Woloszynska et al. 2001). Whether the STRs detected within the Abies nad5-4 are the result of such rearrangements is still a hypothesis to test, and for which it might be necessary to sequence complete mtDNA genomes across different lineages for detecting the recombination hotspots and their flanking sequences (see Allen et al. 2007, for an example in maize). Nevertheless, mtDNA recombination has already been reported in conifer hybrid zones, but without identifying the specific mechanism involved (Jaramillo-Correa and Bousquet 2005).

Increased Mutation Rates in the nad5-4 STR Hotspot of Abies

In the Pinaceae, most of the mutations observed in the nad5-4 STR hotspot occurred at different times (Fig. 2). For instance, the second array was inserted after the separation of Cedrus from the other Abietoideae, about 100 Ma (Gernandt et al. 2008; Lin et al. 2010), while most mutations within this STR occurred exclusively in the genus Abies, during the last 80–40 Ma (see Aguirre-Planter et al. 2012 for divergence dates of this genus). As many exclusive substitutions can also be observed within STRs 3 and 4 in Pinus (data not shown), which probably took place during the last 80–70 Ma, after the divergence of this genus (Gernandt et al. 2008). Such heterogeneous patterns of mutation rate (i.e., indel and substitution rates) are not rare in plant mitochondrial genes, and are often taxon- or region-specific. For example, in the gene rps3, decreased mutation rates were observed for the first intron in the Betulaceae (Laroche and Bousquet 1999) when compared to annual plants such as Petunia, while in its third exon, augmented mutation rates were noted in the Podocarpaceae and Araucariaceae relative to other gymnosperms (Ran et al. 2010). However, none of these cases involved the insertion and proliferation of multiple STRs, such as for nad5-4 in the Pinaceae.

Both the indel and substitution rates estimated for nad5-4 were one order of magnitude higher in the STR hotspot than in its flanking regions, or than in other mtDNA genes devoid of STRs (Table 1). Indeed, the rates for this STR hotspot in Abies were of the same order as those previously estimated for the homologous intron among dicots (Laroche et al. 1997), which implies that as many mutations have accumulated in the same mtDNA region during the last 80–40 Ma in a single genus (Abies) as in the last 120–100 Ma among dicot families (Laroche et al. 1997; Chaw et al. 2004). This observation further suggests that the evolutionary pace of typically slowly evolving plant mtDNA regions can be triggered by recurrent STR insertions, such as observed for the nuclear genome (e.g., Amos et al. 2008). This trend indicates an expansion of the role of repeated sequences in the evolution of plant mtDNA: from promoters of intragenomic recombination (e.g., Palmer et al. 2000; Barr et al. 2005; Allen et al. 2007) to mutagenic hotspots (present study).

Accumulation of STRs in the Plant mtDNA Genome

If recurrent STR insertions can increase the mutation rate of a specific plant mtDNA region, it then follows to verify how frequent STR hotspots are across whole plant mtDNA genomes, and if they do tend to accumulate in specific genomic zones and/or particular species. This study revealed that repeated stretches can be found in the mtDNA of all sorts of plants (from mosses and liverworts, to monocots and dicots), with no apparent bias toward a particular group (Table S3; see also Sperisen et al. 2001). Analogous examinations of EST databases revealed similar trends for the nuclear genome (e.g., Wang et al. 1994; Victoria et al. 2011). However, in this study we did find that plants with larger mtDNA genomes tend to have more and larger STRs than plants with smaller genomes, while such a relationship was not apparent for the nuclear genome (Morgante et al. 2002; Victoria et al. 2011). Indeed, the differential accumulation of micro- and minisatellites only accounted for a fraction of the nuclear size variation across species (e.g., Morgante et al. 2002; Morgante 2006), while dissimilar storing of STRs have enlarged fivefold the mtDNA genome of species like cucumber, with respect to their related taxa (e.g., Lilly and Havey 2001; Alverson et al. 2010). However, it must be noted that some species, such as S. moellendorfii, have gathered unusually high amounts of STRs (Table S3), while conserving mtDNA genomes of similar size than that of their relatives (Hecht et al. 2011).

Across plant taxa, STRs seemed more prone to accumulate in intergenic mtDNA regions than in introns (see below), while they are very rare in exons (i.e., only 3 among 342 STRs; Table 2). This pattern was expected given the highly variable and dynamic structure of the plant intergenic mtDNA regions. For instance, comparisons of mtDNA genomes across varieties of wheat or maize (Ogihara et al. 2005; Allen et al. 2007) revealed that gene synteny only exists across very short regions, thus implying a rapid evolution through abundant recombination. Indeed, in maize, such rearrangements seem to add and remove rapidly diverging STRs across the whole mtDNA genome at the short evolutionary time scale (Allen et al. 2007).

Within introns, only a few of the STRs observed were compound or mosaic stretches (i.e., 5 out 309), and all of them were found in conifers (Mitton et al. 2000; Sperisen et al. 2001; Bastien et al. 2003; present study). Outside the Pinaceae, cases of co-located STRs can be found throughout the mtDNA genome of specific taxa, such as S. moellendorfii (especially in intron nad4-2, Hecht et al. 2011), and C. taitungensis (Chaw et al. 2008). In this last species, the recurrent association of the same two elements (Tai and Bpu) across its whole mtDNA genome suggests that STR hotspots could be very common in specific lineages under certain evolutionary circumstances, such as ancestral transposon invasions (Chaw et al. 2008). Given the amount of introns with compound STRs they bear, it could be hypothesized that this phenomenon may also apply to conifers, but this cannot be verified until the first complete mtDNA genomes are sequenced for these taxa.

On the other hand, it has been proposed that STRs and other major mutations that alter genome size can accumulate more readily in species with small mitochondrial effective populations sizes (N e) (e.g., Lynch and Conery 2003; Boussau et al. 2011). Nevertheless, our analyses in Abies did not support such a hypothesis. Indeed, species with restricted or fragmented distributions (and thus small N e) had a similar number of repeats, total STR length, nh and H E, than their counterparts with wider and continuous ranges (and thus larger N e; Fig. 4). These results point to a differential fixation of ancestral polymorphisms in isolated taxa, rather than to the accumulation of new mutations following speciation (Boussau et al. 2011), which should be less frequent in species with long-generation times and large ancestral N e, such as forest trees (Bouillé and Bousquet 2005; see Petit and Hampe 2006 for a review).

Perspectives

In this study, we highlighted the role of STRs as drivers of evolution in the plant mtDNA genome. For instance, we showed that STRs can accumulate and be retained in short regions of the plant mtDNA genome over long periods of time, which differs from the traditional view that small repeated sequences are inserted and deleted on the short-term evolutionary scale, and are seldom retained among distantly related taxa (e.g., Laroche et al. 1997; Allen et al. 2007). Thus, given that STRs are ideal landmarks for surveying structural mtDNA changes (e.g., Lilly et al. 2001; Ogihara et al. 2005; Allen et al. 2007), their study through phylogenetically controlled analyses could expand such comparisons to higher phylogenetic levels, such as attempted herein for a single intron (see also Gugerli et al. 2001; Ran et al. 2010).

Furthermore, we revealed that recurrent STR accumulations can increase the mutation rate of particular mtDNA regions, which is worth investigating at the genome-wide scale. So far, there are no complete mtDNA genomes for conifers, which hamper such a study in these taxa. But this could be easily examined in the Poaceae, for which more than 12 complete genomes are already available (see Table S3, and references therein). However, mtDNA STR variation is still to be surveyed through phylogeographic studies in these species, such as done for some conifers (e.g., Sperisen et al. 2001; Godbout et al. 2005, 2008; Jaramillo-Correa et al. 2008). Primers for specific STRs are already available for Oryza (Honma et al. 2011) and Triticum (Ishii et al. 2006), but some more could be easily designed for other species (e.g., Zea) and transferred to related taxa.

Finally, the differential accumulation of STR hotspots across plants, and its putative correlation with plant life-history traits and ancestral N e remain open for further investigation. Additional completely sequenced mtDNA genomes should also be necessary to formally address these questions.