Mutation Rates in Plastid Genomes: They Are Lower than You Might Think

1For better or worse, our understanding of plastid biology is largely shaped by studies of land plants. This is particularly evident for plastid genetics. For example, >540 of the 679 complete plastid genome sequences in GenBank, as of January 1, 2015, come from embryophytes, despite the fact that most of the known plastid-containing diversity is represented by protists. Nevertheless, major insights into plastid genomes have come from land plants, not the least of which is that plastid mutation rates can exceed those of mitochondria.

2More than 25 years ago, compared plastid, mitochondrial, and nuclear DNA (ptDNA, mtDNA, and nucDNA) sequences of various land plants and found the silent site substitution rate of the mitochondrion to be lower than those of the plastid and nucleus. When noncoding and synonymous sites (collectively called silent sites) are assumed to be neutrally evolving, the silent site divergence (dsilent) between species or distinct populations can provide an entrée into mutation rate. It was, therefore, concluded that the relative levels of dsilent in land plants reflected a lower mutation rate in the mitochondrion than in the plastid or nucleus. At the time of publication, these findings went against the prevailing notion, based on studies of animal genomes, that mtDNA had a high mutation rate.

3Subsequent investigations have supported the conclusions of, and it is generally accepted that for land plants the plastid and nuclear genomes have an ∼3- to 10-fold greater mutation rate than the mitochondrial genome, with some notable exceptions. Consequently, it is sometimes assumed that lineages outside land plants have higher rates of mutation in their plastids as compared with their mitochondria. But recent organelle genome analyses from diverse lineages suggest that the opposite is true. To better understand organelle mutational patterns in plastid-bearing eukaryotes, I examined the available substitution rate data for plastid, mitochondrial, and nuclear genomes and found that an mtDNA/ptDNA mutation rate ratio of >1 can be observed in a diversity of eukaryotic lineages and might represent the norm for plastid-bearing species. If true, this could be explained by more efficient DNA repair mechanisms in plastid versus mitochondrial genomes.

4Until recently, little was known about the relative rates of substitution in plastid, mitochondrial, and nuclear genomes of non-land plant species. This is because the data needed for these types of analyses are difficult to generate, requiring nucleotide sequences from three different genetic compartments for at least two distinct “species” or populations. Moreover, the two species must be closely enough related that the silent site divergence, in all three compartments, has not reached saturation—not trivial requirements when considering that most protists are poorly sampled and poorly studied. However, improvements in high-throughput sequencing technologies and a growing interest in microbial eukaryotes, especially those from marine environments, mean that organelle and nuclear genomic data are accumulating for a variety of algal groups, allowing for accurate measurements of dsilent.

5Relative silent site substitution rate statistics are now available for various “primary” algae (i.e., those whose plastids descend directly from the endosymbiosis of a cyanobacterium), including green algae, red algae, and glaucophytes, as well from groups that acquired their plastids through eukaryotic–eukaryotic endosymbioses, such as haptophytes, stramenopiles, and apicomplexans, all of which have red–algal-derived plastids (table 1). In many cases, these data encompass all three genetic compartments, and are based on analyses of whole organelle genomes and at least 25 nuclear genes of diverse function (table 1). It is important to stress, however, that relative substitution rates do not necessarily reflect absolute substitution rates, which to calculate requires knowledge of the number of generations separating the species being compared. But relative rates do provide an estimate of the frequency of silent site substitutions among compartments within a species, and in cases were dsilent is exceptionally high or low in one compartment, it can be an indication of a high or low absolute rate of substitution.

6There exist different methods and models for estimating rates of nucleotide substitution. The available substitution rate statistics for plastid-bearing species were not all derived using the same methodologies. Here, I have tried to focus on studies that employed the maximum-likelihood (ML) method, which is considered to be among the most accurate methods available for measuring substitution rates, particular between distantly related sequences. In some cases, I reanalyzed published data sets using the ML method (see Materials and Methods). Consequently, the silent site substitution rates in table 1 and figure 1B were calculated using the ML approach, implemented through the program PAML v4. In some instances, however, I did not have easy access to the raw data sets that were used to calculate divergence. In such cases, I do not present the per-site substitution rate values, as they were not always calculated using the ML method. Instead, I simply record whether the observed silent site divergence in the mtDNA exceeded that of the ptDNA, making sure that the same methodologies were used for calculating divergence in both compartments (fig. 1A).

7The relative levels of dsilent differ substantially both among and within plastid-bearing lineages, but with the exception of land plants there is a near-universal tendency for plastid genomes to have lower rates of silent site substitution than their mitochondrial counterparts (table 1 and fig. 1). The same pattern is observed for other categories of nucleotide site as well. For instance, substitutions at nonsynonymous positions (dN) and regions coding for functional RNAs are consistently lower in plastid than in mitochondrial genomes (table 1). The ratio of nonsynonymous to synonymous substitutions (dN/dS), which can be used to gauge the intensity and directionality of natural selection, are generally quite low for both organelle genomes (table 1), indicating strong purifying selection on nonsynonymous sites—but see and for exceptions to this trend.

8In many instances, the silent site substitution rate differences between mitochondria and plastids are extreme (table 1 and fig. 1). In the haptophyte genus Phaeocystis—the members of which can form massive ocean blooms—dsilent of the mtDNA is on average 10 times that of the plastid. Isolates of the halophilic green alga Dunaliella salina show a 13-fold greater silent site substitution rate in the mitochondrion versus the plastid. And in the coccolithophore Emiliania huxleyi, the average dsilent of the mtDNA is ∼10 times that of the ptDNA, going up to ∼30 times greater when only intergenic sites are considered.

9In other cases, the differences between the mitochondrial and plastid compartments are less severe (table 1 and fig. 1). For example, in the red algal genus Porphyra and the glaucophyte genus Cyanophora, dsilent of the mtDNA is ∼4–5 times that of the ptDNA. And for the green algal lineages Mesostigma and Chlamydomonas, the mitochondrial and plastid genomes have similar rates of silent site substitution, although for Mesostigma the rate for the mtDNA is slightly higher than that for the ptDNA.

10Even certain nonphotosynthetic species display lower rates of substitution in their ptDNAs than their mtDNAs (fig. 1). For instance, a large-scale population genomics study of the malaria parasite Plasmodium falciparum (Apicomplexa), involving the sequencing of complete organelle DNAs from 711 distinct isolates, exposed ∼4 times more polymorphisms per silent site in the mtDNA as compared with the ptDNA. Similarly, organelle genome sequencing of geographical isolates of the apicomplexan parasite Babesia bovis uncovered no polymorphisms in the ptDNA, which is ∼35 kb long (∼15 kb of which are silent sites), whereas the mtDNA, which is ∼6 kb long with only ∼3 kb of silent sites, contained multiple polymorphisms.

11The substitution rate data from the nuclear compartment are less variable than those from the organelles, but as with the plastid-versus-mitochondrial comparisons, dsilent for the ptDNA is consistently lower or similar to that of the nucDNA (table 1 and fig. 1). This is true across a wide breadth of plastid-bearing eukaryotes, although for Porphyra red algae and the dinoflagellate Symbiodinium the ptDNA has a slightly higher silent site substitution rate than the nucDNA—but for the former, it is still much lower than the mtDNA.

12Given all of this, it is hard to ignore that for a large range of algae, including those with primary or secondary plastids as well as nonphotosynthetic species, ptDNAs consistently have lower levels of dsilent than mtDNAs. These trends contrast with those observed for many land plants, where plastids typically harbor higher levels of silent site divergence than mitochondria (table 1 and fig. 1).

13If we accept that the relative silent site divergence for genomes within a cell provides a gateway into the relative mutation rates of those genomes, then the data highlighted above have a clear meaning: Many algae have an mtDNA/ptDNA mutation rate ratio of >1, and such a ratio may be common for plastid-bearing eukaryotes as a whole (fig. 1). This is particularly apparent in lineages with red–algal-derived plastids, including haptophytes, stramenopiles, and apicomplexan parasites, where for some species the mtDNA mutation rate is estimated to be 30 times greater than that of the ptDNA (table 1).

14Land plants appear to be an exception among plastid lineages in that many species are predicted to have higher rates of mutation in their ptDNAs versus mtDNAs (table 1 and fig. 1). That said, a number of seed plants have recently been shown to have unprecedentedly high rates of silent site mtDNA substitution, including certain Silene and Ajuga species. In some of these taxa, elevated mitochondrial rates coincide with inflated ptDNA divergence, but for both Silene and Ajuga, the mitochondrial rates are higher than those of the plastid. Even for land plant mtDNAs with very low levels of dsilent, the substitution rate measurements are often based entirely on synonymous sites, and the divergence at noncoding sites can be much higher but often goes unrecorded because these regions can be difficult to align. This suggests that for land plants an mtDNA/ptDNA mutation rate ratio of >1 might not be as uncommon as once thought.

15Why do plastid genomes consistently exhibit lower rates of silent site substitution and have lower predicted mutation rates than their mitochondrial neighbors? Or, rather, why do mitochondria typically have higher mutation rates than plastids? In other respects, mitochondrial and plastid DNAs are very similar. Both are usually uniparentaly inherited and housed in energy-producing organelles, which evolved from the endosymbiosis of free-living bacteria more than a billion years ago; both are highly reduced and have lost or transferred most of their genes to the host nuclear genome; and both are dependent on nuclear-encoded, organelle-targeted proteins for the preservation of crucial biochemical pathways and for essential repair-, replication-, and expression-related functions. Moreover, in plants and algae, various DNA repair, replication, and recombination proteins targeted to the mitochondrion have plastid-targeted paralogs, and a significant proportion are targeted to both organelles.

16The nature and accuracy of these organelle-targeted DNA maintenance machineries can have a huge impact on organelle mutation rates. The fact that parts of these machineries are shared between mitochondria and plastids can explain why some species exhibit similar mtDNA and ptDNA mutation rates. But when considering that the origin and evolution of organelle DNA repair and recombination processes involves a complex history of gene transfer, co-option, duplication, and replacement events, it is not surprising that organelle mutation rates can vary both within and between species. Indeed, mitochondria boast some of the highest and lowest mutation rate estimates of any eukaryotic or bacterial genome.

17One explanation for the substitution rate patterns described here is that for a number of eukaryotes the fidelity and efficacy of mitochondrial maintenance machineries are more variable and capricious than those of plastids. Support for this hypothesis not only comes from the observation that substitution rates are often higher in mitochondrial versus plastid genomes, but also from the observation that mitochondrial chromosomes display a greater breadth of complexity and more severe genomic embellishments than those of plastids. It is thought that the propensity for mtDNAs to adopt such “extreme” architectures has its roots in wayward DNA maintenance pathways.

18Another difference between mitochondria and plastids, which could help explain why the former are prone to higher substitution rates, is that mtDNAs generally contain fewer genes than ptDNAs. Although speculative, a smaller gene complement could permit larger fluctuations in mutation rate because selection against a mutator allele would be proportional to its effect on the rate of deleterious mutations per genome, not per nucleotide.

19There are over 5,000 mitochondrial and plastid genome sequences in the National Center for Biotechnology Information database, making organelle genomes among the most highly sequenced chromosomes. From unraveling the population genetics of malaria parasites to developing biofuels to tracking the history of ancient Arctic vegetation, mtDNAs and ptDNAs are among the most important and widely used genetic markers. They have shaped our understanding of eukaryotic evolution, and been pivotal in the fields of archaeology, forensics, and medicine. It is therefore paramount that we understand the mutational processes impacting these genomes.

20Land plant studies have improved our knowledge of organelle genetics, but the patterns of organelle genome evolution in plants do not necessarily reflect those of other lineages. If it is true that for many microbial eukaryotes the plastid mutation rate is lower than that of the mitochondrion, it could mean that ptDNAs are a more suitable genome for wide-scale comparative analyses, such as those attempting to resolve relationships among distantly related groups or organisms. Conversely, mitochondrial genomes, with their proclivity toward higher mutation rates, could be useful for fine-scale genetic analyses, such as population genetic studies.

21Again, it is important to stress that the studies from which the mtDNA and ptDNA substitution rate data derive sometimes differed in the type and number of loci used and in the methodologies employed for calculating divergence. These differences as well as the differences in organelle genomic architecture within and among the species being compared should be taken into consideration when assessing the major trends presented in this study. As more data on relative rates emerge from poorly studied plastid-containing lineages, they will likely provide an even more dynamic picture of organelle and nuclear mutation rates. It will be interesting to see if algal lineages not included in this study, such as euglenophytes and chlorarachniophytes, both of which have green–algal-derived plastids, also have lower rates of silent site substitution in their plastid versus mitochondrial compartments. I predict that they will.

22Relative substitution rate data were derived or taken directly from the following sources and species/strain comparisons: B. bovis C9.1 versus B. bovis T2Bo; Chlamydomonas reinhardtii versus Chlamydomonas globosa SAG 7.73 (formerly called Chlamydomonas incerta SAG 7.73); Cyanophora paradoxa strain NIES-763 versus CCMP329; D. salina CCAP 19/18 versus D. salina CONC-001; E. huxleyi CCMP373 versus E. huxleyi CCMP1516; Gephyrocapsa oceanica (dozens of geographical isolates); Glaucocystis spp. (∼10 geographical isolates); Heterosigma akashiwo CCMP452 versus H. akashiwo NIES-293; land plants (27 different species); Mesostigma viride strain NIES-296 versus SAG 50-1; Micromonas pusilla CCMP1545 versus Micromonas sp. RCC299 (supplementary table S1, Supplementary Material online); Nannochloropsis gaditana CCMP527 versus Nannochloropsis salina CCMP1776 (supplementary table S2, Supplementary Material online); Ostreococcus sp. RCC809 versus Ostreococcus tauri OTTH0595 (supplementary table S3, Supplementary Material online); Phaeocystis antarctica CCMP1374 versus Phaeocystis globosa Pg-G(A); P. falciparum (hundreds of geographical isolates); Porphyra umbilicalis UTEX LB 2951 versus Porphyra purpurea strain “Avonport”; Pseudochattonella sp. JG8 versus Pseudochattonella farcimen UIO109 versus Pseudochattonella verruculosa NIES-670; and Symbiodinium sp. A2 versus C90. Note: The data sets for Emiliania, Heterosigma, and Symbiodinium were reanalyzed using the ML method, as described below.

23In cases where the substitution rate statistics were recalculated or did not come directly from the literature (e.g., Micromonas, Nannochloropsis, and Ostreococcus; supplementary tables S1–S3, Supplementary Material online), they were estimated as follows. Organelle and nuclear genes were aligned with MUSCLE, implemented through Geneious v7.1.4 (Biomatters Ltd, Auckland, NZ), using default settings. Synonymous and nonsynonymous substitutions were measured with the CODEML program of PAML v4.3, employing the ML method, the F3 × 4 codon model of (options: seqtype = 1, runmode = −2, and CodonFreq = 2 in the codeml.ctl file), and making the proper adjustments for changes in the genetic code. Substitutions in non–protein-coding regions were estimated with BASEML of PAML, using the HKY85 model. Significance levels (P values) of mtDNA versus ptDNA substitution rate differences were taken from the primary literature or calculated using the t-test as implemented in Microsoft Excel (Mac) 2011 v14.4.4.

24Supplementary tables S1–S3 are available at Genome Biology and Evolution online (http://www.gbe.oxfordjournals.org/).