Muller's Ratchet and compensatory mutation in Caenorhabditis briggsae mitochondrial genome evolution
© Howe and Denver. 2008
Received: 13 November 2007
Accepted: 26 February 2008
Published: 26 February 2008
Skip to main content
© Howe and Denver. 2008
Received: 13 November 2007
Accepted: 26 February 2008
Published: 26 February 2008
The theory of Muller' Ratchet predicts that small asexual populations are doomed to accumulate ever-increasing deleterious mutation loads as a consequence of the magnified power of genetic drift and mutation that accompanies small population size. Evidence for Muller's Ratchet and knowledge on its underlying molecular mechanisms, however, are lacking for natural populations.
We characterized mitochondrial genome evolutionary processes in Caenorhabditis briggsae natural isolates to show that numerous lineages experience a high incidence of nonsynonymous substitutions in protein-coding genes and accumulate unusual deleterious noncoding DNA stretches with associated heteroplasmic function-disrupting genome deletions. Isolate-specific deletion proportions correlated negatively with nematode fecundity, suggesting that these deletions might negatively affect C. briggsae fitness. However, putative compensatory mutations were also observed that are predicted to reduce heteroplasmy levels of deleterious deletions. Paradoxically, compensatory mutations were observed in one major intraspecific C. briggsae clade where population sizes are estimated to be very small (and selection is predicted to be relatively weak), but not in a second major clade where population size estimates are much larger and selection is expected to be more efficient.
This study provides evidence that the mitochondrial genomes of animals evolving in nature are susceptible to Muller's Ratchet, suggests that context-dependent compensatory mutations can accumulate in small populations, and predicts that Muller's Ratchet can affect fundamental evolutionary forces such as the rate of mutation.
Evolutionary theory predicts that mutational decay is inevitable for small asexual populations, provided deleterious mutation rates are high enough. Such populations are expected to experience the effects of Muller's Ratchet [1, 2] where the most-fit class of individuals is lost at some rate due to chance alone, leaving the second-best class to ultimately suffer the same fate, and so on, leading to a gradual decline in mean fitness. The mutational meltdown theory [3, 4] built upon Muller's Ratchet to predict a synergism between mutation and genetic drift in promoting the extinction of small asexual populations that are at the end of a long genomic decay process. Regardless of reproductive mode, mitochondrial genomes from most animal species are expected to be particularly sensitive to Muller's Ratchet due to their uniparental inheritance, high mutation rates and lack of effective recombination [3, 5, 6]. The genomic decay effects of Muller's Ratchet have been observed in laboratory evolution experiments with abiotic RNA molecules , biotic RNA viruses , bacteria  and yeast . Indirect evidence for the effects of Muller's Ratchet in nature has resulted from studies on the long-term effects of reduced population sizes on genetic diversity and fitness in amphibians , greater prairie chickens [12, 13] and New Zealand avifauna . Molecular evidence for Muller's Ratchet has resulted from analyses of deleterious tRNA gene structures encoded by mitochondrial genomes  and analyses of Drosophila sex chromosome evolution . However, direct knowledge on the susceptibilities of natural populations to Muller's Ratchet and the molecular mechanisms underlying this process remain enigmatic.
Caenorhabditis briggsae, like Caenorhabditis elegans, is a self-reproducing hermaphroditic nematode species that also produces males capable of outcrossing with hermaphrodites. Analyses of linkage disequilibrium patterns in C. briggsae natural isolates suggest a very low outcrossing rate of ~3.9 × 10-5 . The same study reported population subdivision between C. briggsae strains collected in temperate localities versus those from tropical regions and nuclear silent-site nucleotide diversity (πS) for the tropical isolates was estimated at 2.7 × 10-3 – a number highly similar to global estimates for C. elegans . The C. briggsae isolates from temperate localities, however, showed a remarkably lower mean πS value of 4.0 × 10-5. A direct estimate of the neutral base substitution mutation rate (9.0 × 10-9 per site per generation) is available from C. elegans mutation-accumulation lines  that can be used along with πS data to estimate the effective population size (N e ) . Assuming a common mutation rate between C. elegans and C. briggsae, N e is estimated to be ~63,000 for C. briggsae tropical isolates. For the C. briggsae temperate isolates, a much smaller N e of ~1,000 is estimated. Based on these and other observations, it is hypothesized that C. briggsae only recently (in the last few hundred years) colonized temperate latitudes from small founding populations . Furthermore, there is evidence for a ~2-fold elevated mutation rate in C. briggsae as compared to C. elegans  that leads to correspondingly lower N e estimates for C. briggsae : ~31,500 for tropical populations and ~500 for temperate populations. The combination of very low outcrossing rates, small N e and high mutation rates are expected to render C. briggsae natural mitochondrial lineages susceptible to the effects of Muller's Ratchet-associated deleterious mutation accumulation.
To probe for the effects of Muller's Ratchet in C. briggsae natural populations, we sequenced nearly complete mitochondrial genomes from multiple geographically diverse C. briggsae natural isolates and characterized molecular evolutionary processes by comparing nucleotide diversity patterns in mitochondrial DNA (mtDNA) protein-coding genes between the temperate- and tropical-clade C. briggsae isolates, characterizing heteroplasmic genome deletions using quantitative real-time PCR (qPCR) approaches and evaluating correlations of various natural mitochondrial genome haplotypes with nematode fecundity and fitness.
C. briggsae natural isolate origins and fecundities.
Johannesburg, S. Africa
With the exception of the AT-region, the vast majority of other animal mitochondrial genomes examined are devoid of similar extensive stretches of noncoding DNA . Based on a recent population-genetic theory on the evolution of genome complexity  that predicts higher noncoding DNA accumulation probabilities in species of small population size due to a magnified power of genetic drift, the accumulation and persistence of ψND5 elements in C. briggsae mitochondrial genomes is consistent with the hypothesis that natural C. briggsae lineages experience very small population sizes that are essential for genomic decay associated with Muller's Ratchet. ψND5-1, however, occurs in both C. briggsae and the obligately outcrossing species Caenorhabditis sp. n. 5. The ψND5-1 element might have accumulated in the (presumably gonochoristic) ancestor of C. briggsae and Caenorhabditis sp. n. 5 as a consequence of Muller's Ratchet since nematode mtDNA is inherited through the hermaphrodite/female lineage and is not known to undergo intergenomic recombination. Only two natural isolates from China are presently known and available Caenorhabditis sp. n. 5 ; thus, we are presently unable to effectively characterize the population-genetic environment in which Caenorhabditis sp. n. 5 evolves. Future comparative analyses of ψND5-1 molecular evolution between these two species, however, might provide important insights into the role of outcrossing in eliminating deleterious mutations from natural populations.
Upon further analyzing the ψND5-2 region in the C. briggsae natural isolates, we found that some strains harbor a large heteroplasmic mitochondrial genome deletion (871–887 bp, depending on isolate) that eliminated the 3' end of ψND5-2 and the 5' end (first 786 bp) of the ND5 protein-coding gene (Figure 1 and 3A). Directly repeated DNA sequence tracts in ND5 and the upstream ψND5-2 element flank the heteroplasmic ND5 deletion allele and are likely to play a causative role in promoting the deletions as direct repeats are also associated with heteroplasmic mitochondrial DNA deletions in C. elegans mutation-accumulations lines  and aging nematodes . The observed deletion is expected to strongly and negatively affect ND5 protein-coding function as the deleted sequences encode more than 200 ND5 amino acids, 34 of which are conserved in C. elegans, Drosophila melanogaster and humans. The ψND5-2 region and associated heteroplasmic ND5 gene deletion was initially discovered in long PCR amplifications of C. briggsae natural isolate mitochondrial genome segments. The observation of the heteroplasmic deletion allele across multiple independent long PCR reactions (3 to 5 kb amplicons) that span many mtDNA genes provides strong evidence that the deletion is not a nuclear sequence of mitochondrial origin (numt), but rather a heteroplasmic mtDNA allelic variant. Furthermore, we have searched for the deletion-bearing sequences in the published C. briggsae nuclear genome sequence and found no evidence for the presence of a candidate numt. We also found no evidence for heteroplasmic deletions associated with ψND5-1.
Although the conventional PCR approach suggested extensive among-isolate variation in deletion heteroplasmy levels, a qPCR approach was applied to provide a more quantitative characterization of variation in the proportion of ND5 deletion-bearing genomes to the total among the isolates. qPCRs were carried out using the same genomic DNA samples that were used for the conventional PCR assays (four individuals per isolate). One set of qPCR primers amplified a ND5 region present only in intact genomes (5' end of the gene) and a second amplified a 16S rRNA region present in both deletion-bearing and intact genomes for which there is no evidence of heteroplasmy (see Methods). The nematode-specific proportions of genomes bearing ND5 deletions were calculated by dividing the estimated abundance of intact genomes by that of the total mitochondrial genomes, and then subtracting that number from one. Estimated deletion-bearing genome proportions from qPCR results correlated positively with conventional PCR band scoring data (Spearman rank correlation = 0.74, P < 10-15) and revealed substantial among-isolate variation in ND5 deletion heteroplasmy levels (Figure 3B). ND5 deletion-bearing genomes were observed to accumulate in both the temperate and tropical intraspecific C. briggsae clades. Two isolates (HK105, VT847) showed remarkably high deletion genotype proportions greater than 40%. Interestingly, these two isolates were collected from islands (Table 1) where populations might have been recently established from very small populations where drift is expected to play a particularly dominant role . However, other island isolates (e.g. HK104, JU439) displayed low deletion proportions. Although it cannot be ruled out that changes in heteroplasmy levels occurred while nematodes were maintained in laboratory culture after collection from nature, most (at least 19/24) of the C. briggsae isolates considered here constitute strains recently collected from the wild  that did not spend substantial time in laboratory culture before being stored as cryogenically-preserved frozen stocks. Although similar heteroplasmic mitochondrial genome deletions have been observed in laboratory-bottlenecked C. elegans mutation-accumulation lines  and are associated with rare mitochondrial myopathies in humans , the widespread occurrence and persistence of such deleterious mitochondrial gene deletions in natural populations is unprecedented.
To investigate the potential effects of deletion-bearing molecules on organismal fitness, we next assayed variation in lifetime fecundity in the C. briggsae natural isolates. The mean fecundity observed across all 24 C. briggsae isolates assayed was 138.6 progeny, a number remarkably smaller than that for C. elegans (273.6 progeny ), but consistent with previously-reported fitness disparities between these two species . A significantly negative correlation (Pearson's correlation coefficient = -0.41, P < 0.05) was observed for isolate-specific deletion genotype proportions and nematode fecundities. Furthermore, the two isolates with the highest mean fecundity values were the two from the Kenya clade (ED3092, ED3101 – see Table 1) that do not encode ψND5-2 or experience associated ND5 deletion events. These observations suggest that the accumulation of deletion-bearing mitochondrial genomes might strongly and negatively affect fitness in C. briggsae natural populations. The present study, however, is unable to disentangle the fitness effects of mitochondrial and nuclear loci – it is possible that there exist deleterious nuclear loci that are responsible for reduced fitness in these lines rather than the high deletion levels. We expected to observe higher ND5 deletion proportions in the temperate-clade C. briggsae isolates as compared to those of the tropical clade as a consequence of reduced efficiency of natural selection associated with smaller N e estimates in the former group. However, no significant differences in ND5 deletion heteroplasmy levels were observed between these two groups (P > 0.9, two-tailed t -test).
These ψND5-2 substitutions that likely promote reduced ND5 deletion proportions (DRSeq2, DRSeq3) are observed in the C. briggsae intraspecific temperate clade where N e estimates are much lower than that of the tropical clade  and, thus, natural selection for compensatory mutations is expected to be less efficient. All tropical clade isolates analyzed encode DRSeq1. However, recent work in phage mutation-accumulation experiments suggests an increase in the relative incidence of beneficial mutations as fitness decreases in association with deleterious mutation accumulation . Our results provide empirical support for this hypothesis in natural animal populations – any mutations in ψND5-2 that result in reduced incidences of direct repeat-induced ND5 deletion events are expected to be advantageous. We propose that the occurrence of one highly deleterious mutational event (i.e. the ψND5-2 insertion and associated ND5 deletion events) presents a novel mutational target in the mitochondrial genome where most all new mutations are expected to reduce homology between ψND5-2 and ND5, thus reducing the likelihood of deleterious mitochondrial genome deletions. Consequently, such new mutations that result in increased divergence between ψND5-2 and ND5 can be characterized as advantageous compensatory mutations. The fact that these candidate compensatory mutations were observed in the temperate clade (where N e estimates are relatively small and natural selection is expected to be relatively inefficient) and not the tropical clade (where N e estimates are relatively large and natural selection is expected to be relatively more efficient) is puzzling. Although the evolutionary causes of these putative compensatory mutations in the temperate clade remains mysterious, it is possible that the heteroplasmic deletions are more deleterious in the temperate clade as compared to the tropical clade and that there is a corresponding difference in the selection coefficients associated with the candidate compensatory changes observed in the temperate isolates. Another (probably unlikely) possibility is that the increased fixation probability of these mutations in smaller populations (i.e. temperate-clade isolates) outweighs the higher population mutation rate experienced by larger populations (i.e. tropical-clade isolates). Finally, it is possible that the N e estimates based on nuclear nucleotide diversity data do not accurately reflect the current population-genetic environments experienced by these two intraspecific clades.
After discovery of these putative compensatory mutations, we again searched for elevated ND5 deletion levels in the temperate as compared to the tropical isolates, this time including only those temperate-clade isolates that encode DRSeq1. Although the estimated ND5 deletion levels were on average greater in DRSeq1-encoding temperate isolates (23.0%) versus that of the tropical isolates (13.1%), the difference was not significant (P > 0.1, two-tailed t -test). Fecundities were also highly similar (Table 1) and not significantly different (P > 0.5, two-tailed t -test). Unfortunately, only three of the analyzed temperate-clade isolates encode DRSeq1 (Figure 4), thereby severely limiting our power to detect differences in the propensities of these two intraspecific clades of different N e estimates to accumulate deleterious ND5 deletions. Future efforts surveying a greater number of C. briggsae natural isolates will be required to further probe for the differential effects of Muller's Ratchet between these two clades at this locus.
In addition to investigating the evolutionary dynamics of the unusual ψND5 elements, the nearly complete C. briggsae mitochondrial genome sequences also offer the opportunity to analyze molecular evolution in the rest of the genome and, in particular, the twelve mtDNA protein-coding genes. The relative efficiency of natural selection in preventing the accumulation of deleterious mutations is often quantified by considering the relative rates of fixed nonsynonymous and synonymous substitutions in protein-coding genes [20, 33]. We applied a similar approach to probe for differences between the temperate- (small N e ) and tropical-clade (large N e ) C. briggsae isolates in their propensities for accumulating nonsynonymous substitutions. However, because we are comparing non-fixed nucleotide polymorphism patterns between two intraspecific lineages, we applied the ratio of nucleotide diversity at nonsynonymous sites (πN) to that at synonymous sites (πS) as a measure of relative susceptibilities to (presumed) deleterious nonsynonymous mutation accumulation, using DnaSP . For ten individual mtDNA protein-coding genes, πN/πS values were compared between isolates of the C. briggsae temperate clade to those of the C. briggsae tropical clade. A concatenated data set including codons from all mtDNA protein-coding genes was also analyzed and further compared to a comparable data set from 17 C. elegans natural isolates where there is no evidence of global phylogeographic population structure .
The findings presented here show that in nature C. briggsae mitochondrial genomes are susceptible to the accumulation of unusual deleterious mutations that include the insertion of large noncoding DNA stretches (Figures 1 and 2), function-disrupting heteroplasmic genome deletions (Figure 3) and nonsynonymous substitutions in protein-coding genes (Figure 5). This pattern of molecular evolution is consistent with strong roles for mutation and random genetic drift in shaping mitochondrial genome evolution in natural populations of this species. These observations, coupled with the strong potential negative impact of heteroplasmic ND5 deletions on organismal fitness, also suggest that C. briggsae mitochondrial genomes evolving in nature are vulnerable to the effects of Muller's Ratchet. Although we did observe higher πN/πS ratios for protein-coding genes in the temperate (small N e estimate) versus tropical (larger N e estimate) isolates, we were unable to detect a role for Muller's Ratchet in promoting higher deleterious ND5 deletion levels in the temperate clade. However, we also observed the accumulation of putative compensatory mutations (DRSeq2) in temperate-clade C. briggsae lineages (Figure 4) that likely reduce ND5 deletion levels. The DRSeq2 mutations can only be considered beneficial, however, in the context of the preexisting deleterious ψND5-2 insertion mutation – in fact, virtually any mutation in ψND5-2 that reduces its homology to ND5 is likely to be beneficial to some extent.
mtDNA sequencing and PCR were performed as previously described [6, 22], with the exception that mitochondrial genome sequences were initially amplified as four overlapping PCR products 3–5 kb in size each using the Expand Long Range PCR kit (Roche). e2TAK (Takara) proofreading DNA polymerase was used for all conventional PCRs. For single-worm DNA extractions, individual worms were picked at the L1 larval stage and digested in 18 μL of lysis buffer [6, 22]. 1 μL was then used for each PCR. In C. elegans, L1-stage nematodes are estimated to harbor ~25,000 mitochondrial genomes . MEGA4 was used for sequence alignments and phylogenetic analyses (both neighbor-joining and maximum parsimony approaches were applied) . The maximum-composite likelihood molecular evolution model was used for neighbor-joining analysis. 1,000 bootstrap replicates were performed for all phylogenetic statistical testing. Primer sequences used for PCR and DNA sequencing are available upon request. DNA sequences generated for this study were submitted to Genbank under accession numbers EU407780–EU407805.
qPCR was carried out on an Applied Biosystems 7300 Real Time PCR machine using iTaq SYBR Green Supermix with ROX (Bio-Rad). 1 uL of genomic DNA (diluted 1:5 in water) was used for each qPCR analysis. To estimate the total amount of mtDNA in a sample (both intact and deletion-bearing), control primers were designed to amplify a 102 bp region of the small ribosomal RNA subunit that displayed no evidence of heteroplasmy. A second set of primers was designed to amplify 101 bp in the 5' end of the ND5 gene, within the deleted region. This product is not expected to amplify only in intact genomes. qPCR data were analyzed to estimate the abundances of the two genome types using the linear regression approach offered by the LinRegPCR software . ND5 locus products were found to amplify more efficiently than ribosomal RNA products; thus, all ND5 qPCR values were normalized to account for this disparity. Individual deletion genotype proportions were calculated by dividing the estimated abundance of intact genomes by that of the total mitochondrial genomes, and then subtracting that number from one. Four individual L1-stage nematodes were analyzed per natural isolate – the same nematode genomic DNA samples used for conventional PCR assays.
C. briggsae isolate-specific fecundities were estimated by counting the numbers of progeny produced by individual hermaphrodite worms using standard methods involving Toluidine dye-stained plates . The fecundity assay was carried out on standard OP50 Escherichia coli -seeded NGM agar plates at 20°C. Progeny production was measured for four independent nematodes per isolate. Prior to testing for relationships between fecundity and qPCR data using the Pearson's correlation approach, both data sets were confirmed to fit a normal distribution using the Kolmogorov-Smirnov test for continuous variables.
Aligned mtDNA protein-coding gene sequences were subjected to nucleotide diversity analyses (πN/πS) using DnaSP v4.0 . The genetic code in DnaSP was set to that for flatworm mtDNA which is identical to that for nematode mtDNA . πN and πS were individually calculated for each of the twelve mtDNA protein-coding genes in addition to a concatenated data file that contained all mtDNA protein-coding gene codons – 3,414 total. A C. elegans concatenated mtDNA file (n = 2,785 codons) was also analyzed – data derived from a previous study  and two newly-sequenced genomes from strain JU258 (Madeira Islands, Portugal) and PS2025 (Altadena, CA, USA). Estimates of nucleotide diversity based on theta and the Jukes-Cantor method of calculating π  each yielded highly similar results to those presented in Figure 5. Additional file 3 reports individual πN and πS values.
effective population size
nuclear sequence of mitochondrial origin
quantitative real-time PCR
nonsynonymous-site nucleotide diversity
synonymous-site nucleotide diversity
We thank P. E. Chappell and C. P. Goodall for assistance with qPCR experiments, M. Dasenko and the OSU Center for Genome Research and Biocomputing for assistance with DNA sequencing, M. Ailion, A. Burnell, E. Dolgin, M.-A. Felix and the Caenorhabditis Genetics Center for providing C. briggsae natural isolate strains, W. K. Thomas for providing the C. remanei mitochondrial genome sequence, and C. F. Baer, E. Bakker, A. Cutter, L. A. Dyal, and three anonymous reviewers for comments and advice. We are grateful to the National Institutes of Health and OSU Computational and Genome Biology Initiative for funding support.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.