Flipping the switch on some of the slowest mutating genomes: Direct measurements of plant mitochondrial and plastid mutation rates in msh1 mutants

Abstract

Plant mitochondrial and plastid genomes have exceptionally slow rates of sequence evolution, and recent work has identified an unusual member of the MutS gene family ("plant MSH1") as being instrumental in preventing point mutations in these genomes. However, the effects of disrupting MSH1-mediated DNA repair on "germline" mutation rates have not been quantified. Here, we used Arabidopsis thaliana mutation accumulation (MA) lines to measure mutation rates in msh1 mutants and matched wild type (WT) controls. We detected 124 single nucleotide variants (SNVs: 49 mitochondrial and 75 plastid) and 668 small insertions and deletions (indels: 258 mitochondrial and 410 plastid) in msh1 MA lines at a heteroplasmic frequency of ≥ 20%. In striking contrast, we did not find any organelle mutations in the WT MA lines above this threshold, and reanalysis of data from a much larger WT MA experiment also failed to detect any variants. The observed number of SNVs in the msh1 MA lines corresponds to estimated mutation rates of 6.1 × 10-7 and 3.2 × 10-6 per bp per generation in mitochondrial and plastid genomes, respectively. These rates exceed those of species known to have very high mitochondrial mutation rates (e.g., nematodes and fruit flies) by an order of magnitude or more and are on par with estimated rates in humans despite the generation times of A. thaliana being nearly 100-fold shorter. Therefore, disruption of a single plant-specific genetic factor in A. thaliana is sufficient to erase or even reverse the enormous difference in organelle mutation rates between plants and animals.

Fig 1. Overview comparison between A. thaliana WT (left) and msh1 mutant (right) MA lines.

(A) Examples of phenotypic variation arising in the msh1 MA lines. F9 progeny from sequenced F8 individuals are pictured. Nine seeds were sown per pot, but not all seeds germinated. (B) Mitochondrial (top) and plastid (bottom) variants detected by MA line sequencing. Each column represents an MA line. Each row represents a nucleotide position in the genome. Only positions with variants are shown, so spacing on the vertical axis is not a measure of distance. The color intensity of horizontal lines in each of cell of the heatmaps represents variant frequency, which was normalized for background sequencing error rates by subtracting the mean variant frequency averaged across all WT MA lines. As indicated by the almost completely white panels, we did not detect any variants that passed filtering criteria in the WT MA lines. We detected 892 SNVs and small indels in the msh1 mutant MA lines.

Fig 2. Mitochondrial (left) and plastid (right) mutation spectra in A. thaliana msh1 MA lines.

For mutation rate calculations, SNVs were weighted by heteroplasmic frequency and normalized to the corresponding number of GC or AT base-pairs to account for biased nucleotide compositions in the respective genomes. Note that the y-axis scales differ between the two panels.

Fig 3. The relationship between homopolymer types and indel bias in A. thaliana msh1 MA lines.

(A) The bars represent the counts of short indels observed in A/T homopolymers (left panel) and G/C homopolymers (right panel). Within each panel, mitochondrial indels are shown to the left of the center line, and plastid indels are shown to the right. Both genomes exhibit a deletion bias in A/T homopolymers and an insertion bias in G/C homopolymers. Indels >3 bp in length were extremely rare (S3 Dataset) and not shown in this plot. (B) Frequency of A/T homopolymers (left panel) and G/C homopolymers (right panel) in the A. thaliana mitochondrial (gold circles) and plastid (green triangles) genomes. The fact that the plastid genome is dominated by A/T homopolymers may explain why its overall indel spectrum is deletion-biased given that A/T homopolymers appear to be more deletion-prone than G/C homopolymers in both organelles of msh1 MA lines.