Heterogeneous polymerase fidelity and mismatch repair bias genome variation and composition

Abstract

Mutational heterogeneity must be taken into account when reconstructing evolutionary histories, calibrating molecular clocks, and predicting links between genes and disease. Selective pressures and various DNA transactions have been invoked to explain the heterogeneous distribution of genetic variation between species, within populations, and in tissue-specific tumors. To examine relationships between such heterogeneity and variations in leading- and lagging-strand replication fidelity and mismatch repair, we accumulated 40,000 spontaneous mutations in eight diploid yeast strains in the absence of selective pressure. We found that replicase error rates vary by fork direction, coding state, nucleosome proximity, and sequence context. Further, error rates and DNA mismatch repair efficiency both vary by mismatch type, responsible polymerase, replication time, and replication origin proximity. Mutation patterns implicate replication infidelity as one driver of variation in somatic and germline evolution, suggest mechanisms of mutual modulation of genome stability and composition, and predict future observations in specific cancers.

Figure 1.

Genome-wide replication error positions, rates, and MMR efficiencies. Transitions are indicated by blue shades (light to dark: AT→GC, CG→TA), transversions by reds (light to dark: AT→TA, TA→GC, CG→AT, CG→GC), deletions by greens (light to dark: −A/T, −G/C, multibase), and insertions by purples (light to dark: +A/T, +G/C, multibase). (A) Positions of more than 43,000 mutations, from all eight strains used in this study, plotted along the 16 S. cerevisiae chromosomes. ([‡] Overlaid black boxes are regions excluded from mutation calling; see Supplemental Methods.) (B) Mutation rates, corrected for genomic GC content from MMR-deficient strains. ([N] Mutation count pooled by strain.) (C) As per B, but for MMR-proficient strains. (D) MMR correction efficiencies for substitution errors in four polymerase allelic backgrounds are ratios of MMR-deficient rates to MMR-proficient rates. ([a] Calculated from <10 observed mutations in MMR-proficient strains; see also Supplemental Fig. S2.)

Figure 2.

Polymerase and strand specificity of replication errors. Select polymerase-biased complementary mismatch pairs and mismatch motifs. (A) Schematic example of adjacent replication origins and their effects on lagging-strand–biased mutagenesis. The T-dG:A-dC ratio in vitro is from Nick McElhinny et al. (2008). (B) Diagrams are example preferences for complementary mutation pathways. In most cases, the three variant polymerases have the same preference (black arrows). Disagreements are color-coded by polymerase variant: Pol alpha (pol1-L868M), red; Pol delta (pol3-L612M), green; and Pol epsilon (pol2-M644G), blue. Plots are the fraction of each substitution mutation (f_i) paired with its complement as a function of relative distance between adjacent replication origins. ([N] Mutation count pooled by strain, excluding mutations in origins.) (C) As for B, but for those mutation types observed >50 times in individual MMR-proficient strains. See also Supplemental Figures S3 and S4.

Figure 3.

Sequence specificity of replication errors in the absence of MMR. (A) Nucleotide fractions and sequence logos (Schneider and Stephens 1990; Crooks et al. 2004) for five bases upstream of and downstream from mutations resulting from presumed C-dT mispairs, as calculated from sequences flanking mutations near replication origins (see example schematic). Expected fractions assume 38% G + C content. ([MM] Mismatch position; [N] mutation count pooled by strain.) (B) As per A, but for mutations resulting from presumed G-dT mispairs. (C) Example: An incoming mismatched nucleotide (red) stacks with adjacent pyrimidines (green) in the nascent strand, as indicated by logos in A.

Figure 4.

Variation in mutation rates near nucleosome positions and genes. Mutation rates (blue indicates transitions; red, transversions; green, one-base deletions) plotted versus either (A) the distance from either the nearest nucleosome dyad (in base pairs) or (B) from the nearest coding start (left) or end site (right; in kilobase pairs). Asterisks denote significantly different substitution rates between indicated regions (Pol alpha, red; Pol delta, green; Pol epsilon, blue). Percentages denote the magnitude of substitution excesses. Shaded areas are DNA regions: nucleosome-bound (yellow), shorter and longer than average linkers (orange and red, respectively), intergenic (green), 5′ nucleosome-free (blue), and coding (purple).

Figure 5.

Indel rates in homopolymers in the absence of MMR. (A) A comparison of nucleosome density (red; relative to the +1 peak) and pol2-M644G mmr- AT deletion rates (blue) as a function of distance from translation start sites. (B) Homopolymer densities (relative to maximum density) for various homopolymer lengths as a function of distance from translation start sites.