Mutator suppression and escape from replication error-induced extinction in yeast

Abstract

Cells rely on a network of conserved pathways to govern DNA replication fidelity. Loss of polymerase proofreading or mismatch repair elevates spontaneous mutation and facilitates cellular adaptation. However, double mutants are inviable, suggesting that extreme mutation rates exceed an error threshold. Here we combine alleles that affect DNA polymerase δ (Pol δ) proofreading and mismatch repair to define the maximal error rate in haploid yeast and to characterize genetic suppressors of mutator phenotypes. We show that populations tolerate mutation rates 1,000-fold above wild-type levels but collapse when the rate exceeds 10⁻³ inactivating mutations per gene per cell division. Variants that escape this error-induced extinction (eex) rapidly emerge from mutator clones. One-third of the escape mutants result from second-site changes in Pol δ that suppress the proofreading-deficient phenotype, while two-thirds are extragenic. The structural locations of the Pol δ changes suggest multiple antimutator mechanisms. Our studies reveal the transient nature of eukaryotic mutators and show that mutator phenotypes are readily suppressed by genetic adaptation. This has implications for the role of mutator phenotypes in cancer.

Figure 1. Evidence for a threshold of error-induced extinction.

A) Entry into error-induced extinction. Mutated pol3 alleles were introduced into haploid MSH6 and msh6Δ yeast by plasmid shuffling (Figure S1), and mutation rates were measured by fluctuation assays and calculated using the maximum likelihood method. Each bar represents the spontaneous mutation rate, expressed as canavanine-resistant (Can^r) mutants per cell division, conferred by a specific POL3 allele in MSH6 or msh6Δ cells. Mutation rate values (x 10⁻⁷) are indicated on each column. Error bars show 95% confidence intervals. WT, wild-type POL3; black, MSH6; gray, msh6Δ; X, no growth. B) Synthetic lethality of strong pol3 mutator alleles with msh6Δ. Serial dilutions of haploid yeast containing POL3–URA3 and pol3–LEU2 plasmids were plated on SC FOA medium to select for cells that spontaneously lost POL3–URA3. Similar numbers of cells (∼10⁵, 10⁴ and 10³) were plated for each set of alleles in the MSH6 and msh6Δ strains. Failed growth of msh6Δ cells carrying pol3-D407A or pol3-01 indicates synthetic lethality (right two panels). pol3-F406A and pol3-Y516F also failed to support colony formation in msh6Δ cells (not shown). Note the small size of pol3-D463A msh6Δ colonies.

Figure 2. Escape from error-induced extinction.

A) Emergence of colonies that escape pol3-01 msh6Δ synthetic lethality. Each segment of an FOA-containing SC plate (eight segments per plate) was streaked with an individual colony of POL3–LEU2 POL3–URA3 msh6Δ (left) or pol3-01–LEU2 POL3–URA3 msh6Δ (right) cells to select for loss of the POL3–URA3 plasmid (see Figure S1). Resultant POL3 msh6Δ cells formed abundant visible colonies (left), whereas pol3-01 msh6Δ cells did not (right). Colonies that escape pol3-01 msh6Δ synthetic lethality (eex mutants) arose at low frequency near the outer margins of the plate (circled) where cell densities were highest. Similar results were seen when pol3-F406A or pol3-D407A were shuffled into msh6Δ cells (not shown). B) Antimutator effects of eex alleles encoding second-site changes in Pol δ. Rates of spontaneous mutation to canavanine-resistance (Can^r) conferred by pol3-01, pol3-F406A or pol-D407A alone (filled symbols) and combined with intragenic eex alleles (open symbols) were determined in MSH6 cells. Downward arrows illustrate the reduction in mutation rates (i.e., the antimutator effect) caused by the second-site, amino-acid substitutions indicated beneath each datum point. Error bars show 95% confidence intervals. C) Representative plate from large-scale screen for eex mutants. Approximately 10⁶ cells from multiple independent pol3-01–LEU POL3–URA msh6Δ parent colonies were plated separately in ∼1-cm spots on FOA-containing SC medium. LEU-only and POL3–LEU plasmids were also shuffled into msh6Δ cells as controls. FOA-resistant colonies arose at varied frequencies from each parent clone. Insert, magnified view showing colonies that are candidate eex mutants. plasmid, LEU-only plasmid with no POL3 gene. D) Mutation rates of eex mutants. Rates of spontaneous mutation to FOA-resistance (FOA^r) were measured in a MMR-proficient strain with a chromosomal URA3 reporter gene. Each datum point represents a different POL3 allele. Mutation rates were determined from multiple independent fluctuation analyses of each allele. Confidence intervals (95%) are shown for POL3 and pol3-01. Mutation rates and confidence intervals of individual eex alleles are in Table 1 and Table 2.

Figure 3. Amino-acid changes in Pol δ eex mutants.

Aligned sequences of five B-family DNA polymerases: Saccharomyces cerevisiae Pol δ (S.c. pol d), Mus musculus Pol δ (M.m. pol d), Thermococcus gorgonarius (T.g. pol B), bacteriophage T4 (T4 pol), and herpes simplex virus 1 (HSV1 pol). Secondary structural elements of yeast Pol δ are indicated below the alignment and color coded to depict their domain locations (see Figure 4A): rectangles, α-helices; arrows, β-strands; solid lines, loops; dotted lines, structure not solved. Conserved polymerase (Pol) and exonuclease (Exo) motifs are framed , , . Amino-acid substitutions of interest in yeast Pol δ are placed underneath the alignment at the relevant positions, highlighted according to the following scheme: no highlight, pol3-01,eex mutations; green, pol3-D407A,eex mutations (R470C and T655A in one mutant, T711P in another); orange, pol3-F406A,eex mutations (three substitutions in the same mutant); blue, A894G mutation that rescued slow growth of pol3-D463A msh6Δ cells; yellow, pol3-01 (D321A,E323A); gray, pol3-t (D643N) and G447S (previously identified antimutator alleles; [121], [122]). Residues that increase polymerase fidelity when mutated in T4 or HSV1 are indicated by aqua boxes in the alignment , -, .

Figure 4. Locations of eex amino-acid substitutions in the S. cerevisiae Pol δ structure.

A) Overall distribution of eex amino-acid substitutions. The catalytic subunit of yeast Pol δ is shown as a ribbon diagram with color-coded structural domains: amino, gray; exonuclease (Exo), red; palm, purple; fingers, blue; thumb, green. Other important elements are indicated as follows: DNA template strand, brown sticks; DNA primer strand, yellow sticks; incoming dCTP, green CPK sticks; metal ions, small black spheres; active-site residues, gray CPK sticks extending out from the α-carbon backbone in the palm and exonuclease domains. Residues changed by eex mutations are shown as light blue spheres. The asterisks mark adjacent E642K and D643N eex substitutions located on the solvent-exposed surface of Pol δ. Structure from (Protein Data Bank accession code 3IAY). B) Amino-acid substitutions near the polymerase active site. Palm domain eex residues are shown as space-filling spheres (light blue) and labeled to indicate the amino-acid substitutions. Important non-mutated residues proximal to the eex substitutions are also shown as space-filling spheres (purple). The fingers and thumb domains have been removed for clarity. C) Amino-acid substitutions in the DNA binding track. View looking down the DNA helical axis. The primer (yellow) and template (brown) are held by a series of interactions along the DNA minor groove. eex residues are light blue. Amino acids positioned by eex residues and minor-groove ‘sensing’ residues in the palm domain (K813, K814 and R815; [74]) are shown as space-filling spheres colored according to domain as in panel (A). The three unpaired 5′ nucleotides of the template have been removed for clarity.

Figure 5. Defining the threshold of error-induced extinction.

A) Plasmid shuffling experiments to reveal synthetic interactions between pol3-01,eex alleles and MMR mutations. Ten-fold serial dilutions of yeast containing POL3–URA3 and pol3–LEU2 plasmids were plated on FOA-containing SC medium to select for cells that spontaneously lost POL3–URA3. Similar numbers of colony forming units were plated for each set of alleles in the MMR⁺, msh6Δ and msh2Δ strains. Failed growth indicates synthetic lethality. Small colonies reflect slow-growth phenotypes. Relative mutation rates are the Can^r mutation rates conferred by each pol3 allele relative to wild-type POL3 in MMR-proficient cells (see Table 1 and Table 2). Alleles are listed in decreasing order of mutator strengths. Some alleles with statistically similar mutation rates (as reflected by overlapping confidence intervals) have slightly different relative rates due to mathematical round-off. B) Relationship between growth capacity and CAN1 mutation rate for 62 haploid yeast strains. Colonies of pol3-01,E642K msh6Δ, pol3-01,G204D msh6Δ, pol3-01,H879Y msh6Δ, and pol3-01 msh6Δ cells are shown to illustrate wild-type (+++), moderately defective (++), severely defective (+), and failed (–) growth, respectively. The vertical dashed line indicates the estimated maximal mutation rate compatible with haploid yeast colony formation, which is our functional definition of the replication error threshold. Filled symbols, rates measured by fluctuation analyses. Open symbols, rates estimated as described in the text. Data in brackets with an asterisk (*) are pol3-01,T711A msh6Δ, pol3-01,S968R msh6Δ, pol3-01,G204D msh6Δ, and pol3-01,Y808C msh6Δ.

Figure 6. Mutational robustness of yeast, E. coli, and mice.

Comparison of spontaneous per-base-pair mutation rates of wild-type (WT) and strong mutator strains of yeast (haploid and diploid), E. coli and mouse cells. Gray boxes indicate the mutation rate intervals that coincide with the transition from wild-type growth (leftmost boundary) to failed growth (rightmost boundary). The data for haploid yeast mutators are from Figure 5B, with the left boundary corresponding to the mutation rate of pol3-01,E642K msh6Δ cells (6.5×10⁻⁷) and the right boundary corresponding to the lethal threshold (4.1×10⁻⁶). The mutation rates of pol3-01,G204D msh6Δ, pol3-01,H879Y msh6Δ, and pol3-01 msh6Δ haploid yeast are shown as examples of progressively stronger mutators with slow (++), very slow (+) and no-growth (–) phenotypes, respectively. The data for pol3-01/pol3-01 pms1/pms1 diploid yeast are from Morrison et al. ; pol3-01/pol3-01 pms1/pms1 cells divide very slowly with a growth phenotype presumably in the range of + to ++. The mouse Pold1^e/e Mlh1^Δ/Δ mutation rate is extrapolated from ouabain-resistance rates of cultured Pold1^+/e Mlh1^Δ/Δ fibroblasts as described in Materials and Methods; a growth phenotype between + and ++ is assumed . E. coli mutation rates and growth phenotypes are from Fijalkowska et al. ; mutD5(pGW1842), mutD5 and dnaQ926 exhibit slow (++), very slow (+) and no-growth (–) phenotypes, respectively. The positions of the gray boxes for diploid yeast, mouse and E. coli are estimates based on the mutation rate and growth capacity relationships observed in haploid yeast (Figure 5B). The error thresholds (rightmost boundaries) for diploid yeast and mouse cells are not known. The yeast wild-type rate is the average of multiple independent determinations (data herein and [40], [108], [114], [124]). Wild-type mouse and E. coli mutation rates are from Drake et al. , .