Spontaneous Polyploids and Antimutators Compete During the Evolution of Saccharomyces cerevisiae Mutator Cells

Abstract

Heterozygous mutations affecting DNA polymerase (Pol) exonuclease domains and homozygous inactivation of mismatch repair (MMR) each generate "mutator" phenotypes capable of driving tumorigenesis. Cancers with both defects exhibit an explosive increase in mutation burden that appears to reach a threshold, consistent with selection acting against further mutation accumulation. In haploid yeast, simultaneous defects in polymerase proofreading and MMR select for "antimutator" mutants that suppress the mutator phenotype. We report here that spontaneous polyploids also escape this "error-induced extinction" and routinely out-compete antimutators in evolved haploid cultures. We performed similar experiments to explore how diploid yeast adapt to the mutator phenotype. We first evolved cells with homozygous mutations affecting proofreading and MMR, which we anticipated would favor tetraploid emergence. While tetraploids arose with a low frequency, in most cultures, a single antimutator clone rose to prominence carrying biallelic mutations affecting the polymerase mutator alleles. Variation in mutation rate between subclones from the same culture suggests there exists continued selection pressure for additional antimutator alleles. We then evolved diploid yeast modeling MMR-deficient cancers with the most common heterozygous exonuclease domain mutation (POLE-P286R). Although these cells grew robustly, within 120 generations, all subclones carried truncating or nonsynonymous mutations in the POLE-P286R homologous allele (pol2-P301R) that suppressed the mutator phenotype as much as 100-fold. Independent adaptive events in the same culture were common. Our findings suggest that analogous tumor cell populations may adapt to the threat of extinction by polyclonal mutations that neutralize the POLE mutator allele and preserve intra-tumoral genetic diversity for future adaptation.

Mutagenic mechanisms of cancer-associated DNA polymerase ϵ alleles

“…Indeed, when we analysed the prevalence of mutations in POL2 we found evidence of additional mutations in this gene often co-occurring with the stronger mutators (P301R, M459K, S312F, L439V) but only rarely with the weaker mutators or in wild-type cells ( Supplementary Figure S3A and B ). In line with recent findings ( 33 ) we also identified one spontaneous diploidization event in four isolates of the pol2-M459K haploid strain, which co-occurred with several additional pol2 mutations ( Supplementary Figure S3A, B; Figure S5C , blue circle).…”

“…As previously observed for several other DNA polymerase mutator alleles ( 16 , 24 , 32 ), sporulation of most pol2-ultra msh2Δ heterozygous diploids did not yield viable double mutant strains for the strongest ultra -mutators ( pol2-P301R , pol2-M459K and pol2-S312F ). Microscopic observation of these spores revealed that they germinated to vegetative cells but ceased to divide after a few generations, a phenotype consistent with extreme mutational burden leading to ‘error-induced extinction’ ( 32 , 33 ). By contrast, double mutants for the hypermutator pol2–4 in combination with msh2Δ were obtained from corresponding heterozygous diploids but displayed reduced colony size compared to MMR-proficient controls.…”

“…These effects likely arose from the experimental settings: population expansion through ∼10 4 rather than single cell bottlenecks presumably allowed different clones to grow at different rates, and random sampling at each passage would favour the propagation and analysis of fitter, faster-growing clones, which would have completed more DNA replications and therefore accumulated more neutral or adaptive mutations than slower growing ones. The appearance and selection of anti-mutators suppressor mutations ( 32 , 33 ) could also explain the increased variability in mutation numbers. Indeed, when we analysed the prevalence of mutations in POL2 we found evidence of additional mutations in this gene often co-occurring with the stronger mutators (P301R, M459K, S312F, L439V) but only rarely with the weaker mutators or in wild-type cells ( Supplementary Figure S3A and B ).…”

“…Both strains were homozygous for MSH2 deletion and pol2 mutations, but the former was a MAT a / MAT a diploid—possibly originating from a whole-genome duplication event caused by failed cytokinesis ( 33 )—and the latter a MAT a / MAT alpha diploid—possibly originating from homothallic mating after a rare mating-type switch event ( Supplementary Figure S5A and B ). These results are in line with recent findings in yeast pol2-P301R cells ( 33 ) and in POLE P286R -driven mouse cancer models ( 42 ), and suggest that a transition to the diploid or polyploid state facilitates survival in the face of extreme mutagenesis as expected from the fact that deleterious mutations are frequently recessive and often masked in a heterozygotic diploid state. Similar to what was previously observed for cells carrying the strongest Pol ϵ hypermutators, we also found that both msh2Δ/msh2Δ pol2–4/pol2–4 and msh2Δ/msh2 pol2-L439V/pol2-L439V accumulated additional mutations in POL2 ( Supplementary Figure S3C and D ).…”

A modified fluctuation assay reveals a natural mutator phenotype that drives mutation spectrum variation withinSaccharomyces cerevisiae

“…To compare the mutation spectra between strains, mutations were first classified as one of the 6 general classes of base-substitutions (A>C, A>G, A>T, C>A, C>G, C>T) or as single base-pair insertions or deletions. We then compared the mutation spectra of the two control strains LCTL1 and LCTL2 to all other haploid isolates as well as one spectrum published by Lang et al (2008) (a total of 35 tests) using a pairwise hypergeometric test (Adams and Skopek, 1987), a custom python script (Tracy et al ., 2020). In the first round of this test, the paired mutation counts were arranged in a 2 x 8 contingency table.…”