DNA polymerase ε and δ variants drive mutagenesis in polypurine tracts in human tumors

Abstract

Alterations in the exonuclease domain of DNA polymerase ε cause ultramutated cancers. These cancers accumulate AGA>ATA transversions; however, their genomic features beyond the trinucleotide motifs are obscure. We analyze the extended DNA context of ultramutation using whole-exome sequencing data from 524 endometrial and 395 colorectal tumors. We find that G>T transversions in POLE-mutant tumors predominantly affect sequences containing at least six consecutive purines, with a striking preference for certain positions within polypurine tracts. Using this signature, we develop a machine-learning classifier to identify tumors with hitherto unknown POLE drivers and validate two drivers, POLE-E978G and POLE-S461L, by functional assays in yeast. Unlike other pathogenic variants, the E978G substitution affects the polymerase domain of Pol ε. We further show that tumors with POLD1 drivers share the extended signature of POLE ultramutation. These findings expand the understanding of ultramutation mechanisms and highlight peculiar mutagenic properties of polypurine tracts in the human genome.

Keywords: CP: Cancer; DNA polymerase delta; DNA polymerase epsilon; POLD1; POLE; VUS; colorectal cancer; endometrial cancer; mutation; polypurine tracts; tumor classification.

Figure 1.. Tumors with POLE driver mutations accumulate G>T transversions in polypurine tracts

(A) The mean value of MMF for each of 104 polypurine motifs in tumors with POLE driver mutations (n = 44), POLE VUSs (n = 30), or no POLE or POLD1 mutations (n = 292). Error bars indicate 95% confidence interval (CI). The red 4 symbols indicate mutation positions within the motifs. The lines were drawn to emphasize the differences between motifs with varying positions of the mutated G and to not assume a continuum of MMF values. (B) Comparison of MMF (all 104 motifs combined) between samples with POLE driver mutations (n = 44 samples × 104 motifs = 4,576), POLE VUSs (n = 3,120), and no POLE or POLD1 mutations (n = 30,368) (Kruskal-Wallis, p < 0.001). Each dot displays MMF value for an individual mutational motif. Mean MMF was 51.9 × 10⁻⁶ for samples with POLE drivers, 5.14 × 10⁻⁶ for samples with POLE VUSs, and 0.30 × 10⁻⁶ for samples with no POLE or POLD1 mutations. (C) Contour plot of the mean MMF among samples with POLE drivers showing MMF dependence on the position of mutated site within a motif. (D) The mean MMF for motifs with two or three purines after the mutation in tumors with different POLE driver alleles. MMF values were normalized by the number of G>T transversions in each tumor. Error bars indicate 95% CI. (E) Comparison of MMF for motifs with two or three purines after the mutation in different categories of tumors in regard to POLE and MSI statuses: “POLE drivers, MSS” (n = 851), “POLE drivers, MSI” (n = 161), “no POLE/POLD1 mutations, MSS” (n = 4,508), and “no POLE/POLD1 mutations, MSI” (n = 2,208). Each dot displays the MMF value for an individual mutational motif. Horizontal lines and stars indicate statistical associations between groups. *p < 0.05; ***p < 0.001; ns, no significance.

Figure 2.. Classification workflow

(A) Workflow for the development of tumor classification model. (B) t-SNE plot visualizing different clusters of tumors with POLE drivers and tumors with no POLE/POLD1 mutations from the UCEC EAC cohort using three top-ranked motifs. (C) Classification accuracy based on different features for the training dataset (UCEC EAC) and for two independent datasets (UCEC SCAC and COAD). Error bars show 95% CIs.

Figure 3.. Validation of novel POLE drivers by functional analysis in yeast

(A) Schematic representation of human POLE and POLD1 showing the location of VUSs tested in this work. Exo, exonuclease domain; Pol, DNA polymerase domain. Striped boxes indicate conserved exonuclease and DNA polymerase motifs. (B) Alignment of amino acid sequences of human and yeast proteins around the mutation sites. Human VUSs and analogous yeast variants are shown above and below the alignments, respectively. (C) Identification of mutator variants in tumors with multiple VUSs predicted to carry driver mutations. Mutation rate relative to wild type is shown for yeast strains with pol2 or pol3 mutations mimicking the indicated human variants. (D) VUSs from tumors not predicted to contain driver mutation confer mild or no mutator effects in yeast. In (C) and (D), the pol3-G731R and pol2-C417R haploid strains were inviable, and the mutator effect is shown for heterozygous diploids. All other data are for haploid strains carrying the indicated alleles as the sole source of Pol ε or Pol δ. Data are from Tables S9 and S10. Dashed lines show wild-type mutagenesis levels. Asterisks indicate p < 0.05 by Wilcoxon-Mann-Whitney compared to the wild-type strain (null hypothesis: MR_mutant > MR_wild-type).

Figure 4.. Structural implications of novel POLE driver mutations

(A) Locations of the previously known (orange balls) and new (red balls) driver mutations on the structure of the catalytic core of yeast Pol ε (PDB: 4M8O). Note the location of the E991G substitution (human E978G) in the thumb subdomain of the DNA polymerase domain. DNA with the primer terminus in the polymerase active site is in yellow. (B) E991, E985, and E1048 form a cluster of negatively charged residues at the DNA-binding interface of yeast Pol ε. Surface representation of the catalytic core is shown with the DNA cartoon in orange. Exonuclease domain (cyan) is in the back, partially visible through the cleft in the polymerase domain. DNA-binding arginine and lysine residues are in blue, and the three glutamate residues are in red. A close-up view of the glutamate cluster is shown on the right. (C) The E991G substitution is predicted to increase the positive charge of the DNA-binding interface in the polymerase domain. Red and blue colors display negative and positive electrostatic potentials, respectively. Electrostatic potentials were calculated with the Adaptive Poisson-Boltzmann Solver method and mapped onto the molecular surfaces of wild-type Pol ε (top) and Pol ε E991G (bottom). The mutagenesis wizard of PyMOL was applied to generate the E991G variant. The color intensity was scaled with electrostatic potential values of the surface. (D) Schematic representation of contacts between E991 and three DNA-binding arginine residues in yeast Pol ε. A −1 nucleotide position corresponds to the last base pair in the polymerase active site. See the main text for details. (E) A close-up view of the interaction between E991, R988, and DNA bases at −4 and −5 positions of the primer. Yellow rods show the position of the DNA backbone.