Mutational processes of distinct POLE exonuclease domain mutants drive an enrichment of a specific TP53 mutation in colorectal cancer

Abstract

Cancer genomes with mutations in the exonuclease domain of Polymerase Epsilon (POLE) present with an extraordinarily high somatic mutation burden. In vitro studies have shown that distinct POLE mutants exhibit different polymerase activity. Yet, genome-wide mutation patterns and driver mutation formation arising from different POLE mutants remains unclear. Here, we curated somatic mutation calls from 7,345 colorectal cancer samples from published studies and publicly available databases. These include 44 POLE mutant samples including 9 with whole genome sequencing data available. The POLE mutant samples were categorized based on the specific POLE mutation present. Mutation spectrum, associations of somatic mutations with epigenomics features and co-occurrence with specific driver mutations were examined across different POLE mutants. We found that different POLE mutants exhibit distinct mutation spectrum with significantly higher relative frequency of C>T mutations in POLE V411L mutants. Our analysis showed that this increase frequency in C>T mutations is not dependent on DNA methylation and not associated with other genomic features and is thus specifically due to DNA sequence context alone. Notably, we found strong association of the TP53 R213* mutation specifically with POLE P286R mutants. This truncation mutation occurs within the TT[C>T]GA context. For C>T mutations, this sequence context is significantly more likely to be mutated in POLE P286R mutants compared with other POLE exonuclease domain mutants. This study refines our understanding of DNA polymerase fidelity and underscores genome-wide mutation spectrum and specific cancer driver mutation formation observed in POLE mutant cancers.

Fig 1. Mutational spectrum of distinct colorectal cancer POLE mutants.

(A) Hierarchical clustered heatmap of the frequency of 96 types of mutational contexts within each WGS POLE mutant ranging from light red (0%) to dark red (35% of all mutations). Four groups “MSS (microsatellite stable), V411L, P286R and Other-Exo” were labeled on the far left panel, and total mutation burden was indicated in the right bottom. The MSS spectrum is averaged across the 44 TCGA MSS POLE wild-type WGS samples while the TCGA samples ID for each POLE mutant is shown. Mutants with multiple variants are underlined. (B) Mutational spectrum of four POLE-mutant groups based on 96 mutational contexts, with mutation type indicated on the top panel. (C) Proportion of C>A, C>T and T>G mutation in four POLE mutant groups. The significance was calculated by paired Chi-squared test. Error bars represent +/- 2 SE. (D) Profile of C>T mutations in penta-nucleotide contexts, with genome-wide frequency of each penta-nucleotide indicated at bottom. The detail of context information is indicated in S8 Table. *** denotes P < 0.001.

Fig 2. Association of methylation and mutation in different POLE mutants.

(A) Correlation between mutations per megabase (Mb) at CpG dinucleotides and fractions of CpGs methylated across different POLE mutants and microsatellite stable (MSS) samples. (B) Mutation burden of each mutant was normalized by the mutation rate of V411L in each methylation level.

Fig 3. Sequence context in different methylation bins.

(A) Proportion of each “NNCGN” penta-nucleotide context in different methylation level, with “TTCGN” shadowed. The detail of context information is indicated in S8 Table. (B-D) Correlation of methylation and mutation burden after normalization of penta-nucleotide context composition, with non-normalized data indicated in light blue.

Fig 4. Genome-wide mutational patterns of distinct POLE mutants.

(A) Somatic substitutions at CBSs with a flanking sequence of 1 kilo bp in different POLE mutants. The expected mutation was indicated in light red colour. (B) Mutation profile around transcription start sites in different mutants. Three primary mutation types C>A (red), C>T (blue) and T>G (green) in specific context were showed. Mutation counts were normalized by the number of corresponding context and the abundance of each context was displayed in the far left panel, together with mutation data in 100 bp bins (grey) is shown. (C) Profile of mutation burden across different part of genes in different mutants. Each part of gene was divided into 20 bins and mutation burden was calculated separately. Methylation level of each part was showed in the top panel. (D) Mutational strand asymmetry associated with replication in different mutants. Lower panel of each mutant shows the log2 ratio of each pair of bars.

Fig 5. Mutation hotspots in POLE mutants.

(A) Contingency table of different POLE-mutant and POLE wild type colorectal cancer samples with or without the TP53 R213* mutation. Samples with Sig10 were confirmed by either POLE driver mutation or mutational spectrum clustering. (B) Truncating mutation TP53 R213* was caused by C>T substitution in TT[C>T]GA context. (C) Frequency of 21-bp sequence context centered by mutated cytosine in different POLE mutants, and the penta-nucleotide contexts were indicated in black box. Proportion of TT[C>T]GA mutations in the NN[C>T]GN pentanucleotide context in all POLE P286R and V411L mutants (D) and (E) POLE P286R mutants with and without the TP53 R213* mutation. * < 0.05. ** < 0.01, Student’s t-test.