Replication-associated mechanisms contribute to an increased CpG > TpG mutation burden in mismatch repair-deficient cancers

Abstract

Background: Single base substitution (SBS) mutations, particularly C > T and T > C, are increased owing to unrepaired DNA replication errors in mismatch repair-deficient (MMRd) cancers. Excess CpG > TpG mutations have been reported in MMRd cancers defective in mismatch detection (dMutSα), but not in mismatch correction (dMutLα). Somatic CpG > TpG mutations conventionally result from unrepaired spontaneous deamination of 5'-methylcytosine throughout the cell cycle, causing T:G mismatches and signature SBS1. It has been proposed that MutSα detects those mismatches, prior to error correction by base excision repair (BER). However, other evidence appears inconsistent with that hypothesis: for example, MutSα is specifically expressed in S/G₂ phases of the cell cycle, and defects in replicative DNA polymerase proofreading specifically cause excess CpG > TpG mutations in signature SBS10b.

Methods: We analysed mutation spectra and COSMIC mutation signatures in whole-genome sequencing data from 1803 colorectal cancers (164 dMutLα, 20 dMutSα) and 596 endometrial cancers (103 dMutLα, 9 dMutSα) from the UK 100,000 Genomes Project. We mapped each C > T mutation to its genomic features, including normal DNA methylation state, replication timing, transcription strand, and replication strand, to investigate the mechanism(s) by which these mutations arise.

Results: We confirmed that dMutSα tumours specifically had higher CpG > TpG burdens than dMutLα tumours. We could fully reconstitute the observed dMutSα CpG > TpG mutation spectrum by adding CpG > TpG mutations in proportion to their SBS1 activity to the dMutLα spectrum. However, other evidence indicated that the SBS1 excess in dMutSα cancers did not come from 5'-methylcytosine deamination alone: non-CpG C > T mutations were also increased in dMutSα cancers; and, in contrast to tumours deficient in BER, CpG > TpG mutations were biased to the leading DNA replication strand, at similar levels in dMutSα and dMutLα cancers, suggesting an origin in DNA replication. Other substitution mutations usually corrected by BER were not increased in dMutSα tumours.

Conclusions: There is a CpG > TpG and SBS1 excess specific to dMutSα MMRd tumours, consistent with previous reports, and we find a general increase in somatic C > T mutations. Contrary to some other studies, the similar leading replication strand bias in both dMutSα and dMutLα tumours indicates that at least some of the excess CpG > TpG mutations arise via DNA replication errors, and not primarily via the replication-independent deamination of 5'-methylcytosine.

Keywords: Colorectal cancer; DNA repair; Mismatch repair deficiency; Mutagenesis; Mutation signatures.

Fig. 1

Colorectal cancers included in mutation analysis. A summary of the primary, treatment-naïve colorectal cancers (CRCs) subjected to PCR-free whole-genome sequencing analysis from V18 of the UK 100,000 Genomes Project. mSINGS = Detecting MSI by Next-Generation Sequencing. MMRp = Mismatch repair-proficient. MMRd = Mismatch repair-deficient. POLE = DNA polymerase ε. POLD1 = DNA polymerase δ

Fig. 2

The SBS landscapes of MMRp and MMRd colorectal cancers. The total single base substitution (SBS) burden (A) of mismatch repair-proficient (MMRp, grey) and mismatch repair-deficient (MMRd, orange) colorectal cancers. Also shown are the burdens (B) and activities (C) of the six SBS mutation channels: C > A, C > G, C > T (split into CpG > TpG and non-CpG C > T C > T mutations), T > A, T > C and T > G. D The activities of the “clock-like” SBS mutation signatures SBS1 and SBS5, as well as the MMRd-associated signatures SBS15, SBS26, SBS44, and the potential artefact signature SBS57 in the MMRp and MMRd colorectal cancers. Other mutations signatures include SBS2, SBS7c, SBS10a, SBS13, SBS17a, SBS17b, SBS18, SBS28, SBS40, SBS56, SBS93, and SBS94

Fig. 3

The SBS landscapes of dMutSα and dMutLα colorectal cancers. The total single-base substitution (SBS) burden (A) of MutLα-deficient (dMutLα, purple) or MutSα-deficient (dMutSα, green) colorectal cancers. Also shown are the burdens (B) and activities (C) of the six SBS mutation channels: C > A, C > G, C > T (split into CpG > TpG and non-CpG C > T mutations), T > A, T > C and T > G. D The activities of the “clock-like” SBS mutation signatures SBS1 and SBS5, as well as the MMRd-associated signatures SBS15, SBS21, SBS26, SBS44, and the potential artefact signature SBS57 in the dMutLα and dMutSα colorectal cancers. Other mutation signatures include SBS7c, SBS14, SBS17a, SBS18, SBS20, SBS36, SBS41, SBS93, and mutations not assigned to any pre-existing COSMIC mutation signature

Fig. 4

Mutation spectra of dMutLα and dMutSα colorectal and endometrial cancers. The activities (proportions of all 96 SBS channels) of each specific substitution are shown for dMutLα colorectal cancers (A), dMutSα colorectal cancers (B), and the same for endometrial cancers (C, D)

Fig. 5

Addition of CpG > TpG mutations in proportion to SBS1 channel activities to the observed dMutLα mutation spectrum causes near-identity to the observed dMutSα spectrum in colorectal cancer. The y-axis shows the cosine similarity between the CpG > TpG mutation spectrum of dMutSα CRCs and the same spectrum in dMutLα CRCs when SBS1-associated mutations are proportionally added. As more SBS1 mutations are added (i.e. x-axis values increase), the cosine similarity rises to a peak of 0.999. The effects of adding in SBS15 and SBS44 channels are shown for comparison, with no evidence of an effect for either signature

Fig. 6

De novo mutation signatures extracted from MMRd colorectal cancers. The activities (proportions of all 96 SBS channels) of each specific substitution are shown for the de novo mutation signatures SBS_CRC-MMRd-A (A) and SBS_CRC-MMRd-B (B), extracted from mismatch repair-deficient (MMRd) colorectal cancers. SBS_CRC-MMRd-A and SBS_CRC-MMRd-B, had cosine similarity to one another of 0.675, while the sixteen C > T mutation channels were more similar (cosine similarity 0.736). Also presented are the signature plots approximated from the study by Fang et al. [14], Signature A (C) and Signature B (D). Also shown are the pairwise cosine similarities of the sixteen C > T mutation channels of these de novo signatures with the dMutLα and dMutSα colorectal cancer spectrum, as well as the spectrum of the COSMIC signatures SBS1, SBS15, SBS44, and SBS57 (E). The spectrum of the sixteen C > T mutation channels in SBS_CRC-MMRd-A most closely resembled the COSMIC signature SBS15 (cosine similarity 0.814), while SBS_CRC-MMRd-B most resembled SBS44 (cosine similarity 0.797)

Fig. 7

Replication strand bias of C > T mutations in MMRd CRC and EC. The replication strand (log₂(Leading/Lagging)) bias of CpG > TpG and non-CpG C > T mutations in mismatch repair-deficient (MMRd) colorectal cancers (A) and endometrial cancers (B), classified as MutLα-deficient (dMutLα, purple) or MutSα-deficient (dMutSα, green). P comparing dMutLα to dMutSα from a Wilcoxon test. P comparing CpG > TpG versus non-CpG C > T from a Wilcoxon signed-rank test