Transcription Factors Increase Mutagenesis by Interfering with DNA Mismatch Repair

Abstract

Mutation rates vary substantially across the genome, at scales ranging from megabases to individual nucleotides. At intermediate scales, cancer genomics studies have recently revealed a pattern of hyper-mutation at sites occupied by transcription factor (TF) proteins. This pattern extends across a wide range of factors and cancer types, is not explained by selection mechanisms, and it correlates with decreased DNA repair activity, suggesting that TFs may increase mutagenesis by interfering with DNA repair. This hypothesis is supported by small-scale studies showing direct competition between TFs and repair enzymes, as well as recent studies from our laboratory showing that TFs can bind strongly to DNA lesions. Still, a direct proof of increased mutagenesis resulting fro¬¬m TF binding to lesions and interference with repair is lacking.Focusing on genomic mutations that result from mismatches generated during DNA replication, here we demonstrate the mechanism of TF-induced mutagenesis for a yeast TF, and we provide evidence from genomic data that the mechanism is also present in human cells. First, we used a high-throughput in vitro binding assay to show that TF binding to mismatches leads to reduced binding by the mismatch recognition enzyme, MutSα. The strong positive correlation between TF binding levels and the decreases in MutSα-mismatch binding, observed for both yeast and human TFs, demonstrates direct competition. Next, to assess the mutagenic role of TF-mismatch binding in vivo, we leveraged the power of yeast genetics to modify the lys2 reversion assay for identifying rare, naturally-occurring mutations at a binding site for TF Cbf1. Upon Cbf1 overexpression, the site showed a strong shift towards mutations resulting from mismatches where Cbf1 significantly outcompetes MutSα in vitro. The trend was not observed when the binding site was scrambled or mismatch repair (MMR) was abolished, providing strong support for our hypothesis that mismatch-bound TFs compete with MMR and lead to increased mutagenesis. To test whether the same mutagenesis mechanism takes place in human cells, we examined the patterns of somatic mutations observed at binding sites of c-MYC, a TF highly expressed in cancer cells. Consistent with our results for yeast TF Cbf1, human c-MYC binding sites exhibit a significant shift towards mutations resulting from mismatches where the TF outcompetes MutSα in vitro; the shift was not present in cancer cells deficient in MMR. The lys2 reversion assay is a low-throughput genetic assay, limited to TF binding sites overlapping a stop codon and mutations that convert the stop codon into a sense codon. To overcome these limitations, we adapted the error-corrected high-throughput sequencing method maximum-depth sequencing (MDS), enabling detection of extremely rare variants in a cell population. Incorporating this novel technique allowed us to test our hypothesis more broadly, including for clinically relevant TFs, and to fully characterize the range and frequency of mutations induced by TF binding. In conclusion, our study establishes a novel mechanism for mutation accumulation in regulatory DNA, with significant implications for understanding the evolution of regulatory regions and for the identification of regulatory drivers of tumorigenesis.