DNA mutagenesis driven by transcription factor competition with mismatch repair

The distribution of somatic mutations across the genome is not uniform. Recently, an unexpected pattern of hyper-mutation was reported at binding sites of transcriptional regulatory factors (TFs). In some human cells, a decrease in DNA repair activity was also observed at TF binding sites, leading to the hypothesis that TFs may increase mutagenesis by interfering with the recognition of DNA lesions by repair enzymes, and thus inhibiting repair. However, direct proof of this surprising TF-induced mutagenesis mechanism is lacking. Here, we show that TF binding to DNA mismatch lesions leads to increased mutation rates at TF binding sites by reducing the efficiency of lesion recognition by MutSα, the main enzyme that initiates mismatch repair in eukaryotic cells. We developed a yeast mutagenesis assay to directly observe the accumulation of mutations in a TF binding site. Upon TF overexpression, the binding site exhibited an increased mutation rate, specifically for mutations resulting from mismatches where the TF strongly reduced MutSα binding in vitro. This trend was amplified in cells with an increased rate of misincorporation errors, and it was not observed in mismatch repair-deficient cells. Analyses of human cancer somatic mutation data revealed a pattern similar to that observed in yeast, with mutations resulting from TF-bound mismatches being specifically enriched in mismatch repair-proficient tumors. Taken together, our results demonstrate that in addition to their well-known roles in gene regulation, TFs also play a role in DNA mutagenesis, by directly interfering with the repair of replication errors. Since a majority of cancer mutations originate from unrepaired replication errors, most commonly mismatches, our results suggest that TF interference with mismatch repair will shape the mutation landscape of regulatory DNA in cancer genomes. Competition DNA-binding experiments were performed for yeast MutS⍺ and yeast TF Cbf1, and human MutS⍺ and human TF MYC. Briefly, we designed DNA sequences to form a stem-loop structure that incorporates mismatched base pairs in the stem, and the DNA molecules were synthesized on an Agilent microarray. Then we measured the binding of MutS⍺ to the DNA, either by itself or under direct competition from Cbf1 or MYC. We also measured the binding of Cbf1 and MYC by themselves. For each Watson-Crick and mismatch-containing sequence except for the random sequences, ten replicate spots were used in each chamber, randomly distributed across the microarray surface. For the 200 random DNA sequences, five replicate spots were used. We report the PBM signal intensity of yeast MutS⍺, Cbf1, human MutS⍺, or MYC for each spot.