Mutagenic Repair Outcomes of DNA Double-Strand Breaks

DNA double strand breaks (DSB) are cytotoxic lesions that can lead to genome rearrangements and genomic instability, which are hallmarks of cancer. The two main DSB repair pathways are non-homologous end joining and homologous recombination (HR). While HR is generally highly accurate, it has the potential for gross chromosomal rearrangements (GCRs) that occur directly or through intermediates generated during the repair process. Whole genome sequencing of cancers has revealed numerous types of structural rearrangement signatures that are often indicative of repair mediated by sequence homology. However, it can be challenging to delineate repair mechanisms from sequence analysis of rearrangement end products from cancer genomes, or even model systems, because the same rearrangements can be generated by different pathways. Numerous studies have provided insights into the types of spontaneous GCRs that can occur in various Saccharomyces cerevisiae mutants. However, understanding the mechanism and frequency of formation of these GCR without knowledge of the initiating lesions is limited. Here, we focus on DSB-induced repair pathways that lead to GCRs. Inverted duplications occur at a surprisingly high frequency when a DSB is formed near short inverted repeats in cells deficient for the nuclease activity of Mre11. Similar to previously proposed models, the inverted duplications occur through intra-strand foldback annealing at resected inverted repeats to form a hairpin-capped chromosome that is a precursor to dicentric chromosomes. Surprisingly, we found that DNA polymerase δ proof-reading activity but not the Rad1-Rad10 nuclease is required for inverted duplication formation, suggesting a role for Pol δ in the removal of the heterologous tails formed during foldback annealing. Contrary to previous work on spontaneous inverted duplications, we find that DSB-induced inverted duplications require the Pol δ processivity subunit Pol32 and that RPA plays little role in their inhibition, suggesting that spontaneous inverted duplications arise differently than those induced by DSBs. We show that stabilization of dicentric chromosomes after breakage involves telomere capture through a strand-invasion step mediated by repeat sequences and requires Rad51. Previous work on spontaneous inverted duplications suggested that Tel1, but not Mre11-Sae2, inhibits inverted duplications that initiate from inverted repeats separated by long spacers (> 12 bp). However, we do not find evidence for this requirement. Cells with Tel1 deletion can still resolve hairpins containing loops up to 30 nt long. Furthermore, deletion of Sae2, but not Tel1, increases the frequency of inverted duplications when a DSB is induced near an inverted repeat separated by a 20 bp-long spacer. This highlights another difference between spontaneous and DSB-induced GCRs. Finally, we find that the sequence context of a DSB affects the type of GCR outcome. Inverted repeats are required for the formation of inverted duplications, as the deletion of a DSB-proximal inverted repeat significantly reduces the incidence of this type of rearrangement. Furthermore, introduction of a DSB near short telomere-like sequence is required for chromosome truncations stabilized by de novo telomere addition. The effect of the sequence context can partly explain how repair pathways can be channeled to different mutagenic outcomes. Our results highlight the importance of considering how the initiating lesion can affect the type of resulting GCRs and the mechanisms by which they occur.