Prime editing (PE) is a powerful genome engineering approach that enables the introduction of base substitutions, insertions and deletions into any given genomic locus. However, the efficiency of PE varies widely and depends not only on the genomic region targeted, but also on the genetic background of the edited cell. Here, to determine which cellular factors affect PE efficiency, we carry out a focused genetic screen targeting 32 DNA repair factors, spanning all reported repair pathways. We show that, depending on cell line and type of edit, ablation of mismatch repair (MMR) affords a 2–17 fold increase in PE efficiency, across several human cell lines, types of edits and genomic loci. The accumulation of the key MMR factors MLH1 and MSH2 at PE sites argues for direct involvement of MMR in PE control. Our results shed new light on the mechanism of PE and suggest how its efficiency might be optimised.

Germline mutations in the mismatch repair ( MM R) genes MSH2 , MSH6 , MLH1 and PMS2 are linked to cancer of the colon and other organs, characterised by microsatellite instability and a large increase in mutation frequency. Unexpectedly, mutations in EXO1 , encoding the only exonuclease genetically implicated in MMR, are not linked to familial cancer and cause a substantially weaker mutator phenotype. This difference could be explained if eukaryotic cells possessed additional exonucleases redundant with EXO1. Analysis of the MLH1 interactome identified FANCD2-associated nuclease 1 (FAN1), a novel enzyme with biochemical properties resembling EXO1. We now show that FAN1 efficiently substitutes for EXO1 in MMR assays and that this functional complementation is modulated by its interaction with MLH1. FAN1 also contributes towards MMR in vivo : cells lacking both EXO1 and FAN1 have a MMR defect and display resistance to N -methyl- N -nitrosourea (MNU) and 6-thioguanine (TG). Moreover, FAN1 loss amplifies the mutational profile of EXO1-deficient cells, implying that the two nucleases act redundantly in the same antimutagenic pathway. However, the increased drug resistance and mutator phenotype of FAN1/EXO1-deficient cells are less prominent than those seen in cells lacking MSH6 or MLH1. Eukaryotic cells thus apparently possess additional mechanisms that compensate for the loss of EXO1.

Prime editing is a powerful genome engineering approach that enables the introduction of base substitutions, insertions and deletions, into any given genomic locus. But prime editing, at even the same locus, can exhibit wildly different efficiencies in various cell backgrounds. It is unclear what determines these variations in efficiencies in a given cellular context. Through a focused genetic screen targeting DNA repair factors, we show that the efficiency of prime editing is attenuated by the mismatch repair pathway. The accumulation of the mismatch repair protein MLH1 at sites of prime editing, indicates that mismatch repair acts at these regions to directly counteract the insertion of the edit. Consequently, ablation of mismatch repair yields an up to 17-fold increase in prime editing efficiency across different human cell lines, several types of edits and multiple genomic loci. Our results shed new light on the cellular requirements for prime editing and identify that ablation of mismatch repair increases editing efficiency and fidelity.