Continuous evolution of base editors with improved activity and target compatibility

Base editors use DNA-modifying enzymes targeted by a catalytically impaired CRISPR protein to precisely install point mutations without double-strand breaks or reliance on homology-directed repair. The applicability of base editors is determined by their editing efficiency and target sequence compatibility. Here we report the development and initial application of phage-assisted continuous evolution of base editors (BE-PACE). We used BE-PACE to overcome target sequence context constraints that limit the applicability of APOBEC1 deaminase-derived cytosine base editors (CBEs), and to increase the editing efficiency of CDA1 deaminase-derived CBEs. Evolved APOBEC1-based CBEs exhibit high levels of editing in all tested sequence contexts, including up to 26-fold improvement in editing GC, a disfavored context for wild-type APOBEC1. We also applied BE-PACE to evolve CDA1-based CBEs with substantially higher average activity at suboptimal target sites and expanded editing windows at efficiently edited target sites. Compared to previously described CBEs, both classes of evolved CBEs enabled greatly improved levels of base editing of SNPs associated with genetic deafness, Alzheimer’s disease, or Wolfram syndrome in primary cells or established cell lines. Finally, we present a conceptual model that illuminates the relationship between deaminase activity, base editing efficiency, activity window width, and byproduct formation. These findings establish a system for the rapid evolution of base editors with altered targeting and editing properties.