Reprogramming deaminase substrate specificity for single nucleobase editing

In this study, we set out to reprogram deaminase context specificity to pinpoint editing. We identified multiple nucleic acid-recognition hotspots in the E. coli tRNA-specific adenosine deaminase (TadA). Strategically sampling these recognition hotspots, we first accessed multipotency for C in TadA and subsequently eliminate its A-deamination activity. We further reprogrammed TadAC context specificity through 16 evolution campaigns, each aimed at a defined NCN context, and isolated hundreds of thousands of context-specific cytosine deaminases. Our panel of 16 NCN-specific deaminases covers the full spectrum of all possible minus1 and plus 1 contexts for a target C, offering on demand deaminase choices for editor customization. Our context-specific CBEs corrected 5,866 of 7,196 disease-associated T:A-to-C:G transitions documented by ClinVar with higher accuracy than existing CBEs, often achieving selective editing of a single cytosine out of multiple cytosines in the protospacer without compromising editing potency. We also showcased the application of context-specific base editing for modeling disease-associated C:G-to-T:A transitions using two cancer driver mutations, KRASG12D and TP53R248Q, each demanding selective editing of one cytosine in two consecutive cytosines (ACC and CCG). These context-specific editors, as expected, showed tightly controlled off-target profiles by rejecting most cytosines at potential off-target sites. Bystander-free, single-nucleobase editing, as enabled by reprogramming deaminase context specificity, complements our current editor portfolio and unlocks new potential in base editing.