Development of B-to-A base editors by leveraging translesion synthesis polymerase

Abstract Base editing technology enables single-nucleotide conversions without donor templates or DNA double-strand breaks (DSBs), yet current base editors lack the capacity to efficiently address C-to-A, G-to-A, or T-to-A conversion. Here we report the development of B-to-A base editors (BABEs, B=C/G/T) through co-expressing certain glycosylase-based base editors together with the engineered translesion synthesis (TLS) polymerase polη Q38S variant, achieving about an average of 5-fold increase in B-to-A editing purity and efficiency. The C-to-A base editor (CABE), G-to-A base editor (GABE), and T-to-A base editor (TABE) achieved C-to-A purity up to 69.0%, G-to-A purity up to 57.1% and T-to-A purity up to 91.6%, respectively, in human cells without compromising the editing efficiency. In addition, we enhanced the editing efficiency of gGBE through rational design and further developed more efficient G-to-A base editor (GABEv2), which displayed a 2.1-fold improvement in G-to-A editing purity and efficiency compared to gGBE in mouse embryo. Our study expands current base editor toolbox by providing B-to-A base editors, broadening the applications of base editors in fundamental research and clinical therapeutics.