Fig. 4: Enhancement of gCBE editing activity through protein engineering.

a Schematic diagram of mutagenesis and screening strategy for the engineered gCBE. b Gradual improvement of EGFP activation for each gCBE variants (n = 3 independent biological replicates). WT, wild-type UNG2Δ88. dead, catalytically inactive UNG2Δ88 (carrying D154N and H277N mutations, equivalent to D145N and H268N of UNG1)⁶⁰. ΔNTD: N-terminal domain (NTD) truncation of UNG. c Bar plots showing the on-target DNA base editing at positions with the highest C conversion frequencies at each genomic site in HEK293T cells (n = 3 independent biological replicates). C#: C position with highest on-target base editing frequencies across protospacer positions 1–20. site #: genomic site number. d Bar plots showing the on-target DNA base editing of different positions at three loci with gCBEv2 or CGBE1 (n = 3 independent biological replicates). e On-target base editing frequencies for gCBEv2 at C6 of site 22 in HEK293T cells for the orthogonal R-loop assay (n = 3 independent biological replicates). f gRNA-independent cumulative off-target editing frequencies detected by the orthogonal R-loop assay at each R-loop site. Each R-loop was performed by co-transfection of each base editor, and an SpCas9 sgRNA targeting corresponding site with dSaCas9 and a SaCas9 sgRNA (n = 3 independent biological replicates). All values are presented as mean ± s.e.m. Source data are provided as a Source Data file.