GoKawiila, On-target editing rates of ABE8e and Cas9(K1020D)-TadA8e at seven endogenous genomic loci in HEK293T cells. Editing rates are shown for the A base indicated in parentheses for each target site. b, c, Comparison of genome-wide sgRNA-dependent off-targets (evaluated by dI-profiling) between ABE8e and Cas9(K1020D)-TadA8e in HEK293T cells using HEK4 sgRNA (b) and ABEsite16 sgRNA (c). Each dot represents an off-target A. The relative reductions in total off-target signals for Cas9(K1020D)-TadA8e compared with ABE8e are 83.0% and 69.0% at HEK4 (n = 1,717 sites; b) and ABEsite16 (n = 270 sites; c), respectively. d, Position of K1020 in an AF3-predicted structure of the Cas9-sgRNA-DNA ternary complex, generated using the ABEsite16-OT206 off-target DNA and the ABEsite16-sgRNA (seed 194503). e, f, Average minimum distance over 50 ns MD simulation (n = 1000 frames) between heavy atoms of the residue 1020 (K1020 or K1020D) and ntsDNA at the ABEsite16-OT206 off-target site (e) and the ABEsite16 on-target site (f). Simulations were performed using AF3-predicted Cas9-sgRNA-DNA ternary structures with the ABEsite16 sgRNA (seed 1, 164664; seed 2, 32383). The centre line indicates the median; box limits indicate the 25th and 75th percentiles; the whiskers extend to the minimum and maximum values within 1.5× IQR. Dots represent individual data points. P values were determined by one-sided Wilcoxon signed-rank test. g, h, Minimum distance between the residue 1020 (K1020 or K1020D) and ntsDNA over a 50 ns time course at the off-target site (g) and the on-target site (h) in the MD simulation. Solid lines represent the mean of the minimum distance calculated over an 11-frame sliding window. Shaded regions denote the maximum and minimum values within the corresponding sliding window. i, Distribution of mismatch positions along the protospacer among HEK4 off-target sites grouped by dI-profiling signal reduction in Cas9(K1020D)-TadA8e-treated samples relative to ABE8e-treated samples. Off-targets were ordered by signal reduction, and the top 5%/10% (“most reduced”) and bottom 5%/10% (“least reduced”) groups were selected. The upper two plots show the proportion of mismatch at each protospacer position among sites in the top 5% or bottom 5% group of dI-profiling signal reduction. The lower two plots show the difference in the proportion at each position between the top 5% and bottom 5% groups, and between the top 10% and bottom 10% groups (top minus bottom), suggesting a modest positional trend with relatively more 5′-end mismatches in the most reduced sites. j, On-target editing rates of Cas9 nuclease and Cas9(K1020D) nuclease at four endogenous genomic loci in HEK293T cells. k, On-target editing rates of BE4max and BE4max(K1020D), in which the Cas9 component carries the K1020D mutation, at three endogenous genomic loci in HEK293T cells. The subscripted red C denotes the edited C position within the protospacer and the PAM sequence is underlined. l, m, Comparison of genome-wide sgRNA-dependent off-targets (evaluated by Detect-seq) between BE4max and BE4max(K1020D) in HEK293T cells using HEK4 sgRNA (n = 246 sites; l) and VEGFAsite2 sgRNA (n = 512 sites; m). Each dot represents an off-target site. The relative reductions in total off-target signals for BE4max(K1020D) compared with BE4max are 73.3% and 63.5% at HEK4 (l) and VEGFAsite2 (m), respectively. n-p, On-target and off-target editing by PEs whose Cas9 components are either wildtype or carry the K1020D mutation, using pegRNAs with known off-target loci. Red bases denote the edited positions, and the PAM sequence is underlined. In a, j, k and n-p, each dot indicates an independent biological replicate (n = 2); bars indicate the mean.
Source data