Highly efficient and precise base editing in discarded human tripronuclear embryos

CRISPR/Cas9 is a powerful tool for genome editing (Komor et al., 2017). Recently, it has been employed in several attempts to edit the human embryos (Liang et al., 2015; Kang et al., 2016; Tang et al., 2017). A major technical concern particularly relevant in studies involving human embryos is the potential off-target effects (Callaway, 2016; Plaza Reyes and Lanner, 2017). Consequently, development of safer genome editing strategy in human embryos is highly anticipated (Cyranoski and Reardon, 2015). The offtarget mutation result in part from Cas9-mediated double strand break (DSB) of DNA. Recently, base editing (BE) without the introduction of DSB has been achieved. The key design for BE is to use a catalytically inactive Cas9 to recruit the cytidine deaminase APOBEC to target sequences, leading to conversion of C to T within a window of approximately five nucleotides (Komor et al., 2016). Therefore, BE is apparently determined by additional features of the target sequence and offers a potentially safer approach for genome editing. Here we report the initial technical assessment of applying BE3, base editor 3 (Komor et al., 2016), in discarded human tripronuclear embryos. We targeted two human gene sites, HEK293 site 4 and RNF2 (Komor et al., 2016). BE3 and sgRNAs were prepared in vitro as described (Shen et al., 2014), and microinjected into the cytoplasm of the tripronuclear zygotes with the concentration of one hundred nanogram BE3 and fifty nanogram sgRNA per microliter. The zygotes were collected 48 h after microinjection, with the embryos containing different numbers of cells ranging from 1 to 8 (Table S1). In total, 8 zygotes for each of the two targets (#1–8 for HEK293 site 4, #9–16 for RNF2) were collected (Fig. 1A). Whole genome of each individual sample was amplified and used as the template for further analysis. To detect the efficiency of base editing, the region around the target sites was amplified and analyzed initially by the T7EN1 cleavage assay. For HEK293 site 4, we did not detect any cleavage bands in any of the samples (Fig. S1A). However, sequencing of the bulk PCR products revealed C to Tconversion at the sixteenth base distal from the PAM in 7 of the samples (#1–6, #8) (Figs. 1B and S1B), which is in accordance with the original report in human cell lines (Komor et al., 2016). We cloned 3 (#1–3) of the 8 bulk PCR products and sequenced multiple colonies from each primary product. For PCR products #2 and #3, each clone sequenced displayed C to T substitution, while PCR product #1 yielded one wildtype genotype besides the identical mutation genotypes (Fig. S1C), indicating highly efficient editing. To more carefully analyze the on-target editing effects, deep sequencing was applied to samples #2 and #3. In total, more than 3 M clean reads for each sample were generated. The results showed that only the 16th nucleotide distal from the PAM completely carried C to T conversion with the efficiency as high as 0.97 for sample #3, and 0.99 for sample #2. No other nucleotide alteration was detected (Fig. 1C). Besides, no on-target indel was found (Table S2). These results demonstrated the BE led to highly precise and efficient genome editing in human embryos. The same tests were performed for RNF2. T7EN1 cleavage bands were detected in 7 out of the 8 samples (#9–13, #15–16) (Fig. S2A). Sanger sequencing of PCR products confirmed C to T conversion in the 7 samples with cleavage (Figs. 1D and S2B). To further analyze the editing, 3 samples (#10–12) were selected for genotyping by TA cloning and subsequent sequencing. As reported before (Komor et al., 2016), in most cases, 2 cytosines (at the 18th and 15th nucleotide distal from the PAM) were simultaneously mutated to T, and triple C to T conversion (at the 18th, 15th, and 9th nucleotide distal from the PAM) also occurred (sample #10 and #12) (Fig. S2C). Collectively, these results demonstrated highly efficient and precise on-target base editing by BE3 in human embryos. We next tried to mutate the two genes simultaneously in the tripronuclear zygotes. To avoid possible toxicity, the concentration of each sgRNA was lowered to 25 nanogram per microliter. Nine embryos (#17–25) were collected and the target sites were analyzed by sequencing (Fig. 1A). For HEK293 site 4, the expected substitution in the sixteenth base distal from the PAM was observed in all samples (Fig. S3A), although the wild type genotype was also detectable in a few samples (Fig. S3B). A sample (#18) was randomly selected for on-target analysis by deep sequencing. The results showed that the conversion rate in the 16th C was about 0.68, which was consistent with the results of