Massively Targeted Evaluation of Therapeutic CRISPR Off-Targets in Cells

Abstract Sensitive and high-throughput evaluation of CRISPR RNA-guided nucleases (RGNs) off-targets (OTs) in cells are essential for safety assessment and advancing RGN-based gene therapies. Here we apply a modified library approach, SURRO-seq, for simultaneously evaluating thousands of off-target sites for therapeutic RGNs in cells. The SURRO-seq captures RGN-induced indels in barcoded surrogate off-target sites in cells by a pool of lentiviral vectors and targeted deep sequencing. We first evaluate 170 previously investigated OTs from 11 RGNs with SURRO-seq in HEK293T cells. SURRO-seq captures nearly 100% OTs found by T7E1, most (70%) GUIDEseq-identified OTs, and approximately half (51%) OTs found by CIRCLE-seq. We then apply SURRO-seq to evaluate 7140 OTs from 110 therapeutic RGNs. 105 RGNs are found to introduce significantly detectable indels in 753 OTs, of which 180 OTs have an indel frequency above 3%. Notably, significantly detectable off-target indels are identified in 37 cancer genes including tumor suppressor gene exons. We deep sequence 23 endogenous genome loci in five human cell lines and further validate that SURRO-seq can sensitively capture off-targets with indel frequency below 0.1%. High fidelity RGNs have substantially reduced indel rates introduced at OTs but significantly detectable indels are still found in some OTs. Analyses of OTs with significantly detectable indels indicates that thermodynamically stable wobble base pair (rG•dT) and free binding energy strongly affect RGN specificity. Our study emphasizes the necessity of carefully selecting high fidelity RGNs and evaluating therapeutic RGN to minimize inevitable off-targets.