CRISPR-sub: Analysis of DNA substitution mutations caused by CRISPR-Cas9 in human cells.

ABSTRACT

CRISPR-Cas9 induces DNA cleavages at desired target sites in a guide RNA-dependent manner; DNA editing occurs through the resulting activity of DNA repair processes including non-homologous end joining (NHEJ), which is dominant in mammalian cells. NHEJ repair frequently causes small insertions and deletions (indels) near DNA cleavage sites but only rarely causes nucleotide substitutions. High-throughput sequencing is the primary means of assessing indel and substitution frequencies in bulk populations of cells in the gene editing field. However, it is difficult to detect bona fide substitutions, which are embedded among experimentally-induced substitution errors, in high-throughput sequencing data. Here, we developed a novel analysis method, named CRISPR-Sub, to statistically detect Cas9-mediated substitutions in high-throughput sequencing data by comparing Mock- and CRISPR-treated samples. We first pinpointed 'hotspot positions' in target sequences at which substitution mutations were quantitatively observed much more often (p > 0.001) in CRISPR- versus Mock-treated samples. We refer to the substitution mutations in defined hotspot positions as 'apparent substitutions' and ultimately calculated 'apparent substitution frequencies' for each target. By examining 51 endogenous target sites in HeLa cells, we found that the average apparent substitution frequency was 0.8% in all queries, that apparent substitutions frequently occur near CRISPR-Cas9 cleavage sites, and that nucleotide conversion showed no meaningful nucleotide preference patterns. Furthermore, we generated NHEJ-inhibited cell lines (LIG4-/- ) by knockout of the gene encoding ligase IV and found that the apparent substitution frequencies were significantly decreased in LIG4-/- cells, strongly suggesting that DNA substitutions are generated by the NHEJ pathway.