The effect of sequence mismatches on binding affinity and endonuclease activity are decoupled throughout the Cas9 binding site

ABSTRACT The CRISPR-Cas9 system is a powerful genomic tool. Although targeted to complementary genomic sequences by a guide RNA (gRNA), Cas9 tolerates gRNA:DNA mismatches and cleaves off-target sites. How mismatches quantitatively affect Cas9 binding and cutting is not understood. Using SelexGLM to construct a comprehensive model for DNA-binding specificity, we observed that 13-bp of complementarity in the PAM-proximal DNA contributes to affinity. We then adapted Spec-seq and developed SEAM-seq to systematically compare the impact of gRNA:DNA mismatches on affinity and endonuclease activity, respectively. Though most often coupled, these simple and accessible experiments identified sometimes opposing effects for mismatches on DNA-binding and cutting. In the PAM-distal region mismatches decreased activity but not affinity, whereas in the PAM-proximal region some reduced-affinity mismatches enhanced activity. This mismatch-activation was particularly evident where the gRNA:DNA duplex bends. We developed integrative models from these measurements that estimate catalytic efficiency and can be used to predict off-target cleavage.