Structure-based computational engineering of saCas9 PAM requirement

ABSTRACT SaCas9 is a major gene editing nuclease that is preferred for in vivo applications thanks to its relatively smaller size compared to that of spCas9. One limiting factor for the use of saCas9 is its strict PAM requirement of NNGRRT. In addition, there are limited efforts to reduce saCas9’s off-target editing rates. In Aim 1, we propose a computational approach to relax saCas9’s PAM requirement. We hypothesize that the interactions between key PAM recognition (KPR) residues of saCas9 and the side chains of PAM nucleotides determine the PAM requirement, and that mutating KPR residues by destroying their interactions with the DNA side chain while introducing favorable interactions with the DNA main chain will relax the PAM requirement. We have demonstrated the feasibility to relax the PAM requirement to NNNRRT in preliminary work. Here we will develop and optimize a computational method UniDesign to engineer saCas9s with further relaxed NNNRRN PAM (rr-saCas9s). In Aim 2, we propose to improve rr-saCas9’s safety. Recently we reported the development of mispCas9, in which a small size (36 amino acids) HDR (homology-directed repair)-promoting peptide Brex27 was fused to spCas9. Compared to spCas9, mispCas9 leads to increased knock-in rates as well as reduced off-target insertion and deletion (indel) events. Importantly, Brex27 can be used as a “plug and play” module to improve other gene-editing nucleases as long as their mechanism of action depends on the generation of double-strand breaks (DSB) of the genome. Here we will fuse Brex27 to rr- saCas9 (mirr-saCas9), and conduct experiments to characterize mirr-saCas9, in comparison with rr-saCas9 and saCas9, with a focus on PAM requirement, on-target editing efficacy, and off-target rates. The proposed mirr-saCas9, if successful, will add favorable features to saCas9 including: (i) 16 times more targetable sequences as a result of the relaxed PAM requirement (NNNRRN vs NNGRRT); and (ii) reduced off-target indel rates as a result of enhanced HDR at off-target sites. Furthermore, the validated computational model is adaptable for the improvement of other CRISPR variants for PAM modification and other desirable engineering.