Two recently engineered SpCas9 variants, namely xCas9 and Cas9-NG, show promising potential in improving targeting specificity and broadening the targeting range. In this study, we evaluated these Cas9 variants in the model and crop plant, rice. We first tested xCas9-3.7, the most effective xCas9 variant in mammalian cells, for targeted mutagenesis at 16 possible NGN PAM (protospacer adjacent motif) combinations in duplicates. xCas9 exhibited nearly equivalent editing efficiency to wild-type Cas9 (Cas9-WT) at most canonical NGG PAM sites tested, whereas it showed limited activity at non-canonical NGH (H = A, C, T) PAM sites. High editing efficiency of xCas9 at NGG PAMs was further demonstrated with C to T base editing by both rAPOBEC1 and PmCDA1 cytidine deaminases. With mismatched sgRNAs, we found that xCas9 had improved targeting specificity over the Cas9-WT. Furthermore, we tested two Cas9-NG variants, Cas9-NGv1 and Cas9-NG, for targeting NGN PAMs. Both Cas9-NG variants showed higher editing efficiency at most non-canonical NG PAM sites tested, and enabled much more efficient editing than xCas9 at AT-rich PAM sites such as GAT, GAA, and CAA. Nevertheless, we found that Cas9-NG variants showed significant reduced activity at the canonical NGG PAM sites. In stable transgenic rice lines, we demonstrated that Cas9-NG had much higher editing efficiency than Cas9-NGv1 and xCas9 at NG PAM sites. To expand the base-editing scope, we developed an efficient C to T base-editing system by making fusion of Cas9-NG nickase (D10A version), PmCDA1, and UGI. Taken together, our work benchmarked xCas9 as a high-fidelity nuclease for targeting canonical NGG PAMs and Cas9-NG as a preferred variant for targeting relaxed PAMs for plant genome editing.
