At the risk of becoming structure-centric (three crystal structures in four posts), I couldn’t pass up commenting on the Jinek Lab’s beautiful structure of SpyCas9 in complex with sgRNA and target DNA including an NGG PAM. One big take-home here is that the PAM is read out by two arginines (1333 and 1335) and a lysine (1107). The PAM itself is hybridized to the non-target DNA strand, but the DNA immediately downstream is flipped 180 degrees so that it can be read out by the RNA protospacer! This bit of structural juggling is made possible by a serine, which together with a backbone contact forms an interaction the authors term a “phosphate lock”.
Like any great science this paper raises as many questions as it answers. Mutating the PAM-contacting arginines to alanine abolishes DNA binding, but switching them to the identities found in organisms with non-NGG PAMs doesn’t switch PAM specificity. So is it even possible to make SpyCas9 recognize other PAMs? You can bet that many groups are working on that very problem, and I’m sure this paper has given them some ideas.