Characterization of Staphylococcus aureus Cas9: a smaller Cas9 for all-in-one adeno-associated virus delivery and paired nickase applications

Staphylococcus aureus Cas9 (SaCas9), a 1053 amino acid nuclease, recognizes an NNGRRT protospacer adjacent motif and tolerates NNGRRV only weakly. In a luciferase knockdown assay where an invariant target sequence is placed next to variable PAMs, targeting is strongest at NNGRRT and moderate at NNGRRV, and the same preference holds at the endogenous human VEGFA and B2M loci, where NNGRRT PAMs give substantially higher indel rates than NNGRRV (unpaired t-test P < 0.0001) though NNGRRV still permits some nuclease activity.

SaCas9 matches SpCas9 in cleavage efficiency where their PAM requirements overlap. Directed to identical genomic target sites bearing NGGRR(T) PAMs — with 24-mer spacers beginning in a target-matching 5′ G for SaCas9 against 21-mer spacers for SpCas9 — SaCas9 yields indels at rates comparable to SpCas9 and outperforms it in some cases. At sites with NNGRRV PAMs, which lack SaCas9's preferred terminal T, SaCas9 cleaves significantly less than SpCas9.

SaCas9 works across a range of guide RNA spacer lengths, with 24 to 20 bases the most efficient. Tested with G-initiating sibling gRNAs that share a target site but differ in spacer length, across VEGFA, CCR5 and an integrated GFP reporter in HEK293-GFP cells, 19-mers and 18-mers retain some activity while 17-mers and shorter are inactive.

SaCas9 can be converted to a nickase by inactivating either catalytic domain, using residues identified by aligning it to SpCas9: D10A disables the RuvC domain and N580A disables the HNH domain, the latter corresponding to the SpCas9 N863A substitution.

SaCas9 nickases require a pair of guide RNAs to generate indels, confirming they nick rather than cleave. With a single gRNA, the N580A nickase typically yields no indels and the D10A nickase yields 0% to about 8%; with two gRNAs, indel rates extend up to about 60%. The D10A nickase consistently outperforms N580A.

SaCas9 and its two nickases leave distinguishable repair signatures at the same locus: wild-type SaCas9 indels are predominantly deletions (73.3%), D10A paired-nickase indels split roughly evenly between insertions and deletions (45.1% deletions, 54.9% insertions), and N580A paired-nickase indels are predominantly insertions (78.9%).

The usable offset window for SaCas9 D10A paired nickases is wider than SpCas9's. Guide pairs with offsets below zero between their 5′ ends yield minimal indels, most pairs offset between 0 and 125 bp yield substantial indels, and — in contrast to previously reported SpCas9 behaviour — some pairs offset between 125 and 170 bp still yield modest indel rates.

SaCas9's compact size permits an all-in-one AAV carrying the nuclease and two guide RNA cassettes. The SaCas9 open reading frame is 3159 bp, leaving roughly 1.8 kb within AAV's approximately 5.0 kb packaging limit for regulatory elements and guides; driving SaCas9 from an EFS promoter with a mini poly(A) tail and adding two U6-driven gRNA cassettes gives a functional vector of 4859 bp total. Packaged with a serotype 2 capsid and transduced into HEK293 and HEK293FT cells, these vectors express SaCas9 by western blot and produce indels at the targeted loci, enabling two-guide approaches such as targeted deletions and multiplexed knockouts from a single vector.

SaCas9 is more specific than SpCas9 at a target with known off-target sites, when both are directed by 20-base spacers. GUIDE-seq at VEGFA site 3 recovers many of the previously reported SpCas9 off-targets, some with read counts nearly as high as the on-target site, whereas SaCas9 gives high on-target reads with only single-digit read counts at comparatively fewer off-targets. All eight SaCas9 off-target sites identified were also SpCas9 off-targets, so SaCas9's off-target repertoire at this site is a subset of SpCas9's.