Tan et al., (2019) created a high fidelity SaCas9 using structure guided techniques.
CRISPR-Cas9 proteins have widely been used for a variety of genome editing applications, albeit with serious off-target activities. For a CRISPR-Cas system to work, a guide, CRISPR RNA (cRNA) and a protospacer adjacent motif (PAM) must be present to cut at a site of interest.
Two different Cas9 proteins, one from Streptococcus pyogenes (SpCas9) and one from Staphylococcus aureus (SaCas9) are commonly used in genome editing experiments. The latter is significantly smaller than the former, and can be packaged into viral vectors for delivery, as is common practice for in vivo work.
In order to create highly specific SpCas9 variants, structure-guided techniques modifying amino acid resides in contact with DNA strands, or random mutagenesis strategies with directed selection have been used. These studies, however, have not yet sought to create high-fidelity SaCas9 variants.
In order to create this high fidelity SaCas9, Tan et al., (2019) used structure guided techniques, examining the crystal structure of the SaCas9 guide complex and identifying amino acid residues that formed polar contacts close to the target DNA strand.
Four single amino acid (R245A, N413A, N419A, and R654A) mutant SaCas9s were made using alanine substitutions to remove this polar contact, and their effects were compared against WT-SaCas9 for both canonical and noncanonical PAM sites.
Genome-wide unbiased identification of double-stranded breaks by sequencing (GUIDE-seq) was used to compare both on- and off-target editing effects.
The authors then went on to test the combined effect of all four mutants, constructing a fully-factorial design for all mutants, but this time only for the canonical PAM sites. Instead of including a non-canonical site, the authors opted to include an additional site that was found to have the greatest number of off-target effects among canonical PAM sites from a previous study (Kleinstiver et al., 2015).
Of all the mutated SaCas9s, those containing R245A and N413A performed the best. The quadruple mutant (SaCas9-HF) had the lowest-number of off-target effects, and these effects were shared with the WT-SaCas9.
The authors were able to further demonstrate the utility of these four mutations by enhancing the KKH-SaCas9, an SaCas9 with a broader range then the WT-SaCas9. The resultant KKH-HF-SaCas9 outperformed the unmodified KKH-SaCas9 with increased on-target frequencies and a reduced number of off-target sites across canonical and non-canonical sites.
The SaCas9-HF also worked well when delivered by an adeno-associated vector (AAV) (in this case AAV8 in human retinal pigmented epithelium cells), however it’s on-target efficiency was much lower than the AAV delivered WT-SaCas9 (18.4 to 50.9%). The authors found no differences in performance between WT-SaCas9 and SaCas9-HF when sgRNAs of varying spacer lengths were used, although they did verify Friedland et al. (2015)’s findings that SaCas9s work best with spacers of 20 – 24 bases.
Lastly, the authors compared their newly designed SaCas9-HF with other high-fidelity Cas nucleases. Their SaCas9-HF had comparable on-target and off-target effects as a previously identified high-fidelity SaCas9 variant, S-HF, (identified by Slaymaker et al., 2016) as well as high-fidelity SpCas9s, eSpCas9, SpCas9-HF1 and HyPa-SpCas9.
Overall, it appears that the SaCas9-HF created by Tan et. al. (2019) has improved efficiency across multiple platforms in comparison to its wildtype counterpart. The modified nuclease still needs to be tested for its effects on other target sites, cell types, as well as with other delivery systems.
Furthermore, more sensitive sequencing methods such as circularization for in vitro reporting of cleavage effects by sequencing (CIRCLE-seq) could help detect even low frequency (<0.1%) off-target effects. The study serves as a good jumping off point for the creation of high-fidelity, and potentially also highly-specific SaCas9 based genome editing nucleases.