Useful paper, comparing multiple Cas proteins in mammalian cells. A few thoughts... (thread) genomebiology.biomedcentral.com/articles/10.11…
First, a word of caution on generalizability. All Cas enzymes are expressed with the same NLS / tag arrangements. Totally obvious, experimentally, why this was done. However, can't assume they do (or don't) interfere with function equally.
2) Likewise, 100% understand why they used sites shared across PAM requirements, to hold chromatin influence constant. However, possible that this systematically harmed one Cas at expense of other (e.g. if enforced overlap led to sub-optimal sequence choice for one)
Fig. 2a - expression levels are not necessarily *causal* for observed difference. Could just be small sample size / sequence preference. That, in all cases, difference b/w hi and low gets smaller with optimal length guides is maybe a hint
Before I keep reading, I should point out explicitly that I am *really* enjoying this paper, a really terrific and systematic study!
FInd this quote interesting: "Hence, our results suggest that there is a compromise between cleavage efficiency and specificity of naturally occurring Cas endonucleases" -- were eCas9, Cas9-HF, etc. really able to tease apart this trade-off?
Fig. 3 -- by using the same oligo for repair, and comparing SpCas9 to AsCpf1, the cut sites are in different places, so is it really fair comparison? Course, if you hold *that* constant then you're testing diff. oligos -- hard expt., needs many, many unique seqs to be definitive
(especially since one produces staggered cuts, one blunt)
"thereby suggesting that each genomic locus may have an inherent ssODN
strand preference" -- while this, initially, may seem annoying, since generally one is limited to so few guides to choose from for an editing expt., trying both donor templates isn't that much more work
strand preference" -- while this, initially, may seem annoying, since generally one is limited to so few guides to choose from for an editing expt., trying both donor templates isn't that much more work
In Discussion: "while GUIDE-seq requires the incorporation of blunt double-stranded oligodeoxynucleotides (dsODNs) into DNA breakage sites, which may inherently be biased against the staggered cuts generated by Cpf1." - wondered about this myself
Overall, *very* useful guidance. But please note that even though this was a ton of work, *still* relatively small sample sizes, so for your favorite edit of interest, probably still worth trying several Cas enzymes in-parallel, rather than in-series rounds of trial and error
