Blackjack mutations improve the on-target activities of increased fidelity variants of SpCas9 with 5′G-extended sgRNAs

The 5′ end of the sgRNA is capped inside SpCas9 by Glu1007 and Tyr1013, two residues joined by a surface loop, and this cap is what makes increased fidelity SpCas9 variants intolerant of 5′-G-extended sgRNAs. Deleting the loop makes room for the extension: replacing the segment between L1004 and K1014 with two glycine residues yields Blackjack-SpCas9-HF1 (B-SpCas9-HF1), which is active with 21-nucleotide, 5′-G-extended sgRNAs (21G-sgRNAs). Substituting Glu1007 and Tyr1013 with glycine while leaving the loop in place does not confer 21G tolerance, so it is removal of the loop rather than loss of these side chains that matters.

Blackjack loop-deletion mutations raise the on-target activity of increased fidelity SpCas9 variants with 21G-sgRNAs by up to 17-fold, and do so across eSpCas9, SpCas9-HF1, HypaSpCas9, evoSpCas9 and HeFSpCas9. The gain is not universal: on some targets, such as EGFP target site 22, B-evoSpCas9 loses activity even with a standard 20G-sgRNA.

Blackjack mutations increase target selectivity — lowering activity at some perfectly matched on-target sites — in every increased fidelity SpCas9 variant tested but not in wild-type SpCas9, measured by EGFP disruption across 50 target sites.

Disrupting the sgRNA 5′ cap raises fidelity as well as conferring 21G-sgRNA tolerance, and both effects run through the same interaction. Blackjack mutations increase the fidelity of every SpCas9 variant tested, assayed with 144 singly mismatched sgRNAs across 16 target sequences, and by GUIDE-seq on six targets both a 5′ G-extension of the sgRNA and the Blackjack mutations reduce the number of detected off-target sites and raise the on-target to off-target read ratio relative to wild-type SpCas9. In B-SpCas9, where the capping interaction is already removed, a 5′ G-extension adds little further fidelity.

Because Blackjack mutations both enable 21G-sgRNA use and independently raise fidelity, reverting one of the parent variant's own fidelity mutations back to its wild-type residue cancels the fidelity shift while keeping 21G tolerance. Reverting A661R in B-SpCas9-HF1 produces SpCas9-HF1-plus, whose fidelity matches SpCas9-HF1, and reverting A1003K in B-eSpCas9 produces eSpCas9-plus, whose fidelity matches eSpCas9. Not every reversion lowers fidelity: reverting A497N instead increased the fidelity of B-SpCas9-HF1 and decreased its on-target activity.

eSpCas9-plus and SpCas9-HF1-plus match the on-target activity of eSpCas9 and SpCas9-HF1 with 20G-sgRNAs across 25 EGFP targets while reaching 90% of their own 20G-sgRNA activity when charged with 21G-sgRNAs, against only 10% and 16% respectively for the parent variants. Their fidelity is unchanged from their parents, confirmed across 13 targets with 117 mismatching sgRNAs at 39 positions and by GUIDE-seq at seven sites. Blackjack mutations do not alter SpCas9 expression levels in the steady state.

Across 23 endogenous target sites in the human FANCF and VEGFA loci measured by amplicon resequencing in HEK293 cells, eSpCas9-plus and SpCas9-HF1-plus match their parent variants with 20G-sgRNAs and, with 21G-sgRNAs, reach 88% and 82% of wild-type SpCas9 activity — essentially the same relative level their parents reach with 20G-sgRNAs (93% and 82%). The plus variants therefore work with nearly identical efficiency whether a target requires a 20G- or a 21G-sgRNA.

The plus variants retain activity when delivered as pre-assembled ribonucleoproteins with in vitro transcribed 5′ G- or GG-extended sgRNAs, reaching up to 23-fold higher activity than their parent increased fidelity variants. With fully matching 20G-sgRNAs in ribonucleoprotein form, all variants including the parents show activity similar to wild-type SpCas9 across 31 EGFP target sequences, contradicting an earlier report that increased fidelity variants are strongly attenuated as ribonucleoproteins.

Using a 21G-tolerant Blackjack variant outperforms processing the extra 5′ nucleotide off the guide: across 19 targets the plus variant showed higher activity than either tRNA-based or ribozyme-based sgRNA processing, and both processing approaches reduced cleavage efficiency even for wild-type SpCas9 at some target sequences.

Blackjack variants reach targets that are otherwise unavailable to increased fidelity nucleases. Inserting EGFP downstream of the mouse Shadoo promoter through NHEJ repair offers five usable NGG PAM sites at relevant positions, none of them targetable with a 20G-sgRNA, and pre-screening those five with increased fidelity nucleases identified working target and nuclease combinations for generating the transgenic lines.

eSpCas9-plus works with 21G-sgRNAs expressed at the low levels typical of pooled library screens, making existing libraries usable with a high-fidelity nuclease. In three cell lines each expressing a single 21G-sgRNA from one integrated lentiviral copy, eSpCas9-plus approaches wild-type SpCas9 activity where eSpCas9 does not.