Compact type II-D Cas9 nucleases for efficient and specific genome editing

Abstract

Compact CRISPR nucleases are attractive for therapeutic genome editing because their small coding sequences facilitate delivery by adeno-associated virus. Type II-D Cas9 (Cas9d) enzymes constitute the most compact Cas9 subtype, yet only a few orthologs have demonstrated mammalian genome-editing activity, leaving it unclear whether this activity is general or exceptional. Here, we mined the IMG/M metagenomic database and identified five previously uncharacterized MG102-like Cas9d orthologs (∼950 amino acids) that share the hallmark genomic, sequence, and structural features of type II-D Cas9. Two of them, Cas9d-1 and Cas9d-4, recognized a 5’-NRC-3’ protospacer-adjacent motif and edited endogenous human loci with efficiencies up to 20.1%, exceeding Streptococcus pyogenes Cas9 at one site, while producing deletion-biased outcomes and no detectable off-target activity. Notably, both orthologs edited more efficiently than the sole previously validated member of this lineage, MG102-2, when assayed side by side under identical conditions. These findings establish compact MG102-like Cas9d orthologs as robust and specific genome editors and provide promising, single-AAV– compatible scaffolds for in vivo therapeutic genome editing.