A programmable dual-RNA-guided DNA endonuclease in adaptive bacterial immunity

Cas9 requires both a crRNA and a trans-activating crRNA (tracrRNA) to engage and cleave target DNA. crRNA alone — even when fully complementary to the target — supports neither specific DNA binding nor cleavage; tracrRNA, which base-pairs with the crRNA repeat, is what activates both Cas9's binding to the target and its cleavage activity.

Dual-RNA-guided Cas9 cuts both strands of target DNA at a blunt site three base pairs upstream of the PAM, and unlike the Type I CRISPR Cascade complex, it cleaves supercoiled as well as linear plasmid DNA. Cleavage proceeds at 0.3–1 min⁻¹ under single-turnover conditions — comparable to restriction endonucleases — and a single Cas9-tracrRNA:crRNA complex can act on multiple target DNA molecules rather than being used up in one reaction.

Cas9 uses two separate catalytic domains to cut the two strands of target DNA: the HNH domain cleaves the strand complementary to the crRNA, and the RuvC-like domain cleaves the noncomplementary strand. Inactivating either domain alone — the D10A mutation in RuvC or the H840A mutation in HNH — converts Cas9 into a nickase that leaves the other strand intact, producing a nicked open-circular rather than linear plasmid.

A tracrRNA truncated to just nucleotides 23–48, out of its native ~171-nucleotide length, still supports robust Cas9-catalyzed DNA cleavage when paired with a full-length crRNA. crRNA missing its 3′-terminal 10 nucleotides remains functional under the same pairing, but removing 10 nucleotides from the crRNA's 5′ end abolishes cleavage entirely — the guide's essential core is asymmetric and far smaller than either full-length RNA.

Cas9 target recognition depends on a PAM-proximal "seed" region in the crRNA: at least 13 contiguous base pairs between the crRNA and the target DNA next to the PAM are required for efficient cleavage, while up to six contiguous mismatches are tolerated in the PAM-distal 5′ region of the protospacer. In S. pyogenes cells, single-nucleotide protospacer mutations near the PAM block plasmid elimination by CRISPR/Cas, but equivalent mutations near the distal end do not.

Cas9 cleavage requires an NGG protospacer-adjacent motif (PAM) immediately 3′ of the target sequence on the noncomplementary DNA strand; mutating either guanine sharply reduces both cleavage and stable binding of the Cas9-tracrRNA:crRNA complex to the target. The PAM has no effect on cleavage of single-stranded DNA, indicating its role is to license duplex unwinding and R-loop formation rather than to activate the catalytic step directly.

Cas9, tracrRNA, and the crRNA repeat sequence are coevolved within a given CRISPR system: S. pyogenes Cas9 guided by tracrRNA:crRNA duplexes from more distantly related CRISPR/Cas systems fails to cleave DNA efficiently, and Cas9 orthologs from more distant bacterial species cannot use S. pyogenes tracrRNA:crRNA. A guide RNA generally only functions with the Cas9 ortholog from its own or a closely related system.

Fusing the 3′ end of crRNA to the 5′ end of tracrRNA through a short GAAA tetraloop produces a single chimeric guide RNA — the precursor of the modern sgRNA — that directs Cas9 cleavage as effectively as the natural tracrRNA:crRNA duplex, provided the chimera retains enough tracrRNA-derived sequence beyond the base-pairing region; a shortened version fails to support cleavage.

Changing only the 5′ targeting sequence of the chimeric guide RNA redirects Cas9 cleavage to a new DNA target. Five independently designed chimeric RNAs against different sites in the GFP coding sequence each guided Cas9 to cut the correct target in vitro, with no requirement beyond a GG-containing PAM adjacent to the site — establishing single-guide-RNA-programmed Cas9 as retargetable to essentially any DNA sequence of interest.