Cas9 cannot cut just anywhere - it requires a short motif called a PAM next to its target - and that constraint limits which sites are editable. US11286468B2 - "Engineered CRISPR-Cas9 nucleases with altered PAM specificity," issued March 29, 2022 to The General Hospital Corporation - claims variants that relax or redirect that constraint.

The protospacer-adjacent motif (PAM) is the genome's gatekeeper for Cas9: no PAM nearby, no cut. Wild-type SpCas9 needs an 'NGG' motif, which leaves swaths of the genome inaccessible. The claimed invention re-engineers the protein so it recognizes different PAMs, and claim 1 defines that re-engineering with real precision. It recites a Streptococcus pyogenes Cas9 (SpCas9) protein with at least 80% identity to SEQ ID NO: 1 carrying mutations at all six of a named set of positions - D1135, S1136, G1218, E1219, R1335, and T1337 - and then enumerates the permitted residue combinations at those six positions as a long list of six-letter substitution strings (LRSVQL, LRKIQK, VRKIQK, and so on). The altered PAM recognition is the limitation, and it is tied to specific changes at specific residues that contact the PAM.

"Engineered CRISPR-Cas9 nucleases with altered and improved PAM specificities and their use in genomic engineering, epigenomic engineering, and genome targeting."- U.S. Patent No. 11,286,468 source

The six positions are not arbitrary - they sit in the PAM-interacting domain of SpCas9, the region that reads the motif next to the target. By mutating that cluster, the claimed variants change which motif the protein will accept, which is the molecular basis for the expanded targeting range. Claim 23 mirrors claim 1 for the encoding nucleic acid, and claims 24-27 extend to vectors, regulatory-domain constructs, and host cells, so the estate covers the protein, the gene, the delivery vehicle, and the cell.

The dependent claims show that the engineered protein is a platform, not just a nuclease. Claims 2-3 add the nuclease-dulling mutations (D10A/N, H840A/N/Y) that turn the cutter into a nickase or a dead Cas9. Claims 4-21 then fuse the PAM-altered protein to a catalog of heterologous domains: transcriptional activators (VP16, VP64, VPR), repressors (KRAB, ERD, SID), DNA-methylation modifiers (DNMT, TET1), histone modifiers, base editors - both a cytidine deaminase (claim 17, APOBEC/AID/CDAT) and an adenosine deaminase (claim 18, including engineered TadA) - FokI (claim 21), and biological tethers like MS2 (claim 20). In other words, the altered-PAM scaffold is claimed as the targeting chassis for editing, epigenome engineering, and base editing alike - which is exactly the "genomic engineering, epigenomic engineering, and genome targeting" the abstract names.

Why does PAM engineering matter commercially? Many therapeutically interesting genomic positions sit at sites without a conveniently placed NGG. A variant with broadened PAM recognition makes those positions addressable, expanding the tool's reach. The value of the claim is that expanded addressable space, not a new cutting chemistry. And the claim-scope question is correspondingly precise: it turns on whether an accused variant carries mutations at all six named positions within one of the enumerated residue sets (or their equivalents). Describing this patent as 'a Cas9 patent' badly understates the specificity - the claim is about which residues were changed to read which motif.

The "all six" requirement in claim 1 is the single most important construction point, and it is easy to miss. The claim does not cover any SpCas9 with a mutation somewhere in the PAM-interacting domain; it requires mutations at all six of D1135, S1136, G1218, E1219, R1335, and T1337, with the residues at those positions drawn from one of the enumerated six-letter strings. That conjunctive limitation is a precise fence. A variant mutated at only three or four of those positions, or carrying a six-letter combination not on the list, is outside the literal claim. The patentee traded breadth for definiteness: the claim is exactly as wide as its enumerated residue sets and no wider.

The functional payoff of those six positions is structural. They cluster in the region of SpCas9 that physically reads the PAM, so changing them retunes which motif the protein will tolerate - the molecular basis for the "altered and improved PAM specificities" of the title. The claim thereby protects a re-tuned reading head, and the dependent claims show what gets bolted onto that head. Claims 15-18 are especially notable: they put base editors on the PAM-altered scaffold, naming both cytidine deaminases (APOBEC, AID, CDAT in claim 17) and adenosine deaminases (including engineered TadA in claim 18). That means the grant reaches PAM-flexible base editing, not just PAM-flexible cutting - a meaningful extension given how much of the editing field has moved toward precise base changes.

Claims 28-34 then cover the use: altering a cell's genome (claim 28), with sub-claims for stem cells, embryos, and in vivo settings, and altering a double-stranded DNA molecule in vitro (claims 32-33), including as a ribonucleoprotein complex (claim 34). So the estate spans protein, nucleic acid, vector, host cell, and method, all keyed to the same six-residue definition. For freedom-to-operate, the practical takeaway is that the navigability of this claim is unusually concrete: a competitor checks its variant's residues at six named positions against an enumerated list, and the answer is largely deterministic - a clarity that is rare in the otherwise tangled CRISPR thicket.

For the landscape, PAM-engineering grants are a distinct sub-estate within the CRISPR thicket. They do not compete with the foundational Cas9 composition claims - they layer on top, expanding the tool's reach, and the fusion claims show they layer over the base-editing and epigenome-editing estates too. Any program that needs to edit a PAM-poor site, or to deploy a base editor or activator at such a site, has to navigate this six-residue claim separately from the core machinery patents.