Engineered guide RNA scaffold with position-specific 2′-O-methyl modifications for enhanced in vivo stability

A chemically-modified CRISPR guide RNA (gRNA) scaffold bearing a position-specific 2′-O-methyl modification pattern that extends serum half-life from <5 minutes to ≥48 hours while preserving Cas9 ribonucleoprotein loading kinetics (k_on within 20% of unmodified). Designed for in vivo liver-targeted gene editing in familial hypercholesterolemia and related LDLR loss-of-function indications.

The novelty resides in the specific pattern of modifications across the scaffold — not the use of 2′-O-methyl per se, which is a known nucleoside analog. Uniform scaffold methylation stabilizes the gRNA but cripples Cas9 activity below 15% editing efficiency. Lattice’s pattern leaves the seed region (positions 1–10) and three scaffold loops unmodified and places 2′-O-methyl substitutions at positions 11–20 plus the tetraloop  [founder intake Q3] — the configuration that emerged from a systematic in vitro screen of 240 permutations.

The practical result is 67% in vivo editing efficiency at 1 mg/kg in non-human primate cohorts vs. 12% for the unmodified scaffold  [uploaded NHP cohort report, pp. 8–11] — a 5.6× improvement that translates directly to therapeutic-window LDL-C reduction in the target indication and brings single-dose dosing into feasibility range.

The scaffold is divided into three regions. Positions 1–10 (the seed, which base-pairs to the protospacer and determines Cas9 specificity) are left unmodified — 2′-O-methyl substitution in the seed reduces on-target cleavage by >60% via disruption of the RNA-DNA heteroduplex geometry  [methods §2.1]. Positions 11–20 of the spacer and the tetraloop of the repeat-antirepeat region carry 2′-O-methyl substitutions: these positions do not make base-pair contacts with the protospacer but are the primary substrates for serum exonuclease attack.

Three scaffold loops — identified by H-bond mapping against the Cas9-gRNA cryo-EM structure  [architecture doc §3.3] — are also left unmodified to preserve the RNA tertiary fold that the Cas9 HNH / RuvC domains recognize during ribonucleoprotein assembly. The resulting gRNA is synthesized by solid-phase phosphoramidite chemistry, purified by dual-column HPLC with inline mass-spec QC (see cand-l005), and formulated into hepatocyte-selective LNPs via the cleavable-linker approach in cand-l003 or direct aqueous-core encapsulation.

The prevailing approach in the literature over the past three years has been either (a) uniform scaffold methylation — which extends half-life but collapses editing efficiency below therapeutic relevance — or (b) limited terminal modifications (first and last 2–3 nucleotides), which minimally impact activity but provide only modest half-life extension (<15 minutes). A practitioner of ordinary skill would not have been motivated to leave the seed and three internal loops unmodified while methylating the tetraloop specifically — the tetraloop is conventionally viewed as structurally critical and therefore untouchable.

Further, the systematic in vitro screen of 240 permutations is not reproduced in any single published reference. The closest analogues ([ref-3], [ref-4]) each test fewer than 30 configurations and do not vary position 11–20 against tetraloop substitution orthogonally; the non-obviousness of the Lattice pattern lies in the interaction effect between these two regions, which only surfaced in the full permutation screen.