Efficient prime editing in mouse brain, liver and heart with dual AAVs

Davis et al. (2024) engineered dual-AAV split-intein prime editor systems (v1em and v3em PE-AAV) that overcome AAV cargo limits (~4.7 kb) by splitting the PE3 construct across two viral particles with intein-mediated reassembly. In adult mice, these systems achieved post-mitotic editing efficiencies that substantially exceed prior PE benchmarks: up to 42% in cortex, 46% in liver, 11% in heart. Safety profiling showed no detectable off-target edits and no significant liver enzyme changes, establishing a therapeutic-grade delivery architecture directly applicable to h07.

Key finding

Dual-AAV split PE can edit post-mitotic cortical neurons at 42% unenriched efficiency in vivo — the highest reported as of 2024. Protective mutations for Alzheimer’s (brain) and coronary artery disease (liver) were introduced as proof of therapeutic utility. The v3em variant outperformed v1em in most tissues.

Numbers that matter

ParameterValueUnitsSourceConditions
Cortical neuron editing efficiency42%Abstract / Fig 3Adult mouse, dual-AAV8, v3em PE-AAV, unenriched
Liver editing efficiency46%Abstract / Fig 4Adult mouse, dual-AAV8, v3em PE-AAV, unenriched
Heart editing efficiency11%Abstract / Fig 4Adult mouse, dual-AAV8, v3em PE-AAV, unenriched
Off-target edits detected0countAbstractWhole-genome sequencing
PE payload size (unfragmented)~6.3kbknown PE3 architectureToo large for single AAV
AAV packaging limit~4.7kbknownStandard rAAV

Limitations

  • Adult mouse cortex ≠ OHCs: OHCs are more terminally differentiated and have distinct AAV transduction kinetics. Cochlear data absent.
  • Efficiency varies dramatically by tissue (11% heart vs 42% cortex), suggesting cell-type-specific barriers matter.
  • v3em vs v1em head-to-head only in specific tissues; best variant for cochlea is unknown.
  • No long-term persistence data beyond the acute experiment window reported in abstract.

Connections