Dual-AAV delivering split prime editor system for in vivo genome editing

Zhi et al. (2022) established that prime editors — too large (~6.3 kb) to fit in single AAV vectors — can be split across two AAV particles using Rma intein-mediated protein trans-splicing. They screened split sites and identified split-PE1005 and split-PE1024 as highly active candidates. Dual-AAV1 vectors delivered these editors to human cells; dual-AAV8 vectors achieved in vivo editing of the Dnmt1 gene in adult mouse retina following subretinal injection. The retina is a post-mitotic sensory tissue — closely analogous in cell biology to the cochlea — making this the first published demonstration that split-intein dual-AAV PE works in post-mitotic sensory tissue in vivo.

Author note: This paper was cited in earlier h07 notes as “Villiger 2021” (PMID 34298129). The correct first author is Shengyao Zhi (Sun Yat-sen University). “Villiger 2021” dual-AAV split PE retina paper does not exist in PubMed. Corrected 2026-04-25.

Key finding

Dual-AAV8 + Rma intein split at PE1005 or PE1024 achieves in vivo prime editing in adult mouse retina. The scientific claim (dual-AAV split PE works in post-mitotic sensory tissue) is confirmed correct; only the attribution was wrong.

Numbers that matter

ParameterValueUnitsSourceConditions
Target locus (in vivo)Dnmt1geneAbstractAdult mouse retina, subretinal injection
Delivery vectorDual-AAV8AbstractSubretinal route; post-mitotic photoreceptors / RPE
Splicing mechanismRma intein trans-splicingAbstractProtein reconstitution post-transduction
Active split sites identifiedPE1005, PE1024positionsAbstractHigh-activity candidates from screen
In vitro delivery vectorDual-AAV1AbstractHuman cell line validation
Edit types demonstratedBase transversions + insertionsAbstractMultiple endogenous genomic sites

Limitations

  • Editing efficiency percentage not quantified in abstract; full PMC text (PMC8753371) has the numbers.
  • Dnmt1 is an endogenous mouse locus, not a therapeutic target; proof-of-concept only.
  • Retina ≠ cochlea: injection route (subretinal vs round window membrane) differs; AAV8 tropism for retina may not transfer directly to OHCs (Anc80L65 preferred for cochlea).
  • Epub July 2021; published January 2022 — cited year varies by source.

Connections