Optimized in vivo base editing restores auditory function in a DFNA15 mouse model

Wang et al. (2025) demonstrate full therapeutic chain in a cochlear gene editing model: SchABE8e (a next-generation adenine base editor) delivered via Anc80L65 synthetic AAV to neonatal mice corrects the Pou4f3-Q113* nonsense mutation causing DFNA15 (dominant non-syndromic deafness). In vitro SchABE8e achieved 48.5% editing of the Q113* codon. Neonatal cochlear injection produced near-complete hearing recovery sustained for at least four months. Biosafety analyses were favorable.

Author note: This paper was cited as “Zhang 2025” in prior h07 notes. The first author is Man Wang; Ziyu Zhang is a co-author (explaining the misattribution). PMID 40968144 is confirmed correct for Wang et al. 2025 Nature Communications. Corrected 2026-04-25.

Key finding

Anc80L65 + SchABE8e corrects a cochlear stop mutation in vivo with near-complete functional rescue at 4+ months. This is the closest published proof-of-concept for the h07 delivery chain (AAV → OHCs → precision base correction → sustained hearing recovery), despite using ABE (A→G) rather than PE. It validates the entire pipeline except the edit chemistry.

Numbers that matter

ParameterValueUnitsSourceConditions
SchABE8e in vitro editing efficiency48.5%AbstractCell culture, Pou4f3-Q113* locus
Functional hearing recoveryNear-completeAbstractNeonatal mouse, Anc80L65 injection
Recovery persistence≥4monthsAbstractSustained follow-up
AAV vectorAnc80L65AbstractPreferred for OHC cochlear delivery
Target genePou4f3 (POU4F3)AbstractQ113* nonsense mutation; DFNA15
Edit typeA→G (ABE); A•T→G•CAbstractCannot correct C→A as needed for Misha

Limitations

  • ABE edits A→G; Misha’s STRC variant requires C→A correction on coding strand — different edit chemistry. This paper validates the delivery and functional rescue but not the PE edit type.
  • Neonatal injection; efficacy in older animals (Misha’s age ~5 years) likely lower due to OHC maturation and reduced AAV access.
  • POU4F3 gene therapy (dominant): one allele correction sufficient. Misha’s DFNB16 is recessive; both alleles need consideration.
  • DFNA15 model (dominant): loss of POU4F3 transcription factor. DFNB16 (STRC loss-of-function) is different pathobiology.

Connections