← All hypotheses

Programmable Recombinases: Skip the Split Entirely

EMERGING

Dual-AAV splits the gene in two and hopes both halves land in the same cell. Mini-STRC removes the disordered N-terminal so the gene fits one AAV. Both are workarounds for the same constraint: AAV can't carry more than 4.7 kb.

Programmable recombinases ignore that constraint. They insert full-length DNA at a specific genomic location, independent of the cell's own repair machinery. No splitting. No truncation. Full-length stereocilin, precisely integrated.

In January 2026, Eli Lilly signed a $1.12 billion deal with Seamless Therapeutics for exclusive rights to this technology in hearing loss. That's not a research grant. That's a bet.

How recombinases work

Traditional CRISPR cuts DNA and relies on the cell to repair it. Recombinases are different: they catalyze the insertion directly, no double-strand break needed, no dependence on HDR or NHEJ pathways. The recombinase recognizes a specific target sequence, opens it, and threads the new DNA in. One enzyme, one reaction, one result.

Three approaches to the STRC size problem

Property Dual-AAV Mini-STRC Recombinase
Payload size Full-length (split across 2 vectors) Truncated (3.2-3.5 kb) Full-length (single insertion)
Vectors needed 2 AAV particles 1 AAV particle 1 vector (AAV or LNP)
Integration Episomal (no integration) Episomal (no integration) Site-specific genomic integration
Persistence Years (episomal loss possible) Years (episomal loss possible) Permanent (integrated into genome)
Efficiency bottleneck Co-transduction + recombination (27%) Single transduction (78%) Integration efficiency (unknown)
Clinical status Phase 1/2 (OTOF only) Preclinical (computational) Preclinical (no cochlear data)

The Lilly signal

Eli Lilly spent $2.4 billion on hearing loss in 6 months. $1.12B for Seamless recombinases (Jan 2026) and $1.3B for Rznomics RNA editing (May 2025). That's not a side project. Big pharma sees cochlear gene therapy as a massive market. For context: ~2.3 million patients carry STRC mutations alone. DFNB16 is the second most common genetic hearing loss.

What this means for STRC

If recombinases work in cochlear hair cells, they solve every problem at once: full-length protein, single vector, permanent integration, no episomal loss. Our mini-STRC approach remains valid (smaller payload, proven AAV delivery), but it's a workaround. Recombinases would be the direct solution.

The catch: nobody has demonstrated recombinase-mediated gene insertion in inner ear hair cells. The technology works in cell lines and some in vivo contexts, but cochlear delivery and OHC-specific targeting are unsolved. Timeline: 3-5 years to preclinical data, optimistically.

Reality check: Programmable recombinases are not yet demonstrated in cochlear cells. Lilly's deal covers the technology platform, not a specific STRC program. Integration efficiency, off-target insertion rates, and OHC-specific delivery remain unknown. This is a 5-10 year horizon technology for STRC specifically.

← All hypotheses