Dyno Therapeutics has unveiled a major expansion of its AI-driven gene delivery platform, announcing two new AAV capsids designed for central nervous system and muscle targeting.

 

This positions its platform as a rapidly evolving engine for next-generation genetic medicine.

 

At the core of Dyno’s pitch is a persistent bottleneck in gene therapy: getting genetic payloads to the right tissues safely and efficiently. Many current therapies require high doses, increasing toxicity risk and cost while limiting access. 

 

Dyno says its AI models—trained on billions of in vivo non-human primate measurements—are designed to engineer AAV capsids that improve tissue targeting, reduce liver exposure, enhance cross-species translation, and scale more efficiently for manufacturing.

 

“Each capsid we launch demonstrates the power of combining deep biological insight with massively parallel multiplexed in vivo experiments and state-of-the-art AI models for protein design, and these new capsids raise the bar again,” said Eric Kelsic, CEO and Cofounder of Dyno. 

 

“Every year, partners now have more options and greater confidence in identifying a suitable gene delivery vector for their therapeutic program. This is how it should feel to be rapidly advancing the frontiers of genetic medicine.”

 

Among the headline developments is Dyno-9zh, a capsid engineered for cross-species CNS delivery using ALPL-binding to cross the blood-brain barrier. 

 

The company says it enables broad brain and spinal cord transduction following intravenous administration across mice, non-human primates, and humans, with significantly reduced liver exposure compared to AAV9. 

 

In non-human primates, Dyno-9zh reportedly achieved up to 50% neuronal transduction in the premotor cortex at a dose of 3×10¹³ vg/kg, while maintaining compatibility with established AAV9 manufacturing processes.

 

For CNS delivery via a different mechanism, Dyno-yp2—first disclosed in January 2026—uses TfR1-mediated transport to cross the blood-brain barrier. 

 

In humanized TfR mouse studies, the company reports near-complete CNS transduction across key regions at higher dosing, with up to 98% of neurons in cortex and striatum and 97% in spinal cord transduced at 2×10¹³ vg/kg. It also showed substantially reduced liver biodistribution compared to both AAV9 and competing constructs.

 

On the neuromuscular front, Dyno introduced Dyno-n96, an optimized AAVrh74-based capsid aimed at skeletal and cardiac muscle delivery. 

 

In non-human primates, it reportedly transduced about 74% of skeletal myofibers and 16% of cardiomyocytes at 5.2×10¹² vg/kg—described as a markedly lower dose than that required for some existing muscular dystrophy gene therapies. The company also highlighted satellite cell targeting, suggesting potential applications in gene editing approaches.

 

Updated data for Dyno-bn8, a previously launched neuromuscular capsid, further reinforced high muscle transduction levels, with up to 88% skeletal muscle and 40% cardiac tissue targeting in non-human primates, along with binding to a receptor conserved across species.

 

Beyond delivery vehicles, Dyno also showcased its AI foundation models for protein design. Dyno Psi-1, trained on NVIDIA DGX Cloud, is positioned as an open-weight model capable of generating protein binders as therapeutic candidates. 

 

The company says it has demonstrated sub-nanomolar binding performance and success against difficult targets such as the SARS-CoV-2 receptor-binding domain, outperforming other open-weight approaches in experimental validation.