AI-engineered proteins transport RNA better than lipid nanoparticles

Helmholtz Munich and TUM have assembled natural protein bricks and structures invented by generative AI. The best of the hundred variants tested delivers more RNA with less dose.

The teams of Helmholtz Munich and the Technical University of Munich presented on 2 September 2026, Nature, an unprecedented family of RNA carriers called Synthetic Transfer Vehicles (STV). The principle: to combine functional protein components from living to structural scaffolding generated by artificial intelligence, without seeking to imitate existing forms.

" We did not want to recreate nature, but to design new structures for a specific task: effective delivery of RNA," says Christoph Gruber, team leader at the Stem Cell Research Institute. More than a hundred variants have been screened, and the geometries farthest from those observed in the lifetime have proved to be the most effective.

The successful candidate, STV-C8, transported RNA in cell culture significantly better than the lipid nanoparticles used as a reference, and reached a comparable level of protein production with a significantly reduced RNA dose. In mice, he reached lung tissue without any observed toxic effect or immune reaction; In pigs, it was used to route CRISPR/Cas9 components for a targeted edition of the dystrophin gene in the muscle. Helmholtz Munich received €3.1 million in German public funding under the GO-Bio Next programme to continue preclinical development and prepare for the creation of a spin-off.