Antibodies are among medicine’s most powerful tools, but they have a big limitation: they normally work outside cells. Many of the most damaging proteins in brain diseases do their harm inside neurons. Now researchers have used AI to bridge that gap.
A team at the University of Essex, led by Dr. Caitlin O’Shea, engineered intrabodies — antibody fragments designed to function inside human cells — using AI software developed in the lab of Nobel laureate David Baker. The key insight: electrical charge determines whether an antibody stays stable inside a cell or clumps together. “Antibodies usually have the wrong charge to exist inside cells without sticking together,” O’Shea explained. The work was published in Nature Communications (2026).
Scaling it up
By mapping charge properties across millions of antibodies, the team converted 672 different antibodies into functional intrabodies — and built molecules aimed at proteins implicated in Alzheimer’s, Parkinson’s, motor neurone disease (MND) and Huntington’s. The research was funded by the MND Association, and the team says it will freely share the molecules with other scientists.
Why it matters
If intrabodies can reliably neutralize disease proteins inside neurons, they could open a route to treatments that current antibody drugs can’t reach. This is early-stage research — a design toolkit and proof of concept, not a therapy — but making the molecules openly available could accelerate work across many labs.