AI-built ‘intrabodies’ target Alzheimer’s—with cautious optimism

An international team of scientists, led by researchers at the University of Essex, has engineered microscopic antibody fragments—dubbed intrabodies—that can be manufactured directly inside human cells. Using AI-driven protein design, these fragments bind to misfolded proteins implicated in neurodegenerative diseases like Alzheimer’s, Parkinson’s, and motor neuron disease (MND).

The approach sidesteps a key hurdle in antibody therapies: delivery. Traditional antibodies struggle to penetrate cells, but intrabodies are encoded via gene therapy, allowing cells to produce them internally. Early lab tests show they can latch onto toxic protein aggregates, including tau tangles in Alzheimer’s and TDP-43 in MND.

Yet this remains preclinical work. The study, published in Nature Communications, tested intrabodies in cultured human cells and mouse models—not patients. The team emphasizes that while the mechanism is plausible, efficacy and safety in humans are unproven.

The research leans on AI to accelerate design, but the clinical path is long. Regulatory approval for gene therapies targeting neurodegenerative diseases is notoriously slow, with recent setbacks in similar pipelines. Even if intrabodies perform well in animal trials, scaling to human dosing—and proving they don’t trigger immune overreactions—will take years.

For patients today, nothing changes. Existing treatments for Alzheimer’s, like lecanemab, focus on amyloid plaques, not intracellular targets. Intrabodies, if successful, could complement these by addressing a different disease pathway. But that’s a speculative if.

The study’s sample size is another limit: mouse models and cell lines don’t replicate human brain complexity. As Dr. Tara Spires-Jones, a neurodegeneration expert, notes, ‘Many therapies look promising in mice and fail in humans.’