Scientists injected tau protein from people who had died with Alzheimer’s disease or corticobasal degeneration into mouse brains, and the mice’s own tau adopted the same atomic structure as in each human disease

Scientists injected tau protein from people who had died with Alzheimer’s disease or corticobasal degeneration into mouse brains, and the mice’s own tau adopted the same atomic structure as in each human disease

Michel Goedert and Sjors Scheres in Cambridge led the study with colleagues in Japan. They injected ordinary mice with tau filaments from the brains of people who had died with Alzheimer’s disease or corticobasal degeneration (КБД, a rarer disease involving similar protein breakdown). After 9–12 months, cryo-electron microscopy, which involves freezing a sample and directly imaging its atomic structure, showed that the filaments formed from the mice’s own tau matched the human filaments at the atomic level. The paper was published in Nature on 30 September 2026.

Tau is a protein that normally stabilizes microtubules, the internal tracks along which neurons transport cargo. In more than 20 brain diseases, including Alzheimer’s disease, tau aggregates into insoluble filaments with a distinct shape for each disease. The Cambridge laboratory where Goedert and Scheres work has been described as having “a Nobel laureate on every floor”: the structure of DNA was determined there, Frederick Sanger won the Nobel Prize twice, and César Milstein developed monoclonal antibodies. In 2017–2021, Goedert, Scheres and their Japanese colleagues were the first to determine the atomic structures of tau filaments from the brains of people who had died with five different diseases, establishing reference structures for comparison. These same structures later proved resistant to AI drug design with RFdiffusion3.

The injected human tau disappeared from the mouse brains within a week, but it triggered a chain reaction: the mice’s own tau began copying the shape of the introduced filaments, much as a crystal causes molecules in solution to extend its lattice. This is possible because the region of the protein that folds into the filament has an identical amino acid sequence in mice and humans. The atomic coordinates of the mouse and human filaments matched with a root mean square deviation (RMSD) of 0,906 Å for Alzheimer’s filaments and 0,576 Å for КБД filaments. The atoms in the newly formed filaments occupied almost the same positions as those in the original filaments. Filament shape also determined which cells were affected: Alzheimer’s filaments produced inclusions only in neurons, whereas КБД filaments produced inclusions in both neurons and glia, the brain’s support cells, with the same lesions seen in humans.

The method worked only for tauopathies involving a tau isoform that mice also produce. Adult mice produce just one of the two tau variants, the four-repeat isoform. Alzheimer’s filaments contain a mixture of both variants, while КБД filaments contain only the four-repeat variant, allowing both to propagate in mice. Pick’s disease filaments consist of the three-repeat variant, which adult mice lack, so this method cannot reproduce them.

The mice that developed filaments remained outwardly healthy throughout the 9–12 months of observation, unlike the rapid course of classic prion diseases such as mad cow disease. Macaques injected with tau filaments from progressive supranuclear palsy, another disease involving accumulation of the same protein, developed movement impairments within six months.

Tauopathies do not spread between people through everyday contact as influenza does. Rare cases of transmission have, however, been reported: until the mid-1980s, children received growth hormone extracted from the pituitary glands of deceased donors, before production switched to genetic engineering. Some of these recipients developed an Alzheimer’s-like disease decades later, presumably because the preparation contained tau and amyloid filaments.

The atomic match between the original filaments and those they seeded establishes the prion hypothesis of neurodegeneration as a demonstrated fact for tau. The underlying principle is the same as in mad cow disease and Creutzfeldt-Jakob disease. This finding opens a path to diagnosis and treatment tailored to the protein conformation present in an individual patient.