If a parent's handwriting can carry the same loops and slants for sixty years, it is because the pattern that makes it theirs does not drift much, no matter where it is written down. A study published in Nature on September 30, 2026 asked whether a misfolded protein tied to dementia behaves the same way once it leaves the brain where it formed. Researchers at the MRC Laboratory of Molecular Biology in Cambridge and the Tokyo Metropolitan Institute of Medical Science took tau filaments from people who had died with Alzheimer's disease or with corticobasal degeneration (CBD), injected them into the brains of ordinary, non-transgenic mice, and used cryo-electron microscopy to examine what grew. The new tau filaments, built entirely from the mouse's own normal tau protein, matched the human seed fold for fold: an Alzheimer's seed produced an Alzheimer's fold, a CBD seed produced a CBD fold.
That answers a question the prion hypothesis for tauopathies has carried since it was proposed: does a disease-specific protein shape get copied faithfully in new territory, or reshaped by it? Earlier studies had shown that cryo-EM can tell different tau folds apart in human brain tissue, and that seeded tau can spread in mice engineered to overproduce mutant human tau. This experiment used mice with no such engineering, only their own tau to work with, and it tracked the result over time. Staining methods (AT8, pFTAA, Gallyas-Braak) picked up tau inclusions in CBD-injected mice within a month, and in Alzheimer's-injected mice within three months, with the inclusions multiplying further by six and nine months. The two seed types even sent the pathology to different addresses inside the brain, the way the human diseases do: the Alzheimer's tau built up mainly in neurons, while the CBD tau also showed up in glial cells.
Researchers quoted by the Alzheimer's disease research site Alzforum called the structural match striking, and also set limits on what it proves. Colin Masters of the University of Western Australia said he would want to see numbers showing the seed actually grew over the course of the experiment, and he would want the fold shown stable across several rounds of mouse-to-mouse transfer before calling the case closed on prion-like behavior. Commentary alongside the paper pointed to something else worth sitting with: the mice carrying all this tau pathology looked outwardly fine. Copying a shape faithfully is not the same as proving that shape is what breaks a brain.
For a family watching a parent lose words or movement, the distinction matters. This study did not test any drug, and it did not show that tau spreads this exact way inside a living human brain. What it offers is a mouse that finally uses only its own tau, giving researchers a cleaner stage on which to study how different tau folds might produce different disease patterns, Alzheimer's looking like Alzheimer's, CBD looking like CBD, inside one animal model.