Tau, the protein whose misfolded tangles are a hallmark of Alzheimer's disease and other tauopathies, also performs a normal, healthy function in memory, according to a study published in Nature Communications on May 17, 2026. Researchers at Flinders University, the University of New South Wales and Macquarie University found that phosphorylation of tau at a single site, threonine 205 (T205), helps determine which small cluster of neurons, called an engram, gets recruited to store a given memory in mice, and that this modification is required for the memory to be reliably recalled weeks later, a phenomenon known as remote recall. Mice carrying a mutant form of tau that cannot be phosphorylated at T205 learned new information and recalled it shortly afterward just as well as normal mice, but their ability to recall the same memory after a longer delay was reduced.
The finding matters because tau has long been treated in dementia research chiefly as a threat: something that, once abnormally modified, clumps into the neurofibrillary tangles seen in Alzheimer's disease and in other tauopathies such as frontotemporal dementia. This study, and a related preprint version posted on bioRxiv in December 2023, instead describes a physiological job for tau at the moment a memory is being laid down. The researchers report that when tau in the mice lacked the T205 modification, more neurons than usual were activated locally around the memory trace, suggesting tau normally helps keep the recruited group of cells sparse and precise. Mice lacking a specific enzyme, p38gamma, which attaches the phosphate to T205, showed a similar pattern of impaired remote recall, according to the paper.
The authors are careful to distinguish this proposed normal role from the toxic, disease-associated tau seen in Alzheimer's tangles. In the same study, when a disease-associated pathogenic form of tau was introduced into the active cell ensemble during encoding or during recall, the mice showed amnesia, either forgetting what they had just learned or failing to retrieve a memory formed earlier, depending on the timing. Independent researchers not involved in the study have noted that the results were generated entirely in mice and describe a mechanism, not a treatment; the authors themselves state that the findings do not directly translate to human brain function or dementia, and that any therapeutic implications remain a matter for future research.