Researchers have named a specific molecular carrier for the spread of tau, the protein whose accumulation tracks cognitive decline in Alzheimer's disease and other tauopathies. In a paper published online in Cell on 29 June 2026 (print issue 23 July 2026), Mitali Tyagi, Jason D. Shepherd and colleagues at the University of Utah, with Bradley T. Hyman at Massachusetts General Hospital, report that Arc, an activity-regulated neuronal protein that forms virus-like capsids, binds tau through a direct protein-protein interaction and is required for tau to be released from neurons inside extracellular vesicles (EVs). Tau spread has been mapped anatomically for decades; this is an attempt to name the vehicle rather than the route.
The core experiments are laboratory work. Brain EVs purified from rTg4510 mice, which carry mutant human tau, contained high levels of that tau and could seed tau aggregation; when the same mice were crossed onto an Arc knockout background, their brain EVs contained significantly less human tau and lost seeding potential. In cultured hippocampal neurons and in living mice, the team used a virus expressing green fluorescent protein alongside mutant human tau (P301L) to mark donor cells and then counted neurons that had acquired tau without the marker. Six-month-old wild-type and Arc knockout mice were injected in the medial entorhinal cortex and examined 10 weeks later, six animals per group, three male and three female: transmission to recipient neurons was significantly reduced in the Arc knockouts, with comparable numbers of donor neurons and no baseline difference in neuronal activity or excitability. "When we removed Arc, we saw that the transfer of Tau was severely, severely reduced," Tyagi said in a statement issued by University of Utah Health.
The human component is small and observational. EVs were isolated from frozen post-mortem prefrontal cortex (Brodmann area 8/9) supplied by the Massachusetts Alzheimer's Disease Research Center brain bank: seven controls (aged 59 to 90), one brain at Braak stage 2, and seven at Braak stage 6, the most advanced stage of tau pathology. Arc and human tau were found together in the same vesicles, and EV Arc levels correlated with phosphorylated EV tau, strongly in the Alzheimer's samples (r = 0.89) and as a non-significant trend in the non-Alzheimer's samples (r = 0.74, p = 0.08). A correlation in 15 donated brains at a single point after death cannot establish that Arc drives disease progression in people.
The authors' own data complicate any simple therapeutic reading. Mice lacking Arc accumulated more tau inside their neurons and showed a modest increase in cell toxicity early in disease progression, consistent with EV export acting partly as a disposal route for the sick cell; Arc is also central to normal synaptic plasticity, and the knockouts lacked it from birth rather than having it blocked at a chosen moment. No compound targeting the Arc to tau interaction has been tested in people, and no clinical trial follows from this paper. Shepherd discloses that he co-founded VNV, LLC and holds stock in and consults for Aera Therapeutics, which licenses intellectual property including Arc capsids.