Protein That Tags Toxic Tau for Destruction Could Point to New Dementia Therapies

Right now, somewhere in your brain, a neuron is doing something the others aren’t. It’s making more of a protein called CUL5, and because of that, it’s probably going to survive what kills its neighbours.

That’s the implication, anyway, of a study published in Cell by researchers at UC San Francisco. Martin Kampmann, a professor of biochemistry and biophysics at UCSF, and his team have identified a cellular cleaning system that disposes of toxic tau before it can form the clumps associated with Alzheimer’s disease and other dementias. Neurons with higher levels of CUL5 appear more resilient. Whether that observation can be turned into a drug is the question that comes next, but for a field that has struggled for decades to find anything genuinely new to work with, having a clearly defined cellular target is not nothing.

Tau is a protein that normally helps stabilise the internal skeleton of neurons. In tauopathies, a category that includes Alzheimer’s disease, frontotemporal dementia and Progressive Supranuclear Palsy, tau misfolds and accumulates into clumps, eventually killing the cells that contain it. What’s long been puzzling is that this process isn’t uniform. In brains riddled with tau pathology, some neurons die quickly, others survive well past the point where you’d expect them to. The molecular explanation for that difference has been unclear. It’s precisely that gap the UCSF team set out to close.

The approach was unusually systematic. Kampmann’s group engineered human neurons (grown from stem cells in petri dishes) to produce the toxic tau clumps characteristic of disease. They then used CRISPR gene-editing technology to silence each of the cells’ roughly 20,000 genes, one at a time, watching to see which silenced genes changed how fast the clumps formed. “It’s the first time we’ve been able to screen human neurons for genes that determine their resilience to tau,” Kampmann said.

The screen pointed clearly to CUL5. When the gene was active, tau clumps were suppressed. When it was knocked down, they accumulated. CUL5 is what’s technically called an E3 ubiquitin ligase, which is biologist shorthand for a protein that attaches small molecular tags to other proteins, earmarking them for destruction. It works together with a partner adaptor protein called SOCS4, forming a complex (CRL5SOCS4) that grabs tau at a specific region near its N-terminus, slaps the tag on it and sends it off to the cellular machinery that degrades unwanted proteins. The proteasome, this machinery, then chops it up before it can accumulate. “CUL5 is uniquely suited to getting rid of tau,” Kampmann said.

To check whether any of this was relevant outside a petri dish, the team turned to the Seattle Alzheimer’s Disease Brain Cell Atlas, a database built from postmortem brain samples from deceased Alzheimer’s patients. Real brains, in other words, from people who had actually died of the disease. Even in those samples, some neurons had survived longer than others and those resilient cells showed consistently higher CUL5 expression. The correlation held across multiple brain regions, multiple cell types, and across different tauopathies including FTD and PSP. That degree of consistency across diseases is somewhat unusual; it hints that CUL5 may be doing something fundamental rather than disease-specific.

The screen also turned up a second, connected finding that’s perhaps equally significant for understanding why tau pathology gets worse with age. A set of genes related to oxidative stress, the cellular damage that accumulates as cells burn energy, appeared to influence how “sticky” tau became, how prone to clumping. This stickiness is at least partly a consequence of oxidative damage to tau’s own structure: when the team blocked mitochondrial function (which ramps up reactive oxygen species, the chemical byproducts of energy metabolism), tau began to fragment in a specific and somewhat unexpected way. A 25-kilodalton fragment broke off from the N-terminal end of the protein. This particular fragment turns out to closely resemble something clinicians already measure, a biomarker called NTA-tau detectable in cerebrospinal fluid and blood that is currently used to track Alzheimer’s progression. The suggestion, and it remains only a suggestion at this stage, is that NTA-tau levels in patients might be reporting on mitochondrial stress as much as on tau aggregation per se.

That’s a more complicated picture than it might sound. It means the progression of Alzheimer’s and related diseases may be shaped not just by whether the tau-clearance machinery is working, but by how well the mitochondria are functioning, and how much oxidative damage has accumulated over a lifetime. Ageing, in other words, may degrade neuronal resilience through at least two parallel routes: letting tau clearing slip and simultaneously making tau more prone to clumping in the first place.

For now, the therapeutic implication is simpler than the underlying biology. “Maybe a future therapy could enhance the body’s natural mechanism for avoiding neurodegeneration,” Kampmann said. The idea would be to boost CUL5 activity pharmacologically, or perhaps to increase SOCS4 expression, tipping the balance in favour of tau clearance before clumps can form. “We hope that CUL5 can be the first of many new targets for drug discovery against the dementias,” he added.

Obstacles remain. The screens were done in immature lab-grown neurons that don’t fully replicate adult human brain biology. CUL5 also has roles in immune signalling quite apart from tau, so boosting it indiscriminately might cause problems elsewhere. And the team noted that the resilient neurons in postmortem samples had survived to be sampled partly because they were resilient, which introduces a selection bias that’s hard to eliminate from the analysis. What happens inside cells that are genuinely on the edge, in the earliest phases of disease, is harder to know from postmortem data.

Still, the identification of a protein that appears protective across multiple human tauopathies, confirmed in both controlled experiments and real patient tissue, gives the field something concrete to work with. Dementia research has been hunting for druggable targets outside the well-worn amyloid pathway for a while. CUL5 won’t cure anything tomorrow. But the logic of the biology, a cellular skip collector for misfolded proteins whose activity predicts which neurons live and which ones don’t, is hard to dismiss.

Study link: https://www.cell.com/cell/fulltext/S0092-8674(25)01487-4