More than 55 million people worldwide live with dementia, and Alzheimer’s disease accounts for the majority of cases. For decades, drug developers have aimed almost exclusively at one target: the sticky amyloid-beta plaques that accumulate between neurons. The newest FDA-cleared antibodies, lecanemab and donanemab, can reduce those plaques, but their benefits are modest and come with serious side effects, including brain swelling and microbleeds. Now a study published in Nature Neuroscience in early 2025 suggests the brain may already possess a more elegant solution: star-shaped support cells called astrocytes, switched into high gear by a single protein.
The protein is Sox9, a transcription factor that normally helps guide cell development. When researchers at the Bhatt Lab genetically increased Sox9 levels specifically in astrocytes of mice engineered to develop Alzheimer’s-like pathology, those cells began aggressively engulfing and digesting amyloid-beta plaques. Treated mice performed significantly better on memory tasks than untreated controls, and their brains showed reduced plaque burden in the hippocampus and cortex, regions critical for learning and recall.
Turning up the brain’s existing waste disposal
What makes the finding unusual is its mechanism. Rather than introducing a foreign molecule or antibody, Sox9 amplifies a cleanup process astrocytes already perform. The key downstream player is MEGF10, a receptor astrocytes use during normal brain maintenance to prune unnecessary synapses. Earlier research, including a 2020 study in Nature, confirmed that astrocytes actively phagocytose (consume and digest cellular debris) in the adult brain and that MEGF10 is essential to that process. Separate work established that MEGF10 can also recognize and engulf toxic amyloid-beta species in astrocytes, extending its role well beyond developmental pruning.
The Sox9 study connects these threads. By dialing up a single transcription factor, the researchers activated a cascade that increased MEGF10 expression, boosted plaque-engulfing activity, and converted astrocytes from relatively passive bystanders into aggressive scavengers. Crucially, the astrocytes did not change their fundamental identity or transform into immune cells. They simply did more of what they were already built to do.
Supporting evidence from human cell biology strengthens the case that this pathway is not unique to mice. A single-cell transcriptomic dataset generated by overexpressing Sox9 (alongside a related factor, NFIB) in human pluripotent stem cells mapped the gene programs Sox9 activates during astrocyte differentiation. While that dataset does not demonstrate plaque clearance in human cells, it confirms that Sox9 drives overlapping genetic programs across species, a necessary prerequisite for any future therapy.
Why the species gap still matters
Every result reported so far comes from genetically modified mice or from human astrocytes grown in laboratory dishes. No clinical trial data exist. No one has yet tested what happens when Sox9 is boosted in living human brain tissue affected by Alzheimer’s. The research team’s own institutional press release frames translational next steps as future work, not near-term clinical reality.
Mouse models of Alzheimer’s are useful but limited. The animals used in this study overproduce amyloid-beta, capturing one hallmark of the disease while missing much of the complexity seen in sporadic, late-onset Alzheimer’s in humans. Tau tangles, vascular damage, chronic inflammation, and metabolic dysfunction all contribute to real-world cognitive decline. Whether Sox9-driven astrocyte activation would influence any of those pathological features, or primarily target amyloid plaques alone, remains untested.
There is also the question of collateral damage. Because MEGF10 eliminates healthy synapses as part of its normal job, ramping up its activity could theoretically strip away connections the brain still needs. The treated mice in this study showed preserved cognition, suggesting the balance tipped in favor of plaque removal without catastrophic synapse loss. But whether that balance holds across different disease stages, or in the far more complex circuitry of the human brain, is unknown.
Durability, delivery, and the long road ahead
Even if the biology translates perfectly, practical hurdles remain. The mouse experiments relied on genetic tools to increase Sox9 expression in astrocytes, an approach that does not translate directly to a pill or injection. Any human therapy would likely require a delivery vehicle, potentially an adeno-associated virus (AAV) gene therapy vector or a small molecule capable of crossing the blood-brain barrier and selectively activating Sox9 in astrocytes. Neither option has been tested for this purpose.
Durability is another open question. The mouse work followed animals long enough to demonstrate cognitive protection over standard behavioral testing windows, but Alzheimer’s in humans unfolds over years to decades. It is not clear whether a one-time Sox9 boost would be sufficient, whether repeated interventions would be needed, or how chronic elevation of phagocytic activity might reshape brain circuits over extended periods.
The relationship between astrocyte-driven clearance and the brain’s other plaque-removal systems also needs mapping. Microglia, the brain’s resident immune cells, play their own role in amyloid management, and the brain employs glymphatic drainage and enzymatic degradation as additional defenses. How a Sox9-boosted astrocyte workforce would interact with these parallel systems, especially in later-stage disease when microglia are often chronically inflamed, is an area future studies will need to address carefully.
Some researchers have speculated about pairing astrocyte-targeted strategies with existing anti-amyloid antibodies like lecanemab to achieve faster plaque reduction. The logic is straightforward: antibodies bind amyloid in the fluid surrounding cells, while Sox9-enhanced astrocytes attack plaques from within brain tissue. But no experimental data test that combination, and any dual-therapy approach would face its own safety and regulatory scrutiny.
What this means for the search for better Alzheimer’s treatments
The field has a long history of promising mouse results that collapsed during the jump to human trials. Complexity, heterogeneity, and unforeseen safety problems have derailed candidate after candidate. Against that record, the Sox9 findings are best understood as a compelling proof of concept, not a cure on the horizon.
What sets this work apart is its specificity. Instead of broadly suppressing inflammation or flooding the brain with engineered antibodies, the approach works through machinery the brain already owns. It targets a defined protein in a defined cell type and produces a measurable effect on both pathology and behavior in animal models. That level of mechanistic clarity is what makes it scientifically valuable and worth watching.
For the roughly 7 million Americans currently living with Alzheimer’s, and the millions of caregivers supporting them, the practical implications remain distant. But if future studies can demonstrate that controlled activation of this pathway is safe, durable, and deliverable in humans, astrocytes may shift from background players to a central focus of Alzheimer’s therapy. The next critical steps, expected to unfold over the coming years as of June 2026, will be safety and feasibility studies in larger animal models and, eventually, early-phase human trials.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
