Baylor Scientists Boost Brain's Cleanup Cells With Sox9 Protein, Slashing Alzheimer's Plaques in Already-Sick Mice

Researchers at Baylor College of Medicine have demonstrated that boosting a single protein in the brain's support cells can dramatically slow Alzheimer's disease in mice that already have the hallmark plaques and memory loss of advanced disease — a result that the team says points toward a new class of treatments aimed at the brain's natural waste-disposal system rather than the plaques themselves.

The research, published this week in Nature Neuroscience, focused on Sox9, a transcription factor long known to regulate the development of astrocytes — the star-shaped support cells that vastly outnumber neurons in the human brain. Astrocytes maintain the chemical balance around neurons and have a known but poorly understood role in clearing amyloid-beta, the misfolded protein that accumulates as plaques in Alzheimer's. Lead author Dr. Dong-Joo Choi and corresponding author Dr. Benjamin Deneen, of Baylor's department of neurosurgery, used a viral vector to selectively raise Sox9 levels in astrocytes of two strains of Alzheimer's mouse models that had already developed cognitive impairment and visible plaques. "Increasing Sox9 triggered astrocytes to ingest more amyloid plaques, clearing them from the brain like a vacuum cleaner," Deneen said in a statement.

Over six months, mice with elevated Sox9 outperformed untreated controls on memory tests including the Barnes maze and novel-object recognition, and showed marked reductions in plaque density on histology. Just as importantly, the treated astrocytes themselves became more structurally complex, growing more branches and stronger connections to surrounding neurons — features that decline naturally with aging. Mice in which Sox9 was experimentally lowered in astrocytes showed the opposite effect: faster plaque accumulation and steeper cognitive decline. "We weren't just slowing the disease," Choi said. "We were actively reversing the trajectory in animals that were already sick. That's the population we have to treat in real life — most patients aren't diagnosed until plaques have been forming for years."

The findings stand apart from the dominant Alzheimer's research strategy of the last 20 years, which has aimed to remove or block amyloid-beta directly through monoclonal antibodies such as lecanemab and donanemab. Those drugs have shown modest cognitive benefit in clinical trials but are dogged by side effects including brain swelling and microbleeds, particularly in patients with the high-risk APOE4 gene. By contrast, the Sox9 approach harnesses cells the brain already uses to handle waste. "It's a fundamentally different idea," said Dr. Costantino Iadecola, director of the Feil Family Brain and Mind Research Institute at Weill Cornell Medicine, who was not involved in the study. "You're not adding a foreign antibody to the brain. You're persuading the brain to do its job better."

The Baylor team has begun engineering small-molecule and gene-therapy approaches to raise Sox9 levels in human astrocytes, and Deneen said the lab is in early discussions with industrial partners to advance toward an investigational new drug application. Significant questions remain — including whether the same protein behaves identically in human astrocytes, whether long-term Sox9 elevation could trigger reactive astrogliosis or other inflammation, and whether the approach will work against the tau tangles that accompany amyloid plaques in late-stage disease. The work was funded by the National Institute on Aging, the Cure Alzheimer's Fund, and Baylor's Duncan Neurological Research Institute.