AI Discovers Brain-Penetrating Drug That Clears Alzheimer's Plaques in Mice

Neuroinflammation is increasingly recognized as a central driver of Alzheimer's disease (AD) progression, not merely a consequence of it. Galectin-3 (Gal-3), a lectin protein expressed by activated microglia, has emerged as a key regulator of this inflammatory cascade, making it an attractive therapeutic target. However, developing small molecules that both inhibit Gal-3 and penetrate the blood-brain barrier (BBB) has proven difficult.

Researchers employed an AI-powered virtual screening platform to sift through large chemical libraries, identifying FJMU1887 as a lead candidate with optimized drug-like and pharmacokinetic properties. Laboratory binding assays confirmed FJMU1887 binds Gal-3 with a dissociation constant of 1.55 μM. Critically, the compound was shown to disrupt the Gal-3–TREM2 protein interaction, a pathway implicated in driving microglial-mediated neuroinflammation in AD.

In cell-based experiments using BV-2 microglial cells, FJMU1887 suppressed the pro-inflammatory cytokine TNF-α with an IC₅₀ of approximately 2.36 μM and showed no cytotoxicity. BBB penetration was confirmed through both artificial membrane assays and in situ brain perfusion studies, although partial P-glycoprotein efflux was noted as a limitation to full CNS exposure.

In vivo, 14-month-old 5×FAD mice — a well-established aggressive AD model — treated orally for 30 days showed significantly reduced Gal-3 expression, decreased amyloid-beta burden, attenuated microglial activation, and improved synaptic integrity. Cognitive improvements were observed across multiple behavioral tests in both transgenic and oligomeric Aβ-induced impairment models.

While results are compelling, the study is preclinical and limited to mouse models. Translation to humans will require extensive safety profiling, optimized dosing strategies, and resolution of partial P-glycoprotein efflux to ensure adequate brain exposure.