A research team led by Professor Mi Hee Lim of KAIST reported that a protein better known for regulating genes can suppress the harmful clumping of amyloid-beta, the peptide linked to Alzheimer's disease. The same approach also reduced the toxicity of clumps that had already formed. In a mouse model of Alzheimer's, the treatment reduced protein buildup in the brain and improved recognition memory.
The protein in question is a zinc finger protein. Zinc ions stabilize its folded shape, which allows it to recognize DNA or RNA. The researchers noted that the protein's surface can also bind other biomolecules, including proteins, so they tested whether it could latch onto amyloid-beta directly. Using computer simulations and laboratory experiments, they found that a specific segment of the protein, called ZF2-4, binds to the regions of amyloid-beta that drive its aggregation.
That binding changed how amyloid-beta clumped. Long, tangled fibrils, which are the more toxic form, were suppressed, and less toxic aggregates formed instead. When preformed fibrils were exposed to ZF2-4, they broke into short, thin fragments and became less harmful to cells.
Getting a therapeutic protein into the brain is a separate hurdle, because the blood-brain barrier also blocks useful molecules. The team attached a short peptide to create a brain-penetrant variant, ZF2-4-P9. When given to a transgenic mouse model, amyloid-beta deposits fell by about 53 percent in both the cerebral cortex and the hippocampus, a region central to memory and learning. Activation of microglia, the brain's immune cells, also dropped, by about 29 percent in the cortex and about 50 percent in the hippocampus. Treated mice performed better on a recognition memory test, though no significant improvements appeared in other memory and learning tests.
The authors stress that these results come from animal experiments, and that further studies are needed to evaluate therapeutic effect and long-term safety before the approach could be applied to humans. The work was published online in Advanced Science on September 30, 2026.