Sixty years into its clinical life, metformin still had a secret.
Researchers at Baylor College of Medicine, working with collaborators at Louisiana State University, Nagoya University, and Meiji University in Japan, have found that metformin — the world's most prescribed diabetes drug — works in part through a brain pathway that had never been identified. The mechanism centers on Rap1, a signaling protein in the ventromedial hypothalamus (VMH), and specifically on a subset of neurons called SF1 cells. The findings, published in Science Advances in August 2025 (DOI: 10.1126/sciadv.adu3700), upend a decades-old assumption that metformin's primary glucose-lowering action happened in the liver and gut.
"We found that metformin's ability to reduce blood sugar at clinically relevant doses relies on suppressing Rap1 activity in this specific area of the brain," said Dr. Makoto Fukuda, corresponding author and associate professor of pediatrics-nutrition at Baylor, in comments reported by ScienceDaily. The researchers used genetically engineered mice lacking Rap1 in the VMH to isolate the pathway. When placed on a high-fat diet to model type 2 diabetes, these mice showed no blood sugar improvement from low-dose metformin — but responded normally to insulin and GLP-1 agonists, demonstrating the specificity of the finding.
The most striking result came from a different experiment: delivering metformin directly into the brains of diabetic mice. At doses thousands of times lower than a standard oral prescription, the treatment produced marked blood sugar reductions. "The brain reacts to much lower levels" than the liver or intestines, Fukuda said, adding that this difference in sensitivity may explain why some patients respond to metformin at doses that seem too low for the concentrations needed in peripheral tissues.
SF1 neurons in the VMH proved to be the critical intermediary. Metformin activated these cells, but only when Rap1 was present. In mice engineered to lack Rap1 in those neurons, the drug had no effect — confirming that the pathway from drug to neuron to glucose regulation runs through Rap1 as an essential switch.
The discovery has implications beyond diabetes treatment. Metformin is already prescribed off-label for its apparent brain-aging benefits, and is the centerpiece of the TAME (Targeting Aging with Metformin) trial, a large FDA-approved study examining whether the drug can delay aging-related diseases. If the Rap1-SF1 pathway in the hypothalamus is involved in those neurological effects, it would be the first mechanistic explanation for what has been an empirical observation.
"We plan to investigate whether this same brain Rap1 signaling is responsible for other well-documented effects of the drug on the brain," Fukuda said.
The research was supported by grants from the National Institutes of Health (including R01DK136627, R01DK121970, R01DK093587, R01DK101379, P30-DK079638, R01DK104901, and R01DK126655), the USDA/ARS, the American Heart Association, and the American Diabetes Association.
For a drug introduced in the 1950s and taken by tens of millions of people worldwide, the finding that its most important mechanism was hiding in the brain is a reminder that in biology, familiarity does not mean understanding.
