Adding a small chemical cap to the front of semaglutide, the peptide in Ozempic and Wegovy, kept it lowering blood sugar in obese mice three days after a single dose, a span longer than the unmodified drug holds.
The result, reported online July 17 in Cell Reports ↗, turns on how the receptor responds, not on how tightly the drug binds. It exploits a signaling split that underlies the whole GLP-1 drug class for type 2 diabetes and obesity.
When semaglutide docks onto the GLP-1 receptor ↗, a class B G protein-coupled receptor, the receptor does not do one thing. It fires two separate internal messengers. One is a Gs protein, which switches on cAMP, the second messenger that tells a pancreatic beta cell to release insulin. The other is beta-arrestin, a protein that clamps onto the activated receptor, drags it inside the cell by internalization, and quiets it down. The Gs protein arm does the glucose-lowering work. The beta-arrestin arm blunts the signal and pulls the receptor off the cell surface, which is part of why cells stop responding as strongly over time.
A drug that pushes hard on the Gs protein arm while barely touching the beta-arrestin arm is called biased. The team, working in China, built one by modifying the very tip of the peptide, the N-terminus, either capping it with an acetyl group or swapping in different amino acids. The capped versions, including an acetylated form of semaglutide the authors call Ac-semaglutide and a matching Ac-GLP1, recruited far less beta-arrestin, stayed on the cell surface longer instead of being internalized, and kept cAMP flowing for an extended stretch.
To find the structural cause, the group froze the modified GLP-1 peptide onto the receptor bound to its Gs partner and imaged the complex with cryo-electron microscopy at 2.64 angstrom resolution, sharp enough to trace individual amino acid side chains. The structure pointed to one feature. A loop on the outside of the receptor, extracellular loop 3, swung outward in the biased complex in a way it does not in receptors bound to beta-arrestin-favoring ligands. That single displacement lines up with the change in signaling, which gives medicinal chemists a defined piece of the receptor to target.
The payoff the paper leads with is duration. In diet-induced obese mice, Ac-semaglutide still lowered glucose three days after one injection. Standard semaglutide is already long-acting because it clings to albumin in the blood and resists breakdown, which is how it became a once-weekly shot in people. This work adds a second, independent lever: keep the receptor from swallowing the drug's signal in the first place, and each molecule does more before the cell shuts it down.
The limits are the standard ones for a preclinical structure paper. This is mouse and molecular work, not a human trial, and biased agonism has a mixed track record across other receptors, where the clean split seen in a cell assay does not always survive in a whole animal. Whether trimming beta-arrestin also trims the nausea and vomiting that push people off GLP-1 drugs is a question the study does not test. Beta-arrestin signaling has been tied to GLP-1 side effects in some models and to durable receptor function in others, so less of it does not translate directly to a better drug.
What the paper delivers is a structural map. It names extracellular loop 3 as the switch that governs the bias and shows that an approved drug, semaglutide ↗, can be pushed toward the Gs protein arm with a chemical change small enough to add to an existing molecule. That aims the next round of GLP-1 engineering at how the receptor behaves after the drug arrives, not only at how tightly the drug binds.