Researchers reported a second enzyme that breaks down GLP-1, the gut hormone behind Ozempic and Wegovy, and it does its heaviest work in the brain.
For most of the drug class's history, one enzyme took the blame. DPP-4 clips GLP-1 apart within minutes of release, which is why the natural hormone is useless as a pill and why every successful GLP-1 medicine, up through semaglutide ↗, is engineered to dodge that single scissor. A study published August 28 in Science Advances ↗ says DPP-4 is not the only one cutting.
The group, at the Shanghai Institute of Organic Chemistry (Chinese Academy of Sciences), identified insulin-degrading enzyme, or IDE, as a previously unrecognized GLP-1-degrading protease. IDE is best known for the job its name describes, breaking down insulin. The team found it also cuts GLP-1 at two specific spots on the peptide, and that this cutting, not IDE's effect on insulin, was the part that actually mattered for blood sugar control. In plain terms, an enzyme everyone had filed under insulin turns out to be a major reason GLP-1 does not last.
The two-site fix
To blunt the second scissor, the chemists swapped in D-amino acids at the two cut sites. Proteins are normally built from left-handed (L) amino acids. The mirror-image right-handed (D) versions are chemically the same but sit the wrong way for an enzyme's cutting machinery to grip. Dropping them in at the exact cleavage points is a way to make one bond uncuttable without rebuilding the whole molecule.
The redesigned peptides lasted longer in plasma, in liver and gut fluids, in peritoneal fluid, and, the part the authors lean on, in the central nervous system. One version, built on semaglutide with a single D-serine at position 18, stayed in the blood longer and lowered glucose for longer in mice. When the researchers knocked IDE down, or injected the D-serine semaglutide straight into the brain, the peptide's extra durability tracked with IDE's absence. That is the control that pins the effect on IDE rather than on something else in the mix.
Why the brain part matters
The brain is where the newest bets on GLP-1 drugs sit. Trials are testing the class in Alzheimer's and Parkinson's disease, and the appetite circuits the drugs act on are central, not peripheral. In one recent analysis, semaglutide moved a dementia-risk score ↗ without yet moving the disease. A version that survives longer inside the central nervous system is, in principle, a way to hold a steadier dose at those circuits.
IDE has its own history in that tissue. It also helps clear amyloid-beta, the protein that piles up in Alzheimer's, so an enzyme that governs both GLP-1 and amyloid in the brain is a crowded intersection. Designing a GLP-1 drug to resist IDE quietly leaves the enzyme freer for its other substrates, a knock-on the paper does not chase but a reader should keep in view.
What it is, and what it is not
This is mouse work and biochemistry, not a human trial. "CNS-stable" was shown in part by injecting the peptide directly into the brain, which steps around the harder problem of getting a drug across the blood-brain barrier at all. The readout is glucose lowering, a surrogate, not a disease outcome. It comes from a single group, and D-amino-acid substitution, while a well-worn trick, can change how a peptide binds its receptor, the tradeoff any redesign has to clear.
What the paper delivers is a target and a method: a second enzyme worth designing around, and a two-site swap that held up in animals. peptidemodel hosts semaglutide against the GLP-1 receptor, and the lesson here sits on the chemistry side of that card rather than the receptor side. What turns a fragile gut hormone into a weekly injection is half-life engineering, and this work says the list of enzymes that engineering has to beat just got one longer.