IDE: the second enzyme clearing GLP-1

A study published on 28 August 2026 in Science Advances identifies insulin-degrading enzyme (IDE) as a previously unrecognised protease that cleaves GLP-1 at two specific sites^[1]. The finding, from a team at the Shanghai Institute of Organic Chemistry (Chinese Academy of Sciences), changes the standard account of how the body clears this incretin hormone and opens a design path for GLP-1 drugs that last longer and reach the brain more effectively.

The known picture: DPP-4 is not acting alone

The pharmacology of endogenous GLP-1 has long been dominated by one enzyme: dipeptidyl peptidase 4 (DPP-4). DPP-4 cleaves the first two amino acids from the N-terminus of GLP-1, inactivating it within two to three minutes of release from intestinal L cells^[2]. That short half-life is the main reason drug developers engineered longer-acting analogues. Semaglutide, for instance, carries a fatty-acid chain that binds albumin, shielding the peptide from DPP-4 and extending its plasma half-life to roughly one week^[3]. DPP-4 inhibitors (gliptins) work by blocking this same enzyme.

What the field largely left open was whether other tissue-resident proteases also degrade GLP-1, and if so, which ones and where. The Zhang et al. study addresses that gap directly^[1].

IDE: what it does and where it sits

IDE is a zinc metalloprotease best known for degrading insulin and amyloid-beta peptides. It is expressed widely, including in the liver, intestinal secretome, peritoneal tissue, and the central nervous system. The Zhang et al. paper reports that IDE cleaves GLP-1 at two positions, creating an independent degradation pathway that operates even when DPP-4 is not the primary factor^[1]. Critically, IDE-mediated degradation of GLP-1 had measurable effects on glucose control in the mouse model used, while IDE's degradation of insulin over the same period was not the primary driver of glycaemic change. The authors interpret this as evidence that IDE-GLP-1 degradation is physiologically meaningful, not a background enzymatic activity.

Engineering resistance: D-amino acid substitutions

The practical output of identifying the two IDE cleavage sites is that you now know exactly where to protect the peptide. Zhang et al. introduced D-amino acid substitutions at those positions in both native GLP-1 and in semaglutide itself. D-amino acids are mirror-image versions of the L-amino acids that enzymes normally recognise; a protease that cuts at an L-residue will not process the D-form at the same site^[1].

The lead modified compound, D-Ser18-Semaglutide (semaglutide with a D-serine substitution at position 18), showed extended plasma retention and sustained glucose-lowering in mice compared with unmodified semaglutide^[1]. The engineered peptides also demonstrated improved stability in liver secretome, intestinal secretome, peritoneal fluid, and cerebrospinal fluid conditions, all compartments where IDE is active.

The CNS angle

IDE is expressed in the brain, and GLP-1 receptors are present in regions involved in appetite, cognition, and neuroprotection. The study confirmed IDE's role in the CNS by two routes: IDE knockdown in mice prolonged intracerebral GLP-1 activity, and direct intracerebral injection of D-Ser18-Semaglutide produced sustained central effects^[1]. This matters because one limitation of current GLP-1 receptor agonists is that blood-brain barrier penetration is partial and plasma-protein binding further limits the free fraction. A version of the drug that also resists CNS-resident IDE degradation could, in principle, maintain activity in the brain longer.

The neurological angle is speculative at this stage. The study demonstrates IDE-mediated GLP-1 clearance in the CNS and shows that the modified peptide resists it in mice. It does not demonstrate a clinical benefit for any neurological condition. The broader interest in GLP-1 class drugs for neurodegenerative conditions is active in the literature, but remains investigational and separate from the approved metabolic indications.

What this is not: caveats on scope

This is a preclinical mechanistic paper. All efficacy data come from mouse models; no human pharmacokinetic or safety data for D-Ser18-Semaglutide or any of the other modified peptides in this study exist yet. D-amino acid substitutions can affect immunogenicity, receptor binding geometry, and metabolic fate in ways that require dedicated safety studies before any modified analogue could enter a clinical programme^[1]. The paper establishes a mechanism and a proof-of-concept design strategy, not a clinical candidate ready for trials.

The study also does not show that IDE is more important than DPP-4 in overall GLP-1 clearance. The two enzymes operate in different compartments and likely in sequence rather than in competition. The practical message is that DPP-4 protection alone, the strategy behind semaglutide's fatty-acid chain, leaves a second degradation mechanism unaddressed.

Why this fits the broader GLP-1 design conversation

Drug developers working in the GLP-1 space are already testing oral formulations, weekly and monthly injectables, and fixed combinations such as cagrisema (semaglutide plus cagrilintide). The IDE cleavage finding adds a different dimension: not just extending the peptide's plasma half-life against DPP-4, but also making it stable against a second enzyme that matters particularly in tissues the current generation of drugs reaches incompletely. For readers following semaglutide trial data, this paper is upstream context: it explains one mechanism by which even the most durable GLP-1 analogues may still be cleared faster than optimal in certain compartments.

The approval and regulatory status of semaglutide is unaffected by this research. Ozempic and Wegovy remain authorised under their existing indications. The EMA, FDA, and MHRA have not issued any guidance related to IDE, and no IDE-resistant analogue is in a clinical trial as of the editorial date of this article. This is basic science with a clear translational direction, and the translational step is several years away at minimum.