Incretin receptors: one, two, or three

What an incretin is, and what was measured to establish it

The incretin concept rests on a measured discrepancy: an oral glucose load produces a larger insulin response than an intravenous load matched for circulating glucose. The difference is attributed to hormones released from enteroendocrine cells of the intestinal mucosa in response to nutrient arrival. Two account for most of it — glucose-dependent insulinotropic polypeptide, secreted from K cells of the proximal small intestine, and glucagon-like peptide-1, secreted from L cells more distally.

Drucker's review in Cell Metabolism in 2018 collects the mechanistic work on the GLP-1 receptor: a class B G-protein-coupled receptor, expressed on pancreatic beta cells and at multiple sites in the central and peripheral nervous system, signalling principally through cyclic AMP. The review also sets out the constraint that governs the entire class chemically. Native GLP-1 is cleaved within minutes by dipeptidyl peptidase-4 at the penultimate alanine, which means the native hormone is not a usable pharmacological tool and every compound in this section is an analogue engineered around that cleavage.

The two hormones behave differently in the same assays, and that difference is what made receptor count a design variable rather than a detail. GLP-1 receptor signalling was characterised as glucose-dependent in its insulinotropic action and as reducing gastric emptying; GIP receptor signalling was reported as insulinotropic in healthy physiology but substantially blunted in type 2 diabetes, which is why GIP was set aside for years before it was revisited in combination.

One receptor

Semaglutide is a GLP-1 receptor agonist carrying substitutions at the DPP-4 cleavage site and a fatty-acid chain that binds albumin, the second modification being what extends circulating half-life into the range that made an extended dosing interval feasible for investigators.

Wilding and colleagues published the STEP 1 trial in the New England Journal of Medicine in 2021: a randomised, double-blind, placebo-controlled trial in adults with overweight or obesity without diabetes, with percentage change in body weight as the primary endpoint over sixty-eight weeks. The trial reported a substantially larger mean weight reduction in the treatment arm than in the placebo arm, and gastrointestinal events — nausea and diarrhoea most commonly — as the leading adverse-event category.

Lincoff and colleagues then published SELECT in the same journal in 2023, which is the trial worth noting for anyone assessing the class rather than the endpoint. SELECT was powered on a cardiovascular composite — cardiovascular death, non-fatal myocardial infarction, non-fatal stroke — in adults with overweight or obesity and established cardiovascular disease but not diabetes, and reported a lower incidence of that composite in the treatment arm. It is the point at which the literature stopped being exclusively about body weight.

Two receptors, in two different combinations

Tirzepatide engages both the GIP and the GLP-1 receptor in a single peptide. Coskun and colleagues described the discovery pharmacology in Molecular Metabolism in 2018, reporting receptor-binding and signalling profiles at each receptor and characterising the molecule as imbalanced rather than equipotent — a stronger agonist at the GIP receptor than at the GLP-1 receptor in their assays. That imbalance is a deliberate design position, and whether it is the reason for the clinical separation from GLP-1 monoagonism is not settled by the paper.

Jastreboff and colleagues reported SURMOUNT-1 in the New England Journal of Medicine in 2022, a randomised placebo-controlled trial in adults with obesity without diabetes over seventy-two weeks, again with percentage change in body weight as the primary endpoint. The trial reported larger mean reductions than the published GLP-1 monoagonist trials had, although the comparison is across trials rather than within one, and cross-trial comparison is a weaker inference than the individual results themselves.

Survodutide is also a dual agonist, but the second receptor is different: glucagon rather than GIP. Le Roux and colleagues reported the phase 3 trial in the New England Journal of Medicine in 2026, with body weight as the primary endpoint in adults with obesity. Earlier phase 2 work by the same group generated post-hoc analyses of blood pressure and of subgroup response by sex and baseline body-mass index, and a separate phase 3 programme examined metabolic dysfunction-associated steatotic liver disease as the indication rather than weight — reported by Kaplan and colleagues in Nature Medicine in 2026, with liver histology endpoints.

Three receptors, and why the glucagon arm is the interesting one

Retatrutide adds glucagon receptor agonism to GIP and GLP-1. The inclusion looks wrong at first reading, since glucagon opposes insulin and raises hepatic glucose output, and the argument in the literature is that glucagon receptor agonism raises energy expenditure and mobilises hepatic lipid, with the GLP-1 component constraining the glycaemic consequence. That argument is mechanistic and plausible rather than demonstrated by the trials, which were not designed to partition contributions by receptor.

Jastreboff and colleagues published the phase 2 trial in the New England Journal of Medicine in 2023, randomised and placebo-controlled in adults with obesity, with percentage change in body weight as the primary endpoint at forty-eight weeks. Bajaj and colleagues then published TRANSCEND-T2D-1 in the Lancet in 2026, a double-blind randomised phase 3 trial in people with type 2 diabetes inadequately controlled by diet and exercise, with glycaemic control as the endpoint rather than weight.

The open question in the triple-agonist literature is attribution. No published trial isolates the glucagon receptor contribution in humans, and the receptor-count framing that organises this article is a description of molecular design, not evidence that three receptors are pharmacologically better than two. Network meta-analyses across the class exist and are the appropriate place to look for comparative claims, with the usual caveat that indirect comparison inherits every difference in trial population and duration.

The tolerability record, stated plainly

Gastrointestinal adverse events are the dominant finding in every trial cited above — nausea, vomiting, diarrhoea and constipation, most frequent during dose escalation and the leading reason for discontinuation in the treatment arms. This is consistent across monoagonist, dual and triple compounds, and it is the finding most often compressed to a clause in secondary summaries of the class.

Sodhi and colleagues published a disproportionality analysis in JAMA in 2023 using a pharmacovigilance database, examining reports of gastroparesis, biliary disease, pancreatitis and bowel obstruction among people using GLP-1 receptor agonists for weight loss compared with a different weight-loss agent. The design is hypothesis-generating rather than causal — disproportionality analyses cannot establish incidence and are subject to reporting bias — and it belongs in an honest account for exactly that reason: it is a signal, published in a major journal, that has not been resolved either way.

Bjerre Knudsen and colleagues reported in Endocrinology in 2010 that GLP-1 receptor agonists activated thyroid C-cells in rats and mice, causing calcitonin release and C-cell proliferation. The finding is species-specific in the authors' own analysis, with GLP-1 receptor expression on C-cells differing markedly between rodents and humans, and it remains the origin of the thyroid warning language attached to the class. It is a clean example of a preclinical finding whose translation is genuinely uncertain rather than dismissed.

Two further limits are worth naming. The long-term record thins sharply moving from the monoagonist end of this section to the triple-agonist end: semaglutide has multi-year outcome data, retatrutide and survodutide do not yet. And the trials above were conducted with pharmaceutical-grade drug product, under randomisation, in populations screened against exclusion criteria, with adverse events captured by protocol. None of those conditions attaches to research material, which is not a drug product, is not supplied for use in people, and is supplied here for laboratory use only.