GLP-1 Receptor Agonist Mechanism: cAMP, Glucose, Beta Cells

A GLP-1 receptor agonist acts on the pancreatic beta cell through GLP-1R, a class B1 G protein-coupled receptor whose activation raises cyclic adenosine monophosphate and increases glucose-dependent insulin secretion (2022, PMID 35065096). The load-bearing word is glucose-dependent. These agonists potentiate secretion that glucose has already triggered, and have little or no activity on insulin secretion in the absence of elevated blood glucose, which is the property that separates the class from sulphonylureas on hypoglycaemia risk (2013, PMID 22776039). Everything else in the beta-cell story sits downstream of that conditional switch.

What does the receptor actually do after the agonist binds?

Binding is only the first step, and the steps after it are separable from each other. In cell lines overexpressing human GLP-1R, the oral small molecule ECC5004/AZD5004 bound with an IC50 of 2.4 nM and augmented cAMP signalling without recruiting β-arrestin-2 and without driving receptor internalization (2025, PMID 39495140).

Signal amplitude and receptor trafficking are therefore different axes. A compound can push cAMP hard while leaving the receptor on the plasma membrane, so “activates GLP-1R” describes a family of outcomes rather than one.

The GIP receptor sits on the same beta cell and feeds the same second messenger. GLP-1R and GIPR are both class B1 GPCRs, and activation of either raises cAMP and increases glucose-dependent insulin secretion (2022, PMID 35065096). That shared convergence is why co-agonism is pharmacologically coherent rather than a lucky combination, though the same review flags that the downstream pathways of GLP-1R-GIPR co-agonists are not fully elucidated, because co-targeting two closely related receptors is hard to isolate in vivo.

How potent is the beta-cell effect in measured numbers?

Six measurements from two reports bracket what has actually been quantified at the receptor and at the islet.

Parameter	Value	System measured in	Source
ECC5004 binding, IC50	2.4 nM	cell lines overexpressing human GLP-1R	PMID 39495140
ECC5004 potentiation of glucose-stimulated insulin secretion, EC50	5.9 nM	EndoC-βH5 human beta-cell line	PMID 39495140
ECC5004 potentiation of glucose-stimulated insulin secretion, EC50	0.022 nM	non-human primates, in vivo	PMID 39495140
Liraglutide concentration that enhanced secretion	25 nmol/L	human donor islets, glucose intolerance (n = 7, P = 0.021)	PMID 42350670
Liraglutide effect at the same concentration	none detected	human donor islets, normoglycaemic (n = 7)	PMID 42350670
GLP-1R mRNA against donor HbA1c	progressive decrease as HbA1c rose (P = 0.015)	112 donor islets; normoglycaemic n = 48 vs type 2 diabetes n = 10	PMID 42350670

Table 1: receptor-level and islet-level measurements behind the GLP-1 receptor agonist mechanism of action in pancreas beta cells. Sources: ECC5004 non-clinical and first-in-human characterization, 2025, PMID 39495140; human donor islet and mouse model study, PMID 42350670.

Both ECC5004 EC50 values come from the same report, so the ratio between them is fair arithmetic: 5.9 / 0.022 = 268-fold lower concentration for half-maximal effect in the non-human primate in vivo assay than in the EndoC-βH5 human beta-cell line (2025, PMID 39495140). Assay system, not molecule, produces most of that spread.

Inside the human system, occupancy and function also separate. The functional EC50 of 5.9 nM in EndoC-βH5 cells sits 5.9 / 2.4 = 2.5-fold above the 2.4 nM binding IC50 measured in the GLP-1R-overexpressing line (2025, PMID 39495140).

Why does the same drug do nothing to insulin secretion at normal glucose?

The gate is the glucose signal itself. GLP-1R agonists potentiate glucose-stimulated insulin secretion and show little or no activity on insulin secretion when blood glucose is not elevated, so the pharmacology is self-regulated by ambient glucose in a way that insulin and sulphonylureas are not (2013, PMID 22776039).

Human donor islets reproduce that asymmetry at the tissue level. Liraglutide at 25 nmol/L enhanced glucose-stimulated insulin secretion in islets from donors with glucose intolerance (n = 7, P = 0.021) and produced no measurable effect in normoglycaemic islets (n = 7), even though those normoglycaemic islets retained their response to GLP-1 (7-36) (PMID 42350670).

Where hypoglycaemia does appear in the clinical record for this class, it clusters with a specific co-medication. Rat work suggests sulphonylureas can uncouple the glucose dependence, and clinical studies show that the majority of hypoglycaemia events in patients treated with GLP-1R agonists occurred in those on a concomitant sulphonylurea (2013, PMID 22776039).

Does receptor density change as diabetes progresses?

It falls, and the functional consequence is not a straight line. GLP-1R mRNA decreased progressively as donor HbA1c rose across 112 stratified donor islets (P = 0.015 comparing normoglycaemic donors, n = 48, with type 2 diabetes donors, n = 10), yet the islets that responded to liraglutide were the ones in the intermediate band, donors with HbA1c 6.0-6.4% (PMID 42350670).

The mouse arm of the same study argues that the direct beta-cell route is not always the operative one. In chow-fed mice, liraglutide’s insulin-stimulating effect was abolished in animals with tanycyte-specific GLP-1R knockdown, meaning it required hypothalamic access; after 12 weeks of high-fat diet, direct islet responsiveness returned independent of tanycyte function (PMID 42350670).

At 27 weeks of high-fat diet the picture inverted again. Islets still responded ex vivo while in vivo insulin enhancement was lost, with glucose lowering carried instead by suppressed hepatic gluconeogenesis and increased peripheral glucose uptake (PMID 42350670). Three metabolic states, three dominant mechanisms, one drug.

What do the dual agonists add at the beta cell?

Tirzepatide’s mechanism includes biased signalling at GLP-1R alongside potent GIPR signalling, which makes its beta-cell biomarkers the most direct available test of whether the second receptor changes anything measurable (2022, PMID 35065096).

A post hoc analysis of phase 2 data in 316 subjects with type 2 diabetes across 47 sites in 4 countries compared tirzepatide 1, 5, 10 and 15 mg against dulaglutide 1.5 mg and placebo at 26 weeks (2021, PMID 33236115). HOMA2-B, the beta-cell function index, increased significantly against placebo with dulaglutide and with tirzepatide 5, 10 and 15 mg (P ≤ .02) — the two drug classes were indistinguishable on that index.

Proinsulin ratios separated them where HOMA2-B did not. Proinsulin/insulin and proinsulin/C-peptide ratios fell significantly with tirzepatide 10 and 15 mg against both placebo and dulaglutide (P ≤ .007), and tirzepatide 10 mg lowered HOMA2-IR against both comparators (P = .004) (2021, PMID 33236115). A falling proinsulin ratio is a processing readout: less unconverted precursor leaving the cell per unit of secreted product.

Weight loss did not account for most of the insulin-sensitivity change. Multiple linear regression adjusting for age, sex, metformin, triglycerides and HbA1c found that weight loss explained 13% of the HOMA2-IR improvement on tirzepatide 10 mg and 21% on 15 mg (P ≤ .028), leaving 100 − 13 = 87% and 100 − 21 = 79% attributable to something other than weight change (2021, PMID 33236115). The weight figures those regressions were adjusting against are collected in what the 68-week trials reported by drug.

Can a small molecule reach the same beta-cell target?

ECC5004 did, orally, through 28 days of daily dosing in patients with type 2 diabetes. It was profiled in glucose-stimulated insulin secretion assays in a human beta-cell line and in non-human primates, then taken into a phase 1, double-blind, placebo-controlled first-in-human study of single doses from 1 to 300 mg in healthy volunteers and daily doses of 5, 10, 30 and 50 mg in patients with type 2 diabetes for 28 days (NCT05654831, 2025, PMID 39495140).

Exposure was dose-proportional at doses of 25 mg and above, no serious adverse events were reported, and dose-dependent reductions in glucose and body weight were seen; a 9-month oral toxicity study in non-human primates ran ahead of the human work and also showed dose-dependent body weight changes against control (2025, PMID 39495140).

Route is where the molecule classes diverge rather than at the receptor, since a peptide agonist reaching the same GLP-1R needs heavy formulation work to survive oral delivery — the specific barriers are set out in why oral peptide doses run so high and in how peptides and small molecules differ on route and dose.

The figure that these reports give least often is receptor occupancy at the clinical dose, as distinct from the 5.9 nM EC50 measured in the EndoC-βH5 line or the 2.4 nM binding IC50 in the overexpressing line (2025, PMID 39495140). That is the number to look for when comparing any two GLP-1R agonists on beta-cell engagement rather than on outcome.