The lizard that ate twice a year

Swallow glucose and your pancreas releases far more insulin than if the identical amount were dripped into a vein. The gap is the incretin effect: the gut telling the pancreas that food has landed, before blood sugar has fully risen. The idea is older than the hormone. In 1902 Bayliss and Starling named the first hormone of any kind — secretin, a gut factor — and coined the word hormone for it. By 1906 a London physician was feeding duodenal extract to diabetics on the hunch that the intestine carried a pancreas-stimulating messenger. In 1932 the Belgian physiologist Jean La Barre gave that messenger a name — incretin — for a gut factor that lowered blood glucose. Then the trail went cold for four decades. The chemistry to isolate a peptide present in vanishing quantities simply did not exist yet.

What broke the logjam was a measuring tool, not a molecule. Rosalyn Yalow and Solomon Berson’s radioimmunoassay, developed at the Bronx VA in the late 1950s, could finally detect hormones at picomolar concentrations. Hold that address — the Bronx VA — because the same building returns to this story in a way nobody planned.

The first incretin fell out in the early 1970s: John Brown’s GIP, at first “gastric inhibitory polypeptide,” later rebadged glucose-dependent insulinotropic polypeptide when its real job turned out to be prodding insulin rather than quieting the stomach. But blocking GIP only partly abolished the incretin effect, which meant a second messenger had to exist.

It arrived by inference before it arrived by isolation. When the proglucagon gene was cloned in the early 1980s, Graeme Bell’s group found the sequence encoded not just glucagon but two more glucagon-like peptides tucked into the same precursor — GLP-1 and GLP-2. The “1” was never a version number; it was a position in a gene. (This is the detail behind why there is no GLP-4.) The catch: full-length GLP-1 barely worked. Svetlana Mojsov, working with Joel Habener at Massachusetts General, showed the body clips off the first six residues, and that the truncated form — GLP-1(7–37) — was the potent one. Daniel Drucker then demonstrated it actually stimulated insulin in a glucose-dependent way. By the late 1980s the physiology was gorgeous and the drug was hopeless.

Enter the second character at that Bronx VA. John Eng, an endocrinologist trained in the Yalow tradition of assay-building, had a sideline habit: screening animal venoms for peptides that lit up mammalian receptors. In 1990 a report noted that Gila monster venom caused inflammation of the pancreas — which, to an endocrinologist, reads as: something in there is talking to the exocrine pancreas.

The animal itself is the clue. The Gila monster (Heloderma suspectum) is a desert lizard that eats only a handful of enormous meals a year and fasts, sometimes for months, between them. A pancreas that swings from gorging to starving and back is exactly where you would expect to find an unusually durable metabolic signal. In 1992 Eng isolated it: exendin-4, a 39-residue peptide from the lizard’s salivary venom, about 53% identical to human GLP-1 — close enough to activate the human GLP-1 receptor, foreign enough to behave differently.

The difference that mattered was a single amino acid. Human GLP-1 has alanine at position 2, the residue DPP-4 grabs. Exendin-4 has glycine there. DPP-4 cannot get a grip, so the lizard peptide survives in circulation for hours rather than minutes. Evolution had already solved the exact problem that had defeated the drug developers — in a venom gland, for reasons that have nothing to do with diabetes.

Then the part of the story that gets left out of the press releases: Eng could not get his employer interested. He paid for the patent himself, and only later licensed it to a small California biotech, Amylin Pharmaceuticals. Synthetic exendin-4 — exenatide — became Byetta, and in April 2005 the FDA approved it: the first GLP-1 receptor agonist to reach patients. It was a twice-daily injection with a two-and-a-half hour half-life. Modest by today’s standards, and yet the entire blockbuster class descends from it.

The lizard’s trick was resistance to the enzyme. The competing idea was to keep a near-human peptide and simply stop the body from clearing it. Novo Nordisk took that road with liraglutide (Victoza, 2010): attach a C16 fatty-acid chain so the molecule binds albumin, hides in plasma, and dribbles off slowly — a once-daily shot. This is acylation, and it is the second of the two great half-life strategies, alongside DPP-4 resistance.

Semaglutide (Ozempic 2017; Wegovy 2021) is what happens when you use both pedals at once: a longer C18 di-acid tether with a spacer for tighter, more durable albumin binding, plus an Aib substitution at position 2 to lock out DPP-4 — the same vulnerability the Gila monster had closed with glycine, now closed again with a synthetic residue. The result is a once-weekly drug, and eventually a pill. A two-minute molecule became a seven-day one; the half-life calculator makes the size of that leap concrete.

A tangent worth naming, because it was a real fork in the road: instead of building a resistant agonist, you can just block the enzyme and let the GLP-1 your own gut already makes last longer. That is the DPP-4 inhibitor class — the “gliptins.” They work, they are convenient pills, and they are also modest, because raising endogenous incretin tone a little is not the same as flooding the receptor with a long-acting agonist. The gliptins are the road-not-taken that still got paved; they explain why “more GLP-1 signal” and “a bit less GLP-1 breakdown” are not interchangeable.

Once one receptor was conquered, the field went sideways rather than up. The proglucagon family has three sibling receptors — GLP-1, GIP, glucagon — with complementary metabolic jobs, so the next moves were peptides that hit more than one at once. Tirzepatide (Mounjaro 2022; Zepbound) is a GIP/GLP-1 dual agonist built on that logic. Curiously, whether you turn the GIP receptor on (tirzepatide) or off (maridebart cafraglutide), you still get weight loss — the paradox unpacked in the GIP paradox.

Retatrutide adds the third receptor — glucagon — for a triple agonist whose glucagon arm raises energy expenditure on top of the appetite and insulin arms. And a fourth idea sits alongside the incretins entirely: cagrilintide is a long-acting analog of amylin, the other post-meal satiety hormone, paired with semaglutide as CagriSema. None of this is a bigger number on the same axis; it is more of the gut’s own report card, recruited at once — the theme of the gut’s full report.

Follow the arrows all the way down and they end somewhere concrete: a research-vendor catalog, where the top sellers are coded so plainly the genealogy is almost visible. On the network’s sources, GLP-1 is semaglutide, GLP-2 is tirzepatide, and GLP-3 is retatrutide — a single-, double-, triple-receptor stack sold as a numbered ladder. Every one of them is a great-great- descendant of a lizard’s meal schedule, refined through a fatty acid, a swapped residue, an added receptor.

That vendor shorthand is a convenience, not a chemistry lesson — the “numbers” are receptor counts, not GLP versions, and the real identities matter when you read a COA or a label. What you can and can’t actually source, and how the community got to these molecules ahead of the clinics, are their own stories in what you can actually get and the community found it first.

Settled: the incretin effect is real, GLP-1 receptor agonism lowers glucose and body weight substantially, the glucose-dependence of the insulin effect keeps intrinsic hypoglycemia risk low, and large trials show cardiovascular and renal benefit for the leading agents. The venom-to-blockbuster arc is one of the cleanest “basic biology pays off decades later” stories in modern medicine.

Open: the exact split between gut, brain, and slowed emptying behind the weight loss; long-horizon safety on a population scale; whether GIP is best agonized or antagonized; and the quality of the weight lost — specifically muscle preservation during rapid loss, the question now driving interest in adjacent pathways. The Gila monster handed the field a durable agonist. It did not hand over the answers to what a lifetime on one does.