The GLP-1 discovery deeper history — Holst, Mojsov, and the science before the drug

That something interesting would take another forty years to become Ozempic.

The glucagon gene had been sequenced in 1983 by Graeme Bell and colleagues at the University of Chicago, and the sequence revealed something unexpected: the gene encoded not just glucagon but two additional glucagon-like peptides — GLP-1 and GLP-2 — embedded in the same precursor protein, called proglucagon. In the pancreatic alpha cells, proglucagon is processed primarily into glucagon. In the intestinal L cells, a different enzyme cleaves the precursor differently, releasing the two GLP fragments instead. The biology had been running all along; no one had noticed the second set of products.

Holst's group and, working in parallel, a research team at Massachusetts General Hospital led in part by a biochemist named Svetlana Mojsov, spent the mid-1980s working out the specific structure of GLP-1 and what it did. This is where the history becomes more complicated than most contemporary accounts acknowledge. Mojsov, working at Mass General and later at Rockefeller University, performed critical work characterizing the truncated form of GLP-1 — GLP-1(7-36) amide — that proved to be the biologically active version. Her 1987 paper in the Journal of Clinical Investigation established that this truncated form had potent insulinotropic activity: it caused insulin secretion in a dose-dependent way when tested in both isolated pancreatic cells and in human subjects. The form that ended up mattering pharmacologically — the one that all subsequent GLP-1 analogs were built to mimic — was the one Mojsov characterized.

The concept that emerged from this work was called the incretin effect, and it explained something that had puzzled endocrinologists for decades. When you give someone glucose intravenously — injected directly into the blood — the insulin response is modest. When you give the same person the same amount of glucose orally, the insulin response is two to three times larger. The pancreas responds more robustly to swallowed glucose than to infused glucose. Something in the act of eating, something in the gut, was amplifying the insulin signal. GLP-1 was one of the primary answers. Released from L cells in the small intestine in response to nutrient contact with the intestinal wall, GLP-1 traveled to the pancreas and potentiated the beta cell response to glucose. The gut was talking to the pancreas. The conversation had been happening every time anyone ate anything, for as long as there had been vertebrates with guts and pancreases. No one had been listening.

Daniel Drucker's laboratory at the University of Toronto — and here Toronto appears again in the history of metabolic pharmacology — spent the 1990s working out the receptor biology and the full physiological role of GLP-1. Drucker's group cloned the GLP-1 receptor in 1992, characterized its distribution across tissues, and began to map what happened when it was activated: not just insulin secretion but suppression of glucagon release, slowing of gastric emptying, satiety signaling in the brain, possible effects on cardiac function and neuroprotection. GLP-1, it turned out, was not a single-purpose incretin. It was a postprandial orchestrator with receptors in the pancreas, the stomach, the hypothalamus, the brainstem, the heart, and the kidneys. Its job description was considerably larger than "stimulate insulin."

The therapeutic potential was obvious by the mid-1990s. GLP-1 administered by infusion could normalize blood sugar in people with type 2 diabetes without causing hypoglycemia, because its insulin-stimulating effect was glucose-dependent — it amplified beta cell response only when glucose was actually present. It also suppressed appetite. It also slowed the rate at which the stomach emptied into the intestine, flattening postprandial glucose spikes. The problem was in a four-letter abbreviation: DPP-IV.

Dipeptidyl peptidase-IV is an enzyme found in the blood and on cell surfaces throughout the body, and its affinity for GLP-1 is extraordinary. Within two minutes of entering the circulation, more than half of endogenous GLP-1 is cleaved and inactivated by DPP-IV. The molecule's plasma half-life is approximately two minutes. You cannot infuse a drug every two minutes. You cannot inject something every two minutes. Native GLP-1 was pharmacologically unusable for that reason alone — brilliant biology, terrible pharmacokinetics.

Two paths emerged from this constraint, and both became major pharmaceutical categories. The first was to inhibit DPP-IV itself — to block the enzyme so that endogenous GLP-1 survived longer in the circulation. Sitagliptin, sold as Januvia, was approved by the FDA in 2006 and became the founding member of the DPP-IV inhibitor class. These drugs extend the life of the body's own GLP-1 by a factor of two to three; they produce modest but clinically meaningful improvements in blood sugar control. They are not transformative, because they depend on endogenous GLP-1 production, which is already compromised in people with type 2 diabetes.

The second path was structural: engineer a GLP-1 analog that the DPP-IV enzyme couldn't recognize. The cleavage site on native GLP-1 is at position 2, where an alanine residue gives DPP-IV its grip. Change that residue, or build around it, and you defeat the enzyme. The first successful approach came from an unexpected source.

Exendin-4 is a peptide found in the saliva of the Gila monster, a venomous lizard native to the American Southwest, whose existence in Heloderma suspectum had been noted since the 1990s by John Eng at the Bronx VA Medical Center. Exendin-4 was structurally similar to GLP-1 — similar enough to bind the GLP-1 receptor — but contained different amino acids at DPP-IV's cleavage site. It was naturally DPP-IV resistant, with a half-life measured in hours rather than minutes. Synthetic exenatide, developed by Amylin Pharmaceuticals, became Byetta: the first GLP-1 receptor agonist approved by the FDA for type 2 diabetes, in 2005. It required twice-daily injections and had GI side effects, but it proved that DPP-IV-resistant GLP-1 analogs could work in humans.

Novo Nordisk took the human GLP-1 sequence and solved the half-life problem differently. Liraglutide attached a fatty acid chain to the GLP-1 backbone; the fatty acid bound to albumin in the bloodstream, protecting the molecule from DPP-IV and extending its half-life to thirteen hours. Once-daily injection. Approved in 2010 as Victoza for diabetes, and later at a higher dose as Saxenda for obesity. Then the team extended the fatty acid chain further and added a linker: semaglutide, with a half-life of approximately one week. Once-weekly injection. Approved as Ozempic for diabetes in 2017, and as Wegovy for obesity in 2021, at a higher dose.

The weight loss numbers from the semaglutide trials — a mean of around 15 percent body weight loss over 68 weeks in the STEP trials — were unlike anything the obesity pharmacology field had seen. They were closer to what bariatric surgery achieved than to what any previous drug had produced. The mechanism included the gastric emptying effect, the hypothalamic satiety signaling, possibly direct effects on reward circuits in the brain. The cultural moment that followed — the shortages, the celebrity use, the social media omnipresence of "Ozempic" — was the surface expression of something that had been building in the literature for four decades.

Eli Lilly's tirzepatide, approved as Mounjaro for diabetes in 2022 and Zepbound for obesity in 2023, added a second receptor — the GIP receptor, for glucose-dependent insulinotropic polypeptide, another incretin — to the same molecule. The mean weight loss in trials exceeded 20 percent. The dual agonist outperformed the single agonist, which is consistent with the incretin biology: GLP-1 and GIP are synergistic signals, and engaging both may produce effects that neither alone can match. Triple agonists adding glucagon receptor activity are in development. The field is still accelerating.

In 2023, the Nobel Prize in Physiology or Medicine was awarded to Katalin Karikó and Drew Weissman for mRNA vaccine technology — not for GLP-1. The GLP-1 work was widely discussed in Nobel-adjacent commentary as an example of consequential science that had not yet been honored. Holst's name appeared prominently. Drucker's name appeared. Mojsov's name appeared less often, and when it did, it was frequently in the context of an argument that her contribution to characterizing the active truncated form of GLP-1 had been systematically underweighted in the credit-allocation narratives that precede Nobel consideration. The debate touched on questions that run through the history of science with depressing regularity: how authorship hierarchies work, how laboratory heads accumulate credit for work they supervised but didn't perform, how women in science find their contributions attributed upward. Mojsov spoke about it publicly in ways that were careful and specific. The resolution, if any comes, will matter to the historical record.

What the deep GLP-1 history teaches about basic science is something the news coverage of Ozempic shortages and celebrity weight loss stories mostly doesn't surface: that the molecule commanding global attention in 2023 was characterized by researchers working in basements and animal labs in the early 1980s, driven by curiosity about intestinal peptide fractions that had no commercial application and no obvious therapeutic target. Holst's group was not trying to develop an obesity drug. Mojsov was not imagining a trillion-dollar pharmaceutical market. Drucker was not building toward Novo Nordisk's market capitalization. They were trying to understand what the gut did when it encountered glucose, because that was an interesting biological question and no one had answered it properly.

The translation from basic science to clinical pharmacology took roughly forty years and required not just the biological discoveries but the engineering solutions: the fatty acid conjugation strategy, the half-life extension, the dose optimization, the large-scale clinical trials. None of that happens without the foundational work. And the foundational work happens, if it happens at all, because someone somewhere is willing to spend years characterizing intestinal peptide fractions that no one else considers important.

The gap between the discovery of GLP-1 biology and its arrival as a widely prescribed drug is not unusual — it reflects the normal timeline for basic-to-clinical translation in pharmacology, which runs closer to two to four decades than to the accelerated timelines that crisis medicine (HIV, COVID vaccines) sometimes achieves. What it suggests about the peptide field more broadly is that the molecules being characterized in laboratories today — the receptor agonists, the neuropeptides, the bioregulators — are operating on a translation timeline that may not resolve into clinical tools until the 2040s or 2050s. The curiosity that precedes the application is never wasted. It just looks that way from inside the decades before the drug arrives.