Tirzepatide: How One Molecule Hits Two Receptors

The way a hormone drug usually works is that you take a hormone the body already makes, find the receptor it plugs into, and build a sturdier copy that plugs into the same receptor and lasts longer. The drug this article is about does something greedier: tirzepatide is, in the words of a review titled “The Cardiovascular Effect of Tirzepatide,” “a 39-amino-acid synthetic peptide functioning as a dual GIP/GLP-1 receptor agonist.” One molecule switches on two different receptors, the ones for the gut hormones GIP and GLP-1. (The same review reports that this earned it the nickname “twincretin,” which is a word somebody said out loud and then kept.)

And it works. In the SURMOUNT-1 phase 3 trial in adults with obesity, the drug produced mean weight reductions of 19.5% at a 10mg dose and 20.9% at 15mg. Drugs that hit only the GLP-1 receptor land at approximately 6% to 15% weight loss in clinical trials, according to one pharmacology review, while dual GIP/GLP-1 agonists land at approximately 15% to 20%. So the second receptor seems to be adding something. What exactly it adds is the part the field has not worked out, and it is a stranger gap than it sounds.

How one peptide fits two different receptors

Start with why this is possible at all. A structural paper on incretin receptors describes the GIP receptor and the GLP-1 receptor as “homologous GPCRs,” meaning they are related proteins from the same family, built on similar plans. A peptide hormone is a chain of amino acids, and the receptor reads parts of that chain. If two receptors read chains in broadly similar ways, a single chain can, with enough editing, read acceptably to both.

Here is what the editing looks like, borrowed from a neighboring problem. Structural work on dual agonists of the GLP-1 and glucagon receptors (a different pair, but the same kind of puzzle) found that differences in the first three residues of the peptide determine which receptor it prefers, through how those residues engage a deep binding pocket. The middle of the peptide, meanwhile, recognizes two extracellular loops and the top of the first transmembrane helix. So a few amino acids at the front end do a lot of the choosing, and the rest of the chain does a lot of the holding on.

The relevant legal standard, as I understand it, is not that the molecule fit each receptor perfectly. It is that the molecule fit each receptor to a degree the parties find commercially reasonable, said degree to be determined by the receptor in its sole discretion and reviewed by the regulator at the regulator’s leisure. In practice this means a lot of meetings where a medicinal chemist says something like (I am inventing this) “can we make residue two a little more GIP without making it a lot less GLP-1?” and everyone stares at a binding curve.

The resulting relationship between one molecule and two receptors is, let’s say, blurry. It is committed to the GLP-1 receptor. It also sees the GIP receptor regularly. Everyone involved has agreed this is fine.

What imbalanced agonism means

Once you have one molecule hitting two receptors, you have a ratio. The molecule activates receptor A by some amount and receptor B by some other amount, and because it is a single molecule, those amounts travel together. You cannot take a little extra GIP on Tuesdays (the molecule is not taking requests). The idea behind imbalanced agonism is that this ratio does not have to be one-to-one, and that the best version of a dual agonist might be deliberately lopsided toward one receptor.

So the dial, for the people running a trial, is the dose. Actually, that is not quite right. The dose turns both receptors up at once, in a proportion that was welded into the molecule at the chemistry stage, so the only dial that really sets the balance is the one the chemists turned years before anyone enrolled a patient. By the time you are choosing between 10mg and 15mg, the argument about balance is over.

This is in some sense the opposite of a combination product. CagriSema, for instance, is a fixed-dose combination of two separate molecules, semaglutide and the amylin analogue cagrilintide. Two molecules can in principle be rebalanced by changing how much of each goes in. One molecule cannot. A single-molecule dual agonist is a bit like a two-way player in baseball, the one who pitches and hits: very efficient, and you do not get to trade the pitching half.

How lopsided tirzepatide actually is, and how much of its effect each receptor carries, is not settled by the sources I am working from, and I am not going to make up a ratio. (This is the kind of honest sentence that makes for a shorter article, a problem I have mostly solved by writing longer ones.) One broad review lists “biased agonist design” as one of “the three major axes of next-generation incretin drug development,” which tells you the field thinks the shape of the activation, and not just its existence, is where the next round of gains lives.

To see why one molecule beats the alternative, consider a scenario I am making up. Suppose instead you sold GLP-1 and GIP as two separate injections and let the balance be tuned per patient. Then someone notices that the best ratio might drift with weight, so you add a third vial for adjustments. Then a fourth for weekends. Soon every clinic needs a pharmacist doubling as a bartender who mixes each patient’s weekly incretin cocktail to order, and the trials become impossible to run because no two arms got the same drink. One fixed molecule is less flexible and vastly easier to test, approve and make.

What the SURMOUNT trials showed for tirzepatide

The headline is the SURMOUNT-1 result: 19.5% mean weight reduction at 10mg and 20.9% at 15mg, in adults with obesity. Across the phase 3 SURMOUNT program, the drug showed superior weight reduction compared with GLP-1 monotherapy, and a review of pediatric obesity treatment states the adult case flatly: tirzepatide has shown “greater weight-loss efficacy than selective GLP-1 receptor agonism in adults.”

The comparison holds up when somebody pools the trials. A review in Annals of Internal Medicine looked at 12 GLP-1 receptor agonists and coagonists across 26 randomized controlled trials involving 15,491 adults without diabetes. Tirzepatide came in at 17.8% weight loss, ahead of subcutaneous semaglutide at 13.9%, and behind the triple agonist retatrutide at 24.2% on 12 mg weekly. (Whether that third receptor is worth it gets its own article on this site.)

There is also SURMOUNT-MMO, a trial designed to test whether the drug reduces morbidity and mortality in adults with obesity. Its rationale and design have been published. Its results are not among what I can report here. Weight loss is the number everyone quotes. Hard outcomes are what eventually matter.

Three ways to read the extra weight loss

There are, roughly, three ways to read the gap between a GLP-1-only drug and a GIP/GLP-1 dual.

The first reading is the straightforward one: GIP does its own useful stuff, and stacking a second incretin signal on the first gets you more effect than either alone. This is the story the molecule was built to tell. It has some biology behind it; even rat pancreatic alpha-cells, for instance, carry functional receptors for both hormones, so the two signals plainly overlap in the body. Two signals, more effect. Tidy.

The second reading is that the dual agonist is, mostly, an excellent GLP-1 drug, and the GIP arm is a cooperative passenger whose contribution is real but hard to isolate. I guess the trouble with this reading is also its appeal: a head-to-head trial against a GLP-1-only drug tells you the package works better, not which half of the package is doing the work. The comparison numbers above cannot separate those.

The third reading is the uncomfortable one. Look, there is another drug in development, maridebart cafraglutide, a bispecific antibody that combines GLP-1 receptor activation with GIP receptor blockade. It pushes the GIP receptor in the opposite direction. And it also appears to add weight loss on top of GLP-1 agonism. A paper titled “The Premise of the Paradox” calls this “a central unresolved scientific question in incretin pharmacology,” which is a polite way of saying that switching the receptor on and switching it off both seem to help.

It is not obvious how both can be true. I don’t know, and neither, as of the sources here, does anyone else with certainty. One possibility is that long-term activation and blockade converge on some similar downstream state. Another is that the GIP effect is smaller or stranger than either design assumed. I suppose a third is that the two programs are measuring different things and calling them the same thing. If you designed a molecule around the premise that GIP agonism is good, you would like this question resolved in your favor, and the answer is currently arriving from a competitor whose premise is that GIP antagonism is good.

If you want to find out what your second receptor is really worth, you had better hope the people trying to switch it off keep running their trials.