How Does Tirzepatide Work? GIP and GLP-1 Explained

Tirzepatide works by locking onto two hormone receptors, GIP and GLP-1, and the strength of that grip on each one was measured directly rather than assumed.

IN VITRO

The founding pharmacology paper measured how tightly tirzepatide binds each receptor using Ki, a laboratory measure of binding affinity where a lower value means a tighter grip. Tirzepatide bound the GIP receptor with Ki 0.135 nM (standard error 0.020) and the GLP-1 receptor with Ki 4.23 nM (standard error 0.23). Its GIP receptor affinity is comparable to the body's own GIP hormone, while its GLP-1 receptor affinity is weaker than semaglutide's, reported in the same study at Ki 1.97 nM with a cAMP EC50, a separate measure of how strongly a bound receptor triggers its internal signal, of 0.0571 nM. [1]

Ki and EC50 are related but distinct measurements, and it is worth keeping them separate. Ki, drawn from a binding assay, tells you how much of the compound is needed to occupy the receptor, regardless of what happens next. EC50, drawn from a functional assay such as the cAMP test cited above, tells you how much compound is needed to actually trigger half of the maximum downstream signal once it is bound. A molecule can bind tightly (low Ki) yet trigger a comparatively weak or strong signal once bound, which is why researchers report both figures rather than either alone.

A lower Ki means tighter binding. Affinity is not the same as potency, and neither one by itself predicts what will happen in a person; they describe the molecular starting point for the effects seen later in trials.

Because the two hormone pathways are not simply doing the same job twice. GLP-1 receptor signalling has well-studied effects on glucose-dependent insulin secretion, glucagon suppression, gastric emptying, and appetite circuits in the brain. GIP receptor signalling also boosts insulin secretion, but it additionally acts in fat tissue and in brain regions where its role in body-weight regulation has been a genuine point of scientific disagreement.

Think of GIP and GLP-1 as two separate molecular switches sitting on overlapping but non-identical circuits: flipping the GLP-1 switch mainly touches the stomach, pancreas and appetite centres, while flipping the GIP switch also reaches into fat cells and other brain regions. Building one molecule that flips both switches, rather than combining two separate drugs, is the specific engineering feat behind tirzepatide.

ANIMAL STUDY

The knockout experiments are what turn this from a hypothesis into a tested claim. When researchers bred mice that lacked either the GIP receptor or the GLP-1 receptor, the extra weight loss seen with dual activation in normal mice disappeared. That result is what establishes both receptors as necessary for the enhanced effect in that model, rather than one receptor quietly doing all the work while the other rides along. [1]

It is worth being precise about what a knockout study can and cannot claim. Breeding an animal that entirely lacks a receptor from birth is a blunt instrument: the animal's body may have partially compensated for the missing receptor during development in ways that would not happen if the same receptor were simply blocked for a few weeks in an adult. That is a limitation of the knockout method in general, not a flaw specific to this study, and it is one reason knockout results are treated as strong supporting evidence for a mechanism rather than as a complete substitute for testing the actual drug in normal, unmodified animals and, eventually, in people.

What the mouse work does not establish is also worth stating plainly: it does not measure how much each pathway contributes in humans, and it does not mean the two effects simply add together arithmetically.

In plain terms, it means the drug's push on insulin release fades as blood sugar returns to normal, rather than continuing to push regardless. This is different from some older diabetes medicines that can keep lowering glucose even after it reaches a safe level.

As glucose falls toward a normal range, the signal that triggers insulin release falls with it. That is the mechanistic reason incretin-based medicines like tirzepatide carry a comparatively low risk of hypoglycaemia (dangerously low blood sugar) when used alone.

This glucose-dependence is a built-in feedback loop rather than a separate safety feature bolted on afterward: the same biochemical step that senses how much glucose is present is the step that decides how strongly to signal for insulin release, so as glucose drops, the signal driving insulin secretion drops with it, and insulin release tapers off before glucose falls dangerously low. Some older classes of diabetes medicine act on insulin release through a different, glucose-independent route, which is why they carry a materially higher hypoglycaemia risk when glucose happens to already be low.

HUMAN CLINICAL TRIAL

The trial data are consistent with that reasoning. In SURPASS-2, clinically significant hypoglycaemia below 54 mg/dL occurred in 0.6%, 0.2% and 1.7% of the tirzepatide 5, 10 and 15 mg arms, and in 0.4% of the semaglutide arm. [2] These are small numbers, measured in a population also taking metformin, and they are reported here alongside the comparator rather than in isolation.

Gastric emptying, the rate at which the stomach empties into the intestine, and brain appetite signalling are the other studied contributors to the weight effect. They are also the source of the most common side effects: nausea in 17 to 22% of tirzepatide participants in SURPASS-2, diarrhoea in 13 to 16%, and vomiting in 6 to 10%. [2] In other words, the same pathways behind the benefit are largely believed to be behind the main downside as well, since slower stomach emptying and altered gut signalling are what is thought to both reduce appetite and cause digestive discomfort in different individuals.

One way to check whether a proposed mechanism is real is to see whether the effect scales with dose in the way the mechanism would predict. Here, it broadly does.

HUMAN CLINICAL TRIAL

In SURMOUNT-1, mean body-weight change was minus 15.0% at 5 mg, minus 19.5% at 10 mg and minus 20.9% at 15 mg, against minus 3.1% on placebo, over 72 weeks in 2,539 adults. [3] A graded response across doses is the pattern expected when a receptor-driven mechanism is doing the work, though it is evidence consistent with the mechanism rather than final proof of it.

Notice that the increase flattens between the 10 mg and 15 mg doses, a gap of only 1.4 percentage points, compared with 4.5 points between 5 mg and 10 mg. Dose-response curves commonly plateau like this, while side effects tend to move in the opposite direction: discontinuation due to adverse events was 4.3%, 7.1% and 6.2% across the three doses, against 2.6% on placebo. [3] The practical takeaway is that going from 10 mg to 15 mg buys a comparatively small extra weight-loss benefit for a comparatively larger tolerability cost.

This flattening pattern, technically called a ceiling or saturation effect, is a familiar shape in receptor pharmacology: once a large enough share of available receptors are already occupied and signalling, adding more drug produces progressively smaller gains, because there are fewer unoccupied receptors left to recruit. Seeing that expected shape in the actual trial data is one more thread connecting the laboratory receptor story to the numbers measured in patients.

0.135 nM

GIP receptor Ki

4.23 nM

GLP-1 receptor Ki

2,539

participants in SURMOUNT-1

−17.8 pts

weight change vs placebo at 15 mg

It does not explain why any one individual might respond more or less than average. Mean values describe a treatment arm as a group, not a person, and the threshold data above show a wide spread of outcomes within the very same dose group.

It also does not establish what happens beyond the trial windows studied here, and it does not transfer automatically to any other molecule. A different peptide with different binding affinities is a different pharmacological question, even if it happens to share the same receptor names.

Finally, none of this describes handling a research material in a laboratory setting. A receptor mechanism characterised in mice and measured in supervised clinical trials with medical oversight is not a laboratory protocol.

For the receptor comparison against a triple agonist, read Tirzepatide versus retatrutide.