Skip to content

AMPK

Also called: AMP-activated protein kinase

Written by Reviewed Sep 2026

The cell's fuel gauge. It reads how much of the cell's energy has been spent, and when the reading drops, it switches on the jobs that make energy and switches off the jobs that use it.

What it is

AMPK is an enzyme made of three parts. One part does the work and two parts control it. Its job is to notice when a cell is running short of energy and do something about it. Everything it does follows one rule: nothing costly gets built while the gauge reads low.

The key thing about AMPK is that it does not measure how much energy a cell has. It measures a balance. The cell's energy currency is ATP, and cells keep ATP inside a narrow range, so the raw amount is a poor signal. When a cell spends ATP faster than it makes it, the spent form builds up. An enzyme then turns two spent units into one ATP and one fully spent unit called AMP. Because of the maths of that swap, a small fall in ATP causes a much bigger rise in AMP. So AMP works like an amplifier for energy stress, and that is why the sensor is built around it.

The honest headline for a fasting page is size. Exercise switches on AMPK far more strongly and far more quickly than fasting does, and the muscle samples from people agree with that.

How it works

One of the control parts has pockets that grab energy molecules. When AMP sits in those pockets, it does three things at once. It helps a key spot on the working part get tagged, which switches it on. It shields that tag from being stripped off. And it boosts the enzyme even more once it is on. ADP, the half-spent form, helps with the first two. ATP competes for the same pockets and switches the sensor off. That is what makes it a balance sensor, not just an AMP counter.

The key spot is tagged by an upstream enzyme, mainly LKB1. LKB1 is a tumour-fighting protein that works with two helper proteins. LKB1 does not respond to energy itself. It is always on. What changes is how easily AMPK gets tagged by it, and how well AMPK hangs onto the tag. A second upstream enzyme, CaMKK2, tags the same spot when calcium rises inside the cell. That is how working muscle switches on AMPK, apart from the energy balance.

What AMPK switches on. Fat burning: it lowers malonyl-CoA, which opens the CPT1 gate that lets fat into the mitochondria, the cell's power plants. Sugar uptake: it moves the sugar doorway GLUT4 to the surface of muscle cells, without needing insulin. New mitochondria: it tags PGC-1 alpha, a protein that helps build them. And autophagy: it tags ULK1 to start the recycling process.

What it switches off. mTORC1, the build switch, in two ways. The making of fat and cholesterol. The storing of sugar as glycogen. And protein making.

Note where PGC-1 alpha sits. AMPK tags it one way, and SIRT1 changes it another way. The energy sensor and a second sensor for the cell's chemical state both act on the same protein, by two different kinds of chemistry.

What switches it on

The balance tipping toward AMP. In practice, that means working muscle, low oxygen, and anything that slows energy making or speeds up energy use. Cells also sense sugar supply through a separate route on the digestion sac, and they sense stored sugar (glycogen) through a part of AMPK that grabs carbohydrate.

On the timing
There is no hour in people when AMPK switches on during a fast. The muscle samples from people argue against a dramatic switch-on at all. In the first 8 hours of a fast, the most trusted AMPK reading drifted slightly down, not up, though the change was too small to be sure of. At 36 hours, AMPK activity was not the main finding, and it was lower in trained people than untrained ones. This does not mean AMPK never switches on anywhere in a fasting person. It means the tissue that can be sampled does not show it. The liver, where the energy change is biggest, cannot be sampled.

What has been measured

In people

In a living person, AMPK is measured by testing a needle sample of muscle for two chemical tags: one on AMPK itself and one on its target, ACC. The ACC tag is usually the better reading. It is a direct target, and it sums up activity over a stretch of time, while the AMPK tag is just a snapshot. Measuring the energy balance itself in a muscle sample does not work well, because the energy molecules break down while the sample is being taken and frozen. Within those limits, the result from fasting people is a no, and it is worth saying plainly.

  • A 36 hour fast, with thigh muscle samples at 2, 12, 24 and 36 hours, in trained and untrained people. AMPK activity, and the tag AMPK puts on ULK1, were lower in trained people than untrained people during the fast. The paper does not report a big fasting-driven rise in AMPK in either group. Its overall take is that 36 hours of fasting only mildly affects some of the signals behind autophagy, and that fitness level shapes the response (Journal of Applied Physiology, 2018).
  • The first 8 hours of a fast, in 10 healthy men, with and without about 400 calories of extra arm-cycling exercise. The ACC tag and a tag on SIRT1 both drifted down a little, though the change was too small to be sure of. PGC-1 alpha protein fell during the fast instead of rising, and its gene signal drifted down a little. A marker on the protein p53 rose. The proteins NFE2L2 and NRF1 rose, while TFAM and COXIV did not change. Adding the exercise did not strengthen the response. It actually blunted the rise in the p53 gene signal (Pflugers Archiv, 2021).
  • The same paper says plainly that these building pathways for new mitochondria do not switch on quickly in human muscle, across 10 to 72 hours of fasting. That is the single most useful line in the research on AMPK and fasting in people.

In animals and in cells in a dish

How the gauge works, what switches it on, its side routes for sensing, and its control over new mitochondria were all worked out with purified proteins, cells in a dish and mouse muscle. The most prescribed drug linked to this switch, metformin, has a truly disputed way of working. The simple version of that story was knocked down in mouse liver cells.

  • In rodent muscle, the pathways that build new mitochondria switch on within 6 hours of fasting. The Pflugers Archiv paper (2021) sets this against the human result in the people column. That contrast between species is the point.
  • In a dish, with lab chemistry: LKB1, working with its two helpers, was found to be the upstream enzyme that switches on AMPK (Journal of Biology, 2003). How the control part grabs energy molecules, the side routes for sensing sugar and glycogen, and how DNA damage can switch it on are set out in a current review (Nature Reviews Molecular Cell Biology, 2023).
  • In mouse muscle and cells: AMPK tags PGC-1 alpha directly. That tag is needed for PGC-1 alpha to switch on its own gene and genes further down the line, including the sugar doorway GLUT4 and genes for mitochondria (Proceedings of the National Academy of Sciences, 2007). The exact spots on the protein were not read from the source, so they are not given here.
  • In a dish: AMPK tags ULK1 to switch it on. That is the step that links energy sensing to mitophagy (Science, 2011).
  • In a dish, with isolated mitochondria and cells: metformin blocks one step of the mitochondria's energy line. That lowers ATP making, which raises AMP and ADP (Biochemical Journal, 2000). It builds up in liver cells because a carrier called OCT1 pulls it in. That is why the liver is where it mainly works.
  • In mouse liver cells and in mice: the direct challenge to the simple story. Metformin's main effect in patients is to cut the sugar the liver makes. In mice, that effect happened without LKB1 or AMPK, through a drop in the liver's energy level itself (Journal of Clinical Investigation, 2010). The best current position is this. Metformin lowers the liver's energy level, and that switches on AMPK. But its blood-sugar-lowering effect at least partly works without AMPK.

Why it matters

AMPK is where the popular story and what has been measured in people split apart the most. That split is useful, not disappointing. Fasting is often called a strong AMPK booster, based on work in rodents. In human muscle, across every fast length anyone has sampled, that boost has not been seen. One careful study found the markers drifting the other way in the first 8 hours.

The reason is size. Exercise shifts the energy balance hard, because it burns ATP directly. It also releases calcium, which switches on the same enzyme a second way. Fasting does neither. If what you want is AMPK switched on in muscle, exercise is the much stronger push, and the data from people say so.

What this does not show is that AMPK stays off during a fast in people. The liver is where the energy change is biggest, and where the body actually makes new sugar and ketones. Nobody samples a healthy volunteer's liver. That is a tissue nobody has measured, not a tissue where nothing was found.

A claim you will see repeated

Fasting and metformin are often said to pull the same lever, and metformin gets called a fasting copycat because of it. There are two problems. First, how metformin works is disputed at exactly the step the comparison depends on. In mouse liver cells, its blood-sugar-lowering effect worked without LKB1 or AMPK. Second, the muscle samples from people do not show fasting strongly switching on AMPK in the first place. So the shared lever is claimed at both ends and measured at neither.

Sources

  1. Study in people2018
    Training state and skeletal muscle autophagy in response to 36 h of fasting

    Journal of Applied Physiology

    Untrained and endurance trained people fasted 36 hours and gave thigh muscle samples at 2, 12, 24 and 36 hours. Energy and recycling start signals were lower in the trained people during the fast. Fasting lowered three recycling proteins, but only in untrained people. The authors sum it up this way: 36 hours of fasting only modestly changed muscle recycling.

    Untrained and endurance trained subjects fasted 36 hours with vastus lateralis biopsies at 2, 12, 24 and 36 hours. AMPK Thr172, AKT Thr308 and ULK1 serine 555 phosphorylation were all lower in trained than untrained subjects during the fast. Fasting reduced LC3-I, LC3-II and p62 protein content, and did so in untrained subjects only. The authors' summary is that skeletal muscle autophagy was only modestly affected by 36 hours of fasting.

  2. Study in people2021
    Increasing whole-body energetic stress does not augment fasting-induced changes in human skeletal muscle

    Pflugers Archiv, European Journal of Physiology

    10 healthy men gave thigh muscle samples at the start, 4 hours and 8 hours of a supervised fast. In one fast, they also did 2 hours of arm cycling. Two energy signals dipped, but not clearly. PGC-1 alpha, a power-plant building protein, fell. Some other signals rose and two power-plant proteins did not change. The paper notes that rodent muscle starts building new power plants within 6 hours of fasting. It says human muscle does not do this from 10 to 72 hours.

    10 healthy males, biopsies at baseline, 4 and 8 hours of a supervised fast, one arm adding 2 hours of arm ergometer exercise. ACC and SIRT1 phosphorylation were non-significantly decreased (both P < 0.06), PGC-1 alpha protein decreased, p53 acetylation increased (P < 0.01), NFE2L2 and NRF1 protein increased, TFAM and COXIV were unchanged. The paper states that the rapid activation of mitochondrial biogenic pathways seen in rodent muscle within 6 hours of fasting is absent in human muscle across 10 to 72 hours.

  3. Lab (cells in a dish)2003
    Complexes between the LKB1 tumor suppressor, STRAD alpha/beta and MO25 alpha/beta are upstream kinases in the AMP-activated protein kinase cascade

    Journal of Biology

    In a dish. LKB1, working with two partner proteins, is the main enzyme that switches on AMPK by tagging it at one spot. LKB1 is always on. What changes is how easily AMPK gets tagged and how well it holds on to the tag.

    Cultured cells and biochemistry. LKB1 in complex with STRAD and MO25 is the principal upstream kinase that phosphorylates AMPK at Thr172. LKB1 is constitutively active; what is regulated is AMPK's susceptibility to being phosphorylated and its resistance to dephosphorylation.

  4. Review of studies2023
    New insights into activation and function of the AMPK

    Nature Reviews Molecular Cell Biology

    A review of how AMPK, the cell's low-energy alarm, gets switched on. It covers the classic energy-sensing route and newer ones: sensing sugar at the cell's digestion sac, sensing stored sugar, and damage to the digestion sac or to DNA.

    Current synthesis of AMPK activation, including the nucleotide binding mechanism of the gamma subunit and the non-canonical routes: lysosomal glucose sensing via fructose-1,6-bisphosphate, aldolase and Axin/LKB1, glycogen sensing through the beta subunit carbohydrate binding module, and activation by lysosomal damage and nuclear DNA damage.

  5. Animal study2007
    AMP-activated protein kinase (AMPK) action in skeletal muscle via direct phosphorylation of PGC-1alpha

    Proceedings of the National Academy of Sciences

    In mice and in a dish. AMPK tags PGC-1 alpha directly in muscle. That tag is needed for PGC-1 alpha to switch on its own gene and genes for sugar uptake (GLUT4) and power plants.

    Mouse skeletal muscle and cells. AMPK phosphorylates PGC-1 alpha directly, and that phosphorylation is required for PGC-1 alpha dependent induction of its own promoter and of downstream genes including GLUT4 and mitochondrial genes.

  6. Lab (cells in a dish)2011
    Phosphorylation of ULK1 (hATG1) by AMP-activated protein kinase connects energy sensing to mitophagy

    Science

    In a dish. AMPK tags ULK1 and switches it on. That links a cell's energy level straight to clearing out its power plants.

    Cultured cells. AMPK phosphorylates ULK1 to activate it, connecting cellular energy status directly to the removal of mitochondria.

  7. Lab (cells in a dish)2000
    Evidence that metformin exerts its anti-diabetic effects through inhibition of complex 1 of the mitochondrial respiratory chain

    Biochemical Journal

    In a dish. The diabetes drug metformin blocks one step of the power plant's energy chain, called complex 1. That lowers the cell's ready energy (ATP) and raises the low-energy signals AMP and ADP.

    Isolated mitochondria and cells. Metformin inhibits complex 1 of the mitochondrial respiratory chain, lowering ATP production and raising AMP and ADP.

  8. Animal study2010
    Metformin inhibits hepatic gluconeogenesis in mice independently of the LKB1/AMPK pathway via a decrease in hepatic energy state

    Journal of Clinical Investigation

    In mice and in mouse liver cells. Metformin still cut the liver's sugar output without LKB1 or AMPK. It did it by lowering the liver's energy level itself. This challenges the idea that metformin works by switching on AMPK.

    Mouse hepatocytes and mice. Metformin's suppression of hepatic glucose production occurs without LKB1 or AMPK, through a decrease in hepatic energy state itself. This is a direct challenge to the account in which metformin works by activating AMPK.