AMPK: The Energy Sensor Behind Metabolic Peptide Research

Read enough about metabolic research compounds and one acronym keeps surfacing: AMPK. It is offered as the explanation for why a peptide affects fat metabolism, why a small molecule is called an “exercise mimetic”, and why a drug prescribed to millions might also slow ageing. AMPK is a real and well-characterised enzyme, and most of that attention is deserved — but the shorthand hides a lot, and some of the claims attached to it do not survive contact with the primary literature.

This guide explains what AMPK is, how it is switched on, what it does when it is, and which of the compounds routinely called “AMPK activators” genuinely are. It spends as much space on the contested parts as on the settled ones.

What AMPK is

AMPK is a protein kinase: an enzyme whose job is to attach phosphate groups to other proteins and so change what they do. Its full name, AMP-activated protein kinase, states its defining property. It is not a peptide, not a hormone and not a receptor, which matters because it is frequently discussed in the same breath as research peptides that are all three of those things.

Structurally it is a heterotrimer — three different subunits working as one unit. The catalytic alpha subunit does the phosphorylating and is encoded by either PRKAA1 or PRKAA2. A beta subunit (PRKAB1 or PRKAB2) scaffolds the complex. The gamma subunit (PRKAG1, PRKAG2 or PRKAG3) is the sensor: it binds adenine nucleotides, and which nucleotide is bound decides whether the enzyme is active. Because each slot has two or three options, mammals assemble twelve possible AMPK complexes, which is part of why activating “AMPK” is less simple than it sounds.

The sensing logic is elegant. ATP is the cell’s energy currency; spend it and you are left with ADP and AMP. AMP binding to the gamma subunit allosterically activates AMPK, while ATP promotes the inactive state, and structural work published in 2011 established that ADP also regulates the enzyme. So AMPK does not measure how much energy a cell has in absolute terms — it measures the ratio, which is a far more sensitive readout. Our guide to peptide receptors and GPCRs covers how cells read signals arriving from outside; AMPK is the intracellular counterpart, responding to metabolites instead.

How it was found

AMPK was not discovered as one thing. During the late 1970s, separate groups described kinase activities that inactivated two enzymes of lipid synthesis: HMG-CoA reductase and acetyl-CoA carboxylase. Both activities were stimulated by AMP, and for a decade it was unclear whether these were one enzyme or two.

The question was settled in 1989, when Carling, Clarke, Zammit and Hardie purified the enzyme from rat liver roughly 4,800-fold and showed that both kinase activities co-purified as a single AMP-activated enzyme, with a 63 kDa polypeptide carrying both the catalytic site and the AMP-binding capacity. AMP raised maximal activity three- to five-fold. That paper’s title uses the name the field has used ever since.

The upstream kinases took longer. In 2003, Woods and colleagues identified LKB1 — already known as a tumour suppressor mutated in Peutz-Jeghers syndrome — as the kinase that phosphorylates AMPK’s activation loop. Two years later, Hawley and colleagues showed that calcium could activate AMPK through a second route: in LKB1-deficient cells, calcium ionophores still raised AMPK phosphorylation, and depolarising rat brain slices activated AMPK without any measurable change in adenine nucleotide levels. That second route runs through CaMKK2, and it means AMPK is not purely an energy sensor; it is also a calcium sensor.

How the switch is thrown

Two inputs converge on the same enzyme, and in practice both are usually needed. The allosteric input is AMP binding to the gamma subunit, which directly raises activity and also protects the activating phosphate from being removed. The covalent input is phosphorylation of a threonine in the alpha subunit’s activation loop. The field calls this site Thr172, which is its number in the alpha-2 isoform; in the alpha-1 isoform the equivalent residue is numbered Thr183, so anyone checking a sequence database will meet both numbers for what is functionally one site. Two kinases put the phosphate there: LKB1, working in a complex with the pseudokinase STRADα and the scaffold MO25; and CaMKK2, triggered by a rise in intracellular calcium.

What it turns on, and off

The umbrella description is that activated AMPK boosts catabolic, ATP-generating pathways and shuts down anabolic, ATP-consuming ones. Underneath that umbrella sit specific, well-mapped substrates.

Skeletal muscle is the classic setting for all of this, because its energy demand swings more sharply between rest and work than almost any other tissue; a large review literature covers AMPK during exercise and recovery, in training adaptation and in substrate handling.

Activating AMPK in the laboratory

How you switch AMPK on in an experiment shapes what the experiment can conclude, and the available tools have real limitations.

Most direct activators converge on the same allosteric pocket between the kinase domain and the carbohydrate-binding module, now generally called the ADaM site; a 2013 crystal structure of human AMPK showed a small-molecule activator bound there, stabilising the interaction between those two regions.

The metformin problem

This is the most instructive argument in the field, and it is a good illustration of the difference between a plausible mechanism and a demonstrated one.

In 2001, Zhou and colleagues reported in the Journal of Clinical Investigation that metformin activates AMPK in hepatocytes, with reduced acetyl-CoA carboxylase activity downstream, and proposed that AMPK activation gave “a unified explanation for the pleiotropic beneficial effects of this drug”. For most of the following decade that was the standard account, and it is still the version most often repeated outside the primary literature.

In 2010, Foretz and colleagues tested it the hard way, in the same journal, using hepatocytes lacking AMPK and hepatocytes lacking LKB1. If AMPK were the mechanism, metformin should have stopped working. Instead, cells lacking AMPK “displayed normal glucose production and gluconeogenic gene expression”, and metformin’s inhibition of glucose production was “amplified” in both AMPK- and LKB1-deficient cells, tracking a dose-dependent fall in intracellular ATP. Their conclusion was that metformin inhibits hepatic gluconeogenesis “in an LKB1- and AMPK-independent manner”.

Sixteen years later this is still not fully resolved, and current syntheses split the question rather than answering it. A 2021 review in Endocrine Reviews describes the mechanism by which metformin inhibits hepatic glucose production as “still highly debated”, and notes that the complex-I-inhibition-then-AMPK-activation account has “only been observed in the context of supra-pharmacological (>1 mM) metformin concentrations”. A 2026 review puts it more sharply: for metformin’s acute suppression of hepatic glucose production, “AMPK is therefore not considered fundamentally required”, while AMPK does appear to contribute to the drug’s chronic metabolic actions, mainly through ACC1 and ACC2 phosphorylation.

Where peptides fit — and what isn’t AMPK

AMPK comes up constantly in research-compound writing, and it is worth separating the compounds whose published work genuinely implicates it from those where the association is loose or simply wrong.

MOTS-c: the AMPK link is real, and it is preclinical

MOTS-c is a 16-amino-acid peptide encoded inside the mitochondrial 12S rRNA gene — a peptide written in mitochondrial rather than nuclear DNA. In the 2015 Cell Metabolism paper that characterised it, Lee and colleagues reported that its cellular actions “inhibit the folate cycle and its tethered de novo purine biosynthesis, leading to AMPK activation”, with skeletal muscle as the primary target organ. In mice, MOTS-c treatment prevented age-dependent and high-fat-diet-induced insulin resistance, and diet-induced obesity.

So the AMPK connection is not invented. What it is not is clinical: that paper’s evidence is cell culture and mouse work, and it contains no human efficacy data. Our MOTS-c research guide covers the compound in detail, and MOTS-c, SS-31 and humanin places it alongside the other mitochondrial-derived peptides.

SLU-PP-332: an ERR agonist, not an AMPK activator

SLU-PP-332 is routinely grouped with AMPK activators under the heading of “exercise mimetics”. It is not one. It is a synthetic pan-agonist of the estrogen-related receptor family, acting on ERRα, ERRβ and ERRγ. ERRs are nuclear receptors — transcription factors that bind DNA and change which genes are read. AMPK is a kinase that phosphorylates proteins. These are different classes of target with different experimental signatures, and the published mouse work on SLU-PP-332 is about ERR agonism. Our SLU-PP-332 guide covers what that work actually showed.

5-Amino-1MQ: an NNMT inhibitor, upstream of NAD+

5-Amino-1MQ is also not an AMPK activator, and also not a peptide. It is a small methylquinolinium molecule that inhibits nicotinamide N-methyltransferase. NNMT methylates nicotinamide, diverting it away from NAD+ salvage; inhibit NNMT and intracellular NAD+ rises. That is a genuine metabolic intervention, but it runs through NAD+ availability rather than through the adenine-nucleotide sensing AMPK does. See our 5-Amino-1MQ guide and, for the coenzyme side, NAD+, NMN and nicotinamide riboside.

The cancer question, which cuts both ways

Because AMPK sits downstream of the tumour suppressor LKB1 and restrains anabolic growth, it was proposed to be a tumour suppressor. The evidence has not cooperated: AMPK also appears to help established tumours survive energy stress. Hardie’s 2015 review concluded that AMPK “can therefore be either a friend or a foe in cancer, depending on the context”, and a 2020 review put it as AMPK switching “from tumor suppression to tumor promotion” once tumorigenesis has occurred — implying activators might suit prevention while inhibitors might suit treating some cancers. Anyone presenting AMPK activation as unambiguously protective is overstating the record.

What’s established and what isn’t

Sourcing metabolic reagents in Canada

If you are buying reagents for metabolic-signalling work, the documentation questions are the same as for any research peptide, with one addition specific to this area: several of the compounds involved are not peptides at all, and their certificates should reflect that.

• Check that identity testing suits the molecule. A 16-residue peptide such as MOTS-c is characterised by HPLC purity and a mass-spectrometry identity check; a small molecule like 5-Amino-1MQ, or a coenzyme like NAD+, is a different analytical problem. See our guides to reading an HPLC chromatogram and reading a mass spectrum.
• Match the batch number on the certificate to your vial. Our COA guide covers what else a useful certificate contains.
• Account for net peptide content. Synthetic peptides arrive as salts carrying counter-ions and water, so labelled mass is not peptide mass. Our guide to net peptide content works through it.
• Store cold, dry and dark, and keep material lyophilized until needed. See the peptide storage guide and what lyophilized means.

MOTS-cThe mitochondrial-derived peptide whose 2015 paper reported AMPK activationView product NAD+The redox coenzyme behind the adjacent, sirtuin-based energy-sensing pathwayView product 5-Amino-1MQAn NNMT inhibitor — NAD+ salvage, not AMPKView product SLU-PP-332An ERR pan-agonist, frequently miscategorised as an AMPK activatorView product

Frequently asked questions