Myokines: The Hormones Muscle Releases During Exercise (Irisin, IL-6 and More)

Myokines are signalling molecules that skeletal muscle makes and releases, most of them proteins or peptides, and they are why physiologists now describe muscle as an endocrine organ. The idea that working muscle talks to the rest of the body is old, but the word only dates from 2003, and the list of candidates has grown far faster than the evidence behind most of them.

This guide covers where the term came from, IL-6 as the first myokine, the contested irisin hormone, myostatin as muscle’s own brake, the other exercise hormones on myokine lists, why they are hard to measure, and how strong the evidence is for each.

What myokines are

A myokine is a signalling molecule produced and released by skeletal muscle fibres that acts on the muscle itself, on neighbouring cells or, through the blood, on distant organs: autocrine, paracrine and endocrine actions, in the vocabulary of our peptide hormones guide. In their 2008 review in Physiological Reviews, Bente Klarlund Pedersen and Mark Febbraio describe skeletal muscle as having been identified as an endocrine organ.

Physiologists had long suspected that contracting muscle releases a blood-borne signal, variously called the “work factor” or the “exercise factor”. In 2003, Pedersen and colleagues proposed IL-6 as a candidate and suggested that IL-6 and other cytokines produced and released by skeletal muscle be named myokines. A 2020 Endocrine Reviews overview describes muscle as able to secrete hundreds of myokines that reach the brain, fat, bone, liver, gut, pancreas, blood vessels and skin, while noting that only a few have been allocated a specific function in humans. That gap runs through everything below.

IL-6: the first myokine, and why exercise IL-6 is different

IL-6, a gp130-family cytokine better known as a marker of inflammation, became the founding myokine because its circulating level rises up to 100-fold during exercise. Proving that muscle is the source meant sampling the blood entering and leaving a working limb. In 2000, Steensberg and colleagues had six healthy men perform five hours of one-legged knee-extensor exercise and sampled arterial and femoral venous blood from both legs. The working leg released IL-6, the resting leg did not, and arterial IL-6 rose 19-fold: production in contracting muscle could account for the increase.

Same molecule, different context

Two 2003 experiments in healthy men suggest the exercise signal is not an inflammatory one. In the first, recombinant IL-6 given for three hours at plasma levels typical of strenuous exercise raised the anti-inflammatory molecules IL-1 receptor antagonist (IL-1ra) and IL-10, and cortisol, without raising tumour necrosis factor alpha (TNF-α), although C-reactive protein did rise. In the second, a small endotoxin challenge raised TNF-α at rest, but the rise was totally attenuated when the challenge came during three hours of cycling or of IL-6 infusion. Pedersen and Febbraio’s review describes a paradox: IL-6 is released after exercise, when insulin action is enhanced, yet it is also associated with obesity and reduced insulin action (see our guide to inflammatory mediators).

Blocking IL-6 in a randomized trial

In a 12-week randomized, placebo-controlled trial, adults with abdominal obesity received tocilizumab, an antibody that blocks the IL-6 receptor, or placebo, with or without bicycle exercise training. Training reduced visceral adipose tissue mass, but the “effect of exercise was abolished in the presence of IL-6 blockade”. Few myokines have been tested this directly in people.

Irisin: the contested hormone

In 2012, Bruce Spiegelman’s laboratory reported in Nature that raising PGC-1α in mouse muscle increases FNDC5, a membrane protein that is cleaved and secreted as a hormone the authors named irisin. Irisin pushed white fat cells, in culture and in mice, toward a brown-fat-like program with more UCP1, raised energy expenditure in mice, and was reported to rise with exercise in mice and humans. Our brown adipose tissue guide covers the browning claims. Three findings then complicated the story.

• The human gene is unusual. Human FNDC5 starts with ATA instead of the conserved ATG. In Raschke and colleagues’ 2013 experiments, a human ATA construct made only 1% of the full-length protein made from ATG, FNDC5 mRNA did not change with training in two human studies, and irisin did not brown human preadipocytes.
• The antibodies were not specific. In 2015, Albrecht and colleagues found that four commercial antibodies cross-reacted with other serum proteins and detected no band of the expected size in any biological sample, casting doubt on earlier ELISA values.
• Mass spectrometry found it anyway. Also in 2015, Jedrychowski and colleagues quantified irisin in human plasma by tandem mass spectrometry with isotope-labelled standards. It was translated mainly from the ATA codon, and levels were roughly a fifth higher in people doing aerobic interval training than in sedentary people.

In 2018 the Spiegelman group proposed αV integrins as irisin’s receptor on osteocytes and fat cells, and mice lacking FNDC5 were completely protected from the osteocytic osteolysis that follows ovariectomy. A 2021 Endocrine Reviews assessment by four authors of the 2015 antibody critique counts more than 1,000 irisin papers, warns that many relied on flawed assays, and lists the human start-codon question as unresolved. The defensible summary: irisin is a real peptide in human blood whose functions in people have not been established.

Myostatin: the brake muscle makes for itself

Some myokines say “stop”. In 1997, McPherron, Lawler and Lee described GDF-8, a TGF-β superfamily member expressed specifically in developing and adult skeletal muscle. In mice lacking it, individual muscles weighed two to three times more than normal, through both hyperplasia (more fibres) and hypertrophy (larger fibres), which marked GDF-8 as a negative regulator of muscle mass. The same year, McPherron and Lee linked mutations in the gene, by then called myostatin, to double muscling in cattle. Myostatin predates the word “myokine” but fits its definition.

Four more myokines: BDNF, SPARC, METRNL and apelin

BDNF. Brain-derived neurotrophic factor is usually discussed as a brain growth factor (see our guide to neurotrophic factors). In 2009, Matthews and colleagues found that its mRNA and protein rose in human muscle after exercise. In cell and isolated-muscle experiments it increased AMPK phosphorylation and fat oxidation, an effect lost when AMPK was blocked (see our AMPK guide). But muscle BDNF appeared not to be released into the circulation, so it seems to act within muscle.

SPARC. A single exercise bout increased the expression and secretion of SPARC (secreted protein acidic and rich in cysteine) in muscle in mice and humans. In a chemically induced mouse model of colon tumours, regular low-intensity exercise reduced aberrant crypt foci in normal mice but not in mice lacking SPARC. Human data are limited to release.

METRNL. Meteorin-like is induced in muscle after exercise and in fat on cold exposure. In mice, raising it increased energy expenditure, improved glucose tolerance and switched on beige-fat genes through eosinophils, IL-4 and macrophages in fat.

Apelin. A 2018 study found that apelin production is induced by muscle contraction and falls with age in humans and rodents. In older people, apelin levels were positively associated with the beneficial effects of exercise; in old mice, restoring apelin signalling considerably enhanced muscle function. The human data are associations.

Beyond muscle: exerkines, FGF21 and BAIBA

A 2022 review in Nature Reviews Endocrinology defines exerkines as “signalling moieties released in response to acute and/or chronic exercise”, from muscle (myokines), heart (cardiokines), liver (hepatokines), white fat (adipokines), brown fat and neurons. The word has a Canadian origin: it was introduced in a 2016 review by Adeel Safdar, Ayesha Saleem and Mark Tarnopolsky of McMaster University in Hamilton, Ontario.

FGF21 shows why the wider word is needed. Sampling blood across the liver and across the leg in exercising people, Hansen and colleagues found that the liver secreted FGF21 at rest and rapidly increased its output during exercise, while the leg contributed nothing; glucagon stimulated the release and insulin inhibited it (see our guide to glucagon function). FGF21 is better described as a hepatokine.

BAIBA shows that exercise signals need not be peptides. β-Aminoisobutyric acid, a small-molecule myokine secreted by muscle cells with extra PGC-1α, rises in human plasma with exercise. In mice and cells it raised brown-fat genes in white fat and fatty-acid oxidation in liver cells through PPARα; in people, the link so far is an inverse association with cardiometabolic risk factors.

Mitochondrial-derived peptides are encoded in mitochondrial rather than nuclear DNA. In 2021, Reynolds and colleagues found that exercise raised endogenous MOTS-c in human muscle and blood; the evidence on its effects comes from mice and cells, where giving MOTS-c enhanced physical performance in young, middle-aged and old mice. No MOTS-c was given to people. In ten young men, humanin rose in plasma and muscle after one bout of high-intensity interval cycling. Our MOTS-c guide and mitochondrial peptide overview go further.

The major myokines compared

“Human” below means measured or manipulated in people, not a proven function.

Why myokines are hard to measure

Irisin is the most visible case of a field-wide problem; the 2022 exerkine review calls for more rigorous experimental design so that studies can be compared. Four questions help:

• Where did it come from? A plasma rise does not show that muscle made the molecule; sampling blood entering and leaving a tissue does, as it did for IL-6 and FGF21.
• Is the assay specific? Immunoassays are only as good as their antibodies. Mass spectrometry detects the peptide itself; our LC-MS guide explains how identity is confirmed.
• Is mRNA standing in for protein? A gene’s mRNA in a biopsy is not a cleaved, secreted, active protein, as FNDC5 shows.
• Is it causal, and in which species? Plasma correlations cannot show that a molecule matters; knockouts can, in mice. Human tests like the IL-6 blockade trial are few.

What’s established and what isn’t

Sourcing research peptides for myokine work in Canada

For peptide reagents used alongside myokine work, the documentation questions are standard:

• Confirm identity, not just purity. A purity figure shows how much of a sample is one compound; a mass-spectrometry check shows that it is the right one.
• Match the batch number on the certificate to your vial; our peptide COA guide explains what else to look for.
• Store lyophilized material cold, dry and dark; see the peptide storage guide.
• Know what you are buying: our guide to research peptides in Canada covers research-grade reagents and supplier documentation.

MOTS-c16-residue peptide encoded in the mitochondrial 12S rRNA geneView product HumaninMitochondrial-encoded peptide first described in 2001 as a neuronal rescue factorView product

Frequently asked questions