What Is Irisin? The Exercise Myokine and Its Measurement Problem

Peptide science · Exercise physiology

Irisin is a myokine — a signalling molecule released by working muscle — that was announced in 2012 as the link between exercise and the browning of white fat. It is also one of the most contested molecules in modern endocrinology, and the reason is unusual: the argument has been less about what irisin does than about whether anyone was measuring it correctly.

What is irisin?

Irisin is a 112-amino-acid fragment cleaved from the extracellular portion of FNDC5, a type I membrane protein expressed in skeletal muscle. Its predicted molecular weight is about 12,300; non-glycosylated recombinant irisin runs near 13,000 on a gel, and the glycosylated form — irisin carries two N-linked glycosylation sites — runs as a diffuse band around 20,000.

Two structural facts are worth holding onto. First, the crystal structure shows a fibronectin type III beta-sandwich that forms an unusual continuous beta-sheet dimer, burying a large interface; irisin circulates as a dimer, not a monomer. Second, the irisin sequence is 100 percent identical in human, mouse, rat and cattle — extraordinary conservation that argues the molecule does something, even where the specifics remain disputed.

How irisin is proposed to work

The 2012 Nature paper by Boström and colleagues laid out the chain: endurance exercise raises PGC-1α in muscle, PGC-1α drives FNDC5 expression, FNDC5 is cleaved and secreted as irisin, and irisin acts on white adipose tissue to induce UCP1 and a brown-fat-like thermogenic program. In mice, mildly raising blood irisin increased energy expenditure without any change in movement or food intake.

The receptor took six more years to find. In 2018, Kim and colleagues reported in Cell that integrins of the αV class, particularly αVβ5, mediate irisin signalling on osteocytes and adipocytes. That paper is important for a reason beyond the identification itself: irisin triggered FAK phosphorylation at concentrations as low as 10 pM, one of the very few reported irisin effects occurring at a level that could plausibly exist in blood. Blocking αVβ5 with cyclo-RGDyK abolished the signal.

The measurement problem

Here is where irisin stops being an ordinary hormone story. Two independent problems collided.

The first is genetic. The human FNDC5 gene has an atypical ATA start codon rather than the canonical ATG — it is the only known human gene with an ATA start not conserved in other species. In transfection experiments, an ATA construct produced FNDC5 protein at roughly one percent of the ATG version. If that is what happens in muscle, human irisin should be present at very low levels indeed.

The second is technical. In 2015, Albrecht and colleagues tested four commercial anti-irisin antibodies and found prominent cross-reactivity with unrelated serum proteins and no immunoreactive band of the expected size in biological samples. Their paper was titled, bluntly, “Irisin — a myth rather than an exercise-inducible myokine,” and they concluded that results obtained with commercial ELISA kits were called into question as a class.

The numbers make the point better than any argument. Reported human circulating irisin across the published ELISA literature spans from about 13.5 pg/mL to 14.8 µg/mL — six orders of magnitude. Two lots of the same manufacturer’s kit have correlated at r = 0.22. Different assays run on the same samples have correlated as low as r = 0.03. The original 2012 mouse estimate of roughly 40 nmol/L was later revised by the same group, using mass spectrometry, to about 0.3 ng/mL — a shift of more than 2,500-fold.

What the human evidence shows

The strongest evidence that human irisin exists at all is mass spectrometric. In 2015, Jedrychowski and colleagues used tandem MS with isotope-labelled standards and reported irisin in human plasma at about 3.6 ng/mL in sedentary controls, rising to about 4.3 ng/mL in people completing twelve weeks of high-intensity aerobic training — roughly a 19 percent increase, in a study of six trained and four control subjects. Crucially, their method detected a peptide that can only exist if translation initiates at that unusual ATA codon, which means the codon is at least partly functional.

That study has never been independently replicated at scale, and a third laboratory using the same approach failed to detect irisin in human plasma at all, though it did measure it in cerebrospinal fluid. A partial resolution arrived in 2024, when Witmer and colleagues reported that FNDC5 is in fact translated from a conserved upstream in-frame ATG codon, producing a larger glycosylated precursor.

Beyond existence, the human picture is thin. A meta-analysis of 66 exercise transcriptome datasets found PGC-1α reliably increased by acute exercise but FNDC5 regulated neither by acute exercise, long-term training, nor inactivity in human skeletal muscle. Browning has been replicated in rodent inguinal fat; in humans, exercise has little or no demonstrated effect on browning of abdominal subcutaneous fat, which expresses only minor UCP1 to begin with.

The bone literature is openly contradictory. One group reported that weekly recombinant irisin increased cortical bone density and lowered sclerostin in mice; Kim and colleagues found the opposite direction on sclerostin, and reported that FNDC5-knockout female mice had more trabecular bone and were protected from ovariectomy-induced loss. Work on irisin, BDNF and hippocampal function — including a 2019 Nature Medicine paper in Alzheimer’s models — is interesting but entirely rodent-based.

Two final caveats that apply across the whole field. Most cell-culture “irisin effects” used recombinant protein at 50–2,500 ng/mL, which is 10- to 8,000-fold above the mass-spectrometry estimate of what actually circulates. And the disease associations — lower irisin in type 2 diabetes, higher with obesity — come from ELISA-based meta-analyses whose leading reviewers describe them as of limited validity.

Cautions and considerations

• No irisin-based drug is approved anywhere. Irisin remains an investigational research molecule; registered human studies measure endogenous irisin as a biomarker rather than administering it.
• Recombinant irisin sold for laboratory work is not a therapeutic. Preparations differ in glycosylation state, and glycosylated and non-glycosylated forms behave differently in assays.
• Treat any specific blood level sceptically unless the method is stated. Assay identity is the single most informative detail in an irisin paper.
• Exercise benefits do not depend on this debate. The health effects of endurance training are established independently of whether irisin is their mediator.

Frequently asked questions

Is irisin real?

Its existence is now reasonably well supported: mass spectrometry has detected it in human plasma, a conserved upstream start codon has been identified, and a receptor has been proposed and functionally tested. What remains genuinely unsettled is nearly every quantitative human claim — baseline concentration, responsiveness to exercise, and association with disease.

Does exercise raise irisin in humans?

The one mass-spectrometry study that addressed this reported roughly a 19 percent rise after twelve weeks of aerobic training in a very small group. Transcriptome meta-analysis, by contrast, finds no reliable regulation of the FNDC5 gene by exercise in human muscle. The honest answer is that the effect, if real, is small and not firmly established.

Is irisin a peptide or a protein?

At 112 residues and roughly 12–20 kDa depending on glycosylation, irisin sits above the usual peptide range and is normally described as a small protein or a proteolytically released hormone fragment.

Why do irisin blood tests give such different answers?

Because most commercial kits use polyclonal antibodies that were shown to bind unrelated serum proteins. When an antibody is not specific, the number it returns reflects whatever else it caught, which is why two kits — even two lots of one kit — can disagree by orders of magnitude.

Informational only — not medical advice · 21+