Humanin: the first mitochondrial-derived peptide, and what is actually known about it

Humanin was found by accident, in the best sense. In 2001 a group at Keio University in Tokyo screened a complementary DNA library made from the surviving occipital lobe of a person with Alzheimer’s disease, looking for anything that could rescue nerve cells from death caused by familial Alzheimer genes and by amyloid beta. They found a short open reading frame coding for a 24-amino-acid peptide that abolished that cell death across a wide spectrum of the genes tested, and named it humanin ^[1]. Searches for humanin Alzheimer research trace back to that screen.

The surprise came next: the same open reading frame exists in the mitochondrial genome, inside the gene for 16S ribosomal RNA. Mitochondria, the cell’s energy-producing structures, carry their own small circle of DNA, and humanin was the first peptide shown to be encoded there. That makes it the founding member of the mitochondrial-derived peptides, a family that now also includes MOTS-c, which comes from the neighbouring 12S ribosomal RNA gene, and the small humanin-like peptides or SHLPs, six of which were reported from the same region as humanin in 2016 ^[2].

Most published animal work does not use humanin as it occurs in the body. It uses S14G-humanin, usually written HNG, an analogue in which the serine at position 14 is replaced by glycine, which makes it far more potent in these assays. Read carefully: a headline saying humanin did something in mice usually means HNG did it, and a listing for HNG peptide is a listing for that analogue rather than for the natural sequence. Our guide to what research peptides are covers what the research-use label on a vial means; the compound sold online is a synthetic copy of the natural peptide, or of the HNG analogue.

It is also worth separating humanin from the other mitochondria-branded peptide on this site. SS-31 is a designed drug that binds a lipid in the inner mitochondrial membrane; humanin is a natural signalling peptide that circulates in blood and acts on cell-surface receptors. They are not variations on a theme. The humanin vs MOTS-c comparison is the more useful one, since both are members of the same family.

Two mechanisms dominate the literature. The first is inside the cell. A 2003 Nature paper reported that humanin binds Bax, the protein that punches holes in mitochondria to start apoptosis, and prevents it moving from the cytosol to mitochondrial membranes. Reducing humanin expression made cells more sensitive to Bax, and humanin peptides blocked Bax binding to isolated mitochondria and suppressed cytochrome c release ^[3]. The same year, another group reported that humanin interacts with insulin-like growth factor binding protein 3 to regulate cell survival and apoptosis ^[4].

The second is outside the cell. Humanin is secreted, and a series of studies identified a cell-surface receptor complex involving CNTF receptor alpha, WSX-1 and gp130 through which humanin blocks nerve cell death ^[5]. Downstream it activates the ERK1/2, AKT and STAT3 pathways, with the pattern of signalling differing by age in the hippocampus ^[6]. The STAT3 arm matters for the metabolic work described below.

A third strand connects humanin to the growth hormone and IGF-1 axis in both directions: IGF-1 regulates humanin levels in an age-dependent way ^[7], and humanin and IGF-1 signalling intersect at several points ^[8]. This is one reason humanin keeps appearing in longevity research rather than in a single disease area.

Claims about humanin peptide benefits rest on three kinds of evidence, and it helps to keep them apart: observational measurements in people, cell experiments, and animal treatment studies using HNG.

Metabolism and insulin: a 2009 study found that infusing humanin into the brain ventricles of rodents significantly improved whole-body insulin sensitivity, an effect that depended on hypothalamic STAT-3 signalling, and that intravenous infusion of potent humanin derivatives reproduced it. A single treatment with a potent analogue lowered blood glucose in Zucker diabetic fatty rats. The same paper reported that detectable humanin fell with age in rodent hypothalamus, skeletal muscle and cortex, and that circulating humanin fell with age in both mice and humans ^[9].

Brain and cognition: a 2018 study reported that humanin administration was neuroprotective in human cell models and improved cognition in aged mice. In a human cohort, a variant in the humanin-coding region of the mitochondrial genome was associated with lower circulating humanin, and in a separate nationally representative cohort of older adults the same variant was associated with accelerated cognitive ageing ^[10]. That is a genetic association, not a demonstration that injecting humanin protects the brain.

Lifespan and healthspan: a 2020 study from the same USC group reported that humanin regulates lifespan and healthspan across species and models ^[11], and a 2022 study found that humanin-induced autophagy played a role in skeletal muscle function and lifespan extension ^[12].

Heart and blood vessels: humanin treatment reduced myocardial ischaemia-reperfusion injury in mice ^[13], preserved endothelial function and slowed atherosclerotic plaque progression in ApoE-deficient mice ^[14], and chronic treatment prevented age-related myocardial fibrosis in mice ^[15]. In people, a Mayo Clinic study of 40 patients undergoing coronary angiography found that those with coronary endothelial dysfunction had significantly lower plasma humanin, 1.3 against 2.2 ng/mL, with a positive correlation between coronary blood flow response and humanin levels ^[16]. That is an association measured at a single time point.

Muscle: a 2026 study exposed human skeletal muscle cells to dexamethasone, which causes wasting, and found that both HNG and MOTS-c preserved myotube size, with HNG blunting the stress signalling the steroid switched on ^[17]. This is a cell study in human cells, not a study in humans.

No published trial has given humanin, or HNG, to people. This is the single most important fact on this page and it is routinely obscured by supplier copy citing the animal literature as though it were clinical.

What does exist is a large observational literature measuring circulating humanin and relating it to health. Levels fall with age in humans ^[9]; lower levels accompany coronary endothelial dysfunction ^[16]; a mitochondrial variant associated with lower humanin tracks faster cognitive ageing ^[10]. Humanin has been measured in carotid atherosclerotic plaques ^[18] and studied as a biomarker in a range of conditions. These studies can show that humanin is involved in something. They cannot show that adding more of it helps, and the direction of causation is often unclear: a body under metabolic stress may release more of a protective peptide, which would make high levels a sign of trouble rather than of health.

Exercise is the nearest thing to an intervention that has been studied. In diabetic rats, eight weeks of high-intensity interval training raised serum humanin alongside reductions in oxidative stress and inflammation, and the authors suggested the benefits of the training might be partly mediated by humanin ^[19]. That is a rat study, and it tested exercise, not an injection. The pattern mirrors what is seen with MOTS-c, where exercise raises the body’s own production.

A company built on mitochondrial-derived peptide research, CohBar, did complete a phase 1a/1b study of a peptide analogue, CB4211, in healthy non-obese subjects and people with non-alcoholic fatty liver disease, with 88 participants, finishing in 2021 ^[20]. The registry record does not state which peptide it derives from, and no peer-reviewed report of that trial appeared in our searches, so it cannot be cited as evidence about humanin.

There is no human answer. In rodents, a single dose of a potent humanin analogue lowered blood glucose acutely in diabetic rats, while the insulin-sensitivity work used continuous infusion during a clamp study ^[9]. The cardiac and cognitive benefits reported in mice came from chronic treatment over weeks to months ^{[10] [15]}.

No pharmacokinetic study of humanin in people has been published, so nothing is known about how long an injected dose lasts, how much reaches the brain, or whether the natural peptide survives long enough in blood to do anything at all. That last question is why nearly every animal study uses the HNG analogue instead.

These figures describe animal and cell experiments and are not recommendations; no humanin dosage has been tested in a person. Rodent metabolic work used intracerebroventricular infusion directly into the brain ventricles and intravenous infusion of analogues during clamp studies ^[9], which are surgical research techniques with no home equivalent. Cardiac and cognitive studies in mice used intraperitoneal injection, typically daily or several times weekly over weeks ^{[13] [15]}. Cell studies of human muscle used 10 µM HNG alongside 10 µM MOTS-c ^[17].

Two points follow. First, almost all published dosing is of S14G-humanin rather than humanin, so a protocol built from those papers is a protocol for a different molecule. Second, much of the rodent work delivered the peptide into the brain, and no published study establishes that a subcutaneous injection reproduces any of it. If you are converting a vial’s stated contents into a concentration, our peptide calculator and the humanin calculator page will do it, but the number you put in is the guess, not the arithmetic.

Research suppliers sell humanin and HNG as lyophilised powders for reconstitution and subcutaneous injection. Published animal routes were intraperitoneal, intravenous and intracerebroventricular ^{[9] [13] [15]}. We found no published oral, nasal or topical humanin study in any species, and no human pharmacokinetic data by any route. A 24-amino-acid peptide with two cysteine residues also raises the practical question of whether the material in a vial is correctly folded, which a simple mass measurement will not answer.

No human safety data exist, because no one has given it to a person in a published study. Anything presenting a humanin side effect profile is not describing published research.

The mechanism deserves thought rather than reassurance. Humanin’s central action is to block apoptosis by preventing Bax from reaching mitochondria ^[3]. Apoptosis is how the body removes cells it should not keep, including damaged and pre-cancerous ones. A systemic anti-apoptotic signal is exactly the kind of thing that could, in principle, allow such cells to survive, and nobody has measured cancer incidence in anything given humanin long term. Humanin has been found expressed in carotid plaques ^[18] and, in observational work, has been reported as raised rather than lowered in several disease states, which complicates the simple picture of more being better.

There is also no evidence for the uses humanin is marketed for. We found no study of humanin and body composition, muscle gain in people, fat loss in people, or athletic performance. The human muscle work is a dish experiment about preventing steroid-induced wasting ^[17], not a muscle-building result. Nothing is known about use in pregnancy, effects on fertility, or interactions with medicines.

Humanin is not an approved medicine in any market we cover, and there is no registered clinical trial of the peptide as a treatment. It is sold as a research chemical. What that means for buyers differs by country: see our legal status overview and the pages for the UK, the US and Europe. The direction of travel is described in our report on the FDA advisory committee meeting on compounded peptides.

We have not confirmed humanin’s status on the WADA Prohibited List. Note that its family member MOTS-c is prohibited at all times as an AMPK activator, so anyone subject to testing should check the current list themselves rather than assuming a related peptide is treated the same way.

Humanin is supplied as a freeze-dried powder. Peptides are generally kept cold, dry and away from light, and reconstituted solutions refrigerated and used within a limited period. See how to store peptides and bacteriostatic water. Peptides containing cysteine can oxidise and form unwanted disulfide links in solution, which is a reason to be particularly careful about how long a reconstituted vial is kept.

The first thing to check on a humanin listing is which peptide it actually is. Humanin and S14G-humanin are different molecules with different masses, and a store that uses the names interchangeably has told you something about its record-keeping. A certificate of analysis for natural-sequence humanin should show a mass close to 2687 g/mol.

Compare humanin prices across the suppliers we track, including listings in the UK, the US and Canada. Our guides to reading a peptide COA, third-party peptide testing and how to spot a fake peptide supplier explain what to look for, the stores publishing independent results are listed under third-party tested suppliers, and peptide prices explained covers why the same vial can cost three times as much from one store as another.