The short answer to the MOTS-c question
Mitochondrial-derived peptides are among the more genuinely interesting findings in recent cell biology, and MOTS-c has never been given to a human being in a controlled trial. Both statements are true at once, and holding them together is the entire exercise. The mouse metabolic data is real. The exercise physiology is real. The population genetics are real. The column in the evidence table headed controlled human efficacy has been empty for a decade.
Elamipretide is the contrast case. It is not itself a mitochondrial-derived peptide, but it is a mitochondria-targeted peptide drug developed the long way: a molecular target, animal work, phase 2 in a heart attack population, phase 3 in a muscle disease, phase 2 in an eye disease, and a randomised crossover trial with an open-label extension in an ultra-rare genetic disorder. Most of those studies missed their primary endpoint, and the programme produced a single narrow approval. That is what a real development path costs, and what it looks like when the answer is mostly no.
MOTS-c, humanin and epitalon have not walked that path. What follows is what each one actually has behind it, in what species, measured against what control.
What a mitochondrial-derived peptide actually is
The mitochondrial genome is small and famously crowded: roughly sixteen and a half thousand base pairs encoding thirteen proteins of the respiratory chain, twenty-two transfer RNAs and two ribosomal RNAs. That list was long taken to be complete. Mitochondrial-derived peptides are the claim that it is not, because short open reading frames sit inside the ribosomal RNA genes themselves and encode small bioactive peptides.
Humanin is a twenty-four amino acid peptide read from an open reading frame within the 16S ribosomal RNA gene. MOTS-c, for mitochondrial open reading frame of the 12S ribosomal RNA type-c, is sixteen amino acids read from the 12S gene. A further family, the small humanin-like peptides, was subsequently described in the same 16S region. If the interpretation holds, the mitochondrion is not only a metabolic organelle taking orders from the nucleus; it also writes its own peptide messages and sends them outward.
That is a large claim, and it carries an unresolved technical problem. Mitochondria use a slightly different genetic code from the cytosol, so where an open reading frame is translated changes what peptide it produces. Humanin has two candidate reading frames of different lengths depending on whether translation happens inside the organelle or in the cytoplasm. The field has not settled this, which matters because it determines what molecule researchers should be measuring.
MOTS-c: what has actually been measured, and in what
MOTS-c was described in 2015 in Cell Metabolism. The proposed mechanism is specific rather than vague: it interferes with the folate-methionine cycle, which causes the intermediate AICAR to accumulate, which activates AMP-activated protein kinase. AMPK is the cell's low-energy sensor, and switching it on shifts metabolism toward glucose uptake and fatty acid oxidation. In mice, MOTS-c administration prevented diet-induced obesity and improved insulin sensitivity on a high-fat diet.
A 2021 report extended this into exercise biology. Plasma MOTS-c rose acutely with exercise in humans and mice, the peptide was shown to translocate into the nucleus under metabolic stress and associate with stress-responsive gene regulatory regions, and injections in aged mice improved physical performance. This is where the exercise-mimetic framing comes from, and it is a legitimate reading of a mouse experiment.
The human evidence is of two kinds and neither is interventional. The first is correlational: circulating MOTS-c rises with acute exercise and tracks with metabolic phenotypes in cross-sectional cohorts. The second is genetic. A mitochondrial variant, m.1382A>C, which changes the fourteenth residue of MOTS-c, has been associated with exceptional longevity and metabolic outcomes in Japanese cohorts. That is real evidence about the peptide's biological importance. It is not evidence that injecting it does anything.
What does not exist is the part that would matter for a decision. There is no randomised controlled trial of MOTS-c administration in humans, no published human pharmacokinetic profile, no dose-response curve, no safety database, no adverse event denominator and no defined clinical endpoint anyone has attempted to move. Circulating protocols are constructed, not derived.
Humanin: twenty-five years old and still preclinical
Humanin has an unusually good origin story. It was identified in 2001 from a complementary DNA library made from the occipital lobe of an Alzheimer disease brain, a region comparatively spared by the disease, in a screen for factors that protected neurons from amyloid-beta toxicity. It did protect them, which is why the name stuck.
The mechanism has been worked out in more molecular detail than most research peptides ever receive. Extracellular humanin signals through a trimeric receptor complex involving CNTFR, WSX-1 and gp130 and activates STAT3. Intracellularly it binds the pro-apoptotic protein Bax and prevents its translocation to mitochondria, blocking the commitment step of programmed cell death. A single substitution at position fourteen produces the analogue HNG, far more potent in cell assays and the version used in most animal work.
The human data looks like the MOTS-c human data. Circulating humanin declines with age, is higher in the offspring of centenarians, and varies with disease states. Every one of those is an association measured in blood, not an effect measured after administration, and twenty-five years after discovery there is still no interventional human trial. There is also a mechanistic concern that ages badly in an anti-ageing framing: sustained systemic inhibition of apoptosis is the survival advantage damaged and pre-malignant cells exploit, and nothing in the rodent literature runs long enough to speak to it.
Elamipretide: what a real development path looks like
Elamipretide, also called SS-31 and formerly Bendavia, is a four amino acid aromatic-cationic peptide from the Szeto-Schiller series. It concentrates in the inner mitochondrial membrane and binds cardiolipin, that membrane's signature phospholipid, stabilising cristae architecture and the electron transport chain supercomplexes within them. The result is more efficient ATP production and less electron leak. Unlike most claims made under the heading of mitochondrial support, this is a demonstrable interaction with a named binding partner.
The clinical programme began in cardiology. EMBRACE STEMI tested an infusion during percutaneous coronary intervention in first-time anterior ST-elevation myocardial infarction, on the reasoning that reperfusion injury is a mitochondrial event. It did not reduce infarct size on its enzymatic primary measure, and later work in heart failure established no efficacy signal either.
MMPOWER-3 was the phase 3 trial in primary mitochondrial myopathy: forty milligrams daily by subcutaneous injection over twenty-four weeks, with co-primary endpoints of six-minute walk distance and a fatigue score from a disease-specific symptom instrument. It missed both. A phase 2 trial in dry age-related macular degeneration with geographic atrophy also failed to meet its primary endpoint, though it reported signals on retinal structural measures.
The one that eventually worked was Barth syndrome. TAZPOWER, a randomised crossover trial, did not meet its co-primary endpoints over its twelve-week randomised phase, but its open-label extension, run over a far longer period, reported progressive gains in walking distance, muscle strength and cardiac measures. On that basis, together with the extreme rarity of the disease and the absence of alternatives, the peptide was approved in the United States in 2025 under the brand name Forzinity. One approval, in one ultra-rare disorder, out of a programme spanning four disease areas.
Why the failures are the most useful part of the record
Barth syndrome is caused by mutations in TAFAZZIN, and the resulting defect is specifically a failure of cardiolipin remodelling. Elamipretide binds cardiolipin. The approval landed exactly where the drug's mechanism matched the molecular lesion. That coincidence is not a footnote; it is the central lesson of the whole programme.
Because where the mechanism was plausible but the lesion lay elsewhere, the same molecule did nothing measurable. Primary mitochondrial myopathy is a heterogeneous group of disorders in which mitochondrial function fails for many different reasons, and a phase 3 trial found no benefit. Geographic atrophy involves mitochondrial dysfunction in retinal pigment epithelium, and a phase 2 trial missed its primary endpoint. Reperfusion injury is a textbook mitochondrial event, and the STEMI trial found no reduction in infarct size.
That is the strongest available empirical argument against generic mitochondrial support as a therapeutic category: the best-characterised mitochondrial peptide in existence worked in one specific cardiolipin disease and failed in every broader indication it was tried in. Two honest caveats run the other way. An open-label extension without a control arm is vulnerable to practice effects and regression to the mean, and injection-site reactions were frequent enough across the programme to count as a real tolerability finding. Neither fact would have emerged from preclinical work.
The one MOTS-c molecule that reached a clinical trial
There is a partial exception to the claim that mitochondrial-derived peptides have never entered the clinic. CB4211, developed by CohBar, was an analogue of MOTS-c that completed an early-phase study in obesity and non-alcoholic fatty liver disease. The company reported topline results describing modest effects on body weight and on liver fat in a small population, alongside tolerability issues at the injection site.
What happened next is the informative part. The programme was not carried forward into the larger controlled trials that would have been needed to establish whether those signals were real, and the company subsequently wound down its peptide work. An early-phase signal nobody pursues is weak evidence, and a discontinued programme is data about how the people closest to the molecule read their own results.
It also needs saying that CB4211 was an engineered analogue, not MOTS-c. Analogues are made because the native peptide has properties, usually pharmacokinetic, that make it unsuitable as a drug. Nothing in that study tells a reader what native MOTS-c does in a person.
Epitalon and the second way an evidence base fails
Epitalon sits in the same longevity peptide bucket commercially, and fails a different test. Its problem is not the absence of human data. Human reports exist, including multi-year follow-up studies in elderly cohorts claiming reduced mortality, produced by a St Petersburg gerontology group over decades of work on the pineal extract epithalamin and its synthetic tetrapeptide successor.
The problem is that nearly all of it traces to that one lineage. There is no independent replication by unaffiliated groups, no registered trial run to contemporary standards, no prospective protocol published in advance, and no regulatory review by the FDA, EMA or MHRA. The central mechanistic claim, induction of telomerase, also carries an unstudied hazard: telomerase reactivation is a hallmark of most cancers, and a compound designed to switch it on systemically has never been evaluated for that risk in a controlled setting.
The two failure modes are distinct. MOTS-c and humanin have too little human data to evaluate; epitalon has human data that cannot be checked. Being sold alongside each other under one marketing heading is a commercial fact, not a scientific relationship.
The measurement problem sitting underneath all of it
Most of the human mitochondrial-derived peptide literature consists of measuring how much of the peptide is in someone's blood and correlating that with age, fitness or disease. That entire body of work rests on assay quality, which for these peptides is not firmly established. Reported concentrations typically come from immunoassays whose specificity has not been extensively validated against mass spectrometry, and the peptides are short and present at low abundance, close to the worst case for antibody-based detection.
The consequence is uncomfortable. If an assay is partly recognising something other than the intended peptide, an association between the reported value and metabolic health may be an association with whatever else the antibody binds. This does not invalidate the field, but a human correlation quoted as evidence is doing less work than a reader might assume.
A second complication affects genetic claims. Segments of mitochondrial DNA have been copied into the nuclear genome over evolutionary time, and these nuclear mitochondrial sequences can confound analyses that assume a read came from the organelle. Careful studies control for this. Sales pages do not.
How to read a MOTS-c product page
Almost every effect claimed for MOTS-c in consumer-facing writing is a mouse effect. Reversal of diet-induced obesity, improved insulin sensitivity, restored running capacity in aged animals: those are real published results, and the species column is what gets dropped in translation. The same applies to humanin's neuroprotection claims, which come from rodent models of Alzheimer disease and stroke.
The exercise argument deserves particular scepticism because it inverts the direction of the evidence. MOTS-c rises when people exercise, which makes it a candidate mediator of adaptation and equally a plausible marker of it. It does not follow that administering the peptide reproduces exercise, any more than the rise in lactate during a hard interval means lactate infusions are training. Establishing mediation requires an intervention study, which is exactly what has not been done.
The regulatory position is unambiguous. MOTS-c, humanin and epitalon are unapproved in the United States, the European Union and the United Kingdom, and material sold under those names is labelled research-use-only, which describes what a product may lawfully be marketed for rather than certifying its quality. For competitive athletes, any pharmacological substance without approval from a government health authority falls under the World Anti-Doping Agency's non-approved substances category and is prohibited at all times. Elamipretide is the only compound here with an approval anywhere, and it covers one ultra-rare genetic disease.