When we covered MOTS-c in June, the honest summary was that its mechanism ran through AMPK, its target was unidentified, and its human file was empty. Two of those statements have moved since. The third has not — but a search engine will now tell you otherwise, because a fabricated registry entry has quietly become the most-cited "human trial" in the mitochondrial-peptide space. This roundup separates what actually published in the mitochondrial-derived peptide (MDP) field over the last year from what is circulating as if it had.
What an MDP is — and what the label doesn't tell you
Mitochondrial-derived peptides are defined by where their genes live, not by how they work. They are translated from short open reading frames inside mitochondrial DNA — humanin from a sORF in the 16S rRNA gene, MOTS-c from a 51-bp sORF in the 12S rRNA gene, and the SHLP family (small humanin-like peptides) from the humanin region.
That shared genomic address has encouraged a shared shelf label: "mitokines," peptides mitochondria use to signal the nucleus and distant tissues. The label is useful for the biology and misleading as a pharmacological grouping — and the 2025–2026 literature makes that clearer than ever.
Note also the distinction from SS-31, often filed alongside these compounds: SS-31 is a synthetic peptide targeted to mitochondria, while MDPs are endogenous peptides encoded by them. Same organelle, opposite direction of the arrow.
MOTS-c finally has a named direct target — and it isn't a receptor
The most substantive mechanistic advance came from Kumagai and colleagues in iScience (2024;27(11):111212, PMID 39559755), reporting that MOTS-c directly binds and activates protein kinase CK2.
The details matter for how the claim should be read. Binding was shown by two independent methods — dot blot and surface plasmon resonance — with low-nanomolar reported affinity, and MOTS-c bound the CK2α catalytic subunit specifically, not CK2β. The paper positions CK2 as "a direct and functional binding partner." It does not claim CK2 is MOTS-c's only target, and it does not claim a cell-surface receptor.
That last point is the one to hold onto. As covered in our receptor pharmacology primer, the vocabulary of affinity, efficacy, selectivity and bias belongs to receptor ligands. MOTS-c instead looks like an intracellular protein-interaction partner — closer in architecture to FOXO4-DRI disrupting a protein-protein interface, or thymosin β4 buffering G-actin, than to anything on the growth-hormone or incretin shelves.
The genetics arm is the more interesting half. A naturally occurring mtDNA variant, m.1382A>C, produces K14Q MOTS-c — a single lysine-to-glutamine substitution. The variant bound CK2α roughly 16-fold more weakly and did not activate the kinase in vitro. In cohort data, C-allele carriers showed higher reported prevalence of low muscle mass (18.6% vs 11.3%; adjusted odds ratio 1.75).
Elegant work — a natural loss-of-binding variant whose carriers differ measurably. But it is association data in carriers of an inherited variant: evidence that the MOTS-c/CK2 axis matters in human physiology, not evidence that administering MOTS-c does anything.
Humanin runs on entirely different hardware
Set the MOTS-c picture beside humanin and the "class" dissolves. Humanin has characterized cell-surface receptors: a trimeric complex of CNTFR-α, gp130 and WSX-1 coupling to JAK2/STAT3, plus the G-protein-coupled formyl peptide receptor family member FPRL1 coupling to MAPK. It also has documented intracellular interactions — with BAX, Bim, tBid and IGFBP3 — placing it in the apoptosis-regulation business.
So within one so-called class: one member signals through a defined cytokine-receptor complex and a GPCR; the other has no established receptor at all and appears to act on an intracellular kinase. They share a genome, not a mechanism. Any generalisation beginning "mitochondrial peptides work by…" is describing one member and borrowing credibility for the rest.
One naming trap, in the spirit of our nomenclature primer: much of the humanin literature — including the study below — uses S14G-humanin (HNG), an analog carrying a serine-to-glycine substitution at position 14, not native humanin. Different molecule, different mass (roughly 30 Da lighter), and its results are not automatically humanin's.
Two 2026 muscle papers, both admirably unflattering
Human myotubes. Elhusseiny et al., Physiological Reports 2026;14(4):e70791 (24 February 2026; PMID 41732124) tested both peptides against dexamethasone-induced atrophy in primary human skeletal muscle myotubes — a step up from the C2C12 mouse-line work that dominates this literature.
Dexamethasone reduced myotube area and fusion index and raised the atrophy-associated E3 ligases MURF1 and MAFbx. MOTS-c co-treatment fully preserved both area and fusion index, raised Akt phosphorylation, and blunted MURF1 and STAT3 activation. HNG preserved area but did not restore fusion index or affect E3 ligase expression.
The authors' own limitations section is the most valuable paragraph in the paper: n = 3, with a formal power analysis reporting several endpoints badly underpowered (fusion index at 41.1%, MAFbx at 36.8%) and an explicit caution that those carry substantial type II error risk. That is what rung-1 evidence looks like when reported honestly — see our evidence hierarchy piece for why a positive result in a dish establishes that an interaction is possible, not that it is useful.
Tumour-bearing mice. Jamnick, Bonetto and colleagues, Frontiers in Medicine 2026;13:1838178 (25 May 2026), asked whether MOTS-c protects muscle in the Colon-26 cachexia model. The title gives the finding away: "partially."
MOTS-c was reported to partly rescue quadriceps mass, trend toward protecting gastrocnemius, preserve extensor digitorum longus contractile function, reduce atrogin-1 and MuRF1, and raise PGC-1α. It did not prevent total body weight loss, did not prevent adipose loss, did not protect tibialis anterior mass, and did not fully restore Akt or suppress STAT3 — the same signalling nodes it controlled cleanly in the dish. The authors also note treatment began at tumour inoculation rather than after cachexia was established: a prevention design, not a rescue design. A paper that names its own incompleteness in the title is worth more than three that don't.
The human file: measured, not administered
Here is the part that vendor copy gets backwards. Search ClinicalTrials.gov for MOTS-c or humanin and you do get hits — but read them and a consistent pattern appears: these are studies that measure MDPs as biomarkers, not studies that give them to anyone.
- NCT03431844 (University of Tartu, completed) — observational; humanin isoforms in cardiac muscle and plasma against post-cardiac-surgery complications.
- NCT04027712 (University of Athens) — observational; MOTS-c among markers tracked against mortality in type 2 diabetics with coronary artery disease.
- NCT07638696 (Bahçeşehir Cyprus University) — observational; mitophagy and mtDNA dynamics across Ramadan dry fasting versus time-restricted feeding.
- NCT04013568 (University of Hawaii, completed) — interventional, but the intervention is exercise; MDPs are read out, not administered.
Now the exception. A single registry record describes MOTS-c as an administered investigational agent: NCT07505745, "A Phase 2a, Randomized, Double-blind, Placebo-controlled Study… in Adults With Prediabetes and Overweight/Obesity," study ID EX-MOTS-2A-001, listed as recruiting since 1 April 2026.
Its sponsor is "Hudson Biotech." Its single site is Peking University Shenzhen Hospital. Its central contact address is at beijing-biotech.com. It carries isFdaRegulatedDrug: false.
Readers of our trial registry literacy piece from 14 August will recognise every one of those fingerprints. That record belongs to a batch of registrations sharing one sponsor name, one site, and one contact domain — a batch in which three entries openly declare themselves fictional or mock examples in their own brief summaries, and two others reproduce Eli Lilly trial titles and internal study-ID formats. NCT07505745 is one of the quiet ones that does not self-declare, and those are the entries that get cited.
And it is being cited. Multiple vendor and aggregator pages published in 2026 now reference "the 2026 MOTS-c Phase 2a trial" as evidence the compound is advancing toward human validation. It is not a result, it is not evidence, and on the face of the record it is not a trial.
The accurate statement remains: there is no verifiable completed or ongoing interventional human trial in which MOTS-c or humanin was administered. The peptides have been measured in people extensively. That is a different sentence.
The geroprotector framing, and its weak joint
Sahu et al., International Journal of Peptide Research and Therapeutics 2026;32(1):14 (online 26 December 2025), position MDPs as an emerging class of geroprotectors, mapping humanin, MOTS-c and the SHLPs onto AMPK, mTOR and sirtuin signalling, and noting that circulating MDP levels decline with age.
That declining-with-age observation is the load-bearing claim in most MDP marketing copy, and it deserves the scrutiny we applied to telomerase framing in the longevity class primer. A biomarker that falls with age may be a cause of decline, a consequence of it, or a marker of the mitochondrial mass and turnover that also falls. Correlation with age does not establish a deficiency, and a deficiency framing does not establish that replacement helps.
The documentation angle
Two points specific to this shelf.
MOTS-c is one of the few catalog compounds where UV quantitation actually works. Its 16-residue sequence carries both tryptophan and tyrosine, so A280 is viable — unlike BPC-157, Epithalon, KPV or Selank, where it is noise (see net peptide content). It also carries two methionines, making oxidation a named stability liability.
And the K14Q finding creates an analytical problem worth naming. Lysine and glutamine residues differ in mass by roughly 0.04 Da — so wild-type and K14Q MOTS-c are, for practical purposes, indistinguishable on the unit-resolution intact-mass spectrum that accompanies a typical COA. This joins the list on our chromatogram and mass spectrum piece of real sequence differences a matching mass cannot exclude: epimers at 0 Da, scrambled disulfides at 0 Da, and now a functionally consequential substitution at 0.04 Da. Resolving it takes MS/MS sequencing, not an intact-mass line. More on what analytical documentation can and cannot settle on our quality page.
FAQ
Did MOTS-c's mechanism change? It gained a layer. AMPK activation remains the best-described downstream consequence; CK2 is now a reported direct binding partner upstream of it — a more specific claim than "acts via AMPK." It is one paper, and the authors do not present CK2 as an exclusive target.
Why isn't humanin in the catalog if it's the best-characterised MDP? That's a stocking question, not a scientific one. It appears here because MOTS-c cannot be read in isolation — the humanin comparison is what shows that "mitochondrial-derived peptide" describes a genomic origin rather than a shared mechanism.
Is the 2026 muscle work a reason to expect human results? No. Both papers are what they say: a small, self-declared underpowered human-cell study and a partial protection in one mouse cachexia model. Useful mechanistic evidence, sitting on rungs 1 and 2. Nothing published this year moves MDPs onto the human rung.
This article is educational and for the laboratory research community. Trulogic Labs products are sold for laboratory and research use only and are not for human consumption.