Humanin Research

Humanin is a mitochondrial-derived peptide first reported in 2001 ^[1]. No study registered on ClinicalTrials.gov has ever administered Humanin to human participants ^[2]. There are no human efficacy data, no human safety data and no human pharmacokinetic data for it. The evidence on what humanin does comes from cell culture, from animals, and from measurements of the humanin that people already produce.

Identity and naming

Humanin is a 24-amino-acid peptide encoded within the MT-RNR2 region of mitochondrial DNA ^[1]. Its sequence MAPRGFSCLLLLTSEIDLPVKRRA sits in PubChem under CAS 330936-69-1 ^[3]. The record gives molecular formula C119H204N34O32S2 and molecular weight 2687.2, and that formula computes to that weight ^[3]. The CAS resolves to the same PubChem record in both directions ^[3]. Its two sulfur atoms belong to one methionine and one cysteine, so a single cysteine cannot form an internal disulfide bridge ^[3].

A 21-residue form of humanin also appears in the literature. Mitochondrial and cytosolic ribosomes read the genetic code differently, so mitochondrial translation yields a 21-mer while cytoplasmic translation yields the 24-mer ^[4]. The site of translation has not been fully identified ^[4]. Material sold as Humanin is the 24-mer.

Naming matters more here than for most peptides, because most interventional work uses an analogue rather than the native peptide. S14G-humanin, written HNG or Gly[14]-humanin, is the usual experimental agent, and one prominent aging paper describes it as a potent humanin analogue ^[5]. HNG is reported in the neuroprotection literature to be substantially more potent than native humanin. A potency-driven effect in an HNG study therefore says little about the native peptide at any realistic exposure. Colivelin, HNGF6A and HNF(14) are further distinct molecules built from that sequence. MOTS-c is a separate mitochondrial-derived peptide read from a different frame, and the small humanin-like peptides SHLP1 to SHLP6 are six more molecules again. A finding on any of those is not a finding on humanin.

No human administration evidence for humanin

Seven studies registered on ClinicalTrials.gov mention humanin, and not one of them administers it ^[2]. Each either measures endogenous humanin as a biomarker, or applies an unrelated intervention such as exercise or anesthesia and measures humanin as an outcome ^[2]. There is no first-in-human study, no human dose-ranging study and no human safety study of humanin in that registry ^[2]. Humanin is not approved for human therapeutic use, and field-qualified openFDA queries return no labeling or approval record for it ^[6]. Absence of approval is not itself a safety finding, and means only that nothing has been established to a regulator's standard ^[6].

Three papers carrying humanin in their PubMed records have been retracted, and all three concern analogues rather than the native peptide ^[7]. Two are S14G-humanin studies in Alzheimer's models and one is a colivelin study, and their author lists overlap ^[7]. The concentration of all three in one related group's Alzheimer's model work makes that vein of the analogue literature partly unreliable on its own terms ^[7].

What the animal and cell work shows

Humanin was identified by functional screening of a cDNA library, as a factor rescuing cultured neurons from death caused by familial Alzheimer's disease genes and by amyloid-beta ^[1]. That is an in vitro discovery result and nothing more ^[1].

In aged mice, humanin administration improved cognition, and the same work reported neuroprotective effects in human cell culture ^[8]. Two of its authors declared consulting or stock relationships with CohBar Inc., a company then commercializing this peptide family ^[8]. A later paper found that overexpressing humanin extended lifespan in C. elegans, and that humanin transgenic mice showed overlapping phenotypes ^[5]. Its middle-aged mice were treated with the analogue HNG rather than with humanin, and two of its authors declared consulting and stock interests in CohBar ^[5]. The only mitochondrial-derived peptide candidate CohBar took into humans was CB4211, which is not humanin, and that trial has posted no results ^[9]. CohBar stated that the CB4211 formulation used in the Phase 1b stage was not suitable for further development, and that efforts at an improved formulation had not succeeded ^[25]. The company then reverse merged into TuHURA Biosciences and left the mitochondrial peptide field ^[25].

Human studies measure the body's own humanin

Every human finding below is an observation, not the result of giving anyone humanin. Correlating a measured level with an outcome is a different claim from showing that giving the peptide changes that outcome.

In a randomized exercise study, acute endurance exercise significantly raised circulating humanin, while MOTS-c showed only a non-significant trend ^[10]. Acute resistance exercise raised neither peptide, and group sizes were 10 per arm ^[10]. At rest, humanin correlated positively with age and with MOTS-c, and neither peptide correlated with VO2max, leg strength or muscle mitochondrial DNA copy number ^[10].

Chronic training data are mixed rather than settled. Fifty-five men with impaired glucose regulation were randomized to resistance training, Nordic walking or control for 12 weeks ^[11]. Skeletal muscle humanin protein rose by 35% in the resistance arm, while serum humanin did not change in any group ^[11]. In a randomized trial of 90 women with type 2 diabetes, serum humanin rose significantly after 8 weeks of combined aerobic and resistance training ^[12]. In another randomized trial, 12 of 24 older women were assigned to 12 weeks of water-based resistance training ^[13]. In that group BDNF, IGF-1, FGF21 and GDF-15 rose significantly, and humanin was not among the significant risers ^[13]. A pilot comparison found professional athletes had higher serum humanin than sedentary controls, yet high-endurance athletes had lower humanin than low or moderate endurance athletes ^[14].

Tissue and blood do not always move together. In chronic kidney disease, circulating humanin was increased while its expression in skeletal muscle was reduced ^[15]. Serum humanin in that study also correlated positively with circulating TNF ^[15]. In women with polycystic ovary syndrome, serum humanin was significantly lower than in matched controls, with no difference in skeletal muscle humanin ^[16]. Those authors conclude the circulating reduction is unlikely to reflect the muscle compartment ^[16].

A mitochondrial variant in the humanin-coding region was associated with lower circulating humanin, and in an independent older-adult cohort with accelerated cognitive aging ^[8]. That is an association in observational human data, and it does not show that humanin drives cognition in people ^[8].

The age-direction conflict is unresolved

One line of work reports that humanin levels generally decline with age across species, and that children of centenarians carry much greater circulating levels than age-matched controls ^[5]. The largest human dataset of measured humanin levels points the other way. Across 693 subjects aged 21 to 113, plasma humanin was increased in old age, with the highest levels found in centenarians ^[17]. In that cohort it was associated with worsened handgrip strength, insulin sensitivity and triglycerides, and was inversely correlated with survival in the oldest subjects ^[17]. People with Down syndrome, a model of accelerated aging, carried higher humanin than their siblings ^[17]. Those authors read humanin as a mitochondrial stress response rather than as a benefit ^[17]. The same group confirmed the positive age correlation in a second cohort, and found humanin lower in Alzheimer's disease ^[18]. In that second study FGF21, and not humanin, emerged as the candidate marker of healthy aging ^[18]. The resting correlation in the randomized exercise study also ran positive with age ^[10].

Conflicts of interest sit asymmetrically across this disagreement. The paper reporting the age-related decline declares CohBar consulting and stock interests ^[5], and so does the companion paper from the same laboratory cited here ^[8]. The papers reporting a positive age correlation declare no conflicts ^[10][17][18]. There is no standardized humanin immunoassay, which is a plausible mechanical explanation for the disagreement. That explanation has not been established, and absolute humanin concentrations are not portable between laboratories.

Findings that point the other way

Humanin's core mechanism is suppression of apoptosis, and that property is not uniformly desirable. Systemic administration of native humanin to mice bearing triple negative breast cancer reduced tumor apoptosis, impaired the antitumor effect of chemotherapy, and accelerated tumor growth and spontaneous lung metastases ^[19]. Those authors write that caution should be taken when using exogenous humanin to treat degenerative diseases ^[19]. This is an animal and cell study, and it used the native peptide rather than an analogue ^[19].

Two glioblastoma papers converge on the same direction. Chemotherapy upregulated humanin in glioblastoma cell lines and in primary cultures from patient biopsies, an HNG-derived analogue boosted chemoresistance and cell migration, and silencing humanin reduced chemoresistance ^[20]. An independent group in another country found humanin upregulated in glioblastoma tissue relative to the corresponding normal region, and reported pro-tumoral effects in a mouse model ^[21]. Both treat humanin as a target to inhibit rather than an agent to supply ^[20][21].

More is not straightforwardly better. In 83 hemodialysis patients followed for 24 months, humanin showed a curvilinear association with cardiovascular events, with both low and high levels linked to increased risk ^[22].

The pro-longevity reading is also contested from inside the field. A 2018 commentary argues that mitochondrial-derived peptides exacerbate senescence ^[26]. That piece is a review rather than primary data, so it is a dissenting position and not a result ^[26].

Limitations

No meta-analysis or pooled quantitative synthesis of humanin has been published. Systematic reviews do exist ^[23][24]. A PRISMA review of mitochondrial-derived peptides and exercise included nine studies, and concluded that current evidence is limited, particularly in human studies ^[24].

Material sold as humanin is supplied for laboratory research use only, and nothing on this page is guidance for use in people.

Frequently Asked Questions

Has humanin been given to people in a study? No study registered on ClinicalTrials.gov has ever administered humanin to human participants ^[2]. The seven registered studies that mention humanin either measure it as a biomarker, or apply an unrelated intervention and measure humanin as an outcome ^[2].

Are findings on HNG findings on humanin? They are not. S14G-humanin, written HNG, is a different molecule, and it is the usual experimental agent in interventional humanin work. One prominent aging paper describes HNG as a potent humanin analogue and used it, not humanin, in its mice ^[5]. Colivelin, HNGF6A and HNF(14) are further distinct molecules built from that sequence.

Does humanin rise or fall with age in humans? Human observational work disagrees and the question is unresolved. The largest dataset of measured humanin levels, 693 subjects aged 21 to 113, found plasma humanin increased in old age with the highest levels in centenarians ^[17]. One line of work reports the opposite direction, that humanin generally declines with age ^[5].

References

1. 1

   Hashimoto Y, et al. Proc Natl Acad Sci U S A. 2001;98(11):6336-41. PMID 11371646. DOI 10.1073/pnas.101133498
2. 2

   ClinicalTrials.gov, seven registered studies mentioning humanin: NCT07438002, NCT05506228, NCT07678073, NCT06125249, NCT03431844, NCT04104958, NCT06105229. Records verified against the ClinicalTrials.gov v2 API on 2026-08-28. None lists humanin as an intervention.
3. 3

   PubChem Compound record CID 16131438. CAS 330936-69-1, UNII H975EUX36G, formula C119H204N34O32S2, molecular weight 2687.2. NCBI PubChem PUG-REST, retrieved 2026-08-28.
4. 4

   Gong Z, Tas E, Muzumdar R. Front Endocrinol (Lausanne). 2014;5:210. PMID 25538685. DOI 10.3389/fendo.2014.00210
5. 5

   Yen K, et al. Aging (Albany NY). 2020;12(12):11185-11199. PMID 32575074. DOI 10.18632/aging.103534
6. 6

   US Food and Drug Administration. openFDA drug/label and drug/drugsfda endpoints, queried 2026-08-28. Field-qualified substance-name queries returned NOT_FOUND for humanin.
7. 7

   Retracted publications in the humanin analogue literature: Zhang W, et al. J Pept Sci. 2013;19(3):159-165, PMID 23349038., retraction notice PMID 27282139.; Miao J, et al. Neuropeptides. 2008;42(5-6):557-567, PMID 18929410., retraction notice PMID 27033171.; Yin R, et al. Cell Physiol Biochem. 2016;38(3):1138-1146, PMID 26964005., retraction notice PMID 35771817.
8. 8

   Yen K, et al. Sci Rep. 2018;8(1):14212. PMID 30242290. DOI 10.1038/s41598-018-32616-7
9. 9

   ClinicalTrials.gov NCT03998514. Phase 1a/1b of CB4211, sponsor CohBar, Inc. Verified against the ClinicalTrials.gov v2 API on 2026-08-28: completed, completion date 2021-04-19, enrollment 88, no results posted.
10. 10

    von Walden F, et al. J Appl Physiol (1985). 2021;131(3):1035-1042. PMID 34351816. DOI 10.1152/japplphysiol.00706.2019
11. 11

    Gidlund EK, et al. Physiol Rep. 2016;4(23):e13063. PMID 27923980. DOI 10.14814/phy2.13063
12. 12

    Basereh A, et al. Sci Rep. 2025;15(1):40113. PMID 41249265. DOI 10.1038/s41598-025-23914-y
13. 13

    Hosseini MH, et al. BMC Geriatr. 2026;26(1):573. PMID 41975304. DOI 10.1186/s12877-026-07413-x
14. 14

    Alser M, et al. Rev Cardiovasc Med. 2022;23(5):181. PMID 39077591. DOI 10.31083/j.rcm2305181
15. 15

    Liu C, et al. Am J Physiol Renal Physiol. 2019;317(5):F1122-F1131. PMID 31432706. DOI 10.1152/ajprenal.00202.2019
16. 16

    Kutuk IS, et al. J Endocrinol Invest. 2026;49(3):635-642. PMID 41186863. DOI 10.1007/s40618-025-02747-6
17. 17

    Conte M, et al. J Gerontol A Biol Sci Med Sci. 2019;74(5):600-607. PMID 29955888. DOI 10.1093/gerona/gly153
18. 18

    Conte M, et al. Geroscience. 2021;43(2):985-1001. PMID 33131010. DOI 10.1007/s11357-020-00287-w
19. 19

    Moreno Ayala MA, et al. Sci Rep. 2020;10(1):8542. PMID 32444831. DOI 10.1038/s41598-020-65381-7
20. 20

    Peña Agudelo JA, et al. Cancers (Basel). 2023;15(16):4061. PMID 37627089. DOI 10.3390/cancers15164061
21. 21

    Ha CP, et al. Cell Death Dis. 2024;15(6):464. PMID 38942749. DOI 10.1038/s41419-024-06790-8
22. 22

    Bolignano D, et al. Cardiorenal Med. 2026;16(1):179-189. PMID 41849628. DOI 10.1159/000551517
23. 23

    Coradduzza D, et al. Biology (Basel). 2023;12(4):558. PMID 37106758. DOI 10.3390/biology12040558
24. 24

    Ayubi N, et al. Physiol Int. 2026, online ahead of print. PMID 42640735. DOI 10.1556/2060.2026.00863
25. 25

    CohBar, Inc. company disclosures on CB4211 and on the subsequent corporate reorganization into TuHURA Biosciences. Company statement and SEC filing, not peer-reviewed sources. The formulation statement was confirmed verbatim during source verification; no date or direct quotation is reproduced here.
26. 26

    Mendelsohn AR, Larrick JW. Rejuvenation Res. 2018;21(4):369-373. PMID 30058454. DOI 10.1089/rej.2018.2114