Humanin Peptide Research: Mitochondrial-Derived Cytoprotection, STAT3/BAX Signaling, and Preclinical Neuroprotective Evidence

Humanin peptide research centres on a 24-amino acid mitochondrial-derived peptide (MDP) that functions as an endogenous cytoprotective signal, operating through both intracellular apoptosis suppression and extracellular receptor-mediated survival pathways. It was identified in 2001 by functional expression screening of a cDNA library from the occipital cortex of an Alzheimer’s disease brain, and it has since become the most extensively characterised MDP in preclinical neuroscience, cardiology, and aging research. Its two mechanisms, binding the pro-apoptotic protein BAX inside the cell while activating STAT3 signalling through a gp130-containing receptor complex at the cell surface, distinguish it from neuroprotective peptides that work through a single pathway. Related cytoprotective candidates such as ARA-290 (cibinetide) act through different receptor systems but overlap with humanin in their preclinical anti-inflammatory and neuroprotective profiles.

Disclaimer: For research purposes only. Not for human consumption. Not for diagnostic or therapeutic use. The information presented here is drawn from published scientific literature and is intended solely for educational reference. Humanin is not a compound in the Maple Research Labs catalogue.

Discovery and Molecular Identity

Humanin was discovered by Hashimoto and colleagues, who screened a cDNA library for genes that could abolish neuronal death caused by familial Alzheimer’s disease (FAD) genes. The gene they recovered encodes a short polypeptide that suppressed death induced by the V642I and K595N/M596L (NL) mutants of the amyloid precursor protein, by M146L presenilin 1 and N141I presenilin 2, and by amyloid beta 1-43, while having no effect on death caused by a polyglutamine repeat (Q79) or by mutant superoxide dismutase 1. Transfected humanin cDNA was transcribed, translated, and secreted into the culture medium, and the rescue activity depended on the peptide’s primary structure (Hashimoto et al., 2001, PNAS 98:6336-6341; PMID 11371646). The open reading frame sits within the mitochondrial 16S ribosomal RNA gene, and a follow-up paper in the same year extended the protective spectrum to further FAD presenilin mutants, to APP stimulated by anti-APP antibody, and to amyloid beta 1-42 and 25-35, while confirming that humanin does not protect against glutamate or against a prion fragment (Hashimoto et al., 2001, J Neurosci 21:9235-9245; PMID 11717357).

The primary sequence is MAPRGFSCLLLLTSEIDLPVKRRA, 24 residues with a molecular weight of about 2687 Da for the free acid. Structure-activity work in the J Neurosci paper mapped the active region to Pro3 through Pro19, identified seven residues within it as essential, and showed that Cys8 and Ser14 are both required for rescue activity. Humanin was the first member of what is now a family of mitochondrial-derived peptides that also includes MOTS-c and the small humanin-like peptides SHLP1 to SHLP6 (Lee, Yen and Cohen, 2013, Trends Endocrinol Metab 24:222-228; PMID 23402768). Because the peptide is encoded in the mitochondrial genome, its expression is tied to mitochondrial copy number and stress state, which is the basis for reading circulating humanin as a retrograde signal from the organelle to the rest of the organism.

Receptor Pharmacology and Extracellular Signalling

The receptor question took most of a decade to settle. Hashimoto and colleagues first showed that neuroprotection in F11 neurohybrid cells depends on STAT3 and on tyrosine kinase activity, and that knocking down FPR2, the mouse counterpart of the formyl peptide receptor-like 1 molecule proposed by another group, did not blunt the rescue effect, which pointed toward a cytokine-type receptor (Hashimoto et al., 2005, Life Sci 77:3092-3104; PMID 16005025). The 2009 follow-up identified that receptor: loss-of-function experiments showed gp130, the shared subunit of the IL-6 receptor family, to be essential, overexpression of ciliary neurotrophic factor receptor alpha (CNTFR) or of the IL-27 receptor subunit WSX-1 increased humanin binding to neuronal cells while siRNA knockdown reduced it, and reconstituted binding assays showed that humanin, but not a control peptide, induces hetero-oligomerisation of CNTFR, WSX-1 and gp130 (Hashimoto et al., 2009, Mol Biol Cell 20:2864-2873; PMID 19386761).

Downstream of gp130, humanin activates more than STAT3. Kim and colleagues showed in 2016 that humanin acting through the GP130/IL6ST complex increases phosphorylation of AKT, ERK1/2 and STAT3, with PI3K, MEK and JAK respectively mediating the three arms. The in vivo part of that study is often misreported: it was old mice, not young mice, in which injected humanin increased AKT and ERK1/2 phosphorylation in the hippocampus, an age-specific response in a brain region central to memory (Kim et al., 2016, Oncotarget 7:46899-46912; PMID 27384491).

The formyl peptide receptor story is real but belongs to a different cell type. Ying and colleagues found that synthetic humanin induces chemotaxis of mononuclear phagocytes through human FPRL1 and its murine counterpart FPR2, the same receptors amyloid beta 42 uses to attract and activate phagocytes, and that humanin reduced amyloid beta 42 aggregation and fibril formation in that setting (Ying et al., 2004, J Immunol 172:7078-7085; PMID 15153530). Read together with the 2005 knockdown result, the current picture is that FPRL1/FPR2 mediates humanin’s effects on immune cells while the CNTFR/WSX-1/gp130 complex mediates neuronal rescue.

Intracellular Anti-Apoptotic Mechanisms

Humanin’s intracellular activity is independent of any receptor. Guo and colleagues showed that humanin binds BAX and prevents its translocation from the cytosol to mitochondria, that reducing endogenous humanin by siRNA sensitises cells to BAX and increases BAX translocation to membranes, and that humanin peptides block BAX association with isolated mitochondria and suppress cytochrome c release in vitro. They also noted that the mitochondrial genome carries an identical open reading frame and that the mitochondrial version of the peptide binds and suppresses BAX as well, which led them to propose that the gene arose in mitochondria as a defence against BAX (Guo et al., 2003, Nature 423:456-461; PMID 12732850).

A second intracellular partner is insulin-like growth factor binding protein 3 (IGFBP-3). Ikonen and colleagues cloned humanin as an IGFBP-3 binding partner in a yeast two-hybrid screen, confirmed the interaction by pull-down, ligand blot and co-immunoprecipitation from mouse testis, mapped the binding site to the 18-residue heparin-binding domain of IGFBP-3, and showed that the F6A humanin mutant does not bind. Humanin blocked IGFBP-3-induced apoptosis in A172 glioblastoma cells but not in SH-SY5Y cells or primary cortical neurons, and in primary neurons IGFBP-3 actually potentiated humanin’s rescue of amyloid-beta-treated cells, so the interaction is context dependent rather than a simple neutralisation (Ikonen et al., 2003, PNAS 100:13042-13047; PMID 14561895).

The S14G Analogue (HNG) and Later Derivatives

Native humanin is active in the micromolar range in most cell assays, which is inconvenient for in vivo work. Substituting glycine for serine at position 14 yields [Gly14]-humanin, written HNG or S14G-HN, which is roughly 1,000-fold more potent in vitro, the figure Tajima and colleagues use when describing it (Tajima et al., 2005, J Neurosci Res 79:714-723; PMID 15678515). Terashita and colleagues worked out why: humanin carrying D-serine at position 14 is also two to three orders of magnitude more potent than the native L-serine peptide, whereas D-serine at position 7 has no effect, so the S14G gain appears to mimic a post-translational isomerisation at Ser14. The same paper showed that Ser7 is required for humanin self-dimerisation, that S7A abolishes activity, and that combining these observations produced a derivative fully protective at 10 pM against 25 µM amyloid beta 1-43 (Terashita et al., 2003, J Neurochem 85:1521-1538; PMID 12787071).

Further engineering produced AGA-(C8R)HNG17, a 17-residue derivative in which the cysteine is replaced by arginine and an Ala-Gly-Ala extension is added, protective at 10 pM, and colivelin, in which the activity-dependent neurotrophic factor peptide is fused to the N-terminus of AGA-(C8R)HNG17. Colivelin suppressed death induced by FAD genes and amyloid beta 1-43 at 100 fM, signalling through a CaMKIV pathway triggered by the ADNF portion and a STAT3 pathway triggered by the humanin portion, and intracerebroventricular colivelin prevented the spatial working memory impairment produced by repeated amyloid beta injection in mice (Chiba et al., 2005, J Neurosci 25:10252-10261; PMID 16267233). HNG remains the form used in most in vivo studies.

Preclinical Neuroprotection Data

The first in vivo evidence came from the amyloid-beta injection model. Tajima and colleagues gave S14G-HN by intracerebroventricular injection to mice made amnesic with amyloid beta 25-35 and found that 50 pmol or more prevented the impairment of short-term spatial working memory in the Y-maze, whereas 5 nmol of the inactive S14A mutant did not, matching the in vitro structure-activity relationship. S14G-HN also prevented the prolonged latency in a water-finding task, preserved cholinergic neuron numbers in the basal forebrain and striatum close to normal, and its effect on scopolamine-induced amnesia was blocked by the tyrosine kinase inhibitor genistein (Tajima et al., 2005; PMID 15678515).

Niikura and colleagues then tested a genetic model. Triple transgenic mice carrying APPswe, tauP301L and PS1M146V received intranasal S14G-HN for three months. Treatment ameliorated cognitive impairment in male mice, lowered brain amyloid beta levels markedly in both sexes by ELISA and immunohistochemistry, and increased expression of the amyloid-degrading enzyme neprilysin in the outer molecular layer of the hippocampal formation, with neprilysin activity also raised in vitro (Niikura et al., 2011, PLoS One 6:e16259; PMID 21264226). Note that this was intranasal delivery in a triple transgenic line, not intracerebroventricular delivery in APP/PS1 mice as some secondary sources state.

In stroke, Xu and colleagues subjected mice to 75 minutes of middle cerebral artery occlusion followed by 24 hours of reperfusion. Pretreatment with 0.1 µg HNG intracerebroventricularly 30 minutes before ischaemia reduced infarct volume from 56.2 ± 3.0 percent to 26.1 ± 1.4 percent, post-treatment at four hours of reperfusion still reduced it to 45.6 ± 2.6 percent, and 1 µg given intraperitoneally one hour before ischaemia or two hours into reperfusion was also effective, with improved neurological scores, less TUNEL-positive apoptosis and less activated PARP, and ERK activation implicated in the mechanism (Xu et al., 2006, Stroke 37:2613-2619; PMID 16960089). The consistency across an amyloid injection model, a transgenic model and an ischaemia model supports the view that humanin acts on a shared downstream death pathway rather than on a disease-specific target.

Cardioprotective Research

Muzumdar and colleagues moved humanin into the heart. Male C57BL/6J mice underwent 45 minutes of left coronary artery occlusion and 24 hours of reperfusion, with HNG or vehicle given intraperitoneally one hour before or at the time of reperfusion. HNG reduced infarct size relative to the area at risk in a dose-dependent fashion, with the maximal effect at 2 mg/kg, improved left ventricular ejection fraction and preserved ventricular dimensions at one week, increased phosphorylation of AMPK and of endothelial nitric oxide synthase in the heart, attenuated BAX and Bcl-2 changes after reperfusion, and improved cardiomyocyte survival against daunorubicin in vitro (Muzumdar et al., 2010, Arterioscler Thromb Vasc Biol 30:1940-1948; PMID 20651283). Researchers comparing cardioprotective peptides may also want the SS-31 (elamipretide) literature, which targets cardiolipin on the inner mitochondrial membrane rather than the BAX pathway, and the BPC-157 ischaemia work.

The large-animal result is more qualified than it is usually presented. Sharp and colleagues tested HNG at 2 mg/kg in a porcine model of myocardial ischaemia and reperfusion. With 60 minutes of ischaemia, HNG reduced infarct size, which shows the cardioprotection translates beyond rodents. When ischaemic time was lengthened toward more patient-relevant durations, the infarct-sparing effect at the same dose was abolished, and the authors call for dose optimisation before any further translation (Sharp et al., 2020, JACC Basic Transl Sci 5:699-714; PMID 32760857). A 2022 minireview from the Muzumdar group summarises the proposed mechanisms, which include effects on autophagy, ER stress, oxidative stress and inflammation in addition to the AMPK and BAX pathways (Gong, Goetzman and Muzumdar, 2022, Biochim Biophys Acta Gen Subj 1866:130066; PMID 34896254).

Aging, Metabolism and Healthspan

The metabolic work also comes from the Muzumdar and Cohen groups. Using clamp methodology in rats, continuous intracerebroventricular infusion of humanin significantly improved whole-body insulin sensitivity, the effect was associated with activation of hypothalamic STAT3 and was negated by co-inhibiting hypothalamic STAT3, intravenous infusion of potent humanin derivatives reproduced the insulin-sensitising effect, inhibition of hypothalamic STAT3 abolished the hepatic effect of the intravenous analogue (consistent with the absence of any direct effect on primary hepatocytes), and a single treatment with a potent analogue lowered blood glucose in Zucker diabetic fatty rats (Muzumdar et al., 2009, PLoS One 4:e6334; PMID 19623253). The hepatic actions, in other words, are centrally mediated.

Yen and colleagues connected humanin to lifespan directly. In C. elegans, overexpressing humanin was sufficient to increase lifespan, dependent on daf-16/FOXO. Humanin transgenic mice showed overlapping phenotypes and increased protection against toxic insults. Treating middle-aged mice twice weekly with HNG improved metabolic healthspan parameters and reduced inflammatory markers. Across multiple species humanin levels generally decline with age, but they are stable in the naked mole-rat, a model of negligible senescence, and circulating humanin is much higher in the children of centenarians than in age-matched controls, while being reduced in Alzheimer’s disease and in MELAS (Yen et al., 2020, Aging 12:11185-11199; PMID 32575074). An earlier paper from the same group showed that humanin administration improves cognition in aged mice and that a mitochondrial SNP in the humanin-coding region, rs2854128, is associated with lower circulating humanin and with accelerated cognitive aging in a nationally representative cohort of older adults (Yen et al., 2018, Sci Rep 8:14212; PMID 30242290). No paper in this literature reports a specific percentage decline in plasma humanin between defined age bands; the published statement is that levels decline with age.

Key Research Findings

Humanin is a 24-residue peptide encoded within the mitochondrial 16S rRNA gene, identified by functional screening for suppression of FAD-gene and amyloid-beta neurotoxicity, with Cys8 and Ser14 essential and the active domain spanning Pro3 to Pro19 (Hashimoto et al., 2001, PNAS and J Neurosci). It protects neurons through a CNTFR/WSX-1/gp130 receptor complex that activates STAT3, AKT and ERK1/2, and it acts on phagocytes through FPRL1/FPR2 (Hashimoto 2009; Kim 2016; Ying 2004). Inside the cell it binds BAX and prevents its translocation to mitochondria and the release of cytochrome c, and it binds the heparin-binding domain of IGFBP-3 (Guo 2003; Ikonen 2003).

The S14G analogue HNG is about 1,000-fold more potent than the native peptide, a gain that mimics D-isomerisation at Ser14; Ser7-dependent dimerisation is required for activity (Terashita 2003; Tajima 2005). HNG prevented amyloid-beta-induced memory impairment in mice at 50 pmol intracerebroventricularly, lowered brain amyloid beta and improved cognition in triple transgenic mice after three months of intranasal treatment, and cut infarct volume from 56 to 26 percent in murine stroke (Tajima 2005; Niikura 2011; Xu 2006). In murine myocardial ischaemia-reperfusion, HNG reduced infarct size with a maximal effect at 2 mg/kg intraperitoneally via AMPK and eNOS phosphorylation; in pigs the same dose worked after 60 minutes of ischaemia but not after longer ischaemia (Muzumdar 2010; Sharp 2020). Central humanin improves insulin sensitivity through hypothalamic STAT3, and humanin overexpression extends C. elegans lifespan through daf-16/FOXO (Muzumdar 2009; Yen 2020). No controlled human intervention trial of humanin or HNG has been published; all efficacy data are from cell culture and animal models.

Analytical Considerations for Research-Grade Humanin

Humanin’s sequence presents two specific analytical points. First, the single cysteine at position 8 is essential for activity (Hashimoto et al., 2001, J Neurosci), and a free thiol is susceptible to oxidation during storage and handling, either to a disulfide-linked dimer or to oxidised cysteine species. A reversed-phase HPLC purity figure alone does not distinguish these from the monomer reliably, so the mass spectrometry identity result on the certificate of analysis should confirm the mass of the reduced monomer. Second, because native humanin and HNG differ by a single serine-to-glycine substitution, a mass difference of 30 Da, and differ in potency by three orders of magnitude, the identity test must be read closely enough to establish which sequence a lot actually contains. A lot sold as native humanin that is in fact HNG, or the reverse, would produce results a thousand-fold off expectation in any functional assay.

Storage of lyophilised humanin follows the general rules for cysteine-containing peptides: desiccated at -20 °C or below, protected from light, and reconstituted aliquots frozen in single-use volumes rather than cycled, since both oxidation and freeze-thaw stress are cumulative. The degradation pathways guide covers the chemistry in detail.

Note on sources. This article was rewritten in September 2026 after a citation audit. The earlier version attributed the STAT3 requirement to a 2009 paper in the Journal of Neuroscience Research with sample sizes and p-values that do not appear in any humanin paper (the STAT3 work is Hashimoto 2005, Life Sciences; the receptor identification is Hashimoto 2009, Molecular Biology of the Cell), named a first author for the Oncotarget signalling paper who is not on it and reversed its age finding, described the Niikura 2011 study as intracerebroventricular HNG in APP/PS1 mice with a 40 percent escape-latency improvement when it was intranasal treatment in triple transgenic mice with no such figure, placed Muzumdar 2010 in the American Journal of Physiology rather than ATVB, attributed the lifespan and healthspan study to “Cohen 2020” with an n=5 that is not in the abstract, cited a “Tajima 2019 Molecules” review, a “Yao 2021 BBA” metallothionein mechanism and a 10 nM humanin-IGFBP-3 affinity that could not be located, stated a 40 percent age-related decline in plasma humanin that no cited paper reports, and claimed protection in prion-induced neurodegeneration when the 2001 J Neurosci paper reports the opposite. Every citation above was checked against the indexed abstract on 18 September 2026.

Related research resources: NAD+ and cellular aging research and epithalon preclinical longevity research.

For research purposes only. Not for human consumption. Not for diagnostic or therapeutic use.

Maple Research Labs submits batches of its catalogue compounds to an independent laboratory for HPLC purity analysis and publishes the measured result for each tested batch on its Certificates of Analysis page, which also lists batches still awaiting a report. Humanin itself is not a catalogue compound; the mitochondrial-derived peptide currently listed is MOTS-c. Orders placed by 2 PM PT (Mon–Fri) ship the same business day, from BC with tracking. Browse the full research peptide catalog.

For peer-reviewed research on this topic, visit PubMed.