What Is SS-31?

Published August 2026 · Patriot Labs Research Team · ~17 min read

A four-residue peptide with five names

Almost every compound in this catalogue carries an alias or two. SS-31 carries five, and the reason is a compressed history of the molecule's trajectory.

SS-31 is the laboratory designation. The SS stands for Szeto–Schiller, after Hazel H. Szeto and Peter W. Schiller, whose collaboration produced a small series of tetrapeptides in the early 2000s. The number is an index within that series — not a generation number or a potency rating. Elamipretide is the international nonproprietary name assigned once the compound entered formal drug development. MTP-131 and Bendavia were the development codes used by Stealth BioTherapeutics; both appear in the older literature, sometimes interchangeably with the INN.¹³ Forzinity is the brand name of the finished pharmaceutical product that received United States accelerated approval in September 2025.²³

Those names span a research chemical, a clinical candidate and an approved drug product, and the guide below has to keep all three apart. The material catalogued here as SS-31 is research-grade lyophilised peptide, not the approved product.

The literature is mid-sized. A PubMed search for elamipretide on 10 August 2026 returned a count of 355 records, of which 46 also matched randomized; cardiolipin AND SS-31 returned 38.²⁴ That is larger than the literature behind most catalogue peptides and far smaller than that behind a mainstream drug class — big enough to contain real disagreements.

The chemistry: an aromatic-cationic tetrapeptide

Written out, SS-31 is D-Arg-Dmt-Lys-Phe-NH₂, where Dmt is 2′,6′-dimethyltyrosine. Four residues, three unusual features, and every one of them is load-bearing.

The alternating motif. The residues alternate between basic and aromatic: arginine, dimethyltyrosine, lysine, phenylalanine. The 2004 paper describes the design explicitly — “the structural motif of these peptides centers on alternating aromatic and basic amino acid residues.”¹ The result is a molecule simultaneously strongly cationic and substantially hydrophobic, an unusual combination at this size. It is amphipathic not because it folds into a helix with a polar face, the way most amphipathic peptides work, but because the alternation is built into a chain too short to fold at all.

The D-arginine. Proteases evolved to cleave L-amino acid backbones, and a D residue at position one blocks aminopeptidase attack at the most exposed end of a very short chain. Combined with the C-terminal amide — the –NH₂, which removes the free carboxylate that carboxypeptidases recognise — both ends are chemically defended. For a four-residue peptide that is the difference between a measurable half-life and none. The general logic is covered in what peptides are and in how peptides degrade.

The dimethyltyrosine. Two methyl groups flanking the phenolic hydroxyl. Sterically hindered phenols are better hydrogen-atom donors and give more stable phenoxyl radicals than unhindered ones. In the original framing this residue was the scavenging element: the 2004 paper says dimethyltyrosine provides “scavenging properties,” and reports that analogs lacking it “did not inhibit mitochondrial ROS generation or swelling and failed to prevent myocardial stunning.”¹ That sentence is the origin of the analog comparison below, and its interpretation has since been revisited.

How SS-31 differs from its siblings

The Szeto–Schiller series was a small family, because the researchers were varying one design element at a time. Szeto's 2006 review in The AAPS Journal is the standard entry point.²

Why it goes where it goes

The mitochondrial inner membrane is the most electrically polarised membrane in the cell, with proton pumping by the respiratory chain building a potential difference across it, negative on the matrix side. The dominant strategy for delivering a molecule there exploits exactly this: attach a lipophilic cation — conventionally a triphenylphosphonium (TPP⁺) group — to the payload, and it is driven across electrophoretically into the matrix. MitoQ, a ubiquinone joined to TPP⁺ by an alkyl linker, is the best-known example. Rottenberg's 1984 paper, “Membrane potential and surface potential in mitochondria: uptake and binding of lipophilic cations,” sits in that tradition.⁴

The approach works, and it has a structural weakness. Accumulation is a function of the potential. A mitochondrion that is depolarised — because it is ischaemic, damaged, uncoupled, or simply the sick organelle in a heterogeneous population — accumulates less of the compound than a healthy one. A potential-driven delivery system delivers least where the pathology is worst.

SS-31 does not work that way. The 2004 report described the peptides as concentrating around 1000-fold in the inner mitochondrial membrane,¹ and the biophysical evidence attributes that localisation to lipid composition rather than to voltage. The clearest statement comes from a 2020 study in the Journal of Biological Chemistry by Mitchell and colleagues, working with model and mitochondrial membranes: SS-31 “partitions into the membrane interfacial region with an affinity and a lipid binding density that are directly related to surface charge.”⁹ The peptide is not being pulled through a membrane by a field; it is binding to an interface whose anionic character it is complementary to.

That last row is the conceptual pivot. For a TPP⁺ conjugate the membrane is plumbing; for SS-31, on the current model, the membrane is the target. The Mitchell study proposes that “tuning of surface charge” could underpin the peptide's activity, including by altering “the distribution of ions and basic proteins at the interface.”⁹

Cardiolipin, cristae and supercomplexes

Cardiolipin is genuinely strange as phospholipids go. Most membrane phospholipids have one glycerol backbone and two acyl chains; cardiolipin has two phosphatidyl groups joined by a third, bridging glycerol — a dimeric lipid with four acyl chains and two phosphate groups. It carries a net negative charge and, in eukaryotes, is found almost exclusively in the inner mitochondrial membrane. That distribution is not incidental: the mitochondrion is where the bacterial ancestor's lipid stayed.

Three structural roles are repeatedly invoked, and are reviewed by Paradies and colleagues in a 2014 article titled “Functional role of cardiolipin in mitochondrial bioenergetics.”⁸

Cristae geometry. Four tails and a small head group give cardiolipin a conical shape, which favours negative curvature. The inner membrane is not a smooth sac; it is folded into cristae with tight curvature at the rims, and a lipid that prefers curvature is enriched where curvature happens. That architecture is what packs a very large amount of respiratory-chain-bearing membrane into a small organelle.

Respiratory supercomplex assembly. The complexes of the electron transport chain do not float independently; they associate into higher-order assemblies, sometimes called respirasomes, and cardiolipin is implicated as a structural glue at complex interfaces. The functional argument is that this shortens the diffusion path for mobile carriers between complexes, making electron transfer more efficient and less leaky.

The cytochrome c interaction. Cytochrome c is a soluble, positively charged haem protein that shuttles single electrons between complex III and complex IV along the outer face of the inner membrane, and it binds cardiolipin electrostatically. Under some conditions that interaction changes character: the protein's conformation loosens and it can act as a peroxidase toward cardiolipin itself, oxidising the very lipid holding it in place and releasing itself from the membrane. The 2004 paper opens by noting that reactive oxygen species “induce dissociation of cytochrome c from cardiolipin on the inner mitochondrial membrane”¹ — that problem is why the programme started.

Against that background the proposed action of SS-31 is easy to state: the peptide is cationic, cardiolipin is anionic and concentrated at the inner membrane, and the peptide partitions into that interface. Birk and colleagues published the case in 2013 as “The mitochondrial-targeted compound SS-31 re-energizes ischemic mitochondria by interacting with cardiolipin,”⁵ followed in 2014 by “Targeting mitochondrial cardiolipin and the cytochrome c/cardiolipin complex to promote electron transport and optimize mitochondrial ATP synthesis.”⁶ Szeto's companion 2014 review is titled “First-in-class cardiolipin-protective compound as a therapeutic agent to restore mitochondrial bioenergetics.”⁷ All three are cited here as bibliographic records, because their publishers do not expose abstracts through the metadata services used for this guide.

The current model, stated in the 2025 editorial accompanying the regulatory approval, is that elamipretide “binds to cardiolipin on the inner mitochondrial membrane, stabilizing respiratory chain supercomplexes, enhancing electron transport efficiency, and reducing reactive oxygen species production.”²³ Note the causal direction: reduced radical production is downstream of better-organised electron transport, not the primary event. An electron handed cleanly from one complex to the next does not get the chance to reduce oxygen prematurely. Less leak means less superoxide, without anything scavenging anything.

How the mechanism story changed

Most catalogue peptides have one mechanistic story that gets repeated. SS-31 has two, and the transition between them is a well-documented case of a field revising its explanation of a molecule while continuing to work on it.

Phase one, roughly 2004–2012: the targeted antioxidant. The founding paper is titled “Cell-permeable peptide antioxidants targeted to inner mitochondrial membrane…”¹ Szeto's 2006 reviews are “Cell-permeable, mitochondrial-targeted, peptide antioxidants”² and “Mitochondria-targeted peptide antioxidants: novel neuroprotective agents.”³ The logic was clean and of its time: reactive oxygen species are generated at the respiratory chain; general antioxidants fail because they never reach the site of generation in useful concentration; therefore build a scavenger with a mitochondrial postcode. The dimethyltyrosine was the scavenger, the aromatic-cationic motif the postcode, and the 2004 finding that analogs lacking dimethyltyrosine lost activity looked like confirmation.¹

Phase two, roughly 2013 onward: the cardiolipin-binding structural modulator. By 2013 the titles have changed — “re-energizes ischemic mitochondria by interacting with cardiolipin,”⁵ “targeting mitochondrial cardiolipin and the cytochrome c/cardiolipin complex,”⁶ “first-in-class cardiolipin-protective compound.”⁷ By 2018 Szeto and Liu write about “cardiolipin-targeted peptides” that “rejuvenate mitochondrial function, remodel mitochondria, and promote tissue regeneration during aging.”¹² By 2020 the biophysics paper frames the question in terms of lipid binding, surface electrostatics and lipid packing, while noting that, as of that writing, “little is known regarding how SS peptides interact with or alter the physical properties of lipid bilayers.”⁹ The same year a proteomic study in PNAS came at it from another direction entirely, as “Mitochondrial protein interaction landscape of SS-31.”¹⁰

Three things are worth saying about this transition, because it is easy to over-read.

First, it is a genuine reframing rather than a rebranding, because the two accounts predict different things. An antioxidant should show effects that track with the radical-scavenging chemistry of its phenol and be substitutable by other scavengers delivered to the same place; a structural modulator should show effects that track with membrane binding and lipid organisation. But the reframing does not require that the earlier observations were wrong. Reduced reactive oxygen species output is common to both models; what differs is whether it is cause or consequence, and the 2025 summary places it second in the chain, after supercomplex stabilisation.²³

Second, the SS-20 comparison that once looked like clean evidence for scavenging now looks ambiguous. Removing dimethyltyrosine removes a hindered phenol, but it also removes a bulky aromatic residue from a motif whose function, on the newer model, is interfacial binding. A substitution that changes two things at once cannot cleanly attribute an effect to one of them. That is not a criticism of the 2004 work; it is what happens when a control designed under one hypothesis is re-read under another.

Third, neither model has been shown to be the whole story. The 2020 biophysics paper offers “nonexclusive mechanisms” rather than one, and presents its calcium finding as a proof of concept.⁹ A compound whose mechanism is still argued over twenty years on is a compound to describe carefully.

The preclinical literature

The preclinical work clusters into four domains. All of it is animal, ex-vivo or cell-culture work, and none of it establishes anything about people.

Cardiac ischaemia–reperfusion. The first domain, and still the largest. The 2004 paper included an ex-vivo heart model, reporting that the peptides “potently improved contractile force” after reperfusion and that analogs without dimethyltyrosine “failed to prevent myocardial stunning.”¹ The rationale is specific to reperfusion: the burst of radical production when oxygen returns to ischaemic tissue is exactly the setting in which cardiolipin peroxidation and cytochrome c release are proposed to matter.

Heart failure. Sabbah and colleagues published a chronic study in dogs with advanced heart failure in Circulation: Heart Failure in 2016.¹³ A large-animal model with chronic treatment is a stronger design than an acute rodent experiment, and this line of work is what carried the compound into the cardiac clinical programme. Sabbah later reviewed the Barth syndrome cardiomyopathy case in Heart Failure Reviews.¹⁴

Renal. The 2013 cardiolipin paper appeared in a nephrology journal for a reason: the ischaemic kidney was one of the main models in which the compound was characterised.⁵ Renal tubular cells are unusually mitochondria-dense.

Skeletal muscle and aging. The most-cited experiment here is Siegel and colleagues' 2013 paper in Aging Cell. Using optical and ³¹P magnetic resonance spectroscopy in young and old mice, the authors report that “age-related declines in resting and maximal mitochondrial ATP production, coupling of oxidative phosphorylation (P/O), and cell energy state (PCr/ATP) were rapidly reversed after SS-31 treatment, while SS-31 had no observable effect on young muscle.”¹¹ Two details matter. The timescale was one hour after a single injection — far too fast for anything transcriptional, and consistent with a direct physical effect on membranes. And the effect was age-dependent, which is what a compound that restores an organised state — rather than one that adds function — would be expected to show. Related material sits in the guide to MOTS-c, another mitochondria-associated peptide with a very different origin.

A 2024 preprint from an overlapping group is worth flagging for its title alone: elamipretide “improves cardiac and skeletal muscle function during aging without detectable changes in tissue epigenetic or transcriptomic age.”²⁶ It is cited here as a bibliographic record only, but the framing is a useful corrective to the assumption that functional improvement in an old animal means the animal is biologically younger.

Neurodegeneration models. The thinnest of the four domains in the retrievable record. Szeto's 2006 review, “Mitochondria-targeted peptide antioxidants: novel neuroprotective agents,” is the entry point,³ cited here by title only. The rationale — neurons are long-lived, post-mitotic and mitochondria-dependent — is not evidence, and the honest summary is that this work has not produced a clinical programme comparable to the cardiac or myopathy ones.

The clinical record, told straight

Vendor writing about SS-31 tends to mention that elamipretide reached late-stage trials and received an FDA approval, and to leave it there. The full record is more interesting and considerably less flattering.

Everything in this section describes trials of a pharmaceutical product conducted under regulatory supervision, and is reported here as history. It is not a claim about research-grade material.

Read the table as a whole and a pattern emerges. The compound is consistently described as well tolerated. The primary endpoints in the two largest programmes — primary mitochondrial myopathy and heart failure with reduced ejection fraction — were not met. The positive signals come from small pilot studies and open-label extension phases, the designs most vulnerable to bias.

That is not an unusual shape for a mitochondrial therapeutic. The available endpoints are functional and effort-dependent — walk distances, fatigue scales, imaging volumes — and noisy relative to the effect sizes plausible from a bioenergetic intervention over weeks. A trial can be negative because the compound does nothing, or because the endpoint could not detect what it did. The literature has not distinguished those possibilities.

The 2025 accelerated approval

On 19 September 2025 the United States Food and Drug Administration granted accelerated approval to elamipretide for Barth syndrome.²³ The FDA's list of 2025 novel drug approvals records the brand name Forzinity, the active ingredient elamipretide, an approval date of 9/19/2025, and an indication given verbatim as “To improve muscle strength in patients with Barth syndrome weighing at least 30 kg.”²²

Barth syndrome is why this approval is mechanistically coherent rather than arbitrary. It is a rare X-linked disorder caused by mutations in tafazzin, the enzyme that remodels cardiolipin's acyl chains after synthesis; the result is abnormal cardiolipin and, downstream, cardiomyopathy, skeletal myopathy and neutropenia.²³ A compound proposed to bind cardiolipin, applied to a disease of cardiolipin remodelling, is about as direct a mechanistic match as drug development produces.

Three qualifications belong with that sentence, and the editorial published alongside the approval states all three. The randomised crossover trial “resulted in no significant improvement in the 6-minute walk test or fatigue scores”; the sustained benefits were “observed during a 168-week open-label extension”; and “as a condition of accelerated approval, a confirmatory trial is required.”²³ Accelerated approval is a conditional pathway, not a final verdict, and an approval resting substantially on open-label extension data is exactly the case the confirmatory-trial requirement exists to handle.

The distinction that matters here. An approved pharmaceutical product is one manufacturer's finished dosage form, made under pharmaceutical quality systems, characterised in a regulatory submission and licensed for a named indication in a named population. Research-grade SS-31 is none of those things. The 2025 approval attaches to that drug product and that file; it confers no approval, no indication and no human-use status on research material, including the SS-31 catalogued here, which remains a laboratory reagent sold for in-vitro research only.

Bench practice: handling a lyophilised tetrapeptide

SS-31 is supplied as a lyophilised powder. What follows is general laboratory practice for research peptides — not a protocol, and not instruction in use.

Why it is freeze-dried. Water is the reagent in most peptide degradation chemistry. Removing it arrests hydrolysis, slows oxidation and produces a solid that tolerates shipping and long storage far better than any solution. The trade-off is that the powder is hygroscopic: an amorphous cake pulls moisture from humid air, reintroducing the chemistry the drying removed. That is why allowing a cold vial to equilibrate before opening is standard practice. The reconstitution guide covers the procedure and the storage guide the conditions.

Chemistry specific to this sequence. The C-terminal amide and D-arginine protect against enzymatic attack and do nothing against non-enzymatic chemistry — a protease-resistant peptide is not a chemically stable one. The phenolic hydroxyl on the dimethyltyrosine is oxidisable; hindered phenols are good hydrogen-atom donors, which is the point of including one, and it also makes that residue the most likely oxidation site. And with three positive charges at neutral pH, solubility and adsorption to container surfaces are pH-dependent, which matters most in low-concentration work where losses to vial walls are proportionally larger. The general failure modes are treated in how peptides degrade.

Once in solution, the clock restarts. Cold, dark, minimal freeze–thaw cycling of stock and aliquoting rather than repeatedly re-entering one vial are the standard mitigations. Cycling matters more than most people expect, because each freeze concentrates solutes in the shrinking unfrozen fraction and drives local pH excursions as buffer components crystallise at different points.

Identity is an experimental variable. A four-residue peptide with a D-amino acid and a non-standard methylated residue is a harder synthesis than its length suggests, and the plausible impurities — epimerised arginine, incomplete methylation, deletion sequences, residual trifluoroacetate — are not visible by inspection. Nothing about a white powder distinguishes 95% material from 99%. That is what a certificate of analysis is for.

What the evidence does not establish

SS-31 has a more substantial evidence base than most compounds in this catalogue and, at the same time, a better-documented record of failure. Both facts have to survive the summary.

The mechanism is a model, not a settled fact. The cardiolipin account is well argued and now standard, but the 2020 biophysical study proposes “nonexclusive mechanisms” and presents its central calcium result as a proof of concept.⁹ The field revised its explanation once already. There is no principled reason to assume the current version is final.

Two large clinical programmes missed their primary endpoints. MMPOWER-3, with 218 participants over 24 weeks, did not meet its primary endpoints.¹⁷ PROGRESS-HF did not improve its imaging endpoint at four weeks.¹⁹ The randomised crossover phase in primary mitochondrial myopathy did not reach statistical significance on its primary endpoint.¹⁶ These are not ambiguous results being spun negatively; they are the outcomes of the largest and best-controlled studies conducted on this compound.

The positive signals come from the weakest designs. The renal study enrolled 6 treated and 8 placebo participants and is described by its own authors as a pilot.²⁰ The Barth syndrome benefits that supported the approval were observed in an open-label extension, not in the blinded crossover phase.^21,23 Open-label extensions have no placebo arm, no blinding, and a population self-selected by having tolerated and continued treatment. They generate hypotheses; they do not test them.

An accelerated approval is conditional and narrow. It covers one rare genetic disorder of cardiolipin remodelling, in a defined population, and carries a confirmatory-trial requirement.^22,23 It is not evidence for any other setting, and given that trials in two other settings were negative, generalising from it runs against the available data.

Animal results do not transfer, and the aging data show why. The Siegel study reversed age-related bioenergetic deficits in mice within one hour and had no observable effect in young animals.¹¹ That is an elegant result in one rodent model measured by magnetic resonance spectroscopy. The 2024 preprint from an overlapping group reports functional improvement in aged mice “without detectable changes in tissue epigenetic or transcriptomic age”²⁶ — a reminder that functional and biological-aging readouts can move independently, and that neither is a claim about people.

In-vitro concentrations are chosen by the experimenter. Cell and isolated-mitochondria work applies the peptide at defined concentrations in defined media. Those are experimental parameters, not findings.

Potential-independence is a mechanistic argument, not a demonstrated advantage. That SS-31 localises by lipid interaction rather than electrophoresis is well supported biophysically.⁹ That this produces better outcomes than potential-driven targeting is a hypothesis; no trial cited here compared SS-31 against a TPP⁺-conjugated compound.

Research-grade material is not the studied article. Every trial above used a pharmaceutical product manufactured under a regulatory file. Purity, counterion content, impurity profile and formulation all differ between such a product and a research reagent, and no published result transfers automatically across that gap.

Nothing above should be read as a claim about any Patriot Labs product. SS-31 is catalogued here as a research chemical, and the references cited describe published research and regulatory history rather than product performance.

Frequently asked questions

What does the SS in SS-31 stand for? Szeto–Schiller, after the two researchers whose collaboration produced the series. The 31 is a compound index, not a generation or potency number. The same molecule appears as elamipretide, MTP-131 and Bendavia, and the approved pharmaceutical product is branded Forzinity.

How is SS-31 different from MitoQ and other mitochondria-targeted compounds? Most mitochondrial targeting uses a lipophilic cation, usually triphenylphosphonium, which accumulates in the matrix because the inner membrane is polarised. SS-31 is reported to partition into the membrane interfacial region with affinity and binding density directly related to surface charge⁹ — so the localisation principle does not weaken in a depolarised organelle, and the peptide associates with the membrane rather than entering the matrix.

What is cardiolipin and why does it keep coming up? A dimeric phospholipid with four acyl chains and two negative charges, found almost exclusively in the inner mitochondrial membrane and implicated in cristae curvature, supercomplex assembly and cytochrome c binding.⁸ It is also the anionic surface the cationic peptide is proposed to bind.

Was SS-31 always described as a cardiolipin binder? No, and this is the most interesting thing about it. From 2004 to roughly 2012 the published framing was “mitochondria-targeted antioxidant,” with the dimethyltyrosine residue as the radical scavenger.^1,2,3 From 2013 onward the framing shifts to cardiolipin binding and structural modulation.^5,6,7 On the newer model, reduced radical output is a consequence of more efficient electron transport rather than the primary action.²³

If it was approved by the FDA, does that mean the trials worked? Not in the general sense. The approval is an accelerated approval for Barth syndrome, granted on 19 September 2025 and carrying a confirmatory-trial requirement.^22,23 The larger trials in primary mitochondrial myopathy and in heart failure with reduced ejection fraction did not meet their primary endpoints.^17,19 Both facts are part of the same record.

Is it approved for human use? No. Elamipretide received FDA accelerated approval in September 2025 as a prescription pharmaceutical product for Barth syndrome under the brand name Forzinity, but that approval covers one manufacturer's finished drug product under a specific regulatory file. SS-31 supplied by Patriot Labs is a research chemical, sold strictly for in-vitro research and laboratory use only. It is not for human or veterinary consumption, and nothing in this guide describes how to use it.

All Patriot Labs products are sold strictly for in-vitro research and laboratory use only. Not for human or veterinary consumption. This guide is educational and describes peptide chemistry and published research in general terms; it is not medical advice, does not describe how to use any product, and the references cited do not constitute a product claim.