Inside the machine that makes almost all of your energy.

NR / NMN, refill the NAD⁺ pool

Established: blood NAD⁺ rises in essentially every competent trial, NR 1,000 mg/day raised it ~60% over 6 weeks (n=30, ages 55–79); NMN 300–900 mg/day raised NAD⁺ dose-dependently over 60 days (n=80). Not shown: NR at 2,000 mg/day for 12 weeks in 40 obese insulin-resistant men did not improve insulin sensitivity, glucose disposal, or energy expenditure despite higher NAD⁺, and pooled analyses in adults over 60 find no clear benefit for muscle mass, grip, or walking speed. Raising the fuel gauge is not the same as fixing the engine.

Urolithin A, clear the broken units

The best functional dataset of any mitochondrial supplement, and still a cautionary tale. In ATLAS (n=88 untrained adults, 4 months), hamstring strength rose ~12% vs placebo and biopsy mitophagy proteins increased, but the primary endpoint, peak power output, did not improve significantly. In older adults (n=66), muscle endurance improved while 6-minute walk and maximal ATP production did not. Both flagship trials were run by the sponsor, Amazentis.

CoQ10, the shuttle itself

Strong in disease: in Q-SYMBIO (n=420, moderate–severe heart failure, 300 mg/day for 2 years), major cardiovascular events were 15% vs 26% on placebo (HR 0.50) and all-cause death 10% vs 18%. Weak for optimization: fatigue benefits in meta-analyses come mostly from ill populations (heart failure, MS, fibromyalgia). Nothing yet shows a healthy person with adequate CoQ makes more ATP by adding more. It also interacts with warfarin.

Metformin, the brake nobody mentions

Part of metformin’s glucose-lowering action is mild Complex I inhibition. In the MASTERS trial (n≈94 older adults, 14 weeks of resistance training), lean-mass gains were +1.95% on placebo vs +0.41% on metformin. In a second RCT (n=53, 12 weeks aerobic), metformin blunted VO₂max gains and abolished the training-induced increase in muscle mitochondrial respiration (~25% in placebo). For a publication about going beyond baseline, that trade-off is the story.

PQQ, “builds new mitochondria”?

The biogenesis claim comes from rodents and cell culture. Human trials are a handful of small studies (largest n≈64) reporting modest cognitive-domain signals, and no human trial has ever measured mitochondrial biogenesis, density, or function as an outcome. The popular slogan “PQQ builds new mitochondria, CoQ10 runs the existing ones” is a memorable mechanism story with no human mechanistic evidence behind it.

Exercise, the benchmark everything chases

Training does in humans what the supplements are trying to mimic: raises PGC-1α, mitochondrial respiration, and mitophagy flux, measured directly in muscle. Cardiorespiratory fitness shows a dose-dependent, no-upper-limit association with lower mortality in 122,007 adults (elite fitness HR 0.20 vs low fitness), and VO₂max is trainable at every adult age. Any mitochondrial content that leads with a pill before training has the evidence hierarchy upside down. Try the “Trained athlete” preset, that’s this card, visualized.

MOTS-c, the exercise signal with no human trial

Why people care: MOTS-c is genuinely unusual. It is a 16-amino-acid peptide encoded in the 12S rRNA region of mitochondrial DNA, not nuclear DNA. Circulating levels fall with age in human cohorts, and a genetic variant (m.1382A>C) has been linked to metabolic traits and longevity in Japanese cohorts. In mice it improves insulin sensitivity and exercise capacity, and exercise itself raises circulating MOTS-c in humans.

What is missing: there is no human trial of MOTS-c itself. The only trial people cite is for a modified analog, CB4211 from CohBar, in 88 volunteers over 28 days. It cleared safety, showed no clear benefit, and the sponsor dissolved and the program stopped. The FDA flagged immune reaction risk for this class. Everything else is animal work, cell work, and observational links between natural levels and health.

Community practice, anecdotal: 5 to 10 mg subcutaneously several times a week as a mitochondrial exercise mimetic, often pre-workout, per the corpus. That protocol has never been tested as a protocol.

SS-31 / elamipretide, the membrane stabiliser

Mechanism: SS-31 is a synthetic aromatic-cationic tetrapeptide (D-Arg-Dmt-Lys-Phe-NH2). It concentrates roughly 1,000-fold in the inner membrane, binds cardiolipin, and stabilises the cardiolipin and cytochrome c interaction. Unlike a normal antioxidant, it does not mop up ROS after they form. It reduces leak at the source by keeping the chain organised, and it makes the permeability transition pore harder to open under stress. It helps struggling mitochondria most, and does comparatively little in healthy young ones in preclinical work.

The approval, stated precisely: on 19 September 2025 the FDA granted accelerated approval to elamipretide as Forzinity, to improve muscle strength in adult and paediatric patients with Barth syndrome weighing at least 30 kg. Barth syndrome is an ultra-rare genetic disorder of cardiolipin remodelling (TAZ gene), so the drug and the disease match at the exact lipid. Continued approval depends on a confirmatory trial.

SS-31, the trial record behind the approval

TAZPOWER (Barth syndrome): the randomised blinded phase enrolled just 12 patients in a crossover (40 mg daily subcutaneously for 12 weeks per period) and did not reach significance on its primary endpoints in that blinded phase. Ten patients continued into an open-label extension. At week 168, the extension reported a 96.1 metre improvement in 6-minute walk distance (p = .003) and improved knee extensor strength, the intermediate endpoint the accelerated approval rests on. Small, open-label at the decisive stage, and in a disease with no prior approved treatment.

MMPOWER-3 (primary mitochondrial myopathy): the larger, cleaner test went the other way. In 218 adults over 24 weeks at the same 40 mg daily dose, both primary endpoints missed. The difference in 6-minute walk was -3.2 metres in favour of placebo (95% CI -18.7 to 12.3, p = 0.69) and fatigue differed by -0.07 (p = 0.37). Heart failure and dry age-related macular degeneration trials for elamipretide also missed their primary endpoints.

Older adults signal: in 39 adults aged 60 to 85 with low muscle mitochondrial function, a single treatment gave a short term rise in maximal ATP production capacity (percentage change p = .045, absolute change p = .055) that was absent at day 7, with no significant benefit on the prespecified hand muscle fatigue test (PMID 34264994). A measurable blip in ATP capacity is not a lasting energy benefit.

Methylene blue, the bypass that trades efficiency for flow

Mechanism: methylene blue cycles between its oxidised form and its reduced form, leucomethylene blue. It accepts electrons from NADH via cellular reductases and donates them directly to cytochrome c. That skips Complexes I, II and III entirely, so electrons can still reach Complex IV and oxygen when the front of the chain is damaged or blocked. The trade is built into the diagram: the bypassed complexes are the proton pumps, so a bypassed electron earns less ATP than a full journey. It also inhibits monoamine oxidase, which is a primary drug action, not a side footnote.

Evidence: methylene blue is FDA approved for acquired methemoglobinaemia, a blood disorder, and has a long clinical history as a dye and treatment in specific hospital settings. For cognition, longevity and healthy mitochondrial optimisation, there is no large, well powered, double blind RCT in healthy people showing benefit. The human cognition and neuroprotection literature is early, small and mixed, and the strongest claims rest on the same bypass mechanism shown above plus animal work.

Safety, the part the marketing skips: methylene blue is a potent MAO inhibitor. Combined with SSRIs, SNRIs, other MAO inhibitors, tricyclics, tramadol, triptans or other serotonergic drugs it has caused serotonin syndrome, including deaths, which prompted an FDA Drug Safety Communication in 2011. In people with G6PD deficiency it can cause acute haemolytic anaemia and is contraindicated for that group. Its redox action also flips with dose: antioxidant behaviour at low exposure can become pro-oxidant at higher exposure. No dosing instructions are given here, deliberately. This is a prescription medicine with real interactions, not a casual nootropic dye.

Humanin and the other mitochondrial-derived peptides

Humanin (24 amino acids, encoded in the 16S rRNA region) was the first mitochondrial-derived peptide found, in 2001. It is cytoprotective in cell and animal work: it binds pro-apoptotic proteins such as Bax, blocks cytochrome c release, and helps preserve membrane potential under stress. In isolated cardiac mitochondria exposed to hydrogen peroxide, its potent analog HNG reduced ROS in part by decreasing Complex I activity. That is a protective trade, less flow for less damage, and it is the opposite of a simple energy booster story.

The small humanin-like peptides (SHLP1 to 6) sit in the same family and are earlier still. For humanin, SHLPs, BPC-157, TB-500 and injectable GHK-Cu combined, the corpus found zero modern controlled human efficacy trials for mitochondrial or energy outcomes. BPC-157 had 35 preclinical studies to 1 small retrospective clinical series in a 2024 systematic review. That does not prove they do nothing. It proves no one has shown, in a controlled human trial, that they move this chain.