Mitochondria & cellular energy

Food is broken into small carbon units entering the citric acid cycle in the mitochondrial matrix. The cycle's real product is not ATP; it is reduced electron carriers, NADH and FADH2. Those hand their electrons to a chain of protein complexes in the inner membrane, and the electrons travel down that chain toward oxygen, the final acceptor, which becomes water.

The surprise is what the chain does with the released energy. It does not make ATP directly. It pumps protons out of the matrix, building an electrochemical gradient across the inner membrane. That gradient, the protonmotive force, is the actual energy store. ATP is made only when protons are let back in through a separate enzyme, ATP synthase, whose central stalk physically rotates as they pass. This is the chemiosmotic hypothesis, proposed in 1961 and initially treated as heresy because it broke with the assumption that a chemical intermediate had to link the two halves ^[3].

Consequences follow directly. Because the link is a gradient rather than a molecule, anything letting protons leak back without passing through ATP synthase uncouples respiration from ATP synthesis: oxygen is still consumed, but the energy leaves as heat. Brown fat does this deliberately; chemical uncouplers do it well enough to have killed people. And because the chain is a series circuit, a blockage anywhere backs everything up behind it.

CoQ10 and NAD+ come up constantly in supplement writing and their roles are precise. CoQ10, or ubiquinone, is the lipid-soluble carrier ferrying electrons between complexes inside the membrane; it is part of the chain rather than a vitamin for it ^[9]. NAD+ is the oxidised carrier collecting electrons from the citric acid cycle. Neither is a stimulant and neither adds energy; they are parts of the conveyor.

No electron transfer is perfect. A fraction of electrons react with oxygen prematurely to make superoxide, from which hydrogen peroxide and the rest follow. Complexes I and III are the main sources ^[5]. This is the free radical story covered in antioxidants and oxidative stress.

Two details are usually left out. First, superoxide production is not a fixed tax on respiration; it depends on the state the mitochondrion is in. Output is highest when the protonmotive force is high and the CoQ and NADH pools are reduced, which is the situation in an organelle not actively making ATP. One working hard, with a lower gradient because ATP synthase is spending it, leaks proportionally less ^[5]. Idling is dirtier than working.

Second, and more important for supplement reasoning, the same review is blunt that rates measured in isolated mitochondria cannot be extrapolated to a living cell, because the governing variables are hard to measure in vivo, and that such extrapolations are misleading ^[5]. A great deal of confident writing about oxidative damage rests on exactly that step.

Practically, a modest ROS signal is how a cell learns it has been worked. That is the basis of the training response, and why blanket high-dose antioxidant use around exercise can blunt adaptation rather than protect it. These molecules are byproduct and messenger at once.

Mitochondrial dysfunction sits on the standard list of hallmarks of ageing: it appears with age, accentuating it accelerates ageing, and improving it extends healthy life in models ^[6]. Several things change together: fewer and structurally altered mitochondria, reduced respiratory capacity per unit, accumulated damage to mitochondrial DNA, and impaired quality control.

Mitochondrial DNA is unusual and that is the root of most of it. Each cell carries hundreds to thousands of copies of a small circular genome encoding 37 genes, 13 of them subunits of the respiratory chain; everything else is imported. Because it sits close to the superoxide source, replicates independently of the cell cycle, and has less protective packaging than nuclear DNA, it accumulates mutations readily, and those can expand clonally to produce patches of respiratory-deficient tissue ^[2].

The strongest causal evidence is a mouse. Knock-in animals expressing a proofreading-deficient mitochondrial DNA polymerase accumulate three to fivefold more point mutations plus deletions, and develop reduced lifespan with premature weight loss, hair loss, spinal curvature, osteoporosis, anaemia and heart enlargement ^[7]. That established that these mutations can cause ageing phenotypes in a mammal.

What it did not establish is that this is the mechanism of ordinary ageing, and here the honest account diverges from the popular one. The mutator mouse carries a mutation burden far above what normal ageing produces, and follow-up work found its phenotype tracked replication errors and respiratory dysfunction more closely than oxidative damage, which is awkward for the classical free radical theory. The defensible position: mitochondrial function clearly declines with age and clearly matters, mitochondrial DNA damage contributes, and the simple story in which oxidative damage to that DNA drives ageing is not supported as stated. Chronic stress, inactivity and poor sleep all worsen the picture; see the HPA axis and cortisol and sleep and circadian rhythm.

The encouraging half is that this population is not fixed. Mitochondria are built, fused, split and destroyed continuously, and the balance answers demand. Biogenesis is the building of new mass, coordinated largely through the coactivator PGC-1alpha. Mitophagy is the selective destruction of damaged organelles, and it matters as much, because one that is not cleared keeps leaking superoxide.

Exercise drives both, with an effect size nobody argues about. It is worth being precise about what training changes, because a proteomic study of high-intensity interval training in human muscle found the answer is not simply more of everything. Content rose, but hundreds of transcripts, proteins and lipids changed non-proportionally to that rise; the data suggested improving electron flow into oxidative phosphorylation mattered more for ATP generation than increasing the abundance of machinery, and did not support the idea that training works through supercomplex assembly ^[8]. Training remodels the organelle qualitatively, not just quantitatively.

That is also a useful lens for supplement claims. Almost every compound below is sold on a biogenesis marker: more PGC-1alpha, more mitochondrial DNA copies, higher citrate synthase. Those measure content, and content is only part of what decides whether a tissue works better.

CoQ10 has the most substantial clinical record, and its best result is not a cognitive one. Q-SYMBIO randomised 420 patients with moderate to severe heart failure to 300 mg per day or placebo on top of standard therapy for two years. Short-term endpoints at 16 weeks were negative. The two year composite of major adverse cardiovascular events was reached by 15% on CoQ10 against 26% on placebo, with lower cardiovascular and all-cause mortality ^[10]. Failing heart muscle has high mitochondrial demand and measurably low CoQ10, which is the pattern across this field: replacement helps most where something is genuinely depleted ^[9].

PQQ is sold as a biogenesis agent and the underlying work is real but preclinical. In mouse hepatocytes, 10 to 30 micromolar PQQ raised citrate synthase and cytochrome c oxidase activity, mitochondrial DNA content and respiration through CREB phosphorylation and PGC-1alpha expression; knocking down either abolished the effect ^[11]. A clean mechanism in a dish. Human outcome trials are small and short.

NAD+ precursors are the most heavily marketed and the human data the most sobering. In a randomised double-blind crossover trial, 12 aged men took 1 g of nicotinamide riboside daily for 21 days. The muscle NAD+ metabolome rose, confirming the compound reaches its target. Muscle bioenergetics did not change. What changed was a downregulation of energy metabolism pathways in the transcriptome plus a fall in circulating inflammatory cytokines ^[12]. The anti-inflammatory signal may be the real story; the bioenergetic claim was not supported. The same caution covers NMN.

Urolithin A, a gut metabolite of pomegranate ellagitannins that stimulates mitophagy, has one properly powered trial: 66 adults aged 65 to 90, 1,000 mg daily for four months. Both primary endpoints, six minute walk distance and maximal ATP production by magnetic resonance spectroscopy, were not better than placebo. Secondary muscle endurance measures and several plasma biomarkers did improve ^[14]. A mixed result reported honestly, and still stronger than most of this category has.

Acetyl-L-Carnitine carries long-chain fatty acids across the inner membrane to be burned, a genuine and specific role; its trials sit mostly in older or fatigued groups. Creatine works elsewhere, buffering ATP rather than making it, and six weeks at 5 g per day improved working memory and reasoning in vegetarian young adults ^[15]. MOTS-c is the most speculative and the most interesting: a 16 amino acid peptide encoded inside the mitochondrial 12S rRNA gene, acting on skeletal muscle by inhibiting the folate cycle and activating AMPK, which prevented diet-induced obesity and insulin resistance in mice ^[13]. No human data yet. Not medical advice.

Three questions separate the substantial claims from the decorative ones.

Was anything measured in a person? Much of this literature is cell culture at concentrations oral dosing never reaches, or rodent work at doses that do not scale. That work is where mechanisms come from, and it is not evidence of a human effect. PQQ is the clearest case of a real mechanism whose translation is still thin ^[11].

Was the endpoint content or function? Mitochondrial DNA copies, PGC-1alpha and citrate synthase all measure how much machinery is present. Whether the tissue does more work is separate, and the two come apart: training changed composition out of proportion to quantity ^[8], while nicotinamide riboside raised the intended metabolite without moving bioenergetics at all ^[12].

Was the population depleted? Repletion helps where something is missing and does little where it is not. That is the honest reading of the CoQ10 record: a strong result in failing heart muscle, measurably low in it ^[9][10], and weaker results in healthy people. The same logic covers carnitine and NAD+ precursors. Absorption decides the rest; see half-life and bioavailability.

What is genuinely not known deserves stating plainly. There is no established way to measure a healthy person's mitochondrial function outside a research setting, and commercial tests claiming to are not validated. No supplement has been shown to extend human healthspan through a mitochondrial mechanism. And whether the age-related decline is mainly a cause of ageing or largely a consequence remains open, which is why the hallmarks framework treats mitochondrial dysfunction as interconnected with the other eleven rather than upstream ^[6].