← Surfer · The Kit · Sage hub

Sage — A Metabolic Hypothesis of Pyrroloquinoline Quinone

Good Doctor · The Kit · Strength pillar (inventory side) · Frugal Philosophy / Metabolic Hypothesis Series · October 2026 · Val Zudans, MD · hypothesis · trial doses as citation context only · source disclaimer intact

Milligram figures below are published-trial and safety-assessment citation context only — not a personal recommendation and not a regimen. Companion pages: Sage — Methylene Blue (flux layer) and Sage — Biogenesis.

A Metabolic Hypothesis of Pyrroloquinoline Quinone

EnSurf Strength pillar · candidate inventory signal

Frugal Philosophy / Metabolic Hypothesis Series

Disclaimer

This essay presents a metabolic hypothesis and a synthesis of the available literature on pyrroloquinoline quinone. It is written for educational and exploratory purposes within the framework of mitochondrial bioenergetics and the EnSurf system. It is not medical advice, a treatment protocol, or a recommendation for self-administration. Pyrroloquinoline quinone is not an approved drug. Human evidence for mitochondrial biogenesis is absent. Cognitive trials are small, short, and often industry-linked. The European Food Safety Authority has assessed a disodium salt for food-supplement use at 20 mg/day in healthy adults, excluding pregnant and lactating women; that assessment is not a claim of efficacy, and it does not clear use in children or in kidney disease. Any use requires individual medical evaluation. The author assumes no liability for actions taken on the basis of this discussion.

A Cofactor That Never Became a Vitamin

Pyrroloquinoline quinone began as an unnamed prosthetic group in bacteria. In 1964 Jens Gabriel Hauge, working on bacterial glucose dehydrogenase, described a third redox cofactor after nicotinamide and flavin, and guessed that it was a naphthoquinone. Anthony and Zatman found the same unknown factor in alcohol dehydrogenase. In 1979 Salisbury and colleagues, and independently Duine and colleagues, extracted the group from the methanol dehydrogenase of methylotrophs and solved the structure: a tricyclic ortho-quinone, first called methoxatin. Bacteria use it as the cofactor of glucose and methanol dehydrogenases. Mammals do not have those enzymes, and they do not have a dedicated pyrroloquinoline quinone enzyme of their own.

In 2003 Kasahara and Kato announced in Nature that the molecule was a new vitamin, the cofactor of an enzyme they placed in lysine degradation. The claim did not survive contact with the pathway. In 2005 Rucker and colleagues reported that the lysine-degradation markers the vitamin argument required — aminoadipate-semialdehyde dehydrogenase activity and plasma aminoadipate — do not move with dietary pyrroloquinoline quinone status. Felton and Anthony showed, in the same exchange, that the sequence databases used to find a mammalian binding site had mislabeled β-propeller domains as pyrroloquinoline quinone motifs. A vitamin is a cofactor an animal cannot make and cannot do without, attached to an enzyme that fails in its absence. That standard was not met. What the deprivation literature actually shows is narrower, and more useful to a metabolic hypothesis: mice raised on chemically defined diets very low in the molecule grow with fewer mitochondria and lower expression of mitochondrial genes, and putting it back restores both. Accessory factor is the right name. Vitamin is not.

Food contains it at nanograms per gram. Kumazawa’s 1995 survey found free pyrroloquinoline quinone in every food tested, from about 4 to 61 ng/g, highest in natto, then parsley, green pepper, kiwi, and spinach. The European Food Safety Authority, assessing a proposed supplement use of 20 mg/day, judged that dose to be at least 250 times estimated background intake from food. A bowl of natto is not a 20 mg intervention. The supplement dose is a pharmacological multiple of dietary exposure. The hypothesis has to be stated at that multiple, or it is a different hypothesis — a repletion story the food data do not support.

The commercial rediscovery has followed the path methylene blue took, one register down. An old, simple, off-patent redox molecule is easy to mock and easy to sell, and the sentence that sells it is the least interesting sentence available: antioxidant. Radical quenching is a property of the quinone. It is shared with a large number of molecules that do not change the number of mitochondria. The metabolic reading is different. Pyrroloquinoline quinone is a catalytic redox cofactor whose useful event, in the systems where it has been shown, is a transcriptional request for more respiratory inventory. The mockery and the marketing agree in stopping at the surface. The surface is not the mechanism.

Mechanism of Action in Energy Metabolism

Pyrroloquinoline quinone cycles between an oxidized ortho-quinone and a reduced quinol. The couple sits positive of the NADH/NAD+ couple, so the oxidized form can accept electrons from NADH. It is not consumed in the reaction. Like methylene blue, it is catalytic rather than stoichiometric. Unlike methylene blue, it does not sit in the respiratory chain as an alternative carrier from Complex I to cytochrome c.

The dehydrogenase observation is the first mechanism that earns a place. Akagawa and colleagues reported that the molecule binds lactate dehydrogenase and accelerates oxidation of NADH to NAD+, with a corresponding push from lactate toward pyruvate. Related cycling has been described with other dehydrogenases. If that binding is real in tissue, the immediate consequence is a nudge of the NAD+/NADH ratio toward oxidation, at a cytosolic enzyme rather than at the chain. NAD+ is the co-substrate of the sirtuins. SIRT1 deacetylates PGC-1α, the coactivator that assembles the mitochondrial biogenesis program. SIRT3 works inside the organelle on the enzymes that make the gradient. A cofactor that regenerates NAD+ at a dehydrogenase is, in principle, feeding the same node that fasting, exercise, and nicotinamide riboside are said to feed, by a different route. Reports that the molecule also raises nicotinamide phosphoribosyltransferase, the salvage enzyme that rebuilds NAD+, belong in the same paragraph and at the same grade: suggestive, not a human tissue measurement. The bridge is the right kind of bridge. It is not yet a demonstrated one in people.

The mechanism the hypothesis stands on is the biogenesis program, and it has a necessary-pathway experiment behind it. Chowanadisai, Bauerly, Tchaparian, Wong, Cortopassi, and Rucker, in the 2010 Journal of Biological Chemistry paper, exposed mouse Hepa1–6 hepatocytes to 10–30 μM pyrroloquinoline quinone for 24–48 hours. Citrate synthase activity rose. Cytochrome c oxidase activity rose. MitoTracker signal rose. Mitochondrial DNA copy number rose. Cellular respiration rose. Upstream, cAMP response element-binding protein was phosphorylated at serine 133, the PGC-1α promoter was activated, and PGC-1α rose at mRNA and protein. Downstream, nuclear respiratory factors 1 and 2 and the mitochondrial transcription machinery — TFAM, TFB1M, TFB2M — moved in the expected direction. Small interfering RNA against CREB, or against PGC-1α, abolished the increase in citrate synthase, MitoTracker signal, and mitochondrial DNA. The CREB pulse was transient, largely gone within eight hours, which is what a signal should look like. The organelle change required a day or two, which is what a transcriptional program should look like.

There is no mammalian pyrroloquinoline quinone receptor in this account. The candidate input is the redox cycle itself. Cycling can generate a small amount of hydrogen peroxide. Peroxide in that range is a hormetic signal, the same class of signal contracting muscle uses, not an injury and not an antioxidant story. Nrf2 dissociation from Keap1 sits on the same class of input and accounts for a parallel arm, the antioxidant-response program, which is cytoprotection rather than biogenesis. A later disk-degeneration study reported Nrf2 nuclear translocation and offered a docking model against the Keap1 interface. That arm may be real. It is not the arm the knockdown defined. Calling the molecule an antioxidant inverts the order: the useful event, if the cell data hold, is a controlled oxidation that the nucleus reads as permission to print mitochondria.

What the molecule does not do should be stated at the same rank. It does not bypass Complexes I and III. It does not hand electrons to cytochrome c. It does not, on the timescale of minutes, restore proton pumping at Complex IV in a chain blocked by rotenone. The same hepatocyte paper reported protection against rotenone, 3-nitropropionic acid, antimycin A, and azide. That protection is consistent with more respiratory capacity and better redox buffering after a day of biogenesis signaling. It is not evidence of a shunt. A cell that needs electrons moved this afternoon will not get them from pyrroloquinoline quinone.

Empirical Evidence in Biological Systems

The deprivation experiments are the in vivo counterpart of the hepatocyte result. Work from Rucker’s group — including Stites, Bauerly, and Steinberg — found that a diet very low in pyrroloquinoline quinone, particularly across the neonatal and perinatal window, lowers hepatic mitochondrial content and the expression of mitochondria-related genes, and that repletion restores them. Older deficiency studies described slower growth, fragile skin, and poorer reproductive performance in mice. Those phenotypes are real in the papers and are not the claim of this essay. The claim is the mitochondrial one: inventory tracks the cofactor when the rest of the diet is defined. A molecule that changes mitochondrial number when it is removed and restored has cleared a bar most antioxidant supplements never approach.

Rodent models of lipotoxic and inflammatory liver stress, including work from Jonscher and colleagues, report attenuation of those phenotypes with supplementation. The direction is consistent with the biogenesis and NAD+ account. It is not independent proof of the CREB requirement, which remains a cell result. Exercise is the physiological agonist of the same PGC-1α axis. A synergy between training and the cofactor is plausible and has been proposed. It has not been shown as an added gain in mitochondrial copy number over training alone.

Across these systems the consistent theme is inventory, not flux. The molecule does not rescue a blocked chain in the way methylene blue does. It asks a competent nucleus, in a permissive energetic state, to build more of the chain. That request fails when CREB or PGC-1α is removed. It has not been shown to succeed in a human muscle.

Evidence in Human Systems

The human record is smaller than the marketing sentence, and it measures different things.

Harris and colleagues, in the Journal of Nutritional Biochemistry in 2013, studied ten adults in crossover. A single dose of 0.2 mg/kg was used for kinetics over 48 hours. A daily dose of 0.3 mg/kg — about 20 mg in a 70 kg adult — was given for three days. Standard clinical chemistries did not move. C-reactive protein and interleukin-6 fell in each of the ten subjects. Matrix metalloproteinase-9 tracked with them. Urinary trimethylamine N-oxide and other methylated amines fell, and the urinary metabolite pattern was read as consistent with a shift in mitochondria-related metabolism. Three days, no biopsy, no copy-number assay. It is the first systemic human signal, and it is an inflammatory and metabolite signal. It is not a demonstration that liver or muscle printed mitochondria.

Nakano and colleagues, in 2012, gave 20 mg of the disodium salt to seventeen adults in an open-label trial for eight weeks and reported improvement on all six subscales of the Profile of Mood States short form, together with fatigue, quality-of-life, and sleep questionnaires. There was no placebo arm. The trial is usable as a tolerability and symptom note. It is not evidence of biogenesis, and it is not a cognitive trial in the controlled sense.

Combination trials do not isolate the molecule. Itoh and colleagues, in 2016, reported memory and fatigue improvements with pyrroloquinoline quinone plus coenzyme Q10 in middle-aged adults. The trial was industry-linked and the intervention was two redox molecules. An effect there cannot be assigned to inventory signaling by pyrroloquinoline quinone alone.

The cleanest cognitive trial is Shiojima and colleagues, 2022, in the Journal of the American Nutrition Association. Sixty-four healthy Japanese adults aged 40 to under 80, who reported becoming more forgetful, were randomized to 21.5 mg/day of a disodium salt or placebo for twelve weeks. Fifty-eight completed. The active arm improved on Cognitrax domains of composite memory, verbal memory, reaction time, complex attention, cognitive flexibility, executive function, and motor speed, and on the DECO forgetfulness questionnaire and the Japanese Mini-Mental State Examination. No adverse events were recorded. The ingredient supplier was involved. Mitochondrial DNA, respiration, and citrate synthase were not measured. A cognitive change at twelve weeks is on a clock compatible with a transcriptional program. It is also on a clock compatible with sleep, inflammation, and expectation. The trial cannot say which.

Baltic and colleagues, in 2024, randomized thirty-four older adults with mild cognitive impairment to a mixture of 20 mg pyrroloquinoline quinone plus dihydrogen-producing magnesium, or placebo, for six weeks. Brain-derived neurotrophic factor rose in the active arm. The treatment-by-time interaction for that marker was not significant. An orientation subscale of the cognitive battery showed an interaction. Molecular hydrogen is itself an experimental redox intervention. The trial is not a pyrroloquinoline quinone trial.

The grade, stated once, is this. Biogenesis is shown in hepatocytes and inferred from deprivation and repletion in mouse liver. A human anti-inflammatory and metabolite signal exists at n = 10 over three days. A human cognitive signal exists in one twelve-week placebo-controlled trial of fifty-eight completers and in smaller open-label and combination studies, several of them run by the companies that supply the salt. No published randomized trial has shown the rise in muscle or leukocyte mitochondrial DNA copy number that the marketing sentence requires. Until that measurement is made, “pyrroloquinoline quinone grows mitochondria in people” is an extrapolation. The extrapolation is reasonable. It is not a result.

Safety is a separate ledger, and it is cleaner than methylene blue’s. The molecule is not a monoamine oxidase inhibitor. It is not a serotonergic precipitant. G6PD deficiency is not a described exclusion. The European Food Safety Authority in 2017 found no safety concern for a disodium salt at a proposed 20 mg/day in healthy adults, against a no-observed-adverse-effect level of 100 mg/kg/day in a 90-day rat study of one branded salt, a margin of exposure of about 340. Clinical studies reviewed by the panel ran as high as 100 mg/day for as long as 24 weeks without raising a safety concern. Pregnant and lactating women were excluded from the intended use. High-dose animal studies found urinary crystals and protein that informed the no-effect level; the human trials did not comprehensively assess kidney function. That is a reason for caution in known kidney disease. It is not evidence that 20 mg injures a healthy kidney. Loose stools and headache appear occasionally in the trial record and are not a limiting toxicity at the studied dose.

Inventory, Not Flux

Methylene blue and pyrroloquinoline quinone are both catalytic redox molecules, and that is where the resemblance ends. Collapsing them into “mitochondrial support” is the error this series exists to avoid.

Methylene blue occupies the flux layer. At low concentration it accepts electrons from NADH, or from the flavin of Complex I, and donates them to cytochrome c, bypassing damaged or congested segments of Complexes I and III. Proton pumping at Complex IV continues. The NADH/NAD+ ratio falls. Substrate-level phosphorylation at succinyl-CoA ligase can proceed when oxidative phosphorylation is compromised. The dose-response is hormetic, with a productive limb in the nanomolar to low-micromolar mitochondrial range and an inversion (Window C) when excess oxidized dye abstracts electrons from the chain. In this series, methylene blue is discussed as chronic catalytic support at low amounts (Window A: the 8–25 mg described in the methylene blue hypothesis), not the acute 0.5–4 mg/kg window used in cognitive studies (Window B). This is information, not a recommendation; any methylene blue use is physician-guided. The cost is real: monoamine oxidase A inhibition, a serotonergic interaction, a G6PD exclusion.

Pyrroloquinoline quinone occupies the inventory layer. It does not donate into the chain. Over a day or two, in a hepatocyte with intact CREB and PGC-1α, it increases the number and the respiratory capacity of mitochondria. It has no monoamine oxidase liability and no G6PD exclusion. It also has no Rodriguez-style acute imaging study, and no human biopsy. If methylene blue is a jumper cable on the chain already installed, pyrroloquinoline quinone is a work order to the nucleus. A work order does not start a stalled engine this afternoon.

They are stacked only if the deficit is at both layers: a chain that is bottlenecked now, and an inventory that is thin. A person whose problem is sleep debt and insulin resistance does not have a pyrroloquinoline quinone deficiency. A person whose problem is a measured Complex I failure does not have a biogenesis deficiency as the first lesion. Diet, light, sleep, and load remain first. Mitophagy is the constraint on the inventory claim. Printing mitochondria without clearing the failed ones increases mass and may not increase function. Pyrroloquinoline quinone is not a mitophagy signal. The fasting and circadian windows that drive clearance are still required, or the work order produces clutter.

Dose, Relative to Diet and to the Chain

The studied human range is 10–20 mg of the disodium salt per day. Twenty milligrams is the open-label Nakano dose, the Harris daily dose in a 70 kg adult, and the Baltic mixture dose. Shiojima used 21.5 mg. That is the range this hypothesis is about. It is not a milligram-per-kilogram pulse, and it should not be borrowed from the methylene blue acute window. The cell work used 10–30 μM in the medium for a day or two. Plasma and tissue concentrations after 20 mg oral in humans are not well described — the European assessment noted that absorption, distribution, and excretion data are limited — so the micromolar cell exposure should not be assumed to be the tissue exposure. The hypothesis does not require that it is. A transcriptional program can be tripped by a transient nuclear signal below the medium concentration of a hepatocyte experiment.

Food will not supply the dose. At the Authority’s estimate, 20 mg is at least 250 times background. Pretending that fermented soy is a clinical intervention is numerology. Intermittent dosing has no rationale here and was not how the trials were run. Daily, in the 10–20 mg range, is the exposure the human data actually describe. Climbing toward the 100 mg/day upper end of the safety reviews buys no demonstrated biogenesis and leaves the animal renal signal less comfortably far away.

The salt matters only as identity. Trials used specified disodium salts. A molecule this cheap does not need a proprietary story, and a proprietary story is not a mechanism. What is required is a characterized salt, not an aquarium-grade analogy — pyrroloquinoline quinone does not have methylene blue’s industrial-dye problem — and a decision not to treat a twelve-week cognitive questionnaire as a mitochondrial assay.

Timing within the Circadian and Training Window

Mitochondrial biogenesis is itself circadian. PGC-1α, NAD+/NADH, and the hours in which new respiratory machinery is permitted to accumulate all move with the clock. The timing question is simpler than it is for methylene blue, because this molecule is not an acute shunt and not a mild monoamine oxidase inhibitor.

The trials that reported anything gave it once daily, Nakano’s after breakfast. That is a compliance clock, not a mechanistic one. A defensible working placement is with a meal, away from the attempt to use it as a pre-session stimulant. Strength training is the physiological PGC-1α agonist, and the reactive oxygen species it uses as a signal are largely NADPH-oxidase–derived, not a pool this cofactor is known to quench at 20 mg. There is no evidence at the studied dose that it blunts the training pulse. There is equally no evidence that it adds copy number on top of a program already saturated by load. The informative user is not the athlete whose biogenesis is already driven, and not the insulin-resistant, sleep-restricted adult whose NAD+ and AMPK state will not run the program. It is the middle case: inventory thin, signal still intact.

These considerations are provisional, because there is no human pharmacokinetic series adequate to say what remains in muscle the next morning. Do not expect it to occupy a chain this afternoon.

Role within the EnSurf System

EnSurf organizes mitochondrial optimization into three practical pillars. Fuel supplies the right electrons — animal fats and proteins, fermented microbial accessible carbohydrates, creatine, exogenous ketones, periodic fasting. Circadian aligns cortisol, light, and sleep so the machinery is not asked to perform at the wrong hour. Strength comprises the interventions that change the capacity and efficiency of the machinery itself: resistance training, targeted mito-nutrient stacks, and agents that restore electron flow.

Pyrroloquinoline quinone is a candidate for Strength, on the inventory side of that pillar, not on the flux side. Load is the physiological work order. Fasting and sleep set the NAD+ and the mitophagy window in which a new unit is worth printing. Methylene blue remains the direct chemical intervention for a congested chain. Near-infrared light photo-activates the same cytochrome oxidase methylene blue supports chemically. Pyrroloquinoline quinone does not converge on Complex IV. It converges, if the hepatocyte result holds in people, on the decision to make more cytochrome oxidase. The combination is coherent only as layers: fuel, clock, load, clearance, then — if inventory is the named constraint — a cofactor that has asked a mouse liver for more mitochondria. It is a poor choice for acute energy, a poor choice as an antioxidant story, and a poor choice as an eye drug.

The retina is worth one paragraph and no more, because the temptation to borrow it from the methylene blue essay is obvious. The retina has the highest mitochondrial density of any tissue, and ganglion cells die when Complex I is poisoned or the gradient fails under pressure. Pyrroloquinoline quinone protects cultured cells against the same class of chain poisons. That is a reason to predict metabolic protection of a failing retina. It is not a drop in intraocular pressure, not a substitute for pressure control, and not a result in any human eye. An electroretinogram and a structural measure could falsify the prediction. Until they do, the eye is a hypothesis, not an indication.

Because the molecule is old, cheap, and, as a structure, unpatentable, it fits the frugal philosophy that underpins EnSurf. Because its primary action in the best experiment is a request for more of the machine that keeps the proton-motive force, rather than the forcing of a receptor, it sits comfortably next to the Experience Nature frame used in this series: health as graded coherence, metabolic dysfunction as progressive dissociation at the mitochondrial interface. Printing the organelle is association paid for in ATP. The cell spends energy to build the structure that makes energy. A bacterial cofactor that mammalian nuclei appear to read as permission is interesting exactly to the extent the permission can be measured. In a human muscle, it has not been.

Closing

Pyrroloquinoline quinone does not invent a new biology. It is a cofactor bacteria still use, which a mammalian nucleus can, under the conditions of one knockdown experiment, read as permission to transcribe the mitochondrial program. The permission is shown in a hepatocyte and inferred in a deprived mouse. It is not shown in a human biopsy. The cognitive questionnaires are real and small. The antioxidant sentence is a distraction. Twenty milligrams, the amount used by Nakano, in the Harris daily arm, and in the Baltic mixture, is enough chemistry for a fair test of the hypothesis, and little enough to remain frugal — provided it is not asked to do methylene blue’s job, and provided diet, light, sleep, and load are already in place. The oldest question in this series still governs the newest cofactor. Electrons move from food to oxygen, or they do not. A work order for more of the machine is not the same act as keeping the electrons on the machine one already has.

A working hypothesis, open to correction and revised as new evidence arrives. Last revised October 17, 2026.

Related: Strength · Biogenesis · Mitophagy · Methylene Blue · NAD precursors · CoQ10 · Coach — PQQ · Surfer — Kit · Sage — Fuchs’ Dystrophy

See the notice at the top of this page. Working hypothesis, not medical advice and not a dosing recommendation. Pyrroloquinoline quinone is not an approved drug. Any use requires individual medical evaluation. Milligram figures above are published-trial and safety-assessment citation context only. Methylene blue is discussed only as a comparison; any methylene blue use is physician-guided.