Pyrroloquinoline Quinone (PQQ): Redox Chemistry, CREB–PGC-1α, and the Limits of Clinical Evidence
From a bacterial quinone cofactor and cellular mitochondrial biogenesis to cardiac and brain models, small human studies, and toxicological limits.
From a bacterial quinone cofactor and cellular mitochondrial biogenesis to cardiac and brain models, small human studies, and toxicological limits.
Pyrroloquinoline quinone (PQQ, methoxatin) occupies an unusual space between bacterial enzymology and experimental mammalian biology. It is a genuine redox cofactor for several bacterial dehydrogenases; in cell culture and animal studies, PQQ exposure changes measures of mitochondrial function. These observations do not establish PQQ as a human vitamin, a universal intramitochondrial antioxidant, or a proven treatment for ischaemia, neurodegeneration, or metabolic disease. The distinction between chemical possibility, a measured cellular response, and a clinical outcome is essential for interpreting the literature that follows.
The history of PQQ began with bacterial dehydrogenases whose activity could not be explained by the familiar NAD(P)- and flavin-dependent schemes. Studies of methylotroph quinoproteins led to the isolation of a new low-molecular-weight cofactor; the structural work of Salisbury and colleagues established the name “methoxatin.” In methanol dehydrogenase, the cofactor is held in the protein’s active site with the participation of a metal ion and transfers reducing equivalents from the substrate to the physiological acceptor. This enzyme-bound environment is fundamentally different from free PQQ added to culture medium or taken orally: concentrations, redox partners, and competing reactions differ between these settings.
A report by Kasahara and Kato proposing PQQ as a cofactor for a mammalian lysine-catabolism enzyme prompted discussion of a “new vitamin.” A subsequent critical analysis by Felton and Anthony challenged the identification of that cofactor. Rodent dietary studies in which mitochondrial DNA content and some metabolic measures changed show that the model is biologically sensitive to diet composition; on their own, however, they do not establish an obligatory PQQ-dependent reaction, a deficiency syndrome, or a human requirement. “Vitamin B14” remains an unofficial name, and the “longevity vitamin” hypothesis is a framework for studying long-term nutritional effects, not recognition of PQQ as essential.
Bacteria synthesise PQQ, while plants and foods may contain it in small amounts. The absence of a known bacterial biosynthetic pathway in mammals is compatible with dietary intake, but does not prove that microbial PQQ is an evolutionarily established signal in nuclear–mitochondrial communication. This is an attractive hypothesis that would require evidence for the availability of the specific compound in tissues, its recognition, and a physiological response that depends on its in-vivo concentration.
PQQ has the molecular formula and a molar mass of about as the free acid. Its fused aromatic system contains an ortho-quinone moiety and three carboxyl groups. Positions C4 and C5 participate in two-electron reduction of the quinone to the hydroquinone form; a one-electron step yields a semiquinone intermediate. The proportions of protonated and deprotonated forms depend on pH and the surrounding medium, and the disodium salt differs from the free acid in mass and solubility. Structural formula, analytical concentration, and supplement mass therefore should not be conflated without specifying the chemical form.
Oxidised and reduced forms can interconvert repeatedly when electron donors and acceptors are present. But the number of cycles is a property of a precisely defined experiment with specified reagents and observation time, not an invariant “capacity” of the molecule in tissue. The original comparison of “20,000 PQQ cycles versus a few cycles for vitamins C and E” does not specify equivalent conditions and is not supported by the cited work by Paz as a universal biological constant. Ascorbate and tocopherol are also regenerated by coupled cellular systems, so ranking them in such a table is scientifically unsound.
Quinone chemistry is not the same as unconditional scavenging of free radicals. Under different conditions, electron transfer among PQQ, amino acids, proteins, and oxygen can accompany either oxidant reduction or the formation of reactive oxygen species. Reactions in cells depend on reagent concentrations, the rates of competing enzymatic pathways, localisation, availability of reducing agents, and metal-catalysed side reactions. Direct “scavenging” formulas for superoxide, the hydroxyl radical, and peroxynitrite should therefore not be treated as proof that oral PQQ reaches the relevant mitochondrial microdomain and outperforms superoxide dismutase or peroxiredoxins there.
Three carboxyl groups increase the molecule’s polarity and make its acid–base state important for solubility, protein binding, and membrane passage. Even if ionised forms predominate near neutral pH, the proportion of each microstate cannot be inferred without measured values in that particular medium; the source’s values of , , and should not automatically be treated as cytosolic constants. The quinone potential also depends on protonation: the Nernst equation relates the observed potential to the activities of oxidised and reduced forms and , rather than defining one universal for every compartment. This physical chemistry matters when considering possible pro-oxidant reactions: electron availability differs for free quinone, a bound cofactor, and a complex amino-acid adduct.
In a quinoprotein dehydrogenase, the protein orients the substrate, PQQ, and metal ion, then transfers electrons along the designated physiological route. Free PQQ lacks this catalytic scaffold. “Pseudoenzyme” is therefore only an analogy for a reaction in solution, not the identification of a new human enzyme. Even if an elementary rate constant for reaction with a radical is high, its contribution to cellular flux depends on compared with competing reactions catalysed by SOD, catalase, peroxiredoxins, and the glutathione system in the same volume.
| Level of observation | What can be concluded | What cannot be concluded |
|---|---|---|
| Bacterial quinoprotein | Protein-bound PQQ participates in catalytic electron transfer. | Free PQQ performs the same cycle in human mitochondria. |
| Reaction in solution | Quinone, semiquinone, and quinol forms are possible; rates depend on the medium. | A solution constant describes tissue protection after oral intake. |
| Cell culture | Exposure may change ROS, respiration, and gene expression. | A measured change proves disease prevention in people. |
The key study by Chowanadisai and colleagues was conducted mainly in mouse liver Hepa1-6 cells. Exposure to PQQ at – for – hours increased citrate synthase and cytochrome-c oxidase activity, Mitotracker staining, the mitochondrial-to-nuclear DNA ratio, and oxygen consumption. The authors observed CREB phosphorylation at Ser133, activation of the Ppargc1a promoter, and an increase in PGC-1α mRNA and protein. siRNA knockdown of either CREB or PGC-1α abolished induction of mitochondrial markers, providing substantially more than a simple correlation: in this model, both components were required for the response to PQQ. The experiment does not establish that an ordinary oral dose in humans produces comparable exposure in hepatocytes.
CREB is a transcription factor that, after phosphorylation at Ser133, can recruit the coactivators CBP/p300 in particular promoter contexts. The canonical cAMP–PKA–CREB scheme helps explain a possible upstream part of the signal, but observing phosphorylated CREB does not by itself prove direct binding of PQQ to adenylyl cyclase, a specific G-protein subunit, or the necessity of every intervening step. The model does not justify equating “CREB is activated” with an identified molecular receptor for PQQ. In the phosphorylation reaction, PKA uses ATP to transfer a phosphate group to a protein, and PQQ is not a substrate:
Timing also matters: an early change in transcription-factor phosphorylation, a later change in Ppargc1a mRNA, an increase in protein, and markers of altered mitochondrial content after a day represent different levels of a causal chain. If only endpoint markers after 24–48 hours are measured, the exact minute-scale kinetics of cAMP or adenylyl cyclase activity cannot be reconstructed. Establishing a direct pathway would require selective inhibitors or genetic disruption of the proposed kinase, quantitative measurement of the second messenger, and controls for nonspecific cellular stress. The source presents a cAMP spike at 10–15 minutes as established, although the cited Chowanadisai abstract confirms CREB Ser133 and dependence on CREB, not that time course.
PGC-1α, the product of PPARGC1A, does not bind DNA on its own. It coactivates nuclear transcription factors, including NRF1, GABP/NRF2, ERR, and PPAR; the set of partners varies by tissue and metabolic state. It is important to distinguish GABP, historically called “nuclear respiratory factor 2,” from NFE2L2/NRF2, an antioxidant-response factor: the same abbreviation “NRF2” refers to different proteins. In the cell study, PQQ increased NRF1 and GABP activity and expression of Tfam, Tfb1m, and Tfb2m. This supports a link between nuclear transcription and mitochondrial response, but does not mean that every respiratory-chain protein subunit was measured to increase.
The TFAM precursor is produced on cytosolic ribosomes, directed through the TOM and TIM complexes into internal compartments, and acts in the matrix after processing. It packages mitochondrial DNA into nucleoids and participates in transcription initiation with mitochondrial RNA polymerase POLRMT and TFB factors. The human mitochondrial genome encodes 13 oxidative-phosphorylation polypeptides: seven ND subunits of complex I, cytochrome b of complex III, three COX subunits of complex IV, and ATP6/ATP8 of complex V; complex II is encoded entirely by the nuclear genome. Increased Tfam expression and a higher mitochondrial-to-nuclear DNA ratio do not establish automatic increases in all 13 products, correct assembly of the complexes, or greater ATP output per oxygen molecule.
The import and assembly sequence should be distinguished by compartment. Nuclear-encoded TFAM crosses the outer membrane through the TOM complex and the inner membrane through the import machinery; its targeting peptide is then cleaved in the matrix, where TFAM interacts with mitochondrial DNA. The mitochondrial promoters LSP and HSP drive transcription of polycistronic precursors, which are then processed and translated by mitochondrial ribosomes. Functional complexes also require dozens of nuclear-encoded subunits, their import, assembly factors, iron–sulphur centres, haem groups, and a suitable lipid environment. TFAM and mitochondrial-DNA measurements are therefore an important but incomplete part of the evidence for biogenesis.
Cristae shape, cardiolipin content, MICOS, the balance of fusion through MFN1/MFN2/OPA1 and fission through DRP1, as well as PINK1/Parkin-dependent mitophagy, can influence mitochondrial-network quality. But the general physiological relevance of these structures is not direct evidence that PQQ altered each of these nodes in Chowanadisai’s experiment. “Biogenesis” should be described using the measures actually studied—enzyme activities, staining, mitochondrial DNA, and respiration—recognising that no single marker measures the completed assembly of functional organelles.
Electrons from NADH and FADH₂ enter the respiratory chain through complexes I and II, then pass through ubiquinone, complex III, cytochrome c, and complex IV to oxygen. Complexes I, III, and IV establish the proton electrochemical gradient used by ATP synthase to phosphorylate ADP. Under some conditions, reduced intermediates transfer an electron to oxygen and form superoxide; the sites and rate of this leak depend on substrates, membrane potential, oxygen conditions, and the direction of electron flow. A universal fraction of “0.2–2% of oxygen” does not describe all tissues and states. During reperfusion, high membrane potential alongside succinate accumulation can support reverse electron transport to complex I and a transient ROS burst. This is a specific reperfusion-model mechanism, not an inevitable effect of cellular ageing.
The proton-motive force combines electrical and chemical components: , given a specified gradient direction and temperature. A value near per pH unit applies only at a particular temperature and should not be inserted as an unconditional constant. Prolonged oxygen deprivation, calcium overload, and lipid oxidation can disrupt coupling, lower , and drive ATP synthase to operate in reverse. More mitochondrial markers, however, do not guarantee better coupling efficiency or lower ROS production.
Mitochondrial permeability transition (mPTP) is a calcium- and redox-sensitive increase in inner-membrane permeability that leads to matrix swelling, loss of , and, in severe injury, outer-membrane rupture. Cyclophilin D (PPIF) regulates the likelihood of transition; participation of ANT and ATP-synthase components remains under discussion, and the canonical “VDAC–ANT–CypD as one proven channel” scheme is too definite. Cytochrome c release into the cytosol can enable assembly of an apoptosome with APAF1, dATP/ATP, and procaspase-9, whereas severe energetic collapse favours necrotic injury. The outcome depends on stress intensity and residual ATP.
In isolated cardiomyocytes, Tao and colleagues showed that prior PQQ exposure attenuated the ROS signal and depolarisation caused by ; in a comparison experiment, cyclosporin A also prevented loss of potential. The result is consistent with an effect on processes associated with permeability transition, but does not prove direct binding of PQQ to cyclophilin D, preservation of particular ANT thiols, normalisation of NCLX, or “pore blockade” in humans. The source’s reported three- to four-fold shift in the calcium-opening threshold is not supported by the cited primary experiment and has been excluded. Cardiolipin and inner-membrane contacts also matter structurally, but their role in pathophysiology should not be turned into a measured PQQ effect without a separate experiment.
SIRT1 acts predominantly in the nucleus and cytosol, while SIRT3 acts in the mitochondrial matrix. Their deacetylase reactions depend on and yield nicotinamide and O-acetyl-ADP-ribose; SIRT1 can alter PGC-1α acetylation, while SIRT3 can affect some mitochondrial enzymes, including SOD2 and IDH2. Huang’s study in HepG2 cells reported changes in SIRT1/SIRT3 expression and activity after PQQ. Measuring sirtuins in this model does not establish that PQQ increases availability in humans, directly deacetylates every listed protein, or enhances the entire respiratory chain through one obligatory mechanism. A causal claim would require independent measurements of acetylation on specific substrates and pathway fluxes, plus genetic disruption of the proposed component.
A separate discussion of CD38 and NAD⁺ in the context of apigenin helps distinguish NAD⁺ abundance, sirtuin activity, and clinical effect: similar biochemical vocabulary does not make two supplements interchangeable.
During coronary ischaemia, oxidative phosphorylation falls, lactate and protons accumulate, and ATP and phosphocreatine are depleted. Impaired Na⁺/K⁺-ATPase, SERCA, and ion-exchanger function contributes to cytosolic and mitochondrial calcium overload. Reperfusion restores oxygen and normalises pH, but excess succinate and a disrupted electron-transport chain can produce a sharp redox shift; together with calcium, this facilitates permeability transition. This framework explains the targets chosen in experiments, but it does not turn every antioxidant into an effective treatment for myocardial infarction.
Zhu and colleagues studied rats after ligation of the left anterior descending coronary artery. One model involved two hours of ischaemia without reperfusion; another involved 17 or 30 minutes of occlusion followed by two hours of reperfusion. PQQ was given intraperitoneally before occlusion or intravenously at the start of reperfusion, not orally at a typical dietary dose. Under these conditions, the authors reported a smaller infarct area and improved early pressure measures. In the reperfusion model, the published ratios were 18.4 ± 2.3% and 25.6 ± 3.5% for prophylactic administration and administration at reperfusion, respectively, versus 38.1 ± 2.6% in controls. These values belong to the authors’ protocol and normalisation method; the source’s original figures of 55.4 → 26.1% do not match its abstract. The results do not prove lower mortality, improved long-term ejection fraction, or equivalence to clinical reperfusion therapy.
In Tao’s experiment, isolated adult rat cardiomyocytes were exposed to after pre-incubation with PQQ. The measured ROS, viability, and gave a consistent signal of cellular protection; however, acute toxic stress in a dish does not reproduce the vascular, immune, and haemodynamic components of infarction. A separate comparison with metoprolol in rats cannot be read as proven clinical superiority: the interventions differ in dose, route, and mechanism. The source’s claims about months-long preservation of LVEF, post-infarction remodelling, and particular LVEDP ranges are not supported by the cited short-term experiments and are not repeated here as established findings.
Early studies by Murase and Yamaguchi examined nerve growth factor (NGF) production in astroglial or other experimental systems after exposure to PQQ and its derivatives. NGF acts through TrkA and p75NTR, whereas BDNF is associated mainly with TrkB; downstream signalling can involve PI3K–Akt and MAPK–ERK survival and plasticity cascades. A cellular NGF response, however, is not equivalent to an established increase in NGF/BDNF in the human brain, restoration of basal forebrain cholinergic neurons, or treatment of polyneuropathy. The model and concentration context are particularly important for BDNF: an effect cannot be transferred automatically from NGF studies.
6-hydroxydopamine damages catecholaminergic cells through oxidative reactions and quinoid products, while rotenone inhibits complex I and provokes energetic and redox stress. Studies by Hara, Qin, and Martin describe neuroprotective signals from PQQ in cell cultures and toxin models, including measures of ROS, viability, and animal motor behaviour. Toxin models help dissect Parkinson’s disease mechanisms, but do not reproduce the full chronic human pathology; an outcome in these models cannot be translated into a promise to preserve substantia nigra neurons in patients. Direct binding to DJ-1, restoration of tyrosine hydroxylase, and particular behavioural effects must be tied to their individual experiments rather than combined into one universal pathway.
With recombinant α-synuclein, Kobayashi and colleagues observed inhibition of fibril formation after adding PQQ; Kim and colleagues studied amyloidogenic proteins, including Aβ, in vitro. Thioflavin fluorescence, electron microscopy, and structural methods describe aggregation in a test tube, but a lower signal from mature fibrils does not guarantee the absence of toxic soluble oligomers, much less microglial clearance. Claims of covalent protection of Aβ Lys16/Lys28, complete depolymerisation of existing plaques, and therapeutic effects in Alzheimer’s disease therefore require different evidence. Zhang’s cell model reported attenuation of Aβ toxicity, but does not establish the free PQQ concentration needed in the human brain.
Aizenman and colleagues showed an interaction between PQQ and a redox-modulated component of the NMDA-receptor response. Receptor cysteine residues alter channel sensitivity to the redox environment; excessive Ca²⁺ entry during glutamate excitotoxicity activates calpains, phospholipases, and nitric oxide synthase. This is a plausible experimental context, but does not confirm the source’s proposed exact stoichiometry for direct oxidation of a particular cysteine pair by one PQQ molecule, safe selectivity over physiological plasticity, or efficacy in stroke.
AMPK responds to cellular energy status and, through phosphorylation of ACC, reduces malonyl-CoA production; malonyl-CoA inhibits CPT1 and thereby regulates access of long-chain acyl groups to mitochondrial β-oxidation. In muscle, AMPK signalling can promote GLUT4 translocation through regulation of TBC1D4/AS160 and Rab proteins. PGC-1α, in turn, coactivates oxidative-metabolism programmes through PPAR and other factors. This physiological network explains which outcomes should be measured in studies of PQQ, but does not itself show that PQQ directly activates AMPK at Thr172, restores IRS1 sensitivity in people with diabetes, or increases clinical glucose utilisation. Such conclusions require tissue fluxes, metabolic challenges, and controlled studies in humans.
The enzymology of fatty-acid entry is covered in a separate article on carnitine palmitoyltransferase CPT1. Its mechanism provides a useful boundary for PQQ: a transcriptional hypothesis does not replace measurements of acyl transport and β-oxidation rates.
Mitochondrial stress may sustain inflammation through ROS, mitochondrial-DNA release, NF-κB, and the NLRP3 inflammasome. NLRP3 activation promotes caspase-1 maturation and processing of IL-1β/IL-18; inhibition of IκBα–NF-κB signalling could theoretically alter expression of many inflammatory genes. But changes in serum CRP or IL-6 in healthy participants cannot be attributed to direct suppression of NLRP3 in their microglia without relevant tissue data. Inflammaging describes a broad age-related phenomenon, not an established indication for PQQ.
The clinical evidence base for PQQ is much narrower than the range of preclinical models and varies substantially in design. Not every publication listed in the source can be called a randomised, double-blind trial: some studies are small, some are open-label, and some use combination products. Their endpoints include biomarkers, cognitive-test results, and self-reports—not prevention of myocardial infarction, dementia progression, or death. PQQ exposure, product form, control group, and study population should be considered separately when interpreting results.
Harris and colleagues studied a small group of healthy volunteers after a single dose and brief repeated intake of PQQ; the work on metabolomic and inflammatory markers was exploratory. Changes in some inflammatory measures and urinary metabolites were reported, but the absence of an adequately controlled clinical endpoint does not support a conclusion that PQQ treats chronic inflammation. A change in organic-acid excretion is not a direct measure of mitochondrial coupling efficiency or ATP synthesis in muscle. The source gives the wrong identifier for this study: the correct PMID is 24231099, DOI 10.1016/j.jnutbio.2013.07.008. Its exact claim of a “45% reduction in hs-CRP over 76 hours” is therefore not repeated as a reproducible clinical effect without confirmation from the protocol and a controlled comparison.
In the publication by Itoh and colleagues, 41 healthy older participants entered a randomised, double-blind, placebo-controlled study of PQQ disodium salt. Changes in selected cognitive measures in a sample of this size are hypothesis-generating and require attention to multiple testing, baseline values, effect size, and replication by an independent group. The brief primary abstract does not support every detail in the source about prefrontal haemodynamics and a specific “improvement in Stroop”; related NIRS experiments cannot automatically be folded into this trial. Nakano’s 2009 publication on PQQ and CoQ10 is considered separately: when a group receives a combination, comparison only with placebo does not identify the contribution of either ingredient or their interaction. Accordingly, a “2.2-fold synergy” is not an established effect of PQQ alone.
Nakano and colleagues also followed 17 adults after eight weeks of PQQ intake and reported changes in subjective sleep, fatigue, and mood scales. The work is a useful signal for planning a controlled trial, but the open-label design and self-reports are vulnerable to participant expectations and changes over time. The Pittsburgh Sleep Quality Index (PSQI) measures subjective sleep quality, not polysomnographic sleep-stage architecture; it cannot establish “deeper sleep” or normalisation of circadian rhythm. Specific PSQI, sleep-latency, LF/HF, and cortisol values in the source require checking against the paper’s tables and are therefore not repeated here as verified results.
| Primary study | Design and measured outcome | Permissible inference |
|---|---|---|
| Harris 2013 | Small, short-term group of healthy volunteers; inflammatory and urinary metabolomic measures. | A biomarker signal is possible; clinical benefit is not established. |
| Itoh 2016 | 41 healthy older participants, randomised placebo-controlled trial; cognitive tests. | Independent replication and assessment of multiple comparisons are needed. |
| Nakano 2012 | 17 adults, eight weeks, subjective scales without convincing placebo control. | A change in physiological sleep architecture cannot be claimed. |
Comparing these findings with melatonin’s mitochondrial and circadian mechanisms is useful precisely because the endpoints differ: shared terms such as “mitochondria” and “sleep” do not imply a common, proven therapeutic effect.
The bacterial biosynthetic pathway for PQQ begins with the short peptide PqqA, which contains Glu and Tyr residues. PqqD presents this precursor to PqqE, a radical-SAM family enzyme; formation of a C–C bond between the residues creates the basis of the future quinone scaffold. PqqB and PqqC participate in subsequent transformations, with the final step involving oxidative formation of the mature cofactor. The composition and organisation of gene clusters differ among bacteria; the shorthand “pqqA–E” describes a core scheme, not one industrial strain or a complete list of accessory genes. Production may involve fermentation, purification, and conversion to the disodium salt, but the source’s specific claims about chemical-synthesis yields, solvent composition, invariant purity above 99%, and an exclusive “gold standard” technology are not supported by its biochemical sources.
The distinction between the free acid and disodium salt matters when converting dose mass, assessing solubility, and comparing assays. A food-safety assessment applies to a defined ingredient, manufacturing process, specification, and intended intake level; it is not approval of PQQ as a treatment for cardiac or neurological disease. PQQ occurs in foods at low concentrations, but analysis is complicated by low abundance, protein adducts, and sample-preparation conditions. For humans, gaps remain in the mass balance for absorption, protein binding, tissue distribution, metabolites, and renal excretion across different chemical forms. The source’s specific figures of “60–80% absorption,” a “peak concentration at 1–2 hours,” a “half-life of 8–12 hours,” and its claim that PQQ crosses the blood–brain barrier are not supported here by clinical pharmacokinetic data with a clear method, so they are not carried into recommendations.
Nakano and colleagues studied BioPQQ disodium salt using acute administration, preliminary 14- and 28-day series, and 13 weeks of repeated oral dosing in rats. They reported signs of renal effects in some high-dose series in females, including changes in relative kidney weight, protein, and crystals in urine; these findings resolved during recovery. For the 13-week protocol, they reported a NOAEL of 100 mg/kg body weight per day, which was the highest dose tested in that series. The source’s statement that this particular study tested 100, 200, and 400 mg/kg conflicts with the primary abstract.
A NOAEL is specific to an animal species, chemical salt, duration, and set of measured outcomes. Allometric scaling can be one step in toxicological assessment, but does not provide a “55-fold clinical safety margin” without selecting uncertainty factors and considering chronic exposure and human data. Older reports of renal injury with other routes or higher doses should not be dismissed as a “myth”; it is equally unjustified to call PQQ a proven human nephroprotectant. The current clinical evidence cannot establish the safety of months-long high doses, use during pregnancy, use in renal impairment, or use alongside complex medication regimens. The source’s suggestion of monitoring every six months and a threshold of 40 mg/day lacks a validated protocol and is not presented here as a recommendation.
Physiological complementarity between PQQ and CoQ10 is plausible: ubiquinone transfers electrons between complexes I/II and III, whereas PQQ changes biogenesis-related signalling in a cell model. But it has not been shown that oral PQQ “doubles mitochondria,” directly reduces ubiquinone in the inner membrane, or that the combination clinically outperforms either compound in a factorial comparison. The same applies to stacks with acetyl-L-carnitine, resveratrol, NMN/NR, and α-lipoic acid: complementary pathways on paper do not determine the benefit, safety, or dose of a mixture. A separate article on acetyl-L-carnitine helps clarify acetyl-group transfer; it does not automatically transport long-chain fatty acids through CPT1, as the original “stack” claim suggested.
PQQ is a chemically defined bacterial redox cofactor and a useful experimental tool for studying cellular bioenergetics. The strongest mechanistic basis for the mitochondrial-biogenesis hypothesis comes from measurements in Hepa1-6 cells with CREB and PGC-1α knockdown; animal models of cardiac and nervous tissue add preclinical signals but use different routes of administration and injury contexts. Human data remain limited to small and inconsistently controlled studies with intermediate outcomes. Future work should connect measured tissue exposure to a target, confirm the effect in an independent factorial design, and prespecify clinically meaningful endpoints. Until then, neither “vitamin status” nor treatment protocols for myocardial infarction, dementia, or diabetes follow from the literature reviewed here.
This is a narrative review of PQQ based on a collection of 35 publications. In the 4 October 2026 editorial revision, selected key studies on the cellular CREB–PGC-1α axis and experimental cardioprotection, toxicology, and bibliographic discrepancies were checked individually. No systematic search under a predefined protocol, meta-analysis, or formal risk-of-bias assessment of each study was conducted; the reference list should not be considered exhaustive.
Redox chemistry, gene-expression changes in a cell line, and effects in animal models are considered separately from clinical outcomes. A culture concentration or an animal dose does not define an effective regimen for humans. When interpreting small and heterogeneous human studies, the population, chemical form, duration, and endpoint matter. Mechanistic coherence does not establish long-term clinical benefit or tissue pharmacokinetics.
PQQ is an established cofactor for several bacterial enzymes; an essential PQQ-dependent human enzyme and a deficiency syndrome have not been established.
The key CREB and PGC-1α experiment was conducted in mouse liver cells at a defined exposure. Comparable tissue effects after ordinary oral intake have not been demonstrated in humans.
A smaller injury area was shown in rats given PQQ parenterally. Clinical efficacy and a therapeutic dose after myocardial infarction have not been established.
There are small, inconsistently controlled studies using tests or self-reports. They do not demonstrate dementia prevention or a change in physiological sleep architecture.
No. It is limited to a specific chemical form, animal species, and study duration; separate data are needed to establish human safety and a specific dose.