NAD+: the currency of cellular time

NAD+ is nicotinamide adenine dinucleotide — a small molecule with a long name, and a job you already met if you ever drew respiration. Two nucleotides share a pair of phosphates: adenine on one side, a vitamin-B3 ring called nicotinamide on the other. Adenine is the same base DNA uses; nicotinamide is what vitamin B3 leaves behind. The interesting atom is carbon 4 of that ring. It can take a hydride — a proton and two electrons, written H− — and become NADH, then hand it back. That swap is the whole trick. The oxidised form is the hydride acceptor; the reduced form is the hydride donor. Almost every dehydrogenase you met in first-year biochemistry is a commentary on that single carbon. Otto Warburg and Arthur Harden were arguing about it a hundred years ago, long before anyone had a word for sirtuin. Molecular weight 663.43, formula C21H27N7O14P2, CAS 53-84-9, biologically the β anomer. Those numbers are here because this is a real chemical with a mass.

In short. NAD+ is a small helper molecule with a long name. It picks up electrons at one carbon of a vitamin-B3 ring, then hands them on.

The day job is still redox, and that is still most of what this molecule does, which I find quietly thrilling. Glyceraldehyde-3-phosphate dehydrogenase in glycolysis takes NAD+ and a phosphate and a three-carbon aldehyde and hands back NADH and a high-energy acyl phosphate. Malate dehydrogenase in the TCA cycle does the same exchange on a four-carbon acid. The acyl-CoA dehydrogenases of β-oxidation, the pyruvate dehydrogenase complex, isocitrate dehydrogenase 3, α-ketoglutarate dehydrogenase: hydride out, NADH in the pool. Complex I of the inner mitochondrial membrane is where most of that NADH is cashed. NADH:ubiquinone oxidoreductase, forty-five or so subunits in mammals, flavin and iron-sulphur clusters, a proton pump. The chain runs. Oxygen is the terminal acceptor at Complex IV. Peter Mitchell's gradient is the product. ATP synthase is the turbine. Whole-body, you turn over something like forty to sixty kilograms of ATP a day — a number Peter Rich put on paper in 2003, and a body-weight of phosphate. That is why a millimolar cofactor sitting on this path is how food becomes fuel you can spend.

In short. Most NAD+ work is ordinary metabolism: food electrons in, energy out. You recycle tens of kilograms of ATP a day on the back of that pool.

What changed, and the reason this cofactor escaped the textbook and landed in ageing institutes, is that three enzyme families consume NAD+ as a substrate rather than recycle it as a coenzyme. Sirtuins deacylate lysines — they take acyl badges off proteins — and spend one NAD+ per lysine. PARP1 polymerises ADP-ribose onto DNA-damage foci and can empty a millimolar pool in minutes. CD38 is a NADase, an enzyme that hydrolyses NAD+, whose expression climbs with age and with inflammation; Verdin's group made that the paper a lot of later work still stands on. Salvage through NAMPT tries to refill what those three spend. Nicotinamide riboside and nicotinamide mononucleotide feed the same pool from different rungs. Imai and Guarente put SIR2, then SIRT1, on caloric restriction. Brenner mapped the NR kinase route. Sinclair's NMN mouse papers exist, and the translational talk around them is contested, and we need both of those facts in view. What follows is the map of the pool: how it is spent, how it is rebuilt, and why mitochondria notice first.

In short. Some enzymes use NAD+ up instead of recycling it. Gene-silencing enzymes, DNA-repair enzymes, and an age-linked hydrolase all punch holes in the pool. Refill is a separate problem.

We stock freeze-dried β-NAD+, 1000 mg, at least 98% by HPLC, because that is the cofactor the sirtuin, PARP and CD38 papers actually weigh into a tube. We stock MOTS-c because it is a 16-mer a mitochondrion translated from its own 12S rRNA — Lee, Kim, Cohen, Cell Metabolism, 2015. Sharing an organelle is not sharing a mechanism: a dinucleotide and a mitochondrial peptide can live next door and still do different jobs. The freeze-dried cake is a reagent for a bench assay. Where an intramuscular appointment exists, that is a different product at eLIVEate, a different company, a different legal object. Intact NAD+ outside a cell is a charged molecule of 663 daltons, and a charge like that does not stroll across a plasma membrane. A drip and a weighed standard are different questions, and it helps to keep them apart. From here on we will stay with the biochemistry — how the pool is spent, how it is rebuilt, and why mitochondria notice first.

In short. The shelf holds freeze-dried NAD+ for the bench, and a mitochondrial peptide nearby. They share an organelle, not a mechanism, and the cake is for a tube.

Cozymase was the old name. Harden and Young, 1906, found a heat-stable fraction that alcoholic fermentation required in addition to the protein. Warburg isolated the nicotinamide nucleotide in the 1930s and showed that the hydride landed on the ring. For most of the twentieth century that was the whole story, and it was already a large story: every hydride-transfer dehydrogenase, the respiratory chain, the reason a cell has a measurable NADH fluorescence. You can still run a cuvette assay in which alcohol dehydrogenase and NAD+ make NADH and you watch 340 nanometres. The extinction coefficient is 6,220 M−1 cm−1. Generations of undergraduates have written that number in a lab book. It still earns its keep. It mattered because the pool is a reagent the cell cannot do without, present at hundreds of micromolar to low millimolar depending on the compartment, and because the ratio of the two redox states is a readout of whether the chain is keeping up with the dehydrogenases. That cuvette is not nostalgia. It is how we first learned to see the coin.

In short. For a century NAD+ was just the helper that moves electrons in metabolism. That job is still most of its work. A simple light reading at 340 nanometres still measures the reduced form.

Let's stay with the dehydrogenases for a minute, because they are why this pool exists at all. GAPDH is the glycolytic step that cannot run without NAD+. If the cytosol is reduced — high NADH, low NAD+ — GAPDH slows, glycolysis backs up, and the cell dumps pyruvate to lactate via lactate dehydrogenase in order to regenerate NAD+ in the cytosol. That is the Warburg observation as a nucleotide problem, rather than a poster about cancer. Malate dehydrogenase is the TCA step that is near equilibrium; the mitochondrial NAD+/NADH ratio sets how hard that step pulls. The β-oxidation spiral produces NADH and QH2 on every turn; a liver that cannot reoxidise them will stall fatty-acid oxidation and start making ketones or, worse, store the fat. Isocitrate dehydrogenase 3 and the α-ketoglutarate dehydrogenase complex are the other TCA hydrides. Pyruvate dehydrogenase is the gate from glycolysis into the matrix. All of them want oxidised NAD+. Complex I is how they get it back. The redox picture did not become untrue when sirtuins were cloned. It became the floor the signalling budget is built on.

In short. Enzymes in sugar and fat burning all need the oxidised form. If the pool is stuck reduced, those pathways stall. The newer signalling jobs sit on top of that older fact.

Leonard Guarente's laboratory, working in yeast, found that SIR2 — silent information regulator 2 — was an NAD+-dependent deacetylase, and that extra copies extended replicative lifespan. Imai, sitting in that lab, then in his own, showed that the mammalian orthologue SIRT1 used NAD+ the same way, and that the availability of the cofactor gated a chromatin and metabolic programme people had been describing, loosely, as caloric restriction. The 2000 Nature paper on SIR2 and NAD, the subsequent mammalian papers, the reviews Guarente and Imai kept writing as the field filled up: those are still the papers I would put in your hand first. Caloric restriction, in the organisms where it extends life, raises NAD+/NADH and often raises NAMPT. Sirtuins then have something to spend. That is a measured physiology. It is not the only chapter of restriction — AMPK, mTOR, insulin/IGF, autophagy all move — and treating sirtuins as the whole story flattens a serious enzyme family. Treating them as irrelevant is how you miss PGC-1α deacetylation and the FOXO stress programme. We can hold both of those thoughts at once.

In short. A yeast gene showed this cofactor also pays for gene silencing. Human versions need it too. Eating less can raise the pool.

The promotion to a signalling budget is a stoichiometric fact, not a metaphor, and that is the bit worth getting excited about. A dehydrogenase binds NAD+, takes a hydride, and releases NADH. The nicotinamide nucleotide is recycled. A sirtuin binds NAD+ and an acyl-lysine, transfers the ADP-ribose to the acyl oxygen, and releases nicotinamide plus O-acyl-ADP-ribose plus a deacylated lysine. The NAD+ is gone. A PARP binds NAD+, nicks it at the nicotinamide-ribose bond, and adds ADP-ribose to a protein or to a growing ADP-ribose polymer. The NAD+ is gone. CD38 hydrolyses NAD+ to nicotinamide and ADP-ribose, or cyclises it to cyclic ADP-ribose, a calcium messenger. The NAD+ is gone. Three families, three products, one shared cost. When the cost is paid on a timescale of DNA-repair minutes, or on a timescale of years of CD38-high immune cells, the mitochondria notice because the same pool feeds Complex I. That sentence is why a redox cofactor sits in ageing biology. The pool is a budget being raided by two kinds of enzyme.

In short. Recycling enzymes borrow NAD+ and give it back. Signalling enzymes spend it. Once spent, it has to be rebuilt. That rebuild is slower than a DNA-repair burst can spend.

The NADH/NAD+ ratio is therefore doing two jobs at once, and we have to keep those jobs separate or the measurement turns into fog. As a redox gauge it reports whether dehydrogenases are outrunning the chain. A typical mitochondrial matrix sits more oxidised than the cytosol; the malate-aspartate shuttle and the glycerol-phosphate shuttle are how those two pools talk without the dinucleotide itself crossing the inner membrane. As a signalling budget the absolute concentration of NAD+, not only the ratio, decides whether SIRT1, SIRT3 and PARP1 have substrate. You can imagine a cell with a healthy ratio and a small pool: redox fine, sirtuins hungry. You can imagine the reverse: a large, reduced pool, Complex I congested, PARP still able to spend. Assays that report 'NAD+' without saying total versus ratio, and without saying which compartment, are reporting a blur. The fluorescent biosensors — SoNar, Peredox, the genetically encoded NADH/NAD+ probes — exist because the blur was costing people papers. Use them, or use LC-MS on fractionated extracts, or say plainly that you measured a homogenate.

In short. The balance between the empty and loaded forms tells you if metabolism is keeping up. The total amount tells you if the spending enzymes still have enough to work with. Those are two different measurements.

Seven mammalian sirtuins, SIRT1 through SIRT7, three compartments. SIRT1 and SIRT6 and SIRT7 are nuclear; SIRT2 is largely cytosolic and can visit the nucleus in G2/M; SIRT3, SIRT4 and SIRT5 are mitochondrial. All of them are NAD+-dependent lysine deacylases, with some of the family moonlighting as ADP-ribosyltransferases — SIRT4 and SIRT6 in particular. The chemistry is the same spend: NAD+ in, nicotinamide and an O-acyl-ADP-ribose out, lysine deacylated. The acyl can be acetyl, succinyl, malonyl, long-chain fatty acyl, depending on the isoform and the paper. Nicotinamide is a product inhibitor, which is why a NAMPT block raises nicotinamide and quiets sirtuins from both sides of the reaction. The family is old. Bacteria have Sir2 homologs. You are looking at a redox nucleotide that was recruited, early, as a way for chromatin and metabolic enzymes to listen to the energy state of the cell. That recruitment is the promotion in the previous heading. The day job did not stop.

In short. Humans have seven NAD+-spending deacetylases, in the nucleus, the cytosol and the mitochondria. Each spent molecule is gone. The family is ancient. Energy state is what they are listening to.

SIRT1 is the isoform the caloric-restriction literature cannot stop talking about, and for once the attention is partly earned. Nuclear. Deacetylates histones (H3K9, H4K16 among the named sites), p53, FOXO transcription factors, PGC-1α, NF-κB subunits, a list that is long because a lysine deacetylase with that address will find clients. PGC-1α deacetylation is the mitochondrial-biogenesis sentence: SIRT1 takes the acetyls off, PGC-1α works with NRF1 and ERRα, nuclear genes for respiratory subunits and for TFAM get written, the organelle census can rise. FOXO deacetylation leans the cell toward stress resistance and, in some tissues, toward autophagy. p53 deacetylation quiets a pro-apoptotic programme. All of that is NAD+-gated. When the nuclear pool is thin, SIRT1 is a slower enzyme, not a missing gene. Pharmacological SIRT1 activators have been a messy literature — resveratrol as a direct activator did not survive contact with the assays — and we do not need to re-try that case here. The enzyme is real. The cofactor dependence is real. A 1000 mg cake of β-NAD+ is not a SIRT1 activator. It is the substrate the enzyme spends.

In short. SIRT1 lives in the nucleus and takes acetyl badges off proteins that run mitochondria and cell-death decisions. It can only work if NAD+ is plentiful.

SIRT3 is the mitochondrial deacetylase, and it is the isoform a respiratory-chain story actually needs. It deacetylates Complex I subunits, SOD2, long-chain acyl-CoA dehydrogenase, IDH2, a matrix neighbourhood of fuel and of superoxide. Knock it out and you get a hyperacetylated mitochondrial proteome and a mouse that mishandles fuel under stress. SIRT4 and SIRT5 share the organelle; SIRT5 prefers succinyl and malonyl over acetyl, which is why the word deacylase, not deacetylase, is the better family name. SIRT6 sits on chromatin, has DNA-repair and glucose-transcription jobs, and is one of the cleaner longevity orthologues when you look across mouse genetics — SIRT6-overexpressing mice have been reported to live longer, a claim that has survived more scrutiny than most. SIRT7 is nucleolar. The point of the census is not to memorise seven one-liners. It is to stop writing 'sirtuins' as if they were one enzyme in one place. Compartment is the variable. NAD+ supply in that compartment is the other variable. A homogenate NAD+ number does not tell you whether SIRT3 had substrate this morning.

In short. Different family members live in different rooms of the cell. The mitochondrial one tunes fuel burning and the superoxide mop. A blended measurement of NAD+ does not tell you which room was empty.

PARP1 is the other nuclear spend, and it is faster. Poly(ADP-ribose) polymerase 1 binds DNA breaks, nicks NAD+, and polymerises ADP-ribose onto itself and onto nearby proteins — histones, repair factors, a focus you can see with an antibody. The polymer is a flag. Repair machinery reads it. Then a glycohydrolase (PARG) takes it off. The NAD+ does not come back from that cycle as NAD+; it comes back, at best, as nicotinamide plus ADP-ribose, which salvage has to rebuild. A genotoxic hit can drop cellular NAD+ by millimolar amounts in minutes. Dawson, Berger, Kraus, the PARP literature is not subtle about this. In an ischaemic neuron the same burst is how PARP1 kills: NAD+ collapse, glycolytic stall at GAPDH, energy failure on top of the break. PARP inhibitors exist as oncology drugs because BRCA-deficient tumours need PARP1 to limp through replication. Those drugs also spare NAD+. That is a side-effect in a cancer trial and a tool in a metabolism lab. If you are about to claim a NAD+ restoration phenotype in a dish, you need to know whether you have a PARP1 problem, a salvage problem, or both.

In short. When DNA breaks, PARP1 spends NAD+ very fast to flag the damage. A bad enough hit can empty the pool in minutes. Repair then has to rebuild the cofactor from the leftovers.

CD38 is the age-associated NADase. Camacho-Pereira, Chini, Verdin, Cell Metabolism 2016, is the paper to put on the bench: CD38 expression rises with age in mice, the rise tracks the fall in tissue NAD+, and CD38-knockout animals keep more of the pool. The enzyme is a type II and type III membrane protein, ecto-facing in many of its incarnations, and it hydrolyses NAD+ to nicotinamide and ADP-ribose, or cyclises it to cADPR. Immune cells carry a lot of it. Inflamed tissue carries more. CD157/BST1 is a related NADase. The drain is chronic rather than a minutes-scale burst, which is a different experimental object from PARP1. A NAMPT activator or an NR supplement is trying to refill a bucket that CD38 is still holing. Some of the restoration literature reads more cleanly once you put a CD38 antibody on the blot. Verdin has been saying this out loud for a decade. The ageing-NAD+ story that talks only about making more, and skips the enzyme that is chopping, is a supply-side sketch of a supply-and-demand problem.

In short. CD38 is an enzyme that chops NAD+ up, and it becomes more common with age and inflammation. Refilling the pool while this enzyme is rising is pouring into a bucket with a hole.

A fourth drain deserves a sentence because axons made it famous. SARM1 is an NAD+ hydrolase that, once activated by a drop in NMN-consuming enzymes at an injured axon, collapses axonal NAD+ and kills the distal segment. Essuman, DiAntonio, Milbrandt, 2017, and the papers that followed: a TIR domain that turned out to be an enzyme. Wallerian degeneration is, on this reading, an NAD+ catastrophe with a named hydrolase. It is not the ageing-of-the-whole-animal story, and it is not how a whole animal ages. It is the cleanest demonstration that a NADase can be a cell-death pathway, and that the pool is not a vitamin you are a bit low on. It is a substrate whose sudden absence is lethal to a process. The three drains in the heading — sirtuins, PARP1, CD38 — are the ones the ageing literature actually argues about. SARM1 is the existence proof that the chemistry can be a fuse.

In short. In damaged nerve fibres another NAD+-cutting enzyme can empty the local pool and kill the fibre. Ageing of a whole body is a slower leak of the same chemistry.

Competing drains matter as much as salvage because the phenotype of 'low NAD+' is not one phenotype. A PARP1 burst is acute, nuclear-first, glycolytic-stall-adjacent. A CD38-high immune infiltrate is chronic, partly extracellular, and will eat precursor as well as the dinucleotide. A sirtuin programme running hard is a spend you might actually want — restriction, exercise, a genuine fast — and the refill through NAMPT is part of the adaptation. Measuring a lower NAD+ in aged tissue and writing 'give more' skips the interesting question. Measuring which drain is open, in which compartment, in which cell type, is how a paper earns its keep. FK866 inhibits NAMPT and will collapse the pool from the refill side; olaparib and its cousins close the PARP1 hole; CD38 antibodies and inhibitors exist in the literature and in the clinic for other reasons. Those are tools. Use them as tools. The diagram that follows is the bucket with three holes and a named bottleneck on the inlet.

In short. Low NAD+ is not one disease. A sudden DNA-repair spend, a slow age-linked leak, and a healthy spend during fasting are three different stories. Name the hole before you talk about the tap.

Mammalian cells build NAD+ three ways, and they do not use them equally. De novo synthesis starts from tryptophan, climbs the kynurenine path, and arrives at quinolinic acid, which QPRT turns into nicotinic acid mononucleotide. The Preiss-Handler path starts from nicotinic acid — the niacin of a flush — and meets the same nicotinic acid mononucleotide, then NAAD, then NAD+ via NAD synthetase. Salvage starts from nicotinamide, the product every consuming enzyme releases, and it is the workhorse in most tissues. NAMPT (nicotinamide phosphoribosyltransferase) transfers a phosphoribosyl group from PRPP onto nicotinamide and makes NMN. NMNAT isoforms adenylate NMN to NAD+. That is the loop the drains feed, and the loop that has to keep up. A liver has more de novo capacity than a neuron. A neuron lives on salvage. If you only remember one enzyme from this heading, remember NAMPT. The kinetic bottleneck of mammalian NAD+ salvage is a Km, a protein abundance, and a reason FK866 is lethal to the pool.

In short. Cells can make NAD+ from tryptophan, from niacin, or by recycling leftover nicotinamide. Most tissues live on the recycle path. The first enzyme on that path is the bottleneck.

NAMPT was cloned as a cytokine, pre-B-cell colony-enhancing factor, visfatin, a confusing decade of names, before Revollo, Imai and colleagues put it where it belongs: the rate-limiting enzyme of NAD+ salvage. Intracellular NAMPT (iNAMPT) is the one that matters for the pool in most cells. There is an extracellular form (eNAMPT) that has been argued over as a secreted enzyme and as a ligand; that argument is not settled enough to build a protocol on, so we will leave it open. What is settled is that NAMPT abundance and activity set how fast nicotinamide becomes NMN, that the enzyme is feedback-sensitive, and that a pharmacological block collapses NAD+ on a timescale of hours in a dish. AMPK phosphorylates and can stabilise NAMPT. Circadian clocks write the gene in some tissues — Ramsey, Bass, and the Northwestern papers on CLOCK/BMAL1 and NAD+. Restriction and exercise raise it in the tissues those papers actually measured. Age, in several depots, lowers it. Supply side. Demand side is CD38 and PARP. Both sides, always.

In short. NAMPT is the enzyme that turns leftover nicotinamide back into the next precursor. Block it and the pool falls in hours. Fasting, clocks and exercise can raise it. Age often lowers it.

NMNAT is the second step, and it is where compartmentation stops being a footnote. NMNAT1 is nuclear. NMNAT2 is cytosolic and Golgi-associated, and it is the isoform axons cannot do without — its loss is how NMN accumulates and SARM1 fires. NMNAT3 is mitochondrial, or at least that is the textbook assignment; the extent to which mitochondria make their own NAD+ from imported NMN versus importing NAD+ itself is a live experimental argument, not a closed diagram. Three genes, three addresses, one chemical reaction: NMN + ATP → NAD+ + PPi. A nuclear PARP1 burst is, among other things, a local NAMPT-NMNAT1 problem. A mitochondrial SIRT3 programme is a local NMNAT3-and-import problem. Adding NMN to a medium is not the same experiment as adding NAD+, and neither is the same as overexpressing NAMPT. If your blot cannot say which NMNAT you have, you are not yet doing compartmental salvage. You are working in a mixed soup, and that is a starting measurement rather than an address.

In short. The second recycling enzyme comes in three postcodes: nucleus, cytosol, mitochondrion. Which postcode you fill decides which job gets paid. Dumping precursor into the medium is not the same as filling one room.

Nicotinamide riboside is the Brenner contribution, and it is the cleanest precursor story in the last twenty years. Charles Brenner identified NR as a vitamin, mapped the nicotinamide riboside kinases NRK1 and NRK2, and showed that NR is phosphorylated to NMN and then adenylated to NAD+. Trammell, Brenner, Nature Communications 2016: oral NR raises the human NAD+ metabolome, with nicotinic acid adenine dinucleotide as a distinctive marker. The chemistry is tidy. NR is uncharged relative to NAD+, it uses nucleoside transporters, and it bypasses NAMPT, which is the point if NAMPT is the bottleneck you are trying to walk around. NRK2 is enriched in muscle, which is why some of the muscle papers look better than the whole-body papers. NR is also in milk, which is a fact and not a marketing department. What NR is not is a guarantee that a clinical endpoint will move once the metabolome has. That gap is the next heading. Hold the chemistry here: NR → NMN → NAD+, NRK, bypass of NAMPT.

In short. Nicotinamide riboside joins the recycle path after the bottleneck enzyme. It changes NAD+ metabolites in people. A disease outcome is a separate question.

NMN is one rung down, and it is the Sinclair-adjacent molecule, which means you have to read more carefully. Yoshino, Mills, Imai, Cell Metabolism 2011: NMN treated diet- and age-induced diabetes in mice, insulin sensitivity, a mitochondrial neighbourhood. Mills, Yoshino, Imai, Cell Metabolism 2016: longer NMN administration in mice, vascular and metabolic endpoints. Grozio, Imai, Nature Metabolism 2019: Slc12a8 as a NMN transporter, a claim that has been disputed as well as cited, which is how a transporter paper ought to age. Extracellular NMN can be dephosphorylated by CD73 to NR, then imported as NR, which is a competing route and a reason a 'NMN receptor' sentence is never free. The mouse work is large. Insulin sensitivity, endothelial function, stem-cell maintenance in specific niches: those are the reported assays, with the usual caveats of strain, dose, endpoint and the lab that ran them. Translating that literature into a human capsule is a further step, and how the molecule enters a human cell is still an open experimental question. The chemistry remains: NMN is the NAMPT product and the NMNAT substrate.

In short. NMN is the middle rung of the recycle ladder. Mouse papers on insulin and blood vessels are real. How the molecule enters cells is still argued.

Intact NAD+ as a precursor is the awkward object, and it is the object on the shelf, so the awkwardness has to be said out loud. The dinucleotide is charged, about 663 daltons, and does not stroll through a plasma membrane. Connexin 43 hemichannels have been proposed as a route in some cells. CD38 and CD73 on the cell surface will eat extracellular NAD+ and leave nicotinamide, ADP-ribose, NMN, NR — a cloud of smaller pieces some of which can then enter. Adding β-NAD+ to a medium is therefore, in many dishes, an experiment about ectonucleotidases and salvage, not about NAD+ going in as itself. Isolated mitochondria, permeabilised cells, and in-vitro enzyme assays are the settings where the intact cofactor is the ligand you think it is. A lyophilised 1000 mg cake is a reagent for those settings, and for any assay that needs a weighed, HPLC-characterised dinucleotide. It is not a cleverly packaged NR. The next-but-one heading is the sentence about infusions. The chemistry belongs here: precursor identity is a mechanism, not a brand.

In short. Whole NAD+ is charged and does not easily cross a cell membrane. Surfaces often chop it first. In a test tube the intact molecule is what you weigh out.

Complex I wants oxidised NAD+. That is a structural statement. The flavin of NADH:ubiquinone oxidoreductase oxidises NADH, the electrons walk a chain of iron-sulphur clusters, ubiquinone is reduced, four protons are pumped. If NAD+ is scarce, dehydrogenases cannot mint NADH, and Complex I has nothing to oxidise. If NADH is plentiful and the chain is blocked, the matrix sits reduced, superoxide leak at I and III rises, and the TCA cycle backs up at the NAD+-dependent steps. Either failure shows up first in the tissues that already run the organelle hardest: heart, brain, skeletal muscle, brown fat when it is on, kidney cortex. A hepatocyte will notice too, but a hepatocyte has more glycolytic latitude. The ageing-NAD+ literature keeps finding mitochondrial phenotypes — oxygen consumption, membrane potential, PGC-1α targets, acylcarnitines — because the organelle is a NAD+ customer twice: once as Complex I, once as SIRT3. Two bills, one pool, or at least one pool that is not freely mixed with the cytosol.

In short. The first machine of the respiratory chain spends the loaded form of NAD+. Tissues that live on mitochondria feel a shortage first. Heart, brain and muscle are the obvious ones.

The mitochondrial NAD+ pool is not in free equilibrium with the cytosol. The inner membrane does not let the dinucleotide cross as such. How the matrix is topped up is still being mapped: NMN import, a dedicated NAD+ transporter in some species (SLC25A51 / MCART1 is the mammalian candidate, Kory, Mootha, Nature 2020, and independent papers the same year), NMNAT3 on the inside. The malate-aspartate shuttle moves reducing power without moving NAD+. So does the glycerol-phosphate shuttle, dumping electrons on Q and skipping Complex I. A cytosolic PARP1 burst can starve the nucleus and the cytosol without immediately emptying the matrix; a CD38-high extracellular drain can starve the precursors before any compartment sees them. Compartmentation is why a whole-cell NAD+ kit is a starting measurement and not a mitochondrial paper. Digitinin fractionation, isolated mitochondria, matrix-targeted biosensors: those are the tools. Use them if the word mitochondrion is going in your title.

In short. Mitochondria keep their own NAD+ stash, behind a membrane the molecule cannot freely cross. A whole-cell average can look fine while the organelle is hungry, or the reverse.

Peter Rich, Biochemical Society Transactions, 2003, is the citation for a number this page keeps using because it is the right size. A human adult turns over on the order of forty to sixty kilograms of ATP per day. The standing pool is about fifty grams. Recycled, not stored. That implies something like 10²¹ hydrolyses a second, while you sit there. Almost all of that phosphate is minted on the inner membrane, which means almost all of it is downstream of NADH oxidation at Complex I or of FADH2/QH2 entry at II and the Q-cycle. NAD+ is not the ATP. It is the hydride coin the chain wants oxidised so that the chain can pump the protons the turbine spends. A 1000 mg cake of β-NAD+ is about 1.5 millimoles, about 9 × 10²⁰ molecules, coincidentally near a second of that flux as a count, and not a second of that flux as a physiology, because the cake is not in the matrix and the molecule does not become ATP. Scale is why Rich belongs next to a 1000 mg listing. A gram of cofactor is not a day's phosphate. It is a standard you can weigh.

In short. You recycle roughly a body-weight of ATP every day. NAD+ is the electron-carrying coin mitochondria need to make it. A gram in a vial is not a day's energy.

A drained mitochondrial pool has a look, and the look is not mystical. Oxygen consumption falls because Complex I is underfed. Membrane potential sags. TCA intermediates rearrange; citrate and α-ketoglutarate are often the ones people quote. Fatty-acid oxidation stalls and acylcarnitines accumulate. Superoxide rises, not always, but often enough that SOD2 acetylation — a SIRT3 substrate — shows up on a blot. PGC-1α stays acetylated if the nuclear pool is thin too, so biogenesis does not answer the failure. mtDNA, sitting next to a leakier chain with a thinner repair budget, picks up damage faster than the nucleus does; that is the neighbouring mitochondria essay, and it is why a NAD+ problem and a second-genome problem are allowed to travel together without being the same problem. None of this lets us write 'NAD+ restores youthful mitochondria' as a product sentence. It does let us measure OCR, Δψ, acylcarnitines, SIRT3 clients and a NAD+ number in the same experiment. If you only have the last, you have a metabolome, not a mitochondrion.

In short. When mitochondrial NAD+ runs low, fuel burning slows, the voltage sags, fat oxidation stalls and the superoxide mop can lag. Measure those things together. A NAD+ number alone is not a mitochondrion.

MOTS-c is the other catalogue object in this neighbourhood, and the neighbourhood has to stay a neighbourhood. MRWQEMGYIFYPRKLR, sixteen residues, translated from an open reading frame in mitochondrial 12S rRNA — an RNA that was supposed to be a ribosome, not a message. Lee, Kim, Cohen, Cell Metabolism 2015: metabolic homoeostasis in mice, AMPK, the folate–methionine cycle. Kim, Lee, Cell Metabolism 2018: nuclear translocation under metabolic stress. Reynolds, 2021: exercise-induced, age-dependent physiology. AMPK is a fuel-gauge kinase that phosphorylates ULK1 and TSC2 and acetyl-CoA carboxylase; it is one door away from autophagy and from mitochondrial biogenesis via PGC-1α. Sitting on AMPK is not sitting on Complex I, and it is not sitting on NAMPT. A 16-mer written inside the organelle and a dinucleotide the organelle spends are two different jobs. We stock both because the papers are real. We will not write them as a protocol, and we will not write them as each other.

In short. MOTS-c is a short peptide the mitochondrion writes from its own RNA. It talks to a fuel-gauge kinase. That is next door to NAD+, not the same molecule. Two jobs.

The mouse literature is large enough to be a field, and it is not a cartoon. Yoshino, Mills, Yoon, Imai, Cell Metabolism 2011: NMN, diet- and age-induced diabetes, insulin sensitivity, a mitochondrial set of readouts. Gomes, Shin-ichiro Imai and colleagues, and the Sinclair laboratory's 2013 Cell paper on NMN and mitochondrial communication in ageing muscle — the paper that put the 'NAD+ restoration' sentence into the bloodstream of the popular press, with a PGC-1α/SIRT1/HIF-1α story that other labs have both used and argued with. Mills, Yoshino, Imai, Cell Metabolism 2016: twelve months of NMN in mice, body weight, energy expenditure, insulin sensitivity, plasma lipids, eye function, a panel rather than a single heroic endpoint. Das, Sinclair, Cell 2018: NMN and capillary density, SIRT1 in endothelium, an exercise-adjacent vascular phenotype in old mice. Fang, Bohr, and colleagues: PARP1, NAD+, and DNA-repair neighbourhoods in Cockayne and xeroderma models, where the drain is the point. Canto, Auwerx: NR in mice, mitochondrial unfolded-protein response, a different lab, an overlapping metabolome. You can set a press release aside and still have to cite the figures.

In short. In mice, NMN and nicotinamide riboside have improved insulin handling, blood-vessel density and some mitochondrial readouts in named papers. That literature is real. It is also mouse work.

David Sinclair's NMN mouse work exists. The translational talk around it is contested. We need both sentences in view, and neither one cancels the other. The 2013 Cell paper, the later Science and npj papers, the company-adjacent interviews, the book: a serious enzyme family and a serious cofactor were asked to carry a public story about reversing ageing that the human data have not yet earned. Other serious people — Imai, Brenner, Verdin, Auwerx, Baur — have kept publishing, often more quietly and with more attention to which drain is open. What we can usefully say is simpler than a personality dispute. The mouse phenotypes are in print; the mechanisms (SIRT1, SIRT3, PARP1-sparing, CD38 as the age-NADase, NAMPT as bottleneck) are in print; the leap from a C57BL/6 mouse drinking NMN to a human clinical claim is a leap, and leaps are where fields get sloppy. Read the figures. Read the doses. Read whether the control was the right drain.

In short. One laboratory made NAD+ restoration famous, and the publicity ran ahead of the human evidence. The mouse figures are still in print. Publicity is not a mechanism.

Human NR trials are smaller and more mixed, which is what you should expect when you perturb a node this central in an animal that also has a diet, a sleep schedule and a CD38-positive immune system. Trammell, Brenner, Nature Communications 2016: oral NR is bioavailable, the NAD+ metabolome moves, NAAD is a marker. Martens, Seals, Nature Communications 2018: NR in healthy middle-aged and older adults, blood pressure and aortic stiffness as the endpoints that moved in that cohort, a modest study, a real vascular conversation. Elhassan, Philp, Cell Reports 2019: NR in aged men, skeletal-muscle NAD+ metabolome up, a set of circulating inflammatory cytokines down, mitochondrial function in muscle not dramatically remodelled. Other trials have moved the metabolome and left VO2 max, insulin sensitivity and strength where they were. That pattern is information. You can fill a pool and not own the phenotype the pool sits under. The tables are more interesting than the headlines, and they are where I would start.

In short. In people, nicotinamide riboside often changes NAD+ metabolites. Blood pressure moved in one study. Fitness and insulin numbers often have not.

Yoshino, Yoshino, Imai, Klein, Science 2021 is the human NMN paper we actually have to cite, because it is a controlled trial rather than a metabolome anecdote. Prediabetic women, postmenopausal, NMN for ten weeks, hyperinsulinaemic–euglycaemic clamps — the gold-standard insulin-sensitivity assay, not a fasting insulin number. Skeletal-muscle insulin sensitivity improved; liver and adipose insulin sensitivity, in that study, did not. The effect was modest, the cohort was specific, and the paper did not claim to have reversed ageing. That is what honesty looks like at this node. Other human NMN studies are smaller still, some industry-adjacent, some reporting NAD+ metabolome changes and some reporting well-being scores that a clamp would not recognise. Dose, duration, sex, menopausal status, baseline NAD+, and whether anyone measured CD38: the covariates are the story. We sell β-NAD+ for the bench, so the useful move is to read this trial as written rather than to stretch it into a protocol.

In short. A 2021 human NMN trial used a proper insulin clamp in prediabetic women and found a modest improvement in muscle, not in liver or fat. Specific cohort, modest effect, no claim to have reversed ageing.

Why a central node can move without owning the phenotype is not a mystery. NAD+ sits under redox metabolism, under sirtuin programmes, under PARP repair, under CD38-positive immune cells, under circadian NAMPT, under whatever the animal ate and whether it slept. Raise the pool and you have changed a denominator. The numerator is still the drain, the tissue, the receptor sheet, the mitochondria you actually have, the mtDNA heteroplasmy, the insulin receptor, the endothelial nitric-oxide synthase. Mouse rooms are 22 °C, which is cold for a clothed mouse and a chronic brown-fat stimulus; human rooms are not. Mouse NMN doses, scaled by body-surface area or by the number someone picked, are often higher than a capsule a person swallows. Strain is a variable. Microbiome is a variable, and some NAD+ precursors are also microbial currency. The restoration literature is at its best when it names those. It is at its weakest when it treats a metabolome shift as a clinical endpoint. We will stay with the best of it.

In short. NAD+ sits under too many jobs for one top-up to fix a whole body. Mice are not people. A changed metabolite is not a changed life.

Pellagra is the historical control, and it belongs in a restoration heading so that niacin does not get forgotten. Goldberger, the South, a diet of maize, four Ds: dermatitis, diarrhoea, dementia, death. Nicotinic acid and nicotinamide cure pellagra because they refill a NAD+ pool that was empty from the dietary side. That is a deficiency disease, a vitamin, a public-health victory. Ageing is not pellagra. Aged tissue is not a population with no niacin. The pool falls, in specific tissues, for reasons of drain and of NAMPT, against a dietary background that would have kept Goldberger's patients well. Ageing-NAD+ is not a vitamin deficiency in that sense, even though NR and NMN do move human metabolomes. The clinical gap is the gap between pellagra and a clamp in a prediabetic woman. Hold both ends. Niacin cured a true empty tank. Restoration here is trying to top up a tank that is leaking for other reasons.

In short. Niacin cures pellagra because that is a true vitamin deficiency. Ageing is not pellagra. The pool can still sag in old tissue for other reasons, and topping it up is not the same victory.

The catalogue listing is lyophilised β-NAD+, 1000 mg, ≥98% by HPLC, molecular weight 663.43, formula C21H27N7O14P2, CAS 53-84-9. A white to off-white cake in a vial, a certificate of analysis, a reconstitution kit. It is the same carbon skeleton every dehydrogenase, sirtuin and PARP paper names. It is a laboratory reagent. The form is the form you weigh into a sirtuin assay, a PARP assay, a cycling assay, a set of isolated mitochondria, a standard curve for LC-MS. Purity is a chromatogram, not a feeling. Water, light and freeze–thaw are the enemies of a nicotinamide nucleotide in solution; the lyophilised solid is how you store it. Those sentences are not a dose, a route or a schedule. If you cannot yet say how many nanomoles went into a well, the experiment has not started. The 1000 mg on the label is how much cake is in the vial, and the legal class of the object is a research solid.

In short. The vial is freeze-dried NAD+, a gram of it, purity on a chromatogram, for weighing into experiments. That is a laboratory chemical, not a dose or a schedule for a person.

Pharmacokinetics of intact NAD+ are why the cake and the appointment are different questions. Charged, 663 daltons, poor passive permeability. Ectonucleotidases on endothelium and on immune cells. CD38 facing out. CD73. Connexins as a disputed door. Plasma half-life of an intravenous bolus is short; what you measure downstream is often nicotinamide, NR, NMN, ADP-ribose, a cloud, not a tidy rise in mitochondrial NAD+ in the tissue you cared about. Intramuscular administration is a different curve again, a depot, a local ecto-enzyme landscape, a set of unknowns a clinic has to own as a clinic. Oral NAD+ is, on the chemistry, mostly a precursor experiment by the time the intestine and the liver have seen it. The research solid does not inherit any of those curves. It was not formulated to. It was freeze-dried to be a standard. If you want a precursor with a published human metabolome, that literature is NR and NMN, oral, and it is not this listing. If you want an enzyme assay, this listing is the cofactor.

In short. Intact NAD+ in blood is quickly chopped and poorly admitted into cells. A freeze-dried standard for a tube does not inherit a drip's pharmacokinetics. Different object, different question.

eLIVEate is a separate company. Where an intramuscular NAD+ appointment exists on that diary, it is a supervised administration of a clinic-compounded or licensed preparation, with a consent form and a clinician. Patriot Peptides does not take a commission on that booking, does not write the appointment as a reconstitution of the catalogue cake, and does not pretend the two legal objects share a pharmacokinetic file. Same carbon skeleton, in the sense that β-NAD+ is β-NAD+. Different product, different company, different regulator's attention. Adults can want both a paper and an appointment. Keeping them unmixed is how a research reagent stays a research reagent, and how a clinic stays a clinic. The physiology in the headings above does not depend on this paragraph. The label on the vial does, because it is the legal class of the object you are holding.

In short. A clinic injection of NAD+, where it exists, is a different product at a different company. The catalogue cake is the laboratory solid. Same molecule on paper. Different law in the room.

Two mitochondrial reagents on the same shelf, two certificates, two questions. The 1000 mg NAD+ cake is a dinucleotide for redox, sirtuin and PARP work. The MOTS-c listing is MRWQEMGYIFYPRKLR, HPLC-characterised, a 16-mer for AMPK and one-carbon neighbourhoods. Retatrutide, if it is on your bench, is occupancy at GLP-1R, GIPR and GCGR, and organism-level fuel demand then changes; mitochondrial flux will follow because flux follows fuel, not because a triple agonist is a Complex I ligand. Three floors, three named objects. We will sell you those objects. We will not write a protocol that pretends a gram of cofactor is a 16-mer or a weekly incretin. Neighbourhood, here, is a courtesy on a reading list. It is not a combination claim. Three named jobs, three named tools. Name the receptor, the cofactor or the reading frame before you stack them, and the experiment stays honest.

In short. NAD+, a mitochondrial peptide, and a triple gut-hormone agonist live near energy biology. They occupy three different things, and they are not one juice.

Decide what you are measuring before you thaw the cake. Total NAD(H) is a pool size. The NADH/NAD+ ratio is a redox state. They can move in opposite directions. A cycling assay (alcohol dehydrogenase, or a commercial kit built on the same idea) reports a number from a homogenate and is only as good as the extraction; acid extraction destroys NADH and keeps NAD+, alkali the reverse, and people have been getting that wrong since the 1960s. LC-MS, with 13C internal standards, is how a metabolome paper earns the word metabolome: NAD+, NADH, NADP+, NADPH, NMN, NR, nicotinamide, NAAD, ADP-ribose, a family. NADPH is not NADH. The extra phosphate on the adenosine ribose is a different pool, anabolic, pentose-phosphate, glutathione reductase, and mixing the two is a way to ruin a figure. Biosensors — SoNar, Peredox, LigA-based probes, matrix-targeted variants — give you ratio in a living compartment, which is the measurement the homogenate could not make. Pick one, and write which one.

In short. Say whether you are measuring how much NAD+ there is, or how much of it is loaded with electrons. Those can move apart. Do not mix it with NADP+.

Pharmacological controls are how you name the hole. FK866 (daporinad) inhibits NAMPT; the pool falls from the refill side, typically over hours, and a rescue with NR (which bypasses NAMPT) is the specificity check. A PARP inhibitor — olaparib is the one a hospital already knows — closes the PARP1 drain; if your NAD+ collapse was a genotoxic burst, this drug will spare the pool and change the interpretation. CD38 inhibitors and blocking antibodies exist in the literature; 78c is a named small molecule from the Chini neighbourhood. Nicotinamide at high millimolar is a product inhibitor of sirtuins and a precursor at the same time, which is why it is a clumsy single-handed tool and a fine reason to be careful. Sirtuin inhibitors (EX-527 for SIRT1, 3-TYP for SIRT3) close a spend without refilling the pool. If a phenotype survives FK866 but dies with a sirtuin inhibitor, you were never looking at pool size. You were looking at an enzyme. That is a better paper.

In short. Use drugs that block refill, DNA-repair spending, or the age-linked chopping enzyme, so you can tell which hole you are studying.

Machines, named, because 'we measured NAD+' is not yet a methods line. A plate reader at 340 nm is the undergraduate cycling assay. A mass spectrometer — Q Exactive, TQ-XS, whatever your core actually has — with a HILIC or a porous-graphitic-carbon method is how you separate the family. A Seahorse XF (or an Oroboros O2k, if you are serious about isolated mitochondria) is oxygen consumption, the Complex I invoice. A Clark electrode still works. TMRM or rhodamine dyes for membrane potential, with FCCP and oligomycin as the brackets. A Clark-type or a fluorescent ROS probe if you are going to mention superoxide, with the usual caveats that DCF is a rumour and MitoSOX is only as good as your controls. Westerns for SIRT1, SIRT3, NAMPT, CD38, PAR, PGC-1α, acetyl-p53, acetyl-SOD2. qPCR for the PGC-1α targets (TFAM, COX4I1, NDUFS1) if biogenesis is the claim. None of that is glamorous. All of it is how you keep a cofactor story attached to a measurement. The cake on the shelf is for the assays that need the standard.

In short. Name the machine. A mass spectrometer for the metabolites, an oxygen-consumption instrument for the mitochondria, blots for the enzymes. The vial is a standard for those assays.

Compartmentation is the variable most NAD+ papers under-report, and it is the variable that decides whether you have a SIRT3 story or a PARP1 story. Digitinin at a titrated concentration will poke the plasma membrane and leave mitochondria, if you are careful and if you check citrate synthase and LDH as markers. Isolated mitochondria, prepared on a sucrose or Percoll gradient, are the classical object; they will tell you about matrix NAD+ only if you did not leak them on the way. Nuclear preparations are dirtier than people admit. Matrix-targeted and nuclear-targeted biosensors skip the fractionation and ask the living cell, which is why they earned their keep. SLC25A51 knockdown or knockout is how you ask whether mitochondrial NAD+ import is required for your phenotype. NMNAT isoform knockdowns are how you ask which postcode's last step you needed. If this sounds like a lot of work, that is because a dinucleotide with three spends and three salvage addresses is a lot of work. A kit on a homogenate is allowed as a scout. It is not enough, on its own, for a mitochondrial title.

In short. NAD+ lives in separate rooms of the cell. Fractionate, or use sensors aimed at one room, or delete the importer. A blended soup measurement cannot tell a mitochondrial story on its own.

A blot of a SIRT1 substrate is not a NAD+ measurement. Acetyl-p53, acetyl-PGC-1α, acetyl-FOXO, H3K9ac: those report whether a deacetylase with that client was active, which is a function of enzyme abundance, enzyme localisation, the acyl-lysine supply, nicotinamide product inhibition, and NAD+ substrate. You can have a fat pool and a missing enzyme. You can have enzyme and no cofactor. PAR polymer is the equivalent honesty test on the PARP1 side: if you claim a PARP1 drain, show PAR, show it go away with a PARP inhibitor, and then show the NAD+ number move. CD38 protein or activity, same rule. The restoration literature that reports a phenotype and a whole-cell NAD+ kit, and nothing about which drain or which sirtuin client, is the literature that does not replicate. Design the assay as if a sceptical colleague will have to believe it. That colleague is who we are writing for.

In short. A mark on a sirtuin target is not a NAD+ number. Show the enzyme, the cofactor, and the client. Papers that skip this step fail later.

Extraction timing and redox-quench are the unglamorous half of an honest assay, and they ruin more figures than the wrong precursor. NADH oxidises in air, in a poorly buffered extract, on a warm bench. NAD+ can be lost to a hydrolase you did not know you had lysed into the tube. Snap-freeze, cold methanol, the acid/alkali split already named, internal standards added at the moment of quench not at the moment of injection: those are the moves. A time course after a genotoxic hit is minutes, not an overnight treatment. A time course after FK866 is hours. A time course after NR in a mouse is a plasma curve plus a tissue curve, and they are not the same shape. If you add β-NAD+ to a medium and harvest at twenty-four hours, you have done an ectonucleotidase-plus-salvage experiment unless you have shown otherwise. Write the time. Write the quench. Write the standard. The cake in the vial is only as informative as the minute you chose to stop the chemistry.

In short. The loaded form oxidises on the bench. Freeze fast, add the standard at the stop, and write the time point.

Cell-type choice is a control, not a convenience. A HEK293 well will tell you whether a construct and a precursor move a metabolome. It will not tell you what a CD38-high macrophage does to a neighbour's pool, and it will not tell you what a myotube does with NRK2. Primary hepatocytes still have NAMPT and a de novo path; neurons mostly do not. Endothelial cells are a SIRT1-and-nitric-oxide neighbourhood; they are not a SIRT3-and-β-oxidation neighbourhood. Aged mouse tissue is not a late-passage dish, and a late-passage dish is not ageing, it is a Hayflick neighbourhood with its own NAD+ story (PARP, senescence, CD38 on the contaminating glia if you are in brain culture). If the claim is ageing, the animal has to be old, the tissue has to be named, and the cell sort, if you can afford it, has to be done, because a whole-liver NAD+ number is a hepatocyte diluted by immune cells that may be the actual CD38. That last sentence is Verdin's point, restated as a design request.

In short. Pick the cell that has the job you are claiming. A kidney-cell line is not a muscle fibre, and a mixed old liver is partly immune cells chopping NAD+. Ageing work needs old tissue, named.

The node is conserved, which is the only reason a yeast silencing gene, a worm lifespan paper, a mouse clamp and a human metabolome can sit in one essay without being a collage. SIR2 spent NAD+ in Saccharomyces. The seven mammalian sirtuins still do. PARP1 is a eukaryotic elaboration on a break. CD38 is a vertebrate NADase that ageing and inflammation both write. NAMPT is the bottleneck a mammal actually lives on. NRKs let a nucleoside walk around it. Complex I has wanted oxidised NAD+ since there was a proton-motive force to build. You can run this argument in a bacterium, with different names, and the hydride at nicotinamide C4 will still be the chemistry. Conservation is not a licence to treat a mouse figure as a human protocol. It is a licence to take the biochemistry seriously enough to measure it, in the organism you have, with the controls the drains require. The popular story got loud because the node is central. The work got hard for the same reason.

In short. From yeast to humans, cells spend this cofactor to silence genes, flag broken DNA, and run mitochondria. A mouse result is not automatically a human plan.

The public papers are the reading list, and they are short enough to actually read. Imai and Guarente, Nature 2000, SIR2 as an NAD+-dependent deacetylase. Revollo, Imai, JBC 2004, NAMPT as the mammalian bottleneck. Bieganowski and Brenner, Cell 2004, NR as a vitamin. Camacho-Pereira, Chini, Verdin, Cell Metabolism 2016, CD38 as the age-NADase. Yoshino, Imai, Cell Metabolism 2011, NMN in diabetic mice. Trammell, Brenner, Nature Communications 2016, NR in humans. Yoshino, Klein, Science 2021, NMN and a clamp in prediabetic women. Kory, Mootha, Nature 2020, SLC25A51 as the mitochondrial NAD+ transporter. Lee, Cohen, Cell Metabolism 2015, MOTS-c, so the neighbourhood stays named. Rich, 2003, so the ATP scale stays honest. Kraus and Berger on PARP, so the minutes-scale drain stays in the picture. That is a fortnight of evenings, not a guru. The restoration headlines will still be there when you come back, and they will look smaller.

In short. A short stack of named papers covers the enzyme, the bottleneck, the vitamin precursor, the age-linked leak, the mouse work, the human clamps and the mitochondrial importer. Read those before any headline.

What you should leave with is a map, not a shopping list. NAD+ is a hydride carrier at nicotinamide C4, and that is still most of the pool's day. Sirtuins, PARP1 and CD38 spend it, one molecule at a time. NAMPT salvages. NR and NMN feed. Mitochondria notice first because Complex I and SIRT3 both sit on the pool, and because the matrix is not in free exchange with the rest of the cell. Mice, given NMN or NR, show metabolic and vascular phenotypes that are in print. Humans, given NR or NMN, show metabolome shifts that are in print and clinical endpoints that are mixed. The 1000 mg cake is lyophilised β-NAD+ for the assays that map demands. MOTS-c is a different object on the same campus. The appointment at another company is a different legal object. If your experiment needs the cofactor, weigh it, quench it, and name the drain. If it needs a precursor, the precursor papers are cited above. If it needs a medicine, this catalogue does not sell one.

In short. Leave with the map: electron carrier, three spends, one bottleneck, mitochondria first, mixed human outcomes. The gram in the vial is for the assay that map requires.

Ageing, on this node, is a budget that started to leak faster than it was refilled. CD38 up, NAMPT down in some tissues, PARP1 busier in a genome that has had more years to be nicked, sirtuins still trying to deacylate a proteome that is more acetylated than it was. That sentence is a research programme. It is not a diagnosis, and it is not a product. The programme runs on HPLC-characterised reagents, on mice whose age you state, on clamps rather than well-being scores, and on a willingness to publish the trial that did not move the endpoint. We will keep the cofactor on the shelf at a serious aliquot because that programme is real and because the molecule it spends is not a mystery. The currency of cellular time is a dinucleotide. Time, in a cell, is also telomeres and proteostasis and mtDNA heteroplasmy and a hundred other clocks. This one you can weigh.

In short. Ageing on this topic looks like a budget leaking faster than it is refilled. That is a research programme, not a diagnosis or a treatment for time.

Research-use-only. Not for human consumption / not a medicine. The lyophilised β-NAD+ on this listing is a laboratory reagent, HPLC-characterised at ≥98 percent, labelled for in-vitro work: a standard curve, a sirtuin tube, a PARP assay, a set of isolated mitochondria whose oxygen consumption you actually record. The physiology in the paragraphs above is public, cited, and older than the vial. Use it to design the experiment you have the controls for, with the drain named, the compartment named, and the time point written down. Read Imai, read Verdin, read Brenner, then weigh the cake. We will sell you the cofactor. The legal class of this listing is that first sentence, and then the work is ordinary: a tube, a quench, a number you can defend. Cellular time is a set of rates. This rate you can measure, in a tube, with a chromatogram on the bench beside it.

In short. The vial is a research chemical for experiments, not a medicine and not food. The biology is public. Weigh it, measure the rate, and keep the claim the size of the chromatogram.