NAD+: Redox Coenzyme, Signalling Substrate and Compartmental Pools

NAD+ is a dinucleotide made of an adenosine monophosphate unit and a nicotinamide mononucleotide unit bridged by pyrophosphate, giving formula C21H27N7O14P2 and a molecular mass of 663.43 g/mol. PubChem catalogues the compound as CID 5892 under the title Nadide, with InChIKey BAWFJGJZGIEFAR-NNYOXOHSSA-N and CAS number 53-84-9.

The plus sign is charge notation rather than typographic decoration. Nicotinamide attaches to its ribose through a beta-N-glycosidic bond at the ring nitrogen, making that nitrogen quaternary and giving the ring a permanent positive charge. The neutral free acid is written C21H27N7O14P2 and the pyridinium cation C21H28N7O14P2+, which is why two masses, 663.4 and 664.4 g/mol, circulate in chemical databases for one laboratory material.

Polarity descriptors account for most practical handling behaviour. PubChem computes an XLogP of minus 6 and a topological polar surface area of 321 square angstroms, with seven hydrogen-bond donors, eighteen acceptors and eleven rotatable bonds. A molecule that polar does not traverse a lipid bilayer unaided, which is why carriers and extracellular cleavage steps dominate the trafficking literature. Older names survive in method sections: coenzyme I, codehydrogenase I, and diphosphopyridine nucleotide, abbreviated DPN.

NAD+ participates in two mechanistically distinct classes of reaction. In oxidoreductase catalysis the nicotinamide ring accepts a hydride ion at carbon 4 to become NADH, then surrenders it again, leaving the coenzyme intact and the pool conserved. In signalling catalysis the glycosidic bond is severed, nicotinamide departs, and the molecule ceases to exist.

The redox couple sits near minus 0.32 volts at pH 7, and compartments hold it at different set points: review work quotes free cytosolic NAD+ to NADH ratios of roughly 60 to 700 against approximately 7 to 8 in the mitochondrial matrix. Consumption is the destructive route, and Hoyland and colleagues quantified its consequences in Nature Metabolism in 2024. Cell lines expressing poly-ADP-ribosyltransferase activity in single compartments lost up to 50 percent of cellular NAD+, and flux modelling showed the lowered concentration kinetically restricting consumption while biosynthesis held constant, so the system settled at a lower steady state.

NAD+ is consumed by three enzyme families: poly-ADP-ribose polymerases, which transfer ADP-ribose onto acceptor proteins; sirtuins, which strip acyl groups from lysine residues and release 2-O-acyl-ADP-ribose; and the glycohydrolases CD38, CD157 and SARM1, which cleave the molecule outright. All three liberate free nicotinamide as a product.

Reported Michaelis constants separate these families by roughly an order of magnitude: near 15 to 25 micromolar for CD38, roughly 20 to 97 micromolar for PARP1, and a wider 94 to 888 micromolar span for SIRT1. Those constants are not coincidental. Cambronne and colleagues placed free NAD+ at approximately 100 micromolar in nucleus and cytosol and 230 micromolar in mitochondria using a compartment-targeted fluorescent sensor. Consuming enzymes therefore sit where activity tracks the pool rather than saturating against it, and as the pool falls the low-Km glycohydrolases keep turning over while sirtuin catalysis drops away first.

SARM1 is the unusual member of the group, a regulated switch rather than a constitutive hydrolase. Its armadillo-repeat domain carries an allosteric site for which nicotinamide mononucleotide and NAD+ compete, so the enzyme effectively reads the ratio between them. Huang and colleagues reported cryo-electron microscopy structures in PNAS in 2025 in which a biosynthetic mononucleotide mimic captured two intermediate conformations and the fully active state; an engineered disulfide bond blocking the transition held the enzyme inactive.

NAD+ arises by three biosynthetic routes: de novo synthesis from tryptophan through the kynurenine pathway, the Preiss-Handler route from nicotinic acid, and the salvage route that recycles nicotinamide released by consuming enzymes. Salvage carries most of the flux in mammalian tissue, which makes nicotinamide phosphoribosyltransferase the rate-limiting step.

Zhang and colleagues reviewed that enzyme in Cell Death Discovery in 2025, describing conversion of nicotinamide to nicotinamide mononucleotide in the presence of magnesium, phosphoribosyl pyrophosphate and ATP, and summarising age-associated reductions across organs of 10 to 50 percent. Three adenylyltransferases complete the final step in distinct locations: NMNAT1 in the nucleus, NMNAT2 in cytosol and Golgi, NMNAT3 in mitochondria. Since the molecule is too polar to cross a bilayer, compartmental supply needs transport, and Luongo and colleagues identified SLC25A51 in Nature in 2020 as the mammalian mitochondrial NAD+ transporter. Where NMNAT3 is present, mitochondria also cleave the coenzyme reversibly into nicotinamide mononucleotide and ATP, holding a virtual pool rebuilt on demand.

NAD+ measured in human whole blood behaves differently from NAD+ measured in tissue, and work published across 2025 and 2026 has sharpened that distinction. Tretowicz and colleagues reported in Nature Metabolism in 2026 that whole-blood NAD+ remained stable with age and across lifestyle interventions in seven independent human cohorts.

That analysis did register the expected change after nicotinamide riboside supplementation, so its narrow conclusion was that blood NAD+ is of questionable use as a biomarker of ageing. Tissue is a separate question: Vinten and colleagues noted in Nature Metabolism in 2025 that an age-related decline in humans has been observed consistently in only a limited number of studies, while Guo and colleagues quantified whole-brain NAD at approximately 0.4 millimolar by phosphorus-31 spectroscopic imaging at 7 tesla.

Raising the circulating pool has proved easier than shifting an outcome. Wu and colleagues randomized 58 community-dwelling adults with long COVID in a trial reported in eClinicalMedicine in 2025; nicotinamide riboside at 2,000 mg per day raised NAD+ 2.6- to 3.1-fold, yet no significant between-group differences emerged on the cognitive primary outcomes or on fatigue severity. Tolerability of the coenzyme itself also differs from that of its precursors, as Reyna and colleagues documented in Frontiers in Aging in 2026.

NAD+ is quantified by four broadly different approaches: ultraviolet absorbance, enzymatic cycling assays, chromatography coupled to mass spectrometry, and genetically encoded sensors. Each answers a different question, and their numbers are not interchangeable, because they report total pools, free pools, or compartment-specific pools respectively.

Absorbance is the oldest and least specific. Both forms absorb near 260 nm, the oxidized species at about 16,900 M-1 cm-1 at 259 nm, while only the reduced form absorbs at 340 nm, at 6,220 M-1 cm-1 at 339 nm, which is what makes dehydrogenase assays workable in a plate reader. Mass spectrometry is the current reference method for biological matrices, and the 2026 whole-blood work carries a point worth borrowing: validation must model the variability that real sample handling introduces, not merely instrument precision. Whole blood, plasma and isolated cells return different numbers from one donor, so the matrix belongs beside the value.

NAD+ is supplied as a lyophilized solid. The most directly useful published handling data concerns solution stability, and Wolfe and colleagues measured it systematically in Molecules in 2024, holding 2 millimolar cofactor in 50 millimolar buffer at pH 8.5 at 19 and 25 degrees Celsius for 43 days.

Buffer identity mattered more than temperature across that window. In Tris the 260 nm peak of the oxidized cofactor fell by only 4 percent over 43 days at 19 degrees, while in HEPES the same cofactor degraded almost entirely, with sodium phosphate falling in between. Buffer choice belongs in the method record as a named variable rather than boilerplate.

That paper also flags an analytical trap. Determining concentration directly from absorbance was described as not straightforward, because degradation products absorb in the same ultraviolet region as the intact cofactor, so a steady-looking peak can conceal turnover; the authors paired spectroscopy with an enzymatic activity assay for exactly this reason. The structural liabilities follow from the chemistry: the pyrophosphate bridge is hydrolysable and the beta-N-glycosidic bond is both the enzymatic cleavage site and a base-labile linkage, so purity records should resolve nicotinamide and ADP-ribose as expected degradation species rather than reporting a single main-peak percentage.