NADH Measurement: Enzymatic Cycling, Fluorescence, LC-MS

A perchloric acid extract and a potassium hydroxide extract pulled from the same flask of cells will return two different numbers, and both of them will be right. That isn't a paradox. It's the chemical foundation every NADH measurement is built on.

Published values for the same tissue often differ by an order of magnitude between papers, and the method is almost always the reason.

How is NADH measurement performed in the laboratory?

NADH measurement uses three method families: enzymatic cycling assays that amplify signal through alcohol dehydrogenase and a tetrazolium dye, fluorescence detection of the reduced cofactor's native emission near 460 nm, and LC-MS/MS quantification against stable-isotope internal standards. Selective acid or alkaline extraction separates NAD+ from NADH before any of them runs.

The common oversimplification is that these three readouts are interchangeable. They aren't. A cycling kit reports the total extractable dinucleotide pool, autofluorescence reports free NAD(P)H inside living cells, and LC-MS reports molar amounts against a labelled standard. This page covers extraction chemistry, each assay format in turn, the NAD+/NADH ratio as a measured versus a calculated quantity, and the interconversion errors that quietly wreck otherwise clean data.

What does a complete NADH measurement workflow actually involve?

Every quantitative NADH measurement moves through five stages: quench, extract, clarify, assay, normalise. Skipping or rushing any one of them changes the answer more than swapping instruments does.

Nicotinamide adenine dinucleotide exists in two interconverting forms. NAD+ is the oxidised form that accepts a hydride ion; NADH is the reduced form that carries it. Together they shuttle reducing equivalents through glycolysis, the tricarboxylic acid cycle and oxidative phosphorylation, and they serve as consumed substrate for sirtuins, PARP enzymes and CD38. Background on that biology sits on the NAD research hub and the broader NAD overview page.

Quenching stops metabolism. Cultured cells are typically washed and hit with ice-cold extraction solvent directly in the well; tissue is freeze-clamped in liquid nitrogen within seconds of excision. The urgency is real: NADH is consumed and regenerated on a timescale of seconds.

Extraction releases the cofactor and simultaneously destroys the enzymes that would keep turning it over. Clarification removes protein and particulates. The assay stage produces signal. Normalisation converts that signal into something comparable, usually per milligram of protein by BCA, per cell count, or per unit DNA.

Total cellular NAD(H) in mammalian cells is commonly reported in the low hundreds of micromolar, with NADH representing a small minority of the pool.

Why does the extraction solvent decide what you can detect?

NAD+ and NADH have mirror-image stability profiles. NAD+ is stable in acid and degrades in alkali; NADH is stable in alkali and degrades in acid. Differential extraction exploits that difference deliberately.

An acid extraction, classically 0.5 to 1 M perchloric acid or hydrochloric acid, preserves NAD+ while destroying NADH. The extract is then neutralised, often with potassium hydroxide or potassium carbonate, which precipitates potassium perchlorate out of solution. An alkaline extraction with sodium or potassium hydroxide does the reverse. Running paired extractions on split samples is the oldest and still one of the most defensible approaches.

Most commercial colorimetric kits use a different trick on a single neutral extract. One aliquot is read as-is to give total NAD(H). A second aliquot is heated, commonly at 60 degrees Celsius for around 30 minutes, which decomposes NAD+ while leaving NADH intact. NAD+ is then obtained by subtraction. The arithmetic is simple and the failure mode is sneaky: because NAD+ usually dominates the pool, a small percentage error in the total measurement propagates into a large percentage error in the subtracted NADH value.

Here's the detail most protocols bury. Neutralisation pH is not a formality. An extract parked even briefly on the acidic side of neutral will lose NADH. Checking the pH of neutralised extracts with a microelectrode, not an indicator strip, is the cheapest insurance available.

How does an enzymatic cycling assay generate signal from picomole amounts?

Enzymatic cycling turns a tiny analyte pool into a large, time-dependent colour change by making each dinucleotide molecule do the same job over and over. That amplification is why plate-based kits can work in the picomole range without a mass spectrometer.

The standard chemistry pairs alcohol dehydrogenase with excess ethanol. The enzyme reduces NAD+ to NADH; the NADH then transfers electrons, usually via phenazine methosulfate or diaphorase, to a tetrazolium salt such as WST-8 or MTT, producing a coloured formazan read at roughly 450 to 570 nm. The regenerated NAD+ re-enters the cycle. Because the cofactor is catalytic rather than consumed, signal accumulates linearly with time. Kinetic reads across several time points are more robust than a single endpoint.

Cofactor specificity is a genuine strength here. Alcohol dehydrogenase uses NAD(H) and not NADP(H), so an ADH-based cycling assay doesn't report the phosphorylated pool.

The interference that trips people up isn't biological, it's chemical. Tetrazolium dyes are reduced directly by ordinary reducing agents. Ascorbate, dithiothreitol, beta-mercaptoethanol and free thiols including glutathione will all generate formazan with no enzyme involved. A no-enzyme control well, run on every sample rather than once per plate, is the only way to catch it, and it is the single most frequently omitted control.

What does autofluorescence near 460 nm really report?

NADH is natively fluorescent and NAD+ is not. Excite a sample around 340 to 360 nm and the reduced cofactor emits broadly near 440 to 470 nm, which makes label-free imaging of living, unfixed cells possible.

That selectivity has one large caveat: NADPH is spectrally indistinguishable from NADH. Any honest microscopy paper therefore labels the signal NAD(P)H rather than NADH.

Fluorescence lifetime imaging microscopy, FLIM, adds a dimension that intensity alone can't. Free NAD(P)H in solution decays quickly, while enzyme-bound NAD(P)H decays considerably more slowly, typically in the low nanoseconds. Fitting the decay to two components yields the relative proportion of free versus bound cofactor. Two-photon excitation around 700 to 750 nm is widely used to reach deeper into tissue with less ultraviolet photodamage.

What autofluorescence does not give you is an absolute molar concentration. Intensity depends on excitation power, detector gain, optical path, scattering and local protein binding. Many groups report the optical redox ratio instead, pairing NAD(P)H emission against flavin adenine dinucleotide autofluorescence.

Used well, this is a spatially resolved, non-destructive complement to extraction-based NADH measurement. Used as a substitute for quantification, it overpromises.

How does LC-MS/MS handle NAD+ quantification and the reduced form?

LC-MS/MS is the reference technique when absolute molar numbers matter, because it separates the analytes chromatographically and identifies them by mass transition.

Both cofactors are highly polar and retain poorly on conventional reversed-phase columns, so methods generally use hydrophilic interaction liquid chromatography on an amide or zwitterionic stationary phase, or ion-pairing reversed-phase chemistry. Detection runs in multiple reaction monitoring mode. The positive-mode transition for NAD+ from the protonated molecule at m/z 664 to the adenine fragment at m/z 136 is the most widely published pair; the reduced form is commonly followed from its protonated molecule near m/z 666.

Stable-isotope-labelled internal standards are not optional at this level. Carbon-13 or nitrogen-15 labelled NAD+ corrects for extraction recovery and for the ion suppression that polar matrices reliably produce in electrospray sources. Without one, a clean-looking chromatogram can still be quantitatively wrong.

The reduced cofactor is the harder analyte. It oxidises in the autosampler, it is sensitive to mobile-phase pH, and it can convert in the source. Keeping the sample tray chilled, limiting queue length, and bracketing batches with calibrants is standard practice. The same instrumentation logic covered in mass spectrometry peptide verification and in HPLC versus mass spectrometry for peptides applies directly to cofactor work.

Which assay format fits which research question?

No single platform wins on every axis. The table below compares the five formats that appear most often in the literature, so a method can be matched to the question rather than to whatever sits on the bench.

What does the NAD+/NADH ratio describe, and why do published values disagree?

The NAD+/NADH ratio is not one number. It is at least two, and conflating them explains most of the apparent contradictions in the literature.

A whole-tissue or whole-cell extract measures every molecule present, including the large fraction of NADH bound tightly to dehydrogenases. Ratios calculated from extracts therefore tend to come out in the single digits to low tens. The free, enzymatically available cytosolic ratio is a completely different quantity, and classic estimates derived from near-equilibrium reactions place it in the hundreds. The discrepancy reflects the fact that most of the reduced cofactor is protein-bound and metabolically invisible.

The indirect route to the free ratio uses metabolites in rapid equilibrium with the cofactor. Lactate dehydrogenase links the lactate-to-pyruvate ratio to the free cytosolic NAD+/NADH ratio; beta-hydroxybutyrate dehydrogenase links the beta-hydroxybutyrate-to-acetoacetate ratio to the free mitochondrial ratio. Both calculations assume the relevant reaction sits at equilibrium.

Genetically encoded ratiometric biosensors sidestep the extraction problem entirely by reporting free ratio inside an intact, compartment-specific environment, in real time.

The practical rule for reporting is simple. State which quantity was measured, by which method, in which compartment. A ratio quoted without that context cannot be compared with anyone else's.

Where do NADH measurement workflows usually go wrong?

Most failures happen before the plate reader or the mass spectrometer ever sees the sample. Interconversion during handling is the dominant error source, and it is systematic rather than random, which means replicates look beautifully consistent while being uniformly wrong.

The first failure is quench delay. Detaching adherent cells by trypsinisation, spinning them, and resuspending them takes minutes, during which oxygen tension, substrate supply and pH all change. Extracting directly in the culture vessel removes the problem.

The second is enzymatic turnover after lysis. NAD-consuming enzymes, particularly CD38 and the PARP family, remain catalytically competent in a crude lysate. An extraction solvent that denatures protein instantly is doing double duty as a quench; a gentle detergent lysis is not.

The third is sample storage. Repeated freeze-thaw cycles degrade both species, unevenly. Extracts should be aliquoted once and held cold. The same stability logic that governs peptide storage best practices, degradation pathways and the lyophilised versus liquid stability comparison applies to cofactor standards, and is summarised for NAD material on the NAD storage page.

The fourth is normalisation. Protein content drops in stressed or dying cells, so a per-milligram denominator can manufacture an apparent rise in cofactor where none exists. Reporting two denominators, protein and cell number, costs almost nothing and makes an NADH measurement far harder to misread.

How is an NADH measurement method validated and documented?

Validation turns a working protocol into a defensible number. The framework most bioanalytical laboratories reference is the ICH M10 guideline on bioanalytical method validation.

For cofactor work, the practical checklist covers calibration curve range and fit, lower limit of quantification, within-run and between-run precision expressed as coefficient of variation, accuracy against spiked controls, extraction recovery, matrix effect assessed by post-extraction spiking, and stability under bench-top, freeze-thaw and long-term storage conditions. Carryover between high and low samples deserves its own check.

Documentation is where most NADH measurement write-ups fall short. A reproducible methods section names the quench solvent and its temperature, the extraction acid or base and its molarity, the neutralisation agent and final pH, whether NAD+ was measured directly or by subtraction, the internal standard used, the normalisation denominator, and whether the reported figure is a total pool or a free ratio.

The same documentation discipline governs the quality of the reference material a method depends on. Reading a certificate of analysis properly matters here; the guides on how to read a peptide COA, HPLC peptide purity and research-quality purity standards cover what an analytical certificate should and should not tell you. Real Peptides publishes lot certificates of analysis at its COA library.

Can NAD levels be measured in blood or inside a living subject?

Yes, with heavy caveats on both the sample type and the interpretation. Whole blood is not a convenient proxy for tissue.

In blood, the dinucleotide pool sits overwhelmingly inside erythrocytes rather than in plasma. That makes haemolysis a decisive pre-analytical variable: a slightly haemolysed plasma sample will read high for reasons that have nothing to do with biology. Collection tube chemistry, time to centrifugation and storage temperature all move the result. Dried blood spot methods coupled to LC-MS have been described and reduce some handling variability.

In living tissue, magnetic resonance spectroscopy at high field strength has been used to estimate NAD+ and NADH signals non-invasively in brain and muscle, and genetically encoded biosensors provide compartment-resolved ratio dynamics in cell and animal models. Neither approach replaces an extraction-based NADH measurement for absolute pool size.

All compounds discussed on this page are sold for laboratory research use only and are not for human consumption. What that designation means in practice is covered on the research use only page, alongside the regulatory summary at is NAD legal and the half-life discussion at NAD half-life. Readers who arrived here from a companion-animal context should talk to their veterinarian; nothing on this page is guidance for any living subject.

What if the assay behaves unexpectedly?

What if total pool looks unchanged but the ratio shifted?

Treat that as a plausible real result rather than an artefact, then confirm it with an orthogonal method. Redox state can move substantially while total pool size stays flat, because interconversion is fast and pool turnover is comparatively slow. Confirming a cycling-kit ratio with LC-MS/MS, or with a lactate-to-pyruvate calculation, is the standard way to decide whether the shift is chemistry or subtraction error.

What if no-enzyme control wells develop colour?

Stop and audit every buffer that touched the sample for reducing agents. Dithiothreitol, beta-mercaptoethanol, ascorbate and free thiols reduce tetrazolium dyes directly, producing formazan without any cofactor present. The fix is either a reducing-agent-free lysis buffer or a desalting step before the plate, and the no-enzyme control must then be run on every sample, not once per plate.

What if the reduced-form peak shrinks across an LC-MS batch?

Suspect oxidation in the autosampler before suspecting the column. The reduced cofactor degrades in warm trays and under acidic mobile-phase conditions, which produces a time-dependent downward drift that mimics a biological effect. Chilling the tray, shortening the queue, bracketing with fresh calibrants and randomising sample order across the batch will separate instrumental drift from genuine differences.

The uncomfortable truth about comparing published NAD numbers

Let's be direct about this: most cross-study comparisons of NAD+ and NADH values are not valid, and the problem is methodological rather than biological. An extract-based total pool, a free cytosolic ratio inferred from lactate and pyruvate, a FLIM bound-fraction and an LC-MS molar concentration are four different measurements wearing the same name. Pooling them into a single narrative produces conclusions the underlying data cannot support. The useful comparison is always within a study, within a method, against a concurrent control. Anything else is a literature review of assay choices masquerading as biology.

How this page was researched

The content here is drawn from the published analytical-chemistry and metabolism literature indexed in PubMed, from assay technical documentation describing enzymatic cycling and heat-decomposition protocols, from HILIC and ion-pairing LC-MS/MS method papers reporting multiple reaction monitoring transitions for the dinucleotide cofactors, and from the ICH M10 guideline on bioanalytical method validation. No first-party laboratory testing was performed for this article, and no numeric values are presented here as original data. Where the literature reports ranges rather than fixed figures, ranges are given.

Further reading across the NAD library on this site: the NAD FAQ, the history of NAD research, the reported benefits summary, the 2026 news roundup, the PARP and DNA repair overview, the NAD versus SS-31 comparison, the MOTS-c versus NAD comparison, the addiction research page, the men over 40 research summary, the is NAD worth it analysis, the acetaminophen and NAD discussion, the oral taste page, shipping and cold chain information, wholesale NAD enquiries, and the company location page.

The thing worth carrying away from all of this is that an NADH measurement is a statement about a method as much as about a sample. Two laboratories studying identical cells, one extracting in perchloric acid and reading a cycling kit, the other imaging lifetime decay in the living dish, will publish numbers that cannot be reconciled and neither will have made a mistake. The discipline isn't in choosing the best assay. It's in writing down precisely which quantity was captured, so the next person can tell whether their number and yours are even describing the same thing.