As shown in Fig. 5, inbred, chow-fed male mice supplemented with NAD+ precursor vitamins by gavage and euthanized at ∼2 pm produced hepatic NAD+ metabolomic data with little variation. However, blood samples from people exhibited greater variation, due to differing baseline levels of metabolites and variable pharmacokinetics, both of which are likely due to genetic and nutritional changes between subjects (Fig. 8). In PBMCs, eight key metabolites were quantified in at least 10 subjects at all time points at each dose. For each metabolite, we plotted average concentration as a function of dose and time, calculated whether NR elevated that metabolite, plotted the averaged peak concentration of the metabolite as a function of dose, and calculated the dose-dependent AUC of the metabolite attributable to NR supplementation. Unlike a drug metabolism study in which metabolites appear only after administration, most NAD+ metabolites are present before supplementation, such that the AUC attributable to supplementation is a time-zero baseline-subtracted AUC. Thus, it is possible to calculate not only the AUC rise in metabolites but also the per cent increase in AUC in these metabolites attributable to dose-dependent NR supplementation.
Collapsing the data into pre-dose versus 24 h levels of each metabolite at all doses, NR significantly elevated PBMC NAD+ (Fig. 8b), MeNam (Fig. 8d) and Me2PY (Fig. 8e) and significantly elevated PBMC NAAD (Fig. 8f) at 8 h. In contrast, NR did not produce a statistically significant all-dose elevation of NMN (Fig. 8a) or Nam (Fig. 8c) at any time point.
The averaged peak concentration of MeNam (Fig. 8d), Me2PY (Fig. 8e) and NAAD (Fig. 8f) increased monotonically with increased NR doses. Of these metabolites, only NAAD was below the detection limit in individuals before they took NR, qualifying this metabolite as a biomarker of supplementation. Nam (Fig. 8c) exhibited no tendency towards higher cellular concentrations with higher doses of NR. NMN tended to rise (Fig. 8a) and NAD+ rose (Fig. 8b) to higher concentrations of ∼2 and 20 μM, respectively, in people taking 300 and 1,000 mg doses of NR versus people taking 100 mg doses. Thus, 100 mg supplementation produced an average ∼4±2 μM increase in PBMC NAD+, whereas the higher doses produced average ∼6.5±3.5 μM increases in PBMC NAD+. No sex differences were discovered.
As was first seen in the n=1 human experiment and in mouse liver experiments, NAAD is the most sensitive biomarker of effective NAD+ supplementation because it is undetectable in the blood of people before supplementation. At all doses, the peak shape of NAAD indicated that NAD+ metabolism is most greatly boosted at 8 h with significant supplementation at 4 h and significant supplementation remaining at 24 h. At the 8 h peak, the average concentration of NAAD was elevated to 0.56±0.26, 0.74±0.27 and 1.24±0.51 μM in PBMCs from volunteers taking 100, 300 and 1,000 mg single doses of NR, respectively.
We plotted pre-dose-subtracted AUCs of each metabolite as a function of dose of NR. With the exception of Nam, the levels of which were unaffected by NR, NR produced or tended to produce dose-dependent elevation of the entire NAD+ metabolome (Fig. 8). In plasma, levels of meNam, me2PY and me4PY also rose in a dose-dependent manner and were identified at concentrations similar to those in the PBMC fraction. The methylated and oxidized Nam derivatives were accompanied by low levels of NAR, which increased with higher doses of NR. Urinary metabolites were similar to plasma metabolites.
In Supplementary Table 4, the average 24 h baseline-subtracted AUC of each metabolite is expressed as a percentage increase in that metabolite at each dose of NR. Once again, Nam and NMN showed essentially no increase in blood cell concentrations with respect to baseline concentrations: averaged AUCs never rose by 50% above baseline. However, NAD+, meNam, me2PY and NAAD rose or tended to rise in dose-dependent manners. The effect size of the rise in NAAD (∼2,900%) was much greater than the effect sizes of the rise in me2PY (∼600%), meNam (∼200%) or NAD+ (∼90%). AUC increases of NAAD, me2PY and meNam achieved statistical significance with respect to lower doses of NR.

Discussion

Despite >75 years of human use of NA and Nam11 and >10 years of preclinical NR research6, there has never been a quantitative metabolomic or pharmacokinetic comparison of the three NAD+ precursor vitamins in any system. In terms of elevation of mouse liver NAD+, we discovered that NR is more orally bioavailable than Nam, which is more orally bioavailable than NA (Fig. 5b). The three precursors also differ in the degree to which they promote accumulation of ADPR, a measure of sirtuin and other NAD+-consuming activities. As shown in Fig. 5j, the ability of NR to elevate ADPR exceeded that of Nam by ∼3-fold. This validates NR as the favoured NAD+ precursor vitamin for increasing NAD+ and NAD+-consuming activities in liver.
NR, Nam and NA each have unique pharmacokinetic profiles in mouse liver, both in terms of kinetics of NAD+ formation and the population of NAD+ metabolites as a function of time. As shown in Fig. 5d, Nam is the only vitamin precursor of NAD+ that produces elevated hepatic Nam 15 min after oral administration and, as shown in Fig. 5g, NA is the only precursor that produces elevated NA 15 min after oral administration. These data exclude the possibility that all three vitamins are utilized through the Preiss-Handler pathway in liver or that oral NR is used exclusively as Nam. Additionally, we have demonstrated that utilization of both NR and extracellular NMN are limited by activity of the NR kinase pathway48.
When PBMCs were analysed from the first person to ingest NR, NAAD was observed to increase at least 45-fold from a baseline of less than 20 nM to a peak value of nearly 1 μM. This occurred concomitant with a rise in NAD+ from ∼18.5 to 50 μM. NAAD was also observed to be elevated in liver when mice were orally administered NAD+ precursor vitamins. In addition, NR led to striking elevation of NAAD in the heart, a tissue that increases NAD+ metabolism without increasing steady-state NAD+.
Surprisingly, NA, the only precursor expected to proceed to NAD+ through an NAAD intermediate, produced the least NAAD. Indeed, although Nam and NR never produced peaks of hepatic NA or NAR, both produced peaks of hepatic NAAD during the periods in which these compounds elevated hepatic NAD+. The temporal basis of the NAAD excursions suggested that elevating NAD+ (Fig. 5b) not only stimulates accumulation of NAD+-consumption products ADPR (Fig. 5j), Nam (Fig. 5d), MeNam (Fig. 5e) and Me4PY (Fig. 5f), but also stimulates retrograde production of NAAD (Fig. 5i) and NAMN (Fig. 5h). According to this view, as the rate of NAD+ synthesis increases, a previously unknown activity would deamidate NAD+ to NAAD. Alternatively, similar conditions could result in NMN deamidation, giving rise to NAMN and NAAD.
In mouse liver, the apparent flux through this pathway is quite significant: the NR-driven peak of NAAD amounted to 10% of the NR-attributable peak of NAD+. Production of high levels of NAAD from NAD+ could therefore account for the NR-stimulated peak in NAMN because NAMN adenylytransferase is a reversible enzyme49. Striking elevation (∼100-fold) of NAAD was also seen in the heart of mice supplemented with NR after 6 days of IP administration (Fig. 6).
The hypothesis that NAAD is formed from NR in vivo was tested by administering NR labelled in the Nam and ribosyl moieties. As shown in Fig. 7, NR stimulates appearance of double-labelled NAAD (8% of total) at the same time in which 5% of NAD+ is double-labelled. The biochemical basis for a potential NAD+ deamidation reaction is unknown. However, glutamine-dependent NAD+ synthetase is irreversible35,36. One intriguing possibility is that NAAD is formed by the long-sought enzyme that forms intracellular NAADP50. According to this view, an NADP deamidase may be responsible for formation of NAADP—this same activity might deamidate NAD+ at high concentrations forming NAAD. Unlike ADPR and methylated Nam waste products, NAAD is not only a biomarker of elevated NAD+ metabolism but also a reserve metabolite that contributes to elevated NAD+ over time.
Finally, in the first clinical study of NR, we established that blood NAD+ metabolism is increased by single 100, 300 and 1,000 mg doses of NR without dose-dependent increases in PBMC Nam or serious adverse events.
In people, as in mice, NAAD is the most sensitive biomarker of boosting NAD+. While 1,000 mg of NR elevated PBMC NAD+ from ∼12 to ∼18 μM and generated a ∼90% increase in 24 h AUC, NAAD was elevated from below the limit of quantification to ∼1 μM and generated a 2,900% increase in 24 h AUC. The ability to detect NAAD in human samples is expected to aid conduct of clinical testing of NR. Availability of over-the-counter supplements can complicate clinical trials because patients may enrol to obtain compounds they expect to bring benefits and be inclined to take supplements in case they are assigned to placebo. Detection of NAAD should therefore be incorporated in phase II and III trials to eliminate the confounding effects of off-study NR use.

Methods

Materials and reagents

NR Cl was produced under GMP conditions. Me2PY and Me4PY were purchased from TLC PharmaChem Inc. (Vaughan, Ontario, Canada). All other unlabelled analytes were purchased from Sigma-Aldrich (St Louis, MO) at highest purity. Internal standards [18O1]- Nam and [18O1]-NR were prepared as described51,52. [18O1-D3]-MeNam was prepared by alkylation of [18O1]-Nam with deuterated iodomethane. 13C-NA and [D4]-NA were purchased from Toronto Chemical Research (Toronto, Ontario, Canada) and C/D/N Isotopes, Inc. (Pointe-Claire, Quebec, Canada), respectively. To prepare [13C, D1]-NR, we first converted 13C-NA to 13C-Nam (ref. 53) and d-[2-D1]-ribose (Omicron Biochemicals, South Bend, IN) to the labelled d-ribofuranose tetraacetate54. The labelled D-ribofuranose-tetraacetate and Nam were then used to synthesize double-labelled NR55. [13C]-labelled nucleotides standards were prepared by growing yeast in U-13C-glucose and extracting as described9.

Pilot human experiment

After overnight fasting, a healthy 52-year-old male self-administered 1,000 mg of NR Cl orally at 8 am on 7 consecutive days. Blood and urine were collected for quantitative NAD+ metabolomic analysis. The participant took 0.25 g of NA to assess sensitivity to flushing and self-reported painful flushing that lasted 1 h. No flushing was experienced on NR. The study was submitted for approval by the University of Iowa institutional review board (IRB), which ruled it not subject to human subjects research on the basis of informed self-administration56.

Mice

For gavage experiments, 12-week-old male C57Bl/6J mice (Jackson Laboratories, Bar Harbour, ME) were housed 3–5 mice per cage on a chow diet (Teklad 7013) for one week before the experiment. Body weight-matched groups were randomly assigned to be given either 185 mg NR Cl per kg body weight (n=3) or equimole amounts of NA (n=4) or Nam (n=4) by saline gavage. On each experimental day, a saline injection (n=3) was performed and served as time point zero and an additional saline gavage (n=3) timecourse was performed. To avoid circadian effects, timecourses were established such that all tissue harvests were performed at ∼2 pm. Double-labelled NR (n=3) was also administered by gavage against a saline control (n=3). With protocols approved by the University of Iowa Office of Animal Resources, mice were live-decapitated and the medullary lobe of the liver was freeze-clamped at liquid nitrogen temperature. For IP experiments, 6–8-week-old, male C57Bl/6J mice were injected with either PBS or 500 mg NR Cl per kg body weight for 6 days. With protocols approved by the Institutional Animal Care and Use Committee of University of Utah, mice were anaesthetised by chloral hydrate and livers and hearts were freeze-clamped at liquid nitrogen temperature. Tissues were stored at −80 °C before analysis.

Clinical trial

A randomized, double-blind, three-arm crossover pharmacokinetic study of oral NR chloride was performed at 100, 300 and 1,000 mg doses (Clinicaltrials.gov Identifier NCT02191462). Twelve healthy, non-pregnant subjects (six male and six female) between the ages of 30 and 55 with body mass indices of 18.5–29.9 kg m−2 were recruited and randomized to one of three treatment sequences after screening, passing eligibility criteria and providing informed consent. Overnight fasted subjects received a single morning dose of either 100 mg, 300 mg, or 1,000 mg of NR on 3 test days separated by 7-day periods in which no supplement was given. To evaluate pharmacokinetics, blood was collected and separated into plasma and PBMC fractions for analysis of the NAD+ metabolome at pre-dose and again at 1, 2, 4, 8 and 24 h. Urine was collected pre-dose and in 0–6 h, 6–12 h and 12–24 h fractions. Safety, vitals, biometrics, complete blood counts and a comprehensive metabolic panel were assessed at time zero and 24 h after each dose. The study was reviewed and approved by the Natural Health Products Directorate, Health Canada and IRB Services, Aurora, Ontario. Written informed consent was obtained from each subject at the screening visit before all study-related activities.
Exclusion criteria: women who were pregnant, breastfeeding or planning to become pregnant during the course of the trial; use of natural health products/dietary supplements within 7 days before randomization and during the course of the study; use of vitamins or St John's Wort 30 days before study enrolment; use of supplements containing NR within 7 days before randomization and the course of the study; use of nutritional yeast, whey proteins, energy drinks, grapefruit and grapefruit juice, dairy products, alcohol for 7 days before the study; consumption of >2 standard alcoholic drinks per day or drug abuse within the past 6 months; smoking; blood pressure ⩾140/90; use of blood pressure medications; use of cholesterol-lowering medications; metabolic diseases or chronic diseases; use of acute over-the-counter medication within 72 h of test product dosing; unstable medical conditions as determined by the qualified investigator; immune compromised conditions including organ transplantation or human immunodeficiency virus; clinically significant abnormal lab results at screening (for example, aspartate transaminase and/or alanine transaminase >2 × upper limit of normal (ULN), and/or bilirubin >2 × ULN); planned surgery during the course of the trial; history of or current diagnosis of any cancer (except successfully treated basal cell carcinoma or cancer in full remission >5 years after diagnosis); history of blood/bleeding disorders; blood donation in the previous 2 months; participation in a clinical research trial within 30 days before randomization; allergy or sensitivity to study supplement ingredients or to any food or beverage provided during the study; cognitive impairment and/or inability to give informed consent; any other condition which in the qualified investigator's opinion may have adversely affected the subject's ability to complete the study or its measures or which may have posed significant risk to the subject.

Sample preparation and targeted quantitative metabolomics

Dual extractions were carried out for complete analysis of the NAD+ metabolome. For analysis of NR, Nam, NA, MeNam, Me2PY and Me4PY (group A analytes), samples were spiked with 60 pmol of [18O1]-Nam, [18O1]-NR and [D3, 18O1]-MeNam and 240 pmol [D4]-NA (internal standard (IS) A). For analysis of NAD+, NADP+, NMN, NAR, NAMN, NAAD and ADPR (group B analytes), samples were dosed with 13C-yeast extract (IS B) as described9.
Human samples. One hundred microlitre of urine was mixed with 20 μl of IS A in 5% (v/v) formic acid or IS B in water for the analysis of group A and B analytes, respectively. Fifty microlitre of ice-cold methanol was added and the mixture vortexed before centrifugation at 16.1kg at 4 °C for 10 min. Supernatants were injected without further dilution and analysed as described below. Standard curves and quality controls for the complete analysis were prepared in the same manner as described for urine samples but in water.
To quantify group A analytes in plasma, 100 μl of plasma was added to 20 μl of IS A prepared in 5% (v/v) formic acid and mixed with 400 μl of ice-cold methanol. The mixture was allowed to sit on ice for 20 min then centrifuged as described for urine. After drying under vacuum overnight at 35 °C, the sample was reconstituted in 100 μl of water. To quantify group B analytes, 100 μl of plasma was added to 10 μl of IS B in water and mixed with 300 μl of acetonitrile with vortexing for 15 s. After briefly resting on ice, the samples were centrifuged as above. Supernatants were applied to Phenomenex Phree SPE cartridges (Torrance, CA) and the flow-through collected. 200 μl of aqueous acetonitrile (four volumes acetonitrile: one volume water) was also applied and the flow-through collected. The flow-through from both steps was combined and dried via speed vacuum. Samples were reconstituted in 60 μl of water. Standard curves and quality controls for both analyses were prepared in donor plasma (University of Iowa DeGowin Blood Center, Iowa City, IA) and extracted using the same method employed for plasma samples.
PBMC fractions were thawed on ice and simultaneously extracted for both A and B analyses when possible. 100 μl of sample was added to either 20 μl of IS A in 5% formic acid (v/v) or 10 μl IS B in water for quantification of group A and B analytes, respectively. Samples were then mixed with 300 μl of acetonitrile and vortexed for 15 s. Samples were shaken for 5 min at 40 °C then centrifuged as described above. For group A analytes, supernatants were dried via speed vacuum overnight at 35 °C after this step. For group B analytes, supernatant was applied to Phenomenex Phree SPE cartridges and treated in the same manner as described above for quantification of group B analytes in plasma. Immediately before analysis, samples were reconstituted in either 100 μl of 10 mM ammonium acetate with 0.1% formic acid (for group A analyte quantification) or 100 μl of 5% (v/v) aqueous methanol (for group B analyte quantification). Standard curves were prepared in water and processed in the same manner as samples.
Murine samples. For the IP experiment, male C57Bl/6J mice were injected with either saline or NR Cl (500 mg kg−1) for 6 days. On the day of tissue harvests, mice were anaesthetized by chloral hydrate before collection of liver and heart both of which were freeze-clamped in liquid nitrogen. All tissues were stored at −80 °C before extraction.
Murine liver and heart obtained by freeze-clamp were pulverized using a Bessman pulverizer (100–1,000 mg size) (Spectrum Laboratories, Rancho Dominguez, CA) cooled to liquid N2 temperatures. Each pulverized liver and heart sample was aliquoted (5–20 mg) into two liquid N2 cooled 1.5 ml centrifuge tubes, which were stored at −80 °C until analysis. Before extraction, murine liver sample identities were masked and sample order randomized.
Before extraction, IS A and IS B were added to separate aliquots resting on dry ice for quantification of group A and B analytes, respectively. Samples were extracted by addition of 0.1 ml of buffered ethanol (three volumes ethanol: one volume 10 mM HEPES, pH 7.1) at 80 °C. Samples were vortexed vigorously until thawed, sonicated in a bath sonicator (10 s followed by 15 s on ice, repeated twice for liver and thrice for heart), vortexed, then placed into a Thermomixer (Eppendorf, Hamburg, Germany) set to 80 °C and shaken at 1,050 r.p.m. for 5 min. Samples were centrifuged as described above. Clarified supernatants were transferred to fresh 1.5 ml tubes and dried via speed vacuum for two h. Before LC-MS analysis, samples were resuspended in 40 μl of 10 mM ammonium acetate (>99% pure) in LC-MS-grade water. Sample preparation following [13C1, D1]-NR administration differed only in the following respect. In lieu of IS A, 60 pmol of [D4]-Nam and [D3, 18O1]-MeNam and 240 pmol of [D4]-NA were added to sample. Standard curves were prepared in water without extraction.
LC-MS. Separation and quantitation of analytes were performed with a Waters Acquity LC interfaced with a Waters TQD mass spectrometer operated in positive ion multiple reaction monitoring mode as described9. MeNam, Me2PY and Me4PY were added to the analysis and detected using the following transitions: MeNam (m/z 137>94, cone voltage=8 V, collision energy=20 V); Me2PY (m/z 153>107, cone voltage=44 V, collision energy=22 V); and Me4PY (m/z 153>136, cone voltage=24 V, collision energy=14 V). For the analysis of urine, plasma and murine liver, heart and blood cells, group A analytes were separated as described for the acid separation9. In blood cells, group A analytes were separated on a 2.1 × 150 mm Synergy Fusion-RP (Phenomenex, Torrance, CA) using the same gradient and mobile phase as described for the acid separation9. For human samples, group B analytes were separated using the mobile phase and gradient as previously described for the alkaline separation9. Murine liver and heart extracts were analysed using a slightly altered alkaline separation on a 2.1 × 100 mm Thermo Hypercarb column. Specifically, flow rate was increased to 0.55 ml min−1 and run time shortened to 11.6 min. Separation was performed using a modified gradient with initial equilibration at 3% B, a 0.9 min hold, a gradient to 50% B over 6.3 min, followed by a 1 min wash at 90% B and a 3 min re-equilibration at 3% B.
Analytes in human plasma were quantified by dividing their peak areas by IS peak areas and comparing the ratio to a background-subtracted standard curve. Analytes in all other matrices were quantified by dividing their peak areas by IS peak areas and comparing the ratio to a standard curve in water. Urinary metabolites were normalized to creatinine concentrations. Hepatic metabolites were normalized to wet liver weights.
For both human and mouse samples, samples were transferred to Waters polypropylene plastic total recovery vials (Part # 186002639) after extraction or preparation and stored in a Waters Acquity H class autosampler maintained at 8 °C until injection. In all cases, 10 μl of extract was loaded onto the column.
For analysis of enrichment in murine liver, metabolites were separated following the same LC procedure described above and detected using a Waters Premier Q-TOF operated in positive ion, full-scan mode. Leucine enkephalin was infused to ensure high mass accuracy. Enrichment data were corrected for natural isotope abundance based on theoretical isotope distribution, 13-carbon abundance skew and the purity of the labelled standard (3/97% [13C1]-NR/[13C1, D1]-NR). Quantification was performed on the Waters TQD as described above and used to determine the quantity of non-labelled and labelled metabolites. Separation and quantification of analytes was performed with a Waters Acquity LC interfaced with a Waters TQD mass spectrometer operated in positive ion multiple reaction monitoring mode. Enrichment analysis was performed with a Waters Q-TOF Premier mass spectrometer operated in positive ion, full-scan mode with the same LC conditions as described for non-enrichment experiments.

Statistical analyses

Statistical analyses were performed in GraphPad Prism version 6.00 for Windows, (La Jolla, CA). Sample sizes were sufficiently powered (1—β=0.8, α=0.05) to detect at least a 2-fold difference in NAD+ concentration. Murine liver data were analysed using two-way analysis of variance, whereas human blood cells were analysed using a repeated measure, two-way analysis of variance. Holm–Sidak and Tukey's multiple comparisons tests were performed when comparing more than two groups. AUCs in blood cells were calculated after subtracting pre-dose metabolite concentrations of each experimental series. For mouse data, AUCs were calculated as described57 after subtracting the saline group for that day and propagating error. Data are expressed as means±s.e.m. Group variances were similar in all cases. A P value<0.05 was considered significant.

Data availability

Metabolomic data have been deposited in Metabolights (https://www.ebi.ac.uk/metabolights/) under accession code MTBLS368.

Acknowledgments

This study was supported by grants from the Roy J. Carver Trust, ChromaDex and National Institutes of Health (R21-AA022371) to C.B., from the National Institutes of Health (R01-HL108379) to E.D.A. and from the Biotechnology & Biological Sciences Research Council (BB/N001842/1) to M.E.M. We thank Dale Wilson and staff at KGK Synergize, Inc. (London, Ontario, Canada) for conducting the n=12 clinical study.

References