Circadian Control of the NAD+ Salvage Pathway by CLOCK-SIRT1

See allHide authors and affiliations

Science  01 May 2009:
Vol. 324, Issue 5927, pp. 654-657
DOI: 10.1126/science.1170803


Many metabolic and physiological processes display circadian oscillations. We have shown that the core circadian regulator, CLOCK, is a histone acetyltransferase whose activity is counterbalanced by the nicotinamide adenine dinucleotide (NAD+)–dependent histone deacetylase SIRT1. Here we show that intracellular NAD+ levels cycle with a 24-hour rhythm, an oscillation driven by the circadian clock. CLOCK:BMAL1 regulates the circadian expression of NAMPT (nicotinamide phosphoribosyltransferase), an enzyme that provides a rate-limiting step in the NAD+ salvage pathway. SIRT1 is recruited to the Nampt promoter and contributes to the circadian synthesis of its own coenzyme. Using the specific inhibitor FK866, we demonstrated that NAMPT is required to modulate circadian gene expression. Our findings in mouse embryo fibroblasts reveal an interlocked transcriptional-enzymatic feedback loop that governs the molecular interplay between cellular metabolism and circadian rhythms.

Accumulating evidence reveals intriguing links between the circadian clock and cellular metabolism (13), which, at least in part, rely on epigenetic control and chromatin remodeling (4). The circadian regulator CLOCK has an intrinsic acetyltransferase activity, which enables circadian chromatin remodeling by acetylating histones (5) and nonhistone proteins, including its own partner BMAL1 (6). The histone deacetylase (HDAC) that counterbalances the histone acetyltransferase function of CLOCK is SIRT1 (7, 8), an enzyme whose activity is dependent on intracellular NAD+ levels (9). Although NAD+ is SIRT1’s natural cosubstrate, the reduced form of NAD+ (NADH) and the by-product of NAD+ consumption, nicotinamide (NAM), repress the activity of SIRT1 (9) and generate an enzymatic feedback loop on the HDAC function of this enzyme. Two main systems determine NAD+ levels in the cell, the de novo biosynthesis from tryptophan and the NAD+ salvage pathway (10). A critical step of the latter pathway is controlled by the enzyme nicotinamide phosphoribosyltransferase (NAMPT) (11), also known as visfatin or pre–B cell colony-enhancing factor (PBEF), which catalyzes the first step in the biosynthesis of NAD+ from NAM. NAMPT is implicated in cellular metabolism, senescence, and survival in response to genotoxic stress (1214).

Because of the role of SIRT1 in modulating clock function (7), we reasoned that circadian tuning may be achieved by NAD+ oscillating levels. Wild-type (WT) mouse embryo fibroblasts (MEFs) were serum-entrained, and cellular NAD+ levels were measured by liquid chromatography coupled to tandem mass spectrometry (LC/MSn) from total extracts prepared at various time intervals after serum shock (15). Cellular NAD+ showed a reproducible circadian oscillation of about 2.5-fold (Fig. 1A). The average NAD+ concentration of 25 pmol/μg protein (~60 μM) found in MEFs is in keeping with recent reports for rat axons using LC coupled to an ultraviolet detector (9 pmol/μg protein) (16), mouse erythrocyte using LC/MS/MS (368 μM) (17), and HEK293 cells using LC with matrix-assisted laser desorption/ionization (MALDI) MS (365 μΜ) (14). Note that cellular NAD+ oscillation is in phase with SIRT1 deacetylase activity and its phase is opposite to that of acetylation of histone H3 and BMAL1 (7).

Fig. 1

Circadian oscillation of NAD+ is controlled by the clock. (A and B) Cellular NAD+ was extracted from serum-entrained MEFs derived from WT (A), clock/clock mutant mice (c/c), and Cry1/Cry2 double-KO mice (cry) (B) at indicated time points and analyzed by LC/MSn. Three independent experiments were performed, and representative results are shown. All data are means ± SEM of three independent samples. (C) Average NAD+ levels. (means ± SEM of >50 independent samples). (D) Cellular NAM was extracted from serum-entrained MEFs derived from WT, c/c, and cry mice at indicated time points and analyzed by LC/MSn. (E) Average NAM levels (means ± SEM of >50 independent samples).

NAD+ levels did not oscillate in entrained MEFs originated from the clock/clock mutant mice (c/c) and from the arrhythmic Cry1/Cry2 double-knockout (KO) mice (cry) (Fig. 1B and fig. S1), which demonstrated that NAD+ oscillation is driven by the circadian clock. Total cellular NAM levels measured by LC/MSn also showed oscillation, albeit more moderate, with a phase opposite to that of NAD+. Also NAM oscillation was abolished in the MEFs from c/c and cry mice (Fig. 1D). Thus, the circadian clock controls the levels of these metabolites. This analysis also provided a clue: NAD+ and NAM levels are significantly lower in mutant MEFs than in WT cells (for NAD+ only 1.1 pmol/μg protein in c/c MEFs, about 5% of that in WT MEFs) (Fig. 1, C and E). This notion points to an involvement of the NAD+ salvage pathway and, because of its dynamic regulation, specifically to NAMPT, whose prominent function in NAD+ production has been demonstrated (1820). Furthermore, increased flux through the NAD+ salvage pathway is responsible for sirtuin-dependent responses even under conditions of unaltered steady-state NAD+ levels (12).

Next, we analyzed the expression of NAD+ salvage pathway metabolic enzymes. The nicotinamide mononucleotide adenylyltransferases, NMNAT1, NMNAT2, and NMNAT3, are central in NAD biosynthesis; they catalyze the adenylation of nicotinamide mononucleotide (NMN) or nicotinic acid mononucleotide (NaMN) by using the adenosine monophosphate (AMP) moiety of adenosine triphosphate to form NAD+ or nicotinic acid dinucleotide, respectively (Fig. 2A). The expression of these three enzymes is mildly oscillatory (fig. S2). Thus, our attention focused on NAMPT, which operates as rate-limiting enzyme in NAD+ production within the NAD+ salvage pathway (10, 14, 21). Expression of Nampt is robustly circadian in livers from WT mice, whereas oscillation is virtually absent in c/c mice (Fig. 2B). Rhythmic expression is observed also in serum-entrained WT MEFs, paralleling Dbp oscillation. Again, Nampt oscillation is abolished in c/c MEFs (Fig. 2C).

Fig. 2

The circadian clock controls Nampt gene expression. (A) Scheme of the NAD+ salvage pathway in mammals. (B) Nampt gene expression in livers from light-entrained WT and c/c mutant mice as measured by quantitative polymerase chain reaction (Q-PCR). Nampt gene expression at Zeitgeber time (ZT) 15 in liver from WT mice was set to 1. (C) Nampt and Dbp gene expressions in serum-entrained MEFs from WT and c/c mutant mice were quantified by Q-PCR. Expression at time 0 in WT MEFs was set to 1. All data are means ± SEM of three independent samples.

The Nampt promoter contains three putative, highly conserved, E-box sequences (CACGTG) (Fig. 3, A and B). Using promoter mutagenesis and transient expression in cultured cells, we demonstrate that the Nampt promoter is readily activated by CLOCK:BMAL1 through the E-boxes (Fig. 3C). Next, using chromatin immunoprecipitation (ChIP), we showed that CLOCK:BMAL1 physically and specifically associates to the E-boxes on the Nampt promoter in a time-dependent manner (Fig. 3, D and E), consistent with Nampt circadian expression. We have recently demonstrated that SIRT1 is in a complex with CLOCK:BMAL1 (7). Dual cross-linking ChIP assays showed that SIRT1 binds to the E-boxes in a time-dependent manner, following the circadian timing of CLOCK:BMAL1 recruitment (Fig. 3, D and E). Thus, as previously shown for Dbp and Per2 (7), CLOCK and SIRT1 contribute to circadian chromatin remodeling at the Nampt promoter. As NAD+ intracellular levels directly influence the HDAC activity of SIRT1 (2224), an enzymatic-transcription feedback loop seems to operate, in which NAD+ levels determine the oscillatory synthesis of NAMPT, the key enzyme in the NAD+ salvage pathway.

Fig. 3

Regulation of the Nampt promoter by CLOCK:BMAL1 and SIRT1. (A) Schematic diagram of regulatory elements in human Nampt promoter. TSS (transcription start site) primer region for Q-PCR, arrowheads; truncated forms of Nampt promoter used in (C), arrows. Putative transcription factor-binding sites are indicated: HRE, hypoxia response element; SP1, specificity protein 1; CRE, cyclic AMP–response element; AP-1, activator protein 1; GRE, glucocorticoid receptor response element. (B) Conserved E-boxes (capitalized) are shown. Numbers indicate positions from human TSS. (C) Schematic diagram of Nampt promoter constructs. Effect of CLOCK:BMAL1 (+CL/BM; black bars) on luciferase activity is shown. Luciferase activity of CLOCK:BMAL1 on the pGL4.10 was set as 1. All data are means ± SEM of three independent samples. (D) Representative results of ChIP assays analyzed by semiquantitative PCR. Dual cross-linked nuclear extracts isolated from MEFs after 16 or 24 hours serum shock and subjected to ChIP assay with antibodies against SIRT1, CLOCK, or BMAL1 or no antibody (ctrl). No antibody and 3′R primers were used as controls for immunoprecipitation and PCR, respectively. (E) Quantification of ChIP by Q-PCR. Q-PCR was performed on the samples described in (D). All data are means ± SEM of three independent samples.

To address whether NAMPT is critical in modulating clock function, we took advantage of FK866, a low-molecular-weight specific NAMPT inhibitor. FK866 lowers cellular NAD+ and NAM levels over a prolonged length of time (18) (figs. S3 and S4). Pharmacological inhibition of NAMPT significantly modifies the circadian expression of Per2 and Dbp in serum-stimulated MEFs (Fig. 4A) and causes an earlier onset of the circadian peak for both genes by 3 to 4 hours and increases the amplitude of oscillation by 30 to 40%. This effect of FK866 is highly similar to the one caused by inhibition of SIRT1 (7). Thus, we predicted that blocking NAMPT would modify BMAL1 acetylation, a regulatory event critical to clock function (6). Using the antibody against acetyl-BMAL1 recently developed in our laboratory (7), we showed that inhibition of NAMPT induces a considerable increase and a broader peak of Lys537 acetylation (Fig. 4B). This BMAL1 acetylation profile is basically equivalent to the one observed in MEFs and livers from Sirt1−/− mice (7).

Fig. 4

NAMPT modulates the circadian clock. (A) Per2 and Dbp gene expression levels in serum-entrained MEFs treated with 10 nM FK866 or ethanol (EtOH) as control (solvent: ctrl) were analyzed by Q-PCR. Highest value for each gene in EtOH-treated MEFs was set to 1. All data are means ± SEM of three independent samples. (B) BMAL1 Lys537 acetylation profile in serum-entrained MEFs treated with 10 nM FK866 or EtOH. Extracts prepared from indicated time points and BMAL1 acetylation were monitored (7). Input samples were probed with BMAL1- and GAPDH-specific antibodies. (C) Scheme of the transcription-enzymatic interplay by which the circadian machinery governs the intracellular levels of NAD+. The NAD+-dependent deacetylase SIRT1 is thereby controlling the oscillatory synthesis of its own coenzyme.

How intimate is the link between cellular metabolism and the circadian clock? Specifically, are metabolic pathways regulating the circadian machinery (3, 25, 26), or are molecular elements of the clock controlling metabolism (3, 27, 28)? The findings reported here demonstrate that both pathways exist in the cell, and each utilizes a discrete molecular mechanism of control in which the enzyme NAMPT occupies a pivotal position. Our results reveal the interlocking of two auto-regulatory systems, in which a classical transcription circadian loop is coupled to an enzymatic feedback loop (Fig. 4C). These findings point to the oscillation of NAD+ as a key regulatory step in the modulation of rhythms in metabolic tissues and peripheral clocks. Note that FK866 may be used for treatment of diseases that may be implicated in deregulated apoptosis or act as a sensitizer for genotoxic agents (18). Thus, our findings could have multiple implications including avenues to pharmacological intervention.

Supporting Online Material

Materials and Methods

Figs. S1 to S4


References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. Thanks to D. Piomelli for insightful discussions and help. Thanks also to L. Guarente, K. Yamagata, and members of the Sassone-Corsi's laboratory for help, reagents, and support. Y.N. was supported by the Japan Society for the Promotion of Science Postdoctoral Fellowships for Research Abroad. S.S. was supported by a postdoctoral fellowship from the American Heart Association, Western States Affiliate. Work supported by the Cancer Research Coordinating Committee of the University of California and NIH (R01-GM081634-01).
View Abstract

Navigate This Article