Nicotinamide Phosphoribosyl Transferase/Pre-B Cell Colony-Enhancing Factor/Visfatin Is Required for Lymphocyte Development and Cellular Resistance to Genotoxic Stress¹

Nicotinamide phosphoribosyl transferase (Nampt)⁴/pre-B cell colony-enhancing factor (PBEF)/visfatin is a multifunctional, ubiquitously expressed protein with potential immunoregulatory properties. This protein was originally identified as a likely cytokine involved in B cell development and hence designated PBEF (1). Pbef belongs to a group of genes defining an immune “activation signature”, being widely expressed within the immune system and positively regulated during in vivo and in vitro immune responses (2). Global gene expression studies have revealed that the expression of this gene is increased in virtually all activated immune cells, including T cells, B cells, monocytes, macrophages, dendritic cells, and neutrophils (3, 4, 5, 6, 7, 8, 9, 10). Extracellular expression of this protein was found to increase under inflammatory conditions, and recombinant forms of PBEF have been shown to exert an antiapoptotic effect on activated neutrophils (11), suggesting a potentially important role for PBEF in immunoregulation. More recently, PBEF has been renamed “visfatin” and described as an adipokine highly expressed by visceral fat (see Ref. 12 for review). Although its mode of action remains controversial (13), soluble visfatin has been found in the serum, a finding that has been interpreted as corroborating its potential role in cell-to-cell communication.

Based on the high similarity between Pbef and the nadV gene of Hemophilius ducrey, we previously demonstrated that PBEF also represents the mammalian Nampt, an enzyme catalyzing the condensation of nicotinamide with phosphoribosyl pyrophosphate, representing the first step in the salvage pathway allowing recycling of nicotinamide to NAD (14, 15). Unlike previous reports, however, Nampt has been found by us and others to be intracellular and not actively secreted, casting some doubts on its potential role as a cytokine (14, 16). Following a series of elegant in vitro and in vivo studies, Imai and colleagues have recently reconciled these two apparently opposing views (17). Based on the analysis of transfected and naturally expressing cells, these authors have suggested that this usually intracellular protein can be actively secreted by selected tissues (such as adipocytes). Both forms of the proteins retain enzymatic activity, leading to the biosynthesis of both intracellular and extracellular nicotinamide mononucleotide, depending on protein location (17).

A growing body of data indicates a possible relationship between NAD metabolism, cell survival, and inflammation, in particular in experimental settings in which cells are exposed to oxidative and/or genotoxic stress (18, 19). Genotoxic damage induced by oxidative stress, ionizing radiations, or alkylating agents activates a series of cellular responses aimed at repairing DNA damage and restoring cellular viability. Poly(ADP-ribosylation) mediated by the abundant nuclear enzyme poly(ADP-ribose) polymerase-1 (PARP-1) represents a well-characterized immediate response to DNA insult. PARP-1 binds to DNA strand breaks and catalyzes the transfer of successive units of the ADP-ribose moiety from NAD to several nuclear proteins, including PARP-1 itself. This posttranslational modification has been shown to facilitate DNA repair and it therefore plays a protective role in response to moderate genotoxic stress (18, 19). However, sustained PARP-1 activation caused by extensive DNA damage promotes a nonapoptotic form of cell death possibly contributing to inflammation (20). In keeping with this conclusion, structurally unrelated pharmacological PARP-1 inhibitors are of remarkable therapeutic efficacy in experimental models of inflammatory-related diseases (21, 22), and PARP-1^−/− mice are protected from endotoxic shock (23). Collectively, these findings suggest that excessive PARP activation may promote an inflammatory response by causing excessive cell death.

We previously documented that PBEF/Nampt expression is increased upon lymphocyte activation (14). To directly evaluate the role of Nampt in regulating lymphocyte development and survival during genotoxic stress, we have generated a mouse strain lacking Nampt expression in the T and B cell lineage using mice expressing a conditional allele of Nampt (Nampt-flox) and transgenic mice expressing the Cre recombinase under the control of an hCD2 promoter. We demonstrate herein that in contrast to its putative B cell-specific cytokine role, expression of Nampt is critically required for the development of both T and B lymphocytes. Hemizygote mice displayed reduced levels of intracellular NAD, in keeping with the important and rate-limiting role of Nampt in the biosynthesis of this coenzyme (15). Using a combination of genetic and pharmacological tools, we demonstrate in this work that Nampt regulates cellular sensitivity to genotoxic agents, and that this functional property requires its phosphorybosyltransferase enzymatic activity. Collectively, these data indicate that the nicotinamide salvage pathway represents the main biosynthetic route to NAD in T and B lymphocytes, and that Nampt plays an important role in regulating cell survival in response to stress.

A P1 artificial chromosome containing the gene encoding Nampt (see Fig. 1,A) was obtained from Geneservice (24). SacI and XhoI libraries were generated from this P1 artificial chromosome. An 8.3-kb SacI fragment was used as a 5′ recombination arm. Two contiguous XhoI fragments were used to generate a 10.4-kb 3′ recombination arm. The floxed region was amplified by PCR. This fragment encompasses exons 5 and 6 that encode amino acids located in the catalytic site and are required for enzymatic activity. These fragments were assembled into the pFRT-NEO-LOX2 plasmid (T. Van Reeth and C. Szpirer, unpublished) to generate the targeting construct (Fig. 1,B). The targeting vector was electroporated into E14 embryonic stem (ES) cells followed by G418 selection. Correctly targeted ES cells were identified by Southern blot after EcoRV restriction of genomic DNA and using a 5′ external probe. Several correctly targeted ES cell clones were identified, as shown in Fig. 1, C and F. Chimeric mice were generated using standard protocols (BioVallée). These mice were intercrossed with Rosa-FLP mice to delete the Neo selection marker, generating a floxed allele (Fig. 1 D). Positive/floxed mice were then interbred with hCD2-Cre mice to generate Nampt^flox/flox-Cre, Nampt^flox/+-Cre, and littermate control mice. Wild-type, floxed, and deleted alleles were identified by PCR using primers P1, TTC CAG GCT ATT CTG TTC CAG; P2, TCT GGC TCT GTG TAC TGC TGA; and P3, CCA ACC CAG ATT TCC AGC TA.

Thymus and lymph nodes were collected from 6-wk-old mice. Single-cell suspensions were prepared and stained with anti-TCRα/β- allophycocyanin, anti-CD19-PE, anti-CD4-PECy7, and anti-CD8-FITC Abs (eBioscience). The cells were analyzed on a FACSCanto II flow cytometer using the FACSDiva software (BD Biosciences).

The open reading frame of mouse Nampt was PCR-amplified and cloned into the pMSCV vector, upstream of an IRES-enhanced GFP (eGFP) reporter sequence. Site-directed mutagenesis of Nampt (D313A) in this plasmid was performed using a standard PCR-based protocol. To inhibit Nampt expression by RNA interference, short hairpin RNAs (shRNAs) were constructed in the pSicoR vector (25) against the following target sequences: shRNA 1, GAACTTTGTTACACTTGAATT; shRNA 2, GGGAATTGCTCTAATTAAATT; and scrambled shRNA, GGAATCTATTGTCCATATATT. Total cellular RNA was isolated from cells with TRIzol reagent (Invitrogen), according to the manufacturer’s instructions. RNA was reverse-transcribed by Moloney murine leukemia virus reverse transcriptase, using oligo(dT) primers. Specific mRNAs were amplified using the following primers: tryptophan 2,3-dioxygenase (TDO) forward, TGACACGCTCATGACCAAAT; TDO reverse, CCTTGTACCTGTCGCTCACA; nicotinic acid phosphoribosyltranferase (Napt) forward, TACTTGGGGCTAGAGGAGCA; Napt reverse, CAGACTCTAGCCAGGGCATC; Nampt forward, ATCTTTACACAGGACACCAGCGGGGAAC; Nampt reverse, TACATACGCACAGGCGCACACATAGG; ribosomal protein L32 (Rpl32) forward, GGCACCAGTCAGACCGATAT; Rpl32 reverse, CAGGATCTGGCCCTTGAAC.

The NIH-3T3 mouse fibroblast cell line, the embryonic kidney 293T human cell line, and the murine macrophage-like RAW 264.7 cell line were grown in DMEM supplemented with 5% FCS, 1 mM sodium pyruvate (Invitrogen), 2 mM l-glutamine (Invitrogen), essential amino acids (Invitrogen), and 0.05 mM 2-ME. Nicotinamide and nicotinic acid were obtained from Sigma-Aldrich. Peritoneal exudate cells were harvested from C57BL/6 mice (Harlan Nederland) 5 days following an i.p. injection of thioglycolate (1 ml of a 4% solution). Twenty-four hours after plating 293T cells in 6-well plates (10⁵ cells per well), the pMSCV-Nampt (wild-type and mutant) or the empty control was transiently transfected using Lipofectamine 2000 reagent (Invitrogen). The same plasmids were used to transfect the retroviral packaging cell line Phoenix (26), and the conditioned supernatants were collected 2 days later and used to infect NIH-3T3 cells. Infected cells were identified by eGFP expression and positively selected by fluorescent-activated cell sorting, generating stable cell lines. The pSicoR-shRNA plasmids targeted to Nampt and controls were transfected into 293T cells in combination with the pCMV-R8.91 and pMD2.G plasmids (encoding packaging and envelope proteins, respectively) to produce lentiviruses, following a previously described protocol (27). NIH-3T3 cells were infected at a multiplicity of infection of 15. Efficiency of infection was assessed by flow cytometry using the eGFP reporter expression and was determined to be >90%. The expression level of Nampt in the cell lines was analyzed by Western blotting using the 14A5 rat mAb, as previously described (14). Anti-actin rabbit polyclonal Ab was purchased form Sigma-Aldrich.

Cells (cell lines or primary cells) were incubated for 15 min in the presence of N-methyl-N′-nitro-N-nitrosoguanidine (MNNG) (Sigma-Aldrich), washed with PBS, and incubated overnight in fresh complete medium. Alternatively, cells were incubated overnight in the presence of dexamethasone, H₂O₂ or of TNF-α (30 ng/ml) (Roche) combined with cycloheximide (2 μg/ml) (Sigma-Aldrich). In some experiments, cells were preincubated for 6 h in medium containing FK866 (obtained from the National Institute for Mental Health, Chemical Synthesis and Drug Supply Program). After MNNG treatment, cells were again incubated in the presence of FK866. To assess viability, culture supernatants were collected, and cells were trypsinized, washed in PBS containing 1% BSA, and stained with propidium iodide (10 μg/ml). Percentage of cellular viability was determined by flow cytometry based on cell size (forward scatter) and propidium iodide exclusion.

Cells were cultured for 24 h on glass slides in 24-well plates. After MNNG treatment, cells were washed in PBS and fixed for 5 min at 4°C in a freshly prepared solution containing methanol and acetone (50:50). Cells were washed three times in PBS containing 0.05% Tween 20 and incubated overnight at 4°C with anti-apoptosis-inducing factor (AIF) Ab (Santa Cruz Biotechnology). A goat anti-rabbit IgG coupled to Alexa 594 (Molecular Probes) was used as a secondary reagent. Nuclei were counterstained with DAPI and cells were analyzed by fluorescent microscopy.

To determine the enzymatic activity of wild-type and mutant (D313A) mouse Nampt, both forms of the enzyme were transiently transfected into 293T cells. Cellular protein extracts were prepared in a 0.01 M NaH₂PO₄/Na₂HPO₄ buffer (pH 7.4), incubated in the presence of the AS2 polyclonal Ab specific for mouse Nampt (14), and immunoprecipitated using protein A-coupled Sepharose beads (Amersham Pharmacia Biotech). Immunoprecipitation beads were washed three times in the lysis buffer and split into two aliquots. One of these was used to test immunoprecipitation efficiency by Western blotting and the other one to assess Nampt enzymatic activity, as previously described (14). Briefly, the beads were incubated in 500 μl of a reaction mix containing 50 mM Tris (pH 8.8), 2 mM ATP, 5 mM MgCl₂, 0.5 mM 5-phosphoribosyl-1-pyrophosphate, and 5 μM [carbonyl-¹⁴C]nicotinamide (American Radiolabeled Chemicals). Reactions were incubated for 2 h at 37°C. Production of labeled nicotinamide mononucleotide from [¹⁴C]nicotinamide was analyzed by a precipitation-filtration assay by adding 50 μl of the reaction mix to 2 ml of acetone. This sample was then passed through an acetone-presoaked Whatman GF/A filter, the filter was rinsed three times with 2 ml acetone, and radioactivity ([¹⁴C]nicotinamide mononucleotide) was counted.

Intracellular NAD concentrations were determined by an enzymatic cycling assay following a previously described protocol (28), with slight modifications. Approximately 5 × 10⁵ cells were lysed by freeze/thaw in liquid nitrogen in 200 μl of a buffer containing 100 mM Na₂CO₃ and 20 mM NaHCO₃. Cellular extracts were centrifuged and pellets were discarded. Samples (20 μl/well, in triplicate) were placed in 96-well MaxiSorp plates (Nunc). A cycling buffer containing the following compounds was prepared: 125 mM Tris-HCl (pH 8.8), 1.25 mM phenazine ethosulfate (Sigma-Aldrich), 0.625 mM MTT (Sigma-Aldrich), 0.25 mg/ml alcohol dehydrogenase (Sigma-Aldrich), and 1.25% BSA. This cycling buffer was prewarmed at 37°C, and 160 μl was added to each well. The cycling reaction was initiated by adding in each well 20 μl of 6 M ethanol prewarmed at 37°C. The plate was incubated at 37°C and the OD at 570 nm was measured after 5, 10, 15, and 20 min using an ELISA plate reader. Serial dilutions of NAD were used as a standard.

To circumvent the early embryonic lethality of mice lacking Nampt expression (Ref. 17 and our own unpublished observation), a conditional null allele was generated by introducing LoxP sites that flank exons 5 and 6 of Nampt, as illustrated in Fig. 1. Conditional mutant animals lacking Nampt expression in the lymphocyte lineage were obtained by interbreeding with a transgenic mouse expressing the Cre recombinase under the control of the hCD2 promoter (hCD2-Cre, hereafter referred to as Cre) that drives the expression of the Cre transgene in T lymphocytes in the thymus beginning at the CD4⁺CD8⁺ double-positive cells and in B lymphocytes from the pre-B cell stage (29). Nampt^flox/flox-Cre and Nampt^flox/+-Cre mice were born at the expected Mendelian ratio and developed normally to adulthood. The cellularity and subset composition of the major lymphoid organs (including thymus, spleen, and lymph nodes) did not differ between Nampt^flox/+- Cre and control (including wild-type (wt), Nampt^flox/flox, and wt-Cre) mice (see Fig. 2, A–E, supplemental Fig. S1,⁵ and data not shown). In marked contrast, Nampt^flox/flox-Cre mice displayed a marked reduction in thymic cell numbers, in excess of 95%, with an almost complete blockade at the level of double-negative T cells (Fig. 2,A). Flow cytometry analysis of spleen (data not shown) and lymph node cell suspensions (Fig. 2, B–E) revealed a near complete loss of peripheral T and B cells in these mice. These results suggest an important role for Nampt in the differentiation of both lymphocyte subsets. To determine whether the few lymphocytes present in the periphery of Nampt^flox/flox-Cre mice were cells that escaped the Cre-mediated deletion of the floxed Nampt allele, or were cells that could develop and survive in the absence of Nampt, we performed PCR analysis of genomic DNA extracted from lymph nodes or from the tail of corresponding mice (Fig. 2,F). While the floxed allele was almost completely deleted in the lymph nodes of heterozygous Nampt^+/flox-Cre mice (Fig. 2,F, lanes 13–16), the floxed alleles were still present in the lymph nodes of homozygous Nampt^flox/flox-Cre mice (Fig. 2 F, lanes 17–20). This observation suggests that peripheral lymphocytes in these mice derived and expanded from rare progenitors in which the floxed alleles were not completely ablated by the Cre recombinase. These observations indicate that Nampt expression is required for optimal B and T cell development.

To evaluate the constitutive role of Nampt in controlling intracellular NAD levels in vivo, we analyzed expression of Nampt protein and intracellular NAD concentrations in Nampt^flox/+-Cre cells expressing a single Nampt allele (referred to as Nampt hemizygous). Western blot analysis on total thymus extracts revealed a moderate but consistent reduction in Nampt protein expression in hemizygous mice when compared with Nampt^flox/flox mice (Fig. 3,A). Accordingly, these mice displayed a significant reduction in intracellular NAD levels (∼60% of control values, Fig. 3,B). This finding is in keeping with the concept that Nampt represents the rate-limiting step allowing NAD biosynthesis from nicotinamide (15) and strongly suggests that the major developmental block observed in mice lacking Nampt expression in the lymphocyte lineage is due to the inability of these cells to synthesize NAD from alternative precursors. In keeping with this hypothesis, wild-type thymocytes were found to selectively express Nampt, lacking detectable levels of mRNAs encoding Napt and TDO, the enzymes allowing nicotinic acid and tryptophan, respectively, to contribute to NAD biosynthesis (Fig. 3,E). Accordingly, splenocytes were found exquisitely sensitive to pharmacological inhibition of the nicotinamide salvage pathway, as shown by their reduced survival in vitro when exposed to graded doses of the Nampt inhibitor FK866 (30, 31) (Fig. 3,F). The inhibitory properties of FK866 were antagonized by high doses of nicotinamide (known to reverse the inhibitory activity of FK866 on Nampt, see Ref. 31) but not by exogenous nicotinic acid (Fig. 3,G). In contrast, both nicotinamide and nicotinic acid were found to rescue the Nampt- and Napt-expressing human cell line THP-1 (31) from FK866-induced cell death, confirming that exogenous nicotinic acid is a bona fide NAD precursor for Napt-expressing cells (Fig. 3,H). To evaluate the possible functional relevance of the decreased NAD content observed in lymphocytes expressing a single Nampt allele, thymocytes from both control and Nampt hemizygous mice were exposed for 15 min to the alkylating agent MNNG, which is known to induce cell death in a PARP-1- and NAD-dependent manner (20). After 16 h of incubation in fresh media, cell viability was analyzed by propidium iodide staining and flow cytometry (Fig. 3,C). As a control, thymocytes were exposed to the synthetic glucocorticoid dexamethasone (Fig. 3 D), which is known to induce apoptosis in a PARP-1-independent manner (32, 33). Reduced expression of Nampt led to increased sensitivity of thymocytes to MNNG-induced, but not to dexamethasone-induced, cell death, suggesting that Nampt regulates the ability of cells to adapt to genotoxic stress.

To further confirm the close relationship between Nampt levels and sensitivity to genotoxic stress, we undertook a detailed analysis using established cell lines suitable for in vitro manipulations, thus allowing the controlled expression of wild-type and/or mutant forms of the Nampt protein. Several lines of NIH-3T3 murine fibroblasts expressing high or low levels of the Nampt protein were generated using a recombinant retroviral vector approach. Expression of the Nampt was evaluated by Western blot, and a representative experiment is shown in Fig. 4,A. Cells transduced with Nampt-encoding vector expressed increased levels of the corresponding protein, while cells transduced with Nampt-specific shRNAs displayed lower levels of the target protein, in the same order of magnitude as previously observed in hemizygous vs wild-type thymocytes. Two shRNA constructs directed against distinct regions of the mRNA led to a similar reduction in intracellular Nampt, while neither the empty vector nor a vector encoding a scrambled shRNA corresponding to shRNA 1 affected Nampt expression levels. As expected, intracellular NAD levels correlated with Nampt expression. Cells overexpressing Nampt displayed a 15–25% increase in total intracellular NAD level, whereas shRNA 1 expression led to a reduction in NAD levels ranging from 20 to 40% when compared with control cells (see Fig. 4 B for a representative experiment). Note that Nampt appears as a closely spaced doublet in NIH-3T3 cells, indicative of a possible posttranslational modification of the protein that was not further investigated in this study. All cell lines examined to date maintained a stable phenotype following long-term in vitro culture and were indistinguishable from control cells in terms of cell viability and proliferative capacity when maintained in complete media (data not shown).

To confirm the observations performed in thymocytes expressing reduced levels of Nampt, the transduced cell lines were exposed to MNNG as previously described (Fig. 5). Cells expressing low levels of this enzyme displayed a remarkable increased sensitivity to MNNG-induced cell death (Fig. 5,A), while infection of cells with a scrambled, control shRNA did not alter cell sensitivity to genotoxic stress (Fig. 5, G and H). Conversely, cells overexpressing Nampt were partially resistant to MNNG (Fig. 5,B). A similar response was observed when cells were exposed to the oxidative agent H₂O₂, also known to induce cell death in a PARP-dependent manner (34) (Fig. 5, C and D). Of note, decreased expression of the Nampt did not sensitize cells to apoptosis induced by TNF-α (Fig. 5,E), while cells overexpressing the Nampt were not protected against TNF-α-induced cell death, and they often displayed increased sensitivity to this proapoptotic agent (Fig. 5 F, an observation that was not pursued futher during this study).

PARP-1-mediated cell death has been described as a caspase-independent event, initiated by the translocation of AIF from the mitochondria to the nucleus (34). To evaluate whether Nampt activity was able to inhibit mitochondrial AIF release in response to MNNG, control cells and cells overexpressing Nampt were treated with MNNG, immunostained using anti-AIF Abs, and analyzed by fluorescence microscopy. As shown in Fig. 6, MNNG treatment led to the nuclear localization of AIF in control cells (note the double-stained nuclei in Fig. 6,B), while cells transfected with a Nampt construct displayed a cytoplasmic pattern of AIF staining compatible with its mitochondrial localization (Fig. 6 D). This observation indicates that Nampt modulates cell viability in response to MNNG by affecting an early event preceding AIF release.

To clearly demonstrate that the ability of Nampt to regulate cell death in response to genotoxic agents is linked to its role in NAD metabolism, a mutant protein lacking the phosphoribosyltransferase activity was generated. The mutant (D313A) was designed based on amino acid sequence alignment with several type II phosphoribosyltransferases and structural data available for the quinolinate phosphorybosyltransferase from Thermotoga maritima and Mycobacterium tuberculosis (35). Human 293T cells were transfected with the constructs encoding wild-type or mutant murine Nampt. Both plasmids led to expression of equivalent amounts of mouse Nampt as judged by Western blot using a murine Nampt-specific Ab (Fig. 7,A). The overexpressed, wild-type, or mutant mouse Nampt was immunoprecipitated and its enzymatic activity was evaluated in the presence of radiolabelled nicotinamide. As shown in Fig. 7,B, the mutant D313A Nampt protein lacked detectable enzymatic activity and, accordingly, did not increase intracellular NAD levels upon transfection (Fig. 7,C). In fact, transfection of the D313A mutant led to a small but significant decrease (−10%, p = 0.0103) in intracellular NAD content. This observation is best explained in light of the finding that the Nampt enzyme is a dimer in solution (36), suggesting that a catalytically inactive enzyme may act as a dominant-negative mutant. In agreement with the previous observations, only cells overexpressing a catalytically active form of Nampt were significantly protected against MNNG-induced cell death (Fig. 7 D). A similar result was observed after stable expression of wild-type and D313A Nampt in NIH-3T3 cells (data not shown).

To further confirm the role of the enzymatic activity of Nampt in the regulation of sensitivity to genotoxic stress, we used a pharmacological inhibitor of Nampt (FK866) (30, 31) Addition of FK866 in the culture media generally led to a drastic reduction in intracellular NAD (Figs. 8A, 9, A and C, and Ref. 30). NIH-3T3 cells were incubated in control media or in media supplemented with FK866 and treated with MNNG as previously described. FK866 did not affect cell viability in the control group, while it caused reduced survival of cells treated with low doses of MNNG (Fig. 8,B). Inhibition of PARP-1 activity by phenanthridinone (Ptd) prevented cell death induced by both MNNG and a combination of MNNG and FK866 (Fig. 8,C), confirming the predominant role of PARP overactivation in this experimental setting. The use of FK866 also allowed us to evaluate the effect of reduced intracellular NAD levels on cells of different origins such as the murine, macrophage-like RAW 264.7 cell line and ex vivo-collected thioglycolate-elicited peritoneal murine macrophages. As shown in Fig. 9, FK866 led to a drastic reduction in intracellular NAD levels (Fig. 9, A and C) and enhanced sensitivity to MNNG-induced toxicity (Fig. 9, B and D) in both cell types, further confirming the important role of the enzymatic activity of Nampt in regulating cell sensitivity to genotoxic stress. Therefore, the observations reported herein strongly suggest that Nampt affects cellular sensitivity to stress through its ability to regulate intracellular NAD levels.

The main objective of this study was to better understand the physiological role of the multipurpose protein Nampt/PBEF/visfatin, whose expression has been found to be up-regulated during immune and inflammatory responses in both human and animal models. Although originally described as a soluble cytokine potentially involved in B cell development (PBEF) (1), or glucose metabolism (visfatin) (12, 17), the data presented in this study further support the concept that this protein is a major determinant of intracellular NAD biosynthesis in cells of the immune system (14). In particular, expression levels of this protein in thymocytes and in cell lines correlated with levels of intracellular NAD (Figs. 3, 4, and 7), indicating a major role of the nicotinamide salvage pathway in NAD biosynthesis in lymphocytes. Of note, liver cells expressing a single Nampt allele have been recently shown to express control levels of NAD (17), in keeping with the notion that these cells can synthesize NAD from alternative precursors such as nicotinic acid and tryptophan (37). The observed embryonic lethality of Nampt KO mice (Ref. 17 and our own unpublished observations), the lack of lymphocyte survival in the absence of Nampt expression, and the reduced intracellular levels of NAD in mice expressing a single Nampt allele confirm that 1) the major role of Nampt in vivo is to catalyze the first enzymatic reaction allowing NAD biosynthesis from the nicotinamide precursors, and 2) lymphocytes are strictly dependent on the nicotinamide salvage pathway for NAD biosynthesis, a finding that is further confirmed by the high sensitivity of thymocytes to a Nampt-specific inhibitor (see Fig. 3). Based on these considerations, the major developmental arrest observed in Nampt^flox/flox-Cre mice is best explained by assuming that Cre-mediated deletion of both Nampt alleles abolishes the NAD biosynthetic capacity of developing cells, thus causing their premature death. It is noteworthy that expression of Nampt is ubiquitous, while both TDO and Napt, allowing NAD biosynthesis from tryptophan and nicotinc acid, respectively, display a more limited tissue distribution.

The observation that lymphoid cells expressing a single Nampt allele display altered NAD metabolism led us to further investigate the possible developmental and functional consequences of reduced intracellular NAD levels. Much to our surprise, cells expressing reduced intracellular NAD concentrations secondary to altered Nampt levels did not display an altered developmental program in vivo. In particular, no developmental anomalies were found in thymocytes from Nampt single allele-expressing mice (as assessed by analysis of CD4, CD8, CD25, and CD44 expression on thymocytes; see Fig. 2 and supplemental Fig. S1). Moreover, peripheral lymph nodes from Nampt hemizygous mice displayed a normal size and cellularity (65.5 ± 10.8 × 10⁶ vs 75.8 ± 0.4 × 10⁶ cells, mean from pooled axillary, inguinal, and mesenteric lymph nodes from, respectively, three wt and four single Nampt allele-expressing mice). Peripheral T cells from Nampt hemizygous mice were also found to adequately respond to stimulation to conventional agonists (such as anti-CD3 and anti-CD28 Abs, see supplemental Fig. S2). Similarly, established cell lines incubated in the presence of FK866 and displaying a nearly complete loss in intracellular NAD levels (Figs. 8 and 9) retained cell viability for several days in culture (data not shown and Ref. 31). Collectively, these data suggest that cells can support wide variations in intracellular NAD levels for an extended period of time, with no apparent functional consequences under conventional culture conditions.

Based on the large body of evidence suggesting a major role for NAD in controlling cell survival in response to PARP-activating cellular insults (18, 19), we therefore investigated whether cells expressing altered NAD metabolism displayed an increased sensitivity to environmental stressors. Nampt levels appeared to regulate the ability of cells to respond to cellular stress mediated by PARP-activating insults, leading to an uncontrolled intracellular NAD consumption. Accordingly, both genetic (Figs. 3 and 5) and pharmacological (Figs. 8 and 9) manipulations causing decreased Nampt activity strongly sensitized cells to genotoxic stress. In agreement with previous reports (38, 39), overexpression of an active form of the Nampt enzyme led to increased cellular resistance to DNA-alkylating agents (Figs. 5 and 7), indicating an important role for this enzyme in regulating cellular resistance to NAD-depleting situations. Collectively, our observations indicate that while cells seem to cope with large variations in intracellular NAD levels under standard conditions, they appear extremely sensitive to variations in Nampt levels and/or activity when exposed to genotoxic stress. Although the experimental conditions generally used to evaluate cell resistance to stress (exposure to chemicals) are of dubious physiological significance, it is generally assumed that they help revealing the capacity of cells to resist to natural sources of chronic oxidative stress, as generated by oxidative metabolism or possibly inflammatory conditions. Further studies, however, are warranted to examine the role of Nampt in a more chronic and physiological setting.

The mechanism whereby Nampt regulates cellular sensitivity in response to PARP-activating stresses remains to be clearly established. In particular, several independent observations point to a possible role for sirtuins (39, 40, 41, 42), a class of NAD-dependent enzymes, in regulating cell survival in response to stress. This conserved family of enzymes comprises seven members in mammals (SIRT1 to SIRT7) that use NAD as a cosubstrate to catalyze the deacetylation and/or mono-ADP-ribosylation of several intracellular targets (43, 44). Since poly(ADP-ribosyl)ation reactions lead to a decrease in intracellular NAD (the sirtuin substrate) and the concomitant increase in nicotinamide (a well-described sirtuin inhibitor), it has been suggested that this family of enzymes may represent key sensors of the NAD/nicotinamide intracellular status (42). However, the precise identification of individual sirtuin members regulating cell sensitivity in response to stress has been difficult to establish. In particular, both nuclear (SIRT1 (45, 46) and SIRT6 (47)) and mitochondrial (SIRT3 and SIRT4 (39)) sirtuins have been found to regulate cell survival in response to stress. Moreover, whereas increased Nampt expression has been found to protect cells against DNA-alkylating agents in a sirtuin-dependent manner, Nampt overexpression also confers resistance to apoptosis induced by topoisomerase inhibitors, but in a sirtuin-independent fashion (39), suggesting a complex relationship between sirtuin activity and resistance to genotoxic stress.

Collectively, the available observations indicate a major role for intracellular NAD levels in protecting cells against a variety of insults. Of note, it has been recently demonstrated that Nampt plays a major role in preserving cell viability by increasing mitochondrial NAD biosynthesis (39). Since large differences in relative concentrations of mitochondrial vs cytosolic NAD may occur in distinct cell types (48), further work may be required to establish whether the mechanism by which NAD regulates cell survival is unique or variable according to tissue origin.

In any case, our observations concur with a previous report and indicate that increased Nampt expression inhibits AIF nuclear localization, indicating that Nampt regulates a protective step upstream of this important apoptotic pathway (39).

The high expression of Nampt by cells of the immune system is therefore best explained by assuming that this protein represents an adaptive biological response to cope with an increased NAD consumption, rather than a potential cytokine. Indeed, 1) in contrast to most cytokines/chemokines known to date, this protein is highly conserved through evolution (14) and is constitutively expressed by most tissues (data not shown and Ref. 1); 2) it lacks a conventional signal sequence and, despite being found as a soluble protein in the serum, only a few cell types have been found to release this protein in the supernatant (17, 49); and 3) the ability of this protein to interact with a membrane-borne receptor has been recently challenged (17). Finally, it is noteworthy that expression of Nampt by nonlymphoid tissues is not sufficient to rescue T and B cell development in Nampt^flox/flox-hCD2 Cre mice, suggesting that, at least for lymphocytes, Nampt regulates NAD metabolism in a cell-autonomous fashion, a conclusion difficult to rationalize with a putative role of Nampt as a soluble cytokine.

We thank Thierry Van Reeth for providing the pFRT-NEO-LOX2 plasmid and Rosa-FLP mice, and Georgette Vansanten for cell sorting.

The authors have no financial conflicts of interest.