The copper-sensing transcription factor Mac1, the histone deacetylase Hst1, and nicotinic acid regulate de novo NAD+ biosynthesis in budding yeast

NADH (NAD+) is an essential metabolite involved in various cellular biochemical processes. The regulation of NAD+ metabolism is incompletely understood. Here, using budding yeast (Saccharomyces cerevisiae), we established an NAD+ intermediate–specific genetic system to identify factors that regulate the de novo branch of NAD+ biosynthesis. We found that a mutant strain (mac1Δ) lacking Mac1, a copper-sensing transcription factor that activates copper transport genes during copper deprivation, exhibits increases in quinolinic acid (QA) production and NAD+ levels. Similar phenotypes were also observed in the hst1Δ strain, deficient in the NAD+-dependent histone deacetylase Hst1, which inhibits de novo NAD+ synthesis by repressing BNA gene expression when NAD+ is abundant. Interestingly, the mac1Δ and hst1Δ mutants shared a similar NAD+ metabolism–related gene expression profile, and deleting either MAC1 or HST1 de-repressed the BNA genes. ChIP experiments with the BNA2 promoter indicated that Mac1 works with Hst1-containing repressor complexes to silence BNA expression. The connection of Mac1 and BNA expression suggested that copper stress affects de novo NAD+ synthesis, and we show that copper stress induces both BNA expression and QA production. Moreover, nicotinic acid inhibited de novo NAD+ synthesis through Hst1-mediated BNA repression, hindered the reuptake of extracellular QA, and thereby reduced de novo NAD+ synthesis. In summary, we have identified and characterized novel NAD+ homeostasis factors. These findings will expand our understanding of the molecular basis and regulation of NAD+ metabolism.

The NAD ϩ pool is maintained by multiple NAD ϩ biosynthesis pathways, which are conserved from bacteria to humans. Depending on the cell types, growth conditions, and availability of specific NAD ϩ precursors, one pathway may dominate the others. In yeast, NAD ϩ can be synthesized de novo from tryptophan or salvaged from intermediates, such as NA, NAM, and NR (Fig. 1A). In the de novo pathway, tryptophan is converted to QA through a series of enzymatic reactions catalyzed by Bna2, Bna7, Bna4, Bna5, and Bna1 ( Fig. 1A) (16). QA is then phosphoribosylated by Bna6, producing nicotinic acid mononucleotide (NaMN). Because several steps in the de novo pathway require molecular oxygen as a substrate (Bna2, Bna4, and Bna1), cells grown under anaerobic conditions rely on the salvage pathways for NAD ϩ synthesis (16). In the NA/NAM salvage pathway, yeast cells retrieve NAM from NAD ϩ consumption reactions or uptake NA from the environment via NA transporter Tna1, leading to NaMN production. NaMN is also the converging point of the de novo pathway and NA/NAM salvage, which is converted to NAD ϩ by NaMN adenylyltransferases (Nma1/2) (17) and glutamine-dependent NAD ϩ syn-thetase (Qns1) (18) (Fig. 1A). NR salvage also contributes to NAD ϩ synthesis. In yeast, NR is assimilated into NMN or NAM, catalyzed by the NR kinase Nrk1 (19) and nucleosidases Urh1/Pnp1/Meu1 (15), respectively. NMN is directly converted to NAD ϩ via NMN adenylyltransferases (Nma1/2) (17), whereas NAM merges into the NA/NAM salvage pathway. It has also been shown that small NAD ϩ precursors, such as NR, NA, NAM, and QA, constantly exit and re-enter yeast cells (20 -22), which constitutes an extended NAD ϩ precursor pool.
The complex and dynamic flexibility of NAD ϩ precursors makes studying NAD ϩ metabolism complicated. For example, NAM can both replenish NAD ϩ pools and inhibit the activity of NAD ϩ -consuming enzymes. In addition, metabolites of the de novo pathway appear to have additional function. The de novo pathway is also known as the kynurenine pathway (KP) or tryptophan degradation pathway, and alterations of the KP metabolites have been linked to several brain disorders (11,23). Interestingly, KP metabolites have been shown to exhibit both neuroprotective (kynurenic acid (KA)) and neurotoxic (QA and 3-hydroxykynurenine (3-HK)) effects (11,23,24) (Fig. 1A). KA is produced from kynurenine (KYN) by the KYN aminotransferase. Yeast cells also produce KA from KYN by the Aro8/9 aminotransferases (25); however, the function of KA in yeast remains unclear.
Studying the regulation of specific NAD ϩ biosynthetic pathways has been challenging in part due to the redundancy and interconnections among them. Employing the properties of yeast cells that constantly release and retrieve small NAD ϩ precursors (20 -22), we have previously carried out precursor-specific genetic screens to identify novel NAD ϩ homeostasis factors in yeast (1,21,26). In this study, we developed a QA release-based reporter system targeting the de novo branch of NAD ϩ metabolism. The hypothesis was that cells with abnormal de novo NAD ϩ synthesis activities would show altered QA release. The mac1⌬ mutant was among the top hits that showed increased QA release. MAC1 encodes a copper-sensing transcription factor (27)(28)(29), and our studies are the first to link Mac1 to NAD ϩ homeostasis. Here, we characterized the mac1⌬ mutants as well as additional factors that regulate the de novo pathway. Our studies help provide a molecular basis underlying the interconnection and cross-regulation of NAD ϩ biosynthesis pathways.

A cell-based reporter assay to identify factors that modulate the de novo pathway
Saccharomyces cerevisiae cells have been reported to secrete QA (22). We exploited this phenomenon and established a cross-feeding assay using the bna4⌬npt1⌬nrk1⌬ mutant as "recipient cells" (which depend on QA for growth) and mutants of interest as "feeder cells." In this system, recipient cells cannot grow on standard growth media (which lack QA). When feeder cells are placed in proximity, feeder cell-released QA supports recipient cell growth by "cross-feeding." This assay determines relative levels of total QA released by feeder cells and can be considered as readout for the de novo pathway activity. As shown in Fig. 1B, recipient cells were spread onto synthetic minimal (SD) medium plates as a lawn. Next, WT, bna1⌬, bna4⌬, and bna6⌬ feeder cells were spotted on top of the lawn and allowed to grow. Recipient cells near the feeder spots appeared as satellite colonies after 3 days, and the number and size of the satellite colonies positively correlate with the amount of QA released from the feeders. According to the function of Bna proteins in the de novo pathway (Fig. 1A), we expected bna1⌬ and bna4⌬ cells to release no QA and bna6⌬ cells to release more QA. Indeed, as shown in Fig. 1B, both bna1⌬ and bna4⌬ feeder cells failed to support the growth of the recipient cells. In contrast, bna6⌬ cells supported the growth of recipient cells more robustly than WT cells. We also tested a few mutants that have been associated with regulation of the de novo pathway. The NAD ϩ -dependent histone deacetylase Hst1 represses BNA gene expression. During NAD ϩ deprivation, decreased Hst1 activity results in de-repression of BNA genes (30). Cells lacking NPT1 and TNA1 are defective in NA/NAM salvage and NA transport, respectively, and thus have decreased NAD ϩ levels (31,32). In addition, Tna1 was reported to also function as a QA transporter (22). Fig. 1C showed that hst1⌬, npt1⌬, and tna1⌬ mutants all released more QA. These results demonstrated that mutants with altered QA levels and de novo activities could be identified using this system.

Transcription factor Mac1 is a novel NAD ؉ homeostasis factor
Next, we used the haploid yeast deletion collection as feeder cells to identify mutants with altered QA release (Fig. 1D). After incubation at 30°C for 3 days, we scored the cross-feeding activity (which indicates the level of QA release) of each mutant by comparing the diameter of the cross-feeding zones with that of the WT. A relative QA release score was assigned to each mutant with the WT baseline score set at 0. Scores of Ϫ4, Ϫ3, and Ϫ2 indicate decreased QA release; scores of ϩ2, ϩ3, and ϩ4 indicate increased QA release. To eliminate false-positives, mutants with scores of Ϫ4, Ϫ3, 3, and 4 (169 mutants) were re-examined. A total of 81 mutants passed the secondary screens on both SD and synthetic complete (SC) growth media (Table S1). Among the 12 mutants that show strongest QA release (score ϩ4), most are known NAD ϩ homeostasis factors, including hst1⌬, npt1⌬, pnc1⌬, nat3⌬, and their interacting partners (Table S1). MAC1 encodes a copper-sensing transcription factor (27)(28)(29), which has not been associated with NAD ϩ homeostasis. To verify that observed phenotypes are due to the featured mutations and not to secondary cryptic mutations in the deletion collection, we reconstructed all deletion mutants used in this study. Fig. 1E showed that mac1⌬ released more QA, and deleting HST1 did not further increase QA release in the mac1⌬ mutant. Increased QA release correlates with increased levels of extracellular QA determined by quantitative liquid assays ( Fig. 2A). Similar results were obtained using either bna4⌬npt1⌬nrk1⌬ ( Fig. 2A, left) or bna1⌬npt1⌬nrk1⌬ ( Fig. 2A, right)  Mac1 is a NAD ؉ homeostasis factor and 3-hydroxyanthranilic acid (3-HA) (Fig. 1A). Interestingly, cells did not appear to retain excess QA intracellularly ( Fig. 2A). Thus, increased QA may be converted to NAD ϩ or released extracellularly. To examine whether Mac1 and Hst1 function in the same pathway to regulate NAD ϩ homeostasis, we determined the levels of NR, NA/NAM, and NAD ϩ /NADH in WT, mac1⌬, hst1⌬, and hst1⌬mac1⌬ cells. Similar to the hst1⌬ mutants, mac1⌬ cells showed moderate increases in intracellu- Figure 1. Identification of yeast mutants with increased de novo NAD ؉ synthesis activity. A, simplified model of Saccharomyces cerevisiae NAD ϩ synthesis pathways. The de novo NAD ϩ synthesis is mediated by Bna2, -7, -4, -5, -1, and -6, leading to the production of NaMN. Yeast cells also produce NaMN from the NA/NAM salvage pathway, which is then converted to NAD ϩ by Nma1/2 and Qns1. NR salvage contributes to NAD ϩ synthesis in part depending on the NA/NAM salvage pathway. L-TRP, L-tryptophan; NFK, N-formylkynurenine; NaAD, deamido-NAD ϩ . Abbreviations of protein names are shown in parentheses. Bna2, tryptophan 2,3-dioxygenase; Bna7, kynurenine formamidase; Bna4, kynurenine 3-monooxygenase; Bna5, kynureninase; Bna1, 3-hydroxyanthranilate 3,4-dioxygenase; Bna6, quinolinic acid phosphoribosyl transferase; Nma1/2, NaMN/NMN adenylyltransferase; Qns1, glutamine-dependent NAD ϩ synthetase. B, deletions of BNA1 and BNA4 abolish QA release, whereas deletion of BNA6 increases QA release. QA-dependent recipient cells (npt1⌬nrk1⌬bna4⌬) were plated onto SD plates as a lawn. Next, WT, bna1⌬, bna4⌬, and bna6⌬ feeder cells were spotted on top of the lawn and allowed to grow for 3-4 days at 30°C. The number and size of the satellite colonies surrounding the feeder spots positively correlate with the amount of QA released from the feeders. C, deletions of HST1, NPT1, and TNA1 increase QA release. D, overview of the genetic screen to identify mutants with increased QA release. Haploid single-deletion mutants (feeder cells) were spotted onto a lawn of QA-dependent recipient cells, whose growth relies on QA released from the feeder cells in a dose-dependent manner. E, deletions of MAC1 and HST1 confer increased QA release. For clarity, inverse images are shown.
Mac1 is a NAD ؉ homeostasis factor lar levels of NR (Fig. 2B) and NA/NAM (Fig. 2C). Increased expression of NA and NR transporters was reported in hst1⌬ mutants (30,33), which could contribute to observed increases of intracellular NR and NA/NAM. The hst1⌬ mutant has also been shown to maintain higher steady-state NAD ϩ levels (30). Consistent with this study, mac1⌬, hst1⌬, and hst1⌬mac1⌬ mutants showed WT levels of NAD ϩ /NADH during early logphase growth (6 h) (Fig. 2D); however, as cells entered late log phase (16 h), these mutants showed higher NAD ϩ /NADH levels when compared with WT cells (Fig. 2E). This result suggested that mac1⌬-and hst1⌬-induced QA production was indeed coupled to increased de novo NAD ϩ synthesis. However, these mutants might also uptake more NA from the medium due to increased TNA1 expression (Fig. 3B), which could contribute to observed increases in NAD ϩ /NADH levels. We therefore determined NAD ϩ /NADH levels in cells grown in NA-free medium to eliminate the contribution of NA. As shown in Fig. 2F, an increase in NAD ϩ /NADH level was still observed in mac1⌬ and hst1⌬ cells. Interestingly, we were able to observe this NAD ϩ /NADH increase in early log-phase cells grown in NA-free medium. These results were in line with the observations that the presence of NA (and/or NA salvage to NAD ϩ ) inhibits de novo activity (30,34). Overall, these studies showed a connection between increased QA production and increased de novo NAD ϩ synthesis activity in mac1⌬ and hst1⌬ cells. QA levels were determined in both growth media (released) and cell extracts (intracellular). Feeding of collected lysate and growth media to QA-dependent recipient cells was conducted in triplicate. Relative QA levels were determined using both bna4⌬npt1⌬nrk1⌬ (left) and bna1⌬npt1⌬nrk1⌬ (right) as QA-dependent recipient cells. B, the hst1⌬ and mac1⌬ mutants show increased NR levels. C, the hst1⌬ and mac1⌬ mutants show increased intracellular NA and NAM levels. D, measurements of intracellular NAD ϩ levels. Deleting HST1 and MAC1 does not increase NAD ϩ levels in log-phase cells (6 h). E, deleting HST1 and MAC1 significantly increases NAD ϩ levels in late log-phase cells (16 h). F, deleting HST1 and MAC1 significantly increases NAD ϩ levels in log-phase cells (6 h) grown in NA-free media. Error bars, S.D. The p values are calculated using Student's t test (*, p Ͻ 0.05).

Mac1 and Hst1 co-repress de novo NAD ؉ synthesis gene expression
Mac1 is a copper-regulated transcription factor, which activates the expression of genes involved in high-affinity copper transport in response to copper deprivation (27)(28)(29). Our studies showed that deleting MAC1 increases de novo NAD ϩ syn-thesis activity and that the mac1⌬ and hst1⌬ mutants shared a similar effect on NAD ϩ homeostasis (Fig. 2). To understand how Mac1 regulates NAD ϩ metabolism, we carried out gene expression profile and differential gene expression studies in mac1⌬, hst1⌬, and WT cells (three biological triplicates of each strain were analyzed). A total of 370 genes were differentially

Mac1 is a NAD ؉ homeostasis factor
expressed in mac1⌬ and WT cells (255 up-regulated, 115 down-regulated), and 498 genes were differentially expressed in hst1⌬ and WT cells (342 up-regulated, 156 down-regulated) (Files S1 and S2). Gene ontology (GO) term enrichment analysis showed that de novo NAD ϩ metabolism-associated GO terms were significantly enriched for the 498 differentially expressed genes in hst1⌬ and WT cells. For the 370 differentially expressed genes in mac1⌬ and WT cells, "copper ion import" and "iron ion homeostasis" GO terms were significantly enriched, which were in line with reported Mac1 function. In addition, "de novo NAD ϩ biosynthetic process from tryptophan" and "quinolinate biosynthetic process" GO terms were also enriched, supporting a role for Mac1 in de novo NAD ϩ metabolism. KEGG pathway mapping analysis also identified "tryptophan metabolism" as the most significantly enriched pathway (first, ranked by p values) in genes co-upregulated by mac1⌬ and hst1⌬. Consistent with these studies, expression of the de novo NAD ϩ metabolism-associated BNA genes (BNA1, -2, -4, -5, and -6) was up-regulated in hst1⌬ and mac1⌬ cells (Fig. 3A, left). On the other hand, high-affinity copper transporter CTR1 was specifically down-regulated in mac1⌬ cells (Fig. 3A, right). These studies showed that in addition to being a transcription activator, Mac1 appeared to repress BNA gene expression.
Differential expression of additional NAD ϩ homeostasis genes was also observed in mac1⌬ and hst1⌬ cells (Fig. 3A), including NAD ϩ intermediate transporters TNA1 and NRT1, and components of the phosphate-sensing PHO signaling pathway (26). Co-up-regulation of TNA1 (NA and QA transporter) and NRT1 (NR transporter) in hst1⌬ and mac1⌬ cells (Fig. 3A, left) was in line with the function of Hst1 as a NAD ϩ -regulated transcriptional repressor of NAD ϩ homeostasis genes, and with the fact that hst1⌬ and mac1⌬ cells shared similar NAD ϩ phenotypes. Expression of phosphatases (PHO5, -11, -12) and high-affinity phosphate transporter (PHO84) was increased in hst1⌬ cells (Fig. 3A, left); however, it was either unchanged or decreased in mac1⌬ cells (Fig. 3A, right). To further understand the roles of these factors in NAD ϩ metabolism and to validate the expression profile results, we first carried out qPCR analysis in WT, mac1⌬, and hst1⌬ cells. Fig. 3B showed that mac1⌬ mutation increased BNA gene expression to a similar extent as seen in the hst1⌬ mutant. The expression patterns of BNA, PHO, and CTR1 genes were in agreement with expression profile analysis (Fig. 3, A and B). Next, we examined whether altered PHO gene expression indeed correlated with PHO signaling activities and whether it contributed to increased de novo NAD ϩ synthesis. Previous studies associated PHO activation with increased NR and NA/NAM production (21,26). We determined PHO activity by measuring the activity of rAPase (repressible acid phosphatase) Pho5, a periplasmic phosphatase activated by PHO signaling (35). As shown in Fig. 3C, PHO activity was moderately increased in the hst1⌬ mutants but not in the mac1⌬ mutant, which was consistent with observed PHO gene expression patterns (Fig. 3, A and B). Interestingly, observed PHO activation in hst1⌬ cells appeared partially independent of the conventional PHO transcription factors Pho4-Pho2 complex, because deleting PHO4 did not completely reduce PHO activation in hst1⌬ cells. This study suggested that PHO activation is not essential for enhanced de novo NAD ϩ synthesis in mac1⌬ cells, and that Hst1 can activate specific PHO downstream targets independent of Pho4-Pho2.
It has been shown that during copper deprivation, Mac1 protein is stabilized and turns on copper homeostasis genes, including CTR1, whereas high copper concentrations facilitate the degradation of Mac1 (36,37). Copper deprivation was also reported to induce BNA gene expression in strains defective in copper homeostasis (29,38,39). We therefore examined whether copper alterations would affect BNA gene expression in WT cells. As controls, expression of CTR1 (Fig. 3D, top right) was induced by copper deprivation (0 M), which was repressed by nutritional level (10 M, normal) and a high level (250 M) of copper. Interestingly, both low-and high-copper conditions appeared to induce the expression of a few BNA genes (Fig. 3D). Consistent with this, we showed that cells grown in media containing a nutritional copper level (10 M) produced the lowest amount of QA and that both copper deprivation (0 M) and high copper stress (Ն250 M) increased QA production (Fig.  3E). High copper stress facilitates Mac1 degradation, which could explain the de-repression of BNA genes and QA production. However, copper might affect BNA gene expression independent of Mac1 because copper deprivation also moderately increased BNA gene expression and QA production. Interestingly, NAD ϩ and NADH levels did not correlate with increased QA production under copper stress conditions (Fig. 3F). In fact, both NAD ϩ and NADH levels were slightly reduced by high copper stress (Fig. 3F). It is possible that NAD ϩ and its derivatives were consumed to offset the damages and redox changes caused by copper stresses. Together, these studies demonstrated a link between copper stress, de novo QA production, and NAD ϩ metabolite homeostasis.
We next asked how might Mac1 and Hst1 co-repress BNA genes. Because mac1⌬ and hst1⌬ did not show a synergistic effect on NAD ϩ homeostasis (Figs. 1E and 2 (A-E)), they likely function in a linear pathway to regulate BNA genes. As a transcription activator, Mac1 may activate HST1 expression, which then represses BNA gene expression. This scenario was unlikely because expression of HST1 was not altered in mac1⌬ cells and vice versa (Fig. 3B). Mac1 has been shown to bind to the copperresponse element (CuRE), TTTGC(TG)C(A/G) sequences, which are present in the promoters of copper homeostasis genes (40,41). However, this consensus Mac1-binding motif is absent in the promoters of BNA genes. Hst1 and associated proteins have been reported to bind to the promoter of BNA2 (42). We therefore examined whether Mac1 required Hst1 and vice versa to bind to the BNA2 promoter. To address this question, ChIP studies of various BNA2 promoter fragments using HA-tagged Hst1 (Hst1-HA) and Mac1 (Mac1-HA) were carried out. Fig. 4A (top) showed Mac1 and Hst1 protein levels in strains used for this study. Deleting HST1 did not significantly alter Mac1 protein levels; likewise, deleting MAC1 also did not alter Hst1 protein levels. Note that Mac1 was overexpressed for the ChIP studies because the Mac1 protein level was low and unstable in cells grown under nutritional levels of copper (standard conditions). Because all mac1⌬-associated NAD ϩ homeostasis phenotypes were observed under standard conditions, we overexpressed Mac1 to enhance the sensitivity and to allow Mac1 is a NAD ؉ homeostasis factor further studies to be carried out under normal copper conditions. Fig. 4A (bottom) illustrates BNA2 promoter fragments (BNA2-a, -b, -c, -d, and -e) examined in ChIP studies. As shown in Fig. 4B, Hst1 was bound to all five BNA2 promoter fragments, with the most significant binding activity near the BNA2-e region. Deleting MAC1 did not significantly alter Hst1 binding to BNA2 promoter fragments, indicating that Hst1 did not require Mac1 to bind to BNA2 promoter (Fig. 4B). On the other hand, Mac1 only showed significant binding near the BNA2-e region (Fig. 4C). Interestingly, Mac1 binding to the BNA2-e region was abolished in hst1⌬ cells, indicating that Hst1 is essential for Mac1 binding. It remained possible that Mac1 also bound to other regions of the BNA2 promoter; however, our assay conditions were not sensitive enough to detect the binding. It was also likely that the Mac1 overexpression might hinder its binding activity. This was less likely because cells overexpressing Mac1 still showed normal QA release (Fig. 4D,  left), indicating that overexpression did not significantly affect Mac1 function at BNA promoters. Among Hst1-associated protein complexes, the Hst1-Sum1-Rfm1 complex appeared to play a key role in BNA repression because both sum1⌬ and rfm1⌬ mutants were also identified in our screen and showed increased QA release (Fig. 4D, right). Overall, these studies suggested that Mac1 works with the Hst1-associated repressor complex to regulate BNA genes, and deleting MAC1 or HST1 alone is sufficient to abolish the repression.

Regulation of de novo NAD ؉ synthesis by NA salvage
An unexpected phenomenon was observed while developing the QA cross-feeding reporter system, which indicated that de novo NAD ϩ synthesis is also regulated by an Hst1/Mac1-independent mechanism. In the reporter system, QA released by the feeder cells (mutants of interest) supported the growth of the QA-dependent recipient cells (readout for QA release). When grown on standard rich medium YPD, all strains tested, including the hst1⌬ and mac1⌬ mutants, appeared to be defective in QA production and failed to support the recipient cell growth (Fig. 5A, panel 1). On the other hand, cells grown on minimal synthetic medium SD were able to produce QA in an Hst1-and Mac1-regulated manner (Fig. 5A, panel 2). NA was shown to repress de novo NAD ϩ synthesis, and the amount of NA differed significantly in standard SD (0.4 mg/liter, ϳ3.3 M) and YPD (ϳ13-61 mg/liter, ϳ100 -500 M). However, NA-mediated repression was suggested to require NA salvage into NAD ϩ and NAD ϩ -dependent activation of Hst1 (30). These observations prompted us to examine whether NA played a role in this seemingly Hst1/Mac1-independent repression. As shown in Fig. 5A (panels 2-5), QA levels released by WT, mac1⌬, and hst1⌬ cells indeed inversely correlated with NA levels (0 mg/liter, NA-free SD; 0.4 mg/liter, SD/normal; Ն61 mg/liter, high NA) in growth media. In contrast, NAM enhanced QA release, and it did not further increase QA release

Mac1 is a NAD ؉ homeostasis factor
in hst1⌬ and mac1⌬ cells (Fig. 5A, panel 6). Although NAM can be deamidated into NA, it also inhibits the activity of the Sir2 family deacetylases (43,44). Therefore, NAM was anticipated to induce hst1⌬-like phenotypes. These results indicated that high NA-mediated inhibition of QA production did not require Hst1 and Mac1 (Fig. 5A, panels 2-4). However, elevated QA release observed in WT cells grown in NA-free media could be rescued by restoring NAD ϩ levels via supplementing NR (Fig. 5B), and that NR-mediated rescue required functional HST1 (Fig. 5B). Thus, it appeared that NA could inhibit de novo pathway activity in both an NAD ϩ /Hst1-dependent (Fig. 5B) and an NAD ϩ /Hst1-independent manner (Fig. 5A), depending on the NA levels in growth media.
Interestingly, NA-induced inhibitions were not completely due to repression of BNA genes. In WT cells, the expression of most BNA genes inversely correlated with the levels of NA in growth media (Fig. 5C). However, in hst1⌬ and mac1⌬ cells, elevated BNA expression was not altered by NA treatments Figure 5. NA inhibits the activity of the de novo pathway. A, levels of QA production inversely correlate with the concentrations of NA present in the growth media. Results show released QA levels in WT, hst1⌬, and mac1⌬ cells grown in the following media: YPD, NA-free SD, SD (normal), and SD supplemented with NA (ϩNA) or NAM (ϩNAM). NAM is a competitive inhibitor of the NAD ϩ -dependent deacetylases. B, NA inhibits QA production in an NAD ϩ /Hst1-dependent manner. Elevated QA release in WT cells grown in NA-free media was rescued by restoring NAD ϩ levels via supplementing NR (at 20 M, a NAD ϩ precursor). Hst1 was required for the rescue. C, most BNA gene expression in WT cells shows an inverse correlation with NA levels in growth media. Cells were grown in NA-free, SD (normal) and ϩNA (at 123 mg/liter) media. Relative gene expression determined by qPCR (normalized to TAF10) is shown. D, BNA gene expression in hst1⌬ and mac1⌬ cells is not altered by NA levels. Cells were grown in NA-free, SD (normal), and ϩNA (at 123 mg/liter) media. Relative gene expression determined by qPCR (normalized to TAF10) is shown. E, QA uptake is inhibited by NA in a dose-dependent manner, which is alleviated by increasing QA concentrations. The QA-dependent npt1⌬nrk1⌬bna4⌬ mutant was grown in SD containing different levels of QA and NA. The extent of growth (A 600 ) of this mutant correlates with its QA uptake efficiency. F, NA-mediated inhibition of QA production can be rescued by supplementing specific intermediates of the de novo pathway. Results show relative QA production in WT cells grown in SD supplemented with NA (40 mg/liter) and L-Kyn (Fig. 5D), indicating that mac⌬ and hst1⌬ cells still produced elevated levels of QA regardless of NA levels. This suggested that high NA might actually hinder growth of the QA-dependent recipient cells, masking the QA cross-feeding readout (Fig.  5A, panels 1, 3, and 4). The NA transporter Tna1 has been reported to also transport QA (22). It was shown that the tna1⌬ mutant failed to transport QA (1-3.3 M) and thus exhibited growth defects in NA-free media under anaerobic conditions in which the de novo NAD ϩ pathway is completely blocked and the cells depend on QA transport for growth (22). Because Tna1 can transport both NA and QA, it is possible that high NA competes with QA for Tna1-mediated transport. To address this, we examined whether high NA could inhibit QA uptake by exploiting the QA-dependent npt1⌬nrk1⌬bna4⌬ recipient cell mutant. This mutant cannot utilize NA (due to npt1⌬ mutation); therefore, its de novo pathway gene expression is not altered by NA treatments. As a result, the extent of cell growth (A 600 ) of this mutant in media containing different levels of QA and NA correlates with its QA uptake efficiency. As shown in Fig. 5E, NA indeed inhibited QA uptake in a dose-dependent manner. At 5-10 M NA (approximate NA levels in SD), no inhibition of QA uptake was observed. High NA (Ն100 M) showed the most significant inhibition of QA uptake when QA was supplemented at 1-5 M, which was similar to reported QA levels released by WT cells in SD (22). Observed NA inhibitions of QA uptake could be alleviated by increasing QA concentrations (at 10 -50 M) (Fig. 5E), suggesting that NA competes with QA for Tna1-mediated transport. It remains possible that NA may also inhibit intracellular QA utilization. One possible target is the enzymatic activity of Bna6, a quinolinate phosphoribosyl transferase (QPRT) that converts QA into NaMN (Fig. 1A). However, it has been shown that NA does not inhibit QPRT activity (45,46). Together, these results suggested that high NA could also inhibit the de novo pathway activity by competing with QA transport, which is independent of NA salvage into NAD ϩ and Hst1-mediated BNA gene repression.

Mac1 is a NAD ؉ homeostasis factor
Nevertheless, Hst1-mediated BNA gene repression remains the major mechanism of NA/NAD ϩ -induced inhibition of de novo QA/NAD ϩ synthesis. To further understand this regulation, we first determined which BNA gene was most sensitive to NA inhibition using QA production as readout. A simplified de novo QA production pathway was illustrated in the bottom panel of Fig. 5F. The QA-dependent npt1⌬nrk1⌬bna1⌬ mutant was used as the recipient cells whose growth correlates with the levels of QA released by the WT donor cells. As shown in Fig. 5F (top), levels of QA released by WT donor cells were inhibited by NA (at Յ40 mg/liter NA to minimize its inhibition of QA uptake). This inhibition could be rescued by supplementing L-KYN or 3-HK (Fig. 5F, top) suggesting that the expression of BNA2 or BNA7 in the WT donor cells was most sensitive to NA-induced Hst1-mediated repression (Fig. 5F,  bottom). Because BNA7 expression appeared less sensitive to NA alterations (Fig. 5C) and hst1⌬ mutation (Fig. 5D), we directly examined BNA2. As shown in Fig. 5G, overexpressing BNA2 was sufficient to increase QA release (left) and rescue NA inhibition (right) in WT cells. As controls, overexpressing BNA4 did not show significant effects on QA release. These results indicate that BNA2 expression is most sensitive to Hst1and NA-mediated repression, and therefore it is likely a ratelimiting step in de novo QA production.

Discussion
In this study, we identified Mac1 as a novel NAD ϩ homeostasis factor in yeast. Mac1 was previously characterized as a copper-sensing transcription activator (27)(28)(29). Cells lacking MAC1 shared similar NAD ϩ phenotypes with the hst1⌬ mutant. The NAD ϩ -dependent histone deacetylase Hst1 has been suggested to be an NAD ϩ sensor, which inhibits de novo NAD ϩ synthesis by repressing BNA gene expression when NAD ϩ is abundant (Fig. 6) (30). We showed that deleting either MAC1 or HST1 was sufficient to abolish BNA gene repression (Fig. 3, A and B). Although BNA gene promoters lack Mac1binding sequences CuRE, Mac1 proteins likely associate with BNA promoters with help from the Hst1 complex (Fig. 4C). On the other hand, Hst1 binding to the BNA2 promoter does not require functional Mac1 (Fig. 4B), indicating that binding of the Hst1 complex precedes Mac1 at the BNA2 promoter. Together, these results suggest that Mac1 proteins work with the Hst1containing repressor complexes (Hst1, Rfm1, and Sum1) formed at the BNA promoters to repress gene expression. It remains unclear how Mac1 proteins work with the Hst1 complexes to mediate BNA gene repression. Although Mac1 and Hst1 complex components did not appear to directly interact in systematic pulldown studies (47,48), they may interact under Figure 6. A model depicting the regulation of QA production and the de novo pathway activity by NA/NAD ؉ -dependent Hst1, copper-regulated Mac1, and NA-inhibited QA transport. NAD ϩ can be synthesized de novo from tryptophan via the BNA proteins (box 2) or salvaged from intermediates and small precursors such as NA, NAM, and NR (box 1). For clarity, this diagram centers on the regulation of the de novo pathway (shaded area). Under NAabundant conditions, NA/NAM salvage is the preferred NAD ϩ biosynthesis route, and BNA genes are silenced by the NAD ϩ -dependent Hst1. NAM can be deamidated into NA and contributes to NAD ϩ biosynthesis, but it also inhibits Hst1 activity. The copper-sensing transcription factor Mac1 appears to work in concert with the Hst1-containing repressor complex to repress BNA genes. Yeast cells also release and re-uptake small NAD ϩ precursors, such as NA, NAM, NR, and QA; however, the mechanisms are not completely understood. It has been suggested that vesicular trafficking and vacuolar function play a role in the production and release of NAD ϩ precursors. Uptake of NAD ϩ precursors is mostly mediated by specific transporters, including Nrt1 (for NR) and Tna1 (for NA and QA). High levels of NA in the growth media hinder the re-uptake of released QA, thereby reducing the de novo activity.

Mac1 is a NAD ؉ homeostasis factor
specific conditions. Mac1 and Sum1 were reported to share a common interacting partner, histone H4 (Hhf1) (47,48), suggesting a possible interaction model. Interestingly, both Mac1 and Hst1 proteins showed higher binding activities at the BNA2 promoter regions near the start codon (Fig. 4, B and C), suggesting that Mac1 and Hst1 binding to these regions is more critical for BNA2 gene repression. Further studies are required to understand how Mac1 contributes to Hst1-mediated BNA gene repression.
Our studies also uncovered additional NA-mediated regulation of the de novo pathway. The NAD ϩ pool is maintained by NA and NR salvage pathways and de novo NAD ϩ synthesis from tryptophan (Fig. 6). Under NA-abundant conditions, NA salvage is the preferred NAD ϩ synthesis route (30,34), which leads to Hst1 activation and BNA gene repression. As cells consume NA, eventually NA/NAD ϩ deprivation decreases Hst1 activity, leading to de-repression of BNA genes, QA production, and de novo NAD ϩ synthesis. Supporting this, cells grown in NA-free media showed increased BNA gene expression (Fig.  5C) and QA production (Fig. 5A). Supplementing NR to NA/NAD ϩ -deprived cells was sufficient to restore their NAD ϩ levels and repression of QA production (Fig. 5B). It appears that QA production occurs before complete depletion of NA/NAD ϩ because QA release is detectable in log-phase WT cell culture, which is further increased in stationary phase culture (22). Most excess QA (if not converted to NAD ϩ ) is released extracellularly (Fig. 2A) and can re-enter cells via Tna1 (also a high-affinity NA transporter) (22) as the NA levels decline. Supporting this, the high QA-producing hst1⌬ and mac1⌬ mutants showed more significant increases in NAD ϩ levels when NA was depleted in the growth media (Fig. 2, E and  F). Thus, NA inhibits de novo NAD ϩ synthesis both by repressing QA production (Fig. 5, B and C) and by competing with QA transport (Fig. 5E); the latter is particularly significant when cells are grown in rich media, which contain high levels of NA (ϳ100 -500 M) (Fig. 5A, panel 1). We have also shown BNA2 expression is most sensitive to NA-mediated repression (Fig.  5F), and BNA2 is a rate-limiting step in QA production (Fig.  5G). Overall, these studies provide further understanding of the dynamic interactions between NA salvage and de novo NAD ϩ synthesis.
Additional nutritional stresses also play a role in regulating de novo NAD ϩ synthesis. Copper has been shown to negatively regulate Mac1-dependent transcription activation. Under copper deprivation, Mac1 remains stable and activates expression of genes in the high-affinity copper transport system (36,37,49). Upon copper repletion, copper binding to Mac1 causes an intramolecular interaction between its N-terminal DNA binding domain and C-terminal trans-activating domain, resulting in a transcriptionally inactive state. Under high-copper conditions, Mac1 is quickly degraded to avoid copper toxicity (36,37,49). Thus, the copper-bound form of Mac1, although transcriptionally inactive, is likely to bind to and repress BNA gene promoters because our studies were carried out under nutritional copper levels. High copper stress caused Mac1 degradation (36,37,49), which was in line with increased BNA gene expression (Fig. 3D) and QA production (Fig. 3E). Interestingly, copper deprivation also increased BNA expression (Fig. 3D) and QA production (Fig. 3E). It is possible that the transcriptionally active form of Mac1 binds to CuRE promoters more efficiently than to the BNA promoters during copper deprivation. Alternatively, copper may regulate BNA gene expression independent of Mac1. Together, these studies demonstrate that a dynamic Mac1 pool is shared between the Hst1-containing repressor complexes at BNA promoters and the transcriptionactivating complexes at CuRE-containing promoters. However, NAD ϩ levels did not appear to correlate with increased QA production under copper stress conditions (Fig. 3F). It is likely that NAD ϩ is consumed or converted to NAD ϩ derivatives, such as NADH, NADP ϩ , and NADPH, to maintain proper redox state and to prevent and counteract copper stress-induced oxidative damages. In addition, copper may inhibit the activities of NAD ϩ metabolic enzymes. For example, copper was shown to inhibit the activity of QPRT (50,51). It is also possible that under stress conditions, it is beneficial for cells not to increase NAD ϩ levels to maintain an optimal metabolic state.
It appears that once QA is produced from de novo synthesis, it is either released to the extracellular environment or converted to NAD ϩ (Fig. 6). In yeast, released QA can re-enter cells to support NAD ϩ synthesis via the NA transporter (22). Yeast cells also release and re-uptake other small NAD ϩ precursors that arise from NA/NAM salvage and NR salvage, such as NA, NAM, and NR (1,21,52). It has been suggested that vesicular trafficking and vacuolar function play a role in the production and release of these NAD ϩ precursors (1,21,52). However, the detailed mechanisms are not completely understood. There appears to be a storage pool for NAD ϩ precursors generated from the salvage pathways. For example, cytoplasmic NR is released to the environment or transported into the vacuole for storage if not converted to NAD ϩ . Likewise, vacuolar NR may exit the vacuole to support NAD ϩ synthesis in the cytoplasm. The equilibrative nucleoside transporter Fun26 (human lysosomal hENT homolog) controls the balance of NR and possibly other nucleosides between the vacuole and the cytoplasm (26). It is possible that vacuolar degradation of NAD ϩ intermediates coincides with NAD ϩ salvage. NAD ϩ intermediates may enter the vacuole through vesicular transport and autophagy and are then broken down into smaller precursors for storage or reuse. On the other hand, when excess QA is made in the cytoplasm, it is mostly released and not stored in the vacuole. Interestingly, the fun26⌬ mutant showed increased QA production/release (Table S1). This is likely due to the blockage of NR salvagemediated NAD ϩ synthesis because NR is trapped inside the vacuole in the fun26⌬ mutant. We have shown that NAD ϩ depletion causes BNA gene de-repression and that NR supplementation restores the repression in an Hst1-dependent manner (Fig. 5B). In line with this, cells lacking the NAM deamidase PNC1 also showed increased QA production (Table S1). The pnc1⌬ mutation causes accumulation of NAM (21), which is anticipated to inhibit Hst1 activity. Therefore, increased QA production observed in the fun26⌬ and pnc1⌬ mutants was likely due to loss of Hst1-mediated repression of BNA genes. Together, these studies demonstrate a complex interconnection between the de novo and salvage pathways.

Mac1 is a NAD ؉ homeostasis factor
In summary, our studies have uncovered novel NAD ϩ homeostasis factors. Hst1 and Mac1 may play important roles in the cross-regulation of de novo NAD ϩ synthesis and other nutrient-sensing pathways. Copper levels determine the conformation and stability of the Mac1 proteins, resulting in corresponding changes in de novo QA production and copper homeostasis. In addition, increased phosphate-sensing PHO activity was also observed in the hst1⌬ mutant. Activation of PHO has been associated with increased NR (26) and NA/NAM (21) production, suggesting that Hst1 may also affect NR and NA/NAM salvage activities via regulating specific components in the PHO pathway. Recent studies have demonstrated that NAD ϩ metabolism is a therapeutic target for several human diseases. This strategy was more effective when specific defects in NAD ϩ biosynthesis were identified and associated with the progression of diseases (9,10). Supplementation of specific NAD ϩ precursors can also be combined with the use of genetic modifications and inhibitors of specific NAD ϩ biosynthesis steps to help channel the precursors' flow through a more efficient NAD ϩ synthesis route (9,53,54). Recent studies have also shown that inhibiting the activities of de novo pathway enzymes, such as tryptophan 2,3-dioxygenase (TDO; Bna2 in yeast) (55) and kynurenine 3-monooxygen (KMO; Bna4 in yeast) (56), as well as increasing the ratio of neuroprotective KA over neurotoxic QA or 3-HK, may help to alleviate a few neurological disorders (24,55). Understanding the molecular basis and interconnection of multiple NAD ϩ metabolic pathways is important for the development of disease-specific therapeutic strategies. Overall, our studies contribute to the understanding of the regulation of NAD ϩ metabolic pathways and may provide insights into the underlying mechanisms of disorders associated with aberrant NAD ϩ metabolism.

QA cross-feeding spot assays
The npt1⌬nrk1⌬bna1⌬ and npt1⌬nrk1⌬bna4⌬ mutants (which depend on QA for growth) were used as "recipient cells," and yeast strains of interest were used as "feeder cells." First, recipient cells were plated onto SD as a lawn (ϳ10 4 cells/cm 2 ). Next, ϳ2 ϫ 10 4 cells of each feeder cell strain (2 l of cell suspension made in sterile water at A 600 of 1) were spotted onto the lawn of recipient cells. Plates were then incubated at 30°C for 3 days. The extent of the recipient cell growth indicates the levels of QA released by feeder cells.

Genetic screen using the yeast deletion collection
The haploid yeast deletion collection (ϳ4,500 strains) established in the BY4742 strain was acquired from Open Biosystems (63). To screen for mutants with altered QA release, 2 l of each strain was directly taken from the frozen stock and then spotted onto SD plates spread with the npt1⌬nrk1⌬bna4⌬ recipient cells at a density of ϳ10 4 cells/cm 2 . After incubation at 30°C for 3 days, we scored the cross-feeding activity (which indicates the level of QA release) of each mutant by comparing the diameter of the cross-feeding zones with that of the WT. Mutants were assigned a score of Ϫ4 through 4 (WT ϭ 0). Scores of Ϫ4, Ϫ3, and Ϫ2 indicate decreased QA release; scores of ϩ2, ϩ3, and ϩ4 indicate increased QA release. To eliminate false-positives, mutants with scores of Ϫ4, Ϫ3, 3, and 4 (169 mutants) were re-examined. A total of 81 mutants passed the secondary screen on both SD and SC growth media (Table S1).

Measurement(s) of NAD ؉ , NADH, QA, NR, and NA/NAM
Total intracellular levels of NAD ϩ and NADH were determined using enzymatic cycling reactions as described (62). Levels of NAD ϩ intermediates (QA, NR, and NA/NAM) were determined by a liquid-based cross-feeding bioassay as described previously (20,21,26) with modifications. To prepare cell extracts for intracellular NAD ϩ intermediates determination, ϳ150 A 600 unit (1 A 600 unit ϭ 1 ϫ 10 7 cells/ml) donor cells grown to late logarithmic phase in SC (ϳ16 h of growth from an A 600 of 0.1) were collected by centrifugation and lysed by beadbeating (Biospec Products) in 800 l of ice-cold 50 mM ammonium acetate solution. After filter sterilization, 100 l of clear extract was used to supplement 8-ml cultures of recipient cells with starting A 600 of 0.05 in SC. To determine extracellular NAD ϩ intermediates levels, supernatant of donor cell culture was collected and filter-sterilized, and then 500 l was added to 7.5 ml of recipient cell culture with total starting A 600 of 0.05 in SC. A control culture of recipient cells in SC without supplementation was included in all experiments. For measuring relative QA levels, npt1⌬nrk1⌬bna4⌬ and npt1⌬nrk1⌬bna1⌬ mutants were used as recipient cells. The npt1⌬bna6⌬pho5⌬ recipient cells were used to measure relative NR levels. To measure relative NA/NAM levels, the bna6⌬nrk1⌬nrt1⌬ recipient cells were grown in NA-free SC. After incubation at 30°C for 24 h, growth of the recipient cells (A 600 ) was measured and normalized to the cell number of each donor strain. A 600 readings were then converted to concentrations of QA, NR, and NA/NAM using the standard curves established as described previously (Fig. S1) (20, 21).

Mac1 is a NAD ؉ homeostasis factor Gene expression profile analysis
Approximately 40 A 600 unit cells grown to early logarithmic phase in SD (6-h growth from an A 600 of 0.1) were collected by centrifugation. Total RNA was extracted using the GeneJET RNA purification Kit (Thermo Scientific). High-quality DNAfree RNA for sequencing was made using the RNA Clean and Concentrator kit (Zymo Research). For each sample, 10 l of RNA (at 100 ng/l) was used to generate 3Ј-Tag-Seq libraries (a single initial library molecule per transcript, complementary to 3Ј-end sequences, was generated). The libraries were then sequenced by single-end sequencing on the HiSeq 4000 (DNA Technology Core, University of California, Davis, CA). Three biological triplicates of each strain were sequenced. Differential gene expression testing was conducted using a single-factor analysis of variance model in the limma-voom Bioconductor pipeline (Bioinformatics Core, University of California, Davis, CA). Prior to analysis, genes with expression less than 1 count per million reads were filtered, leaving 6,023 of 7,126 genes. The multidimensional scaling plot shows the distance between samples based only on normalized counts. In this experiment, the WT samples clustered together and were very different than the mutants, as expected. GO enrichment analyses of differential gene expression results were conducted using Kolmogorov-Smirnov tests in the Bioconductor package topGO. For each GO term, the KS test tests whether p values for genes annotated with that GO term are smaller than for genes not annotated with that GO term. KEGG enrichment analyses were conducted using Wilcoxon rank sum tests, in conjunction with the Bioconductor package KEGGREST. For each KEGG pathway, the Wilcoxon rank-sum test tests whether p values for genes in that pathway are smaller than genes not in that pathway. Enrichment analyses of genes that are significantly up or significantly down in both mac1⌬ and hst1⌬ mutants relative to WT were obtained by KEGG and GO enrichment analyses of gene lists from Venn diagrams. Fisher's exact test was used to test whether the number of genes in the Venn list in a given pathway/GO term was greater than would be expected by chance.

Quantitative PCR (qPCR) analysis of gene expression levels
Approximately 40 A 600 unit cells grown to early logarithmic phase in SD (6 h of growth from A 600 of 0.1) were collected by centrifugation. Total RNA was isolated using the GeneJET RNA purification Kit (Thermo Scientific), and cDNA was synthesized using the QuantiTect reverse transcription kit (Qiagen) according to the manufacturer's instructions. For each qPCR, 50 ng of cDNA and 500 nM each primer were used. The qPCR was run on a Roche LightCycler 480 using LightCycler 480 SYBR Green I Master Mix (Roche Applied Science) as described previously (64). The average size of the amplicon for each gene was ϳ150 bp. The target mRNA transcript levels were normalized to TAF10 transcript levels.

rAPase activity assay
The rAPase liquid assay was carried out as described (35) using cells grown to late log phase in SC. In brief, 2.5 A 600 unit cells were harvested, washed, and resuspended in 150 l of water. Next, 600 l of substrate solution (5.6 mg/ml p-nitro-phenylphosphate in 0.1 M sodium acetate, pH 4) was added to cell suspension, and the mixture was incubated at 30°C for 15 min. The reaction was stopped by adding 600 l of ice-cold 10% TCA. 600 l of this final mixture was then added to 600 l of saturated Na 2 CO 3 to allow color (neon yellow) development. Cells were pelleted to acquire the supernatant for A 420 determination. The rAPase activities were determined by normalizing A 420 readings to total cell number (A 600 ).

Protein extraction and Western blot analysis
A total of 100 ml of cells were grown in SD to early logarithmic phase (A 600 of ϳ1) and collected by centrifugation. The cell lysate was obtained by bead-beating in lysis buffer (50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 1% Nonidet P-40 (Sigma), and protease inhibitor mixture (Pierce TM ). The protein concentration was measured using the Bradford assay (Bio-Rad), and 30 g of total protein was loaded in each lane. After electrophoresis, the protein was transferred to a nitrocellulose membrane (Amersham Biosciences TM Protran TM , GE Healthcare). The membranes were then washed and blotted with either anti-HA antibody (Invitrogen) or anti-␤-actin antibody (Abcam). Protein was visualized using anti-mouse IgG antibody conjugate to the horseradish peroxidase (GE Healthcare) and the ECL reagents (Amersham Biosciences TM , GE Healthcare). The chemiluminescent image was analyzed using the Amersham Biosciences Imager 600 (GE Healthcare) system and software provided by the manufacturer.

ChIP assay
Approximately 500 A 600 unit cells grown to early logarithmic phase in SD were cross-linked with 1% formaldehyde for 30 min at room temperature and stopped by adding glycine to a final concentration of 125 mM. Cells were pelleted by centrifugation and washed two times with cold TBS (20 mM Tris-HCl, pH 7.5, 150 mM NaCl). Cells were lysed by bead beating in 1 ml of FA-140 lysis buffer (50 mM HEPES, 140 mM NaCl, 1% Triton X-100, 1 mM EDTA, 0.1% sodium deoxycholate, 0.1 mM PMSF, 1ϫ protease inhibitor mixture (Pierce)) (33). The cell lysate was drawn off the beads and centrifuged at a maximum speed (13,200 rpm) for 30 min at 4°C. The chromatin pellet was resuspended in 1 ml of FA-140 lysis buffer and sonicated on ice eight times with 20-s pulses using a Branson 450 Sonicator (output control set at 1.5 and duty cycle held at constant) to shear chromatin to an average length of ϳ500 bp. Sonicated chromatin solution was centrifuged twice at 10,000 rpm for 10 min at 4°C. The supernatant was then aliquoted into three tubes (labeled "input," "IP," and "no-Ab"). The IP samples were incubated overnight at 4°C with anti-HA mAb (ab1424, Abcam) at a dilution of 1:150. Both IP and no-Ab samples were incubated with 60 l of ChIP-grade protein G beads (Cell Signaling Technology) for 2 h at 4°C and then washed as described (33). DNA was then eluted from the beads two times with 125 l of elution buffer (50 mM Tris-HCl, pH 8, 10 mM EDTA, 1% SDS). The combined DNA solution and input samples were incubated at 65°C overnight to reverse the cross-linking. The purified DNA samples were analyzed by qPCR, and results were compared with a standard curve prepared from input DNA. The amount of immunoprecipitated specific promoter DNA Mac1 is a NAD ؉ homeostasis factor