Cellular redox state constrains serine synthesis and nucleotide production to impact cell proliferation

Abstract

The de novo serine synthesis pathway is upregulated in many cancers. However, even cancer cells with increased serine synthesis take up large amounts of serine from the environment1 and we confirm that exogenous serine is needed for maximal proliferation of these cells. Here we show that even when enzymes in the serine synthesis pathway are genetically upregulated, the demand for oxidized NAD+ constrains serine synthesis, rendering serine-deprived cells sensitive to conditions that decrease the cellular NAD+/NADH ratio. Further, purine depletion is a major consequence of reduced intracellular serine availability, particularly when NAD+ regeneration is impaired. Thus, cells rely on exogenous serine consumption to maintain purine biosynthesis. In support of this explanation, providing exogenous purine nucleobases, or increasing NAD+ availability to facilitate de novo serine and purine synthesis, both rescue maximal proliferation even in the absence of extracellular serine. Together, these data indicate that NAD+ is an endogenous limitation for cancer cells to synthesize the serine needed for purine production to support rapid proliferation.

Many cancers exhibit altered serine metabolism^2,3,4, including upregulated expression of de novo serine synthesis pathway enzymes⁵. The gene encoding the first enzyme in the serine synthesis pathway, phosphoglycerate dehydrogenase (PHGDH), exhibits copy number gain in some cancers^5,6, most frequently in breast cancer and melanoma^5,7. Serine synthesis enzymes can also be transcriptionally upregulated by the transcription factor ATF4, and many non-small cell lung cancers exhibit ATF4-mediated PHGDH upregulation downstream of deregulated NRF2⁸. There is also evidence that environmental serine availability can be critical for proliferation of some tumor cells^6,7,9,10,. In fact, environmental serine limitation can slow the growth of some tumors⁹, and even cells with upregulated serine synthesis appear to rely on extracellular serine for optimal proliferation¹⁰.

Serine is a precursor for many classes of biomolecules, and a major donor of one-carbon (1C) units, which can support methylation reactions and nucleotide synthesis. One 1C unit is needed to produce deoxythymidine monophosphate (dTMP), and two 1C units are needed to produce inosine monophosphate (IMP), the precursor for both AMP and GMP (Supplementary Fig. 1a). Indeed, purine levels are decreased in cells cultured without exogenous serine¹⁰, and blocking de novo serine synthesis can deplete nucleotides in some cells¹¹.

Why cells differentially increase serine synthesis and whether the source of intracellular serine matters for cell proliferation remains unclear. Examining the metabolic constraints of serine synthesis could provide insight. A difference between serine consumption and synthesis is demand for redox cofactors: while extracellular serine uptake does not use NAD+, serine synthesis from glucose requires oxidation steps that consume NAD+ and produce NADH (Supplementary Fig. 1a).

Regeneration of NAD+ to support oxidation reactions and build biomass can be a metabolic constraint for cancer cell proliferation and tumor growth in some contexts^12,13,14. This suggests that NAD+ availability could constrain flux through the serine synthesis pathway and necessitate exogenous serine consumption. In support of this notion, withdrawing extracellular serine can increase cancer cell sensitivity to the biguanides metformin and phenformin¹⁵, both of which inhibit NAD+ regeneration via Complex I of the mitochondrial electron transport chain (ETC)¹⁴. Furthermore, inhibiting NAD(H) production via the salvage pathway inhibits serine synthesis in breast cancer cells with PHGDH copy number gain¹⁶. NAD+ is also required to produce the 1C units needed in purine and dTMP synthesis (Supplementary Fig. 1a), raising the possibility that NAD+ insufficiency could exacerbate the effect of serine deprivation on nucleotide synthesis. Indeed, 1C unit production is impaired in cells with defective mitochondrial respiration¹⁷, a condition that causes decreased NAD+ availability. The cellular NAD+/NADH ratio can therefore dictate the metabolic processes available to cells¹⁸, and could influence dependence on extracellular serine consumption.

We evaluated the importance of extracellular serine consumption in cells in which serine synthesis is upregulated as a result of PHGDH gene copy number gain (MDA-MB-468)⁵ or increased via transcriptional regulation (A549)⁸. We also examined cells with low serine synthesis (MDA-MB-231)^5,11 in which PHGDH was or was not exogenously expressed to increase serine synthesis (MDA-MB-231 pLHCX PHGDH and MDA-MB-231 pLHCX EV, respectively)¹¹ (Fig. 1a). In each of these cell lines, as well as in other cells with high PHGDH expression, removing serine from the media substantially decreased proliferation (Fig. 1b and Supplementary Fig. 1b,c). PHGDH-expressing cells cultured in serine-free media maintain robust serine synthesis but cannot sustain the same levels of intracellular serine when extracellular serine is not available (Supplementary Fig. 1d, e). Together, these data indicate that even high expression of serine synthesis enzymes cannot sustain maximal proliferation.

Figure. 1: Withdrawal of exogenous serine limits proliferation and is exacerbated by inhibition of mitochondrial respiration.

a, Western blot analysis to assess expression of serine synthesis enzymes in A549 and MDA-MB-468 cells, and in MDA-MB-231 cells with (pLHCX PHGDH) and without (pLHCX EV) PHGDH expression as indicated. These experiments were repeated 3 times with similar results. b, Proliferation rates of the indicated cancer cells cultured in media with serine (+ Serine), or without serine (- Serine). c, Proliferation rates of A549 and MDA-MB-468 cells, and of MDA-MB-231 cells with (pLHCX PHGDH) and without (pLHCX EV) PHGDH expression at the indicated concentration of rotenone, in the presence or absence of exogenous serine as indicated. Data shown are mean (+/− standard deviation) of 3 biological replicates.

Serine deprivation has been shown to enhance the antiproliferative effects of the biguanides phenformin and metformin¹⁵, both of which inhibit Complex I of the mitochondrial ETC¹⁴. To examine whether increased sensitivity to serine withdrawal is generally observed with Complex I inhibitors, and whether this is affected by upregulated serine synthesis, we exposed cells to increasing concentrations of rotenone and metformin in the presence or absence of serine. Withdrawal of serine from the media increased sensitivity of cells to Complex I inhibitors (Fig. 1c and Supplementary Fig. 1f). We confirmed that expression of serine synthesis enzymes was not affected by mitochondrial inhibition (Supplementary Fig. 2a). These data suggest that extracellular serine can be limiting for cell proliferation even in cells with increased serine synthesis, and that this limitation can be exacerbated by inhibiting mitochondrial Complex I.

Production of nucleotides is an important fate of serine carbon to support proliferation^10,11,19. To assess whether cells prefer to use either extracellular serine or newly synthesized serine to produce nucleotides, we traced the fate of 6-¹³C-glucose or 3-¹³C-serine. For cells cultured in 6-¹³C-glucose, newly synthesized serine has a mass value of M+1, and generates an M+1 1C unit. 6-¹³C-glucose can also label the ribose backbone of nucleotides. For cells cultured in 3-¹³C-serine, consumed serine has a mass of M+1, and generates an M+1 1C unit (Supplementary Fig. 1g). 1C units contribute to pyrimidine synthesis only when adding a methyl group to make dTMP. Thus, when cultured in the presence of 6-¹³C-glucose, whether dTMP is M+1 labeled relative to dCMP shows whether 1C units are derived from serine that was synthesized from glucose, while dTMP labeling from 3-¹³C-serine reflects dTMP production from exogenous serine. Consistent with only a small fraction of the serine in MDA-MB-468 cells being derived from glucose when extracellular serine is available (Supplementary Fig. 1e), the increased amount of dTMP labeled from 6-¹³C-glucose compared to dCMP labeled from 6-¹³C-glucose (resulting from 6-¹³C-glucose-derived 1C units) is small relative to the amount of dTMP labeled from extracellular serine (Supplementary Fig. 1g). This argues that cells do not preferentially use serine derived from one source for 1C unit production. It also raises the question of why extracellular serine is important for proliferation even when cells have genetically upregulated synthesis and unlimited access to glucose. Because glucose is abundant in cell culture media and these cells excrete most glucose carbon as lactate¹, serine synthesis is unlikely to be limited by carbon substrate availability.

De novo serine synthesis and serine uptake have different redox cofactor requirements. Unlike the consumption of exogenous serine, serine production from glucose involves two oxidation reactions that consume NAD+ and produce NADH, and thus requires transfer of electrons to a waste product for every serine molecule produced (Fig. 2a). Thus, flux through serine synthesis should be sensitive to the cellular NAD+/NADH ratio. Of note, mitochondrial respiration allows transfer of electrons to oxygen and regenerates NAD+ from NADH. Thus, decreased NAD+/NADH ratio could explain why cells without access to exogenous serine are more sensitive to Complex I inhibition. To test this possibility, we compared serine production from U-¹³C-glucose in cells with and without Complex I inhibition (Supplementary Fig. 1d). While untreated serine-deprived cells displayed robust serine labeling from glucose, rotenone treatment completely eliminated U-¹³C-glucose incorporation into serine (Fig. 2b). As expected, rotenone treatment decreased the cellular NAD+/NADH ratio (Supplementary Fig. 4a). We next supplemented rotenone-treated cells with the exogenous electron acceptor α-ketobutyrate (AKB), which can be reduced to α-hydroxybutyrate (AHB) to regenerate NAD+ (Supplementary Fig. 2b)¹³. Indeed, providing AKB restored serine synthesis from glucose in rotenone-treated cells (Fig. 2b). We confirmed that cells took up AKB and used it to produce AHB (Supplementary Fig. 2c), and that AKB supplementation restored the NAD+/NADH ratio (Supplementary Fig. 4a), but did not cause cells to further upregulate expression of serine synthesis enzymes (Supplementary Fig. 2a). Addition of AKB also rescued proliferation of rotenone-treated and metformin-treated cells cultured in the absence of exogenous serine (Fig. 2c, Supplementary Fig. 2d,e, and Supplementary Fig. 5b). Together, these data indicate that NAD+ availability can limit serine synthesis and cell proliferation, particularly when the ability to transfer electrons to oxygen is impaired.

Figure. 2: Electron acceptor availability limits serine synthesis and cell proliferation.

a, Schematic showing the relationship between serine synthesis, the one-carbon (1C) cycle, and nucleotide synthesis (dTMP and purines), highlighting the NAD+-consuming steps involved in serine synthesis and 1C unit generation. Also shown is how NAD+ can be regenerated from NADH via mitochondrial respiration, or by the reduction of α-ketobutyrate (AKB) to α-hydroxybutyrate (AHB). Ser, serine; Gly, glycine; 3PG, 3-phosphoglycerate; THF, tetrahydrofolate. b, Fractional labeling of serine from U-¹³C-glucose in A549 cells cultured for 48 hours in serine-free media with or without 20 nM rotenone, and with or without AKB as indicated. Labeling was determined by LCMS analysis. c, Proliferation rates of the specified cells cultured in media containing 20nM rotenone, with or without serine and AKB, as indicated. Data shown are mean (+/− standard deviation) of 3 biological replicates.

Producing one-carbon (1C) units for nucleotide synthesis relies on functional mitochondrial respiration because mitochondrial NAD+ is needed to generate 1C units¹⁷ (Fig. 2a), as is the case in most cells²⁰. Since serine is a major 1C unit donor, impaired serine synthesis following Complex I inhibition could exacerbate a decrease in 1C unit production. To test this possibility, we assessed how metabolite levels change in cells cultured in serine-replete compared to serine-free media, with or without rotenone treatment. Indeed, purine nucleotides were among the most depleted metabolites in serine-deprived cells, and rotenone treatment further exacerbated purine nucleotide depletion (Fig. 3a). To directly test whether de novo purine synthesis is impaired in serine-deprived cells, we measured the incorporation of ¹⁵N from amide-¹⁵N-glutamine into purines. Because the amide nitrogen of glutamine is incorporated in two steps of AMP synthesis and three steps of GMP synthesis, any newly synthesized adenylate or guanylate nucleotides have mass values of M+2 and M+3, respectively (Fig. 3b). Cells cultured in serine-replete media used de novo synthesis to turn over their purine pools by 48 hours (Fig. 3c and Supplementary Fig. 3a). In contrast, cells cultured in serine-free media displayed decreased de novo synthesis, leading to purine depletion over time. Consistent with previous reports, the purine synthesis intermediates GAR and AICAR also accumulated in serine-deprived cells¹⁰ (Fig. 3d). As these intermediates each precede 1C-requiring reactions in purine synthesis, this accumulation is consistent with insufficient 1C units limiting purine synthesis. Restoring 1C units by supplementing cells with formate alleviated this accumulation (Fig. 3d), as noted previously^10,17.

Figure. 3: Purine nucleotide production downstream of serine metabolism constrains cell proliferation when NAD+ regeneration is impaired.

a, Hierarchical clustering and heat map indicating relative metabolite levels measured using LCMS from A549 cells cultured for 48 hours in media with or without serine and with or without 20 nM rotenone as indicated. b, Schematic showing how ¹⁵N label is incorporated from amide-¹⁵N-glutamine into the purines AMP and GMP. Two ¹⁵N labels are incorporated into newly synthesized AMP, while three ¹⁵N labels are incorporated into newly synthesized GMP. c, Total levels, and labeling of ATP and GTP from amide-¹⁵N-glutamine as determined by LCMS in A549 cells cultured in media with serine, or in media lacking serine for the indicated times. d, Levels of the purine synthesis intermediates GAR and AICAR measured using LCMS in A549 cells cultured with or without serine and with or without formate as indicated. GAR and AICAR each precede steps in purine synthesis that involve addition of a 1C unit derived from formate. GAR, glycineamide ribonucleotide; AICAR, 5-aminoimidazole-4-carboxamide ribonucleotide. e, Total levels and labeling of ATP and GTP from 5-¹³C-hypoxanthine in A549 cells cultured with or without serine, 20 nM rotenone, and 5-¹³C-hypoxanthine as indicated. Metabolites were analyzed by LCMS. f, Proliferation rates of A549 cells cultured with or without serine, formate, and/or hypoxanthine, and treated with DMSO or 20 nM rotenone as indicated. Data shown are mean (+/− standard deviation) of 3 biological replicates.

Salvage of purine nucleobases avoids the demand for serine-derived carbon to synthesize purines de novo. When 5-¹³C-hypoxanthine was provided, this was the primary carbon contributor to both adenylate and guanylate nucleotides regardless of serine availability (Fig. 3e and Supplementary Fig. 3b). Importantly, salvage of hypoxanthine rescued purine levels in serine-starved cells. Moreover, providing either formate to restore 1C unit availability or hypoxanthine to directly restore purine levels partially rescued cell proliferation in serine-free media (Fig. 3f). Formate or hypoxanthine supplementation also improved the proliferation of serine-deprived cells following rotenone or metformin treatment (Fig. 3f, Supplementary Fig. 3c,d, and Supplementary Fig. 5b). Importantly, formate or hypoxanthine supplementation did not restore the NAD+/NADH ratio (Supplementary Fig. 4a), suggesting that maintaining purine synthesis is a metabolic requirement downstream of NAD+ availability, and a reason why NAD+ is limiting for proliferation.

Serine synthesis requires NAD+ in the cytosol, so coupling this pathway to mitochondrial electron transport requires redox shuttles to transport electrons across the mitochondrial membranes. The malate-aspartate shuttle is one contributor to redox shuttling between the mitochondria and the cytosol, and disrupting this shuttle by knockdown of aspartate-glutamate carrier 1 (AGC1) expression²¹ did not affect proliferation of cells grown without exogenous serine, but did increase sensitivity of serine-deprived cells to Complex I inhibition (Supplementary Fig. 4b). These data are consistent with the malate-aspartate shuttle being important to maintain cytosolic NAD+ when mitochondrial NAD+ regeneration is impaired.

NAD+ availability can also be affected by nicotinamide salvage. Indeed, inhibiting nicotinamide salvage can inhibit serine synthesis in PHGDH-amplified breast cancer cells¹⁶. We therefore investigated whether nicotinamide salvage can increase oxidized NAD+ availability for serine synthesis. Addition of nicotinamide mononucleotide (NMN) did not restore proliferation of cells cultured without exogenous serine, and did not change the cellular NAD+/NADH ratio (Supplementary Fig. 4c). This may indicate that the NAD+/NADH ratio is a stronger constraint on serine synthesis than is the total amount of NAD(H) in cells.

We next questioned whether basal NAD+ availability limits cell proliferation in the absence of exogenous serine. Indeed, increasing NAD+ availability by providing AKB increased proliferation of cells cultured in serine-free conditions (Fig. 4a and Supplementary Fig. 5a,b). To evaluate whether NAD+ availability constrains purine synthesis in serine-deprived cells, we measured U-¹³C-glucose incorporation into purines in cells grown in serine-free or serine-replete media with or without rotenone. In these experiments, newly synthesized nucleotides are labeled with a mass value of M+5 because the ribose sugar contains 5 ¹³C carbons from U-¹³C-glucose. Moreover, this tracing can distinguish the source of serine used for purine production. When produced from glucose-derived serine, glycine is M+2 labeled, and each 1C unit is M+1 labeled. Thus, purines with mass values of M+6 or greater were produced using glucose-derived serine (Fig. 4b)¹¹.

Figure. 4: Electron acceptors are an endogenous limitation for serine synthesis and proliferation in the absence of exogenous serine.

a, Proliferation rates of A549 and PHGDH-expressing MDA-MB-231 cells (pLHCX PHGDH) cultured in serine-replete or serine-free media with or without AKB as indicated. b, Schematic of how U-¹³C-glucose carbon can contribute label to serine and purines. Newly synthesized purines incorporate ribose containing 5 ¹³C from labeled glucose. If serine that is synthesized from ¹³C-labeled glucose contributes to purine synthesis, up to 4 additional ¹³C-labeled carbons can be incorporated, because glycine produced from glucose-derived serine can contribute 2 ¹³C-labeled carbons, and 10-formyl-THF from glucose-derived serine can contribute up to 2 ¹³C labels. Thus, M+5 labeling suggests the purine is newly synthesized using exogenous serine, while species with >5 ¹³C labels suggests the purine is synthesized using glucose-derived serine. c, Total levels and labeling of ATP and GTP from U-¹³C-glucose in A549 cells grown with or without serine, 20 nM rotenone, and AKB as indicated. Data shown are mean (+/− standard deviation) of 3 biological replicates.

By 48 hours of labeling, cells cultured in the presence of serine had purines that were mostly M+5 labeled (Fig. 4c and Supplementary Fig. 6a–d), suggesting that most of the serine used for purine synthesis in these cells is from the environment. Again, cells grown in the absence of serine had decreased purine pools. Furthermore, because serine-deprived cells must use glucose-derived serine for purine production, their purines had mass values of M+6 or greater. Consistent with NAD+ being limiting for serine synthesis and the production of 1C units for purines, rotenone treatment further decreased purine levels and decreased the levels of purines with mass values of M+6 or greater. For both untreated and rotenone-treated cells grown in serine-free media, addition of AKB increased intracellular purine levels, and increased contribution of glucose-derived serine to purines, resulting in increased labeling of purines with mass values of M+6 or greater. Notably, expression of serine synthesis enzymes was necessary for this rescue, as MDA-MB-231 cells required PHGDH expression to produce purines using synthesized serine (Supplementary Fig. 6b). These data confirm that increasing NAD+ availability can restore purine production using serine synthesized from glucose, and suggest that demand for NAD+ constrains the use of serine synthesis to support purine production and allow maximal cell proliferation.

The metabolic network available to cancer cells is defined by the enzymes and regulatory factors expressed, and thus is influenced by both the cell type which gave rise to the cancer, and the genetic and signaling changes which drive the cancer^22,23. In addition, cancer cells in different tissues are exposed to different nutrient environments²⁴. This suggests that different cancers experience different metabolic constraints, and may underlie why cancers show differential sensitivity to drugs targeting metabolic processes predicted to be important in all proliferating cells^14,25.

We find that the extent to which cells can utilize the serine synthesis pathway is dictated by NAD+ availability. Mitochondrial respiration allows net regeneration of NAD+ from NADH, and this ability to regenerate NAD+ is an important way that respiration supports cell proliferation^12,13,14. Previous work has focused on aspartate production as a precursor for proteins and nucleotides that could be limited by NAD+ regeneration^12,13,26. The finding that serine metabolism for purine synthesis can also be limited by NAD+ regeneration demonstrates that aspartate production is not the only important use of NAD+ to support proliferation. Many NAD+-consuming reactions compete for NAD+, and whether production of aspartate, the generation of serine-derived 1C-units, or other processes requiring NAD+ as a cofactor are more limiting will depend on the cancer context. For example, if environmental serine is limited, then NAD+ consumption in serine synthesis may become required and select for adaptations to circumvent other NAD+-utilizing pathways.

The demand for extracellular serine is consistent with studies showing that dietary serine limitation can slow the growth of some, but not all, tumors. Indeed, dietary serine limitation appears to be more affective for slowing the growth of tumors arising in tissues with lower serine content^6,9. Low-serine tissue environments appear to select for tumors that genetically upregulate serine synthesis⁶. While this increases intracellular serine and supports faster tumor growth⁶, the requirement for serine synthesis in these tumors could render them vulnerable to NAD+ depletion. Further, high rates of serine synthesis in these tumors might make NAD+ less available for other pathways needed to synthesize oxidized biomass, resulting in other metabolic limitations. Similarly, our data suggest that tumors growing in environments where NAD+ availability is more limited may be more susceptible to serine deprivation. In line with this idea, a serine-deficient diet enhances the efficacy of metformin treatment in an allograft cancer model¹⁵. Recognizing that forcing increased serine production can synergize with drugs such as metformin that inhibit NAD+ regeneration, not because they limit the carbon available for ATP production¹⁵ but rather because they cause a redox imbalance, suggests that other NAD+-depleting drugs could also enhance response to dietary serine limitation. By extension, a better understanding of how differential pathway use by cancer cells creates metabolic liabilities could suggest additional opportunities to target metabolism for therapeutic benefit.

Methods

Cell Culture

All cell lines were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM) (GIBCO) supplemented with 10% heat inactivated fetal bovine serum at 37°C with 5% CO₂. Pyruvate or nucleotides were not included in the culture media except when indicated. Concentrations of metabolites added to treatment media are listed in Supplementary Table 1.

Proliferation Rates

Cells were trypsinized, counted and plated in 6-well plates (Denville) in 2 mL DMEM with 10% FBS and incubated overnight. Cells were plated at 20,000 cells per well (A549, 143B, H1299) or 40,000 (MDA-MB-468, MDA-MB-231, T.T, Carney, BT-20, H522, HS578T) cells per well in 6 well plates. The following day, one 6-well plate of each cell line was counted to determine the initial number of cells at the time of treatment. Cells were washed three times with 2 mL PBS and 4 mL of treatment media was added. Treatment media was made with 10% dialyzed FBS. Serine-free media was made by adding a mix of amino acids at DMEM concentrations lacking serine to DMEM without pyruvate or amino acids. The number of cells seeded for each cell line allowed for exponential growth over the course of the assay. After 4 days of treatment, final cell counts were measured using a Cellometer. The following formula was used to calculate proliferation rate:

$$\text{Doublings\hspace{0.17em}per\hspace{0.17em}day}=\text{log}2\left(\text{Final\hspace{0.17em}cell\hspace{0.17em}count\hspace{0.17em}}\left(\text{day\hspace{0.17em}}5\right)/\text{Initial\hspace{0.17em}cell\hspace{0.17em}count\hspace{0.17em}}\left(\text{day\hspace{0.17em}}1\right)\right)/4\left(\text{days}\right)$$

Western Blots

10⁶ cells were plated in 10 cm plates and incubated overnight to allow cells to adhere. The following day, cells were washed three times with PBS and 10 mL of treatment media containing 10% dialyzed FBS was added. After culturing cells for 24 hours in treatment media, protein lysates were prepared: cells were trypsinized, pelleted, and washed twice with PBS. Cells were then resuspended in ice-cold RIPA buffer with protease inhibitor. Lysed cells were centrifuged at maximum speed at 4°C and the protein lysate supernatant was removed and stored at −80°C. Proteins were separated using SDS-PAGE (12% acrylamide gels) and transferred to a nitrocellulose membrane using a wet transfer method. Membranes were blocked for 60 minutes using 5% bovine serum albumin (BSA) in Tris buffered saline with Tween (TBST). Membranes were incubated in primary antibody overnight at 4°C. The following primary antibodies were used: vinculin (Cell Signaling Technology), PHGDH (Sigma), PSAT (Abnova), and PSPH (Sigma Prestige). Antibodies were diluted in 5% BSA in TBST (vinculin and PSAT, 1:1000; PHGDH, 1:500; PSPH, 1:500). The following day, membranes were washed three times with TBST on a rocker for 10 minutes. Secondary antibodies were applied for 60 minutes. Anti-rabbit (Cell Signaling Technology) secondary antibody was used at a dilution of 1:5000 and anti-mouse (Cell Signaling Technology) secondary antibody was used at a dilution of 1:10000. Membranes were then washed again three times with TBST for 10 minutes on a rocker. Signal was detected using film. Uncropped Western blots are shown in Supplementary Figure 7.

LCMS Analysis

100,000 cells were plated in 6-well plates in 2 mL DMEM with 10% FBS and incubated overnight. The following day, cells were washed three times with PBS and 4 mL of treatment media was added. All treatment media was made with 10% dialyzed FBS. After the indicated time period, polar metabolites were extracted from cells: plates were placed on ice, cells were washed with ice-cold blood bank saline, and 500 μl of ice-cold 80% methanol in water with 250nM ¹³C/¹⁵N labeled amino acid standards (Cambridge Isotope Laboratories, Inc.) was added to each well. Cells were scraped, each sample was vortexed for 10 minutes at 4°C, and then centrifuged at maximum speed for 10 minutes at 4°C. Samples were dried under nitrogen gas and resuspended in 25 μl of a 50/50 acetonitrile/water mixture. Metabolites were measured using a Dionex UltiMate 3000 ultra-high performance liquid chromatography system connected to a Q Exactive benchtop Orbitrap mass spectrometer, equipped with an Ion Max source and a HESI II probe (Thermo Fisher Scientific). Samples were separated by chromatography by injecting 2–10 μl of sample on a SeQuant ZIC-pHILIC Polymeric column (2.1 × 150 mm 5 μM, EMD Millipore). Flow rate was set to 150 μl/min, temperatures were set to 25 °C for column compartment and 4 °C for autosampler sample tray. Mobile Phase A consisted of 20 mM ammonium carbonate, 0.1% ammonium hydroxide. Mobile Phase B was 100% acetonitrile. The mobile phase gradient (%B) was set in the following protocol: 0–20 min.: linear gradient from 80% to 20% B; 20–20.5 min.: linear gradient from 20% to 80% B; 20.5–28 min.: hold at 80% B. Mobile phase was introduced into the ionization source set to the following parameters: sheath gas = 40, auxiliary gas = 15, sweep gas = 1, spray voltage = −3.1kV, capillary temperature = 275 °C, S- lens RF level = 40, probe temperature = 350 °C. Metabolites were monitored in full-scan, polarity-switching, mode. An additional narrow range full-scan (220–700 m/z) in negative mode only was included to enhance nucleotide detection. The resolution was set at 70,000, the AGC target at 1,000,000, and the maximum injection time at 20 msec. Relative quantitation of metabolites was performed with XCalibur QuanBrowser 2.2 (Thermo Fisher Scientific) using a 5 ppm mass tolerance and referencing an in-house retention time library of chemical standards. Metabolite measurements were normalized to the internal ¹³C/¹⁵N labeled amino acid standard and to cell number.

Metabolomic Data Analysis

100,000 cells were plated in 6-well dishes and allowed to adhere overnight. The following day, cells were washed three times with PBS and 4mL of the specified media was added. After 48 hours, polar metabolites were extracted and analyzed using a Dionex UltiMate 3000 ultra-high performance liquid chromatography system connected to a Q Exactive benchtop Orbitrap mass spectrometer, as described in “LCMS Analysis.” Metabolite peak integration was performed as above using XCalibur QuanBrowser 2.2, and peak areas were normalized to an internal standard (¹³C/¹⁵N labeled amino acid standard) and to cell number. Differential and statistical analysis was performed using MetaboAnalystR. The following parameters were used: data normalization: normalization to sample median; data transformation: log normalization; data scaling: autoscaling. Hierarchical clustering in MetaboAnalystR was performed using the following parameters: similarity measure: Euclidean distance; clustering algorithm: Ward’s linkage.

NAD+/NADH Measurements

NAD+/NADH measurements were performed using the NAD/NADH Glo Assay (Promega) with a modified version of manufacturer instructions as reported previously¹³. Cells were plated as done for proliferation assays and incubated overnight. Cells were treated as indicated for 6 hours prior to preparation of cell extracts. For extraction, cells were washed 3 times in ice cold PBS, extracted in 100 μL ice cold lysis buffer (1% Dodecyltrimethylammonium bromide (DTAB) in 0.2 N NaOH diluted 1:1 with PBS). Each sample was flash-frozen in liquid nitrogen and immediately stored at −80°C. To measure NADH, 20 μL of sample was moved to PCR tubes and incubated at 75°C for 30 min where basic conditions selectively degrade NAD+. To measure NAD+, 20 μL of the samples was moved to PCR tubes containing 20 μL lysis buffer and 20 μL 0.4 N HCl and incubated at 60°C for 15 min, where acidic conditions selectively degrade NADH. Samples were then allowed to equilibrate to room temperature and quenched by neutralizing with 20 μL 0.25 M Tris in 0.2 N HCl (for NADH) or 20 μL 0.5 M Tris base (for NAD+). Manufacturer instructions were followed thereafter to measure NAD+/NADH.

Data Availability

The raw data that support the findings of this study are available from the corresponding author upon request. Raw data for Figure 1a, Supplementary Figure 1c, and Supplementary Figure 2a (uncropped Western blots) are shown in Supplementary Figure 7.

Reporting Summary

Further information about research design is available in the Nature Research Reporting Summary linked to this article.