Basic leucine zipper transcription factor, ATF-like (BATF) regulates epigenetically and energetically effector CD8 T-cell differentiation via Sirt1 expression

Abstract

CD8 T cells play a critical role in protection against viral infections. During effector differentiation, CD8 T cells dramatically change chromatin structure and cellular metabolism, but how energy production increases in response to these epigenetic changes is unknown. We found that loss of basic leucine zipper transcription factor, ATF-like (BATF) inhibited effector CD8 T-cell differentiation. At the late effector stage, BATF was induced by IL-12 and required for IL-12–mediated histone acetylation and survival of effector T cells. BATF, together with c-Jun, transcriptionally inhibited expression of the nicotinamide adenine dinucleotide (NAD⁺)-dependent deacetylase Sirt1, resulting in increased histone acetylation of the T-bet locus and increased cellular NAD⁺, which increased ATP production. In turn, high levels of T-bet expression and ATP production promoted effector differentiation and cell survival. These results suggest that BATF promotes effector CD8 T-cell differentiation by regulating both epigenetic remodeling and energy metabolism through Sirt1 expression.

CD8 T cells are one of the most important components of protective immunity against viral infections, and understanding their development is necessary for generating effective vaccines (1–3).

During effector differentiation, CD8 T cells dramatically change gene expression and cellular metabolism (4). In naïve T cells, most energy is generated in mitochondria through oxidative phosphorylation. Upon stimulation, T cells massively increase glucose uptake and produce ATP by glycolysis to support rapid proliferation. At the late effector stage, inflammatory cytokines promote effector CD8 T-cell differentiation through chromatin remodeling (5, 6), which increases ATP consumption to sustain gene expression and subsequent protein synthesis. However, how effector T cells increase energy production in response to these epigenetic changes is unknown.

Basic leucine zipper transcription factor, ATF-like (BATF), a member of the AP-1 family, forms heterodimers with Jun and blocks AP-1 transactivation (7–9). Unlike other AP-1 family members, BATF expression is restricted to lymphoid organs, and is induced in T cells and natural killer T cells upon stimulation. Although BATF-deficient mice show impaired Th17 and follicular Th differentiation and antibody (Ab) production (10, 11), the function of BATF in CD8 T-cell differentiation is unknown.

In this study, we examined the role of BATF in effector CD8 T-cell development by generating knock-in mice. We show that BATF is induced by IL-12 and required for IL-12–mediated histone acetylation and survival of effector CD8 T cells. We demonstrate that BATF with c-Jun transcriptionally inhibits expression of the NAD⁺-dependent deacetylase Sirt1. Furthermore, BATF promotes effector CD8 T-cell differentiation by regulating both epigenetic remodeling and energy metabolism through Sirt1 expression.

Results

BATF Deficiency Inhibits Effector CD8 T-Cell Differentiation.

To examine the role of BATF in CD8 T-cell responses in vivo, we generated a reporter mouse strain by inserting GFP cDNA into the Batf locus (Fig. S1 A and B). These Batf^gfp/gfp mice lacked BATF protein expression (Fig. S1C). Consistent with previous reports (10, 11), Batf^gfp/gfp mice demonstrated a slight increase in splenic B-cell numbers (Fig. S1 D and E).

To induce antigen-specific effector CD8 T cells in vivo, we targeted a model antigen, ovalbumin (OVA), to DEC-205⁺ dendritic cells (DCs) with CD40 ligation (12–14). By incorporating antigen within a monoclonal Ab to DEC-205, an endocytic receptor expressed on DCs, targeted antigen is taken up by DCs and presented to T cells (12). CD40 ligation induces DC maturation, and mature DCs provide “signal 2” (costimulatory signal) and “signal 3” (inflammatory cytokine signal) to T cells by expressing high levels of B7 molecules and IL-12 (5, 15–17). IL-12 promotes effector CD8 T-cell differentiation, probably through chromatin remodeling by histone acetylation (5). At day 3 after priming, similar numbers of OVA-specific tetramer⁺ CD8 T cells were detected in the spleens of Batf^+/+ and Batf^gfp/gfp mice (Fig. S2A). At day 5, however, OVA-specific tetramer⁺ CD8 T cells increased in Batf^+/+ mice but not in Batf^gfp/gfp mice (Fig. S2A). At day 7, the number of OVA-specific tetramer⁺ CD8 T cells was much lower in Batf^gfp/gfp mice than in Batf^+/+ mice (Fig. 1A and Fig. S2A). Similarly, the number of OVA-specific IFN-γ⁺ CD8 T cells was much lower in Batf^gfp/gfp mice than in Batf^+/+ mice (Fig. 1A). Moreover, the frequency of CD62L^low and CD127 (IL-7R)^low subsets in Batf^gfp/gfp tetramer⁺ CD8 T cells was lower than in Batf^+/+ tetramer⁺ CD8 T cells (Fig. S2B). These results suggest that BATF deficiency inhibits effector CD8 T-cell differentiation at the late effector phase.

Fig. 1.

BATF deficiency inhibits effector CD8 T-cell differentiation. (A) Batf^gfp/gfp and Batf^+/+ mice were primed with anti-DEC:OVA+anti-CD40 and 7 d later, OVA-specific CD8⁺ T cells were examined by K^b/OVA257–264 tetramer or peptide-stimulated intracellular IFN-γ staining. The percentage (Left) and total number (Right) of OVA-specific CD8 T cells per spleen are indicated. (B) mRNA levels of the indicated genes in CD8 T cells stimulated with anti-CD3 in combination with anti-CD28 and IL-12 for 72 h. Data are presented relative to expression levels in unstimulated cells. Data are representative of three independent experiments.

To demonstrate the CD8 T-cell–intrinsic function of BATF in vivo, we transferred Batf^+/+ and Batf^gfp/gfp CD8 T cells into Rag1^−/− mice and immunized them with anti-DEC:OVA+anti-CD40 (Fig. S2C). After 7 d of priming, the number of OVA-specific tetramer⁺ CD8 T cells was much lower in the mice transferred with Batf^gfp/gfp CD8 T cells than in the mice transferred with Batf^+/+ CD8 T cells. These results suggest that Batf^gfp/gfp CD8 T cells have an intrinsic defect in their ability to differentiate into effector CD8 T cells.

IL-12 with Costimulation Induces BATF at the Late Effector Stage.

Upon T-cell stimulation, BATF was slowly up-regulated and peaked at 72 h, whereas c-Fos and c-Jun were down-regulated (Fig. S3A). To examine the effect of signals 2 and 3 on BATF expression, CD8 T cells were stimulated with anti-CD3 in combination with anti-CD28 and IL-12. CD3/CD28 coligation induced BATF, but CD3 ligation alone did not (Fig.1B). IL-12 treatment alone also slightly increased BATF (Fig. S3B). However, IL-12 treatment together with CD3/CD28 ligation induced much higher BATF expression than CD3/CD28 ligation alone at 72 h (Fig. 1B), whereas induction was not observed at 48 h (Fig. S3C). These results suggest that IL-12 with costimulation induces BATF at the late effector stage. Unlike BATF, IL-12 treatment with CD3/CD28 ligation slightly increased c-Jun but not c-Fos (Fig. 1B).

BATF Is Required for IL-12-Induced Cell Survival and Histone Acetylation.

IL-12 is known to promote survival of effector CD8 T cells (5). To examine the effect of BATF on IL-12–induced cell survival, Batf^gfp/gfp and Batf^+/+ CD8 T cells were stimulated with anti-CD3/CD28 with or without IL-12. IL-12 had no effect on cell survival and proliferation at 48 h (Fig. 2A and Fig. S4 A and B). At 72 h, however, IL-12 treatment with CD3/CD28 ligation increased the viability of Batf^+/+ CD8 T cells, but not Batf^gfp/gfp CD8 T cells compared with CD3/CD28 ligation alone (Fig. 2A and Fig. S4B). These results suggest that BATF is required for IL-12–induced effector CD8 T-cell survival.

Fig. 2.

BATF is required for IL-12-mediated cell survival and histone acetylation. (A) The percentage of PI^low population in Batf^gfp/gfp and Batf^+/+ CD8 T cells stimulated with anti-CD3/CD28 with and without IL-12. (B) The percentage of acetylated H3^high population in Batf^gfp/gfp and Batf^+/+ CD8 T cells stimulated with anti-CD3/CD28 with and without IL-12. Data are representative of three independent experiments.

To examine the effect of BATF on IL-12–mediated histone acetylation, Batf^gfp/gfp and Batf^+/+ CD8 T cells were stimulated with anti-CD3/CD28 with or without IL-12, and the per-cell level of diacetylated histone H3, a marker of open chromatin, was examined by intracellular staining (18) (Fig. 2B and Fig. S4C). IL-12 had no effect at 48 h, but at 72 h, IL-12 treatment with CD3/CD28 ligation increased the percentage of acetylated H3^high population in Batf^+/+ CD8 T cells, but not in Batf^gfp/gfp CD8 T cells relative to CD3/CD28 ligation alone. These results suggest that BATF is required for IL-12–mediated histone acetylation.

BATF Promotes Histone Acetylation of the T-Bet Locus.

There is accumulating evidence that IL-12 promotes effector CD8 T-cell differentiation through Tbx21 (T-bet) expression (19, 20). We therefore examined the histone acetylation level of the Tbx21 locus using chromatin immunoprecipitation (ChIP) analysis (Fig. 3A). Batf^+/+ CD8 T cells stimulated with anti-CD3/CD28+IL-12 showed much higher histone H3 acetylation levels of the Tbx21 locus than Batf^gfp/gfp CD8 T cells. When stimulated with anti-CD3/CD28+IL-12, Batf^+/+ CD8 T cells showed higher mRNA expression of T-bet and its target genes, such as Perforin, IFN-γ, and IL-12Rβ2, than Batf^gfp/gfp CD8 T cells (Fig. 3B). These results suggest that BATF promotes histone acetylation of the Tbx21 locus, resulting in high expression of T-bet and its target genes. Consistent with the FACS data showing that BATF-deficiency reduces cell viability (Fig. 2A), Batf^gfp/gfp CD8 T cells showed lower Bcl2 expression than Batf^+/+ CD8 T cells (Fig. 3B).

Fig. 3.

BATF promotes histone acetylation of the T-bet locus. (A) Batf^gfp/gfp and Batf^+/+ CD8 T cells were stimulated with anti-CD3/CD28+IL-12 for 60 h and immunoprecipitated with antiacetyl H3 or control IgG. Purified ChIP and input DNA were analyzed by semiquantitative PCR (Left) and real-time PCR (Right). ChIP DNA level was normalized to the level of input DNA. (B) mRNA levels of the indicated genes in Batf^gfp/gfp and Batf^+/+ CD8 T cells stimulated with anti-CD3/CD28 alone or with IL-12 for 72 h. Data are presented relative to expression levels in unstimulated Batf^+/+ CD8 T cells. Data are representative of two independent experiments.

BATF Overexpression Increases ATP Production.

To examine the mechanism by which BATF promotes histone acetylation, a stable BATF-expressing murine T-cell line was established using EL4 cells (Fig. S5A), which show AP-1 family gene expression profiles similar to those observed in activated T cells (Fig. S5B). EL4/BATF cells showed higher levels of acetylated H3 than EL4 cells, suggesting that BATF promotes histone acetylation (Fig. 4A). Among the histone deacetylase family genes, Sirt1 was significantly down-regulated in EL4/BATF cells relative to EL4 cells (Fig. 4B). Sirt1 is a NAD⁺-dependent histone deacetylase (21, 22). NAD⁺ is the cosubstrate of Sirt1 and acts as a coenzyme in redox reactions where it continuously cycles between NAD⁺ and NADH forms to produce ATP without being consumed (23, 24). EL4/BATF cells showed increased levels of cellular NAD⁺, but not NADH, relative to EL4 cells (Fig. 4C). EL4/BATF cells showed higher ATP levels, higher glyceraldehydes-3-phosphate dehydrogenase (GAPDH) activity, and higher mitochondrial membrane potential (ΔΨm) than EL4 cells (Fig. 4 D–F). In addition, EL4/BATF cells showed higher glucose consumption and lactate production than EL4 cells (Fig. 4G). These results suggest that BATF increases cellular NAD⁺, which in turn increases ATP production by promoting glycolysis and mitochondrial activity. The mRNA levels of other NAD-consuming enzymes were also examined (Fig. S5C). Parp1 was slightly lower in EL4/BATF cells than in EL4 cells.

Fig. 4.

BATF overexpression increases ATP production. (A) Acetylated H3 levels in EL4 (filled curve) and EL4/BATF (open curve) cells. (B) Comparison of histone deacetylase family gene expression between EL4 and EL4/BATF cells. (C) Cellular NAD⁺ and NADH levels, (D) cellular ATP levels, and (E) GAPDH activity in EL4 and EL4/BATF cells. (F) ΔΨm of EL4 (filled curve) and EL4/BATF (open curve) cells. (G) EL4 and EL4/BATF cells were plated in 24-well plates at 1 × 10⁵ cells/mL. After 18 h of culture, glucose (Left) and lactate (Right) levels were measured in culture supernatant. Data are representative of three independent experiments. *P < 0.05; **P < 0.01; NS, not significant.

BATF Promotes Cell Survival by Increasing NAD⁺ Levels.

We then examined the effect of BATF on ΔΨm of primary T cells (Fig. 5A). After 72 h of stimulation, IL-12 treatment with CD3/CD28 ligation increased the percentage of TMRE^high population in Batf^+/+ CD8 T cells, but not in Batf^gfp/gfp CD8 T cells relative to CD3/CD28 ligation alone. These results suggest that BATF is required for IL-12–mediated maintenance of ΔΨm. In addition, the loss of ΔΨm and cell death were correlated (Figs. 5A and 2A), suggesting that only cells that maintain ΔΨm can survive. The surviving Batf^gfp/gfp and Batf^+/+ CD8 T cells showed similar ATP levels (Fig. S6A), but a significant difference in their NAD⁺ levels was observed (Fig. 5B). In synergy with CD3/CD28 stimulation, IL-12 treatment increased NAD⁺ levels in Batf^+/+ CD8 T cells but not in Batf^gfp/gfp CD8 T cells, whereas no difference in NADH levels was observed. These results suggest that BATF increases cellular NAD⁺ levels. Moreover, treatment with NAD⁺ increased the viability of Batf^gfp/gfp CD8 T cells stimulated with anti-CD3/CD28+IL-12 but did not affect Batf^+/+ CD8 T cells (Fig. 5C), suggesting that NAD⁺ rescues BATF-deficient CD8 T cells from cell death.

Fig. 5.

BATF promotes cell survival by increasing NAD⁺ levels. (A) The percentage of TMRE^high population in Batf^gfp/gfp and Batf^+/+ CD8 T cells stimulated with anti-CD3/CD28 with and without IL-12. (B) Cellular NAD⁺ and NADH levels in Batf^gfp/gfp and Batf^+/+ CD8 T cells stimulated with anti-CD3/CD28+IL-12 for 60 h. (C) Batf^gfp/gfp and Batf^+/+ CD8 T cells were stimulated with anti-CD3/CD28+IL-12 for 24 h, and NAD⁺ was added for 48 h. Cell viability was examined by PI staining. (D) Batf^gfp/gfp and Batf^+/+ CD8 T cells were plated in 96-well plates at 1 × 10⁶ cells/mL and stimulated with anti-CD3/CD28 with and without IL-12 for the indicated times. Glucose (Left) and lactate (Right) levels were measured in culture supernatant. Data are representative of two independent experiments. *P < 0.05; NS, not significant.

NAD⁺-consuming enzymes such as Sirt1 and Parp1 use NAD⁺ as a substrate, and their inhibitors increase NAD⁺ levels in activated T cells (25). Treatment with the Sirt1 inhibitors nicotinamide (Nam) and Sirtinol, but not with the Parp1 inhibitor 6-(5H)-phenanthridinone (PHE), increased the viability of Batf^gfp/gfp CD8 T cells to Batf^+/+ CD8 T-cell levels (Fig. S6B), suggesting that Sirt1 inhibition promotes the survival of activated T cells. These results are consistent with the finding that loss of Sirt1 increases T-cell activation (26). Treatment with the Sirt1 inhibitors, but not with NAD⁺, increased the viability of Batf^+/+ CD8 T cells stimulated with anti-CD3/CD28+IL-12 (Fig. S6B and Fig. 5C), suggesting that Sirt1 inhibition may also promote effector T-cell survival through factors other than NAD⁺.

To examine the role of BATF in metabolic flux, Batf^gfp/gfp and Batf^+/+ CD8 T cells were stimulated with anti-CD3/CD28 with or without IL-12, and the levels of glucose and lactate in culture supernatant were examined (Fig. 5D). IL-12 had no effect on glucose or lactate levels at 48 h. At 72 h, IL-12 treatment with CD3/CD28 ligation increased glucose consumption but not lactate production in Batf^+/+ CD8 T cells relative to CD3/CD28 ligation alone, suggesting that IL-12 promotes oxidative phosphorylation. By contrast, IL-12 treatment with CD3/CD28 ligation did not increase either glucose uptake or lactate production in Batf^gfp/gfp CD8 T cells relative to CD3/CD28 ligation alone. These results suggest that BATF plays an important role in IL-12–induced metabolic shift to oxidative phosphorylation.

BATF with c-Jun Transcriptionally Inhibits Sirt1 Expression.

Batf^gfp/gfp and Batf^+/+ CD8 T cells showed similar Sirt1 expression levels, when stimulated with anti-CD3/CD28+IL-12 (Fig. S6C). Because T cells increase ATP consumption upon antigen stimulation (4), we hypothesized that T cells expressing high levels of Sirt1 might die from energy collapse, whereas T cells expressing low levels of Sirt1 might selectively survive. To examine the role of BATF in Sirt1 expression without T-cell activation (i.e., under baseline conditions with low levels of ATP consumption), Batf^gfp/gfp and Batf^+/+ CD8 T cells were treated with IL-12 alone (Fig. 6A). Batf^gfp/gfp CD8 T cells showed higher Sirt1 expression than Batf^+/+ CD8 T cells. Memory T cells have demonstrated resistance to apoptosis (27). When stimulated with anti-CD3/CD28+IL-12, Batf^gfp/gfp CD44^high memory CD8 T cells showed significantly higher Sirt1 expression than Batf^+/+ CD44^high memory CD8 T cells (Fig. S6D). These results suggest that BATF inhibits Sirt1 expression.

Fig. 6.

BATF with c-Jun negatively regulates Sirt1 transcription. (A) Sirt1 mRNA expression in Batf^gfp/gfp and Batf^+/+ CD8 T cells stimulated with IL-12 for 48 h. Data are presented relative to expression levels in unstimulated Batf^+/+ CD8 T cells. (B) The 5′ regulatory region of Sirt1. A VISTA plot showing the comparison of the Sirt1 loci between mouse and either human, rat, chicken, or dog. (C) ChIP analysis of BATF binding to the Sirt1 locus. Batf^gfp/gfp and Batf^+/+ CD8 T cells were stimulated with anti-CD3/CD28+IL-12 for 60 h and subjected to ChIP analysis using control and anti-BATF Abs. (D and E) Luciferase activity of 293T cells transfected with the indicated expression vectors and the Sirt1 reporter construct (D) or mutation-containing reporter constructs (E). Data are representative of two independent experiments. *P < 0.05.

We then investigated the molecular mechanism by which BATF regulates Sirt1 expression. The Sirt1 locus has no conserved noncoding sequences in the 5′ upstream region except the promoter region (Fig. 6B). ChIP analysis using Batf^+/+ CD8 T cells stimulated with anti-CD3/CD28+IL-12 revealed that BATF specifically bound to the Sirt1 promoter region (Fig. 6C). To examine whether BATF regulates Sirt1 transcription, 293T cells were transfected with the Sirt1 reporter construct and BATF expression vector (Fig. 6D). Expression of c-Jun decreased Sirt1 transcription. Coexpression of BATF and c-Jun, but not BATF expression alone, decreased Sirt1 transcription, suggesting that BATF inhibits Sirt1 transcription in concert with c-Jun. To identify the regulatory elements important for negative regulation by these molecules, 293T cells were transfected with BATF and c-Jun expression vectors and reporter constructs containing mutations (Fig. 6E). Mutation of the AP-1 motif (position −186) decreased basal promoter activity and abolished the suppressive effect of c-Jun and BATF. These results suggest that the AP-1 motif (position −186) in the Sirt1 promoter region is important for basal Sirt1 transcription and negative regulation by c-Jun and BATF.

Discussion

Our study has revealed a unique mechanism of BATF-mediated effector CD8 T-cell differentiation (Fig. S7). We found that BATF deficiency inhibits effector CD8 T-cell differentiation. IL-12 with costimulation induces BATF in activated T cells at the late effector phase. IL-12 is one of the inflammatory cytokines that can provide a third signal required for effector CD8 T-cell differentiation. We show that BATF is required for IL-12–mediated histone acetylation and survival of effector CD8 T cells. BATF, together with c-Jun, transcriptionally inhibits Sirt1 expression, resulting in increased histone acetylation of the T-bet locus and elevated cellular NAD⁺ levels, which increase T-bet expression and ATP production, respectively. In turn, high levels of T-bet expression and ATP production promote effector differentiation and cell survival. Thus, BATF promotes effector CD8 T-cell differentiation by regulating both epigenetic remodeling and energy metabolism through Sirt1 expression.

We demonstrate the molecular mechanism by which BATF promotes histone acetylation. We identified Sirt1 as one of the target genes of BATF. Sirt1 is a class III histone deacetylase that is uniquely dependent on NAD⁺ for catalysis, and plays an important role in epigenetic gene silencing (21, 22). BATF, together with c-Jun, negatively regulated Sirt1 transcription. We found that BATF was recruited to the Sirt1 promoter region, and that the AP-1 motif (position −186) in the Sirt1 promoter region was important for basal Sirt1 transcription and negative regulation by BATF and c-Jun.

We show that BATF promotes histone acetylation of the T-bet locus, resulting in high expression of T-bet and its target genes. T-bet is the master regulator of type I effector differentiation both in CD4 and CD8 T cells (19, 28). By increasing T-bet expression, BATF promotes effector T-cell differentiation. It has been reported that T-bet expression is considerably enhanced and sustained in the presence of IL-12 (19, 29). We show that IL-12 induces BATF, and that BATF in turn induces high levels of T-bet expression through chromatin remodeling. These results suggest that BATF plays an important role in IL-12–induced T-bet expression.

Our study also revealed the mechanism by which BATF promotes effector T-cell survival. During effector differentiation, the balance between ATP production and ATP consumption determines T-cell fate: death or survival. BATF inhibited Sirt1 expression, resulting in increased levels of cellular NAD⁺. High levels of NAD⁺ drive both glycolysis and the TCA cycle, followed by oxidative phosphorylation, resulting in rapid and efficient ATP production. Thus, BATF promotes effector T-cell survival by preventing energy collapse. Importantly, BATF is induced by IL-12 at the late effector stage. In response to IL-12–induced epigenetic changes, BATF increases ATP production to sustain gene expression and subsequent protein synthesis.

Intriguingly, BATF deficiency induced high levels of Sirt1 expression in memory CD8 T cells upon stimulation, but not in naïve CD8 T cells. Unlike naïve T cells, memory T cells are resistant to apoptosis (27) and do not require a third signal to develop effector functions (6). We speculate that naïve and memory T cells may have different energy requirements for effector differentiation and survival. In primary CD8 T-cell response, we demonstrated that BATF acted downstream of IL-12 and promoted effector CD8 T-cell differentiation through T-bet up-regulation. In chronic infection, Quigley et al. (30), reported that the immunoinhibitory receptor PD-1 caused CD8 T-cell exhaustion by up-regulating BATF, whereas Kao et al. (31) reported that PD-1 expression was repressed by T-bet. Further investigation is required to determine the role of BATF in memory responses and the signaling pathways that induce BATF.

Although Sirt1 has been reported to modulate T-cell activation (26), the administration of Sirt1 inhibitors or activators is associated with many side effects because Sirt1 is expressed in many cell types. Because BATF expression is restricted to activated lymphocytes, BATF may be a promising therapeutic target to control effector T-cell generation for vaccine development and immunotherapies.

Materials and Methods

Mice.

Mice were used and maintained under specific pathogen-free conditions according to the guidelines of the institutional animal care committee at Tokyo Medical and Dental University. C57BL/6 (B6) mice were obtained from Japan SLC. Rag1^−/− mice (B6 background) were obtained from The Jackson Laboratory. Generation of the Batf^gfp allele is described in SI Materials and Methods.

Cell Preparation, Stimulation, Flow Cytometry, Real-Time PCR, ChIP Assay, and Luciferase Assay.

Detailed methods and primer sequences are provided in SI Materials and Methods and Tables S1 and S2.

Metabolic Assays.

Intracellular ATP content was measured using the luciferin-luciferase method (ATP Bioluminescence Assay Kit HS II; Roche) as recommended by the manufacturer. NAD⁺ and NADH levels were measured from whole-cell extracts using the NAD⁺/NADH Quantification kit (Biovision), according to manufacturer instructions. GAPDH activity was measured using the Kdalert GAPDH Assay kit (Ambion). Glucose and lactate levels in cell culture supernatants were measured using Glucose Assay kit II (Biovision) and Lactate Assay kit II (Biovision), respectively.

Statistics.

All data are expressed as mean ± SD (n = 3 or more). Statistical analysis was performed using the unpaired two-tailed Student's t test (*P < 0.05; **P < 0.01; NS, not significant).