Involvement of GDH3-encoded NADP+-dependent Glutamate Dehydrogenase in Yeast Cell Resistance to Stress-induced Apoptosis in Stationary Phase Cells*

Background: Gdh1 and Gdh3 are glutamate-synthesizing isofunctional NADP-GDH in S. cerevisiae. Results: Stationary phase-specific GDH3 expression and degradation of Gdh1 were responsible for the Gdh3-dependent glutamate supply and resistance to stress-induced apoptosis in stationary phase. Conclusion: Gdh3 plays a role distinct from Gdh1 by rendering cells resistant to stress and aging. Significance: This provides mechanistic insight into apoptosis and protein degradation in response to stress. Glutamate metabolism is linked to a number of fundamental metabolic pathways such as amino acid metabolism, the TCA cycle, and glutathione (GSH) synthesis. In the yeast Saccharomyces cerevisiae, glutamate is synthesized from α-ketoglutarate by two NADP+-dependent glutamate dehydrogenases (NADP-GDH) encoded by GDH1 and GDH3. Here, we report the relationship between the function of the NADP-GDH and stress-induced apoptosis. Gdh3-null cells showed accelerated chronological aging and hypersusceptibility to thermal and oxidative stress during stationary phase. Upon exposure to oxidative stress, Gdh3-null strains displayed a rapid loss in viability associated with typical apoptotic hallmarks, i.e. reactive oxygen species accumulation, nuclear fragmentation, DNA breakage, and phosphatidylserine translocation. In addition, Gdh3-null cells, but not Gdh1-null cells, had a higher tendency toward GSH depletion and subsequent reactive oxygen species accumulation than did WT cells. GSH depletion was rescued by exogenous GSH or glutamate. The hypersusceptibility of stationary phase Gdh3-null cells to stress-induced apoptosis was suppressed by deletion of GDH2. Promoter swapping and site-directed mutagenesis of GDH1 and GDH3 indicated that the necessity of GDH3 for the resistance to stress-induced apoptosis and chronological aging is due to the stationary phase-specific expression of GDH3 and concurrent degradation of Gdh1 in which the Lys-426 residue plays an essential role.

tarate and ammonia using NAD(H) and NADP(H) as cofactors. The enzyme is present in almost all living organisms and plays a pivotal role in important cellular processes, such as the tricarboxylic acid (TCA), ammonia management, and energy metabolism (1). Although the equilibrium of mammalian GDH reaction favors the synthesis of glutamate, this reaction hardly occurs in mammalian cells because it requires high ammonia (K m ϭ 12-25 mM) and ␣-ketoglutarate (K m ϭ 1.0 -2.0 mM) levels not likely to be seen under base-line conditions (1)(2)(3). When a high ammonia concentration prevails, however, glutamate synthesis via reductive amination of ␣-ketoglutarate by GDH may function as a detoxification process (4). The generation of ␣-ketoglutarate via oxidative deamination of glutamate leads to the production of NAD(P)H, GTP, and ATP through the TCA cycle in mitochondria. Therefore, GTP and ATP are allosteric inhibitors of GDH, whereas ADP is an activator (5).
In all mammals, except for humans and some closely related species, GDH is encoded by a single gene. However, humans and other primates have two distinct genes, GLUD1 and GLUD2, that encode two isoforms of GDH, hGDH1 and hGDH2, respectively (6,7). GLUD1 is widely expressed in almost all human tissues including liver, brain, pancreas, and kidney, but not muscle. In pancreatic ␤-cells, immoderate generation of ␣-ketoglutarate due to an activating mutation of hGDH1 leads to increased insulin exocytosis through ATP overproduction (8 -10). The activity of hGDH1 in ␤-cells is repressed by ADP-ribosylation catalyzed by SIRT4, one of seven homologs of yeast Sir2, which subsequently results in the down-regulation of insulin secretion (11). GLUD2 is expressed predominantly in a limited range of tissues including retina, brain, and testis (7). Despite the high similarity between hGDH1 and hGDH2, as they share all but 16 of their 505 amino acid residues, they show definite differences in their basic catalytic activities and allosteric regulation (3,5,12,13). Thus, the enzymes may contribute differentially to cellular processes. Deregulation of the activity of hGDH2 caused by an S445A substitution in the regulatory domain enhances glutamate oxidation, which results in enhanced nigral cell degeneration (14).
In contrast to mammals in which the reductive amination of ␣-ketoglutarate by GDH does not occur to an appreciable extent, the yeast Saccharomyces cerevisiae cannot only biosynthesize glutamate but also utilize it via the reactions catalyzed by three distinct GDH isoenzymes. The NAD ϩ -dependent GDH (NAD-GDH; Gdh2) encoded by GDH2 catalyzes reversible oxidative deamination of glutamate to ␣-ketoglutarate and ammonia (15). Glutamate anabolism via amination of ␣-ketoglutarate is catalyzed by two different NADP ϩ -dependent GDH (NADP-GDH), Gdh1 and Gdh3, encoded by GDH1 and GDH3, respectively (16). Yeast cells lacking Gdh2 show impaired glutamate utilization and poor growth in minimal glucose media containing glutamate as a nitrogen source (15). Although Gdh2 is responsible for the reversible deamination of glutamate, it is not involved in glutamate biosynthesis during normal growth (15). On the contrary, Gdh1 and Gdh3 catalyze glutamate biosynthesis through the reductive amination of ␣-ketoglutarate (16). An alternative pathway for glutamate biosynthesis is accomplished by the combined activities of GLN1encoded glutamine synthetase and the GLT1-encoded glutamate synthase (17,18). However, both GDH3 and GLT1 are dispensable for yeast growth in minimal glucose medium containing ammonia as a sole nitrogen source, indicating that Gdh1 is the primary enzyme for glutamate biosynthesis (16). Gdh1 uses ␣-ketoglutarate at a higher rate than does Gdh3 and almost solely contributes to the NADP-GDH activity under fermentative growth conditions with glucose as the sole carbon source. However, during post-diauxic growth, the Gdh1/Gdh3 ratio decreases, and the majority of the total NADP-GDH activity is due to Gdh3, even though GDH1 transcription proceeds during this growth phase (19). This phenomenon is in accordance with a previous observation in that NADP-GDH is degraded during glucose starvation (20).
In the present study, we examined the differential roles of two NADP-GDH, Gdh1 and Gdh3, in sustaining stress resistance. Our results indicate that Gdh3, but not Gdh1, is responsible for tolerance to stress-induced apoptosis in stationary phase cells, as there is stationary phase-specific expression of GDH3 and degradation of Gdh1.

EXPERIMENTAL PROCEDURES
Yeast Strains, Media, and Transformation-S. cerevisiae strains used in this study are listed in Table 1. The YPD medium consisted of 1% yeast extract, 2% peptone, 75 M adenine sulfate, and 2% glucose. The synthetic complete dextrose (SCD) medium consisted of 0.67% yeast nitrogen base without amino acids (Difco, Detroit, MI), 0.14% yeast synthetic drop-out medium supplement without leucine and uracil (Sigma-Aldrich), and 2% glucose. When necessary, SCD was supplemented with 2 mM uracil and 1 mM amino acids (glutamate and leucine). For solid media, 2% agar (Difco) was added. Transformation of yeast strains was performed using the lithium acetate method (21).
Plasmids-Plasmids used in this study are listed in Table 1. For construction of YCpGdh1, a 2.4-kb DNA fragment containing the promoter and coding sequence of GDH1 was PCR-amplified from the genomic DNA of an S. cerevisiae WT strain (BY4741) with the primers P1-1 and P1-2 ( Table 2). The resulting fragment was cloned into the low copy yeast/Escherichia coli shuttle vector YCp111. In parallel, to construct YCpGdh3, a 2.4-kb fragment containing the complete GDH3 gene was PCRamplified with the primers P3-1 and P3-2 (Table 2) and cloned into YCP111. To construct YEpGdh2, a 4.3-kb fragment containing the complete GDH2 gene was amplified by PCR with the primers P2-1 and P2-2 (Table 2) and cloned into the high copy yeast/E. coli shuttle vector YEp352.
Two chimeric genes were constructed by a reciprocal exchange of the GDH1 and GDH3 promoter regions. For construction of YCpP GDH1 -Gdh3, a 1.0-kb fragment containing the promoter region of GDH1 (gdh1 P ) was PCR-amplified with the primers P1-1 and P1-3 (Table 2). A 1.4-kb fragment containing the complete coding sequence of Gdh3 (gdh3 ORF ) was amplified with the primers P3-3 and P3-2 ( Table 2). The two PCR products were fused by overlapping PCR, and the resulting 2.4-kb PCR product (gdh1 P ::gdh3 ORF ) was cloned into YCp111 to yield YCpP GDH1 -Gdh3. Similarly, for construction of YCpP GDH3 -Gdh1, a 1.0-kb fragment containing the promoter region of GDH3 (gdh3 P ) was amplified with the primers P3-1 and P3-4 ( Table 2). A 1.4-kb fragment containing the complete coding sequence of Gdh1 (gdh1 ORF ) was amplified with the primers P1-4 and P1-2 ( Table 2). The two PCR products were combined by overlapping PCR, and the resulting 2.4-kb PCR product (gdh3 P ::gdh1 ORF ) was cloned into YCp111 to yield YCpP GDH3 -Gdh1. The genes coding for FLAG-tagged derivatives of Gdh1 and Gdh3 were constructed as follows. A 2.3-kb PCR fragment containing the promoter and coding region of GDH1 followed by a FLAG-coding sequence was amplified with the primers P1-5 and P1-6 ( Table 2) and cloned into YCp111 to yield YCpGdh1-FLAG. A 2.6-kb PCR fragment containing the promoter and coding region of GDH3 followed by the FLAG-coding sequence was amplified with the primers P3-5 and P3-6 ( Table 2) and cloned into YCp111 to yield YCpGdh3-FLAG.
To construct the LacZ reporter plasmids for analysis of the expression of GDH1 and GDH3, the promoter regions and short N-terminal coding sequences (6 -8 amino acids) of GDH1 (gdh1 PЈ ) and GDH3 (gdh3 PЈ ) were fused in-frame to the lacZ gene. A 0.9-kb gdh1 PЈ fragment was PCR-amplified with the primers P1-7 and P1-8 ( Table 2). The resulting PCR product was fused in-frame to a promoter-less lacZ gene to construct the gdh1 PЈ ::lacZ fusion in YEp353 (YEpP GDH1 -LacZ). In parallel, a 1.2-kb DNA fragment containing the gdh3 PЈ fragment was PCR-amplified with the primers P3-5 and P3-9 ( Table 2). The PCR fragment was then fused in-frame to a promoterless lacZ gene to form a gdh3 PЈ ::lacZ fusion in YEp353 (YEpP GDH3 -LacZ).
Survival Tests-For cell survival experiments, yeast cells were grown in SCD broth at 30°C for 2 days (unless otherwise indicated) and suspended in PBS to a concentration of 1.0 ϫ 10 8 cells ml Ϫ1 . Two-milliliter samples were taken and subjected to heat (50°C, 30 min) or oxidative stress (1 mM H 2 O 2 , 1 h). One hundred-microliter aliquots of each sample were taken and serially diluted with 10-fold steps. For the colony-forming unit (CFU) assay, a 100-l aliquot of each dilution was spread onto an SCD plate, and colonies were scored after incubation at 30°C for 3 days.
Test for Apoptotic Markers-TUNEL assays were performed as described by Madeo et al. (22). In brief, cells were fixed with 3.7% formaldehyde, digested with lyticase (Sigma-Aldrich), applied to a polylysine-coated slide, treated with 0.3% H 2 O 2 to block endogenous peroxidases, permeabilized with 0.1% Triton X-100, incubated with 10 l of TUNEL reaction mixture (Roche Diagnostics) for 60 min at 37°C, and incubated with 10 l of Converter POD (Roche Diagnostics) for 10 min. A coverslip was mounted with a drop of Kaiser's glycerol gelatin (Merck, Darmstadt, Germany). Bright-field images of cells were acquired using a BX51 universal research microscope (Olympus Corp., Tokyo) equipped with a DP11 digital camera (Olympus).
To monitor the levels of intracellular reactive oxygen species (ROS), cells were stained with 10 g ml Ϫ1 2Ј,7Ј-dichlorodihydrofluorescein diacetate (DCFH-DA) for 2 h with shaking at 30°C (23). For analysis of nuclear fragmentation, cells were fixed in 3.7% formaldehyde and stained with 0.5 g ml Ϫ1 DAPI (Sigma-Aldrich) (22). Externalization of phosphatidylserine was assayed using FITC-coupled annexin V (BD Biosciences) as described by Madeo et al. (22). Cells were suspended in digestion buffer, and cell walls were digested as described above. The spheroplasts were stained with both FITC-annexin V and 500 g ml Ϫ1 propidium iodide (PI; BD Biosciences).
Fluorescence microscopy was performed under a BX51 universal research microscope (Olympus) equipped with a DP11 digital camera (Olympus) using the appropriate filter sets (FITC filter for DCFH-DA and FITC-annexin V; DAPI filter for DAPI; and rhodamine filter for PI staining). To determine the frequencies of morphological phenotypes (TUNEL, DCFH-DA, DAPI, FITC-annexin V, and PI), at least 300 cells from three independent experiments were evaluated.
NADP-GDH and ␤-Gal Assay-To prepare whole cell extracts for NADP-GDH assays, yeast cells grown in liquid medium were harvested by centrifugation and washed twice with distilled water and once with extraction buffer (0.1 M potassium phosphate (pH 7.5), 1 mM EDTA, 1 mM DTT, 1 mM PMSF, and 50 l ml Ϫ1 N-␣-p-tosyl-L-lysine chloromethyl ketone) (24). Cell pellets were stored at Ϫ20°C. Soluble extracts were prepared by mechanical disruption of cells suspended in extraction buffer by agitation with glass beads in a vortex mixer (five cycles of 1 min of agitation and 1 min of incubation on ice) followed by brief centrifugation to clarify the lysate. NADP-GDH activity was assayed by measuring the oxidation of NADPH according to the method of Doherty (25). One unit of NADP-GDH activity corresponds to 1 mol of NADP ϩ produced/min. Protein was measured by the method of Lowry et al. (26) using BSA as a standard.
␤-Gal assays were performed using a yeast ␤-gal assay kit (Thermo Scientific) according to the manufacturer's instructions. ␤-Gal activity was calculated using the following equation: ␤-gal activity ϭ (1000 ϫ A 420 )/(t ϫ V ϫ OD 660 ) in which t ϭ time (in minutes) of incubation and V ϭ volume of cells (ml) used in the assay.
Glutathione and Glutamate Assays-Intracellular levels of the total glutathione pool, glutathione (GSH) and glutathione disulfide (GSSG), were determined by measuring the rate of 2-nitro-5-thiobenzoic acid formation from 5,5Ј-dithiobis-(2nitrobenzoic acid) (Sigma-Aldrich) in the glutathione recycling system (27). Five microliters of cell lysate was added to 1 ml of 100 mM phosphate buffer (pH 7.5) containing 0.6 mM 5,5Ј-dithiobis-(2-nitrobenzoic acid), 5 mM EDTA, 0.2 mM NADPH, and 1 unit ml Ϫ1 glutathione reductase (Sigma-Aldrich). The rate of increase in A 412 was monitored. To determine the levels of GSSG, NADPH consumption was monitored by measuring the rate of decrease in A 340 . Intracellular levels of glutamate were determined using a colorimetric glutamate analysis kit (R-Biopharm, Darmstadt, Germany) according to manufacturer's instructions.

Deletion of GDH3 Causes Increased Sensitivity to Thermal and Oxidative Stress in Stationary Phase Cells and Accelerates
Chronological Aging-To address the differential roles of the two NADP-GDH, Gdh1 and Gdh3, we examined the resistance against heat and oxidative stress of NADP-GDH yeast mutants and their derivatives that harbor either ectopically expressed GDH1 or GDH3 genes. Although the survival rates of the WT, ⌬gdh1, and ⌬gdh1⌬gdh3/YCpGdh3 yeast strains grown in SCD for 2 days were ϳ70% after 30 min of exposure to heat stress at 50°C, less than 10% of the initial population of the ⌬gdh3 (BY4741-YAL062W), ⌬gdh1⌬gdh3/YCp111, and ⌬gdh1⌬gdh3/YCpGdh1 strains survived following the same treatment (Fig. 1A). In addition, only 10 -15% of the initial population of ⌬gdh3, ⌬gdh1⌬gdh3/YCp111, and ⌬gdh1⌬gdh3/ YCpGdh1 cells cultured for 2 days survived after a 1-h exposure to 1 mM H 2 O 2 , whereas more than 75% of the WT, ⌬gdh1, and ⌬gdh1⌬gdh3/YCpGdh3 cells subjected to the same treatment remained alive. These results indicate that deletion of GDH3 causes increased sensitivity to both heat and oxidative stress in stationary phase cells, whereas impairment of GDH1 leads to no significant stress-sensitive phenotype. Additionally, the stress sensitivity of the Gdh3-null mutants was not suppressed by ectopic expression of GDH1.
Intracellular ROS is a mediator of chronological aging in yeast (29,30). Accordingly, we have shown previously that ROS accumulation caused by depletion of intracellular glutamate in the yeast cells lacking Cit1, which catalyzes the first step of the TCA cycle, leads to accelerated chronological aging (31). To determine whether the deletion of GDH1 or GDH3 affects the survival of cells in chronologically aged yeast cultures, WT and mutant cells were cultured in SCD for 13 days, and the surviving cells were quantified by CFU assay. Although ϳ50% of the initial populations of the WT, ⌬gdh1, and ⌬gdh1⌬gdh3/YCpGdh3 strains remained viable after a 7-day culture, only 2-5% of the ⌬gdh3, ⌬gdh1⌬gdh3/YCp111, and ⌬gdh1⌬gdh3/YCpGdh1 cells survived after the same culture period (Fig. 1B). In addition, the survival rates of ⌬gdh3, ⌬gdh1⌬gdh3/YCp111, and ⌬gdh1⌬gdh3/YCpGdh1 cultures reached their minimum levels after 7 days and then increased continuously up to 20 -30% of the initial population during a subsequent 6-day cultivation. This result indicates that the Gdh3-null strains experienced a much faster and more severe aging process than did their isogenic WT population. Furthermore, Gdh3-null cells were subjected to accelerated aging-dependent cell death followed by adaptive regrowth upon long-term cultivation. Our data support a previous study that reported that during the chronological aging process of yeast, premature apoptotic death promotes the regrowth of a subpopulation of better adapted mutants (29).
Oxidative Stress-induced Death of Stationary Phase Gdh3null Cells Is Caused by ROS-mediated Apoptosis-To clarify whether the increased sensitivity of the Gdh3-null cells to heat and oxidative stress is caused by a higher tendency of the mutant populations toward stress-induced apoptosis, we looked for cytological and biochemical features of apoptosis in the mutant cells. We analyzed the cells for the presence of intracellular ROS, which are both necessary and sufficient for inducing apoptosis in yeast (32). As a probe for ROS production, we used DCFH-DA, which can be oxidized to the fluorescent chromophore 2Ј,7Ј-dichlorofluorescein primarily by the action of peroxide (H 2 O 2 ). As seen in the fluorescence micrographs (Fig.  2, A and D), a vast majority of the ⌬gdh3 (72%) and ⌬gdh1⌬gdh3 cells (88%) grown for 2 days in SCD broth at 30°C were fluorescent after a 1-h exposure to 10 mM H 2 O 2 . On the other hand, the WT and ⌬gdh1 cells did not show any detectable levels of 2Ј,7Ј-dichlorofluorescein fluorescence under these conditions. Therefore, these results indicate that deletion of GDH3 causes oxidative stress-dependent accumulation of ROS in stationary phase cells.
Because ⌬gdh1 and ⌬gdh1⌬gdh3 cells exposed to oxidative stress have elevated ROS levels, we assayed for nuclear fragmentation, a well established cytological hallmark of apoptosis. A normal, single-round nucleus was detected by DAPI staining in WT and mutant cells under normal conditions (untreated) (Fig. 2, B and D). On the other hand, ϳ66 and 72% of the ⌬gdh3 and ⌬gdh1⌬gdh3 cells, respectively, displayed irregularly shaped and fragmented nuclei at 1 h after exposure to H 2 O 2 . In contrast, only ϳ20% of the similarly treated WT and ⌬gdh1 strains had abnormal nuclei.  DECEMBER 28, 2012 • VOLUME 287 • NUMBER 53

Role of Gdh3 in Resistance to Stress-induced Apoptosis
We also examined nuclear DNA fragmentation, another feature of apoptosis, using TUNEL analysis in which fluorescent nucleotides were added to the 3Ј-OH ends of the DNA fragments making the phenomenon visible by fluorescence microscopy. None of the WT or mutant cells had TUNEL-positive nuclei under normal conditions (untreated) (Fig. 2, B and D). However, after 1 h of exposure to H 2 O 2 , ϳ58 and 62% of ⌬gdh3 and ⌬gdh1⌬gdh3 cells, respectively, showed intensely stained nuclei indicating DNA strand breakage. On the other hand, the nuclei of both the WT and ⌬gdh1 cells remained unstained or only slightly stained after the same treatment.
In yeast cells as well as mammalian cells, translocation of phosphatidylserine, which is predominantly located on the inner leaflet of the plasma membrane under normal conditions, to the outer leaflet serves as a sensitive marker for the early stages of apoptosis (33). For detection of phosphatidylserine in the outer layer of the cytoplasmic membrane, spheroplasts formed from the cells subjected to oxidative stress were stained with both FITC-annexin V and PI. The stained cells were then observed by fluorescence microscopy. When exposed to H 2 O 2 for 1 h, a portion of the stationary phase cells from the ⌬gdh3 and ⌬gdh1⌬gdh3 strains were stained exclusively with FITCannexin V, indicating that they were undergoing an early stage of apoptosis (Fig. 2C). In addition, some of the oxidatively stressed cells were dually stained with both FITC-annexin V and PI, suggesting the late apoptotic phases or necrosis. On the contrary, only a negligible portion of the similarly treated WT and ⌬gdh1 cells were stained solely with FITC-annexin V and thus were considered to be apoptotic. These results suggest that stationary phase Gdh3-null cells are much more susceptible to oxidative and thermal stress-induced cell death (bearing the structural attributes of apoptosis) than are the isogenic WT strains.
Stationary Phase Gdh3-null Cells Exhibit GSH and Glutamate Depletion, Which Leads to Increased Susceptibility to Oxidative Stress-induced Apoptosis-GSH synthesized from glutamate, glycine, and cysteine by the action of ␥-glutamylcysteine FIGURE 2. Yeast Gdh3-null cells display apoptotic markers under stress conditions. A, ROS accumulation in yeast cells exposed to oxidative stress. Yeast cells (WT, ⌬gdh1, ⌬gdh3, and ⌬gdh1⌬gdh3) grown in SCD for 2 days were exposed to 10 mM H 2 O 2 for 1 h. Cells were stained with DCFH-DA and examined by fluorescence microscopy. Bar, 10 m. DIC, differential interference contrast. B, nuclear or DNA fragmentation in yeast cells exposed to oxidative stress. Yeast cells prepared and treated as described in A were stained with DAPI or TUNEL reagent and examined by fluorescence microscopy. Bar, 10 m. C, phosphatidylserine externalization in yeast cells exposed to oxidative stress. Yeast strains prepared and treated as described in A were stained with FITC-annexin V and PI and examined by fluorescence microscopy. Bar, 10 m. D, percentage of yeast cells exhibiting the typical hallmarks of apoptosis after exposure to oxidative stress (10 mM H 2 O 2 , 1 h). The number of cells with accumulated ROS (DCFH-DA(ϩ)), fragmented nuclei (DAPI(ϩ)), and fragmented DNA (TUNEL(ϩ)) were determined from ϳ500 -700 cells in three independent experiments. Values are the mean Ϯ S.E. *, p Ͻ 0.001; **, p Ͻ 0.005 (two-tailed Student's t test versus WT). synthase (Gsh1) and GSH synthase (Gsh2) is one of the most prevalent reducing thiol compounds functioning as an antioxidant in nearly every aerobic organism (34). Accordingly, depletion of GSH or its precursor, glutamate, causes oxidative or thermal stress-induced apoptosis in yeast cells (31).
To determine whether the ROS accumulation in stressed Gdh3-null cells is facilitated by depletion of GSH, which is used by many peroxidases to scavenge H 2 O 2 and a multitude of organic hydroperoxides (35), we evaluated the levels of GSH in the mutant cells (Table 3). When ⌬gdh3 and ⌬gdh1⌬gdh3 cells were grown for 2 days in SCD containing 1 mM glutamate, the levels of GSH were ϳ26 and 29% of that in WT cells, respectively. In contrast, ⌬gdh3 and ⌬gdh1⌬gdh3 mutants exhibited GSH levels similar to those of WT cells when grown in SCD supplemented with a higher concentration of either glutamate or GSH (88 and 92% in the presence of 10 mM glutamate and 76 and 80% in the presence of 10 mM GSH, respectively). Accordingly, the intracellular levels of glutamate were similar to those of GSH. Glutamate concentrations in the cells of ⌬gdh3 and ⌬gdh1⌬gdh3 mutants were 20% less than that of WT cells when cultured in SCD containing 1 mM glutamate. The depletion of glutamate in ⌬gdh3 and ⌬gdh1⌬gdh3 mutant cells recovered to at least 75% of the normal level when the medium was supplemented with either 10 mM GSH or 10 mM glutamate.
We examined the effect of exogenous GSH and glutamate on stress-induced apoptosis of the yeast strains. In accordance with the restoration of the intracellular GSH and glutamate levels, ⌬gdh3 and ⌬gdh1⌬gdh3 cells grown in SCD medium supplemented with 10 mM GSH or glutamate showed almost the same level of survival after a 1-h exposure to 10 mM H 2 O 2 as did WT and ⌬gdh1 cells (Fig. 3A). In addition, when stained with DCFH-DA, ⌬gdh3 and ⌬gdh1⌬gdh3 cells showed only slightly higher levels of DCF fluorescence compared with WT and ⌬gdh1 cells under oxidative stress conditions. This indicates significant attenuation of stress-induced ROS accumulation in the presence of 10 mM GSH or glutamate (Fig. 3B). Together, these data suggest that the deletion of GDH3 results in the depletion of intracellular GSH, which, in turn, induces the accumulation of ROS and accompanying apoptotic cell death during stress conditions.

The Increased Susceptibility of Stationary Phase Gdh3-null Cells to Thermal and Oxidative Stress-induced Apoptosis Is
Suppressed by GDH2 Deletion-The NAD-GDH (Gdh2) encoded by GDH2 catalyzes oxidative deamination of gluta-

. Exogenous supply of GSH or glutamate suppresses the hypersusceptibility of yeast Gdh3-null cells to stress-induced apoptosis mediated by ROS accumulation.
A, effect of exogenous GSH and glutamate on survival of yeast cells exposed to oxidative stress. Yeast strains (WT, ⌬gdh1, ⌬gdh3, and ⌬gdh1⌬gdh3) grown in SCD with or without 10 mM GSH or glutamate for 2 days were exposed to 10 mM H 2 O 2 for 1 h. Surviving cells were evaluated by CFU assays. Values are the mean Ϯ S.E. of three independent experiments. B, effect of exogenous GSH and glutamate on ROS accumulation in yeast cells exposed to oxidative stress. Yeast cells prepared and treated as described in A were stained with DCFH-DA and examined by fluorescence microscopy. The number of DCFH-DA-positive cells was estimated in fluorescence images and total cells in the corresponding differential interference contrast images. Approximately 500 -700 cells were observed in three independent experiments. Values are the mean Ϯ S.E. *, p Ͻ 0.005 (two-tailed Student's t test versus untreated). mate to ␣-ketoglutarate and ammonia (15), hence the reverse reaction of glutamate synthesis from ␣-ketoglutarate catalyzed by either Gdh1 or Gdh3. To investigate how Gdh2 function is related to stress-induced apoptosis, we analyzed the effect of GDH2 deletion and the presence of ectopic GDH2 on the hypersusceptibility of Gdh3-null mutants. The introduction of the ⌬gdh2 mutation caused increased resistance to both thermal and oxidative stress in Gdh3-null strain (Fig. 4A). The survival rates of ⌬gdh2⌬gdh3 cells grown for 2 days in SCD were ϳ55% after a 1-h exposure to oxidative stress (10 mM H 2 O 2 ) and 47% after a 30-min exposure to thermal stress (50°C), whereas the survival rates of ⌬gdh3 cells were ϳ15 and 5%, respectively. Correspondingly, ⌬gdh2⌬gdh3 cells exposed to oxidative and thermal stress exhibited significantly lower levels of ROS accumulation (ϳ44 and 62%, respectively) compared with ⌬gdh3 cells (ϳ76 and 92%, respectively) when monitored by fluorescence microscopy using DCFH-DA (Fig.  4B). In addition, the levels of GSH and glutamate in ⌬gdh2⌬gdh3 cells were ϳ2.6and 2.9-fold higher than that of ⌬gdh3 cells, respectively, when the cells were grown for 2 days in SCD containing a basal level of glutamate (1 mM) ( Table 3). On the contrary, all of the yeast strains containing ectopic GDH2 in high copy vectors (WT/YEpGdh2, ⌬gdh1/YEpGdh2, ⌬gdh2/YEpGdh2, and ⌬gdh3/YEpGdh2) showed reduced resistance against both thermal and oxidative stress-induced apoptosis, although to a limited extent, compared with their corresponding host strains (Fig. 4A). Thus, it appears that deletion of GDH2 compensates for the depletion of intracellular glutamate and GSH followed by stress-induced ROS accumulation and apoptotic cell death in stationary phase Gdh3-null cells, whereas ectopic expression of GDH2 enhances the depletion of intracellular glutamate and GSH. These results are in agreement with the fact that Gdh2 catalyzes the oxidative deamination of glutamate to ␣-ketoglutarate resulting in decreased intracellular glutamate and GSH levels.

GDH1, but Not GDH3, Is Responsible for the Resistance against Stress-induced Apoptosis in Logarithmic Phase Cells-
To determine the differential roles of GDH1 and GDH3 in sustaining resistance to stress-induced apoptosis, we monitored the ectopic expression of gdh1 PЈ ::lacZ and gdh3 PЈ ::lacZ hybrid genes in WT/YEpP GDH1 -LacZ and WT/YEpP GDH3 -LacZ strains, respectively. Considerably higher levels of ␤-gal activity (Ն1500 units) were detected in the WT/YEpP GDH1 -LacZ cell extracts, regardless of the growth stage, than in the WT/ YEpP GDH3 -LacZ cell extracts (Fig. 5). The level of ␤-gal activity in WT/YEpP GDH3 -LacZ cells was low (ϳ250 units) after a 24-h culture into late logarithmic phase but gradually increased dur-

. Deletion of GDH2 suppresses the hypersusceptibility of yeast Gdh3-null cells to stress-induced apoptosis mediated by ROS accumulation.
A, effect of ⌬gdh2 mutation on survival of yeast cells exposed to thermal or oxidative stress. Yeast strains (WT, ⌬gdh1, ⌬gdh2, ⌬gdh3, ⌬gdh2⌬gdh3, WT/YEpGDH2, ⌬gdh1/YEpGDH2, ⌬gdh2/YEpGDH2, and ⌬gdh3/YEpGDH2) grown in SCD for 2 days were exposed to either thermal (50°C, 30 min) or oxidative stress (10 mM H 2 O 2 , 1 h). Surviving cells were evaluated by CFU assays. Values are the means Ϯ S.E. of three independent experiments. *, p Ͻ 0.001 (two-tailed Student's t test versus ⌬gdh3). B, effect of the ⌬gdh2 mutation on ROS accumulation in yeast cells exposed to thermal or oxidative stress. Yeast cells prepared and treated as described in A were stained with DCFH-DA and examined by fluorescence microscopy. The number of DCFH-DA-positive cells was estimated in fluorescence images and total cells in corresponding differential interference contrast images. Approximately 500 -700 cells were observed in three independent experiments. Values are the mean Ϯ S.E. *, p Ͻ 0.001; **, p Ͻ 0.05 (two-tailed Student's t test versus ⌬gdh3). ing the following 48-h period up to levels 4-fold higher than during the logarithmic phase (ϳ1,000 units). Therefore, GDH1 was expressed consistently throughout all of the growth periods at a relatively high level, whereas GDH3 was expressed at negligible levels during the logarithmic phase but increased gradually to a higher level during stationary phase. These results are in agreement with a previous report indicating that GDH3 transcription is repressed by glucose and is induced only under respiratory conditions or during the stationary phase (36), whereas transcription of GDH1 in cells grown with glucose is regulated by the transcriptional activators Gln3, Hap2, and Hap3 (37).
To determine whether the transcription profiles of GDH1 and GDH3 are consistent with the subsequent gene expression events, we analyzed the enzymatic activities and protein levels of Gdh1 and Gdh3 in WT cells and a variety of mutant strains. The levels of NADP-GDH activity in the exponential phase (24 h) cells of ⌬gdh3, ⌬gdh1⌬gdh3/YCpGdh1, and WT strains were ϳ10-fold higher than in ⌬gdh1 and ⌬gdh1⌬gdh3/YCpGdh3 cells. Meanwhile, the levels of NADP-GDH activity in the sta-tionary phase (48 h) cells of ⌬gdh1, ⌬gdh1⌬gdh3/YCpGdh3, and WT strains were at least seven times higher than in ⌬gdh3 and ⌬gdh1⌬gdh3/YCpGdh1 cells (Table 4). In immunoblotting, Gdh1 protein was detected in the exponential phase (24 h) cells of ⌬gdh3, ⌬gdh1⌬gdh3/YCpGdh1, and WT strains but not in the stationary phase (48 h) cells of the strains (Fig. 6A). On the contrary, Gdh3 protein was detected in the stationary phase cells of ⌬gdh1, ⌬gdh1⌬gdh3/YCpGdh3, and WT strains but not in their exponential phase cells. In addition, a substantial level of Gdh1-FLAG signal was detected in ⌬gdh1⌬gdh3/YCpGdh1-FLAG cells during the exponential phase (12 and 24 h), but no signal was identified during the stationary phase (48 and 72 h) (Fig. 6B). However, ⌬gdh1⌬gdh3/YCpGdh3-FLAG cells showed only negligible levels of Gdh3-FLAG signal during the logarithmic phase followed by a marked increase in Gdh3-FLAG signal to the highest level during stationary phase. Together, these results suggest that a large majority of the NADP-GDH activity is Gdh1-dependent until the cells reach late logarithmic phase and then becomes Gdh3-dependent during the stationary phase. . Gdh3 protein is stable throughout all growth stages, whereas Gdh1 is subjected to stationary phase-specific degradation. A, immunoblot analysis of the Gdh1 and Gdh3 proteins in yeast cells at different growth stages. Yeast strains (WT, ⌬gdh1, ⌬gdh3, ⌬gdh1⌬gdh3/YCp111, ⌬gdh1⌬gdh3/Gdh1, ⌬gdh1⌬gdh3/YCpGdh3) were grown in SCD, and samples were taken after 24 and 48 h of culture. Cell extracts were immunoblotted and probed with anti-Gdh1, anti-Gdh3 and anti-tubulin antibodies. B, immunoblot analysis of the FLAG-tagged Gdh1 and Gdh3 proteins in yeast cells carrying the promoterswapped derivatives of GDH1 and GDH3 at different growth stages. Yeast strains (⌬gdh1⌬gdh3/YCpGdh1-FLAG, ⌬gdh1⌬gdh3/YCpGdh3-FLAG, ⌬gdh1⌬gdh3/ YCpP GDH1 -Gdh3-FLAG, and ⌬gdh1⌬gdh3/YCpP GDH3 -Gdh1-FLAG) were grown in SCD and samples were taken after 12, 24, 48, and 72 h of culture. Cell extracts were immunoblotted and probed with anti-FLAG and anti-tubulin antibodies.

Gdh1, but Not Gdh3, Is Subjected to Degradation in Stationary Phase Cells in Which the Lys-426 Residue Plays an Essential
Role-Interestingly, the results of the NADP-GDH activity assays (Table 4) and immunoblotting (Fig. 6) obtained from the stationary phase cells exhibited a significant discrepancy with the results of the ␤-gal reporter assay (Fig. 5) in that only negligible levels of Gdh1 protein and its corresponding NADP-GDH activity were detected despite relatively high levels of GDH1 transcription. To further investigate the cause for such discrepancy, we monitored the activity and protein levels of NADP-GDH in ⌬gdh1⌬gdh3 cells carrying either of the promoter-swapped derivatives of GDH1 or GDH3. Although the levels of NADP-GDH activity in the YCpP GDH1 -Gdh3 transformants were almost the same as those in the WT cells regardless of growth stage, the NADP-GDH activity in cells carrying YCpP GDH3 -Gdh1 was negligible (Table 4). Accordingly, no signal for the Gdh1-FLAG protein was detected in ⌬gdh1⌬gdh3/ YCpP GDH3 -Gdh1-FLAG cells throughout all growth stages, whereas strong signals for the FLAG-tagged Gdh3-FLAG protein was observed in ⌬gdh1⌬gdh3/YCpP GDH1 -Gdh3-FLAG cells in all growth stages (Fig. 6B). In agreement with these results, both the late logarithmic (24 h) and stationary (48 h) cells of the ⌬gdh1⌬gdh3/YCpP GDH1 -Gdh3 transformant had levels of survival after a 1-h exposure to 10 mM H 2 O 2 similar to those of WT cells, whereas the survival rates of ⌬gdh1⌬gdh3/ YCpP GDH3 -Gdh1 cells showed only basal levels of resistance against oxidative stress (Fig. 7). This result indicates that ectopic expression of gdh1 P ::gdh3 ORF confers consistent resistance to stress-induced apoptosis in ⌬gdh1⌬gdh3 cells throughout all growth stages, whereas the ectopic expression of gdh3 P ::gdh1 ORF cannot protect the cells from stress-induced damage. Taken together, these results suggest that the Gdh3 protein is stable throughout all growth stages, whereas Gdh1 is subjected to stationary phase-specific degradation.

DISCUSSION
S. cerevisiae is the first microorganism described in which the NADP-GDH activity is encoded by two genes (16). It has been claimed that coordinated regulation of the two NADP-GDH isoenzymes enables a balanced utilization of ␣-ketoglutarate for glutamate synthesis during diauxic growth and eventually improves the efficiency of glutamate biosynthesis (19). However, the physiological significance of this apparent redundancy has not yet been fully addressed. In the present study, we attempted to determine the differential role of the two isoforms of NADP-GDH, Gdh1 and Gdh3, in the resistance to stress-induced apoptosis and chronological aging.
The initial clue as to the involvement of Gdh3 in the protection against stress-induced apoptosis was based on the observation that Gdh3-null cells, but not Gdh1-or Gdh3-null cells, were hypersensitive to oxidative and thermal stress compared with WT cells (Fig. 1A). We observed several typical morphological and cytological hallmarks of apoptosis, such as ROS accumulation, nuclear and DNA fragmentation, and phosphatidylserine translocation, in stationary phase Gdh3-null cells following oxidative stress (Fig. 2). A similar phenomenon was observed in Gdh3-null cells exposed to thermal stress (data not shown). In addition, impairment of GDH3 resulted in a higher susceptibility to chronological aging-induced cell death followed by regrowth of a subpopulation that consisted of cells better adapted to prolonged culture conditions (Fig. 1B).
Cells should be equipped with an efficient ROS-scavenging enzymatic system to protect against oxidative damage. Superoxide dismutase catalyzes the destruction of superoxide free radicals to oxygen and H 2 O 2 (38), which is in turn reduced to H 2 O by catalase or peroxidase. The principal enzyme for H 2 O 2 detoxification is generally considered to be glutathione peroxidase (GPx), which requires GSH as a reducing power (39), rather than catalase, because catalase has a much lower affinity than GPx for H 2 O 2 (40). Thus, GSH is an essential metabolite for stress resistance. When grown in SCD medium, Gdh3-null cells were subject to GSH depletion, which was relieved by an exogenous supply of GSH or glutamate (Table 3). Furthermore, the hypersusceptibility of the stationary cells of Gdh3-null strains to thermal and oxidative stress-induced apoptosis, which is mediated by ROS, was suppressed by exogenous GSH or glutamate (Fig. 3) or by deletion of GDH2 (Fig. 4). Thus, Gdh3 plays a pivotal role in preventing the stress-induced ROS accumulation and subsequent apoptotic events by supplying glutamate, one of the precursors for GSH biosynthesis, in stationary phase cells. It is also suggested that the major form of the ROS accumulated in the stationary phase Gdh3-null cells exposed to oxidative and thermal stress is H 2 O 2 , the substrate of GPx. In a human B-lymphoma cell line, decreased GSH alone can act as a potent early activator of apoptotic signaling. Increased ROS production following mitochondrial GSH depletion irreversibly commits cells to apoptosis (41). In addition, both glutamate and glutamine have a significant role not only in the resistance of cells to apoptosis but also in promoting cell proliferation (42). Glutamate can rescue a breast carcinoma cell line from apoptotic cell death (43). Taken together, in both yeast and mammalian cells, glutamate is required for the first committed step in the synthesis of GSH and acts as a suppressor of stress-induced apoptosis. Thus, the present results indicate that yeast strains lacking NADP-GDH could be used as a model system for studying the mechanisms of apoptotic or proliferative defects related to glutamate metabolism in mammals.
Despite the high level of sequence homology between Gdh1 and Gdh3, the transcription patterns of GDH1 and GDH3 are significantly different from each other. Our data from the ␤-gal reporter assays indicate that GDH3 transcription occurs mainly during the stationary phase, whereas GDH1 is transcribed consistently throughout all growth periods (Fig. 5). Accordingly, GDH3 transcription is strongly repressed by glucose and is highly induced under respiratory conditions and during the stationary phase (36,37,44). This implies that Gdh3 may play a crucial role in sustaining oxidative phosphorylation. On the other hand, GDH1 transcription is controlled by transcriptional activators exclusive of either nitrogen (Gln3 and Gcn4) or carbon metabolism (HAP complex) and occurs independently of growth stage and glucose repression (37).
Although consistently high levels of GDH1 transcription were observed regardless of the growth stage (Fig. 5), negligible levels of Gdh1 protein (Fig. 6A) and its NADP-GDH activity (Table 4) were detected during the stationary phase, suggesting that Gdh1 is subjected to stationary phase-specific degradation. The activity and protein levels of the NADP-GDH in ⌬gdh1⌬gdh3 cells carrying ectopic copies of the promoterswapped derivatives of GDH1 or GDH3 were in agreement with our hypothesis: whereas the NADP-GDH activity of Gdh3 (YCpP GDH1 -Gdh3) and the protein level of Gdh3-FLAG (YCpP GDH1 -Gdh3-FLAG) were consistently high, the NADP-GDH activity of Gdh1 (YCpP GDH3 -Gdh1) and Gdh1-FLAG (YCpP GDH3 -Gdh1-FLAG) protein levels were trivial throughout all growth periods (Table 4 and Fig. 6B). In further support, a previous study reported that transfer of S. cerevisiae cultures to medium deficient in a readily utilizable carbon source results in proteolysis of NADP-GDH. However, it has not been determined which of the two NADP-GDH, Gdh1 or Gdh3, is subjected to such degradation (20).
We asked how it is that only Gdh1, but not Gdh3, is subjected to stationary phase-specific degradation despite the extremely high degree of sequence homology between the two enzymes. Thus, we evaluated the importance of degradation of the lysinerich Box420 region of Gdh1 (Box420 Gdh1 ) that contains a sequence distinct from that of the corresponding Box420 region of Gdh3 (Box420 Gdh3 ) (Fig. 8A). Substituting alanine for the lysine residues in Box420 Gdh1 and tracking the stability of the mutant derivatives of Gdh1 revealed that only one lysine residue (Lys-426) of the four lysine residues concentrated in Box420 Gdh1 was necessary for the stationary phase-specific degradation of Gdh1 (Fig. 8). A previous study identified Lys-325 and Lys-371 of Gdh1 as ubiquitinated residues through a proteomics approach to enrich, recover, and identify ubiquitin conjugates from S. cerevisiae lysates (45). Although almost all of the ubiquitin-modified lysine residues are exposed at the surface of the molecule and are readily accessible from the outside, Lys-371 is located in a hydrophobic stretch, PPKAA, and is buried inside the protein (46). Thus, this position may be involved in the degradation of misfolded Gdh1 molecules. However, Gdh1 and Gdh3 share a striking homology throughout their amino acid sequences, except for their Box420 regions. The sequences surrounding Lys-325 and Lys-371 in Gdh1 are almost identical to those centered at the corresponding lysine residues, Lys-326 and Lys-372, in Gdh3. Thus neither of the lysine residues is responsible for the stationary phasespecific degradation of Gdh1. The significance of the ubiquitination of Lys-325 and Lys-371 requires further investigation. Future studies will focus on determining whether the degradation of Gdh1 is mediated by the ubiquitin-proteosome pathway in which Lys-426 provides a specific binding site for ubiquitin.
During the stationary phase, yeast cells acquire a variety of features, including a dramatic reduction in the overall rate of growth and protein synthesis, accumulation of the storage carbohydrate glycogen, and increased resistance to a variety of environmental stresses such as oxidative stress and heat shock (47,48). Specifically, the rate of protein synthesis drops ϳ300fold upon entry into stationary phase (48,49), which is an essential characteristic for stationary phase survival (50). Thus, it seems that a variety of amino acids become unnecessary for protein synthesis in stationary phase cells. In parallel, cells protect themselves from increasing environmental stress-induced accumulation of ROS during stationary phase. Among the ROS-scavenging systems, the GSH system, which consists of GSH, GPx, and glutathione reductase, is probably the most important intracellular defense mechanism. GPx catalyzes the reduction of H 2 O 2 and oxidizes GSH to GSSG. GSSG is then reduced back to GSH by glutathione reductase. Hence, the ability of the cell to reduce GSSG or synthesize GSH from glutamate is the key to how effectively the cell can eliminate ROSmediated cell damage (51). In our previous study, we showed that the hypersusceptibility of yeast cells lacking Cit1 to stressinduced apoptosis mediated by ROS does not result from the depletion of reducing power required for glutathione reductase reaction but, instead, is due to an insufficient supply of glutamate, a precursor of GSH biosynthesis (31). Therefore, yeast cells in stationary phase require glutamate for their GSH supply rather than for protein synthesis. The two isofunctional NADP-GDH of S. cerevisiae differ in allosteric properties and rates of ␣-ketoglutarate utilization. Specifically, Gdh1 exhibits a 3-fold higher rate of ␣-ketoglutarate utilization than does Gdh3 (19). Thus, Gdh1 is more suitable for functioning as a major glutamate-producing enzyme during the exponential phase in which substantial amounts of amino acids, including glutamate, are necessary for protein synthesis. On the other hand, Gdh3 seems to be more suitable during the stationary phase in which glutamate is mainly required for GSH biosynthesis. Therefore, it may be more beneficial to the cells to substitute Gdh1 with Gdh3 through the stationary phase-specific expression of GDH3 and simultaneous degradation of Gdh1 after exiting from the exponential growth phase.