Expression of a Glutamate Decarboxylase Homologue Is Required for Normal Oxidative Stress Tolerance in Saccharomyces cerevisiae *

The action of γ-aminobutyrate (GABA) as an intercellular signaling molecule has been intensively studied, but the role of this amino acid metabolite in intracellular metabolism is poorly understood. In this work, we identify a Saccharomyces cerevisiae homologue of the GABA-producing enzyme glutamate decarboxylase (GAD) that is required for normal oxidative stress tolerance. A high copy number plasmid bearing the glutamate decarboxylase gene (GAD1) increases resistance to two different oxidants, H2O2 and diamide, in cells that contain an intact glutamate catabolic pathway. Structural similarity of the S. cerevisiae GAD to previously studied plant enzymes was demonstrated by the cross-reaction of the yeast enzyme to a antiserum directed against the plant GAD. The yeast GAD also bound to calmodulin as did the plant enzyme, suggesting a conservation of calcium regulation of this protein. Loss of either gene encoding the downstream steps in the conversion of glutamate to succinate reduced oxidative stress tolerance in normal cells and was epistatic to high copy number GAD1. The gene encoding succinate semialdehyde dehydrogenase (UGA5) was identified and found to be induced by H2O2 exposure. Together, these data strongly suggest that increases in activity of the glutamate catabolic pathway can act to buffer redox changes in the cell.

␥-Aminobutyric acid (GABA) 1 is a metabolite of glutamate that is a major inhibitory neurotransmitter in animals (see Ref. 1 for a review). GABA is generated by the decarboxylation of L-glutamate by the enzyme glutamate decarboxylase (GAD) and has been implicated in hormone release from endocrine cells of several target tissues in mammals (2). Biosynthesis and secretion of GABA is a critical step in assuring normal neural function, and defects in metabolism of this glutamate derivative have been linked with clinical manifestations including epilepsy and Parkinson's disease (3,4).
Regulation of GABA levels is a consequence of the balance between GABA synthesis, catalyzed by GAD, and GABA degradation, catalyzed by the concerted action of the enzymes GABA transaminase and succinate semialdehyde dehydrogenase (SSADH) (reviewed in Ref. 5). Human genetic diseases have been linked to loss of each enzyme in this pathway, underscoring its importance in normal physiology (6). The specific mechanisms that regulate the activity of these GABA metabolic enzymes are not well described except for plant GAD (7). This GAD enzyme has been found to bind the Ca 2ϩ regulatory protein calmodulin (CaM), a binding step that acts to elevate the specific activity of the plant GAD (8), perhaps by inducing dimerization (9). This tight regulation is important in plants, as loss of CaM-regulated GAD activity elicits developmental abnormalities in plants (10).
Although the role of GAD-produced GABA as an intercellular signal is well established, relatively little is known of the intracellular function of this glutamate metabolite. In this work, we present initial characterization of the GAD/GABA metabolic pathway in the yeast Saccharomyces cerevisiae. The use of GABA as a nitrogen source in S. cerevisiae has been well studied, especially with regards to the transcriptional control of the GABA transaminase and GABA transporter loci (11)(12)(13). UGA1 encodes the GABA transaminase (14), whereas UGA4 produces the GABA transporter (15). Genetic evidence argues that UGA2 may encode the enzyme SSADH (16), but definitive data are lacking to substantiate this suggestion.
Here we provide evidence that the YMR250w locus is the gene that encodes the S. cerevisiae glutamate decarboxylase. Data are also presented that YBR006w encodes the SSADH enzyme in this organism. Importantly, we find that the S. cerevisiae L-glutamate metabolism pathway is involved in oxidant tolerance. These experiments provide important evidence that, in addition to the roles of GABA as a neurotransmitter and endocrine regulator, glutamate/GABA metabolism is a key contributor to the ability of cells to tolerate oxidative insult.
Isolation of the GAD1 Gene-A YEp24-based genomic library (21) was introduced into SEY6210 by a high efficiency transformation method (22). Ura ϩ transformants were selected on minimal medium and tested for oxidant hyper-resistance by replica plating onto YPD medium containing diamide. Ura ϩ and diamide-resistant colonies were then streaked onto SC media containing a range of diamide concentrations. A plasmid conferring increased levels of resistance to diamide was recovered and designated HRD12. This plasmid was re-introduced into SEY6210, resulting in increased diamide and H 2 O 2 tolerance.
Plasmids-HRD12 is 9.8 kb of yeast genomic DNA cloned as a Sau3AI fragment into the BamHI site of YEp24. The YEp352-GAD1 plasmid was constructed by inserting a 3.6-kb Asp718 fragment containing the GAD1 gene into Asp718-digested YEp352. A 3.0-kb SmaI-SalI fragment was inserted into YEp351 to form YEp351-GAD1. The Asp718 fragment from YEp352-GAD1 was moved into pBluescript KSII ϩ to form pBS-GAD1. The coding sequence of GAD1 was PCRamplified (primers: CCG AGA TCT ATG TTA CAC AGG CAC GGT TC and CCG GTC GAC TCA ACA TGT TCC TCT ATA GT) and cloned into pTrcHis2-TOPO (Invitrogen Corp., Carlsbad, CA) generating the plasmid pBG1. The GAD1 coding sequence was then moved into BamHI/ SalI-digested pGEX-KG (23) as a BglII/SalI fragment (pBG2) to form the GST-GAD1 fusion. The plasmid pUGA4-lacZ was constructed by PCR. A 579-base pair fragment of the UGA4 promoter and translation start signal was produced with an upstream primer (GCG AAT TCT  TTG GGA TCT ATT TTT CTC TTT AG) and a downstream primer  (CGG GAT CCT TAT TCT CGT TTT TGC TTG A) corresponding to positions Ϫ550 to ϩ29. The resulting product was cloned into the lacZ fusion plasmid pSEYC102 (24) as an EcoRI-BamHI fragment. All PCR products were sequenced to ensure that no errors occurred during amplification.
Gene Disruption-A SmaI-SalI fragment from pUC19-HIS3 was inserted into SmaI-SalI-digested pBS-GAD1 to produce a gene disruption of GAD1. The resulting plasmid was designated p⌬GAD1 and replaced the entire GAD1 coding sequence with the HIS3 gene. The p⌬GAD1 plasmid was digested with AgeI and XbaI prior to transformation into SEY6210. His ϩ transformants were selected and purified. Genomic DNA was prepared (25), and the correct integration event was verified by PCR and Southern blotting. A representative gad1⌬ mutant was used in these studies and designated YSC11. Disruption of the UGA1 gene was produced by a PCR-based method using the primers TCT ATT TGT GAA CAA TAC TAC CCA GAA GAG CCA ACC AAA CCA ACT GTT AAT TGT ACT GAG AGT GCA CC and TCT ATT TGT GAA CAA TAC TAC CCA GAA GAG CCA ACC AAA CCA ACT GTT AAT TGT ACT GAG AGT GCA CCA T. These primers were used to amplify via the underlined sequences the HIS3 gene from plasmid pRS303 (26). The resulting PCR product replaces the UGA1 coding sequence from ϩ53 to ϩ1370 with the HIS3 gene. His ϩ transformants were selected and purified. As described above, the correct integration event was verified by PCR using a primer upstream of the UGA1 deletion (TTC GCG CTA TCT CGA TTT CTA CCT A) and in the middle of HIS3 (TTC TTC GAA GAA ATC ACAT TAC TTT ATA TA). The UGA5-lacZ fusion gene was obtained from the Yale Genome Analysis Center in plasmid form and used to transform the uga1⌬ strain to analyze expression of UGA5 in response to elevation of GAD1 copy number. Disruption of the UGA5 gene was accomplished as above using the primers: ACT TTG AGT AAG TAT TCT AAA CCA ACT CTA AAC GAC CCT AAT TTA TTC AGT TGT ACT GAG AGT GCA CCA T and TTA AGT GTT GAC ATT TTT AGA AAA GAC ATA TGC TGC TAA ACC AAA CTC AGG GTA TTT CAC ACC GCA TA. The resulting PCR product replaces the UGA1 coding sequence from ϩ53 to ϩ1259 with HIS3. His ϩ transformants were selected and purified. The correct integration event was verified from genomic DNA by PCR using a primer upstream of the UGA5 deletion (ACC CAC CGG AGA GGG CAA AGG TAA A) and in the middle of HIS3 (TTC TTC GAA GAA ATC ACA TTA CTT TAT ATA).
Characterization of Yeast GAD-Glutathione S-Transferase (GST) Fusion Protein-E. coli transformants expressing the GST-Gad1p fusion protein were grown overnight, ϳ16 h, at 37°C in LB with 100 g/ml ampicillin. Overnight cultures were diluted 1:100 in fresh LB with 100 g/ml ampicillin and incubated at 25°C with shaking at 200 rpm. After 4 h, isopropyl-1-thio-␤-D-galactopyranoside was added to a final concentration of 100 M, to induce expression of the recombinant protein, incubation continued 16 h. Bacterial cells were collected by centrifugation at 6000 ϫ g for 10 min. Pellets were immediately frozen in liquid N 2 and stored at Ϫ80°C. As a positive control, rGAD1 from Arabidopsis thaliana was prepared as described previously by Turano and Fang (27), and, as negative control, a control GST was prepared. Pellets containing cells were resuspended in 5-7 ml of extraction buffer (50 mM Tris-HCl, pH 7.5, 0.2 mM EDTA, 2.5% (v/v) glycerol, 2 mM DTT, 0.05 mM pyridoxal 5Ј-phosphate, 0.05% (v/v) Triton, 10 M leupeptin, 20 g/ml lysozyme, and 1 mM PMSF. The cells were lysed by sonication with 12 2-s bursts at 150 watts. Cellular debris was removed by centrifugation at 12,000 rpm for 20 min. The supernatant was filtered through a 0.45-m filter. CaCl 2 , DTT, and PMSF were added to final concentrations of 1, 2, and 1 mM, respectively. The supernatant was loaded on to a CaM-Sepharose column (Amersham Pharmacia Biotech, ϳ1-ml bed volume), pre-equilibrated in binding buffer (25 mM Tris-HCl, pH 7.5, 150 mM NaCl, 2.5% (v/v) glycerol, 1 mM CaCl 2 , 2 mM DTT, 10 M leupeptin, and 1 mM PMSF). The column was washed with 20 bed volumes of wash buffer minus calcium (25 mM Tris-HCl, pH 7.5, 150 mM NaCl, 2.5% (v/v) glycerol, 2 mM DTT, 10 M leupeptin, and 1 mM PMSF). CaM-binding proteins were eluted with (50 mM Tris-HCl, pH 7.5, 2 mM EDTA, 150 mM NaCl, 2.5% (v/v) glycerol, and 1 mM PMSF). Samples were immediately assayed as described above and analyzed by SDS-PAGE and immunoblot analysis. Protein concentrations, estimates of GAD activity, SDS-PAGE, silver staining, and immunoblot analysis were conducted as described by Turano and Fang (27), except that immunoreactive peptides were detected using a chemiluminescent detection system (SuperSignal West Dura, Pierce).

Cloning of a Diamide and H 2 O 2
Resistance-conferring Locus-As part of our ongoing studies on oxidative stress in S. cerevisiae, we carried out a screen of a high copy number plasmid library to identify genes that were capable of increasing resistance to the oxidants diamide and H 2 O 2 . Diamide causes oxidative stress by shifting the majority of intracellular glutathione from the reduced form to the oxidized form (28), whereas H 2 O 2 elicits oxidative damage to proteins, lipids and nucleic acids (29).
The wild-type yeast strain SEY6210 was transformed with a S. cerevisiae YEp24 genomic DNA library. Approximately 20,000 Ura ϩ transformants were recovered and tested for diamide tolerance by replica plating to YPD plates containing various concentrations of diamide. Colonies that survived increased concentrations of diamide were purified and plasmids recovered. The recovered plasmids were then reintroduced into SEY6210 to further assess their diamide tolerance. Plasmids that were able to reproducibly confer resistance to diamide were also tested for resistance to H 2 O 2 . We focused our studies on a plasmid that conferred resistance to both diamide and H 2 O 2 that was designated HRD12, for hyper-resistance to diamide. The genomic DNA insert carried by the HRD12 plasmid was ϳ9.8 kb (data not shown).
To localize the gene responsible for resistance to diamide and H 2 O 2 from HRD12, we first sequenced both ends of the genomic DNA insert. We then performed a BLAST search of the S. cerevisiae genomic sequence (30) using these DNA sequences to determine the bounds of the segment of cloned genomic DNA and found that this segment of genomic DNA originated from chromosome XIII. Further computer analysis detected the presence of several open reading frames (ORFs) present in this segment of genomic DNA. One of these ORFs (YMR250w) was subcloned as a 3.6-kb Asp718 fragment into YEp352 (31), and the resulting construct was designated YEp352-GAD1. This construct was then transformed into wild-type cells and found to confer the same level of tolerance to diamide and H 2 O 2 as HRD12 (data not shown). A BLAST search of the GenBank data base with the predicted amino acid sequence of YMR250w indicated the presence of sequence similarity with glutamate decarboxylase proteins. The two most closely related sequences were the Petunia hybrida GAD with 39% and A. thaliana GAD with 38% identity, respectively (Fig. 1). Key lysine and histidine residues were conserved in the plant and yeast sequences, again supporting the identification of YMR250w as encoding the S. cerevisiae glutamate decarboxylase enzyme, and we have designated this locus as GAD1.
Biochemical Characterization of Gad1p-To examine the biochemical properties of S. cerevisiae Gad1p, the open reading frame from YMR250w was expressed in E. coli as a fusion protein. The recombinant protein was purified by affinity chromatography (Fig. 2) using a CaM-Sepharose column. CaMbinding proteins were eluted with 2 mM EGTA. Purified CaMbinding proteins were analyzed by silver staining SDSpolyacrylamide gels and by immunoblot analysis. A 91-kDa protein was found to be retained by the CaM-Sepharose column from extracts expressing the GST-Gad1p fusion protein. This result agreed well with the combined molecular masses of GST (26.5 kDa) and the estimated molecular mass of the deduced amino acid sequence from the yeast Gad1p (66 kDa). To demonstrate that GST did not bind to CaM, protein extracts from E. coli overexpressing GST were subjected to CaM affinity purification and no proteins were found to be bound by this resin. To confirm the identity of the CaM-binding peptide, affinity-purified peptides were subjected to immunoblot analysis with antiserum to Petunia GAD (7). The 91-kDa fusion peptide crossreacted with the Petunia GAD antiserum. Western blot analysis of crude extracts expressing GST or the GST-Gad1p fusion also demonstrated that the Petunia antiserum would specifically detect the yeast protein (data not shown). Collectively, these results confirm that the recombinant fusion protein was expressed in E. coli and that the yeast Gad1p was responsible for CaM binding of the fusion peptide. The purified recombinant fusion protein was assayed for GAD activity, and no activity was observed. Several attempts, including excision of the Gad1p from GST fusion protein or expression of the yeast Gad1p as an unfused protein, were unsuccessful. Possible reasons for the failure to detect enzymatic activity of this recombinant protein include unique properties of the yeast enzyme and possible differences in cofactor requirements (see below).
Normal H 2 O 2 Resistance Requires the Presence of the GAD1 Locus-To examine the physiological role of Gad1p, a strain lacking the GAD1 gene was constructed. A fragment from Ϫ530 to ϩ2482 base pairs, relative to the GAD1 translation start, was replaced with the HIS3 gene. The resulting plasmid, p⌬GAD1, was digested with AgeI and XbaI before being transformed into the wild-type strain SEY6210. His ϩ transformants Protein extracts from bacterial cells expressing either a fusion protein between GST and Gad1p or GST alone were subjected to chromatography on calmodulin-Sepharose beads. Specifically bound material was eluted with EGTA and analyzed by SDS-PAGE, followed by visualization of total eluted proteins using silver stain or Western blotting using an anti-Petunia GAD antiserum. Bacterially expressed A. thaliana GAD (rGad1p) was used as a positive control for immunological crossreaction. Molecular mass standards are indicated on the right.
were recovered and confirmed to contain the desired gene disruption allele. The resulting S. cerevisiae strain was designated YSC11.
Growth of gad1⌬ cells on rich medium was indistinguishable from wild-type cells. Differences in growth do arise, however, when growth is compared on medium containing oxidants. We assayed the growth of wild-type and gad1⌬ strains on YPD plates containing either diamide or H 2 O 2 . Unlike wild-type cells, the gad1⌬ strain was unable to grow on plates containing 3 mM H 2 O 2 (Fig. 3) or on increased diamide concentrations (data not shown). These findings indicate that GAD1 is required for normal resistance to oxidants.
Normal Oxidant Tolerance Requires a Functional Glutamate Metabolism Pathway in Yeast-Our finding that oxidant resistance of S. cerevisiae is proportional to the gene dosage of GAD1 could be explained by the action of this enzyme to form GABA, which in turn acts to signal a stress response. Alternatively, increased GAD production could simply elevate the catabolism of glutamate. Increased glutamate catabolism is believed to allow bacterial cells to tolerate low pH conditions in the medium through the removal of acid by elimination of glutamate (32). Importantly, elevated flux through the glutamate catabolic pathway defined by GAD, GABA transaminase, and SSADH could also elevate production of NADPH, an important antioxidant (33)(34)(35). NADPH can be produced by the action of SSADH (16). If this last model is correct, then the entire glutamate catabolic pathway would be predicted to be required for the observed increase in oxidative stress tolerance provided by enhanced production of Gad1p.
To test this idea, we wanted to construct strains that lacked the S. cerevisiae GABA transaminase and SSADH loci. UGA1 encodes the GABA transaminase (14), but the SSADH-encoding gene in this organism has not yet been identified. A mutation called uga2 was isolated that eliminated SSADH activity, but the gene corresponding to this mutation was never isolated (16). A BLAST analysis of the protein data base from S. cerevisiae using the E. coli SSADH protein indicated that the YBR006w ORF shared 52% identity with the bacterial protein (Fig. 4). The YBR006w-encoded protein also exhibited 47% identity with human SSDH. Based on this high degree of sequence identity and other properties (see below) of YBR006w, we designated this locus UGA5 and believe it encodes the S. cerevisiae SSADH.
Deletion mutant strains were constructed that lacked either UGA1 or UGA5. Primers were generated to amplify a PCR product that replaced nucleotides ϩ53 to ϩ1370 of the UGA1 coding region with the HIS3 gene employing the same strategy used above to disrupt GAD1. This same method was used to substitute nucleotides ϩ53 to ϩ1259 of UGA5 with the HIS3 gene. The uga1-⌬1::HIS3-containing strain and the uga5-⌬1::HIS3-containing strain were designated YSC29 and YBB1, respectively.
The ability to utilize GABA as the sole nitrogen source in the medium defines the UGA series of genes in S. cerevisiae (16). To provide support for our placement of UGA5 in the glutamate catabolic pathway, we tested the ability of the uga1⌬ and uga5⌬ strains to grow with either ammonium sulfate or GABA as the nitrogen source in the medium. These strains, along with the isogenic wild-type, were pre-grown in synthetic complete medium and then diluted into minimal medium with either 0.1% ammonium sulfate or GABA as the nitrogen source. Growth was monitored over 12 h to assess the ability of each strain to grow. Strains grown in medium with ammonium sulfate as the nitrogen source were able to grow normally (Fig.  5). However, in GABA media, only wild-type cells were able to grow. Wild-type cells reached an optical density of nearly 2, whereas uga1⌬ and uga5⌬ cells only grew to an absorbance value of 0.16. This result is consistent with our identification of the UGA5 locus as encoding S. cerevisiae SSADH since uga1⌬ and uga5⌬ mutant strains have similar defects during growth on GABA-containing medium.
We next compared the oxidative stress resistance phenotypes of cells lacking GAD1, UGA1, or UGA5. These strains were also transformed with a high copy number plasmid that contains the wild-type GAD1 gene. In a strain containing UGA1 and UGA5, elevation of GAD1 gene dosage increases oxidative stress tolerance (Fig. 3). To determine whether this increased resistance required the presence of genes down- FIG. 3. The glutamate catabolic pathway is required for normal oxidative stress tolerance. A, the enzymatic steps for conversion of glutamate to succinate in S. cerevisiae is shown with the corresponding structural genes listed below. B, strains with either high (2-m GAD1) or low (GAD1) dosages of the GAD1 gene along with the indicated UGA1 and UGA5 alleles were analyzed for their ability to tolerate H 2 O 2 by spot test assay. Equal numbers of cells of each genotype were placed on rich medium (YPD) or the same medium containing 3 mM H 2 O 2 and allowed to grow at 30°C. stream in the glutamate catabolic pathway, we assayed the ability of these strains to tolerate H 2 O 2 (Fig. 3). Strains were grown to mid-log phase and assayed for tolerance to varying concentrations of H 2 O 2 .
Loss of either UGA1 or UGA5 reduced H 2 O 2 resistance of cells with single or multiple copies of the GAD1 locus. Thus, the oxidative stress phenotypes of uga1⌬ and uga5⌬ mutant strains were epistatic to that conferred by high copy number GAD1, again supporting both our placement of UGA5 downstream from GAD1 in the glutamate catabolic pathway and the requirement for this pathway in the effect of Gad1p on oxidative stress tolerance.
UGA5 Expression Is Induced by Oxidants and GABA-The UGA1 gene and the GABA transporter-encoding UGA4 locus are expressed at a 30 -100-fold higher levels when GABA is used as the nitrogen source rather than ammonium sulfate (11)(12)(13). This induction involves binding of the transcriptional regulatory protein Gln3p to the promoter elements of these GABA-regulated loci (36). Gln3p recognizes elements centered around a GATAA sequence, a binding site found upstream of genes involved in metabolism of poor nitrogen sources like GABA (12). There are two close relatives of the GATAA sequence in the UGA5 5Ј-flanking region, suggesting that this gene, like UGA1 and UGA4, might be transcriptionally regulated by the nitrogen source in the medium. To examine this possibility, we obtained a lacZ fusion constructed by transposon insertion into UGA5 from the Yale Genome Analysis Center (37). This transposon insertion produced an in-frame fusion between lacZ and UGA5 at codon 67 and was carried in a diploid strain to allow the presence of a functional copy of the gene as well as the UGA5-lacZ fusion gene. This strain was grown in medium using either ammonium sulfate or GABA as the nitrogen source and ␤-galactosidase activity determined.
UGA5-lacZ expression increased by ϳ240-fold when GABA replaced ammonium sulfate as the nitrogen source (Fig. 6). This large induction of gene expression is similar to those reported previously for UGA1 and UGA4, other GABA metabolic proteins (11)(12)(13). Since UGA5 is required for normal H 2 O 2 tolerance (Fig. 3), expression of the UGA5-lacZ fusion was also examined in the presence of this oxidant. A nearly 3-fold higher level of UGA5-dependent ␤-galactosidase activity was found when this strain was exposed to 1 mM H 2 O 2 , supporting the idea that UGA5 expression may be increased to elevate flux through the glutamate catabolic pathway and increase NADPH pools in the cell.
Genetic Evidence Supporting the Role of Gad1p as the S. cerevisiae Glutamate Decarboxylase Protein-Since we believe that GAD1 encodes the S. cerevisiae glutamate decarboxylase protein, elevation in the copy number of this gene would be expected to increase the ability of a cell to produce GABA. To test the ability of changes in GAD1 gene dosage to influence intracellular GABA level, we utilized an indirect assay for GABA production. UGA4 encodes the GABA transporter protein and is highly induced upon an increase in the level of GABA in the medium (15). We constructed an UGA4-lacZ gene fusion to facilitate measurement of UGA4 expression and introduced this fusion gene into several different strains. These strains were generated to provide varying levels of intracellular GABA by either allowing (UGA1) or blocking (uga1⌬) consumption of GABA. We anticipated that GABA levels should increase as the UGA1 gene is removed and GAD1 gene dosage level increases as GABA is expected to be produced at a higher rate but can no longer be broken down. Strains carrying the UGA4-lacZ fusion gene and varying in their copy number of GAD1 and UGA1 were grown in medium containing ammonium sulfate or glutamate as the nitrogen source and then assayed for expression of UGA4 (Table I).
When GABA breakdown was blocked by the presence of a uga1⌬ allele, UGA4-lacZ expression was found to increase by nearly 5-fold when the GAD1 copy number was elevated in the presence of glutamate as a nitrogen source. If ammonium sulfate replaced glutamate as the nitrogen source, UGA4-dependent ␤-galactosidase levels were much lower and only increased from 0.4 units/OD to 1.3 units/OD when GAD1 was present on a high-copy-number plasmid. Although UGA4 expression was induced when glutamate was used as the nitrogen source rather than ammonium sulfate in a UGA1 genetic background, the extent of this induction was not significantly affected by elevation of GAD1 copy number, consistent with the intact glutamate catabolic pathway limiting the accumulation of GABA. We also assayed the response of the UGA5-lacZ fusion gene described above carried in uga1⌬ cells to varying the copy number of GAD1. A uga1⌬, UGA5-lacZ strain produced 196 Ϯ 42 units/OD ␤-galactosidase in the presence of high copy GAD1, FIG. 4. Uga5p shares strong sequence similarity with plant and bacterial SSADH enzymes. A computer alignment performed as described above was carried out to analyze the extent of sequence conservation between S. cerevisiae Uga5p (YBR006w gene product), A. thaliana SSADH, and E. coli GabD. Identical residues are indicated as boxes. Note the N-terminal extension on the plant protein that is believed to serve a as mitochondrial targeting signal. which was reduced to 12 Ϯ 2.5 units when GAD1 was maintained at chromosomal levels. These data support identification of GAD1 as encoding the S. cerevisiae glutamate decarboxylase enzyme and indicate that coordinate induction of UGA1 and UGA5 is likely to occur when Gad1p activity increases with accompanying GABA accumulation. DISCUSSION The participation of glutamate decarboxylase proteins in intercellular signaling pathways has received a great deal of attention due to the important role of GABA in neurotransmission (1) and in plant development (10). Our work provides insight into the intracellular involvement of GAD in oxidative stress tolerance. Increasing the gene dosage of the S. cerevisiae GAD1 locus produced an increased tolerance to two different oxidative agents, diamide and H 2 O 2 . This increased tolerance was strictly dependent on the presence of the intact glutamate catabolic pathway leading to the production of succinate from glutamate. Genetic elimination of either enzymatic reaction downstream from glutamate decarboxylase rendered cells hypersensitive to oxidants. Genetic diseases have been described in humans that are associated with loss of either GABA transaminase or SSADH enzymes, further underscoring the importance of maintenance of the intact glutamate catabolic pathway (6).
The identification of GAD1 as encoding S. cerevisiae glutamate decarboxylase was made on three independent criteria. First, the highest degree of sequence similarity was found between Gad1p and GAD proteins. Second, an antiserum directed against a plant GAD enzyme cross-reacted with bacterially expressed Gad1p. Finally, an increase in the gene dosage FIG. 5. Both UGA5 and UGA1 are required for use of GABA as a nitrogen source. An isogenic series of strains with the indicated UGA1 or UGA5 alleles was tested for the ability to grow in minimal medium using either ammonium sulfate or GABA as the sole nitrogen source. Strains were grown to saturation in ammonium sulfate medium and then diluted into fresh minimal medium containing the nitrogen source indicated on the right. Cultures were incubated at 30°C with shaking and the optical density at 600 nm (A 600 ) was determined each hour.
FIG. 6. UGA5 expression is regulated by nitrogen source and oxidative stress. A, a diagram of the UGA5 locus on chromosome II is shown. The potential GATAA binding sites required for Gln3p-dependent gene regulation are indicated upstream of the ATG for the UGA5 open reading frame. The approximate position for insertion of the lacZ-containing transposon is shown. B, a diploid cell carrying a single copy of the UGA5-lacZ gene fusion was grown in minimal medium with either ammonium sulfate or GABA as the sole nitrogen source. The ammonium sulfate-grown culture was split in two aliquots, and one was treated with the addition of 1 mM H 2 O 2 to induce oxidative stress. UGA5-dependent ␤-galactosidase activities were then determined for all strains. The numbers above the bars represent the average of two assays for two different transformants.

TABLE I
Elevation in GAD1 copy number induces UGA gene expression Cells containing a normal (GAD1) or high copy number (2-m GAD1) dosage of the GAD1 gene and either an intact (wild-type) or defective (uga1⌬) copy of UGA1 were grown in media using the indicated nitrogen source to mid-log phase. Each strain was also transformed with a UGA4-lacZ gene fusion carried on a low copy number plasmid. UGA4dependent ␤-galactosidase activity was determined in each case.  of GAD1 elicited an increase of the expression of a gene known to be responsive to GABA levels (UGA4). Another interesting biochemical similarity between the plant and yeast enzymes was the observation that Gad1p was able to bind to calmodulin like several plant GAD enzymes (7, 8, 27, 38 -40). This finding suggests that perhaps S. cerevisiae GAD activity will be controlled by calcium/calmodulin as are these plant enzymes and may provide an explanation for the failure to detect enzyme activity of the yeast enzyme in vitro. These experiments suggest that, although Gad1p can bind to the mammalian calmodulin protein, perhaps proper regulation of the yeast enzyme requires the authentic S. cerevisiae calmodulin, a possibility currently under investigation. Evidence is also provided that the S. cerevisiae succinate semialdehyde dehydrogenase is encoded by the YBR006w locus that we propose be designated UGA5. As mentioned above, earlier work defined a mutation called uga2 that eliminated SSDH activity, but the locus defined by this mutation was not described (16). Genetic analysis will be required to determine whether YBR006w and uga2 are allelic. Several independent pieces of evidence support identification of UGA5 as the SSDH locus in S. cerevisiae. UGA5 shares the highest degree of sequence similarity with SSDH enzymes from other organisms of any ORF in the genome, it is required for utilization of GABA as a nitrogen source and it is highly inducible when GABA is present in the medium.

UGA4-lacZ expression
SSADH is also a key participant in the ability of elevated Gad1p levels to influence oxidative stress tolerance. Since loss of either UGA1 or UGA5 is epistatic to amplification of the GAD1 locus, we argue that the entire glutamate catabolic pathway is required for the observed increase in oxidant resistance conferred by high copy number GAD1 plasmids. SSADH enzymes produce either NADH or NADPH during their oxidation of succinate semialdehyde to succinate (41). This production of reduced nicotinamide adenine dinucleotide can then be coupled to buffering redox changes that would otherwise occur in the presence of oxidants like H 2 O 2 or diamide.
The precise nature of the reduced nicotinamide adenine dinucleotide is not known in the case of S. cerevisiae SSADH. Measurements of the enzymatic properties of the yeast SSADH found that this protein had a 2.5-fold higher activity when assayed in the presence of NAD ϩ rather than NADP ϩ (16). Since the [NADPHϩNADP ϩ ]/[NADHϩNAD ϩ ] ratio is 35 in yeast (42), it seems more likely that NADP ϩ will serve as the cofactor for the SSADH reaction in this organism. NADPH is critical for maintenance of normal redox balance and is required for regeneration of reduced glutathione and thioredoxin by glutathione reductase and thioredoxin reductase, respectively (43). Previous work has shown that activity of the pentose phosphate pathway, which also leads to production of NADPH, is critical for normal oxidative stress tolerance (33,34). Ensuring proper levels of key antioxidants like glutathione and NADPH is crucial for maintenance of normal oxidative stress resistance (33,34,44). The data described here indicate that activity of the glutamate catabolic pathway is also an essential contributor to the intracellular pool of antioxidants.
The plant and human SSADH enzymes carry out their catalysis in the mitochondrion (41,45). We believe this is unlikely to be the case for the S. cerevisiae protein. As expected for a protein targeted to the mitochondrion, both the plant and human SSADH proteins possess N-terminal targeting sequences (41,45). However, the alignment of yeast, bacterial, human, and plant enzymes indicates that the yeast protein does not exhibit an N-terminal targeting extension and instead shows strong sequence similarity with the bacterial enzymes. These data suggest that the S. cerevisiae protein will be found in an extra-mitochondrial location. Localization of the yeast proteins is the focus of future work.