Importance of Glutamate 279 for the Coenzyme Binding of Human Glutamate Dehydrogenase*

Although the structure of glutamate dehydrogenase (GDH) has been reported from various sources including mammalian GDH, there are conflicting views regarding the location and mechanism of actions of the coenzyme binding. We have expanded these speculations by photoaffinity labeling and cassette mutagenesis. Photoaffinity labeling with a specific probe, [32P]nicotinamide 2-azidoadenosine dinucleotide, was used to identify the NAD+ binding site within human GDH encoded by the synthetic human GDH gene and expressed inEscherichia coli as a soluble protein. Photolabel-containing peptides generated with trypsin were isolated by immobilized boronate affinity chromatography. Photolabeling of these peptides was most effectively prevented by the presence of NAD+ during photolysis, demonstrating a selectivity of the photoprobe for the NAD+ binding site. Amino acid sequencing and compositional analysis identified Glu279 as the site of photoinsertion into human GDH, suggesting that Glu279 is located at or near the NAD+ binding site. The importance of the Glu279 residue in the binding of NAD+ was further examined by cassette mutagenesis with mutant enzymes containing Arg, Gly, Leu, Met, or Tyr at position 279. The mutagenesis at Glu279 has no effects on the expression or stability of the different mutants. The K m values for NAD+ were 10–14-fold greater for the mutant GDHs than for wild-type GDH, whereas the V maxvalues were similar for wild-type and mutant GDHs. The efficiency (k cat/K m ) of the mutant GDH was reduced up to 18-fold. The decreased efficiency of the mutants results from the increase in K m values for NAD+. In contrast to the K m values for NAD+, wild-type and mutant GDHs show similarK m values for glutamate, indicating that substitution at position 279 had no appreciable effect on the affinity of enzyme for glutamate. There were no differences in sensitivities to ADP activation and GTP inhibition between wild-type and mutant GDH, suggesting that Glu279 is not directly involved in allosteric regulation. The results with photoaffinity labeling and cassette mutagenesis studies suggest that Glu279 plays an important role for efficient binding of NAD+ to human GDH.

Glutamate dehydrogenase (GDH; 1 EC 1.4.1.3) catalyzes the reversible reaction of 2-oxoglutarate to L-glutamate using NADH or NADPH (1). There are three types of GDH that vary according to the coenzyme they use: NAD(H)-specific GDH, NADP(H)-specific GDH, and GDH with mixed specificity. The bacterial and fungal NADP ϩ -linked and vertebrate dual specificity GDHs have six identical subunits, with a subunit size between 48 kDa (Escherichia coli) (2) and 55-57 kDa (vertebrate) (1,3), whereas the NAD ϩ -linked enzymes have either four identical subunits with a size of ϳ115 kDa (Neurospora crassa) (4) or six identical subunits with a subunit size of ϳ48 kDa (Clostridium symbiosum) (5). NAD(H)-specific enzymes are believed to participate mainly in the catabolism of glutamate, but NADP(H)-specific enzymes have a mainly anabolic role. Unlike GDH from primitive organisms, mammalian GDH uses both forms of coenzyme with comparable efficacy, and the anabolic/catabolic balance is therefore tightly controlled by a complex network of allosteric regulators.
The structures of GDHs from microbial and mammalian sources have been reported previously (6 -11). The largest difference between mammalian GDH and bacterial GDH is a long antenna domain in mammalian GDH formed by the 48-amino acid insertion starting at residue 395 (10 -12). Mammalian GDH is strictly regulated by allosteric activators and inhibitors (1)(2)(3). GTP inhibits enzyme turnover over a wide range of conditions by increasing the affinity of the enzyme for the product, making product release rate-limiting under all conditions in the presence of ADP (10 -14). ADP is a potent activator decreasing product affinity (11,15,16). In contrast to vertebrate GDH, bacterial GDH is not regulated by the allosteric regulators (4,5). Therefore, it has been suggested that the antenna domain that is unique to mammalian GDH has important roles in allosteric regulation (10,11). However, it is well documented that the regulatory pattern of GDH is very complicated (1,3,(15)(16)(17)(18), and therefore it cannot be completely explained only by the existence of the antenna region. For instance, ADP is generally considered as an activator of mammalian GDH (11,18,19), but it can also inhibit GDH activity under some conditions, such as low pH (3,15). The most recent study shows that GTP binds to GDH from E. coli at an allosteric site and reverses the destabilizing effects of the coenzyme (20). This result strongly suggests that the unique 48-amino acid antenna region in mammalian GDH may not be wholly responsible for the observed regulation of ADP and GTP.
It has been a major goal to identify the substrate and regulatory binding sites of GDH. Identification of the nucleotide binding sites of a variety of proteins has been advanced by the use of nucleotide photoaffinity analogues that selectively insert into a site upon photoactivation with ultraviolet light. For instance, [ 32 P]2N 3 NAD ϩ was shown to be a valid active-site probe for several proteins (21)(22)(23)(24). GTP and ADP binding sites of bovine GDH have been reported using [ 32 P]8N 3 GTP and [ 32 P]8N 3 ADP, respectively (25)(26)(27)(28). The ATP binding site of adenylate kinase, creatine kinase, and the protein unique to cerebrospinal fluids of Alzheimer's patients has also been successfully identified using 2N 3 ATP and 8N 3 ATP (29,30). Recent studies of the x-ray structure of bovine liver GDH indicate that Glu 275 (Glu 279 in human enzyme) forms a hydrogen bond with the coenzyme (10,11). However, no information about the importance of this residue in the direct binding with coenzyme has been reported. Studies of the effects of site-directed mutagenesis at the Glu 279 site on the affinity and kinetics of coenzyme binding are necessary to obtain such information.
Recently, a 1557-bp gene that encodes human GDH has been synthesized and expressed in E. coli in our laboratory (31). Using this synthetic human GDH gene, reactive amino acid residues for catalysis (31), GTP base binding (32), and ADP base binding (33) have been identified by cassette mutagenesis. In the present work, we report identification of an NAD ϩ binding site of human GDH by a combination of cassette mutagenesis and photolabeling. For the present study, the mutant human GDHs containing Gly, Leu, Met, Arg, or Tyr at the Glu 279 site have been expressed in E. coli as a soluble protein and characterized. Our data place the NAD ϩ binding domain within a proposed catalytic cleft defined in the crystal structure of GDH (10,11,34,35). To our knowledge, this is the first report by site-directed mutagenesis showing an involvement of Glu 279 of mammalian GDH in NAD ϩ binding.

EXPERIMENTAL PROCEDURES
Materials-NADH, NAD ϩ , 2-oxoglutarate, ADP, and L-1-tosylamido-2-phenylethyl chloromethyl ketone-treated trypsin were purchased from Sigma. Human GDH gene (pHGDH) has been chemically synthesized and expressed in E. coli as a soluble protein in our laboratory as described elsewhere (31). 2N 3 NAD ϩ and [ 32 P]2N 3 NAD ϩ were synthesized using NMN adenylyltransferase according to the same method described previously (21). NMN adenylyltransferase was kindly provided by Dr. Mathias Ziegler (Institut fü r Biochemie, Freie Universitä t Berlin). Precast gels for SDS-PAGE were purchased from Novex. All other chemicals and solvents were of reagent grade or better.
Photolabeling of GDH-Photolabeling of wild-type human GDH was performed by the method of Kim and Haley (21), with a slight modification. For saturation studies, wild-type GDH (100 g) in 10 mM Tris acetate, pH 8.0, containing 12 mM glutarate was incubated with various concentrations of [ 32 P]2N 3 NAD ϩ in Eppendorf tubes for 5 min. Glutarate was included in the reaction mixture because it has been reported by equilibrium dialysis and initial rates that glutarate makes NAD ϩ bind to GDH more tightly (39 -41). For competition studies, 100 g of enzyme was incubated with various concentrations of NAD ϩ for 10 min in the same buffer before the addition of 100 M [ 32 P]2N 3 NAD ϩ and then incubated with the photoprobe for 5 min as described above. The samples were irradiated twice with a hand-held 254 nm UV lamp for 90 s at 4°C. The reaction was quenched by the addition of ice-cold trichloroacetic acid (final concentration, 7%). The reaction mixtures were kept on the ice bath for 30 min and centrifuged at 10,000 ϫ g for 15 min at 4°C. The pellets were washed and resuspended with 10 mM Tris acetate, pH 8.0. The remaining free photoprobe, if any, was further removed from the protein by exhaustive washing using Centrifree (Amicon), and 32 P incorporation into protein was determined by liquid scintillation counting.
Tryptic Digestion and Isolation of Photolabeled Peptide-To determine the site modified by [ 32 P]2N 3 NAD ϩ , 2.0 mg of wild-type GDH in 10 mM Tris acetate, pH 8.0, was incubated with 100 M [ 32 P]2N 3 NAD ϩ in the presence of 1 mM GTP, 12 mM glutarate, and 10 M EDTA for 5 min at 4°C. The mixtures were irradiated twice for 90 s. The reaction was quenched by the addition of ice-cold trichloroacetic acid (final concentration, 7%) and kept at 4°C for 30 min. The protein was precipitated by centrifugation at 10,000 ϫ g for 15 min at 4°C, and the pellet was resuspended in 75 mM NH 4 HCO 3 , pH 8.5, containing 2 M urea. The enzymes were proteolyzed by the addition of 15 g of trypsin and kept at room temperature for 3 h, after which time 15 g of trypsin was added again. After an additional 3 h at room temperature, 20 g of trypsin was added, and the digestion mixture was kept at 25°C overnight. To validate that the isolated peptide was specific for the NAD ϩ binding site and could be protected from photomodification, the GDH proteins were photolysed in the presence of 300 M NAD ϩ and proteolyzed as described above.
The tryptic digested peptides were applied to a boronate column equilibrated with 0.1 M ammonium acetate, pH 9.0. Unmodified peptides were washed with the same buffer, and photolabeled peptides were eluted with an 80-ml gradient of pH 9.0 to pH 5.0 in 0.1 M ammonium acetate. The absorbance of the fractions was measured at 220 nm, and the photoincorporation was determined by liquid scintillation counting. The fractions containing photolabeled peptides were desalted, freeze-dried, resuspended in 0.1% trifluoroacetic acid, isolated by reversed-phase HPLC using a Waters C 18 column, and sequenced by the Edman degradation method described previously (28).
Construction and Characterization of Glu 279 Mutants-Cassette mutagenesis at the Glu 279 site was performed using a synthetic human GDH gene, pHGDH (31). The plasmid DNA (5 g) was digested with NsiI and EspDI to remove the 46-bp fragment that encodes amino acids 275-290, which was replaced with five 46-bp synthetic DNA duplexes containing a substitution on both DNA strands at positions encoding Glu 279 to make E279G, E279L, E279M, E279R, and E279Y mutants. Glu 279 mutants were identified by DNA sequencing using plasmid DNA as a template. Each of these mutants has been expressed in E. coli and purified to homogeneity as described below and has had its steady-state kinetic parameters determined. The gene expression level of Glu 279 mutant proteins in the crude extracts was examined by Western blot and compared with that of wild-type GDH.
Purification and Characterization of Mutant Proteins-Fresh overnight cultures of DE3/pHGDH were used to inoculate 1 liter of LB containing 100 g ampicillin/ml. DE3/pHGDH was grown at 37°C until A 600 reached 1.0, and then isopropyl-1-thio-␤-D-galactopyranoside was added to a final concentration of 1 mM. After isopropyl-1-thio-␤-Dgalactopyranoside induction, DE3/pHGDH was grown for an additional 3 h at 37°C and then harvested by centrifugation. Cell pellets were suspended in 100 ml of 100 mM Tris-HCl, pH 7.4, 1 mM EDTA, and 5 mM dithiothreitol and lysed with a sonicator. The wild-type and mutant GDHs were purified according to the method developed in our laboratory (3,31). Because the wild-type and mutant GDHs were readily solubilized, no detergents were required throughout the entire purification steps. The purified enzymes were analyzed by SDS-PAGE and recognized by Western blot using monoclonal antibodies previously produced in our laboratory against bovine brain GDH (42). Protein concentration was determined by the method of Bradford (43) using bovine GDH as a standard.
Enzyme Assay and Kinetic Studies-Because E. coli has only an NADP ϩ -dependent GDH (36), the enzyme assay was performed with NAD ϩ in the direction of glutamate oxidation in 50 mM Tris-HCl, pH 9.5, containing 14 mM NAD ϩ , 2.6 mM EDTA, and 1 mM ADP at 25°C. The reaction was started by the addition of glutamate to 25 mM final concentration. One unit of enzyme was defined as the amount of enzyme required to reduce 1 mol NAD ϩ /min at 25°C. For determination of K m and V max values, the assays were carried out by varying the substrate under investigation while keeping the other substrate and reagents at the optimal concentration indicated above. The K m and V max values were calculated by linear regression analysis of double-reciprocal plots and given along with standard errors. Catalytic efficiency was estimated by use of the equation (44). Effects of allosteric regulators on GDH activities were examined by incubating the enzyme with the allosteric effectors at various concentrations in the assay buffer at 25°C. At intervals after the initiation with the effectors, aliquots were withdrawn for the assay of GDH activity.

RESULTS
Specific Binding of 2N 3 NAD ϩ to Human GDH-The role of Glu 279 in the direct binding of NAD ϩ to human GDH was examined by photoaffinity labeling with [ 32 P]2N 3 NAD ϩ . To show specificity of the photoprobe-protein interaction, saturation of photoinsertion should be observed. To demonstrate saturation effects with the photoprobe, the purified enzymes were photolabeled with increasing concentrations of [ 32 P]2N 3 NAD ϩ in the presence of glutarate. Under the experimental conditions described, saturation of photoinsertion with [ 32 P]2N 3 NAD ϩ occurred at around 95 M photoprobe (Fig. 1). In all photolabeling experiments, the ionic strength was kept low to enhance binding affinity because we have observed in general that the lower the ionic strength, the tighter the binding of nucleotide photoaffinity probes and the more efficient the photoinsertion. The results in Fig. 1 indicate the saturability of the NAD ϩ -specific site of GDH with this photoprobe and therefore decrease the possibility of nonspecific photoinsertion. To further demonstrate specific labeling of GDH, wild-type GDH was photolabeled with [ 32 P]2N 3 NAD ϩ in the presence of increasing NAD ϩ concentrations. As shown in the results of the competition experiments (Fig. 2), NAD ϩ was able to protect photolabeling from 10 M [ 32 P]2N 3 NAD ϩ at concentrations in the range of known K d values (41). Approximately 85% protection was observed with 150 M NAD ϩ for wild-type GDH (Fig. 2). These results show the specificity and utility of [ 32 P]2N 3 NAD ϩ as a good probe for determining the NAD ϩ binding site.
Tryptic Digestion of Photolabeled Proteins and Boronate Affinity Column-To identify the peptides modified by [ 32 P]2N 3 NAD ϩ , wild-type GDH was photolabeled twice in the absence and presence of 300 M NAD ϩ and digested with trypsin. To reduce any possible nonspecific labeling and, at the same time, optimize the specific labeling of the enzymes, 100 M [ 32 P]2N 3 NAD ϩ was used, which is the concentration at which photoinsertion approaches saturation. The photolabeled proteins were separated from most of the noncovalently bound nucleotide by acid precipitation and proteolyzed by trypsin. Taking advantage of the existence of the two cis-hydroxyl groups on the ribose groups of the photoincorporated probe, the photolabeled peptides were isolated by using immobilized boronate column chromatography. It has been shown that immo-bilized boronate can be successfully used to fractionate adenine and pyridine nucleotides (21,45) and nucleosidyl peptides (46,47). The tryptic digested GDHs were applied to the boronate column equilibrated with 0.1 M ammonium acetate, pH 9.0. After washing the column with ammonium acetate buffer, pH 9.0, the radioactive peptides were eluted with a pH gradient of 9.0 to 5.0 (Fig. 3). One major radioactive peak around pH 6.5 was observed for wild-type GDH. The peptides exhibited an unusual UV spectrum with a maximum absorption of 262 nm and a shoulder near 278 nm, which verifies that an adduct of the [ 32 P]2N 3 NAD ϩ photoprobe is still covalently attached to the peptides. Photolabeling of the peptide was prevented by the presence of NAD ϩ during photolysis. When 300 M NAD ϩ was originally present in the incubation mixture, ϳ 90% of the radioactivity of the peak was eliminated, as shown in Fig. 3. These results demonstrate the selectivity of the photoprobe and suggest that the radioactive peak represents a peptide in the NAD ϩ binding domain of human GDH.
When the radioactive eluates from the boronate column were subjected to reversed-phase HPLC, one major radioactive peak was clearly observed (data not shown). The radioactive peptide was collected and identified by amino acid sequence analysis.
Sequence Analysis of Photolabeled Peptide-The amino acid sequence analysis revealed that the peak fractions contained the amino acid sequence CIAVGXSDGSIWNPDGI. The sequences obtained were also compared with those of various GDHs (Table I). As judged by comparison with the amino acid sequence of mammalian GDHs, this site was identified as residues 274 -290 of human GDH. The symbol X indicates a position for which no phenylthiohydantoin amino acid could be assigned. The missing residue, however, can be designated as a photolabeled Glu because the sequences including the Glu residue in question have a complete identity with those of the other GDH species known. The amino acid composition of the photolabeled peptide revealed that the peptide had a composition that was compatible with that of the tryptic peptide spanning residues 274 -293, with the exception that there was a significant reduction in Glu (Table II). On the basis of information obtained on the amino acid sequence determination and composition analysis of the isolated peptide, we suggest that the attachment site of [ 32 P]2N 3 NAD ϩ is Glu 279 . The impor- Wild-type GDH in reaction buffer was photolysed with the indicated concentrations of [ 32 P]2N 3 NAD ϩ , and 32 P incorporation into protein was determined by liquid scintillation counting (see "Experimental Procedures" for details). Relative 32 P incorporations were expressed relative to each control.

FIG. 2. The effect of NAD ؉ on [ 32 P]2N 3 NAD ؉ phosphoincorporation into GDH.
Wild-type GDH in reaction buffer was photolysed with 100 M [ 32 P]2N 3 NAD ϩ in the presence of the indicated concentrations of NAD ϩ . Relative 32 P incorporation into protein was determined and expressed as described in Fig. 1. tance of the Glu 279 residue in the binding of NAD ϩ was further examined by cassette mutagenesis at position 279.
Construction, Expression, and Purification of Glu 279 Mutants-The 46-bp NsiI/EspDI fragment in pHGDH was re-placed with five different 46-bp synthetic DNA duplexes containing a substitution on both DNA strands at positions encoding Glu 279 . These substitutions made mutant proteins E279G, E279L, E279M, E279R, and E279Y at position 279. The five mutants were designed to have different size, hydrophobicity, and ionization of the side chains at each position. Analysis of crude cell extracts by Western blot showed that Glu 279 mutant plasmids encoding an amino acid substitution at position 279 directed the synthesis of a protein that interacted with monoclonal antibodies raised against GDH at almost identical levels for all Glu 279 mutants and wild-type GDH (Fig. 4). These results indicate that the mutagenesis at Glu 279 has no effects on expression or stability of the different mutants. The mutants also appeared to be as stable as wild-type human GDH, on the basis of their stability toward proteolysis and retention of activity upon prolonged storage at 4°C. In addition, the mutant enzymes were purified homogeneously by the same method as the wild-type enzyme (data not shown), indicating that no gross conformational change in the enzyme had occurred.
Kinetic Properties for Mutant GDH-To evaluate the kinetic properties for the mutants, the K m values for the individual substrates were determined. Detailed investigation of the catalytic activities of the mutant enzymes revealed an ϳ10 -14fold decrease in the respective apparent K m values and an 11-18-fold decrease in the k cat /K m values for NAD ϩ compared with those of wild-type GDH (Table III). Although the mutations at position 279 produce an increase in the apparent K m values for NAD ϩ , all the mutants and the wild-type enzyme show similar k cat values of the same order of magnitude (Table  III). Therefore, the dramatic reductions in the catalytic efficiencies (k cat /K m ) of the Glu 279 mutants primarily reflect changes in the K m values for NAD ϩ and suggest that the mutations at the Glu 279 site reduced the affinity of the enzyme for NAD ϩ binding. These results suggest that Glu 279 plays an important role for efficient binding of the coenzyme to human GDH.
The K m values for glutamate for wild-type and Glu 279 mutants were also determined in the standard assay mixture at various concentrations of glutamate. In contrast to the K m values for NAD ϩ , the apparent K m values for glutamate obtained from Lineweaver-Burk plots increased only slightly (3.05 and 3.44 -4.12 mM for wild-type and mutant GDHs, respectively) (Table IV). These slight changes may be due to a local conformational change in substrate binding by the mutant enzymes. However, the similarities in k cat /K m values (14 -19 s Ϫ1 mM Ϫ1 ) for glutamate between wild-type and mutant enzymes suggest that substitution at position 279 might have no appreciable effect on the affinity of the enzyme for glutamate.    ADP and GTP Effects on Human GDH Activity-Finally, the effects of ADP and GTP, well-known allosteric regulators, on the activities of wild-type and mutant GDHs were compared. There were no differences between wild-type and mutant GDHs in sensitivity to ADP activation at concentrations between 0.1 and 1.0 mM and GTP inhibition at concentrations between 1 and 50 M (Table V). These results indicate that Glu 279 is not responsible for the allosteric regulation of human GDH by ADP or GTP. DISCUSSION The construction of a synthetic gene encoding human GDH will enable us to generate a large number of site-directed mutations at several positions in the coding region. The high level of GDH expression as a soluble protein in E. coli will facilitate the purification of large quantities of mutant proteins for biochemical and structural studies. This combination of genetic and biochemical techniques could be used to address a broad range of questions related to the structure and function of human GDH. In the present work, we identified an adenine binding domain peptide of the NAD ϩ binding site of human GDH using cassette mutagenesis of the synthetic human GDH gene and photoaffinity probe [ 32 P]2N 3 NAD ϩ . [ 32 P]2N 3 NAD ϩ is a probe that, on photolysis, generates a very reactive nitrene that has the capacity of photoinserting into any residue. The data showing decreased photoinsertion by addition of NAD ϩ demonstrate that photoinsertion occurs only by the bound form of [ 32 P]2N 3 NAD ϩ . This indicates that proximity controls photoinsertion and that the residues modified are within the adenine binding domain. In addition, pre-photolysis followed by immediate addition of GDH did not lead to covalent labeling (data not shown), eliminating the existence of any long-lived chemically reactive intermediate that could be involved in covalently modifying enzymes. The selectivity and specificity have been successfully utilized to locate the specific base binding domains of the nucleotide binding site peptides of many proteins (23)(24)(25)(26)(27)(28)(29)(30).
The specificity of [ 32 P]2N 3 NAD ϩ and the utility of this probe as a good candidate for determining the NAD ϩ binding site were further supported by the following observations. First, in the absence of activating light, 2N 3 NAD ϩ is a substrate for GDH (data not shown). The ability to mimic a native compound before photolysis has an advantage over determination of the enzyme function after modification. Second, the photoinsertion into GDH was saturated with [ 32 P]2N 3 NAD ϩ . Saturation of photoinsertion with [ 32 P]2N 3 NAD ϩ occurred at around 95 M photoprobe (Fig. 1). Third, the prevention of photoinsertion of [ 32 P]2N 3 NAD ϩ by NAD ϩ demonstrates that the photoprobe is inserted into a specific NAD ϩ site within GDH (Fig. 2). In addition, the sites of attachment of the photoaffinity label were more precisely defined by generating small peptide fragments of the labeled protein and separating the labeled peptides. For separation of the 32 P-labeled peptide fragments generated in the proteolytic digest, immobilized boronate column chromatography was carried out before HPLC separation. The boronate chromatography greatly reduces the possibility of any nonphotolabeled peptide co-eluting on HPLC with the photolabeled peptide, which could give misleading results.
There were differences in the biochemical properties between wild-type GDH and Glu 279 mutants. The K m values for NAD ϩ increased 10 -14-fold in the Glu 279 mutants compared with those of wild-type GDH (Table III). Although the mutations at position 279 produce changes in the apparent K m values for NAD ϩ , all the mutants and wild-type enzyme show similar k cat values of the same order of magnitude (Table III). The results from the Western blot analysis (Fig. 4) show that the mutagenesis at the Glu 279 site has no effects on expression or stability of the different mutant GDHs. Therefore, the dramatic reductions in the catalytic efficiencies (k cat /K m ) of the Glu 279 mutants primarily reflect changes in the K m values for NAD ϩ and suggest that Glu 279 plays a role in NAD ϩ binding. The importance of Glu 279 in NAD ϩ binding is supported by the interesting information that the invariant, functional residue Asp is found in the ␤-pleated sheet region of four known dehydrogenases and is conservatively changed to Glu 279 in mammalian GDHs (34). With these dehydrogenases, Asp is proposed to be involved in hydrogen bond formation with the O-2Ј atom of the ribose group, which could not occur, due to charge repulsion, with NADP ϩ . This probably accounts for the inability of NADP ϩ to be a substrate for these dehydrogenases, whereas it is a good substrate for mammalian GDH. This concept is supported by the observation that the equivalent position of Glu 279 in NADP ϩ -specific dihydrofolate reductase is replaced by a positively charged Arg residue. It is likely that the reason for Glu replacement of Asp in GDH is to produce a different active site conformation that allows both NAD ϩ and NADP ϩ binding.
The crystal structure of C. symbiosum GDH has been  aligned with the primary sequence of bovine liver GDH (8 -11, 48). The structures of C. symbiosum and mammalian GDHs were suggested to be similar due to considerable identity and the conservation of 13 glycine residues, which probably conserves the structure among species, and to consist of two domains. The first domain has been proposed to contain the catalytically important residues, whereas the second domain contains a ␤-␣-␤ motif that is responsible for NAD ϩ binding (8). The primary sequence of bovine liver GDH also suggests the presence of a ␤-␣-␤ motif corresponding to the motif within the bacterial subunit. The photolabeled peptide of human GDH identified in this work is located within the proposed NAD ϩ binding domain of bovine liver GDH (10,11,34,35). Therefore, this sequence is expected to interact with the adenine ring of NAD ϩ . This site is also near the active site portion of clostridial GDH (49). A large motion of the NAD ϩ binding domain is associated with substrate binding and is required for catalysis in clostridial GDH (8,50). However, the similarities of K m values for glutamate between wild-type and Glu 279 mutants (Table IV) suggest that the mutagenesis at position 279 does not severely affect the affinity of human GDH for the substrate. Very recently, Smith et al. (51). have reported that the ␣1 and ␣2 helices associated with the NAD ϩ binding domain in human GDH undergo large conformational changes and that all of these conformational changes are absorbed before the beginning of the first ␤-strand of the glutamate binding domain that remains fixed during these movements. Therefore, in contrast to clostridial GDH, the NAD ϩ binding site in human GDH may not be directly associated with the glutamate binding site. It has been suggested that the GTP site lies between the NAD ϩ binding domain and the antenna, whereas the ADP site is under the pivot helix and behind the glutamate binding domain (11). The presence of an additional coenzyme binding site for bovine liver GDH has been proposed by a number of investigators (52,53). Based on the recent crystal structure of bovine liver GDH, it has been suggested that the second NAD ϩ site is equivalent to the ADP activation site and that NADH, NAD ϩ , and ADP bind to the same site (9 -11, 51). These suggestions, however, raise a paradox because NADH inhibits the enzyme, whereas NAD ϩ and ADP activate the enzyme. It should be mentioned that various chemical reagents affect NADH inhibition by binding to disparate sites of the enzyme, yet none of the residues are near the bound coenzyme. These results suggest that the second coenzyme binding to the ADP site may not be wholly responsible for the observed NADH inhibition. For the last 20 years, the sequence identities and kinetic properties of mammalian GDHs from various sources including human liver, human brain, rat brain, mouse brain, chicken liver, bovine brain, and bovine liver have been reported by many researchers. It is very interesting that no other mammalian GDHs except bovine liver GDH have been known to show the second coenzyme site, although the sequence identities between mammalian GDHs are extremely high. Very recently, the structure of apo human GDH has been determined, but no evidence for the second coenzyme binding site has been provided (51). To our knowledge, no studies by site-directed mutagenesis for the coenzyme binding site(s) of the mammalian GDHs have been reported yet, although the importance of Lys 286 at the NADP site of GDH from Salmonella typhimurium has been reported using site-directed mutagenesis (54). The possibility of photolabeling into the second NAD ϩ binding site was eliminated in this study because at the concentration of 2N 3 NAD ϩ used, there is approximately 1 coenzyme molecule bound/submit according to the equilibrium dialysis data, and this is at the active site (39 -41). In addition, there were no differences between wild-type and Glu 279 mutant GDHs in sensitivity to ADP activation and GTP inhibition ( Table V), suggesting that Glu 279 is not responsible for the second coenzyme binding to the putative ADP site.
In conclusion, the enzyme efficiencies (k cat /K m ) were 11-18fold lower for Glu 279 mutants than for the wild-type. The decreased efficiency of the Glu 279 mutants results from the increase in K m for NAD ϩ , consistent with a role for the Glu residue at position 279 that enhances the binding of NAD ϩ . The results with cassette mutagenesis and photoaffinity labeling techniques suggest that Glu 279 is required for efficient base binding of NAD ϩ to human GDH.