A New Class of Glutamate Dehydrogenases (GDH)

A new class of glutamate dehydrogenase (GDH) is reported. The GDH of Streptomyces clavuligerus was purified to homogeneity and characterized. It has a native molecular mass of 1,100 kDa and exists as an α6 oligomeric structure composed of 183-kDa subunits. GDH, which requires AMP as an essential activator, shows a maximal rate of catalysis in 100 mmphosphate buffer, pH 7.0, at 30 °C. Under these conditions, GDH displayed hyperbolic behavior toward ammonia (K m , 33 mm) and sigmoidal responses to changes in α-ketoglutarate (S0.5 1.3 mm;n H 1.50) and NADH (S0.5 20 μm; n H 1.52) concentrations. Aspartate and asparagine were found to be allosteric activators. This enzyme is inhibited by an excess of NADH or NH4 +, by some tricarboxylic acid cycle intermediates and by ATP. This GDH seems to be a catabolic enzyme as indicated by the following: (i) it is NAD-specific; (ii) it shows a high value of K m for ammonia; and (iii) when S. clavuligerus was cultured in minimal medium containing glutamate as the sole source of carbon and nitrogen, a 5-fold increase in specific activity of GDH was detected compared with cultures provided with glycerol and ammonia. GDH has 1,651 amino acids, and it is encoded by a DNA fragment of 4,953 base pairs (gdh gene). It shows strong sequence similarity to proteins encoded by unidentified open reading frames present in the genomes of species belonging to the genera Mycobacterium,Rickettsia, Pseudomonas, Vibrio,Shewanella, and Caulobacter, suggesting that it has a broad distribution. The GDH of S. clavuligerus is the first member of a class of GDHs included in a subfamily of GDHs (large GDHs) whose catalytic requirements and evolutionary implications are described and discussed.

Accordingly, GDHs, as well as aminotransferases, are enzymes that participate in the reapportionment of ␣-amino groups inside cells, playing a central role in the regulation of the flux of intermediates between different biosynthetic and catabolic pathways (3,4).
All GDHs reported to date are oligomeric enzymes (1)(2)(3) and, depending on the subunit size and monomer composition, can be grouped into the following two different structural types: hexameric GDHs (with six identical subunits of about 50 kDa) and tetrameric GDHs (with four identical subunits having a molecular mass close to 115 kDa). The hexameric type includes the NAD-or NADP-GDHs from most known bacterial species (8), the NADP-GDHs from lower eukaryotes (9), and the NAD(P)-GDHs from vertebrates (10). The tetrameric type includes the NAD-GDHs found in lower eukaryotes (9). Dehydrogenase enzymes that utilize glutamate, leucine, valine, and phenylalanine have been recognized as common members of a large superfamily (11)(12). Thus, in line with proper hierarchical ordering, GDHs should be considered as a family of enzymes that includes two subfamilies as follows: small GDHs (S_GDHs, which contain the two member classes 50 I kDa GDHs and 50 II kDa GDHs) and large GDHs (L_GDHs, which until now contained only the 115-kDa GDHs). Some interesting considerations of the evolutionary relationship between tetrameric and hexameric GDHs have been ventured (13)(14), and this topic area can now be reevaluated and expanded with the current discovery of a more diverse and complex GDH membership than previously known.
Despite the fact that GDH has an ubiquitous distribution in nature, this enzyme had not been previously detected in the important industrial species Streptomyces clavuligerus (15). For this reason, it was proposed that the catabolism of glutamic acid in S. clavuligerus does not require the participation of this enzymatic activity (15). In contrast with the latter conclusion, we found that when the assimilation of glutamic acid by S. clavuligerus cultured in a medium containing this amino acid as the carbon and nitrogen source was studied, a GDH activity was, in fact, present. Here we report the purification, characterization, and sequence of the encoding gene of this enzyme, thus providing a first documentation of a previously unknown class of GDH. Phylogenetic and evolutionary relationships are discussed.

EXPERIMENTAL PROCEDURES
Materials-Molecular biology products were supplied by Amersham Pharmacia Biotech. Biochemicals and reagents were obtained from Sigma.
Strains and Culture Media-S. clavuligerus NRRL-3585 (ATCC 27064) was obtained from the American Type Culture Collection. Spores were produced as reported (16) and kept frozen at Ϫ20°C in 20% w/v glycerol (17). DNA manipulations and sequence analysis were carried out as reported previously (18).
The medium used for the growth of S. clavuligerus was a seed medium (19) containing glycerol (10 g liter Ϫ1 ) and/or glutamic acid (5 g liter Ϫ1 ) as carbon sources. Incubations were carried out as reported (20).
GDH Assay-GDH activity (reductive amination) was analyzed at 30°C by following the oxidation of NADH at 340 nm in the presence of ␣-ketoglutaric acid and ammonia (21)(22)(23). The reaction mixture contained ␣-ketoglutarate (15 mM); NH 4 Cl (25 mM), NADH (150 M), the essential activator AMP (1 mM), and 25 l of enzyme solution (between 4 and 15 milliunits). All substrates were dissolved in 100 mM HK 2 PO 4 / H 2 NaPO 4 at pH 7.0. In some experiments, effectors were added to the reaction mixture as specified. GDH reactions (1 ml) were started by adding ␣-ketoglutarate to the assay mixture, and the transformation of NADH into NAD ϩ was followed at 340 nm in a Shimadzu UV-120-02 spectrophotometer. One international unit (IU) of enzyme activity is defined as the catalytic activity leading to the consumption of a micromole of NADH per min. Usually, the activity is given as milliunits (10 Ϫ3 units). Specific activity is indicated as milliunits/mg protein.
When required, GDH activity was assayed in the direction of oxidative deamination. In such cases the concentrations of glutamic acid and NAD ϩ were 15 and 2 mM, respectively. Protein was measured by the method of Bradford (24).
Purification of GDH from S. clavuligerus-S. clavuligerus mycelia cultured as above (19 -20) were harvested after 78 h of growth, filtered, and washed with sterile saline solution (1 l). Wet cells (25 g, obtained from 3 liters of culture) were resuspended in 200 ml of 25 mM phosphate buffer, pH 7, containing 10% (w/v) glycerol (PG buffer) and disrupted either by sonication or by using a Braun MSK mechanical disintegrator. Cell debris was eliminated by centrifugation (37,000 ϫ g, 30 min, 4°C); the pellet was discarded, and the supernatant fluid, containing the GDH activity, was precipitated with ammonium sulfate (35-47.5% saturation). The precipitate was dissolved in PG buffer and passed through a Sepharose G-25 PD10 column (equilibrated with the same buffer) to eliminate the excess of ammonium sulfate. Fractions containing maximal activity were mixed, and the total volume (75 ml) was applied to a DEAE-Sephacel (Amersham Pharmacia Biotech) column (2.0 ϫ 18 cm) equilibrated with PG buffer. The column was washed with 150 ml of this buffer. Elution was carried out with a KCl gradient (0.16 -0.28 M). Aliquots of 2 ml were collected at a flow rate of 20 ml h Ϫ1 . GDH activity eluted between 50 and 125 ml, showing a peak in the fraction 48. These fractions were mixed and precipitated with (NH 4 ) 2 SO 4 (65% saturation). The precipitate was dissolved in 2 ml of PG buffer and applied to a Sepharose CL-6B (Amersham Pharmacia Biotech) column (3 ϫ 128 cm). Aliquots of 6 ml were collected (flow rate 20 ml h Ϫ1 ) and assayed for GDH activity. The enzyme eluted between fractions 79 and 83, showing a peak in tube 81. By using this procedure the enzyme was purified to homogeneity (225-fold). The results of a typical purification following this procedure are summarized in Table I.
GDH Amino Acid Sequences-The amino-terminal amino acid or tryptic peptide sequences were determined as reported before (18).
Fast Gene Amplification Procedure-The isolation of the gene encoding GDH in S. clavuligerus was performed by a strategy that allows the rapid amplification of a single gene from a genomic library, the only requirement being that an oligopeptide belonging to the protein, or an oligonucleotide sequence corresponding to its encoding gene, must be known previously. The principle of the method is based on the PCR (27). Two primers flanking the left and the right sides of the polylinker of the vector used to construct the S. clavuligerus genomic library were designed. Later, an oligonucleotide deduced from the amino terminus of the protein was combined with either the left or the right primers, and PCR amplification was carried out using a sample of the total DNA obtained from the genomic library as template (28).
PCR was carried out in a Perkin-Elmer DNA Thermal Cycler 2400. Each independent reaction (50 l) contained the following: 75 mM Tris-HCl buffer, pH 9; 50 mM KCl; 20 mM (NH 4 ) 2 SO 4 ; 100 ng of purified genomic library DNA; 0.4 M each independent primer; 2 mM MgCl 2 ; 0.4 mM dNTPs, and a mixture of thermostable DNA polymerase (2 units) from Thermus thermophilus (Biotools, Spain) and Pfu DNA polymerase (1 unit) from Pyrococcus furiosus (Promega). The annealing temperature was close to 70°C; the extension time was 4 -5 min, and the number of cycles was 35.
Bioinformatic Analysis-Multiple alignments were obtained with the ClustalW program (29) included in the BioEdit multiple alignment tool (30). Phylogenetic trees were computed by use of the maximum parsimony method (bootstrap scores for 500 iterations) (31) and drawn with the PHYLO_WIN program (32). Alternatively, results were compared by use of the neighbor-joining method (33) using either the PHY-LO_WIN program (bootstrap scores for 500 iterations) or using the PHYLIP program (bootstrap scores for 1000 iterations) (34). The phylogenetic tree for prokaryotes was constructed by extraction of a subtree using methods facilitated by the Ribosomal Data Base Program.

RESULTS AND DISCUSSION
Physiological Aspects-S. clavuligerus is a filamentous bacterium belonging to the Actinomycetales order, a group of microbes that is associated with broad biotechnological applications and of great industrial interest (35). Most of the strains included in the genus produce a variety of secondary metabolites that have clinical and pharmacological relevance (36). For example, S. clavuligerus synthesizes different ␤-lactam compounds, including penicillins, cephamycin, and the ␤-lactamase inhibitor clavulanic acid (37)(38). The biosynthetic pathways of these compounds are strictly regulated, and most of them are under a mechanism of control directly related to the carbon or nitrogen source supplied to the medium (19, 39 -41). Many efforts have been made to elucidate their biosynthetic pathways as well as the molecular basis of the regulatory mechanisms controlling them (19, 40 -43). It has been shown that ␤-lactam antibiotic production is regulated by the nitrogen source used for growth in S. clavuligerus (34,44). Accordingly, the influence of several enzymes related to nitrogen metabolism (GDH; glutamine synthetase, or glutamate synthase) on the biosynthesis of such compounds has been studied (15,(45)(46). Although attempts to assay GDH in S. clavuligerus were unsuccessful (15), glutamine synthase and glutamate synthase have been studied and partially characterized (15,(45)(46). In light of these results, it has been suggested that in S. clavuligerus the function of GDH is performed by alanine dehydrogenase, an enzyme that is also involved in nitrogen catabolism (15).
In contrast to the latter results, we observed the presence of a GDH that catalyzes in vitro the conversion of ␣-ketoglutarate and ammonia into glutamate (reductive amination) as well as the reverse reaction (oxidative deamination). The enzyme was found to be induced under nutritional conditions that demand the catabolic function ( Fig. 1). This enzyme is NAD-specific, requires AMP as an essential activator, and it is strongly inhibited by Tris. These unusual characteristics undoubtedly explain why GDH was not detected before in S. clavuligerus. A chemically defined medium containing glycerol and glutamic acid as the carbon and nitrogen sources, respectively, supports excellent growth of S. clavuligerus (Fig. 2). Under these culture conditions, GDH achieved a maximal specific activity in the early phase of growth, and this level was sustained throughout the exponential phase of growth and even for a significant portion of the stationary phase. After an elapsed time of 100 h, rapid enzyme degradation was apparent. In view of these results, bacterial cultures were harvested at 78 h for the purification of GDH (see Table I).
Physicochemical and Enzymological Properties-Purified GDH (see Table I) migrated in SDS-10% PAGE as a single band with a molecular mass of 179 Ϯ 7 kDa (Fig. 3). According to Sepharose CL-6B column chromatography, the native enzyme has a mass of 1,084 Ϯ 55 kDa. PAGE analysis, carried out under non-denaturating conditions, revealed a molecular mass of 1,135 Ϯ 39 kDa (data not shown). These results indicate that the native form of the enzyme is probably a homohexamer. By contrast, all the hexameric GDHs reported to date (NAD-or NADP-GDHs) have a monomeric molecular mass of about 50 kDa (14,47). Therefore, GDH from S. clavuligerus represents a new class of glutamate dehydrogenase, i.e. a second class within the large subfamily of GDHs (L_GDHs, see below), designated as GDH L_180.
Under the above conditions, the reactions were linear between 1.5 and 22 min when the quantities of GDH added to the assays ranged between 20 and 1 milliunits, respectively (data not shown). When substrate saturation profiles were obtained, GDH showed hyperbolic behavior toward increasing ammonia concentration ( Fig. 5c) but sigmoidal responses to increasing changes in both ␣-ketoglutarate (Fig. 5a) and NADH (Fig. 5b) concentrations. Recently, the GDH of Psychrobacter sp. TAD1 was reported to exhibit similar phenomena of positive cooperativity (54). Unlike the S. clavuligerus GDH, the Psychrobacter GDH is NADP-specific and is a member of the S_50 I class. It is perhaps of interest that in correlation with the positive cooperativity of substrate binding, the Psychrobacter GDH is thus far the only GDH to share the variant Rossmann motif (GXGXXS) for dinucleotide binding that is typical of the GDH L_180 class. Interestingly, ammonia and NADH caused inhibition of GDH above a given threshold concentration. Thus, the addition to standard reactions (in the absence of Asp, see "Experimental Procedures") of NH 4 ϩ at a concentration higher than 100 mM caused inhibition of GDH activity (about 10%), whereas if the concentration was 350 mM or higher, no activity was detected. Likewise, NADH began to inhibit GDH activity at a concentration of 200 M (10%) whereas at 500 M, inhibition exceeded 95%. Similar data were noted for GDH of Azospirillum brasilense (55). As occurs with other GDHs (51), the reductive amination reaction of S. clavuligerus GDH is inhibited by the reaction products (NAD ϩ and glutamic acid) as well as by NADP ϩ (see Table III).
Further kinetic studies allowed the calculation of the K m , S 0.5 , and Hill coefficient values in each case (see Table II).
Additional analysis revealed that aspartic acid (see Fig. 5, a and b, and Table III) and asparagine (Table III) are allosteric activators of S. clavuligerus GDH. Although it has been reported that different molecules (AMP, ADP, GTP, ATP, NADP ϩ , and succinate) could be allosteric modulators of different GDHs (51, 56 -58), to our knowledge, this is the first description of aspartic acid and asparagine as allosteric activators of glutamate dehydrogenase.
In order to characterize the substrate specificity of this enzyme, some of the substrates of GDH were replaced during the assay by close structural molecules. Thus, when ␣-ketoglutarate (reductive amination) or glutamic acid (oxidative deamination) was replaced by other ␣, ␤, or ␥ ketoacids or by different amino acids (those indicated in Table III), respectively, no GDH activity was detected. This indicates specificity of the enzyme for ␣-ketoglutarate and for glutamic acid. Moreover, when NAD(H) was replaced by NADP(H), no catalysis was detected, therefore indicating that this enzyme is an NAD(H)-specific GDH.
In other experiments, AMP was omitted from the assays or replaced by other nucleotides. In its absence, no GDH activity was detected, indicating that AMP, at least in the assay conditions tested, is an essential activator for this enzyme. Although a similar activating effect of AMP on various GDHs has been observed elsewhere (53, 57, 59), AMP has not previously been found to be essential for activity. In Escherichia coli AMP is an activator at micromolar concentrations, but it is an inhib-  itor at millimolar concentrations (56). The NAD-GDH of Clostridium difficile is also inhibited by AMP, as well as by ADP and ADP-ribose (52).
When AMP was replaced by ADP, 20% of activity was detected, but when AMP was substituted by other mono-, di-, or trinucleotides, no activity was measured. Furthermore, when ATP and GTP were tested as effectors, they inhibited GDH activity (Table III). Vertebrates possess a GDH that is subject to allosteric inhibition by GTP (57). Study of the influence of other molecules on the GDH activity revealed that some intermediates of the Krebs cycle (isocitrate, succinate, fumarate, malate, and oxalacetate) as well as glyoxylic acid and glutaric acid are also inhibitors (see Table III). Reports of inhibition caused by different central metabolic intermediates on GDH activities have appeared in the literature a number of times, but no common pattern of inhibition has been established (58,60).
Enzymatic assays of GDHs are usually performed either in Tris-HCl or in phosphate buffer (22)(23). However, we observed that the GDH activity of S. clavuligerus was not detected when cell-free extracts of this bacterium were prepared in 100 mM Tris-HCl. These data suggested that Tris could act as an inhibitor of the GDH. Analysis of this compound as an effector of GDH activity revealed that 50 mM Tris causes 81% of inhibition, whereas at 80 mM Tris, no activity was detected. Furthermore, other Tris structural analogs (or close related molecules) caused a similar degree of inhibition. Thus, when added as effectors to the GDH assay mixture (at a concentration of 50 mM), ethylamine, aminoethylpropanol, ethanolamine, methoxyethylamine, and aminoethylpropanediol caused inhibitions of up to 60% (Table III). The fact that AMP is an essential activator together with the inhibition caused by Tris very likely explain previous failures to detect this enzymatic activity in S. clavuligerus cell-free extracts (15).
Isolation and Characterization of the Gene Encoding GDH in S. clavuligerus-The molecular mass of the GDH, deduced from the amino acid sequence encoded in the gdh, is 183,354 Da, in agreement with the molecular mass estimated for each monomer by SDS-PAGE (179 Ϯ 7 kDa, see above).
The gene encoding GDH in S. clavuligerus (gdh) was isolated from a -GEM-12 genomic library using a very simple and efficient strategy (see "Experimental Procedures") that allows the rapid amplification of a gene from a genomic library with the only requirement being that an oligonucleotide deduced from the amino terminus of a protein (for the whole gene) or deduced from an internal part of the protein (for the partial amplification of the gene) is known (28). By using this method, we PCR-amplified different bands that, when sequenced, revealed that they include an ORF of 4,953 base pairs (gdh gene) encoding a protein of 1,651 amino acid residues (Fig. 6). Analysis of the amino-terminal sequence (MQTKLDEAKAEL-LARAARV) of the purified protein as well as other internal peptides isolated after tryptic digestion (AAVADLVHIEAL-AGSGR and VVGEGGNLGLTQLGR) confirmed that all were present in the protein (see Fig. 6).
Furthermore, comparative analysis of the amino acid sequence of this protein with those included in the data bases without a functional assignment revealed that the central domain is quite similar to known GDHs, suggesting that the protein encoded by this gene is capable of GDH function, as expected. Additionally, a strong similarity (see Table IV) throughout the entire sequence was observed when S. clavuligerus GDH was compared with presumptive proteins encoded by genes present in the genomes of species belonging to the genera Mycobacterium, Rickettsia, Pseudomonas, Vibrio,    Table IV). It seems readily apparent that the GDH of S. clavuligerus represents a new class of GDH exhibiting some breadth of distribution in nature. This hexameric GDH has a monomer size that corresponds to neither the typical 50-kDa size of all the hexameric GDHs reported to date nor to the 115-kDa size of tetrameric GDHs. Thus, according to the criteria for oligomeric structure and monomer size, three different types of GDHs (␣ 6 -50, ␣ 6 -180, and ␣ 4 -115 kDa) exist. Hierarchical homology grouping (see below) reveals the existence of four different well defined classes (50 I , 50 II , 115, and 180 kDa). These sort into two widely spaced subfamilies as follows: the small GDHs (S_GDHs, including 50 I and 50 II classes) and the large GDHs  relationship between the hexameric S_GDH subfamily and the tetrameric GDH L_115 class has been assessed (14). Insertions and deletions that comprise the essential alignment variations were found to be between or at the ends of the elements of secondary structure. Substantial insight has been obtained into the understanding of the relationship between secondary structure, the variation in particular sequence, and the unique insertions/deletions on the one hand, and with the subunit interactions that dictate oligomer size on the other hand. The eventual crystallographic structure of the GDH L_180 protein will be a fascinating addition to this analysis because it is hexameric (similar to the less closely related S_GDH subfamily members), yet it is much more similar to the tetrameric GDH L_115 class members in terms of overall identity and the gen- The GDH domain itself has two subdomains as follows: the glutamate-binding region (subdomain I) and the dinucleotide-binding region (subdomain II). Subdomain I is indicated as a region with gray color; subdomain II is indicated with yellow color; glutamate-binding residues are in red; residues contacting the dinucleotide are indicated in blue, and residues contacting both glutamate and dinucleotide (K in subdomain I and N and G in subdomain II) are indicated in black boxes. ␤␣␤ fold is indicated as a box. The central 520-amino acid GDH region within the 1651-residue length of the S. clavuligerus is shown to highlight functionally important residues implicated by their conservation in comparison with rigorously studied model proteins. The approximate boundaries of the GDH domain are between amino acid residues 764 and 1283. The core GDH region extends from residue 815 to residue 1253. eral configuration of insertions and deletions (relative to the S_GDH subfamily).
Structural Analysis of S. clavuligerus GDH, Key GDH Residues-The inferred organization of the S. clavuligerus GDH domain is illustrated in Fig. 6. The conservation of residues known to be critical for GDH function in the thoroughly studied systems of Clostridium symbiosum (61), P. furiosus (62), Thermotoga maritima (63), Thermococcus litoralis (64), E. coli (65), and B. taurus (66) provides a strong basis for functional assignments. The following analysis is based upon multiple sequence alignments of the S. clavuligerus sequence with the foregoing sequences in which the three-dimensional structure has been established and in which the role of conserved residues has been identified in detail. Fig. 6 shows the approximate boundaries of subdomain II that specifies the dinucleotide-binding region (yellow). This is flanked upstream by the major portion of subdomain I (gray), the glutamate-binding region. On the carboxyl-terminal side of subdomain II is the remainder of subdomain I (helix ␣-17 in C. symbiosum). The essential "core" GDH is defined by the common critical elements preserved in the larger superfamily (Glu/Leu/Phe/Val dehydrogenase) that include leucine dehydrogenase, valine dehydrogenase, and phenylalanine dehydrogenase (11)(12). The latter three closely related families lack both the amino-terminal distal and the carboxyl-terminal portions of subdomain I, which in C. symbiosum includes a bundle of ␣-helixes that pack together (the five amino-terminal ␣-helixes and the C-terminal ␣-helix).
Residues implicated in the catalytic mechanism are Lys-875 (Lys-113), Lys-885 (Lys-125), and Asp-951 (Asp-165). The corresponding residue positions of C. symbiosum are given in parentheses. Lys-875 recognizes the 1-carboxyl group of glutamate; Lys-885 enhances the nucleophilicity of the essential water molecule, and Asp-951 is involved in proton transfer to and from glutamate during catalysis. Both Arg-853 (Arg-93) and Lys-885 interact with Asp-951 to form highly conserved intrasubunit ion pairs. The latter salt bridges probably stabilize the three-dimensional structure around the active site (63).
Arg-849 (Lys-89), Gly-850 (Gly-90), Ala-950 (Pro-164), Val-1207 (Val-377), and Ser-1210 (Ser-380) function in the binding of side chain atoms. Note that Arg-849, which interacts with the side chain carboxyl of glutamate, is an arginine residue in all GDH 115 and GDH 180 proteins, whereas the equivalent residue in all GDH 50 proteins is lysine. The conservative replacement of the critical Lys-89 (C. symbiosum) residue by an arginine residue (Arg-849 in S. clavuligerus) might be expected to be accompanied by a parallel change in another member of the suite of residues which interact with the C-5 of glutamate. Table V shows the identity of five residues that have been shown to interact with the C-5 of glutamate in the two 50-kDa classes of GDH. Indeed, it appears that a conservative change in either of two residues can compensate for the alteration of Lys-89. In the GDH L_115 class the equivalent of Ala-163 is uniformly glycine. On the other hand in the GDH L_180 class the equivalent of Thr-193 is uniformly serine (see Fig. 7). Four of the five glycine residues that have been shown to influence the shape of the active-site pocket are conserved as follows: Gly-850 (Gly-90), Gly-851 (Gly-91), Gly-883 (Gly-123), and Gly-1206 (Gly-376).
A classical ␤␣␤ fold for dinucleotide binding is shown near the amino-terminal portion of subdomain II. The typical GXGXXG motif is replaced by GXGXXS in S. clavuligerus. Overall in the L_180 group this motif is either GXGXXS or GXGXXA. Ahead of the aforementioned Rossmann fold is a conserved threonine 992 (Thr-209) that influences the conformation of the glycosidic bond of the nicotinamide ring. The latter together with Gly-1206 (Gly-376), Val-1207 (Val-377), and Ser-1210 (Ser-380) dictate the stereospecificity of the hydride transfer. In this region Asn-1203 (Asn-373) hydrogen bonds to Lys-875 (Lys-113) across the GDH domain interface.
We note that analysis of the active site residues in the Glu/Leu/Phe/Val dehydrogenase superfamily using the program PROSITE has resulted in identification of the consensus sequence (LIV)XXGG(SAG)KX(GV)XXX(DNST)(PL) (PS00074). The lysine residue corresponds to S. clavuligerus

TABLE V Comparison of critical C-5 carbon contacts of glutamate within different GDH classes
The 1st column includes both classes of GDH S_50, and residue numbers are those of C. symbiosum as the reference organism. The equivalent residue numbers are given in the 3rd column for S. clavuligerus in parentheses. Arrows indicate parallel conserved changes in the two GDH_L classes in comparison with the residues indicated in the 1st column. 50 residue 885 marked in Fig. 7. The GDH L_180 class does not fit the pattern perfectly, the corresponding residues being (I)XXvG(AS)KX(G)XXX(kNr)(rfqk). The lowercase residues represent ones that do not fit the consensus. An inspection of Fig. 7 shows that the region of greatest absolute conservation of neighboring residues is in a region of primary sequence where both glutamate and dinucleotide contacts are made. In this region (around S. clavuligerus residue 1210 marked in Fig. 7) the consensus pattern is NXXGVXXSXXE.
Beyond the Core GDH-The core GDH is quite well defined (61-66) with respect to the atomic interactions that dictate glutamate binding (within subdomain I) and dinucleotide binding (within subdomain II). Detailed structural information about interface regions that are responsible for oligomerization within subdomain I are also elucidated. Beyond the structural basis for catalytic competency and for oligomerization, some insight is emerging about the structural basis for regulatory properties. Different GDH types can be quite variable with FIG. 7. Multiple alignment of the core GDH domain. The four classes of the GDH family are shown in each frame as blocks from top to bottom: GDH L_115, GDH L_180, GDH S_50 II , and GDH S_50 I . The approximate beginning of the core GDH region is shown with a bent arrow. Residues conserved throughout the entire family are highlighted in yellow; those conserved throughout subfamily L_GDH are highlighted in pink; and those conserved throughout subfamily S_GDH are highlighted in blue. Residues that are located within 6 Å of any atom of the bound glutamate substrate or dinucleotide cofactor with respect to the P. furiosus and C. symbiosum sequences (71) are indicated at the bottom of each frame with E or OE symbols, respectively. The four residues indicated by q are located within 6 Å of both glutamate and dinucleotide. Asterisks associated with a corresponding residue number at the top refer to S. clavuligerus residues. The GXGXX(G/A/S) motif for the dinucleotide helix turn (Rossmann fold) is 1022 GXGXXS 1027 for S. clavuligerus. The equivalent but unaligned glycine residues in the GDH L_115 group where the motif is GXXGXXG (14) are joined with solid bent lines. respect to whether they are sensitive to allosteric control. Where such control exists, marked variety is evident with respect to the complexity of the control. The mammalian GDHs exemplify a case where complex effector control is accompanied by a somewhat larger size (about 10%) of the monomer (66). An insertion of about 40 amino acids is placed just beyond the carboxyl-terminal end of the core GDH region, the end point of the alignment shown in Fig. 7 (i.e. the novel insert begins at about residue 454 for the Homo sapiens GDH). This 48-amino acid insert has been described as an antenna that serves as an intersubunit communication conduit and intimately influences negative cooperativity and allosteric regulation promoted by GTP, NADH, and ADP.
Chlorella sorokiniana exemplifies another case where additional sequence is associated with regulatory complexity (33).
In this most fascinating of systems, an amino-terminal extension consisting of two additional ␣-helixes (relative to C. symbiosum) exist. Differential transcript splicing yields two subunit types, differing only at the amino terminus. The two homohexamer types differ dramatically in the affinity for ammonia in the reductive amination activity. Differential regulation of the relative levels of the two subunit types, derived from a single gene, provides the potential for the fine-tuned modulation of ␣-ketoglutarate amination to an impressive degree since all ratios of heterohexamers presumed to have intermediate catalytic properties can be generated. The function of the large amount of amino acid sequence that flanks the GDH domain of both GDH L_115 kDa and now GDH L_180 is completely unknown. Aside from the similarity with other members of the same GDH class, these regions showed no similarity Relationships between the Four Homolog Classes of GDH-To establish the phylogenetic relations between the different GDHs reported to date, a comparative study of all these proteins (Tables IV and VI) was performed. Fig. 8 illustrates an unrooted tree obtained by maximum parsimony in which the input sequences were restricted to the core GDH region (refer to Figs. 6 and 7). Fig. 9 shows an unrooted tree obtained by the neighbor-joining methodology of all complete sequences available that were not truncated and that did not contain obvious sequencing errors. Either tree shows a hierarchical grouping whereby two subfamily components of the GDH family each contain two classes of GDH. The most divergent members of GDH L_180 class exhibit about 65% identity, whereas the most divergent members of GDH L_115 class display about 50% identity. Particular sequences representing each of the latter two classes are related to one another at a level of only about 25% identity. Subfamily S_GDH also contains two classes, GDH S_50 I and GDH S_50 II . The most divergent members of GDH S_50 I class exhibit about 54% identity, whereas the most divergent members of GDH S_50 II class exhibit about 43% identity. Any given representative of each of the latter classes are related to one another at a level of about 30% identity. When representative members of each of the two GDH subfamilies are compared with one another, significant similarity is not evident when a routine comparison is performed (e.g. using NCBI's Blast 2 sequences), even though homology is clearly apparent based upon conservation of crucial amino acid residues (see Fig. 7). At difference with the analysis reported in the Fig. 7 (using the GDH motifs), Fig. 9 compares the complete sequences of the different GDHs found in the data bases. It can be observed that in both comparisons four different GDH groups were established, reinforcing the convenience of defining two subfamilies between GDHs, since the GDH motifs in S_50 I and S_50 II differ from each other just about as much as L_115 and L_180 (see Fig. 7).
Evolutionary History of GDH-Within subfamily L_GDHs, members of GDH L_115 are found only in fungi and protozoans, whereas the phylogenetic range of GDH L_180 is thus far restricted to bacteria. Within subfamily S_GDHs, members of GDH S_50 I are present in the bacteria, fungi, protozoans, and algae. Although higher plant chloroplasts apparently lack GDH, Chlorella possesses a thoroughly characterized chloroplast-localized species of GDH S_50 I (67). Members of the GDH S_50 II class are present in Archaea, Bacteria, mitochondria of vertebrates, and mitochondria of higher plants. Fig. 10 illustrates the phylogenetic distribution of GDHencoding genes in selected members of the Archaea and Bacteria that have been organized on a 16 S rRNA tree. Since the genomes of most of these organisms are completely sequenced, one can be relatively confident not only that representatives of every known homolog class has been detected but that GDH genes not found really are absent. C. sorokiniana represents the position of chloroplast 16 S rRNA. Mitochondrial 16 S rRNA (not shown) would be approximately at the divergence level of R. prowazekii. Class GDH S_50 II clearly enjoys the broadest level of phylogenetic distribution, being the only GDH representative present in Archaea. It follows that this is likely to have been the ancestral GDH species. The S_50 I species of GDH appeared early after divergence of the Bacteria and Archaea at about the level of Deinococcus (which has both paralogs). The joint presence of both paralogs is seen elsewhere, e.g. species of Neisseria and Bordetella. Although E. coli possesses It is not surprising that those organisms that have a minimal genome size that is associated with restricted metabolic repertoires (e.g. Mycoplasma, Ureaplasma, Chlamydia, Treponema, and Borrelia) lack all types of GDHs. In line with contemporary conclusions about genome reduction in such organisms, it seems certain that these organisms lost GDH genes in relatively recent times. Fig. 10 also shows the distribution of leucine/valine/phenylalanine dehydrogenases. These appear to be of relatively infrequent occurrence in the Bacteria. It is apparent from inspection of Fig. 8 that in the Bacteria there has been a dynamic progression of evolutionary events involving paralog acquisition and paralog loss. Bacillus subtilis possesses two copies of GDH S_50 II . Even closely spaced phylogenetic progressions such as the gamma group of Proteobacteria (from P. aeruginosa to Salmonella typhi in Fig. 8)  An examination of the multiple alignment shown in Fig. 8 shows that GDH L_115 and GDH L_180 are sister classes within subfamily L_GDH, whereas GDH S_50 I and GDH S_50 II are sister classes within subfamily S_GDH. These relationships are specifically tied to the GDH domain. The non-GDH sequence in the GDH L_180 class has no significant similarity with the non-GDH sequence in the GDH L_115 class. The functions of the non-GDH regions in both classes of L_GDH are completely mysterious at present, but they must have evolved independently from a common ancestor having the unique sequence elements of the L_GDH subfamily. Within both of the GDH L_180 and GDH L_115 classes, the GDH domain is distinctly more conserved than the non-GDH regions. It is interesting that GDH L_180 has retained the structural features that dictate the hexameric subunit configuration usually seen in the S_GDH subfamily, even though it has a much greater overall similarity with the GDH L_115 class of GDH that exhibits a tetrameric subunit configuration.
Proposed Evolutionary Scenario-The most ancient GDH species is likely to have been GDH S_50 II . It is broadly distributed throughout the Archaea and Bacteria. It also is present in Eukaryotes where higher plants and vertebrates express it as a mitochondrion-localized species. This presumably was acquired via endosymbiosis, probably from a member donor within the ␣ division of Proteobacteria. Following the divergence of Archaea and Bacteria, gene duplication produced GDH S_50 I at a time between the divergence of Thermotoga and Deinococcus. Even though the GDH species present in contemporary Chlorella chloroplasts and in contemporary Synechocystis are different class of GDHs, the tree dynamics are clearly consistent with the presence of the Chlorella-type GDH in an ancestral cyanobacterium that may have had both GDH S_50 classes. Indeed, cyanobacteria as a group are sufficiently diverse that one could reasonably expect to find both GDH S_50 classes in other modern genera of cyanobacteria whose genomes are not yet sequenced. The phylogenetic divergence of different cyanobacteria exceeds the phylogenetic breadth of the enteric bacteria, where a considerable diversity of GDH distribution is apparent (Fig. 10).
An active-site motif of the GDH domain (see S. clavuligerus residue marked 875 in Fig. 7 for reference) is LXXXQXXKN in the GDH S_50 I class (except for Agaricus bisporus and C. symbiosum, which exhibit LXXXQXXKD). This motif varies in the GDH domain of class GDH S_50 II, where it is LXXXMXXKc/t (except for S. sulfataricus which is LXXXMXXKN). We suggest that this motif is a molecular flag (as depicted by the distribution of motif variation in Fig. 8) that reflects the evolutionary steps leading to the contemporary GDH phylogenetic distribution (Fig. 10). The L_GDH subfam- FIG. 9. Unrooted bootstrap tree for GDH by the neighbor joining method. The tree was constructed with the software package PHYLIP. The blue branch indicates the ␣ 4 -115 class of GDHs; the red branch represents the ␣ 6 -180 class, and the black branch shows the ␣ 6 -50 subfamily of GDHs (two different groups could be observed, from the enzyme numbers 18 -46 and 47-77). The numbers correspond to the GDHs included in Tables IV and VI. ily uniformly exhibits the LXXXQXXKN motif present in the GDH S_50 I class, and thus, we propose that GDH S_50 I was the immediate precursor to the L_GDH subfamily. Subsequent fusions led to the increased sizes of the modern L_GDH subfamily; these fusions must have been of independent origin since no similarities of the non-GDH regions are apparent. GDH L_115 may have been acquired by lower eukaryotes and protozoans via endosymbiosis since nuclear genes with this origin do not necessarily make gene products targeted to intracellular organelles (68). So far, no member of the Bacteria has been found to possess a gene corresponding to the likely evo-lutionary intermediate, i.e. encoding an approximately 50-kDa GDH that corresponds more closely to the GDH domain of the L_GDH subfamily rather than of the S_GDH subfamily.
The ancestral GDH S_50 II species may have possessed broad substrate specificity. Indeed, broad substrate specificity is a property of some contemporary species of GDH S_50 II (e.g. bovine GDH) (see Ref. 69 and references therein). The leucine/ valine/phenylalanine dehydrogenase families may have originated from GDH S_50 II in the Gram-positive lineage via gene duplication followed by substrate specialization as formulated by the recruitment hypothesis (70). Consistent with this is the presence of LXXXMXXKT as an active-site motif for all leucine/ valine/phenylalanine dehydrogenases ( Fig. 7 and 8).
To summarize, this analysis reveals the following: (i) the nitrogen metabolism of S. clavuligerus involves a new type of GDH (genetically and biochemically characterized) that requires a new classification for the different types of GDHs; (ii) all GDHs reported up to date belong to two different subfamilies (small GDHs and large GDHs); (iii) each subfamily subdivides into two sister classes (S_50 I or S_50 II and L_115 or L_180); and (iv) ␣ 4 -115 and ␣ 6 -180 may have had a closer evolutionary relation with the 50 I (␣ 6 ) class (enzymes 18 -46) than with the 50 II (␣ 6 ) class (enzymes 47-77) (see Table VI).