The Crystal Structure of Aquifex aeolicus Prephenate Dehydrogenase Reveals the Mode of Tyrosine Inhibition*

TyrA proteins belong to a family of dehydrogenases that are dedicated to l-tyrosine biosynthesis. The three TyrA subclasses are distinguished by their substrate specificities, namely the prephenate dehydrogenases, the arogenate dehydrogenases, and the cyclohexadienyl dehydrogenases, which utilize prephenate, l-arogenate, or both substrates, respectively. The molecular mechanism responsible for TyrA substrate selectivity and regulation is unknown. To further our understanding of TyrA-catalyzed reactions, we have determined the crystal structures of Aquifex aeolicus prephenate dehydrogenase bound with NAD+ plus either 4-hydroxyphenylpyuvate, 4-hydroxyphenylpropionate, or l-tyrosine and have used these structures as guides to target active site residues for site-directed mutagenesis. From a combination of mutational and structural analyses, we have demonstrated that His-147 and Arg-250 are key catalytic and binding groups, respectively, and Ser-126 participates in both catalysis and substrate binding through the ligand 4-hydroxyl group. The crystal structure revealed that tyrosine, a known inhibitor, binds directly to the active site of the enzyme and not to an allosteric site. The most interesting finding though, is that mutating His-217 relieved the inhibitory effect of tyrosine on A. aeolicus prephenate dehydrogenase. The identification of a tyrosine-insensitive mutant provides a novel avenue for designing an unregulated enzyme for application in metabolic engineering.

The TyrA protein family consists of dehydrogenase homologues that are dedicated to the biosynthesis of L-tyrosine. These enzymes participate in two independent metabolic branches that result in the conversion of prephenate to L-tyrosine, namely the arogenate route and the 4-hydroxyphenylpyruvate (HPP) 3 routes. Although both of these pathways utilize a common precursor and converge to produce a common end-product, they differ in the sequential order of enzymatic steps. Through the HPP route, prephenate is first decarboxylated by prephenate dehydrogenase (PD) to yield HPP, which is subsequently transaminated to L-tyrosine via a TyrB homologue (8). Alternatively, through the arogenate route, prephenate is first transaminated to L-arogenate by prephenate aminotransferase and then decarboxylated by arogenate dehydrogenase (AD) to yield L-tyrosine (9 -11) (see Fig. 1A).
There are three classes of TyrA enzymes that catalyze the oxidative decarboxylation reactions in these two pathways. The enzymes are distinguished by the affinity for cyclohexadienyl substrates. PD and AD accept prephenate or L-arogenate, respectively, whereas the cyclohexadienyl dehydrogenases can catalyze the reaction using either substrate (12).
To ensure efficient metabolite distribution of the pathway intermediates, TyrA enzymes are highly regulated by various control mechanisms, including feedback inhibition, and genetic regulation by the Tyr operon (13)(14)(15)(16). In some cases, L-tyrosine competes directly with substrate, be it prephenate or L-arogenate for the active site of arogenate or cyclohexadienyl dehydrogenases (14,(17)(18)(19). The product HPP can also serve as an efficient competitive inhibitor with respect to prephenate (20). Additionally, at the protein level PDs are only shown to be regulated at distinct allosteric sites or domains to modulate their activity. For example, the results of kinetic studies on the bifunctional Escherichia coli chorismate mutase-prephenate dehydrogenase (CM-PD) have indicated that this enzyme likely possesses a distinct allosteric site for binding tyrosine (21). In contrast, the Bacillus subtilis PD is the only enzyme reported to be competitively inhibited by HPP and L-tyrosine but is also noncompetitively inhibited by L-phenylalanine and L-tryptophan (12,22). Additional regulatory control is thought to originate through a C-terminal aspartate kinase-CM-TyrA domain of the B. subtilis PD (23).
Biochemical analyses of PD from E. coli CM-PD have provided a framework for understanding the molecular mecha-nism of the TyrA enzymes. The E. coli PD-catalyzed reaction proceeds though a rapid equilibrium, random kinetic mechanism with catalysis as the rate-limiting step (24). Additionally, studies of the pH dependence of the kinetic parameters V and V/K indicate that a deprotonated group facilitates hydride transfer from prephenate to NAD ϩ by polarizing the 4-hydroxyl group of prephenate, whereas a protonated residue is required for binding prephenate to the enzyme⅐NAD ϩ complex (25). The conserved residues His-197 and Arg-294 have been identified through extensive mutagenesis studies to fulfill these two roles (26,27). Further analyses of the activities of wild-type protein and site-directed variants in the presence of a series of inhibitory substrate analogues support the idea that Arg-294 binds prephenate through the ring carboxylate (26).
The structures of AD from Synechocystis sp. and PD from Aquifex aeolicus (both in complex with NAD ϩ ) have been reported by Legrand et al. (28) and by our group (29), respectively. Analyses of these structures have provided structural information on the conserved histidine and arginine residues. The structure A. aeolicus PD has also led to the identification of other active site residues that may play a role in enzyme catalysis, most notably Ser-126, which we propose facilitates catalysis by orienting the catalytic histidine and the nicotinamide moiety of NAD ϩ into their catalytically efficient conformations. Ambiguities can arise from examination of the binary complexes, because prephenate has only been modeled in the active site. For example, analysis of the AD structure by Legrand et al. (28) places Arg-217 (equivalent to Arg-294 in E. coli and Arg-250 in A. aeolicus) too far from the active site to play a role in prephenate binding. Thus, the full complement of interactions between prephenate and TyrA proteins are still largely unknown, as are the interactions of the enzymes with L-tyrosine.
To further investigate the importance of residues involved in ligand binding, specificity, and catalysis, we have carried out co-crystallization studies of A. aeolicus PD with NAD ϩ and prephenate, with NAD ϩ and 4-hydroxyphenylpropionate (HPpropionate), a product analogue, and with NAD ϩ and L-tyrosine. Accordingly, this study provides the first direct evidence that L-tyrosine binds to the active site of a prephenate dehydrogenase. We have investigated the role of Ser-126, His-147, His-217, and Arg-250 through the kinetic analysis of site-directed mutants and structural analysis of the co-crystal complexes. To understand the role of active site residues in substrate selectivity, comparative structural analysis of AD and PD was also conducted. The current study provides a basis for understanding the mechanism of substrate selectivity between the different classes of TyrA enzymes and details how A. aeolicus PD can accept prephenate as substrate and L-tyrosine as a competitive inhibitor.

MATERIALS AND METHODS
Chemicals and Reagents-Prephenate was prepared as described previously (30), while NAD ϩ (free acid) was obtained from Roche Applied Science. L-Tyrosine was from ICN, whereas HPP and HPpropionate were from Aldrich. The keto form of HPP was prepared as outlined in Lindblad et al. (31). All other reagents were of molecular biology grade and were pur-chased from Sigma, Bioshop, or BDH. Oligonucleotides used for site-directed mutagenesis were purchased from Integrated DNA Technologies (Coralville, IA).
Site-directed Mutagenesis-The expression plasmid (⌬19PD) encoding residues 20 -311 of the A. aeolicus VF5 PD protein (AAC07589 from Genbank TM ) has been previously described (19). Site-directed mutagenesis was carried out using the QuikChange site-directed mutagenesis kit (Stratagene), whereby complimentary oligonucleotides containing the desired mutations for ⌬19PD were used. Table 1 summarizes the mutants generated and the respective oligonucleotides used for mutagenesis. All ⌬19PD mutants were verified by DNA sequencing.
Protein Expression and Purification-All constructs of A. aeolicus VF5 PD were expressed and purified as previously described (29) with the following modifications. Cells harboring recombinant mutant ⌬19PD were disrupted by French press followed by sonication, and the heat treatment step was omitted. Following chromatography of the thrombin-treated enzyme on nickel-nitrilotriacetic acid resin, PDs were subjected directly to size exclusion chromatography on a Superdex 200 column.
Determination of Enzyme Activity and Dissociation Constants for Ligand Binding-Enzyme activity of A. aeolicus PD in the presence of NAD ϩ and prephenate was monitored in a 1-ml reaction cuvette containing 50 mM HEPES and 150 mM NaCl at pH 7.5 as previously described (19). Briefly, the reaction mixture was incubated at 55°C for 2 min, and then the reaction was initiated by the addition of enzyme. The production of NADH was followed at 340 nm spectrophotometrically. The turnover numbers and Michaelis constants for substrates were obtained by fitting initial velocity to the Michaelis-Menten equation. Inhibition constants (K i ) for the dissociation of HPP, HPpropionate, or L-tyrosine from the enzyme⅐NAD ϩ complex were obtained by fitting initial velocity data to the equation for linear competitive inhibition. These data were obtained by varying prephenate at four concentrations ranging from below to above the K m (in the presence of 2 mM NAD ϩ ) and HPP from 0.2 to 1.0 mM and L-tyrosine ranging from 0 to 5 mM. Percent residual activity as a function of L-tyrosine concentration was monitored as described by Bonvin et al. (19). All kinetic data were fitted to the appropriate rate equations by using the computer programs of Cleland (32) or GraFit (Version 5.0, Leatherbarrow). Changes in fluorescence emission as a function of prephenate concentration were used to determine dissociation constants of the prephenate from the binary complex as described by Bonvin et al. (19). Protein concentration was Crystallization of PD-The initial crystallization condition was determined with a sparse matrix crystallization (Hampton Research Crystal ScreenI TM ) screen at room temperature using the hanging drop, vapor-diffusion technique. The optimized co-crystallization condition consists of 48% 2-methyl-1,3 propanediol and 100 mM HEPES at pH 7.8, and co-crystals were obtained by supplementing the protein solution (at a concentration of 12 mg/ml) with 5 mM NAD ϩ and 10 mM of HPpropionate or prephenate, or 2.25 mM L-tyrosine.
X-ray Diffraction and Structure Determination-X-ray diffraction data were collected from single crystals at a temperature of 100 K in a nitrogen stream on beamline SBC19 at a wavelength 0.9794 Å at the Argonne National Laboratory, Advanced Photon Source. The diffraction data were processed and scaled with the HKL3000 suite of programs (DENZO/SCALEPACK) (33). The enzyme-ligand structures were sufficiently isomorphous with the previously described ⌬19PD⅐NAD ϩ -complexed structure (29) to allow the immediate use of the RIGID routine of CNS (Crystallography and NMR system) (34), to correctly place the search model. Model visualization and rebuilding were done with the program O (35).
The remainder of the model was manually built with O, and simulated annealing refinement was subsequently conducted with CNS after every round of model building. In the later stages of model rebuilding and structure refinements, Coot and Refmac with TLS refinement were, respectively, used for the ⌬19PD⅐NADH⅐HPP, ⌬19PD⅐NAD ϩ -HPpropionate, and ⌬19PD⅐NAD ϩ -L-tyrosine structures, until each of the model converged (36 -38). All molecules of PD were rebuilt prior to every round of refinement, and non-crystallographic symmetry constraints were not applied during refinement. NAD ϩ molecules and ligands were fitted to the unaccounted electron density in each molecule in the tetramer after the second round of refinement. PDB, topology, and parameter files for NAD ϩ , tyrosine, HPpropionate, and HPP were obtained from the HIC-up server (41). Representative figures from the crystal structure were produced with PyMOL.

RESULTS AND DISCUSSION
The primary objectives of this study were to identify active site residues that are directly involved in the catalytic and reg-ulatory mechanisms of PD and to determine the mechanism that contributes to substrate selectivity among enzymes in the TyrA family. Co-crystallization was conducted in the presence of NAD ϩ and either prephenate, HPpropionate, or L-tyrosine. Sitedirected mutagenesis studies targeted residues Ser-126, His-147, His-217, and Arg-250, which are considered important for either the enzyme-catalyzed reaction or for conferring ligand selectivity.
Crystallization and Structural Summaries of ⌬19PD⅐ NADH⅐HPP, ⌬19PD⅐NAD ϩ -HPpropionate, and ⌬19PD⅐NAD ϩ -L-Tyrosine-Crystals of ⌬19PD⅐NAD ϩ -HPP, ⌬19PD⅐NAD ϩ -HPpropionate, and ⌬19PD⅐NAD ϩ -L-tyrosine grew under identical precipitant and pH crystallization conditions. These crystals belong to space group P2 1 2 1 2 1 , and their respective structures were determined by molecular replacement. All atomic models display excellent overall stereochemistry as judged by the Ramachandran plot; for the structures of ⌬19PD complexed with NADH-HPP, NAD ϩ -HPpropionate, and NAD ϩ -L-tyrosine Ͼ99% of the residues are in the allowed regions. The structure of ⌬19PD⅐NADH⅐HPP, ⌬19PD⅐NAD ϩ -HPpropionate, and ⌬19PD⅐NAD ϩ -L-tyrosine were refined to 2.15, 2.25, and 2.0 Å, respectively. Data collection and refinement statistics are summarized in Tables 2 and 3, respectively. Although continuous, interpretable electron density was observed for the vast majority of these structures, the following residues were excluded in the model due to poor or absent densities: for the ⌬19PD⅐NADH⅐HPP structure, residues 20 -25 and residues 310 -311 for chain A, residues 20 -27 and 310 -311 for chain B, residues 20 -24 and 306 -311 for chain C, residues 20 -26 and 309 -311 for chain D; ⌬19PD⅐NAD ϩ -HPpropionate crystal structure, residues 20 -25 and 311 in Chain A, 20 -27 and 310 -311 in Chain B, 20 -24 and 311 in Chain C, and 20 -26 and 309 -311 in Chain D; for the ⌬19PD⅐NAD ϩ -L-tyrosine crystal structure, residues 20 -25 and 311 in Chain A, 20 -29 and residues 308 -311 in Chain B, 20 -29 and 306 -311 in Chain C, and 20 -26 and 311 in Chain D. Poor quality of electron density in these regions of the maps contributed to the higher than normal refinement factors, R free and R work ( Table 3).
Crystals of ⌬19PD⅐NAD ϩ -HPpropionate and ⌬19PD⅐ NAD ϩ -L-tyrosine were obtained from co-crystallization studies with 5 mM NAD ϩ and 10 mM of HPpropionate or 2.25 mM L-tyrosine, respectively. Interestingly, the ⌬19PD⅐NADH⅐HPP crystals were obtained from co-crystallization experiment containing 5 mM NAD ϩ and 10 mM prephenate. The presence of NADH in the protein crystal was determined spectrophoto-  MAY 8, 2009 • VOLUME 284 • NUMBER 19

Structure of Prephenate Dehydrogenase
metrically at 340 nm. The presence of HPP and NADH instead of prephenate and NAD ϩ , in the structure, indicates that prephenate was enzymatically converted to HPP during the co-crystallization studies. Fig. 1B shows the chemical structure of each representative ligand used in structure determination of ⌬19PD. All ligand types were located in the F o Ϫ F c difference electron density map after the initial round of refinement. Accordingly, these molecules were built into the model prior to the second round of refinement (Fig. 2, A-C).
Only one molecule of the product or product analogue is consistently identified per dimer. In contrast, both subunits in the dimer contain a molecule of NAD ϩ or NADH. For example, for each dimer of the ternary complex ⌬19PD⅐NADH⅐HPP, one subunit contains a molecule of NADH and one molecule of product (we define this as a "paired" occupancy), whereas the other subunit contains only one molecule of NAD ϩ , but no product (we defined this as the "unpaired" occupancy). It is unlikely that interdimer interaction in the crystal packing prevents the binding of HPP to both binding sites simultaneously, because the structure was obtained from co-crystallization studies. The presence of NAD ϩ in the active site suggests that NAD ϩ binds first followed by prephenate; similarly, HPP needs to be released before NADH from the active site. This is highly speculative: this model is supported by the fact that we never identified that HPP, in the absence of NADH, occupied the active site. It is also likely that NAD ϩ binding is required for crystallization. The observation of paired and unpaired occupancy of product analogue in the dimer is indicative that substrate binding and product release are ordered, which is consistent with kinetic data obtained for arogenate dehydrogenase from Synechosystis sp. in which arogenate preferentially binds first (14). However, this finding contrasts the random kinetic mechanism proposed from the analysis of initial velocity and dead-end and product inhibition studies of PD and AD from a number of organisms, including E. coli and Arabidopsis thaliana (17,19,24,28). Conformational Shifting upon Substrate Binding-Superimposition of the crystal structure of PD indicated that substrate binding induces a global conformational change on the dimer. The dimeric protein is in a closed conformation with only NAD ϩ bound and the binding of HPP or a product analogue moves the two subunits away by about 5Å (Fig. 3, A  and B, inset). This global conformational change is imparted mainly on the unpaired monomer. Therefore, the unpaired monomer is in the "open" conformation ( Fig. 3A). We speculated that this global conformational change could negatively influence the binding of ligands, product  The binding of the product analogue induces structural changes within the PD subunits (Fig. 3, A and B, inset). A com-parison of ⌬19PD⅐NAD ϩ ⅐tyrosine structure with the ⌬19PD⅐NAD ϩ structure indicated that the overall structures of the subunits are similar to each other; however, in all cases the same conformational changes were observed in regions that comprise the active site (Fig.  3B, inset).
The most notable shift occurs in the loop located between residues 149 and 156, a region which defines the hydrophobic region, or the "wall," of the active site. A C ␣ trace of this region reveals that Gly-151 shifts 2.5-3.4 Å away from this pocket upon substrate binding. The inclusion of HPP, HPpropionate, or L-tyrosine molecule in the active site demonstrates that this shift is necessary to avoid steric clash between Gly-151 and the side chain of the product or product analogues (Figs. 3B (inset), 4A, 4B, and 4C). In addition, this region contains Glu-153, a residue that is suggested to participate in the gated mechanism through coordination of an ionic network with Asp-247 and Arg-250. Moreover, we have proposed, based on our previous structural analyses, that Glu-153 modulates binding through this ionic network by regulating substrate access to the active site (29). Positional analysis of the Glu-153 side chain reveals  that the ionic interaction with Arg-250 is maintained with substrate binding, whereupon the release of the product this interaction is lost.
Other secondary structures that have shifted upon ligand binding include ␣8, ␣9, ␣10, and ␣11 (residues 214 -266). Two functionally important residues, His-217 and Arg-250, are contained within these regions. With respect to the binary subunit, the side chain of Arg-250 in the paired subunit is shifted 1.3 Å closer to the active site pocket and closer to the side-chain carboxylate of the respective ligand.
Location of the A. Aeolicus PD Active Site-A comparison of the structures of A. aeolicus PD with each of the three ligands, HPP, HPpropionate, and L-tyrosine, helps to identify residues involved in substrate binding and catalysis, and in the regulation of enzyme activity by inhibitor, L-tyrosine. All of the ligands studied have a common C4-hydroxyl group, but they vary in the properties of the side chain at the C1 position. These variations at the side chain dictate the binding properties of the different ligands to A. aeolicus PD (Fig. 1B).
HPP, the immediate product, acts as a linear competitive inhibitor with respect to prephenate for the reaction catalyzed by ⌬19PD (data not shown). The dissociation constant of HPP from the enzyme⅐NAD ϩ complex of 118 Ϯ 14 M is essentially identical to the Michaelis constant for prephenate (K m of 135 Ϯ 12 M) even though HPP is aromatic rather than a cyclohexadiene and lacks the ring carboxyl group associated with prephenate (Table 4). Based on the structural similarities between these two compounds, we propose that the interactions between HPP and active site residues of PD should closely reflect those with prephenate. Initial velocity data obtained by varying prephenate (102-680 M) in the presence of 2 mM NAD ϩ , and increasing concentrations of L-tyrosine also fit well to the equation for linear competitive inhibition (data not shown), with a K i of 15.9 Ϯ 1.3 M for tyrosine.
Architecture of the Substrate Binding Site-The position of the active site of A. aeolicus PD is consistent with our previous report from modeling with prephenate (29). The active site lies adjacent to the nicotinamide moiety of NAD ϩ , which is expected for efficient hydride transfer from prephenate to the C4 position of the nicotinamide ring. Amino acids from both subunits of the dimer contribute to the prephenate binding site; however, a large portion of the binding pocket is contained within one subunit, specifically at the inter-domain cleft (29).
The active site can be arbitrarily divided into three regions based on residue composition and interactions with the product analogues; a polar region consisting of residues His-147, Ser-126, His-214, Ser-213, Phe-209, and His-205, a hydrophobic region consisting of residues Ile-149, Ala-150, Gly-151, Thr-152, His-217, Phe-221, Met-258, and Trp-259, and an ionic region consisting of residues Glu-153, Arg-250, Ile-251, and Asp-247Ј (Figs. 4A, 4B, 4C, and 5A). Collectively, the physiochemical properties of these three regions in the active site reflect the interactions with the different groups of the substrate. The C-4 hydroxyl group of the ligand is located in the polar region, whereas the side chain participates in interactions with residues in the ionic region.
Role of His-147 in the Reaction Mechanism-Interactions between the C4-hydroxyl group of the different ligands (HPP, HPpropionate, and L-tyrosine) and active site residues are conserved. The C-4 hydroxyl hydrogen bonds with His-147, a highly conserved catalytic group, which is part of a hydrogen bonding network between Ser-126 and Ser-213 and a highly conserved water molecule (WAT1) (Figs. 2A, 2B, 2C, 4A, 4B, and 4C). Additionally, C4 of the ligand is also located within 2.5 Å from the N1 portion of the nicotinamide ring of NAD ϩ . These interactions are consistent with kinetic data from E. coli CM-PD, which implicates His-197 (corresponding to A. aeolicus His-147) as a key catalytic group. As observed for the E. coli enzyme, pH activity profiles of ⌬19PD revealed that a deprotonated group, most likely His-147 with a pK of ϳ6.8, is required for catalysis (data not shown). Kinetic analysis was also conducted on the A. aeolicus PD H147N mutant to confirm its role in catalysis ( Table 4). The H147N mutant is essentially inactive but binds prephenate with apparent affinity similar to the wildtype enzyme, as determined by its kinetic parameters and by thermodynamic measurements of the quenching of tryptophan fluorescence emission by prephenate (data not shown) ( Table  4). These findings further support the catalytic role of His-147 in polarization the C4-hydroxyl group of prephenate to facilitate hydride transfer from prephenate to NAD ϩ .
Role of Ser-126 in the Reaction Mechanism-The role of Ser-126 was further investigated by mutagenesis and kinetic analyses. Ser-126 coordinates an H-bonding network with His-147, the N1 atom of NAD ϩ and C4-hydroxyl group, and thus may play a role directly in catalysis by bringing together these groups in a catalytically competent conformation. In addi- a Values were calculated from initial rates using a minimum of five NAD ϩ concentrations ranging from one-half K m up to 10-fold K m for the variable substrate. When prephenate was the variable substrate NAD ϩ was fixed at 2 mM, while when NAD ϩ was the variable substrate, prephenate was fixed at 11 mM (WT, S126A), 18 mM (H217A), 16 mM (H217N), and 6 mM (R250Q). b NAD ϩ was kept at 2 mM, and prephenate was varied from 50 to 600 M using 630 g of protein. Parameters for NAD ϩ were not determined. c ND, not determined. tion, this H-bonding interaction could also contribute to the binding of the ligands studied and prephenate in the active site. As expected, S126A yielded both a 15-fold reduction in k cat , a 10-fold increase in the K m value for prephenate, and just a 2-fold increase in K i , 37.1 Ϯ 5.0 M for tyrosine, whereas the binding of NAD ϩ remained unchanged (Table  4). This observation is consistent with the proposed role of Ser-126 in coordinating the H-bonding interaction with prephenate and A. aeolicus PD in a catalytically competent geometry (Fig. 4A).
Role of Wat-1 in the Reaction Mechanism-Wat-1 is a highly conserved water molecule that participates in the H-bonding network with the ligand. Interestingly, this water molecule is only observed in the ternary complex of the enzyme with NAD ϩ and the ligands studied. Specifically, Wat-1 is shown to bridge the interaction between Ser-113, Ser-126, and the C-4 hydroxyl of the ligand (Figs. 2A, 2B, 2C, 4A, 4B, and 4C). We propose that Wat-1 may serve two mechanistic roles: Wat-1 may be participating solely in the binding interaction with the ligand and/or Wat-1 could participate in modulating the properties of His-147 in the catalytic mechanism.
Role of Arg-250 in the Reaction Mechanism-The carboxylate of the propionyl side chain is conserved between, HPP, HPpropionate, and L-tyrosine (Fig. 1B). Analysis of the ⌬19PD structure in complex with each of the three ligands revealed that this carboxylate directly interacts with the guanidinium group of Arg-250 and is positioned in proximity of this group by an ionic network (Fig. 4, A-C).
In these PD-ligand structures, Arg-250 is ordered with excellent electron density. In the absence of a ligand, Arg-250 is disordered judging by the lack of its representative electron density thus suggesting that it has an important binding role. Kinetic analysis of the R250Q mutant displayed a 10-fold increase in the K m for prephenate and a 20-fold increase in K i , 370 Ϯ 42 M, for tyrosine relative to the wild-type enzyme without significant change in the enzyme's affinity for NAD ϩ or its turnover rate (Table 4). Together these findings are in agreement with Arg-250 being important, but it is not critical, for the binding of the different ligands via the side-chain carboxylate group. Mutating Arg-250 did not eliminate prephenate or tyrosine binding. This is indicative that one or more other active site residues are also important for ligand binding.
Role of His-217 in the Reaction Mechanism-The most important differences between the C1-side chain of the ligands are: HPP has a keto group in addition to the conserved carboxyl, L-tyrosine possesses an amino group, and HPpropionate is lacking a group at the corresponding position. The structure of the ⌬19PD⅐NADH⅐HPP complex shows that the keto group on this propionyl side chain can potentially make H-bonding interaction with the main-chain amide of Gly-244 and is also positioned near a H-bonding network, which is mediated by a water molecule, Wat-2 (Fig. 4A). This H-bonding network includes N⑀2 of His-217, -OH of Ser-254, the keto and the carboxyl groups of HPP propionyl side chain and Wat-2 (Fig. 4A). The N␦1 of His-217 is also H-bonding to the main-chain carbonyl of Ser-213. These H-bonding interactions by both N␦1 and N⑀2 of His-217 would indicate that it is in the protonated state. To further understand the role of His-217, H217A and H217N mutants were designed and kinetically analyzed. Substitutions of His-217 to either alanine or asparagine produce mutant enzymes that are not inhibited by tyrosine but also have significant changes in their kinetic parameters for the ⌬19PD-catalyzed reaction. The K m for prephenate was increased, by 40-fold and 30-fold, for the H217A and H217N mutants, respectively, thus indicating the importance of His-217 in prephenate selectivity (Table 4). However, these substitutions also coincided with 10-to 20-fold decrease in k cat thus indicating that this interaction with the keto group of the substrate may assist in HPpropionate makes similar interactions as HPP except that the propionyl keto is absent and no interaction is observed with Wat-2 and the propionyl side chain. C, PD-NAD ϩ -tyrosine (silver)-bound structure. Tyrosine makes similar interactions at the C4 -OH and the propionyl carboxylate as HPP except that the amino group is interacting with the main-chain carbonyl of Thr-152.
positioning prephenate in a catalytically competent conformation. This change in kinetic properties for both H217A and H217N mutants can also be attributed in part to structural perturbation on the active site presumably due to disruption of the hydrophobic stacking of His-217 with the neighboring Trp-259 and Ile-251. We observed that Ile-251 and Trp-259 participate in a stacking interaction with His-217, and this interaction is lost when the ligand is not present (Fig. 4, A and B). We also determined that the binding of NAD ϩ has improved, and the K m for NAD ϩ was lowered by 7-fold for both His-217 variants ( Table 4). Mutation of the equivalent residue (His-257), based on sequence alignment, in E. coli CM-PD (H257A) had very little effect on both the K m for prephenate, a 2-fold increase, and the k cat , a 3-fold decrease, with no effect on NAD ϩ binding (27). Interestingly, this E. coli H257A mutant is also insensitive to tyrosine, even at high millimolar concentrations of tyrosine, and no inhibition was observed (data not shown). The inability of tyrosine to inhibit the E. coli H257A mutant provides further supports for the role of this conserved histidine in tyrosine binding. Tyr-303 from E. coli CM-PD aligned with Trp-259 from A. aeolicus PD in a primary sequence, indicating that the composition of these two active sites is not completely conserved. Therefore, such variation in the active site residues among PDs from different organisms could explain the varying levels of perturbations in kinetic properties between the H217A and H257A mutants from A. aeolicus PD and E. coli CM-PD.
The importance of His-217 is further illustrated by the co-crystal structure of HPpropionate with ⌬19PD. HPpropionate lacks the side-chain keto group and as such cannot participate in the hydrogenbonding interaction at this position (Fig. 4B). Superimposition of HPP and HPpropionate in the active site clearly show that the propionyl side chains are not superimposable; the HPpropionate side chain is slightly shifted closer to His-217 (Fig. 5A). Kinetic constants for the interaction of HPP to E. coli PD is ten times better than that for HPpropionate, 0.18 mM versus 1.8 mM (26). These findings are in agreement with our hypothesis that one or more residues in the microenvironment of His-217 are functioning as a determinant for the selective binding of ligands to PD active site. Ligands lacking the keto group can still bind but with lower affinity because of increased dynamics of their side chain.
The structure of ⌬19PD in complex with tyrosine revealed a different set of H-bonding interactions with the amino group of tyrosine compared with those with the keto group of HPP (Fig. 4, A and C). In this complex, the amino group of L-tyrosine is pointing in the opposite direction compared with that of the corresponding keto group of HPP. As a result, the amino group of L-tyrosine is directed away from His-217 and from the Gly-244 amide backbone by approximately 180 degrees and instead is interacting with the mainchain carbonyl group of Thr-152 (Figs. 4A, 4C, and 5A). Superimposition of the ligands in the active site of PD revealed that the side chain of L-tyrosine has been shifted away from His-217 by 3.2 Å, compared with the propionyl side chain of HPP, and the resulting distance between the amino group of the alanyl side chain and His-217 is 4.1 Å.
H217A/N Mutants Are Insensitive to Tyrosine Inhibition-Superimposition of the three ligands in the A. aeolicus PD active site revealed that the propionyl side chains adopt a different conformation depending on the presence of a keto, HPP, or an amino group, tyrosine. This observation in conjunction with the kinetic studies of the His-217 mutants led us to speculate that the micro- , and tyrosine (silver) illustrates the substrate-specific interactions. The keto group of HPP is directly pointing toward His-217. The amino group of Tyr is pointing away from His-217 and is interaction with main-chain carbonyl of Thr-152. B, the PDH active site (yellow) with a bound HPP molecule (green) superimposed with the putative ADH active site (silver). The functionally important active site residues identified in PDH are conserved in ADH, in parentheses. A major difference between the two active sites is the presence/absence of His-217 and of bulky groups adjacent to the substrate-binding site. The absence of these bulky groups in the ADH active site produces a large pocket. C, a C␣ superimposition of the PDH (blue) and ADH (magenta) monomeric structures. environment around His-217 is important for tyrosine binding. We further speculated that the ionization state of His-217 and its interaction with other active site residues might dictate the nature of ligands that can bind to the active site of PD. Based on the H-bonding interactions at both N␦1 and N⑀2 discussed above for His-217, we speculated that the side chain of this histidine is in the protonated state. This hypothesis is consistent with our observation for the HPP-bound and L-tyrosine-bound structures, in which the keto group of HPP is pointing toward the N⑀2 of His-217 and the amino group of L-tyrosine is pointing away from the histidine and instead is interacting with the main-chain carbonyl of Thr-152 (Fig. 4C, 5A). It is likely that this charge state of His-217 would produce a repulsive effect, which directs the interaction between the main chain carbonyl of Thr-152 and the amino group of the bound tyrosine.
Tyrosine inhibition studies were conducted to evaluate the role of His-217 and other residues in the binding of L-tyrosine to PD. L-Tyrosine was shown to have an inhibitory effect on both wild-type enzyme and on S126A, but had a reduced inhibitory effect on the activity of R250Q (Fig. 6). R250Q activity is inhibited by L-tyrosine less effectively than the WT enzyme, because the interaction is absent between the carboxylate of the bound tyrosine and Arg-250. The most interesting finding though is the inhibitory effect of L-tyrosine was completely eliminated by mutating His-217 to either an alanine or an asparagine (Fig. 6). Indeed, the binding of prephenate is also affected by the alanine and asparagine substitutions, which may suggest that mutating His-217 could result in structural perturbation of the active site, which has a more pronounced effect on tyrosine binding. In addition, our results from mutational analysis of the E. coli enzyme show that the H257A mutation (equivalent to H217A in A. aeolicus) also completely eliminates feedback inhibition by L-tyrosine, and its Michaelis constant for prephenate is relatively unperturbed. We envision that removing His-217 eliminates the repulsive effect from the microenvironment, which is responsible for directing the interaction between the amino group of tyrosine and the main-chain carbonyl of Thr-152.
Structural Comparisons of AD and PD-The availability of an AD structure permits structural comparisons between L-aroge-nate-specific and prephenate-specific TyrA dehydrogenases for better understanding of the molecular mechanism responsible for substrate binding in the TyrA family of proteins. To this end, we compared the ternarycomplex⌬19PD⅐NADH⅐HPP, ⌬19PD⅐NAD ϩ -HPpropionate, and ⌬19PD⅐NAD ϩ -tyrosine structures with that of Synechocystis sp and superimposed the structure of ⌬19PD⅐NADH⅐HPP with the Synechocystis sp. AD structure (PDB ID: 2F1K) (28) (Fig. 5, A-C).
The monomers superposed over 272 equivalent C ␣ atoms and have an root mean square deviation value of 2.4 Å. The overall structures PD and AD are quite similar, because their secondary structures have been maintained (Fig. 5C). Regions exhibiting significant deviations include portions of the loop between ␤6 and ␣6 and helices ␣8, ␣9, ␣10, and ␣11 (28, 29) (Fig. 5C). Overall, the active site of AD is more open and accessible, relative to that of binary ⌬19PD⅐NAD ϩ complex (Fig. 5B). This is due to the fact that ␣10, ␣11, and the ␤6 -␣6 loop region, which comprise the base and wall of the pocket, respectively, are shifted ϳ3.1-6.5 Å away from the active site (Fig. 5C) (28,29). Analysis of the PD and AD active sites reveals that functionally important residues are conserved (Fig. 5, A and B). For example, the catalytic histidine, His-147 in PD, and the important binding arginine, Arg-250 in PD, are equivalently positioned in both structures. Moreover, the serine residues that are shown to bind to the C4-hydroxyl group of prephenate are also spatially conserved. Other common active site residues include Gly-151, Thr-152, His-205, Ser-213, and His-214 (numbering corresponds to the ⌬19PD from A. aeolicus) (Fig. 5B). Given the importance of the backbone carbonyl of Thr-152 in the binding of L-Tyr, perhaps this group may also bind the amine moiety of L-arogenate. The conservation of functionally important residues indicates that L-arogenate may bind to the AD active site in a similar fashion to that of prephenate in PD. Superimposing the HPP molecule from the ⌬19PD⅐NADH⅐HPP structure into the AD active site supports this hypothesis. Based on the latter study, we infer that it is indeed possible for L-arogenate to interact with Ser-92, His-112 and Arg-217, which correspond to Ser-126, His-147, and Arg-250 in ⌬19PD. Our proposed interaction with Arg-217 contradicts Legrand et al. (28) who have suggested that this arginine is too far from the active site pocket to play a role in substrate binding.
Despite these similarities in the active site, there are notable differences. Most notably, there is a large pocket adjacent to the superposed HPP molecule in the AD active site (Fig. 5, A and B). This pocket results from the absence of bulky Trp and His residues (Trp-259 and His-217 in PD) and the position of ␣10 and ␣11 in the AD structure (Fig. 5, A-C). These missing residues could provide the required binding interactions for prephenate. Interestingly, Synechocystis sp. AD is lacking a histidine corre-FIGURE 6. Effect of L-tyrosine on the prephenate dehydrogenase activity of A. aeolicus ⌬19PD. Proteins include: WT (F) and Ser-126Ala (f),His-217Ala (), His-217Asn (ϩ), Arg-250Gln (ࡗ). Reactions were performed at 55°C in 50 mM HEPES, 150 mM NaCl (pH 7.4) and using concentrations of prephenate at ϳ4 ϫ K m with NAD ϩ fixed at 2 mM. Activity is expressed as a percentage of the value obtained in the absence of tyrosine. Right graph, effect of tyrosine at higher concentrations up to 8.5 mM. WT ⌬19PD (F), His-217 3 Ala (), and Arg-250 3 Gln (ࡗ) variant proteins. Reactions were performed at 55°C as described above except that prephenate concentration for H217A/N was kept at ϳK m . sponding to His-217 and is not inhibited by tyrosine. This finding is completely in line with our speculative role of His-217 in conferring selectivity for tyrosine binding.
Biological and Biochemical Relevance-This study describes the functional role of active site residues of proteins in the TyrA family, mainly, residues that are involved in ligand selectivity, binding, and catalysis. Comparative structural analysis with existing structures in the PDB revealed that the catalytic histidine residue is conserved among proteins in the TyrA family. The arginine residue that is involved in binding the different ligands through the side-chain carboxylate is also conserved. The most significant finding, however, is that the mutation of His-217 prevents L-tyrosine inhibitions, and we speculated that it could dictate the selective binding of different ligands to the PD active site. Notably, the presence of His-217 allows L-tyrosine to function as a competitive inhibitor with respect to prephenate. Substitution of the histidine residue at position 217 completely eliminated the inhibitory effect of L-tyrosine on A. aeolicus PD activity. This novel finding has direct implications on future metabolic engineering of downstream pathways in plants. It is likely that tyrosine could function as a feedback regulator by inhibiting PD at elevated intracellular concentrations. The mutation of residue the equivalent His-217 could produce a constitutively active PD, which will result in the unregulated production of HPP and subsequently tyrosine. Tyrosine, aside from being an essential component in proteins, is known to be an essential precursor of the biosynthesis of benzylisoquinoline alkaloids. Benzylisoquinoline alkaloid is diversified toward the production of Ͼ2500 secondary metabolites, and in addition benzylisoquinoline alkaloid itself is used as a precursor for the synthesis of pharmaceuticals, including morphine and codeine (39).