Crystal Structure of Binary and Ternary Complexes of Archaeal UDP-galactose 4-Epimerase-like l-Threonine Dehydrogenase from Thermoplasma volcanium*

Background: The catalytic mechanism of UDP-galactose 4-epimerase-like l-threonine dehydrogenase (GalE-like l-ThrDH) is unknown. Results: Four crystal structures of archaeal GalE-like l-ThrDH in the presence of NAD+, an inhibitor, and two substrates were determined. Conclusion: Tyr137 is essential for substrate binding and catalysis by the enzyme. Significance: This study provides new insight into substrate recognition by GalE-like l-ThrDH. A gene from the thermophilic archaeon Thermoplasma volcanium encoding an l-threonine dehydrogenase (l-ThrDH) with a predicted amino acid sequence that was remarkably similar to the sequence of UDP-galactose 4-epimerase (GalE) was overexpressed in Escherichia coli, and its product was purified and characterized. The expressed enzyme was moderately thermostable, retaining more than 90% of its activity after incubation for 10 min at up to 70 °C. The catalytic residue was assessed using site-directed mutagenesis, and Tyr137 was found to be essential for catalysis. To clarify the structural basis of the catalytic mechanism, four different crystal structures were determined using the molecular replacement method: l-ThrDH-NAD+, l-ThrDH in complex with NAD+ and pyruvate, Y137F mutant in complex with NAD+ and l-threonine, and Y137F in complex with NAD+ and l-3-hydroxynorvaline. Each monomer consisted of a Rossmann-fold domain and a C-terminal catalytic domain, and the fold of the catalytic domain showed notable similarity to that of the GalE-like l-ThrDH from the psychrophilic bacterium Flavobacterium frigidimaris KUC-1. The substrate binding model suggests that the reaction proceeds through abstraction of the β-hydroxyl hydrogen of l-threonine via direct proton transfer driven by Tyr137. The factors contributing to the thermostability of T. volcanium l-ThrDH were analyzed by comparing its structure to that of F. frigidimaris l-ThrDH. This comparison showed that the presence of extensive inter- and intrasubunit ion pair networks are likely responsible for the thermostability of T. volcanium l-ThrDH. This is the first description of the molecular basis for the substrate recognition and thermostability of a GalE-like l-ThrDH.

from conventional L-ThrDHs, which belong to the medium chain dehydrogenase reductase family (12,13). Structural comparison of F. frigidimaris L-ThrDH with E. coli and human GalEs showed a topological difference in three loops around the substrate and NAD ϩ binding site; in F. frigidimaris L-ThrDH, loops 1 and 2 insert toward the active site cavity, creating a barrier that prevents the binding of UDP-glucose, and the NAD ϩ binding loop, which tightly holds the adenine ribose moiety of NAD ϩ in the E. coli and human GalEs, is absent in the F. frigidimaris L-ThrDH (12, 14 -16). By contrast, we found that the two signature sequences in the GalE, a YXXXK motif in which the conserved tyrosine plays a key role in catalysis, and a GXXGXXG motif, which is located near the cofactor binding pocket, are strictly conserved in the F. frigidimaris enzyme. This suggests the reaction catalyzed by F. frigidimaris L-ThrDH is probably similar to that catalyzed by GalE. Notably, however, the catalytic residue and the mechanism underlying the catalytic reaction of F. frigidimaris L-ThrDH remain uncertain due to the enzyme instability.
Within the genomic sequence of the thermophilic archaeon Thermoplasma volcanium we found a gene (open reading frame identification number TVN0454) whose predicted amino acid sequence exhibits 44% identity with that of F. frigidimaris L-ThrDH. In this study we expressed the gene, characterized the enzyme produced, and show that the enzyme is a thermostable GalE-like L-ThrDH. We also determined four crystal structures for this enzyme at 1.77-2.07 Å resolution in the presence of NAD ϩ , an inhibitor, and two substrates. This is the first description of the molecular basis of the substrate recognition and thermostability of an archaeal GalE-like L-ThrDH. Our findings provide new insight into the catalytic mechanism of GalE-like L-ThrDH.

EXPERIMENTAL PROCEDURES
Cloning and Protein Expression-The gene encoding L-ThrDH (ORF ID: TVN0454, the gene information is available in the Kyoto Encyclopedia of Genes and Genomes (KEGG) database) was amplified by PCR. The oligonucleotide primers used to amplify the L-ThrDH gene fragment were 5Ј-TGA-TAACATATGATCCTAGTTACCGGCTCA-3Ј, which contains a unique NdeI restriction site overlapping the 5Ј initiation codon, and 5Ј-AAGGAAGGATCCTCAGAGGGCATGTTT-ACC-3Ј, which contains a unique BamHI restriction site proximal to the 3Ј end of the termination codon. Chromosomal T. volcanium DNA was isolated as described previously (17) and used as the template. The amplified 1.0-kb fragment was digested with NdeI and BamHI and ligated with the expression vector pET11a (Novagen) linearized with NdeI and BamHI to generate pETVN0454, which was then used to transform E. coli strain BL21 (DE3) codon plus RIL (Stratagene). The transformants were cultivated at 37°C in 100 ml of Luria-Bertani medium until the optical density at 600 nm reached 0.6. Expression was then induced by adding 1.0 mM isopropyl-␤-D-thiogalactopyranoside to the medium, and cultivation was continued for an additional 3 h at 37°C.
Purification of Recombinant L-ThrDH-All of the purification steps were carried out at room temperature. E. coli cells harvested from the 100 ml of culture (about 1.0 g (wet weight)) were used as the starting material for purification of L-ThrDH. To prepare the crude extract, the cells were washed twice and suspended in 10 mM potassium phosphate buffer (pH 7.0). The cells were then disrupted by sonication and centrifuged at 20,000 ϫ g for 10 min. The resultant supernatant, which served as the crude extract, was heated at 70°C for 10 min in the presence of 0.2 M Na 2 SO 4 (except for the R134S and E153N mutant enzymes), and the denatured proteins were removed by centrifugation (20,000 ϫ g for 10 min). The supernatant from that step was dialyzed against 10 mM potassium phosphate buffer (pH 7.0), and the remaining enzyme solution was applied to a Red-Sepharose CL-4B column (10 ϫ 110 mm) equilibrated with 10 mM potassium phosphate buffer (pH 7.0). After washing the column with the same buffer, the enzyme was eluted with a 100-ml linear gradient of 0 -0.5 M NaCl in the same buffer. The active fractions were pooled, and the enzyme solution was dialyzed against 10 mM potassium phosphate buffer (pH 7.0) containing 50 mM NaCl.
Determination of Enzyme Activity-Enzyme activity was assayed spectrophotometrically using a Hitachi U-2910 spectrophotometer equipped with a thermostat. The standard reaction mixture for oxidation consisted of 100 mM glycine-NaOH buffer (pH 10.0) containing 25 mM L-threonine, 1.25 mM NAD ϩ , and the enzyme in a final volume of 1.0 ml. The pH values of the buffer and threonine solutions were adjusted at room temperature and used for the assay without temperature compensation. After warming the reaction mixture by incubation for 3 min at 50°C without the coenzyme, the reaction was started by adding the coenzyme. The appearance of NADH was monitored from the absorbance at 340 nm (extinction coefficient ⑀ ϭ 6.22 mM Ϫ1 ⅐cm Ϫ1 ). One unit (U) of activity was defined as the amount of enzyme producing 1.0 mol of NADH per min at 50°C under the standard assay conditions, and the specific activity was expressed in U⅐(mg of protein) Ϫ1 . The protein concentration was determined using the Bradford method; bovine serum albumin served as the standard (18). The enzyme activity of GalE was assayed as previously described (13).
Polyacrylamide Gel Electrophoresis and Molecular Mass Determination-SDS-PAGE (12% acrylamide slab gel, 1-mm thick) was carried out using the procedure of Laemmli (19), after which the protein band was stained with Coomassie Brilliant Blue R-250.
Analysis of N-terminal Amino Acid Sequences-Approximately 3 g of the purified enzyme were subjected to SDS-PAGE as described above and then electroblotted onto a polyvinylidene difluoride membrane, after which the membrane was stained with Coomassie Brilliant Blue R-250 and destained.
To determine the N-terminal amino acid sequences, the protein band was then subjected to automated Edman degradation using a PPSQ-21A Protein Sequencer (Shimadzu).
Stability, pH Optimum, and Kinetic Parameters-To determine the effect of temperature on its stability, the enzyme was incubated for 10 min at different temperatures in 10 mM potassium phosphate buffer (pH 7.0) containing 50 mM NaCl. The residual activity was determined using the standard assay method. To determine the effect of pH on its stability, the enzyme was incubated for 20 min in buffer at different pH values (50°C), and the remaining activity was again determined using the standard assay method. The buffers (100 mM) used for these assays were citrate (pH 3.0 -6.0), potassium phosphate (pH 6.0 -7.5), Tris-HCl (pH 7.5-9.0), and glycine-NaOH (pH 9.0 -11.5). The same buffers were also used to determine the optimal pH for enzyme activity at 50°C. The initial velocity was determined by varying the concentration of one substrate while keeping the concentrations of the other substrates constant, as previously described (20).
Crystallization and Data Collection-Initial screening for crystallization was carried out using Cryo 1 and 2 (Emerald Biostructures) at 20°C using the sitting-drop vapor diffusion method. Crystals of the wild-type and Y137F mutant enzymes were grown in sitting drops composed of 1 l of enzyme solution (wild type, 40 mg/ml; Y137F, 25 mg/ml) containing 1 mM NAD ϩ mixed with 1 l of mother liquor containing 100 mM cacodylate buffer (pH 6.4), 50% (v/v) 2-methyl-2,4-pentandiol (MPD), and 5% (w/v) poly(ethylene glycol) 8000. Crystals of NAD ϩ /pyruvate-bound wild-type enzyme were prepared by soaking the crystals in reservoir solution containing 0.3 M pyruvate for 2 h. Crystals of NAD ϩ /L-threonine-bound and NAD ϩ / L-3-hydroxynorvaline-bound Y137F mutant enzyme were prepared by soaking the crystals in reservoir solution containing 0.1 M L-threonine for 8 h or 0.1 M DL-3-hydroxynorvaline for 7 h. Data were collected using an ADSC CCD detector system on the BL-5A, BL-17A, AR-NW12A, and AR-NE3A beamlines at the Photon Factory in Tsukuba, Japan. All diffraction measurements were carried out on crystals cryoprotected with 50% (v/v) MPD and cooled to 100 K in a stream of nitrogen gas. The data were processed using HKL2000 (21).
Phasing and Refinement-The initial phases of the wild-type enzyme structure were determined by molecular replacement using the program MOLREP in the CCP4 program suite (22); the structure of chain A of L-ThrDH from F. frigidimaris (Protein Data Bank entry 2YY7) served as the search model. Further model building was performed with the program COOT (23), and refinement to a resolution of 2.07 Å was carried out using Refmac5 (24) and CNS (25). The NAD ϩ molecule was clearly visible in both the A -weighted 2 F o Ϫ F c and F o Ϫ F c density maps and was included in the latter part of the refinement. Model geometry was analyzed using MOLPROBITY (26). Based on the model of the wild-type enzyme, the structures of the NAD ϩ /pyruvate-bound wild-type (1.78 Å), NAD ϩ /L-threonine-bound Y137F (1.85 Å), and NAD ϩ /L-3-hydroxynorvaline-bound Y137F (1.77 Å) enzymes were refined using difference Fourier maps. Two NAD ϩ molecules and two MPD molecules were included in the refinement. The data collection and refinement statistics are listed in Table 1.
Structure Analysis-Hydrogen bonds were identified using the program CCP4 mg (27). Ion pairs (the cut-off distance was 4.0 Å (28)) were identified using the WHAT IF web server (29). Molecular graphics were created using PyMOL.

RESULTS AND DISCUSSION
Gene Expression and Enzyme Purification-Within the sequence of the T. volcanium genome, we identified a gene, TVN0454 (954 bp at position 440460 -441413 of the entire genome) whose predicted amino acid sequence showed 44% identity with that of F. frigidimaris L-ThrDH. After transforming E. coli with pETVN0454, an expression vector harboring the gene, the transformant cells exhibited a high level of L-ThrDH activity, and the enzyme was readily purified from the cell crude extract in two simple steps: heat treatment and affinity chromatography of NAD(P) ϩ -binding proteins. About 15 mg of the purified enzyme was obtained from 1 g (wet weight) of cells. The purified enzyme showed a single protein band on SDS-PAGE, and its N-terminal sequence was determined to be MILVTGSSGQ, which corresponds to that predicted from the TVN0454 gene sequence and confirms that TVN0454 encodes the T. volcanium L-ThrDH. The specific activity of the purified wild-type and mutant enzymes was estimated to be about 40.1 U⅐mg Ϫ1 (wild-type), 0.2 U⅐mg Ϫ1 (Y137F), 37.1 U⅐mg Ϫ1 (R134S), and 36.6 U⅐mg Ϫ1 (E153N) at 50°C.
Molecular and Catalytic Properties of T. volcanium L-ThrDH-SDS-PAGE showed the subunit molecular mass of T. volcanium L-ThrDH to be about 35 kDa, which is consistent with the molecular weight (35,843) calculated from the amino acid sequence. That the native molecular mass determined by gel filtration was about 77 kDa suggests the native enzyme is a homodimer. Evaluation of the catalytic activity (oxidation of L-threonine) at temperatures ranging from 20 to 100°C revealed the enzyme to be maximally active at around 70°C. The enzyme was also somewhat thermostable, retaining more than 90% of its activity after incubation for 10 min at temperatures up to 70°C (supplemental Fig. 1). Evaluation of the catalytic activity at different pH values revealed the enzyme to be maximally active at around pH 10.0. The enzyme was stable over a wide range of pH values, losing no activity when incubated at pH values between 4.5 and 9.5 for 20 min at 50°C.
We also examined the effect of various chemicals on the enzyme activity. T. volcanium L-ThrDH was unaffected by EDTA, LiSO 4 , MgCl 2 , MnCl 2 , CaCl 2 , NiCl 2, CoCl 2, BaCl 2 , HgCl 2 , CdSO 4 , CuSO 4 , ZnCl 2 , or iodoacetic acid (each at 1 mM). On the other hand, the enzyme was partially inhibited by pyruvate, retaining 80% of its activity in the presence of 10 mM pyruvic acid sodium salt. This is consistent with the earlier observation that pyruvate acts as a competitive inhibitor of L-threonine (K i ϭ 23.2 mM) with the F. frigidimaris L-ThrDH (13). T. volcanium L-ThrDH showed typical Michaelis-Menten kinetics; for oxidation, the K m values for NAD ϩ and L-threonine were 0.185 and 17.5 mM, respectively.
The main chain coordinates of the T. volcanium L-ThrDH/ NAD ϩ and L-ThrDH/NAD ϩ /pyruvate monomers were basically the same as the F. frigidimaris L-ThrDH monomer (r.m.s.d. ϭ 1.1-1.2 Å for the C␣ atoms of 306 residues; PDB code 2YY7) (Fig. 1, B and C), as was assembly of the dimer. In addition, there was no apparent conformational difference among the four T. volcanium L-ThrDH structures studied (r.m.s.d. ϭ 0.38 -0.41 Å for the C␣ atoms of 311-315 residues), which suggests substrate binding does not affect the mutual orientation of the two domains of the T. volcanium enzyme. The gross movements might be precluded by crystal packing.
Structural Features Underlying Thermostability-F. frigidimaris L-ThrDH rapidly inactivates at 50°C (13) and is thus much less stable than the T. volcanium enzyme. Therefore, the factors contributing to the relative thermostability of T. volcanium L-ThrDH were analyzed by comparing its structure to that of the F. frigidimaris enzyme. Structural studies of hyperthermophilic proteins have shown that the number of ion pairs and the formation of ion pair networks contribute significantly to the thermostability of these enzymes (32)(33)(34). Using a cutoff distance of 4.0 Å between oppositely charged residues, we calculated that T. volcanium L-ThrDH contains 26 intrasubunit ion pairs, whereas F. frigidimaris L-ThrDH contains only 13 intrasubunit ion pairs. In addition, two major intersubunit ion pair networks between the A and B subunits were observed in T. volcanium L-ThrDH: Arg 132 -Glu 119 -Arg 134 -Glu 153 Ј and Arg 132 Ј-Glu 119 Ј-Arg 134 Ј-Glu 153 (the primes indicate the neighboring subunit in the dimer) ( Fig. 2A). In F. frigidimaris L-ThrDH, however, the charged Glu 119 , Arg 132 , Arg 134 , and Glu 153 are replaced by Thr 125 , Glu 138 , Ser 140 , and Asn 159 , respectively (Fig. 2B). Because Arg 134 is involved in both interand intrasubunit ion pairs and Glu 153 is involved with intersubunit ion pairs ( Fig. 2A), we constructed two T. volcanium L-ThrDH single point mutants (R134S and E153N) and compared their thermostability with that of the wild-type enzyme. We found that the mutant enzymes were significantly less thermostable, with both mutants retaining full activity only at 50°C or less (supplemental Fig. 1). This suggests the presence of the extensive inter-and intrasubunit ion pair networks is likely responsible for the thermostability of T. volcanium L-ThrDH.
Active Site and L-Threonine Binding Model-Within all four T. volcanium L-ThrDH structures studied, the electron density corresponding to the NAD ϩ coenzyme bound within the active site was very clear, which enabled us to place the ligand with reasonable accuracy (supplemental Fig. 2). Moreover, in the initial electron density map of the NAD ϩ /pyruvate-bound wild-type enzyme, we observed an extra density within the active site cavity, and after construction and refinement of the peptide chain, a pyruvate molecule could be modeled into that density. However, we could not determine precise orientation of the pyruvate molecule, as pyruvate is an approximately symmetrical molecule. In our pre- dictive model (Fig. 3A), two oxygen atoms of the C1 carboxylate are situated within hydrogen bonding distance of the side chains of Ser 74 , Thr 112 , and Tyr 137 and the main chain amide proton of Ser 74 . The oxygen atoms of the C2 carbonyl group are within hydrogen bonding distance of the side chain of Ser 74 , a proton in the main chain amide of Thr 178 , and a water molecule (WAT 427). Among these residues, Tyr 137 is strictly conserved in GalE-like L-ThrDHs and GalEs. As mentioned above, Y137F mutation nearly abolished the activity of T. volcanium L-ThrDH, which suggests Tyr 137 plays a critical role in the catalytic reaction of GalElike L-ThrDHs (see below).
Given the structure of pyruvate, we sought to predict the structure of the L-threonine molecule within the active site of T. volcanium L-ThrDH. Based on the structure of pyruvate, we tried the modeling of L-threonine binding into the active site of T. volcanium L-ThrDH. However, we could not postulate a sin-gle binding mode of L-threonine molecule. On the other hand, using the inactive T. volcanium L-ThrDH Y137F mutant, which trapped the substrate within the active site, we were able to determine the substrate-bound structures of the enzyme in complex with L-threonine or L-3-hydroxynorvaline. In both cases, the map clearly defined the precise orientation of the L-threonine and L-3-hydroxynorvaline (Fig. 3, B and C); two oxygen atoms of the C␣ carboxylate are situated within hydrogen bonding distance of the side chains of Ser 74 and Thr 179 and the main chain amide proton of Ser 74 , Thr 178 , and Thr 179 . The oxygen atoms of the C␤ hydroxyl group are within hydrogen bonding distance of the side chain of Thr 112 . In addition, the C␣ amino group is within hydrogen bonding distance of the side chain of Thr 179 , the carbonyl group of the nicotinamide ring, and a water molecule. All of the residues are conserved in F. frigidimaris L-ThrDH, except Thr 112 is replaced by a homologous serine residue (Fig. 4).
Insight into ThrDH Reaction-With short chain dehydrogenase reductase family enzymes, the Thr/Ser-Tyr-Lys catalytic triad is thought to be important for catalysis. The side chain oxygen of the tyrosine residue functions as an acidbase catalyst for proton transfer, and the serine residue plays a subsidiary role for stabilization of substrate binding (35). In addition, the lysine residue has two critical functions; it interacts with the hydroxyl groups of the nicotinamide ribose (cofactor binding), and it lowers the pK a value of the hydroxyl group in the side chain of the tyrosine residue (35).
In the E. coli GalE, it has been reported that the mutation of Lys 153 to Met or Ala abolishes the UMP dependence of reductive inactivation by sodium cyanoborohydride. This suggests that the positive electrostatic field would stabilize the phenolate form of Tyr 149 and thereby account for a low value of pK a (14). This Thr/Ser-Tyr-Lys catalytic triad is strictly conserved in E. coli and human GalEs and F. frigidimaris and T. volcanium L-ThrDHs (Fig. 4). Thus, a common catalytic mechanism should be predictable among short chain dehydrogenase reductase family enzymes.
We recently proposed two possible catalytic mechanisms for F. frigidimaris L-ThrDH based on the L-threonine binding models deduced from the structures of bound glycerol and the glucose moiety of the UDP-glucose in the E. coli and human GalEs (12). Based on the model deduced from the structure of the glycerolbound enzyme, we proposed that the reaction proceeds through a proton shuttle mechanism in which Tyr 143 provides the driving force for general acid-base catalysis and Ser 118 mediates the proton relay as in the case of E. coli GalE (36,37). With that mechanism, the conserved Tyr 143 cannot function as the active-site base because the observed distance (4.9 Å) between the O of Tyr 143 and the ␤-hydroxyl group of the L-threonine is too large for direct interaction. By contrast, based on the L-threonine binding model deduced from the structure of UDP-glucose, Tyr 143 lies within hydrogen-bonding distance (2.7 Å) of the ␤-hydroxyl group of the L-threonine, so that this tyrosine residue could serve directly as the active-site base as in the case of human GalE (16). At that time, we could not determine which was the correct catalytic mechanism of F. frigidimaris L-ThrDH.
Within the structures of the T. volcanium L-ThrDH Y137F/ NAD ϩ /L-threonine and L-3-hydroxynorvaline, we observed that the O of Tyr 137 , which corresponds to Tyr 143 in F. frigidimaris L-ThrDH, lies within hydrogen-bonding distance (2.4 -2.6 Å) of the ␤-hydroxyl group of the L-threonine or L-3-hy-droxynorvaline (Fig. 3D). This means that Tyr 137 could serve directly as the active-site base. In addition, the distance between the O␥ of Thr 112 and the ␤-hydroxyl of the L-threonine A, the NAD ϩ /pyruvate-bound wild-type enzyme (magenta) is shown. B, the NAD ϩ / L-threonine-bound Y137F mutant enzyme (white) is shown. C, the NAD ϩ /L-3-hydroxynorvaline-bound Y137F mutant enzyme (gold) is shown. D, superposition of the active site pocket of the NAD ϩ /L-threonine-bound Y137F mutant with that of the NAD ϩ /L-3-hydroxynorvaline-bound Y137F mutant is shown. For clarity, the Tyr 137 residue was positioned based on the orientation of Phe 137 in Y137F. Atoms are colored as in Fig. 1. The C4 atom of the pyridine ring (a hydride acceptor site) and the si-and re-faces are labeled. The L-threonine and L-3-hydroxynorvaline molecules are shown as stick models in white. WAT, water. In panels A-C, the final A -weighted F o Ϫ F c omit electron density maps for the ligands are shown at the 1 level.
(2.8 Å) or L-3-hydroxynorvaline (2.8 Å) as well as the distance between the side chain hydroxyls of Thr 112 and Tyr 137 (3.0 -3.2 Å) are in good agreement with those observed in the UDPglucose based model of F. frigidimaris L-ThrDH. Taken together, these findings suggest the catalytic mechanism underlying the T. volcanium L-ThrDH reaction likely proceed through the following steps: 1) abstraction of the ␤-hydroxyl hydrogen of L-threonine via direct proton transfer driven by Tyr 137 and 2) transfer of a hydrogen from the ␤-carbon (C3) of the L-threonine to C4 of the NAD ϩ (si-face), forming L-2-amino-3-oxobutyrate and NADH (Fig. 5).