T. Tamura, Bioproduction Research Institute, National Institute of Advanced Industrial Science and Technology (AIST), 2-17-2-1, Tsukisamu-higashi, Toyohira, Sapporo 062-8517, Japan Fax: +81 11 857 8980 Tel: +81 11 857 8938 E-mail: email@example.com
Bacillus megaterium IAM 1030 (Bacillus sp. JCM 20016) possesses four d-glucose 1-dehydrogenase isozymes (BmGlcDH-I, -II, -III and -IV) that belong to the short-chain dehydrogenase/reductase superfamily. The BmGlcDHs are currently used for a clinical assay to examine blood glucose levels. Of these four isozymes, BmGlcDH-IV has relatively high thermostability and catalytic activity, but the disadvantage of its broad substrate specificity remains to be overcome. Here, we describe the crystal structures of BmGlcDH-IV in ligand-free, NADH-bound and β-d-glucose-bound forms to a resolution of 2.0 Å. No major conformational differences were found among these structures. The structure of BmGlcDH-IV in complex with β-d-glucose revealed that the carboxyl group at the C-terminus, derived from a neighboring subunit, is inserted into the active-site pocket and directly interacts with β-d-glucose. A site-directed mutagenic study showed that destabilization of the BmGlcDH-IV C-terminal region by substitution with more bulky and hydrophobic amino acid residues greatly affects the activity of the enzyme, as well as its thermostability and substrate specificity. Of the six mutants created, the G259A variant exhibited the narrowest substrate specificity, whilst retaining comparable catalytic activity and thermostability to the wild-type enzyme.
Database The atomic coordinates and structure factor amplitudes for BmGlcDH-IV in ligand-free form, in complex with NADH, in complex with d-glucose, G259A mutant in ligand-free form, and A258F mutant in complex with d-glucose and NADH were deposited in the RCSB Protein Data Bank (http://www.rcsb.org) under the accession codes 3AUS, 3AUT, 3AUU, 3AY6 and 3AY7, respectively
NAD(P)+-dependent glucose 1-dehydrogenase (GlcDH) isozymes from Bacillus megaterium (EC 220.127.116.11; BmGlcDHs) catalyze the oxidation of β-d-glucose to d-glucono-1,5-lactone, using NAD(P)+ as a cofactor. BmGlcDH is a member of the short-chain dehydrogenase/reductase (SDR) superfamily and has been used clinically to examine blood glucose levels. In addition, BmGlcDH is believed to have great potential in the regeneration of NAD(P)H for the biocatalytic synthesis of fine chemicals [1–3]. Two other types of GlcDHs have been used for blood glucose detection. One is a pyrroloquinoline-quinone-containing GlcDH (EC 18.104.22.168; PQQ-GlcDH), which is a homodimer with a subunit weight of ∼ 50 kDa . Although PQQ-GlcDH is self-sufficient and does not require additional cofactors, it lacks thermostability and shows broad substrate specificity toward various aldohexoses . Some attempts have been made to overcome these problems by using directed evolution and site-directed mutagenesis [6,7]. Another GlcDH is an FAD-containing enzyme (EC 22.214.171.124; FAD-GlcDH) that has been isolated from two organisms, namely Aspergillus oryzae  and Burkholderia cepacia ; furthermore, the gene encoding FAD-GlcDH from B. cepacia has been cloned, and the enzyme has been characterized in detail. The B. cepacia FAD-GlcDH does not require additional cofactors and is relatively more thermostable than the PQQ-GlcDHs. However, the disadvantages of its relatively low catalytic activity at room temperature and its broad substrate specificity remain to be resolved . Site-directed mutagenesis studies on FAD-GlcDH have been performed to improve its substrate specificity . In contrast to these two types of GlcDHs, BmGlcDH requires the addition of NAD+ to the assay mixture. However, BmGlcDH is currently applied for clinical assay because of its relatively narrow substrate specificity toward d-glucose [11–13].
BmGlcDH has the typical tetrameric structure of SDR enzymes  and is composed of identical subunits, each with a subunit mass of 28 kDa. So far, several BmGlcDHs have been characterized [11–13], and their corresponding genes have been cloned from various Bacillus species. These enzymes have a single Rossmann-fold domain known as an NAD(P)+-binding motif. The active form of BmGlcDH is a homotetramer, which dissociates into inactive monomer or dimer at pH values above 9, in the absence of a high concentration of NaCl [15–17]. Several BmGlcDH mutants exhibiting enhanced thermostability at an alkaline pH and/or under low-salt conditions have been reported previously [18–22]. B. megaterium IAM1030 has four BmGlcDH isozymes (I–IV) that have been enzymatically characterized . Among these four isozymes, BmGlcDH-IV has high thermostability and exceptionally high catalytic activity toward d-glucose. However, weak activity of this enzyme toward several other monosaccharides such as d-xylose, d-mannose, d-galactose and d-glucosamine has also been detected. More accurate measurements of blood glucose levels are therefore anticipated once the substrate specificity of BmGlcDH-IV is altered. Many structures of SDR enzymes have been reported in complexes with substrates/inhibitors such as sugars, steroids and alcohols [23–26], whereas no structural information for d-glucose binding is yet available. Herein, we have determined the X-ray structures of the substrate-free, NADH-bound and d-glucose-bound forms of BmGlcDH-IV. These structures reveal that the C-terminal carboxyl group derived from a neighboring subunit directly interacts with d-glucose. Based on the structures, we have performed site-directed mutagenesis of BmGlcDH-IV and have obtained a useful mutant that showed improvement of d-glucose specificity while retaining similar thermostability to the wild-type enzyme.
Results and Discussion
Structures of wild-type BmGlcDH-IV
All BmGlcDH-IV crystals in the ligand-free form, NADH-complex and d-glucose-complex belong to the same C2 space group. The structure of BmGlcDH-IV in complex with d-glucose was determined by the molecular replacement method. As a search model, a monomer of GlcDH from Bacillus megaterium IWG3 (BmGlcDH-IWG3; PDB code 1GCO) was used . The structure of BmGlcDH-IV in ligand-free form and in complex with NADH was determined by the molecular replacement method with a monomer of d-glucose-bound BmGlcDH-IV as a search model. All models were refined to a resolution of 2.0 Å. A Ramachandran plot showed that almost 90% of residues were in the most favored region, while there were no residues in the disallowed region. The data collection and model refinement statistics are summarized in Table 1.
Table 1. Crystallographic data. Data for the highest resolution shell are provided in parentheses. Rmerge = ∑h∑i |Ih,i − 〈Ih〉|/∑h∑iIh,i, where 〈Ih〉 is the mean intensity of a set of equivalent reflections. Rwork = ∑|Fobs − Fcalc|/∑Fobs for the 95% of the reflection data used in the refinement. Fobs and Fcalc are observed and calculated structure factor amplitudes, respectively. Rfree is the equivalent of Rwork except that it was calculated for a randomly chosen 5% test set excluded from the refinement.
The asymmetric unit contains two BmGlcDH-IV monomers (subunits A and B) that are related by a non-crystallographic twofold symmetry. The homotetramer having a 222-point-group symmetry was generated from the dimer in the asymmetric unit with crystallographic twofold symmetry. The resultant tetrameric structure is shown in Fig. 1. No major conformational differences were observed between the ligand-free, NADH-bound and d-glucose-bound structures. The BmGlcDH-IV monomer comprises 261 amino acid residues and shares high sequence homology (82.4%) with BmGlcDH-IWG3, whose quaternary structure has been previously reported in d-glucose-free form . The quaternary structure of BmGlcDH-IV in ligand-free form is very similar to that of BmGlcDH-IWG3, with an rmsd of 0.6 Å (for 261 Cα atoms). The BmGlcDH-IV subunit consists of seven-stranded central parallel β-sheets sandwiched by two arrays of three α-helices. BmGlcDH-IV represents a typical Rossmann fold in the core structure containing a GXXXGXG sequence, which is a characteristic fingerprint of the nucleotide-binding motif of the SDR enzyme [28,29].
In our current structure of the NADH complex, an Fo − Fc electron density map for NADH is clearly visible in chain A, but is rather ambiguous in chain B. Thus, the NADH model was built only in chain A. BmGlcDH-IV showed much lower KM values for NAD+ than NADP+, while BmGlcDH-IWG3 showed much larger KM values for NAD+ than NADP+, as has been previously reported by others [13,14]. Our current structure shows that the cofactor-binding pocket is very similar to that of BmGlcDH-IWG3 except for residues located near an adenine moiety. In BmGlcDH-IWG3, the side chain of the Arg39 residue in the β2α3 turn is probably important for the specificity of the NADP+ molecule to form a salt bridge with the phosphate group of NADP+. In contrast, there were no similar basic residues close to the 2′-phosphate group in BmGlcDH-IV. The corresponding residue is a Tyr39 that lies near the adenine ring, and probably plays a role in decreasing the enzyme’s affinity toward NADP+ (Fig. S1).
To understand the structural mechanism of d-glucose recognition of BmGlcDH-IV, we have determined the structure of BmGlcDH-IV in complex with d-glucose. Crystals of the d-glucose complex grew only when the crystallization solution contained both NADH and d-glucose. However, the electron density for NADH was completely missing. Steric hindrance between the reduced nicotinamide ring of NADH and the C1 hydrogen of d-glucose may stimulate the release of NADH as described below. The unambiguous electron density for the bound sugar indicated that the conformation of the pyranose ring is in a chair-type configuration (Figs 2 and S2). Amino acid residues forming hydrogen bonds with d-glucose are listed in Table S1, together with their bond distances. The C1 hydroxyl forms hydrogen bond interactions with the side chains of Tyr158 and Ser145. These residues are part of a catalytic triad (Tyr-Lys-Ser) in SDR enzymes. It is believed that Lys decreases the pKa of the hydroxyl group in Tyr that acts as a catalytic base [30,31]. The Ser residue also functions in the stabilization of the enzyme–substrate complex during the reaction. In addition, clear electron density maps showed that the trapped d-glucose is in the β form with the C1 hydroxyl group in the equatorial configuration (Fig. S2). These results are consistent with previous reports that BmGlcDH acts on β-d-glucose, but not on α-d-glucose . In addition, the side chains of Glu96 and Asn196 form hydrogen bonds with the C2 and C4 hydroxyl groups of β-d-glucose. The side chains of Lys199 and His147 also interacted with the C4 and C6 hydroxyl groups, respectively. It is interesting to note that the C-terminal carboxyl group (in Gly261) derived from a neighboring subunit is inserted into the active-site pocket and directly interacts via hydrogen bonds with C4 and C6 hydroxyl groups in the bound β-d-glucose. The side chain amino group of Lys199 is also located near the C-terminal carboxyl groups at a distance of < 3.0 Å. No crystal contacts are found in the vicinity of the substrate-binding pocket. These elaborate interatomic interactions probably play a critical role in stabilizing the conformation of the active-site and the bound β-d-glucose.
Since no structural differences were observed between the NADH and d-glucose complexes, it is possible to build a model of the putative ternary complex of BmGlcDH-IV-d-glucose-NADH by superimposing the NADH and d-glucose complexes. In the model, the C1 and C4 hydroxyl groups in the nicotinamide ring of NADH are in very close proximity (2.2 Å apart), suggesting that steric hindrance could have excluded the binding of NADH in the d-glucose-bound structure. The proximity between C4 in the nicotinamide ring and C1 in β-d-glucose also indicates that the binding position and orientation of β-d-glucose is not an artifact, and reflects the transient enzyme–substrate Michaelis intermediary complex that precedes hydride transfer from C1 in d-glucose to C4 in the nicotinamide ring.
Target residue selection for mutational analysis of BmGlcDH-IV C-terminal region
The C-terminal region of several SDR enzymes is reportedly involved in catalytic activity [32,33]. Our current structures of BmGlcDH-IV also reveal that the C-terminal carboxyl directly interacts with substrate and stabilizes the structure of the active site (Fig. 2). To investigate the role of the C-terminal region in catalytic activity and substrate specificity, we attempted to disrupt the conformation of the C-terminal region in BmGlcDH-IV by using site-directed mutagenesis. Gly261 (at the C-terminus), Gly259 and Ala258 were chosen as target residues for mutational analysis, while Arg260 was ruled out because it is involved in the inter-subunit association through electrostatic interactions with Asp208 (Fig. 2). We have created a total of five single mutants as follows. For Gly261 and Gly259, we have created alanine and valine variants (G261A, G261V, G259A and G259V). Because there were no spaces for the methyl and isopropyl groups in the side chains of the variants (Fig. 2), these mutations are expected to disrupt the conformation of the C-terminal region and consequently affect the enzyme’s activity and/or substrate specificity. In contrast, the methyl side chain in Ala258 is fully exposed to the solvent; thus, we created an A258F variant that is also expected to thermodynamically destabilize the C-terminal region. Additionally, we also created a C-terminal deletion mutant (ΔG261) to assess the importance of the C-terminal carboxyl group.
Comparison of enzymatic properties between BmGlcDH-III, BmGlcDH-IV and BmGlcDH-IV variants
First, we examined the pH dependence of the d-glucose oxidizing activity of BmGlcDH-III, -IV and the variants. Generally, wild-type BmGlcDH-III and BmGlcDH-IV show higher activity at alkaline pH because the proton-abstracting power of Tyr158 from the substrate is probably stronger at higher pH values . However, the results showed that the optimal pH for five variants was around pH 6.0; furthermore, the activity drastically decreased at alkaline pH (Fig. 3). This might reflect a relaxation of the tetrameric assembly in the assay mixture at alkaline pH due to destabilization of the C-terminal region, which creates a part of the inter-subunit interface. However, gel filtration analysis suggested that both wild-type BmGlcDH-IV and its mutants exist as a tetramer in solution. Another possibility is that a different reaction pathway predominates for these mutants. Similar pH dependence was also reported with uronate dehydrogenase of the SDR family ; the authors proposed that the novel cationic intermediate does not require initial proton abstraction by the catalytic triad described above. The same reaction mechanism might also underlie BmGlcDH-IV oxidizing activity.
Based on the results of pH dependence tests, the specific activities for d-glucose and other monosaccharides and the kinetic analysis for d-glucose were determined at both pH 6.0 and pH 8.0 (Tables 2 and 3). The specific activities of mutants G259V, G261A, G261V and ΔG261 for d-glucose were severely decreased at both pH 6.0 and pH 8.0. Kinetic analysis also indicated that KM values for these mutants increased by 100–1000-fold, demonstrating the harmful influence of these mutations on enzyme performance (Table 2). However, the specific activity of the G259A mutant was slightly lower, but is still comparable with wild-type BmGlcDH-III and BmGlcDH-IV. Notably, the activity of G259A mutant with other analogous sugars was significantly decreased; in particular, its activity with d-xylose, d-mannose and d-galactose was not detected at pH 8.0. In contrast, the A258F mutation resulted in a slightly lowered activity but with considerably broadened substrate specificity. Kinetic analysis revealed that the KM value for A258F with d-glucose slightly decreased to ∼ 45% of the Vmax value. We also examined the thermostability of wild-type BmGlcDH-III as well as BmGlcDH-IV and its variants by measuring its specific activity for d-glucose after incubation for 0–40 min at 65 °C (Fig. 4). The results clearly show that the thermostability of A258F and ΔG261 deteriorated markedly, and the level of activity was below that of BmGlcDH-III. The remaining activity of the G259V mutant was also lower than that of wild-type BmGlcDH-IV. The other three mutants retained almost the same level of remaining activity. The results suggest that the C-terminal carboxyl of Gly261 is important for both enzyme activity and thermostability. In summary, the G259A mutant is the most useful mutant, exhibiting the narrowest substrate specificity whilst retaining its thermostability and specific activity at a practical level.
Table 2. Kinetic parameters of BmGlcDH-III, -IV and its mutants at 37 °C.
aKM and Vmax were estimated by extrapolation with substrate concentrations ranging from 280 to 1400 mm.
11 ± 0.3
320 ± 11
63 ± 4.0
820 ± 34
3.1 ± 0.3
230 ± 19
12 ± 0.8
810 ± 33
4.3 ± 0.2
110 ± 2.0
9.8 ± 1.0
310 ± 23
81 ± 7.1
480 ± 23
380 ± 16
960 ± 41
1.4 × 103 ± 96
510 ± 38
4.1 × 102 ± 35
410 ± 25
1.1 × 103 ± 170
260 ± 38
Table 3. Comparison of substrate specificity of BmGlcDH-III, -IV and its mutants at 37 °C. ND, not detected. Values in parentheses refer to the specific activity (initial rate of enzyme reaction at each substrate concentration of 148 mm) (U·mg−1).
Substrate specificity (%)
Structures of the G259A and A258F mutants
To understand the structural basis for the improved or broadened substrate specificity in the mutants, crystal structures of G259A and A258F mutant were also determined. The G259A crystals were obtained in the same space group as the wild-type BmGlcDH-IV under similar crystallization conditions with 4 mm NADH and 400 mm d-glucose. However, the electron density for d-glucose and NADH was missing. The electron density for the C-terminal residues 259–261 (chain A) and 260–261 (chain B) was also completely absent. These results suggest that these residues at the C-terminal region are protruded into the solvent region, and that the conformation of the C-terminal region is disordered (Fig. 5). Overall conformations except for these C-terminal residues are similar to that of wild-type BmGlcDH-IV, with rmsd of 0.2 Å for Cα atoms of the residues 1–258. We also determined the structure of the A258F mutant, which showed significantly broadened substrate specificity. A new crystal form belonging to the P21 space group was obtained under the same crystallization conditions as wild-type BmGlcDH-IV. The asymmetric unit contains one tetramer, and the four active sites were all occupied by both β-d-glucose and NADH. This is quite unexpected because NADH in the wild-type BmGlcDH-IV structure in complex with d-glucose is missing. Superimposing wild-type BmGlcDH-IV and the A258F mutant revealed no major conformational changes, although there was very slight movement of β-d-glucose toward Trp152 (< 0.4 Å) in the mutant structure to keep an appropriate distance between d-glucose and NADH. The distance between C1 in d-glucose and C4 in the nicotinamide ring is in the range of 3.1–3.3 Å. Two of four Phe258 residues in the tetramer are exposed to the solvent, while the other two Phe258 residues make a crystal contact with the Pro203-Val204 region of the symmetry-related neighboring tetramer. However, the crystal packing force does not influence the conformation of the C-terminal region, because the C-terminal residues including Phe258 in all subunits were well ordered in a similar manner to wild-type BmGlcDH-IV (Fig. 6).
The structural superimposition of all crystallographically independent BmGlcDH-IV subunits showed that one α-helix (αFG1; the nomenclature of the secondary structure elements follows the convention [14,25]) in the vicinity of the active site is rather mobile (Fig. 6). In particular, the residues of αFG1 in the structure of the G259A mutant have relatively high B-factor values (∼ 60 Å2) and move slightly away from the active site, probably due to the lack of electrostatic interactions between Arg257 and Asp208. The side chain amino group of Lys199 derived from the αFG1 makes hydrogen bonds to both O4 of d-glucose and the C-terminal carboxyl in the structure of wild-type enzyme and A258F mutant (Fig. 2). In the structure of G259A mutant, the side chain of Lys199 also moves away from the active site. The observation implies that the mobility of the C-terminal region and the αFG1 might be related to sugar-binding selectivity of the enzyme. Although the structures of the two mutants provided no clear explanation for the observed changes in the properties of the mutant enzyme, our mutational studies suggested that catalytic activity and substrate specificity could be accurately manipulated by introducing mutations at the C-terminal region. A more useful mutant might be generated using saturated mutagenesis on these C-terminal residues.
We have determined the structures of BmGlcDH-IV in ligand-free form, in complex with NADH and in complex with d-glucose. The structure of the d-glucose complex revealed that the C-terminal region of the polypeptide chain is inserted into the active-site pocket and that the C-terminal carboxyl group directly interacts with bound d-glucose. Mutational analyses demonstrated that disrupting the conformation of the C-terminal region greatly affected the catalytic activity, pH dependence and substrate specificity of the enzyme. Of the six mutants we created, the G259A mutant showed the most improved substrate specificity, while retaining its thermostability as well as a reasonable activity level against d-glucose. In some SDR enzymes such as GlcDHs and alcohol dehydrogenases, the C-terminal region is found to lie in close proximity to the active-site pocket and is important for the catalytic activity of the enzymes. Site-directed and/or saturated mutagenesis studies on the C-terminal loop in SDR enzymes might thus be a good strategy to obtain useful mutants having preferable substrate specificities.
Synthetic oligonucleotides were obtained from Hokkaido System Science Co. Ltd (Sapporo, Japan). All enzymes used for genetic manipulation were purchased from New England Biolabs Japan (Tokyo, Japan), Takara Bio Inc. (Otsu, Japan) and Stratagene (La Jolla, CA, USA). All chemicals were purchased from Sigma-Aldrich (St. Louis, MO, USA), Wako Pure Chemicals Inc (Osaka, Japan).
Construction of plasmids for wild-type BmGlcDH-III, -IV and its mutants
To express recombinant BmGlcDH-III and BmGlcDH-IV with N-terminal His-tags, the corresponding genes (gene bank accession number D10625 and D10626, respectively) were cloned into the NdeI and XhoI sites of vector pET28a (Novagen). Bacillus megaterium IAM1030 has four GlcDH isozymes (BmGlcDH-I–IV), and the nucleotide sequences of gdhIII and gdhIV are almost identical. Therefore, the upstream and downstream regions of gdhIV were initially amplified using PCR with Pfu Turbo DNA polymerase and the specific primers. The B. megaterium IAM1030 genome DNA was used as the template. Finally, gdhIII and gdhIV were amplified using PCR with Pfu Turbo DNA polymerase and the specific primers. BmGlcDH-IV single mutants were prepared using inverse PCR with appropriate synthetic oligonucleotides as primers, and pET28a-gdhIV as the template.
Expression and purification of the recombinant BmGlcDH-III, -IV and its mutants
Transformed Escherichia coli BL21-CodonPlus(DE3)-RIL cells were grown at 37 °C in 100 mL of LB medium supplemented with 20 μg·mL−1 kanamycin and 34 μg·mL−1 chloramphenicol. This culture was then inoculated at a ratio of 100 mL of overnight culture to 900 mL of LB medium. Gene expression was induced by the addition of 0.4 mm isopropyl thio-β-d-galactoside, and the culture was incubated overnight at 30 °C. The cells were harvested from the culture by centrifugation at 3000 g for 10 min. The resultant cell pellet was then resuspended in buffer A (50 mm sodium phosphate, 300 mm NaCl, pH 8.0) containing 1 mg·mL−1 lysozyme and about 50 U·mL−1 benzonase. The cell suspension was incubated on ice for 30 min and sonicated for 10 min. After centrifugation at 27 200 g for 20 min, the supernatant was loaded onto a nickel affinity column (HIS-SELECT resin; Sigma-Aldrich (St. Louis, MO, USA)) equilibrated with buffer A. Bound proteins were washed with buffer B (50 mm sodium phosphate, 300 mm NaCl and 10% glycerol, pH 6.0) and eluted using a linear gradient of 0–0.4 m imidazole in buffer B. The purified fractions were pooled and dialyzed against 50 mm sodium phosphate, 2 m NaCl and 20% glycerol (pH 6.5) and stored at −20 °C.
The activities of wild-type BmGlcDH-III, BmGlcDH-IV and its variants were spectrophotometrically assayed by measuring the rate of reduction of NAD+ in a 3 mL assay mixture at 340 nm at 37 °C. One unit of activity was defined as a reduction of 1 μmol of NADH per minute. Protein concentrations were determined using the Bradford assay with BSA as a standard . The assay mixture contained 148 mm of appropriate substrates, 85.3 mm Tris/HCl pH 8.0 or 85.3 mm potassium phosphate pH 6.0, and 3.66 mm NAD+ for activity measurements. To examine the pH dependence of the enzyme activity, 85.3 mm potassium phosphate buffer (pH range from 6.0 to 7.5) or 85.3 mm Tris/HCl buffer (pH range from 8.0 to 9.0) was used for the assay. The KM and Vmax values for d-glucose were estimated from Michaelis–Menten kinetics after triplicate experiments with varying d-glucose concentrations (2.4–1400 mm; for details see Table 2). To examine the thermostability of wild-type BmGlcDH-III, BmGlcDH-IV and its variants, the remaining activity for d-glucose was determined after incubation for 0, 20 and 40 min at 65 °C. The assay mixture for the thermostability examination contained 0.1 mg·mL−1 enzyme in 50 mm sodium phosphate pH 8.0 and 2.0 m NaCl.
For the crystallization experiments, protein concentration was adjusted to 10.0 mg·mL−1 in 20 mm Bistris pH 6.5 and 200 mm NaCl. For the crystallization of the cofactor complex, 4 mm NADH was added to the sample followed by an overnight incubation at 4 °C. Crystallization screens were performed using the commercially available sparse-matrix screens (Hampton Research (Aliso Viejo, CA, USA) and Emerald BioStructure (Bainbridge Island, WA, USA)), and diffraction-quality crystals suitable for X-ray structure analysis were obtained by optimizing the buffer pH and precipitant concentration. BmGlcDH-IV in ligand-free form and in complex with NADH was crystallized using a reservoir solution containing 0.1 m sodium cacodylate pH 6.5, 0.2 m magnesium chloride and 50% poly(ethylene glycol) 200 by using the hanging-drop vapor diffusion method. BmGlcDH-IV in complex with d-glucose was crystallized using a reservoir solution containing 0.1 m sodium cacodylate pH 6.5, 0.2 m magnesium chloride, 51% poly(ethylene glycol) 200 and 400 mm d-glucose by the hanging-drop vapor diffusion method at 293 K. G259A mutant crystals were obtained using a micro-seeding technique in a drop with reservoir solution containing 0.1 m sodium cacodylate pH 6.5, 0.2 m magnesium chloride, 50.5% poly(ethylene glycol) 200 and 400 mm d-glucose. Crystals of A258F mutant were obtained using the hanging-drop vapor diffusion method with reservoir solution containing 0.1 m sodium cacodylate pH 6.5, 0.2 m magnesium chloride, 50% poly(ethylene glycol) 200 and 400 mm d-glucose. For crystallization of these mutants, 4 mm NADH was added to the sample, followed by overnight incubation at 4 °C.
X-ray diffraction studies and structure determination
The X-ray diffraction data were collected using a charge-coupled device detector (ADSC) under cryogenic conditions (100 K) at beamline AR-NW12A, AR-NE3A or BL-5A, Photon Factory (Tsukuba, Japan). These data were indexed, integrated and merged using the hkl2000 program package . Crystals of wild-type BmGlcDH-IV and G259A mutant belong to the monoclinic C2 space group with unit-cell dimensions a = 65 Å, b = 129 Å, c = 64 Å and β = 111°. The A258F mutant was crystallized in different crystal form: the space group P21 with unit-cell parameters a = 57 Å, b = 127 Å, c = 89 Å and β = 95°. The structure of BmGlcDH-IV was solved via the molecular replacement method by using the program molrep  and by applying the structure of BmGlcDH-IWG3 (PDB code 1GCO)  as a search model. Atomic coordinates and atomic displacement parameters were refined with refmac5 program , and a manual model fitting was performed using the graphic program coot . The stereochemical quality of the final refined model was verified using the program procheck . Data collection and refinement statistics are provided in Table 1. The atomic coordinates and structure factor amplitudes for BmGlcDH-IV in ligand-free form, in complex with NADH, in complex with d-glucose, G259A mutant in ligand-free form, and A258F mutant in complex with d-glucose and NADH were deposited in the RCSB Protein Data Bank (http://www.rcsb.org) under the accession codes 3AUS, 3AUT, 3AUU, 3AY6 and 3AY7, respectively. Molecular illustrations were prepared using the program pymol (DeLano Scientific; http://pymol.sourceforge.net/).
We thank beamline scientists at Photon Factory (Tsukuba, Japan) for their kind assistance with X-ray diffraction experiments. We also thank N. Tamura for technical support. This work was supported by a grant from the National Institute of Advanced Industrial Science and Technology (AIST), Japan.