* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Functional analysis of hyperthermophilic endocellulase from
Proteolysis wikipedia , lookup
Drug design wikipedia , lookup
Protein–protein interaction wikipedia , lookup
Amino acid synthesis wikipedia , lookup
Evolution of metal ions in biological systems wikipedia , lookup
Oxidative phosphorylation wikipedia , lookup
Biosynthesis wikipedia , lookup
Photosynthetic reaction centre wikipedia , lookup
Ligand binding assay wikipedia , lookup
Deoxyribozyme wikipedia , lookup
Homology modeling wikipedia , lookup
Nuclear magnetic resonance spectroscopy of proteins wikipedia , lookup
X-ray crystallography wikipedia , lookup
Enzyme inhibitor wikipedia , lookup
Structural alignment wikipedia , lookup
NADH:ubiquinone oxidoreductase (H+-translocating) wikipedia , lookup
Biochemistry wikipedia , lookup
Metalloprotein wikipedia , lookup
Biochem. J. (2011) 437, 223–230 (Printed in Great Britain) 223 doi:10.1042/BJ20110292 Functional analysis of hyperthermophilic endocellulase from Pyrococcus horikoshii by crystallographic snapshots Han-Woo KIM and Kazuhiko ISHIKAWA1 National Institute of Advanced Industrial Science and Technology (AIST), Health Research Institute, 1-8-31, Midorigaoka, Ikeda, Osaka 563-8577, Japan, and Biomass Technology Research Center, 3-11-32, Kagamiyama, Higashi-Hiroshima, Hiroshima 739-0046, Japan A hyperthermophilic membrane-related β-1,4-endoglucanase (family 5, cellulase) of the archaeon Pyrococcus horikoshii was found to be capable of hydrolysing cellulose at high temperatures. The hyperthermophilic cellulase has promise for applications in biomass utilization. To clarify its detailed function, we determined the crystal structures of mutants of the enzyme in complex with either the substrate or product ligands. We were able to resolve different kinds of complex structures at 1.65–2.01 Å (1 Å = 0.1 nm). The structural analysis of various mutant enzymes yielded a sequence of crystallographic snapshots, which could be used to explain the catalytic process of the enzyme. The substrate position is fixed by the alignment of one cellobiose unit between the two aromatic amino acid residues at subsites + 1 and + 2. During the enzyme reaction, the glucose structure of cellulose substrates is distorted at subsite − 1, and the β1,4-glucoside bond between glucose moieties is twisted between subsites − 1 and + 1. Subsite − 2 specifically recognizes the glucose residue, but recognition by subsites + 1 and + 2 is loose during the enzyme reaction. This type of recognition is important for creation of the distorted boat form of the substrate at subsite − 1. A rare enzyme–substrate complex was observed within the low-activity mutant Y299F, which suggested the existence of a trapped ligand structure before the formation by covalent bonding of the proposed intermediate structure. Analysis of the enzyme– substrate structure suggested that an incoming water molecule, essential for hydrolysis during the retention process, might be introduced to the cleavage position after the cellobiose product at subsites + 1 and + 2 was released from the active site. INTRODUCTION has been provided by visualization of a covalent glycosyl-enzyme intermediate by X-ray crystallography of enzyme–ligand complex structures formed with saccharide-substrate analogues substituted with fluoride at the 2-position of the glucopyranoside [8–10]. This covalent intermediate was also observed between a native saccharide substrate and mutant enzyme [11–13]. Therefore the double-displacement mechanism involving the formation of a covalent intermediate has been generally accepted in this class of enzymes. Structural data necessary to elucidate the catalytic mechanism of hyperthermophilic endocellulase have long been awaited. In a previous study, we prepared a protein crystal using truncated forms of EGPh [14], and the protein structure of EGPh was determined by X-ray crystallography at high resolution [15]. However, we were not able to obtain any useful X-ray diffraction data of protein complexed with either substrate or product ligand. In the present study, under different crystallization conditions, we succeeded in preparing an alternative crystal for analysis of the enzyme–ligand complex structure, in which substrate or product ligand was competently bound to the active centre. Three molecules of EGPh in the crystallographic asymmetric unit were identified, allowing us to obtain many different types of the complexed structure. Crystallographic snapshots of the sequential events during the enzymatic reaction were obtained from various mutant enzyme–ligand complex structures. These snapshots described the detailed catalytic process of EGPh. Using these structural data, we analysed the detailed role of the subsite structure, as well as the catalytic mechanism of hyperthermophilic endocellulase. Cellulase is a key enzyme involved in the degradation of βglucan cellulose biomass. A hyperthermophilic cellulase would be particularly useful in industrial applications. A hyperthermophilic archaeon, Pyrococcus sp., was isolated from a hydrothermal volcanic vent in the ocean [1,2]. Hyperthermophilic β-1,4endocellulases (endo-type cellulases) have been identified in the genome sequences of several hyperthermophilic archaea. The first hyperthermophilic endo-type cellulase (family 12) was found in Pyrococcus furiosus; however, this enzyme exhibited relatively low activity towards cellulose [3]. We previously isolated a unique membrane-related hyperthermophilic endocellulase from Pyrococcus horikoshii (termed EGPh; family 5) exhibiting relatively high activity towards cellulose [4,5]. A membrane-related hyperthermophilic β-1,4-glucosidase has also been isolated from P. horikoshii [6]. Based on our current knowledge, it appears that archaea use a primitive mechanism, utilizing two enzymes to obtain glucose from cellulose. The family 5 glycosyl hydrolase that belongs to clan GHA (glycoside hydrolase clan A) is a large superfamily with divergent substrate specificity (http://www.cazy.org/GH5.html). The superfamily has also been referred to as the 4/7 superfamily because the three key residues are adjacent asparagine/glutamate residues at the end of β-strand 4 and a glutamate residue at the end of β-strand 7. Retention is the accepted catalytic mechanism of the family 5 enzymes, although the fundamental mechanism has been the subject of debate for many years. Evidence of the doubledisplacement mechanism originally suggested by Koshland [7] Key words: cellulose, endocellulase, endoglucanase, hydrolytic mechanism, hyperthermophile, Pyrococcus horikoshii. Abbreviations used: EGAc, family 5 endocellulase E1 from Acidothermus cellulolyticus ; EGPh, hyperthermophilic endocellulase from Pyrococcus horikoshii ; G2, cellobiose; G4, cellotetraose; PSA, phosphoric acid-swollen Avicel; rmsd, root mean square deviation; WT, wild-type. 1 To whom correspondence should be addressed (email [email protected]). c The Authors Journal compilation c 2011 Biochemical Society 224 H.-W. Kim and K. Ishikawa EXPERIMENTAL RESULTS Construction and preparation of truncated protein and site-directed mutagenesis Crystallization of the enzyme–ligand complex The protein crystallization and structural analysis were conducted using EGPhN5C5, which is the truncated protein lacking five amino acid residues from both the N- and C-terminal ends of EGPh (Gene ID PH1171, EC 3.2.1.4) [16]. Point-mutant enzymes were prepared by site-directed mutagenesis using the QuikChange® mutagenesis method (Stratagene). All mutant genes were inserted into the expression vector pET11a (Novagen). The constructed plasmids were introduced into Escherichia coli strain BL21(DE3) for recombinant protein expression. Expression and purification of the recombinant enzymes was carried out following a method described previously [5,16]. Crystallization The purified proteins were dialysed against 50 mM Tris/HCl buffer (pH 8.0) and then concentrated to 20 mg·ml − 1 . Crystallization was performed using the hanging-drop vapourdiffusion method. The drops consisted of equal volumes (1.5 μl) of the protein and reservoir solutions. The crystal of the enzyme– ligand complex was prepared from EGPhN5C5 using a reservoir solution consisting of 1.5 M ammonium phosphate and 0.1 M Mes buffer (pH 6.5). The crystals were obtained over a period of approximately 3 days at 22 ◦ C. The crystal of the EGPhN5C5– ligand complex was co-crystallized under the same conditions using a final concentration of 100 mM cellobiose (G2). In the case of the mutants (E201A, E342A and Y299F), the complex with cellotetraose (G4) was obtained by soaking a crystal for 2 h in the used reservoir solution containing 10 mM G4. Data collection and processing The crystals were collected with a CryoLoopTM (Hampton Research) and immediately flash-cooled at 100 K in a nitrogen cryostream. Diffraction data of crystals for the complex were collected at BL38 and BL44, SPring-8 (Harima, Japan), and processed and scaled using the HKL2000 and CCP4 (Collaborative Computational Project, Number 4, 1994) [17] packages. All model-building stages were performed with Coot [18]. The modelling of the structure of the enzyme– ligand complex was performed with Coot, and the refinement was performed with REFMAC5 [17] and CNS [19]. The dictionary of ligand is derived from the HIC-Up database (http://xray.bmc.uu.se/hicup/). After the incorporation of the ligand model in the F o − F c omit map, refinement of the enzyme– ligand complex was carried out using REFMAC5 or CNS. The diffraction data statistics and the crystallographic refinement statistics are summarized in Table 1. Figures were produced using PyMOL (http://www.pymol.org). In a previous study [14], we prepared the crystal of the truncated protein EGPhC5 of EGPh lacking five amino acid residues from the C-terminus, and determined the apo-enzyme structure [15]. However, we could not obtain X-ray diffraction data suitable for determination of the structure of the enzyme–substrate ligand complex. Failure to obtain suitable diffraction data might have been caused by the presence of zinc ions tightly bound between the two catalytic glutamate residues, which present an obstacle for the entrance of ligand in the active site. In the crystal screening using EGPhN5C5 (the truncated protein lacking five amino acid residues from both the N- and C-terminal ends of EGPh) [16], we could prepare a new crystal with reagent number 48 [2.0 M ammonium phosphate and 0.1 M Tris/HCl (pH 8.5)] in the crystal screening kit (Hampton Research), thus optimizing the crystallization conditions as described in the Experimental section. The new crystal belongs to the space group C2 with unit cell dimensions a = 162 Å (1 Å = 0.1 nm), b = 58 Å and c = 138 Å, and β = 109 Å. We identified three molecules of EGPhN5C5 in the crystallographic asymmetric unit. The rmsd (root mean square deviation) of the Cα atom was less than 0.5 Å among the three molecules (hereafter referred as MolA, MolB and MolC respectively), suggesting that they were almost identical. These structures of EGPhN5C5 are referred to as WT (wild-type) EGPh in the present paper. In the packing of the crystal form, the speculative active site of MolA and MolB does not exhibit sufficient space for binding of cello-oligosaccharide substrate. In MolC, however, sufficient space for substrate binding was observed between MolC and MolB. We prepared crystals of the EGPhN5C5–cellobiose complex (WT–G2) under these conditions, and determined its structure. The structure was refined within 1.9 Å (Table 1). Unlike family 5 endocellulase E1 from Acidothermus cellulolyticus (hereafter referred to as EGAc) [22], we did not observe an electron-density map for any condensed cello-oligosaccharides in the active site of WT–G2. This result was consistent with HPLC analysis of the product from the reaction mixture containing G2 (results not shown). In the structure of WT–G2, three protein molecules (MolA, MolB and MolC) in the crystallographic asymmetric unit exhibited a high F o − F c electron-density map corresponding to G2 at subsites + 1 and + 2. Figure 1 shows the structure of MolA, MolB and MolC. The supposed catalytic residues (Glu201 and Glu342 ) were located at the non-reducing end of the bound G2. The G2 at subsites + 1 and + 2, [Glc( + 1) and Glc( + 2)], was mainly stacked by Trp273 and Tyr304 . In the structure of G2 between Trp273 and Tyr304 , the non-polar surface was co-planar with the hydrophobic part of Glc( + 1) and Glc( + 2). The electron density corresponding to G2 at subsites − 2 and − 1 was poor compared with subsites + 1 and + 2. However, the existence of some substance at the subsites could be sufficiently recognized by the F o − F c omit map contoured at 3.0 σ in MolA, MolB and MolC (Figure 1). Activity measurement The hydrolytic activity of the enzymes toward 0.5% Avicel (Merck) was determined by measuring the amount of reduced sugars released from substrate at 85 ◦ C in 100 mM acetate buffer (pH 5.5). The reduced sugars were determined by a modified Somogi–Nelson method [20] using glucose as a standard. PSA (phosphoric acid-swollen Avicel), used as a substrate, was prepared by swelling in 85% phosphoric acid, and then by removing phosphoric acid by filtration and centrifugation according to the method described previously [21]. c The Authors Journal compilation c 2011 Biochemical Society Michaelis complex structure of the catalytic residue mutants To examine the enzyme–substrate complex (Michaelis complex) structure, we prepared crystals of the inactive mutants, E201A and E342A, in which Glu201 and Glu342 were replaced by alanine. The crystals were soaked with G4 in the crystallization solution as described in the Experimental section. The refined protein structures revealed no distinct change in comparison with the apo-enzyme structure. Functional analysis of hyperthermophilic cellulase Table 1 225 Statistics of data collection and refinement Values for the highest-resolution shell are given in parentheses. Statistic Data collection Wavelength (Å) Space group Unit-cell parameters (Å) Matthews coefficient (Å 3 ·Da − 1 ) Solvent content (%) Subunits per asymmetric unit Resolution range (Å) Number of observed reflections Total number of unique reflections <I/ σ (I )> Multiplicity R merge * Completeness (%) Refinement Resolution used in refinement R cryst (%)† R free (%)‡ Rmsd of Bond distance (Å) Bond angle (◦ ) Average B factor of Protein molecules (Å2 ) Ligand molecules (Å2 ) Ramachandran plot In most favoured regions (%) In disallowed regions (%) PDB code WT–G2 E201A–G4 E342A–G4 Y299F–G4 0.9 C2 a = 162.2, b = 58.4, c = 138.3, β = 109.2 2.38 48.26 3 50–1.90 (1.97–1.90) 281105 96770 24.1 (3.1) 3.1 (2.5) 0.042 (0.210) 94.9 (85.9) 1.0 C2 a = 162.3, b = 58.3, c = 138.3, β = 109.6 2.33 47.32 3 50–2.02 (2.10–2.02) 269527 83759 12.1 (2.7) 3.4 (3.1) 0.072 (0.287) 95.1 (82.1) 1.0 C2 a = 162.5, b = 58.6, c = 139.1, β = 109.7 2.36 47.93 3 50–2.01 (2.05–2.01) 209856 82731 16.4 (2.6) 2.9 (2.2) 0.066 (0.328) 88.3 (77.8) 0.9 C2 a = 163.1, b = 58.4, c = 138.2, β = 109.5 2.35 47.67 3 50–1.65 (1.68–1.65) 464216 147323 16.4 (3.5) 3.3 (2.6) 0.064 (0.308) 94.9 (84.8) 50.00–1.90 (2.02–1.90) 22.8 (35.4) 24.1 (36.2) 28.99–2.00 (2.13–2.00) 19.6 (25.0) 24.8 (33.5) 35.25–2.01 (2.14–2.01) 22.8 (37.6) 26.1 (40.7) 34.90–1.65 (1.69–1.65) 18.8(30.6) 22. 6 (33.7) 0.007 1.7 0.022 1.8 0.007 1.6 0.025 2.0 37.0 51.9 29.7 39.2 47.9 54.1 24.84 31.71 85.1 0.0 3AXX 85.1 0.0 3QHN 85.4 0.1 3QHM 98.0 0.0 3QHO *R merge = hkl i |Ii (hkl ) − <I (hkl )>| hkl i Ii (hkl ), where Ii (hkl ) is the i -th intensity measurement of reflection hkl , including symmetry-related reflections, and <I (hkl )> is their average. †R cryst = hkl |F o − F c |/ hkl F o , where F o and F c are the observed and calculated structure factor amplitudes of reflection hkl respectively. ‡R free is calculated as the R cryst , using F o that were excluded from the refinement (5% of the data). In the structures of E201A–G4 and E342A–G4, four glucose residues of G4 were clearly observed in the F o − F c omit map from subsite − 2 to subsite + 2 in the active-site cleft of MolA and MolB (Figure 2). In MolC, however, G4 was observed at subsites − 4 to − 1, and G2 was exhibited at subsites + 1 to + 2 on the calculated electron-density map (Figure 2). The Glc( − 4) of MolC was stacked by Arg44 of MolB in the crystal form that was lying on Trp82 of putative subsite − 4 of MolC. In the mutational analysis, the activity of the W82A mutant was 25% lower than that of WT enzyme (Table 2). In the other structures of E201A–G4 and E342A–G4, one G4 unit was observed along the active-site cleft, and each glucose residue interacted with the stacking residues (Phe69 , Trp273 and Tyr304 ) at subsites − 2, + 1 and + 2 respectively (Figure 2). However, the conformations of Glc( − 1) at subsite − 1 were significantly different in the structures of MolA. The substratebinding pattern in MolB of E201A–G4 was almost identical with that in MolA and MolB of E342A–G4 (Figure 2). In these structures, Glc( − 1) does not interact with the corresponding stacking residue Trp377 . In the case of MolA of E201A–G4 (Figure 2), however, Glc( − 1) interacted with Trp377 at subsite − 1, so that the sugar conformation was changed from the relaxed chair form to the distorted boat (1,4 B) form of the β-anomer. Figure 3 shows the hydrogen-bond interaction between Glc( − 1) and the conserved residues (Glu201 , Tyr299 and Glu342 ) in the three structures. In MolA and MolB of E342A–G4, the 2 -OH group of Glc( − 1) hydrogen-bonds with the putative proton-donor oxygen (O2) of Glu201 (Figures 2 and 3A). In MolB of E201A–G4, a unique hydrogen bond between the 2 -OH group of the Glc( − 1) and oxygen (O1) of Glu342 was observed (Figure 3B). In MolA of E201A–G4, however, the 2 -OH group of Glc( − 1) hydrogenbonds with two oxygens (O1 and O2) of Glu342 (Figure 3C). The sequence of (A)→(B)→(C) in Figure 3 suggests a sequence of snapshots that represent the formation of the stable Michaelis complex structure. During the enzyme reaction, the hydrogen bond between the 2 -OH group of Glc( − 1) and the proton-donor oxygen (O2) of Glu201 is shifted to another bond between the 2 -OH group of Glc( − 1) and Glu342 . In the following stage, a stable complex structure (the distorted boat form) is formed by hydrogen-bonding with Glu342 and Tyr299 (Figure 3C). In the superimposed structures of three enzyme–ligand complex structures (WT–G2, E342A–G4 and E201A–G4), the location of Glc( + 1) and ( + 2) was slightly different, even though they were inside the same subsite position (Figure 4). The position of the bound G2 unit at Glc( + 1) and ( + 2) shifted towards the active centre by the formation of the distorted boat form of Glc( − 1) (Figure 4). Despite this positional shift of the G2 unit, a common hydrogen bond was observed between the 2 -OH group of Glc( + 2) and the carboxy group of Gln306 (Figure 4). To understand the role of Gln306 , we examined the activity of mutant Q306A. The mutant Q306A exhibited only 25% of the enzymatic activity exhibited by the WT against the substrate PSA (Table 2). Michaelis complex structure of the mutant Y299F In the structure of MolA in E201A–G4 (Figure 3C), Tyr299 is located within hydrogen-bond distance (2.97 Å) of the c The Authors Journal compilation c 2011 Biochemical Society 226 H.-W. Kim and K. Ishikawa Å. Furthermore, a rare hydrogen bond between the 2 -OH group of Glc( − 1) and the carboxy group of Glu201 (2.80 Å) was found in the structure of Y299F–G4 (Figure 6A). Structural change in the subsite − 1 position by catalytic mutation The overall G4 conformation of MolA in E201A–G4 was identical with that of Y299F–G4, but Glc( − 1) of MolA in E201A–G4 showed a subtle shift (∼0.44 Å) towards Glu201 replaced by alanine (Figure 5) that influences the glycosidic linkage oxygen of the scissile bond in the substrate (G4) structure. The shift of Glc( − 1) seems to result from the lack of a spatial barrier caused by the absence of a side chain on the Glu201 residue. In the structure of Y299F–G4, the 2 -OH group of Glc( − 1) exhibited hydrogen-bonding with Glu201 and Glu342 , the key catalytic residues. The 2 -OH group of Glc( − 1) showed hydrogen-bonding with the carboxy group of Glu342 , and was simultaneously located within hydrogen-bonding distance (2.84– 3.05 Å) of the proton-donor oxygen (O2) of Glu201 (Figure 6A). As mentioned above, this rare hydrogen-bond pattern has not been reported in any other glycosyl hydrolase. DISCUSSION Figure 1 Crystallographic snapshots of the EGPhN5C5–cellobiose (WT– G2) complex The active sites of three structures in the asymmetric unit are shown in divergent-eyed stereo view. The F o − F c omit maps of the subsites − 2 to + 2 contoured at 3.0 σ were calculated prior to incorporation of G2 at the subsites + 1 and + 2 in the model structures. endocyclic oxygen (O5) of Glc( − 1), and simultaneously contacts (2.55 Å) the nucleophilic oxygen (O1) of Glu342 . To clarify a functional role for Tyr299 in the catalytic action of EGPh, we determined the structure of Y299F complexed with G4. The mutant Y299F exhibits slight hydrolytic activity towards PSA, producing G2 (Table 2). The crystal structure of Y299F–G4 was resolved at 1.65 Å. In the active-site cleft of Y299F–G4, the whole G4 structure was clearly observed in the F o − F c omit map. Subsites − 2 to + 2 were covered by the bound G4, and the structure of Glc( − 1) also exhibited the distorted boat form similar to that of MolA of E201A–G4 (Figures 5 and 6A). However, a strong negative electron density was found in the scissile bond of the substrate structure model from the final refinement (Figure 6B). Figure 5 shows the superimposed structures from E201A– G4(MolA), Y299F–G4 and WT–G2(MolA). The overall structure of Y299F–G4 shows no significant difference with MolA of E201A–G4 structure (Figure 5). In the superimposed structure of Y299F–G4 and the other structures, however, a noticeable change was observed in the χ angle of Glu201 (Figure 5). The difference of the χ angle of Glu201 was also not observed in the structure of Y299F. A noticeable change was induced by the binding of G4 in Y299F. The distance between the oxygen of the carboxy group in Glu201 and the C1 atom of Glc( − 1) was estimated to be 3.10 c The Authors Journal compilation c 2011 Biochemical Society In a previous study [15], we determined the crystal structure of EGPhC5, but failed to obtain useful data to indicate the structure of the enzyme–substrate or enzyme–product complex. This seems to have been caused by the presence of zinc ions tightly bound between the two catalytic glutamate residues, which obstructed the entrance of ligand into the active-site cleft. An alternative crystal of EGPhN5C5, which was prepared in a crystallization solution without any bivalent metal cations, was used for the analysis of the enzyme–ligand structures. Unlike the previous crystal [14], this new crystal involved three protein molecules in the crystallographic asymmetric unit. These protein structures were almost identical, but their active sites exhibited some differences, which allowed us to obtain various crystallographic snapshots of the enzyme–ligand complex structure. These snapshots aided our understanding of the substrate-binding and catalytic mechanism of the enzyme. Michaelis complex structure In E201A–G4 and E342A–G4, the structures of MolC exhibited a different substrate-binding pattern from those of MolA and MolB. Two substrate molecules were separately bound on both sides of the catalytic centre in MolC (Figure 2). This substrate-binding pattern, which has not been observed in other family 5 enzymes, revealed the additional subsites − 4 and − 3, which indicate that the main substrate-binding site of EGPh stretches from subsite − 4 to subsite + 2. That W82A exhibits low activity towards PSA (Table 2) suggests that Trp82 functions as a stacking residue at subsite − 4. It was also confirmed that Trp82 is conserved in subfamily 1 of family 5 cellulases (results not shown). The finding of the additional substrate-binding site (subsites − 3 and − 4) was due to the sufficient space for substrate binding at the active site of MolC in the crystal form. In both MolA and MolB, the substrate-binding space (subsites − 3 and − 4) was blocked by the interface of the neighbour molecule in the crystal form. The structures of MolA and MolB in E342A–G4 (Figure 2) provide an insight into the early stage of the substrate-binding process (Figures 3 and 4). The result that Glc( − 1) in E342A–G4 exhibits no stacking interaction with the corresponding aromatic residue (Trp377 ) indicates that Glu342 is involved in the stacking interaction between Trp377 and Glc( − 1). The structure of Glc( − 1) of MolB Functional analysis of hyperthermophilic cellulase Figure 2 227 Conformation of cellotetraose (G4) bound to the inactive mutants (E201A and E342A) Left-hand panels in MolA show divergent-eyed stereo views of the model structures and F o − F c omit maps (contour level 3.0 σ ) calculated prior to incorporation of G4 in the model. MolC of E342A–G4 also represents a F o − F c omit map (contour level 3.0 σ ) calculated without the contribution of G4 and G2. The structures of E201A–G4 and E342A–G4 are divided into two horizontal sections respectively. in E201A–G4 exhibits hydrogen-bonding between the 2 -OH group of Glc( − 1) and the O1 atom of Glu342 . Interestingly, this structure is similar to that of Glc( − 1) of MolA and MolB in E342A–G4 (Figures 2 and 3). Therefore Glu342 appears to attract a glucose moiety through the interaction between the 2 -OH group of Glc( − 1) and Glu342 in the initial substrate-binding stage. After this pulling action, the following stacking interaction of Glc( − 1) and Trp377 seems to change Glc( − 1) from the normal chair form into the distorted boat form (Figure 3). The resulting distorted boat form would have a tendency to revert to the initial chair form because of its unstable structural conformation. The electron density at subsites + 1 and + 2 in the WT– G2 structure was stronger than that of subsites − 1 and − 2, indicating that subsites + 1 and + 2 have more stable and stronger binding affinity towards the substrate than the other subsites. Although this strong binding affinity exists at subsites + 1 and c The Authors Journal compilation c 2011 Biochemical Society 228 Figure 3 H.-W. Kim and K. Ishikawa Interactions of the Glc( − 1) and the catalytic residues in the active centre of the inactive mutants (E201A and E342A) The refined structures from the enzyme–ligand complexes exhibit various hydrogen-bond interactions between the catalytically important residues (Glu201 , Tyr299 and Glu342 ) and the proximal glucose unit Glc( − 1). The catalytic residues are shown as a cyan stick model. Possible hydrogen bonds are indicated by broken lines. Figure 4 Comparison of the ligand-binding pattern in the active site The structures of ligand from the three enzyme–ligand complexes [WT–G2 (green stick), E201A–G4 (yellow stick) and E342A–G4 (white stick)] were superimposed based on the conserved residues of family 5 glycosyl hydrolases in the active centre. The superimposed structures are shown in divergent-eyed stereo view. Table 2 Mutational analysis of the EGPhN5C5 enzyme Enzyme Relative activity (%) against PSA EGPhN5C5 (WT) W82A Q306A Y299F W377A 100 75 25 <1 <1 + 2, stretching of the glucose moiety at the position was observed, as shown in Figure 4. In addition, since the position of Glc( − 2) was not changed in these structures (Figure 4), the recognition of glucose residue at subsite − 2 is specific. The movement of the bound glucose residues at subsites + 1 and + 2, and the specific binding at subsite − 2, are important for forming the distorted boat form of Glc( − 1) at subsite − 1. Characterization of the catalytic process From the structural analysis, a narrower cleft was observed at subsites + 1 and + 2, when compared with that of subsites − 1 and − 2 at the active site. At subsites + 1 and + 2, two aromatic amino acid residues (Trp273 and Tyr304 ) stack one G2 unit and seem to fix the position of the substrate. During the hydrolytic reaction process, the glucose residue of cellulose is distorted at subsite − 1, and the β-1,4-glucoside bond of cellulose is twisted between c The Authors Journal compilation c 2011 Biochemical Society Figure 5 Divergent-eyed stereo views of the superimposed structures representing the slight conformational change from three enzyme–ligand complexes The structures of MolA of E201A–G4 (yellow stick), Y299F–G4 (white stick) and WT–G2 (green stick) were superimposed based on the conserved residues. The glycosidic linkage oxygen of the scissile bond in MolA of E201A–G4 is within hydrogen-bond distance with the proton-donor residue, Glu201 , of WT–G2. In Y299F–G4, the χ angle of Glu201 was significantly changed compared with that of WT–G2. subsites − 1 and + 1 perpendicular to the cellulose fibre. The final stage of the reaction is the release of the G2 from subsites + 1 and + 2. Functional analysis of hyperthermophilic cellulase Figure 6 Divergent-eyed stereo view of the active centre in the structure of Y299F–G4 (A) Hydrogen-bond interaction between Glc( − 1) and the catalytic residues. The interaction exhibits a rare hydrogen bond between the 2 -OH group of the Glc( − 1) and the proton-donor oxygen (O2) of Glu201 . (B) Electron-density maps corresponding to Glc( − 1) and Glc( + 1) in the structural analysis of Y299F–G4. Postive density of the F o − F c omit map, calculated prior to incorporation of G4 in the model, is shown in green mesh (contour level 3.0 σ ). The negative density of F o − F c map is in blue (contour level 3.0 σ ), which is generated from final refinement of the model structure involving G4 in the active centre. The model refinement was performed using REFMAC 5 TLS refinement. That glucose and cellotriose were not released from the reducing end of cello-oligosaccharides by EGPh indicates that the affinity of subsite + 2 in EGPh is strong, and that the binding of a glucose residue at subsite + 2 is important for catalytic activity. The strong affinity of substrate (G2 unit) at subsites + 1 and + 2, and the strong negatively charged environment at the catalytic centre (Glu342 ) characterize EGPh. This characteristic may explain why EGPh has no condensation reaction activity, unlike EGAc [22]. Role for Tyr299 in the retaining mechanism The critical role of the conserved amino acid residues in family 5 enzymes were confirmed through mutational analysis [16]. Among these conserved residues, Trp377 and Tyr299 showed structural and functional compatibility with family 5 cellulase as well as family 18 chitinase [23]. The interaction between Tyr299 and the endocyclic oxygen (O5) of Glc( − 1) observed in the structure (Figure 3C) has been reported in the 2-fluoro glucosyl-enzyme structure (covalent intermediate structure) of family 1 myrosinase [24] and family 11 xylanase [25]. This interaction was proposed to be an oxygen–oxygen interaction, not a hydrogen bond, which might serve to stabilize the positively 229 charged oxocarbonium-like state through charge redistribution with the negatively charged nucleophilic residue [25]. To clarify the functional role of Tyr299 , we constructed a Y299F mutant and elucidated its structure complexed with G4. The mutant Y299F exhibits slight hydrolytic activity towards G4. However, the Y299F–G4 structure revealed a unique complex structure involving the key catalytic residues (Glu201 and Glu342 ) and G4. The strong negative electron density that was generated from the final refinement involving G4 indicates that the bound G4 was digested between Glc( − 1) and Glc( + 1). The structure in Figure 6 represents the rare oxocarbonium-like structure trapped by the carboxy group of Glu201 of the mutant Y299F. Interestingly, the structure of Glc( − 1) with the distorted boat form showed a rare hydrogen bond between the 2 -OH group and the protondonor oxygen (O2) of Glu201 , which causes a noticeable change in the χ angle of Glu201 (Figures 5 and 6A). The χ angle of Glu201 observed in the Y299F–G4 structure was not observed in the Y299F structure. Furthermore, no change in the χ angle of Glu201 was observed in the other mutants. This rare hydrogen bond might be due to a change of the hydrogen-bond network associated with protonation of Glu342 caused by the absence of the OH group from Tyr299 . These results indicate that the structure of Y299F–G4 is a unique intermediate structure in the hydrolysis of G4 by Y299F. The pH activity profile of the Y299F mutant was shifted towards alkaline pH, supporting the fact that protonation of Tyr299 is important for catalytic action [16]. Thus the intermediate states following the cleavage might not proceed in an ordered sequence because of the hydrogen-bond interaction between the 2 -OH group of Glc( − 1) and the proton-donor oxygen (O2) of Glu201 in Y299F. The results indicate that Tyr299 may be the protondonor residue and mediate the formation of the enzyme–substrate intermediate during the catalytic process. Furthermore, it was clarified that the OH group in Tyr299 is not necessarily important for stabilizing the positively charged oxocarbonium-like state or the distorted boat form. The carboxy group in Glu201 seems to compensate for the role of Tyr299 . Unlike the case of other enzymes [8–10,12,26], we could not obtain any crystallographic snapshot of an intermediate structure with a covalent bond between enzyme and ligand in the structural analysis of EGPh. It seems to be difficult for EGPh to form an intermediate because of the steric hindrance between the 2 -OH group of Glc( − 1) and the carboxy group of Glu342 (Figure 3C). Therefore, during the catalytic process, EGPh seems to attract and position the substrate at the active site through the hydrogenbond interaction, as suggested by the sequence of (A)→(B)→(C) (Figure 3). After formation of the enzyme–substrate complex (C), positional change of Glu342 accompanied by the movement of Tyr299 seems to be induced to form the proposed intermediate with a covalent bond. The structure of WT–G2 involves the final stage in the catalytic events. In MolA, the electron density of G2 at subsites − 1 and − 2 was recognized at 3.0 σ , although it was not clear (Figure 1). This poor density is due to its disorder or heterogeneity. MolB and MolC also have a planar and relatively strong unassigned electron density at subsites − 1 and − 2. In MolB, we clearly observed one water molecule in close proximity (2.50 Å) to Glu201 (Figure 1, MolB). The position of the water molecule was compatible with that of the O1 atom of the β-faced 1 -OH group of Glc( − 1) in MolA (Figure 1, MolB and Figure 3C). This poor electron density at subsites − 2 and − 1 in MolB seems to represent the intermediate at the final stage. In these structures, the position of the bound G2 at subsites + 1 and + 2 was clearly observed. The position of the G2 shifts to the outside (approximately 2.10 Å), compared with the structure of the E201A–G4, E342A–G4 and Y299F–G4 complexes (Figure 5). Therefore, at the final stage, an c The Authors Journal compilation c 2011 Biochemical Society 230 H.-W. Kim and K. Ishikawa incoming water molecule essential for hydrolysis via the retaining mechanism seems to be introduced to the cleavage site after the bound product G2 moves towards the outside of the active cleft. During the steady state of enzyme catalysis, the intermediate form is easily changed to a more stable state. Therefore obtaining protein crystals of the intermediate form is typically very difficult. To address this problem, the intermediate structure has generally been studied using chemical quenching methods involving the use of enzyme mutations or substrate analogues [9,10,25,27]. However, the hyperthermophilic nature of EGPh allowed us to study the intermediate forms more easily. The optimal temperature for EGPh is >100 ◦ C. This is significantly higher than the temperature required for crystallization (22 ◦ C). Thus, during the co-crystallization of the enzyme and the ligands, the catalytic reaction might appear to be suppressed by temperature quenching [27], resulting in a rare intermediate form that was relatively well maintained in crystalline form. AUTHOR CONTRIBUTION Han-woo Kim performed the experiments and designed the experiments. Kazuhiko Ishikawa contributed to the experimental design and conceived the study. ACKNOWLEDGEMENTS We thank Ms Mai Nakatsuka for her help with the preparation of enzymes, and Dr Koich Uegaki, Dr Tsutomu Nakamura, Dr Yasunobu Wada and Mr Yuji Kado for their kind suggestions. The X-ray diffraction experiments were carried out with the approval of the Japan Synchrotron Radiation Research Institute (Hyogo, Japan). We thank to Professor Henry Nelson of Kinki University (Nara, Japan) for his helpful comments and critical reading of this manuscript prior to submission. FUNDING This work was supported by the Japan Society for the Promotion of Science postdoctoral fellowship programme. REFERENCES 1 Lipp, J. S., Morono, Y., Inagaki, F. and Hinrichs, K. U. (2008) Significant contribution of Archaea to extant biomass in marine subsurface sediments. Nature 454, 991–994 2 Gonzalez, J. M., Masuchi, Y., Robb, F. T., Ammerman, J. W., Maeder, D. L., Yanagibayashi, M., Tamaoka, J. and Kato, C. (1998) Pyrococcus horikoshii sp. nov., a hyperthermophilic archaeon isolated from a hydrothermal vent at the Okinawa Trough. Extremophiles 2, 123–130 3 Bauer, M. W., Driskill, L. E., Callen, W., Snead, M. A., Mathur, E. J. and Kelly, R. M. (1999) An endoglucanase, EglA, from the hyperthermophilic archaeon Pyrococcus furiosus hydrolyzes β-1,4 bonds in mixed-linkage (1→3),(1→4)-β-D-glucans and cellulose. J. Bacteriol. 181, 284–290 4 Ando, S., Ishida, H., Kosugi, Y. and Ishikawa, K. (2002) Hyperthermostable endoglucanase from Pyrococcus horikoshii . Appl. Environ. Microbiol. 68, 430–433 5 Kashima, Y., Mori, K., Fukada, H. and Ishikawa, K. (2005) Analysis of the function of a hyperthermophilic endoglucanase from Pyrococcus horikoshii that hydrolyzes crystalline cellulose. Extremophiles 9, 37–43 Received 14 February 2011/28 April 2011; accepted 11 May 2011 Published as BJ Immediate Publication 11 May 2011, doi:10.1042/BJ20110292 c The Authors Journal compilation c 2011 Biochemical Society 6 Matsui, I., Sakai, Y., Matsui, E., Kikuchi, H., Kawarabayasi, Y. and Honda, K. (2000) Novel substrate specificity of a membrane-bound β-glycosidase from the hyperthermophilic archaeon Pyrococcus horikoshii . FEBS Lett. 467, 195–200 7 Koshland, D. E. J. (1953) Stereochemistry and the mechanism of enzymatic reactions. Biol. Rev. 28, 416–436 8 White, A., Tull, D., Johns, K., Withers, S. G. and Rose, D. R. (1996) Crystallographic observation of a covalent catalytic intermediate in a β-glycosidase. Nat. Struct. Biol. 3, 149–154 9 Vocadlo, D. J., Davies, G. J., Laine, R. and Withers, S. G. (2001) Catalysis by hen egg-white lysozyme proceeds via a covalent intermediate. Nature 412, 835–838 10 Notenboom, V., Birsan, C., Nitz, M., Rose, D. R., Warren, R. A. and Withers, S. G. (1998) Insights into transition state stabilization of the β-1,4-glycosidase Cex by covalent intermediate accumulation in active site mutants. Nat. Struct. Biol. 5, 812–818 11 Czjzek, M., Ben David, A., Bravman, T., Shoham, G., Henrissat, B. and Shoham, Y. (2005) Enzyme-substrate complex structures of a GH39 β-xylosidase from Geobacillus stearothermophilus . J. Mol. Biol. 353, 838–846 12 Hovel, K., Shallom, D., Niefind, K., Belakhov, V., Shoham, G., Baasov, T., Shoham, Y. and Schomburg, D. (2003) Crystal structure and snapshots along the reaction pathway of a family 51 α-L-arabinofuranosidase. EMBO J. 22, 4922–4932 13 Kuroki, R., Weaver, L. H. and Matthews, B. W. (1993) A covalent enzyme-substrate intermediate with saccharide distortion in a mutant T4 lysozyme. Science 262, 2030–2033 14 Kim, H. W., Mino, K. and Ishikawa, K. (2008) Crystallization and preliminary X-ray analysis of endoglucanase from Pyrococcus horikoshii . Acta Crystallogr. Sect. F Struct. Biol. Crystallogr. Commun. 64, 1169–1171 15 Kim, H. W. and Ishikawa, K. (2010) Structure of hyperthermophilic endocellulase from Pyrococcus horikoshii . Proteins 78, 496–500 16 Kang, H. J. and Ishikawa, K. (2007) Analysis of active center in hyperthermophilic cellulase from Pyrococcus horikoshii . J. Microbiol. Biotechnol. 17, 1249–1253 17 Collaborative Computational Project, Number 4 (1994). The CCP4 suite: programs for protein crystallography. Acta Crystallogr. Sect. D Biol. Crystallogr. 50, 760–763 18 Emsley, P. and Cowtan, K. (2004) Coot: model-building tools for molecular graphics. Acta Crystallogr. Sect. D Biol. Crystallogr. 60, 2126–2132 19 Brunger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., Grosse-Kunstleve, R. W., Jiang, J. S., Kuszewski, J., Nilges, M., Pannu, N. S. et al. (1998) Crystallography and NMR system: a new software suite for macromolecular structure determination. Acta Crystallogr. Sect. D Biol. Crystallogr. 54, 905–921 20 Hiromi, K. and Ono, S. (1963) Time course of consecutive reactions catalyzed by exo-enzymes. J. Biochem (Tokyo). 53, 164–166 21 Wood, T. M. (1988) Preparation of crystalline, amorphous, and dyed cellulose substrates. Methods of Enzymol. 160, 19–25 22 Sakon, J., Adney, W. S., Himmel, M. E., Thomas, S. R. and Karplus, P. A. (1996) Crystal structure of thermostable family 5 endocellulase E1 from Acidothermus cellulolyticus in complex with cellotetraose. Biochemistry 35, 10648–10660 23 van Aalten, D. M., Komander, D., Synstad, B., Gaseidnes, S., Peter, M. G. and Eijsink, V. G. (2001) Structural insights into the catalytic mechanism of a family 18 exo-chitinase. Proc. Natl. Acad. Sci. U.S.A. 98, 8979–8984 24 Burmeister, W. P., Cottaz, S., Driguez, H., Iori, R., Palmieri, S. and Henrissat, B. (1997) The crystal structures of Sinapis alba myrosinase and a covalent glycosyl-enzyme intermediate provide insights into the substrate recognition and active-site machinery of an S-glycosidase. Structure 5, 663–675 25 Sidhu, G., Withers, S. G., Nguyen, N. T., McIntosh, L. P., Ziser, L. and Brayer, G. D. (1999) Sugar ring distortion in the glycosyl-enzyme intermediate of a family G/11 xylanase. Biochemistry 38, 5346–5354 26 Davies, G. J., Dauter, M., Brzozowski, A. M., Bjornvad, M. E., Andersen, K. V. and Schulein, M. (1998) Structure of the Bacillus agaradherans family 5 endoglucanase at 1. 6 Å and its cellobiose complex at 2. 0 Å resolution. Biochemistry 37, 1926–1932 27 Barends, T. R., Bultema, J. B., Kaper, T., van der Maarel, M. J., Dijkhuizen, L. and Dijkstra, B. W. (2007) Three-way stabilization of the covalent intermediate in amylomaltase, an α-amylase-like transglycosylase. J. Biol. Chem. 282, 17242–17249