From Structure to Function: Insights into the Catalytic Substrate Specificity and Thermostability Displayed by Bacillus subtilis Mannanase BCman

doi:10.1016/j.jmb.2008.03.068

J. Mol. Biol. (2008) 379, 535–544

Available online at www.sciencedirect.com

From Structure to Function: Insights into the Catalytic Substrate Specificity and Thermostability Displayed by Bacillus subtilis Mannanase BCman Xiao-Xue Yan, Xiao-Min An, Lu-Lu Gui and Dong-Cai Liang⁎ National Laboratory of Biomacromolecules, Institute of Biophysics, Chinese Academy of Sciences, 15 Datun Road, Chaoyang District, Beijing 100101, P. R. China Received 23 December 2007; received in revised form 16 March 2008; accepted 25 March 2008 Available online 7 April 2008

BCman, a β-mannanase from the plant root beneficial bacterium Bacillus subtilis Z-2, has a potential to be used in the production of mannooligosaccharide, which shows defense induction activity on both melon and tobacco, and plays an important role in the biological control of plant disease. Here we report the biochemical properties and crystal structure of BCman-GH26 enzyme. Kinetic analysis reveals that BCman is an endo-βmannanase, specific for mannan, and has no activity on mannooligosaccharides. The catalytic acid/base Glu167 and nucleophile Glu266 are positioned on the β4 and β7 strands, respectively. The 1.45-Å crystal structure reveals that BCman is a typical (β/α)8 folding type. One large difference from the saddle-shaped active center of other endo-β-mannanases is the presence of a shallow-dish-shaped active center and substrate-binding site that are both unique to BCman. These differences are mainly due to important changes in the length and position of loop 1 (Phe37-Met47), loop 2 (Ser103-Ala134), loop3 (Phe162-Asn185), loop 4 (Tyr215-Ile236), loop 5 (Pro269-Tyr278), and loop 6 (Trp298-Gly309), all of which surround the active site. Data from isothermal titration calorimetry and crystallography indicated only two substrate-binding subsites (+ 1 and − 1) within the active site of BCman. These two sites are involved in the enzyme's mannan degradation activity and in restricting the binding capacity for mannooligosaccharides. Binding and catalysis of BCman to mannan is mediated mainly by a surface containing a strip of solvent-exposed aromatic rings of Trp302, Trp298, Trp172, and Trp72. Additionally, BCman contains a disulfide bond (Cys66fCys86) and a special His1-His23-Glu336 metalbinding site. This secondary structure is a key factor in the enzyme's stability. © 2008 Elsevier Ltd. All rights reserved.

Edited by R. Huber

Keywords: endo-β-mannanase; substrate specificity; thermostability; shallowdish-shaped active center; X-ray crystallography

Introduction Plant cell walls, which are composed of cellulose and hemicellulose, are the richest renewable energy substances on earth.1 Mannan, glucomannan, galac*Corresponding author. E-mail address: [email protected]. Abbreviations used: M1, mannose; M2, mannobiose; M3, mannotriose; M4, mannotetraose; M5, mannopentaose; M6, mannohexaose; ITC, isothermal titration calorimetry; EDTA, ethylenediaminetetraacetic acid.

tomannan, and galactoglucomannan are the major polysaccharides that constitute hemicellulose. They can be cooperatively catalyzed and hydrolyzed into soluble mannooligosaccharides by hemicellulases such as endo-mannanase and exo-mannose glucosidase. Effective exploitation of oligosaccharides of 2– 10 monosaccharide molecules linked through β-1,4 glycosidic bonds can provide a sustainable source of cheap raw materials for agriculture, livestock, and various industries.2–5 β-Mannanase is an endoenzyme that can hydrolyze both mannooligosaccharides and polysaccharide mannan linked by a β1,4 glycosidic bond.6 Mannooligosaccharides can be produced by β-mannanase hydrolysis of β-mannan-

0022-2836/$ - see front matter © 2008 Elsevier Ltd. All rights reserved.

536

Crystal Structure and Enzyme Kinetics of GH26 Enzyme BCman

Table 1. Enzyme kinetic parameters of wild-type and mutant BCman Substrate

kcat (min− 1)

Km

kcat/Km

Locust bean gum Konjac powder Ivory nut mannan Mannobiose Mannotriose Mannotetraose Mannopentaose Mannohexaose Locust bean gum Locust bean gum Locust bean gum

3672 4507 668 Noa Noa Noa Noa — 118.5 Noa 1589

10.2 mg ml− 1 14.2 mg ml− 1 15.9 mg ml− 1 Noa Noa Noa Noa — 35.7 mg ml− 1 Noa 12.5 mg ml− 1

3.6 × 102 mg ml− 1 min− 1 3.2 × 102 mg ml− 1 min− 1 4.2 × 10 mg ml− 1 min− 1 Noa Noa Noa Noa 4.0 × 10− 1 mM− 1 min− 1 3.3 mg ml− 1 min− 1 Noa 1.3 × 102 mg ml− 1 min− 1

BCman Wild type Wild type Wild type Wild type Wild type Wild type Wild type Wild type E167A E266A H1A-23A

—, not calculated. a Catalytic activity not assayed. E167A and E266A are single-site mutants, while HA1-H23A is the twin-site mutant.

type plant gums such as locust bean gum, guar gum, and konjak powder. This process can accelerate the growth of human enteric beneficial bacterium and excite the plant's defense system.7 According to primary sequence homology, βmannanase is a member of the glycoside hydrolase subfamily GH5 and GH26.8 The three-dimensional structure has been reported for two β-mannanases of the GH26 subfamily, which show primary sequence homology greater than 40%. The first structure published was PCMan26 A (1J9Y) from Cellvibrio japonicus,9 for which a series of conservative residues was identified. However, since there was no ligand binding, the active site was not identified. Subsequently, the structures [Protein Data Bank (PDB) code 1ODZ] of PCMan26 and the mannobiose (M2) complex revealed that there are four substrate-binding sites within the active center of β-mannanases, which are responsible for the effective catalysis of oligosaccharides and polysaccharides.10 In addition, the crystal structure (PDB code 2BVT) of CtLic26A-mannotriose (M3) complex from Cellulomonas fimi was investigated.11 The results indicated that CtLic26A contains five substrate-binding sites within its active center for effective catalysis of mannooligosaccharides and polysaccharides, but that it cannot catalyze M2. Research on the subfamilies of GH-A glycoside hydrolases has shown that changes in the active-site residues and/or in the loops around the active site will cause changes in the substrate specificity, the catalytic activity, and even the mechanism of catalysis.12–15

BCman is an endo-β-1,4 mannanase derived from Bacillus, having primary sequence identity of 27% and 24% compared to PCMan26A and CtLic26A, respectively.9,11 The enzyme can hydrolyze konjak powder and locust bean gum to produce mannooligosaccharide, but not a monosaccharide.7 Kinetic analysis (Table 1) demonstrates that BCman cannot degrade short mannooligosaccharides. Results of isothermal titration calorimetry (ITC) (Table 2) and crystal structural analysis reveal that the substratebinding subsites of BCman are completely different from that of PCMan26A and CtLic26A. Additionally, BCman has stronger resistance to heat and proteinases than the reported β-mannanases.9,11,12 The stability of mannanase is one reason for its popularity in paper manufacturing, food industries, and livestock feeds. Biochemical analysis confirms that the disulfide bond of mannanase is a key factor in its thermal stability and resistance to proteolytic attack.16 However, the stability of mannanase has still not been explained in terms of its threedimensional structure. Here we report the enzymatic characterization and three-dimensional structure of BCman, with information on its catalytic substrate specificity and thermostability.

Results and Discussion Biochemical properties of BCman Analysis of the substrate specificities of BCman showed that the enzyme displays no activity against

Table 2. Affinity of BCman for mannooligosaccharides as determined by ITC Protein BCman BCman BCman BCman BCman BCman

Ligand a

Mannose Mannobiosea Mannotriosea Mannotetraosea Mannopenosea Mannohenose

Ka (M− 1)

ΔG (kcal mol− 1)

ΔH (kcal mol− 1)

TΔS (kcal mol− 1)

n

0.23 ± 0.1 1.50 ± 0.1 2.11 ± 0.03 4.30 ± 0.1 1.56 ± 0.1 117 ± 13.8

— — — — — − 2.7 ± 0.2

— — — — — − 36.4 ± 3.7

— — — — — −33.7 ± 0.1

1.0 ± 0.0 1.0 ± 0.0 1.0 ± 0.0 1.07 ± 0.0 1.0 ± 0.0 1.0 ± 0.0

—, not calculated. a An estimate of the affinity is given, because the affinity is below the range for accurate determination.

Crystal Structure and Enzyme Kinetics of GH26 Enzyme BCman

xylose, carboxymethyl cellulose, arabinose, or βglucans, trace activity against ivory nut mannan, and significant activity against galactomannan and glucomannan. BCman also displays activity against dyed polysaccharides, which is consistent with the designation of BCman as a β-mannanase displaying endo-activity. The kinetic parameters of BCman (Table 1) show that the enzyme is approximately eightfold more active against konjak powder and locust bean gum compared with ivory nut mannan. This is in contrast to the results reported by Jerome et al.11 and Stålbrand.17 The glycoside branched chain of mannan possibly plays a role in the process by which BCman degrades polysaccharide. Analysis of the reaction products released by BCman after hydrolysis of ivory nut mannan showed M2 and mannotetraose (M4) in a ratio of 2:1 (Fig. 1b). This is in contrast to the report by Jérôme et al.,11 in which mannose (M1) and M2 were found to be the end products of β-mannanase degradation of ivory nut mannan. More important, our results (Table 1) indicate that BCman cannot catalyze the short mannooligosaccharides, including M2, M3, M4 (Fig. 1a), and mannopentaose (M5), and that it has very low catalytic efficiency for mannohexaose (M6), 1/50,000 that of CtLic26A catalysis of M6 (at 112 s− 1 mM− 1). According to these data, we think that BCman is notably different from other members of GH26 in terms of substrate specificity and catalytic activity and it may also have a unique active center. Analysis of the biophysical properties of BCman showed that the enzyme has a pH optimum of 4.5, and the enzyme activity is stable between pH 2.5 and 8.5. Enzyme activity is greatest at 60 °C, with

537

activity remaining above 80% between 50 and 80 °C. After incubation for 2 h with pancreatic proteinases, BCman retained more than 90% of its catalytic activity, demonstrating a high degree of resistance to proteolytic inactivation. These results show that BCman has higher thermal stability than most reported β-mannanases.9,11,12 The use of ITC to estimate the number of subsites in BCman ITC was used to quantify the affinity of BCman for mannooligosaccharides and determine the thermodynamics of ligand binding. The Ka values (Table 2) of BCman show that the enzyme has at least a 100-fold greater affinity for M6 than for M5. The affinity of BCman for M5 and all other oligomannose ligands with lower degrees of polymerization was in fact lower than our limits of detection. The data suggest that BCman interacts increasingly with mannan having a degree of polymerization of 6 or greater. We also found that BCman exhibits similar binding constants for M2, M3, M4, and M5, suggesting that BCman has a ligand-binding region that can accommodate two M1 moieties. These data, combined with information on the crystal structure (see below), suggest that there are two subsites (+1 and − 1) within the active site of BCman for effective hydrolysis of mannan. The capacity of the purified BCman proteins (wild-type and two single-site mutants) to bind insoluble mannan polysaccharides from locust bean gum was studied by incubating the proteins with the polysaccharides and determining the protein concentration in the filtrate (unbound protein). The

Fig. 1. HPLC analysis of the end products of BCman catalysis of M4 and ivory nut mannan. After treatment of M4 (a) and ivory nut mannan (b) with an appropriate amount of BCman for a given time, the enzymolysis products were assayed (see Materials and Methods). The arrows indicate the retention times of standard samples: (1) M1, (2) M2, (3) M3, and (4) M4. The retention time of M4 is 10.2 min.

538

Crystal Structure and Enzyme Kinetics of GH26 Enzyme BCman

results show that the relative binding of native BCman, E266A, and E167A to locust bean gum are 93%, 89%, and 91%, respectively. The finding of similar affinities of both wild-type BCman and the mutants indicates that catalytic residue mutants do not affect the binding of polysaccharides. Studies of the enzyme kinetics (Table 1) combined with information on the active-site structure (see below) also lends support to the assumption that there are two subsites within the BCman active site, and that − 1 is glycone and + 1 is aglycone. We hypothesize there is another binding factor, in addition to the + 1 and − 1 subsites, contributing to the BCman binding specificity to polysaccharides too. Three-dimensional structure of BCman The structure of BCman was solved using mercury single-wavelength anomalous dispersion data to 2.0 Å combined with native data to 1.45 Å (Table 3). The P1 crystal form contains two monomers (A and B), and the structural model contains 672 residues. Both molecules A and B comprise 336 residues. The overlapping r.m.s.d. for the main-chain Cα between the two molecules is 0.2 Å. The contact surface between the two molecules within the asymmetric unit contains the loops at the N- and C-termini of molecule A, as well as the loops between β3 and α4 and between β4 and α5 of molecule B. Size-exclusion chromatography and dynamic light-scattering experi-

Table 3. Data collection, phasing and refinement statistics for BCman structures

Data collection statistics Wavelength (Å) Space group Unit cell parameters a, b, c (Å) α, β, γ (°) Resolution (Å) No. unique reflections Redundancy Completeness (%) Rmerge (%) I/σ(I) No. molecules per asymmetric unit Refinement statistics Non-hydrogen atoms Protein Water Metal ion 5 Glycerol molecules Rwork (%) Rfree (%) r.m.s.d. from ideal Bond length (Å) Bond angle (Å) B-factors (Å2) Protein Water

Native

Hg-BCman

1.5418 P1

1.5418 P1

a = 45.1, b = 58.8, c = 63.5 α = 83.0, β = 82.4, γ = 77.6 8–1.45 (1.48–1.45) 18,955 5.9 90.8 (80.1) 3.4 (18.2) 51.9 (11.3) 2

a = 45.7, b = 59.1, c = 63.8 α = 83.0, β = 82.2, γ = 77.6 20–1.95 (2.0–1.95) 10,050 4.6 95.1 (91.9) 5.9 (10.5) 43.9 (18.4) 2

6346 672 909 2 30 15.8 19.0 0.007 1.136 11.32 21.92

ments (not shown) indicate the protein is a monomer in solution. Its functional unit is also a monomer. The structure of BCman is a classical (β/α)8 “TIM” barrel folding type of the GH-A glycoside hydrolase family. Its interior contains a hydrophobic core formed by the folding of eight β-strands (β1–β8) that are parallel in opposite directions and eight α-helices (α1–α9) that are exposed to the solvent around the exterior. Comparison of primary sequence and three-dimensional structure shows that BCman belongs to the GH26 subfamily. The 1.45-Å crystal structure reveals that BCman is different from other members of the GH26 in terms of three-dimensional structure. Primarily, it has an α-helix (α9) that is formed by eight amino-acid residues between β3 and α4. This α-helix is unique to BCman and occurs in a position that contains the loop region in other members of the GH26 family. Secondly, loops within the BCman structure account for 53% of the total structure. Among these, the loops that connect β2 and α3, β3 and α4, β4 and α5, and β5 and β6 are all composed of 20–30 amino acid residues. This structural feature means that the catalytic domain of BCman has greater flexibility, which is likely to influence its function. Additionally, there is a stable disulfide bond between residues Cys66 and Cys86 within BCman, a feature not previously observed in the three-dimensional structure of members of the GH26 subfamily. Finally, the BCman molecule contains a metalbinding motif, His1-His23-Glu336. The metal ion within this motif tightly binds to both His1 at the Nterminus and Glu336 at the C-terminus of the BCman molecule, effectively stabilizing the flexible loops at both ends of the molecule. We believe that the disulfide bond and metal-binding motif not only play an important role in the effective enzyme function, but also increase rigidity within the molecule. The catalytic center of BCman A DALI search on the protein shows that BCman is most similar to the Pseudomonas cellulosede βmannanase PCMan26A (1J9Y), C. fimi β-mannanase CtLic26A (2BVY), and Clostridium thermocellum laminarinase CfMan26A (2BVD) with r.m.s.d. values of 2.3, 2.5, and 2.3 Å, respectively, for the main chain Cα. Three-dimensional structure of BCman shows that it has a unique, shallow-dish-shaped active center that is different from that of other members of GH26. The active site of endo-β-mannanase is best described as a channel (Fig. 2a). The open ends of the active site allows the enzyme to “ride” on the chain of hemicellulose molecules, then to cleave polysaccharide chains internally and generate carbohydrate products with various degrees of polymerization. Obviously, the shallow-dish-shaped active center of BCman is unable to “ride” on the substrate sugar chain and perform random cleavage (Fig. 2b). According to the three-dimensional map (Fig. 3a), in the active region of BCman there are six key loops involved in binding in the unique shallow-dish-

Crystal Structure and Enzyme Kinetics of GH26 Enzyme BCman

539

Fig. 2. Comparison of active regions of BCman and members of the GH26 subfamily. (a) Model of the solventaccessible surface of the homologous structure PCMan26A, which shows the saddle-shaped, open active center of the GH26 subfamily. M2 sugar molecules are displayed in green. (b) Model of the solvent-accessible surface of BCman. The colored circle of residues indicates the shallow-dish-shaped active center of BCman. The catalytic groups Glu167 and Glu266 are in red. Loop 1 (green), in the form of a double spatial arrangement, obliquely blocks the end of the active site. Blue loops 2–6 and loop 1 are gathered round the active site to provide a hydrophobic platform for the binding of mannan.

shaped active site: loop 1 (Phe37-Met47), loop 2 (Ser103-Ala122), loop 3 (Phe162-Asn185), loop 4 (Tyr215-Ile236), loop 5 (Pro269-Tyr278), and loop 6 (Trp298-Gly309). The most significant differences between the BCman and other endo-β-mannanases are loop 1 connecting β2 and α2 and loop 3. In the homologous structure of other members of GH26, loop 1 deviates from the active site and forms the walls of the saddle-shaped crack with loops 5 and 6. The open saddle-shaped fracture provides space for the penetration of sugar substrates. In BCman, loop 1, turning toward the wall of the potential saddleshaped crack, inserts and blocks the active site in the form of a double spatial arrangement. Tyr40 on loop 1 and His109 on loop 2 use hydrogen bonding to stably block the rear end of the saddle-shaped pocket and thus hinder the entry of sugar molecules. There is no change in length of loops 3 and 4, lying at the front of the BCman active site, compared to the corresponding loops in the homologous structure. However, their positions deviate towards the active center of the enzyme and thus block the front of the saddle-shaped passageway. In addition, loops 2, 5, and 6 in BCman comprise 20, 10, and 12 amino acids, respectively, and thus are all obviously shorter in comparison with loop 2 (Ser141-Tyr177), loop 5 (Ile323-Gln336), and loop 6 (Val360-Arg384) in the homologous structure, which form the opposite walls of the passageway. These six loops within BCman create a circular dish edge around the active site of the enzyme, forming the shallow-dish-shaped active center specific to BCman (Fig. 2b). Comparison of the active site of BCman with the PCMan26A M2 complex (PDB code 1ODZ) reveals significant similarity between the + 1 and − 1 subsites

of these two enzymes (Fig. 4a). However, the − 2 site in PCMan26A is occupied by Trp163 with a negative charge, while in BCman it is occupied by Lys117 with a positive charge. The hydrophobic stacking offered by the aromatic side chain of Trp163 within the PCMan26A molecule is significant for determination of the binding direction and acceptance of substrate molecules. Thus, this residue is key in ensuring that the − 2 site can effectively bind to the M2 sugar. In BCman, the distance between the − 2 site and the side-chain amino group of Lys117 at the − 2 site is 5.3 Å, which is twofold greater than the distance between Trp162 and the − 2 site (2.82 Å); therefore, Lys117 of BCman has no affinity interaction with the M2. In addition, Tyr40 within BCman replaces Asp111 in PCMan26A. Steric hindrance occurs between the aromatic hydroxyl oxygen of Tyr40 and the M2 pyranose ring C6, preventing binding between the substrate M2 and BCman. Comparison of the active site of BCman with CtLic26A-M3 is shown in Fig. 4b. The active residues of BCman and CtLic26A are highly conserved at the + 1 and − 1 subsite. However, at the − 2, − 3, and − 4 subsites of BCman, there is no interaction between the amino acid residues and substrate M3. This significant difference is mainly due to the loop connecting β7 and α7. In CtLic26A, the flexible loop formed by Trp322-Phe331 is parallel to the substrate binding direction. Ala323 and Phe325, located on loop 6, are very important for binding between the − 2 and − 3 sites and the long-chain oligosaccharides M4 and M5. In contrast, the loop formed by Trp298-Ser303 in BCman is much shorter than the corresponding loop in CtLic26A, and thus cannot provide amino acid residues for interaction

540

Crystal Structure and Enzyme Kinetics of GH26 Enzyme BCman

Fig. 3. Shallow-dish-shaped active center of BCman. (a) The divergent wall-eyed stereo of the differences in the six loops of the active centers of BCman and another GH26 member. Overlapping map of BCman and the homologous structure PCMan26A (1J9Y). The direction is the same as that in Fig. 2b. BCman loops are in red, while those of PCMan26A are blue. (b) The stereo view of the structural elements involved in the BCman active center. Residues are shown in ball-and-stick model. The conserved catalytic residues (Glu167, Glu266) and other hydrogen-bonding polar residues that are around the exposed tryptophan residues (including Trp72, Trp172, Trp298, and Trp302) together form the active center of BCman. The catalytic residues are shown in blue, the hydrophilic and charged residues in brown, and the exposed tryptophan residues in green.

with the substrate. In addition, the aromatic ring of Tyr40 in BCman is only 1.53 Å away from the M3 pyranose ring C6 located at the − 2 site, and steric hindrance prevents M3 from binding at site − 2. Therefore, we have demonstrated that there are two subsites (+1 and − 1) within the active site of BCman for effective hydrolysis of mannan. This conclusion is strongly supported by the results of ITC.

Identification of the catalytic residues and binding residues of polysaccharide The solvent-accessible surface of BCman contains a unique shallow-dish-shaped active center (Fig. 2b). Similar to other members of the GH26, the active site of BCman contains the same conserved residue motif, His166-Glu167-Glu266 (Fig. 3b). To determine the influence of residues Glu167 and Glu266 on

Crystal Structure and Enzyme Kinetics of GH26 Enzyme BCman

541

Fig. 4. The overlap of the subsites in the BCman and other GH26 enzyme's active center. (a) Diagram of overlapping active centers of BCman and PCman26A-M2 complex structure (1ODZ). BCman is indicated in blue, PCMan26A in purple, and the M2 at the − 1 and −2 subsites is red. (b) Diagram of overlapping active centers of BCman and the Ctlic26AM3 complex structure (2BVT). BCman is indicated in blue, Ctlic26A in purple, and the M3 at the − 2, −3, − 4 subsites is red. BCman residues discussed in the text are labeled with the equivalent residue of PCman26A or Ctlic26A in parentheses.

BCman catalysis of sugar substrates, we assayed the kinetics parameters for single-site mutants of both E167A and E266A (Table 1). Mutant E266A showed no catalytic activity for locust bean gum, confirming that Glu266 is the nucleophile. E167A has a catalytic activity of 3.3 mg ml−1 min− 1, which is 100-fold lower than that of the wild-type enzyme (3.6 × 102 mg ml− 1 min− 1), and is the hydrolysis acid/base. Kinetic analysis of polysaccharide hydrolysis showed that BCman hydrolyzed locust bean gum approximately eightfold more efficiently than ivory nut mannan (Table 1). These data indicate that the branched chain of the mannan possibly influences the enzyme degradation of polysaccharides. It has been suggested that the binding of xylan-binding domain 1 to xylan is mediated primarily by a surface containing a strip of solvent-exposed aromatic rings, which are spaced such that they are able to form multiple stacking interactions with a cellulose chain, plus several hydrogen-bonding polar residues.18 We analyzed the amino acid residues around the BCman active site according to the expanding width of galactomannan (d ≈ 10 Å). Our results indicate that all of the residues, which can create potential hydrogen bonds with the substrate, are in loops around the active center, including Trp302, Glu301, Asp300, Asn299, Trp298, Gln270, Tyr242, Asp218, Phe173, Trp172, Phe116, Trp72, Arg70, Asp43, and Phe37 (Fig. 3b). BCman contains four solvent-exposed tryptophan rings, Trp302, Trp298, Trp172, and Trp72. Trp172 and Trp298 are conserved throughout family GH26 (Fig. 4a), and Trp172 has been demonstrated to be involved in binding mannooligosaccharides rather than polysaccharides.9 To investigate the proposed role of

the four tryptophans (Trp302, Trp298, Trp172, and Trp72), we assayed the affinity for polysaccharides of three mutant BCman enzymes: W172A, W298A, and BCm4 (a four-site mutant in which the four tryptophans were replaced by alanine). Results showed that the affinities for locust bean gum are much lower in the mutants (36%, 28%, and 0%, respectively, compared to wild type). Our research suggests that both Trp172 at +1 and Trp298 at −1 are very important for binding and catalysis. In addition, the binding of BCman to mannan is mediated mainly by a surface containing a strip of solvent-exposed aromatic rings of Trp302, Trp298, Trp172, and Trp72. In addition, surrounding the four aromatic rings in the binding site are a number of hydrophilic and charged residues, namely, Glu301, Asp300, Asn299, Gln270, Tyr242, Asp218, Arg70, and Asp43. All are oriented such that their side-chain groups would be able to form hydrogen bonds to a polysaccharide, thus playing a definite role in the binding between BCman and mannan and influencing the catalytic function. In this study, we have determined the likely binding characteristics and catalytic processes of BCman. In the first phase of the catalytic reaction, the solventexposed tryptophan coplanar side chains stack against the sugar rings of the polysaccharide, and the aforementioned hydrogen-bonding polar residues stabilize the binding through hydrogen bonds to a polysaccharide stacking over the exposed tryptophans. In the second stage of the reaction, at the −1 subsite, the glycosidic bond is cleaved through a combination of nucleophilic attack of the anomeric carbon by the nucleophile Glu266 and protonation of the glycosidic oxygen by the catalytic acid/base

542

Crystal Structure and Enzyme Kinetics of GH26 Enzyme BCman

Glu167, generating a glycosyl–enzyme intermediate. In the last stage of the reaction, the enzyme–substrate is hydrolyzed by water attacking the anomeric carbon of the glycosyl group, and polysaccharides are degraded to mannooligosaccharides. The structural basis of thermal stability One of the prominent features of BCman as a major hemicellulase is that it has higher thermal stability than most reported β-mannanases.9,11,12 We carried out mercaptoethanol reduction of disulfide bond, after which the catalytic activity of the enzyme was 34 U/mL, which is slightly lower than that of the native protein. The thermal stability of the enzyme decreased markedly with increasing temperature (data not shown) and lost activity at 80 °C. These results demonstrate the importance of the disulfide bond on the thermal stability of BCman. The three-dimensional structure of BCman showed that molecules A and B within the asymmetric unit both bind to a metal ion. As shown in Fig. 5 (molecule A), the metal atom is coordinated by α-NH2 at the Nterminus, main-chain carbonyl O, and the side-chain N of His1, Glu336 (OE1), Glu336 (OE2), and His23 (NE2). Molecule B is identical to A. Inductively coupled plasma atomic emission spectrometry analysis showed that the metal ion was Zn2+ (and part of Ni2+). We believe that the metal-binding site links the flexible loops at the N-terminus and C-terminus together, thus reducing the freedom of both ends of the peptide chain and increasing the rigidity of the protein. Such a metal ion binding site has not been reported for the crystal structure of GH-A glycoside hydrolases. In addition, enzyme activity assays showed that Zn2+ (1 mM) and Ni2+ (1 mM) did not affect activity after dialysis against Tris–HCl (pH 8.0),

0.05 mM ethylenediaminetetraacetic acid (EDTA). Dialysis against six changes Tris–HCl (pH 8.0) was done to remove EDTA completely. To further discuss the function of the metal-binding sites in BCman, residues His1 and His23 were both mutated to alanine residues. Results for the enzyme activities of wild-type and mutant H1A-23A are shown in Table 1. The catalytic activity of mutant H1A-23A for locust bean gum was 1.3 × 102 mg ml− 1 min− 1, which is slightly lower than that of the wildtype protein (3.6 × 102 mg ml− 1 min− 1). The thermal stability of the mutant was significantly decreased, with the half-life decreasing from 1 h to 20 min at 80 °C. These observations support our suggestion that the binding motif of the six-coordinate Zn with the His-Glu-His within the BCman molecule contributes to its high thermal stability.

Materials and Methods Cloning and expression Using primers detailed under the Supplementary Data, the region of the MAN gene (GenBank AY827489) encoding residues 27–367 of the mature protein, defined hereinafter as BCman, was amplified from Bacillus subtilis Z-2 genomic DNA by PCR. The PCR product was ligated to the NcoI and BamHI sites of vector pET22b (Novagen) with a pelB signal peptide to generate P01. BCman encoded by P01 contains a C-terminal His6 tag. Escherichia coli BL21(DE3) harboring P01 was cultured in LB containing 100 μg/mL ampicillin at 37 °C to mid-exponential phase (A550 = 0.6), at which point isopropyl-β-D-thiogalactopyranoside was added to a final concentration of 0.5 mM and the cultures were incubated for a further 4 h. Following this incubation, the cells were harvested by centrifugation and were resuspended in the sucrose osmosis as described.19 A centrifugation step was performed to remove debris and unbroken cells. The supernatant was then equilibrated in binding buffer (50 mM Tris–HCl, 500 mM NaCl, 20 mM imidazole, pH 8.0) and was loaded on a HisBind column (Amersham Biosciences). Elution was performed using a stepwise increase of imidazole (20–300 mM). Further purification was carried out by size-exclusion chromatography on a HiLoad 26/60 Superdex 200 column (Amersham Biosciences) eluted with buffer (50 mM Tris–HCl, pH 8.0, 500 mM NaCl, 1 mM DTT). BCman protein was concentrated to 20 mg/ml using a Millipore 10-kDa cutoff spin column, and the buffer was exchanged by concentrating and diluting the sample three times with double-distilled water. The purified protein was adjusted to a final concentration of 20 mg/mL by a BCA protein analysis kit according to the manufacturer's instructions. Construction of site-specific variants

Fig. 5. Unique metal-binding motif of the octahedron with six-coordination. His1 lies in the first position in the loop at the N-terminus, Glu336 is in the last position in the loop at the C-terminus, and His 23 is in the helix α1. The Zn2+ shown in brown creates a stable binding mode involving an octahedron and six-coordination with α-NH2 at the N-terminus, main-chain carbonyl O and the sidechain NE1 of His1, the side-chain NE2 of His23, Glu336 (OE1), and Glu336 (OE2).

Site-directed mutagenesis employing a QuikChange kit (Stratagene), primers listed in the Supplementary Data, and P01 as the template DNA was used to construct mutants of the man gene, encoding active-site variants of BCman. All mutants were sequenced to ensure that only the designed mutation had been introduced into the man gene. For this study, we constructed two single-site mutants, E167A, E266A, W172A, and W298A; a two-site mutant, H1A-23A; and a four-site mutant BCm4.

Crystal Structure and Enzyme Kinetics of GH26 Enzyme BCman Enzyme assays Sugar substrates M2, M3, M4, M5, and M6 were purchased from Megazyme. M1, locust bean gum, and ivory nut mannan were obtained from Sigma, and konjak powder from the Chinese Academy of Agricultural Sciences. The activity of BCman against various polysaccharides was determined as described previously13 by detecting the release of reducing sugars using the Somogyi–Nelson reagent.20 Standard assays for BCman were carried out at 37 °C in 50 mM sodium phosphate, 10 mM citric acid buffer, pH 7.0, containing 1 mg/ml bovine serum albumin. Kinetic analysis of BCman against various substrates was done as described by Hogg et al.9 Hydrolysis of ivory nut mannan was done by incubating 1 mg/mL ivory nut mannan with 4 μM BCman for 4 h. For kinetic analysis, at least six substrate concentrations that straddled the Km were employed, and non-linear regression was used to estimate Km and Kcat. High-performance anion-exchange chromatography with pulsed amperometric detection was used to assay the end products of the enzyme catalysis. The pH optimum of BCman was determined as described by Louise et al.12 Temperature stability was determined as described by Andrews et al.21 The affinity of BCman for insoluble polysaccharides was measured as follows: purified enzyme (0.5 g/L) was mixed with an equal volume of 5% (w/v) polysaccharide in 50 mM sodium phosphate buffer (pH 7.0) for 15 min at 0 °C. The affinity of the mannanase for insoluble polysaccharides was measured as described previously.22 Unbound material was quantified by measuring protein concentration in the filtrate after the insoluble polysaccharides and mannan had been recovered by filtration. Isothermal titration calorimetry ITC measurements were made as described by Charnock et al.23 Briefly, the measurements were made at 25 °C following standard procedures using a Microcal Omega titration calorimeter. Proteins were dialyzed extensively against 50 mM sodium phosphate buffer, pH 7.0, and the dialysis buffer was used for heats of dilution controls. During a titration experiment, the protein sample (100 μM), stirred at 400 rpm in a 1.4-mL reaction cell, was injected with 26 successive 10-μL aliquots of a 5–10 mM solution of ligand at appropriate intervals. Control experiments were performed under identical conditions by injection of ligand into buffer alone (to correct for heats of dilution of the ligand) and injection of buffer into the protein solution (to correct for heats of dilution of the protein). Integrating heat effects, after correction for heats of dilution, were analyzed by non-linear regression using a simple single-site binding model using the standard Microcal ORIGIN 7.0 software package. For each thermal titration curve, this formula yields estimates of the apparent number of binding sites on the protein, the Ka, and the enthalpy of binding (ΔH°). Other thermodynamic properties were calculated using the standard thermodynamic equation: RT ln Kd = ΔG° = ΔH° − TΔS°. Crystallization and data collection The crystallization experiments were carried out at 290 K according to the hanging-drop vapor-diffusion method. The initial crystallization conditions were 20% polyethylene glycol (PEG) 4000, 100 mM sodium acetate (pH 5.6). Crystal optimization identified the best crystal growth conditions as

543

7.5% PEG 4000, 80 mM sodium chloride, and 100 mM sodium citrate (pH 5.12). Data were collected using a Rigaku Research image plate at a wavelength of 1.5418 Å and a temperature of −180 °C. DENZO and SCALEPACK were used to carry out indexing and integration.24 The protein crystal exhibits a P1 space group, and each asymmetric unit comprises two molecules. Statistical results for the data quality are shown in Table 3. Structure solution and refinement Phase angle analysis of the BCman crystal structure was carried out using the SIRAS method with HgCl2 as the crystal-soaked heavy-atom reagent. The soaking conditions were as follows: 22% PEG 4000, 100 mM sodium chloride, 100 mM sodium acetate (pH 5.12), 0.01 mM HgCl2, and a soaking time of 68 h. Data resolution for the Hg atom derivative was 2.0 Å. Using the SHELXD program to identify the heavy-atom site, we found a total of six mercury atoms in an asymmetric unit. MLPHARE was employed to refine the heavy-atom sites and calculate the experimental phase angle. The overall quality factor of the phase angle finally obtained is 0.31. By adopting the solvent-smoothing steps used in the DM program to improve the phase angle, the overall quality factor increases to 0.68. In inputting the BCman amino acid sequence, Arp/ wARP25 and OASIS200626 were combined for automatic construction of the model, while OASIS2006 was used to further optimize the phase site. Within the resolution range 8–1.45 Å, a combination of REFMAC27 and COOT28 was adopted for careful revision of the initial model obtained by automatic modeling. Detailed crystallography and statistical parameters of the final model are shown in Table 3. PROCHECK analysis of the final BCman model shows that 90.5% of the residues fall within the best allowed regions of the Ramachandran map, while no residue is located in non-allowed regions. The structure is well ordered: the overall Wilson plot B-factor is 11 Å2 and the main chain of all residues (1–336) fits well with the electron density map. The His-tag at the C-terminus of both molecules A and B (the P1 crystal form contains two monomers) is not observed in the density map, and therefore is not included in the model. The figures were created using PyMol†. Protein Data Bank accession numbers The coordinates and structure factors for BCman have been deposited in the RCSB PDB (accession code 2QHA).

Acknowledgements This study was supported by grants (No. 2002CB713801 and No. 2006CB806501) from the National Program on Key Basic Research Project (973 Program) and the National Protein Project (No. 2006CB090102 and No. 2007CB914302). Additional support was provided by the National Natural Science Foundation of China (No. 30600102). We are grateful to Professor Li-qun Zhang of the Department of Plant † http://www.pymol.org

544

Crystal Structure and Enzyme Kinetics of GH26 Enzyme BCman

Pathology (CAU, China) for providing the B. subtilis Z2 strain, Zhao-feng Luo at the Center of Biotechnology (USTC, China) for excellent technical support in ITC, and Yi Han at the Institute of Biophysics (CAS, China) for his assistance in data collection.

14.

Supplementary Data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/ j.jmb.2008.03.068

References 1. Suurnakki, A., Tenkanen, M., Buchert, J. & Viikari, L. (1997). Hemicellulases in the bleaching of chemical pulps. Adv. Biochem. Eng./Biotechnol. 57, 261–287. 2. Sims, I. M., Craik, D. J. & Bacic, A. (1997). Structural characterisation of galactoglucomannan secreted by suspension-cultured cells of Nicotiana plumbaginifolia. Carbohydr. Res. 303(1), 79–92. 3. Waldroup, P. W. (2003). Int. J. Poult. Sci. 2(1), 28–31. 4. Heldt-Hansen, H. P., Kofod, L. V., Budolfsen, G., Nielsen, P., Huttel, S. & Bladt, T. (1996). Pectins and Pectinases. In Progress in Biotechnology (Visser, J. & Voragen, A. G. J., eds), vol. 14, pp. 463–474, Elsevier, Amsterdam. 5. Oxenboll Sørensen, S., Pauly, M., Bush, M., Skjøt, M., McCann, M. C., Borkhardt, B. & Ulvskov, P. (2000). Pectin engineering: modification of potato pectin by in vivo expression of an endo-1, 4-beta-D-galactanase. Proc. Natl Acad. Sci. USA, 97, 7639–7644. 6. Schombug, D. & Salzmann, M. (1991). Enzyme Handbook (4), pp. 1–5, Springer-Verlag Heideberg, Berlin. 7. Zhang, Q.-X. (2006). Preliminary study of mannose oligosaccharides on plant resistance induction and cloning and expression of a β-mannanase gene from Bacillus subtilis Z-2. China doctor’s degree paper. China National Knowledge Infrastructure, CNKI: CDMD:1.2006.042383. 8. Henrissat, B. A. (1991). A classification of glycosyl hydrolases based on amino acid sequence similarities. J. Biochem. 280(2), 309–316. 9. Hogg, D., Woo, E. J., Bolam, D. N., McKie, V. A., Gilbert, H. J. & Pickersgill, R. W. (2001). Crystal structure of mannanase 26A from Pseudomonas cellulosa and analysis of residues involved in substrate binding. J. Biol. Chem. 276(33), 31186–31192. 10. Valérie, M.-A. D., David, L. Z., Garib, N. M., Harry, J. G., Lóránd, S., Dominik, S. et al. (2002). Substrate distortion by a β-mannanase: snapshots of the Michaelis and covalent-intermediate complexes suggest a B2,5 conformation for the transition state. Angew. Chem. Int. Ed. 41(15), 2824–2827. 11. Jérôme, L. N., Lars, A., Dominik, S., Henrik, S. & Leila, L. L. (2005). The structure and characterization of a modular endo- β-1,4-mannanase from Cellulomonas fimi. Biochemistry, 44, 12700–12708. 12. Louise, E. T., Victoria, A. M., Nicola, L. S., Claire, D., Gideon, J. D. & Harry, J. G. (2007). Mannose foraging by Bacteroides thetaiotaomicron: structure and specificity of the beta-mannosidase, BtMan2A. J. Biol. Chem. 282(15), 11291–11299. 13. Dias, F. M., Vincent, F., Pell, G., Prates, J. A., Centeno, M. S., Tailford, L. E. et al. (2004). Insights into the

15.

16. 17. 18.

19. 20. 21.

22.

23.

24.

25. 26.

27.

28.

molecular determinants of substrate specificity in glycoside hydrolase family 5 revealed by the crystal structure and kinetics of Cellvibrio mixtus mannosidase 5A. J. Biol. Chem. 279(24), 25517–25526. Edward, J. T., Arun, G., Catarina, I. P. D. G., José, A. M. P., Victoria, A. M., Natalie, F. et al. (2005). How family 26 glycoside hydrolases orchestrate catalysis on different polysaccharides: structure and activity of a Clostridium thermocellum lichenase, CtLic26A. J. Biol. Chem. 280, 32761–32767. Ana, L. C., Arun, G., José, A. M. P., David, N. B., Harry, J. G., Virgínia, M. R. P. et al. (2004). The family 11 carbohydrate-binding module of Clostridium thermocellum Lic26A-Cel5E accommodates β-1,4- and β-1,31,4-mixed linked glucans at a single binding site. J. Biol. Chem. 279, 34785–34793. Wu, J. & He, B. W. (2000). Purification and properties of β-D-mannanase from Nocardioform actinomycetes. Acta Microbiol. Sin. 16(2), 227–230. Stålbrand, H. (2003). Handbook of Food Enzymology, pp. 961–969, Marcel Dekker, Inc., New York. Peter, J. S., David, N. B., Alan, C., Antonio, C., Geoffrey, P. H., Harry, J. G. & Michael, P. W. (1999). A family IIb xylan-binding domain has a similar secondary structure to a homologous family IIa cellulose-binding domain but different ligand specificity. Structure, 7, 853–864. Ausubel, F. M., Brent, R. & Kingston, R. E. (1989). Short Protocols in Molecular Biology, 3rd edit., John Wiley and Sons, New York, NY. Somogyi, M. (1952). Notes on sugar determination. J. Biol. Chem. 195, 19–23. Andrews, S. R., Taylor, E. J., Pell, G., Vincent, F., Ducros, V. M., Davies, G. J. et al. (2004). The use of forced protein evolution to investigate and improve stability of family 10 xylanases. The production of Ca2+-independent stable xylanases. J. Biol. Chem, 279, 54369–54379. Ana, C. F., Carlos, M. G. A. F., Harry, J. G., Geoffrey, P. H., Tito, H. F. & Luis, M. A. F. (1999). Homologous xylanases from Clostridium thermocellum: evidence for bifunctional activity, synergism between xylanase catalytic modules and the presence of xylan-binding domains in enzyme complexes. Biochem. J. 342, 105–110. Charnock, S. J., Bolam, D. N., Turkenburg, J. P., Gilbert, H. J., Ferreira, L. M., Davies, G. J. & Fontes, C. M. (2000). The X6 “thermostabilizing” domains of xylanases are carbohydrate-binding modules: structure and biochemistry of the Clostridium thermocellum. Biochemistry, 39, 5013–5021. Otwinowski, Z. & Minor, W. (1997). Processing of X-ray diffraction data collected in oscillation mode. Methods in Enzymology, Macromolecular Crystallography, Part A, vol. 276, pp. 307–326, Academic Press, New York, NY. Perrakis, A., Morris, R. & Lamzin, V. S. (1999). Automated protein model building combined with iterative structure refinement. Nat. Struct. Biol. 6, 458–463. Yao, D. Q., Huang, S., Wang, J. W., Gu, Y. X., Zheng, C. D., Fan, H. F. et al. (2006). SAD phasing by OASIS2004: case studies of dual-space fragment extension. Acta Crystallogr., Sect. D: Biol. Crystallogr. 62, 883–890. Murshudov, G. N., Vagin, A. A. & Dodson, E. J. (1997). Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallogr., Sect. D: Biol. Crystallogr. 53, 240–255. Emsley, P. & Cowtan, K. (2004). Coot: model-building tools for molecular graphics. Acta Crystallogr., Sect. D: Biol. Crystallogr. 60, 2126–2132.