Chitin oligosaccharide binding to a family GH19 chitinase from the moss Bryum coronatum

Authors


T. Fukamizo, Department of Advanced Bioscience, Kinki University, 3327-204 Nakamachi, Nara 631-8505, Japan
Fax: +81 742 73 8976
Tel: +81 742 73 8237
E-mail: fukamizo@nara.kindfai.ac.jp
M. Sørlie, Department of Chemistry, Biotechnology and Food Science, Norwegian University of Life Sciences, PO Box 5003, N-1432 Ås, Norway
Fax: +47 64 96 59 01
Tel: +47 64 96 59 02
E-mail: morten.sorlie@umb.no

Abstract

Substrate binding of a family GH19 chitinase from a moss species, Bryum coronatum (BcChi-A, 22 kDa), which is smaller than the 26 kDa family GH19 barley chitinase due to the lack of several loop regions (‘loopless’), was investigated by oligosaccharide digestion, thermal unfolding experiments and isothermal titration calorimetry (ITC). Chitin oligosaccharides [β-1,4-linked oligosaccharides of N-acetylglucosamine with a polymerization degree of n, (GlcNAc)n, = 3–6] were hydrolyzed by BcChi-A at rates in the order (GlcNAc)6 > (GlcNAc)5 > (GlcNAc)4 >> (GlcNAc)3. From thermal unfolding experiments using the inactive BcChi-A mutant (BcChi-A-E61A), in which the catalytic residue Glu61 is mutated to Ala, we found that the transition temperature (Tm) was elevated upon addition of (GlcNAc)n (= 2–6) and that the elevation (ΔTm) was almost proportional to the degree of polymerization of (GlcNAc)n. ITC experiments provided the thermodynamic parameters for binding of (GlcNAc)n (= 3–6) to BcChi-A-E61A, and revealed that the binding was driven by favorable enthalpy changes with unfavorable entropy changes. The change in heat capacity (ΔCp°) for (GlcNAc)6 binding was found to be relatively small (−105 ± 8 cal·K−1·mol−1). The binding free energy changes for (GlcNAc)6, (GlcNAc)5, (GlcNAc)4 and (GlcNAc)3 were determined to be −8.5, −7.9, −6.6 and −5.0 kcal·mol−1, respectively. Taken together, the substrate binding cleft of BcChi-A consists of at least six subsites, in contrast to the four-subsites binding cleft of the ‘loopless’ family 19 chitinase from Streptomyces coelicolor.

Database

Chitinase, EC 3.2.1.14

Abbreviations
GlcNAc

N-acetylglucosamine

(GlcNAc)n

β-1,4-linked oligosaccharides of GlcNAc with a polymerization degree of n

ITC

isothermal titration calorimetry

LOGNAc

N-acetylglucosaminono-1,5-lactone (Z)-oxime

PUGNAc

N-acetylglucosaminono-1,5-lactone O-(phenylcarbamoyl)-(Z)-oxime

XOS

xylo-oligosaccharide

Introduction

Chitin, a β-1,4-linked polysaccharide of N-acetylglucosamine (GlcNAc), is hydrolyzed by chitinases (EC 3.2.1.14) that are widely distributed in living organisms and are responsible for self-defense, growth, morphogenesis, cuticle destabilization and stress tolerance [1–5]. A number of chitinase and chitinase-like genes have been isolated from living organisms, and the gene products have been characterized. Although chitinases from plant origins were classified into at least five classes (classes I–V) based on their deduced amino acid sequences [6,7], nowadays a simple classification according to the CAZy database is more widely used for all chitinolytic enzymes [8]. The chitinases are classified into families GH18 and GH19, both of which can be subdivided based on their domain arrangements and sequence deletions. The first three-dimensional structure of a plant chitinase was reported for 26 kDa family GH19 chitinase from barley seeds [9]. The barley enzyme is composed of two lobes, both of which are rich in α-helical structures, with the substrate binding cleft between the lobes. Several loops are found at both ends of the binding cleft (‘loopful’). The structure of the ‘loopful’ family GH19 chitinase from papaya in a complex with GlcNAc monomers has also been reported, and two GlcNAc molecules were found to bind separately to subsites −2 and +1, respectively, in the complex structure [10]. This structure has been used to build a plausible model of a complex with (GlcNAc)4. On the other hand, family 19 chitinases lacking the loops at the two ends of the binding cleft (‘loopless’ chitinase) have been isolated from two bacteria, Streptomyces griseus HUT6037 and Streptomyces coelicolor A3(2), and the evergreen conifer Norway spruce [11–13]. Based on the three-dimensional structures, it was suggested that the substrate binding clefts of the ‘loopless’ enzymes are shorter than those of ‘loopful’ enzymes [14,15]. The crystal structure of a ‘loopless’ family GH19 chitinase from S. coelicolor A3(2) and analysis of the products from chitooligosaccharide degradation by this enzyme revealed that the substrate binding cleft consists of four subsites [14]. However, no quantitative data for binding of (GlcNAc)n to the ‘loopless’ family GH19 chitinases have been reported to date. The subsite arrangement of the ‘loopless’ chitinase can be delineated not only from structural data but also from quantitative binding experiments.

Recently, Taira et al. [16] have isolated and characterized a family GH19 enzyme from a moss species, Bryum coronatum, designated as BcChi-A. They showed that the molecular weight of BcChi-A is relatively small compared with those of ‘loopful’ GH19 chitinases due to the deletion of several loops. Thus, BcChi-A is a ‘loopless’ family GH19 chitinase like that from S. coelicolor A3(2) rather than a ‘loopful’ GH19 enzyme such as the GH19 chitinase from barley seeds. Interestingly, when the shorter oligosaccharide substrate (GlcNAc)4 was used as the substrate, BcChi-A exhibited a much higher hydrolytic activity (1000-fold) than the ‘loopful’ GH19 chitinase from rye seeds [17]. In this study, to gain insight into the mechanism of (GlcNAc)n binding to the ‘loopless’ family GH19 enzymes, (GlcNAc)n binding to BcChi-A was examined by oligosaccharide digestion, thermal unfolding experiments and thermodynamic analysis using isothermal titration calorimetry (ITC).

Results

Chitin oligosaccharide hydrolysis

The time-courses of enzymatic hydrolysis of (GlcNAc)n (= 3–6) by BcChi-A are shown in Fig. 1. From (GlcNAc)6, BcChi-A produced (GlcNAc)3, (GlcNAc)2 and a lesser amount of (GlcNAc)4. The product (GlcNAc)4 appears to be hydrolyzed further to (GlcNAc)2 (Fig. 1A). The frequency of cleavage to (GlcNAc)2 + (GlcNAc)4 appears to be comparable with that of cleavage to (GlcNAc)3 + (GlcNAc)3. Equal amounts of (GlcNAc)2 + (GlcNAc)3 were produced from (GlcNAc)5 and only (GlcNAc)2 was from (GlcNAc)4 (Fig. 1B,C). (GlcNAc)3 was hydrolyzed to GlcNAc and (GlcNAc)2 (Fig. 1D). In the early stage of (GlcNAc)3 hydrolysis, the inverting BcChi-A enzyme exclusively produced the α-anomer of (GlcNAc)2 and β-anomer of GlcNAc, indicating that the reducing end residue of (GlcNAc)3 makes contact with amino acid residues forming subsite +1 and its β-form is selectively recognized by the subsite (Fig. 1E). BcChi-A hydrolyzed chitin oligosaccharides at rates in the order (GlcNAc)6 > (GlcNAc)5 > (GlcNAc)4 >> (GlcNAc)3, suggesting that the longer the chain length of the substrate, the higher the affinity of the oligosaccharides to the enzyme. No transglycosylation products were observed in any cases.

Figure 1.

 Experimental time-courses of (GlcNAc)n (= 3–6) digestion by BcChi-A. Enzyme and substrate concentrations were 0.04 μm and 4.67 mm, respectively, except for the reaction with (GlcNAc)3, in which the enzyme concentration was increased 10-fold. The enzyme reaction was conducted in 50 mm sodium acetate buffer pH 5.0 at 40 °C. Symbols: open circle, (GlcNAc)1; square, (GlcNAc)2; triangle, (GlcNAc)3; diamond, (GlcNAc)4; cross, (GlcNAc)5; closed circle, (GlcNAc)6. Lines were obtained by roughly following the experimental data points. (A) Substrate (GlcNAc)6. (B) Substrate (GlcNAc)5. (C) Substrate (GlcNAc)4. (D) Substrate (GlcNAc)3. (E) HPLC analysis of size and anomeric configuration of products formed from (GlcNAc)3 by BcChi-A. Bold numbers 1–6 indicate (GlcNAc)1–6.

Unfolding curves of BcChi-A-E61A in the presence of (GlcNAc)n

BcChi-A-E61A was successfully produced and purified by the methods described previously [16]. The unfolding curves of BcChi-A-E61A were obtained by monitoring CD at 222 nm in the absence or presence of (GlcNAc)n (= 1–6), as shown in Fig. 2. Although some of the experimental data points deviated from the theoretical curves obtained by assuming a simple two-state transition (Fig. 2; the early stage of the transition in the absence of the ligand), most data points for the individual unfolding experiments fitted well to the corresponding theoretical curves. Since the unfolding transitions were found to be irreversible, we could not determine the thermodynamic parameters and therefore evaluated the structural stability only from the transition temperatures (Tm) of the major transition zones. As judged from elevation of TmTm), the thermal stability increased when (GlcNAc)n (= 2, 3, 4, 5 or 6) was added to BcChi-A-E61A solution, by 1.5, 2.7, 4.0, 5.3 or 6.2 °C, respectively (Table 1). In the presence of GlcNAc, the Tm value was essentially unchanged. The elevation of Tm appears to be almost proportional to the polymerization degree of the (GlcNAc)n added. The stabilization effects induced by the oligosaccharides should be derived from the specific binding of (GlcNAc)n (= 2–6) to BcChi-A-E61A, suggesting again that the longer the chain length of the ligand, the stronger the ligand binding to the enzyme.

Figure 2.

 Thermal unfolding curves of BcChi-A-E61A obtained in the absence and presence of (GlcNAc)n (= 1–6). The transitions were followed by CD at 222 nm. The unfolding conditions are described in Experimental procedures. Solid lines indicate the theoretical curves obtained with the two-state transition model. Symbols: open circles, no ligand; upside-down triangle, (GlcNAc)1; square, (GlcNAc)2; triangle, (GlcNAc)3; diamond, (GlcNAc)4; cross, (GlcNAc)5; closed circle, (GlcNAc)6.

Table 1.   The transition temperature Tm of thermal unfolding and Tm elevation ΔTm upon the addition of (GlcNAc)n (= 1–6) for BcChi-A-E61A.
ProteinTmΔTm
BcChi-A-E61A76.8
+(GlcNAc)176.5−0.3
+(GlcNAc)278.31.5
+(GlcNAc)379.52.7
+(GlcNAc)480.94.0
+(GlcNAc)582.15.3
+(GlcNAc)683.06.2

ITC analysis of (GlcNAc)n binding to BcChi-A-E61A

The binding of oligomeric substrates to BcChi-A was studied by ITC at 30 °C and pH 7.0. Figure 3 shows typical ITC thermograms and theoretical fits to the experimental data for (GlcNAc)6, (GlcNAc)5, (GlcNAc)4 and (GlcNAc)3 binding. Theoretical fits were obtained and compared with experimental data using a nonlinear least-squares algorithm (accompanied by the ITC system) by varying the binding affinity constant (Ka), the number of binding sites, i.e. the stoichiometry of the reaction (n), and the enthalpy change of ligand binding (ΔHr°). For all chitin oligosaccharides, all fits yielded n between 0.9 and 1.1 indicating a one to one stoichiometry. At this temperature and pH, (GlcNAc)6 was found to bind to the enzyme with a Kd of 0.77 ± 0.06 μm (Table 2). The binding was clearly enthalpy driven (ΔHr° = −9.5 ± 0.1 kcal·mol−1) with a small entropy penalty (−TΔSr° = 1.0 ± 0.1 kcal·mol−1 and ΔSr° = −3 ± 1 cal·K−1·mol−1). An exception is (GlcNAc)3 binding, where an increase in the enthalpic contribution compared with (GlcNAc)4 and a relatively higher unfavorable entropic contribution are observed (Table 2).

Figure 3.

 Thermograms (upper panels) and binding isotherms with theoretical fits (lower panels) obtained for the binding of (A) (GlcNAc)6, (B) (GlcNAc)5, (C) (GlcNAc)4 and (D) (GlcNAc)3 to BcChi-A-E61A.

Table 2.   Thermodynamic parameters for binding of (GlcNAc)n (= 3–6) to BcChi-A-E61A at 30 °C and pH 7.0 as determined by ITC.
SubstrateKd (μm)ΔGr° (kcal·mol−1)ΔHr° (kcal·mol−1)TΔSr° (kcal·mol−1)ΔSr° (cal·K−1·mol−1)
(GlcNAc)60.77 ± 0.06−8.5 ± 0.1−9.5 ± 0.11.0 ± 0.1−3 ± 1
(GlcNAc)51.8 ± 0.3−7.9 ± 0.1−8.1 ± 0.40.2 ± 0.4−1 ± 1
(GlcNAc)416 ± 1−6.6 ± 0.1−7.6 ± 0.31.0 ± 0.3−3 ± 1
(GlcNAc)3230 ± 20−5.0 ± 0.1−8.3 ± 0.63.3 ± 0.6−11 ± 2

(GlcNAc)6 binding to BcChi-A-E61A was also analyzed at pH 6.0 and 8.0. The results (Table 3) showed that the change in pH had insignificant effects on binding, showing slight increases in the Kd values and little variation in the enthalpic and entropic terms.

Table 3.   Thermodynamic parameters for binding of (GlcNAc)6 to BcChi-A-E61A at 30 °C and at various pH values as determined by ITC.
pHKd (μm)ΔGr° (kcal·mol−1)ΔHr° (kcal·mol−1)TΔSr° (kcal·mol−1)ΔSr° (cal·K−1·mol−1)
6.01.0 ± 0.1−8.3 ± 0.1−8.7 ± 0.10.4 ± 0.1−1 ± 1
7.00.77 ± 0.06−8.5 ± 0.1−9.5 ± 0.11.0 ± 0.1−3 ± 1
8.00.91 ± 0.05−8.4 ± 0.1−9.7 ± 0.21.3 ± 0.1−4 ± 1

Measurements of the temperature dependence of ΔHr° for (GlcNAc)6 binding to BcChi-A-E61A in the temperature range between 20 and 37 °C yielded a straight line of slope −105 ± 8 cal·K−1·mol−1Cp,r°) at pH 7.0 (Table 4 and Fig. 4). By recognizing that the entropy of solvation is close to zero for proteins near 385 K, ΔCp can be related to the solvation entropy change (ΔSsolv°) of the binding reaction at = 30 °C as described by

image(1)
Table 4.   Parameterization of the entropic term at 30 °C and pH 7.0. Data for ΔCp,r° are derived from the temperature dependence of ΔHr°.
ΔCp,r° (cal·K−1·mol−1)ΔSr° (cal·K−1·mol−1)ΔSmix° (cal·K−1·mol−1)ΔSsolv° (cal·K−1·mol−1)ΔSconf° (cal·K−1·mol−1)
−105 ± 8−3 ± 1−825 ± 2−20 ± 2
Figure 4.

 Temperature dependence of (GlcNAc)6 binding to BcChi-A-E61A at pH 7.0. The plots of ΔHr° versus temperature yield the change in heat capacity (ΔCp°) as the slope. The ΔCp° value was calculated to be −105 ± 8 cal·K−1·mol−1.

Furthermore, the mixing entropy change (ΔSmix°) of the reaction can be calculated as a ‘cratic’ term, a statistical correction that reflects mixing of solute and solvent molecules and the changes in translational/rotational degrees of freedom:

image(2)

Finally, the reaction entropy change (ΔSr°) can be viewed as the sum of ΔSsolv°, ΔSmix° and the conformational entropy change (ΔSconf°) (Eqn 3). ΔSconf° details the change in side-chain and backbone conformational entropy associated with binding.

ΔSr° can be parameterized into three terms as shown in Eqn (3) [18].

image(3)

Thus, ΔSsolv° may be derived from ΔCp,r° [18–21] and ΔSmix represents a fixed known ‘cratic’ term [18], meaning that ΔSconf° can be derived from ΔSr°. The results summarized in Table 4 show that at pH 7.0 ΔSsolv° is equal to 25 ± 2 cal·K−1·mol−1 (−TΔSsolv° = −7.6 ± 0.6 kcal·mol−1), the loss of translational entropy ΔSmix° = −8 cal·K−1·mol−1 (−TΔSmix° = 2.4 kcal·mol−1) and the entropic effect of conformational changes, ΔSconf°, is equal to −20 ± 2 cal·K−1·mol−1 (−TΔSconf° = 6.1 ± 0.6 kcal·mol−1).

Discussion

Crystal structures of both ‘loopless’ and ‘loopful’ family GH19 chitinases from various plant species, including barley seeds, jack bean, mustard greens, papaya, Norway spruce and rice, have been solved by different research groups [10,15,22,23]. Nevertheless, no crystal structure of a family GH19 enzyme complexed with a GlcNAc oligosaccharide is available at present. The structure of a GH19 chitinase in a complexed state was obtained only for ‘loopful’ papaya chitinase with two separately bound GlcNAc monomers [10]. Furthermore, quantitative analysis of the substrate binding to GH19 chitinases has not been conducted yet, and the mechanism of substrate recognition has never been understood. In this study, we report the first quantitative data on the substrate binding to a ‘loopless’ family GH19 chitinase, which were obtained by kinetic and thermodynamic strategies.

We compared the observed thermodynamic values with those of other carbohydrate binding systems such as glycoside hydrolases and lectins. First, the Kd value for (GlcNAc)6 binding to BcChi-A is approximately 3-fold and 6-fold higher respectively than that observed for the GH18 chitinases Chit-42 from Trichoderma harzianum and ChiB from Serratia marcescens. Moreover, all three chitinases demonstrate a clear proportionality between chain length and binding strength as Kd values increase with decreasing length of the oligomers. In comparison, (GlcNAc)6 binds 100-fold more strongly to BcChi-A than to human lysozyme (0.77 versus 73 μm).

Furthermore, for (GlcNAc)6 binding at pH 7.0 and 30 °C, the driving force is the enthalpy change (ΔHr° = −9.5 ± 0.1 kcal·mol−1) and the solvation entropy change (−TΔSsolv° = −7.6 ± 0.6 kcal·mol−1). The latter is almost negated by an unfavorable conformational entropy change (−TΔSconf° = 6.1 ± 0.6 kcal·mol−1). Such thermodynamic signatures are also observed for methyl 3-O-(α-d-mannopyranosy1)-α-d-mannopyranoside binding to the lectin concanavalin A [24], xylo-oligosaccharides (XOS) to xylanases [25] and chitotriose to lysozyme [26]. The latter system deviates to a certain extent with a much more negative conformational entropy change compared with the positive solvation entropy change (−TΔSsolv° = −8 kcal·mol−1 versus −TΔSconf° = 14 kcal·mol−1). When the binding of 18 inhibitors to the family GH1 β-glucosidase TmGH1 was studied, it was shown that, even though 11 had positive entropy changes, all 18 had negative enthalpy changes [27]. An interesting case is the binding of N-acetyl gluconolactam, N-acetylglucosaminono-1,5-lactone O-(phenylcarbamoyl)-(Z)-oxime (PUGNAc) and N-acetylglucosaminono-1,5-lactone (Z)-oxime (LOGNAc) to a bacterial O-GlcNAcase homologue. The lactam and PUGNAc binding is clearly driven by favorable enthalpy changes with little entropic contribution, but the LOGNAc binding is facilitated by an increase in entropy with no contribution by enthalpy [28]. For (GlcNAc)6 binding to a family 18 chitinase (chitinase B of S. marcescens), the thermodynamic signatures differ remarkably from the ones discussed. Here, binding is accompanied by an enthalpic penalty (ΔHr° = 1.2 kcal·mol−1) and it is desolvation (−TΔSsolv° = −13 kcal·mol−1) that drives the binding with a non-contributing conformational entropy change (−TΔSconf° = −0.1 kcal·mol−1) [29]. For allosamidin binding to S. marcescens chitinase B, the enthalpic penalty is even larger (ΔHr° = 3.8 kcal·mol−1) and the strong binding is facilitated by favorable conformation changes (−TΔSconf° = −11 kcal·mol−1) and, to a smaller extent, desolvation (−TΔSsolv° = −5 kcal·mol−1). These two examples appear to be anomalies in carbohydrate interactions with respect to having unfavorable enthalpic terms.

The binding of (GlcNAc)6 to BcChi-A was independent of pH. For interaction, this cannot only be due to the fact that there are no titratable groups on the ligands, but must also imply that the titratable groups remaining in the catalytic center of BcChi-A after mutating the catalytic Glu61 to a non-titratable Ala are not significantly titrated in the pH 6.0–8.0 range.

BcChi-A hydrolyzed chitin oligosaccharides at rates in the order (GlcNAc)6 > (GlcNAc)5 > (GlcNAc)4 >> (GlcNAc)3, suggesting that the longer the chain length of the substrate, the higher the affinity of the oligosaccharides to the enzyme (Fig. 1). A similar suggestion was obtained from the thermal unfolding experiments shown in Fig. 2. The transition temperature of thermal unfolding (Tm) was elevated by the addition of (GlcNAc)n (= 2, 3, 4, 5 and 6). The Tm elevations were found to be proportional to the polymerization degree of the (GlcNAc)n added. It is clear that the Tm elevation (ΔTm) is derived from an increase in the number of interaction sites between BcChi-A and (GlcNAc)n (= 2–6). The Tm data listed in Table 1 are also consistent with the substrate-size dependence of binding free energy change listed in Table 2. All of these data support the idea that BcChi-A has at least six subsites.

In a previous paper [16] it has been shown that (GlcNAc)6 productively binds to the enzyme through three types of productive binding modes, from −4 to +2, from −3 to +3, and from −2 to +4 subsites of BcChi-A. (GlcNAc)5 binds to subsites from −3 to +2 and from −2 to +3. From the oligosaccharide digestion experiments (Fig. 1C), (GlcNAc)4 was found to productively bind to −2 to +2 subsites. (GlcNAc)3 binds to −2 to +1 subsites to be split into (GlcNAc)2 and GlcNAc (Fig. 1E). The binding modes are summarized in Fig. 5 together with the binding free energy changes. Thus, the binding free energy change of subsite +2 is estimated to be −1.6 kcal·mol−1 by subtraction of the free energy change of (GlcNAc)3 (−5.0 kcal·mol−1) from that of (GlcNAc)4 (−6.6 kcal·mol−1). The difference between the binding free energy changes for (GlcNAc)4 and (GlcNAc)6 is calculated to be −1.9 kcal·mol−1, suggesting that there are at least two more subsites that contribute with a ΔGr° ∼ −0.9 kcal·mol−1 each. These values correspond well with XOS binding to xylanase where the free energy change contribution from individual subsites is ∼ −0.7 kcal·mol−1 [25]. Both (GlcNAc)n–chitinase and XOS–xylanase interactions experience a more negative ΔGr° with respect to an increase in substrate length due to a decrease in ΔHr°, in spite of smaller magnitudes for the chitinase, ΔHr° ∼ −0.9 kcal·mol−1 per unit outside the −2 to +2 subsites compared with ΔHr° ∼ −3 kcal·mol−1 per unit for the xylanase outside the −2 and −1 subsite [25]. The binding of (GlcNAc)3 deviates somewhat from the binding of the longer chitooligosaccharides in that the enthalpic contribution is 0.7 kcal·mol−1 (see Table 2) more favorable than observed for (GlcNA)4. This is not surprising as there are likely to be many favorable interactions between the sugars and BcChi-A in the −2 to +1 subsites to overcome the loss of free energy associated with the necessary distortion of the sugar bound in the −1 subsite to allow for hydrolysis. Such thermodynamic signatures have been observed previously for various chitinases by both direct calorimetric measurements [30] and calculations of binding free energies from kinetic measurements [31,32] as discussed below. A subsite energy mapping has also been done for the (GlcNAc)nS. marcescens chitinase B interactions. In that work, ΔGr° for each subsite–sugar unit interaction varies, with −2.7 kcal·mol−1 for subsite +2, ∼ −1 kcal·mol−1 for subsites +3 and +4, and −0.3 kcal·mol−1 for subsite −3 [30]. In addition, binding to +2 and +3 is associated with favorable entropy gains and unfavorable enthalpy gains despite some enthalpically favorable stacking interactions with Trp and Phe residues. It is suggested that favorable conformational changes and unfavorable intermolecular bond breaks within the protein are triggered by the binding of substrate to these subsites, which is responsible for the observed thermodynamic values [29]. Moreover, binding free energies have been estimated for individual sugar-binding subsites using a purely kinetic approach for (GlcNAc)n interactions with both a family GH18 chitinase [31] and a family GH19 chitinase [32]. The former yielded −3.8, +3.1, −2.5, −3.0, +0.8 and −1.8 kcal·mol−1 for the −2 to +4 subsites, respectively, while the latter gave −0.4, −4.7, +3.4, −0.5, −2.3 and −1.0 kcal·mol−1 for the −3 to +3 subsites, respectively. It is interesting to observe that the two subsites outside −2 to +2 of the latter (family GH19 enzyme) contribute with a free energy change upon binding of −1.4 kcal·mol−1, which is close to the −1.9 kcal·mol−1 observed in our work. Moreover, it is apparent that binding to the −1 subsite for both family GH19 and family GH18 chitinases is associated with a positive free energy change that is most probably due to the necessary distortion of the sugar bound in subsite −1.

Figure 5.

 Binding modes and binding free energy changes of chitin oligosaccharides to a family GH19 chitinase from the moss Bryum Coronatum.

Zolotnitsky et al. [25] demonstrated that the ΔCp values can be used to evaluate the contribution of stacking-hydrophobic interactions to xylosaccharide binding to a family GH10 xylanase. They estimated that one aromatic residue located in the catalytic cleft of the enzyme contributes about −100 to −150 cal·K−1·mol−1 to ΔCp. The change in heat capacity (ΔCp = ΔΔHT) for (GlcNAc)6 binding to BcChi-A-E61A was −105 ± 8 cal·K−1·mol−1, possibly suggesting that an aromatic residue of BcChi-A contributes to binding through a hydrophobic interaction with (GlcNAc)6. In fact, as shown in Fig. 6, one tryptophan residue (Trp103) is found near the catalytic residue Glu61 and its side chain is surface exposed within the putative catalytic cleft of the modeled three-dimensional structure of BcChi-A. Our result is consistent with the participation of this residue in substrate binding. This tryptophan residue appears to be similar to Trp62 of hen egg white lysozyme in its localization (Fig. 6). Trp62 is involved in substrate binding of the lysozyme, and significantly contributes to the enzymatic activity [33]. It is likely that Trp103 is involved in the substrate binding of BcChi-A in a similar manner to that of Trp62 of the lysozyme. To confirm this hypothesis and to obtain more insight into the substrate binding mechanism of BcChi-A, a crystal structure determination of the BcChi-A–(GlcNAc)n complex is currently in progress.

Figure 6.

 The three-dimensional model of BcChi-A (light green) was superimposed on the crystal structure of hen egg white lysozyme (cyan) in complex with (GlcNAc)6 (grey) (PDB: 1sfb) [41] (using just the proton donor residues, Glu61 in BcChi-A and Glu35 in hen egg white lysozyme, as reference) with the Pair Fitting wizard of the molecular visualization program pymol.

Experimental procedures

Chemicals

(GlcNAc)n (= 1–6) oligosaccharides were purchased from Seikagaku Biobusiness Co. (Tokyo, Japan). Other reagents were of analytical grade commercially available.

Protein expression and purification

Site-directed mutagenesis was done using the QuickChange site-directed mutagenesis kit (Stratagene, La Jolla, CA, USA). The primers used for the Glu61→Ala mutation were 5′-CTTGGTAACATCAACCAGGCATCCGGAGGGTTGCAGTTTA-3′ and 5′-TAAACTGCAACCCTCCGGATGCCTGGTTGATGTTACCAAG-3′ (the mutation site is underlined). The wild-type and mutated BcChi-A (BcChi-A-E61A) were successfully produced and purified by the methods described previously [16]. We confirmed that the profile of the CD spectrum of BcChi-A-E61A is nearly identical to that of the wild-type and that BcChi-A-E61A is completely inactive toward any chitinase substrates. Protein concentration was determined by reading absorbance at 280 nm, using an extinction coefficient of BcChi-A obtained from the equation proposed by Pace et al. [34].

HPLC-based determination of the reaction time-course

The reaction products from the chitinase-catalyzed hydrolysis of (GlcNAc)n (= 3, 4, 5 or 6) were quantitatively determined by gel filtration HPLC. The enzymatic reaction was performed in 50 mm sodium acetate buffer, pH 5.0, at 40 °C. Enzyme and substrate concentrations were 0.04 and 4.75 mm, respectively, except for the reaction with (GlcNAc)3 in which the enzyme concentration was increased 10-fold. To completely terminate the enzymatic reaction at a given incubation time, a portion of the reaction mixture was mixed with an equal volume of 0.1 m NaOH solution and immediately frozen in liquid nitrogen. The resultant solution was applied to a gel filtration column of TSK-GEL G2000PW (Tosoh, Tokyo, Japan) and eluted with distilled water at a flow rate of 0.3 mL·min−1. Oligosaccharides were detected by ultraviolet absorption at 220 nm. Peak areas obtained for individual oligosaccharides were converted to molar concentrations, which were then plotted against reaction time to obtain the reaction time-course.

HPLC analysis of anomers of the enzymatic products from (GlcNAc)3

The anomeric form of the enzymatic products was determined by HPLC. Enzymatic hydrolysis of (GlcNAc)3 was carried out in 50 mm sodium acetate buffer, pH 5.0, at 25 °C. Concentrations of BcChi-A and (GlcNAc)3 were 0.4 μm and 4.75 mm, respectively. After 4, 10, 15 and 20 min of incubation, a portion of the reaction mixture was directly injected into a TSK-GEL Amide 80 column (Tosoh), and the elution was performed with acetonitrile/H2O (7 : 3) at a flow rate of 0.7 mL·min−1. The substrate and enzymatic products were detected by ultraviolet absorption at 220 nm. The splitting mode of (GlcNAc)3 was qualitatively estimated from the α/β ratio of each oligosaccharide product in the HPLC profiles [35].

Thermal unfolding experiments

To obtain the thermal unfolding curve of the protein, the CD value at 222 nm was monitored using a Jasco J-720 spectropolarimeter (JASCO, Tokyo, Japan) (cell length 0.1 cm), while the solution temperature was raised at a rate of 1 °C·min−1 using a temperature controller (PTC-423L, Jasco). To facilitate comparison between unfolding curves, the experimental data were normalized as follows. The fraction of unfolded protein at each temperature was calculated from the CD value by linearly extrapolating the pre- and post-transition baselines into the transition zone, and was plotted against the temperature. Final concentrations of the enzyme and (GlcNAc)n were 8 μm and 8 mm, respectively.

ITC experiments

ITC experiments were performed with a VP-ITC system from Microcal Inc. (Northampton, MA, USA) [36]. Solutions were thoroughly degassed prior to experiments to avoid air bubbles in the calorimeter. Typically, the BcChi-A-E61A solution in 20 mm sodium phosphate buffer (pH 6.0, 7.0 and 8.0) at 30 °C was placed in the reaction cell with a volume of 1.42 mL, and the ligand solutions in the identical buffers were placed in the ITC syringe. For examining the temperature dependence, ITC measurements were performed at pH 7.0, while the temperature was varied from 20 to 25, 30 and 37 °C. Concentrations of the protein and the ligand solutions used for the ITC experiments are listed in Table 5. For all titrations, 8 μL aliquots were injected into the reaction cell at 200-s intervals with a stirring speed of 260 r.p.m. The titrations were completed after 38 injections. The shape of the ITC binding curve is determined by the so-called Wiseman c value [36], which can be expressed as

image(4)

where n is the stoichiometry of the reaction, Ka is the equilibrium binding association constant and [M]t is the protein concentration. For three of the four ligands ITC experiments could be optimized to give a c value in the range 10 < < 1000. This ensures that Ka can be determined from the Wiseman binding isotherm. Titration of (GlcNAc)3 and BcChi-A-E61A yielded a c value of 0.2. Binding thermodynamics can be obtained using ITC even when c is in the range 0.01 < < 10 if a sufficient portion of the binding isotherm is used for analysis [37]. This is achieved by ensuring a high molar ratio of ligand to protein at the end of the titration, accurate knowledge of the concentrations of both ligand and receptor, an adequate level of signal-to-noise in the data, and known stoichiometry. All of these conditions were satisfied, with the possible exception of the stoichiometry issue. It is conceivable that two (GlcNAc)3 molecules could bind simultaneously. However, the two-site binding model did not provide a satisfactory fit between experimental and theoretical data, whereas the one-site binding model did (see below).

Table 5.   Concentrations of the protein and the ligand solutions used for the ITC experiments.
LigandLigand concentrationsBcChi-A-E61A (μm)
(GlcNAc)6250 μm15
(GlcNAc)5250 μm15
(GlcNAc)41.2 mm55
(GlcNAc)37 mm55

Analysis of calorimetric data

ITC data were collected automatically using the microcal origin v.7.0 software accompanying the VP-ITC system [36]. Prior to further analysis, data were corrected for heat of dilution by subtracting the heat remaining after saturation of binding sites on the enzyme. These heat values had the same magnitudes as that for titrating ligand into buffer alone. Data were fitted using a nonlinear least-squares algorithm and a single-site binding model employed by the origin software that accompanies the VP-ITC system. All data from the binding reactions fitted well with the single-site binding model yielding the stoichiometry (n), equilibrium binding association constant (Ka) and the reaction enthalpy change (ΔHr°) of the reaction. The value of n was found to be between 0.9 and 1.1 for all reactions. The reaction free energy change (ΔGr°) and the reaction entropy change (ΔSr°) were calculated from the relation

image(5)

Errors are reported as standard deviations of at least three experiments at each temperature. A description of how the entropic term is parameterized has been described in detail previously [38,39].

Computer-aided modeling of the three-dimensional structure

swiss model, a knowledge-based protein modeling tool [40], was used to predict the tertiary structure of BcChi-A from the known X-ray structure of a bacterial family 19 chitinase from S. coelicolor A3(2) (Protein Data Bank entry 2CJL) [14]. After modeling, the entire structure was visualized using the program pymol (http://www.pymol.org/). To superimpose the modeled BcChi-A structure on the crystal structure of hen egg white lysozyme in complex with (GlcNAc)6 (PDB entry, 1sfb) [41], the Pair Fitting wizard of the pymol software was used.

Acknowledgements

We are grateful to Dr Karl J. Kramer, USDA-ARS Center for Grain and Animal Health Research, for his critical reading of the manuscript. Thanks are also due to Mr Mitsuru Kubota, Atsushi Urasaki and Takuya Nagata for their technical assistance. This research was supported by FY 2010 Researcher Exchange Program between JSPS and The Research Council of Norway.

Ancillary