Identification of the Catalytic Residues in Family 52 Glycoside Hydrolase, a -Xylosidase from Geobacillus stearothermophilus T-6*

From the Departments of ‡Food Engineering and Biotechnology and §Chemistry and the ‡‡Institute of Catalysis Science and Technology, Technion-Israel Institute of Technology, Haifa 32000, Israel, the Department of Inorganic Chemistry and The Laboratory for Structural Chemistry and Biology, The Hebrew University of Jerusalem, Jerusalem 91904, Israel, and the **Architecture et Fonction des Macromolécules Biologiques, UMR 6098, CNRS and Universités d’Aix-Marseille I and II, 31 Chemin Joseph Aiguier, 13402 Marseille cedex 20, France

␤-D-Xylosidases (EC 3.2.1.37) are exo-type glycoside hydrolases that hydrolyze short xylooligosaccharides to xylose units. The enzymatic hydrolysis of the glycosidic bond involves two carboxylic acid residues, and their identification, together with the stereochemistry of the reaction, provides crucial information on the catalytic mechanism. Two catalytic mutants of a ␤-xylosidase from Geobacillus stearothermophilus T-6 were subjected to detailed kinetic analysis to verify their role in catalysis. The activity of the E335G mutant decreased ϳ10 6fold, and this activity was enhanced 10 3 -fold in the presence of external nucleophiles such as formate and azide, resulting in a xylosyl-azide product with an opposite anomeric configuration. These results are consistent with Glu 335 as the nucleophile in this retaining enzyme. The D495G mutant was subjected to detailed kinetic analysis using substrates bearing different leaving groups (pK a ). The mutant exhibited 10 3 -fold reduction in activity, and the Brønsted plot of log(k cat ) versus pK a revealed that deglycosylation is the rate-limiting step, indicating that this step was reduced by 10 3 -fold. The rates of the glycosylation step, as reflected by the specificity constant (k cat /K m ), were similar to those of the wild type enzyme for hydrolysis of substrates requiring little protonic assistance (low pK a ) but decreased 10 2fold for those that require strong acid catalysis (high pK a ). Furthermore, the pH dependence profile of the mutant enzyme revealed that acid catalysis is absent. Finally, the presence of azide significantly enhanced the mutant activity accompanied with the generation of a xylosyl-azide product with retained anomeric configuration. These results are consistent with Asp 495 acting as the acid-base in XynB2.
␤-D-Xylosidases (EC 3.2.1.37) are hemicellulases that hydrolyze short xylooligosaccharides into single xylose units. These enzymes are part of an array of glycoside hydrolases responsible for the complete degradation of xylan (1). This polymer is the major hemicellulosic polysaccharide in the plant cell wall representing up to 30 -35% of the total dry mass (2). Hemicellulases, together with cellulases, have a key role in the carbon cycle in nature, because they are responsible for the complete degradation of the plant biomass to soluble saccharides. These in turn can be used as carbon or energy sources for microorganisms and higher animals. Hemicellulases have attracted much attention in recent years because of their potential industrial uses in biobleaching of paper pulp (2,3), bioconversion of lignocellulose material to fermentative products (4), improvement of animal feedstock digestibility, and recently in the field of oligosaccharide and thioglycoside synthesis (5,6).
The glycosidic bond is one of the most stable bonds in nature, with a half-life of over 5 million years (7). Glycoside hydrolases can accelerate the hydrolysis of the glycosidic bond by more than 10 17 -fold, making them the most efficient catalysts known. The enzymatic hydrolysis of the glycosidic bonds occurs via two major mechanisms, giving rise to either an overall retention or an inversion of the anomeric configuration. In both mechanisms, the hydrolysis usually requires two carboxylic acids, which are conserved within each glycoside hydrolase family, and proceeds through oxocarbenium ion-like transition states. Inverting glycosidases use a single displacement mechanism with the assistance of general acid and general base residues. Retaining glycosidases follow a two-step double displacement mechanism as shown in Fig. 1, involving two catalytic residues, one functioning as a nucleophile and the other functioning as an acid-base (8).
Identification of the key active site residues is of great importance because it provides crucial information regarding the enzymatic catalytic mechanism and allows rational protein design for novel applications, such as for enzymatic synthesis (5). Candidates for the catalytic residues are primarily identified by searching for conserved carboxylates (Glu or Asp) throughout multiple amino acid sequence alignment. These conserved residues are replaced to a noncarboxylic residue, and the generated mutants are subjected to detailed kinetic analysis using substrates bearing different leaving groups, azide rescue analysis, and pH dependence activity profiles of the mutants and the wild type. In some cases, identification can be accomplished by labeling the catalytic residues using mechanism-based inactivators and affinity labels (9).
Based on amino acid sequence similarities, ␤-D-xylosidases are currently divided into families 3, 39, 43, 52, and 54 of glycoside hydrolases (10,11). These families together with all other glycoside hydrolase families can be readily accessed at the constantly updated web site afmb.cnrs-mrs.fr/CAZY. Although the identities of the catalytic residues for most of these ␤-D-xylosidase families are already known (12)(13)(14)(15)(16), no such information is available for family 52.
Previously, we reported the cloning and purification of a ␤-xylosidase from Geobacillus stearothermophilus T-6 (XynB2) showing homology to family 52 glycoside hydrolases. Its stereochemical course of hydrolysis showed that the configuration of the anomeric carbon was retained, indicating that a retaining mechanism prevails in family 52 glycoside hydrolases (17). Because the ␤-xylosidase from G. stearothermophilus T-6 can be readily overexpressed and purified, it can serve as an excellent representative of family 52 glycoside hydrolases for the identification of the two key active site residues. This paper describes a detailed kinetic analysis of the putative acid-base and nucleophile mutants of XynB2, using substrates bearing different leaving groups, chemical rescue, and pH dependence profiles. The study provides for the first time unequivocal identification of the two catalytic residues of family 52 glycoside hydrolases.
Mutagenesis, Protein Expression, and Purification-The xynB2 gene (GenBank TM accession number AJ305327) from G. stearothermophilus T-6 was cloned in the pET9d vector, overexpressed in Escherichia coli BL21(DE3), and purified as previously reported (17). Site-directed mutagenesis was performed using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA). The mutagenic primers for the mutations were as follows (the mutated nucleotides are in bold type): D495G, 5Ј-GGAAATCACAACGTACGGGAGTTTGGATGTTTCTCTT-GG-3Ј and 5Ј-CCAAGAGAAACATCCAAACTCCCGTACGTTGTGATT-TCC-3Ј; E335G, 5Ј-GCCGATTTGGGTCGTTAACGGCGGCGAGTACC-GGATGATG-3Ј and 5Ј-CATCATCCGGTACTCGCCGCCGTTAACGAC-CCAAATCGGC-3Ј; E335A, 5Ј-GCCGATTTGGGTCGTTAACGCCGGC-GAGTACCGGATGATG-3Ј and 5Ј-CATCATCCGGTACTCGCCGGCGT-TAACGACCCAAATCGGC-3Ј; and E335Q, 5Ј-GCCGATTTGGGTCGT-TAACCAGGGCGAGTACCGGATGATG-3Ј and 5Ј-CATCATCCGG-TACTCGCCCTGGTTAACGACCCAAATCGGC-3Ј. The mutagenic primers were designed to include the mutation and when possible a restriction site to allow easy identification of the mutation. All of the mutations were created by a double base pair substitution to avoid translational misincorporation during protein synthesis by the host cell. The mutated genes were sequenced to confirm that only the desired mutations were inserted, and the proteins were overexpressed and purified as described for the wild type.
Kinetic Studies-Steady state kinetic studies were performed by following the absorbance changes in the UV-visible range, using an Ultrospec 2100 pro spectrophotometer (Amersham Biosciences) equipped with a temperature-stabilized water circulating bath. Initial hydrolysis rates were determined by incubating 500 l of different substrate concentrations (ranging from 0.1 to 7 K m where applicable) in 100 mM phosphate buffer (pH 7.0) containing 1 mg/ml bovine serum albumin at 40°C within the spectrophotometer until thermal equilibration was achieved. The exact temperature inside the cuvette was verified using a thermocouple. The reactions were initiated by the addition of 100 l of appropriately diluted enzyme, and the release of the phenolderived product was monitored at the appropriate wavelength. For very low K m values, the initial rates were measured with special care. For highly reactive substrates, blank mixtures containing all of the reactants except the enzyme were used to correct for spontaneous hydrolysis of the substrates. Sodium azide and formate were added to the reaction mixtures where mentioned. The extinction coefficients used at pH 7.0 and 40°C and the wavelength monitored for each substrate were as follows: 2,5-dinitrophenyl, 420 nm, ⌬ʦ ϭ 3.68 mM Ϫ1 cm Ϫ1 ; 3,4-dinitrophenyl, 400 nm, ⌬ʦ ϭ 11.15 mM Ϫ1 cm Ϫ1 ; 2,4,6-trichlorophenyl, 312 nm, ⌬ʦ ϭ 3.97 mM Ϫ1 cm Ϫ1 ; 4-nitrophenyl, 420 nm, ⌬ʦ ϭ 7.61 mM Ϫ1 cm Ϫ1 ; 2-nitrophenyl, 420 nm, ⌬ʦ ϭ 1.91 mM Ϫ1 cm Ϫ1 ; 4-methylumbelliferyl, 355 nm, ⌬ʦ ϭ 2.87 mM Ϫ1 cm Ϫ1 ; 3-nitrophenyl, 380 nm, ⌬ʦ ϭ 0.455 mM Ϫ1 cm Ϫ1 . The values of K m and k cat were determined by nonlinear regression analysis using the program GraFit 5.0 (19).
pH dependence studies were carried at 40°C with pNPX 1 as a substrate. Mixtures containing 600 l of 1 mg/ml bovine serum albumin and different concentrations of substrate solutions in the appropriate buffer were prewarmed until the reaction was initiated by the addition of 200 l of appropriately diluted enzyme. The buffers used were at a final concentration of 100 mM and were: citric acid-Na 2 HPO 4 (pH 4.5-6.5), phosphate buffer (pH 6.0 -8.0), and Tris-HCl buffer (pH 7.5-8.5). The pH range employed in this study included only pH values for which the enzyme was stable for at least 5 min. The reactions were monitored continuously at 40°C, and upon completion the actual pH was measured to verify that the pH had not changed. The release of p-nitrophenol was monitored at 400 nm, and the mM extinction coefficients for pnitrophenolate were determined at pH 4.63, 5.35, 6.02, 6.53, 6.93, 7. Catalysis by retaining glycosidases proceeds via a double displacement mechanism involving two catalytic residues functioning as a nucleophile and an acid-base. The kinetic constants are as follows: XynB2-E335G or XynB2-D495G, 10 mM of 2,5-DNPX, and 1 M sodium azide in a final volume of 10 ml of 100 mM phosphate buffer, pH 7.0. The mixtures were incubated at 40°C, and the reaction was monitored by TLC. TLC analysis was performed using precoated plates (Silica Gel 60 F 254 , 0.25 mm; Merck), and MeOH/CHCl 3 1:4 as the running solvents. The spots were visualized by charring with a yellow solution containing 120 g of (NH 4 )Mo 7 O 24 ⅐4H 2 O and 5 g of (NH 4 ) 2 Ce(NO 3 ) 6 in 800 ml of 10% H 2 SO 4 . After complete hydrolysis of the substrate (ϳ5 h), the mixtures were lyophilized, and the resulting solid was extracted with methanol (4 ϫ 5 ml). The extracts were combined and evaporated to dryness. The crude material was purified by flash chromatography (MeOH:CHCl 3 , 1:9) on a silica gel (Merck; 63-200 mesh) column to yield the pure product as a white solid (40 mg). 1 H NMR and 13 C NMR spectra were recorded at an ambient temperature on a Bruker Avance 500 MHz spectrometer. The mass spectrum was obtained on a TSQ-70B mass spectrometer (Finnigan Mat) by negative chemical ionization in isobutane or on a Bruker Daltomics Apex-III (ICR-MS) by the method of electrospray ionization. Fourier transform infrared spectroscopy (FTIR) was recorded on a Bruker vector 22 spectrometer.

RESULTS
Site-directed Mutagenesis-We have recently described the purification and stereochemical course of hydrolysis of a ␤-xylosidase (XynB2) from family 52 glycoside hydrolases (17). In that report we suggested that Glu 337 and Glu 413 are involved in catalysis. However, further examination of these mutants revealed that they are not catalytic residues. To identify the catalytic residues of XynB2, the amino acid sequences of family 52 glycosidases were aligned, and many conserved carboxylic residues were revealed. These were replaced via site-directed mutagenesis to glycine, and the generated mutants were screened for reduction in activity. During the course of the work, it was apparent that part of the mutants exhibited inconsistent behavior. As it turned out, inconsistency arose from two main reasons: (a) contaminations from translational misincorporation as was observed previously (20) and (b) contaminations from the purification procedure. To avoid these potential problems, all of the mutations were created by a double base pair substitution, and special care was taken during all of the purification procedures. Following systematic replacement of the conserved carboxylic residues in XynB2, two mutants, E335G and D495G, that reside in highly conserved regions were promising candidates for the catalytic pair. Therefore, these were overexpressed, purified, and subjected to extensive kinetic analysis as described below.
Catalytic Properties of the Glu 335 and Asp 495 Mutants-Glu 335 was replaced with Gly, Ala, or Gln, and the catalytic properties of the E335G, E335A, and E335Q mutants using pNPX and 2,5-DNPX as substrates were determined and summarized in Table I. The k cat values measured for the E335G, E335A, and E335Q mutants were all significantly reduced and are about 10 Ϫ6 -fold of wild type activity with both pNPX and 2,5-DNPX as substrates. The very low activity precluded reliable K m determination.
Similarly, Asp 495 was replaced with Gly, and the kinetic constants of the D495G mutant were determined using different aryl ␤-D-xylopyranosides with different leaving groups (Table II and Fig. 2). The k cat values of the D495G mutant were 10 3 -fold lower than for the wild type. For both the wild type enzyme and the D495G mutant, k cat values were invariant for hydrolysis of all of the substrates. Although the K m values of the wild type enzyme were roughly similar with all substrates, with the D495G mutant these values increased as the substrate reactivity decreased (increasing pK a ). Consequently, the decrease in the specificity constant (k cat /K m ) values for the D495G mutant was more pronounced, because the substrates are less reactive.
Chemical Rescue of the Catalytic Mutants-Rate acceleration by exogenous nucleophilic anions is the most definitive tool for the identification of the catalytic residues (21). In the presence of 1.4 M azide, k cat values increased by ϳ4, 33, and 400 times for E335Q, E335A, and E335G mutants, respectively (Table I), indicating that the effect of azide decreases as the side chain is longer. The addition of 2.3 M formate resulted in a 10 3 -fold increase of k cat for the E335G mutants (Table I). Both azide and formate accelerated the reaction in a concentration-dependent manner.
The kinetic constants of hydrolysis of 2,5-DNPX by the D495G mutant in the presence of different concentrations of azide were measured and are plotted in Fig. 3. Both k cat and K m increased with increasing azide concentrations until leveling off at about 0.5 M azide. Consequently, the k cat /K m values remained unchanged. The effect of azide was also tested for the hydrolysis of substrates with different leaving groups at substrate saturating conditions (Fig. 4). All of the k cat values increased with increasing concentrations of azide until reaching a plateau. Finally, the kinetic constants for hydrolysis of various aryl ␤-D-xylopyranosides were determined in the presence of 0.63 M azide (Table II and Fig. 5).
Characterization of Reaction Products in the Presence of Sodium Azide-The formation of a glycosyl-azide product is a powerful diagnostic tool for identifying the catalytic residues, and determining its anomeric configuration is useful for distinguishing between the acid-base and the nucleophile. TLC analysis of the reaction mixture containing E335G, 2,5-DNPX, and azide revealed the formation of a new product (R f ϭ 0.5) distinct from xylopyranoside (R f ϭ 0.18) and 2,5-DNPX (R f ϭ 0.51). This new product was isolated and identified as ␣-Dxylopyranosyl azide, as determined by 1 H NMR (Fig. 6a), 13  Likewise, TLC analysis of the reaction mixture containing D495G, 2,5-DNPX, and azide revealed the formation of a new product (R f ϭ 0.42) distinct from xylopyranoside (R f ϭ 0.18) and 2,5-DNPX (R f ϭ 0.51). This new product was isolated and identified as ␤-D-xylopyranosyl azide, as determined by 1 H NMR (Fig. 6b), 13   pH Dependence-The k cat values of the wild type enzyme and the D495G mutant for hydrolysis of pNPX were determined at different pH values in the range of 4.5-8.5 (Fig. 7). The pH activity profile of the wild type enzyme showed strong dependence upon pH changes with pK a values of Ͻ4 and 7.3. Conversely, no such dependence is observed for the mutant enzyme within these pH values. The activity of the mutant at pH levels lower than 4 could not be determined because the enzyme was insoluble.

DISCUSSION
Glu 335 Is the Catalytic Nucleophile-In retaining glycoside hydrolases, the nucleophilic residue is extremely important for carrying out efficient catalysis. The nucleophile attacks the anomeric carbon to form a covalent enzyme intermediate, which is sufficiently stable to permit the diffusion of the leaving  -D495G (q). a, plot of log(k cat ) versus pK a of the aglycone. b, plot of log(k cat /K m ) versus pK a of the aglycone. The initial rates were determined at 40°C in 100 mM phosphate buffer, pH 7.0. group from the active site and subsequently the entrance of a water molecule. Furthermore, the nucleophile is important for maintaining the correct ionization state of the acid-base catalyst, and in some cases, for assistance in stabilizing the oxocarbenium ion-like transition state by creating strong hydrogen bond with the sugar 2-hydroxyl (22). Thus, replacing the nucleophile with a noncarboxylic residue usually severely affects the enzymatic catalysis and in some cases leads to nondetectable activity (21).

FIG. 2. Brønsted plots of the hydrolysis of aryl ␤-D-xylopyranosides catalyzed by XynB2 (E) and XynB2
The k cat values measured for the E335G, E335A, and E335Q mutants were all drastically reduced with activities of about 10 6 -fold lower than the wild type for hydrolysis of both pNPX and 2,5-DNPX. This magnitude of decrease in activity is typical for nucleophile mutants and was also observed in other retaining glycoside hydrolases (21). However, the substantial decrease in the catalytic activity is insufficient for the unequivocal assignment of Glu 335 as the catalytic nucleophile. There are several examples where single mutations in proposed catalytic residues of glycoside hydrolases resulted in reduced or undetectable activity (23,24), whereas crystallographic and biochemical analyses showed that these residues are not involved directly in catalysis (25,26). To unambiguously identify Glu 335 as the catalytic nucleophilic residue of XynB2, the rescue methodology was applied. In this procedure (Fig. 8a), the catalytic activities of the putative catalytic mutants are monitored in the presence of small nucleophilic anions, such as azide or formate. The small anion can enter the vacant place created by the elimination of the nucleophilic residue and attack the anomeric carbon of the sugar substrate to form a glycosyl-azide product (when azide is added) with inverted anomeric configuration. Rate acceleration of the mutant in the presence of external anions is a strong indication that the mutation is indeed in the catalytic residue.
Although no rate enhancement was observed for the wild type enzyme, the hydrolysis rates of 2,5-DNPX by the E335G mutant significantly accelerated with increasing concentrations of azide and formate.  smaller, preventing azide facile penetration. Interestingly, formate had a greater effect on rate acceleration, although it exhibits lower nucleophilicity. This was also observed with the nucleophile-less mutants of the ␤-glucosidase from Agrobacterium faecalis (27) and the 1,3-1,4-␤-glucanase from Geobacillus licheniformis (28). It is possible that the strong resemblance between formate and the missing carboxylate nucleophile allows better accommodation of formate in the created cavity. Similar chemical rescue of activity in the presence of azide and formate was observed for many other retaining glycoside hydrolases (20,(27)(28)(29)(30).
Azide can rescue the activity of both the nucleophile and acid-base mutants. However, the anomeric configuration of the glycosyl-azide product is inverted or retained for the nucleophile and acid-base mutant, respectively (Fig. 8). To verify the role of Glu 335 as the catalytic nucleophile, azide was added to the reaction mixture containing the E335G mutant and 2,5-DNPX. The isolated product revealed the formation of 1-azido-1-deoxy-␣-D-xylopyranoside (Fig. 6a). The inverted anomeric configuration of the xylopyranosyl-azide product is exactly as would be expected if Glu 335 is the catalytic nucleophile. Together, these results provide unambiguous confirmation for the assignment of the conserved Glu 335 as the catalytic nucleophile of XynB2 and, by extension, of the other members of family 52 glycoside hydrolases.
Catalytic Properties of the D495G Mutant-Catalysis by retaining glycosidases proceeds via a double displacement mechanism involving two catalytic residues (nucleophile and acidbase) as shown in Fig. 1. According to this mechanistic pathway, substitution of the acid-base catalytic residue by mutation with a nonacidic residue should affect the rates of both chemical steps, although the effect on each step will be different. The effect on the glycosylation step will depend strongly on the leaving group ability of the aglycone. The rates of hydrolysis for substrates with a poor leaving group should be more affected than those with a good leaving group. Therefore, if this step is rate-limiting, k cat values will vary with substrate reactivity. The deglycosylation step, however, will be affected equally for all substrates carrying different leaving groups because the same glycosyl-enzyme intermediate is hydrolyzed during this step. Thus, if this step is rate-limiting, the k cat values will be invariant with substrate reactivity. The k cat values of the D495G mutant were reduced by 10 3 -fold, consistent with Asp 495 acting as a catalytic residue (Table II). The Brønsted plot of log(k cat ) versus pK a shown in Fig. 2a reveals that for the wild type enzyme, the deglycosylation is the ratelimiting step because k cat values are invariant for all substrates. Likewise, this step is also rate-limiting for the D495G mutant for hydrolysis of substrates with pK a Ͻ 8. Thus, for these substrates the deglycosylation step was more affected upon the removal of Asp 495 . The 10 3 -fold reduction of this step, as evident by the relative k cat values of the wild type and the mutant (Table II), is consistent with Asp 495 acting as a general base catalyst. Such a large rate reduction in the deglycosylation step was also observed for other acid-base mutants from the ␣-L-arabinofuranosidase of G. stearothermophillus (31) and the family 1 ␤-glucosidases of A. faecalis (32) and Streptomyces sp. (33). Furthermore, hydrolysis of these substrates by the mutant D495G resulted in very low K m values (Table II), suggesting that a glycosyl enzyme intermediate accumulates, as would be expected if the second step is rate-limiting. Interestingly, hydrolysis of 3-nitrophenyl ␤-D-xylopyranoside (pK a ϭ 8.39) by the mutant resulted in k cat that is approximately three times lower than the k cat values for hydrolysis of substrates with pK a Ͻ 8, suggesting that in this case the glycosylation step is, at least partially, rate-limiting. This claim will gain more support upon applying chemical rescue with exogenous anions, as will be described in the following section.
The effect of the mutation on the glycosylation step can be obtained by comparing the specificities (k cat /K m ) of the wild type and the D495G mutant (Table II). Because k cat /K m ϭ k 1 k 2 /(k Ϫ1 ϩ k 2 ), and k 1 and k Ϫ1 are the rate constants for the Michaelis substrate-enzyme complex association and deassociation, respectively, it is likely that k cat /K m reflects the first glycosylation step. For hydrolysis of substrates with good leaving groups, k cat /K m are similar for both the wild type and the mutant. However, for substrates with poor leaving groups, the k cat /K m values of the D495G mutant were two orders of magnitude smaller than those of the wild type. Thus, the replacement of Asp 495 had little effect on the hydrolysis rates of the first step for substrates with good leaving groups that require little or no acid assistance. Conversely, the first step was severely affected with poor leaving groups substrates that require strong acid assistance. This behavior is consistent with Asp 495 acting as a general acid catalyst. More information regarding proton assistance during the glycosylation step can be obtained from the Brønsted plot of log(k cat /K m ) versus the pK a of the leaving group. The Brønsted plot shown in Fig. 2b reveals good correlation between the pK a and log(k cat /K m ) with a slope of ␤ 1g ϭ Ϫ1 for XynB2-D495G. The corresponding correlation for XynB2 was derived only for poor substrates with pK a of Ͼ8 because a biphasic relationship was observed, probably reflecting a change in the enzyme-substrate association rate constant, which becomes rate-limiting for the most highly reactive substrates (33,34). Thus, for the wild type enzyme a slope of ␤ 1g ϭ Ϫ0.82 was observed (data not shown), and regardless of the reason for this biphasic nature, it is clear that the reaction catalyzed by the mutant is much more dependent on the aglycone leaving group ability. The larger value of ␤ 1g for the mutant as compared with that of the wild type suggests a higher amount of negative charge on the glycosidic oxygen of the leaving group in the glycosylation transition state. This difference probably results from little proton donation by the mutant, exactly as would be expected in the absence of acid catalysis.
Chemical Asp 495 as the acid-base catalyst can also be verified by the use of the azide rescue methodology (Fig. 8b). Small anions can accelerate the second step, therefore increasing k cat values if the second step is rate-limiting, and in this case a ␤-glycosyl azide product is expected. Indeed, in the presence of increasing concentrations of azide, the k cat values significantly increased for hydrolysis of 2,5-DNPX (a substrate for which deglycosylation is rate-limiting) by XynB2-D495G (Fig. 3). Interestingly, the rates of hydrolysis increased until reaching a plateau, which probably reflects a change in the rate-limiting step from deglycosylation to glycosylation step. Thus, the value of k cat at the plateau reflects the rate of the glycosylation step, which is not affected by azide. This increase in activity is markedly high, with k cat values increasing by almost 10 3 -fold, approaching the rates of the wild type enzyme. Rate enhancements by azide were also observed for other acid-base mutants, although in these cases only up to a 3 ϫ 10 2 -fold increase was achieved (32,35). The presence of azide also affected the K m values, and these increased dramatically until leveling off (Fig. 3). For retaining glycoside hydrolases K m ϭ (k Ϫ1 ϩ k 2 )k 3 /(k 2 ϩ k 3 )k 1 . This K m value is small for hydrolysis of a good substrate, such as 2,5-DNPX (pK a ϭ 5.15), by XynB2-D495G, because the deglycosylation step is severely affected (k 3 is low), and the glycosylation step is virtually unaffected (k 2 is high relative to k 3 ). However, because k 3 increases relative to k 2 in the presence of azide, the K m values also increase to a point at which k 2 Ͻ Ͻ k 3 and K m ϭ (k Ϫ1 ϩ k 2 )/k 1 reaching the highest possible K m value that can be obtained for hydrolysis of any given substrate. Although K m and k cat increase in the presence of azide, k cat /K m , which reflects the glycosylation step, is unaffected (Fig. 3), confirming that azide affects primarily the deglycosylation step.
As mentioned above, the presence of azide changed the ratelimiting step from deglycosylation to glycosylation, with k cat values probably reflecting the rate of hydrolysis of the first glycosylation step at the plateau. Hydrolysis by XynB2-D495G of additional substrates with different leaving groups in the presence of azide also resulted in rate enhancement until reaching a similar plateau (Fig. 4). However, the plateau level of k cat values drops as the pK a of the leaving group elevates, and higher concentrations of azide are required for reaching this plateau, because the substrate is more reactive. The correlation between the plateau level and the pK a values provides additional confirmation that indeed glycosylation becomes the rate-limiting step upon the addition of azide. The Brønsted plot relating log(k cat ) values obtained in the absence and presence of 0.63 M azide (already at the plateau) and the pK a of the leaving group shows again the change in the rate-limiting step from deglycosylation to glycosylation because the k cat values vary with substrate reactivity upon the addition of azide (Fig. 5a). The plot reveals good correlation between the pK a and log(k cat ) in the presence of azide with a slope of ␤ 1g ϭ Ϫ1, which is exactly the same as derived from the Brønsted plot of log(k cat / K m ) in the previous section (Fig. 2b). The Brønsted plot of log(k cat /K m ) shown in (Fig. 5b), which reflects the glycosylation step, reveals similar relationships in the absence or presence of azide, verifying again that azide has no effect on the glycosylation step. As speculated in the previous section, the predominantly rate-limiting step for hydrolysis of 3-nitrophenyl ␤-Dxylopyranoside (pK a ϭ 8.39) by the D495G mutant is glycosylation because azide did not affect the reaction rate and the K m value (Table II).
Finally, the reaction mixture containing the D495G mutant, 2,5-DNPX, and azide revealed the formation of ␤-D-xylopyranosyl azide (Fig. 6b). The fact that the anomeric configuration of the xylopyranosyl-azide product is retained provides definite evidence that Asp 495 is indeed the catalytic acid-base. Together, these results are consistent with the assignment of Asp 495 as the catalytic acid-base.
pH Dependence Profiles of XynB2 and XynB2-D495G-The assignment of Asp 495 as the acid-base catalyst can also be done by testing the pH dependence profiles for the wild type and the D495G mutant. The pH dependence profile of the wild type presented in Fig. 7 is a typical bell-shaped curve, suggesting that hydrolysis requires one group in its protonated form and the other in its deprotonated form, as observed for many other glycoside hydrolases. Thus, enzymatic catalysis depends upon two ionizable amino acid residues with pK a values of Ͻ4 and 7.3, which are ascribed to the catalytic pair, with the higher pK a attributed to the acid-base catalyst and the lower pK a assigned to the nucleophile (36). However, with the D495G mutant no dependence in activity was observed at high pH values, suggesting that the protonated group had been removed exactly as would be expected from a glycoside hydrolases lacking its acid catalyst. Unfortunately, determination of activity at pH values lower than 4.5 was precluded because the enzyme precipitated rapidly. Similar behavior was also observed for many other acid-base mutants from different glycoside hydrolases families (12,16,31,32,37,38). Collectively, these results provide unequivocal confirmation for the assignment of the conserved Asp 495 as the catalytic acid-base of the family 52 XynB2.
In conclusion, Glu 335 and Asp 495 are the catalytic nucleophile and acid-base, respectively, of the family 52 XynB2. The large rate reduction for the E335G mutant, together with chemical rescue of activity, and the formation of a xylosyl-azide product with inverted configuration provide unequivocal evidence for the assignment of Glu 335 as the catalytic nucleophile. With the D495G mutant, the large decrease in the deglycosylation step together with the large decrease in the glycosylation step for hydrolysis of poor substrates indicates that both general acid and general base catalysis were severely affected. Further, rate enhancement by chemical rescue accompanied with the generation of a xylosyl-azide product with retained configuration, together with the absence of acid catalysis as indicated by pH dependence profiles provide unambiguous confirmation for the assignment of Asp 495 as the catalytic acid-base. Thus, these results allow for the first time direct identification of the catalytic residues of XynB2, and by extension, those of all members of family 52 glycoside hydrolases.