Substrate Specificity in Glycoside Hydrolase Family 10

The Pseudomonas family 10 xylanase, Xyl10A, hydrolyzes β1,4-linked xylans but exhibits very low activity against aryl-β-cellobiosides. The family 10 enzyme, Cex, fromCellulomonas fimi, hydrolyzes aryl-β-cellobiosides more efficiently than does Xyl10A, and the movements of two residues in the –1 and –2 subsites are implicated in this relaxed substrate specificity (Notenboom, V., Birsan, C., Warren, R. A. J., Withers, S. G., and Rose, D. R. (1998)Biochemistry 37, 4751–4758). The three-dimensional structure of Xyl10A suggests that Tyr-87 reduces the affinity of the enzyme for glucose-derived substrates by steric hindrance with the C6-OH in the –2 subsite of the enzyme. Furthermore, Leu-314 impedes the movement of Trp-313 that is necessary to accommodate glucose-derived substrates in the –1 subsite. We have evaluated the catalytic activities of the mutants Y87A, Y87F, L314A, L314A/Y87F, and W313A of Xyl10A. Mutations to Tyr-87 increased and decreased the catalytic efficiency against 4-nitrophenyl-β-cellobioside and 4-nitrophenyl-β-xylobioside, respectively. The L314A mutation caused a 200-fold decrease in 4-nitrophenyl-β-xylobioside activity but did not significantly reduce 4-nitrophenyl-β-cellobioside hydrolysis. The mutation L314A/Y87A gave a 6500-fold improvement in the hydrolysis of glucose-derived substrates compared with xylose-derived equivalents. These data show that substantial improvements in the ability of Xyl10A to accommodate the C6-OH of glucose-derived substrates are achieved when steric hindrance is removed.

Glycoside hydrolases cleave glycosidic bonds by acid-base catalysis via either a single or a double displacement mechanism, leading to inversion or retention of anomeric configuration, respectively (1). Using a combination of primary structure homology and hydrophobic cluster analysis, these enzymes have been grouped into more than 80 families (2,3). A feature of this sequence-based classification is that enzymes in a given family catalyze glycosidic bond cleavage with the same mechanism, display a common protein fold, and are therefore believed to have evolved from a common ancestor (4). As more three-dimensional structures become available and sequencesimilarity programs become more sensitive, a number of "clans" of distantly related glycoside hydrolases have also been discovered, which, in the case of retaining enzyme families, have their catalytic nucleophile and acid-base residues in the same relative positions (4,5).
The three-dimensional structures of Cex and Xyl10A are (␣/␤) 8 barrel proteins in which the two catalytic residues involved in the retaining mechanism are situated at the end of ␤-strands 4 and 7 (13)(14)(15). The substrate binding cleft of Xyl10A contains seven xylose-binding subsites (13), three glycone (-1 to -3) and four aglycone (ϩ1 to ϩ4), whereas Cex contains three glycone and two aglycone subsites (12). Inspection of the -2 and -1 subsites of Xyl10A and Cex reveals differences that could explain the higher efficiency of the Cel-lulomonas enzyme against PNPG 2 compared with the Pseudomonas xylanase. In Cex Gln-87, located in the -2 subsite, becomes disordered when glucose enters this site in the enzyme, suggesting that the amino acid impedes access of the C-5 hydroxymethyl group of the hexose sugar (14). The corresponding residue in Xyl10A, Tyr-87, is both more bulky and potentially less flexible than Gln-87 and is likely to hinder access of the C-5 hydroxymethyl group of glucose to the active site of Xyl10A (Figs. 1 and 2). In the -1 subsite of Cex, Trp-281 is repositioned when the enzyme binds a glucose moiety at the -1 subsite, indicating that this residue is making unfavorable contacts with the C-5 hydroxymethyl group of the sugar (14). Xyl10A has a direct equivalent, Trp-313, but this is involved in a close hydrophobic interaction with a leucine residue, Leu-314 (Figs. 1 and 2). It is likely that this residue reduces the flexibility of Trp-313 such that the binding of glucose-derived substrates is incurred at greater energetic penalty, as is also seen for the Streptomyces lividans Xyl10A (SlXyl10A), in the accompanying paper by Ducros et al. (16). An overview of the movement of amino acids in Xyl10A, Cex and SlXyl10A, when cellobiose binds at the -1 and -2 subsites, is shown in Fig. 1.
In this paper, we show that Tyr-87 and Leu-314 in Xyl10A impede access of glucose residues to the -2 and -1 subsites of the enzyme, respectively, providing an explanation for the different relative activities of Xyl10A and Cex toward aryl glycosides. The data obtained showed that removal of Tyr-87 from Xyl10A increased the catalytic efficiency of the enzyme against PNPG 2 and decreased the efficiency of hydrolysis of the corresponding xylobioside (PNPX 2 ), whereas the L314A mutation reduced the catalytic efficiency of the xylanase against PNPX 2 but not against PNPG 2 . We also propose that the unusually high catalytic efficiency of Cex toward aryl-␤-glycosides may merely reflect the favored binding of aromatic aglycones in the ϩ1 subsite. In light of the activities of these enzymes toward a range of substrates, the description of some family 10 enzymes as "multifunctional" should perhaps be reevaluated.
The Xyl10A derivative that encoded the double mutant L314A/Y87F was generated by digesting pCM1 (encodes L314A) with ClaI and HindII, removing the 300-base pair sequence that makes up the 5Ј region of Xyl10A and inserting the corresponding ClaI/HindIII restriction fragment from pCM2 (encoding Y87F) into pCM1. The complete sequences of the Xyl10A mutants were determined using an ABI 373 DNA sequenator using the M13 forward and reverse primers, to ensure that only the desired mutations had been introduced into the xylanase gene.
Purification of Recombinant Proteins-Xyl10A and its derivatives were purified essentially as described previously (19). Appropriate recombinant strains of E. coli JM83 were grown to stationary phase, and the periplasm was prepared by the osmotic shock method (19). Xyl10A was prepared from the periplasm by ion exchange chromatography using a DEAE-TrisAcryl Plus M column, and a linear 400 ml of 0 -200 mM NaCl gradient in 10 mM Tris-HCl buffer, pH 8.0, was used to elute the enzyme. The purity of Xyl10A in fractions containing xylanase activity was evaluated by SDS-polyacrylamide gel electrophoresis (21). To purify Cex, cultures of BL21 (DE3):pLysS containing pETcex were grown to mid-exponential phase (A 550 0.5) at 25°C, after which, isopropyl-␤-D-galactopyranoside was added to a final concentration of 1 mM, and the bacterial cells were incubated, without shaking, at 25°C for a further 6 h. Cex contained an N-terminal His 10 tag, and was therefore purified, from cell-free extracts, by nickel ion affinity chromatography, using a Talon TM metal affinity column as described previously (22).
Assays-The substrates used in this study were obtained as follows: oat spelt xylan, medium viscosity carboxymethyl cellulose (CMC), bar-  (14), S. lividans Xyl10A (16), and P. cellulosa (13). Upon formation of the covalent cellobiosyl enzyme intermediate in Cex and SlXyl10A residues in both the -2 and -1 subsites move (shown in red) in order to accommodate the C6-OH groups of glucose. Trp-313 and Tyr-87 in the P. cellulosa enzyme would appear to generate a similar steric hindrance, but Trp-313 may be less able to move given its close packing interaction with Leu-314.

RESULTS AND DISCUSSION
Experimental Rationale-The three-dimensional structures of the S. lividans Xyl10A (SlXyl10A), and the C. fimi enzyme Cex, in complex with their 2-deoxy-2-fluoro-␣-D-xylobiosyl and 2-deoxy-2-fluoro-␣-D-cellobiosyl enzyme intermediates (14 -16) were used as "templates" for the design of site-directed mutants of the highly homologous P. cellulosa Xyl10A, our attempts at trapping a similar intermediate on the P. cellulosa enzyme having proved unsuccessful. Modeling indicated that two residues in particular, Tyr-87 and Trp-313 in the -2 and -1 subsites of the enzyme, respectively, would hinder binding of D-glucose-derived substrates though steric clashes with the C6-OH of D-glucose (Fig. 2). Furthermore, the equivalent of Trp-313 in the C. fimi enzyme Cex (Trp-281) undergoes a conformational change in order to accommodate the C6-OH of D-glucose (14), but in Xyl10A, this movement seemed likely to be restricted due to a tight hydrophobic interaction with Leu-314. A similar restriction is imposed by an arginine in SlXyl10A described in the accompanying paper by Ducros et al. (16). P. cellulosa Xyl10A mutants of Tyr-87, Trp-313, and Leu-314 were therefore constructed in order to test these hypotheses. Mutant enzymes were purified to homogeneity (Fig. 3), and their global structural integrity was evaluated by CD spectroscopy (Fig. 4). The spectra of all mutants analyzed in this study were indistinguishable from that of wild type Xyl10A, indicating that the mutations had not caused a significant perturbation to the secondary structure of these proteins.
Tyr-87 in the -2 Subsite-In both the C. fimi enzyme Cex and SlXyl10A, a glutamine residue in the -2 subsite moves to accommodate D-glucose-derived (as opposed to D-xylose-derived) substrates in the -2 subsite of the enzyme. On Xyl10A, the side chain of Tyr-87 is structurally equivalent to that of Gln-87 in Cex. The greater bulk and reduced conformational freedom of a tyrosine in this position could decrease the catalytic efficiency of the P. cellulosa enzyme on these D-glucosederived substrates. In order to test this hypothesis, we created a Y87A mutant of Xyl10A and evaluated the catalytic efficiency of the enzyme against aryl-␤-glycosides. The Y87A mutation indeed caused an approximately 3-fold increase in the catalytic efficiency (k cat /K m ) of PNPG 2 hydrolysis and a 6-fold reduction in PNPX 2 cleavage (Table I). Interestingly, the decrease in catalytic efficiency against PNPX 2 was primarily due to a reduction in k cat , whereas the increase in efficiency against PNPG 2 was mainly the result of a smaller K m value. The decrease in k cat for PNPX 2 hydrolysis suggests that modification of Tyr-87 did not influence the capacity of the substrate to bind the enzyme but had a direct affect on the actual rate of catalysis. A possible explanation for these data is that the interaction of the substrate with the -2 subsite of the xylanase is different in the mutant and wild type xylanase. Although the mutation to Tyr-87 does not influence the affinity of the arylglycoside for the enzyme, it does have a negative affect on the capacity of the substrate to form a productive complex with the enzyme. This could be due to the substrate adopting a slightly different position in the active site of the mutant enzyme (compared with wild type Xyn10A), such that the xylose moiety at the -1 subsite is farther away from the critical catalytic nucleophile and acid base residues. Alternatively, the position of the xylose at the -1 subsite diminishes the interactions with the enzyme that are critical for the sugar to adopt its semichair/skew boat transition state conformation. In addition to influencing the -2 subsite, the mutations to Tyr-87 could affect the catalytic efficiency of PNPX 2 hydrolysis by causing a subtle distortion in the crucial -1 subsite. To evaluate this latter possibility, the activities of wild type and mutant forms of the enzyme against MUX, a substrate that does not occupy the -2 subsite, were evaluated. The data showed that Y87A was 2-fold 2 K. Johnson, M. Underwood, and M. L. Sinnott, manuscript in preparation. less active than the wild type enzyme against MUX. These results indicate that the decrease in activity against PNPX 2 , on removal of Tyr-87, could partly be explained by a subtle distortion at the active site, resulting in a general reduction in catalytic efficiency.
The increase in PNPG 2 hydrolysis in the two Tyr-87 mutants was primarily a result of a decrease in K m , rather than an increase in k cat . These data indicate that removal of the aromatic amino acid increased the capacity of glucose to bind at the -2 subsite, hence the reduction in K m .The increased affinity of the hexose moiety for Xyl10A is likely to be the result of a decrease in steric hindrance between the C5 hydroxymethyl group of the sugar and the phenolic side chain of Tyr-87.
To investigate whether removal of the OH group of Tyr-87 improved the accessibility of glucopyranoside units to the -2 subsite of Xyl10A, the catalytic efficiency of Y87F was compared with native Xyl10A. The mutant enzyme displayed similar catalytic activities as Y87A against PNPX 2 and PNPG 2 , with a 3-fold decrease and 4-fold increase in catalytic efficiency, respectively, against the two substrates. The observation that the Y87F mutation did not diminish the activity of the enzyme against MUX indicated that substitution of Tyr for Phe did not significantly perturb the -1 subsite.
Collectively, these results suggest that Tyr-87, and in particular its OH group impedes the access of glucopyranoside units to the -2 subsite of Xyl10A, and hence removal of this group increases the catalytic efficiency of the enzyme against aryl-␤-cellobiosides. The loss of activity of Y87F against PNPX 2 presumably reflects a change in the position of xylose at the -2 subsite, such that the location of the substrate in the critical -1 subsite is not optimal for efficient catalysis. Inspection of the three-dimensional structure of two related family 10 xylanases, C. fimi Cex (15) and S. lividans SlXyl10A (16), covalently attached, via the catalytic nucleophile, to 2-deoxy-2-fluoroxylobiose (X 2 F) reveals that the amide side chain of Gln-87, the residue corresponding to Tyr-87, makes an H-bond with the ring oxygen of the xylopyranoside in Cex but that this residue is displaced in SlXyl10A and may make no significant contribution to binding (16). An overlay of the Xyl10A structure with that of Cex suggests that Tyr-87 in the P. cellulosa xylanase is in an appropriate position to form a similar bond between the tyrosine hydroxyl group and the ring oxygen of the sugar in the -2 subsite (Fig. 2).
Leu-314 and Trp-313 in the -1 Subsite-In the C. fimi and S. lividans family 10 enzyme complexes containing covalent cellobiosyl enzyme intermediates, it is clear that a tryptophan that forms a lid over the -1 subsite must have undergone conformational change in order to accommodate the extra bulk of a glucose residue in this position (14,16). In Cex, the lack of a bulky amino acid side chain behind Trp-281 (equivalent to Xyl10A Trp-313) allows this aromatic amino acid more flexibility than Trp-313 in Xyl10A, and conformational change may be achieved with no additional perturbation. Inspection of the -1 subsite of the P. cellulosa enzyme indicated that Trp-313 is likely to impede access of the C-5 hydroxymethyl group of glucose to the -1 subsite of Xyl10A in an analogous way to that observed in Cex and SlXyl10A. In the P. cellulosa enzyme, however, Trp-313 is held in a rigid position by a close hydrophobic interaction with Leu-314, (Fig. 2). A similar situation is observed in SlXyl10A, in which the Trp is held through a stacking interaction with Arg-275, and consequently, displacement of the Trp is accompanied by much more substantial conformational changes around the -1 subsite (Fig. 1 and Ref 16).
To investigate whether Leu-314 was preventing the binding of glucopyranoside units at the -1 subsite by reducing the mobility of Trp-313, the L314A mutation was introduced into Xyl10A, and the catalytic properties of the resultant enzyme were determined. The removal of Leu-314 caused a dramatic decrease in the catalytic efficiency of Xyl10A against PNPX 2 but had no significant effect on the capacity of the enzyme to hydrolyze PNPG 2 (Table I). These results indicate that the leucine residue plays a pivotal role in binding xylopyranoside at the -1 subsite of the enzyme but is apparently less critical for the hydrolysis of glucopyranosides. As Leu-314 is located "behind" Trp-313 in the -1 site of Xyl10A, it is unable to form a direct interaction with a sugar residue in this subsite. It is likely, therefore, that the aliphatic amino acid is mediating its effect by influencing the position of Trp-313, which interacts with sugar residues at the -1 subsite via a hydrophobic interaction (Fig. 1). Removal of the Leu-314 side chain may alter the position of the aromatic residue such that it no longer interacts with a xylopyranose unit at the -1 site. These data support the view that Leu-314 influences the position of the surface aromatic residue Trp-313.
The observation that L314A does not substantially influence PNPG 2 hydrolysis is interesting. Two possible explanations for these data present themselves: (i) the new position adopted by Trp-313 does not influence its capacity to interact with a glucopyranose moiety at the -1 subsite, or (ii) the L314A mutation does disrupt the structure of the -1 subsite, reducing its affinity for sugar residues, but this loss in substrate affinity is at least partly compensated by the increased flexibility afforded to Trp-313, resulting in a decrease in steric hindrance of the C-5 hydroxymethyl group of glucose.
To investigate more directly the role of Trp-313 at the -1 site, the W313A mutation was introduced into Xyl10A, and the catalytic properties of the mutant enzyme were evaluated. The data showed that the enzyme was 1000 and 150 times less active than wild type Xyl10A against PNPX 2 and PNPG 2 , respectively. These results provide further support for the view that the L314A mutation is mediating its effect by diminishing the capacity of Trp-313 to interact with D-xylose units at the -1 subsite. These data also indicate that this aromatic residue plays an important role in PNPG 2 hydrolysis. Removal of Trp-313 caused a substantial reduction in both substrate affinity and the rate of catalysis. These results indicate that the aromatic residue plays a critical role in the binding of both glucose and xylose at the -1 subsite. This decrease in substrate binding in W313A leads to a loss in catalytic activity as the interactions between the sugar at the -1 subsite and the tryptophan side chain are critical for the xylose and glucose moieties to adopt its transition state conformation. The W313A data are difficult to reconcile with the L314A results that showed no effect on PNPG 2 hydrolysis. The interpretation that L314A does affect the catalytic efficiency on PNPG 2 but that this is in some way balanced by the increased flexibility of the Trp is consistent with both the L314A and W313A results.
Do Leu-314 and Tyr-87 Explain Sugar Specificity at the -1 and -2 Subsites?-The data described above, together with the three-dimensional structures of Cex and SlXyl10A in complex with both xylobiosyl and cellobiosyl enzyme intermediates, indicate that Leu-314 and Tyr-87 play pivotal roles in defining substrate specificity at the -1 and -2 subsites of Xyl10A. To test this hypothesis the L314A/Y87F double mutant of the enzyme was constructed, and its catalytic efficiency against aryl-␤-disaccharides was determined. The data (Table I) showed that the catalytic efficiency of the mutant xylanase against PNPX 2 and PNPG 2 was 1200-fold lower and 5-fold higher, respectively, than wild type Xyl10A. The double mutation therefore results in a 6500-fold change in substrate specificity between the two aryl-␤-glycosides, such that L314A/ Y87F preferentially hydrolyzes PNPG 2 , whereas native Xyl10A exhibits a much higher catalytic efficiency against PNPX 2 than the corresponding cellobioside. The double mutation reduced catalytic efficiency against PNPX 2 by decreasing both the affinity of the enzyme for the substrate and diminishing the rate of catalysis. Based on the single mutation data, it is likely that the modification to the -1 and -2 subsites caused a reduction in substrate affinity and the capacity to form a productive enzyme-substrate complex, respectively. The primary affect of the L314A/Y87F mutation against PNPG 2 is to cause a significant increase in the affinity of the enzyme for the substrate, resulting in an elevation in the rate of catalysis, presumably by increasing the capacity of the sugar at the -1 subsite to adopt the transition state conformation. These results thus support the view that substrate specificity at the -1 and -2 subsites is primarily defined by Leu-314 and Tyr-87, respectively. It is apparent from structural studies of family 10 enzymes that the corresponding residue to Xyl10A Tyr-87, in other family 10 enzymes, also plays an important role in the binding of glucose at the -2 subsite, as first proposed by Rose and coworkers (14,15). However, it is less clear whether the amino acids that equate to Leu-314 are pivotal in defining ligand specificity at the -1 subsite. In Cex, there is no bulky amino acid side chain behind Trp-281; this space is instead occupied by a valine, Val-282 (29). Not only is the valine a smaller amino acid, but the main chain conformation in this region is also different from Xyl10A, which results in a different position for Val-282, such that the side chain points away from the Trp aromatic ring. In SlXyl10A, Trp-274 (equivalent to Xyl10A Trp-313) stacks against Arg-275 (30), which restricts the movement of the aromatic residue to accommodate the C-5 hydroxymethyl group of glucose. However, as shown in the accompanying paper (16), when the hexose sugar occupies the -1 FIG. 4. CD spectroscopy of native and mutant forms of Xyl10A. Native Xyl10A (green) and the mutants L314A (red), Y87F (blue), L314A/Y87F (purple), and W313A (black) were subjected to CD spectroscopy as described previously (26). subsite, both Arg-275 and Trp-274 become disordered, suggesting that the flexibility in the location of the basic amino acid enables the aromatic residue to adopt a new position to accommodate glucose (Fig. 1). These data predict that the ratio of PNPX 2 /PNPG 2 catalytic efficiency for the three enzymes would be Cex Ͼ SlXyl10A Ͼ Xyl10A, which is consistent with the observed relative activities of 225, 700, and 3000, respectively (16). Do Subsites -1 and -2 Exhibit Similar Affinities for Glucose?-Although the -1 and -2 subsites of family 10 enzymes can both accommodate glucose, it is unclear whether they exhibit similar affinities for the hexose sugar. To investigate substrate specificity at the -1 and -2 subsites, the activities of wild type Xyl10A against MUX 2, MUXG, and MUG 2 were evaluated. The data showed that the enzyme had similar activities against MUXG and MUX 2 but was substantially less active against MUG 2 (Table I). These results show that the -1 subsite plays a more important role in defining substrate specificity than the -2. This is consistent both with mutations in the -1 subsite and with a previous study that showed that another family 10 xylanase exhibited relative activities similar to that of Xyl10A against the three aryl-␤-disaccharides (31).
Xylan Hydrolysis-To evaluate the effect of disrupting the -1 and -2 subsites on xylan hydrolysis, the capacity of Y87A, Y87F, L314A/Y87F, L314A, and W313A to hydrolyze xylan were determined. The data (Table II) showed that mutations that reduce the catalytic efficiency of Xyl10A against PNPX 2 caused a smaller decrease in xylan hydrolysis. This was particularly apparent in the L314A and W313A mutants, for which the ratio of xylan/PNPX 2 catalytic efficiency changed from 3.5 in the wild type enzyme to 120 and 950, respectively. A likely explanation for these data is that the enzyme is able to utilize binding energy from its many additional subsites that compensate for loss of interactions in the -2 and -1 subsites, as has been observed for many similar mutations (12,26). It is interesting to note that both the W313A and L314A mutations caused an apparent increase in substrate affinity and a decrease in the rate of catalysis. It is possible that although the enzyme, by virtue of its numerous subsites, can bind xylan efficiently, these mutations diminish the capacity of xylose, at the -1 site, to adopt its transition state conformation, and/or its position in the mutant enzymes is not optimal for attack by the nucleophile and acid-base catalytic residues. Thus, although the substrate binds efficiently to the enzyme, its capacity to form a productive complex with the xylanase is reduced. This leads to a decrease in K m as dissociation due to hydrolysis occurs less frequently than in the wild type enzyme. These results are similar to several previous studies that showed that removal of the acid-base catalyst of glycoside hydrolases decreased the rate of catalysis via a lowering of k 2 (k 2 is the rate constant for the enzyme glycosylation step) but also caused a substantial reduction in K m as the decrease in k 2 trapped the enzyme-bound substrate (12,26).
Activity against ␤-Cellobiosyl Fluoride-In addition to the substrate specificity differences between Cex and Xyl10A, re-flected in the catalytic efficiency on PNPG 2 /PNPX 2 , a further notable feature of Cex is its apparent ability to hydrolyze aryl-␤-glycosides substantially more efficiently than other family 10 enzymes. This is a confusing observation because Cex shows almost equal catalytic efficiency to other family 10 enzymes on xylotriose, which occupies the same (-2 to ϩ1) subsites but is approximately 30-fold more efficient against PNPX 2 than the P. cellulosa xylanase (Table I; Ref. 13 and 24). Similarly, Cex is approximately 400-fold more active than Xyl10A against PNPG 2 (12,26).
One possible explanation for this is that Cex can bind aryl-␤-aglycones at the ϩ1 subsite much more efficiently than Xyl10A and other family 10 members. To test this hypothesis, we determined the capacity of both enzymes to cleave ␤-cellobiosyl fluoride, the small aglycone fluorine of which is unlikely to contribute to binding in the ϩ1 subsite. The k cat /K m values for Cex and Xyl10A were 71 and 5.0 min Ϫ1 mM Ϫ1 , respectively. The catalytic efficiency of Xyl10A against the fluoro substrate was approximately 2-fold higher than its catalytic efficiency against PNPG 2 and approximately three times lower than its capacity to hydrolyze 2,4-dinitrophenyl-␤-cellobioside (DNPG 2 ) (12,26). Because the pK a of the leaving group of DNPG 2 (2,4dinitrophenol) is 4.2 and the fluoride component of ␤-cellobiosyl fluoride is 3.2, the higher catalytic efficiency against DNPG 2 strongly suggests that the 2,4-dinitrophenol moiety makes a binding interaction with the ϩ1 subsite in Xyl10A.
In contrast to Xyl10A, the difference in catalytic efficiency of Cex against the fluoro and aryl-␤-cellobiosides is much larger, with the enzyme hydrolyzing DNPG 2 and PNPG 2 98 and 18 times more efficiently, respectively, than 1-fluoro-␤-cellobioside (12,26). These data strongly indicate that the aryl-moieties in DNPG 2 and PNPG 2 bind tightly to the ϩ1 subsite of Cex, and thus, this enzyme hydrolyzes the aryl-␤-glycosides much more efficiently than the fluoro substrate. These data indicate that the unusually high catalytic efficiency of Cex against PNPG 2 , PNPX 2 , and DNPG 2 is not a reflection of efficient binding of the sugar residues in the glycone region of the active site; it is simply due to the enzyme exhibiting relatively high affinity for phenolic groups in the aglycone region of the substrate binding cleft.
The true difference in catalytic efficiency between Cex and other family 10 xylanases, such as Xyl10A, against cellobiosidebased substrates is just 25-fold. The difference in activity of Xyl10A and Cex against larger cellulosic substrates such as cellohexaose and ␤-glucan is also approximately 25-fold (Table  III), suggesting that the difference in activity of the two enzymes against these substrates primarily results from the differential specificities at the -1 and -2 subsites.
What Is the Real Cellulase Activity of Family 10 Xylanases?-The observation that family 10 xylanases exhibit detectable a The activities of the polysaccharides and cellohexaose were relative to soluble oat spelt xylan and xylopentaose, respectively. The concentration of the substrates was 0.2%, and the concentration of enzyme varied from 10 pmol to 1 mol. A value of Ͻ3 ϫ 10 Ϫ5 indicates that no enzyme activity was detected under the assay conditions used. b ASC, acid-swollen cellulose.
activity against aryl-␤-cellobiosides has led to the suggestion that these enzymes exhibit cellulase/exocellulase activity. Based on its PNPG 2 /PNPX 2 catalytic efficiency, Cex was reported to have 50-fold lower activity against cellulose than xylan (32), whereas more recent data suggest that the enzyme exhibits 9-fold higher activity against xylan compared with soluble cellulose (15). To evaluate the relative xylan/cellulose activity of Cex and Xyl10A, we used an oat spelt xylan substrate that was poorly substituted (6% of xylose monomers) with arabinose side chains, xylopentaose, CMC, barley ␤-glucan, acid-swollen cellulose, and cellohexaose. The data, displayed in Table III, showed that the activities of Xyl10A and Cex against ␤-glucan were approximately 10 4 and 10 3 times lower, respectively, against soluble oat spelt xylan. The two enzymes were Ͼ10 4 less active against CMC and acid-swollen cellulose compared with the hemicellulosic substrate. Xyl10A and Cex were 2 ϫ 10 6 and 2 ϫ 10 5 times more active against xylopentaose, respectively, compared with cellohexaose. Inspection of the initial cleavage pattern of acid swollen cellulose hydrolysis by Cex and Xyl10A revealed the presence of several different cellulo-oligosaccharides that were subsequently broken down to cellobiose and glucose (data not shown). These results clearly demonstrate that the two family 10 xylanases have extremely low cellulase activity, and the hydrolysis of this polymer is via a typical endo-mode of attack. The negligible cellulase activity exhibited by Cex is surprising, as the ratio of PNPX 2 /PNPG 2 catalytic efficiency is only 250-fold. The data suggest that the distal regions of the substrate binding cleft do not accommodate glucose molecules as efficiently as the -1 and -2 subsites. This may either result from steric hindrance of the C6-OH groups in the additional subsites or reflect the different secondary structures of the two polysaccharides. In cellulose, interchain hydrogen bonding occurs between C6-OH and C2-OH and between C3-OH and C5-O, resulting in each residue being rotated 180°relative to its neighbor. The lack of a C6-OH/C2-OH in xylan results in the molecule having a left-handed, three fold helical structure, as the xylose moieties are orientated 120°with respect to each other (33). Thus, it is not surprising that a substrate binding cleft that can accommodate a helical polysaccharide interacts extremely weakly to a polymer that has a planar structure. A possible reason why the enzyme can bind glucose more easily at the proximal region of the substrate binding cleft, in addition to the accommodation of the C6-OH groups, is that the characteristic hydrogen bonding pattern of polymeric substrates is frequently distorted in the -2 to ϩ1 subsites, where catalysis occurs (34,35).
The data presented in this report, which suggest a 10 4 -10 5fold difference between the activities on xylan and cellulose, are in contrast with that of Notenboom et al. (15). This could reflect the difference in the xylan substrates used in the two studies. Birchwood xylan, employed by Notenboom et al. (15) can be heavily acetylated (36), which is likely to significantly diminish the activity of xylanases against this substrate. In contrast, the xylan used in this report is not heavily substituted and, in the case of xylopentaose, contains no side chains. We propose, therefore, that family 10 xylanases do not exhibit significant cellulase activity and that the capacity of these enzymes to attack aryl-␤-cellobiosides should not be taken as an indication of their cellulolytic prowess.
Conclusions-Data presented in this report show that Tyr-87 plays an important role in determining the substrate specificity of Xyl10A for aryl-␤-xylobiosides and cellobiosides. The OH group of the phenolic ring may form a beneficial interaction with a xylopyranose unit at the -2 subsite but reduces the affinity of the subsite for glucose by preventing sufficient space for the C-5 hydroxymethyl group of the hexose sugar. Leu-314 plays an important role in positioning Trp-313 in the -1 subsite, so this amino acid can bind to a xylopyranose unit. This report demonstrates how subtle differences in the structure of proximal glycone subsites of glycoside hydrolases can cause substantial differences in the substrate specificities of these enzymes.