The Mechanisms by Which Family 10 Glycoside Hydrolases Bind Decorated Substrates*

Endo-β-1,4-xylanases (xylanases), which cleave β-1,4 glycosidic bonds in the xylan backbone, are important components of the repertoire of enzymes that catalyze plant cell wall degradation. The mechanism by which these enzymes are able to hydrolyze a range of decorated xylans remains unclear. Here we reveal the three-dimensional structure, determined by x-ray crystallography, and the catalytic properties of the Cellvibrio mixtus enzyme Xyn10B (CmXyn10B), the most active GH10 xylanase described to date. The crystal structure of the enzyme in complex with xylopentaose reveals that at the +1 subsite the xylose moiety is sandwiched between hydrophobic residues, which is likely to mediate tighter binding than in other GH10 xylanases. The crystal structure of the xylanase in complex with a range of decorated xylooligosaccharides reveals how this enzyme is able to hydrolyze substituted xylan. Solvent exposure of the O-2 groups of xylose at the +4, +3, +1, and -3 subsites may allow accommodation of the α-1,2-linked 4-O-methyl-d-glucuronic acid side chain in glucuronoxylan at these locations. Furthermore, the uronic acid makes hydrogen bonds and hydrophobic interactions with the enzyme at the +1 subsite, indicating that the sugar decorations in glucuronoxylan are targeted to this proximal aglycone binding site. Accommodation of 3′-linked l-arabinofuranoside decorations is observed in the -2 subsite and could, most likely, be tolerated when bound to xylosides in -3 and +4. A notable feature of the binding mode of decorated substrates is the way in which the subsite specificities are tailored both to prevent the formation of “dead-end” reaction products and to facilitate synergy with the xylan degradation-accessory enzymes such as α-glucuronidase. The data described in this report and in the accompanying paper (Fujimoto, Z., Kaneko, S., Kuno, A., Kobayashi, H., Kusakabe, I., and Mizuno, H. (2004) J. Biol. Chem. 279, 9606-9614) indicate that the complementarity in the binding of decorated substrates between the glycone and aglycone regions appears to be a conserved feature of GH10 xylanases.

The microbial hydrolysis of the plant cell wall into its constituent sugars is one of the major mechanisms by which organic carbon is utilized in the biosphere. The enzymes that catalyze this process are thus of considerable biological and industrial importance. Endo-␤-1,4-xylanases (xylanases), which are one of the key components of the repertoire of enzymes that catalyze plant cell wall degradation, hydrolyze the ␤-1,4 glycosidic bonds linking the xyloside units that comprise the backbone of the polysaccharide xylan (1). Furthermore, the O-2 and/or O-3 of xylosides within the xylan backbone may be decorated with acetyl and arabinofuranosyl units, whereas 4-O-methyl-D-glucuronic acid (4-O-MeGlcA) 1 is exclusively ␣-1,2-linked (1). The extent and nature of such decoration varies between plant species. For complete degradation the arabinofuranosyl and acetyl side chains are removed from the polysaccharide by arabinofuranosidases and acetyl xylan esterases, respectively (1), whereas the ␣-glucuronidases, which release 4-O-MeGlcA, act only on xylooligosaccharides in which the xylose at the nonreducing end is decorated with the uronic acid (2,3).
Xylanases, which are generally located in glycoside hydrolase families (GH) 10 and 11 (CAZy website 2 and Ref. 4), hydrolyze glycosidic bonds by acid base-assisted catalysis via a double displacement mechanism leading to retention of anomeric configuration at the site of cleavage (5). The crystal structures of xylanases show that GH10 enzymes fold into a (␤/␣) 8 -barrel (6,7), whereas family 11 enzymes are ␤-jelly roll proteins (8). Consistent with their "endo" mode of action, the substrate binding cleft of xylanases extends along the length of the proteins and can accommodate from four to seven xylose residues (9,10). Each region that accommodates xylose moieties are known as subsites, which are given a negative or positive number dependent on whether they bind the glycone or aglycone region of the substrate, respectively, with glycosidic bond cleavage occurring between the Ϫ1 and ϩ1 subsites (11). The structures of xylanases in complex with oligosaccharides and inhibitors have revealed detailed information on the interactions of these enzymes with their substrates in the proximal, Ϫ2 and Ϫ1 glycone subsites (12,13). The data show that at the Ϫ1 subsite the O-3 of xylose makes a hydrogen bond with a highly conserved lysine and histidine, whereas O-2 interacts * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The atomic coordinates and structure factors ( with an asparagine and a glutamate (which acts as the catalytic nucleophile) that are invariant in GH10 xylanases. At the Ϫ2 subsite the xylose interacts with residues that are also highly conserved in family 10 xylanases with the O-2 of the sugar hydrogen bonding to a glutamate and tryptophan, the O-3 to an asparagine, and the endocyclic oxygen to a lysine. There is a paucity of information, however, on the mechanism by which the aglycone region of the substrate binding cleft of these enzymes interacts with the xylan backbone, although the three-dimensional structure of Cellvibrio japonicus xylanase 10A (CjXyn10A) in complex with xylopentaose bound to subsites Ϫ1 to ϩ4 has been described (7,15). These studies showed that a highly conserved aromatic residue stacks against the xylose at the ϩ1 subsite, and although hydrophobic stacking interactions at the ϩ3 and ϩ4 subsites were the primary mechanism of protein-substrate recognition, these amino acids are not invariant in GH10 glycoside hydrolases, suggesting that xylan binding in the distal aglycone region of this enzyme family is variable. One of the fundamental differences between the two major families of xylanases is that the GH11 enzymes hydrolyze unsubstituted regions of xylan, whereas the corresponding family 10 glycoside hydrolases are able to attack decorated forms of the polysaccharide (16). The mechanism by which the side chains of decorated xylans are accommodated in the active site of GH10 xylanases is largely unknown. Schmidt et al. (17) solved the structure of a non-natural decorated xylooligosaccharide comprising 1,2-(4-deoxy-␤-L-threo-hex-4enopyranosyluronic acid)-␤-1,4-D-xylotriose. Although the backbone xylose residues were evident at subsites Ϫ1 to Ϫ3, the side chain was not observed, presumably because it was highly disordered.
To probe the structural basis for the capacity of GH10 xylanases to hydrolyze decorated substrates, we have analyzed the biochemical properties of a family 10 enzyme from Cellvibrio mixtus (CmXyn10B) and determined its crystal structure in complex with unsubstituted and decorated substrates at resolutions from 1.7 to 1.4 Å. CmXyn10B comprises an N-terminal signal peptide and a 360-residue GH10 catalytic module; the enzyme does not contain noncatalytic accessory modules typical of plant cell wall-degrading enzymes (18) and is shown to be the most active GH10 xylanase described to date. Its capacity to bind decorated substrates is conferred partly by the exposure of O-2 and O-3 groups at selected glycone and aglycone subsites and partly through productive interactions with the 4-O-MeGlcA side chains. The data provided here and in the accompanying paper by Fujimoto et al. (19) focus on a GH10 xylanase that contains an N-terminal lectin domain (19), reveal complementarity in the binding of decorated substrates between the glycone and aglycone regions of the active site, and demonstrate how these binding modes mediate synergy with the xylan accessory enzyme, ␣-glucuronidase.

EXPERIMENTAL PROCEDURES
Bacterial Strains, Culture Conditions, and Plasmids-The Escherichia coli strain TUNER:pLysS (Novagen) was used in this study. The bacterium was cultured in Luria broth (LB) at 37°C with aeration unless otherwise stated. The plasmid pCF1 is a recombinant of pET21b, which encodes the mature form of CmXyn10B corresponding to residues 11-379. The xylanase gene xyn10B (GenBank TM accession number AF049493) was inserted into the pET vector at the NdeI and XhoI restriction sites such that the translation stop codon was removed and the gene was in-frame with the His 6 tag encoding sequence supplied by the vector (18). Thus recombinant CmXyn10B contains a C-terminal His 6 tag.
Generation of CmXyn10B Mutants-Derivatives of CmXyn10B were generated using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) according to the manufacturer's instructions, using pCF1 as template DNA and the appropriate primers. The complete sequences of the DNA encoding the xylanase mutants were determined using an ABI 377 DNA Sequenator employing T7 forward and reverse primers and the custom sequencing primers 5Ј-AACGGCGCGCGATG-TACAAGTCG-3Ј and 5Ј-CAGCTGGTGAAAGAAGTAGCGC-3Ј to confirm that only the desired mutations had been introduced. During the course of this study errors in the original manual sequencing of xyn10B, which resulted in 12 amino acid changes in the encoded enzyme, were detected and a corrected version of this gene has now been submitted to GenBank TM .
Expression and Purification of CmXyn10B-E. coli strain TUNER: pLysS harboring pCF1 was cultured in LB supplemented with 50 g ml Ϫ1 ampicillin (1000 ml in 2-liter conical baffle flasks) at 37°C and 180 rpm to mid-exponential phase (A 600 nm 0.6). The culture was cooled and incubated at 30°C before expression of CmXyn10B was induced by the addition of isopropyl ␤-thiogalactopyranoside to a final concentration of 0.2 mM. The culture was incubated for a further 5 h. The cells were then harvested by centrifugation at 4500 ϫ g for 10 min at 4°C and resuspended in 1/40th volume 20 mM Tris/HCl buffer, pH 8.0, containing 300 mM NaCl. Cells were lysed by sonication and centrifuged (25,000 ϫ g) for 15 min at 4°C to produce cell-free extract. CmXyn10B was purified from cell-free extract by immobilized metal affinity chromatography as described previously (20) using Talon TM resin (Clontech, Palo Alto, CA). The protein eluted from the matrix was dialyzed against 3 ϫ 500 volumes of 10 mM Tris/HCl buffer, pH 8.0 (Buffer A) and was then applied to a 3-ml DEAE Tris-Acryl column (15 ϫ 50 mm). CmXyn10B was eluted with a linear 0 -500 mM NaCl gradient in Buffer A. The recovered protein was concentrated to ϳ1 ml using a Vivaspin 10-kDa molecular mass cut-off centrifugal concentrator (VIVASCIENCE, Hannover, Germany) and then applied to a High Load TM Superdex TM 75 column (16 ϫ 600 mm; Amersham Biosciences) and eluted in Buffer A containing 150 mM NaCl. The columns were run at 1 ml min Ϫ1 using the Bio-Rad Biologic system. The purity of the protein was evaluated by SDS-PAGE. Protein concentration was determined from the calculated molar extinction coefficient of the enzyme at 280 nm (67,850 M Ϫ1 cm Ϫ1 ).
Preparation of Oligosaccharides-Unsubstituted xylooligosaccharides were purchased from Megazyme International (Bray, County Wicklow, Ireland) and arabinoxylotriose (AX 3 ) in which arabinofuranose is ␣-1,3-linked to the internal xylose moiety in xylotriose was a kind gift from Satoshi Kaneko. Aldotetraouronic acid (MX 3 ) in which 4-O-MeGlcA acid is ␣-1,2-linked to the nonreducing sugar of xylotriose was prepared as follows. Glucuronoxylan (1 g; Megazyme Int.) was digested to completion with 1 mg CmXyn10B in a 5-ml reaction in 50 mM sodium phosphate buffer, pH 7.0, at 37°C for 18 h. MX 3 was purified by Dowex X-100 (Sigma) chromatography in which the bound decorated oligosaccharide was eluted with a 0 -500 mM sodium chloride gradient in 50 mM sodium phosphate buffer, pH 7.0. The MX 3 was then further purified by size exclusion chromatography using a Bio-gel P2 column (2.5 ϫ 100 cm) in which the oligosaccharide was eluted in distilled water at a flow rate of 0.2 ml min Ϫ1 . The identity of the product as MX 3 was verified by digestion with the ␣-glucuronidase GlcA67A from C. japonicus (21) followed by Dionex-HPLC analysis, which revealed xylotriose and 4-O-MeGlcA as the reaction products. Arabinoxylobiose (AX 2 ), in which the arabinofuranose is ␣-1,3-linked to the nonreducing xylose unit in xylobiose, was prepared as follows. Wheat arabinoxylan (1 g; Megazyme Int.) was digested to completion with 1 mg of CmXyn10B in a 5-ml reaction in 50 mM sodium phosphate buffer, pH 7.0, at 37°C for 18 h. AX 2 was purified by size exclusion chromatography as described above. The identity of the product as AX 2 was verified by digestion with the arabinofuranosidase Abf51A from C. japonicus (22) and Dionex-HPLC analysis, which revealed xylobiose and arabinose as the reaction products.
Enzyme Assays-The activity of the GH10 xylanases against aryl ␤-glycosides was determined as described previously (23). Xylanase activity was performed essentially as described by Charnock et al. (23) except that the release of reducing sugar was determined using the Somogyi-Nelson reagent (24). Each assay was performed in triplicate. To evaluate the activity of the two xylanases against xylooligosaccharides, progress curves were carried out as described previously, and the data were used to determine k cat /K m following the equation of Matsui et al. (25). The bond cleavage frequency and k cat /K m data obtained from these experiments were used to calculate the ⌬G of xylose binding at each of the subsites following the method of Suganuma et al. (26).
All data were collected from single crystals at 120 K, over an oscillation range of 135°with a ⌬ of 0.5°. Data for CmXyn10B E262S in complex with xylopentaose were collected in the home laboratory using a MarResearch image plate detector on a Rigaku rotating anode RU-200 x-ray generator with a Cu target operating at 50 kV and 100 mA and focusing x-ray optics (Osmics). Data for CmXyn10B E262S in complex with AX 3 were collected at the Daresbury Synchrotron Radiation Source (SRS) on beamline PX-14.2. Data for CmXyn10B E262S in complex with AX 2 and MX 3 were collected on the European Synchrotron Radiation Facility (ESRF) beamline ID-29. SRS and ESRF data were collected using ADSC Quantum-4 and Quantum-210 charge-coupled device detectors, respectively.
Structure Solution and Refinement-All data were processed and scaled with the HKL suite (27). All other computing used the CCP4 suite (28) unless otherwise stated. All crystals belonged to space group P2 1 2 1 2 1 with approximate unit cell dimensions a ϭ 47 Å, b ϭ 68 Å, and c ϭ 105 Å.
The structure of CmXyn10B E262S in complex with xylopentaose was solved by molecular replacement with the program AMoRe and with data in the resolution range 20 -3.0 Å and an outer radius of Patterson integration of 25 Å. The protein atoms from the catalytic core domain of the C. japonicus Xyn10A (protein data bank 1clx) was used as the search model. AMoRe generated one solution corresponding to one molecule in the asymmetric unit. This gives a V M of 2.0 A 3 D Ϫ1 and a solvent content of 36.6%. Prior to refinement and model building, 5% of the observations were set aside for cross-validation analysis (29) and were used to monitor various refinement strategies such as the weighting of geometrical and temperature factor restraints as well as the insertion of solvent water during maximum likelihood refinement. REFMAC and ARP/wARP (CCP4 suite) were used to build the correct sequence into electron density automatically. Manual corrections of the model using the X-FIT routines of the program QUANTA (Accelrys) were interspersed with cycles of maximum likelihood refinement using REFMAC. Solvent molecules were added automatically using ARP/ wARP and inspected manually prior to deposition.
The refined structure of the CmXyn10B E262S complexed with xylopentaose was used as the starting model for the refinement of the decorated oligosaccharide complexes. The same 5% of observations was maintained as the R free set in each case. Ideal coordinates for stereochemical dictionaries for the substituted xylooligosaccharides were generated using the energy minimization CHARMm function in QUANTA.
Structures were validated using PROCHECK (30). Coordinates have been deposited with the Macromolecular Structures Database (Table I).

RESULTS
Biochemical Properties of CmXyn10B-CmXyn10B was purified to electrophoretic homogeneity, and its biochemical properties were determined (Tables II and III). The enzyme displays catalytic properties typical of a GH10 xylanase. The activity of CmXyn10B against xylooligosaccharides increases with the degree of polymerization (d.p.) of the substrate, although the difference in the hydrolysis rate between xylohexaose and xylopentaose is modest, implying that the sixth binding site (ϩ4 subsite) interacts only weakly with the oligosaccharide substrate. The hydrolysis of xylotetraose exclusively into xylobiose demonstrates that the enzyme contains two subsites on either side of the point of cleavage, but as no xylotriose was generated, the xylanase does not contain a significant Ϫ3 subsite (Fig. 1). The catalytic efficiency of CmXyn10B and its mode of action against different xylooligosaccharides and aryl-␤-xylosides (Table IV) shows that the Ϫ2 and ϩ2 subsites have binding energies of 9.0 and 3.7 kcal mol Ϫ1 , respectively. The maximum binding energy at the ϩ3 is 1.1 kcal mol Ϫ1 , assuming that the increased activity of CmXyn10B against xylopentaose, compared with xylotetraose, is due entirely to binding at this location.
Against both poorly and highly decorated xylans, CmXyn10B initially generates a range of oligosaccharides that progressively become smaller in size as the enzyme reactions progress (data not shown). Against both wheat arabinoxylan and glucuronoxylan, the unsubstituted end products of the reactions comprise xylose and xylobiose. An oligosaccharide that did not co-migrate with linear xylooligosaccharides was also observed in the two reactions, and these products were purified by size exclusion chromatography and treated with a GH51 arabinofuranosidase, a GH67 ␣-glucuronidase, and a GH39 ␤-xylosidase. The oligosaccharide produced from glucuronoxylan was hydrolyzed by the ␣-glucuronidase into xylotriose and 4-O-MeGlcA but was not cleaved by the other two enzymes. The arabinofuranosidase cleaved the saccharide generated from arabinoxylan into xylobiose and arabinose with a stoichiometry of 1:1, whereas the GH67 and GH39 enzymes did not attack the oligosaccharide (data not shown). Thus the major decorated oligosaccharide released from glucuronoxylan by CmXyn10B consists of xylotriose with 4-O-MeGlcA linked ␣-1,2 to the xy- lose at the nonreducing end (the GH67 enzyme removes side chains only from the nonreducing end of xylooligosaccharides; (Ref. 21)), whereas the major substituted product derived from wheat arabinoxylan is an arabinofuranose moiety linked ␣-1,3 to xylobiose (wheat arabinoxylans are decorated with arabinose primarily at the 3Ј position; Ref. 1)). Typical of GH10 enzymes, CmXyn10B hydrolyzes aryl-␤-cellobiosides (10) but displays no significant activity against cellulosic substrates.
Comparison of the Activity of CmXyn10B with Other GH10 Xylanases-CmXyn10B, similar to Cellulomonas fimi Xyn10A (CfXyn10A; formerly known as Cex), displays considerably higher activity against small oligosaccharides compared with other reported GH10 enzymes (Table II). Against xylohexaose, however, CmXyn10B exhibits similar activity to other family 10 xylanases such as CjXyn10A, which likely reflects stronger binding in the distal subsites of the C. japonicus enzyme (Table II). A particularly interesting feature of CmXyn10B is that against both poorly substituted and extensively decorated xylans, the enzyme is considerably more active than all other GH10 enzymes analyzed to date (Table II). These data are rather surprising, as the xylanase displays similar activity against xylohexaose to CjXyn10A, implying that the C. mixtus Xyn10B enzyme contains features that specifically facilitate xylan hydrolysis.
Crystal Structure of CmXyn10B in Complex with Linear Xylooligosaccharides-To investigate the mechanism by which CmXyn10B interacts with substrate, the crystal structure of an inactive mutant of the enzyme in which the catalytic nucleophile had been substituted with serine (E262S) was solved in complex with xylopentaose was by molecular replacement using CjXyn10A as the search model (Fig. 2a). The model refined to a final R-factor of 0.14 and R free of 0.19. The enzyme has a classic (␤/␣) 8 -barrel fold typical of Clan GH-A enzymes, which include GH10 xylanases (6,7). The three-dimensional structure of CmXyn10B and the catalytic domain of the xylanase SoXyn10A described in the accompanying paper (19) are very similar. The catalytic acid-base (Glu-157) and nucleophile (Glu-262) residues are at the ends of ␤-strands 4 and 7, respectively, a feature that is conserved in all Clan GH-A glycoside hydrolases (31). The data in Fig. 2b reveal the electron density for molecules located in both the glycone and aglycone binding sites, that can be modeled as xylotriose and xylotetraose, respectively. The nonreducing sugar of the trisaccharide does not appear to contact CmXyn10B; its conformation is fixed by the intrachain hydrogen bonding network of the xylooligosaccharides.
At the Ϫ2 subsite the O-2 of the xylose moiety hydrogen bonds with the side chains of Glu-67 and Trp-328, and the O-3 interacts with Asn-68. The endocyclic oxygen of the internal xylose makes hydrogen bonds with Lys-71 and Gln-111. These interactions are almost invariant in GH10 xylanases, and substitution of any of these residues with alanine greatly reduces the activity of these enzymes against xylooligosaccharides (23).
Similar to the Ϫ2 subsite, the interactions of the Ϫ1 subsite of CmXyn10B with the reducing sugar of xylotriose are very similar to other GH10 xylanases. The xylose residue is perpendicular to the ring plane of the highly conserved tryptophan Trp-336. The sugar O-1 (in ␣ conformation) interacts with Gln-231 and His-233. O-2 hydrogen bonds with Asn-156 and a water, which provides a link to the nucleophile mutant (Ser-262), whereas the O-3 interacts with Lys-71 and His-104. The activity of CmXyn10B against 4-nitrophenyl-␤-D-cellobioside is ϳ2000 times lower than against the corresponding xylobioside, indicating that the C-5 hydroxymethyl group of glucose cannot easily be accommodated at the Ϫ1 and Ϫ2 subsites. It is well established that accommodation of glucosides at the Ϫ1 subsite demands conformational change in a flanking tryptophan residue (12). In CmXyn10B, the conformation of this residue, Trp-336, is restricted by a network of hydrophobic interactions with Phe-340, Leu-337, and Pro-341 (Fig. 3), explaining why the enzyme displays poor activity against 4-nitrophenyl-␤-Dcellobioside. In CfXyn10A Trp-281, which corresponds to Trp-336 in CmXyn10B, is more flexible, as the indole side chain does not interact with other amino acids in the xylanase (32). This conformational freedom of the aromatic residue provides an explanation of why the C. fimi enzyme is more active against aryl-␤-cellobiosides than the C. mixtus glycoside hydrolase. The poor activity of C. japonicus Xyn10C 3 against 4-nitrophenyl-␤-D-cellobioside compared with CmXyn10B may further reflect structural differences at the Ϫ2 subsite. In the C. japonicus enzyme, Tyr-340 makes a steric clash with the C-5-CH 2 OH of glucose, whereas in the C. mixtus xylanase the equivalent residue, Gln-111, does not restrict the binding of glucose at this subsite.
The ϩ1 subsite interacts with the nonreducing sugar of xylotetraose located in the aglycone region of the substrate binding cleft of CmXyn10B. The xylopyranose moiety is sandwiched between two hydrophobic walls, with Phe-336 and Phe-  The substrates used were as follows: X 3 , xylotriose; X 4 , xylotetraose; X 5 , xylopentaose; X 6 , xylohexaose; PNPX 2 , 4-nitrophenyl-␤-D-xylobioside; PNPX, 4-nitrophenyl-␤-D-xylopyranose; PNPG 2 , 4-nitrophenyl-␤-D-cellobioside; OX, oat spelt xylan; RAX, rye arabinoxylan; WAX, wheat arabinoxylan; GX, glucuronoxylan; -, not determined; ND, no activity detected.  340 on one side and Tyr-200 on the other. In contrast, xylose located at the ϩ1 subsite of the other GH10 xylanases characterized to date "stack" against a single aromatic residue (either phenylalanine or tyrosine), which is equivalent to Tyr-200 (6,7). The hydrophobic interactions between CmXyn10B and both the ␣ and ␤ faces of the sugar located at the ϩ1 subsite may explain why this enzyme displays unusually high activity against small xylooligosaccharides. The binding mode at ϩ1 represents the major difference in the network of interactions in the proximal subsites between the C. mixtus enzyme and other GH10 xylanases. Subsite mapping shows that the enzyme does not display unusually high binding energy at the Ϫ2 subsite (Table IV), and thus interactions at the Ϫ1 and/or ϩ1 subsites are likely to mediate increased catalytic efficiency against small substrates. As the interactions with substrate at the Ϫ1 subsite of CmXyn10B appear to be identical to other GH10 xylanases, we propose that the additional hydrophobic interactions between Phe-336 and Phe-340 and the xylose residue in the ϩ1 subsite are responsible for the increased activity of the C. mixtus xylanase against small substrates; to investigate this hypothesis we have characterized the mutant F340A. The mutant F340A does indeed display greatly reduced activity against xylotriose, xylotetraose, xylohexaose, glucuronoxylan, and 4-nitrophenyl-␤-xylobioside (Tables II and III). It is interesting to note that the mutation causes only a small decrease in activity against 4-nitrophenyl-␤-D-cellobioside, and this may well reflect increased mobility of Trp-336, which must undergo a conformational change to accommodate the C-5 hydroxymethyl group of glucose at the Ϫ1 subsite, as Phe-340 restricts the mobility of the indole side chain of the tryptophan (discussed above). At the ϩ2 and ϩ3 subsites Trp-272 "stacks" against the xylosyl moiety. The equivalent residue in CjXyn10A, Tyr-255, interacts with substrate only at the ϩ3 subsite (10), illustrating the variation in the structure of the aglycone region of the substrate binding cleft in GH10 enzymes. At the ϩ3 subsite, in addition to the stacking interaction with Trp-272, Glu-273 makes a hydrogen bond with the O-2 of the xylose moiety. At the ϩ4 subsite the O-3 of the xylose residue interacts only with the side chain of Ser-270. The variation in substrate recognition in the distal region of the aglycone binding cleft of GH10 xylanases, alluded to above, is further illustrated by the accompanying paper (19), which demonstrates a lack of substrate binding at the ϩ3 subsite of SoXyn10A. The structure of CmXyn10B in complex with xylopentaose bound to both the glycone and aglycone region of the substrate binding cleft provides the first glimpses of the conformation of xylan when bound to all subsites of a GH10 xylanase. Xylotriose bound at subsites Ϫ3 to Ϫ1 is twisted into a 3-fold helical structure (Fig.  2b), consistent with the conformation of xylan as determined by fiber x-ray crystal analysis (33). At subsites ϩ1 to ϩ3 the xylotetraose substrate is also in a 3-fold helical conformation. However, at the ϩ4 subsite (the reducing end) the sugar is at an angle to the rest of the chain (Fig. 2b). This twist in the substrate chain reflects the observation that the distal region of the aglycone binding cleft is considerably more enclosed than other GH10 enzymes. This is due to a large "excursion" at the The glycosides in parentheses refer to the substrates used to determine binding energies at each subsite, and the superscript values define the k cat /K m values for cleaving the specific glycosidic bond (which are numbered from the reducing end; data were obtained from Fig. 1 and Table II), which were used to obtain subsite binding energies as described under ''Experimental Procedures.'' PNPX 2 , 4-nitrophenyl-␤-D-xylobioside; PNPX, 4-nitrophenyl-␤-D-xylopyranose; X 3 , xylotriose; X 4 , xylotetraose; X 5 , xylopentaose. end of strand ␤-7, comprising residues 268 -304, which brings a number of residues into contact with the ϩ2 to ϩ4 subsites, notably Trp-272 which stacks above the xylose ϩ2, Glu-273, which hydrogen bonds to the sugar at ϩ3 and Ser-27, which interacts with the ϩ4 saccharide.
Crystal Structure of CmXyn10B in Complex with Decorated Xylooligosaccharides-To investigate how decorated xylans are accommodated in GH10 xylanases, the crystal structures of the inactive nucleophile mutant, E262S, of CmXyn10B in complex with various decorated xylooligosaccharides were solved. The oligosaccharides comprised 4-O-MeGlcA ␣-1,2 linked to the xylose moiety at the nonreducing end of xylotriose (MX 3 ), arabinofuranose ␣-1,3 linked to the central xylose of xylotriose (AX 3 ), and arabinofuranose ␣-1,3 linked to the nonreducing xylose in xylobiose (AX 2 ). A schematic of how these substrates bind to Xyn10B in the crystal complexes is displayed in Fig. 4.   FIG. 2. Three-dimensional structure of the Xyn10B from C. mixtus. a, the protein schematic is "color-ramped" from the N terminus (blue) to the C terminus (red) and is shown in divergent (wall-eyed) stereo. The single Mg 2ϩ ion is shown as a shaded sphere and the two xylooligosaccharide ligands in ball-and-stick representation. This figure was drawn with MOLSCRIPT (34). b, the electron density of xylopentaose in complex with E262S. The map, which is shown in divergent (wall-eyed) stereo, is a maximum likelihood-weighted 2F obs Ϫ F calc synthesis contoured at 1 (0.45 electrons Å Ϫ3 and 0.47 electrons Å Ϫ3 , respectively), and is drawn with BOBSCRIPT (35).
FIG. 3. The ؊1 subsite of C. mixtus Xyn10B. A possible explanation for the poor tolerance of glucosides in the Ϫ1 subsite comes from the steric restraints "behind" Trp-336, which prevent the movement of this residue necessary to accommodate the C-5-CH 2 OH group of glucosides (14,32). This figure, shown in divergent (wall-eyed) stereo, was drawn with BOBSCRIPT (35). 3 Complex-The structure of CmXyn10B E262S in complex with MX 3 (Fig. 5a) was refined to a final R-factor of 0.16 and R free of 0.20. The 4-O-MeGlcA of the decorated xylooligosaccharide is accommodated in the ϩ1 subsite, which is entirely consistent with the biochemical properties of not only CmXyn10B but other GH10 xylanases, including SoXyn10A described in the accompanying paper (19), as MX 3 is the major substituted product generated from glucuronoxylan by these enzymes (16). No 4-O-MeGlcA can be modeled in the Ϫ3 subsite, as the electron density is diffuse; this is consistent with the observation that it does not interact with the enzyme but is simply accommodated (the O-2 of the xylose moiety at this location is solvent-exposed). It is interesting to note, however, that although the uronic acid also does not interact with the Ϫ3 subsite of SoXyn10A, clear electron density for 4-O-MeGlcA is evident at this subsite (19). In CmXyn10B the electron density of 4-O-MeGlcA at the ϩ1 subsite is clearly defined, indicating that the side chain has a more restricted conformation because of direct interactions with the aglycone region of the active site (Fig. 5a). The hydroxyl group of Tyr-200 and the OD-2 of Asp-161 make hydrogen bonds with the O-2 of 4-O-MeGlcA, and the pyranose sugar ring in the ϩ1 subsite makes hydrophobic interactions with the aromatic ring of Phe-340. In contrast, studies reported in the accompanying paper show that when MX 3 and MX 2 are bound in the aglycone region of the substrate binding cleft of SoXyn10A, the electron density for the 4-O-MeGlcA is not apparent, suggesting that the side chain is disordered (19). The difference between CmXyn10B and SoXyn10A may reflect generally weaker binding of the decorated substrate to the distal aglycone binding sites of the Streptomyces enzyme, as evidenced by the absence of a xylose moiety at the ϩ3 subsite. The tighter binding of substrate at the ϩ1 subsite of CmXyn10B through the additional hydrophobic interactions mediated by Phe-340, both with the backbone xylose and the 4-O-MeGlcA side chain, may well contribute to the clear electron density for the uronic acid.

CmXyn10B-MX
In GH10 xylanases the equivalent residue to Tyr-200 (stacks against the xylose sugar at the ϩ1 subsite; see above) is either tyrosine or phenylalanine. To investigate whether the identity of the aromatic residue at the ϩ1 subsite influences the catalytic activity of CmXyn10B, the biochemical properties of Y200F were investigated. Removal of the hydroxyl group does not influence activity against xylooligosaccharides; however, it does cause a 3-fold reduction in catalytic efficiency against glucuronoxylan (Table II). The crystal structure of CjXyn10B reveals a hydrogen bond between the OH of Tyr-200 and the OD2 of Asp-161, and the resultant restriction in the mobility of the respective side chains may facilitate their capacity to interact with the uronic acid side chain. These results indicate that the tyrosine-phenylalanine "polymorphism" at the ϩ1 subsite of GH10 xylanases may influence the capacity of these enzymes to hydrolyze xylans decorated with 4-O-MeGlcA.
Inspection of CmXyn10B E262S in complex with substrates shows that the O-2 hydroxyls of the xylose moieties at the Ϫ2, Ϫ1, and ϩ2 subsites point directly "into" the enzyme surface (Fig. 6), demonstrating that the enzyme most likely cannot accommodate O-2-substituted sugars in these subsites. In contrast, the O-2 hydroxyls of the xylosides at the ϩ4 and ϩ3 subsites are pointing into solvent. Thus one would predict that these hydroxyl groups, in addition to those in Ϫ3 and ϩ1, may accommodate an O-2 substituent, which is entirely consistent with the data presented in the accompanying paper (19).
CmXyn10B-Arabinoxylooligosaccharide Complexes-The model for CmXyn10B E262S in complex with AX 2 was refined to a final R-factor of 0.15 and R free of 0.19. The structure reveals xylose moieties bound in the Ϫ1 and Ϫ2 subsites, with arabinofuranose ␣-1,3 linked to the nonreducing xylose (Fig.  5b). The electron density shows that the arabinofuranose moiety is statically disordered in two conformations. In one of these conformations only, O-2 hydrogen bonds with Glu-67 (a highly conserved residue in GH10 xylanases); this is the only interaction the arabinofuranose makes with the protein. In the second conformation (which corresponds to the orientation seen in the AX 3 complex below), the arabinofuranose makes no direct in-teractions with the protein. In the aglycone region there is very weak density for a xyloside in ϩ2 and perhaps an arabinofuranoside at ϩ1, but the density is too weak to allow modeling with any confidence. These data are consistent with the accompanying paper, in which the electron density for the arabinofuranosyl side chain was not observed in the SoXyn10A-AX 2 complex (19).
The model for CmXyn10B E262S in complex with AX 3 was refined to a final R-factor of 0.15 and R free of 0. 19. The structure shows xylotriose bound in the Ϫ3 to Ϫ1 and ϩ1 to ϩ3 subsites, with arabinofuranose bound only in the Ϫ2 subsite (Fig. 5c), similar to the SoXyn10A-AX 3 complex described in the accompanying paper (19). The arabinofuranoside moiety is in one discrete ordered conformation and does not interact directly with the protein but hydrogen bonds both to solvent and through a 2.7-Å hydrogen bond from O-3 to the endocyclic O-5 of the Ϫ3 subsite xyloside. Xylotriose bound at subsites ϩ1 to ϩ3 subsites may be a trace contaminant; AX 3 would not be able to bind at these subsites, as steric clashes would prevent decoration of the O-3 of xylose at the ϩ2 subsite (Fig. 5c). It should be noted, however, that as SoXyn10A was used to generate AX 3 , a decorated xylose does appear to bind at the ϩ2 subsite of this enzyme. Indeed, in the crystal structure reported in the accompanying paper (19) the xylose at ϩ2 in the SoXyn10A/AX 3 complex is displaced out of the active site, which would enable the arabinose side chain to be accommodated, and it is likely that such flexible binding of xylose at this subsite may be facilitated by the absence of a ϩ3 subsite. In CmXyn10B the xylose is located in a deeper position, within the ϩ2 subsite, and is likely to display reduced flexibility as adjacent sugars interact at the ϩ1 and ϩ3 subsites. Thus the Cellvibrio enzyme is less likely to accommodate a decorated xylose at ϩ2 and will therefore generate AX 3 less frequently than SoXyn10A, if at all.
Conclusions-CmXyn10B is the most active GH10 xylanase reported to date. The increased catalytic activity displayed by this enzyme is unlikely to reflect interactions in the glycone region of the substrate binding cleft, as the Ϫ1 and Ϫ2 subsites of CmXyn10B are extremely similar to other GH10 enzymes. Although additional hydrophobic interactions at the ϩ1 and ϩ2 subsites may mediate enhanced activity against small oligosaccharides, the reasons for the enhanced activity against xylans are unclear. This report, in conjunction with the accompanying paper (19), reveals the binding mode of decorated xylooligosaccharides to GH10 xylanases. The three-dimensional structure reveals how the uronic acid side chains linked to the O-2 of xylose moieties can be accommodated at the ϩ1 subsite. Solvent accessibility of O-2 in Ϫ3 and ϩ3/ϩ4 suggests that these subsites could also accommodate O-2 decorations. It is significant that this pattern of xylan decoration tolerated by CmXyn10B facilitates synergy with the accessory enzyme ␣-glucuronidase, illustrating how the substrate specificities of these enzymes have co-evolved. Thus, by ensuring that 4-O-MeGlcA can be accommodated at ϩ1 but not ϩ2, CmXyn10B cleaves glucuronoxylan to release oligosaccharides in which the sugar at the nonreducing end is decorated with 4-O-MeGlcA. This is entirely consistent with the mode of action of ␣-glucuronidases, which only remove the uronic acid when it is linked to the nonreducing end of xylooligosaccharides (2,21). If the side chain was accommodated at ϩ2, then a ␤-xylosidase would be required to remove xylose moieties from the nonreducing end before the ␣-glucuronidase could hydrolyze the ␣-1,2linked 4-O-MeGlcA. A GH39 ␤-xylosidase was recently shown not to be capable of removing the nonreducing xylose when linked to a decorated xylose moiety (data not shown); thus it is possible that decorated xylooligosaccharides in which the uronic acid side chain is linked to sugar n ϩ 1 (when n is the sugar at the nonreducing end) may be a "dead end" reaction product that cannot be further attacked by either the xylanases (4-O-MeGlcA cannot be accommodated at Ϫ1 or Ϫ2) or the accessory enzymes. The accommodation of a xylose at the ϩ4 and ϩ1 subsites that is decorated with the uronic acid will generate a reaction product that can be further hydrolyzed, as the 4-O-MeGlcA side chains at n and n ϩ 3 of the xylooligosaccharide can now be located at subsites Ϫ3 and ϩ1. A similar accommodation of decorated substrates by the Streptomyces xylanase is described in the accompanying paper (19). It is clear that the enzymatic consortia that exist to degrade this complex and recalcitrant substrate have co-evolved to facilitate a remarkable "synergy." The subtle accommodation of decorations, honed by the topology of the substrate binding cleft, ensures complementary between glycone and aglycone regions such that the reaction products derived from glucuronoxylan can be further attacked by the xylanase and are the optimal substrates for the accessory enzymes.