GH 11 (glycoside hydrolase family 11) xylanases are predominant enzymes in the hydrolysis of heteroxylan, an abundant structural polysaccharide in the plant cell wall. To gain more insight into the protein–ligand interactions of the glycone as well as the aglycone subsites of these enzymes, catalytically incompetent mutants of the Bacillus subtilis and Aspergillus niger xylanases were crystallized, soaked with xylo-oligosaccharides and subjected to X-ray analysis. For both xylanases, there was clear density for xylose residues in the −1 and −2 subsites. In addition, for the B. subtilis xylanase, there was also density for xylose residues in the −3 and +1 subsite showing the spanning of the −1/+1 subsites. These results, together with the observation that some residues in the aglycone subsites clearly adopt a different conformation upon substrate binding, allowed us to identify the residues important for substrate binding in the aglycone subsites. In addition to substrate binding in the active site of the enzymes, the existence of an unproductive second ligand-binding site located on the surface of both the B. subtilis and A. niger xylanases was observed. This extra binding site may have a function similar to the separate carbohydrate-binding modules of other glycoside hydrolase families.
- Aspergillus niger xylanase
- Bacillus subtilis xylanase
- glycoside hydrolase family 11
- inactive mutant
- substrate-binding subsite
Heteroxylan is a polysaccharide found in the plant cell wall, especially in the secondary wall, where it is the major non-cellulosic polysaccharide . It is composed of a homopolymeric linear backbone of β-1,4-linked D-xylopyranosyl units, which can be replaced with L-arabinofuranosyl, acetyl, glucuronic, 4-O-methylglucuronic and/or p-coumaric residues depending on its source . A consequence of this diversity is the need for a large variety of co-operatively acting enzymes to achieve complete hydrolysis of heteroxylan. These can be found both in the plant, where they play a role in seed germination, and in fungi and bacteria, where they initialize a pathogenic attack. Hydrolysis of the xylan backbone is performed mainly by endo-1,4-β-xylanases (EC 22.214.171.124; xylanases). Most xylanases can be found in GH 10 (glycoside hydrolase family 10) and GH 11, a classification based on amino acid sequence homology of the catalytic domains .
GH 10 xylanases have a (β/α)8-barrel as a catalytic domain and typically contain one or more carbohydrate-binding domains, which increase the effective concentration of the active site on polymeric substrates . The structure of GH 11 xylanases has been described as a partially closed right hand. It consists of only one domain folding into two β-sheets, which are packed against each other, and one α-helix. The two β-sheets are strongly twisted and form a cleft on one side of the protein in which the active site is situated. This cleft is covered by a long loop region, which is called the thumb region and partly closed on one side by the cord, a long irregular loop with a well-defined structure . In contrast with GH 10 xylanases, no carbohydrate-binding modules are present. A consequence of the difference in structure is their difference in substrate specificity. The active site of GH 10 xylanases is a shallow groove, which is reflected in their specificity towards a lower number of unsubstituted consecutive xylose units. In contrast, GH 11 xylanases show higher affinity towards a larger number of unsubstituted consecutive xylose units because of their cleft-shaped active site . Not only between but also within each family, there are differences in substrate specificity [7,8].
The reaction mechanism of both GH 10 and 11 xylanases is a general acid–base mechanism resulting in retention of the anomeric configuration in the product. It involves two acidic amino acid residues, one acting as an acid/base and one as a nucleophile. For xylanases of GH 10 and 11, these two catalytic residues are glutamate residues [9,10].
The active site of GH 11 xylanases contains many aromatic residues that are important for substrate binding. They can make hydrophobic stacking interactions with the sugar residues and their hydroxy groups can form hydrogen bonds . Using the substrate-binding subsite nomenclature proposed by Davies et al. , the subsites are labelled from –n to +n, where –n are the glycone subsites and +n the aglycone subsites. Hydrolysis takes place between the −1 and +1 subsites . For GH 11 xylanases, the glycone subsites have been crystallographically characterized for Bacillus circulans xylanase [13–15], Trichoderma reesei xylanase II , Bacillus agaradhaerens xylanase [17,18] and Chaetomium thermophilum xylanase . Especially the −1 and −2 subsites have been well characterized. In contrast, to date, the characterization of the aglycone subsites was only based on modelling [20,21]. Also, no structural information on sugars spanning the −1 and +1 subsites is available. This is due to the fact that, in native xylanase, any bound xylan chain is immediately hydrolysed and the aglycone part lost due to the low affinity of the aglycone-binding subsites.
Hence, to explore the aglycone subsites and gain more insight into the substrate specificity of different GH 11 xylanases, we followed the strategy used by Wakarchuk et al.  to produce an inactive mutant. The mutation of the acid/base glutamate residue to an alanine residue prevents the first step in the reaction mechanism resulting in the inactivation of the xylanase without distortion of the substrate conformation. In the present study, we have replaced both the acid/base glutamate residue of Bacillus subtilis xylanase, Glu-172, and that of Aspergillus niger xylanase (Swiss-Prot accession number P55328), Glu-170, by an alanine residue. This gives us an example of both an alkaline (B. subtilis) and an acidophilic (A. niger) xylanase . Crystals of these inactive xylanases were soaked with xylo-oligosaccharides and the structures of the complexes were solved by X-ray crystallography, providing insight into the substrate binding of xylanases at both the glycone and aglycone subsites. Interestingly, we found a second, unproductive, ligand-binding site located on the surface of both xylanases, possibly providing a hint towards explaining the difference in activity and/or functionality between different xylanases.
Oligonucleotide primers were purchased from Proligo Primers and Probes. Restriction enzymes were from Roche Diagnostics and Pfu DNA polymerase for PCR was from Fermentas. Escherichia coli strain TOP10F′ from Invitrogen was used for transformation of DNA constructs, except in the case of the pCR4®-TOPO® vector from Invitrogen, for which E. coli TOP10 from Invitrogen was used. E. coli WK6 was used as host strain for heterologous expression with expression vector pQE-Ec . Expression in Pichia pastoris was performed with expression vector pPICZαC (Invitrogen) containing the α-factor signal sequence from Saccharomyces cerevisiae to direct the expressed protein to the medium. The Easy Select Pichia Expression kit (Invitrogen) was used for transformation of competent P. pastoris X33 cells.
Construction of the inactive mutants
For B. subtilis xylanase, the E172A mutant was generated using the megaprimer method  with pQE-En-XynA as a template . In the case of A. niger xylanase, a pCR4®-TOPO® vector containing the A. niger xylanase gene (Swiss-Prot accession number P55328) was used as a template in a new PCR with primers AnExf02 and AnExr02 to remove a remaining part of the signal sequence. The megaprimer was amplified using the forward primers BsXynAf02 and AnExf02 for B. subtilis and A. niger respectively together with reverse primers containing the mutation (BsE172Ar for B. subtilis and AnE170Ar for A. niger) (Table 1). Amplification of the megaprimer product was obtained by 35 cycles of 1 min at 95 °C, 90 s at 58 °C and 2 min at 72 °C, and 1 cycle of 15 min at 72 °C. The PCR products were gel-extracted (QIAquick gel extraction kit; Qiagen), purified (MSB Spin PCRapace kit; Invitek) and used as a megaprimer in a subsequent PCR to complete the coding sequence. For the final PCR, the conditions used were as described by Smith and Klugman . Briefly, 2 μg of megaprimer product was used with 2 ng of template and 1 μM of the flanking primer containing a stop codon: BsXynAr03 and AnExrstop for B. subtilis xylanase and A. niger xylanase respectively. The latter primer was added after 5 cycles of denaturation at 95 °C for 1 min and extension at 72 °C for 3 min. The remaining 35 cycles of the PCR programme were 1 min at 95 °C, 90 s at 52 °C and 2 min at 72 °C, and 1 cycle of 15 min at 72 °C. The resulting product was cloned in a pCR4®-TOPO® vector after gel purification (Qiagen) and addition of 3′ A-overhangs with SuperTaq polymerase (SphaeroQ) (2.5 units, 72 °C and 10 min). The presence of the mutation and the absence of other mutations were verified by DNA sequencing, and the mutant B. subtilis xylanase gene (E172A) was subcloned in the BglII site of the pQE-Ec vector, whereas the mutant A. niger xylanase gene (E170A) was subcloned as a BglII fragment in the BsmBI site of the pPICZαC vector. The ligation mixture was used to transform E. coli TOP10F′ cells. A sequence-verified pPICZαC E170A construct, linearized with PmeI, was used to transform competent P. pastoris X33 cells according to the instructions of the manufacturer. All the primers used are further characterized in Table 1.
Recombinant expression of the B. subtilis E172A xylanase mutant
Protein expression of the E172A xylanase mutant was performed in transformed E. coli WK6 cells according to instructions of the QIAexpress expression system (Qiagen). Cultures were grown in LB (Luria–Bertani) medium supplemented with 100 μg/ml ampicillin and 2% D(+)-glucose (37 °C, shaking at 225 rev./min) until the attenuance (D) at 600 nm reached approx. 0.5. Expression was induced with 0.2 mM IPTG (isopropyl β-D-thiogalactoside) for 16 h at 16 °C. Lysate was prepared according to the manufacturer's instructions (Qiagen). To improve the protein yield, 3 cycles of freeze–thawing (–80 to 25 °C) were included. Protease inhibitor Pefablock SC (VWR International) was added to a final concentration of 1 mM.
The purification of the E172A mutant was carried out in two chromatographic steps. The obtained lysate was dialysed overnight at 4 °C against 25 mM sodium acetate buffer (pH 5.0) and then concentrated on a SP-Sepharose Fast Flow column (GE Healthcare) equilibrated with the same buffer. Bound proteins were eluted using a linear salt gradient of 0–1.0 M sodium chloride. Protein fractions containing E172A were pooled, desalted and dialysed against 25 mM sodium acetate buffer (pH 5.0) containing 0.2 M sodium chloride. The sample was then loaded on to the TAXI (Triticum aestivum xylanase inhibitor) affinity column. Bound proteins were eluted with 0.25 M Tris/HCl (pH 12.0). The eluate was neutralized immediately with 1.0 M acetic acid and dialysed against 25 mM sodium acetate buffer (pH 5.0). After each purification step, protein purity was verified by SDS/PAGE.
Recombinant expression of the A. niger E170A xylanase mutant
A single P. pastoris X33 colony harbouring the E170A encoding gene was used to inoculate buffered minimal glycerol-complex medium (pH 6.0) (10 ml) containing 0.35 M sodium chloride and incubated for 24 h (30 °C; shaking at 250 rev./min). The volume was increased to 500 ml in a 2 litre flask and incubated overnight (30 °C; shaking at 250 rev./min). A pre-induction transition phase was included whereby 0.1 vol. of 10% (v/v) glycerol/13.4% (w/v) yeast nitrogen base was added to the primary culture. The resulting mixture was incubated for an additional 3–5 h until the number of cells/ml reached 0.4×109 to 0.6×109. The cell culture was harvested by centrifugation at 2500 g for 10 min. To induce protein expression, the cells were suspended in a buffered minimal methanol-complex medium (pH 6.0) to 2.5×109 cells/ml and 1% (v/v) methanol was added every 24 h. The induction was performed in baffled flasks over 73 h with vigorous shaking (20 °C, 250 rev./min). The cultures were harvested by centrifugation at 2500 g for 10 min.
The recombinant protein-containing supernatant was dialysed overnight against 0.25 M Tris/HCl buffer (pH 8.5) prior to purification. The mutant protein was purified with anion-exchange chromatography using a Q-Sepharose Fast Flow column equilibrated with the same buffer. Bound proteins were eluted using a linear salt gradient of 0–1.0 M sodium chloride. The protein was then dialysed against 25 mM sodium acetate buffer (pH 5.0).
Crystallization and soaking experiments
Both B. subtilis and A. niger inactive xylanase mutants were concentrated to 10 mg/ml in 25 mM sodium acetate (pH 5.0) and crystallized at 4 °C using the hanging-drop vapour diffusion method. Protein and precipitant solutions were mixed 1:1 (v/v) in the drop and equilibrated against 0.7 ml of precipitant solution (Structure Screens 1 and 2 of Hampton Research). For the B. subtilis E172A xylanase, a crystal appeared after 8 months in 0.2 M ammonium acetate and 0.1 M Tris/HCl buffer (pH 8.5) and 30% (v/v) propan-2-ol (hereafter referred to as XBS1) and after 1 year in 0.1 M imidazole (pH 6.5) and 1.0 M sodium acetate trihydrate (XBS2), whereas for the A. niger E170A xylanase, mutant crystals appeared after 3–4 days in 0.1 M sodium chloride and 0.1 M Hepes buffer (pH 7.5) and 1.6 M ammonium sulfate (XAN1). The crystals were used in soaking experiments with xylotetraose (Megazyme) (for the XBS1 crystals), xylopentaose (Megazyme) (for the XAN1 crystals) or a mixture of arabinoxylo-oligosaccharides with an average degree of polymerization of 5 and an arabinose to xylose ratio of 0.52 (AXOS-5-0.52; for the XBS2 crystals). These AXOS were derived from WPC (Wheat Pentosan Concentrate; Pfeifer & Langen) and were kindly provided by Katrien Swennen (Laboratory of Food Chemistry and Biochemistry, Katholieke Universiteit Leuven). In all cases, a supersaturated solution [30% (w/v) sugar] was used and soaking times ranged from 1 to 5 min. Crystals were then transferred briefly to a cryoprotectant composed of precipitant solution supplemented with 30% glycerol and flash-cooled and stored under liquid nitrogen before data collection.
Data collection, structure solution and refinement
All X-ray diffraction data were collected at 100 K at the BW7a beam line of DESY (Deutsches Elektronen Synchrotron), EMBL-Hamburg, Hamburg, Germany. The diffraction images were visualized by using XDisplayF, processed by using DENZO and scaled and merged by using SCALEPACK from the HKL suite of programs . All further computing used the CCP4 suite  unless otherwise stated. The structures were solved by molecular replacement using the program Phaser  for the XBS1 and XAN1 data, whereas Molrep  was used for the XBS2 data. The B. circulans xylanase with PDB entry 1BCX  and the A. niger xylanase with PDB entry 1UKR  served as models for the B. subtilis and A. niger xylanase structures respectively. Cycles of refinement and model building were performed using Refmac5  and Coot  respectively. After several cycles of refinement, residual positive density in the Fo–Fc and 2Fo–Fc electron density maps revealed the presence of several bound sugar molecules, which were included in the models, followed by further refinement and manual addition of water molecules. The final structures were evaluated using SfCHECK  and rmsd (root mean square deviation) values were calculated using the program SuperPose . All data collection and refinement statistics are shown in Table 2. The Figures were drawn using the program PyMOL (DeLano Scientific; http://pymol.sourceforge.net/).
Overall structure of the soaked inactive xylanases
Insight into the binding of the substrate in the active site of enzymes is of prime importance in understanding the substrate specificity and related biochemical parameters. However, for GH 11 xylanases, only the binding of xylose residues in the glycone subsites of the active site has been crystallographically demonstrated. To explore the aglycone subsites, we produced inactive mutants of the B. subtilis and A. niger xylanases. This was done by mutating the acid/base glutamate residue to an alanine residue and thus preventing the first step in the reaction mechanism. For the B. subtilis xylanase, this resulted in the E172A mutant, and for A. niger xylanase, in the E170A mutant.
For the B. subtilis xylanase mutant, crystals appeared under two crystallization conditions and are further referred to as XBS1 and XBS2. XBS1 crystals were soaked with xylotetraose, while XBS2 crystals were soaked with a mixture of arabinoxylo-oligosaccharides with an average degree of polymerization of 5 and an average arabinose to xylose ratio of 0.52 (AXOS-5-0.52). For the A. niger xylanase mutant, one type of crystal could be grown (XAN1), which was soaked with xylopentaose. All structures could be solved by molecular replacement. Inspection of the electron density maps revealed clear density for xylose residues in the active-site subsites of the XBS1 and XAN1 structures (Figures 1a–1c). However, no extra interpretable density could be observed in the active site for the XBS2 structure, indicating that the soak with AXOS-5-0.52 was not successful. This structure is further used as reference, i.e. ligand-free structure. For both B. subtilis xylanase structures, the N-terminal amino acid could not be modelled because of the lack of good electron density, suggesting that this region is disordered.
Electron densities corresponding to the bound xylo-oligosaccharides were not only found in the catalytic cleft. Instead, the electron density map also revealed a well-defined density corresponding to a xylo-oligosaccharide on the surface of the protein, for both the B. subtilis and A. niger xylanases.
The thumb region
The overall structure of the ligand-bound xylanases is very similar to those of ligand-free structures, as observed earlier for other species. When comparing both XBS1 and XBS2 with PDB entry 2B46 , a B. subtilis xylanase not susceptible to TAXI inhibition in complex with xylobiose, the rmsd for the Cα atoms is 0.7 and 0.5 Å respectively (1 Å=0.1 nm). A superposition of the ligand-bound XBS1 and PDB entry 2B46 structures, on the one hand, and the unbound XBS2 structure, on the other hand, reveals no displacement of the thumb region as can be deduced from the distance between the Cα atoms of AspBS-119 and AsnBS-35. This distance is 16.3 Å for XBS1 and 17.1 Å for XBS2 and 2B46. This is in contrast with earlier studies [16,37], which state that xylanases adopt a closed conformation upon ligand binding. The conformation of the enzyme observed here is, however, necessary, since in the closed conformation a xylose residue in the −3 subsite would clash with the tip of the thumb region.
For A. niger xylanase XAN1 compared with chain A of the ligand-free wild-type structure (PDB entry 1UKR) , the rmsd is 0.3 Å for the Cα atoms. Also, for the XAN1 structure, no structural differences in the thumb region were observed. The distance between the Cα atoms of ProAN-119 and AspAN-37 is 11.3 Å for the ligand-free structure and 10.8 Å for the ligand-bound structure. This is in contrast with the opinion of Tahir et al.  that the presence of a smaller thumb region (as in XAN) should not play a role in the motility of the thumb region upon substrate binding, i.e. that a smaller thumb region has an open–closed movement upon substrate binding.
Xylo-oligosaccharide bound in the active site
The XBS1 crystals contained two protein molecules in the asymmetric unit, named A (XBS1A) and B (XBS1B). Both monomers bind xylotetraose, but for XBS1A, xylose residues were observed in the −3, −2 and −1 subsites, whereas for XBS1B, they were observed in the −2, −1 and +1 subsites. As can be seen in Figures 1(a) and 1(b), the xylose co-ordination was almost identical in the two monomers for the −2 and −1 subsites. The XAN1 crystal contained one monomer in the asymmetric unit. For this structure, only in the −2 and −1 subsite a xylose residue could be fitted in the electron density. The missing xylose residues probably extend beyond the active-site cleft and are too mobile to be seen in the electron density. Table 3 and Figure 2 give an overview of all the interactions formed between the xylanases and its substrate. In what follows, the different subsites will be discussed in detail, starting with the −2 and −1 subsites, which are characterized best, followed by the −3 and +1 subsites, for which crystallographic data are presented for the first time.
The −2 subsite
At the −2 subsite, for both the B. subtilis (Figure 2a) and A. niger (Figure 2b) xylanases, an aromatic ring (TrpBS-9 and TyrAN-10) stacks against the xylopyranose ring which is in the chair conformation. Also, the hydrogen-bond patterns are very similar for both xylanases. Two tyrosine residues (TyrBS-69–TyrAN-70 and TyrBS-166–TyrAN-164) are in a similar position and make hydrogen bonds with the OH2 and OH3 groups of the xylose. The length of one of these bonds (the hydrogen bond between TyrBS-166 and the OH2 group of xylose) differs by 0.4 Å between XBS1A and XBS1B. This makes the bond in the XBS1B a stronger one. A glutamine residue (GlnBS-7 and GlnAN-8) forms a weaker hydrogen bond with the OH3 group of xylose, while the main-chain atoms of the serine and proline residues at the tip of the thumb region form weaker hydrogen bonds with the endocyclic O and OH1 of the xylose.
The −1 subsite
In the −1 subsite, the interactions between the sugar and both the ProBS-116/ProAN-119 and the nucleophilic GluBS-78/GluAN-79 are analogous. The carbonyl group of the proline residue is hydrogen-bonded to the OH3 group of xylose, and the glutamate residue forms two hydrogen bonds with the OH2 of xylose. In A. niger xylanase, AspAN-37 (which is responsible for its acidic character) is hydrogen-bonded to the endocyclic oxygen and the OH1 group, while the analogous AsnBS-35 (making this xylanase alkaline) forms only one hydrogen bond with OH1. This residue occupies a slightly different position than in the structures without the bound ligand. For both the B. subtilis and A. niger xylanases, the side chains of AsnBS-35 and AspAN-37 are turned by 90° in the absence of ligand. For the B. subtilis xylanase, this makes the distance between the residue and OH1 longer (from 3.3 to 4.3 Å) and, consequently, impairs the formation of a hydrogen bond. For the A. niger xylanase, this rotation does not seem to cause any differences in the length of the hydrogen bonds between AspAN-37 and the xylose residue. In A. niger xylanase, TyrAN-81 and GlnAN-129 are also hydrogen-bonded to the xylose. In addition, the latter residue also forms a 2.8 Å hydrogen bond with Oϵ of the nucleophile. In the B. subtilis xylanase, ArgBS-112 makes several hydrogen bonds with the OH2 and OH3 groups of the xylose, while, in A. niger xylanase, the analogous residue (ArgAN-115) is turned away from the xylose. All these interactions make the xylose adopt a skew boat conformation in the B. subtilis xylanase (Figure 2c). Superposition with the native xylanase gives insight into the position and interactions of the acid/base glutamate residue. To this end, the A monomer of XBS1 was superposed with 2B46  (Figure 2d) and XAN1 was superposed with the A monomer of 1UKR  (Figure 2d). For both xylanases, Oϵ of GluAN-170/GluBS-172 forms a strong hydrogen bond with the xylose, which creates an ideal situation to initiate the reaction.
The −3 subsite
From the bound structure, subsite −3, which is only observed in XBS1A (Figure 2e), seems to have a weak binding energy, which probably reflects a less significant subsite. While subsites −2 and −1 show many interactions indicating tight binding, the xylose moiety of the −3 subsite forms only one weak hydrogen bond (the OH4 group of xylose with the carbonyl group of IleBS-118). This interaction causes the xylose to adopt a boat conformation. To prevent the tip of the thumb region from clashing with the bound xylose residue, the thumb region cannot adopt the closed conformation as proposed in earlier studies [16,37].
The +1 subsite
XBS1B had a clear electron density corresponding to a xylose in the +1 subsite (Figure 2f), which makes this the first observation of a xylose residue in this subsite for GH 11 xylanases, showing substrate spanning the −1/+1 subsites. The xylose moiety has a twist boat conformation, which is mainly due to the hydrogen bonds between TyrBS-80 and the OH1 and OH3 groups of the xylose. An additional hydrogen bond is formed between the OH2 of xylose and the amide group of GlyBS-173. For the xylose to bind, TyrBS-174 must flip away from the substrate. Furthermore, TyrBS-88, further down the aglycone subsites, is also flipped. Superposing the XBS1B with the native xylanase (2B46)  revealed a clash between the xylose in the +1 subsite and the acid/base glutamate residue. In this structure, the latter is in the down position. To visualize the interactions between the xylose and the acid/base glutamate residue when this residue is in the up position, a superposition was made with the Trichoderma reesei xylanase II structure 1XYO from crystals grown at pH 4.5 [39,40]. This structure revealed no clashes between the xylose in the +1 subsite and the acid/base glutamate residue, indicating that this residue has to be in the up position for the enzyme to bind substrate.
Xylo-oligosaccharide bound to the surface
For both the B. subtilis and A. niger xylanases, there was a well-defined electron density corresponding to a xylo-oligosaccharide at the surface of the protein. For XBS1A and XBS1B, three xylose residues could be build in the density, whereas, for the A. niger structure, there was density for four xylose residues. In both structures, the xylo-oligosaccharide is bound to the finger region of the xylanase, but in a different position (Figure 3). For the A. niger xylanase, it is bound to the tip of the finger region between β-strands A2 and A3, while for the B. subtilis xylanase it is found at the knuckles of the finger region between β-strands A4 and A5. For convenience, the binding subsites at the surface were numbered I, II and III starting from the reducing end of the xylo-oligosaccharide. Comparison between the xylanase structures shows that the direction in which the xylan is bound to the surface is the same: the reducing end is near the tip of the fingers, while the non-reducing end is near the knuckles. Also, the binding of xylan to the surface does not seem to cause structural differences.
For B. subtilis xylanase, there seem to be three subsites at the surface to which xylose residues bind in their chair conformation. This is most obvious in XBS1A (Figure 4a), where a xylose residue is found in all three subsites. The IBS subsite is formed by AsnBS-54, GlyBS-56, AsnBS-181 and ThrBS-183, which form hydrogen bonds to the xylose residue, whereas in the IIBS subsite, AsnBS-54 and AsnBS-141 form the hydrogen bonds to the xylose. Table 4 gives an overview of all the interactions. The existence of the IIIBS subsite could be deduced from the electron density of a third xylose in XBS1A. This xylose makes a hydrophobic stacking interaction with TrpBS-185. A. niger xylanase has a serine residue at this position, making this interaction impossible. XBS1B (Figure 4b) contains an extra xylose residue beyond the IBS subsite, which does not interact with the protein, but has clear electron density and B-factors analogous to the other xylose residues bound to the surface. Comparison of this region of B. subtilis with the A. niger xylanase shows some structural differences. GlyBS-56 and AsnBS-141, which make important hydrogen bonds, both constitute insertions compared with A. niger xylanase.
For the A. niger structure, the presence of four xylose residues could be deduced from the electron density data (Figure 4c). The xylose residue in the IAN subsite interacts with the carboxy groups of AsnAN-13 and GlyAN-15, whereas in the IIAN subsite, GluAN-31 and AspAN-32 form the interaction partners. GluAN-31 is, together with AspAN-16, also important in the IIIAN subsite. In analogy with B. subtilis xylanase, all xylose residues adopt the chair conformation. Table 4 lists details on these interactions. In this region too, there are structural differences between the B. subtilis and A. niger xylanases. The β-strand A2 in A. niger is much shorter compared with the analogous β-strand in B. subtilis. Hence, there is no corresponding residue for AsnAN-13, which is in the loop region at the end of β-strand A2, in B. subtilis. In analogy with the III subsite of B. subtilis xylanase, the xylose in the IIIAN subsite makes a hydrophobic stacking interaction, but with a tyrosine residue, TyrAN-29, instead of a tryptophan residue. B. subtilis xylanase does not have an aromatic residue at the corresponding position. The fourth residue is situated beyond subsite IIIAN, but does not form any interactions with the protein.
For the first time, crystallographic data of xylo-oligosaccharides bound to incompetent B. subtilis and A. niger xylanases are obtained. The data show the substrate spanning the −1/+1 subsites and revealed an extra binding site at the surface.
The xylose residue in the −1 subsite of B. subtilis xylanase seems to be slightly distorted towards a β-2S0 skew conformation (Figure 2c). This is one of the two (5S1 and 2S0) conformations flanking the 2,5B conformation (which has been observed earlier [14,17]) in the skew-boat pseudorotational series and provides support for the Deslongchamps-type itinerary for the glycosylation step of GH 11 (β-2S0→E3-TS→α-4C1 glycosyl enzyme intermediate) postulated by Nerinckx et al. . The proposed deglycosylation step would then go through a re-entry into the skew-boat pseudorotational series: α-5S1→E4-TS→β-1C4 .
For both the B. subtilis and A. niger xylanases, superposition of substrate-bound and -unbound structures does not show the expected difference between the open and closed conformations observed previously for Trichoderma reesei xylanase [16,37]. This may indicate that this difference in conformation is species-dependent and, therefore, is not a general property of GH 11 xylanases. For A. niger, this difference is probably due to the shorter thumb region, whereas for B. subtilis, it is due to the −3 subsite, as the xylose residue at this subsite would clash with the tip of the thumb region if the xylanase adopts the closed conformation upon substrate binding.
For A. niger xylanase, the main end-products from the hydrolysis of xylo-oligosaccharides are xylobiose and xylotriose , whereas for B. subtilis, there is no information on the hydrolysis of xylo-oligosaccharides. As can be seen from the superposition of XBS and XAN, there is little difference in substrate binding in the glycone subsites. This may point out that differences between hydrolysing products would be due to the aglycone subsites or to subtle changes at the glycone subsites. This needs further analysing.
Inspection of the different subsites may further explain the failed attempt to soak XBS2 with AXOS-5-0.52. It was seen that for the −2, −1 and +1 subsites, no arabinose substitution on the xylan backbone is acceptable because of steric hindrance: the tip of the thumb region makes the active site cleft nearly tunnel-shaped near the −2 and −1 subsites. Hence, AXOS-5-0.52 with an average arabinose to xylose ratio of 0.52 probably has too many substitutions to allow for binding.
For both the B. subtilis and A. niger xylanases, a xylo-oligosaccharide was found to be bound to the surface. This is also seen for a family 8 xylanase from Pseudoaltermonas haloplanktis having an (α/α)6 fold . Also within GH 11, this was already observed for an A. niger xylanase  and a B. circulans xylanase . These results suggest that this extra binding site has a physiological meaning and is not only due to the high concentrations during the soaking experiment. A possible function, previously suggested by Törrönen et al. , may be one similar to the separate carbohydrate-binding domains of other glycoside hydrolase families, i.e. increasing the effective concentration of the active site on polymeric substrate. This was recently confirmed by Ludwiczek et al. , who revealed the presence of a xylan-specific secondary binding site at the surface of B. circulans xylanase using NMR-monitored titrations. They proposed that the active site and the secondary binding site function co-operatively to enhance the activity towards longer substrates .
Superposition of the subsites at the surface shows large differences between the xylanases of B. subtilis and A. niger in the position where the xylose residues bind. This corresponds to the assumption of Ludwiczek et al. , who suggest that different regions of the serine/threonine-rich surface in different xylanases also function as secondary binding sites. Not only the location of the secondary binding site differs, but also the position where arabinose substitution is possible varies. For B. subtilis xylanase, in the IIBS and IIIBS subsites, arabinose substitution is possible. For A. niger xylanase, in the IAN subsite, an arabinose is possible only on the OH3 of the xylose residue. In the IIAN subsite, there is no possibility for arabinose substitution, and, in the IIIAN subsite, both OH2 and/or OH3 can be replaced. These differences may suggest that the binding of xylan to the surface may play a role in substrate specificity.
Earlier mutation studies of Moers et al.  suggested a role in substrate selectivity for some surface-exposed aromatic residues in B. subtilis xylanase. Mutation of TrpBS-185 to an alanine residue revealed a significantly lower selectivity of the enzyme towards water-unextractable arabinoxylans . The binding site at the surface observed here strongly supports the proposed role of this region as playing a role in substrate selectivity and specificity, although further mutation analysis should be performed.
We thank the staff of the EMBL/DESY Hamburg Outstation for the provision of synchrotron facilities and skilful technical assistance and financial support through the I3 contract with the European Commission for support of access for external users. We also thank Dr Wim Nerinckx (Universiteit Gent, Gent, Belgium) for the discussion on sugar conformations. This work was funded by the Flemish IWT (Instituut voor de aanmoediging van Innovatie door Wetenschap en Technologie in Vlaanderen; funding of the SBO project Impaxos and the scholarship to S. R.), the Flemish FWO (Fonds voor Wetenschappelijk Onderzoek Vlaanderen; post-doctoral fellowship to A. R.) and the ‘Bijzonder Onderzoeksfonds K. U. Leuven’ (postdoctoral fellowship to S. V. C.).
Abbreviations: GH 11, glycoside hydrolase family 11; rmsd, root mean square deviation; TAXI, Triticum aestivum xylanase inhibitor
- © The Authors Journal compilation © 2008 Biochemical Society