Note The atomic coordinates and structure factors of BSX·TAXI-IA (PDB code 2B42) and BSX·rTAXI-IIA (PDB code 3HD8) are deposited in the Protein Data Bank, Research Collaboratory for Structural Bioinformatics, Rutgers University, New Brunswick, NJ, USA (http://www.rcsb.org)
C. M. Courtin, Laboratory of Food Chemistry and Biochemistry and Leuven Food Science and Nutrition Research Centre (LFoRCe), Katholieke Universiteit Leuven, Kasteelpark Arenberg 20 - bus 2463, B-3001 Leuven, Belgium Fax: +32 16 321997 Tel: +32 16 321917 E-mail: firstname.lastname@example.org
Triticum aestivum xylanase inhibitor (TAXI)-type inhibitors are active against microbial xylanases from glycoside hydrolase family 11, but the inhibition strength and the specificity towards different xylanases differ between TAXI isoforms. Mutational and biochemical analyses of TAXI-I, TAXI-IIA and Bacillus subtilis xylanase A showed that inhibition strength and specificity depend on the identity of only a few key residues of inhibitor and xylanase [Fierens K et al. (2005) FEBS J272, 5872–5882; Raedschelders G et al. (2005) Biochem Biophys Res Commun335, 512–522; Sørensen JF & Sibbesen O (2006) Protein Eng Des Sel19, 205–210; Bourgois TM et al. (2007) J Biotechnol130, 95–105]. Crystallographic analysis of the structures of TAXI-IA and TAXI-IIA in complex with glycoside hydrolase family 11 B. subtilis xylanase A now provides a substantial explanation for these observations and a detailed insight into the structural determinants for inhibition strength and specificity. Structures of the xylanase–inhibitor complexes show that inhibition is established by loop interactions with active-site residues and substrate-mimicking contacts in the binding subsites. The interaction of residues Leu292 of TAXI-IA and Pro294 of TAXI-IIA with the −2 glycon subsite of the xylanase is shown to be critical for both inhibition strength and specificity. Also, detailed analysis of the interaction interfaces of the complexes illustrates that the inhibition strength of TAXI is related to the presence of an aspartate or asparagine residue adjacent to the acid/base catalyst of the xylanase, and therefore to the pH optimum of the xylanase. The lower the pH optimum of the xylanase, the stronger will be the interaction between enzyme and inhibitor, and the stronger the resulting inhibition.
Endo-β-1,4-d-xylanases (xylanases, E.C. 18.104.22.168) hydrolyse β-1,4-linkages between the d-xylosyl residues of arabinoxylans in cereal grain cell walls, releasing (arabino)xylo-oligosaccharides of different lengths . Based on sequence similarities and hydrophobic cluster analysis, most xylanases are classified in glycoside hydrolase families (GH) 10 and 11, with only a minority categorized in GH5, 7, 8 and 43 (http://www.cazy.org) . GH11 xylanases have a molecular mass of approximately 20 kDa and display a β-jelly roll structure in which the substrate-binding groove is formed by the concave face of the inner β-sheet. The structure has been likened to a right hand, with a two-β-strand ‘thumb’ forming a lid over the active site. The active site is thus located in the ‘palm’ with two conserved glutamate residues located on either side of the extended open cleft [7,8].
Despite their high structural and sequence similarities, the pH optima of GH11 xylanases vary considerably from acidic values (as low as 2) to alkaline values (as high as 11). The pH-dependent enzymatic catalysis by GH11 xylanases has been well studied. It has been demonstrated that the pH optima of the xylanases are correlated with the nature of the residue adjacent to the acid/base catalyst. In xylanases that function optimally under acidic conditions (pH < 5), this residue is aspartic acid, whereas it is asparagine in those that function optimally under more alkaline conditions (pH ≥ 5) [9–11].
Plants have evolved different classes of proteinaceous inhibitors with the ability to counteract xylanases secreted by phytopathogens. To date, three distinct types of proteinaceous xylanase inhibitors have been isolated from wheat: Triticum aestivum xylanase inhibitor (TAXI) , xylanase-inhibiting protein  and thaumatin-like xylanase inhibitor . These classes of inhibitors show remarkable structural variety leading to different modes and specificities of inhibition. TAXI-type inhibitors inhibit bacterial and fungal xylanases belonging to GH11 . They are inhibitors with a high isoelectric point and occur in two molecular forms: an intact form with a molecular mass of approximately 40 kDa; and a processed form, consisting of two polypeptides of approximately 10 and 30 kDa, held together by one disulfide bond [15,16]. High-resolution 2D electrophoresis in combination with MS/MS analysis has identified large families of isoforms of TAXI-type inhibitors in wheat grain . The amino acid sequences of TAXI-I and TAXI-II isoforms share a high degree of identity (UniProt accession nos: Q8H0K8, Q53IQ2, Q53IQ4 and Q53IQ3), but both types of inhibitors show different inhibition strengths and xylanase-inhibitor specificities. TAXI-I proteins show activity against a broad range of GH11 xylanases (Table 1) such as Bacillus subtilis xylanase A (BSX) and Aspergillus niger xylanase A (ANX), the latter being inhibited to a greater extent than the former [18,19]. TAXI-II proteins have a very high inhibition capacity against BSX, but are distinguished by the lack of activity against some other xylanases, such as ANX [2,18,19]. Two TAXI-I genes (encoding TAXI-IA  and TAXI-IB ) and two TAXI-II genes (encoding TAXI-IIA and TAXI-IIB)  were cloned and recombinantly expressed in Pichia pastoris.
Table 1. Summary of literature data on TAXI-I and TAXI-II activities towards different glycoside hydrolase family 11 xylanases.
a Inhibition activities were determined by measuring residual xylanase activities using a colorimetric Xylazyme AX method with wheat arabinoxylan, at 30 °C and pH 5.0, as described by Gebruers et al. . b–d Inhibition activities were determined by measuring residual xylanase activities using a dinitrosalicylic acid reducing group assay with wheat arabinoxylan, at 42 °C and pH 4.2 (b) , at 30 °C and pH 4.5 (c) , or at 30 °C and pH 5.5 (d) . e +++, very strong inhibition; ++, intermediate inhibition; +, weak inhibition; n.i., not inhibited.
The 3D structure of TAXI-IA has been thoroughly characterized . TAXI-IA consists of a two-β-barrel domain divided by an extended open cleft and displays structural homology with the pepsin-like family of aspartic proteases. The structure of TAXI-IA in complex with ANX (ANX·TAXI-IA) revealed a direct interaction of the inhibitor with the active-site region of the enzyme and further substrate-mimicking contacts with binding subsites filling the whole substrate-docking region . The His374TAXI-IA imidazole ring is located directly between the two catalytic glutamate residues of ANX and makes additional interactions with Asp37ANX, Arg115ANX and Tyr81ANX. In the −1 glycon subsite, contacts are made between Phe375TAXI-IA and Thr376TAXI-IA and the xylanase, while, in subsite −2, Leu292TAXI-IA mimics perfectly the position of a xylose bound in this subsite. On the aglycon side, TAXI-IA interferes with subsites +1 and +2 and prevents access to the aglycon end through steric hindrance. Mutational studies identified amino acids in the active site and in the thumb region of BSX and of TAXI-type inhibitors that are crucial for xylanase-inhibitor interaction [1–4].
Structural information on TAXI isoforms other than TAXI-IA is not available. Crystallographic analysis of TAXI-II, and of its interaction with xylanases, in particular, could provide an explanation for its divergent inhibition specificity and verify the previously described hypothesis that its specificity depends on the identity of only a few residues [2,4]. The structures of TAXI-IA and recombinant TAXI-IIA described here, in complex with GH11 BSX (BSX·TAXI-IA and BSX·rTAXI-IIA, respectively), allowed identification of the structural determinants for the different TAXI-type xylanase inhibition strengths and specificities. Fine-tuned criteria could be deduced for the evaluation of TAXI-type inhibition specificity, with a predictive power on both the inhibitor as well as the enzyme side.
Interaction interface of the BSX·TAXI-IA complex
For the description of TAXI-IA in the complex, the secondary structure elements are denoted as described previously . In the BSX·TAXI-IA complex, five TAXI-IA loop regions (LNiNj, LHdCk, LHfCs, LHhCy and LCzCterm) are responsible for an extensive network of interactions, resulting in a total buried accessible surface area of 1248 Å2 (Fig. 1A). The TAXI-IA loop LCzCterm protrudes between the thumb and the fingers of the xylanase, inducing a displacement of the thumb-like loop compared with an uncomplexed BSX structure [protein data bank (PDB) code 2Z79]  (Fig. 1B). The shortest active site cleft-spanning distance of 5.6 Å (Pro116BSX Cγ to Trp9BSX Nε1) is lengthened to 8.8 Å upon formation of a complex with TAXI-IA. The opening of the substrate-binding cleft is accompanied by side-chain re-arrangements at the basis of the thumb. Re-orientation of the Thr110BSX side-chain, rotated 102° around the χ1-torsion angle, results in the loss of a hydrogen bond with the side-chain of Gln127BSX, which subsequently is involved in a close interaction with the main-chain carbonyl oxygen of Phe375TAXI-IA. A further cascade of conformational changes upon association of TAXI-IA and BSX is observed in the aglycon-binding sites of the xylanase, determined by Tyr174BSX (subsites +1 and +2) and Tyr88BSX (subsite +3) . Driven by the presence of TAXI-IA, the aromatic side-chain of Tyr174BSX is pushed back to be re-oriented parallel to the xylanase surface, stabilized in its newly acquired position by the Asn63BSX side-chain that underwent a similar conformational change. Asn63BSX in turn pushes Tyr88BSX outwards, from pointing into the substrate-binding cleft towards the solvent. As a result of the new orientation of Tyr174BSX, the side-chain of Gln175BSX is no longer stabilized and becomes solvent exposed. The enlarged total buried accessible surface area in the BSX·TAXI-IA complex compared with the ANX·TAXI-IA complex (1248 Å2 versus 992 Å2, ) can mainly be ascribed to the conformational changes in the BSX aglycon subsites as they lead to a better fit with the inhibitor.
Interaction interface of the BSX·rTAXI-IIA complex
TAXI-IA and rTAXI-IIA have a highly similar basic architecture (Fig. 2). Much as for TAXI-IA, the rTAXI-IIA molecule has an overall two-β-barrel domain topology with a six-stranded antiparallel β-sheet that forms the backbone. For reasons of uniformity, the nomenclature denoting the TAXI-IA secondary structure  is used in the description of the rTAXI-IIA structure. Compared with the native TAXI-IA sequence, rTAXI-IIA possesses two extra amino acids at the N-terminus, which is reflected in the numbering. Also, it has six additional amino acids at the C-terminus.
rTAXI-IIA loop regions LNiNj, LHdCk, LHfCs, LHhCy and LCzCterm are involved in an extensive network of interactions with BSX residues in the active-site cleft and the thumb region (Fig. 1C). Binding of rTAXI-IIA results in the burial of an accessible surface area, of 1203 Å2, at the interface. rTAXI-IIA binding induces a partial opening of the BSX hand, with a net lengthening of 3.1 Å of the distance between Pro116BSX Cγ at the tip of the thumb and Trp9BSX Nε1 at the fingers, much as for the BSX·TAXI-IA complex (Fig. 1D). Several re-arrangements take place at the base of the thumb, with the establishment of a close hydrogen bond between main-chain Phe377rTAXI-IIA oxygen and Gln127BSX Nε1 as the main driving force. The position of Tyr174BSX in the aglycon subsites (subsites +1 and +2), however, is different from that in the BSX·TAXI-IA complex. In the BSX·rTAXI-IIA complex Tyr174BSX is highly stabilized through a hydrophobic stacking interaction with Pro375rTAXI-IIA and is therefore found in a different conformation than the uncomplexed xylanase structure and the BSX·TAXI-IA complex. When looking at the residues contributing to the interface area, again the behaviour of Tyr174BSX is most aberrant. Whereas complexation with TAXI-IA results in the burial of 65 Å2 of the Tyr174BSX solvent-accessible surface, upon rTAXI-IIA binding Tyr174BSX is much better stabilized by interaction with Pro375rTAXI-IIA, burying 100 Å2. For Tyr88BSX, the re-orientation and stabilization in its new position are identical to what was observed for BSX·TAXI-IA. Another striking difference with the BSX·TAXI-IA complex is the nature and contribution to the total contact area of Pro294rTAXI-IIA, compared with that of Leu292TAXI-IA (6.8% and 10.1%, respectively).
Structural basis for the inhibition of BSX by TAXI-IA and rTAXI-IIA
In the BSX·TAXI-IA structure the imidazole side-chain of His374TAXI-IA is located directly between the two catalytic glutamate residues of BSX (Fig. 3A). In this position, the Nε2 atom of the imidazole side-chain is highly stabilized through hydrogen-bonded contacts with Glu172BSX Oε2 (2.9 Å), Glu172BSX Oε1 (3.0 Å) and Tyr80BSX Oζ (2.8 Å), while the more positive Nδ1 atom is involved in a weak electrostatic interaction with the negatively charged Glu78BSX Oε2 over a distance of 3.7 Å and in a water-bridged contact with the Pro116BSX main-chain O. Moreover, the main-chain His374TAXI-IA N is tightly bonded to Asn35BSX Nδ2 (2.6 Å) and the main-chain Phe375TAXI-IA O is hydrogen-bonded to Gln127BSX Nε2 (2.7 Å).
To assess the interactions of TAXI-IA with the glycon-binding subsites of BSX, the superimposition of the BSX·TAXI-IA complex with the structure of a catalytically inactive B. subtilis xylanase mutant complexed with xylotriose (PDB code 2QZ3)  was inspected (Fig. 3A*). The His374TAXI-IA Nε2 atom nearly coincides with the xylose C1 atom in subsite −1, and, in subsite −2, five Leu292TAXI-IA atoms (N, Cα, Cβ, Cγ and Cδ1) get close to the atomic positions of C5, O5, C1, C2 and O2 of the xylose in subsite −2. In this way, Leu292TAXI-IA accomplishes an efficient burial of the hydrophobic surface of Trp9BSX, constituting subsite −2, resulting in a tight binding through a significant hydrophobic effect. Furthermore, as a consequence of the conformational changes of Tyr174BSX and Tyr88BSX in the aglycon subsites of BSX, induced upon binding of TAXI-IA, additional interactions can be observed. Contacts between Gln187TAXI-IA main-chain O and Tyr174BSX Oζ (2.9 Å), Gln187TAXI-IA main-chain O and Asn63BSX Nδ2 (3.2 Å), and Gln190TAXI-IA Nε2 and Tyr88BSX Oζ (2.9 Å), further stabilize the complex by induced fit and physically block the binding of substrate in the aglycon subsites (Fig. 4A). Interactions between the thumb region of BSX and TAXI-IA are established through Asp320TAXI-IA Oδ2 and Asp121BSX Oδ2 (3.4 Å), and Glu354TAXI-IA Oε1 and Arg122BSX Nζ2 (2.8 Å). Asp11BSX Oδ2, located in the outer finger region, interacts with Arg371TAXI-IA Nε (3.7 Å).
Although the interactions between the inhibitor key residues His376rTAXI-IIA and Phe377rTAXI-IIA and the xylanase active site are very similar to those of the BSX·TAXI-IA structure, rTAXI-IIA induces a slightly larger distortion of the active-site architecture, reflected in somewhat longer intermolecular distances. The His376rTAXI-IIA imidazole side-chain is hydrogen-bonded with its Nε2 atom to the acid/base catalyst Glu172BSX Oε2 (2.9 Å) and Glu172BSX Oε1 (2.9 Å), while the positive Nδ1 atom points towards the negatively charged nucleophile Glu78BSX Oε2 over a distance of at least 5.2 Å, forming a weak electrostatic interaction (Fig. 3B). Other interactions are nearly invariable with respect to the BSX·TAXI-IA model: a water-bridged contact between His376rTAXI-IIA Nδ and Pro116BSX main-chain O, a hydrogen bond between main-chain His376rTAXI-IIA N and Asn35BSX Oδ2 (2.9 Å), and an interaction between main-chain Phe377rTAXI-IIA O hydrogen-bonded to Gln127BSX Nε2 (3.1 Å).
In the BSX·rTAXI-IIA complex, however, Tyr80BSX is no longer involved in a contact with His376rTAXI-IIA. The superimposition with the structure of the catalytically inactive B. subtilis xylanase mutant complexed with xylotriose (PDB code 2QZ3)  revealed some differences (Fig. 3B*). Whereas Leu292TAXI-IA coincides with the xylose moiety bound in the −2 subsite, in the case of rTAXI-IIA, Pro294rTAXI-IIA is responsible for the substrate mimicry in this subsite. The envelope conformation of Pro294rTAXI-IIA (with N, Cα, Cγ and Cδ coplanar, and Cβ located above this plane) superimposes perfectly on the −2 xylose unit in chair conformation (C1 up, and C2, C3, C5 and O coplanar). This very stable Pro294rTAXI-IIA conformation maximizes the burial of the Trp9BSX side-chain accessible surface. In the aglycon subsites, further inhibitor–enzyme interactions both contribute to complex stabilization and reinforce the occlusion of the substrate-binding positions. Contacts are established between Gln189rTAXI-IIA main-chain O and Asn63BSX Nδ2 (3.1 Å), and between Gln192rTAXI-IIA Nε2 and Tyr88BSX Oζ (2.3 Å) (Fig. 4B). Also, several interactions are made between Arg122BSX in the thumb region and rTAXI-IIA, in particular with Asp322rTAXI-IIA Oδ2 (4.1 Å), Glu356rTAXI-IIA Oε1 (2.8 Å) and Lys317rTAXI-IIA main-chain O (2.8 Å). Finally, interaction is observed between Arg373rTAXI-IIA Nε and Asp11BSX Oδ2 (3.7 Å), similarly to the BSX·TAXI-IA complex.
The strength and specificity of inhibition of TAXI-I- and TAXI-II-type inhibitors differ strongly (Table 1). Analysis of the structures of the BSX·TAXI-IA and BSX·rTAXI-IIA complexes presented here, and of the ANX·TAXI-IA complex described previously (Fig. 3C,C*) , basically reveal the same inhibition mechanism. First, His374/376 completely blocks the active site through intense contacts with the xylanase active site amino acids. Second, parallel to the substrate–enzyme interactions involved in the reaction mechanism of xylanases, the glycon subsites are firmly occupied by strong hydrophobic interactions, perfectly mimicking the natural substrate. Finally, further contacts between TAXI-type inhibitors and xylanase residues constituting the aglycon subsites, prevent the access to the aglycon end through steric hindrance, thus filling the whole substrate-docking region. The above-described interactions of TAXI-type inhibitors with the active site and surrounding regions of the xylanase are in agreement with previously reported results of mutational studies of BSX by Sørensen & Sibbesen  and Bourgois et al. , which are summarized in Table 2. Modification of Glu127BSX in the −1 glycon subsite, involved in a hydrogen-bonding interaction with Phe375TAXI-IA and Phe377rTAXI-IIA, and of Asp11BSX, which interacts with Arg371TAXI-IA and Arg373rTAXI-IIA, resulted in TAXI insensitivity [3,4]. Xylanase mutants, where thumb-region residues Arg122BSX and Asp121BSX, that interact with several residues of TAXI-IA and rTAXI-IIA, were replaced, were less sensitive to TAXI-type inhibitors . The fact that BSX mutants which had decreased inhibitor sensitivities also had decreased enzyme activities [3,4], confirms that TAXI binding is accomplished by substrate mimicry in the active site of the xylanase.
Table 2. Inhibition of BSX mutants by a mixture of TAXI-type inhibitors, as reported by Sørensen & Sibbesen  (A) and by recombinant TAXI-I and TAXI-II, as reported by Bourgois et al.  (B).
a The IC50 value is defined as the half-maximal inhibitory concentration under the conditions of the assay  .
The seemingly minimal disparities between TAXI-IA and rTAXI-IIA, and between the enzyme–inhibitor complexes, suggest that the inhibition strength and specificity of TAXI-IA/TAXI-IIA reside in the subtle difference of only a few amino acid residues. In this study, in-depth analysis of the enzyme–inhibitor complexes allowed identification of two structural features that determine the xylanase–TAXI interaction.
First, based on the structural analysis provided here, the stronger inhibition of ANX than of BSX by TAXI-I, as reported by Gebruers et al. and Fierens et al. [1,19] (Table 1), can be explained as follows. Figure 3A,C shows that the orientation of the His374 side-chain differs between the ANX·TAXI-IA and BSX·TAXI-IA complexes. In contrast to the conformational change observed in TAXI-IA for this His374TAXI-IA upon complexation with ANX , in the BSX·TAXI-IA complex the side-chain has an orientation identical to that in the uncomplexed structure. The basis for the (re)orientation of the imidazole side-chain is found in the mechanism of action of both xylanases. In ANX (or more general: ‘acidic’ xylanases), the side-chain of Asp37ANX has the lowest pKa value of the residues involved in the catalytic action, and hence is negatively charged at the pH optimum . This negative charge is the driving force for the conformational perturbation of His374TAXI-IA upon complexation with the inhibitor. Re-orientation of the histidine allows charge complementarity between the positively charged Nδ1 atom of the imidazole side-chain and the negatively charged Asp37ANX . As a consequence, in the ANX·TAXI-IA complex, the main electrostatic interaction is with the acid/base catalyst, which induces a pH dependency of the inhibition profile. Moreover, the induced fit of TAXI-IA upon complexation with ANX results in a strong salt bridge between the more positively charged Nδ atom of the imidazole side-chain of His374TAXI-IA and the negatively charged Asp37ANX Oδ2 that will substantially contribute to an increased affinity of the inhibitor for the enzyme and complex stabilization. By contrast, in BSX (or ‘alkaline’ xylanases), the pH optimum is not influenced by the asparagine residue adjacent to the acid/base catalyst and, in the complex, the main electrostatic interaction is with the catalytic nucleophile that remains deprotonated throughout a broad pH range. Hence, no conformational changes are needed for TAXI-IA to reach charge compatibility and the pH dependency of the inhibition will be less pronounced. Furthermore, the rather long-distance salt bridge thus formed in the BSX·TAXI-IA complex will not contribute substantially to the affinity and stability of the complex. This could be the basis for the weaker inhibition by TAXI-I of BSX than of ANX. So, one could argue that the lower the pH optimum of the xylanase (i.e. the lower the pKa value of the aspartate residue adjacent to the acid/base catalyst), the more pronounced the induced fit will be, and the stronger the resulting salt-bridge. Thus, the inhibition strength of TAXI-IA seems to depend on the pH optimum of the inhibited xylanase. Earlier results from biochemical testing of TAXI-IA and TAXI-IA His374 mutants, described by Fierens et al. , are in accordance with this conclusion. The lower the pH optimum of the tested xylanase, the more the binding affinity was deleteriously affected by His374 replacement. Binding affinity reduction ranged from a fivefold decrease with BSX to a total lack of interaction with ANX. Moreover, replacement of Asn35BSX with the corresponding Asp37ANX resulted in a BSX mutant with increased TAXI-I sensitivity  (Table 2), validating the above-described theory.
Second, TAXI-II type inhibitors, unlike TAXI-I type inhibitors, do not inhibit ANX. Inhibition of BSX by TAXI-II, by contrast, is stronger than inhibition of this xylanase by TAXI-I . As outlined earlier, comparison of the BSX·TAXI-IA and the BSX·rTAXI-IIA structures shows that the active-site blocking by His374/His376 is relatively well conserved in both complexes. Furthermore, the extra amino acids at the C-terminus of rTAXI-IIA do not directly intervene in xylanase binding, despite the crucial role of the loop LCzCterm in the inhibition interaction. Therefore, to find determinants of the TAXI-IA/TAXI-IIA specificity, a more detailed analysis was performed. The results of this analysis showed discrepancies in the interactions at the −2 (Trp9) BSX-binding subsite. Pro294rTAXI-IIA– as a result of the ring structure – shares more equivalent positions with the xylose −2 sugar ring atoms compared with the Leu292TAXI-IA side-chain atoms and hence accomplishes a mimicry with a higher degree of likeness to the substrate than TAXI-IA, which is also reflected in a slightly better burial of Trp9BSX by Pro294rTAXI-IIA than by the more voluminous Leu292TAXI-IA. This explains the stronger inhibition of BSX by TAXI-IIA than by TAXI-I. Also, ANX has a tyrosine instead of a tryptophan in binding site −2. Pro294rTAXI-IIA is not able to accomplish the same substrate mimicry at the −2 subsite of ANX. These views are in line with previously reported results of affinity tests that were performed by Raedschelders et al.  using engineered rTAXI-IIA and by Bourgois et al.  using engineered BSX. Changing Pro294rTAXI-IIA into leucine, to generate the Leu294/His376 combination present in TAXI-IA, resulted in the ability of rTAXI-IIA to inhibit ANX, while inhibition activity towards BSX fell back to a moderate level. A BSX mutant, where Trp9BSX was exchanged for Tyr10ANX, was no longer inhibited by rTAXI-IIA and displayed a lower TAXI-I sensitivity (Table 3), illustrating the incompatibility between Pro294rTAXI-IIA and Tyr10ANX and the tighter binding between Pro294rTAXI-IIA and Trp9BSX than between Leu294TAXI-IA and Trp9BSX. This confirms the crucial role of Leu294TAXI-IA and Pro294rTAXI-IIA for inhibition specificity.
Table 3. Data collection and refinement statistics of the structures of the BSX·TAXI-IA and BSX·rTAXI-IIA complexes.
a Values in parentheses are for the highest resolution shell. b , where <I(h)> is the mean intensity of symmetry-equivalent reflections. c , where Fo and Fc are the observed and calculated structure factors, respectively. d Root mean square deviations relate to the Engh and Huber parameters.
Resolution limit (Å)a
a =107.89 Å
a =77.35 Å
b =95.33 Å
b =60.30 Å
c =66.31 Å
c =134.19 Å
β = 122.4°
β = 101.48°
Completeness of all data (%)a
Resolution range (Å)
Number of reflections used
Reflections in Rfree set
Number of atoms
Root mean square deviationsd
Bond lengths (Å)
Bond angles (°)
In summary, the first interaction of the inhibitors with the xylanase active site can be identified as the interaction between the residue on position 374 or 376 of TAXI-IA or TAXI-IIA, respectively, and the xylanase amino acid located next to the acid/base catalyst. For the inhibitor, a histidine has been found in all sequences identified so far, with exception of the TAXI-IIB and TAXI-IV sequences (Uniprot accession nos Q53IQ3 and Q5TMB2, respectively) where a glutamine takes position 376. On the xylanase side, the aspartate or asparagine adjacent to the acid/base catalyst determines the pH optimum. This enables us to state that, for the principal xylanase–TAXI interaction, the Asp/His combination results in a higher affinity than the Asn/His combination.
The enzyme–inhibitor contact in the −2 xylanase-binding subsite can be brought back to the residue on positions 292 or 294 of TAXI-IA or TAXI-IIA, respectively, and the xylanase amino acid constituting the −2 subsite. Except for TAXI-IIA (Pro294), the TAXI residue on position 292/294 is a leucine. The nature of the amino acid constituting subsite −2 has been shown to be important for the pH optimum of the xylanase. For ‘acidic’ xylanases, glycon subsite −2 corresponds to a tyrosine, while a tryptophan is found for ‘alkaline’ xylanases . Hence, for the second important xylanase–TAXI interaction, a higher affinity for the Trp/Pro combination than for the Trp/Leu combination is expected. This in turn leads to a much higher affinity than the Tyr/Pro combination.
Although based on only two main interactions, these two criteria nicely rationalize the results of studies performed previously, where inhibition tests were carried out using different native xylanases (Table 1). In spite of the fact that inhibition tests were carried out by different authors under different conditions, for each single xylanase, inhibition by TAXI-I and TAXI-II was tested under the same conditions, allowing comparison. Acidic xylanases, such as ANX, Penicillium funiculosum XynB, Penicillium purpurogenum XynB and Hypocrea jecorina Xyn1, have an aspartate residue adjacent to the acid/base catalyst, and the −2 glycon subsite is formed by a tyrosine. Therefore, the presently defined criteria for the TAXI-inhibition specificity indicate a weak or absent inhibition by TAXI-IIA. When probing these xylanases for their sensitivity against TAXI-type inhibitors, they were indeed less sensitive towards TAXI-inhibition, because they are not, or are only weakly, affected by TAXI-II type inhibitors [18,19,24] (Table 1). For xylanase XynCB1 from Botrytis cinerea, also an acidic xylanase with a pH optimum of 4.5, one would expect a decreased susceptibility for TAXI-II inhibition. Inhibition tests indeed confirm that XynCB1 is inhibited by TAXI-I and not by TAXI-II . Surprisingly, this xylanase contrasts sharply with the other uninhibited xylanases, because, despite its low pH-optimum, the residue next to the acid/base catalyst is an asparagine, and a tryptophan residue constitutes the −2 glycon subsite. Structural analysis of B. cinerea XynCB1 could produce interesting results because additional factors may be involved in the inhibition interaction between this xylanase and TAXI-type inhibitors.
The basic xylanases P. funiculosum XynC, Trichoderma viride xylanase, H. jecorina Xyn2 and BSX are inhibited by TAXI-II type inhibitors [2,18,19,26]. They have an asparagine residue next to the acid/base catalyst in combination with a tryptophan residue in the −2 glycon subsite. An exception is the P. funiculosum xylanase XynC, for which a pH optimum of 5 results from a combination of an aspartate and a tryptophan residue. Both our criteria on the strength and specificity of the inhibition, however, indicate an increased susceptibility of this xylanase for inhibition by TAXI, which is in line with the determined inhibition specificity  (Table 1).
The elucidation of the molecular architecture of complexes of TAXI-IA and TAXI-IIA with xylanases considerably contributes to the understanding of TAXI-type xylanase inhibition. The structures hold key information on the features of TAXI that are indispensable for the inhibitory action. Combined with mutational and biochemical data from previous studies, structural analysis of the xylanase–TAXI complexes provides an integrated view on the inhibition of xylanases by TAXI-type inhibitors.
Materials and methods
Production and purification of xylanases and xylanase inhibitors
TAXI-I (i.e. a mixture of TAXI-IA and TAXI-IB) was purified from wheat whole meal (cv. Soissons) by cation-exchange chromatography and affinity chromatography . The production (in P. pastoris), and purification, of recombinant TAXI-IIA (rTAXI-IIA) were carried out as described by Raedschelders et al. . GH11 BSX was purified from the Grindamyl H640 enzyme preparation (Danisco, Brabrand, Denmark) by cation-exchange chromatography [27,28]. The BSX·TAXI-I and BSX·rTAXI-IIA complexes were prepared by incubation of TAXI-I or rTAXI-IIA with an excess amount of BSX and purified by cation-exchange chromatography, as described by Sansen et al. .
Crystallization of TAXI-I and rTAXI-IIA in complex with BSX
Prior to crystallization trials, the protein solutions were concentrated to approximately 10 mg·ml−1. Crystals of the BSX·TAXI-I complex were grown using the hanging-drop vapor-diffusion method at 277 K, with a reservoir solution containing 0.22 m ammonium sulfate and 25% (w/v) polyethylene glycol 4000 in a sodium acetate buffer (0.1 m, pH 4.6). For the BSX·rTAXI-IIA complex, a fine-tuned condition of 18% (w/v) polyethylene glycol 4000, 0.18 m ammonium sulfate, in 0.1 m sodium acetate buffer pH 4.6, promoted the growth of cube-shaped crystals, suitable for X-ray diffraction data collection. Crystals of complexes were cryoprotected by soaking for 30 s in a drop containing the crystallization condition to which 20% glycerol was added.
Data collection, structure solution and refinement of the BSX·TAXI-IA complex
A high-quality diffraction data set was collected at 100 K using an ADSC Q4R charge-coupled device (CCD) detector at the European Synchrotron Radiation Facility (ESRF, Grenoble, France) on beam line ID14-EH1. Intensity data were indexed and integrated using mosflm  and scaled using scala . The packing density for one inhibitor–enzyme complex molecule in the asymmetric unit of these crystals was 2.6 Å³·Da−1, corresponding to an approximate solvent content of 51.7% . The TAXI-I model (PDB code 1T6E) , together with the Bacillus circulans xylanase structure (PDB code 1C5H) , were used in molecular replacement searches in order to obtain a first model of this protein–protein complex. In two consecutive molecular replacement protocols, the positions of TAXI-I (first) and the xylanase were determined using CNS . Initial rigid-body least-square minimization was followed with cycles of maximum-likelihood refinement, as implemented in REFMAC , refining individual percentage factors after applying a translation, libration and screwrotation (TLS) correction (two TLS groups, i.e. one for each molecule in the asymmetric unit, 20 parameters each), with intermittent manual re-adjustments. Ramachandran statistics indicated that 87.0% of the residues are in the most favored regions and the remaining residues are in the additionally allowed regions. Table 3 lists further data-collection and refinement statistics. Based on well-defined electron density for residues Gly380 and Leu381 it could be concluded that TAXI-IA was present in the complex structure, while TAXI-IB was not. Therefore, the naming ‘TAXI-IA’ was used throughout the manuscript.
Data collection, structure solution and refinement of the BSX·rTAXI-IIA complex
Diffraction data were collected at 100K on a MAR Research CCD area detector (165 mm) using synchrotron radiation at the BW7A beamline (DESY, Hamburg, Germany). Data were processed using mosflm  and scala . According to Matthews  coefficient calculations, the unique and repeating environment in the crystals consisted of two inhibitor–enzyme complex molecules. A packing density of 2.6 Å3·Da−1 and an approximate solvent content of 51.5% were calculated for these crystals. Table 3 lists further data-collection and refinement statistics.
Because of the very high degree of sequence homology between TAXI-IA and rTAXI-IIA (86.4%), on the one hand, and complete sequence identity for BSX, on the other, molecular replacement was the method of choice to obtain preliminary phases for calculating the first BSX·rTAXI-IIA electron density maps. The complete BSX·TAXI-IA model was used as a template for rotation and translation searches in the auto-MR mode of the program molrep . Refinement of the model thus obtained was initiated by rigid-body fitting followed by cycles of maximum-likelihood refinement using REFMAC , with intermittent minor manual re-adjustments. In silico mutations using the molecular visualization program O  of the template molecule TAXI-IA, in order to match the rTAXI-IIA sequence, was performed only when the electron density maps unambiguously indicated to do so. To this end, electron density maps were calculated after the amino acid of interest was mutated to an alanine. Five short rTAXI-IIA portions, invariably turn-regions located at the surface, could not be unequivocally retrieved in the electron density (i.e. residues 43–48, 70–80, 225–228, 264–268 and 336–342). As none of these residues is involved in the interaction with the xylanase, the lack of coordinates for these rTAXI-IIA amino acids did not hamper the protein–protein interface analysis. The same holds true for the residues Arg387rTAXI-IIA–Ser388rTAXI-IIA–Thr389rTAXI-IIA at the C-terminus.
We acknowledge the European Synchrotron Radiation Facility and the EMBL Grenoble Outstation for providing support for measurements at the ESRF under the European Union ‘Improving Human Potential Programme’. Furthermore, we gratefully acknowledge the beam line scientists at EMBL/DESY for assistance and the European Union for support of the work at EMBL Hamburg. The ‘Instituut voor de aanmoediging van Innovatie door Wetenschap en Technologie in Vlaanderen’ (IWT Vlaanderen) (Brussels, Belgium) is thanked for project funding. This study is also part of the Methusalem programme ‘Food for the Future’ at the Katholieke Universiteit Leuven.