Wheat endoxylanase inhibitor TAXI-I inhibits microbial glycoside hydrolase family 11 endoxylanases. Crystallographic data of an Aspergillus niger endoxylanase-TAXI-I complex showed His374 of TAXI-I to be a key residue in endoxylanase inhibition [Sansen S, De Ranter CJ, Gebruers K, Brijs K, Courtin CM, Delcour JA & Rabijns A (2004) J Biol Chem 279, 36022–36028]. Its role in enzyme–inhibitor interaction was further investigated by site-directed mutagenesis of His374 into alanine, glutamine or lysine. Binding kinetics and affinities of the molecular interactions between A. niger, Bacillus subtilis, Trichoderma longibrachiatumendoxylanases and wild-type TAXI-I and TAXI-I His374 mutants were determined by surface plasmon resonance analysis. Enzyme–inhibitor binding was in accordance with a simple 1 : 1 binding model. Association and dissociation rate constants of wild-type TAXI-I towards the endoxylanases were in the range between 1.96 and 36.1 × 104m−1·s−1 and 0.72–3.60 × 10−4·s−1, respectively, resulting in equilibrium dissociation constants in the low nanomolar range. Mutation of TAXI-I His374 to a variable degree reduced the inhibition capacity of the inhibitor mainly due to higher complex dissociation rate constants (three- to 80-fold increase). The association rate constants were affected to a smaller extent (up to eightfold decrease). Substitution of TAXI-I His374 therefore strongly affects the affinity of the inhibitor for the enzymes. In addition, the results show that His374 plays a critical role in the stabilization of the endoxylanase–TAXI-I complex rather than in the docking of inhibitor onto enzyme.
Endoxylanases (EC 184.108.40.206) are key plant or microbial enzymes in the degradation of arabinoxylan (AX) [1,2], an important structural and quality determining nonstarch polysaccharide in cereals. In sequence-based classifications, endoxylanases are mainly grouped into glycoside hydrolase families (GH) 10 and 11 (CAZy database http://afmb.cnrs-mrs.fr/CAZY/) . The catalytic domain of GH11 endoxylanases has a β-jelly roll fold which resembles the shape of a partially closed right hand with ‘finger’, ‘palm’ and ‘thumb’ regions . Its active site is located into the extended open cleft formed by the ‘palm’ region. GH10 endoxylanases, comprising all endogenous plant endoxylanases known to date , have the typical (β/α)8 barrel fold . The active site of both endoxylanase families contains two conserved Glu residues that are involved in substrate hydrolysis via a double displacement mechanism [7,8].
Microbial endoxylanases belonging to both GH10 and GH11 are often used to impact AX functionality in cereal-based biotechnological applications to optimize process parameters, yields and product quality [9–11]. The presence of cereal endoxylanase inhibitors, however, impacts the activity of the added endoxylanases. Two structurally different types of proteinaceous endoxylanase inhibitors have been identified in wheat, i.e. the TAXI- (Triticum aestivum xylanase inhibitor) [12,13] and XIP- (Xylanase inhibiting protein) [13,14] type inhibitors. Two TAXI-type endoxylanase inhibitors have been described, TAXI-I and TAXI-II, both with molecular mass of 40 kDa but differing from one another in pI (8.8 and 9.3, respectively) and endoxylanase specificity . Despite very low sequence homology levels (≈ 15% identity), TAXI-I is structurally homologous with the pepsin-like family of aspartic proteases . It folds as a two β-barrel domain protein with a few helical segments, and the separate domains are divided by an extended cleft. XIP-I, on the other hand, is a glycosylated, monomeric inhibitor with a molecular mass of 29 kDa and a basic pI (8.7–8.9) [14,17]. The inhibitor possesses a (β/α)8 barrel fold and displays structural features typical for GH18 chitinases, but lacks chitinase activity . Sequence data and 3D structures of TAXI-I and XIP-I show no structural homology between both types of inhibitors [16,18–20]. Moreover, these inhibitors have different endoxylanase specificities. XIP-I inhibits microbial endoxylanases from GH10 as well as GH11 , while TAXI-I and TAXI-II only inhibit microbial endoxylanases belonging to GH11 . Experimentally determined Ki values ranging from 5 to 30 nm for TAXI [12,22,23] and from 3 to 610 nm for XIP  suggest tight binding inhibition for both inhibitor types.
The 3D structure of TAXI-I complexed with a GH11 Aspergillus niger endoxylanase revealed both a direct interaction of the inhibitor with the active site region of the enzyme as well as substrate-mimicking contacts filling the whole substrate-docking region . Five TAXI-I loop regions completely cover the deep substrate-binding and active site cleft of the endoxylanase through ionic and hydrophobic interactions and hydrogen bonding (including water-bridged contacts) (Fig. 1A), resulting in the burial of 992 Å2 of accessible surface area. The imidazole ring of TAXI-I His374 that fits into the active site of the enzyme and, more precisely, in between the two catalytic Glu residues involved in substrate hydrolysis is the key residue for endoxylanase inhibition (Fig. 1B). In the −2 glycon subsite of the endoxylanase, Leu292 of TAXI-I perfectly superimposes with xylose.
In spite of the above, profound analysis of the specific contributions of the amino acid residues at the interface of the enzyme–inhibitor interaction is lacking. Mutagenesis studies of a GH11 Bacillus subtilis endoxylanase revealed that both the ‘thumb’ and ‘finger’ regions of the enzyme are important for interaction with TAXI-type inhibitors . Tahir and coworkers  showed that mutation of the pH-optimum related Asp37 in the active site cleft of a GH11 A. niger endoxylanase into Ala completely abolishes interaction with TAXI-I.
Mutational analysis of TAXI-I rather than of its target enzymes is another approach to study the structural requirements for formation of the enzyme–inhibitor complex and for identifying the nature of forces involved in its stabilization. In the present case, His374 of TAXI-I is a critical residue for endoxylanase inhibition because of its many interactions with several endoxylanase residues . Hence, we investigated its role in the endoxylanase–inhibitor interaction by site-directed mutagenesis. Residual endoxylanase inhibition activities of the TAXI His374 mutants were determined. The specific contribution of TAXI-I His374 in endoxylanase interaction was examined using surface plasmon resonance (SPR) analysis. The need for interface charges and the pH stability of the enzyme–inhibitor interaction were studied by IEF titration curves.
Results and Discussion
Recombinant expression, purification and characterization of TAXI[H374A/K/Q] mutants
The role of TAXI-I His374 in endoxylanase inhibition was studied by site-directed mutagenesis of this residue into Ala (A), Gln (Q) or Lys (K). Ala is mostly used for amino acid substitutions because of its small size and aliphatic properties. Crystallographic analysis of the A. niger endoxylanase–TAXI-I complex showed involvement of the two nitrogen atoms of the His374 imidazole ring of TAXI-I in endoxylanase interaction . Moreover, molecular identification of TAXI-II proteins showed the presence of either His or Gln at this position . For that reason, positively charged and basic His374 was replaced by Gln and Lys both containing one nitrogen atom in their side chain. The Gln nitrogen superimposes with the distal imidazole nitrogen while Lys cannot superimpose similarly but, like His, has a positive charge. Mutant TAXI-I proteins were overexpressed in Pichia pastoris and purified to homogeneity as described previously . From a 100 mL culture of each of the yeast transformants, about 3–5 mg of purified protein was obtained, similar to the yield obtained for wild-type TAXI-I . Each of the purified mutants exhibited a single band with molecular mass (≈ 42 kDa) similar to that of the wild-type protein on SDS/PAGE. The isoelectric point of all mutants was at least 9.3. CD spectra (Fig. 2) for mutants TAXI[H374A] and TAXI[H374K] were similar to that of wild-type TAXI-I indicating that there were no major structural perturbations caused by the mutations. No CD spectrum was recorded for TAXI[H374Q]. However, as the protein still showed endoxylanase inhibition activity (cfr. infra), the overall protein structure was expected to be similar like those of wild-type TAXI-I and the mutants TAXI[H374A] and TAXI[H374K].
Endoxylanase inhibition activity of the TAXI-I mutants
Specific endoxylanase inhibition activities of the TAXI-I His374 mutants were determined for a defined set of GH11 endoxylanases (Table 1). Endoxylanase selection was based on high sensitivity for wild-type TAXI-I inhibition [12,15,22]. Endoxylanases from A. niger, Penicillium funiculosum, Trichoderma longibrachiatum and B. subtilis were all strongly inhibited by wild-type TAXI-I with specific inhibition activities ranging from 2900 to 4400 IU·mg−1 protein. The TAXI-I His374 mutants, however, exerted large differences in endoxylanase inhibition activities. A. niger endoxylanase inhibition activity was not detected for any of the three mutants even when, in the inhibition assay, 10 times more inhibitor was used than for wild-type TAXI-I. All mutants still strongly inhibited P. funiculosum endoxylanase with specific inhibition units ranging from 2500 to 2900 per mg protein. T. longibrachiatum M3 endoxylanase was strongly inhibited by TAXI[H374A] and TAXI[H374Q], but only weakly by TAXI[H374K]. The latter TAXI-I mutant exerted low endoxylanase inhibition activities for all GH11 endoxylanases tested with exception of the one from P. funiculosum. The lysine side chain probably does not optimally fit into the active site of the endoxylanases. Both mutants TAXI[H374A] and TAXI[H374Q] showed reduced inhibition activities against B. subtilis and T. longibrachiatum M2 endoxylanases.
Table 1. Specific inhibition units (sd = 10%, n = 3) of TAXI-I and TAXI-I His374 mutants for GH11 endoxylanases (endoxylanase selection was based on high sensitivity for wild-type TAXI-I inhibition [12,15,22]. Endoxylanase inhibition activities were determined with a routinely used variant of the colorimetric Xylazyme-AX method using Xylazyme-AX tablets as substrate. One enzyme unit (EU) corresponded to an increase in absorbance of 1.0 at 590 nm under the conditions of the assay. One inhibition unit (IU) was defined as the amount of inhibitor that, under the conditions of the assay, reduces the A590 of one EU by 50% (to 0.5). Inhibition activities were determined with 0–10 nm wild-type and 0–60 nm mutant TAXI-I.
Real-time interactions of wild-type TAXI-I and TAXI-I His374 mutants with GH11 endoxylanases studied by SPR analysis
SPR measurements were performed to further investigate the specific contribution of TAXI-I His374 to complexation with three industrially important GH11 endoxylanases. Real-time interactions between A. niger, T. longibrachiatum M2, B. subtilis GH11 endoxylanases and wild-type TAXI-I and mutants TAXI[H374A], TAXI[H374K], TAXI[H374Q] were determined. Figure 3 shows representative real-time binding sensorgrams. Equilibrium dissociation constants (KD) were calculated from the ratio of dissociation rate constants over association rate constants (koff/kon) and were in the low nanomolar range (Table 2).
Table 2. Kinetic parameters of the binding of TAXI-I and its mutants to GH11 endoxylanases. Association rate constants (kon), dissociation rate constants (koff) and equilibrium dissociation rate constants (KD = koff/kon) were derived from SPR data using Biacore software. The data were fitted to the 1 : 1 Langmuir binding model. The rate constants represent the average of measurements with three different endoxylanase concentrations performed at least in duplicate (± sd). Measurements were performed at 22 °C in 100 mm sodium acetate pH 5.0.
A. niger endoxylanase
1.96 ± 0.48
0.72 ± 0.15
1.59 ± 0.35
No binding detected
koff (10−4 s−1)
0.72 ± 0.42
39.1 ± 17.6
56.1 ± 13.2
3.77 ± 2.00
545 ± 208
360 ± 82.5
T. longibrachiatum M2 endoxylanase
5.35 ± 2.32
3.37 ± 1.82
3.72 ± 1.73
2.80 ± 1.62
koff (10−4 s−1)
0.82 ± 0.35
14.7 ± 3.43
3.25 ± 0.21
60.0 ± 19.8
1.84 ± 1.05
55.3 ± 30.2
10.8 ± 5.42
551 ± 916
B. subtilis endoxylanase
36.1 ± 10.6
11.7 ± 6.84
25.6 ± 13.3
4.57 ± 1.52
koff (10−4 s−1)
3.60 ± 0.43
15.9 ± 1.90
11.8 ± 1.00
34.4 ± 6.70
1.07 ± 0.32
17.0 ± 7.45
6.58 ± 4.68
81.6 ± 29.9
KD values of the complexes between wild-type TAXI-I and A. niger or B. subtilis endoxylanase (KD[A. niger] 3.77 nm, KD[B. subtilis] = 1.07 nm) were, respectively, five- and 17-fold lower than the inhibition constants (Ki[A. niger] = 20.1 nm, Ki[B. subtilis] = 16.7 nm) previously measured by endoxylanase inhibition assays using water-soluble wheat AX as substrate . Differences may be due to different experimental conditions, including absence/presence of substrate, reaction temperatures (difference of 8 °C) and use of different measurement techniques. A closer look at the SPR results revealed differences in endoxylanase affinity. The wild-type inhibitor has a threefold higher affinity for B. subtilis than for A. niger endoxylanase (KD values of 1.07 nm vs. 3.77 nm, respectively).
Mutation of TAXI-I His374 clearly affected the interaction between enzyme and inhibitor. In contrast to what could be predicted from the obtained specific endoxylanase inhibition activity results, mutants TAXI[H374A] and TAXI[H374Q] still showed affinity (albeit reduced) for A. niger endoxylanase. However, no A. niger endoxylanase interaction was observed for TAXI[H374K]. The kon rate constant for TAXI[H374A] was about three times lower than that of wild-type TAXI-I, while the association rate constant of TAXI[H374Q] was only slightly affected. Large differences were found in the dissociation rate constants with koff values for TAXI[H374A] and TAXI[H374Q] that showed about a 50- and 80-fold increase, respectively, over the constant of wild-type TAXI-I. Consequently, the KD values for both mutants were at least 95-fold higher, indicating weaker affinity of the TAXI-I mutants than wild-type TAXI-I for A. niger endoxylanase. Lower affinity may explain the absence of A. niger endoxylanase inhibition by the TAXI-I His374 mutants in the inhibition assay because the amount of mutated inhibitor used was too low. Indeed, in another experiment, 100-fold excess of mutated inhibitor (TAXI[H374A] or TAXI[H374Q]) slightly reduced the A. niger endoxylanase activity (results not shown).
Reduced affinity was also observed for the interaction of the TAXI-I His374 mutants with T. longibrachiatum M2 and B. subtilis endoxylanases. However, mutation of TAXI-I His374 did more markedly affect the kon rate constants of the interaction with B. subtilis endoxylanase than the kon rate constants of the interactions with A. niger or T. longibrachiatum M2 endoxylanases. At the same time, the increase in koff values was less pronounced for B. subtilis endoxylanase than for either the A. niger or T. longibrachiatum M2 endoxylanase. However, the range of koff increases was still larger than the decrease in their respective kon rate constants.
Hence, mutation of TAXI-I His374 weakened the affinity of the enzyme–inhibitor interaction. This was clearly reflected in the drastic increase in koff and KD constants. The smaller effect on the kon rate constants indicated that TAXI-I His374 is less critical for docking of the inhibitor onto the enzyme. Crystallographic analysis of the TAXI-I–A. niger endoxylanase complex  showed that the two nitrogen atoms of the TAXI-I His374 imidazole ring strongly interact with several endoxylanase residues (Fig. 1B). They are highly stabilized through salt bridge and hydrogen bond interactions with Asp37, a residue known to determine the acidic enzymic pH optimum , and ionic interactions with the nucleophilic catalyst Glu79 of the endoxylanase (Fig. 1B). In addition, strong hydrogen bond interactions were observed for TAXI-I His374 and acid/base catalyst Glu170 of the endoxylanase. As mutation of TAXI-I His374 most affected the koff rate constants, it seems that both nitrogen atoms of the TAXI-I His374 imidazole ring provide additional stabilization once the enzyme–inhibitor complex is formed. This is especially true for A. niger endoxylanase as the KD values of its interaction with all TAXI-I His374 mutants largely increased. Our assumption was convincingly demonstrated by the binding behavior of TAXI-I mutants TAXI[H374A] and TAXI[H374Q] (Table 2). The side chain of Gln (containing one nitrogen atom) probably accounts for additional hydrogen bonding interactions compared to the aliphatic side chain of Ala, resulting in stronger complex formation and lower equilibrium dissociation constants for TAXI[H374Q] than for TAXI[H374A]. The higher KD values for TAXI[H374K] may be due to difficult fitting and stabilization problems of the flexible and long Lys side chain into the endoxylanase active site cleft despite the presence of one positively charged nitrogen atom in its terminal amino group.
Interface charges and pH stability of the enzyme–inhibitor interaction studied by IEF titration curves
The above results indicated that, especially for A. niger endoxylanase, electrostatic interactions are important for stabilization of the enzyme–inhibitor complex. The need for interface charges and the pH-stability of the enzyme–inhibitor interaction were therefore studied in the pH range pH 3.0–9.0 by IEF titration curves of enzyme, inhibitor and their complex. Pre-incubation of wild-type TAXI-I and A. niger endoxylanase at pH 5.0 resulted in the formation of an enzyme–inhibitor complex stable in a pH range from pH 3.0–7.0 (Fig. 4). Above pH 7.0, the complex dissociated. The complexes of TAXI-I with endoxylanases from B. subtilis or T. longibrachiatum M2 were stable in the entire pH range tested from pH 3.0–9.0, respectively (Fig. 4). These results show that the weaker A. niger endoxylanase–TAXI-I complex also is more susceptible to pH–induced changes in interface charges. The pH-dependent stability of this enzyme-inhibitor complex cannot easily be rationalized since stability is reflected in the titration of several residues in the active site of the endoxylanase and TAXI-I itself. However, it can be partially explained by the weakened affinity of the TAXI-I His374 mutants for GH11 endoxylanases indicating that the presence of normally charged TAXI-I His374 is of utmost importance for hydrogen bonding and salt bridge interactions with Glu79, Glu170 and Asp37 of A. niger endoxylanase. These results are in perfect agreement with structural data of the A. niger endoxylanase–TAXI-I complex  and with results of Tahir and coworkers  who found that mutation of the negatively charged and pH-optimum related Asp37 of A. niger endoxylanase into neutral Ala completely abolishes interaction with TAXI-I. Recent data by Raedschelders et al.  showed the functional importance of His374 in the interaction between TAXI-type proteins and A. niger endoxylanase. Moreover, the TAXI combination His374/Leu292 is required for A. niger endoxylanase inhibition by TAXI-type proteins. In a pH range 3.0–9.0, changes in interface charges are less critical for stabilization of the enzyme–inhibitor complexes of wild-type TAXI-I and endoxylanases from B. subtilis or T. longibrachiatum M2 as such complexes were stable between pH 3.0–9.0. Indeed, SPR data showed that the increase in koff rate constants was smaller for B. subtilis and T. longibrachiatum M2 endoxylanases than for interaction with A. niger endoxylanase. Hence, the IEF titration curves were in perfect agreement with the SPR analyses of the endoxylanase–TAXI-I interaction.
Our results suggest that TAXI-I His374 is important for stabilization of the formed enzyme–inhibitor complex. Mutation of this amino acid residue to a variable degree affects the affinity of the enzyme–inhibitor interaction. Inhibition activity remains upon mutation of TAXI-I His374 for endoxylanases from P. funiculosum and T. longibrachiatum M3. This is strong evidence for additional inhibition determinants in the TAXI-I protein. Based on structural data , possible candidates are the neighboring amino acid residues of TAXI-I His374 such as Phe375 and Thr376 and the residues situated on another endoxylanase interaction TAXI-I loop, e.g. amino acid Leu292 as shown recently for A. niger endoxylanase . Determination of the crystal structures of TAXI-I complexed with different GH11 endoxylanases and mutational analysis of enzyme and inhibitor would provide insight in the currently observed differences in enzyme-inhibitor affinity. In this way, the exact mechanism of endoxylanase inhibition by TAXI-I and the specific contributions of the amino acid residues at the interface of the enzyme–inhibitor complex would be unravelled.
Endoxylanase inhibition by TAXI-I is due to a specific, reversible, high-affinity (in the nanomolar range) 1 : 1 stoichiometric interaction between GH11 endoxylanases and TAXI-I. In addition to structural data of an A. niger endoxylanase–TAXI-I complex , site-directed mutagenesis studies showed that TAXI-I His374 plays an important role in the stabilization of the enzyme–inhibitor complex rather than in the docking of the inhibitor onto the enzyme. It is also concluded that TAXI-I His374 is not the sole critical amino acid residue for endoxylanase inhibition. The work proved that substitution of a single specific amino acid residue strongly affects the affinity of the inhibitor for the enzyme. This fact opens new perspectives for development of novel TAXI mutants with adapted endoxylanase specificity.
Primers were from Proligo Primers and Probes (Paris, France) and restriction enzymes were from Roche Diagnostics (Basel, Switzerland). Escherichia coli TOP10 (Invitrogen, Carlsbad, CA, USA) cells were used for cloning while Pichia pastoris strain X33 (Invitrogen) was used for protein expression experiments. GH11 endoxylanases were from Aspergillus niger (M4 from Megazyme (Bray, Ireland), Swissprot P55329), Bacillus subtilis (Grindamyl H640 from Danisco (Brabrand, Denmark), Swissprot P18429), Trichoderma longibrachiatum (formerly T. reesei) (M2 and M3 from Megazyme, Swissprot P36218 and P36217, respectively), Penicillium funiculosum (xynC, GenBank acc. number CAC15487; kind gift from C. Furniss, I.F.R., Norwich, UK) and Trichoderma viride (M1 from Megazyme, Swissprot AJ012718).
Site-directed mutagenesis of TAXI-I His374: plasmid construction and P. pastoris transformation
The contribution of amino acid His374 of TAXI-I to endoxylanase inhibition was studied by mutation of this amino acid residue into alanine, lysine or glutamine. Site-directed mutagenesis was performed using a ‘two-round’ PCR method for mutants H374A and H374K. As His374 is situated near the end of the TAXI-I coding sequence , a first round PCR reaction was performed incorporating the H374A mutation with a reverse primer and amplifying the complete TAXI-I coding sequence except its C-terminal end. In a second PCR reaction, the complete coding sequence was amplified with a reverse primer that overlapped with the first PCR product. PCR reactions were performed in 30 µL using 2 units cloned Pfu DNA polymerase (Stratagene, La Jolla, CA, USA), commercially supplied buffer, 200 µm of each dNTP, 1 µm of each primer and 5 ng of template DNA. The reaction mixtures were incubated for 2 min at 95 °C, followed by 35 cycles of 1 min at 95 °C (dissociation), 90 s at 47 °C (annealing), 2 min at 72 °C (extension), and a final extension step for 20 min at 72 °C on a Mastercycler gradient (Eppendorf, Hamburg, Germany). For the first round of PCR, plasmid pQE16-SSPelB TAXI-I  was used as DNA template together with forward primer XIF1 (Table 3), comprising a BglII restriction site, and reverse mutagenic primer XI019 (Table 3) to incorporate the mutations. The obtained PCR-product was gel-purified using the QIAquick Gel Extraction Kit (Qiagen) and used as DNA template for the final PCR round. The coding sequence of TAXI[H374A] was amplified completely using forward primer XIF1 and reverse primer XIR2 (Table 3), both comprising BglII restriction sites, and the same PCR conditions as above. Again, the PCR product was gel-purified and 3′A-overhangs were added with 1 Unit Supertaq DNA polymerase (SphaeroQ, Leiden, the Netherlands) during 10 min at 72 °C. The PCR product was cloned in the pCR®4-TOPO® TA cloning vector (Invitrogen) according to the manufacturer's instructions. Vectors with insert were retained and sequenced on a 377 DNA Sequencer using ABI PRISM Big Dye Terminator chemistry (Applied Biosystems, Foster City, CA, USA) and vector-specific primers. A BglII-restricted TAXI-H374A insert was cloned into the BsmBI restriction site of the P. pastoris pPICZαC vector (Invitrogen) for protein secretion. Proper insert orientation was verified by restriction digestion and sequencing. Mutant TAXI[H374K] was prepared in the same way. Only for the first round of PCR, forward primer XIF1 and reverse mutagenic primer XI022 (Table 3) were used.
Table 3. Primers used for TAXI-I mutational analysis. Mutagenic bases and restriction sites are in bold and underlined, respectively.
For the construction of the TAXI[H374Q] mutant, the QuikChange® Site-Directed Mutagenesis Kit (Stratagene) was used. One pair of complementary mutant primers XI036 and XI037 (Table 3) was used for the amplification and introduction of mutation H374Q into the pPICZαC-TAXI-I plasmid . All reactions were performed according to the manufacturer's instructions. The obtained vector was sequenced to confirm the desired mutation.
The vectors pPICZαC-TAXI-H374A/K/Q were linearized with PmeI and used for transformation by means of homologous recombination of the P. pastoris X33 genome according to the EasyCompTM Transformation protocol (Invitrogen). Genomic DNA of ZeocinTM-resistant Pichia transformants was isolated  and incorporation of the mutant TAXI-I gene was determined by PCR using vector-specific primers and HotStarTaq DNA polymerase (Invitrogen).
Expression and purification of TAXI-I and TAXI-I mutants
Recombinant X33 TAXI-I and TAXI-I His374 mutants were produced in P. pastoris and purified using cation exchange and gel filtration chromatography as described by Fierens et al. .
Purification of GH11 endoxylanases for IEF titration curve analysis and SPR
Endoxylanases from B. subtilis (extract of 1.0 g Grindamyl H640 in 5 mL 25 mm sodium acetate buffer pH 5.0), A. niger (1.0 mL, M4 from Megazyme) and T. longibrachiatum M2 (1.0 mL, Megazyme) were purified to homogeneity using gel filtration chromatography. Endoxylanase samples were loaded on a Bio-Gel P-30 Gel fine (BioRad, Hercules, CA, USA, 16 mm × 65 cm) column and fractionated at 0.2 mL·min−1 using 250 mm sodium acetate buffer pH 5.0. Highly pure endoxylanase fractions were used for IEF titration curve analysis and SPR.
Protein concentration determination
Protein concentrations of purified proteins were determined by measuring the absorbance at 280 nm. The molar extinction coefficients (m−1·cm−1) were calculated from the amino acid sequences using the protparam tool . The absorbances, corresponding to a protein concentration of 1.000 mg·mL−1, are 4.013 AU (B. subtilis endoxylanase), 2.526 AU (A. niger endoxylanase), 2.464 AU (T. longibrachiatum M2 endoxylanase), 0.763 AU (TAXI-I), 0.764 (TAXI[H374A]), 0.763 AU (TAXI[H374K]) and 0.763 AU (TAXI[H374Q]), respectively.
Endoxylanase inhibition assay
Endoxylanase inhibition activities were determined with a routinely used variant of the colorimetric Xylazyme-AX method as described by Fierens et al. . The endoxylanases were diluted in sodium acetate buffer (25 mm, pH 5.0) with 0.5 mg·mL−1 bovine serum albumin (BSA). Inhibitor fractions were dissolved in 25 mm sodium acetate buffer (pH 5.0). One enzyme unit (EU) corresponded to an increase in absorbance of 1.0 at 590 nm under the conditions of the assay. One inhibition unit (IU) was defined as the amount of inhibitor that, under the conditions of the assay, reduces the A590 of one EU by 50% (to 0.5). All inhibition activity measurements were performed in triplicate.
CD spectra of TAXI-I, TAXI[H374A] and TAXI[H374K] were recorded with a JASCO Spectropolarimeter (J-810) (JASCO Benelux, B.V., Maarssen, the Netherlands) at room temperature using a 0.1 mm quartz cell path length. Three scans in the far-UV (260–180 nm) were recorded for each protein sample and averages are reported. The obtained ellipticities were expressed as molar ellipticities. Pure protein samples of TAXI-I, TAXI[H374A] and TAXI[H374K] were prepared in phosphate buffer (10 mm, pH 6.0) at concentrations of 17.5 µm, 17.0 µm and 16.8 µm, respectively.
Surface plasmon resonance (SPR) analysis
SPR allows direct visualization of real–time interactions and, hence, determination of complex association (kon) and dissociation (koff) rate constants. The kinetics of binding between TAXI-I wild-type, TAXI[H374A/K/Q] mutants and GH11 endoxylanases from A. niger, B. subtilis and T. longibrachiatum M2 were analyzed in real time by SPR using a BIAcore 3000 (BIAcore, Uppsala, Sweden) system. Random amine coupling of TAXI-I wild-type and mutants was carried out by injecting the proteins (10 µg·mL−1 each) in 10 mm sodium acetate, pH 5.0, following preactivation of the carboxymethylated dextran matrix (CM5 sensor chip) using N-hydroxysuccinimide (NHS)/N-ethyl-N′-[3-(diethylamino)propyl]carbodiimide. After injection of the proteins, the residual NHS esters were deactivated by the injection of 25 µL of ethanolamine (1 m, pH 8.5).
BIAcore 3000 sensorgrams [resonance units (RU) versus time] were recorded at a flow rate of 30 µL·min−1 at room temperature (22 °C), using different concentrations of analytes (A. niger endoxylanase: 300 nm, 600 nm, 1200 nm, 2205 nm; B. subtilis endoxylanase: 100 nm, 200 nm, 307 nm; T. longibrachiatum M2 endoxylanase: 50 nm, 100 nm, 200 nm) in sodium acetate running buffer (100 mm, pH 5.0). The runs were at least in duplicate for each analyte concentration using sensor chips coupled with 1200–1400 RU of protein. One RU corresponds to 1 pg of bound protein/mm2. Association and dissociation data were both collected for 8 min. The sensor chips were regenerated at the end of each run by one 10 µL injection of 50 mm sodium hydroxide pH 12.0. Data obtained from parallel flow cells with coupled thrombin activatable fibrinolysis inhibitor  served as blank sensorgrams for subtraction of changes in the bulk refractive index.
The sensorgrams were analyzed using biaevaluation version 3.1 software that provides both numerical integration and global fitting algorithms. The data were fitted to a single-site interaction model [1:1 (Langmuir) binding: A + B ⇆ AB]. Assuming pseudo-first-order interaction kinetics, the rate of complex formation during sample injection is given by d[AB]/dt = kon[A][B]–koff[AB], which may be expressed as dR/dt = konCRmax– (konC+koff)R, where dR/dt is the rate of change of the SPR signal, C is the concentration of analyte, Rmax is the maximum analyte binding capacity in RU, and R is the recorded SPR signal in RU at time t. koff values were determined from the data collected during the dissociation phase (dR/dt = -koffR), while kon values were derived from the above rate equation for complex formation. The equilibrium dissociation (KD) constants were calculated from the kinetic rate constants (KD = koff/kon). Details of the rate equations are described in the BIAevaluation version 3 software manuals.
IEF titration curve analysis
IEF titration curve analysis was performed with the PhastSystemTM (Amersham Biosciences, Uppsala, Sweden). Protein samples were applied on a PhastGel IEF 3–9 (Amersham Biosciences). In the first dimension, carrier ampholytes contained in the gel were subjected to an electric field (2000 V, 2.5mA, 3.5 W, 15 °C and 150 Vh) to generate a pH gradient (3–9). The gel was then rotated clockwise 90°, and the protein sample was applied perpendicular to the pH gradient. The focusing step was then performed (1000 V, 2.5mA, 0.2 W, 15 °C and 40–50Vh). Protein samples (1–5 ng of inhibitor, endoxylanase or endoxylanase–inhibitor complex/well) were solubilized in 25 mm sodium acetate buffer (pH 5.0) and incubated during 30 min at room temperature before applying them onto the gel. Finally, proteins were silver stained as described in Amersham Biosciences' development technique file 210.
The authors thank Dr Y. Engelborghs and M. Hellings (Laboratory of Biomolecular Dynamics, K.U. Leuven, Leuven, Belgium) for helpful discussions and technical assistance concerning the CD spectra measurements. We gratefully acknowledge Griet Compernolle for technical assistance with the SPR analyses. Financial support was obtained from the ‘Bijzonder Onderzoeksfonds’ (K.U. Leuven, Belgium). This study was carried out in the framework of research project GOA/03/10 financed by the Research Fund K.U. Leuven and with financial support from the ‘Fonds voor Wetenschappelijk onderzoek-Vlaanderen’ (F.W.O. Vlaanderen, Brussels, Belgium). A.G., K.G., A.R and S.V.C. are postdoctoral fellows of the F.W.O. Vlaanderen. GBOU project funding by the ‘Instituut voor de aanmoediging van Innovatie door Wetenschap en Technologie in Vlaanderen’ (I.W.T., Brussels, Belgium) is gratefully acknowledged.