IG

LEMa

b

c

d

e

a

ARR3AA

KGHN−S

HcoNXx

Af

A

0h

Journal of Biotechnology 174 (2014) 64–72

Contents lists available at ScienceDirect

Journal of Biotechnology

j ourna l ho me pa ge: www.elsev ier .com/ locate / jb io tec

mpact of an N-terminal extension on the stability and activity of theH11 xylanase from Thermobacillus xylanilyticus

etian Songa,b,c,1, Claire Dumona,b,c, Béatrice Siguiera,b,c,d, Isabelle Andréa,b,c,lena Eneyskayae, Anna Kulminskayae, Sophie Bozonneta,b,c,ichael Joseph O’Donohuea,b,c,∗

Université de Toulouse, INSA,UPS,INP, LISBP, 135 Avenue de Rangueil, F-31077 Toulouse, FranceINRA, UMR792 Ingénierie des Systèmes Biologiques et des Procédés, F-31400 Toulouse, FranceCNRS, UMR5504, F-31400 Toulouse, FranceCNRS, Institut de Pharmacologie et de Biologie Structurale, F-31077 Toulouse, FranceNational Research Center “Kurchatov Institute”, B.P. Konstantinov Petersburg Nuclear Physics Institute, Gatchina, 188350 St. Petersburg, Russia

r t i c l e i n f o

rticle history:eceived 18 July 2013eceived in revised form1 December 2013ccepted 3 January 2014vailable online 15 January 2014

eywords:H11 xylanaseybrid enzyme-terminal region3 subsite mappingecondary binding site

a b s t r a c t

To understand structure–function relationships in the N-terminal region of GH11 xylanases, the 17 N-terminal amino acids of the GH11 xylanase from Neocallimastix patriciarum (Np-Xyn) have been graftedonto the N-terminal extremity of the untypically short GH11 xylanase from Thermobacillus xylanilyticus(Tx-Xyn), creating a hybrid enzyme denoted NTfus. The hybrid xylanase displayed properties (pH andtemperature optima) similar to those of the parental enzyme, although thermostability was lowered, withthe Tm value, being reduced by 5 ◦C. Kinetic assays using oNP-Xylo-oligosaccharides (DP2 and 3) indicatedthat the N-extension did not procure more extensive substrate binding, even when further mutagene-sis was performed to promote this. However, these experiments confirmed weak subsite −3 for bothNTfus and the parental enzyme. The catalytic efficiency of NTfus was shown to be 17% higher than thatof the parental enzyme on low viscosity wheat arabinoxylan and trials using milled wheat straw as thesubstrate revealed that NTfus released more substituted oligosaccharide products (Xyl/Ara = 8.97 ± 0.13compared to Xyl/Ara = 9.70 ± 0.21 for the parental enzyme), suggesting that the hybrid enzyme possesses

wider substrate selectivity. Combining either the parental enzyme or NTfus with the cellulolytic cock-tail Accellerase 1500 boosted the impact of the latter on wheat straw, procuring yields of solubilizedxylose and glucose of 23 and 24% of theoretical yield, respectively, thus underlining the benefits of addedxylanase activity when using this cellulase cocktail. Overall, in view of the results obtained for NTfus,we propose that the N-terminal extension leads to the modification of a putative secondary substratebinding site, a hypothesis that is highly consistent with previous data.

Abbreviations: BWX, birchwood xylan; Dpl-WS, xylanase-depleted wheat straw;-bond, hydrogen bond; In-WS, intact (untreated) wheat straw; LVWAX, low vis-osity wheat arabinoxylan; Np-Xyn, Neocallimastix patriciarum GH-11 xylanase;NP-X2, o-nitrophenyl-�-d-xylobioside; oNP-X3, o-nitrophenyl-�-d-xylotrioside;Tfus, N-terminal fused hybrid xylanase; SBS, secondary substrate binding site; Tx-yn, Thermobacillus xylanilyticus xylanase; X/A ratio, xylose to arabinose ratio; XOS,ylo-oligosaccharides.∗ Corresponding author at: LISBP-Biocatalysis group INSA/INRA UMR 792, 135,venue de Rangueil, 31077 Toulouse, France. Tel.: +33 5 61 55 94 28;

ax: +33 5 61 55 94 00.E-mail address: michael.odonohue@insa-toulouse.fr (M.J. O’Donohue).URL: http://www.lisbp.fr/en/index.html (M.J. O’Donohue).

1 Current address: Institut National de la Recherche Scientifique, INRS-Institutrmand-Frappier, Laval, Québec, Canada H7V 1B7.

168-1656/$ – see front matter © 2014 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.jbiotec.2014.01.004

© 2014 Elsevier B.V. All rights reserved.

1. Introduction

Endo-�-1,4-xylanases (EC 3.2.1.8) that form the glycoside-hydrolase family 11 (GH11) of the CAZy database are true xylanases,because they only hydrolyze �-1,4 xylosidic linkages in xylans(Berrin and Juge, 2008; Collins et al., 2005). In the past, xylanasesfrom family GH11 have been empirically selected for a wide varietyof industrial applications, and are currently in use in sectors suchas paper pulping, food and feed processing (Collins et al., 2005;Kulkarni et al., 1999). Likewise, now that biorefining has become apriority R&D target, the role of xylanases in the conversion of ligno-cellulosic biomass into platform intermediates, such as glucose and

xylose, is increasingly recognized. Recent studies have underlinedthe importance of xylanases (and other hemicellulases) in cellu-lolytic cocktails for the hydrolysis of residual hemicelluloses, whichremain in the biomass after pre-treatment (Gao et al., 2010; Kumar

L. Song et al. / Journal of Biotech

Fig. 1. Superposition of the molecular structures of Tx-Xyn (orange), derived fromcrystallographic data, and NTfus (grey), generated using in silico modelling (thiswork). The key generic features of all GH11 xylanases are indicated. The N-terminalett

a2sa(itp

osAnxr(ta(o(itA�llsTiPtcte

taas

xtension in NTfus, derived from Np-Xyn, is shown in violet. (For interpretation ofhe references to color in figure legend, the reader is referred to the web version ofhe article.)

nd Wyman, 2009; Rémond et al., 2010; Song et al., 2012; Wyman,007). Moreover, xylanases will be important in integrated, or con-olidated, biomass processing strategies, which require completersenals of highly active and efficient biomass-degrading enzymesLynd et al., 2005). Therefore, in order to meet tomorrow’s needsn terms of robust, efficient xylanases, it is vital to acquire a bet-er understanding of structure–function relationships, which is arerequisite for enzyme design and engineering.

The secondary structure of GH11 xylanases is composed ofne �-helix and two large anti-parallel �-pleated sheets. The �-heet A is composed of a maximum of six �-strands denoted1–A6, whereas �-sheet B is composed of nine �-strands, which areamed B1–B9 (Törrönen et al., 1994). The N-terminal part of GH11ylanases is variable in length. In cases in which the N-terminalegion is shorter, the �-strand A1 is absent, being replaced by a loopHavukainen et al., 1996; Törrönen and Rouvinen, 1997). Regardingertiary structure, GH11 xylanases display a compact �-jelly rollrchitecture, which is common to members of the CAZy clan GH-Ci.e. families GH11 and GH12). Conveniently, the overall structuref GH11 xylanases has been likened to a partially closed right handFig. 1), and accordingly the different structural elements have beendentified using anthropomorphic terms, such as fingers, palm andhumb (Törrönen et al., 1994). The fingers are formed by �-sheet

and part of �-sheet B, while the palm is made up of the unique-helix and the twisted �-sheet B. The thumb corresponds to the

oop that connects strands B7 and B8 and finally, the ‘cord’ is aong loop that joins B6 and B9 (Purmonen et al., 2007). The activeite is located within a cleft that is the major feature of the palm.he cleft is partially surrounded by the fingers and the thumb, ands blocked at one extremity by the cord (Ludwiczek et al., 2007;urmonen et al., 2007). The catalytic dyad, which is composed ofwo conserved glutamate residues, is located at the centre of theleft and is surrounded by a number of other conserved amino acidshat are important for substrate binding (Collins et al., 2005; Paëst al., 2012).

The N-terminal region of GH11 xylanases, which is located on

he opposite side of the protein with respect to the cord, is gener-lly defined as the region that extends from the N-terminal aminocid through to the �-strand B3, and hence includes five or sixtrands. Numerous studies have shown that the N-terminal region

nology 174 (2014) 64–72 65

is related to enzyme stability and recent molecular dynamics datahave indicated that unfolding initiates there (Hakulinen et al., 2003;Purmonen et al., 2007; Ruller et al., 2007). To improve the thermo-stability of various xylanases, several groups have focused on theN-terminal region, introducing therein disulphide bonds (Paës andO’Donohue, 2006; Xiong et al., 2004) or arginine-rich sequences(Sung, 2007), or by replacing all (Shibuya et al., 2000; Sun et al.,2005) or part of the N-terminal sequence of mesophilic GH11xylanases (Gao et al., 2013; Yin et al., 2013; Zhang et al., 2010)with that of a thermophilic counterpart. However, the role of theN-terminal region in xylanase activity has been the focus of muchless attention, despite the fact that it forms part of the active sitecleft. Nevertheless, having engineered two disulphide bonds intothe GH11 xylanase from Thermobacillus xylanilyticus (Tx-Xyn), withone located in the N-terminal region, Paës and O’Donohue (2006)reported that the resultant mutant enzyme released more solublesugars from destarched wheat bran than the parental enzyme. Sim-ilarly, a study of the GH11 xylanase from Neocallimastix patriciarum(Np-Xyn) has revealed that this enzyme, which displays an unusu-ally extended N-terminal region (one of the longest reported amongGH11 members), is characterized by elevated catalytic efficiencyon oat spelt xylan (kcat/KM = 1400 s−1 mM−1), which is tentativelyattributed to the fact that the N-terminal extension provides thexylanase with an extra subsite (Vardakou et al., 2008). Finally,in a recent study of Tx-Xyn conducted using random mutagene-sis and enzyme evolution, we have isolated a series of mutants,selected through screening for greater hydrolytic potency on wheatstraw. Among the more remarkable mutations, two N-terminally-localized mutants, Y3W and Y6H, were characterized by widersubstrate specificity, being able to hydrolyse both birchwood xylanand low viscosity wheat arabinoxylan with approximately equalefficiency (Song et al., 2012).

Unlike Np-Xyn, the moderately thermostable Tx-Xyn is charac-terized by an exceptionally short N-terminal region, meaning thatthese two enzymes represent opposite extremes of family GH11with respect to this structural feature. Moreover, Np-Xyn was pre-viously reported to be highly active on oat spelt xylan, with thepresence of strong subsite −3 binding being proposed as a determi-nant of this exceptional catalytic property. Therefore, in an attemptto gain some new insight into the functional role of the N-terminalregion of GH11 xylanases and to improve the catalytic potency ofTx-Xyn, we have created a hybrid enzyme by adding the A1 andB1 strands of Np-Xyn to the N-terminal extremity of Tx-Xyn. Inthis report, we recount our findings, in particular with regard toenzymatic activity of the hybrid enzyme.

2. Materials and methods

2.1. Plasmid construction and bacterial strain

The recombinant plasmids used in the study were as follows:pECXYL-R2, which contains the coding sequence of Tx-Xyn in vec-tor pRSETa (Paës and O’Donohue, 2006) and pNPTX5 which encodeshybrid xylanase NTfus, the construction of which is describedbelow. The Escherichia coli strain JM109 (DE3) (Stratagene, USA)was used for protein expression.

2.2. N-terminal fusion of Tx-Xyn xylanase

An 81 base-length oligonucleotide (primer PL1, shown below)including the coding sequence of A1 and B1 strands of Np-

Xyn (shown in bold as below) was synthesized, based on E. colicodon preference (K12 type strain), using the online optimizerserver (http://genomes.urv.es/OPTIMIZER/). The fusion work wasachieved through a two-step PCR. In the first PCR, the reaction

6 iotech

mtoTsf

tcaTcert

2

QitN

uuut

2

t(pthcodat2ew

2

iatm3

x

6 L. Song et al. / Journal of B

ixture (total volume of 50 �l) contained 10 ng of pECXYL-R2 asemplate, 250 �M of primers PL1 and P2 (shown below), 200 �Mf each dNTP and 1 IU of Phusion polymerase (NEB Inc., USA).he amplification reaction was conducted using the followingequence: 1 cycle at 98 ◦C for 1 min, 25 cycles of 98 ◦C for 10 s, 50 ◦Cor 30 s and 72 ◦C for 15 s, and finally 1 cycle at 72 ◦C, 10 min.

PL1: 5′-GTGGCATATGGCCTTCACCGTTGGTAACGGTCAGAACCAGCACAAAGGTGTTAACGACGGTACCTACTGGCAGTATTGGAC-3′;PS1: 5′-GTGGCATATGGCCTTCACC-3′;P2: 5′-GGATCAAGCTTCGAATTCTTACC-3′.

The second PCR, which amplified the final DNA product, usedhe products of the first PCR as template, with PS1 and P2 oligonu-leotides as forward and reverse primers, respectively (shownbove). The same PCR reaction sequence as above was employed.he final PCR product was purified using the QIAquick PCR purifi-ation kit (Qiagen, Germany), then double-digested by restrictionnzymes NdeI and EcoRI (NEB Inc., USA) at 37 ◦C. The insert wasecovered from SYBR®-Safe stained agarose gel (Qiagen, Germany),hen cloned into the similarly digested pRSETa vector.

.3. Site-directed mutagenesis

Mutants Ser162Tyr in Tx-Xyn and NTfus were obtained using theuikChange mutagenesis kit (Stratagene, La Jolla, CA). The follow-

ng mutagenic primers (Eurogentec) were designed, to introducehe desired mutation into the sequences encoding Tx-Xyn andTfus:

SER162TYR FW: 5′-GGGCAGCAGCTGGTATTACCAGGTGCTCG-3′

SER162TYR REV: 5′-CGAGCACCTGGTAATACCAGCTGCTGCCC-3′

Following replication using the mutagenic primers, DpnI wassed to digest (1 h at 37 ◦C) the template DNA, and the product wassed to transform E. coli DH5� cells. After, DNA sequencing wassed to confirm the successful introduction of the mutation intohe target sequences.

.4. Growth condition and xylanase purification

JM109(DE3) cells bearing the recombinant plasmids were cul-ured and expressed in LB medium at 37 ◦C as described previouslyPaës and O’Donohue, 2006). The proteins, Tx-Xyn and NTfus, wereurified using the routine two-step chromatographic procedurehat employs Q-Sepharose FF, followed by Phenyl Sepharose (GEealthcare, USA) (Paës et al., 2007), working on an ÄKTA FPLC purifi-ation system (GE Healthcare, Uppsala, Sweden). The purificationf Np-Xyn and the mutants thereof was performed as previouslyescribed (Vardakou et al., 2008). The purified enzymes weredjudged homogeneous after examination of a SDS-PAGE. Pro-ein concentrations were determined by measuring absorbance at80 nm and applying the Lambert–Beer equation. Theoretical molarxtinction coefficients were calculated using ProtParam online soft-are (Gasteiger et al., 2005).

.5. Thermoactivity and thermostability assays

Thermoactivity assays were performed at various temperatures,n the range of 50–75 ◦C. For NTfus, activity at 60 ◦C was explored in

pH range from 5.0 to 7.5. All the experiments were performed inriplicate, unless indicated otherwise. Specific activity was deter-

ined on birchwood xylan (5 g l−1, Sigma–Aldrich, USA) using the,5-dinitrosalicylic acid (DNS) method (Miller, 1959).

Thermostability assays were performed by incubating aylanase solution (100 �M) in 10 mM Tris–HCl, pH 8.0 buffer at

nology 174 (2014) 64–72

50 ◦C, 60 ◦C and 70 ◦C for up to 6 h. At regular time intervals, resid-ual xylanase activity was quantified at 60 ◦C. Xylanase half-life(t1/2) was deduced by fitting the curve with the following equa-tion: ln (residual activity) = kt where t is the time and k is the slope,and t1/2 = k−1 ln(0.5) (You et al., 2010).

2.6. Determination of melting temperatures by differentialscanning fluorimetry (DSF)

CFX96 Real-Time PCR Detection System (Bio-Rad) was usedas a thermal cycler and the fluorescence emission was detectedusing Texas Red channel (�exc = 560–590 nm, �em = 675–690 nm).The PCR plate containing triplicates of samples (20 �l per well)were prepared by mixing SYPRO® Orange (Invitrogen, final concen-tration 10×) with protein (6.75 �M) in 20 mM Tris–HCl, 100 mMNaCl, pH 8.0 buffer. Negative controls containing either SYPROor proteins alone were analyzed in parallel. The DSF assay wasconducted using a temperature ramp from 20 ◦C to 99.5 ◦C, withincrements of 0.3 ◦C every 3 s. The apparent melting temperature(Tm) was calculated using the Bio-Rad CFX Manager software.

2.7. Kinetic assays

Kinetic parameters were derived from reactions using birch-wood xylan (BWX) and low viscosity wheat arabinoxylan (LVWAX,Megazyme Inc., Ireland) as substrates. Eight different concen-trations were selected in the range of 0.5–12 g l−1 for eachsubstrate. Initial velocities were measured by following the ratesof appearance of reducing sugar with the DNS method. The kineticparameters (kcat and apparent KM) were calculated using SigmaPlotV10.0 and the non-linear regression algorithm embedded in theenzyme kinetics module.

2.8. −3 subsite mapping

Two end-labelled xylo-oligosaccharides (XOS)-o-nitrophenyl-�-d-xylobioside (oNP-X2) and o-nitrophenyl-�-d-xylotrioside(oNP-X3) synthesized as previously described (Eneyskaya et al.,2003) were used to investigate substrate binding in the −3 subsitesof Tx-Xyn and NTfus. Using a previously described method (Matsuiet al., 1991; Suganuma et al., 1978), the binding affinity of glyconsubsites was calculated using Eq. (1). It was assumed that both Tx-Xyn and NTfus release oNP as the main product from oNP-X2 andoNP-X3 (Pollet et al., 2010) and that the oNP group functionallysubstitutes for the xylosyl moiety that binds in subsite +1. There-fore, assuming i = 3 and n = 4, Eq. (1) can be transformed into Eq. (2),which is suitable to calculate the binding energy of −3 subsite.

�G−i = − RT

4183ln

((kcat/KM) × BCF(Xn−i))(Xn)

((kcat/KM) × BCF(Xn−i))(Xn−1)(kcal · mol−1)

(1)

�G−3 = − RT

4183ln

(kcat/KM)(oNP-X3)

(kcat/KM)(oNP-X2)(kcal · mol-1) (2)

where �Gi, Gibbs free energy (binding energy) of subsite i; kcat/KMof Xn, performance constant for reducing end-labelled XOS withDP n; BCF(Xn−i) of Xn, bond cleavage frequency for (n−i)-mer end-labelled product from all hydrolysis products of DP n end-labelledXOS; R, universal gas constant (8.314 J mol−1 K−1); T, temperature,in K (273.15 + degree Celsius).

The kinetic assays for the hydrolysis of oNP-X2 and oNP-X3 weredetermined at six different substrate concentrations in the rangeof 1–8 mM and 0.5–6 mM, respectively. The final concentrations ofboth xylanases were 35 nM and 15 nM for reactions with oNP-X2

iotech

aap1uv

2I

(pdwssp

spatwi9lecrAufwchacwc3wa3Ad2

ha

C

si

2

wNtd(

L. Song et al. / Journal of B

nd oNP-X3, respectively. The hydrolysis reaction was conducted in quartz cuvette at 60 ◦C, in 500 �l of 50 mM sodium acetate buffer,H 5.8. Liberated oNP was measured by continuous detection (Cary00 spectrophotometer, GE Healthcare) at 380 nm and quantifiedsing a standard curve of free oNP (0.1–1 mM). The kcat and KMalues were deduced using SigmaPlot V10.0.

.9. Evaluation of xylanase-mediated hydrolysis on Dpl-WS andn-WS

Finely milled (0.5 mm diameter in average) wheat strawTriticum aestivum, cv. Apache, France), denoted In-WS, wasrepared as previously described (Song et al., 2010). Xylanase-epleted wheat straw (Dpl-WS) was prepared by incubating In-WSith Tx-Xyn (150 BWX U g−1 biomass) for 72 h at 60 ◦C in 50 mM

odium acetate buffer, pH 5.8 (containing 0.02% NaN3). After, theolid residue was recovered by filtration (Whatman® No.4 filteraper), washed and dried at 45 ◦C.

To measure xylanase activity using In-WS or Dpl-WS as sub-trates, a reaction mixture in 50 mM NaOAc buffer, pH 5.8 wasrepared that contained 2% (w/v) biomass, 0.1% (w/v) bovine serumlbumin (BSA), 0.02% (w/v) NaN3, and an aliquot (final concentra-ion of 10 nmol enzyme g−1 biomass) of Tx-Xyn or NTfus. Hydrolysisas performed at 50 ◦C for 24 h with continuous stirring (250 rpm)

n a screwed-capped glass tube, and then stopped by heating at5 ◦C for 5 min. To analyze the combined effect of xylanase and cel-

ulases on In-WS, reactions were conducted as described above,xcept that Accellerase 1500 (Genencor, Rochester, NY) (0.2 mlocktail per g−1 biomass) was added to the reaction mixture andeactions were buffered at pH 5.0 (i.e. the optimal pH for theccellerase 1500 cocktail). Control reactions (buffered at pH 5.0)sing either Accellerase 1500 or xylanase alone were also per-ormed. For the analysis of monosaccharides, the supernatantsere adjusted to 2 M H2SO4 and incubated for 2 h at 95 ◦C. The

arbohydrate content in hydrolysis samples was determined byigh performance anion exchange chromatography with pulsedmperometric detection (HPAEC-PAD) using a Dionex ICS 3000hromatography system (Sunnyvale, CA, USA). Monosaccharidesere separated on a Dionex CarboPac PA-1 column, working at a

oncentration of 4.5 mM NaOH and a flow rate of 1 ml min−1 at0 ◦C over 25 min. Determination of xylo-oligosaccharides (XOS)as achieved on a Dionex CarboPac PA-100 column thermostated

t 30 ◦C, using a linear gradient of NaOAc (5–85 mM NaOAc) over0 min, in a 150 mM NaOH solution, at a flow rate of 1 ml min−1.ppropriate standards such as l-arabinose, d-xylose, d-glucose and-galactose and XOS displaying DP 2 to 6 between 2 mg l−1 and5 mg l−1 were used to provide quantitative analyses.

The amount of monosaccharides or polymers released duringydrolysis was calculated as a percentage of theoretical yieldsccording to the following equation:

onversion%tot. N = average solubilizedN

theoretical totalN× 100%(w/w)

N represents any monosaccharide (e.g. xylose, glucose) in thetraw, and the “theoretical total N” is the total amount of sugar Nn the straw which is derived from compositional analysis.

.10. Compositional analysis of substrates

The sugar composition of BWX, LVWAX, In-WS and Dpl-WSere performed according to the standard method proposed by

REL (Sluiter et al., 2011). Monosaccharide separation and quan-

ification was performed using a Dionex CarboPac PA-1 column asescribed above and results were compared with published dataHespell and Cotta, 1995).

nology 174 (2014) 64–72 67

2.11. Modelling of hybrid xylanase

A three-dimensional model of the NTfus was constructed usingthe 3D structures coordinates of Tx-Xyn (Harris et al., 1997) andNp-Xyn (PDB ID: 2C1F) as starting points. Following a structuralsuperimposition of the two structures, the 18 amino acid residuesat the N-terminal were extracted from the Np-Xyn structure andlinked covalently to the Thr2 residue, located at the N-terminal ofTx-Xyn. To be consistent with the hybrid NTfus sequence, Met1was deleted and mutation of Lys2 into an alanine was performed.The 3D model of the hybrid NTfus was then further refined usingthe CFF91 forcefield implemented within the DISCOVER module ofInsightII software suite (Accelrys, San Diego, CA, USA). For mini-mization, the CFF91cross terms, a harmonic bond potential, and adielectric of 1.0 were used. An initial minimization with a restrainton the protein backbone was performed using a steepest descentalgorithm followed by conjugated gradient minimization stepsuntil the maximum RMS was less than 0.1. In a subsequent step,the system was fully relaxed. Additionally, the Tx-Xyn structuralmodel described from crystal data was optimized using the sameprocedure. Structural visualization and superposition were per-formed using PyMOL Molecular Graphics System (Version 1.5.0.4Schrödinger, LLC).

3. Results

3.1. Model of NTfus

Examination of an alignment of mature (i.e. without signalpeptide) amino acid sequences of 516 GH11 xylanases (Pfam ID:PF00457), including those of Tx-Xyn and Np-Xyn sequences, clearlyillustrates that both of these enzymes are atypical, with Tx-Xyndisplaying an exceptionally short N-terminal region, while Np-Xynpossesses an extended N-terminal extremity (Fig. 2). Similarly, thesuperposition of the 3D structures of Tx-Xyn and Np-Xyn (Fig. 1)further highlights these variations and reveals that the N-terminalextension in Np-Xyn constitutes two extra N-terminal �-strands(A1 and B1), which according to Vardakou et al. contributes to theformation of a −3 subsite responsible for the high catalytic activityof Np-Xyn (Vardakou et al., 2008).

Since the 3D structures of both Np-Xyn (PDB ID: 2C1F) andTx-Xyn (Harris et al., 1997) are available, it was possible to con-struct using computational methods a 3D model of NTfus, whichwas used as the basis for further analysis in place of an experi-mentally determined structure. Briefly, the �-strands A1 and B1from Np-Xyn were excised in silico and pasted onto the secondamino acid position (Thr2) of Tx-Xyn (Fig. 1). The N-terminal Lysresidue was then mutated in silico to Ala, thus finalizing the modelof NTfus. As expected, the energy minimized NTfus was globallysimilar to that of the parental enzymes, with the N-terminal regiondisplaying a conformation and position highly similar to that foundin Np-Xyn (Figs. 1 and 3A). Nevertheless, one major difference inthe N-terminal region was the absence of the �-strand B1, whichwas replaced by a long loop (denoted L1).

The �-strand A1 in NTfus (from Phe2 to Val4) appears to be suit-ably configured to allow the formation of three hydrogen (H) bondswith the anti-parallel �-strand A2 (Fig. 3B). However, only one H-bond is feasible between the loop L1 and strand B2, mainly becauseof steric hindrances resulting from the presence of Trp20 and Tyr22,which are not present in Np-Xyn (Fig. 3B). Moreover, this lack of H-bonding explains why the region that composes L1 cannot adopt a

�-sheet conformation. However, three possible backbone H-bondswere identified in the �-turn that links L1 to B2, in addition to theother three side-chain H-bonds that involve Lys12 – Tyr19, Asn15– Gln36 and Asp16 – Gln36 (Fig. 3B), these interactions strongly

68 L. Song et al. / Journal of Biotechnology 174 (2014) 64–72

Fig. 2. Partial alignment of the N-terminal sequences of NTfus and other GH11 xylanases. All protein sequences were obtained from Swiss-Prot database and signal peptideswere deleted. With the exception of NTfus, proteins are named according to their microbial origin, thus T. xylanilyticus and N. patriciarum correspond to Tx-Xyn and Np-Xynxylanase, respectively. Secondary structure elements of Np-Xyn are indicated by arrows and appropriate nomenclature. The colour scheme is used to indicate the conservationof amino acids, and white letters on a red background are highly conserved residues. Alignment is prepared by ESPript 2.2 utility (Gouet et al., 1991). N. patriciarum, Neo-callimastix patriciarum PDB code 2C1F; T. xylanilyticus, Thermobacillus xylanilyticus C. thermophilum, Chaetomium thermophilum, PDB code 1H1A; T. lanuginosus, Thermomycesl e 1F5G iger, Ai e web

cL

asaaiptiota

F(Hdnoafi

anuginosus, PDB code 1YNA D. thermophilum, Dictyoglomus thermophilum PDB codeobacillus stearothermophilus; T. flexuosa, Thermopolyspora flexuosa PDB 3MF6; A. n

nterpretation of the references to color in figure legend, the reader is referred to th

ontributing to the stabilization of the conformation adopted by1.

According to Vardakou et al. (2008), in Np-Xyn, three amino-cid side-chains (Gln11, Ile151 and Tyr194) might contribute toubstrate binding in the distal (−3) glycon subsite via water medi-ted hydrogen bonds. In NTfus, the equivalent of these residuesre Gln11, Ile132 (116) and Ser178 (162) (original Tx-Xyn number-ng in brackets). Ile132 is located at the tip of the thumb and itsosition is not affected by the N-terminal fusion, and Gln11, con-ained within the N-terminal extension, is also correctly positioned

n the model structure of NTfus. However, Ser178, though capablef entering into a H-bond network, is relatively distant (8.55 A) fromhe endocyclic oxygen of the −3 xylosyl moiety, whereas to make

water-mediated contact the distance between these should be

ig. 3. (A) Superposition of the N-terminal regions of NTfus (violet) and Np-Xynblue). (B) Putative H-bonding in the N-terminal region (Ala1–Gln36) of NTfus. The-bonds formed by main chains and side-chains are indicated by red and blueashed lines, respectively. The aromatic rings of Trp20(4) and Tyr22(6) (Tx-Xynumbering in brackets) create steric hindrance in the vicinity of the loop L1. In theriginal description of the Tx-Xyn structure, Tyr22(6) and Tyr29(13) were identifieds determinants of ‘sticky patches’. (For interpretation of the references to color ingure legend, the reader is referred to the web version of the article.)

J; B. agaradhaerens, Bacillus agaradhaerens PDB code 1QH6; G. stearothermophilus,spergillus niger PDB code 1UKR; H. jecorina, Hypocrea jecorina PDB code 1ENX. (For

version of the article.)

no greater than approximately 5.1 A (assuming an average H-bondlength of 1.97 A and a maximum dimension for H2O of 1.18 A). Nev-ertheless, encouragingly the in silico mutation of Ser178 to tyrosinereduced the overall distance to 5.5 A, which suggests that a sim-ple site-directed mutagenesis experiment might be sufficient tocomplete the creation of a −3 subsite.

3.2. Design and production of the N-terminal modified xylanase,NTfus

The structural superposition of Tx-Xyn and Np-Xyn was usedto design a fusion protein, NTfus that is composed of the Tx-Xynamino acid sequence combined with the N-terminal extensionfrom Np-Xyn, which represents approximately 17 amino acids (51nucleotides). Briefly, the DNA sequence encoding the first 18 aminoacids of Np-Xyn (Met1 to Gly18) was linked to the Tx-Xyn encodingDNA sequence, starting at codon 2 (encoding T2) (Fig. 2). Regardingthe 51 bp DNA extension, compared to the wild type sequence, twochanges were made. First, in order to ensure the correct processingof the resultant recombinant fusion protein by the E. coli methionylaminopeptidase, the second codon (encoding Lys2 in Np-Xyn) wasaltered to encode alanine and, second, the overall codon usage wasoptimized for expression in E. coli.

After construction, cloning and expression of NTfus in E. coli, itwas found that the protein was soluble and could be purified usingthe standard protocol, previously established for wild type Tx-Xyn.Analysis of the purified NTfus by SDS-PAGE clearly indicated anincrease in molecular weight compared to Tx-Xyn, indicating thatthe fusion was successful (data not shown).

3.3. Characterization of NTfus

The specific activity of NTfus at 60 ◦C on birchwood xylan was1170 U/mg protein, which is approximately 80% that of Tx-Xyn.Measurement of its pH optimum revealed that NTfus was opti-mally active at pH 6.2 and displayed quite stable activity (3.5%variation) over 0.8 pH units, from pH 5.8–6.5 which was similarto the parental enzyme. Moreover, NTfus displayed good stabilityover the pH range 5.0–7.5, with residual activities being 90% (pH5.0) and 85% (pH 7.5) of that measured at pH 6.2.

Analysis of the thermoactivity profile of NTfus revealed a Topt

situated at approximately 67 ◦C, like that of Tx-Xyn (Fig. 4A). How-

ever, above 70 ◦C, the decline of activity for NTfus was apparentlyfaster than that observed for Tx-Xyn.

Regarding the thermostability of NTfus and Tx-Xyn, bothxylanases were highly stable at 50 ◦C, retaining 100% activity even

L. Song et al. / Journal of Biotechnology 174 (2014) 64–72 69

Table 1Enzyme characteristics of Tx-Xyn and NTfus.

Enzyme ε (M−1 cm−1) Mw (kDa) SA (U mg−1) pHopt Topt (◦C) Tm (◦C) 60 ◦C t1/2 (h) 70 ◦C t1/2 (h)

Tx-Xyn 102,790 20.65 1450 5.8–6.0 ∼67 75.9 5.4 0.32NTfus 102,790 22.30 1127 ∼6.2 ∼67 70.9 4.1 0.16

ε, extinction coefficient; SA, specific activity measured on 5 g l−1 birchwood xylan at 60 ◦C; Tm, melting temperature; t1/2, half-life.

Table 2Kinetic parameters of Tx-Xyn and NTfus and Ser to Tyr mutants thereof on oNP-xylosides.

MM parameter Substrate Enzyme

Tx-Xyn Tx-Xyn Ser162Tyr NTfus NTfus-Ser178Tyr

kcat (s−1) oNP-X2 36.5 ± 3.1 59.8 ± 9.2 55.4 ± 6.5 30.2 ± 3.7KM (mM) 5.6 ± 1.0 6.9 ± 2.0 5.2 ± 1.3 5.9 ± 1.4kcat/KM (s−1 mM−1) 6.53 8.75 10.62 5.13

kcat (s−1) oNP-X3 145 ± 23 272 ± 89 238 ± 26 124 ± 21KM (mM) 8.6 ± 2.3 11.6 ± 5.7 8.5 ± 1.5 8.5 ± 2.3kcat/KM (s−1 mM−1) 16.96 23.49 27.84 14.52

�G subsite −3 (kcal mol−1) −0.632 −0.654 −0.639 −0.689

Table 3Sugar composition of the different xylanase substrates (% dry wt).

Substrate Glucose% Xylose% Arabinose% Galactose% X/A ratio

BWX 2.99 ± 0.41 97.01 ± 0.41 – – naLVWAX 0.95 ± 0.02 64.13 ± 0.42 34.92 ± 0.43 – 1.84In-WS 44.51 ± 0.08 26.16 ± 0.14 2.37 ± 0.03 0.44 ± 0.06 10.98Dpl-WS 45.69 ± 0.94 21.92 ± 0.17 2.05 ± 0.07 0.46 ± 0.05 10.75

B heat sn

abdea0brbv(

3t

saaslNscNXNttbXw

used to compare the activity of Tx-Xyn and NTfus on complexbiomass, respectively. Compositional analysis (Table 3) revealedthat Dpl-WS was characterized by less xylose (4.3% dry weight)and arabinose (0.3% dry weight) than In-WS, but the Xyl/Ara ratio

Table 4Kinetic parameters of Tx-Xyn and NTfus on polymeric substrates.

MM parameter Substrate Enzyme

Tx-Xyn NTfus

kcat (s−1) BWX 610.5 ± 20 573.7 ± 3KM app (g l−1) 2.5 ± 0.2 2.3 ± 0.1

WX, birchwood xylan; LVWAX, low viscosity wheat arabinoxylan; In-WS, intact wa, not applicable.

fter 6-h incubation (data not shown). At 60 ◦C, Tx-Xyn showedetter thermostability than NTfus (Fig. 4B), and the half-lives wereetermined as 5.4 and 4.1 h, respectively (Table 1). At 70 ◦C, bothnzymes were subject to rapid inactivation (>90% activity lossfter 1 h), and half-lives were estimated to be 0.16 h (NTfus) and.32 h (Tx-Xyn), respectively (Fig. 4B). Overall, NTfus appeared toe less thermostable than Tx-Xyn. This postulate was corrobo-ated by results from a protein melting experiment performedy differential scanning fluorimetry, which indicated that the Tm

alue of NTfus was 70.9 ◦C, while that of Tx-Xyn was 75.9 ◦CTable 1).

.4. Probing the presence of a −3 subsite in Tx-Xyn and NTfus andheir corresponding mutants Ser162Tyr

To probe for extended substrate binding, two synthetic sub-trates, oNP-X2 and oNP-X3, were employed. For both Tx-Xynnd NTfus, the performance constants (kcat/KM) on oNP-X3 werepproximately 2.6-fold greater than on oNP-X2, mainly due toignificant increases (approx. 4-fold) in kcat. However, the abso-ute value of the performance constant was higher in the case ofTfus (Table 2). Concerning the mutation of the serine residue

upposedly involved in substrate binding in the distal (−3) gly-on subsite (Ser162Tyr in Tx-Xyn and its homolog Ser178YTyr inTfus), this had different impacts on the parental enzymes. In Tx-yn, the mutation Ser162Tyr positively affected kcat, whereas inTfus, the equivalent mutation lowered the kcat values for reac-

ions involving both oNP-X2 and oNP-X3. Finally, calculation of

he additional binding energy that characterizes the interactionetween the four xylanase variants and oNP-X3 compared to oNP-2, revealed similar values in the range −0.63 to −0.69 kcal mol−1,hich is indicative of the presence of a weak binding −3 subsite.

traw; Dpl-WS, xylanase-depleted wheat straw; X/A ratio, xylose to arabinose ratio;

3.5. Determination of kinetic parameters using soluble xylans

To analyze xylanase performance on soluble polymeric sub-strates, BWX and LVWAX were employed whose composition ismarkedly distinct with respect to the amount of arabinose content(Table 3). Indeed, the compositional analysis of BWX used in thisstudy did not reveal any arabinose after acid hydrolysis, whereasthe LVWAX showed an X/A ratio of 1.84.

When acting on BWX, Tx-Xyn and NTfus displayed similarMichaelis–Menten constants (kcat and KM app) (Table 3). However,compared to Np-Xyn, both enzymes displayed a lower turnoverrate. In the case of LVWAX, NTfus displayed a 14% higher turnoverrate than Tx-Xyn, although the value of the Michaelis constantremained highly similar to that of Tx-Xyn (Table 4).

3.6. Hydrolysis of wheat straw and synergy with commercialcellulases

Two kinds of wheat straw samples (In-WS and Dpl-WS) were

kcat/KM app (s−1 l g−1) 240.4 245.2

kcat (s−1) LVWAX 1699.4 ± 96 1944.3 ± 112KM app (g l−1) 5.1 ± 0.1 5.0 ± 0.3kcat/KM app (s−1 l g−1) 333.1 388.4

70 L. Song et al. / Journal of Biotechnology 174 (2014) 64–72

FN

itxaoxmwbavso1mpcMs

ttndd

Fig. 5. Hydrolysis of In-WS and Dpl-WS using various enzyme combinations. The

ig. 4. (A) Thermoactivity and (B) thermostability at 60 ◦C and 70 ◦C of Tx-Xyn andTfus.

n both samples was highly similar. Enzymatic hydrolysis reac-ions were performed in two conditions (pH 5.0 or pH 5.8) usingylanases alone (which operate at pH 5.8), or in combination with

commercial cellulase cocktail, Accellerase 1500, which optimallyperates at pH 5.0 (Fig. 5). In the hydrolysis reactions conducted byylanases alone, neither glucose nor xylose were detected, and theain degradation products were xylobiose and xylotriose, togetherith little xylotetraose and xylopentaose. Conversion of all solu-

le xylo-oligosaccharides into monosaccharides through sulphuriccid treatment showed that the action of Tx-Xyn and NTfus wasery similar in both pH conditions, with NTfus perhaps releasinglightly more xylose: Tx-Xyn solubilized 12.8%tot. xyl and 16.7%tot. xylf In-WS at pH 5.0 and 5.8, respectively, whereas NTfus released5.0%tot. xyl (pH 5.0) and 19.2%tot. xyl at pH 5.8 (Fig. 5). In addition,onosaccharide sugar analysis of hydrolytic products obtained at

H 5.8 revealed that NTfus also released more substituted oligosac-harides (Xyl/Ara = 8.97 ± 0.13) than Tx-Xyn (Xyl/Ara = 9.70 ± 0.21).oreover, analysis of similar reactions using Dpl-WS as substrate

howed that NTfus yielded 29.1% more xylose than Tx-Xyn.When an enzyme mixture of Tx-Xyn and Accellerase was used

o hydrolyze In-WS, 24.5%tot. xyl and 23.6%tot. glu were released into

he aqueous reaction medium (Fig. 5), indicating that both arabi-oxylans and �-1,4 glucans had been targeted. Glucose was alwaysetected as a monosaccharide, probably because Accellerase 1500isplays high �-glucosidase activity. When a similar experiment

nature of the enzyme combination used is indicated by letters on the X-axis. A is thehydrolysis of Dpl-WS at pH 5.8, B is the hydrolysis of In-WS at pH 5.8 and C is thehydrolysis of In-WS at pH 5.0.

was performed using NTfus, the amount of solubilized xylose andglucose was very similar, with values of 28.0% and 24.8%, respec-tively. In comparison, when Accellerase 1500 alone was used, only7.3%tot. xyl and 18.9%tot. glu were solubilized, thus clearly underlin-ing the benefit of using xylanase in combination with Accellerase1500.

4. Discussion

The rationale for this work is derived from the previous obser-vation that the GH11 xylanase from N. patriciarum is highly activeon oat spelt xylan, this elevated activity being correlated with thepresence of an extensive active site, comprised of six subsites, from−3 to +3, and to significant substrate binding in the −3 subsite(−2.1 kcal mol−1) (Vardakou et al., 2008). Contrarily, Tx-Xyn is anexceptionally short GH11 member, lacking at least one N-terminal�-strand compared to other GH11 family members. Inevitably, thisimplies that the active site is shorter, and makes the existence ofa −3 subsite uncertain (Paës et al., 2007). Therefore, to probe thefunctional consequences of this difference, an N-terminal extendedversion of Tx-Xyn has been created, using the N-terminal sequenceof Np-Xyn.

The construction of NTfus procured an active enzyme, imply-ing that the fusion protein is structurally coherent and sufficientlystable. This result is gratifying, because it is well-established thatthe N-terminal region of GH11 xylanases is a “hot spot” for proteinunfolding (Hakulinen et al., 2003; Purmonen et al., 2007).

Regarding the slightly lower stability of NTfus, this can be ten-tatively attributed to a number of factors, including the looseassociation of �-strand A1 with L1, the apparent lack of stabi-lizing forces (e.g. hydrophobic interactions, disulphide bonds orsalt bridges) and the possible failure of the newly introducedN-terminal region to adopt �-strand conformation and thus toextend the original �-sheet (Fig. 3A). However, in the absence of anexperimentally determined structure, these explanations are spec-ulative. Nevertheless, one explanation is perhaps better-founded.Previously it was suggested that the thermostability of Tx-Xynis protein concentration dependent and determined by inter-molecular hydrophobic protein-protein interactions (Harris et al.,

1997). Based on crystallographic observations, it was hypothesizedthat surface exposed aromatic side-chains mediate pair-wise asso-ciation of xylanase monomers, forming stabilized homo-oligomers.Among these residues, Tyr6 and 13 (Tx-Xyn numbering) are located

L. Song et al. / Journal of Biotechnology 174 (2014) 64–72 71

Fig. 6. Comparison of the putative SBS grooves of Tx-Xyn and NTfus. Trp198 (1 8 2) and Thr18 (2) (Tx-Xyn numbering in brackets) are indicated in yellow and red, respectively.T d by do d, the

orobTcittwTa

sTtieXstSincslo

atusastcpc

iemiooT

We thank the China Scholarship Council (CSC) for the attribution

he approximate location of the SBS, on the ‘knuckle-side’ of the fingers, is indicatef the fingers and palm. (For interpretation of the references to color in figure legen

n �-strands B2 and A2, which correspond to Tyr22 and Tyr29,espectively, in NTfus (Fig. 3B). Examination of our in silico modelf NTfus reveals that unlike Tyr6 (Tx-Xyn), Tyr22 is completelyuried inside the structure and, compared to Tyr13 (Tx-Xyn),yr29 presents altered exposure. Plausibly, these subtle differencesould prevent NTfus monomers from engaging in thermostabiliz-ng intermolecular hydrophobic interactions and thus account forhe observed lower thermostability of NTfus. Indeed, this interpre-ation is consistent with the findings of another recent study inhich we showed that the introduction of the mutation Tyr6His in

x-Xyn led to a small reduction in Tm and a 2-fold reduction of t½t 60 ◦C (Song et al., 2012).

At the outset, it was thought that the addition of two extra �-trands to Tx-Xyn might create or reinforce −3 subsite binding.his is apparently not the case, because both the wild type andhe N-terminal extended variant display similar −3 subsite bind-ng energies. Nevertheless, this work does provide experimentalvidence for the presence of a weak −3 subsite binding in Tx-yn, thus confirming a previous prediction made using in silicoubstrate docking analysis (Paës et al., 2007). Strikingly, even fur-her modification of NTfus by the introduction of the mutationer162Tyr/Ser187Tyr did not enhance −3 subsite binding, suggest-ng either that the newly introduced tyrosinosyl side-chains didot adopt the exact positions predicted by modelling, or that otheronformational changes are necessary to form a Np-Xyn-like −3ubsite. While disappointing, this result illustrates the difficultiesinked to rational engineering of enzymes and underlines the limitsf static models.

In a recent study, Paës (2012) provided data showing that alter-tions to the thumb tip can strongly influence kcat, inferring thathumb conformation and mobility are key determinants of prod-ct release from the distal subsites (Paës, 2012). In this study, wehow that other alterations in the vicinity of the distal subsites alsoffect kcat. Indeed, both the extension of the distal side of the activeite cleft and the introduction of a tyrosinosyl side-chain beneathhe −3 xylosyl moiety increase kcat approximately 1.7-fold, but theombination of these two modifications decreases the value of thisarameter to the level of Tx-Xyn, possibly due to net steric over-rowding that might hamper thumb-mediated product expulsion.

Intriguingly, though a −3 subsite was not reinforced, the activ-ty and selectivity of NTfus was altered. While both enzymesxhibited an almost equivalent kcat/KM app value on BWX, on theore arabinose-substituted LVWAX, NTfus revealed a 17% increase

n catalytic efficiency compared to the parental enzyme. More-ver, on wheat straw, a similar albeit less pronounced trend wasbserved, since NTfus released slightly more soluble sugars thanx-Xyn, indicating that it was better able to hydrolyze a wider

ashed arrow. The active site is not visible, because this is formed by the inner side reader is referred to the web version of the article.)

range of arabinoxylans. Significantly, this result recalls the unex-pected outcome of the introduction of a disulphide bond linkingthe N- and C- extremities of Tx-Xyn. The resultant enzyme dis-played better ability to hydrolyze destarched wheat bran (Paësand O’Donohue, 2006). Similarly, the use of a directed evolutionselected Tx-Xyn N-terminal mutants (Tyr3Trp and Tyr6His) thatdisplayed altered substrate selectivity (Song et al., 2012). Finally,the mutation of a C-terminal residue (Trp185) in the GH11 xylanasefrom Bacillus circulans (Bc-Xyn) also altered specificity towardswheat water-unextractable arabinoxylans (Xyl/Ara = 1.96), a resultthat was attributed to the alteration of a secondary substrate bind-ing site (SBS), which supposedly plays a role in selectivity (Moerset al., 2007; Vandermarliere et al., 2008). In the light of these obser-vations, it is noteworthy that one of the most prominent differencesbetween Tx-Xyn and NTfus revealed by in silico modelling is thewidening of the groove that is adjacent to NTfus Trp 198 (Tx-Xyn Trp182), which is the equivalent of Trp185 in Bc-Xyn (Fig. 6).Considering that homologous grooves in other GH11 xylanases(Ludwiczek et al., 2007) have been described as SBS, we proposethat the aforementioned groove in Tx-Xyn is also a SBS and thatthe presence of an extended N-terminal in NTfus alters it, thusimproving ligand binding and modifying substrate selectivity.

5. Conclusion

In conclusion, although the creation of a hybrid enzyme wassuccessfully achieved, the addition of 17 amino acids to the N-terminal of Tx-Xyn apparently did not confer the catalytic cleft withnew subsite binding ability. However, the hybrid enzyme did dis-play altered substrate selectivity and activity on two arabinoxylans,an observation that is highly consistent with previous data, andthus provides further weight to an increasing body of evidence thatpoints to the importance of secondary substrate binding in GH11xylanases. In future work, it will be interesting to submit NTfusto a directed evolution approach in order to further optimize theenzyme, notably with respect to the organization of � strands A1and B2, and to recombine several of the more promising mutationsthat we have identified during the course of our recent studies.

Acknowledgements

of a Ph.D. scholarship to L.S., the Région Midi-Pyrénées for a post-doctoral stipend for B.S. and the Integrated Screening Platform ofToulouse (PICT, IPBS, CNRS – Université de Toulouse) for providingaccess to DSF equipment.

7 iotech

R

B

C

E

G

G

G

G

H

H

H

H

K

K

L

L

M

M

M

P

P

P

2 L. Song et al. / Journal of B

eferences

errin, J.-G., Juge, N., 2008. Factors affecting xylanase functionality in the degrada-tion of arabinoxylans. Biotechnol. Lett. 30, 1139–1150.

ollins, T., Gerday, C., Feller, G., 2005. Xylanases, xylanase families and extremophilicxylanases. FEMS Microbiol. Rev. 29, 3–23.

neyskaya, E.V., Brumer, H., Backinowsky, L.V., Ivanen, D.R., Kulminskaya, A.A.,Shabalin, K.A., Neustroev, K.N., 2003. Enzymatic synthesis of beta-xylanase sub-strates: transglycosylation reactions of the beta-xylosidase from Aspergillus sp.Carbohydr. Res. 338, 313–325.

ao, D., Chundawat, S.P.S., Krishnan, C., Balan, V., Dale, B.E., 2010. Mixture opti-mization of six core glycosyl hydrolases for maximizing saccharification ofammonia fiber expansion (AFEX) pretreated corn stover. Bioresour. Technol.101, 2770–2781.

ao, S.-J., Wang, J.-Q., Wu, M.-C., Zhang, H.-M., Yin, X., Li, J.-F., 2013. Engineer-ing hyperthermostability into a mesophilic family 11 xylanase from Aspergillusoryzae by in silico design of N-terminus substitution. Biotechnol. Bioeng. 110,1028–1038.

asteiger, E., Hoogland, C., Gattiker, A., Duvaud, S., Wilkins, M.R., Appel, R.D., Bairoch,A., 2005. Protein identification and analysis tools in the ExPASy server. In:Wilkins, J.M. (Ed.), The Proteomics Protocols Handbook. Horizon Press Inc.,Totowa, NJ, pp. 571–607.

ouet, P., Courcelle, E., Stuart, D.I., Metoz, F., 1991. ESPript: multiple sequence align-ments in PostScript. Bioinformatics 15, 305–308.

akulinen, N., Turunen, O., Janis, J., Leisola, M., Rouvinen, J., 2003. Three-dimensionalstructures of thermophilic �-1,4-xylanases from Chaetomium thermophilum andNonomuraea flexuosa Comparison of twelve xylanases in relation to their thermalstability. Eur. J. Biochem. 270, 1399–1412.

arris, G.W., Pickersgill, R.W., Connerton, I., Debeire, P., Touzel, J.P., Breton, C.,Pérez, S., 1997. Structural basis of the properties of an industrially relevantthermophilic xylanase. Proteins 29, 77–86.

avukainen, R., Törrönen, A., Laitinen, T., Rouvinen, J., 1996. Covalent binding ofthree epoxyalkyl xylosides to the active site of endo-1,4-xylanase II from Tri-choderma reesei. Biochemistry 2960, 9617–9624.

espell, R.B., Cotta, M.A., 1995. Degradation and utilization by Butyrivibrio fibri-solvens H17c of xylans with different chemical and physical properties. Appl.Environ. Microbiol. 61, 3042–3050.

ulkarni, N., Shendye, A., Rao, M., 1999. Molecular and biotechnological aspects ofxylanases. FEMS Microbiol. Rev. 23, 411–456.

umar, R., Wyman, C.E., 2009. Effects of cellulase and xylanase enzymes on thedeconstruction of solids from pretreatment of poplar by leading technologies.Biotechnol. Prog. 25, 302–314.

udwiczek, M.L., Heller, M., Kantner, T., McIntosh, L.P., 2007. A secondary xylan-binding site enhances the catalytic activity of a single-domain family 11glycoside hydrolase. J. Mol. Biol. 373, 337–354.

ynd, L.R., van Zyl, W.H., McBride, J.E., Laser, M., 2005. Consolidated bioprocessingof cellulosic biomass: an update. Curr. Opin. Biotechnol. 16, 577–583.

atsui, I., Ishikawa, K., Matsui, E., Miyairi, S., Fukui, S., Honda, K., 1991. Subsite struc-ture of Saccharomycopsis alpha-amylase secreted from Saccharomyces cerevisiae.J. Biochem. 109, 596.

iller, G.L., 1959. Use of dinitrosalicylic acid reagent for determination of reducingsugar. Anal. Chem. 31, 426–428.

oers, K., Bourgois, T., Rombouts, S., Beliën, T., Van Campenhout, S., Volckaert, G.,Robben, J., Brijs, K., Delcour, J.A., Courtin, C.M., 2007. Alteration of Bacillus subtilisXynA endoxylanase substrate selectivity by site-directed mutagenesis. EnzymeMicrob. Technol. 41, 85–91.

aës, G., 2012. Thumb – loops up for catalysis: a structure/function investigation ofa functional loop movement in a GH11 xylanase. Comput. Struct. Biotechnol. J.,1–24.

aës, G., Berrin, J.-G., Beaugrand, J., 2012. GH11 xylanases: struc-

ture/function/properties relationships and applications. Biotechnol. Adv.30, 564–592.

aës, G., O’Donohue, M.J., 2006. Engineering increased thermostability in the ther-mostable GH-11 xylanase from Thermobacillus xylanilyticus. J. Biotechnol. 125,338–350.

nology 174 (2014) 64–72

Paës, G., Tran, V., Takahashi, M., Boukari, I., O’Donohue, M.J., 2007. New insights intothe role of the thumb-like loop in GH-11 xylanases. Protien Eng. Des. Sel. 20,15–23.

Pollet, A., Lagaert, S., Eneyskaya, E., Kulminskaya, A., Delcour, J.A., Courtin, C.M., 2010.Mutagenesis and subsite mapping underpin the importance for substrate speci-ficity of the aglycon subsites of glycoside hydrolase family 11 xylanases. Biochim.Biophys. Acta 1804, 977–985.

Purmonen, M., Valjakka, J., Takkinen, K., Laitinen, T., Rouvinen, J., 2007. Moleculardynamics studies on the thermostability of family 11 xylanases. Protein Eng.Des. Sel. 20, 551–559.

Rémond, C., Aubry, N., Crônier, D., Noël, S., Martel, F., Roge, B., Rakotoarivonina, H.,Debeire, P., Chabbert, B., 2010. Combination of ammonia and xylanase pretreat-ments: impact on enzymatic xylan and cellulose recovery from wheat straw.Bioresour. Technol. 101, 6712–6717.

Ruller, R., Deliberto, L., Ferreira, T.L., Ward, R.J., 2007. Thermostable variants of therecombinant xylanase A from Bacillus subtilis produced by directed evolutionshow reduced heat capacity changes. Proteins 70, 1280–1293.

Shibuya, H., Kaneko, S., Hayashi, K., 2000. Enhancement of the thermostabilityand hydrolytic activity of xylanase by random gene shuffling. Biochem. J. 656,651–656.

Sluiter, A., Hames, B., Ruiz, R., Scarlata, C., Sluiter, J., Templeton, D., Nrel, D.C., 2011.Determination of Structural Carbohydrates and Lignin in Biomass.

Song, L., Laguerre, S., Dumon, C., Bozonnet, S., O’Donohue, M.J., 2010. A high-throughput screening system for the evaluation of biomass-hydrolyzingglycoside hydrolases. Bioresour. Technol. 101, 8237–8243.

Song, L., Siguier, B., Dumon, C., Bozonnet, S., O’Donohue, M.J., 2012. Engineering bet-ter biomass-degrading ability into a GH11 xylanase using a directed evolutionstrategy. Biotechnol. Biofuels 5, 3.

Suganuma, T., Matsuno, R., Ohnishi, M., Hiromi, K., 1978. Study of mechanism ofaction of Taka-amylase-A (Aspergillus oryzae) on linear oligosaccharides by prod-uct analysis and computer-simulation. J. Biochem. 84, 293–316.

Sun, J.-Y., Liu, M.-Q., Xu, Y.-L., Xu, Z.-R., Pan, L., Gao, H., 2005. Improvement of thethermostability and catalytic activity of a mesophilic family 11 xylanase by N-terminus replacement. Protein Expr. Purif. 42, 122–130.

Sung, W.L., 2007. Modification of xylanases to increase thermophilicity, thermosta-bility and alkalophilicity. WO/2007/115391.

Törrönen, A., Harkkil, A., Rouvinen, J., 1994. Three dimensional structure of endo-1,4-�-xylanase II from Trichoderma reesei: two conformational states in the activesite. EMBO J. 13, 2493–2501.

Törrönen, A., Rouvinen, J., 1997. Structural and functional properties of low molec-ular weight endo-1,4-�-xylanases. J. Biotechnol. 57, 137–149.

Vandermarliere, E., Bourgois, T.M., Rombouts, S., Van Campenhout, S., Volckaert,G., Strelkov, S.V., Delcour, J.A., Rabijns, A., Courtin, C.M., 2008. Crystallographicanalysis shows substrate binding at the −3 to +1 active-site subsites and at thesurface of glycoside hydrolase family 11 endo-1,4-�-xylanases. Biochem. J. 410,71–79.

Vardakou, M., Dumon, C., Murray, J.W., Christakopoulos, P., Weiner, D.P., Juge, N.,Lewis, R.J., Gilbert, H.J., Flint, J.E., 2008. Understanding the structural basis forsubstrate and inhibitor recognition in eukaryotic GH11 xylanases. J. Mol. Biol.375, 1293–1305.

Wyman, C.E., 2007. What is (and is not) vital to advancing cellulosic ethanol. TrendsBiotechnol. 25, 153–157.

Xiong, H., Fenel, F., Leisola, M., Turunen, O., 2004. Engineering the thermostability ofTrichoderma reesei endo-1,4-�-xylanase II by combination of disulphide bridges.Extremophiles 8, 393–400.

Yin, X., Li, J.-F., Wang, J.-Q., Tang, C.-D., Wu, M.-C., 2013. Enhanced thermostabi-lity of a mesophilic xylanase by N-terminal replacement designed by moleculardynamics simulation. J. Sci. Food Agric. 93 (12), 3016–3023.

You, C., Huang, Q., Xue, H., Xu, Y., Lu, H., 2010. Potential hydrophobic interactionbetween two cysteines in interior hydrophobic region improves thermostability

of a family 11 xylanase from Neocallimastix patriciarum. Biotechnol. Bioeng. 105,861–870.

Zhang, Z.-G., Yi, Z.-L., Pei, X.-Q., Wu, Z.-L., 2010. Improving the thermostability ofGeobacillus stearothermophilus xylanase XT6 by directed evolution and site-directed mutagenesis. Bioresour. Technol. 101, 9272–9278.