Design, expression and functional characterization of a thermostable xylanase from Trichoderma reesei

Xylanases isolated from microorganisms such as the Trichoderma reesei have attracted considerable research interest because of their potential in various industrial applications. However, naturally isolated xylanases cannot withstand harsh conditions such as high temperature and basic pH. In this study, we performed structural analysis of the major T. reesei xylanase (Xyn2), and novel flexible regions of the enzyme were identified based on B-factor, a molecular dynamics (MD) parameter. To improve thermostability of the Xyn2, disulfide bonds were introduced into the unstable flexible region by using site-directed mutagenesis and two recombinant xylanases, XM1 (Xyn2Cys12-52) and XM2 (Xyn2Cys59-149) were successfully expressed in Pichia pastoris. Secreted recombinant Xyn2 was estimated by SDS-PAGE to be 24 kDa. Interestingly, the half-lives of XM1 and XM2 at 60°C were 2.5- and 1.8- fold higher, respectively than those of native Xyn2. The XM1 also exhibited improved pH stability and maintained more than 60% activity over pH values ranging from 2.0 to 10.0. However, the specific activity and catalytic efficiency of XM1 was decreased as compared to those of XM2 and native Xyn2. Our results will assist not only in elucidating of the interactions between protein structure and function, but also in rational target selection for improving the thermostability of enzymes.

Recently, protein engineering techniques such as the random and site-directed mutagenesis were proven to be the most efficient method to improve the thermostability of the GH 11 xylanases [10,11]. In previous studies, most molecular evolution or protein design was focused on the N-terminus because of its complicated molecular structures [12,13]. In GH 11 xylanases, protein flexibility plays a critical role in stabilizing the protein structure, and the modification of protein flexibility has been previously utilized to improve the protein stability [14]. In recent years, some flexible molecules or protein regions were successfully identified by using molecular dynamics (MD) simulation. Importantly, these flexible regions were capable of being stabilized by extra intermolecular forces such as salt bridges, hydrogen bonds, aromatic interactions and disulfide bonds [15], which pave the way to improve the protein stability.
Trichoderma reesei xylanase 2 (Xyn2) is one of the most important xylanases in the GH 11 family and has a right hand β-sandwich structure [16]. Xyn2 is a low-molecular mass (21 kDa) enzyme with an alkaline isoelectric point (pI 9.0) and an optimum activity at pH 4-6. The enzyme rapidly loses its activity when incubated at temperatures over 50˚C since it has no stabilizing disulfide bridges or any thermostabilizing domain [16]. In this study, the flexible region at the end of the N-terminal α-helix has been identified based on structural analysis. This flexible region has been stabilized by introducing a covalent disulfide bond (XM1, Cys 14 -Cys 52 ). Additionally, the α-helix has been indirectly fixed to the β-core by introducing a disulfide bond between β-sheets B5 and B6 (XM2, Cys 59 -Cys 149 ) becuase the α-helix was reported to be associated with the stability of GH 11 family xylanases [17]. We describe the design and expression of the two mutated Xyn2 genes (MX1 and MX2) in Pichia pastoris. Moreover, the enzymatic properties of the mutated enzymes are fully characterized.

Structural analysis of the T. reesei Xyn2
The crystal structure with accession code 3AKQ from the RCSB Protein Database (http:// www.rcsb.org/pdb/explore/explore.do?structureId=3akq) was identical to Xyn2 in amino acid sequences. The B-factors were calculated by Swiss-PDB viewer (http://spdbv.vital-it.ch/) based on the coordinates from RCSB Protein entry 3AKQ and visualized by using PyMOL 1.7 (http://www.pymol.org/).

Cloning and site-directed mutagenesis of the Xyn2 gene
The T. reesei Rut C-30 was cultured in basal medium and subsequently collected for total RNA isolation by TRIzol reagent (Takara Biotechnology Co., Ltd., Dalian) according to the manufacturer's instructions. First-strand cDNA synthesis was carried out with THE PrimeScript II 1st Strand cDNA Synthesis Kit (Takara Biotechnology Co., Ltd., Dalian) according to the manufacturer's instructions. Then, the Xyn2 gene was amplified by using the primer pair: P f (5'-GCTGAATTCCAGACGATTCAGCCCGGCA-3'), and P r (5'-ATGCGGCCGCTTAGCTGA CGGTGATGGAA-3'), cloned into the pMD19-T Simple Vector and named T-Xyn2. The Fast Mutagenesis System TM (TransGen Biotech Co., Ltd, Beijing) was used for construction of the Xyn2 mutants. All the mutant primers used in this study are listed in Table 1. PCR to introduce mutations was carried by using pfu DNA polymerase (TransGen Biotech Co., Ltd., Beijing) based on the T-Xyn2 template. The desired PCR products were purified by Gel Extraction Mini Kit (Omega). The purified DNA was digested with 1 μl of DMT enzymes (TransGen Biotech Co., Ltd, Beijing) for 3 h and then transformed into E. coli DH5α. Transformants were selected by using LB agar plates containing 50 μg/mL of ampicillin. A single colony was isolated and inoculated into 3 mL of LB medium containing 50 μg/mL ampicillin, and plasmids were extracted by using the Plasmid Mini Kit (Omega).

Expression of the native and mutated Xyn2 in P. pastoris
The Xyn2 and its mutants, XM1 and XM2, were excised from recombinant T-vectors by double enzyme digestion using Eco RI and Not I. The target fragments were inserted into the yeast expression vector pPICZαA with the same sticky ends, followed by transformation into E. coli DH5α. The recombinant shuttle plasmid pPICZαA was linearized with Sal I and electroporated into P. pastoris. All P. pastoris transformants were cultured and induced in buffered minimal methanol medium (BMMY) at 30˚C for 72 h. The supernatant of each xylanase was dialysed against McIlvaine buffer (0.2 M sodium phosphate, 0.1 M citric acid, pH 6.0). The dialysate was concentrated to 1 mL by ultrafiltration using a 10 kDa cut-off membrane (Sangon Biotech, Shanghai).

SDS-PAGE analysis
The protein fraction (cell culture supernatant) was boiled for 5 min and then applied to a 12% (w/v) SDS-PAGE gel. The proteins were visualized using Coomassie Brilliant Blue R 250 staining. The total protein concentration was determined by the Bradford method using bovine serum albumin (BSA) as a standard.
c All the mutation sites in the primers were highlight with underline. https://doi.org/10.1371/journal.pone.0210548.t001 Thermostability improvement of T. reesei xylanase sodium phosphate, 0.1 M citric acid, pH 6.0) was added to 360 μl of 1% (w/v) beech wood xylan (Sigma) that was suspended with the same buffer and incubated at 50˚C for 10 min. The reducing sugars hydrolysed by xylanase were quantitated by adding 600 μl of DNS and boiling at 100˚C for 10 min. The absorbance of the solution was then measured at 540 nm by using a visible-spectrophotometer. One unit (IU) of xylanase activity was defined as the amount of enzyme that liberated 1 μmol of reducing sugar equivalent per minute under standard conditions (at pH 6.0 and 50˚C for 10 min).

Validation of disulfide bond formation
The formation of disulfide bonds of the recombinant proteins was measured by using the method described by Yang [18]. In brief, the native and mutated enzymes proteins were pretreated with a concentration gradient of 1,4-Dithiothreitol (DTT) ranging from 1-5 mM in the presence of 1% SDS at 70˚C for 5 min, and then applied to a 12% (w/v) SDS-PAGE gel. To determine the influence of disulfide bond introduction on enzyme thermostability, the purified enzymes were pre-treated with 10 mM DDT at 4˚C for 12 h, and then the residue activity of the enzymes was measured after incubation at 60˚C for 10 min.

Glycosylation prediction and validation of recombinant proteins
The glycosylation of Xyn2 and its mutants was predicted by the ExPASy SIB Bioinformatics Resource Portal (http://www.expasy.org/) with the PROSITE program (http://prosite.expasy. org/) based on the amino acid sequences. Moreover, glycosylation of the recombinant proteins was experimentally verified by SDS-PAGE using purified proteins with or without being pretreated with endo-glycosidase (Endo H f ).

Enzymatic property assays
The optimal temperature was measured by performing the xylanase activity assay at temperatures ranging from 30 to 80˚C. The thermostability of the xylanases was measured by pre-incubating the xylanases in 50 mM McIlvaine buffer (pH 6.0) for 10 min at different temperatures (50 to 80˚C), and residue enzyme activity was measured after cooling on ice. Enzymatic assays at different pH values were performed at the optimal temperature over a pH range of 2.0-10.0. The buffers used were 50 mM McIlvaine buffer (pH 2.0-8.0) and 50 mM of a glycine-NaOH buffer (pH 9.0-10.0). To determine the half-lives (t 1/2 ) of the xylanases, the proteins were incubated in McIlvaine buffer (pH 6.0) at 60˚C, and the residue activity was measured at different time intervals (1, 3, 5, 10, 15, 20, and 30 min). The half-life was calculated by using the equation (1. y = A � e -kt ), which was fitted to specific activity-time curves with Excel and the equation t 1/2 = ln2/k. The Km and k cat values were determined by using the method as described by Jiang et al [19].

Identification of the flexible regions in the T. reesei Xyn2
Flexible regions have been recognized as critical unstable factors for proteins, and the modulation of flexible regions has been previously used to improve the protein stability [15]. However, more studies have focused on engineering the protein termini [12,20], since it is difficult to obtain sufficient information about protein molecular structures. Therefore, the N-terminus was usually selected as the target for protein engineering to improve the thermostability of GH 11 xylanases [12,21]. In addition to the N-terminus, other critical domains or regions related to the stability of the GH 11 xylanases are still unknown. Currently, molecular dynamics simulation offers an efficient approach to identify the flexible regions of proteins [22,23]. In this work, the flexible regions of Xyn2 were identified based on the B-factor, molecular dynamics parameter. The B-factor of the amino acids in Xyn2 were visualized by PyMOL with colors ranging from blue to red (Fig 1A). We show that the fragments of a β-sheet (A3/B5) and an αhelix exhibited larger B-factors than those from other regions of Xyn2. The two regions were selected for protein engineering.

Amino acid mutations in the flexible regions of Xyn2
Previous studies indicated that flexible regions in the protein could be stabilized by the intermolecular forces such as the salt bridges, hydrogen bonds, and aromatic interactions [15]. However, covalent disulfide bonds have been considered to be the most important structures responsible for protein stability [24]. In this study, disulfide bonds were engineered into the flexible regions of Xyn2 by site-directed mutagenesis (Fig 1B). A 600 bp DNA fragment was successfully amplified by using a pair of specific primers. The DNA sequences were subsequently aligned by NCBI, and the results indicated that the desired DNA were identical to T. reesei Rut C-30 Xyn2 (Gen Bank Accession No.EU532196.1). The mutated Xyn2 genes (XM1 and XM2) were constructed by using Fast Mutagenesis System. The mutated genes were sequenced, and the results showed that both genes have a sequence similar to that of the Xyn2 gene (S1 Fig). As expected, the mutated amino acids (XM1, Cys 14 -Cys 52 ; XM2, Cys 59 -Cys 149 ) were successfully introduced into the T. reesei Xyn2. Both genes encode a mature, 190-amino acid Xyn2 protein (S2 Fig). The calculated molecular weight (21 kDa) is consistent with the molecular weight of native Xyn2 isolated from T. reesei [16].

Expression of the Xyn2 and its mutants in P. pastoris
The P. pastoris is the most important heterologous expression host strain used in industrial or medical fields because of several advantages such as promoter strength and ease of achieving a high cell density [25,26]. In this study, the native Xyn2 gene and its mutants (XM1 and XM2) were integrated into the genome of P. pastoris X-33 by the pPICZαA shuttle plasmid and successfully expressed in P. pastoris X-33 (Fig 2). A previous study indicated that the molecular weight of the native T. reesei Xyn2 was 21 kDa. However, the Xyn2 enzyme secreted by P. pastoris has a different molecular weight molecular. The molecular weight of Xyn2 secreted by P. Thermostability improvement of T. reesei xylanase pastoris is 24 kDa as estimated by SDS-PAGE. There is a 3 kDa difference in the molecular weights of Xyn2 proteins secreted by P. pastoris and T. reesei. This difference is caused by N-glycosylation of the proteins since the P. pastoris tends to hyperglycosylate heterologous proteins [27][28][29][30]. Treatment of the recombinant Xyn2 protein with endoglycosidase F produced a novel protein species with a molecular weight of 21 kDa (Fig 2), which is consistent with the native Xyn2 secreted by T. reesei. However, this large, glycosylated protein was efficiently secreted and passed through the yeast cell wall into the culture medium. We also performed the glycosylation site analysis by using the PROSITE program (http://prosite.expasy. org/). We found that the mutated amino acids were not involved in the glycosylation sites, suggesting that glycosylation did not affect the formation of disulfide bonds in mutated Xyn2. However, the glycosylation seemed to affect the activity of the enzymes, and the enzymatic activity of the recombinant proteins was decreased after deglycosylation (Fig 3). This finding is consistent with previous studies showing that glycosylation is required for the full activity of a wide variety of enzymes [31,32]. The influence of glycosylation on protein properties may be attributed to several functions such as ensuring correct protein folding, preventing proteolytic degradation, and facilitating intracellular transportation [31]. Moreover, glycosylation is one of the greatest advantages of the yeast expression systems [25,26].
To validate the formation of the engineered disulfide bonds, the purified enzymes were pretreated with DDT before SDS-PAGE. The presence of disulfide bonds changes the mobility of proteins, and proteins without disulfide bonds tended to bind more SDS, resulting in a decreased migration speed compared to that of proteins with intact disulfide bonds [18]. As shown in Fig 4A,  Thermostability improvement of T. reesei xylanase both the XM1 and XM2 moved slowly on the gels after DDT treatment, indicating that disulfide bonds were successfully introduced into XM1 and XM2. Interestingly, the introduction of disulfide bonds affected the activity of Xyn2 and abolishing the disulfide bonds by using DDT resulted in a significant reduction in its activity (Fig 4B). This finding is consistent with a previous report that disulfide bonds can serve as a switch for protein functions and introducing disulfide bonds to a functional protein may increase both its stability and activity [33].

Effect of amino acid mutation on thermostability and pH stability
Currently, there are two approaches to obtain thermostable xylanases. The first is to discover enzymes from thermophilic microorganisms, and the second is to engineer currently used mesophilic xylanases to create novel enzymes that can withstand the harsh conditions [4]. However, the isolation of thermostable xylanases is a very tedious and difficult task because of their low expression levels. Previous studies indicated that the covalent disulfide bonds played a critical role in stabilizing the flexible parts of proteins, and introducing extra disulfide bonds has successfully improved the thermostability of many protein species [17,34]. However, most studies engineered disulfide bonds on the protein surface or at the N and C terminus. In the present study, we identified an unstable flexible region of Xyn2, and disulfide bonds were introduced to stabilize this region. Compared to native Xyn2, the optimal temperature for the two mutants changed from 50˚C to 55˚C (Fig 5A) and both proteins retained more activity after incubation at this temperature than native Xyn2. Interestingly, native Xyn2 lost nearly all activity after incubation at 70˚C for 10 min. However, both mutated proteins retained approximately 30% activity ( Fig 5B). Moreover, the half-lives of XM1 and XM2 at 60˚C increased 2.5and 1.8-fold, respectively ( Table 2). Both of their half-lives at this temperature were higher than those from previous reported enzymes engineered by Wakarchuk et al [17]. The N-terminus of the GH 11 family xylanases is recognized as a thermostable domain because of its crucial role in stabilizing the β-core, and unfolding of the N-terminus caused direct exposure of the β-core to thermal or acidic conditions [12,13]. Moreover, an α-helix also plays a critical role in stabilizing the GH 11 family xylanases, and the introduction of disulfide bonds to this α-helix significantly increased the stability of xylanases from Bacillus circulans [17]. In this study, the identified unstable region involves both the β-core and the α-helix, and the introduction of disulfide bonds stabilized this region, which led to improvements in the thermostability of T. reesei Xyn2. Interestingly, the introduction of disulfide bonds did not affect the optimal pH of Xyn2 (Fig 5C). However, the XM1 showed a broader pH range than native Xyn2 and XM2, and it retained more than 60% activity at a pH range of 2.0 to 10.0 ( Fig  5D). The result is consistent with a previous report in which introducing disulfide bonds decreased pH sensitivity by reducing the electronic density on the surface of the protein [18]. The thermostability of XM1 and XM2 was lower than several xylanases naturally secreted by thermophilic microorganisms such as Thermotoga maritima and Thermomyces lanuginosus [19,35]. However, the P. pastoris heterologous expression system allows higher expression levels than expression in natural microorganisms [36].

Effect of amino acid mutations on enzymatic activity and kinetic parameters
Changes in the amino acids or conformation of an enzyme protein may result in altered properties such as enzymatic activity and kinetic parameters [37]. The catalytic domains of GH 11 family xylanases consist principally of β-pleated sheets formed into a two-layered trough that surrounds the catalytic site [38,39]. Additionally, the N-terminus of T. reesei xylanases is known to be associated with their thermostability and catalytic activity [27]. In the present study, disulfide bond introduction at the N-terminus (Cys 14 -Cys 52 ) decreased the specific activity of the enzyme (Table 1). Both the K m (15.4 mg/ml) and k cat (179.2 s -1 ) of XM1 were significantly higher than those of native Xyn2. In contrast, fixing the α-helix to the β-core through a disulfide bond between β-sheets B5 and B6 (Cys 59 -Cys 149 ) had little influence on the catalytic efficiency of Xyn2. The decreased catalytic efficiency of XM1 may be associated with the substrate binding capacity since previous studies indicated that the N-terminus of GH 11 family xylanases plays a critical role in substrate binding because of the presence of charged amino acid residues [14,40]. The electronic charges on the surface of the N-terminus may be covered by the introduced disulfide bonds.

Conclusions
In this study, we successfully introduced disulfide bonds into the unstable flexible regions of T. reesei Xyn2 by using site-directed mutagenesis, and two mutated xylanases were successfully expressed in Pichia pastoris. Introduction of the disulfide bonds resulted in elevated thermostability and pH stability in the T. reesei Xyn2. However, disulfide bonds introducing at the Nterminal may affect the catalytic efficiency. Our results not only identified a novel unstable region in the T. reesei xylanases but also offer a potential avenue to improve the stability of the enzyme.