Structural Analysis of a Glycoside Hydrolase Family 11 Xylanase from Neocallimastix patriciarum

Background: Thermophilic xylanases are valuable in many industrial applications.

Results: The structures of a xylanase XynCDBFV and its complex with xylooligosaccharides were determined, and its N-terminal region (NTR) contributes to thermostability.

Conclusion: NTR may stabilize the overall protein folding of XynCDBFV.

Significance: The structural and functional investigation of unprecedented NTR of XynCDBFV provides a new insight into the molecular basis of thermophilic xylanases.

Abstract

The catalytic domain of XynCDBFV, a glycoside hydrolase family 11 (GH11) xylanase from ruminal fungus Neocallimastix patriciarum previously engineered to exhibit higher specific activity and broader pH adaptability, holds great potential in commercial applications. Here, the crystal structures of XynCDBFV and its complex with substrate were determined to 1.27–1.43 Å resolution. These structures revealed a typical GH11 β-jelly-roll fold and detailed interaction networks between the enzyme and ligands. Notably, an extended N-terminal region (NTR) consisting of 11 amino acids was identified in the XynCDBFV structure, which is found unique among GH11 xylanases. The NTR is attached to the catalytic core by hydrogen bonds and stacking forces along with a disulfide bond between Cys-4 and Cys-172. Interestingly, the NTR deletion mutant retained 61.5% and 19.5% enzymatic activity at 55 °C and 75 °C, respectively, compared with the wild-type enzyme, whereas the C4A/C172A mutant showed 86.8% and 23.3% activity. These results suggest that NTR plays a role in XynCDBFV thermostability, and the Cys-4/Cys-172 disulfide bond is critical to the NTR-mediated interactions. Furthermore, we also demonstrated that Pichia pastoris produces XynCDBFV with higher catalytic activity at higher temperature than Escherichia coli, in which incorrect NTR folding and inefficient disulfide bond formation might have occurred. In conclusion, these structural and functional analyses of the industrially favored XynCDBFV provide a molecular basis of NTR contribution to its thermostability.

Introduction

Xylans are the major hemicellulose components in the plant cell wall and account for nearly one-third of all renewable organic carbon source on earth (1, 2). Xylans are heteropolysaccharides composed of β-1,4-glycosidic bond-linked xylose units as a backbone chain that is usually decorated by different side groups such as methyl group, acetyl group, and other sugar molecules (1, 3). Due to its structural complexity, a set of enzymes is required for complete xylan decomposition, including endo-1,4-β-xylanase, β-xylosidase, acetylxylan esterase, arabinase, and α-glucuronidase (3, 4). Among them, the glycoside hydrolase endo-1,4-xylanase (xylanase, EC 3.2.1.8) is the key enzyme that catalyzes random hydrolysis of the xylan backbone to small fragments by cleaving the β-1,4-glycosidic bonds (1). Xylanases have been widely applied in many industries such as feed manufacture, paper and pulp processing, and food industry (5–7). These biotechnological treatments usually involve harsh conditions and demand enzymes with good thermostability, broad pH adaptability, and high specific activity. Therefore, numerous research projects have been carried out in search of novel xylanases with favorable properties and to improve performance of available enzymes via direct evolution or rational design approaches.

To date, various xylanases have been identified from bacteria, fungi, yeasts, plants, and insects (8) and are classified into the glycoside hydrolase (GH)³ family 5, 8, 10, 11, 30, and 43 according to their protein sequence similarities by CAZy database, with a majority belonging to GH10 and GH11. In terms of biotechnological applications, GH11 xylanases have drawn much attention due to their small sizes, high substrate selectivities, and wide ranges of pH and temperature adaptability (9). To date nearly a thousand GH11 xylanases have been characterized from various species, but only 26 structures from bacteria and fungi have been solved (CAZy database). Among them, several are thermophilic/thermostable enzymes, including Dictyoglomus thermophilum XynB (optimal temperature 75 °C) (10), Thermobifida fusca NTU22 Xyl11 (optimal temperature 70 °C) (11), and Thermopolyspora flexuosa XynA (optimal temperature 80 °C) (12) from bacteria and Chaetomium thermophilum Xyn11A (optimal temperature 80 °C) (13) from fungus.

Previously, a GH11 xylanase from an anaerobic ruminal fungus, Neocallimastix patriciarum, was isolated and characterized (14, 15). The catalytic domain of the enzyme (Xyn-CD) exerts optimal activity at pH 6.0, and an alkalophilic mutant XynCDBFV with seven amino acid substitutions was later created via direct evolution using error-prone PCR (16). The recombinant XynCDBFV protein is among the highest active xylanases and possesses an optimal temperature of 65 °C, broad pH adaptability, and remarkable tolerance at pH 10.0 when expressed in E. coli. Therefore, the XynCDBFV is an attractive candidate to be developed as an industrial product. In the present study the recombinant XynCDBFV was expressed in the industrial strain P. pastoris and crystallized. The overall protein fold and ligand complex structure are analyzed in detail. Based on these data, potential factors contributing to the enzyme thermostability are proposed.

EXPERIMENTAL PROCEDURES

Gene Cloning and Mutagenesis

The synthesized gene encoding XynCDBFV, an engineered mutant of Xyn-CD from N. patriciarum (GenBank^TM accession number AF123252) (16), was amplified by using PCR with a forward primer of 5′-CCCGAATTCCAAAGTTTCTGTAGTTCAGCTTCT-3′ and a reverse primer of 5′-CCCGCGGCCGCTTAATCACCAATGTAAACCTTTGCGTA-3′. The gene was then cloned into the vector pPICZαA for the P. pastoris system by EcoRI and NotI to yield pPICZαA/xynCDBFV. The substituted mutants including E109A, C4A, C172A, and C4A/C172A were prepared by using the QuikChange site-directed mutagenesis kit (Agilent) with pPICZαA/xynCDBFV as the template. The genes encoding the deleted mutants of Δ6 (deletion of Gln-1–Ser-6) and Δ11 (deletion of Gln-1–Gly-11) were generated by PCR with full-length xynCDBFV gene as the template. These truncated genes were then cloned into the vector pPICZαA by using EcoRI and NotI to yield pPICZαA/xynCDBFV-Δ6 and pPICZαA/xynCDBFV-Δ11. The sequences of the mutated primers are listed in the supplemental Table S1.

Alternatively, the xynCDBFV gene was amplified by using PCR and cloned into the vector pET32 Xa/LIC for E. coli expression system. This vector has designed a His tag before the N terminus of targeted gene for purification purpose. The specific primers used here were 5′-GGTATTGAGGGTCGCCAAAGTTTCTGTAGTTCAGCT-3′ (forward) and 5′-AGAGGAGAGTTAGAGCCTTAATCACCAATGTAAACCTTTGC-3′ (reverse).

Protein Expression and Purification

These above plasmids, except pET32 Xa/LIC-xynCDBFV, were linearized by PmeI and then individually transformed into X33 strain of P. pastoris by electroporation. The transformants were selected on the YPD (yeast extract peptone dextrose) plates containing 100 μg/ml zeocin (Invitrogen). The selected clones were inoculated and amplified in 50 ml of buffered glycerol-complex medium (BMGY; 1% yeast extract, 2% peptone, 100 mm potassium phosphate, pH 6.0, 1.34% yeast nitrogen base (YNB) with ammonium sulfate without amino acids, 4× 10–5% biotin, and 1% glycerol) at 30 °C for 1 day. Then the culture medium of cultured cells was replaced by 20 ml of buffered methanol-complex medium (BMMY; 1% yeast extract, 2% peptone, 100 mm potassium phosphate, pH 6.0, 1.34% yeast nitrogen base (YNB) with ammonium sulfate without amino acids, 4× 10–5% biotin, and 0.5% methanol) to induce protein expression. For protein purification, the supernatants were collected and dialyzed twice against the buffer containing 25 mm Tris, pH 7.5. In addition, the proteins were treated by endoglycosidase H (New England Biolabs) during the dialysis procedure. The proteins were then purified by FPLC system using diethylaminoethyl (DEAE) column (GE Healthcare) and eluted using a linear gradient of 0–250 mm NaCl in the buffer containing 25 mm Tris, pH 7.5. The purified proteins were finally concentrated to 10 mg/ml in 25 mm Tris, pH 7.5, 150 mm NaCl by using Amicon Ultra-15 Centrifugal Filter Units (Millipore), and the purity (>95%) was checked by SDS-PAGE.

On the other hand the pET32 Xa/LIC-xynCDBFV plasmid was transformed into BL21 (DE3) strain of E. coli. Then the transformed cells were propagated and induced by adding isopropyl 1-thio-β-d-galactopyranoside for protein expression. The protein was purified by FPLC system using a nickel nitriloacetic acid column. The His-tagged protein was eluted using a gradient of 0–250 mm imidazole in the buffer containing 25 mm Tris, pH 7.5, 150 mm NaCl. The eluted protein was then dialyzed against the buffer containing 25 mm Tris, pH 7.5, 150 mm NaCl, with Factor Xa enzyme to remove the His tag. The mixture was loaded onto another nickel nitriloacetic acid column, and the untagged protein was eluted in the buffer without imidazole for purification of untagged protein. The eluted protein was then dialyzed against the buffer containing 25 mm Tris, pH 7.5, for the DEAE column. Next, the purified protein was eluted using a gradient of 0–250 mm NaCl in the 25 mm Tris, pH 7.5, for the DEAE column. The purified proteins were finally concentrated to 10 mg/ml in 25 mm Tris, pH 7.5, 150 mm NaCl, and the purity was checked by SDS-PAGE.

Crystallization and Data Collection

The XynCDBFV and E109A proteins were crystallized by using sitting-drop vapor diffusion method and Crystal Screen kit (Hampton Research). The XynCDBFV protein crystals were obtained from the reservoir solution containing 0.1 m Tris, pH 8.5, and 2 m ammonium sulfate at room temperature for 1 day. The E109A protein crystals were obtained from the reservoir solution containing 0.1 m sodium cacodylate, pH 6.5, 0.2 m ammonium sulfate, 26% PEG8000, and 5% glycerol at room temperature for 2 days. The hexagonal E109A-xylotriose-bound crystal was prepared by soaking with 10 mm xylotriose (Megazyme) in the reservoir solution for 1 h. The cryoprotectants for XynCDBFV and E109A crystals contained 0.1 m Tris, pH 8.5, 2 m ammonium sulfate and 10% glycerol and 0.12 m sodium cacodylate, pH 6.5, 0.24 m ammonium sulfate, 31% PEG8000, and 5% glycerol, respectively. All of the x-ray diffraction data were collected at beam line BL13B1 of National Synchrotron Radiation Research Center in Hsinchu, Taiwan. The diffraction images were processed by using HKL2000 (17).

Structure Determination and Refinement

The crystal structure of XynCDBFV was solved by the molecular replacement method with the program Phaser (18) from the CCP4 suite (19) using the hypothetical XynCDBFV model generated from the structure of Bacillus subtilis B230 Xyn11X (PDB code 1IGO; 47% sequence identity with XynCDBFV) by the SWISS-MODEL website (20, 21) as a search model. Subsequent model building and structural refinement were carried out by using the programs COOT (22) and REFMAC5 (23), respectively. The complex structure of E109A-xylotriose was determined by the molecular replacement method with Phaser using refined XynCDBFV structure as a search model. The structural refinements were finished by the programs COOT (22) and REFMAC5 (23). Some data collection and statistics are summarized in Table 1. All of the structural diagrams were drawn by using PyMOL (24).

TABLE 1.

Data collection and refinement statistics for the XynCDBFV

Values in parentheses are for the outermost resolution shells

	XynCDBFV	E109A	E109A-XTI
Data collection
Wavelength (Å)	1.00000	1.00000	1.00000
Resolution (Å)	25.00-1.27 (1.32-1.27)	25.00-1.32 (1.37-1.32)	25.00-1.43 (1.48-1.43)
Space group	P64	P64	P64
Unit-cell a, b, c (Å)	92.8, 92.8, 42.9	92.9, 92.9, 43.1	93.2, 93.2, 43.4
No. of unique reflections	55625 (5510)	50174 (5000)	39995 (3947)
Redundancy	8.8 (8.7)	9.1 (8.9)	9.1 (9.1)
Completeness (%)	99.8 (99.9)	100.0 (100.0)	99.9 (100.0)
Mean I/σ(I)	49.4 (7.9)	44.8 (3.1)	37.8 (3.1)
Rmerge (%)	5.0 (20.4)	4.3 (46.7)	5.2 (49.2)
Refinement
No. of reflections	52832 (4096)	47540 (3646)	37972 (2875)
Rwork (95% of data)	13.9 (14.7)	14.7 (21.2)	12.8 (19.1)
Rfree (5% of data)	16.6 (17.1)	18.3 (25.4)	16.0 (23.4)
r.m.s.d. bonds (Å)	0.013	0.014	0.013
r.m.s.d. angles (°)	1.65	1.74	1.64
Ramachandran plot (%)
Most favored (%)	98.7	99.1	98.7
Allowed (%)	1.3	0.9	1.3
Disallowed (%)	0.0	0.0	0.0
Mean B (Å2) /atoms
Protein	10.8/1747	12.0/1738	15.6/1743
Water	30.9/332	29.6/351	35.7/273
Ion	19.3/5
Ligand			26.5/47
PDB ID code	3WP4	3WP5	3WP6

Xylanase Activity Assay

The detection of xylanase activity was based on the determination of reducing sugar by using dinitrosalicylic acid method (25, 26). In general, the protein solution in 50 mm sodium acetate buffer, pH 5.3, with a proper concentration was mixed with 1% xylan substrate (beechwood, Sigma) in the proportion of 1 to 9 and then incubated at 55 °C for 10 min. The reaction was stopped by adding 1% dinitrosalicylic acid solution and incubated at 100 °C in boiling water for 10 min. The absorbance of A_{540 nm} was measured for calculation of the enzyme activity. The standard curve for calibrating the enzyme activity was determined by 0–0.6 mg/ml xylose solutions. One unit of activity is defined as the amount of enzyme that releases 1 μmol of reducing sugar equivalent to xylose per minute.

RESULTS

Overall Structure

The crystal structure of XynCDBFV was solved to a resolution of 1.27 Å by molecular replacement using the Xyn11X from B. subtilis B230 as a search model. The data collection and refinement statistics of the wild type, E109A mutant (<0.2% activity, data not shown), and its complex with xylotriose (XTI, E109A-XTI) structures are listed in Table 1. The root mean square deviations (r.m.s.d.) of Cα atoms between XynCDBFV and the E109A mutant is 0.374 Å, indicating that the mutation did not cause a significant alteration in the protein structure. To investigate whether the ligand binding could cause significant conformational change, the ligand-free (XynCDBFV and E109A mutant) and ligand-bound (E109A-XTI) structures were superimposed. The r.m.s.d. between these structures ranged from 0.192 to 0.436 Å for all Cα atoms, suggesting that the ligand binding did not result in significant conformational change.

As shown in Fig. 1A, the protein structure of XynCDBFV displays a β-jelly-roll fold, which is typical of GH11 family enzymes. It comprises two anti-parallel β-sheets, linked by interconnecting loops and α-helices. The topology of GH11 enzymes is also described as the shape of a right hand (27). The two β-sheets mimic the palm and fingers, whereas the loop β12-β13 is like a thumb. As shown in Fig. 1B, there are two disulfide bridges in the XynCDBFV structure, including DS1 (Cys-4–Cys-172), which connects the N-terminal α1 helix to the strand β14, and DS2 (Cys-50–Cys-60), which joins the strand β5 to the strand β6. The tunnel-like active site cleft (Fig. 1C) is formed by the curved inner β-sheet consisting of β2, β3, β6, β15, β8, β9, β10, β12, and β13 strands (Fig. 1, A and C). The catalytic residues Glu-109 and Glu-202 predicted from sequence alignment of GH11 xylanases were found embedded in this region and located on the strand β10 and β15, respectively (Fig. 1, A and D).

FIGURE 1.

The overall structure of ligand-bound XynCDBFV. A, the schematic representation of XynCDBFV is shown with different colors for α-helices (red), β-strands (yellow), and loops (green). The xylotriose (XTI; purple) and xylobiose (XBI; salmon) are taken from the E109A-XTI complex structure and shown as sticks. The side chains of catalytic residues Glu-109 and Glu-202 are shown as pink sticks. B, the model is the same as panel A but is rotated by 90 degrees about a horizontal axis. Two disulfide bonds, DS1 (Cys-4–Cys-172) and DS2 (Cys-50–Cys-60), are shown as sticks with cyan carbons and yellow sulfurs. C, a surface representation of the ligand-bound XynCDBFV is shown with electrostatic potentials colored red for negative charge and blue for positive charge. The representation of the bound ligand is as in panel A. D, a close-up view of panel C. The side chains of Glu-109 and Glu-202 are shown as yellow sticks.

Substrate Binding Site

The electron density maps in the tunnel-like cleft of the E109A-XTI structure clearly indicated the presence of three β-1,4-linked xylosyl moieties in the +1 to +3 subsites and two in the −2 to −3 subsites (Fig. 2A). The sugar molecules were modeled into the wild-type XynCDBFV to display the detailed interactions between the substrates and the surrounding residues (Fig. 2B). Trp-32 and Trp-125 provide stacking forces to sugar units in the subsites −2 and +3, respectively. Trp-32 further forms a hydrogen bond to the hydroxyl group of the −3 sugar. Glu-30, Arg-61, and Tyr-100 form hydrogen bonds to the −2 sugar unit. Asp-57, Tyr-94, Tyr-111, Arg-148, Gln-161, and Glu-202 (one of catalytic residues) form hydrogen bonds to sugar moiety in the +1 subsite. Arg-148 also forms a hydrogen bond to the +2 sugar unit. The −1 sugar moiety was not observed from our crystal structure, but the subsite can be predicted from the superimposition of the XynCDBFV-XTI model and the complex structure of XynII from Trichoderma reesei (PDB code 4HK9) (28). The superimposition indicates that several residues might participate in the −1 subsite formation including Asp-57, Arg-148, Pro-151, and Gln-161 along with two catalytic residues, Glu-109 and Glu-202 (Fig. 2C). Among these ligand-interacting residues, Trp-32, Tyr-100, Tyr-111, Gln-161, and Pro-151 are strictly conserved in GH11 xylanase, indicating their vital roles in catalytic reactions.

FIGURE 2.

The electron density maps of ligands and the detailed interaction networks in the active site of XynCDBFV. A, the refined oligosaccharide models of E109A-XTI are superimposed on the 2F_o − F_c electron density maps, which are contoured at 1 σ and 2.5 σ levels and shown in cyan and red, respectively. B, a stereo view of detailed interactions between the active-site residues of XynCDBFV and ligands is shown with side chains of amino acids as thin green sticks and sugar moieties as gray thick sticks. The dashed lines denote potential hydrogen bonds between the enzyme and ligands. C, a stereo view of detailed interactions of active site residues and ligands in the glycone subsites. XynCDBFV-XTI model and T. reesei xylanase complex structure (PDB code 4HK9) are superimposed and the xylobiose (XBI) from the former and XTI from the latter are shown as stick models and colored in gray and salmon, respectively.

The Unique N-terminal Region of XynCDBFV

From protein sequence alignment and structure superimposition, we found that XynCDBFV carries an extended N-terminal region (NTR) that is unique in the GH11 family (Fig. 3, A and B). The NTR is composed of 11 amino acids and spans the convex side of the palm β-sheet, whereas other GH11 enzymes start from the strand β1 or β2 (Fig. 3B). There are several interactions formed between the NTR and the nearby β-strands including β4, β5, β14, β16, and β17 (Fig. 3C). Two stacking interactions are formed between NTR-Phe-3 and β17-Tyr-218 and between NTR-His-9 and β4-Tyr-43 (Fig. 3C). NTR-His-9 is also at a hydrogen bonding distance to β5-Ser-47 and β16-Asp-215. Notably, NTR-Cys-4 forms the DS1 disulfide bond with β14-Cys-172. Through these interactions, the NTR is stably attached to the catalytic core of XynCDBFV instead of hanging freely as a flexible segment.

FIGURE 3.

The unique feature of NTR. A, sequence alignment of the N-terminal region of XynCDBFV and several GH11 xylanases including PDB codes 1IGO (B. subtilis B230), 2DCJ (Bacillus sp. 41M-1), 1H4H (Bacillus agaradhaerens), 2NQY (Bacillus sp. NCL 87-6-10), 1F5J (D. thermophilum), 2C1F (N. patriciarum), 1M4W (T. flexuosa), 1HIX (Streptomyces sp. S38), 3ZSE (T. fusca), 1PVX (P. varioti Bainier), and 1H1A (C. thermophilum) was performed with ClustalW (47). The secondary structural elements (α-helices and β-strands) of XynCDBFV are shown above the sequences. Strictly conserved residues are highlighted by a red background, and conservatively residues are boxed. The figure was produced by ESPript (48). B, the structure of XynCDBFV is superimposed with other crystal structures of GH11 xylanases. The colors used here are red for XynCDBFV, cyan for 1IGO, magenta for 2DCJ, yellow for 1H4H, light pink for 1F5J, green for 2C1F, blue for 1M4W, purple for 1PVX, orange for 1H1A, dark green for 2NQY, pink for 1HIX, and light yellow for 3ZSE. C, the side chains of residues participating in the detailed interactions of NTR and catalytic core of XynCDBFV are shown as thin sticks and colored in blue and green, respectively. The disulfide bond, labeled as DS1, is shown as thick sticks in yellow. The dashed lines denote potential hydrogen bonds.

Functional Analysis of the NTR

To investigate the functional significance of NTR, we constructed the NTR-truncated mutants, including Δ6 (Gln-1–Ser-6 deletion) and Δ11 (Gln-1–Gly-11 deletion) (Fig. 4A), and analyzed their performance. As shown in Fig. 4B, the enzymatic activities of mutants Δ6 and Δ11 were reduced to 80.9 and 61.5% at 55 °C, respectively, compared with the wild-type protein. Remarkably, the variations between wild-type and mutant enzymes were even larger at higher temperatures. As the wild-type XynCDBFV showed enhanced activity at 65 °C (140%) and 75 °C (151.8%), the relative activities of Δ6 and Δ11 were reduced to 33.7 and 20.7% at 65 °C and 21 and 19.6% at 75 °C (Fig. 4B). These results clearly demonstrated that the NTR plays a role in the catalytic activity of XynCDBFV, and its presence is especially important for the enzyme to exert thermophilic functions. As mentioned, there is a disulfide bond DS1 formed between the NTR and strand β14. To further investigate the role of the DS1 linkage, mutants C4A, C172A, and C4A/C172A were constructed. Similar to the results of the NTR-truncated mutants, the DS1-removed mutants showed lower activities comparing to the wild-type enzyme at 55 °C, and their activities further declined when the tested temperature was increased to 65 °C (39.3–44.8%) and 75 °C (20.8–23.3%) (Fig. 4C). Protein expression levels of all mutants were similar to that of wild type (data not shown). These results suggested that the NTR is important for the catalytic activity and is an essential element for thermophilicity of XynCDBFV, and its function is highly dependent on the presence of disulfide bond DS1.

FIGURE 4.

Enzymatic activity analysis of wild-type and mutant XynCDBFV. A, schematic and schematic diagrams of the NTR-truncated mutants. The disulfide bond (Cys-4–Cys-172) is shown as yellow sticks. B and C, the culture supernatant of P. pastoris expressing wild-type XynCDBFV was tested for activity and compared with those of NTR-truncated mutants (B) and DS1-removed mutants (C). The enzymatic activity was determined at the indicated temperatures, and the relative activity of each sample was presented as a percentage of the value of wild-type XynCDBFV at 55 °C.

In the beginning we tried to use E. coli-expressed protein (E-XynCDBFV) for structural study because the E. coli expression system is easier to handle than P. pastoris expression system. The E-XynCDBFV was easily expressed and purified but failed to be crystallized despite the protein being catalytically active (Table 2). When the purified proteins were subjected to SDS-PAGE analysis, the E-XynCDBFV formed a significant amount of dimer and trimer in the absence of reducing agent, suggesting the presence of one or more free Cys residues on the protein surface (Fig. 5). Reducing agent was supplemented during E-XynCDBFV crystallization to prevent formation of nonspecific disulfide bonds, but no crystal was obtained. However, the crystal structure solved in this study indicated that the four Cys residues in XynCDBFV form two pairs of disulfide bonds, and there is no free thiol group exposed to the bulk solvent. Indeed, the recombinant P-XynCDBFV, which was successfully crystallized, showed much less tendency of dimer formation (Fig. 5).

TABLE 2.

Specific activity of XynCDBFV expressed in E. coli and P. pastoris

	Specific activity at pH 5.3
	55 °C	65 °C	75 °C
	units/mg
E. coli	4009.7 ± 115.6 (82 ± 2.4%)	5747.6 ± 270.2 (117.5 ± 5.5%)	5778.3 ± 73.0 (118.2 ± 1.5%)
P. pastoris	4889.9 ± 218.3 (100 ± 4.5%)	7611.6 ± 382.4 (155.7 ± 7.8%)	7995.3 ± 178.6 (163.5 ± 3.7%)

FIGURE 5.

Protein analysis of E-XynCDBFV, P-XynCDBFV, and C4A/C172A mutant. Recombinant proteins of wild-type and mutant XynCDBFV that were expressed and purified in different hosts were subjected to SDS-PAGE analysis. The empty and filled arrowheads indicate the XynCDBFV dimer and trimer, respectively. Locations of DS1 and DS2 disulfide bonds in the XynCDBFV structure are indicated in the right panel.

Intriguingly, the specific activity of E-XynCDBFV was lower than that of P-XynCDBFV by 20% at 55 °C and only slightly increased at higher experimental temperature (up to 18% increase), whereas the P-XynCDBFV showed a 55.7 and 63.5% increase in specific activity at 65 and 75 °C, respectively (Table 2). These observations were reminiscent of the results of analyzing NTR-truncated and DS1-removed mutants. Therefore, we suspect that the XynCDBFV was misfolded in E. coli to a certain extent so that the Cys residues were exposed. The disruption of the DS2 linkage is rather unlikely because it is embedded in the protein interior and its exposure should cause deleterious effects to the enzyme. It would be more likely that the NTR was incorrectly folded and disrupted DS1 formation. The observation that DS1-removed mutant C4A/C172A of P-XynCDBFV hardly formed a dimer may support our hypothesis (Fig. 5). In conclusion, the failure of obtaining the E-XynCDBFV crystals and the lower activity at higher temperatures might be attributed to the inefficient formation of the DS1 linkage in the E. coli system, and P. pastoris should be a more ideal system to express XynCDBFV protein with better performance.

DISCUSSION

Thermostability is one of the most valued properties for industrial enzymes; thus, investigating the strategies utilized by thermophilic enzymes is an intense area of research. XynCDBFV exhibits remarkably high specific activity, which is appreciated for a thermophilic enzyme and is worthy of further investigations. Here the crystal structure of XynCDBFV was solved at high resolution, revealing an unprecedented NTR element that is functionally related to enzymatic activity and thermophilicity. The underlying cause of NTR-mediated thermophilicity is thus an intriguing issue to address. Previous studies have found correlations between the thermostability/thermophilicity of GH11 enzymes and the increased intramolecular interactions including hydrogen bonding, salt bridges, disulfide bonds, and hydrophobic forces in them (10–13, 29, 30). Although various kinds of interactions were proposed to contribute to enzyme thermostability, it seems that making the protein fold into a more compact state is a fundamental principle. Notably, heat-mediated unfolding of a GH11 enzyme is proposed to initiate from the N-terminal region (31). Therefore, introducing an N-terminal disulfide bond to stabilize the first β-strand has been reported as a useful strategy to enhance enzyme thermostability (32–35). In XynCDBFV, the NTR adheres to several β-strands and spans the back of the outer β-sheet. The Cys-4–Cys-172 DS1 bond on the most N-terminal helix α1 serves as a molecular rivet to fix the NTR to strand β14. Through these interactions, the NTR clamps the catalytic core into a more restrained architecture, which may prevent its destabilization at higher temperatures. This is the first report to reveal an extended NTR feature and its function in the GH11 family, and it shall be interesting to examine whether this element could also provide benefits when incorporated into other GH11 enzymes.

Among the 26 crystal structures of GH11 xylanases, only 6 possess one single disulfide bond (8). Three of them are in an equivalent position that connects the cord and strand β12, including the xylanases from Aspergillus kawachii (36), Aspergillus niger (37), and Scytalidium acidophilum (38). For xylanases from Paecilomyces varioti Bainier (30) and Thermomyces lanuginosus (12), the disulfide bond is located between helix α4 and strand β11. Another xylanase from an uncultured bacterium forms the disulfide bond between the N-terminal strands β1 and β4 (39). The XynCDBFV structure in this study, revealing two disulfide bridges DS1 (Cys-4–Cys-172) and DS2 (Cys-50–Cys-60), is the first GH11 structure that includes multiple disulfide bonds in its catalytic domain. A recent study inferred that XynCDBFV has only one disulfide bond formed between Cys-4 and Cys-172 based on the thiol titration analysis (40). The discrepancy might be a result of different experimental approaches. In the present study the clear electron density map has demonstrated the linkage between Cys-50 and Cys-60 and thus validates the existence of DS2 bond in the XynCDBFV structure. Notably, removing the DS2 bond from XynCDBFV did not significantly affect its thermostability (40). These findings suggest different roles played by DS1 and DS2 in XynCDBFV.

The N-terminal deletion experiments showed the Δ11 mutant has lower activity than the Δ6 mutant at 55 and 65 °C, raising the possibility that one or more residues between Ala-7 and Gln-12 in NTR may also be involved in interactions for XynCDBFV thermostability aside from the DS1 disulfide bond. The most possible candidate is His-9, which provides hydrogen bonding to Ser-47 and Asp-216 as well as stacking force to Tyr-43. The forces provided by His-9 are weaker than the covalent disulfide linkages and are more readily dissociated at higher temperatures, which may explain the similar enzymatic activities exhibited by Δ6 and Δ11 at 75 °C (Fig. 4B). This assumption needs to be further validated in the future.

Our results clearly demonstrated that the P. pastoris system is superior to the E. coli system in obtaining XynCDBFV proteins with better performance. Since it was developed, P. pastoris has become a very useful heterologous expression system for both industrial and academic purposes that numerous soluble enzymes with higher activities and better thermal profiles were produced with this system (41–43). The P. pastoris is particularly attractive in making eukaryotic proteins because yeast cells provide correct post-translational modifications such as phosphorylation, glycosylation, and disulfide bond formation (44, 45). XynCDBFV possesses two putative N-glycosylation sites, and P-XynCDBFV showed a glycosylated pattern (data not shown). But the possibility of glycosylation-mediated thermostability is excluded by the fact that the enzymatic performance was not influenced by deglycosylation treatments (the parental proteins in Fig. 4 versus the deglycosylated protein in Table 2). Instead, the correct folding of NTR and the DS1 linkage formation are likely responsible for the different thermal profiles of E-XynCDBFV and P-XynCDBFV.

Previous and present studies show that utilizing the E. coli expression system yields E-XynCDBFV protein that exhibits its highest activity at no more than 65 °C and is less active than the P-XynCDBFV at every temperatures tested (16). Similar phenomenon was observed in our previous study that analyzed the structure of rhodostomin, which contains six pairs of disulfide bonds. NMR examinations indicated that the E. coli-produced rhodostomin was misfolded despite being capable of inhibiting platelet aggregation (46). In contrast, P. pastoris produced correctly folded rhodostomin with all of six intramolecular disulfide pairings formed, which also exhibited higher activity than that from E. coli. Consequently, choosing a suitable production host is critical for XynCDBFV to exhibit its optimal performance, and the utilization of an industrial strain shall also benefit its further commercial applications. Nevertheless, the enhanced thermostability of XynCDBFV requires the presence of NTR in the first place.