Distinct Roles for Carbohydrate-Binding Modules of Glycoside Hydrolase 10 (GH10) and GH11 Xylanases from Caldicellulosiruptor sp. Strain F32 in Thermostability and Catalytic Efficiency

Abstract

Xylanases are crucial for lignocellulosic biomass deconstruction and generally contain noncatalytic carbohydrate-binding modules (CBMs) accessing recalcitrant polymers. Understanding how multimodular enzymes assemble can benefit protein engineering by aiming at accommodating various environmental conditions. Two multimodular xylanases, XynA and XynB, which belong to glycoside hydrolase families 11 (GH11) and GH10, respectively, have been identified from Caldicellulosiruptor sp. strain F32. In this study, both xylanases and their truncated mutants were overexpressed in Escherichia coli, purified, and characterized. GH11 XynATM1 lacking CBM exhibited a considerable improvement in specific activity (215.8 U nmol⁻¹ versus 94.7 U nmol⁻¹) and thermal stability (half-life of 48 h versus 5.5 h at 75°C) compared with those of XynA. However, GH10 XynB showed higher enzyme activity and thermostability than its truncated mutant without CBM. Site-directed mutagenesis of N-terminal amino acids resulted in a mutant, XynATM1-M, with 50% residual activity improvement at 75°C for 48 h, revealing that the disordered region influenced protein thermostability negatively. The thermal stability of both xylanases and their truncated mutants were consistent with their melting temperature (T_m), which was determined by using differential scanning calorimetry. Through homology modeling and cross-linking analysis, we demonstrated that for XynB, the resistance against thermoinactivation generally was enhanced through improving both domain properties and interdomain interactions, whereas for XynA, no interdomain interactions were observed. Optimized intramolecular interactions can accelerate thermostability, which provided microbes a powerful evolutionary strategy to assemble catalysts that are adapted to various ecological conditions.

INTRODUCTION

Xylanases (endo-1,4-β-xylanase; EC 3.2.1.8) hydrolyze the β-1,4 bond in the xylan backbone of hemicellulose, which yields short xylooligosaccharides or xylose. They are distributed in several glycoside hydrolase families (GH), such as GH5, GH7, GH8, GH10, GH11, and GH43, and most of them belong to GH10 and GH11 according to the CAZy (Carbohydrate Active Enzymes) database (1). Xylanase families differ in their physicochemical properties (molecular mass and isoelectric point [pI]), three-dimensional structures, and catalytic mechanisms (2). Enzymes from GH family 11 have a relatively low molecular weight and a high pI, and they possess a β-jelly roll secondary structure, a double displacement catalytic mechanism, and two glutamates acting as catalytic residues (3). In contrast, members of GH10 typically have a high molecular mass and a low pI and display an (α/β)₈ barrel fold. Xylanases have been found in many organisms, such as bacteria, fungi, algae, and protozoa (4). Xylanases from fungi or mesophilic bacteria generally share a low optimum temperature and poor thermal stability (5). Thermostable xylanases have been characterized from some thermophilic microorganisms, such as Clostridium thermocellum (6), Rhodothermus marinus (7), Thermotoga species (8), Sulfolobus solfataricus (9), Neocallimastix patriciarum (10), and Thermococcus zilligii (11), which exhibit increased specific activity and higher stability.

In recent years, considerable efforts have been taken to understand the mechanisms underlying the thermal stability of proteins from thermophilic organisms that grow at optimal temperatures of more than 60°C (12). The effects of various sequence and structural determinants of thermophilic adaptation have been investigated (13, 14). Usually, thermophilic adaptation of proteins involves optimized packing of hydrophobic interactions and increased density of salt-bridge or charge-assisted hydrogen bonds. In many cases, a combination of the structural features described above is crucial to achieve increased thermostability (14–16). These structural changes contribute to the protein thermophilic adaptation of the potential network of noncovalent interaction within the protein, presumably leading to a balance between overall rigidity, which is important for thermostability, and local flexibility, which is important for activity (17).

Conventionally, noncatalytic carbohydrate-binding modules (CBMs) bind to polysaccharides, bringing biocatalysts close to their substrates (18). Cuskin et al. have reported that the CBM of Bacillus subtilis (BsCBM66) confers enzymatic specificity to nonspecific catalytic domains (19). Several studies on GH10 and GH11 xylanases have indicated that the absence of the N- or C-terminal accessory module reduces or increases the thermal stability of these enzymes (20, 21). To date, analysis of the effect of CBMs and disordered-region amino acids on the catalytic efficiency and thermostability of multimodular xylanases has yielded inconclusive results. Although crystal structure analysis has shown an interaction between the catalytic module and CBM for two GH10 xylanases (6, 22), the contribution of these interdomain interactions to thermal stability has not yet been determined.

Species from the genus Caldicellulosiruptor are excellent consolidated bioprocess candidates for plant polysaccharide deconstruction (23, 24) and have garnered consideration recently. A novel mechanism of biomass deconstruction by cellulases with multiple catalytic domains has been elucidated in C. bescii (25), and an engineered system of C. bescii has been constructed for direct ethanol production from switchgrass without conventional pretreatment (26). Caldicellulosiruptor sp. strain F32 is an anaerobic, extremely thermophilic, and cellulose- and xylan-degrading bacterium which was isolated from biocompost in our laboratory. The multimodular xylanases XynA and XynB, which belong to GH11 and GH10, respectively, were identified from Caldicellulosiruptor sp. strain F32 and exhibit distinct enzymatic properties (27). In this study, both multimodular xylanases and their truncated mutants were expressed in Escherichia coli. The roles for accessory modules in catalytic efficiency and thermostability of the multimodular GH10 and GH11 xylanases were investigated. To the best of our knowledge, chemical cross-linking has not been previously utilized to investigate the intramolecular interaction between the catalytic module and CBM. This study expands our understanding of multimodular enzyme assembly and evolutionary diversity.

MATERIALS AND METHODS

Bacterial strains, plasmids, and culture conditions.

Caldicellulosiruptor sp. strain F32 (CGMCC 1.5183; China General Microbiological Collection Center, Beijing, China) was cultured at 75°C in an anaerobic jar containing modified GS-2 medium (27). Genomic DNA of Caldicellulosiruptor sp. strain F32 was used as the template for gene cloning. E. coli Trans1-T1 phage-resistant, chemically competent cells were used for plasmid maintenance and propagation. E. coli BL21(DE3) chemically competent cells were used for protein expression. His tag expression vector pEASY-E1 was used for gene cloning (all purchased from TransGen Biotech, Beijing, China). E. coli cells containing recombinant plasmids were cultivated at 37°C in LB medium supplemented with 100 μg ml⁻¹ of ampicillin (Sangon Biotech, Shanghai, China).

Cloning, expression, and purification of xylanase and its mutants.

Open reading frames of XynA and XynB (GenBank accession numbers JX030400 and JX030401, respectively) encode signal peptides containing 27 and 35 amino acids, respectively, which were predicted by using SignalP 4.1 Server (http://www.cbs.dtu.dk/services/SignalP/) through online analyses. Thus, oligonucleotide primers for amplifying the coding sequences of these two wide-type xylanases and their truncation mutants were designed (Fig. 1A). Site-directed mutagenesis (XynA-M and XynATM1-M) for replacing the alanine residues at sites 2 and 3 with threonine and serine, respectively, was performed by introducing site-specific amino acid substitutions into the coding sequence using oligonucleotide primers. The chimeric xylanases rXynB, cXynB, and mXynB (Fig. 1) were constructed using splice overlap extension PCR (SOE-PCR). The linker and CBM36 sequences from XynA (residues 196 to 332) were connected to the C terminus of the GH10 module from XynB (321 to 651) to construct the xylanase cXynB. mXynB was constructed by replacing the linker sequence of XynB (residues 294 to 320) with the linker sequence of XynA (196 to 214). All of the primers used in this study were designed according to enzyme domains, which were predicted by using NCBI's Conserved Domain Database (CDD) (28). PCR products were cloned into the pEASY-E1 expression vector in accordance with protocols of the manufacturer.

FIG 1.

Schematic structures and SDS-PAGE protein analysis. (A) Schematic structures for the XynA, XynB, and mutant enzymes. Signal peptide, filled black rectangle; GH, shaded rectangle; CBM, blank rectangle. (B) SDS-PAGE of the XynA, XynB, and mutant enzymes. Lane M, protein molecular mass marker; lane 1, XynA; lane 2, XynATM1; lane 3, XynA-M; lane 4, XynATM1-M; lane 5, XynACBM; lane 6, XynB; lane 7, XynBTM1; lane 8, XynBTM2; lane 9, XynBTM3; lane 10, rXynB; lane 11, XynBCBM1; lane 12, XynBCBM2; lane 13, mXynB; lane 14, cXynB.

The recombinant pEASY-E1 expression vector containing the target sequences was transformed into Transl-T1 phage-resistant, chemically competent cells per the manufacturer's protocol. Accurate inserts were verified by sequence analysis (GenScript, Nanjing, China). Recombinant plasmids were isolated and transformed into E. coli BL21(DE3) for protein expression. Recombinant E. coli cells were cultivated at 37°C in LB medium supplemented with 100 μg ml⁻¹ of ampicillin until the optical density at 600 nm reached 0.5 to 0.6. For induction, 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) was added to the medium, and the culture was incubated for another 16 h in a shaker at 16°C.

Cells were harvested by centrifugation and resuspended in lysis buffer (50 mM NaH₂PO₄, 300 mM NaCl, pH 8.0) supplemented with lysozyme (1 mg ml⁻¹) and protease inhibitor cocktail (Amresco, Solon, OH). The resuspended cells were disrupted by ultrasonication on ice, and crude enzymes were prepared after centrifugation at 4°C and 12,000 × g for 20 min to remove the cell debris. The crude enzymes were heated at 75°C (XynA, XynATM1, XynA-M, XynATM1-M, and XynB) or 60°C (XynBTM1, XynBTM2, XynBTM3, rXynB, cXynB, and mXynB) for 10 min and centrifuged at 4°C. The recombinant proteins were purified from the supernatant by using a nickel-nitrilotriacetic acid (Ni-NTA)-Sefinose column (Sangon, Shanghai, China). The eluent was replaced by PC buffer (50 mM sodium phosphate, 12 mM citrate) by ultrafiltration (10-kDa-cutoff membrane; Millipore, Billerica, MA) at 4°C. The protein obtained as described above was analyzed by SDS-PAGE (Fig. 1B). Protein concentrations were determined by using the Bradford method with bovine serum albumin as the standard (29).

Determination of optimal temperature, pH, and thermostability.

The optimal pH of enzyme was determined using two buffers, 0.2 M acetate buffer (pH 4.0 to 5.6) and PC buffer (pH 6.0 to 8.0). Adequate amounts of the enzyme were incubated with 1% beechwood xylan (Sigma-Aldrich, St. Louis, MO) in different buffers for 10 min at the optimal temperatures of the respective enzymes. The reducing sugars released were measured using a dinitrosalicylic acid (DNS) assay with xylose as the standard (30). Absorbance was read at a wavelength of 540 nm. Optimal temperatures of the respective enzymes were determined by incubating adequate amounts of each enzyme with 1% beechwood xylan in PC buffer for 10 min at temperatures ranging from 50°C to 70°C (XynBTM1, XynBTM2, and XynBTM3) or 65°C to 85°C (XynA, XynA-M, XynATM1, XynATM1-M, and XynB), with increments of 5°C. In the thermostability assay, at different time points, aliquots of samples were periodically withdrawn and the enzyme activity was determined immediately. Residual activity was expressed as the ratio of specific activity measured at each time point to the activity at 0 h.

Enzyme activity assay and xylan hydrolysis.

Specific activities of the enzymes and their mutants against beechwood xylan were determined at 75°C (XynA, XynATM1, XynA-M, XynATM1-M, and XynB) or 60°C (XynBTM1, XynBTM2, and XynBTM3) in PC buffer. Adequate amounts of the enzymes were incubated with 1% beechwood xylan under optimal conditions for 10 min. The reducing sugars released were measured using the DNS method. One unit of enzyme activity was defined as the amount of enzyme that released 1 μmol of reducing sugar per minute. Unless otherwise stated, the reaction was performed in triplicate.

The hydrolysis products of xylan produced by XynA, XynA-M, XynB, XynBTM1, XynBTM2, and XynBTM3 were analyzed using thin-layer chromatography (TLC). Purified enzyme was incubated with 1% (wt/vol) beechwood xylan in PC buffer for 12 h. Products of hydrolysis (2 μl) were spotted on silica gel plates (60F254; Merck, Darmstadt, Germany). The plates were developed using a butan-1-ol-acetic acid-water (2:1:1, vol/vol) solvent system. The color reaction was performed in an oven at 105°C for 30 min after spraying the plates with a methanol-sulfuric acid mixture (4:1, vol/vol). Xylose, xylobiose, xylotriose, xylotetraose, and xylopentaose for standards were purchased from Megazyme (Wicklow, Ireland).

Determination of kinetic parameters of purified enzymes and binding assay.

Michaelis-Menten kinetic parameters of the enzymes against beechwood xylan were determined. Appropriate concentrations of enzymes were incubated with the substrate at various concentrations. The reaction was performed at optimal conditions for each enzyme, and reducing sugars were detected using the DNS method to calculate the initial rates, which were plotted against substrate concentrations. The apparent kinetic parameters were estimated using the Michaelis-Menten equation with GraphPad Prism 5.01 software (San Diego, CA) by employing nonlinear regression (31).

The affinity of recombinant XynA, XynATM1, XynB, XynBTM1, XynBTM2, and XynBTM3 to insoluble xylan was determined. Ten micrograms of each enzyme was incubated with 20 mg insoluble beechwood xylan in PC buffer in a total volume of 70 μl. The mixture was shaken end over end for 1 h at 4°C and then centrifuged at 12,000 × g for 3 min. The supernatant contained the unbound fraction. The pellet was washed three times with 0.2 ml PC buffer and resuspended in 70 μl PC buffer. It then was boiled for 5 min to release the bound protein. The supernatant (unbound protein) and the suspended pellet (bound protein) were examined by SDS-PAGE according to Zhang et al. (32).

Measurement of melting temperature.

Differential scanning calorimetry (DSC) usually is used to investigate the thermal stability of proteins (33, 34). In this study, the assay was performed using a MicroCal VP-Capillary differential scanning calorimeter (Northampton, MA). The cell volume was set at 0.4 ml with heating rates of 60°C min⁻¹. DSC scans were performed at temperatures ranging from 25°C to 110°C. For all measurements, the enzymes were dissolved in PC buffer to achieve a final concentration of 20 μM. A DSC scan corresponding to buffer versus buffer (PC buffer) was used as the instrumental baseline. All of the experimental data were corrected for the calorimetric baseline by subtracting results for the buffer only from those for the buffer scan.

Structure modeling and determination of the quaternary structure of wild-type enzymes.

I-TASSER (http://zhanglab.ccmb.med.umich.edu/I-TASSER) was used to identify structural homologues and to predict the structure of the enzymes (35). Structures of wild-type XynA and XynB were predicted through homology modeling by using the crystal structures of the alkaliphilic XynJ (Protein Data Bank entry 2DCJ; sequence identity, 57%) from Bacillus sp. strain 41M-1 and Xyn10B (Protein Data Bank entry 2W5F; sequence identity, 29%) from Clostridium thermocellum, respectively, as the template. Quaternary structures of XynA and XynB were investigated using size exclusion chromatography with a Superdex 200 column (36, 37). Blue dextran 2000 (2,000 kDa), ferritin (400 kDa), conalbumin (75 kDa), and ovalbumin (43 kDa) (all obtained from Sigma-Aldrich) were used as reference proteins.

Determination of the interactions between GHs and CBMs.

Surface plasmon resonance (SPR) was used to determine the interactions between GH modules and CBMs. An N-ethyl-N′-[3-(dimethylamino) propyl] carbodiimide (EDC)/N-hydroxysuccinimide (NHS)-mediated amine coupling protocol was used to immobilize XynATM1 and XynBTM2 onto a CM5 chip surface in the presence of 50 mM sodium acetate buffer at pH 4.8 and 4.2, respectively. HBS-N buffer (0.01 mM HEPES, 0.15 mM sodium chloride, pH 7.4) containing XynACBM (100 μM) was introduced to pass immobilized XynATM1 over the chip surface. HBS-N buffer containing XynBCBM1 and XynBCBM2 (100 μM), individually or simultaneously, was introduced to pass immobilized XynBTM2 over the chip surface. Adsorption data were automatically collected using the Biacore T100 instrument (GE Healthcare, Uppsala, Sweden) with a chip channel without any treatment as a control.

The binding of GHs and CBMs was measured using isothermal titration calorimetry (ITC) at 25°C with an iTC200 microcalorimeter (Microcal Inc., Northampton, MA). XynATM1 and XynBTM2 were adjusted to a final concentration of 50 mM. XynACBM was injected 20 successive times into a reaction cell containing XynATM1 by using a 2-μl rotating stirrer syringe. XynBCBM1 and XynBCBM2 were injected into XynBTM2 individually or simultaneously. All of the enzymes were dissolved in PC buffer (pH 6.5), and the buffer was used as the control.

Protein-protein interactions were analyzed using chemical cross-linking (38). Glutaraldehyde was added to XynA and XynB at a final concentration of 0.01% (vol/vol). A cross-linking reaction was performed at 22°C for 15 min in a total volume of 40 μl. The reactions were stopped by adding 4 μl of 1 M Tris-HCl (pH 8.0). XynA and XynB were cleaved using cyanogen bromide and hydroxylamine, respectively. The cross-linking and cleavage products were determined by SDS-PAGE. The bands were cut from the gel, digested in gel, and analyzed using liquid chromatography-tandem mass spectrometry (LC-MS/MS) with a Surveyor Plus ion trap mass spectrometer (LTQ XL; Thermo Fisher, San Jose, CA) equipped with a nanospray source (39).

Determination of catalytic efficiencies of purified enzymes.

Catalytic constants of the purified enzymes against xylopentaose were determined using a method described by Han et al. (40). Briefly, xylopentaose (30 μM) was hydrolyzed using XynA, XynATM1, XynB, XynBTM1, and XynBTM2 at the optimal temperature and pH of each enzyme in a final volume of 500 μl, and the reactions were terminated by boiling for 5 min. The final concentration of XynA and XynATM1 was 0.02 nM, and that of XynB, XynBTM1, and XynBTM2 were 1.2 nM. A linear relationship was observed between the hydrolysis rate and xylopentaose concentration (0, 30, 60, and 90 μM). Therefore, it was believed that the initial substrate concentration of 30 μM was well below the K_m. The relationship between hydrolysis rate and hydrolysis time (0, 5, and 10 min) was linear; hence, the hydrolysis reaction was terminated after 5 min. Xylopentaose concentrations at the beginning (S₀) and at termination (S_t, 5 min) were measured using high-performance anion-exchange chromatography with pulsed amperometric detection (HPAEC-PAD; Dionex Corporation, Sunnyvale, CA). These concentrations were used to calculate the catalytic efficiencies of the enzymes, as described by Hu et al. (41).

RESULTS

Caldicellulosiruptor sp. strain F32 XynA and XynB are β-1,4-endoxylanases.

Screening the genome of Caldicellulosiruptor sp. strain F32 (GenBank accession no. APGP00000000) reveals two putative xylanases, XynA and XynB (27). According to domain annotation from CAZy (1), XynA has a GH11 catalytic domain and a family 36 CBM, whereas XynB has two family 22 CBMs and a GH10 domain (Fig. 1A). Sequence analysis by SignalP 4.1 Server revealed that both xylanases have signal peptides in their N termini, which suggested the two xylanases are extracellular proteins.

Both recombinant xylanases were purified to homogeneity according to SDS-PAGE (Fig. 1B). The assay with beechwood xylan showed catalytic activity at 2,601 U mg⁻¹ for XynA and 160 U mg⁻¹ for XynB (Table 1), which indicated the recombinant proteins both possessed endoxylanase activity but different catalytic abilities. Oligosaccharides, including xylobiose, xylotriose, xylotetraose, and xylopentaose, were detected in the xylan degradation mixture catalyzed by XynA and XynATM1. XynB and its CBM-deleted mutants hydrolyzed beechwood xylan to xylose, xylobiose, and xylotetraose (Fig. 2A). This indicated both xylanases act complementarily during xylan degradation.

TABLE 1.

Characteristics of XynA, XynB, and their truncated mutants

Protein	Optimal temp (°C)	Optimal pH	Sp act measured asa:		Km (mg ml−1)	kcat (s−1)	kcat/Km (s−1 mg−1 ml−1)	t1/2b (h)
			U mg−1	U nmol−1
XynA	75	6.5	2,601 ± 43	94.7 ± 1.6	2.9 ± 0.05	131.3 ± 2.3	45.9 ± 0.8	5.5
XynA-M	75	6.5	3,581 ± 72	130.3 ± 2.6	2.4 ± 0.04	83.8 ± 1.4	35.2 ± 0.6	5.5
XynATM1	75	6.5	9,221 ± 96	215.8 ± 2.2	5.8 ± 0.16	304.7 ± 8.4	52.3 ± 1.4	48
XynATM1-M	75	5.6	5,451 ± 74	127.6 ± 1.7	5.3 ± 0.21	294.3 ± 11	55.9 ± 1.3	70
XynB	75	6.5	160 ± 4	12.0 ± 0.3	1.2 ± 0.002	101 ± 0.3	84.2 ± 0.14	42.7
XynBTM1	60	6.5	91 ± 8	5.4 ± 0.47	3 ± 0.003	57 ± 0.04	19.0 ± 0.02	52.2
XynBTM2	60	6.5	60 ± 1	1.4 ± 0.02	7.4 ± 0.02	28 ± 0.08	3.8 ± 0.01	48.5
XynBTM3	60	6.5	35 ± 2	1.4 ± 0.08	6.7 ± 0.09	19.2 ± 0.29	2.9 ± 0.03	55.6
rXynB	70	6.0	54 ± 2	0.97 ± 0.04	3.9 ± 0.2	51 ± 2	13 ± 0.2	1.8
mXynB	70	6.5	47 ± 2	3.5 ± 0.2	3.5 ± 0.2	107 ± 3	30.7 ± 0.25	25
cXynB	60	6.5	53 ± 2	2.8 ± 0.1	4.6 ± 0.05	23 ± 1	4.9 ± 0.1	41

^aThe experiments were performed in triplicate, and data are reported as means ± standard deviations.

^bt_1/2, half-life of enzyme at optimal temperature.

FIG 2.

Hydrolytic products and binding analysis. (A) TLC analysis of hydrolytic products from 1% (wt/vol) beechwood xylan. Lane M is standard xylose (X1), xylobiose (X2), xylotriose (X3), xylotetraose (X4), and xylopentaose (X5). Lane X is 1% beechwood xylan without enzyme. Lines 1, 2, 3, 4, 5, and 6 are the products of 1% beechwood xylan hydrolyzed by XynA, XynATM1, XynB, XynBTM1, XynBTM2, and XynBTM3, respectively. (B) Qualitative binding analysis of the wild-type and mutant enzymes to insoluble xylan. –, unbound protein; +, bound protein.

The substrate affinity of XynA and XynB were determined through a binding assay (Fig. 2B). XynA and XynB, which harbored one or two CBMs, bound tightly to insoluble beechwood xylan. The truncated mutant XynBTM1, harboring one CBM, showed weaker affinity than the wild type. However, the binding capabilities to insoluble substrate for XynATM1, XynBTM2, and XynBTM3, lacking CBM, were completely lost, indicating CBMs play an important role in terms of protein binding to insoluble substrate.

Determination of optimal temperature and pH.

The deletion derivatives of XynA, including XynA-M, XynATM1, and XynATM1-M, showed optimal temperatures at 75°C similar to those of the wild type, whereas those derived from XynB, including XynBTM1, XynBTM2, and XynBTM3, had a significant decrease of optimal temperature (60°C). As for the optimal pH, both wild-type xylanase and most of their truncated mutants showed their highest xylan-degradative activity at the same pH (∼6.5), except that the truncated mutant XynATM1-M had an optimal pH of 5.6 (Table 1). These results suggested the optimal conditions for xylanases are strictly dependent on their functional modules.

Specific activity and kinetic parameters of both xylanases.

The specific activity of the recombination enzyme XynA was 2,601 U mg⁻¹ with beechwood xylan as the substrate, which was 15.3 times higher than that of XynB. On the other hand, the catalytic efficiency (k_cat/K_m) value for XynA was much lower than that for XynB (45.9 versus 84.2 s⁻¹ mg⁻¹ ml). When CBM was deleted, the specific activity of the truncated mutant XynATM1 showed a 1.3-fold increase (215.8 U nmol⁻¹ versus 94.7 U nmol⁻¹). However, the K_m value for XynATM1 was two times that for XynA (5.8 versus 2.9 mg ml⁻¹), which led to only a 0.13-fold increase in the catalytic efficiency (k_cat/K_m; 52.3 versus 45.9 s⁻¹ mg⁻¹ ml). In contrast, the truncated mutants XynBTM1, XynBTM2, and XynBTM3 showed decreased specific activity and catalytic efficiency (k_cat/K_m) compared to those of XynB (Table 1).

Determination of thermostability.

At 75°C, XynA had a half-life of 5.5 h, whereas XynB had a half-life of 42.7 h (Table 1). There was no significant difference in thermal stability between XynA and the N-terminal amino acid mutant XynA-M (Table 1). However, XynATM1 had 50% residual activity after incubating at 75°C for 48 h, indicating an improved tolerance for heat treatment was achieved when the CBM was deleted. Moreover, the thermal stability of XynATM1-M, the N-terminal-amino-acid-substituted mutant of XynATM1, which has a half-life of 70 h at 75°C, was further improved.

For XynB, the optimal temperatures of CBM-deleting mutants deceased to 60°C. Moreover, the thermal stability of these truncated proteins had a more dramatic decline than that of XynB (Table 1). After incubating at 70°C for 24 h, the residual activity of XynBTM1, XynBTM2, XynBTM3, mXynB, and rXynB was 43%, 44%, 53%, 51%, and 1.3%, respectively. Chimeric xylanase, cXynB, harboring CBM36 from XynA and a catalytic module of XynB, showed a half-life of 41 h at 60°C, which indicates it is less thermostable than XynB.

T_m of purified proteins.

The melting temperatures (T_m) of XynA, XynA-M, XynATM1, XynATM1-M, and XynACBM were 86.6, 87.4, 91.7, 91.5, and 81.3°C, respectively (Fig. 3A). The T_ms of proteins XynA and XynATM1 were consistent with their half-lives, indicating the enhanced enzyme thermostability of the latter resulted from its structural stability under high temperature. For all five proteins tested, CBM alone had the lowest T_m value, which might cause the entire enzyme to have poor thermal stability. The T_ms for XynB, XynBTM1, XynBTM2, XynBTM3, XynBCBM1, and XynBCBM2 were 87.2, 79.4, 77.7, 79.5, 83.5, and 82.2°C (Fig. 3B), respectively. In contrast to XynA, CBM deletion negatively affected the thermal stability of XynB, which caused the T_ms of truncated mutants of XynB to decline by 7 to 10°C. We also noticed the T_ms for XynBCBM1 and XynBCBM2 were higher than that of the GH10 catalytic module.

FIG 3.

Melting curves of mutant enzymes from XynA (A) and XynB (B) at temperatures from 60 to 100°C. XynA and XynB, black line; XynA-M and XynBTM1, red line; XynATM1 and XynBTM2, green line; XynATM1-M and XynBTM3, blue line; XynACBM and XynBCBM1, dark yellow line; XynBCBM2, bright yellow line; rXynB, gray line. For all measurements, the protein concentration was 20 μM. Cp, heat capacity.

Structural analysis.

To understand the contribution of different modules for thermal stability of multimodular xylanases, structures of XynA and XynB were modeled based on the structures of XynJ (Protein Data Bank entry 2DCJ) and Xyn10B (Protein Data Bank entry 2W5F), respectively (Fig. 4). The confidence scores of the structures predicted using I-TASSER were given as C-score values (confidence interval, −5, 2) to estimate the quality of the predicted models. In this study, the C-score value for XynA was 1.61, and that for XynB was −2.31, even though Xyn10B (Protein Data Bank entry 2W5F) had the modular arrangement of an N-terminal CBM22 and GH10 catalytic domain at the C terminus, which was different from that of XynB. The structure of XynA was constructed by two independent parts, GH11 and CBM, that were connected with a peptide linker of 19 amino acids (Fig. 4A). The structure of XynB was different from that of XynA, in that two CBMs and the GH10 domain interacted closely with each other, probably through noncovalent forces, including van der Waals and hydrogen bonds (Fig. 4B). Size exclusion chromatography showed that XynA and XynB had molecular masses of 82 kDa and 77.5 kDa, respectively, and suggested that XynA existed as a homodimer and XynB existed as a monomer. From a structural view, deletion of the CBM from XynB may have exposed the hydrophobic groups, thereby reducing the thermal stability and catalytic efficiency.

FIG 4.

Predicted model structures of XynA and XynB. (A) The model structure of XynA using the crystal structure of XynJ (Protein Data Bank entry 2DCJ) as the template. (B) The structure of XynB using the crystal structure of Xyn10B (Protein Data Bank entry 2W5F) as the template. Alpha helices and beta sheets are shown in red and yellow, respectively. Linker sequences between GH and CBM are shown in blue.

Analysis of interdomain interactions.

SPR and ITC were conducted to analyze the interactions between the GH module and CBM. However, results of SPR and ITC did not show any interaction between the catalytic domain and CBM for both XynA and XynB (data not shown), probably because it is too weak to capture. Chemical cross-linking has been proved to be able to detect weak intramolecular interactions (38). In addition, XynA and XynB can be cleaved specifically at Met221 by CNBr and Aln208 by hydroxylamine, respectively, both immediately before or after the linker region. For XynA, chemical cross-linking by using glutaraldehyde did not detect any interdomain interaction, since no differences were observed after cleavage with CNBr for cross-linked protein (Fig. 5A). In contrast, only one band was detected for cross-linked XynB in SDS-PAGE after cleavage with hydroxylamine (Fig. 5B), indicating the presence of intramolecular interactions between the catalytic domain and CBM. The appearance of cleaved protein bands on the gel was confirmed using liquid chromatography-tandem mass spectrometry (LC-MS/MS; data not shown). Cross-linking of XynA by glutaraldehyde produced a faint band, probably because XynA forms a homodimeric structure.

FIG 5.

Detection of interdomain interactions in GH11 XynA and GH10 XynB from the hyperthermophilic bacterium Caldicellulosiruptor sp. strain F32. (A) Chemical cross-linked and cleaved XynA; lane M, protein molecular mass marker; lane 1, XynA; lane 2, XynA cleaved by CNBr; lane 3, XynA cross-linked by glutaraldehyde; lane 4, XynA cleaved by CNBr after glutaraldehyde cross-linking. (B) Chemical cross-linked and cleaved XynB; lane 1, XynB; lane 2, XynB cleaved by hydroxylamine; lane 3, XynB cross-linked by glutaraldehyde; lane 4, XynB cleaved by hydroxylamine after glutaraldehyde cross-linking.

In addition, the effects of interdomain interactions on the catalytic efficiencies of xylanases were investigated using soluble xylopentaose as the substrate. The catalytic efficiency (k_cat/K_m) of XynATM1 (27.76 mM⁻¹ s⁻¹) was comparable to that of XynA (36.15 mM⁻¹ s⁻¹) but was higher than that of XynB (0.45 mM⁻¹ s⁻¹); the catalytic efficiencies of truncated mutant XynBTM1 (0.16 mM⁻¹ s⁻¹) and XynBTM2 (0.18 mM⁻¹ s⁻¹) decreased sharply. These results clearly indicated that deletion of CBMs disturbed the catalytic efficiency of XynB but not of XynA, indicating that interdomain interactions existed only in multimodular XynB.

Site-directed mutagenesis.

Xylanase XynB (GenBank accession number AAC46361.1) from Dictyoglomus thermophilum (DtXynB) has an 82% amino acid identity with XynA from strain F32. The five N-terminal amino acid residue deletions caused a dramatic decrease in thermal stability of DtXynB (42). XynA-M and XynATM1-M were constructed as described above. It was shown that the mutagenesis of the two amino acids did not cause any obvious change in the optimal temperature or pH, except XynATM1-M had a decline in the pH optimum to 5.6 (Table 1). For either the whole enzyme or the core enzyme, the Michaelis constant (K_m) value declined only slightly after site-directed mutagenesis at the N terminus (Table 1). Substitutive modification in the N terminus of XynA did not change its thermostability (Table 1). Nevertheless, the N-terminal mutation of XynA-TM1 accelerated thermostability. The residual activity of XynATM1-M was 76% after heat treatment at 75°C for 48 h, while only 50% residual activity remained for XynATM1.

DISCUSSION

Deletion of the CBM has an entirely different effect on the thermostability of GH10 and GH11 xylanases. The truncated mutant XynATM1, harboring a GH11 catalytic module without CBM, showed improved thermal stability compared with that of XynA. Here, the CBM is a hydrophobic structure linked with the GH11 module by a short peptide linker, which are independently folding polypeptide units without physical interaction, as revealed by the predicted structure (Fig. 4A). Homology modeling suggested that no interdomain interactions existed between the GH module and the CBM in family 11 XynA, which was further confirmed by chemical cross-linking (Fig. 5A). DSC analysis of truncated mutants of XynA proved that the CBM is thermodynamically sensitive and can unfold easily. Therefore, it was suggested that CBM36 is making an unfavorable contribution to the thermostability of the entire enzyme (Table 1). It has been reported that deletion of CBM has increased the thermal stability of GH family 11 xylanase DtXynB from D. thermophilum Rt46B.1 (42) and NfXyn11A from N. flexuosa (43), which is consistent with the results obtained for GH family 11 XynA from strain F32 (Table 2).

TABLE 2.

Thermostability of GH10 and GH11 xylanases and amino acid composition for their linkers

Xylanasea	Thermostabilityb	Linker characteristics
		Locationc	Length (bp)	Pd (%)	S+Te (%)	Gf (%)
Caldicellulosiruptor sp. strain F32
XynA	56%, 4 h, 75°C	GH11/CBM36	19	5.3	63	5.3
XynA-TM1	87%, 4 h, 75°C
D. thermophilum (42)
XynB7	50%, 8 h, 85°C	GH11/CBM36	23	0	60.9	21.7
XynB6	100%, 8 h, 85°C
N. flexuosa (43)
33.4 kDa	50%, 13 min, 80°C	GH11/CBM2	43	16.3	21	25.6
r23.8 kDa	50%, 157 min, 80°C
Caldicellulosiruptor sp. strain F32
XynB	50%, 42 h, 75°C	CBM22/GH10	27	14.8	7.4	3.7
XynBTM2	44%, 24 h, 70°C
T. maritima (20)
rXTM A	70%, 30 min, 80°C	CBM22/GH10	33	9.1	12.1	0
rXTM AΔN	40%, 30 min, 80°C
C. stercorarium (44)
rCBM22-CM-CBM9	90%, 10 min, 70°C	CBM22/GH10	26	3.8	11.5	7.7
rCM-CBM9	60%, 10 min, 70°C
Bacillus sp. (21)
XynC	90%, 30 min, 55°C	CBM22/GH10	34	5.9	20.6	0
XynCΔTD	30%, 30 min, 55°C
C. cellulovorans (45)
XynAd1/2	85%, 4 h, 70°C	CBM22/GH10	25	12	8	0
XynAd2	50%, 4 h, 70°C

^aXynB7 was xylanase with module structure GH11-linker-CBM36; 33.4 kDa was xylanase with module structure GH11-linker-CBM2; XynB6 and r23.8 kDa were xylanases with module structure GH11-linker; rXTMA, rCBM22-CM-CBM9, XynC, and XynAd1/2 were xylanases with module structure CBM22-linker-GH10; rXTM AΔN, rCM-CBM9, XynCΔTD, and XynAd2 were xylanases with module structure linker-GH10.

^bThermostability was determined under the conditions indicated; the percentages represent residual activity divided by the original activity.

^cDesignation of module flanking the linkers.

^dThe proline content of the linkers.

^eThe serine and threonine content of the linkers.

^fThe glycine content of the linkers.

In contrast, the CBM deletion from the GH family 10 xylanase XynB decreased thermal stability (Table 1), which is consistent with DSC analysis. A number of previous results for GH10 xylanases indicated that N- or C-terminal accessory module absence can reduce the thermal stability to the same extent as XynB from Caldicellulosiruptor sp. strain F32 (Table 2) (20, 21, 44, 45). To date, only three crystal structures of GH10 xylanases have been reported together with CBMs, i.e., Cellvibrio japonicus Xyn10C (CjXyn10C; Protein Data Bank entries 1US2 and 1US3) (46), Streptomyces olivaceoviridis Xyn (SoXyn; Protein Data Bank entry 1ISV) (22, 47), and Clostridium thermocellum (CtXyn10B; Protein Data Bank entry 2W5F) (6). In contrast to CjXyn10C, where no direct interaction was observed between the two modules, the catalytic module and CBM interactions in SoXyn and CtXyn10B were observed in the crystal structure. However, the contribution of intramolecular interactions to thermostability has not been demonstrated (20). From the predicted structure of the GH10 XynB (Fig. 4C), the two CBMs interacted intramolecularly with the (α/β)₈ barrel catalytic module via noncovalent forces, forming a hydrophobic core that could increase the enzyme's thermostability. SPR, ITC, cross-linking experiments, and catalytic efficiency analysis against xylopentaose with a truncated mutant strongly indicated the existence of weak interdomain interactions between the catalytic domain and two CBMs in XynB (Fig. 5B). Many factors can affect enzyme thermostability, but few of them can be easily detected (48). For instance, the difference in thermostability between a mesophilic and a thermophilic enzyme can be caused by rather small stabilizing interactions (40 kJ mol⁻¹), for example, interactions involving three or four hydrogen bonds (49).

Upon analysis of the linker region, a high serine and threonine content in the linker sequence of XynA (63%) and DtXynB (61%) was found (Table 2). This property can provide high linker extensions for the two xylanases (50). For NfXyn11A, a 16.3% proline content of the linker sequence also can lead to extended conformation of the enzyme (50, 51). Thus, we speculated that the existence of a high serine and threonine content, or a high enrichment in proline, in the linker sequence can lead to extended peptide conformation of family 11 xylanase (Fig. 4A). This speculation could explain why the GH11 domain did not interact with the CBM domain. Actually, similar linkers in amylases, which are characteristically rich in serine and threonine, particularly those containing proline, have been reported to be rigid (52). In contrast, linkers in most GH10 xylanases have low serine and threonine content, or proline content, forming a flexible structure that allows free movement of the catalytic domain and CBM, enabling them to tightly interact with each other (Table 2). Furthermore, it has been reported that linker regions can affect the overall rigidity of proteins and play a key role in adapting to cold environments (53). Thus, it can be concluded that the linker is important in multimodular enzyme assembling as well as functioning during evolution in various environments.

Site-directed mutagenesis experimental results indicated that changing the terminal amino acids at disordered regions can effectively improve the thermal stability of the core enzyme XynATM1-M compared to that of XynATM1. Previous studies have reported that terminal amino acid residues are crucial for the entire enzyme (42, 54). Liu et al. have reported that disordered terminal amino acid deletions increased regular secondary structural contents, which led to slow decreases in Gibbs free energy in the thermal denaturation process and ultimately enhanced enzyme thermostability (55). In the present study, circular dichroism (CD) spectrum data showed that the substitution of amino acids at the N terminus of XynA and XynATM1 did not cause secondary structure changes in the proteins (data not shown). We speculated that additional hydrogen bonds formed from other amino acids outside the active cavity with the hydroxyl group of the substituted serine and threonine, which would convey the mutant elevated ability to resist heat denaturation over a longer period of time (10 to 48 h) at relatively low temperature (65 and 75°C) or at the early stage of heat treatment (0 to 10 h) at relatively high temperature (85°C). It has been reported for another xylanase, XynCDBFV, that the N-terminal region is attached to the catalytic core by hydrogen bonds, stacking forces, and a disulfide bond, which stabilize the overall protein folding (10).

XynATM1 lacking CBM shows a significant increase in specific activity and thermostability compared to XynA, which might be a result of the lack of irreversible binding of enzymes to the residual substrate. This makes it an attractive candidate for use in the hydrolysis of lignocellulose with high consistency in industrial applications (56). The optimum temperatures of XynA and XynB both were at 75°C, and they also showed high thermal stability at this temperature. These characteristics resulted from long-term natural evolution and adaptation to the growth environment of strain F32. CBMs of XynA and XynB have opposite effects on thermal stability, which may be caused by protein evolution in different enzyme families. Fine-tuning rigidity and flexibility of the linker can optimize intramolecular interactions between GH and CBMs, which provides microbes a powerful evolution strategy to assemble enzymes to adapt to various environmental conditions. Interestingly, XynB is more thermally stable than XynA, which imparts an advantage to the degradation of xylan at elevated temperatures, whereas XynA has the advantage of high catalytic efficiency for the rapid degradation of substrates. In addition, when the initial ratio was 1:1 for XynA (20 nM) and XynB (20 nM), they showed a degree of synergy of approximately 1.3. The complementary enzymatic properties of XynA and XynB, which might have formed through protein thermophilic adaptation, allow strain F32 to effectively deconstruct lignocellulose (Fig. 6). It is worth noting that homologs of GH10 XynA and GH11 XynB are widespread in Caldicellulosiruptor species as well as other lignocellulosic biomass-degrading microorganisms. Genes encoding GH elements, including the modules GH, CBM, and linker, may be transferred from one organism to another via horizontal gene transfer, enabling enzymes to be adapted to extreme environmental conditions. These results highlight the mechanisms underlying the thermal stability of xylanases belonging to GH families 10 and 11 and provide a basis for further protein engineering efforts.

FIG 6.

Schema of xylan hydrolysis by GH11 XynA and GH10 XynB from the hyperthermophilic bacterium Caldicellulosiruptor sp. strain F32. Note that many enzymes are involved in the complete hydrolysis of hemicellulose; however, the figure shows only multidomain endo-1,4-β-xylanases XynA and XynB, which generate xylo-oligosaccharides from xylan.