The codon optimized xylanase gene Xyn and β-xylosidase gene XynB from Pseudothermotoga thermarum DSM5069 were synthesized by Genray Biotech Co., Ltd. (Shanghai, China). Escherichia coli DH5α and Bacillus subtilis WB800 were used for cloning and secretory expression of enzymes. Escherichia coli and B. subtilis were grown in LB media at 37°C. 100 μg/mL ampicillin for E. coli and 50 μg/mL kanamycin for B. subtilis were used for selecting transformed bacteria. Beechwood xylan and p-nitrophenyl-β-D-xylopyranoside (pNPX) were purchased from Sigma to analyze Xyn activity and XynB activity, respectively.

To construct a E. coli/B. subtilis shuttle vector on the basis of pWB980, DNA fragments of pWB980 and pET-20b were amplified by using primers shown in Table 1 (P1-P4). The amplified DNA fragments were digested with enzymes of NdeI and ligated together to obtain pWB980-20b. To secretory express Xyn and XynB in B. subtilis, the mature Xyn and XynB genes were amplified using P5-P6 and P7-P8 primers, respectively. The PCR products were digested with KpnI and XbaI, and then ligated into pWB980-20b, generating pWB980-20b-Xyn and pWB980-20b-XynB. The cloning operations were conducted in E. coli DH5α.

The recombinant plasmids pWB980-20b-Xyn and pWB980-20b-XynB were transformed into the competent cells of B. subtilis WB800. The recombinant bacteria were grown in LB media containing 50 μg/mL Kanamycin at 37°C. The cells were harvested by centrifugation at 10,000 g for 10 min and then sonicated to release the intracellular proteins. The intracellular proteins and extracellular proteins were then heat treated at 60°C for 30 min and centrifuged (10,000 g for 20 min). The resulting supernatants were purified by affinity chromatography on Ä KTATPLC™ with a HiTrap column (GE Healthcare Life Sciences). To confirm the target proteins’ purity, SDS-PAGE was used. Bradford method was used to measure the concentration of the purified enzymes.

Under the standard assay condition, the recombinant enzyme was mixed with substrate beechwood xylan (20 mg) or pNPX (20 mM) in a final volume of 50 mM sodium acetate buffer (pH 5.0). The total reaction volume was 0.5 mL. For Xyn assay, 0.3 mL of 3,5-dinitrosalicylic acid (DNS) was added after incubation at 85°C for 15 min. The solution was boiled for 5 min and then measured at 550 nm. The XynB activity was stopped by adding 0.6 mL of 1M Na₂CO₃ and measured the absorbance at 405 nm. One unit of Xyn or XynB activity was defined as the amount of enzyme releasing 1 µmol reducing sugar or pNPX per minute. The values were given as the averages of three separate determinations.

To determine the effect of pH on enzyme activity, Xyn and XynB were incubated at different pH values (4.0-8.5) using 50 mM citrate buffer (pH 4-6.5) and 50 mM Tris-HCl buffer (pH 6-8.5), respectively. The pH stability of the two enzymes was determined by measuring the remaining activity after incubating the enzymes for 1 h at different pH values. The optimal temperature of Xyn and XynB was examined by standard assay ranging from 55°C to 100°C. The effect of temperature on the stability of the two enzymes was assessed by measuring the residual activity after pre-incubating enzyme at 75–95°C for 0.5 h. All measurements were performed in triplicate.

The influence of metal ions and chemicals on the enzyme activity was conducted by incubating the enzyme with 5 mM (final concentration) MgCl₂, MnCl₂, CaCl₂, CuCl₂, Tris, Tween 60, or EDTA for 1 h before adding substrate to start the reaction. All experiments were performed in triplicate.

The 3D structure of Xyn and XynB was predicted using SWISS-MODEL, respectively. Each enzyme model was validated by different parameters, including Ramachandran’s plot and QMEAN scoring. PROCHECK was then used to further evaluate the results. The 3D structure of xylan substrate was obtained from PubChem. AutoDock was then used for docking study. According to amino acids involved in hydrogen bond formation and binding energy, appropriate docking results were selected. The docking results were finally visualized using PyMol (Thatoi et al., 2022).

To evaluate the synergism between Xyn and XynB, the hydrolysis experiments were investigated by combining Xyn and XynB in 1:1, 2:1, 4:1, and 8:1 ratios with 1% (w/v) beechwood xylan prepared in 50 mM citrate buffer (pH 6.0) at 80°C, 180 rpm in a bench top hybridization incubator. Three sets of hydrolysis were conducted, including Xyn/XynB alone or in combination hydrolysis. The amount of reducing sugars were measured using 3.5-dinitrosalicylic acid (DNS). The following equation was used to calculate the degree of synergism between Xyn and XynB: DS = A mixture ∑ A individual where A _mixture is the reducing sugar content achieved with Xyn and XynB added together, and Σ _individual is the sum of reducing sugar content released with the individual enzymes.

The hydrolysis experiment was carried out by hydrolyzing pretreated corncob and sugarcane bagasse at 10% (w/w) solids loading. The alkali-catalyzed glycerol pretreatment was conducted according to previously described procedures (Long et al., 2020a). The reaction mixtures were shaken at 180 rpm. Two-stage hydrolysis was conducted by first prehydrolyzing the substrates using thermostable Xyn and XynB at 85°C for 6 h, which was then followed with common hydrolysis by using Celluclast 1.5 L and Novozymes 188 at 50°C. Samples for sugar analysis were collected after 4, 8, 12, 24, 48 and 72 h of hydrolysis. Glucose concentration after hydrolysis was measured using HPLC equipped with an Aminex HPX-87H column. At the end of hydrolysis, samples were boiled at 100°C for 10 min to inactivate the enzymes. The supernatants were collected by centrifugation at 10,000 g for 10 min. All experiments were performed twice.

Bacillu subtilis have been developed as one of the promising candidates for heterologous proteins secretion. The expression vector pWB980, derived from pUB110 is commonly used for producing foreign proteins due to its high copy number and high stability. However, the transformation efficiency of B. subtilis as the primary host is low, limiting the application of pWB980 (Zhao et al., 2020). To improve the transformation efficiency of this plasmid, a new B. subtilis-E. coli shuttle vector pWB980-20b was first constructed on the basis of pWB980 in this study. The process of construction was shown in Figure 1. A 1841 bp fragment containing Ap^R gene and the ori for E. coli was amplified from the Gram-negative plasmid pET-20b by PCR, resulting in fragment I. The P43 promoter, Km^R gene, ble, and the plus ori for B. subtilis were amplified from pWB980 by reverse transcriptase PCR, resulting in fragment II. After digestion with NdeI and ligation of the two resulting fragments, shuttle vector pWB980-20b was constructed. pWB980-20b was in the size of 5,631 bp. With introduction of the ori for E. coli, pWB980-20b was capable of conducting the initial cloning steps in E. coli and then transferring the recombinant plasmid to B. subtilis. Transformants with this plasmid can be either selected with Ampicillin in E. coli or Kanamycin in B. subtilis.

To express the xylanase (Xyn, Genbank No. WP_013932897) and the β-xylosidase (XynB, Genbank No. WP_013931466) from Pseudothermotoga thermarum DSM5069 in B. subtilis, the DNA fragments of Xyn and XynB genes, encoding 1,177 and 774 amino acids with predicted pIs of 5.31 and 6.12, were amplified from the optimized synthetic fragments, respectively. Xyn was a modular protein containing three N-terminal carbohydrate binding domains (CBMs), two C-terminal CBMs, and a catalytic domain (CD) (Figure 2). The CD (L487-K812) displayed considerable similarity to catalytic domains of glyosidic hydrolase family 10 (GH10). XynA from Thermotoga sp. Mc24 (Genbank No. WP_038052050) exhibited the highest homology (66.4%). The N-terminal noncatalytic region of Xyn contained three tandem family 4_9 CBMs. Teo et al. reported that the presence of CBM4_9 in xylanase from Roseithermus sacchariphilus resulted in an increase in NaCl tolerance (∼17%) and turnover rate (∼40%) of the catalytic domain (Teo et al., 2019). Kim et al. found that the CBM4_9 family module in 1,4-β-xylanase KRICT PX3 from Paenibacillus terrae HPL-003 was essential for assisting the hydrolysis of insoluble xylan by promoting the binding to xylan (Lim et al., 2016). The C-terminal domain contained two tandem CBM9s, which were located at G828-E987 and G992-T1176. CBM9 played a role in driving the enzymes to their targeted insoluble cellulose and soluble sugars, including soluble hexose sugars, xylan and glucopyranoside-based polymers (Boraston et al., 2001). The β-xylosidase from P. thermarum in this study included a specific glycosyl hydrolase family 3 (GH3) domain. Its non-specific hits included a N-terminal domain β-glucosidase-related glycosidase (BglX). These elements might form the β-D-glucoside glucohydrolase (PRK15098).

Expression plasmids pWB980-20b-Xyn and pWB980-20b-XynB were then constructed and transformed into B. subtilis WB800 to express the enzymes, respectively. To measure the expression level of Xyn and XynB, enzyme activities of the intracellular and extracellular fractions of the recombinant bacteria were determined (Table 2). As shown in Table 2, the total enzymatic activity was 147.2 U/mL for Xyn and 220.8 U/mL for XynB. Both of the enzymatic activities of the two enzymes expressed in B. subtilis were approximately 1.5 times as high as those expressed in E. coli (94.3 U/mL for Xyn and 150.3 for XynB), indicating that B. subtilis was a potential candidate to achieve a high level expression of the enzymes for subsequent application. The recombinant proteins were then purified using Ni²⁺ affinity chromatography after being heated at 60°C for 30 min. The purified Xyn and XynB showed single bands on SDS-PAGE with molecular masses of about 130 kDa and 85 kD, respectively, which were identical to their calculated masses, stating that the target proteins were successfully expressed in B. subtilis (Figure 3).

After purification, biochemical properties of the enzymes (e.g., optimal pH, optimal temperature, and pH thermostability) were tested systematically to determine the optimal conditions for the subsequent synergistic hydrolysis. It showed that the purified Xyn exhibited maximal activity at pH 6.5 and 95°C (Figures 4A, B). The Xyn was stable between pH 5.5 and 8.5, retaining >80% of the initial activity after 2 h of incubation at 85°C. When pH < 5.5, the enzyme was rapidly inactivated (Figure 4C). As for the temperature thermostability, Xyn retained more than 80% of its activity after incubating with 5 mM Ca²⁺ at 90°C for 0.5 h, however, it was observed that this enzyme was unstable at its optimal temperature (95°C) and lost ∼80% of its activity after incubation for 0.5 h (Figure 4D). When various additives were added into the reaction respectively, the activity of Xyn was not affected by MgCl₂, AlCl₃, Tween 60 and EDTA, but slightly affected by Tris. The activity of Xyn was significantly boosted by CaCl₂ (Table 3). With 5 mM of CaCl₂ addition, the Xyn activity was greatly enhanced by 394%. However, Xyn activity was strongly inhibited by CuCl₂, only remaining ∼10% of its activity in the solution. This was due to Cu²⁺ could not only bind thiol groups of amino acids, but also interact with their carboxyl groups or imidazole (Yin et al., 2010). These results suggested that Ca²⁺ might play a role in keeping the enzyme in a catalytically active conformation.

The optimal pH and temperature for purified XynB were 6.5 and 95°C (Figures 4A, B), which possessed the general properties of previously reported β-xylosidase that were more active in slightly acidic solution (pH 4.5-6.5) (Wang and Arioka, 2021; Mercado-Flores et al., 2022; Zafar et al., 2022). The purified XynB was stable over a wide range of pH (5.5-8.5), retaining >80% of its initial activity after 2 h of incubation at 80°C (Figure 4C). When the temperature increased, the remaining activity of XynB decreased rapidly. Similar to Xyn, XynB was also unstable at its optimal temperature (95°C), only retaining ∼20% of the initial activity. However, at 80°C, it showed approximately 80% of its initial activity (Figure 4D). The effects of additives on XynB activity were also investigated (Table 3). XynB activity was strongly inhibited by CuCl₂, but not affected by most of the additives, including MgCl₂, MnCl₂, CaCl₂, Tris, Tween 60 and EDTA. ZnCl₂, CoCl₂ and AlCl₃ slightly inhibited XynB activity. The capacity to resist these metal ions and chemical reagents suggested that XynB could survive in a variety of scenarios.

To further understand the properties of the two enzymes, molecular docking analysis of modeled enzymes with substrates was also conducted. The predicted 3D models of Xyn and XynB were obtained by the SWISS-MODEL, respectively. Both predicted Xyn and XynB models showed >95.0% residues in Ramachandran’s favorable region with QMEAN score less than 2 (Supplementary Figure S1). These results indicated that the predicted models were of good quality and could be used for docking.

The predicted 3D models Xyn and XynB were then docked with the substrate xylan by AutoDock software. The complex with binding affinity less than −5 kcal/mol was selected for the interaction studies. Figure 5A showed the interaction between Xyn and the substrate xylan. It appeared that Glu525, Asn526, Trp774 and Arg784 were all involved in the binding and catalysis of xylan. The bond length between residues and substrate was 2.2, 2.1, 2.0 and 2.1, respectively. These docking results suggested that Glu525, Asn526, Trp774 and Arg784 were critical for Xyn-xylan interaction. The complex was further stabilized by the surrounding hydrophobic amino acids, resulting in the further improvement of enzyme activity and thermostability (Saleem et al., 2021). Compared with the molecular docking results of the xylanase from Thermotoga maritima, same residues were involved in the binding interaction between enzyme and xylan substrate (Yang and Han, 2018). However, the docking results of xylan with xylanase from Geobacillus thermodenitrificans C5 showed the involvement of Glu, Pro, Arg, Trp and His (Irfan et al., 2018). The interaction of xylopentose with xylanase from Thermotoga petrophila RKU1 analyzed through molecular docking also involved Glu525, Asn526 and Trp774 residues with substrate (Shahid et al., 2023). However, residues position is different for all these xylanases.

Additionally, molecular docking study of the XynB-xylan complex was carried out using AutoDock, and the results were shown in Figure 5B. The fissure structure of XynB provided favorable conditions for the protein to release the binding of small molecule substrate. It showed that Val237, Lys238, Val761 and Asn764 were all involved in the binding and catalysis of xylan. Lots of hydrogen bonds were formed by these residues, endowing the enzyme with favorable heat resistance during hydrolysis. Generally, stable hydrogen bonds were formed at distances less than 3.5 Å. The distances of residues to xylan were observably shorter at 1.9-2.8 Å, suggesting that a stably bound XynB-xylan complex. It was reported that heat resistant ability of the enzyme was correlated with the number of hydrogen bonds formed in the reaction (Liu et al., 2023). However, compared with Xyn, which had higher heat resistant ability with fewer hydrogen bonds than that of XynB, indicating that there were still other factors might also influence enzyme thermostability (Nezhad et al., 2022). These results showed some key residues’ information of both enzymes, which could further instruct enzyme directed evolution to improve their properites.

To investigate the synergism of Xyn and XynB, varying ratios were incubated with beechwood xylan. The resulting saccharides were determined by DNS. As shown in Figure 6, Xyn exhibited higher activity than XynB when hydrolyzing beechwood xylan. The reducing sugar concentrations were 18.4 mM for Xyn and 4.8 mM for XynB.

When combining the Xyn and XynB together for xylan hydrolysis, a significant synergistic effect was observed with a considerably increased reducing sugar content. When different Xyn: XynB ratios were used, the degree of synergy was 1.1 (1:1), 1.3 (1.5:1), 1.6 (2:1) and 1.1 (4:1), respectively. Combining the two enzymes in a 2:1 ratio liberated the most reducing sugar from xylan. The total amount of reducing sugar released was about 53% higher than that of Xyn alone. These results were similar to those reported previously, which also observed the synergistic effect of xylanase and β-xylosidase on xylan deconstruction (Zhou et al., 2012; Yang et al., 2014; Fanchini Terrasan et al., 2016). Similar to cellulose degradation, hydrolysis of xylan also required the cooperation of enzymes to work synergistically. Xylanase initially generated soluble oligosaccharides from xylan by directly cleaving glycosidic bonds. The released oligosaccharides were further converted to xylose by β-xylosidase, and simultaneously reduced the inhibition of xylanase by its reaction products (Van Dyk and Pletschke, 2012). Since shorter oligosaccharides are more favorable for generating xylose by β-xylosidase, xylanase requirement is higher than that of β-xylosidase to achieve a high hydrolysis yield. Notably, the addition of Xyn presented a nonmonotonic effect on reducing sugar content for xylan hydrolysis, which was likely due to the acceleration of Maillard reaction by increasing protein loading (Wei et al., 2018). Thus, an over-high addition of enzymes was unfavorable to high temperature hydrolysis.

Since the recombinant Xyn and XynB exhibited obvious synergistic hydrolysis of model substrate (beechwood xylan), plus the commercial cellulase was active only at mild temperature, we next evaluated the synergistic effect of these two enzymes on hydrolysis of pretreated corncob (CC) and sugarcane bagasse (SCB) using two-stage hydrolysis. Chemical compositions of CC and SCB after pretreatment were shown in Table 1. It was apparent that Xyn and XynB pre-hydrolysis could facilitate the hydrolysis yield on both CC and SCB substrates, but the extent of improvement was highly substrate dependent (Figure 7). For example, although CC and SCB substrates contained same amount of xylan, the thermostable Xyn and XynB treatment could improve cellulose conversion by 11% for the CC substrate, while a less improvement (7%) was obtained on the SCB substrate. This was likely caused by the highly branched xylan structure of the SCB restricted the accessibility to enzymes. In addition, it appeared that the hydrolysis boosting effect of Xyn and XynB treatment was much higher at the first 24 h, which was probably due to the increased accessibility of cellulose after Xyn and XynB treatment.