Two Bifunctional β-Xylosidase /α-L-Arabinofuranosidases From GH43 With Different Structures in the xylan Degradation Strain of Paenibacillus Physcomitrellae XB Displayed the Similar xylo-oligosaccharides Degradation Ability

The strain Paenibacillus physcomitrellae XB isolated from moss of Physcomitrella patens was found have the xylan degradation ability, but its degradation characteristics and the related mechanism has not been revealed. In this study, Paenibacillus physcomitrellae XB exhibited different xylan degradation ability under the different substrates of corncob xylan (CCX), oat spet xylan (OSX), wheat our arabinoxylan (AX) and beech wood xylan (BWX). Genomic analysis showed that ~ 38 genes were related to xylan degradation, and quantitative real time RT-PCR showed that two glycoside hydrolase family 43 genes (Pph_0602 and Pph_2344) were up-regulated on 1% CCX and xylose. Substrate-specic experiments with puried proteins Ppxyl43A (Pph_0602) and Ppxyl43B (Pph_2344) revealed that both of them exhibited β-xylosidase activity toward chromogenic substrate p-nitrophenyl–D-xylopyranoside and α-L-arabinofuranosidase activity toward p-nitrophenyl-α-L-arabinofuranoside, indicating at least bifunctionality. Combined their degradation features on the natural substrates of different xylans with the hydrolytic products separated by thin-layer chromatography and high-performance anion exchange chromatography proles, it was found that both Ppxyl43A and Ppxyl43B were with the similar degradation ability on xylo-oligosaccharides (like CCX, OSX, xylohexaose and xylobiose). Both of them even could hydrolyze xylohexaose and xylobiose completely to xylose, but could not hydrolyze BWX and AX to produce xylooligosaccharides or xylose, suggesting they have no endo-xylanase activity and mainly hydrolyze xylo-oligosaccharides by β-xylosidase activity.

Lignocellulosic biomass is the most abundant renewable resource and can be obtained from agricultural residues, herbaceous grasses, and forest harvests. Xylan is one of the major polymeric hemicellulosic constituents of lignocellulosic biomass, and its effective utilization is crucial for the economical production of highly-valued substances [1].
Although many biomass-degrading enzymes characterized to date are from fungal secretomes, such as Thermothelomyces thermophile [10], more recently bacterial strains have been shown to exhibit strong xylan degradation ability and many related xylanolytic enzymes have been shown to have potential application value in biomass conversion [4]. Paenibacillus physcomitrellae strain XB is a strain isolated from moss [11]. Based on its putative xylanolytic gene structure and expression levels in the presence of xylan and xylose, we identi ed two genes from GH43 that encode bifunctional enzymes and analyzed their characteristics. We speculate they may be important for boosting xylo-oligosaccharide hydrolysis e ciency in biomass conversion.

Results
Xylan degradation characteristics of strain XB In order to determine the xylan degradation ability of strain XB, four kinds of xylan from different sources were used as substrates in the assay system. The result showed that much more reducing sugars were produced from the crude enzyme when using 1% OSX and CCX as substrate, with the highest value (3.28 mg/mL and 3.02 mg/mL, respectively) on the fth day. For AX and BWX, there was less reducing sugar detected, and the highest peak appeared on the third (1.59 mg/mL) and fourth day (1.58 mg/mL), respectively (Fig. 1a). Thus, strain XB had the ability to degrade all tested xylans, but its degradation ability was different on different xylan types.
Genomic features and gene prediction of xylanolytic enzyme of strain XB Genome assembly of strain XB resulted in 4 contigs, which were then combined into one scaffold. The genome was 4.83 Mb in size and the G+C content was 51.87 mol%. The NCBI Prokaryotic Genome Annotation Pipeline identi ed 4371 genes and 4198 proteins. Annotation with CAZy revealed strain XB had a relatively high proportion of CAZomes for 210 proteins (Additional le 2: Table S1). The glycoside hydrolase (GH) family accounted for more than 50% (107) of them, followed by the glycosyl transferase family (GT, 36), the carbohydrate esterase family (CE,19) and the carbohydrate binding module family (CBM, 47). Only one axillary activity (AA) family and none of the polysaccharide lyase (PL) families were detected in the CAZomes of strain XB. (Additional le 1: Fig. S1).

Putative xylan degradation gene expression induced by CCX and xylose
The relative expression levels of genes encoding GH8, GH11, GH10, six of GH43 and three GH51 families were measured via RT-qPCR using recA as the reference gene (Additional le 2: Table S2; Fig. 1b). The majority of genes exhibited up-regulation in R2A broth with 1% CCX, and several were over-expressed more than 5-fold (p < 0.01), such as Pph_1610 (GH10), Pph_0648 (GH11), Pph_0602, Pph_2344, Pph_2078 and Pph_3723 of GH43, Pph_4538 (GH51). Only two genes (Pph_0602 and Pph_2344) displayed extremely high expression levels when 1% xylose was added to the medium, up-regulating more than 5-fold (p-value < 0.01). Thus, Pph_0602 and Pph_2344 not only could be induced to overexpress at high levels in a substrate of CCX but also could be prompted by xylose. They were further analyzed based on protein expression, puri cation, and enzymatic characteristics.

Gene cloning and puri cation of recombinant enzymes
The coding region of two GH43 family encoding genes Pph_0602 and Pph_2344 were linked with expression vector pET28a (+) based on the two restriction enzymes Bam HI and Xho I for construction of recombinant proteins Ppxyl43A (pET28a-Pph_0602), and EcoR I and Xho I for recombinant proteins Ppxyl43B (pET28a-Pph_2344) (Additional le 2: Table S2). Ppxyl43A and Ppxyl43B contained open reading frames encoding a protein of 329 (~36.9kDa) and 543 (~62.3kDa) amino acids, respectively (Additional le 2: Table S3). The SDS-PAGE showed single bands consistent with the expected sizes of Ppxyl43A and Ppxyl43B after puri cation and washing with buffer containing 50 and 100 mM imidazole ( Fig. 2a and 2b).
In order to understand the reaction patterns of Ppxyl43A and Ppxyl43B, the reducing sugars and xylose amounts produced from 1% CCX assay showed that BL21 with either recombinant Ppxyl43A or Ppxyl43B could hydrolyze CCX and produce the signi cantly more reducing sugars (Fig.2c) and xylose (Fig. 2d) intracellularly than control (p<0.01). Whether for xylose or reducing sugars, recombinant Ppxyl43A exhibited an increasing trend intracellularly with induction time from 2 h to 18 h, and recombinant Ppxyl43B expressed higher activity levels from 2 h-induction and maintained a relatively stable level up to 18-h induction time. Moreover, Ppxyl43A and Ppxyl43B produced nearly similar amounts of reducing sugars (Fig.2c), while recombinant Ppxyl43B produced much more xylose than Ppxyl43A (Fig.2d). In addition, with a longer induction time (6 h and 18 h), some reducing sugars could be detected from both two recombinants and little xylose could be assayed mainly from recombinant Ppxyl43B extracellularly.

Substrate-speci c and hydrolytic products analyses
The substrate speci city of puri ed Ppxyl43A and Ppxyl43B was determined by measuring the amount of reducing sugars toward several polysaccharides and xylose toward pNP derivatives, including pNPX and pNPAf, as substrates. Ppxyl43A and Ppxyl43B displayed similar substrate degradation ability, and both could hydrolyze pNPX, pNPAf, and various types of polysaccharides, containing CCX, XOS, and OSX. Ppxyl43A was more active toward pNPX, and other polysaccharides, such as CCX, XOS, X2, and X6 (p<0.01) than Ppxyl43B after 10 min's digestion. However, neither Ppxyl43A nor Ppxyl43B could hydrolyze BWX and AX to reducing sugars ( Table 2).
In order to determine the action model of Ppxyl43A and Ppxyl43B, hydrolytic products were further investigated by TLC and HPAEC method. Comparing the TLC pro les of different substrates with their hydrolytic products, CCX and OSX used in this survey mainly contained xylobiose, Xylotriose, Xylotetraose, et cetera-thus, in fact, should belong to xylo-oligosaccharides. When digested with Ppxyl43A and Ppxyl43B, xylose could be produced from CCX and OSX (Figure 3a), and their pro les were similar with those from XOS, but BWX could not be hydrolyzed by either Ppxyl43A or Ppxyl43B to produce xylo-oligosaccharides or xylose ( Figure 3b) and only very little xylose was also produced from AX (Additional le1: Fig. S2). The TLC hydrolysis pro le of X6 and X2 exhibited that both Ppxyl43A and Ppxyl43B could hydrolyze X6 and X2, even completely, to xylose ( Figure 3c). Thus, All TLC pro les showed that Ppxyl43A and Ppxyl43B mainly hydrolyze CCX and OSX via β-xylosidase activity. In addition, although very weak TLC signals of arabinose could be observed from the substrate of pNPAf (Additional le1: Fig. S2), it also re ected Ppxyl43A and Ppxyl43B could hydrolyze pNPAf to arabinose and supposed that both had α-L-arabinofuranosidase activity. Thus, both Ppxyl43A and Ppxyl43B possess at least bifunctional enzyme activity for β-xylosidase and α-L-arabinofuranosidase. The hydrolytic products separated by HPAEC pro les also con rmed xylose could be produced from X6 (Fig. 4a, 4b and 4c) and X2 (Additional le1: Fig. S3) and arabinose was produced from pNPAf ( Fig. 4d, 4e and 4f) after digestion by Ppxyl43A and Ppxyl43B.
In addition, According to the TLC hydrolytic pro le at different reaction times, Ppxyl43A displayed more rapid CCX hydrolytic activity than Ppxyl43B. Xylose could be detected at 3 min post-incubation with Ppxyl43A, and the amount increased signi cantly within 1 h (Fig. 5a). The xylose signal could not be observed until 60 min post-incubation by Ppxyl43B (Fig. 5b).
Effects of pH and temperature on activity and stability The effects of pH on β--xylosidase and α-L-arabinofuranosidase activity of Ppxyl43A and Ppxyl43B further examined using pNPX and pNPAf displayed that the optimum pH value was 6.0 for β--xylosidase but 5.0 for α-L-arabinofuranosidase for Ppxyl43A (Fig. 5a). Ppxyl43A retained more than 60% β-xylosidase activity in pH range 3.0 to 8.0 but only retained ~60% of α-L-arabinofuranosidase activity in pH range 4.0-6.0. The pH stability assay showed that highest stability occurred in pH range 5.0-8.0 for β-xylosidase and 6.0-8.0 for α-L-arabinofuranosidase activity (Fig. 5b). However, for Ppxyl43B, the optimal pH for β--xylosidase was 5.0 and 3.0 for α-L-arabinofuranosidase activity (Fig. 5c). β--xylosidase could act in the widest pH range, retaining 80% or more activity over the entire detected pH range of 2.0 to 9.0; whereas, the relative activity of α-L-arabinofuranosidase was only 60% of total within pH range 3.0-6.0. The pH stability assay showed that both enzymes could retain 60% relative activity from pH 3.0-9.0 (Fig.  5d). Thus, there were notable differences in suitable pH ranges for the two enzymes-Ppxyl43B had a wider pH range and better pH stability than Ppxyl43A.
The optimal temperature of puri ed Ppxyl43A for α-L-arabinofuranosidase activity was 40°C, while that for β--xylosidase was 50°C (Fig. 6a). Thermostability assays showed that Ppxyl43A had a relatively high activity only when the temperature was lower than the optimum and could retain more than 50% activity across the temperature range 10-40°C for α-L-arabinofuranosidase and 20-50°C for β--xylosidase (Fig.  6b). While the optimal reaction temperature of Ppxyl43B was 40°C for both enzymes, β--xylosidase activity level were similar across the entire range of 20-70°C (Fig. 6c), and they displayed high thermostability in temperatures ranging from 10 to 70°C (Fig. 6d). Thus, Ppxyl43B exhibited a wider temperature range and higher thermostability than Ppxyl43A, especially when using pNPX as substrate for β--xylosidase activity.

Effects of metal ion on enzyme activity
Among the 10 kinds of metal ions detected, the β--xylosidase activity of Ppxyl43A and Ppxyl43B could be enhanced by 10 mM Ca 2+ and Mn 2+ by approximately 10% over the control group, slightly inhibited by Fe 3+ and Ni 2+ (10-30% loss of activity) and strongly inhibited by Zn 2+ and Cu 2+ by loss of 70-80% activity.

Kintetic properties
Kinetic parameters such as Vmax and Km were determined from regression lines of Lineweaver-Burk plots using pNPX and pNPAf, respectively ( Table 4). The Vmax and Km values obtained for β--xylosidase of Ppxyl43A based on the substrate of pNPX were 8.64±1.87 U/mg and 2.37±0.54 mM, while those for Ppxyl43B were 13.49±0.27 U/mg and 2.12±0.05 mM, respectively; when using CCX as the substrate, the detected Vmax and Km values were 481.48±159.30 U/mg and 88.44±48.16 mg/ml for Ppxyl43A, and 316.80±217.95 U/mg and 75.16±34.35 mg/ml for Ppxyl43B, and there were no signi cant differences between Km values of two proteins (p>0.05). In addition, the kinetic parameters for pNPAf showed that the average Vmax and Km values were 5.73±5.35 U/mg and 3.03±1.64 mM for Ppxyl43A, were slightly higher than those of Ppxyl43B (0.86±0.60 U/mg and 0.82±0.28 mM) but not signi cantly. Thus, these two proteins showed similar a nity to both chemical and natural substrates.

Discussion
In this study we characterized the xylan degradation ability of strain Paenibacillus physcomitrellae XB and two β--xylosidase/α-L-arabinofuranosidase bifunctional enzymes of GH43 family in strain XB were puri ed and analyzed, this is the rst time to determine its xylan degradation characteristics and mechanism of strain XB. The result showed that it had the ability to degrade all tested xylans, but its degradation ability was different on different xylan types. Of them, BWX and AX used in this study were insoluble, and their structure might be much more recalcitrant than CCX and OSX, hence more di cult to hydrolyze.
Screening the whole genome of strain XB, approximately 38 genes might be related to xylan degradation, which contained the putative encoding genes of endo-xylanase, βxylosidase, α-N-arabinofuranosidase, α-glucuronidase, α-galactosidase, α-xylosidase and peptidoglycan/xylan/chitin deacetylase. Of all these xylanolytic enzymes, β-xylosidase had the highest structural diversity and could hydrolyze more glycosidic bonds than any other xylanolytic enzymes, thus, was considered as a crucial enzyme in xylan degradation. Previous reports revealed that GH43 was the second largest β-xylosidase-containing family, even several of them have been found to be bi/multifunctional [12] and played important roles in the xylan degradation [7,8,13]. However, most of the characterized β-xylosidases, to some extent, inhibited themselves by xylose, arabinose, glucose, and/or other monosaccharides [14,15]. Even the β-xylosidase in some GH43 family members was inhibited 25-66% by D-xylose [12]. However, in this survey both Pph_0602 and Pph_2344 were predicted as β-xylosidase of GH43 family and up-regulated in the medium supplemented with xylose and CCX, suggesting that β-xylosidase expression might be improved by xylose in some cases. Due to this interesting point, their enzymatic characteristics were worthy of further analyses.
Assessment of the reducing sugars and xylose produced by the two recombinants of Ppxyl43A (pET28a-Pph_0602) and Ppxyl43B (pET28a-Pph_2344) revealed that some reducing sugars and xylose could be detected extracellularly with a longer induction time (6 h and 18 h), which re ects that part of Ppxyl43A or Ppxyl43B could be exported outside the cell. However, the two proteins of Ppxyl43A and Ppxyl43B did not possess N-terminal signal peptides according to SignalP prediction results (Additional le 2: Table S3), and therefore, should only express intracellularly. Whereas previous reports found that some proteins can be exported without a classical N-terminal signal peptide, and this kind of secretion is known as leaderless secretion or non-conventional/non-classical secretory pathway [16]. Further prediction by SecretomeP 2.0a Server (SecP scores were 0.91 and 0.84 for Ppxyl43A and Ppxyl43B, respectively) exceeded the threshold of 0.5 for bacterial sequences (Additional le 2: Table S3) supposed that Ppxyl43A or Ppxyl43B might also secret outside the cell by this kind of leaderless pathway. In this case, these proteins may play important roles in CCX degradation by strain XB by exporting of outside the cells. In addition, whether for xylose or reducing sugars, recombinant Ppxyl43A exhibited an increasing trend intracellularly with induction time longer, and recombinant Ppxyl43B expressed relatively stable and higher activity levels from 2 h-to 18-h induction time; moreover, although recombinant Ppxyl43A and Ppxyl43B produced nearly similar amounts of reducing sugars, recombinant Ppxyl43B could produce much more xylose than Ppxyl43A. All of these results implied that the action patterns of Ppxyl43A and Ppxyl43B were not completely same.
The substrate speci c assay showed that both Ppxyl43A and Ppxyl43B possess at least bifunctional enzyme activity for β-xylosidase and α-L-arabinofuranosidase based on the chemical substrate of pNPX and pNPAf. According to the different kinds of natural xylan substrates and their hydrolysis products, it was con rmed that Ppxyl43A and Ppxyl43B mainly hydrolyze CCX and OSX via β-xylosidase activity and could not hydrolyze BWX by endo-xylanase activity, which also suggests that although strain XB has the ability to degrade BWX and AX, this degradation likely needs endo-xylanase activity expressed by other proteins. Further predicting their structure by SWISS-MODEL and InterPro Scan Server, Ppxyl43A showed 65.09% sequence identity with 5glk.1.A, which is a monomer structure of CoXyl43 that belongs to the GH43_AXH-like subgroup [17]. The CoXyl43 structure is composed of a single catalytic domain that consists of a ve-bladed β-propeller [18] and proved to be with β-xylosidase and α-arabinofuranosidase activity to promote plant biomass sacchari cation by degrading xylooligosaccharides into xylose [19]; while Ppxyl43B displayed 57.71% sequence identity with the template of 5zqj.1.A [20] and also had a domain with a ve-bladed β-propellor belonging to the GH43_XybB subgroup with β-1, 4-xylosidase and α-L-arabinofuranosidase activity. In addition, both Ppxyl43A and Ppxyl43B contained three acidic residues (Asp-20, Asp-140 and Glu-227 for Ppxyl43A; Asp-14, Asp-127 and Glu-187 for Ppxyl43B) (Additional le 2: Table S3), which were considered active sites. Thus, the detected bifunctional enzyme activity of Ppxyl43A and Ppxyl43B was generally consistent with the predicted results. Moreover, the kinetic parameters of these two proteins also con rmed their a nity with substrates of bifunctional β-xylosidase/α-L-arabinofuranosidase.
However, the enzyme characteristics of Ppxyl43B exhibited a wider pH and temperature range, better pH stability and thermostability than Ppxyl43A, especially when using pNPX as substrate for β--xylosidase activity. Moreover, the xylose was released much more slowly from CCX when digested by Ppxyl43B (at 60 min post-incubation) than by Ppxyl43A (at only 3 min post-incubation). Further comparing the structure of these two proteins, it was found that besides the domain of a ve-bladed β-propellor, Ppxyl43B also contains an additional β-xylosidase C-terminal Concanavalin A-like domain (GH43_C2), which has a sandwich structure of β strands with a complex topology. Its structure was considered like a long V-shaped groove, forming a single extended substrate-binding surface across the face of the propeller, which probably restricts access to a portion of the active site forming a pocket [12]. Hence, it might be more di cult for substrates to bind to this active site than those without this C-terminal domain, which may be the reason that why xylose was produced much more slowly by Ppxyl43B. This extra Cterminus domain was also considered as an evolutionary remnant that might be unnecessary for catalytic function but might aid in protein stability [21], which might also be the reason that why there was the relatively higher enzyme stability in Ppxyl43B than Ppxyl43A.
In addition, both two enzyme activities of Ppxyl43A and Ppxyl43B could be in uenced by many metal ions, and less of them in uenced enzyme activity for Ppxyl43B. In fact, the β-xylosidase and αarabinofuranosidase activity of recombinant CoXyl43 has been ever reported to be inactivated by zinc and copper ions but were activated by manganese ions, while only β-xylosidase activity was dramatically enhanced by calcium ions [21]. Another GH43 family member of RS223-BX was also found to be enhanced in β-xylosidase activity by calcium ions, cobalt, ferrous, magnesium, manganese and nickel ions [22]. Thus, the effects of cations on the activity of GH43 family have been frequently observed and may be caused by high accessibility through the central tunnel [23]. In this survey, Ppxyl43B had an additional C-terminal domain and less substrate binding sites (Additional le 2: Table S3) than Ppxyl43A might restrict access for substrates to the active site to form a pocket to nish the hydrolysis reaction [12], which might be the reason why cations rarely in uenced the enzyme activity for Ppxyl43B.
In summary, although there were some differences between Ppxyl43A and Ppxyl43B, like their structures, and the optimal enzyme activity conditions (pH, temperature, pH stability and thermostability, and metal ions), they also displayed the similar enzymatic features, such as the ability of degrading natural xylooligosaccharides and chemical substrates based on β-xylosidases or α-L-arabinofuranosidases activity and exporting outside the cell by non-conventional way. In addition, the encoding genes Pph_0602 and Pph_2344 of Ppxyl43A and Ppxyl43B also expressed signi cantly at high levels under the 1% CCX and xylose suggesting that they might be stimulated by xylose or other xylo-oligosaccharides, which would also be a special advantage in their application for biomass conversion by strain XB in the future. Certainly, much more work is needed to understand their reaction mechanisms in detail, and many more xylanolytic genes in strain XB need to be analyzed for function before the application of strain XB on bioconversion of lignocellulosic biomass would be possible.

Conclusions
Paenibacillus physcomitrellae XB was proven to have the ability to degrade different kinds of xylan. Its genomic features showed that approximately 38 genes might be involved in xylan degradation. Two genes of GH43 family (Ppxyl43A and Ppxyl43B) were signi cantly up-regulated under the induction of both xylose and CCX. The enzyme characteristics of puri ed proteins showed that both of them at least were bifunctional enzymes with the similar a nity with the substrates of β-xylosidase and α-Larabinofuranosidase. However, Ppxyl43B displayed higher stability across pHs, temperatures, and in the presence of metal ions than Ppxyl43A, which might be related to an extra C-terminus domain in Ppxyl43B. Future research would be helpful in developing the application of these two proteins in the xylan bioconversion process by strain XB.

Strain cultivation and chemicals
Paenibacillus physcomitrellae XB was maintained at -80°C in R2A broth with 30% glycerin; for use in this study, it was activated by inoculation into fresh R2A medium and cultured at 30°C in a shaker at 180 rpm.

Xylan degradation assay
Paenibacillusphyscomitrellae XB was cultured in R2A broth (pH 6.0) at 30°C for 24 h and then activated by inoculation into R2A medium with 1% CCX, OSX, BWX, and AX, respectively. In order to test the ability to degrade these xylans, the strain precipitate was collected at 2 and 7 days post incubation, centrifuged at 8000 rcf for 15 min at 4°C, and the supernatant, which was used as the crude enzyme, was transferred to a new tube. The assay was performed in 50 mM acetic acid-sodium acetic buffer (pH 5.0) containing 1% substrate. The reaction system used 100 µL of different substrate and 100 µL of each crude enzyme.
After 10 min incubation at 50°C, samples were immediately added into 300 µL of 3, 5-dinitrosalicylic acid (DNS) and boiled on a heating block at 100 °C for 10 min before being cooled on ice for 5 min. Finally, the reducing sugars were determined by DNS colorimetry [24] with xylose as the standard. The absorbance was measured using a microplate reader Varioskan Flash, Thermo Scienti c at 540 nm. All reactions were assayed at least in triplicate.
Genomic DNA extraction, sequencing, and annotation for strain XB To obtain DNA for sequencing libraries, strain XB was rst inoculated into 200 mL of liquid R2A broth, incubated in a shaker at 30℃ for 24 h, and harvested by centrifugation at 13000 rpm for 2 min. Subsequently, genomic DNA extraction, quality detection, sequencing and raw reads processing were performed according to the previous method (Wang et al. 2020). Finally, a de novo genome was assembled using SOAP version 2.04 [25].

Expression analysis of putative xylanolytic genes based on RT-qPCR
Total RNA was extracted from strain XB samples incubated in R2A broth with 1% CCX, 1% xylose and without both of them (control), at 30°C for ve days. All the protocols of RNA extraction, reverse transcription, cDNA synthesis and real-time PCR were performed as the previous report [33]. Primers used for the reference gene recA and 12 xylanolytic genes are shown in additional le 2 (Table S2). The relative fold change of each gene was calculated based on the of 2 -ΔΔCq method.
Ampli cation, cloning, and differential expression of Pph_0602 and Pph_2344 High delity PCR was performed to obtain the full-length of the two highly expressed genes of Pph_0602 and Pph_2344 based on primers with restriction enzyme sites (Additional le 2: Table S2). Then, the two gene fragments and vector pET28a (+) were ligated with T4 DNA ligase (Takara Co.) to construct recombinant plasmids, which were then transformed into E. coli BL21(DE3) competent cells.
Recombinant enzyme induction was performed with 1 mM IPTG at 37°C for 2 h, 6 h and 18 h, and puri cation using Ni 2+ -nitrilotriacetic acid (NTA) metal-chelating a nity chromatography was completed [34]. The purity of the recombinants named Ppxyl43A (pET28a-Pph_0602) and Ppxyl43B (pET28a-Pph_2344) was evaluated via sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Protein concentrations were determined according to Bradford's method using bovine serum albumin as the standard [35]. Finally, the reducing sugar in the crude enzyme was assayed in all recombinants and the control (only with empty vector of pET28a (+)) by DNS method [24]. Xylose was detected by using 50 mM acetic acid-sodium acetic buffer (pH 5.0) containing 200 μL of p-nitrophenol-β-D-xylopyranoside (pNPX) (4 mM) and the crude enzyme. The mixture was incubated at 50°C for 15 min and stopped by 0.4 mL of 2 M Na 2 CO 3 . The xylose was calculated according to the p-nitrophenol (pNP) level measured via absorbance at 410 nm and compared with a standard reference [7].

Substrate speci c detection of recombinants Ppxyl43A and Ppxyl43B
In order to measure the activity of the two puri ed proteins of Ppxyl43A and Ppxyl43B,CCX, OSX, BWX, AX, X6, and X2 were used as the substrates. The reaction was performed in 50 mM acetic acid-sodium acetic buffer (pH 5.0) containing 1% substrate. After adding 100 μL puri ed Ppxyl43A or Ppxyl43B and incubating at 50°C for 10 min, the reducing sugars was detected as aforementioned methods. One unit of enzyme activity was de ned as the amount of enzyme which liberated 1 µmol reducing sugar per min. In addition, 4 mM of pNPX and p-nitrophenol-α-L-arabinofuranoside (pNPAf) were also used as substrates to determine the β-xylosidase and α-L-arabinofuranosidase activity respectively, via the amount of pNP released [7]. One unit of enzyme activity was de ned as the amount of enzyme which liberated 1 µmol pNP per min.
Analysis of hydrolysis products.
Hydrolysis products of X2, X6, pNPAf, AX, CCX, OSX, BWX and XOS were rstly determined by thin-layer chromatography (TLC) on silica gel 60 F254 plates of 20 cmХ20 cm (Sigma) with a mixture of n-butanol, acetic acid, and water (2:1:1) as a solvent system, after incubating 1% (w/v) of them with 0.5 units of PpAxy43A and PpAxy43B at 50°C (pH 5.0). Sugar spots were detected by heating the plates to 110°C after spraying them with 95:5 (v/v) of methanol: ammonium sulfate. Further analysis of the hydrolysis products of X6, X2, and p-NPAf was also completed with high-performance anion exchange chromatography (HPAEC, Dionex ICS-5000+, Thermo Scienti c) with pulsed amperometric detection (PAD) using a CarboPac PA20 column under a gradient elution (mobile phase: 90:10 of 2 mM NaOH and 500 mM NaOH) at 0.3 mL/min.
Effects of pH and temperature on the activity of Ppxyl43A and Ppxyl43B In order to observe the effects of pH on β-xylosidase and α-L-arabinofuranosidase activity, substrates pNPX and pNPAf were used to determine the enzymatic activities of Ppxyl43A and Ppxyl43B at different pHs (ranging from 2.0 to 9.0) at 50 °C using aforementioned methods. The pH was adjusted using Sodium Citrate Buffer (pH 3.0, 4.0 and 5.0), Disodium hydrogen phosphate-sodium dihydrogen phosphate buffer (pH 6.0, 7.0 and 8.0), and Tris-HCl buffer (pH 9.0), respectively. Activity level of 100% was de ned as optimal pH. The effect of temperature on the activity of Ppxyl43A and Ppxyl43B was measured under optimal pH at 10-80°C. Activity of 100% was de ned as the optimal temperature. The pH stability was assessed by incubating the enzyme in 50 mM of each pH buffer at the optimal temperature for 1 h. Thermostability enzyme activities was measured after preincubation of the enzyme at a temperature range from 10°C to 80°C for 1 h without the substrates. All the relative activities were calculated from the ratio of assayed enzyme activity of each treatment and that without treatment.
Effects of metal ions on enzyme activity of Ppxyl43A and Ppxyl43B The effects of metal ions on the activity of Ppxyl43A and Ppxyl43B were determined by using pNPX and pNPAf as substrate. The reaction was performed in 50 mM sodium acetic buffer (pH 5.0) containing 1% substrates and 10 mM corresponding metal compounds. The enzyme activity was determined as the method above described. Residual activity was calculated based on the control without metal compounds addition.

Kinetics and thermodynamics
The Km and Vmax values of Ppxyl43A and Ppxyl43B at optimal pH and optimal temperature (section 2.8) were determined by using pNPX (0.1 to 8.0 mM), pNPAf (0.1 to 8.0 mM) and CCX (0.1 to 2%) as the substrate, respectively. The Michaelis constants were determined from Lineweaver-Burk plots.