Introduction

The term “bioeconomy” refers to economic activities associated with the use of bioproducts. The improved utilization of unused resources is becoming increasingly important in the context of global warming and the energy crisis. Plant cell walls are the most abundant resource on earth, being mainly composed of cellulose, hemicellulose, and lignin, at proportions of approximately one-third each [1]. Xylan, the major hemicellulose, is the second most abundant biomass resource on earth next to cellulose [2]. Xylan consists of a β-1,4-linked xylose main chain with the substitution of α-1,3-linked arabinofuranose and an α-1,2-linked 4-O-methyl-glucuronic acid/glucuronic acid [3]. Because xylan is composed mainly of pentose, having a complex structure composed of several constituent sugars, and is more easily decomposed than cellulose by pretreatments, changing it into a fermentation inhibitor, research about hemicellulose lags behind that on cellulose and lignin and no effective methods to utilize xylan have been developed. However, it is important to use xylanolytic enzymes for the application of xylan because enzymes are molecules that can precisely discriminate the structures of substrates, even for substrates with heterogeneous and complex structures.

For the decomposition of xylan, xylanolytic enzymes such as β-xylanase (EC 3.2.1.8), β-xylosidase (EC 3.2.1.37), α-L-arabinofuranosidase (EC 3.2.1.55), α-glucuronidase (EC 3.2.1.139), and acetyl xylan esterase (EC 3.1.1.72), are needed [2]. Many microorganisms possess numerous different xylanolytic enzymes and are involved in the degradation of biomass in nature. To apply these enzymes for the utilization of xylan, it is essential to elucidate the substrate recognition mechanism of these enzymes; however, the substrate specificity of these enzymes, especially for branches in xylan, remains unclear.

We have studied the structure–function relationship and substrate specificity of xylanolytic enzymes such as β-xylanases [4,5,6,7,8,9,10,11,12,13], α-L-arabinofuranosidases [14, 15], and α-glucuronidases [16]. In this work, we focused on β-xylosidases and analyzed how branches in xylan affect their activity. Carbohydrate-active enzymes are classified in the CAZy database (http://www.cazy.org/) based on their amino acid sequences [17]. β-Xylosidases are classified into glycoside hydrolase (GH) families 1, 3, 5, 30, 39, 43, 51, 52, 54, 116, and 120.

Even substrate specificities of GH3 xylosidases from Aspergillus awamori and Trichoderma reesei toward branched xylooligosaccharides were analyzed [18,19,20], there are still a lot of unknown things regarding the substrate specificity of β-xylosidases.

Many β-xylosidases have been reported to have α-L-arabinofuranosidase activity as a side activity, which is considered to be advantageous for xylan degradation. However, such side activity is disadvantageous in preparing branched oligosaccharides. In this study, since the correlation between the β-xylosidase family and substrate specificity is poorly understood, GH39 and GH52 β-xylosidases from Bacillus halodurans C-125, a useful bacterium for xylan degradation [21], were characterized to clarify the mechanism of their substrate recognition for branches in xylan. The specificity of reducing-end xylose-releasing exo-oligoxylanase (Rex; EC3.2.1.156) belonging to GH8 [22] was also investigated.

Materials and Methods

Protein Expression and Purification

The genes encoding putative β-xylosidases such as rex (GH8), bhxyl39 (GH39), and bhxyl52 (GH52) were cloned from the genome of B. halodurans C-125. The genes were amplified by PCR with KOD-Plus-Neo (Toyobo, Osaka, Japan), using the following primer pairs: (for GH8: forward: CATATGAAGAAAACGACAGAAGGTGCATTTTG, reverse: AAGCTTGTGCCCCTTTG; for GH39: forward: CCATGGAAACAGTAGTTGTAAATGATCGTTC, reverse: AAGCTTATACGAAGGAATCAGCCGATC; and for GH52: forward: CATATGAATCGAGGAGAAACTGGTTATGAAACATG, reverse: GCGGCCGCTATCGGATAAATGGTT). The underlined sequences represent restriction sites. Before insertion of the PCR products into the expression vector, the amplified fragment was sub-cloned by Target Clone-Plus- (Toyobo, Osaka, Japan) and was confirmed by sequencing with a 3130 Genetic Analyzer (Applied Biosystems, Tokyo, Japan). rex, bhxyl39, and bhxyl52 were cloned between the NdeI and HindIII sites of pET30a (Novagen, Darmstadt, Germany), NcoI and HindIII sites of pET28a (Novagen), and NdeI and NotI sites of pET28a, respectively. The obtained pET30-rex, pET28-bhxyl39, and pET28-bhxyl52 plasmids were transformed into Escherichia coli BL21(DE3) (Merck KGaA, Darmstadt, Germany) by electroporation. The transformants were grown in Luria–Bertani medium containing kanamycin at 37 °C with shaking (200 rpm). Isopropyl β-D-thiogalactopyranoside (IPTG) was added to the culture at a final concentration of 0.1 mM when the optical density at 600 nm reached 0.2, and the culture was then shaken (200 rpm) at 25 °C for 22 h. The cells were collected by centrifugation (6000 rpm, 15 min, 4 °C), suspended in 50-mM phosphate buffer (pH 7.0), sonicated for 5 min, and then centrifuged (10,000 rpm, 30 min at 4 °C) to remove insoluble materials. The recovered supernatant was applied to HisTrap HP (GE Healthcare, USA) for purification using a 6× histidine tag at the C-terminus. The active fraction was collected and dialyzed against distilled water. The purity of the obtained purified enzyme was confirmed by SDS-PAGE [23].

Substrates

PNP-glycosides and xylans from beechwood, birchwood, and oat spelts were purchased from Sigma Chemical Company (St. Louis, MO, USA). Xylobiose (X2), xylotriose (X3), xylotetraose (X4), xylopentaose (X5), xyloheptaose (X6), and wheat arabinoxylan (low viscosity; 2cSt.) were obtained from Megazyme International (Wicklow, Ireland). In accordance with a previous report [24], branched oligosaccharides such as O-β-D-xylopyranosyl-(1 → 4)-[O-α-L-arabinofuranosyl-(1 → 3)]-O-β-D-xylopyranosyl-(1 → 4)-β-D-xylopyranosyl-(1 → 4)-β-D-xylopyranose (Ara3Xyl4) and O-β-D-xylopyranosyl-(1 → 4)-[O-4-O-methyl-α-D-glucuronopyranosyl-(1 → 2)]-β-D-xylopyranosyl-(1 → 4)-β-D-xylopyranosyl-(1 → 4)-β-D-xylopyranose (MeGlcA3Xyl4) were prepared as follows. The scheme is shown in Fig. 1. From 30.9 g of sugarcane bagasse, 640 mg of Ara3Xyl4 and 130 mg of MeGlcA3Xyl4 were obtained. The structure of the obtained oligosaccharide was determined by NMR, LC-MS, and crystal structure analysis. GH11 xylanase (SoXyn11B) was prepared in accordance with a previously reported method [12]. The amounts of pentose and hexose were determined as previously reported [25].

Fig. 1
figure 1

Preparation of substituted xylooligosaccharides

Full size image

Enzyme Assay and Substrate Specificity

To evaluate the activity against PNP substrates, a mixture of 25 μL of 2 mM substrate and 20 μL of McIlvaine buffer (pH 5.5 for BhXyl39, pH 6.2 for BhXyl52) was preincubated for 5 min, after which 5 μL of enzyme was added to the solution. The enzyme reaction was performed at 50 °C for 10 min, after which 50 μL of 0.2 M Na2CO3 was added to stop the reaction. Color development was measured by determining absorbance at 405 nm. To evaluate the substrate specificity of BhXyl39 and BhXyl52 to polysaccharides, the reactions were performed in McIlvaine buffer (pH 5.5 for BhXyl39, pH 6.2 for BhXyl52) containing 1% (w/v) substrate and 1-μM enzyme at 50 °C for 10 min. The reaction was stopped by heating the solutions at 100 °C for 20 min. The hydrolytic activity was determined by measuring the amounts of reducing sugars using the Somogyi–Nelson method [26]. The activity against xylooligosaccharides was measured as follows. A reaction mixture containing 30 μL of 100-μM substrate, 35 μL of 0.005% (w/v) L-fucose, 100 μL of McIlvaine buffer (pH 5.5 for GH39, pH 6.2 for GH52), 735 μL of DW, and 100 μL of enzyme (final concentrations of 400 nM for GH39 and 5.5 nM for GH52) was incubated at 30 °C for 0, 10, 20, 40, 60, 80, 100, 150, and 200 min, followed by inactivation by heating at 100 °C for 10 min. The hydrolysis products were analyzed by high-performance anion-exchange chromatography with a pulsed amperometric detection (HPAEC-PAD) system and a CarboPac PA1 column (4 × 250 mm) (Dionex Corp., Sunnyvale, CA) as described previously [6]. The decrease in the peak of oligosaccharide was quantified and the degradation rate was calculated. The substrate specificity of GH8 for branched oligosaccharides was determined as follows. The reaction mixture containing 500 μL of 100-μM substrate, 100 μL of McIlvaine buffer at pH 7.0, 200 μL of DW, 100 μL of 200-μM L-fucose, and 100 μL of 5-μM enzyme was incubated at 30 °C for 24 h. The reaction products were analyzed by Dionex HPLC. Hydrolysis products were also analyzed by MALDI-TOF MS on a REFLEX II (Bruker Daltonics) or a 4800 MALDI-TOF/TOF Analyzer (Applied Biosystems) in positive-ion mode.

Homology Modeling and Ligand-Docking Modeling

The homology modeling was conducted by the program MODELER 9v12 [27]. The models of BhXyl39 and BhXyl52 were calculated in single-template mode using structural models of Thermoanaerobacterium saccharolyticum B6A-RI β-xylosidase [XynB, PDB ID 1PX8, 62.0% amino acid identity with BhXyl39 [28]] and Parageobacillus thermoglucosidasius NBRC 107763 β-xylosidase [PDB ID 4C1P, 73.3% amino acid identity with BhXyl52 [29]], respectively.

MeGlcA3Xyl4 and Ara3Xyl4 binding models of BhXyl39 were constructed manually using COOT [30]. The ligand models were set in the catalytic pocket of the homology-modeled BhXyl39 so that the non-reducing xylose moiety was located at subsite −1, with reference to the position of xylose (X1) in the T. saccharolyticum β-xylosidase structure (XynB, PDB ID 1PX8). The X3 binding model of BhXyl52 was similarly constructed, with reference to the position of X2 in the P. thermoglucosidasius β-xylosidase structure (XynB, PDB ID 4C1P). These docking models were subjected to structure optimization using the program REFMAC5 [31].

X1 and X2 binding models of Rex were constructed by superimposition of the X2 model bound at subsites −1 and −2 from the X2 complex Rex structure (PDB ID 1WU6) onto the X1 complex Rex structure (PDB ID 1WU6) [32]. Structural drawings were prepared using the Cuemol2 program (http://cuemol.osdn.jp/en/).

Results

Expression, Purification, and Properties of BhXyl39

The construct in which bhxyl39 was cloned into pET28 was expressed in IPTG-inducible E. coli BL21(DE3). The recombinant enzyme was purified by affinity chromatography using the His-Tag encoded on the C-terminal of the protein expression product. The purified protein showed a single band on SDS-PAGE, with the expected molecular mass of 42 kDa (Fig. 2a).

Fig. 2
figure 2

SDS-PAGE of purified enzymes a Rex, b BhXyl39, and c BhXyl52. Lane 1, molecular mass marker (1 μg each band); lane 2, purified enzyme (1 μg)

Full size image

The effects of pH and temperature on BhXyl39 activity and stability were determined using PNP-Xyl as the substrate. Maximal enzyme activity was detected at pH 5.5 and 50 °C upon a reaction lasting 10 min. BhXyl39 was stable between pH 5.0 and 8.0 at 30 °C for 30 min and was also stable up to 40 °C during 30 min of incubation at pH 5.5 (Fig. 3).

Fig. 3
figure 3

Enzymatic properties of BhXyl39 and BhXyl52 A: BhXyl39, B: BhXyl52. a: The effect of pH on enzyme activity. b: The effect of temperature on enzyme activity. c: The effect of pH on enzyme stability. d: The effect of temperature on enzyme stability. Open circle, McIlvaine buffer; closed triangle, Atkins–Pantin buffer

Full size image

Expression, Purification, and Properties of BhXyl52

The construct in which bhxyl52 was cloned into pET28 was expressed in IPTG-inducible E. coli BL21(DE3). BhXyl52 was purified by metal affinity chromatography and subjected to SDS-PAGE (Fig. 2b). BhXyl52 was observed as a single band with a molecular mass of approximately 76 kDa, which was in agreement with the molecular mass expected from the amino acid sequence (78,580). The effects of pH and temperature on the activity and stability of the obtained purified enzyme were investigated. The properties of this enzyme are shown in Fig. 3b. The optimal pH of this enzyme was pH 6.2 and the optimal temperature was 50 °C. BhXyl52 was stable between pH 6.0 and 8.0 at 30 °C for 30 min and was also stable up to 40 °C during 30 min of incubation at pH 6.0.

Substrate Specificities of Rex

The activities of Rex for branched oligosaccharides such as Ara3Xyl4 and MeGlcA3Xyl4 were examined (Fig. 4). Rex showed hydrolytic activity for both substrates. MALDI-TOF MS analysis of Ara3Xyl4 and MeGlcA3Xyl4 hydrolysis products detected m/z 569 and m/z 627 as a sodium adduct ion [M + Na]+, respectively, suggesting that Rex released only reducing-end xylose from these oligosaccharides. m/z 437, which corresponds to an oligosaccharide having one arabinose and two xyloses, was also detected among the hydrolysis products of Ara3Xyl4, but this would be derived from contamination of oligosaccharides having branches at different positions. Because the excess amount of enzymes such as X4 was immediately degraded, it is considered that the reaction is very slow even if the enzyme cleaves a xylose linkage next to branched xylose.

Fig. 4
figure 4

Substrate specificities of Rex for substituted xylooligosaccharides A: MALDI-TOF MS analysis of Ara3Xyl4 hydrolysate with Rex. B: MALDI-TOF MS analysis of MeGlcA3Xyl4 hydrolysate with Rex

Full size image

Substrate Specificities of BhXyl39

BhXyl39 hydrolyzed only PNP-Xyl among PNP substrates and it had little effect on polysaccharides. When the enzyme acted on linear xylooligosaccharides, it hardly hydrolyzed X2. The enzyme hydrolyzed X3 at the highest hydrolysis rate, followed by X6 > X5 > X4 (Table 1). When the enzyme acted on branched oligosaccharides, the enzyme hydrolyzed MeGlcA3Xyl4 to xylose and MeGlcA3Xyl3 very rapidly. The hydrolysis rate was significantly higher than that of X3, which had the highest hydrolysis rate among the linear oligosaccharides. In contrast, the activity toward Ara3Xyl4 to release non-reducing-end xylose was almost the same as that of X4.

Table 1 kcat/Km of BhXyl39 and BhXyl52 for various xylooligosaccharides
Full size table

Substrate Specificities of BhXyl52

BhXyl52 also only acted on PNP-Xyl among the PNP substrates. No increase in reducing power was observed when the enzyme was incubated with various xylans (data not shown). When comparing the hydrolysis rates of linear xylooligosaccharides, the enzyme most rapidly hydrolyzed X2 and X3 at almost the same rate. The rates gradually decreased with increasing oligosaccharide length as follows: X2 = X3 > X4 > X5 > X6 (Table 1). Moreover, BhXyl52 did not hydrolyze the branched oligosaccharides, even when the amount of enzyme was increased 1000 times (data not shown).

Discussion

B. halodurans C-125 has been reported as an alkaline xylanase-producing bacterium [21]. Sequence analysis of genomic DNA of this strain has been completed and its xylanolytic enzyme system has been clarified. In this study, we focused on β-xylosidases because the substrate specificity of these enzymes against substituted oligosaccharides was poorly understood. Despite B. halodurans C-125 possessing β-xylosidase genes classified into GH1, 8, 39, 43, and 52 as candidates for study, GH1 and 43 were excluded from the targets because they are multi-functional enzymes and hydrolyze the linkages of both main and side chains.

To date, 13 GH39 xylosidases, including BhXyl39, have been characterized. As for BhXyl39, there are three reports of the characterization of recombinant BhXyl39 from three different research groups. In a report by Smaali et al. [33], the optimal pH of the enzyme was 7.5 at 55 °C. The enzyme hydrolyzed only PNP-Xyl among the tested PNP-glycosides and had no activity for birchwood xylan (the activity for PNP-Glc was detected, but it was 0.32% of that for PNP-Xyl). The activity of the enzyme for GH11 xylanase hydrolysate of wheat bran arabinoxylan was also investigated, but the hydrolysis rate and specificity for the contained individual oligosaccharides were not clarified. According to a report by Wagschal et al. [34], BhXyl39 showed optimal activity at pH 6.5, and the enzyme cleaved PNP-Araf and PNP-Arap in addition to PNP-Xyl. It was also described that only the release of xylose was observed by BhXyl39 from xylans from rye, wheat, oat-spelt, beech, and birch. In addition, Liang et al. [35] reported that the optimal activity of BhXyl39 was observed at pH 7.0 and 35 °C–45 °C, and it hydrolyzed PNP-Xyl, as well as ONP-Gal, PNP-Araf, PNP-Man, and PNP-Glc. A report has also described that a small amount of xylose was produced from beechwood xylan. In our experiments, we obtained results that differed from those in these previous reports. The optimal pH of BhXyl39 was 5.5 and the optimal temperature was 50 °C (Fig. 2). The optimal pH may shift significantly depending on the temperature at which measurement is performed. Regarding substrate specificity, synthetic substrates other than PNP-Xyl and polysaccharides were not hydrolyzed by BhXyl39. Since the activities on the substrates except for PNP-Xyl were extremely low in all cases, if activity was recognized at all, it is considered that this enzyme prefers a low-molecular-weight substrate and has almost no side activity other than β-xylosidase activity. Similar results were obtained with other characterized GH39 enzymes; therefore, the properties of not acting on a polymer substrate and hydrolyzing only xylosyl linkages in oligomeric substrates are considered to be properties common to the GH39 family. However, no members of the GH39 family of enzymes have had their substrate specificity tested for linear xylooligosaccharides of different lengths and branched oligosaccharides.

On the other hand, GH52 β-xylosidases have hardly been characterized. According to the CAZy database, to date only nine GH52 β-xylosidases have been reported. A β-xylosidase from Aeromonas caviae ME-1, one of the characterized enzymes, has been shown to degrade only PNP-Xyl among PNP substrates and it hydrolyzed a mixture of X1–X3 into X1. The enzyme showed optimal activity at pH 6.0 and 50 °C, and was stable in the range of pH 5–8 and below 40 °C [36]. The other characterized enzyme GH52 β-xylosidase from Geobacillus stearothermophilus had optimal activity at pH 5.5 and 70 °C and was stable in the range of pH 5–6.5 and at temperatures below 65 °C, and it was reported not to produce X1 from beechwood xylan [37]. The GH52 xylosidase from B. stearothermophilus T-6 showed maximal activity at 65 °C and pH 5.6–6.3. Catalytic amino acids were identified by sequence alignment and site-directed mutagenesis, but only activity against PNP-Xyl has been investigated [38]. BhXyl52 had an optimal pH of 6.2 and an optimal temperature of 50 °C. The enzyme did not hydrolyze chromogenic substrates other than PNP-Xyl and did not act on polysaccharides. Such substrate specificity was considered to be a property common to GH52 β-xylosidases.

Throughout the study of β-xylosidases, hydrolysis rates for limited linear oligosaccharides have been reported. However, only a few studies have been performed about the substrate specificity of a series of linear xylooligosaccharides and for branched oligosaccharides [39]. As described above, no studies have been performed of the substrate specificity for a series of linear xylooligosaccharides and branched oligosaccharides on GH39 and GH52. It is known that xylan has side chains of α-1,3-linked arabinofuranose and α-1,2-linked 4-O-methyl-glucuronic acid/glucuronic acid [3]. This difference has a significant effect on the three-dimensional structure. Since arabinose extends in the same direction as the main chain, the arabinose side chain and the glucuronic acid side chain have different effects regarding the recognition of xylose adjacent to branched xylose. In fact, GH10 xylanase produced oligosaccharides with different branched structures in the case of an arabinose side chain and a glucuronic acid side chain [9, 24]. With the exception of Rex, β-xylosidases generally recognize the non-reducing end of the substrates. Therefore, recognition of non-reducing-end xylose adjacent to branched xylose is undisturbed by the glucuronic acid side chain, but arabinose interferes with the recognition of the xylose residue. In fact, it has been reported that the mold enzyme (GH3) does not hydrolyze the linkage of xylose adjacent to a branch in oligosaccharides having an arabinose side chain [18]. However, since the GH3 enzyme has α-L-arabinofuranosidase activity as a side activity, it is unclear whether this enzyme does not really cleave such a linkage. We have confirmed that GH3 β-xylosidases icluding Trichoderma reesei enzyme released non-reducing terminal xylose then arabinose side chain when the substrates were incubated with large amount of enzyme that instantly degrades linear xylooligosaccharides (unpublished data). This problem of side activity makes it difficult to analyze the properties of β-xylosidase with respect to branched oligosaccharides, and the substrate specificity of β-xylosidase for branched oligosaccharides has yet to be clarified. Therefore, in the present study, the characteristics of GH39 and GH52 β-xylosidases for branched oligosaccharides were clarified in detail for the first time.

As shown in this paper, β-xylosidase activity changes with variation in the length of the main chain. Therefore, hydrolysis activities were compared using the same length of main chain. This led to some interesting results. For example, substrates with glucuronic acid side chains were preferred in GH39, but arabinose side chains did not affect the activity. Moreover, since it does not hydrolyze X2, it was considered that the GH39 enzyme recognizes three or more longer sugars. We used a structure model of BhXyl39 to understand how the enzyme discriminates substrates. The binding structures of BhXyl39 with MeGlcA3X4 and Ara3Xyl4 are shown in Fig. 5a and b, respectively. The hydroxyl groups at the 2- and 3-positions of the subsite +1 xylose face the solvent surface surrounded by a large pocket-like structure. It was thought that the arabinose residue fits in a large pocket structure, does not cause steric hindrance, and can be hydrolyzed the same as X4. In contrast, the carboxyl group of the glucuronic acid residue appears to form a hydrogen bond with the amino group of Lys235. The interaction between the carboxyl group of glucuronic acid and the amino group of basic amino acids is often observed in xylan-degrading enzymes [40, 41].

Fig. 5
figure 5

Sugar binding structures of Rex, BhXyl39, and BhXyl52 Xylooligosaccharide binding surface structural models at the catalytic site of β-xylosidases. a Xylose and xylobiose binding structure of Rex. b MeGlcA3Xyl4 binding structure of BhXyl39A. c Ara3Xyl4 binding structure of BhXyl39A. d Xylotriose binding structure of BhXyl52A. Xylose, MeGlcUA, and arabinofuranose moieties are drawn with green, cyan, and magenta ball-and-stick models. Catalytic residues and substrate binding residues (partial) are drawn with red and orange ball-and-stick models. Subsites are indicated by −2, −1, +1, +2, and +3. Oxygen atoms of C-2 or C-3 hydroxyl groups of xylose moiety at subsite −1 (+1 in Rex) are indicated by red numbers

Full size image

GH30 xylanase was shown to significantly decrease the hydrolytic activity of xylan when the interaction was removed by amino acid mutation [11]. Therefore, BhXyl39 shows higher affinity for oligosaccharides having a glucuronic acid side chain and selectively hydrolyzes them more than other oligosaccharides (Table 1).

On the other hand, the hydroxyl groups at the 2- and 3-positions of subsite +1 xylose were buried in the substrate binding pocket of BhXyl52 (Fig. 5c), suggesting there are no space subsite +1 xylose accommodate side chains. Thus, the linkage of non-reducing-end xylose with substituted xylose could not be cleaved by BhXyl52 (Table 1).

Rex has already been studied in detail [22, 32], but its specificity for branched oligosaccharides remained unknown. As such, the degradation products of X4, MeGlcA3Xyl4, and Ara3Xyl4 by Rex were analyzed by HPLC and MS. It was confirmed that only one xylose was released from the reducing end as MALDI-TOF MS analysis detected m/z 627 from MeGlcA3Xyl4 and m/z 569 from Ara3Xyl4. When the enzyme acted on Ara3Xyl4, m/z 437, which corresponds to the loss of two xylose residues, was detected, but no significant peak was observed in the HPLC analysis, suggesting that this reaction was absent or occurred at a significantly low level. Valenzuela et al. [42] reported that Rex-like enzyme released only one xylose from the reducing end of MeGlcA3Xyl4. Our results are the same as in this report. According to the sugar binding structure of Rex mutant with X2, the hydroxyl group at the 2-position of subsite −1 xylose was hydrogen-bonded to the hydroxyl group of Arg266, the hydroxyl group at the 3-position hydrogen-bonded to Ala70 and Thr69 via water, and there was no space to accommodate the side chains (Fig. 5d). The hydroxyl groups at the 2- and 3-positions of subsite −2 xylose are both hydrogen-bonded to Tyr244 via water, but both face the solvent side and there is space to accommodate side chains. It was suggested that Rex produces oligosaccharides retaining one xylose from the branch to the reducing end by accommodating arabinose and glucuronic acid side chains at subsite −2.

In conclusion, we investigated the substrate specificity of β-xylosidases belonging to different GH families and found that the characteristics were significantly different depending on the particular family. Xylan is acetylated in nature and the acetyl modification at the hydroxyl groups at the 2- and 3-positions of the xylan backbone cause steric hindrance to affect xylanolytic enzyme activity [43]. Since the effect of steric hindrance by acetyl groups were the same as that by arabinose or glucuronic acid modifications, it is considered that GH39 β-xylosidases are not affected by the acetylation of the second xylose residue from the non-reducing end, in contrast GH52 enzymes could not hydrolyze such substrate. These findings are useful for understanding how microorganisms use these enzymes to decompose biomass in nature, and are also useful for utilizing xylan-degrading enzymes. Although hemicellulose is shunned for its heterogeneity, our results strongly suggest that microorganisms use enzymes to strictly discriminate between hemicellulose structures in nature. Since GH52 has extremely strict substrate specificity for branches in xylan, it is considered to be a useful tool for preparing branched oligosaccharides from natural materials and for analyzing the structures of oligosaccharides. We would like to emphasize once again that it is very important to understand the properties of hemicellulases for branched structures in hemicelluloses to expand the range of applications of hemicellulose, which is a breakthrough in the utilization of biomass.