**1. Introduction**

Xylan is one of the most abundant polymers in plant biomass. The polymer consists of a β-1,3/1,4-linked xylopyranose backbone with side branches such as *O*-acetyl, α-4-*O*-glucuronic acid, α-l-arabinofuranose, *p*-coumaric acid, or ferulic acid at C-2 or C-3 positions [1]. Due to such complexity, a complete hydrolysis of xylan requires synergism of various xylanolytic enzymes including endo-β-1,4-d-xylanase, β-d-xylosidase, α-d-glucuronidase, α-l-arabinofuranosidase, and acetylesterase [2]. Among these enzymes, endo-β-1,4-xylanase (E.C. 3.2.1.8) plays a crucial role in hydrolyzing the β-1,4-glycosidic bonds of the xylan to form xylo-oligosaccharides (XOs) and xylose. According to the Carbohydrate-Active Enzyme (CAZy) database (http://www.cazy.org) [3], endo-β-1,4-xylanases are currently grouped in glycosyl hydrolase (GH) 5, 8, 10, 11, 30, 43, 51, 98, and

141 families. The majority of endo-β-1,4-xylanases belong to GH10 and GH11 families, which are distinctive of their respective origin, molecular properties, and protein structure [4].

Xylanases are produced by a diverse range of organisms, which include fungi, bacteria, yeast, algae, protozoa, crustaceans, and insects. Fungal and bacterial xylanases are important due to their superior properties, which could potentially be applied in industrial processes [5]. As summarized in a review article [6], xylanases are utilized for the delignification of paper pulp, modification of cereal food, improvement of digestibility of animal feedstock and production of xylo-oligosaccharides for pharmaceutical industries. Selection of the types of xylanases for these applications is based on suitability. For instance, thermostable alkaline xylanases are applicable for efficient biobleaching of pulp and paper [7], while thermostable acidic xylanases are applicable for animal feed processes [8]. Xylanases which are active and stable in low or high pH values are suitable for hemicellulosic biomass saccharification [5]. Xylanases with an optimum activity at low temperature and alkaline pH are applicable in detergent formulation additives [9]. Earlier reports suggested that xylanases obtained from psychrophilic species could improve the quality of bread and fruit juices [10,11]. Xylanases from halophilic microorganisms often exhibit salt-tolerance, which can be used for wastewater treatment and marine/saline food preparation [12].

The protein architecture of endo-β-xylanases comprises a glycoside hydrolase catalytic domain that is sometimes associated with one or more carbohydrate-binding modules (CBMs) [13]. Endo-β-xylanases without a CBM have also been reported [14–17]. CBMs do not contribute directly to catalytic mechanisms. However, they play a role in carbohydrate recognition and binding. The presence of CBM binding allows carbohydrate-active enzymes to concentrate on the polysaccharide surface and improve the overall catalytic efficiency [18]. There are currently 84 families of CBMs. These CBMs display considerable variations in substrate specificity against crystalline cellulose, non-crystalline cellulose, chitin, β-1,3/1,4-glucans, starch, glycogen, xylan, mannan, galactan, and inulin [19]. In the CAZy database, various families of CBMs were appended with GH10 xylanases, predominantly from GH 1, 2, 3, 4, 6, 9, 13, and 22, as well as from GH 10, 15, 35, and 37.

A rare halo-thermophilic bacterium, initially designated as *Rhodothermaceae* bacterium RA (NCBI taxonomy ID: 1779382), was isolated from a hot spring in Langkawi (6◦25 22.31" N, 99◦48 48.97" E), Malaysia [20]. The bacterium exhibited a low identity of 16S rRNA (89.3%) and ANI value (79.3) to the closest strain *Rhodothermus marinus* DSM 4252T. This information indicates that *Rhodothermaceae* bacterium RA might represent a new genus in the family *Rhodothermaceae* [20,21]. In 2019, Park et al. reported a strain MEBiC09517T isolated from a port in South Korea [22]. MEBiC09517<sup>T</sup> was proposed as the first member of the new genus and the authors suggested that the strain be classified as *Roseithermus sacchariphilus* gen. nov., sp. nov. Due to high similarity (ANI value of 96.2%) between *Rhodothermaceae* bacterium RA and strain MEBiC09517T, we propose that our strain is a subspecies of *Roseithermus sacchariphilus*. To differentiate both strains, we renamed our bacterium *Roseithermus sacchariphilus* RA. In this study, a xylanase gene (*xynRA2*) was cloned from this bacterium. The study aims to describe the biochemical properties of this enzyme and to understand the effects of CBM truncation on the xylanase.

#### **2. Results and Discussion**

#### *2.1. Bioinformatic Analysis*

Numerous xylanases have been discovered from extreme habitats such as soda lakes, marine sediments, and hot springs [23–25]. We previously isolated *R. sacchariphilus* RA from a saline hot spring (45 ◦C, pH 7.1, 13,000 mg/L for Cl<sup>−</sup> ion, and 7900 mg/L for Na+) [26]. A complete genome sequencing elucidated that strain RA harbors 57 GHs affiliated to 30 families [21]. Two non-homologous xylanases were identified and designated as XynRA1 and XynRA2 respectively. The xylanase XynRA1 (Genbank: ARA95075.1) has 379 amino acids and lacks a CBM, while the xylanase XynRA2 (Genbank: ARA92359.1) consists of 813 amino acids and a CBM. XynRA2 was chosen for further study, as we were interested in the function of the CBM attached to this xylanase.

XynRA2 and the putative xylanase annotated in the genome of *R*.*sacchariphilus* strain MEBiC09517<sup>T</sup> are homologs with the identity of 98.6%. The protein sequence of XynRA2 has the identity of 50–65% with xylanases from *Rubrivirga marina*, *Verrucomicrobiae* bacterium DG1235, *Lewinella nigricans*, *Catalinimonas alkaloidigena*, *Fibrisoma* sp. HYT19, *Rhodohalobacter* sp. SW132, *Cellvibrio* sp. PSBB006, and *Ignavibacteria* bacterium GWC2\_36\_12. The xylanase XynRA2 shares less than 50% identity with xylanases from *Rhodothermaeota* bacterium MED-G12, *Hymenobacter chitinivorans*, and *Siccationidurans arizonensis*. The xylanases mentioned earlier including that of *R*. *sacchariphilus* MEBiC09517T were deduced from genome annotation and have not been heterologously expressed and characterized. A phylogenetic tree utilizing a neighbor-joining algorithm was built to show a relationship of XynRA2 with selected counterparts (Figure 1a). In comparison to the well-characterized xylanases, XynRA2 is 73.1% in identity to that of xylanase Xyn10A produced by *R*. *marinus* DSM 4252 and 52.3% to a xylanase Xyl2091 from *Melioribacter roseus* P3M-2 [27,28]. The sequence identity is low between XynRA2 to other well-studied enzymes, including xylanases that are from *Fusarium graminearum* (38%) [29], *Trichoderma reesei* QM6α (28%) [30], *Bacillus stearothermophilus* T-6 (26%) [31], and *Thermoanaerobacterium saccharolyticum* B6A-RI (25%) [32].

**Figure 1.** (**a**) Protein dendrogram of XynRA2 and its close homologs. The signal peptide sequences of the proteins were not included in the phylogenetic analysis. Asterisks (\*) denote xylanases which have been characterized, otherwise represent genome annotated xylanases. (**b**) Schematic domains arrangement of the respective proteins identified by InterPro. Dotted-line box represents the truncated region in XynRA2ΔCBM.

A mature XynRA2 protein sequence comprises a CBM (Gln36–Asn198), a catalytic domain (Glu396–Tyr712), and a CTD (Trp735–Val810) (Figure 1b). The homology model of CBM clearly denoted the β-sandwich structure formed by eleven anti-parallel β-strands, while that of the catalytic domain is a typical TIM-barrel consisting of eight alternating β-strands and β-helices (Figure 2a,b). Similar to xylanases from *R*. *marinus* (Xyn10A) and *M*. *roseus* (Xyl2091), these enzymes possess a GH10 catalytic domain. From the multiple sequence alignment with six GH10 xylanases with crystal structures, the putative catalytic residues for XynRA2 were identified as Glu520 and Glu635. The linker region connecting the GH10 domain and CBM comprises 198 residues.

It is likely that in the wild-type *R. sacchariphilus* RA, XynRA2 is exported across the cytoplasmic membrane by the Sec pathway due to the presence of a signal peptide (Met1 to Ala33). In addition, XynRA2 has a CTD that enables the protein to be secreted across the outer membrane by T9SS. The T9SS protein secretion pathway is also known as Por secretion system (PorSS) [33,34] which was discovered in *Porphyromonas gingivalis* for secreting a potent protease gingipains [34]. Besides being identified in *R. sacchariphilus* RA, we noticed that in another genome sequencing project, some other annotated cellulases and hemicellulases incorporated a CTD; however, there has been little research describing the actual function of T9SS to GH enzymes. The two closest homologs of XynRA2, the Xyn10A from *R. marinus* and Xyl2091 from *M. roseus* also possessed a similar CTD [28,35]. The CTD possesses five short motifs, in which Motif B, Motif D and Motif E are important for the extensive modification by T9SS [36,37]. By aligning the CTD region of XynRA2 with other xylanases, the well-conserved Gly residues were identified in Motif B and Motif D, whereas Arg substituted the almost-conserved Lys in Motif E [37] (Figure 2c). Proteins that possessed the CTD were found to be cell-anchoring or rely on CTD for secretion, such as Xyn10A from *R*. *marinus* DSM 4252 as well as SprB, RemA, and ChiA from *Flavobacterium johnsoniae* [35,36]. Collectively, this suggests that XynRA2 could be a cell-anchoring enzyme. However, further experimental validation is required.

**Figure 2.** Putative structure of (**a**) CBM4\_9 and (**b**) GH10 catalytic domain of XynRA2. The models were colored with the rainbow scheme (blue N-terminus, follows by green, yellow, and red C-terminus); (**c**) multiple sequence alignment of XynRA2 CTD with the counterpart of Xyn10A from *R*. *marinus*, Xyl2019 from *M*. *roseus*, ChiA from *F*. *johnsoniae*, as well as CTD proteins from *P*. *gingivalis* and *Parabacteroides distasonis*. Amino acid stretch for Motif B, D, and E are indicated by red, blue, and yellow boxes, respectively. Asterisks (\*) indicate fully conserved amino acids while colon (:) indicates amino acid groups of similar properties.

Based on an InterPro analysis, the CBM of XynRA2 was annotated as CBM4\_9. The closest biochemically characterized xylanase (Xyn10A) from *R*. *marinus* DSM 4252 has two dissimilar CBM4\_9s arranged in tandem (Figure 1b), which were denoted as "CBM4-1" and "CBM4-2" in the original article [38]. Another close homolog, a characterized xylanase (Xyl2091) from *M*. *roseus*, also possessed a CBM4\_9 [28]. Interestingly, the amino acid stretch of the CBM4\_9 from *R. sacchariphilus* RA is only 70% and 51% identical to *R*. *marinus* and *M*. *roseus* counterparts respectively, suggesting that the affinity of the three enzymes against hemicellulose might be different. Different families of CBMs such as CBM6\_36 for XynG1-1 [39], CBM13 for XynAS27 [40], and dual CBM9-CBM22 for XynSL3 [24] were often reported in GH10 xylanases. According to the CAZy, other CBMs associated with xylanases

are from families 1, 2, 3, 10, 15, 35, and 37. The CBM4 family from xylanases usually binds to xylan β-glucan, and/or amorphous cellulose [41,42]. We anticipated the substrate specificity of CBM4\_9 in XynRA2 to be similar. Several reports have shown that the removal of the CBMs affected the biochemical properties of their partnering xylanases [39,40,43]. Therefore, we constructed a mutant enzyme (designated as XynRA2ΔCBM) by deleting the CBM4\_9 but retaining the linker connecting the CBM to the catalytic domain to evaluate the effect of its truncation on the xylanase.

#### *2.2. Expression of Recombinant XynRA2 and XynRA2*Δ*CBM*

The gene fragments encoding for mature XynRA2 (2349 bp) and XynRA2ΔCBM (1857 bp) were cloned in pET28a(+), expressed in *E. coli* BL21 (DE3) and purified using Ni-NTA columns. The purified enzymes migrated as two distinct bands around 90 kDa and 70 kDa on SDS-PAGE, which were consistent with the theoretical molecular weight of XynRA2 (89.5 kDa) and XynRA2ΔCBM (68.5 kDa) respectively (Figure 3a).

**Figure 3.** Molecular properties of purified XynRA2 and XynRA2ΔCBM. (**a**) SDS-PAGE analysis of purified XynRA2 and XynRA2ΔCBM. **M**, BenchMark™ Protein Ladder; **A**, purified XynRA2; **B**, purified XynRA2ΔCBM; (**b**) effect of pH in the range of 2–11; (**c**) effect of temperature from 20–90 ◦C; (**d**) thermostability at 70 ◦C across 120 min; (**e**) effect of NaCl from 0–5.0 M concentration.

#### *2.3. Biochemical Characterization of XynRA2 and XynRA2*Δ*CBM*

#### 2.3.1. Effect of pH and Temperature

Using beechwood xylan as the substrate, the purified XynRA2 had maximum activity at pH 8.5 and retained a relatively high activity between pH 7–9. Truncation of the CBM broadened the pH profile (pH 5–9) with the optimum pH shifted to 6.0 (Figure 3b). Similarly, CBM4\_9 truncation changed the optimum pH from 7.5 to 7.0 for a xylanase PX3 from *Paenibacillus terrae* HPL-003 [44]. The working pH for the mutant PX3 also narrowed to pH 5–10, while the native PX3 had an active pH ranging from 3–12. The optimum pHs of xylanase Xyn10A from *R*. *marinus* and Xyl2091 from *M*. *roseus* were 7.5 and 6.5 respectively [28,45], while that of truncated counterparts was not reported.

The optimum temperature for the activity of native XynRA2 and XynRA2ΔCBM was 70 ◦C. Overall, the temperature profiles for both enzymes were identical (Figure 3c). To evaluate the thermostability, XynRA2 and XynRA2ΔCBM were incubated at 70 ◦C without substrate for a specific interval prior to measuring residual activity. The half-life of both XynRA2 and XynRA2ΔCBM at 70 ◦C was approximately 45 min; however, XynRA2ΔCBM was more sensitive to prolonged temperature treatment (Figure 3d). The optimum temperatures of Xyn10A and Xyl2091 were 80 ◦C and 65 ◦C respectively and their half-lives were about 90 min (80 ◦C) and 160 min (60 ◦C), respectively. Truncation

of the CBM in Xyn10A from *R*. *marinus* also resulted in a decrease in thermostability, indicating that the CBM with this xylanase also contributed to enzyme stability [46]. Truncation of the CBM from xylanases from *Streptomyce rochei* L10904 (Srxyn10) [43], *Paenibacillus campinasensis* G1-1 (XynG1-1) [39], and *Streptomyces* sp. S27 (XynAS27) [40] showed that removal of the CBM did not affect the optimum temperature of xylanases. However, the truncated versions of XynG1-1 and XynAS27 displayed a significant decrease in thermostability [39,40]. In contrast, the CBM-truncated variant of Srxyn10 from *S*. *rochei* L10904 exhibited a substantial increase in thermostability at 60–70 ◦C despite sharing similar optimum temperature with its native counterpart [43].

#### 2.3.2. NaCl Tolerance

The *R. sacchariphilus* RA was capable of growing in media containing a high concentration of NaCl [20]. Since XynRA2 is probably expressed as an extracellular cell-bound enzyme, we decided to investigate the effect of NaCl on xylanase activity. Multiple xylanases are known to exhibited moderate halo-tolerance, but only limited reports have demonstrated extreme halo-tolerance ability as displayed by XynRA2 (Table 1). The relative activity of XynRA2 and XynRA2ΔCBM was slightly enhanced when the catalytic reactions were supplemented with 1.0 M NaCl. Notably, XynRA2 retained 94% of initial activity at 4.0 M, and 71% at 5.0 M NaCl. Although the mutant XynRA2ΔCBM was more salt-sensitive, the enzyme retained the relative activity of 79% at 4.0 M and 54% at 5.0 M (Figure 3e). The reason for the lower halo-tolerance is unknown. In addition, there is a lack of literature elucidating the relationship between CBM and halo-xylanase activity.

A homology model of XynRA2 catalytic domain demonstrated a high distribution of acidic amino acids on the protein surface resulting in an overall negative electrostatic potential (Figure 4), which might explain the excellent protein stability in high NaCl concentration. Theoretically, halo-tolerant enzymes contain more acidic residues (Asp and Glu) than non-polar residues (Val, Ile, Leu, Met, and Phe). Halo-tolerant enzymes are also enriched with small residues (Ala, Val, Ser, and Thr) but lack Lys residue [47]. It has been proposed that excess acidic residues could facilitate the weakening of hydrophobicity or strengthening of hydrophilic forces on the enzyme surface, which increases water-binding capacity and prevent proteins aggregation at high salt concentration [48,49].


**Table 1.** The reported halo-tolerant xylanases and their activity in high concentration of NaCl. Enzymatic reactions carried out at 0 M of NaCl was treated as 100%.

ND: not determined.

**Figure 4.** Predicted electrostatic potentials on the surface of XynRA2 catalytic domain from (**a**) top and (**b**) bottom views. The ligand (xylobiose) is bound at the catalytic binding pocket of XynRA2 as indicated in (**a**). Red and blue indicate negative and positive electrostatic potentials respectively.

#### 2.3.3. Enzyme Kinetics

The specific activities and the turnover rate (*k*cat) of the purified XynRA2 and XynRA2ΔCBM were determined by reacting the enzymes with soluble beechwood xylan. The specific activities of XynRA2 and XynRA2ΔCBM were 300 U/mg and 160 U/mg respectively. The *k*cat of the native and mutant enzymes were 24.8 s−<sup>1</sup> and 15.7 s−1, respectively. We found that the truncation of CBM significantly affected the performance of the enzymes. This finding was in consistent with XynG1-1 from *P*. *campinasensis* that CBM truncation reduced the *k*cat by 20% [39]. Removal of CBM alone did not affect the *k*cat of XynAS27 from *Streptomyces* sp. S27. However, truncating CBM together with the linker reduced *k*cat value by 25% [40]. On the other hand, the xylanase variant of Srxyn10 with a CBM truncation had a three-fold higher specific activity on beechwood xylan than its native counterpart [43].

#### 2.3.4. Substrate and Product Specificities

The purified XynRA2 and XynRA2ΔCBM were incubated with various substrates before analyzing them using HPLC. Generally, XynRA2 and XynRA2ΔCBM showed similar substrate specificities. Both enzymes were active on beechwood xylan, oat-spelt xylan, and xylo-oligosaccharides (XOs) such as X6, X5, X4 but not on X3 and X2. Except for xylose-based carbohydrates, the enzymes were unable to hydrolyze glucose-, maltose-, and arabinose-derived polymers such as carboxymethylcellulose (CMC), Avicel™, starch, pullulan, d-cellobiose, and arabinan. The results indicated that the enzymes did not possess either a cellulase or arabinase activity, suggesting that XynRA2 is a specific GH10 xylanase. This is in agreement with a recent statistical study that showed most of the characterized GH10 xylanases were mono-specific (96.8%, *n* = 350) towards xylanosic substrates [4].

We compared the product formation pattern of XynRA2 and XynRA2ΔCBM by reacting the purified enzymes with beechwood xylan and oat-spelt xylan (Figure 5). Upon reacting XynRA2 with beechwood xylan, the products constituted a mixture of XOs ranging X6, X5, X4, X3, and X2 at the beginning of the reaction (15 min). After a prolonged hydrolysis (24 h), xylobiose (X2) was accumulated as the primary product together with detectable X3 and X1 (Figure 5a), and the product formation pattern for XynRA2ΔCBM against beechwood xylan was shown in Figure 5b. For reactions of 15 min and 24 h, the product profile for XynRA2ΔCBM was similar to that of XynRA2. Yet, the ratio of X3 and X2 was slightly different in the 1 h and 3 h reactions. Previous reports on xylanase rXTMA from *Thermotoga maritima* and xylanase A from *Caldibacillus cellulovorans* also showed a variation in the profiles of XOs produced by native and CBM-depletion xylanases [55,56].

Although the same reaction conditions were used with oat-spelt xylan, we obtained lower sugar yields, probably due to the physical structure of oat-spelt xylan which consisted of both insoluble and soluble fractions. Furthermore, the product profiles were also different for beechwood xylan and oat-spelt xylan. After prolonged reaction, X4, X3, and X2 were accumulated as the major products (Figure 5c). Interestingly, oat-spelt xylan was a poor substrate for XynRA2ΔCBM (Figure 5d), as reported for other xylanases [39,40,57,58]. A lower activity against oat-spelt xylan might be due to the inefficient binding onto the substrate, as a result of CBM truncation in XynRA2ΔCBM. It has also been recurrently reported that the truncation of CBMs affects catalytic efficiency of GHs towards other insoluble substrates but not the soluble counterparts [18,19].

**Figure 5.** Product analysis of xylan degradation. (**a**,**b**) products formed from hydrolysis of beechwood xylan by XynRA2 and XynRA2ΔCBM, respectively; (**c**,**d**) products formed from hydrolysis of oat-spelt xylan by XynRA2 and XynRA2ΔCBM, respectively. The product peaks shown in this figure were normalized.

#### **3. Materials and Methods**
