The amino acid sequences with the first 100 similarities to QsGH13 sequences were obtained using NCBI BLASTp.² Clustal was used for amino acid multiple sequence alignment of QsGH13 and its homologs³ (Larkin et al., 2007). Secondary structure alignment results were obtained using ESpript based on multiple sequence alignment results and QsGH13 spatial coordinates⁴ (Robert and Gouet, 2014). The maximum likelihood estimation (MLE) in MEGA X was used to construct the phylogenetic tree (Kumar et al., 2018).

Qs. SW-135 (basionym: Erythrobacter seohaensis sp. SW-135) (Taxonomy ID: 266951, GenBank Accession: GCA_002795865.1) (Kwon et al., 2007; Xu et al., 2020; Supplementary Figure 1), belonging to the family Erythrobacteraceae, was isolated from deep-sea sediment samples that were collected from the East Pacific Seamount region. The QsGH13-encoding gene (qsgh13) was identified and cloned from the genome of Qs. SW-135. To be specific, the qsgh13 gene was amplified by PCR (forward: 5′-GGCGGATCCATGAGCGGCAAGCTGCCTTG-3′, BamHI, reverse: 5′-GCGGAGCTCTCATGTGTCGGTCTCCAGGATGA-3′, SacI) and inserted into the expression vector pSMT3 (Li et al., 2012) using the BamHI and SacI restriction sites.

The recombinant E. coli BL21(DE3) plus/pSMT3-qsgh13 was cultured in LB medium containing 50 μg/ml of kanamycin and 34 μg/ml of chloramphenicol at 37°C. When the OD₆₀₀ reached 0.8, the recombinant protein expression was induced by 0.5 mM isopropyl-β-D-thiogalactopyranoside (IPTG) for 20 h at 16°C (Shen et al., 2019). Cells were harvested by centrifugation at 5,000 rpm for 15 min at 4°C. The cell precipitation was resuspended in a lysis buffer [50 mM Tris, pH 8.0; 500 mM NaCl; 1% (v/v) glycerol; 10 mM imidazole; 1 mM β-Me; 0.2 mM PMSF] and was disrupted on ice with an ultrasonic crusher (SCIENIZ, Ningbo, China). The separated supernatant was purified by Ni-NTA affinity chromatography. After the recombinant protein with 6xHis-Sumo tag was digested with Ulp1 enzyme, the target protein was obtained in the flow-through fractions and was concentrated with a 50-kDa concentrator. Then QsGH13 protein was purified in a buffer (20 mM Tris, pH 7.4; 100 mM NaCl; 2 mM DTT) by gel-filtration chromatography (Superdex 200 16/600, GE, United States). The protein was determined by SDS-PAGE, and the concentration was measured by the method of Bradford with bovine serum albumin (BSA) as a standard (Huang et al., 2016).

The enzymatic activity and substrate specificity of wild-type QsGH13 were determined by p-nitrophenol method (Irma et al., 1992). The 100-μl standard reaction buffer consisted of 1 mM pNPαGlu, 20 mM Gly-NaOH buffer (pH 10.0), and an appropriate amount of purified enzyme. The enzymatic activity was determined by continuously measuring the amount of released p-nitrophenol at the absorbance of 405 nm using Multiskan FC (ThermoFisher, MA, United States), at 45°C, and the inactivated enzyme solution was used as the blank control. All results were carried out in three independent experiments. The absorbance of 405 nm was measured every 2 min, and the total reaction time was 30 min.

The substrate specificity of the enzyme was determined using p-nitrophenyI-β-D-glucopyranoside (pNPβGlu), p-nitrophenyl-α-D-glucopyranoside (pNPαGlu), pNP-α-L-arabinopyranoside (pNPαArap), p-nitrophenyl -β-D-lactopyranoside (pNPβLac), p-nitrophenyI-α-D-galacto pyranoside (pNPαGal), p-nitrophenyI-β-D-galactopyranoside (pNPβGal), p-nitrophenyI-β-D-mannopyranoside (pNPβ Man), p-nitrophenyI-β-D-xylopyranoside (pNPβXyl), and p-nitrophenyI-β-D-cellobioside (pNPβCel) as substrates, respectively. All the substrates were purchased from Shanghai Yuanye Bio-Technology Co., Ltd.

The kinetic parameters were examined using pNPαGlu as a substrate at different concentrations varying from 0.025 to 2.0 mM. The Michaelis–Menten constant (K_m) and maximum velocity (V_max) were analyzed by the Michaelis–Menten equation using GraphPad Software (GraphPad prism5, United States) (Huang et al., 2019).

The optimum temperature of the enzyme was determined with different temperatures varying from 4 to 70°C. Buffers with different pH were used to determine the optimal pH of the enzyme, including 20 mM citrate buffer (pH 3.0–pH 6.5), 20 mM phosphate buffer (pH 6.5–pH 7.5), 20 mM Tris–HCl buffer (pH 7.5–pH 9.0), and 20 mM Gly-NaOH buffer (pH 9.0–pH 13.0).

The influence of metal ions on the enzyme activity was determined by adding 10 mM of cations (Ba²⁺, Ca²⁺, Co²⁺, Cu²⁺, K⁺, Fe²⁺, Fe³⁺, Mg²⁺, Mn²⁺, Na⁺, Ni²⁺, Sr²⁺, Zn²⁺) and ethylene diamine tetraacetic acid (EDTA) to the standard reaction buffer. The effect of organic solvents on enzyme activity was investigated in the presence of 10% (v/v) acetonitrile, acetone, dimethyl sulfoxide, ethanol, formic acid, isopropanol, methanol, respectively. The influences of the detergents on enzyme activity were examined by using 10% (v/v) Triton X-100, Triton X-114, Tween 20, Tween 80, and SDS. The enzyme activity assay was carried out under optimal standard reaction buffer with 20 mM Gly-NaOH buffer (pH 10.0) at 45°C, and the enzyme activity in the blank group was defined as 100% without additives.

The salt tolerance of enzyme was measured in the presence of 1.0, 2.0, 3.0, 4.0, or 5.0 M NaCl, and the enzyme activity in the blank group without NaCl was defined as 100%. The tolerance of enzyme on products were tested by adding 0.5, 1.0, 1.5, 2.0, 2.5, and 3.0 M glucose to 20 mM Gly-NaOH buffer (pH 10.0). In the absence of the substrate, the thermostability of the enzyme was determined by measuring the residual activity at the optimum conditions after incubating the enzyme for different periods of time at 4°C, respectively. Similarly, in the absence of the substrate, the alkali resistance of enzyme was measured by testing the residual activity at the optimum conditions after incubating the enzyme for 2, 24 h in buffers of different pH at 4°C.

The enzymatic activities of QsGH13 toward maltose, sucrose, isomaltose, panose, and isommaltose were determined by HPLC. First, glucose, maltose, isomaltose, panose, and isommaltose trisaccharide were dissolved in mobile phase to establish the standard curve. Second, the reaction buffer contained 15% substrate (w/v), 1 mg/ml of QsGH13, and 20 mM Gly-NaOH buffer (pH 10.0), which were used to catalyze the reaction at 45°C. Meanwhile, QsGH13 in the reaction buffer was replaced by the deactivation of QsGH13, which served as a negative control. The reaction of the samples were terminated by high-temperature denaturation, and the supernatant was collected by centrifugation at 12,000 rpm for 10 min. The samples were detected by Thermo Scientific™ Hypersil™ APS-2 HPLC with 72% acetonitrile and 28% water as mobile phase. Glucose concentration was calculated according to the standard curve.

The purified QsGH13 was concentrated to 10 mg ml^–1 using Amicon Ultra 50K ultrafiltration devices (Merck Millipore, Darmstadt, Germany). The crystallization of QsGH13 was prepared by handing drop vapor diffusion methods by mixing 1 μl of QsGH13 (10 mg ml^–1) with an equal volume of reservoir solution at 4°C. It was grown in a condition of 0.1 M Tris–HCl (pH 8.5), 0.2 M NaCl, and 20% PEG 3,350. The crystals were briefly soaked with the cryoprotectant solution, and then flash cooled directly in liquid nitrogen at −173°C. The x-ray diffraction data were collected at the BL17U beamline of SSRF (Shanghai Synchrotron Radiation Facility, Shanghai, China). The data were processed and scaled using XDS (Kabsch, 2010). The collected data and processing details are shown in Table 1.

The structure of QsGH13 was determined by molecular replacement method with Phaser (McCoy et al., 2007), using the crystal structure of α-glucosyltransferase XgtA from Xanthomonas campestris WU-9701 (PDB entry: 6AAV, 51% identity to QsGH13) (Watanabe et al., 2020) as a search model. The refinement data of QsGH13 was performed using REFMAC and Phenix (Liebschner et al., 2019). Coot was used to further modify and adjust the structure (Emsley and Cowtan, 2004). The refinement statistics data of QsGH13 are shown in Table 1.

The disordered area of QsGH13 sequence was predicted using IUPred2A.⁵ The catalytic pocket of QsGH13 was predicted using POCASA.⁶ The visualization of the tunnel in the QsGH13 structure used CAVER 3.0 PyMol plugin. The docking model between QsGH13 and its substrate was predicted with AutoDock 4 (Morris et al., 2009). All of the 3D structures were analyzed and drawn using PyMOL program (Version 2.0 Schrödinger, LLC).

Nucleotide sequence analysis showed that the open reading frame of 1,587 bp qsgh13 gene sequence from deep-sea sediment metagenomic screening encodes a protein of 528 amino acids with a theoretical molecular weight of 59.3 kD. Phylogenetic tree sequence analysis based on NCBI, PDB, and CAZy databases showed that the protein QsGH13 belongs to the GH13 family (Supplementary Figure 1), and predicted that the protein QsGH13 had α-glucosidase characteristics. To explore the physicochemical characterizations of protein QsGH13, we constructed the gene qsgh13 into the recombinant plasmid with an N-terminal His₆-SUMO label in E. coli DH5α and expressed the protein QsGH13 in E. coli BL21 (DE3) plus. Cells were cultured and enriched, and protein His₆-SUMO-QsGH13 was extracted. After the His₆-SUMO tag was removed by Ulp1 enzyme, the protein QsGH13 was purified by gel-filtration chromatography to 95% homogeneity. Pure QsGH13 (12.6 ± 2.9 mg) was obtained from 1-L cultures. SDS-PAGE was used to show the molecular weight of QsGH13 (Figures 1A,B). QsGH13 was found to specifically hydrolyze p-nitrophenyl-α-D-glucopyranoside (pNPαGlu), but no other substrates (Figure 1C). The enzymatic activities of QsGH13 toward maltose was determined by HPLC. Maltose (substrate) and glucose (product) came out at the peak positions of 12.68 and 8.69 min, respectively (Figures 1D,E and Supplementary Figure 2). Therefore, QsGH13 was identified as α-glucosidase.

We have studied the optimum reaction conditions of QsGH13 toward pNPαGlu. QsGH13 was able to maintain 80% catalytic activity toward pNPαGlu at a temperature range of 40–55°C and a pH range of 8.0–11.0; QsGH13 showed the highest catalytic activity at 45°C and pH 10.0 (Figures 2A,B). The catalytic activity of QsGH13 was almost unchanged when stored at 4°C for 24 h (Supplementary Figure 3A). The catalytic activity of QsGH13 was decreased with the increase in incubation time in different pH buffers (Supplementary Figure 3B). The enzymatic activity of QsGH13 was 25.41 U/mg, and the K_m value was 0.2952 ± 0.0322 mM (Figure 2C).

Furthermore, we investigated the tolerance ability of QsGH13 to the addition of different cations, detergents, organic solvents, and different concentrations of NaCl and glucose added into the reaction buffer under optimum temperature and pH conditions (Figures 2D–H). The α-glucoside activity of QsGH13 was completely eliminated with the addition of Cu²⁺, and only about 10–30% was retained with Zn²⁺, Ni²⁺, Co²⁺, and 50–80% activity remained with K⁺, Fe²⁺, Fe³⁺, and the activity of QsGH13 reached more than 100% with Mg²⁺, Na⁺. Moreover, the α-glucosidase activity of QsGH13 was comparative with the blank under the addition of 10% (v/v) organic solvents (methanol, formic acid, ethanol, isopropanol, acetonitrile, glycerol, and DMSO) or 10% (v/v) detergent (SDS Triton X-T114, Triton X-100, Tween20, and Tween80); it was attenuated. Especially, it is completely inactivated after the addition of formic acid (Figures 2E,F). In addition, the α-glucosidase activity of QsGH13 was decreased with the increase in NaCl concentration. The α-glucosidase activity of QsGH13 was less than 40% of the blank group under the addition of 4 M NaCl (Figure 2G). QsGH13 still retained 30% activity compared with the blank group under the condition of 2 M glucose, but with an increase in glucose concentration, its activity was gradually abolished (Figure 2H).

The crystal structures of QsGH13 molecules were fully built to 2.2-Å resolution with the satisfied R_work and R_free values of 21.83 and 27.38%, respectively. According to the diffraction dataset, the QsGH13 crystal belongs to the monoclinic space group P2₁2₁2₁, with unit-cell parameters a = 58.815 Å, b = 129.920 Å, c = 161.304 Å, α = γ = β = 90°, which contains two monomers per asymmetric unit (Table 1). Phylogenetic tree analysis showed that QsGH13 belonged to α-glucosidase (EC 3.2.1.20), GH13 family, subfamily 23 (Supplementary Figure 1). QsGH13 has typical and conserved characteristics of the GH13 family (Figure 3A), where the catalytic domain A (residues 1–107, 176–463) is composed of a typical TIM barrel, sandwiched between the loop-rich domain B and the conserved domain C, and it is mainly composed of (β/α) eight-barrel structures, including eight α-helical and eight β-fold. The loop-rich domain B (residues 108–175) is derived from the domain A, including an α-helix and three β-fold, which are involved in the formation of QsGH13 dimers and forms catalytic pockets with domain A. Domain C (residues 464–525) is composed of multiple highly conserved β-folds that stabilizes the entire protein structure (Shen et al., 2015; Figures 3B,C). In order to achieve catalytic activity, dimers formed by two monomers are essential in the GH13 family of enzymes (Watanabe et al., 2020; Wangpaiboon et al., 2021). The QsGH13 monomers formed dimers mainly through five hydrogen bonds and 105 non-bonded contacts (Supplementary Figure 4). After the superposition of two monomers, the root-mean-square deviation (r.m.s.d.) value was 0.250 Å using the DALI server⁷ (Holm, 2020). It shows that the two chains are very similar in structure and has no obvious biological significance, so we only study one monomer.

Multiple sequence alignments showed that QsGH13 had high homology with other members of the GH13 family, such as XgtA (PDB entry: 6AAV), an α-glucosyl transfer enzyme derived from Xanthomonas campestris WU-9701 (Watanabe et al., 2020); HaG (PDB entry: 3WY1), an α-glucosidase produced from Halomonas sp. H11 (Shen et al., 2014); GSJ (PDB entry: 2ZE0), an α-glucosidase from Geobacillus sp. strain HTA-42 (Shirai et al., 2008); and MalL (PDB entry: 4m8u), an oligo-1,6-glucosidase from bacillus subtilus (Hobbs et al., 2013; Supplementary Table 1). The similarity of amino acid sequence from high to low is 51.3, 47.8, 39.6, and 37.5% (Figure 4).

The overall structures of these GH13 families were similar through superimposing the overall structures of QsGH13 on XgtA, HAG, and GSJ, as indicated by r.m.s.d. values between Cαs in these three structures and the QsGH13 (0.680, 0.722, and 1.426 Å, respectively). However, as shown in Figure 5A, their β-α loop 4 regions were significantly different. The β → α loop 4 of QsGH13 is located above the catalytic pocket (Figure 5B), and it is associated with substrate specificity. It controls substrate entry by maintaining structural integrity around the catalytic pocket, and its amino acid sequence is poorly conserved. The β → α loop 4 of QsGH13 has some differences in composition and structure compared with the other three proteins. Amino acid sequence analysis showed that QsGH13 has the highest similarity with XgtA (51.3%); however, the amino acid sequence at 203–242 of QsGH13 just shows a homology of 17.4% (4/23 amino acids are identical) with that of XgtA and a homology of 21.7% with that of HaG. The size of β → α loop 4 of QsGH13 is between GSJ and HaG (Figure 5C). The β → α loop 4 of QsGH13 is demonstrated to be more dynamic than other regions of QsGH13 and the β → α loop 4 of HaG and XgtA by IUPred2A analysis⁸ (Erdõs and Dosztányi, 2020). Furthermore, the more dynamic β → α loop 4 of QsGH13 may be involved in the acquisition and recognition of regulatory substrates (Supplementary Figure 5). The structural differences between QsGH13, XgtA, HaG, and GSJ at the β → α loop 4 region may, therefore, result in the differences observed for the formation of byproducts during α-glucosylation reaction.

The predicted results of the QSGH13 catalytic pocket are consistent with the structure comparison results, and the channel volume is 672 ų, and the average VD is 3.70238 (Figures 6A,B). As shown in Figure 6C, the β → α loop 4 of HaG covers most of the entrance of the catalytic pocket, hinders the entry of long-chain substrates, and almost only hydrolyzes disaccharides (Figure 6C). Compared with HaG, the entrance region of the catalytic site of QsGH13 is relatively large, and the two tunnels are connected to the active center, but the channel is relatively narrow, and the larger substrate cannot pass through (Figure 6D).

QsGH13 forms a deep and narrow catalytic pocket of 14 residues in (β/α) eight-barrel structures (TIM) away from the dimeric interface. Among these residues, there are eight residues, including Tyr65, His105, Arg200, Glu266, Asp202, Phe292, His328, and Asp329, that are conserved in the GH13 family. Six residues, Asp62, Phe147, Phe166, Thr203, Phe206, and Arg396, are conserved between QsGH13 and HaG (Figure 7A). The enzymatic hydrolysis of glycosidic bond takes place via general acid catalysis that requires two critical residues. In the catalytic pocket of HaG, Asp202, Glu271 (3wy4-Gln271), and Asp333 constitute the catalytic site, of which Asp202 is the nucleophilic residue, Glu271 is the proton acceptor/donor, and Asp333 helps to stabilize Glu271 residue (Figure 7B). Similarly, the structure of QsGH13-maltose formed by docking shows not only how maltose is bound prior to hydrolysis but also the roles of Glu266 as a proton donor for interglycosidic oxygen of maltose and Asp202 as a catalytic nucleophile (Figure 7C). The distance between the proton acceptor/donor (Glu266) and the catalytic nucleophile (Asp202) is 6 Å, and similar conditions have been observed in other retaining glycosidases (Davies and Henrissat, 1995).