Identification and Characterization of a Thermostable GH36 α-Galactosidase from Anoxybacillus vitaminiphilus WMF1 and Its Application in Synthesizing Isofloridoside by Reverse Hydrolysis

An α-galactosidase-producing strain named Anoxybacillus vitaminiphilus WMF1, which catalyzed the reverse hydrolysis of d-galactose and glycerol to produce isofloridoside, was isolated from soil. The α-galactosidase (galV) gene was cloned and expressed in Escherichia coli. The galV was classified into the GH36 family with a molecular mass of 80 kDa. The optimum pH and temperature of galV was pH 7.5 and 60 °C, respectively, and it was highly stable at alkaline pH (6.0–9.0) and temperature below 65 °C. The specificity for p-nitrophenyl α-d-galactopyranoside was 70 U/mg, much higher than that for raffinose and stachyose. Among the metals and reagents tested, galV showed tolerance in the presence of various organic solvents. The kinetic parameters of the enzyme towards p-nitrophenyl α-d-galactopyranoside were obtained as Km (0.12 mM), Vmax (1.10 × 10−3 mM s−1), and Kcat/Km (763.92 mM−1 s−1). During the reaction of reverse hydrolysis, the enzyme exhibited high specificity towards the glycosyl donor galactose and acceptors glycerol, ethanol and ethylene glycol. Finally, the isofloridoside was synthesized using galactose as the donor and glycerol as the acceptor with a 26.6% conversion rate of galactose. This study indicated that galV might provide a potential enzyme source in producing isofloridoside because of its high thermal stability and activity.


Introduction
α-Galactosidases (α-D-galactoside galactohydrolases; EC 3.2.1.22) are exoglycosidases that catalyze the hydrolysis of the terminal non-reducing α-galactosyl residue of various substrates [1]. α-Galactosidases have been classified into six glycoside hydrolase (GH) families namely GH4, GH27, GH36, GH57, GH97, and GH110 based on structure and sequence similarity, but most of them belong to the GH27 or GH36 families, which share common evolutionary origins and reaction mechanism [2]. A majority of GH36 α-galactosidases are reported from bacterial sources with high molecular mass and multimeric nature [3]. α-Galactosidases are known to be potentially useful in diverse applications. In the pharmaceutical industry, they have been shown to be effective against Fabry disease [4]. Additionally, α-galactosidases, which act as hydrolases in nature, can be used in the food industry, such as the hydrolysis of galactosyl residues from raffinose to improve the crystallization of sucrose [5]. Studies conducted over the years have shown that α-galactosidases can mediate transglycosylation to produce a series of important compounds [6].

Obtaining α-Galactosidase-Secreting Strains
Six α-galactosidase-producing strains were found by observing the blue single col-

Obtaining α-Galactosidase-Secreting Strains
Six α-galactosidase-producing strains were found by observing the blue single colony on the primary screening plate, and enzyme activity was detected in the supernatant of lysate but not in the supernatant of fermentation. Among them, a strain with the highest activity to catalyze the synthesis of isofloridoside by reverse hydrolysis was selected. The 16s rDNA sequence of the target strain, which was named as A. vitaminiphilus WMF1, possessed the highest homology (99.0%) with that of A. vitaminiphilus (NCBI GenBank accession no. NR_108379). As far as we know, no report has been published on α-galactosidase from Anoxybacillus sp. Therefore, the research on the biochemical characterizations of α-galactosidase from Anoxybacillus sp. is of great significance.

Sequence and Structure Analysis
The genomic DNA sequence of A. vitaminiphilus was found in NCBI, and a putative α-galactosidase gene was noted; however, the gene has not been cloned, expressed, and characterized yet. According to this sequence, the primers were designed and then the α-galactosidase gene from A. vitaminiphilus WMF1 was cloned. The sequencing analysis showed that galV showed 75.3% identity with the putative α-galactosidase from A. vitaminiphilus (WP_111643960.1). Additionally, galV shared the identity of 91.4% with the uncharacterized α-galactosidase from Bacillus alveayuensis (WP_044748107.1), followed by the α-galactosidase from Geobacillus sp. MR (73.4%, WP_171355420.1) and Alkalihalobacillus akibai (67.0%, WP_035664793.1). The α-galactosidase from G. stearothermophilus (AAG49421.1), which has been experimentally characterized, shared 75.3% identity with galV and was used as the template for modeling ( Figure 2). The theoretical molecular mass of galV was calculated to be 83.8 kDa. No signal peptide was found in galV, which was consistent with the aforementioned result that the α-galactosidase was an intracellular enzyme. A catalytic domain belonging to GH36 α-galactosidase was observed in the sequence (from Glu328 to Glu627), indicating that galV should be a GH36 family α-galactosidase. Moreover, galV was found to contain the consensus motif LFVL/MDDGWFG of GH36 family α-galactosidases [16]. Residues D478 and D548 are the putative nucleophile and catalytic acid/base in the motif KWD and SDXXDXXXR of galV, respectively [16].
The putative structure of galV showed the typical GH36 organization, which comprised three parts: N-terminal domain, catalytic domain with a conserved (β/α) 8 -barrel topology and similar active sites, and C-terminal domain [17]. The N-terminal domain (residues 1-327), which was connected to the catalytic domain, consisted of a β-super sandwich and terminated in a long α-helix. The catalytic domain (residues 328-627) showed a (β/α) 8 -barrel fold containing the putative nucleophile and proton donor, Asp478 and Asp548 ( Figure 3). The least conservative of the three domains, the C-terminal domain (residues 628-727), showed a β-sandwich structure, which contained an α-helix and eight βfolds. Furthermore, galV was presumed to have a symmetrical tetramer structure because the template α-galactosidase was tetrameric, which was observed in a previous study [17]. The putative structure of galV showed the typical GH36 organization, which comprised three parts: N-terminal domain, catalytic domain with a conserved (β/α)8-barrel topology and similar active sites, and C-terminal domain [17]. The N-terminal domain (residues 1-327), which was connected to the catalytic domain, consisted of a β-super sandwich and terminated in a long α-helix. The catalytic domain (residues 328-627) showed a (β/α)8-barrel fold containing the putative nucleophile and proton donor, Asp478 and Asp548 (Figure 3). The least conservative of the three domains, the C-terminal domain (residues 628-727), showed a β-sandwich structure, which contained an α-helix and eight β-folds. Furthermore, galV was presumed to have a symmetrical tetramer structure because the template α-galactosidase was tetrameric, which was observed in a previous study [17].

Expression and Purification of galV
Recombinant α-galactosidase was abundantly expressed in E. coli. The recombinant protein was approximately 80 kDa on a 12% SDS-PAGE gel, which was in agreement with

Expression and Purification of galV
Recombinant α-galactosidase was abundantly expressed in E. coli. The recombinant protein was approximately 80 kDa on a 12% SDS-PAGE gel, which was in agreement with the calculated molecular mass (83.8 kDa) of galV ( Figure 4) and in the range of the molecular weight (70-100 kDa) of most GH36 α-galactosidases [18]. The native molecular mass of the enzyme was 320 kDa as determined by gel filtration, suggesting a homotetramer structure, which was consistent with the previously speculated structure, and the same result was also found in the α-galactosidase from Paecilomyces thermophila [19].

Expression and Purification of galV
Recombinant α-galactosidase was abundantly expressed in E. coli. The recombinant protein was approximately 80 kDa on a 12% SDS-PAGE gel, which was in agreement with the calculated molecular mass (83.8 kDa) of galV ( Figure 4) and in the range of the molecular weight (70-100 kDa) of most GH36 α-galactosidases [18]. The native molecular mass of the enzyme was 320 kDa as determined by gel filtration, suggesting a homotetramer structure, which was consistent with the previously speculated structure, and the same result was also found in the α-galactosidase from Paecilomyces thermophila [19].

Biochemical Characterization of galV
The optimal pH of galV was found to be approximately 7.5, which was consistent with the previous finding that GH36 enzymes functioned optimally at neutral or alkaline pH [20]. Figure 5 shows that galV manifested the maximum activity at 60 • C, which was in agreement with the α-galactosidase from Paceilomyces thermophila [19] and higher than that reported for α-galactosidase from Bifidobacterium breve (37 • C) [21] and Aspergillus oryzae YZ1 (50 • C) [22]. Under the optimal conditions, galV showed a specificity of 70 U/mg against pNPG, which was higher than that of α-galactosidase from Carnobacterium piscicola (2.3 U/mg) [23] and Lactobacillus fermenti (2.19 U/mg) [24], but lower than that produced by Aspergillus oryzae YZ1 (76.9 U/mg) [22] and Penicillium janczewskii zalesk (667 U/mg) [25].
in agreement with the α-galactosidase from Paceilomyces thermophila [19] and higher than that reported for α-galactosidase from Bifidobacterium breve (37 °C) [21] and Aspergillus oryzae YZ1 (50 °C) [22]. Under the optimal conditions, galV showed a specificity of 70 U/mg against pNPG, which was higher than that of α-galactosidase from Carnobacterium piscicola (2.3 U/mg) [23] and Lactobacillus fermenti (2.19 U/mg) [24], but lower than that produced by Aspergillus oryzae YZ1 (76.9 U/mg) [22] and Penicillium janczewskii zalesk (667 U/mg) [25]. The enzyme was stable over a slightly alkaline pH range between 6.0 and 9.0, which was consistent with the α-galactosidase from Bacillus stearothermophilus NCIM 5146 [26]. Contrary to our results, the α-galactosidase from Penicillium sp. F63 CGMCC 1669 [16] and Penicillium janczewskii zaleski [27] had optimum activity in the acidic pH range. The neutral or weak alkaline pH form of α-galactosidase is suitable for the hydrolysis of soy milk, since an acidic pH leads to the deposition of soy protein and gives milk its sour taste [26]. The thermostability of galV was also measured. About 78% of its original activity was retained after incubation at 60 °C for 2 h, which was consistent with the α-galactosidase from thermophilic microorganisms, such as the α-galactosidase from Rhizomucor miehei [28] and Dictyoglomus thermophilum sp [29]. Moreover, galV was more stable than most GH36 α-galactosidases, such as the α-galactosidase from Bacillus megaterium [18], Yersinia pestisbiovar Microtus str. 91,001 [30], Aspergillus oryzae YZ1 [22] and Paceilomyces thermophila [19]. Thus, the enzyme showed activity and stability over a broad range of temperature, which made it a potential candidate in various industrial processes. The enzyme was stable over a slightly alkaline pH range between 6.0 and 9.0, which was consistent with the α-galactosidase from Bacillus stearothermophilus NCIM 5146 [26]. Contrary to our results, the α-galactosidase from Penicillium sp. F63 CGMCC 1669 [16] and Penicillium janczewskii zaleski [27] had optimum activity in the acidic pH range. The neutral or weak alkaline pH form of α-galactosidase is suitable for the hydrolysis of soy milk, since an acidic pH leads to the deposition of soy protein and gives milk its sour taste [26]. The thermostability of galV was also measured. About 78% of its original activity was retained after incubation at 60 • C for 2 h, which was consistent with the αgalactosidase from thermophilic microorganisms, such as the α-galactosidase from Rhizomucor miehei [28] and Dictyoglomus thermophilum sp [29]. Moreover, galV was more stable than most GH36 α-galactosidases, such as the α-galactosidase from Bacillus megaterium [18], Yersinia pestisbiovar Microtus str. 91,001 [30], Aspergillus oryzae YZ1 [22] and Paceilomyces thermophila [19]. Thus, the enzyme showed activity and stability over a broad range of temperature, which made it a potential candidate in various industrial processes. Table 1 presents the effects of metal ions and reagents on galV. The enzyme activity drastically decreased to 2.37%, 4.86%, and 3.93% of the original activity in the presence of Fe 2+ , Ni 2+ , and Fe 3+ , respectively, while Ca 2+ , Mn 2+ , and Zn 2+ considerably inhibited the activity. The drastic mitigation of galV activity was seen in the presence of Cu 2+ (0.16% residual activity), which was also reported for the α-galactosidase from Aspergillus terrus GR [31]. Na + , K + , Li + , and Mg 2+ did not affect the enzymatic action, which was similar to that observed in Humicola sp [32] and Alicyclobacillus sp. A4 [33]. Reagents such as CTAB, SDS, and acetonitrile had a strong inhibitory effect on the enzyme activity. Most proteins lose the tertiary and quaternary structures under the action of SDS due to the strong denaturation of SDS [34]. Unlike the α-galactosidase from Bacillus megaterium, the organic solvents DMSO and methanol had no significant effect on the enzyme activity [18]. The tolerance of galV to alcohol may make it easier to construct a solvent-free system.
In this case, a high concentration of acceptor such as alcohol is beneficial to the reverse hydrolysis reaction, resulting in a high yield. The kinetic parameter values of the galV were obtained using the Lineweaver-Burk plot with certain concentrations of pNPG. The K m , V max , and K cat /K m for pNPG were 0.12 mM, 1.10 × 10 −3 mM s −1 , and 763.92 mM −1 s −1 , respectively. The kinetic parameters of α-galactosidases have been studied extensively. Accordingly, the catalytic efficiency (K cat /K m ) of galV toward pNPG was forty-fold that of the α-galactosidase from Rhizomucor miehei [28]. Furthermore, the α-galactosidase from Bacillus megaterium possessed K m and K cat /K m values of 0.42 mM and 610 mM −1 s −1 , respectively [18], and the α-galactosidase from Sphingomonas sp. had the K m of 2.2 mM and K cat /K m of 233 mM −1 s −1 [35]. In addition, the α-galactosidase from Irpex lacteus owned the K m of 1.2 mM and K cat /K m of 1900 mM −1 s −1 [36]. Compared with other α-galactosidases, galV was moderate in its activity to catalyze the hydrolysis of pNPG.
As shown in Figure 6, the degradation of raffinose and stachyose by galV was performed and analyzed by TLC. Most of the raffinose was rapidly degraded into sucrose and galactose in 5 min, and the residue was completely hydrolyzed in 10 min (Figure 6a). For the hydrolysis of stachyose, the degradation of the tetrasaccharide stachyose produces the intermediate trisaccharide raffinose in the initial hydrolysis process, which indicates that galV is an exoglycosidase [42]. The formed raffinose was completely converted to galactose and sucrose as the final product in 20 min (Figure 6b). The difference in efficiency of hydrolysis of the two oligosaccharides catalyzed by galV was in agreement with the result that galV showed higher substrate specificity for raffinose than stachyose. The complete hydrolysis of raffinose was faster than that of stachyose, probably because there is one more α-1, 6-galactose bond in stachyose than in raffinose [36]. the intermediate trisaccharide raffinose in the initial hydrolysis process, which indicates that galV is an exoglycosidase [42]. The formed raffinose was completely converted to galactose and sucrose as the final product in 20 min (Figure 6b). The difference in efficiency of hydrolysis of the two oligosaccharides catalyzed by galV was in agreement with the result that galV showed higher substrate specificity for raffinose than stachyose. The complete hydrolysis of raffinose was faster than that of stachyose, probably because there is one more α-1, 6-galactose bond in stachyose than in raffinose [36].

Reverse Hydrolysis of galV
The capability of galV to synthesize glycosides by reverse hydrolysis was investigated, using sugars as a donor and alcohols and sugar alcohols as an acceptor ( Table 4). The result indicated that D-galactose and glycerol were the best substrates for reverse hydrolysis catalyzed by galV. No synthetic product was observed but ethanol and ethylene glycol were used as acceptors, with similar relative galactose conversion rates (about 85.2% and 88.6% respectively). Many studies showed that glycerol was a good acceptor. For instance, α-galactosidase from Penicillium oxalicum SO exhibited high acceptor specificity towards glycerol [11]. In addition, a previous study indicated that mono-alcohols were a poor acceptor compared with ethylene glycol and glycerol [12]. However, in our study, ethanol was also a good acceptor. Therefore, this enzyme has great potential for application in synthesizing alkyl glycosides.

Reverse Hydrolysis of galV
The capability of galV to synthesize glycosides by reverse hydrolysis was investigated, using sugars as a donor and alcohols and sugar alcohols as an acceptor ( Table 4). The result indicated that D-galactose and glycerol were the best substrates for reverse hydrolysis catalyzed by galV. No synthetic product was observed but ethanol and ethylene glycol were used as acceptors, with similar relative galactose conversion rates (about 85.2% and 88.6% respectively). Many studies showed that glycerol was a good acceptor. For instance, α-galactosidase from Penicillium oxalicum SO exhibited high acceptor specificity towards glycerol [11]. In addition, a previous study indicated that mono-alcohols were a poor acceptor compared with ethylene glycol and glycerol [12]. However, in our study, ethanol was also a good acceptor. Therefore, this enzyme has great potential for application in synthesizing alkyl glycosides. Table 4. Acceptor specificity of α-galactosidase.

Acceptor
Relative Activity (%) Relative activity was calculated in relation to the conversion rate of galactose using glycerol as the acceptor, which was considered as 100%. Values are the mean ± SD of three independent experiments. Figure 7 shows the time-course for synthesizing isofloridoside using the low-cost ingredients D-galactose and glycerol. The content of D-galactose and glycerol decreased with the extension of reaction time, resulting in a gradual increase in the content of isofloridoside in the time progression of synthesis. There was no significant increase in the content of isofloridoside after reaction for 24 h. The final conversion rate of galactose was 26.6% without the optimization of reaction conditions (Figure 8a). The structure of isofloridoside was identified by LC−MS (Figure 8b). Mass spectra showed a peak with [M+Na] + molecular ions of 277.0, which confirmed that the product was isofloridoside (m/z 254). In addition, galactosyl glycerol was synthesized by transglycosylation, using activated sugar melibiose or pNPG as the substrate [43]. However, the use of these expensive substrates is not practical in producing galactosyl glycerol. On the contrary, the reverse hydrolysis reaction does not require activated sugar, the ingredients needed are cost effective, and the product is single, which is more suitable for industrial production. Wang used the α-galactosidase from Alicyclobacillus hesperidum to catalyze the synthesis of isofloridoside by reverse hydrolysis, the galactose conversion was 23% after optimizing pH, temperature, and galactose and glycerol concentration [13]. In the future, effective methods should be used to improve the content of isofloridoside, making it more suitable for expanding production, such as protein engineering on the enzyme and optimization of the reaction parameters.

Synthesis of Isofloridoside
isofloridoside was identified by LC−MS (Figure 8b). Mass spectra showed a peak with [M+Na] + molecular ions of 277.0, which confirmed that the product was isofloridoside (m/z 254). In addition, galactosyl glycerol was synthesized by transglycosylation, using activated sugar melibiose or pNPG as the substrate [43]. However, the use of these expensive substrates is not practical in producing galactosyl glycerol. On the contrary, the reverse hydrolysis reaction does not require activated sugar, the ingredients needed are cost effective, and the product is single, which is more suitable for industrial production. Wang used the α-galactosidase from Alicyclobacillus hesperidum to catalyze the synthesis of isofloridoside by reverse hydrolysis, the galactose conversion was 23% after optimizing pH, temperature, and galactose and glycerol concentration [13]. In the future, effective methods should be used to improve the content of isofloridoside, making it more suitable for expanding production, such as protein engineering on the enzyme and optimization of the reaction parameters.

Materials
E. coli DH5α and pMD 19-T vector for gene cloning, and E. coli BL21 (DE3) and pET-28a (+) for gene expression of α-galactosidase were preserved in our laboratory. Restriction endonuclease, pfu DNA polymerase and T4 DNA ligase were purchased from

Cloning, Sequence and Structure Analysis of α-Galactosidase Gene
The galV-encoding gene was amplified using the forward primer (5 -CATGCCATGGTTAGC ACGCCTTCAGCCTCC-3 with restriction NcolI site) and reverse primer (5 -CCGCTCGAGATGG GGATTATATATAATGA-3 with restriction XhoI site), and ligated to the pMD 19-T vector, transformed into E. coli DH5α and sequenced by GENEWIZ, Inc (Suzhou, China). The plasmid was digested with NcolI and XhoI and ligated to the pET-28a (+), which was used as an expression backbone and transformed into E. coli BL21 (DE3).
The three-dimensional structure of α-galactosidase from A. vitaminiphilus AWM1 was determined with a Swiss Model server using the α-galactosidase from Geobacillus stearothermophilus (PDB-ID: 4fnq) as the template and optimized based on the energy minimization.

Expression and Purification of Enzymes
The inoculum was prepared by transferring loopfuls of fresh strains cultured on a Luria broth (LB) agar plate into an LB medium containing kanamycin, followed by incubation at 37 • C for 12 h. The inoculation amount of 2% (v/v) was transferred to a fresh LB medium containing kanamycin for 90 min at 37 • C and induced with Isopropyl β-D-Thiogalactoside (IPTG) at a final concentration of 0.1 mM at 20 • C for 20 h. The sediment strain of the culture broth was resuspended in 50 mM Na 2 HPO 4 -NaH 2 PO 4 (pH 7.5) after centrifugation (12000 rpm, 4 • C, 20 min) and lysed by ultrasonication on ice (work time: 10 min, work/interval time: 3 s/5 s and ultrasonic output power: 200 W). The supernatant of total lysate, which was directly used as crude α-galactosidase for purification, was obtained by centrifugation.

Enzyme Assay
pNPG (10 mM) was incubated with an enzyme sample in 50 mM NaH 2 PO 4 -Na 2 HPO 4 buffer (pH 7.5) at 35 • C for 10 min (working volume of 250 µL). Then the absorbance of the released p-nitrophenol at 410 nm was determined. One unit of enzyme activity (U) was defined as the amount of the enzyme required to liberate 1 µmol of p-nitrophenol per minute. The enzyme was deactivated by boiling for 5 min.

Biochemical and Kinetic Properties of galV
For optimal pH, the enzyme activity was measured at various pH values ((pH 3-10), using citrate buffer (pH 3-6), phosphate buffer (pH 6-8) and Tris-HCl buffer (pH 8-10)) at 35 • C for 10 min. Under stable pH conditions, the enzyme was pre-incubated at various pH values (pH 3-10) and at 35 • C for 120 min. For optimal temperature, the enzyme activity was measured at various temperatures (30-80 • C) and pH 7.5 for 10 min. Under stable temperature conditions, the enzyme was pre-incubated at various temperatures and pH 7.5 for 120 min.
Kinetic parameters were determined by performing enzymatic reactions at 35 • C, with pNPG (0.01-20 mM) in 50 mM NaH 2 PO 4 -Na 2 HPO 4 buffer (pH 7.5) as the substrate. The products were monitored as described earlier, and the reaction rate was calculated. The catalytic constant (K cat ) and specificity constant (K cat /K m ) were calculated using K m and V max determined from the Lineweaver-Burk plot.

Substrate Specificity in a Hydrolysis Reaction
The substrate specificity of galV towards artificial substrates (pNPG and pNP-β-Dgalactopyranoside) was measured in the standard assay as described above. For natural substrates, the reaction mixture consisting of 10 mM oligosaccharide or 0.1% guar gum in 50 mM NaH 2 PO 4 -Na 2 HPO 4 buffer (pH 7.5) was incubated at 35 • C for 30 min (working volume of 2 mL). When raffinose, stachyose and guar gum were used as substrates, the enzyme activity was determined by measuring the reducing sugar using 3,5-dinitrosalicylic acid (DNS) method with galactose as a standard [44]. For melibiose, lactose and D(+)-cellobiose, the released glucose was determined by the glucose oxidase-peroxidase method with a commercial kit (Biosino, Beijing, China). One unit of the enzyme activity was defined as the amount of enzyme required to produce 1 µmol of reducing sugar or glucose per minute.
The hydrolysates of raffinose and stachyose were analyzed. A mixture of purified galV (4.7 unites/mL) and 10 mM raffinose or 10 mM stachyose in 50 mM NaH 2 PO 4 -Na 2 HPO 4 buffer (pH 7.5) was incubated at 35 • C for 30 min. Aliquots of the solution were sampled at different intervals and boiled for 5 min. Then the reaction products were analyzed by thin-layer chromatography (TLC). Hydrolysates were loaded on silica gel G plates (10 cm × 10 cm) and developed twice using n-propanol/acetic acid/water (10:15:1, v/v/v). The plate was sprayed with a mixture of methanol:sulphuric acid (4:1), followed by heating at 115 • C for 10 min to detect sugar spots.

Substrate Specificity in a Reverse Hydrolysis Reaction
The synthetic substrate specificity of α-galactosidase was investigated by mixing 0.3 M donor, 3 M acceptor, and 5 units/mL of the enzyme, giving a final volume of 10 mL by adding 50 mM Na 2 HPO 4 -NaH 2 PO 4 (pH 7.5). The reaction mixture was incubated at 35 • C for 24 h. It was then boiled for 5 min to deactivate the enzyme after incubating for 24 h. When D-galactose was used as the donor, the acceptors were alcohols (methanol, ethanol, ethylene glycol (1,2-ethanediol), glycerol and 1-butanol) and sugar alcohols (xylitol, inositol, D-sorbitol and mannitol). The donors used were sugars (D-galactose, D-(-)-arabinose, Dxylose, D-fructose, L-sorbose, N-acetyl-D-glucosamine, and glucose) with glycerol as the acceptor. The products were evaluated by HPLC. All reactions were performed in triplicate. The results were reported as mean ± standard deviation (SD).

Time-Course for Isofloridoside Synthesis
Galactose (0.3 M), glycerol (3 M) and 5 units/mL of the enzyme were mixed, and 50 mM Na 2 HPO 4 -NaH 2 PO 4 (pH 7.5) buffer was added to make the volume of the reaction solution 10 mL. The reaction was carried out at 55 • C, and 200 rpm for 36 h, and the samples were taken at regular intervals (0, 2, 4, 6, 8, 10, 15, 20, 24, and 36 h). The samples were transferred to boiling water for 5 min to inactivate the enzyme. After filtration, they were analyzed by high-performance liquid chromatography (HPLC) and liquid chromatographymass spectrometry (LC−MS).

Conclusions
In this study, we successfully discovered and heterologously expressed a thermostable α-galactosidase of the GH36 family from A. vitaminiphilus WMF1. galV showed high activity for reverse hydrolysis with D-galactose as the donor and glycerol, ethanol, and ethylene glycol as acceptors. Finally, isofloridoside was synthesized using the low-cost galactose as the donor and glycerol as the acceptor. The conversion rate of galactose was 26.6% without optimization, which provided a potential enzyme to produce isofloridoside. Furthermore, galV could be considered as a good candidate additive for the food and feed industry due to its high thermal stability and tolerance to organic solvents.  Data Availability Statement: All data included in this study are available from the corresponding author by request.