Bifidobacterium adolescentis ATCC15703, isolated from the intestine of an adult, was purchased from BeNa Culture Collection. The p-nitrophenyl glycoside derivatives, including p-nitrophenyl-β-D-glucopyranoside (pNPβGlc), p-nitrophenyl-α-D-glucopyranoside (pNPαGlc), p-nitrophenyl-β-D-galactopyranoside (pNPβGal), p-nitrophenyl-α-D-galact opyranoside (pNPαGal), p-nitrophenyl-β-D-xylopyranoside (pNPβXyl), p-nitrophenyl-α-D-xylopyranoside (pNPαXyl), p-nitrophenyl-α-L-arabinofuranoside (pNPαAraf), and p-nitro phenyl-α-L-arabinopyranoside (pNPαArap), were purchased from Yuanye Bio-Technology Co., Ltd. (Shanghai, China). The other chemicals, including sophorose, laminaribiose, cellobiose, gentiobiose, Rb1, and Rd, were purchased from Solarbio Science & Technology Co., Ltd. (Beijing, China).

The coding region of the genes encoding the three β-glucosidases was amplified from the genomic DNA of B. adolescentis ATCC15703 using polymerase chain reaction with the primers shown in Table 1 and cloned into the pET-28a vector. The recombinant plasmids were sent to Comate Bioscience Co., Ltd., for sequencing. The resulting constructs of pet-28a-babgl1a, pet-28a-babgl3b, and pet-28a-babgl1a were transformed into E. coli BL21 (DE3) and grown in Luria–Bertani medium. The expression of the target protein was induced with 0.5 mM isopropyl β-D-1-thiogalactopyranoside at 16°C for 20 h. The recombinant protein was purified using Ni-Sepharose 6 Fast Flow affinity chromatography as described previously (Zhu et al., 2015). The enzyme purity and molecular mass were determined using sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and the gels were both stained with Coomassie Brilliant Blue R250 (Conway et al., 2016) to visualize proteins. The protein concentration was determined using bovine serum albumin as the standard and the Coomassie Brilliant Blue method (Bradford, 1976).

The enzyme activity of the three β-glucosidases was determined by measuring the hydrolysis of the substrate pNPβGlc (30 mM) (Strahsburger et al., 2017). The reaction mixture of 200 μl, composed of 10 μl β-glucosidases (0.3 mg/ml), 20 μl pNPβGlc (30 mM), and 170 μl buffer (pH 7.0), was incubated at 37°C for 10 min. Then, the reaction was terminated by adding 50 μl Na₂CO₃ (0.5 M). The amount of p-nitrophenol (pNP) liberated from pNPβGlc was determined by measuring the absorption at 405 nm by BioTek ELx808 microplate reader (Winooski, VT, United States). The calculation of enzyme activity was based on pNP standard curve. One unit (U) of β-glucosidase activity was defined as the amount of enzyme that released 1 μmol of pNP per minute under the assay conditions.

The optimum pH was determined by monitoring the enzyme activity with pNPβGlc at different pH conditions (Dan et al., 2020). The optimal pH of the enzyme for pNPβGlc hydrolysis was determined at 37°C and pH 3.5–11.0 using 50 mM NaAc (pH 3.5–6.0), Na₂HPO₄–NaH₂PO₄ (pH 6.0–8.0), and Gly-NaOH (pH 8.0–11.0) buffers. A total of 200 μl of the reaction mixture containing 10 μl β-glucosidases (0.3 mg/ml), 20 μl pNPβGlc (30 mM), and 170 μl of different buffers was incubated at 37°C for 10 min. The reaction mixture without enzyme was used as a blank, and then the reaction was terminated by adding 50 μl Na₂CO₃ (0.5 M). The pH stability was determined by measuring the residual enzyme activity after incubation in buffers with varying pH at 4°C for 12 h.

The optimum temperature was determined by measuring the enzyme activity at optimal pH in the temperature range of 20–70°C. A total of 200 μl of the reaction mixture containing 10 μl β-glucosidases (0.3 mg/ml), 20 μl pNPβGlc (30 mM), and 170 μl optimal buffers was incubated at different temperatures for 10 min. The reaction mixture without enzyme was used as a blank, and then the reaction was terminated by adding 50 μl Na₂CO₃ (0.5 M). The thermal stability of BaBgl1A was determined by measuring the residual enzyme activity after incubation at 25–45°C for up to 2 h, with sampling after every 20 min. The thermal stabilities of BaBgl3A and BaBgl3B were determined by measuring the residual enzyme activity after incubation at 35–55°C for up to 2 h, with sampling after every 20 min.

The stability of the three β-glucosidases, when coincubated with factors that potentially affect enzyme activity, was evaluated (Jiang et al., 2021). Metal ions and other chemicals, including NaCl, KCl, MgCl₂, CaCl₂, CuCl₂, BaCl₂, HgCl, MnCl₂, FeCl₂, ethylenediaminetetraacetic acid (EDTA), and sodium dodecyl sulfate, were studied in this work. The enzymatic activities were tested in the presence of 5 or 25 mM (final concentration) of metal ions or other chemicals for 20 min at optimum pH and temperature. The residual activity of the three β-glucosidases was determined using pNPβGlc as a substrate as described before, and the activities are expressed as a percentage of the activity obtained in the absence of the compound. The effect of polyols on three β-glucosidases when incubated with the substrate in the presence of 2 M polyols, including glucose, xylose, mannose, galactose, arabinose, and fructose, was determined. For assessing the residual activity, the enzyme was initially incubated at optimum pH and temperature for 20 min. Then, the residual activity was determined at optimum temperature using pNPβGlc as substrate. Enzyme activity without any additive was included as the control (100%).

Substrate specificity was determined by measuring the enzyme activity on different artificial substrates (pNPβGlc, pNPαGlc, pNPβGal, pNPαGal, pNPβXyl, pNPαXyl, pNPαAraf, and pNPαArap) and disaccharides (sophorose, laminaribiose, cellobiose, and gentiobiose). The reaction mixture of 200 μl, which was composed of 10 μl β-glucosidases (0.3 mg/ml), 20 μl of different artificial substrates (30 mM), and 170 μl of optimal buffers, was incubated at optimal temperature for 10 min. The reaction mixture without enzyme was used as a blank. The reaction was stopped by the addition of 50 μl Na₂CO₃ (0.5 M). For artificial substrates, the absorbance of the mixture was measured at 405 nm. The activities over disaccharides were measured following the 3,5-dinitrosalicylic acid method described previously (Zhu et al., 2016).

To investigate the bioconversion ability of the β-glucosidases derived from B. adolescentis, two different ginsenosides (Rb1 and Rd) were used as substrates. The enzyme (final concentration of 0.1 mg/ml in 50 mM buffer of optimum pH) was allowed to react with 5 mg/ml of ginsenosides (Rb1 and Rd) at 37°C for 2 h. The reaction was terminated by boiling the sample for 10 min to remove protein. Samples were taken, filtered, and analyzed using thin-layer chromatography (TLC) and-high performance liquid chromatography (HPLC). For TLC, GF₂₅₄ silica gel plates were used, which were developed in a solvent system consisting of butyl alcohol/ethyl acetate/water (4:4:1, v/v/v) (Zheng et al., 2021). The spots on the TLC plates were visualized by spraying with 5% (v/v) H₂SO₄ and identified by comparing with the ginsenoside standards. A Shimadzu HPLC system (CTO-20A pump and SPD-20AVD UV detector) was used for the analysis. The separation was performed on a Unitary-C18 column (5 μm, 4.6 mm × 250 mm). The mobile phase consisted of water (solvent A) and acetonitrile (solvent B). The elution gradient consisted of 70% solvent A, followed by 100% solvent B for 0–70 min. The flow rate was 0.4–0.8 ml/min, the injection volume was 20 μl, and the detection wavelength was 203 nm (Kim et al., 2022).

The sequence of the three β-glucosidases—BaBgl1A, BaBgl3A, and BaBgl3B—from B. adolescentis ATCC15703 (with GenBank accession numbers BAF40068.1, BAF39975.1, and BAF39978.1) contains 391, 751, and 809 amino acids with a theoretical molecular mass of 44.2, 81.1, and 87.8 kDa, respectively. The deduced amino acid sequence of BaBgl1A was similar to that of the GH1 family. The multiple amino acid sequence alignment indicated that BaBgl1A exhibited identities with some characterized GH1 β-glucosidases, such as β-glucosidases from Alicyclobacillus acidiphilus (45%), Exiguobacterium antarcticum B7 (44%), and Clostridium cellulovorans (41%) (Jeng et al., 2011; Zanphorlin et al., 2016; Delgado et al., 2021). While the deduced amino acid sequences of BaBgl3A and BaBgl3B were similar to those of the GH3 family (CAZy database)¹, the multiple amino acid sequence alignment indicated that BaBgl3A and BaBgl3B exhibited identities with some characterized GH3 β-glucosidases, including β-glucosidases from Bacteroides ovatus, Bifidobacterium longum subsp. longum KACC 91563, and Listeria innocua Clip11262 (Masahiro et al., 2016; Rikuto et al., 2017; Yan et al., 2018). Signal sequences were absent in the three β-glucosidases. The multiple amino acid sequence alignment indicated that the amino acid sequences of the three recombinant β-glucosidases were similar (Supplementary Figure 1), such as Phe71, Pro72, Ala92, Glu128, and Asp129. These amino acids might be present in the main domain required for glucosidase activity. The structure of these β-glucosidases was evaluated by PyMOL software. The results indicated that the active sites of BaBgl1A were Glu159 and Glu306, while the active sites of BaBgl3A and BaBgl3B were both composed of Asp and Glu, the sites of BaBgl3A were Asp232 and Glu417, and the sites of BaBgl3A was Asp306 and Glu549. In addition, BaBgl3A and BaBgl3B were more similar to each other than BaBgl1A, with a similarity level of 32%. This may be because both BaBgl3A and BaBgl3B were members of the GH3 family, while BaBgl1A was from the GH1 family.

To investigate the enzymatic properties of the three β-glucosidases, the enzymes were heterologously expressed in E. coli BL21 (DE3) with high enzyme production efficiency. The recombinant β-glucosidases were expressed as soluble proteins in E. coli BL21 (DE3) and showed single bands after purification using a Ni²⁺ affinity column of approximately 44, 81, and 88 kDa, respectively, which were close to their theoretical molecular mass (Figure 1). To compare the expression of the three enzymes, enzyme activity was determined by measuring the increase in absorbance of the reaction mixture at 405 nm using pNPβGlc as the substrate (Table 2). BaBgl1A was purified 2.3-fold, with 85.1% yield and 71.2 U/mg specific activity; BaBgl3A was purified 1.2-fold, with 9.2% yield and 3.9 U/mg specific activity; and BaBgl3B was purified 1.3-fold, with 14.1% yield and 7.6 U/mg specific activity. Based on the results of purification, it could concluded that the properties of BaBgl1A were better than those of BaBgl3A and BaBgl3B.

The biochemical properties of the three β-glucosidases were examined from pH 3.5 to 11.0 with pNPβGlc as the substrate (Figure 2). The maximum activity of BaBgl1A, BaBgl3A, and BaBgl3B was observed at pH 6.0, 6.5, and 6.5, respectively. BaBgl1A and BaBgl3B were more stable than BaBgl3A. After pre-incubation at 4°C for 24 h, BaBgl1A had over 60% of the enzyme activity retrieved at pH 4.5–7.0, and BaBgl3B had over 60% of the enzyme activity retrieved at pH 4.0–8.0 (Figures 2D,F). The effect of temperature on enzyme activity was investigated at optimal pH. As shown in Figure 3, maximum activities of BaBgl1A, BaBgl3A, and BaBgl3B were observed at 30, 45, and 50°C, respectively. According to optimal temperature, the temperature stability of BaBgl1A was investigated by varying the temperature from 25 to 45°C at optimal pH and those of BaBgl3A and BaBgl3B were investigated by varying the temperature from 35 to 55°C (Figures 3A–C). The results showed that BaBgl3A exhibited poor thermal stability. More than 80% of its activity remained after treatment at over 35°C for 1 h; however, the enzyme activity decreased below 80% after incubation at all the temperature ranges for 2 h (Figure 3E). BaBgl1A and BaBgl3B had better thermal stability than BaBgl3A, as their enzyme activity remained over 80% at a wider temperature range (Figures 3D,F). This indicated that the optimal conditions for BaBgl1A, BaBgl3A, and BaBgl3B were pH 6.0 and 30°C, pH 6.5 and 45°C, and pH 6.5 and 50°C, respectively.

For determining the efficiency of hydrolysis, the effects of metal ions and reagents on BaBgl3A, BaBgl1A, and BaBgl3B activity were further investigated. As shown in Figure 4, at the concentration of 5 mM, only Mn²⁺ and EDTA slightly activated BaBgl1A, and Ca²⁺ slightly activated BaBgl3A. In contrast, Ca²⁺, Cu²⁺, and Ba²⁺ significantly inhibited BaBgl1A and BaBgl3B. At the concentration of 50 mM, all the metal ions and reagents used in this study had little or inhibitory effect on the activities of the three enzymes.

Furthermore, different monosaccharides were added to the reaction system to investigate the effect of polyols on the three β-glucosidases. Most monosaccharides could increase BaBgl1A and BaBgl3B activity, while mannose and fructose acted as competitive inhibitors for BaBgl1A and BaBgl3B. Mannose could significantly inhibit their activity by more than 70% at 1-M monosaccharide concentration (Figures 5A,C). Glucose and xylose could increase the activity of only BaBgl3A by 10% at 1-M monosaccharide concentration (Figure 5B). The residual activities of the enzymes after incubation at optimal conditions for 120 min were evaluated. As shown in Figure 5D, the addition of glucose and galactose to the reaction resulted in a stronger protection effect of enzyme activity, and BaBgl1A still retained more than 80% vitality. Similar results could be seen in Figures 5E,F; compared to that observed in the blank control, glucose and xylose could increase the residual activity of BaBgl3A, while xylose and arabinose could increase the residual activity of BaBgl3B. The results mentioned above indicate that glucose, xylose, and galactose can be used as candidates for β-glucosidase activity enhancers in B. adolescentis.

The activities of the three recombinant enzymes from B. adolescentis were assessed using aryl-glycosides and disaccharides. The results (Table 3) showed that BaBgl1A, BaBgl3A, and BaBgl3B exhibited the best activity toward pNPβGlc when aryl-glycosides were used as substrates. In addition, BaBgl3A was also active on pNPβXyl and pNPαArap, and BaBgl3B was also active on pNPβGal, pNPβXyl, and pNPαArap. These results indicated that BaBgl1A had better substrate specificity than BaBgl3A and BaBgl3B. When disaccharides were used as substrates, BaBgl1A, a GH1 enzyme, showed good activity on sophorose [glucose-β-(1→2)-glucose] and medium activity on laminaribiose [glucose-β-(1→3)-glucose] and cellobiose [glucose-β-(1→4)-glucose], while there was no activity on gentiobiose [glucose-β-(1→6)-glucose]. The linkage preference of BaBgl1A for disaccharides was β-1, 2 > β-1, 3 > β-1, 6 > > > β-1, 4. The GH3 enzymes BaBgl3A and BaBgl3B exhibited good activity on laminaribiose [glucose-β-(1→3)-glucose] and cellobiose [glucose-β-(1→4)-glucose], respectively. They also showed a different activity on other disaccharides. Therefore, the disaccharide substrate selectivity of the three β-glucosidases from B. adolescentis differed. BaBgl1A might mainly act as β-1,2-glucosidase, BaBgl3A might mainly act as β-1,3-glucosidase, and BaBgl3A might mainly act as β-1,4-glucosidase.

To further verify the substrate selectivity of the three recombinant β-glucosidases, ginsenosides Rb1 and Rd, which contain different glucose linkage types (β-glucose, β-1,2-glucose, and β-1,6-glucose), were used as the substrates. The results of TLC and HPLC revealed that, in the presence of BaBgl1A and BaBgl3A, ginsenoside Rb1 was converted into different metabolites, while BaBgl3B could not hydrolyze the ginsenosides, indicating that BaBgl3A and BaBgl1A are ginsenoside-hydrolyzing enzymes (Figure 6A). To verify the bioconversion pathway and efficiency of BaBgl3A and BaBgl1A, individual ginsenosides were used as substrates for the biotransformation, and the products were analyzed using HPLC. As shown in Figure 6B, Rb1 could be converted into Rd and Gyp XVII by BaBgl3A, with yields of 9.65 and 5.65%, respectively. This suggested that BaBgl3A can remove the outer glucose linked to the β-1,6 linkage of the C-20 position and that linked to the β-1,2 linkage of the C-3 position in ginsenoside Rb1, while it had no activity toward Rd (Figures 6B,C). This indicated that BaBgl3A had weak activity toward the β-1,2-/β-1,6 linkage of glucose. Similarly, BaBgl1A could hydrolyze the β-1,2 linkage of the C-3 position in ginsenosides Rb1 and Rd, producing the minor ginsenoside Gyp XVII and F₂ with yields of 100 and 72.16%, respectively (Figures 6B,C). The transformation was almost completed in 2 h, which suggested that BaBgl1A had strong activity toward the β-1,2 linkage of glucose.