The T. trogii S0301 strain was routinely maintained on the GYP medium (2% glucose, 0.5% yeast extract, 0.5% tryptone, and 0.1% MgSO₄⋅7H₂O) at 4°C at the Biotechnology Research Center of Life Science and Technology College, Kunming University of Science and Technology (Yan et al., 2014; Zhang et al., 2021).

The genome and putative mRNA sequencing data of T. trogii S0301 strain presented here are associated with NCBI BioProject PRJNA480364 and BioSample SAMN09635320 (Liu et al., 2019). The structure of GH3 gene family members was analyzed using the online software Gene Structure Display Server,¹ and GH3 protein domains were identified and annotated using the Simple Modular Architecture Research Tool.² BioEdit software was used for sequence alignment, and the software package MEGA 6³ was used for phylogenetic analysis. According to the methods described in our previous study, gene expression profiles of the GH3 gene family members were analyzed under the three culturing conditions, Highley’s basal salt medium containing 1% (w/v) glucose (1% G), 0.5% (w/v) glucose, and 0.5% (w/v) ball-milled oak woods (0.5% LG), or 1% (w/v) ball-milled oak woods (1% L), respectively (Liu et al., 2019).

Total RNA was extracted from 6-day-old fresh hyphae of T. trogii S0301 on the GYP medium using TRIzol RNA Isolation Reagents (Promega, United States). Then, the first-strand cDNA was synthesized using HiScript II Q RT SuperMix for qPCR (+gDNA wiper) (Vazyme Biotech, China). The open reading frame of TtBgl3 (T_trogii_12914) was then amplified by PCR from the cDNA with the specific primers 5′-tctagaATGTCGCGCGACTTCCTCG-3′ (TtBgl3-F) and 5′ctcgagCACCCCGTTCCATGTGAATC-3′ (TtBgl3-R). After digesting by EcoRI and XhoI, the resulting PCR fragments were cloned to the prokaryotic expression vector pET-28b which was previously digested by the same restriction enzymes. After sequencing, the resulting plasmid, named pET-28b-TtBgl3, was transformed into the E. coli strain Rosetta (DE3) and the single colony containing the cDNA of TtBgl3 was obtained (Zhang et al., 2021).

The expression of the recombinant enzyme was performed by adding 0.1 mM isopropyl-β-D-thiogalactopyranoside (IPTG) to the bacterial suspension at the initial concentration of OD600 = 0.7. After 18 h at 16°C, the induced cells were harvested by centrifugation at 4°C for 5 min at 10,000 g and washed twice with PBS buffer (pH 7.4). After sonication and centrifugation at 12,000 g for 10 min, the supernatants was obtained as the crude enzyme solution. To purify the recombinant TtBgl3, the crude enzyme solution was applied onto a Ni-NTA column (Sangon Biotech, China). The TtBgl3 protein was eluted from the column with midazolam buffer by gradient elution. Enzyme fractions were analyzed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) with a 5% (w/v) stacking gel and a 12% (w/v) separating gel. After electrophoresis, SDS-PAGE was stained with Coomassie blue R-250. Protein concentrations were determined using a bicinchoninic acid (BCA) protein assay kit (Tiangen Biotech, China) (Zhang et al., 2021).

The β-glucosidase activity was determined with 4-nitrophenyl β-D-glucopyranoside (pNPG) as the substrate as described previously. In brief, a reaction mixture containing 5 μL of 10 mM pNPG and 5 μL of appropriately diluted crude enzyme in phosphate citrate buffer (200 mM, pH 6.0) was evaluated at 50°C for 10 min. And the reactions were stopped by adding 150 μL of 1 M Na₂CO₃ solution. Then, the increase in the absorbance at 405 nm was recorded, and one unit of activity was defined as the amount of enzyme that released 1 μmol of pNPG per minute under the test conditions. All the experiments were performed in triplicate (Qin et al., 2018; Zhang et al., 2021).

To determine the optimal pH of the recombinant TtBgl3 activity, the reaction mixtures were preincubated with 200 mM disodium hydrogen phosphate-citric acid buffer (pH 2.4–8) at 50°C for 10 min. To determine the optimal temperature of the recombinant TtBgl3 activity, the reaction mixtures were pre-incubated at given temperatures (30–80°C) at the optimum pH for 10 min. For the pH and thermal stability analysis, the enzyme was preincubated with different buffers (pH 3.0–11.0) at 37°C for 2 h or at the optimal pH at 50 and 60°C for different time durations without pNPG. Then, the activity of the reaction mixture or the pretreated enzyme was measured under the optimum pH and temperature according to the standard enzyme determination. An identical amount of enzymes placed on ice was used as the positive control and set as 100%. All assays were performed in triplicate (Li et al., 2018).

The reaction mixture containing 200 mM disodium hydrogen phosphate-citric acid buffer (pH 5.0), 1 mM pNPG and 50 U/mL of TtBgl3, was mixed with Fe³⁺, Sr²⁺, NH⁴⁺, Ba²⁺, K⁺, Na⁺, Co²⁺, Mg²⁺, Ni²⁺, Ca²⁺, Mn²⁺, Cu²⁺, and Zn²⁺ at the final concentration of 10 mM to determine the effect of metal ions on the activity of TtBgl3 (Akram et al., 2018). After incubation at 50°C for 10 min, the increase in the absorbance at 405 nm was measured according to the standard enzyme determination. The activity of the reaction mixture without metal ions was recorded as 100%.

The kinetic parameters (K_m, V_max, and K_cat) of the recombinant TtBgl3 were estimated in disodium hydrogen phosphate citric acid buffer (pH 5.0, 200 mM) at 55°C for 5 min, with different concentrations of pNPG (0.2–1.5 mM) as the substrates. The data were calculated and analyzed according to the Lineweaver–Burk method (Li et al., 2018).

We selected phenols (gastrodin), coumarins (esculin), isoflavones (daidzin), and flavonoids (baicalin) as the transformation substrates of TtBgl3 to analyze the transformation ability of the recombinant TtBgl3 to different phenolic glycosides. The reaction mixture (400 μL) consisted of 200 mM sodium phosphate buffer (pH 6.0), 1 mg/mL substrate and 80 U/mL recombinant TtBgl3. After incubating the reaction mixture for 12 h at 37°C, the reaction was stopped using 400 μL of methanol. The mixture was centrifuged at 10,000g for 30 min at 4°C. The supernatant was analyzed by UHPLC-ESI-Q-TOF-MS (Agilent Technologies, United States) to confirm the identity of the transformation products. The optimal dosage of the recombinant TtBgl3 was determined by adding 1, 5, 10, 20, 40, 60, and 80 U/mL of the recombinant TtBgl3 to the reaction system for 12 h at 37°C. The optimal transformation time of the recombinant TtBgl3 was determined by incubating the transformation system after adding 10 U/mL recombinant TtBgl3 at 37°C for 0.5, 1, 2, 4, 8, and 12 h. All samples were analyzed by high-performance liquid chromatography (HPLC).

All the transformation samples were analyzed by HPLC-UV on a Shimadzu Analytical Instrument (Shimadzu, Japan). The separation was performed on a YMC-pack ODS-A column (4.6 × 250 mm; i.d., 5 μm; YMC Co., Ltd., Japan) using a mobile phase of water with 0.05% phosphoric acid (A) and acetonitrile (B) at a flow rate of 1 mL/min. The isocratic profile of gastrodin was as follows: 0–25 min, 3% B. The gradient profile of esculin was as follows: 0–20 min, 20% B; 20–25 min, 20–50% B. The gradient profile for the separation of daidzin and baicalin was as follows: 0–1 min, 5% B; 1–25 min, 5–70% B. The column temperature was set at 30°C. The detection wavelength of gastrodin, esculin, daidzin, and baicalin was set at 220, 348, 250, and 280 nm, respectively (Yan et al., 2016; Chang et al., 2018).

Homology modeling of TtBgl3 was performed using SWISS-MODEL software⁴ based on the crystal structure of TtBgl3 (PDB accession number 3abz.1.A). The 2D structures of phenolic glycosides (gastrodin, esculin, daidzin, and baicalin) were obtained from the PubChem database.⁵ The optimized structures of four phenolic glycosides were then used for docking with TtBgl3, using AutoDock Vina (version 1.1.2). The structure was visualized and analyzed using PyMOL.

The pET-28b-TtBgl3 plasmid was used as the template for site-directed mutagenesis to introduce H85A and K467L point mutations into TtBgl3 using a Fast Site-Directed Mutagenesis Kit (Tiangen Biotech, China), with subsequent verification by DNA sequencing. Mutagenic primer pairs were as follows: 5′-CGGAGTTCGCGGCTCCTCTCTCTTCGTTTCCAC-3′ (TU-A-F) and 5′-AGAGAGGAGCCGCGAACTCCGTTTGGGC CGTC-3′ (TU-A-R) for H85A; 5′-GCAAGACACTCGTGTCA TACTGAATGACTTC-3′ (TU-B-F) and 5′-ATGACACGAG TGTCTTGCAGAACGGTCTCC-3′ (TU-B-R) for K467L. The mutagenic nucleotides are shown in italic type. The expression of the mutant protein and transformation system toward gastrodin was assessed as described previously.

The results of all experiments were based on the average of three independent experiments ± standard deviation (SD), and the statistical significance was determined using the unpaired-samples t-test and one-way analysis of variance.

The T. trogii genome contains 10 genes (Supplementary Figure 1) that can be classified as the GH3 gene family (Liu et al., 2019). Multiple sequence alignment (Supplementary Figure 2) and phylogenetic tree analysis showed that the coding sequences of all the identified GH3 genes except T_trogii_08451 possessed the conserved Glyco_hydro_3, Glyco_hydro_3_C, and Fn3-like domains, and clustered with the putative β-glucosidas, whereas T_trogii_08451 lacked the Fn3-like domains and clustered together with CbsA, a member of N-acetyl beta-glucosaminidase (Figure 1A). When cultivated on the GYP liquid medium containing wood powder (0.5%, w/v) and glucose (0.5%, w/v) as the carbon sources (Liu et al., 2019) for 2 (LG_–2), 5 (LG_–5) and 8 days (LG_–8), the expression pattern of 10 predicted β-glucosidases in T. trogii S0301 could be divided into three groups. The increase in four GH3 transcripts T_trogii_ (08451, 01689, 07129, and 08766) in Group I and four GH3 transcripts (T_trogii_08757, 00566, 01687, and 14312) in Group III was in line with the level of glucose consumption and maintained at a higher level during the lignocellulosic utilization stage (LG_–8), while the transcription of two GH3 transcripts (T_trogii_12914 and 00539) in Group II was only up regulated at the point of glucose depletion (LG_–5) (Figure 1B). These results indicated that GH3 family members of Group II might mainly be involved in the carbon source conversion stage, while members of Group I and III were the enzymes involved in lignocellulose degradation.

To further explore the characteristics of two GH3 members in Group II, the cDNAs of T_trogii_12914 and 00539 were cloned and prokaryotically expressed. The recombinant protein coded by T_trogii_12914, with the calculated molecular weight value of 97 kDa (Figure 2), had significant activity toward pNPG, and did not exhibit hydrolytic activity toward pNP-GlcA, pNPX, cellobiose, and cellotriose, which further confirmed that T_trogii_12914 coded a typical GH3 glucosidase, named TtBgl3. In addition, the expressed proteins coded by T_trogii_ 00539 existed in the form of inclusion body and was inactive toward pNPG.

The enzymatic properties of the recombinant TtBgl3 were determined using pNPG as the substrate. The optimum pH ranged from pH 5.5 and 6.0, with >80% of the maximum activity at pH 5.0–7.0 (Figure 3A and Table 1). The optimum temperature for TtBgl3 was 50°C (Figure 3B and Table 1). TtBgl3 was stable at a wide range of pH from 5.0 to 10.0 with more than 95% of maximum activity after 2 h of treatment (Figure 3C). TtBgl3 retained 90% of their initial activity after 5 h incubation at 50°C (Figure 3D). However, β-glucosidase activity decreased rapidly at temperatures above 60°C (Figure 3). The specific activity, K_m, V_max, and K_cat values of TtBgl3 for pNPG at 50°C and pH 5.5 were determined to be 341.5 U/mg, 0.55 mM, 263.16 μM/mg/min, and 164.5, respectively (Table 1). Catalytic efficiency (K_cat/K_m) of TtBgl3 against pNPG were 297.6 (Table 1). TtBgl3 activity increased with Fe³⁺ and decreased with Ni²⁺, Ca²⁺, Mn²⁺, Cu²⁺, Zn²⁺, while Sr3+2+, Ba₂₊, K⁺, Na⁺, Ca²⁺, and Mg²⁺ did not show any obvious effect (Table 2).

To test the ability and their potential applications of TtBgl3 in biotransformation of phenolic glycosides, we selected gastrodin, esculin and daidzin with β-glucoside bond and baicalin with β-glucuronide bond as the substrates of TtBgl3, and analyzed the conversion ability of recombinant TtBgl3 to phenolic glycosides (Figure 4). Based on HPLC analysis, the reaction system of phenolic glycosides produced new product peaks after TtBgl3 treatment (Figure 5). Each compound was analyzed by mass spectrometry (MS) to further confirm the structures of the transformation products (Figure 6). After comparing the molecular weight of each product, it was found that gastrodin, esculin, and daidzin all lost one glucose molecule and were converted into p-hydroxybenzyl alcohol, esculetin, and daidzein, respectively. In addition, baicalin lost a glucuronic acid molecule and converted to baicalein. By analyzing the chemical structure of four phenolic glycosides, we inferred that TtBgl3 acted on the β-glucoside/glucuronide bond of phenolic glycosides and transformed them into aglycon.

We optimized the enzyme concentration and transformation time at 37°C to further improve transformation efficiency (Supplementary Figures 3, 4). The optimal transformation parameters for three natural compounds were as follows: 60 U/mL for 8 h, 5 U/mL for 2 h, and 5 U/mL for 1 h at 37°C for gastrodin, esculin, and daidzin at the concentration of 500 g/mL with the transformation efficiency of 0.23, 0.92, and 0.31 mM/h, respectively. Under optimized conditions, the three substrates of gastrodin, esculin, and daidzin were completely hydrolyzed by TtBgl3, and the catalytic efficiency reached 100% (Table 3). The hydrolysis efficiency of baicalin was slightly lower, which was about 50%. When the substrate concentration was 500 g/mL, the yield of esculetin was the highest, 329.71 g/mL, while the yield of baicalein was 36.22 g/mL. Therefore, TtBgl3 not only acted on the β-glucoside bond of phenolic glycosides, but also acted on the β-glucuronide bond of phenolic glycosides, and the former was better.

The 3D structure of TtBgl3 was established by online homologous modeling of the SWISS-MODEL database to predict the structure of TtBgl3. The QSQE, Seq identity, and GMQE were 0.4, 41.73, and 0.73, respectively, indicating that the structural model was reliable. As shown in the structural model (Figure 7A), the enzyme contained three domains (β/α)8-barrel domain, α/β sandwich domain, and fibronectin type III domain.

The modular docking procedure was used to analyze the binding sites of different phenolic glycosides with the recombinant TtBgl3 to understand the substrate specificity of the TtBgl3 (Figures 7B,C). The binding energies of proteins and small molecules were calculated using AutoDock Vina 1.1.2 software. After docking, the absolute values of binding constants of gastrodin, esculin, and daidzin was 7.9, 8.2, and 9.1, respectively. The substrates gastrodin, esculin, and daidzin could dock into the catalytic cleft of TtBgl3 well when the absolute values of binding constants were more than 7. These substrates were mainly combined with enzymes by hydrogen bonding, although the structures of the compounds were different. The substrate gastrodin containing a β-glucoside bond had two enzyme catalytic sites. It was speculated that the two residues of His85 and Lys467 were the enzyme-binding sites of the substrate containing a β-glucoside bond (Figures 7B,C). According to the β-glucosidase activity assays, the H85A mutant lost its catalytic activity with 8.83 U/mg protein, which is 2.62% of the activity of wild-type (337.4 U/mg protein) (Table 4). And the K476L mutant showed 23.83% of the activity of wild-type (Table 4). Further transformation study showed that the H85A and K476L mutants could cause a marked decrease in transformation rate with 0.18 and 6.41%, respectively, compared with complete transformation of gastrodin at the concentration of 500 μg/mL (Table 4). It was further confirmed that the two residues of His85 and Lys467 were essential for the catalytic hydrolysis activity of the recombinant TtBgl3.