Isolation and Characterization of β-Glucosidase-Producing Bacteria for Compound K Production via Biotransformation

2.1. Bacterial strains and culture conditions

In this study, a total of 28 bacterial strains were selected from the laboratory-maintained microbial strain library, including E10 and E33 (isolated from leaf surfaces), G4 (isolated from humus-rich soil), K1, K3, K3-1, K3-4, K4, and K7-1 (isolated from waste feathers), L1, L4, L31, L36, L49, L55, L66, and L77 (isolated from Kimchi), N1, N2, N4, N6, and N7 (isolated from rhizosphere soil), P3 and P15 (isolated from agricultural field soil), W62 and W69 (isolated from dairy products), and Y1 and Y9 (isolated from paddy soil). Unless otherwise stated, strains isolated from Kimchi were cultivated in De Man-Rogosa-Sharpe (MRS) medium (Difco, BD, USA), whereas the remaining strains were cultured in Luria-Bertani (LB) medium (Difco, BD, USA) at 30°C and 200 rpm for 24 h. Short-term storage was performed at 4°C, and long-term preservation was conducted at −80 °C in the presence of 20% (v/v) glycerol.

2.2. Screening and quantitative assay of β-glucosidase activity

Primary qualitative screening for β-glucosidase-producing strains was performed using esculin agar (HiMedia, India) according to the method of Lee et al.(2014). Strains forming dark brown halos were selected as putative β-glucosidase producers.

Quantitative β-glucosidase activity was determined following the method of Matsuura and Obata (1993) with p-nitrophenyl-β-D-glucopyranoside (pNPG) as a chromogenic substrate. Culture broths were centrifuged at 12,000 × g and 4°C for 10 min, and the supernatants were filtered through 0.22 μm syringe filters to obtain crude enzyme extracts. Crude enzyme solution (500 μL) was mixed with 500 μL of 10 mM pNPG prepared in 0.05 M sodium acetate buffer (pH 5.0) and incubated at 40°C for 30 min. The reaction was terminated by adding 1 mL of 0.5 M Na₂CO₃, and the released p-nitrophenol (pNP) was measured at 405 nm using a UV-Vis spectrophotometer (Ultrospec 4000; Pharmacia Biotech, UK). Heat-inactivated crude enzyme (100°C, 10 min) was used as a negative control. One unit of β-glucosidase activity was defined as the amount of enzyme causing an absorbance change of 0.001 per minute under the assay conditions.

2.3. Molecular identification of selected strains

Strains exhibiting high β-glucosidase activity were identified by 16S rRNA gene sequencing. The 16S rRNA gene was amplified by polymerase chain reaction (PCR) using the universal primers 27F (5’-AGAGTTTGATCCTGGCTCAG-3’) and 1492R (5’-GGTTACCTTGTTACGACTT-3’) (Lane, 1991). The PCR products were purified and sequenced by a commercial sequencing service (Genotech, Daejeon, Republic of Korea). The obtained sequences were analyzed using BLASTN against the NCBI GenBank database to determine their closest phylogenetic relatives. Multiple sequence alignment was performed using Clustal X (version 1.81) (Thompson et al., 1997). Phylogenetic analysis was conducted using MEGA X software, and a phylogenetic tree was constructed based on the neighbor-joining method with 1,000 bootstrap replications to assess branch reliability (Kumar et al., 2018).

2.4. Optimization of β-glucosidase production

To enhance β-glucosidase production, the effects of basal medium type, initial pH, and incubation temperature were systematically evaluated according to the method of Shin et al. (2015). For strain L66, the basal media tested were MRS broth, MRS supplemented with 0.1% (w/v) soluble starch (MRSS), MRS supplemented with 0.1% (w/v) glutinous rice flour (MRSG), yeast extract-based mineral medium (YM), YM supplemented with 0.1% (w/v) soluble starch (YMS), and YM supplemented with 0.1% (w/v) glutinous rice flour (YMG). For strain K7-1, the basal media tested were tryptic soy broth (TSB), TSB supplemented with 0.1% (w/v) soluble starch (TSS), TSB supplemented with 0.1% (w/v) glutinous rice flour (TSG), nutrient broth (NB), NB supplemented with 0.1% (w/v) soluble starch (NBS), NB supplemented with 0.1% (w/v) glutinous rice flour (NBG), Luria-Bertani broth (LB), LB supplemented with 0.1% (w/v) soluble starch (LBS), and LB supplemented with 0.1% (w/v) glutinous rice flour (LBG). The composition of YM medium was yeast extract (1%, w/v), KH₂PO₄ (0.1%, w/v), MgSO₄ (0.05%, w/v), KCl (0.05%, w/v), FeSO₄ (0.01%, w/v), and MnCl₂ (0.05%, w/v). All media were adjusted to pH 7.0 prior to sterilization.

Following medium screening, the effects of initial pH (4.0-9.5) and incubation temperature (25-35°C) on β-glucosidase production were further investigated under otherwise identical cultivation conditions.

2.5. Biochemical characterization of crude β-glucosidase

Crude enzyme preparations were used for biochemical characterization. Optimal temperature was determined over the range of 25-60°C at 5°C intervals. Optimal pH was evaluated using 0.05 M buffers ranging from pH 3.0 to 11.0, including sodium acetate/sodium citrate (pH 3.0-5.5), potassium phosphate (pH 6.0-8.0), Tris-HCl (pH 7.0-9.0), and glycine-NaOH/bicarbonate-NaOH (pH 9.0-11.0). Thermal stability was assessed by incubating enzymes at 4-100°C for 1 h prior to activity measurement. pH stability was determined after incubating enzyme samples in buffers of different pH values at 30°C for 1 h, followed by residual activity measurement.

2.6. Enzymatic bioconversion of ginsenosides

Red ginseng concentrate (Korea Ginseng Corp., Korea) was used as the substrate. The concentrate was diluted with sterile distilled water to obtain a total ginsenoside concentration of 5 mg/mL and filtered through a 0.22 μm syringe filter. Crude enzyme and substrate solutions were mixed at a ratio of 1:1 (v/v) and incubated under the optimal pH and temperature determined in Section 2.5 at 100 rpm for 1-7 days in the dark. To ensure comparability among reactions, the crude enzyme was added to achieve a final β-glucosidase activity of 5.0 U/mL in the reaction mixture, as determined by the pNPG assay described in Section 2.2. Samples were collected at regular intervals for product analysis.

2.7. TLC analysis of biotransformation products

Biotransformation products were analyzed using silica gel 60 F₂₅₄ TLC plates (Merck, Germany) according to the method of Chi et al. (2005). Reaction mixtures (5 mL) were extracted with saturated n-butanol (2 mL) by vigorous mixing, and the upper organic phase containing ginsenosides was collected for TLC analysis. The developing solvent system consisted of chloroform: methanol: water (65:35:10, v/v/v), and the lower (chloroform-rich) phase was used for development after phase separation. Plates were sprayed with 10% sulfuric acid in ethanol and heated at 110°C for 5 min. Compound K was identified by comparison with authentic standards (AceEMzyme, Korea). For semi-quantitative comparison of band intensity, TLC plates were photographed under identical illumination and camera settings, and densitometric analysis was performed using ImageJ (NIH, USA). Band intensities were quantified as integrated optical density after local background subtraction and were normalized to the total lane signal to enable relative comparison among reaction times and between strains.

2.8. Data analysis

All experiments were performed in triplicate. Data are presented as mean ± standard deviation (SD). Statistical significance was evaluated using one-way analysis of variance followed by Tukey’s post hoc test (p < 0.05).

3.1. Screening of β-glucosidase-producing bacteria

A total of 28 isolates were screened for β-glucosidase activity using an esculin hydrolysis assay. On esculin agar, β-glucosidase-positive strains produce characteristic dark brown to black zones resulting from the formation of iron-esculetin complexes following esculin cleavage, a reaction that has been widely applied as a preliminary screening method for β-glucosidase producers (Lee et al., 2014). Except for strains E33, K3, N2, N4, and Y9, all tested isolates formed distinct dark brown zones surrounding the colonies on esculin agar, indicating positive β-glucosidase activity in the primary screening assay (Fig. 1). The esculin-positive isolates were subsequently subjected to quantitative β-glucosidase activity measurement using pNPG as a chromogenic substrate. Hydrolysis of pNPG releases p-nitrophenol, which can be quantified sensitively after alkalinization, enabling reliable comparison among strains (Strahsburger et al., 2017; Kim et al., 2018). Among the tested isolates, strains L66 and K7-1 exhibited the highest β-glucosidase activities and were therefore selected for further investigation (Fig. 2). The combined use of plate-based screening and quantitative assays enhanced selection reliability by minimizing false positives associated with colony pigmentation, diffusion variability, and growth-related artifacts (Strahsburger et al., 2017).

Fig. 1.

Primary qualitative screening of β-glucosidase-producing strains on esculin agar plates. Dark brown to black halos surrounding colonies indicate positive β-glucosidase activity.

Fig. 2.

Quantitative β-glucosidase activities of esculin-positive strains measured using pNPG as a substrate.

3.2. Molecular identification and phylogenetic analysis

The taxonomic identities of strains L66 and K7-1 were determined based on 16S rRNA gene sequence analysis. Pairwise nucleotide alignment revealed that strain L66 shared 99% sequence identity with Lactiplantibacillus plantarum, while strain K7-1 exhibited 99% sequence identity with Priestia aryabhattai. These results indicate a high degree of nucleotide conservation between the selected isolates and their closest reference strains. Phylogenetic analysis of the 16S rRNA gene sequences further supported the sequence similarity results. As shown in Fig. 3, strain L66 clustered with the type strain of L. plantarum, whereas strain K7-1 grouped closely with the type strain of P. aryabhattai. Because 16S rRNA gene sequencing has limited discriminatory power among closely related species within the L. plantarum complex, strain L66 was conservatively identified at the species level as L. plantarum. Both isolates were clearly separated from closely related reference taxa, confirming their distinct phylogenetic positions.

Fig. 3.

Phylogenetic tree based on 16S rRNA gene sequences showing the taxonomic positions of strains L66 (A) and K7-1 (B) among related reference strains.

3.3. Optimization of culture conditions for β-glucosidase production

To enhance enzyme productivity, the effects of basal medium type, incubation temperature, and initial pH were systematically evaluated for strains L66 and K7-1. Carbon source composition and nitrogen availability are critical determinants of β-glucosidase expression, as polysaccharides and oligosaccharides frequently act as metabolic inducers (Garcia et al., 2015; Neesa et al., 2020).

For strain L66, MRSG supported the highest enzyme production (Fig. 4A), whereas for strain K7-1, maximal activity was observed in NBS medium (Fig. 4B). These results indicate strain-specific substrate utilization patterns and regulatory responses, consistent with previous reports on heterogeneous induction mechanisms among microbial β-glucosidases (Garcia et al., 2015). Similar induction effects of starch-based substrates have been reported in solid-state and submerged fermentations of β-glucosidase-producing microorganisms (Neesa et al., 2020).

Fig. 4.

Effects of basal media on β-glucosidase production by strains L66 (A) and K7-1 (B).

Incubation temperature and initial pH also significantly influenced enzyme production (Fig. 5). Both strains exhibited maximal production at 30-35°C (Fig. 5A and B), which is consistent with optimal enzyme secretion observed in mesophilic β-glucosidase producers (Yan et al., 2016; Neesa et al., 2020). With respect to initial pH, strain L66 showed an optimal range under mildly acidic to neutral conditions (Fig. 5C), whereas K7-1 maintained relatively high productivity across a broader pH range (Fig. 5D). Such strain-dependent pH responses have been attributed to differences in regulatory networks, membrane transport systems, and extracellular enzyme stability (Michlmayr et al., 2010; Li et al., 2025).

Fig. 5.

Effects of incubation temperature and initial pH on β-glucosidase production by strains L66 and K7-1. A, L66 (temperature); B, K7-1 (temperature); C, L66 (initial pH); D, K7-1 (initial pH).

3.4. Physicochemical characterization of crude β-glucosidase

The biochemical properties of crude β-glucosidase preparations were evaluated by determining temperature and pH optima, as well as thermal and pH stabilities. Both enzymes exhibited maximal catalytic activity at approximately 45-50°C (Fig. 6A and B), which is consistent with reported optima of microbial β-glucosidases used for glycoside hydrolysis and biotransformation processes (Lee et al., 2012; Yan et al., 2016; Li et al., 2025). The high β- glucosidase activity and favorable catalytic properties observed for strain L66 are in agreement with previous reports highlighting the important role of L. plantarum in food fermentation and functional compound biotransformation (Paventi et al., 2025). Notably, enzyme activity from strain K7-1 declined sharply at elevated temperatures, whereas L66-derived enzyme retained higher activity over a wider temperature range, suggesting superior structural stability of the L66 enzyme. Similar strain-dependent thermal tolerance has been reported for crude β-glucosidase preparations (Michlmayr et al., 2010; Garcia et al., 2015). Both enzymes showed highest activities near neutral to slightly alkaline pH (pH 7.5; Fig. 6C and D), consistent with buffer-dependent modulation of β-glucosidase activity through ionization changes in catalytic residues and substrate-binding microenvironments (Yan et al., 2016; Strahsburger et al., 2017). Stability assays further distinguished the two preparations: the L66 enzyme retained substantial residual activity after moderate thermal pretreatment, whereas the K7-1 enzyme exhibited a narrower thermal stability range before rapid inactivation (Fig. 7A and B). Both enzymes maintained appreciable activity over a broad pH range, although stability decreased under strongly alkaline conditions (Fig. 7C and D), which is relevant for prolonged bioconversion processes in which pH drift can occur (Garcia et al., 2015; Yan et al., 2016). Comparable thermostable and pH-tolerant β-glucosidases with industrial potential have been reported in metagenomic and environmental studies, supporting the practical relevance of such physicochemical properties (Jeilu et al., 2024).

Fig. 6.

Effects of reaction temperature and pH on β-glucosidase activity from strains L66 and K7-1. A, L66 (temperature); B, K7-1 (temperature); C, L66 (pH); D, K7-1 (pH).

Fig. 7.

Thermal and pH stabilities of β-glucosidase from strains L66 and K7-1. A, L66 (thermal stability); B, K7-1 (thermal stability); C, L66 (pH stability); D, K7-1 (pH stability).

These enzymatic characteristics are directly relevant to ginsenoside biotransformation because compound K is produced via sequential deglycosylation of major PPD-type ginsenosides (Lee et al., 2012; Shin et al., 2015), and enzymes with broad operational stability can enhance conversion efficiency in crude enzyme systems (Zhang et al., 2019; Sharma and Lee, 2020). Accordingly, the crude β-glucosidase preparations from strains L66 and K7-1 represent practical biocatalyst candidates for compound K production with operational simplicity compared with purified or recombinant enzyme systems.

3.5. TLC analysis of ginsenoside biotransformation

Time-dependent changes in ginsenoside profiles during biotransformation by strains L66 and K7-1 were analyzed using TLC (Fig. 8). In both strains, gradual reduction of major PPD-type ginsenosides (Rb1, Rb2, and Rc) was accompanied by a progressive increase in the compound K band, indicating sequential deglycosylation.

Fig. 8.

Time-course TLC profiles of ginsenoside biotransformation by strains L66 (A) and K7-1 (B). Compound K was identified by comparison with authentic standards. ST, authentic standards; CK, compound K.

In strain L66 (Fig. 8A), the compound K band intensified markedly by day 7, whereas bands corresponding to precursor ginsenosides, particularly Rc and Rb2, were substantially weakened, and intermediate metabolites such as Rd and F2 decreased at later time points. This pattern suggests that strain L66 efficiently catalyzed both intermediate and terminal hydrolysis steps leading to compound K accumulation. In contrast, strain K7-1 (Fig. 8B) also exhibited time-dependent compound K formation; however, relatively stronger residual bands corresponding to Rc, Rb2, and Rd were observed even after 7 days. This persistence indicates that, although K7-1 possesses β-glucosidase activity toward major ginsenosides, its conversion efficiency for late-stage steps appears lower than that of L66 under the tested conditions.

The observed biotransformation patterns are consistent with the well-established sequential deglycosylation pathway of PPD-type ginsenosides, in which Rb1, Rb2, and Rc are converted to Rd and F2, followed by the formation of compound K (Chi and Ji, 2005; Chi et al., 2005; Quan et al., 2013; Upadhyaya et al., 2016). Similar TLC profiles, characterized by progressive disappearance of precursor bands and accumulation of compound K, have been reported for efficient compound K-producing food-derived and intestinal microorganisms (Quan et al., 2013; Upadhyaya et al., 2016). Notably, the progressive reduction of intermediate metabolites in strain L66 suggests the presence of a highly coordinated β-glucosidase system capable of catalyzing both intermediate and terminal hydrolysis reactions. Such a “complete-type” deglycosylation system has been associated with high compound K productivity in previous studies using Lactobacillus, Bifidobacterium, and Paecilomyces strains (Yang et al., 2015; Kim et al., 2020).

From a physiological perspective, compound K is known to be generated predominantly through intestinal microbial metabolism and is rarely detected in the absence of bacterial hydrolysis (Akao et al., 1998; Deng et al., 2024). The TLC pattern observed for strain L66 closely resembles that of highly active intestinal bacteria, in which limited accumulation of intermediates and preferential compound K formation are observed. Therefore, strain L66 may function as an effective in vitro model for intestinal-type ginsenoside metabolism.

Collectively, these results demonstrate clear strain-dependent differences in ginsenoside biotransformation efficiency, with strain L66 exhibiting more efficient precursor conversion and dominant compound K accumulation than strain K7-1.