Engineering of Cyclodextrin Glucosyltransferase from Paenibacillus macerans for Improved Regioselectivity and Product Specificity Toward Hydroxyflavone Glycosylation

Abstract

The regioselective glycosylation and product specificity of hydroxyflavonoid compounds profoundly influences their biological activity and stability, offering significant therapeutic potential. However, most cyclodextrin glucosyltransferases (CGTases) inherently lack regioselectivity and product specificity for flavone glycosylation. Herein, a CGTase from Paenibacillus macerans was engineered for enhanced glycosylation regioselectivity and product specificity by combining molecular docking analysis and saturation mutagenesis strategies. K232L (favoring 4′-and 6-hydroxyflavones) and K232V (favoring 7-hydroxyflavone) were identified with distinct preferences. In addition, H233Y (preferring for 4′-hydroxyflavones), H233T (preferring for 6′-hydroxyflavones), and H233K (preferring for 7′-hydroxyflavones) also demonstrated distinct regioselectivity. These variants further exhibited enhanced hydrolytic activity, enabling the efficient production of short sugar-chain glycosides. Molecular dynamics (MDs) simulations revealed that the variants adopted optimized catalytic conformations with increased loop region flexibility near the binding pocket, enhancing substrate accessibility. These findings underscore the pivotal roles of K232 and H233 in broadening the substrate scope of CGTase and offer valuable guidance for enzyme engineering targeting regioselective glycosylation.

1. Introduction

Hydroxyflavones, a class of naturally occurring flavonoid compounds, are renowned for their potent bioactive properties, including anti-oxidant, anti-inflammatory, and anticancer activities [1,2,3,4]. Most hydroxyflavones are abundant in fruits, vegetables, and medicinal herbs. They are critical in combating the cellular damage associated with aging and chronic diseases due to their efficacy in scavenging free radicals and mitigating oxidative stress [5,6]. The pharmacological profiles of hydroxyflavones can be finely tuned through glycosylation at specific hydroxyl positions. For instance, glycosylation at the 3-position of the C-ring is able to attenuate antitumor activity, whereas methoxylation or glycosylation at the A-ring can significantly augment cytotoxicity [7,8,9]. These dual effects underscore the therapeutic potential of hydroxyflavones and highlight the necessity for the precise regulation of glycosylated regioselectivity.

Chemical glycosylation methods face significant challenges (e.g., harsh reaction conditions, poor regioselectivity, and complex byproduct formation), which hinder the efficient modification of sensitive compounds (like flavonoids) [10]. For instance, although baicalein can be glycosylated by chemical approaches to produce baicalein-6-α-glucoside, these methods typically require extensive purification steps [11]. Similarly, achieving improved solubility for quercetin-3-O-glycosides through chemical glycosylation is often resource-intensive and inefficient [12]. In contrast, enzymatic glycosylation offers a sustainable alternative, operating under mild conditions with superior regioselectivity and simplified downstream processing. Therefore, enzymatic glycosylation approaches are advantageous for flavonoid modification [13,14].

Cyclodextrin glucosyltransferase (CGTase, EC 2.4.1.19), a member of the glycosyl hydrolase family 13 (GH13), has emerged as a versatile enzyme with potential in transglycosylation, cyclization, and hydrolysis [15,16]. Its adaptability has been demonstrated in glycosylating smaller bioactive molecules, such as hydroxyflavones and daidzein-7-O-glucoside, highlighting its applicability beyond its natural cyclodextrin substrates [17,18]. The substrate-binding pocket of CGTase plays a critical role in determining catalytic efficiency and regioselectivity, with key residues such as Y89, F195, and N193 identified as essential for substrate interactions [19,20,21]. However, CGTase exhibits limited regioselectivity for non-cyclodextrin substrates, particularly at specific hydroxyl groups. For instance, when CGTase was used for the glycosylation of steviol and p-nitrophenyl, product confounding occurred due to the lack of glycosylation regioselectivity [22,23]. To overcome these limitations, strategies involving systematic mutagenesis and rational protein design have been employed to enhance the regioselectivity of CGTase. By fine-tuning the substrate-binding region, researchers have improved the ability to glycosylate flavonoids, enabling the production of derivatives with enhanced stability, bioavailability, and therapeutic potential. These advances establish CGTase as a valuable tool for carbohydrate glycosylation, particularly in pharmaceutical and nutraceutical applications [24,25,26,27].

Our previous studies have demonstrated the flavonoid glycosylation potential of CGTase from Paenibacillus macerans (PmCGTase), but with limited regioselectivity and product specificity [28,29]. To address this, we engineered PmCGTase for enhanced glycosylation regioselectivity and product specificity, and explored its mechanism in this study. Here, structural modifications were designed to enhance the regioselectivity of PmCGTase for three monohydroxy flavones (4′-, 6-, and 7-hydroxyflavone). Molecular docking identified key residues in the substrate-binding pocket, and alanine scanning of 10 candidate residues pinpointed K232 and H233 as critical determinants of regioselectivity and activity. Saturation mutagenesis subsequently yielded variants (K232L and K232V) which exhibited distinct regioselectivity and activity profiles to hydroxyflavones. Molecular dynamics (MDs) simulations further elucidated the structural and catalytic adaptations of these variants, providing mechanistic insights into their enhanced performance. These findings offer valuable insights for the engineering of CGTase to improve its glycosylation regioselective and product specificity, thereby streamlining the purification process and laying the groundwork for the industrial production of hydroxyl compounds glycosylation.

2. Results and Discussion

2.1. Identification of Key Residues for Regioselective Glycosylation of PmCGTase

To investigate the regioselectivity of PmCGTase in hydroxyflavone glycosylation, three monohydroxy flavones, including 6-hydroxyflavone (6-HF), 7-hydroxyflavone (7-HF), and 4′-hydroxyflavone (4′-HF), were selected as substrates. Initial enzymatic assays revealed that the wild type (WT) exhibited low catalytic efficiency toward all three substrates and lacked a clear substrate preference, indicating limited regioselectivity (Figure 1).

To identify the residues responsible for substrate binding and regioselectivity, molecular docking simulations were performed with three hydroxyflavones (Figure S1). For 6-HF, K232 formed hydrogen bonds (1.86 Å) and π-alkyl interactions (5.36 Å), while H233 contributed π-π interactions (5.12 Å and 4.16 Å). With 7-HF, K232 established π-cation interactions (2.81 Å), and H233 maintained π-π interactions (5.26 Å). In the case of 4′-HF, only H233 interacted with the substrate via π-π interactions (5.15 Å). Additional interactions were observed with other residues, as follows: Y100 exhibited π-π T-shaped interactions with 6-HF and 7-HF, while D372 formed hydrogen bonds with 4′-HF, and Y260 contributed hydrogen bonding with 6-HF. These findings suggest that the position of the hydroxyl group in the flavone scaffold strongly influences the catalytic contributions of these residues, highlighting the potential role of K232, H233, and others in determining regioselectivity.

To validate the importance of these residues, alanine scanning was conducted for all interaction sites (e.g., Y100, F183, D229, K232, H233, E258, Y260, D329, D372, andR376) identified in the molecular docking analysis. Most residues lost catalytic activity upon mutation to alanine, indicating their essential role in substrate binding and catalysis. However, K232A and H233A exhibited enhanced catalytic activity compared with WT (Table S2). K232A achieved a conversion of 17.6% for 4′-HF and 18.5% for 6-HF, representing increments of 19.9% and 16.1% compared with WT. Similarly, H233A showed a conversion of 10.6% for 4′-HF and 24.8% for 6-HF, corresponding to increases of 2.9% and 22.4%, respectively. These findings suggest that K232 and H233 not only influence substrate recognition but also play an important role in catalytic conformations that enhance specific glycosylation reactions. While other residues, such as F183 and R376, retained some activity after alanine substitution, their impact was less pronounced. Consequently, K232 and H233 were selected for saturation mutagenesis to explore their full potential in modulating regioselectivity and catalytic efficiency.

2.2. Regioselectivity Modulation via Saturation Mutagenesis of K232 and H233

The saturation mutagenesis of K232 and H233 revealed distinct regioselective preferences and substantial improvements in catalytic efficiency. K232L exhibited strong preferences for glycosylation at the 4′- and 6-hydroxyl positions, achieving conversions of 36.1% and 33.6%, respectively. These conversions correspond to 4.7-fold and 13.9-fold increases compared with WT (Figure 2a). On the other hand, K232V showed a significant preference for glycosylation at the 7-hydroxyl position of 7-HF, with a conversion of 34.6%, representing an 8.3-fold increase over WT (Figure 2a). Similarly, H233 displayed substrate specific regioselectivity. H233K preferentially catalyzed glycosylation at the 7-hydroxyl position of 7-HF with a conversion of 18.7%, a 4.5-fold increase compared with WT. H233Y and H233T, in contrast, favored glycosylation at the 4′-and 6-positions, achieving conversions of 15.7% and 12.4%, representing 2.1-fold and 5-fold increases, respectively (Figure 2b). These results demonstrate that both K232 and H233 are pivotal in determining the regioselectivity of CGTase, with mutations at these residues leading to significant shifts in substrate orientation and catalytic outcomes.

2.3. Enhanced Specificity for Short-Chain Glycoside Products by K232 and H233 Variants

To further characterize the transglycosylation products, LC-MS analysis was performed to confirm glycosylation at the hydroxyl sites of the substrates. The reactions resulted in glycosylated products containing one or more glucose moieties attached to the substrate, forming short- or long-chain glycosides (Figure 1 and Figure S2). For instance, when using K232L to catalyze the glycosylation of 4′-HF, four distinct glycoside products were identified, each representing a different degree of glycosylation with one to four glucose moieties linked. Similarly, K232L catalyzed the glycosylation of 6-HF, yielding four glycoside products of varying sugar chain lengths. In contrast, K232V catalyzed the glycosylation of 7-HF, producing two glycoside products with shorter sugar chains (Figure 1). The LC-MS spectra revealed m/z corresponding to the expected glycoside products, further confirming the regioselectivity and glycosylation efficiency of the variants.

Compared with WT, which exhibited moderate specificities for short-chain glycoside products (SCGP) (48% for 4′-HF, 43% for 6-HF, and 58% for 7-HF), K232 and H233 demonstrated significantly enhanced SCGP specificity. K232L achieved SCGP specificities of 93% and 85% when catalyzing 4′-HF and 6-HF, respectively. Similarly, K232V exhibited 92% specificity for SCGP with 7-HF. H233 variants also displayed remarkable SCGP specificity, as follows: H233Y achieved 97% specificity with 4′-HF, H233T showed 91% specificity with 6-HF, and H233K demonstrated 96% specificity with 7-HF (Figure 3). In addition, compared with the WT, the overall conversions of variants were significantly elevated correlating with an increase in SCGP specificity (Figure 3). This indicates that a high SCGP specificity is favorable for enhancing the overall glycosylation conversion. Typically, the flavone with a short-chain glycoside is considered to be more stable than that with a long-chain glycoside [29].

These results underscore the critical roles of K232 and H233 in enhancing the substrate selectivity and product specificity in CGTase-catalyzed glycosylation reactions. By favoring the production of SCGP and overall conversion, these engineered variants provide valuable tools for the efficient and precise modification of flavonoid substrates, offering significant potential for industrial applications.

2.4. Enhanced Hydrolytic Activity and Altered Functional Profiles of K232 and H233 Variants

As shown in

Table 1. Analysis of catalytic activities of PmCGTase WT and variants.

Enzymes	Relative Activity (%)
	Cyclization	Hydrolytic	Disproportionation
WT	100	100	100
K232L	79 ± 2.1	147 ± 3.2	105 ± 1.8
K232V	75 ± 1.2	172 ± 2.3	96 ± 2.4
H233K	83 ± 2.3	144 ± 1.5	98 ± 2.9
H233T	89 ± 3.4	130 ± 2.3	164 ± 2.4
H233Y	94 ± 2.3	151 ± 2.0	153 ± 1.4

Each value represents the mean of three independent measurements, with deviations from the mean being less than 5%. The relative activity of WT for cyclization, hydrolytic, and disproportionation activities is defined as 100%.

, all variants exhibited higher hydrolytic activities and lower cyclodextrin activities compared with the WT. Specifically, the K232V variant demonstrated a 72% increase in hydrolytic activity and a 25% decrease in cyclodextrin activity. In addition, the H233T, H233Y, and K232L variants showed enhanced disproportionation activities, whereas the K232V and H233K variants displayed disproportionation activities similar to those of the WT.

2.5. MD Simulations Analysis Revealing the Improved Properties of Variants

To investigate the molecular mechanisms underlying the enhanced regioselectivity of PmCGTase variants, the crystal structure of CGTase from P. macerans (PDB: 4JCL) was utilized for structural analysis. Then, 50 ns molecular dynamics (MDs) simulations were performed using monohydroxy flavones and maltose as ligands. RMSD analysis confirmed that the systems reached equilibrium after 20 ns, and average conformations of the WT and variants in the stable catalytic state (20–50 ns) were extracted for further analysis (Figure S3).

MD simulations revealed structural adaptations that enhanced regioselectivity. The catalytic mechanism of CGTase was mediated by three conserved residues (D229, E258, and D329) characteristic of the α-amylase family.

Specifically, E258 acted as an acid catalyst by donating protons during catalysis, D229 stabilized the reaction intermediate as a nucleophile, and D329 formed hydrogen bonds with the hydroxyl groups of substrates to stabilize the transition state [15]. In K232L and K232V, shortened catalytic distances between E258 and the hydroxyl oxygens of flavonoid substrates were observed, improving substrate activation. For K232L, the catalytic distances were 3.3 Å and 2.9 Å for 4′-HF and 6-HF, respectively, while K232V showed a catalytic distance of 3.3 Å for 7-HF. These reduced distances facilitated more effective substrate activation, directly correlating with the observed regioselectivity and enhanced catalytic efficiency (Figure 4).

The variants exhibited improved residue interactions with the substrate, contributing to enhanced stability during catalysis. For K232L, the π-π interactions between Y195 and the flavonoid acceptor were strengthened, stabilizing the substrate and supporting regioselective glycosylation (Figure 5). This is consistent with the previous studies identifying Y195 as a critical residue influencing hydrolytic activity and cyclization specificity in α-CGTase [30]. Additionally, K232L displayed π-π interactions between A230 and the substrate, with side-chain interactions influencing the cyclodextrin specificity [31]. The enhanced interactions likely contributed to the increased conversion of K232L and K232V, reinforcing the functional significance of these residues in substrate stabilization and regioselectivity.

The dynamic loop region (W259–T266) emerged as a key determinant of regioselectivity. Mutations at K232 disrupted interactions within the loop, particularly between residue 232 and W259, resulting in increased loop flexibility (Figure S4). The improved flexibility of this loop enlarges the substrate-binding pocket and facilitates better accommodation of substrates at the +1 binding site, which may be critical for regioselective glycosylation. Notably, Y260 exhibited the largest fluctuation with 4′-HF (1.8 Å), while D264 showed increased flexibility with 6-HF (0.5 Å). These loop dynamics directed substrates into specific catalytic orientations, enabling precise regioselective glycosylation (Figure 6).

Although the cyclization activity of K232L and K232V decreased, their hydrolytic activity increased by 47% and 72%, respectively (Table 1). This enhanced hydrolytic activity promoted the formation of short-chain glycosides, aligning with the regioselective tendencies of these variants. The interplay between loop flexibility, catalytic distance, and substrate interactions provides a mechanistic explanation for their enhanced regioselectivity and catalytic efficiency.

3. Materials and Methods

3.1. Strains and Chemicals

Hydroxyflavones were purchased from Titan Company (Shanghai, China), and soluble starch was obtained from Aladdin (Shanghai, China). HPLC-grade acetonitrile was supplied by Titan Company (Shanghai, China), while all other reagents were procured from local suppliers.

3.2. Saturation Mutagenesis

Saturation mutagenesis was performed using pET-28a(+) CGT (preserved in our laboratory) as a template. Polymerase chain reactions (PCRs) were conducted with KOD DNA polymerase, using primers for alanine scanning (Table S1) and saturation mutagenesis (Tables S3 and S4). The amplified PCR products were digested with Dpn I and transformed into E. coli BL21(DE3) (TakaRa, Dalian, China). Positive clones were screened by colony PCR and verified through DNA sequencing.

3.3. Expression and Purification of PmCGTase and Its Mutants

WT and mutants were inoculated into Luria-Bertani (LB) medium containing 0.05 mg/mL kanamycin and incubated at 37 °C with shaking at 180 rpm for 10 h. Cultures were then transferred to Terrific Broth (TB) medium (yeast extract: 24 g/L, tryptone: 12 g/L, potassium dihydrogen phosphate: 16.43 g/L, dipotassium hydrogen phosphate: 2.314 g/L, glycerol: 5 g/L) supplemented with 0.05 mg/mL kanamycin, using a 4% inoculum. Cultures were grown at 30 °C with shaking at 120 rpm until the OD600 reached 0.6–0.8, at which point isopropyl β-D-1-thiogalactopyranoside (IPTG) was added to a final concentration of 0.2 mM for protein induction. The cultures were further incubated at 16 °C with shaking at 120 rpm for 16 h.

After incubation, cells were harvested by centrifugation at 8000 rpm for 5 min. The pellets were resuspended in PBS buffer, disrupted by ultrasonication, and the supernatant containing soluble proteins was collected by centrifugation at 4 °C and 8000 rpm. Proteins were purified using nickel affinity chromatography, and their purity was confirmed by SDS-PAGE. Protein concentrations were determined using the NanoDrop method.

3.4. Enzyme Activity Assays

The cyclization activity of CGTase was measured using the methyl orange method [32]. Briefly, 100 μL of appropriately diluted enzyme solution was mixed with 100 μL of PBS (50 mM, pH of 6.0), followed by the addition of 50 μL of maltodextrin solution (40 g/L, prepared in PBS, pH of 6.0). The mixture was incubated at 40 °C for 10 min, after which 250 μL of HCl (1 M) was added to terminate the reaction. Then, 150 μL of methyl orange solution (0.5 mM, prepared in PBS, pH of 6.0) was added, and the mixture was thoroughly mixed and allowed to stand at room temperature for 20 min. Absorbance was measured at 505 nm, with a blank control lacking enzyme solution. One unit of enzyme activity was defined as the amount of enzyme required to produce 1 μmol of α-cyclodextrin per minute.

The transglycosylation activity was determined using 4,6-ethylidene-para-nitrophenyl-α-D-maltoheptaoside (EPS) as a substrate [16]. A mixture of 300 μL of EPS (4 mM) and 300 μL of maltose (0.2 M) was preheated at 50 °C for 10 min before adding 100 μL of enzyme solution. The reaction was stopped by adding 50 μL of HCl (3 M), followed by neutralization with 50 μL of NaOH (3 M). Afterward, 100 μL of α-glucosidase was added, and the reaction was incubated at 60 °C for 1 h. Finally, 100 μL of Na2CO3 (1 M) was used to adjust the pH to 8.0. The product, paranitrophenol, was detected at 400 nm. One unit of enzyme activity was defined as the amount of enzyme required to convert 1 μmol of EPS per minute.

The hydrolytic activity was measured using the dinitrosalicylic acid (DNS) method [16]. A mixture of 50 μL of enzyme solution and 200 μL of soluble starch (10 g/L, prepared in sodium acetate buffer, pH of 5.5) was incubated at 50 °C for 10 min. The reaction was stopped by adding 500 μL of DNS solution, followed by heating at 100 °C for 7 min. After cooling and dilution, the absorbance was measured at 540 nm. A blank control without enzyme solution was prepared. One unit of activity was defined as the amount of enzyme required to produce 1 μmol of reducing sugar per minute.

The conversion was calculated based on substrate consumption using the following formula:

$$\mathrm{conversion}={\displaystyle \frac{\mathrm{Mi}-\mathrm{Mr}}{\mathrm{Mi}}}\times 100\%$$

In this formula, Mi represents the initial concentration (mM) of the monohydroxy flavone substrate and Mr represents the concentration (mM) of the remaining substrate.

3.5. Analysis and Identification of Hydroxyflavone Glycosylated Products

The transglycosylation reaction was conducted in a 1 mL system containing 200 μL of hydroxyflavone (0.2 mg/mL, dissolved in DMSO), 600 μL of soluble starch (40 mg/mL, prepared in PBS, pH of 6.0), and 200 μL of enzyme solution. Reactions were carried out at 40 °C with shaking at 120 rpm for 24 h. Products were analyzed using HPLC equipped with a Diamomisil C18 column (250 × 4.6 mm, 5 μm). A gradient elution method was employed, with solution A (0.1% phosphoric acid in water, v/v) and solution B (acetonitrile) as the mobile phases. The flow rate was 0.8 mL/min, and the concentration of solution B increased from 15% to 85% over 15 min. Detection was performed at 254 nm, and conversion rates were calculated based on the decrease in substrate peak area.

The molecular weights were analyzed using a WATERS MALDI SYNAPT Q-TOF MS in ESI-mode. Chromatographic analysis was conducted on a WATERS ACQUITY UPLC with a BEH C18 column (2.1 × 150 mm, 1.7 µm). A gradient elution method was employed, utilizing solution A (0.1% phosphoric acid in water, v/v) and solution B (acetonitrile) as the mobile phases. The analysis conditions were as follows: 0–40 min (100% A), 41–45 min (30% B), 46–50 min (80% B), 51–55 min (100% B), 40–41 min (10% B), 41–50 min (10% B), and 51–55 min (100% A). The injection volume was 5 μL and the flow rate was 0.3 mL/min, with a detection wavelength of 200–400 nm and a column temperature of 45 °C. The mass spectrometer scanned a range from m/z 20 to 2000.

3.6. Molecular Docking and MD Simulation

The crystal structure of CGTase from P. macerans (PDB ID: 4JCL, with 100% homology to PmCGTase in this study) was employed as the template for molecular docking and variant models. Molecular docking and MD simulations were performed using GROMACS 2019.6 with the Amber ff19SB force field. Protein–ligand complexes were prepared with AmberTools2023, and systems were solvated using TIP3P water molecules. The systems were minimized using 20,000 steps of the steepest descent followed by 10,000 steps of the conjugate gradient. Equilibration was carried out under NPT conditions at 303 K and 1 atm. Production simulations were conducted for 50 ns with trajectory files written every 1 ps. Covalent bonds involving hydrogen were constrained, and a time step of 2 fs was used.

4. Conclusions

In this study, the structural and functional mechanisms underlying the enhanced regioselectivity of CGTase variants engineered at K232 and H233 were elucidated. Herein, K232L and K232V were demonstrated to improve regioselectivity and catalytic efficiency by shortening catalytic distances, optimizing substrate positioning, and strengthening π-π interactions with key residues such as Y195 and A230. Similarly, H233 variants exhibited various regioselectivity, with H233Y and H233T preferentially catalyzing 4′-HF and 6-HF, and H233K favoring 7-HF. These structural adaptations were further supported by increased loop flexibility in the dynamic region W259–T266, which facilitated substrate accommodation at the +1 binding site and improved catalytic orientations. In addition, the variants of K232 and H233 also displayed various product specificity for the adherent sugar chain. K232L, K232V, and H233 exhibited SCGP specificities exceeding 85%, compared with the moderate specificities observed in WT. Moreover, the hydrolytic activity of the variants was significantly increased, promoting the efficient formation of SCGPs while reducing cyclization activity. Finally, MD simulations further revealed the mechanism of improved regioselective glycosylation for variants.

In this study, the groundwork is laid for the design of CGTase variants with tailored regioselectivity and product specificity, enabling the efficient synthesis of functional glycosides for pharmaceutical and nutraceutical applications. Moreover, it holds promise for enhancing the product purity in the CGTase-catalyzed glycosylation of flavonoid, steviol, and various other polyhydroxy compounds, thereby simplifying the purification process and paving the way for industrial production.