Continuous Synthesis of Polydatin by Dual Enzyme Coupling Reaction and Its Kinetic Study in Microreactors

Abstract

Resveratrol is a promising ingredient in functional food products, but its low bioavailability and solubility hinder its application. Polydatin, a 3-OH glycosylation of resveratrol, has been shown to exhibit enhanced bioavailability and more favorable physicochemical properties. In this work, a continuous-flow microreactor was developed to synthesize polydatin using resveratrol and a mutant glycosyltransferase (UGT_BS) as the substrate and biocatalyst, respectively. All reaction orders were determined to elucidate the possible reaction mechanism. To further reduce the process costs, a dual enzyme coupling reaction system was developed to enable the in situ regeneration of UDP-Glc.

1. Introduction

Stilbenes represent a group of phytochemicals ubiquitously found in nature, known for their multifaceted therapeutic potentials ranging from antibacterial to antitumor capabilities [1,2]. Among them, resveratrol (1) is the most extensively studied, exhibiting these broad pharmacological activities alongside notable cardioprotective and neuroprotective effects [3,4]. However, the clinical translation of 1 is hampered by its poor bioavailability, which is attributed to low water solubility, chemical instability (e.g., photo-isomerization and oxidation), and rapid metabolism [5]. Polydatin (resveratrol-3-O-β-D-glucoside (2), a natural glycoside of resveratrol, exhibits improved bioavailability and solubility compared to its aglycone. Consequently, it has demonstrated potent therapeutic efficacy in treating metabolic diseases [6], notably including the alleviation of non-alcoholic fatty liver disease [7] and significant antioxidative activity [8].

Further, 2 is primarily obtained by extraction from Polygonum cuspidatum, where it is a major bioactive constituent [9]. However, these extraction processes suffer from high solvent consumption, complex purification steps, and the limited availability of plant resources, which hinder their industrial application [10,11,12]. The glycosylation of 1 with biocatalysts is an appealing research goal for producing 2. However, the technical challenge of this glycosylation is controlling the regioselectivity, as 1 consists of three phenolic hydroxyl groups (3-OH, 5-OH, and 4′-OH), which are located in a similar chemical environment (Scheme 1). Over the past decades, considerable efforts have been devoted to developing diverse biocatalysts capable of selectively converting 1 into 2 [13,14].

In our previous work, a UDP-dependent glycosyltransferase from Bacillus subtilis strain 168 (UGTBS) was identified as a promising catalyst for the glycosylation of 1. To further enhance its catalytic performance, directed evolution was applied, resulting in a triple mutant (Y14I/I62G/M315W) that exhibited markedly improved 3-OH glycosylation activity toward 1 and afforded 2 in 91% yield [15]. However, its productivity remains limited, making it difficult to meet market demand. To overcome this bottleneck and enhance productivity, new technologies are required to intensify the process.

Recently, microfluidics has become a popular research area due to its wide range of applications, such as chemical transformations [16], biocatalytic processes [17], nanomaterials [18], and drug delivery [19]. The transitions from conventional stirred-reactor vessels to continuous-flow microreactors show significant improvements in mass and heat transfer, mixing efficiency, safety and reproducibility [20,21]. In the pharmaceutical industry, an increasing number of small-molecule-based active pharmaceutical ingredients have shown higher synthetic efficiency in microreactors [22,23,24,25,26], such as antitumor drugs [27,28,29], antibiotics [30,31], anti-HIV medications [32,33,34] etc. On the other hand, microfluidics, with tight residence-time control, fast mixing, and intensified heat/mass transfer at low reagent usage, has proven effective for measuring reaction kinetics [35]. Determination of intrinsic kinetics can reveal the rate-determining step, which not only helps chemists understand the reaction mechanism more profoundly but also benefits chemical engineers in identifying optimum reactor configurations and process conditions [36].

To intensify the process and gain deeper insight into the underlying kinetic behavior, we leveraged the advantages of microreactor technology. A continuous-flow microreactor platform was constructed to quantitatively investigate the kinetics of the glycosylation reaction. The apparent kinetic parameters, including the reaction orders with respect to each reactant, were first determined. Furthermore, a dual-enzyme cascade was established within the microreactor for the production of 2, followed by systematic optimization of the key reaction parameters.

2. Materials and Methods

2.1. Chemicals

1, uridine 5-diphosphate disodium salt (UDP), UDP-Glc, 3,5-dinitrosalicylic acid (DNS), authentic standard of 2, sucrose, and methanol were obtained from Macklin (Shanghai, China). PrimeSTAR HS DNA polymerase and Dpn I restriction endonuclease were supplied by Takara Bio Inc. (Dalian, China). We used Origin 2024 to draw the figures and perform data fitting.

Solution A (5 mL): 50 mM Tris-HCl + 2 mM Resveratrol + 1 mM UDP + 0.8 mM sucrose + 10% (V/V) DMSO. Solution B (5 mL): Glycosyltransferase (150 mU/mL) + sucrose synthase AtSuSy (200 mU/mL) solution. Solution C: methanol. As the reactant concentrations were occasionally adjusted during the experiments, the ratio described above represents only one of the suitable conditions. The reaction mixture was maintained at approximately pH 8.0.

The glycosyltransferase used in this study is the triple mutant Y14I/I62G/M315W, which was derived from the wild-type UGTBS gene (GenBank accession no. KU500621.1) originally cloned from Bacillus subtilis 168. The mutant was constructed via site-directed mutagenesis and expressed in Escherichia coli BL21 (DE3), as described in our previous work [15].

The sucrose synthase AtSuSy (GenBank accession no. NM_001036838), used for the dual-enzyme coupling system, originates from Arabidopsis thaliana and was expressed in E. coli BL21 (DE3). Both recombinant strains were constructed and stored in our laboratory.

2.2. Process Optimization and Kinetics Research of Continuous Synthesis of 2 in Microreactor

The microreactor system for the synthesis of 2 is shown schematically in Figure 1. Two syringe pumps (LSP02-2A, Longer, Baoding, China) with a repeat precision of ≤±1% were used to continuously inject solutions A and B into the microreactor system. The two solutions were heated (or cooled) to a pre-set temperature (15–35 °C) in a water bath before being mixed in the first micro-mixer (IDEX, 0.5 mm bore, PEEK). The extension tube attached to the mixer was used to conduct the glycosylation reaction; the residence time was controlled by changing the length of the reaction tube or the volumetric flow rate of solutions A and B. The second micro-mixer was connected with the reacting tubes, and piston pump (JJRZ-01020 S, Hangzhou Jingjin, Hangzhou, China) was used to pump methanol to quench the glycosylation reaction. The effluent was collected after the microreactor system had been operated for 3–5 times the predetermined residence time to ensure steady-state conditions, and samples were obtained by filtration. The outer diameters (OD) of all PEEK capillaries were 1.590 mm, inner diameters (ID) were all 0.500 mm, and the internal volume of 1 m reaction tube was 0.196 mL. The yields of 2 were determined by the HPLC external standard method.

2.3. Sample Analysis

To reduce the potential variability associated with storage, samples were maintained at 2–8 °C and analyzed within 4 h by HPLC. Detailed chromatographic conditions are provided in Table S1. All experiments were performed in triplicate under identical conditions. The relative error was controlled to be less than 3%. The reported values are averages, as shown in Figure 2, Figure 3, Figure 4 and Figure 5. Product 2 and by-product 3 were separated and purified using thin-layer chromatography, and their NMR and MS spectra were provided in the Supplementary Information.

3. Results and Discussion

3.1. Kinetics Research of Continuous Synthesis of 2 in Microreactor

3.1.1. The Effect of Total Flow Rate on the Reaction Performance

To initiate the study, we examined the effect of volumetric flow rate in the microreactor, using capillaries of different lengths (2, 4, 6, 8, and 10 m) to evaluate how mixing efficiency influenced the conversion rate of 1 (x) at a constant residence time of 10 min. As shown in Figure 2, the conversion rate of 1 increased sharply with increasing total flow rate, indicating that the reaction was strongly influenced by mixing efficiency at low flow rates. When the total flow rate exceeded 78.5 μL/min, the yield reached a plateau, suggesting that the system became governed primarily by intrinsic reaction kinetics rather than mass transfer. Accordingly, the volumetric flow rate was maintained above 78.5 μL/min in subsequent experiments.

3.1.2. The Reaction Order of 1

According to classical kinetic theory, the governing rate equations can be expressed as follows:

$${\displaystyle \frac{{dC}_1}{dt}}=-k_1{C_1}^{\alpha }{C_{UDPG}}^{\beta }{C_{UGT}}^{\gamma }-{ k}_2{C_1}^{\alpha }{C_{UDPG}}^{\beta }{C_{UGT}}^{\gamma }$$

(1)

Because sucrose was present in large excess, the UDP-Glc concentration was assumed not to limit the rate. In addition, the glycosyltransferase concentration was constant during the reaction; therefore, Equation (1) can be simplified to:

$${\displaystyle \frac{{dC}_1}{dt}}=K{C_1}^{\alpha }$$

(2)

$$K=-(k_1+k_2){C_{UDPG}}^{\beta }{C_{UGT}}^{\gamma }$$

(3)

where K is a constant. If the order of 1 is first, the rate law is:

$${\displaystyle \frac{{dC}_1}{dt}}=C_{10}{\displaystyle \frac{dx}{dt}}=K{C}_1=K{C}_{10}(1-x)$$

(4)

where $x$ is the conversion rate of 1, $C_{10}$ is the initial concentration of 1. This equation can be solved by integration as:

$$\ln \left(1-x\right)=Kt$$

(5)

Therefore, the linearity of $\ln \left(1-x\right)$ versus t with an intercept at the origin would confirm first-order kinetics. In contrast, for a second-order dependence on 1, Equation (4) applies, and plotting 1/(1 − x) against t should result in a straight line.

$$1/\left(1-x\right)=1-Kt$$

(6)

Figure 3a,b present the plots of $\ln \left(1-x\right)$ versus t and $1/(1-x)$ versus t, respectively.

3.1.3. The Reaction Order of UDP-Glc

To determine the reaction order with respect to UDP-Glc, experiments were performed using a range of initial UDP-Glc concentrations. The initial concentration of 1 was kept sufficiently high so that it did not affect the measured rate. Because the yield was maintained below 10%, UDP-Glc depletion was negligible; consequently, conversion increased linearly with residence time, and the fitted lines passed through the origin. The initial rates were therefore calculated from the slopes in Figure 4a. As shown in Figure 4b, the rate exhibited a strong linear dependence on the initial UDP-Glc concentration (R² = 0.996), indicating first-order kinetics in UDP-Glc (β = 1).

3.1.4. The Reaction Order of Glycosyltransferase

To determine the reaction order with respect to UGT_BS, experiments were conducted over a range of initial UGT_BS concentrations. The initial concentration of 1 was maintained at a sufficiently high level to avoid influencing the observed rate. Because the yield was kept below 20%, enzyme depletion was negligible; accordingly, conversion varied linearly with residence time, and the fitted lines passed through the origin. The initial reaction rates were therefore obtained from the slopes of the plots in Figure 5a. As shown in Figure 5b, a strong linear dependence of the rate on the initial UGT_BS concentration (R² = 0.996) indicates first-order kinetics in UGT_BS (γ = 1). Kinetic analysis indicated that the apparent reaction orders for 1, UDP-Glc, and UGT_BS are all close to unity, which is consistent with an ordered Bi–Bi mechanism. In this mechanism, the substrates bind sequentially to the enzyme to generate a ternary complex prior to glycosyl transfer and product release.

3.2. Optimization of Operating Conditions Under Dual Enzyme Conditions

To reduce the overall process cost, a dual-enzyme coupled system was established (Figure 6). In recent years, two-enzyme and multi-enzyme cascade catalysis has attracted increasing attention for the synthesis of fine chemicals, particularly in continuous-flow formats where in situ cofactor regeneration can be readily implemented for glycoside production. For instance, Liu reported high conversion using co-immobilized enzymes in a packed-bed reactor, although such packed-bed configurations may suffer from mass-transfer limitations and scale-up challenges [37]. In contrast, our microreactor-based glycosyltransferase–sucrose synthase cascade enables continuous polydatin synthesis with enhanced mixing, precise residence-time control, rapid parameter optimization, and straightforward numbering-up scalability, thereby supporting regioselective and efficient continuous production. More broadly, in vitro cascade systems provide a green alternative to conventional chemical routes for manufacturing functional molecules.

In this study, the cascade integrates the UGT_BS triple mutant Y14I/I62G/M315W (Bacillus subtilis 168) with AtSuSy. The two reactions are coupled via UDP-Glc: AtSuSy catalyzes sucrose and UDP to generate fructose and UDP-Glc, and the UGT_BS mutant subsequently uses UDP-Glc to glycosylate resveratrol, producing polydatin. Given the high cost of UDP-Glc, coupling these steps enables UDP-Glc regeneration in situ, substantially lowering the production cost of 2.

3.2.1. The Effect of Reaction Temperature on Glycosylation

Under dual-enzyme coupling conditions, we first evaluated the effect of temperature on the glycosylation performance. As shown in Figure 7, both conversion and selectivity were highest at 35 °C. Further increasing the temperature above 35 °C led to pronounced decreases in conversion and selectivity, likely due to the partial loss of enzyme activity and/or stability at elevated temperatures.

3.2.2. The Effect of the Resveratrol Concentration on Glycosylation

To maximize polydatin productivity in the microreactor, the influence of the initial resveratrol concentration on glycosylation yield and selectivity was examined. As shown in Figure 8, the conversion decreased progressively as the substrate concentration increased. Therefore, an initial resveratrol concentration of 1 mM was selected for subsequent experiments.

3.2.3. The Effect of the UDP Concentration on Glycosylation

In the dual-enzyme coupling system, UDP and sucrose were supplied to generate UDP-Glc; therefore, the UDP loading was optimized. As shown in Figure 9, maintaining UDP at 0.05–0.1 mM markedly improved both conversion and selectivity. In contrast, when the UDP concentration exceeded 1 mM, the conversion of 1 decreased, which may be attributed to excess UDP inhibiting the cascade, for example, by suppressing UDP-Glc formation and/or causing product (UDP) inhibition of the glycosyltransferase.

Consistent with the trends observed in Figure 8 and Figure 9, elevated resveratrol levels or excess UDP resulted in a noticeable loss of conversion, indicating deviations from ideal first-order behavior under these conditions. Notably, our kinetic analysis was based on initial-rate measurements, where product accumulation was minimal; thus, effects such as UDP-mediated product inhibition and the approach to saturation kinetics were not explicitly captured. These kinetic limitations likely account for the reduced conversion observed at higher substrate or UDP concentrations.

3.2.4. The Effect of the Residence Time on Glycosylation

Finally, the effect of residence time was evaluated. As shown in Figure 10, the conversion increased markedly as the residence time was extended to 40 min, and then approached a plateau with little further improvement. Accordingly, a residence time of 40 min was selected for subsequent experiments.

4. Conclusions

In this study, a continuous-flow microreactor process was developed for the enzymatic synthesis of polydatin from resveratrol. Kinetic analysis showed apparent first-order dependences on resveratrol, UDP-Glc, and glycosyltransferase (α = β = γ = 1), consistent with a sequential substrate-binding pathway. To reduce the cost associated with UDP-Glc, a glycosyltransferase–sucrose synthase coupled cascade was implemented in the microreactor to enable in situ UDP-Glc regeneration. Under continuous operation, key parameters were systematically optimized: the best performance was obtained at 35 °C, with an initial resveratrol concentration of 1 mM, a UDP loading of 0.1 mM, and a residence time of 40 min. Benefiting from the advantages of flow processing (enhanced mixing, precise residence-time control, and facile numbering-up), this microreactor-based cascade provides an efficient and potentially industrially relevant strategy for continuous polydatin production.