Evaluation of the Biological Properties of an Optimized Extract of Polygonum cuspidatum Using Ultrasonic-Assisted Extraction

Phytochemicals are natural compounds found in plants that have potential health benefits such as antioxidants, anti-inflammatory and anti-cancer properties, and immune reinforcement. Polygonum cuspidatum Sieb. et Zucc. is a source rich in resveratrol, traditionally consumed as an infusion. In this study, P. cuspidatum root extraction conditions were optimized to increase antioxidant capacity (DPPH, ABTS+), extraction yield, resveratrol concentration, and total polyphenolic compounds (TPC) via ultrasonic-assisted extraction using a Box–Behnken design (BBD). The biological activities of the optimized extract and the infusion were compared. The optimized extract was obtained using a solvent/root powder ratio of 4, 60% ethanol concentration, and 60% ultrasonic power. The optimized extract showed higher biological activities than the infusion. The optimized extract contained 16.6 mg mL−1 resveratrol, high antioxidant activities (135.1 µg TE mL−1 for DPPH, and 230.4 µg TE mL−1 for ABTS+), TPC (33.2 mg GAE mL−1), and extraction yield of 12.4%. The EC50 value (effective concentration 50) of the optimized extract was 0.194 µg mL−1, which revealed high cytotoxic activity against the Caco-2 cell line. The optimized extract could be used to develop functional beverages with high antioxidant capacity, antioxidants for edible oils, functional foods, and cosmetics.


Introduction
Nutraceutical is a term derived from "nutrition" and "pharmaceutics", used for compounds isolated from herbal products with biological activity. These compounds provide health benefits, especially for preventing and treating diseases such as cancer, diabetes, cardiovascular and neurological disorders [1,2]. These diseases are associated with the generation and accumulation of reactive oxygen species (ROS) produced by cellular oxidative stress [3,4]. Antioxidant compounds can inhibit or decrease oxidation processes that affect biomolecules such as proteins, lipids, and DNA [5]. Antioxidants can protect cells against oxidation by blocking the initiation phase of radical production or neutralizing radicals. Commonly, herbal plants contain antioxidant properties due to the presence of bioactive compounds. Polygonum cuspidatum Siebold & Zucc. belongs to the Polygonaceae family and grows widely in Asia and North America. It has been used for centuries in China and Japan as an herbal medicine to treat inflammatory diseases, hepatitis, tumors, diarrhea, arthralgia, Table 1 shows the experimental values of antioxidant capacity (DPPH and ABTS+), TPC, resveratrol concentration, and extraction yield obtained by UAE from the interaction variables: solvent/root-powder ratio (X 1 ), ethanol concentration (X 2 ), and ultrasonic power (X 3 ). According to the experimental values, the scavenging capacities for DPPH and ABTS+ ranged from 51.1 to 135.1 µg TE mL −1 and 119.5 to 230.4 µg TE mL −1 , respectively. The data obtained for total polyphenolic compounds ranged from 5.8 to 33.2 mg GAE mL −1 . The resveratrol content in crude extract and the extraction yield ranged from 12.0 to 16.7 mg mL −1 and 2.74% to 12.43%, respectively. A reduced quadratic polynomial was used to predict the experimental data, shown in Equation (1).  Table 2 shows the ANOVA results for the responses and interactions of the independent variables. Regarding antioxidant activities (ABTS + and DPPH), variables X 1 and X 2 2 showed negative effects, while variable X 2 had highly significant positive effects (p < 0.001). Meanwhile, variable X 3 did not show a significant effect; however, it contributed positively to the antioxidant activities. The analysis indicated that the determinant coefficients (R 2 ) were 0.9612 for ABTS + and 0.9593 for DPPH. The regression model explained 96.12% and 95.93% of the responses for ABTS + and DPPH, respectively (p < 0.05). The TPC showed a linear impact for variables X 1 and X 2 , with an R 2 = 0.8581 , able to explain 85.81% of the fitted regression model (p < 0.05). Ethanol concentration was the most critical parameter for increasing the resveratrol content in the extract. The lineal variable X 2 showed a highly significant positive effect (p < 0.001), while variable X 2 2 showed a highly significant negative effect (p < 0.001). According to the F-value, the impact of the linear X 2 is higher than the quadratic X 2 , resulting in a positive effect on resveratrol extraction. Variables X 1 , X 3 , X 1 X 2 , X 2 X 3 , significantly impacted the resveratrol content, showing an excellent coefficient of correlation of the predicted model (R 2 = 0.9850). The extraction yield indicated a linear effect for X 1 and X 3 and a quadratic impact for X 2 , an R 2 = 0.8993. Data analysis of the extraction yield showed that the mathematical model could predict the effect of the interactions for variables X 1 , X 2 , and X 3 on UAE from P. cuspidatum.
Resveratrol content ( Figure 2B,C) showed a highly significant impact (p < 0.001), with resveratrol content increasing to 16.7 mg mL -1 when the ethanol concentration was higher than 50%. These results suggest that solvent/root-powder ratio, ultrasonic power, and interactions impacted the resveratrol concentration. The maximum extraction yield ( Figure  2D), corresponding to 12.43%, reached a lower solvent/root-powder ratio (4 mL g -1 to 7 mL g -1 ) and high ethanol concentration (45 to 60%).
The regression coefficients to calculate the predicted response for the antioxidant activities of ABTS+ and DPPH, TPC, resveratrol content, and extraction yield were performed using a reduced second-order polynomial equation (Table 3). TPC showed a maximum concentration of 33.2 mg GAE mL −1 at higher ethanol concentrations (45% to 60%) and low solvent/root-powder ratios (4 mL g −1 to 7 mL g −1 ) ( Figure 2A).
Resveratrol content ( Figure 2B,C) showed a highly significant impact (p < 0.001), with resveratrol content increasing to 16.7 mg mL −1 when the ethanol concentration was higher than 50%. These results suggest that solvent/root-powder ratio, ultrasonic power, and interactions impacted the resveratrol concentration. The maximum extraction yield ( Figure 2D), corresponding to 12.43%, reached a lower solvent/root-powder ratio (4 mL g −1 to 7 mL g −1 ) and high ethanol concentration (45 to 60%).
The regression coefficients to calculate the predicted response for the antioxidant activities of ABTS+ and DPPH, TPC, resveratrol content, and extraction yield were performed using a reduced second-order polynomial equation (Table 3).  3D plots for interaction variables for (A) TPC (X 1 X 2 ), (B) resveratrol content (X 1 X 2 ) y (C) resveratrol content (X 2 X 3 ), and (D) extraction yield (X 2 X 3 ). X 1 = solvent/root-powder ratio, X 2 = ethanol concentration, and X 3 = ultrasonic power. Table 3. Reduced polynomial equations for antioxidant capacity, TPC, resveratrol content, and extraction yield.
The optimal predicted value was obtained in experimental run 14, corresponding to a solvent/root-powder ratio of 4, 60% ethanol concentration, and 128.5 W of ultrasonic power, corresponding to 58 KJ/g. The predicted values for the antioxidant activity of DPPH and ABTS+ corresponded to 136.625 µg TE mL −1 and 227.575 µg TE mL −1 , respectively. For TPC, the expected value was 34.068 mg GAE mL −1, and resveratrol content and extraction yield were 15.56 mg mL −1 and 11.36%, respectively. The predicted desirability value for optimal extraction of P. cuspidatum was 0.9661 ( Figure 3). The experimental conditions were validated using an independent experiment, finding 136.2 µg TE mL −1 , 195.4 µg TE mL −1 , 29.95 mg GAE mL −1 , 16.72 mg mL −1 , and 12.3% for DPPH, ABTS+, TPC, resveratrol, and extraction yield, respectively.  The optimal predicted value was obtained in experimental run 14, corresponding to a solvent/root-powder ratio of 4, 60% ethanol concentration, and 128.5 W of ultrasonic power, corresponding to 58 KJ/g. The predicted values for the antioxidant activity of DPPH and ABTS+ corresponded to 136.625 µg TE mL -1 and 227.575 µg TE mL -1 , respectively. For TPC, the expected value was 34.068 mg GAE mL -1, and resveratrol content and extraction yield were 15.56 mg mL -1 and 11.36%, respectively. The predicted desirability value for optimal extraction of P. cuspidatum was 0.9661 ( Figure 3). The experimental conditions were validated using an independent experiment, finding 136.2 µg TE mL -1 , 195.4 µg TE mL -1 , 29.95 mg GAE mL -1 , 16.72 mg mL -1 , and 12.3% for DPPH, ABTS+, TPC, resveratrol, and extraction yield, respectively.
Analysis of the infusion of P. cuspidatum root powder indicates that the values for the antioxidant activities of DPPH and ABTS+ were 103.75 µg TE mL -1 and 147.78 µg TE mL -1 , respectively. In addition, the TPC value was 0.4024 mg GAE mL -1 . The resveratrol content and extraction yield were 0.044 mg mL -1 and 0.340%, respectively.

Cytotoxic Assay and EC50 Value
The optimized extract was diluted (1:110) in MEM (Minimal Eagle's Medium), and the infusion was used to directly compare the cytotoxicity on colorectal cancer cells for 24 h. The results obtained for the optimized extract demonstrate that the viability of Caco-2 cells decreases at low concentrations in a dose-depending manner (1.24 µg mL -1 to 0.03 µg mL -1 ), and the EC50 value corresponded to 0.125 ± 0.008 (R 2 = 0.9913) ( Figure 4A). In contrast, the infusion extraction of P. cuspidatum showed low viability with an estimated EC50 of 0.03 ± 0.002 µg mL -1 (R 2 = 0.9892) ( Figure 4B). Figure 5A shows that Caco-2 cells treated with optimized extraction were smaller and had low cell confluency compared to Analysis of the infusion of P. cuspidatum root powder indicates that the values for the antioxidant activities of DPPH and ABTS+ were 103.75 µg TE mL −1 and 147.78 µg TE mL −1 , respectively. In addition, the TPC value was 0.4024 mg GAE mL −1 . The resveratrol content and extraction yield were 0.044 mg mL −1 and 0.340%, respectively.

Cytotoxic Assay and EC 50 Value
The optimized extract was diluted (1:110) in MEM (Minimal Eagle's Medium), and the infusion was used to directly compare the cytotoxicity on colorectal cancer cells for 24 h. The results obtained for the optimized extract demonstrate that the viability of Caco-2 cells decreases at low concentrations in a dose-depending manner (1.24 µg mL −1 to 0.03 µg mL −1 ), and the EC 50 value corresponded to 0.125 ± 0.008 (R 2 = 0.9913) ( Figure 4A). In contrast, the infusion extraction of P. cuspidatum showed low viability with an estimated EC 50 of 0.03 ± 0.002 µg mL −1 (R 2 = 0.9892) ( Figure 4B). Figure 5A shows that Caco-2 cells treated with optimized extraction were smaller and had low cell confluency compared to untreated cells ( Figure 5C). Cells treated with the infusion of P. cuspidatum show higher confluency and apoptotic bodies, characteristic of the induction of apoptotic cell death ( Figure 5B), compared to those treated with the optimized extract ( Figure 5C). untreated cells ( Figure 5C). Cells treated with the infusion of P. cuspidatum show higher confluency and apoptotic bodies, characteristic of the induction of apoptotic cell death ( Figure 5B), compared to those treated with the optimized extract ( Figure 5C).
(A) (B)   Figure 6 shows the chromatogram of the optimized extract from P. cuspidatum obtained through MS analysis. The spectra revealed two main peaks attributed to stilbene compounds: piceid (RT = 5.71) and resveratrol (RT = 7.55). Thus, these results indicate that piceid and resveratrol are the major bioactive components in the optimized extract. untreated cells ( Figure 5C). Cells treated with the infusion of P. cuspidatum show higher confluency and apoptotic bodies, characteristic of the induction of apoptotic cell death ( Figure 5B), compared to those treated with the optimized extract ( Figure 5C).

Compounds Identified by UPLC-Mass Spectrometry (MS)
(A) (B)   Figure 6 shows the chromatogram of the optimized extract from P. cuspidatum obtained through MS analysis. The spectra revealed two main peaks attributed to stilbene compounds: piceid (RT = 5.71) and resveratrol (RT = 7.55). Thus, these results indicate that piceid and resveratrol are the major bioactive components in the optimized extract.  Figure 6 shows the chromatogram of the optimized extract from P. cuspidatum obtained through MS analysis. The spectra revealed two main peaks attributed to stilbene compounds: piceid (RT = 5.71) and resveratrol (RT = 7.55). Thus, these results indicate that piceid and resveratrol are the major bioactive components in the optimized extract.

Discussion
Bioactive compounds have medicinal benefits and are extracted from fresh or dried plants using different extraction methods and solvents [26]. Solvents impact the recovery and quality of bioactive compounds, but can degrade them at high temperatures [27]. Water is used to prepare beverages with an affinity for hydrophilic molecules, such as phenolic compounds, proteins, and carbohydrates. Studies have reported that the low dielec-

Discussion
Bioactive compounds have medicinal benefits and are extracted from fresh or dried plants using different extraction methods and solvents [26]. Solvents impact the recovery and quality of bioactive compounds, but can degrade them at high temperatures [27]. Water is used to prepare beverages with an affinity for hydrophilic molecules, such as phenolic compounds, proteins, and carbohydrates. Studies have reported that the low dielectric constant and low polarity of hot water increase the diffusion of compounds, improving the extraction of lipophilic bioactive compounds. On the other hand, ethanol is used to extract bioactive compounds, such as polyphenols, flavonoids, alkaloids, and terpenes [28,29]. Several factors, such as plant type, ethanol concentration, and extraction time, impact the extraction yield. The ethanol-water concentration increases the polarity of the solvent, improving the efficiency of the extraction of nonpolar phytochemicals [30].
Usually, resveratrol is extracted with 95% ethanol from the root of P. cuspidatum using a refluxing method followed by liquid-liquid extraction with organic solvents [25]. Kuo et al. [18] found that when applying a temperature of 70 • C, 60% ethanol concentration, and 120 W ultrasonic power, the amount of resveratrol was 3.9 mg/g. Jia et al. [31] reported high flavonoid concentration (94.5%) using UAE-and CCRD-extracted flavonoids from P. cuspidatum using an ethanol concentration of 60%, solid-liquid ratio of 1:20 g mL −1 , extraction temperature of 45 • C, extraction time of 34 min, and ultrasonic power of 80 W. Ruan et al. [32] reported an extraction yield value for resveratrol and emodin of 2.564 ± 0.121 mg mL −1 and 2.804 ± 0.108 mg mL −1 , respectively, and a high scavenging capacity. Our results showed that resveratrol content at the optimized extract conditions was higher than those reported in the literature.
The standard methods to evaluate the antioxidant capacity of foods employ the stable radicals DPPH and the cation ABTS+. Results obtained from antioxidant assays reported the necessary antioxidant concentrations to reduce radicals [33]. Becze et al. [34] obtained extracts with a high antioxidant capacity of 34.623 µg AAE mL −1 and 182.4 µL of resveratrol content from F. regala leaves. UAE of Arachis repens, known as peanut ass, showed high concentrations of trans-resveratrol and total polyphenolics, demonstrating a high level of DPPH free radical inhibition (70%). Polyphenolic compounds can reduce the Folin-Ciocalteu reagent under alkaline conditions, which yields a colored product [35]. El Moussaoui et al. [36] evaluated the antioxidant activity and total polyphenols of extract from Whitania frutescens L., and found that roots are 67 times richer in polyphenols (53.3 ± 1.2 mg GAE/g).
The optimized ethanolic extract showed higher cytotoxic activity than the infusion extract cont due to the higher resveratrol concentration found in the ethanolic extract. Similar results were found by Youmbi and coworkers [37]. They found that crude extract obtained from the leaves of Brucea antidysenterica induces cell death mediated by caspases on drug-resistant cancer cells, such as CCRF-CEM, a human leukemia cell line, at a low concentration (from 12.42 µg mL −1 to 38.70 µg mL −1 ). Other authors reported that ethanol extracts of P. cuspidatum demonstrated efficacy on the loss of viability of cells in oral, breast, and ovarian cancer at lower concentrations, EC 50 : <50 µg mL −1 [17,38,39].
On the other hand, tea or infusion is widely used for its beneficial properties and low cost. Traditionally, P. cuspidatum is commonly used as part of traditional Chinese medicine, and is prepared as a herbal infusion in boiling water for a few minutes. Kosovic et al. [40] evaluated the stability of resveratrol from Vitis vinifera L. at high temperatures, finding a better release of stilbenes such as trans-resveratrol. However, if a prolonged time was applied, the polyphenolic compounds could be degraded. The effect of green tea on rat hepatocyte cells was evaluated by Schmidt et al. [41], and the results show that cell viability decreased at 500 µg mL −1 of the extract. Aqueous extract of Reynoutris japonica Houtt root, a Polygonaceae family, was not cytotoxic in SK-Hep1 and Huh7 cell lines, but in vitro results indicated that at a concentration of 20 µg mL −1 inhibited wound recovery and invasion in Huh7 cells [42].
The main compounds in P. cuspidatum root extracts have been investigated using chromatographical analysis. Yi et al. [43] reported that the P. cuspidatum rhizome is rich in stilbenes (piceid and resveratrol) and anthraquinones (emodin-8-O-β-D -glucoside, and emodin). These compounds are commonly used as an indicator of quality assessment for herbal extracts. Vastano et al. [44] evaluated two varieties of P. cuspidatum root extracts (Hu Zhang and Mexican Bamboo). They identified three stilbene glucosides: piceatannol glucoside, resveratroloside, and piceid. By comparison, our results showed that the optimized extract of P. cuspidatum root contains piceid and resveratrol as the major bioactive compounds.

Optimization of Ultrasonic-Assisted Extraction (UAE)
P. cuspidatum powder was dispersed in ethanol, according to Table 1. Then, dispersions were sonicated using an Elmasonic p70H ultrasonic bath (37 kHz, Elma Schmidbauer GmbH, Singen, Germany) following the conditions of the Box-Behnken design. After applying the ultrasound, the samples were centrifuged at 4000 rpm for 15 min. The supernatant was collected and stored at −20 • C until analysis. As a control, an infusion of 10 g of root powder was dispersed into 200 mL of boiling water; afterward, the solution was filtrated to remove sediment and stored at −20 • C until analysis.

Experiment Design Strategy and Statistical Analysis
RSM optimized the UAE extraction parameters for DPPH, ABTS+, TPC, resveratrol concentration, and extraction yield from P. cuspidatum. In a three-level, three-factorial Box-Behnken design, including five replicates of the central point, 17 runs were analyzed in random order ( Table 4). The independent variables were solvent/root-powder ratio (X 1 , mL g −1 ), ethanol concentration (X 2 , %), and ultrasonic power (X 3 , W). Temperature and extraction time were fixed at 45 • C and 30 min, respectively. RSM was applied to obtain the optimal UAE conditions for each raw material. A second-order polynomial equation derived from the RSM was used to calculate the predicted response (Equation (1)): Y is the response variable, X i is the coded or uncoded value for the factors evaluated, β 0 is a constant, β i is the main effect of the coefficient for each variable, β ij represents the interaction effect coefficients, and E is the error of the model. ANOVA evaluated significant interactions of the model (p < 0.05). The coefficient of determination was quantified (R 2 and adjusted R 2 ). The predicted values for antioxidant activity (DPPH and ABTS+), TPC, resveratrol content, and extraction yield were maximized to establish the optimal UAE via RSM. An additional extraction was conducted using the optimal predicted conditions to verify the model's suitability. Applied energy was estimated according to Strieder et al. [45]: energy = [ultrasonic power × time extraction]/mass of the sample. The regression coefficients were used to generate 3D surface plots from the fitted polynomial equation. Also, the regression coefficients were used to visualize the relationship between the response variable and experimental levels and predict the optimum conditions.

Determination of Extraction Yield
The extraction yield (%) was quantified by vaporization utilizing 2 mL of crude extract in an oven at 40 • C until the dry mass was obtained. The results were shown as a mass of total extractable solids per 100 g of dry plant material (%).

Resveratrol Quantification Using Spectrophotometry UV-Vis
UV-Vis spectroscopy was used to assess the concentration of resveratrol in the crude extracts. Samples were compared with a standard curve of resveratrol in ethyl alcohol. Ethyl alcohol was utilized as a blank for background correction. Absorbance was read at 312 nm using a multiwell plate reader (TECAN infinite pro-200, Trading AG, Steinhausen, Switzerland).

DPPH Radical Scavenging Assay
For the DPPH stable radical scavenging assay, samples (ethanolic samples 1:400, infusion 1:2), blank (ethyl alcohol), or Trolox standard were added to a 96-well microplate. Subsequently, DPPH solution was added to each well and allowed to stand for 30 min in the dark before reading absorbance at 517 nm. All experiments were realized in triplicate. The capacity to scavenge the DPPH radical was presented as µg of Trolox equivalent (TE) mL −1 [46].

ABTS+ Assay
For the ABTS+ test, in a 96-well microplate the sample (ethanolic samples 1:400, infusion 1:2), blank (ethyl alcohol pure), or Trolox standard were added. Subsequently, the ABTS+ working solution was added to each well and allowed to stand for 5 min in the dark before reading absorbance at 734 nm [22]. All experiments were realized in triplicate. The results for antioxidant activity in the ABTS+ radical test were presented as µg of Trolox equivalent (TE) mL −1 [46].

Caco-2 Cell Culture and Cytotoxic Assay
The cytotoxic assay for the optimal extraction of P. cuspidatum was evaluated using a resazurin reduction assay [48]. The optimal extract was diluted with MEM to obtain different concentrations. The Caco-2 cell line (HTB-37 ATCC), a human colorectal adenocarcinoma, was cultured in MEM supplemented with 10% FBS and 1% penicillin/streptomycin. Briefly, cells were incubated until reaching~80% confluence and seeded in a 96-well plate at 5 × 10 3 cells per well at 37 • C and 5% CO 2 . After 24 h, MEM was replaced with 100 µL of the extracts at different concentrations and incubated for 24 h (37 • C and 5% CO 2 ). After that, 20 µL of resazurin solution (0.2 mg mL −1 ) was added and incubated for 3 h at 37 • C. The fluorescence emitted by resorufin, metabolized from resazurin by viable cells, was measured (excitation 560 nm, emission 590 nm). A negative control (MEM medium) and positive control (DMSO 10%) were used. Percentage viability and effective concentration 50 (EC 50 ) of crude extracts and the infusion of P. cuspidatum were calculated using OriginPro 2018 Software (Origin Lab Corporation, Northampton, MA, USA). The cell viability was calculated as shown: % cell viability = I s /I c × 100% where: I s is the absorbance of the cells exposed to the extracts and I c is the absorbance of the cells without exposure to the extracts

Chromatography Analysis by Mass Spectrometry (MS/MS)
Mass spectrometry (MS) analysis was performed using a Waters Xevo TQ-S instrument (Waters, Milford, MA, USA). The conditions used for identifying the bioactive compounds from the optimized extract by comparing fingerprint and MS data were those previously reported [49,50]. Resveratrol and piceid were identified using multiple reaction monitoring (MRM), a sensitive method of targeted MS. For acquisition and data processing, MassLynx V4.1 software was employed (Waters, Milford, MA, USA).

Statistical Analysis
The results obtained were analyzed using ANOVA and Tukey's means comparison analysis. Data from the Box-Behnken experimental design were examined using the least square multiple regression methodology to fit the polynomial models for UAE optimization. Data analysis and response surfaces were conducted using Statgraphics Centurion XVI. I software (Statistical Graphics Corp., Manugistics, Inc., Cambridge, MA, USA). Significance was established at p ≤ 0.05. All experiments were conducted in triplicate.

Conclusions
Ultrasonic-assisted extraction from P. cuspidatum roots was effectively developed. The extraction efficiency, antioxidant activity, resveratrol content, and total polyphenolic compounds could be further improved through BBD experimental design and response surface methodology. A second-order model was obtained to describe the relationship between the responses and their interactions with independent variables X 1 (solvent/root powder ratio), X 2 (ethanol concentration), and X 3 (ultrasonic power at 45 • C for 30 min). The evaluated parameters affected antioxidant activity (DPPH, ABTS+), total polyphenolic compounds, resveratrol content, and extraction yields from the crude extract from P. cuspidatum root. The optimized extract showed higher cytotoxic activity than the traditional infusion of P. cuspidatum root against Caco-2 colorectal cancer cells. Due to the complex mixture in the optimized extract of P. cuspidatum, future research is required to identify and quantify all the bioactive compounds and to determine the mechanism by which this extract induces cell death.