ORIGINAL ARTICLE Year : 2014  Volume : 10  Issue : 37  Page : 5764 Optimization of quercitrin and total flavonoids extraction from Herba Polygoni Capitati by response surface methodology Fengwei Ma^{1}, Yang Zhao^{1}, Xiaojian Gong^{1}, Yu Xie^{2}, Xin Zhou^{1}, ^{1} The Research Center for Quality Control of Nature Medicine; Key Laboratory for Information System of Mountainous Area and Protection of Ecological Environment of Guizhou province, Guizhou Nomal University, Guiyang, Guizhou, China ^{2} Guizhou Warmen Pharmaceutical Co., Ltd., Guiyang, Guizhou, China Correspondence Address: Objective: To optimize the conditions for extraction of quercitrin and total flavonoids (TF) from Herba Polygoni Capitati (Touhualiao in Chinese) by using response surface methodology (RSM). Materials and Methods: A central composite design (CCD) was adopted to investigate the effects of three independent variables including solvent composition (%), solventmaterial ratio (ml/g) and extraction time (min) on the responses, quercitrin and TF yields. Results: The optimized conditions of the extraction are as follows: Ethanol concentration, 65.63%; solventmaterial ratio, 10.55:1 (ml/g); extraction time, 54.33 min. The established mathematical model described the factors of experimental parameters well and provided a statistically accurate prediction of the optimum yields of quercitrin and TF. Conclusion: The experimental values agreed with those predicted by the established mathematical model, thus indicating the suitability of the model employed and the success of RSM in optimizing the extraction conditions.
Introduction Herba Polygoni Capitati, the aerial part or whole plant of Polygonum capitatum Buch.Ham. ex D. Don, has been used as one of the Traditional Chinese Medicines (TCMs) for a long time in China, especially in some ethnic minority regions. The active chemical constituents of Herba Polygoni Capitati were identified as flavonoids, phenolic acids and tannins, [1],[2],[3] in which flavonoids were considered as the major bioactive compounds. Pharmacological and clinical studies indicated quercetin and its glucosides were bioactive compounds possessing anticarinogenic, [4],[5] antioxidative [6],[7] and enzymemodulating activities. [8],[9],[10] Epidemiological studies also show an inverse relationship between high intake of flavonoids (mainly quercetin) and cardiovascular disease. [11] Extracts of TCMs have been used in traditional cures and herbal remedies for centuries throughout the world [12],[13] and a great deal of valuable experiences have been accumulated in this therapeutic system. Ephedrine, an amphetaminelike stimulant used as decongestant for asthma medication, was extracted and isolated from the Chinese medicinal herb Ephedrae herba.[14] The extract of Ginkgo biloba, especially the standardized EGb 761 was one of the most widely used herbal remedies for dementia and cognitive impairment. [15] Since, many factors such as solvent composition, extraction time, extraction temperature, solventmaterial ratio may significantly influence the extraction efficacy, extraction parameters need to be optimized to get more ingredients of prophylactic or therapeutic value in human subjects. [16] Of all the compounds presented in Herba Polygoni Capitati, the phenolic acids and flavonoid components had received most attention and were considered to be of importance for their pharmacological effects. [17] The extraction of Gallic acid (GA) or TF from Herba Polygoni Capitati in onefactoratatime approach alone, [18] or associated with orthogonal design, [19] has been reported previously. The onefactoratatime approach, in which only one factor is variable at a time while keeping all other factors constant, [20] is time consuming and might cause misleading conclusions as it does not include interactive effects among factors. Orthogonal design is useful in test design [21] but it does not offer prediction from the model equation. Response surface methodology (RSM), a collection of statistical and mathematical techniques, can overcome these drawbacks, since it accounts for possible interaction effects between variables. [22],[23] The response surface plot and contour plot of the response as a function of the independent parameters can be obtained by RSM. [24] In the present study, the RSM approach was applied to optimize the extraction conditions (solvent proportion, solventmaterial ratio, time of extraction, etc.) from Herba Polygoni Capitati in order to maximize the yields of quercitrin and TF simultaneously. Materials and Methods Materials and chemicals Herba Polygoni Capitati, botanically identified by professor Deyuan Chen, was obtained from Guizhou Warmen Pharmaceutical Corporation, Guizhou, China. The whole plant of Herba Polygoni Capitati was dried in a universal oven with forced convection (1012AB Taijin, China) at 50°C for 4 h. The dried sample was ground in a rotary mill (DFY200 Wenling, China) and the powder was sieved. Particles with sizes between 10 and 40 mesh (0.3~2mm, i.d.) were collected for the study. The reference compounds quercitrin (111538200403) and rutin (0809303) were purchased from the National Institute for the Control of Biological and Pharmaceutical Products of China (Beijing, China). Acetonitrile and methanol were HPLC grade (Tedia, USA). Water was Robust drinking pure water (Le Bai Shi Food and Beverage Ltd., Guangdong). Formic acid was MS grade (Roe Scientific Inc, USA). All other chemicals and solvents were of analytical grade and commercially available. Sample preparation Herba Polygoni Capitati powder of 2g was accurately weighed and extracted under certain conditions, thereafter the extract was centrifuged at 3500 rpm for 10 min. The supernatant was collected and filtered through filter paper for subsequent analysis. Apparatus The analytical systems used in this study were as follows: The UVVis spectra were recorded on a doublebeam spectrophotometer (Cary 100Varian) with cells of 1 cm path length. A DIONEX/UltiMate 3000 series UHPLCDAD system consisting of a vacuum degasser, a quaternary pump, an autosampler, a thermostated column compartment and a DAD detector (DIONEX, ThermoFisher, Sunnyvale, USA) were used for acquiring chromatograms. Determination of total flavonoid content The total flavonoid content (TFC) was determined according to the aluminum colorimetric method described previously [25],[26],[27],[28] with modifications. Briefly, an aliquot (0.5ml) of each plant extract was mixed with 4.3ml of related solvent, followed by adding 0.1ml of 10% aluminum chloride (AlCl 3 ) and 0.1ml of 1M sodium acetate (CH 3 COONa). Rutin was used as the reference standard. After incubating at room temperature for 30 min, the reaction mixture was measured at 405nm against a blank. The TFC value of each sample was calculated and expressed as milligrams of rutin equivalents per gram of dried material (DM). Chromatographic separation HPLC analyses of the extracts were carried out by using DIONEX/UltiMate 3000 series. The separation was performed on a ZORBAX SBC 18 column (150 mm × 4.6 mm, 5 μm; Agilent, CA, USA) with a flow rate of 1.0ml/min. The mobile phase consisted of a combination of A (acetonitrile) and B (0.2% formic acid in water). The linear gradient was from 5% to 15% A (v/v) at 5 min, to 16% A at 10 min, to 20% A at 20 min and was then held at 20% A to 22 min. Each run was followed by equilibration time of 5 min. The wavelength of the UV detector was set at 260 nm and the temperature of the column oven was maintained at 30°C. Samples were filtered before injection (PTFE syringe filter; 0.45 μm; Hanbon Ltd, China; 013045). The data was collected and analyzed with Chromeleon 7.1 software. Preparation of standard solutions Standard stock solutions of rutin and quercitrin were prepared in methanol, at concentrations of 190.8 μg/ml and 106.0 μg/ml, respectively. The standard curve of quercitrin and TF were calibrated by using the linear least squares regression equation derived from the peak area and absorbance, respectively, concentrations of quercitrin and TF in the samples were calculated according to the standard curve. Selection of appropriate extraction conditions Before the development of the study by using RSM, a set of preliminary tests were performed to select the relevant independent variables and determine their experimental ranges. The initial step of the preliminary experiment was to select an appropriate method for extraction for Herba Polygoni Capitati. Three different approaches, namely maceration, refluxing and sonication extraction were tested. The second step of the preliminary experiment was to select an appropriate extraction solvent for Herba Polygoni Capitati. Using a series of extraction solvents varying in range of 0100% (ethanol or methanol in water), the best solvent was determined. Thirdly, the factor of the solventmaterial ratio (6:1, 8:1, 10:1, 12:1, 14:1; v/m) on extraction was investigated by using ethanol: water (70:30, v/v) as the solvent on the condition of sonication extraction for 30 min at room temperature. Fourthly, a proper particle size was selected for further investigation. Finally, duration of extraction (10 min, 20 min, 30 min, 40 min, 50 min, 60 min) was also tested. Based on these results, three levels (lower, middle, upper) of each relevant variable were selected for central composite design (CCD) to analyze the response pattern and to establish a model. Statistical analysis The experimental data obtained from CCD procedures were analyzed by using RSM to fit the following secondorder polynomial model (Eq. (1) and regression coefficients obtained. [INLINE:1] Where X 1 , X 2 ,…, X k are the encoded independent variables affecting the response Y; β0, βi (i = 1, 2,…, k), βii (i = 1, 2,…, k) and βij (i = 1, 2,…, k) are the regression coefficients for intercept, linear, quadratic, and interaction terms, respectively; k is the number of variables. The Design Expert version 8.0.5b (STATEASE Inc.) software was used for regression and graphical analysis of the experimental data. The quality of the model's fitness was evaluated using the coefficients of determination (R 2) and analysis of variance (ANOVA). Response surfaces and contour plots were developed using the fitted quadratic polynomial equation. Results and Discussions System precision, linearity of HPLC or UVVis Quercitrin and TF were designated as the markers for evaluation of extraction efficiency. HPLC profiles of the reference standard and Herba Polygoni Capitati extract were shown in [Figure 1]. The calibration curve of peak area against level of quercitrin was y = 47.622 x  0.2231 with linear range from 0.0331.06 μg (R 2 = 1.0000, y = peak area and x = quercitrin amount). The calibration curve of absorbance against the amount of TF was y = 0.0029 x  0.0088 with linear range from 74.32267.12 μg (R 2 = 0.9996, y = absorbance and x = TF amount).{Figure 1} Determination of parameters and the selection of the levels Effect of the extraction methods 0Three different methods of extraction, including maceration, refluxing and sonication extraction were tested with the following parameter values: particle size, 10 mesh; solventmaterial ratio, 10:1 (ml/g); duration of extraction, 30 min for refluxing and sonication extraction, 12h for maceration extraction; solvent composition, 70% aqueous ethanol; temperature, room temperature. The amount of quercitrin and TF extracted per gram of DM was presented in [Figure 2]. ANOVA for the experimental results showed significant differences among these three extraction methods (P < 0.01). Meanwhile, the sonication extraction method gave the highest yields of quercitrin and TF, so it was selected as our extraction method.{Figure 2} Effect of extraction solvents The solvent is a key factor affecting the recovery of analytes. In the present study, water, different concentrations of methanol and different concentrations of ethanol were tested. As shown in [Figure 3], the extraction efficiency of 60% aqueous ethanol was the highest. Therefore, 50%, 60% and 70% ethanol were selected as the lower, middle and upper levels of the solvent composition for further RSM investigation.{Figure 3} Solventmaterial ratios The impact of solventmaterial ratios on the extraction was tested with five values (6:1, 8:1, 10:1, 12:1, 14:1; v/m). A suitable solventmaterial ratio value would be selected based on the rules [29] that the value should neither be too small (the bioactive compounds cannot be completely extracted up) nor too high (the processing cost would be unbearable). As presented in [Figure 4], an increasing trend of the quercitrin and TF yields was observed, reaching the highest values when the ratio was about 10:1. A probable explanation was that an increase in solventmaterial ratio may augment diffusivity of the solvent into the cells and enhance desorption of the phenolic compounds from the cells. [30] In the present study, a slight decrease in quercitrin and TF yields were observed when the ratio value was higher than 10:1, which might be due to a bigger loss in product collection or process operation. Therefore, the solventmaterial ratio was set at 10:1 as the middle level for RSM in subsequent study.{Figure 4} Effect of particle size Particle size is another important variable to be considered. Generally, extraction efficiency increases with the decrease of particle size [Figure 5]. In the present study, particle size was set at 0.450.9 mm in diameter.{Figure 5} Effect of extraction time Selection of an appropriate extraction time was the final step in our preliminary study. Different lengths of extraction at 10, 20, 30, 40, 50 and 60 min were tested to examine the optimal value. The results showed that the yields of quercitrin and TF increased with the prolongation of extraction time from 10 to 50 min [Figure 6]. At the level of 50 min, the quercitrin and TF yields reached the highest amount. Hence, 40, 50 and 60 min were selected as the lower, middle and upper levels of extraction time.{Figure 6} Optimization of extraction parameters by RSM Multiple linear regression results and analysis of the adequacy of the fitted model A threefactor and threelevel rotatable CCD consisting of 20 experimental runs was employed with 6 replicates at the center point. The effects of unexplained variability were minimized by randomizing the order of experiments. The independent variables were the solvent composition (X 1 , %, v/v, ethanol/water), the solventmaterial ratio (X2 , ml/g) and the extraction time (X 3 , min) with quercitrin and TF yields as indicators. A fixed particle size (0.450.9 mm) was chosen. The quercitrin and TF yields of all runs are shown in [Table 1]. The multiple linear regressions using the secondorder polynomial model (Eq. (1)) were performed based on these results.{Table 1} An ANOVA of independent variables shown in [Table 2] indicates that all these three independent variables significantly affect the extraction efficiency (P < 0.05). The regression parameters of the fitted quadratic models with corresponding coefficients of multiple determinations (R 2) were shown in [Table 3]. Good fittings were achieved and the variability of the response was explained by the model. The high R 2 values, 0.94 and 0.96 for quercitrin and TF, respectively, imply the experimental data confirm the compatibility with the data predicted by the model. The low value of the coefficient of variation (1.92% for quercitrin, 3.28% for TF) indicates that the results obtained from the fitted model were reliable. The adjusted coefficient of determination (R 2 Adj.) value reconstructs the expression with all the *significant terms included and high value of the R 2 Adj. (R 2 Adj. equals to 87%, 92%, respectively) supports the significance of the model.{Table 2}{Table 3} The lackoffit testing was used to verify the adequacy of the model's fitness. ANOVA for the lackoffit test was not significant with regard to the quercitrin and TF yields ( P > 0.05), indicating that the model could adequately fit the experimental data [Table 4].{Table 4} Analysis of response surfaces Since; the models show good fittings, the response values are sufficiently explained by the regression equation. The response surface curves were plotted to demonstrate the interactions of the independent variables and to determine the optimal value of each independent variable for the maximum response. The relationship between independent and dependent variables was illustrated in threedimensional (3D) response surface plots and twodimensional (2D) contour plots. The 3D response surface and the 2D contour plots were provided as graphical representations of the regression equation [Figure 7] and [Figure 8]. Each contour curve represented an infinitive number of combinations of two independent variables while keeping the other independent variable at the stationary point. In that way, a 3D response surface can be readily visualized. [Figure 7]a shows the 3D response surface function developed by the model for solvent composition and solventmaterial ratio. The solvent composition demonstrated quadratic effects on the response. The content of quercitrin reaches the highest value when 65% ethanol is used. Similarly, the solventmaterial ratio demonstrated quadratic effects on quercitrin at the same time. The results of the effects of solvent composition and solventmaterial ratio on TFs are shown in [Figure 8]a. Solvent composition displayed a quadratic effect on the response yielding maximum between 60% and 70% ethanol. Similarly, when solvent composition was fixed with increase of solventmaterial ratio, TF increased gradually and reached the highest value when the solventmaterial ratio was about 10:1.{Figure 7}{Figure 8} The effects of independent variables solvent composition and extraction time on the response values of quercitrin and TF yields are shown in [Figure 7]c, [Figure 7]d, [Figure 8]c, [Figure 8]d, respectively. The interactive relationship between the two independent variables can be easily understood by examining the contour plots generated by keeping the other independent variable, solventmaterial ratio, constant at the stationary point [Figure 7]d and [Figure 8]d]. An increase in the yields of quercitrin and TF were significantly achieved with the increases of solvent composition and extraction time. However, quercitrin and TF yields no longer increase when the independent variables exceeded certain values. An interaction of solvent composition and extraction time was not statistically significant in affecting the quercitrin yield, but it had significant (P < 0.01) effect on extraction of TF. [Figure 7]e and [Figure 8]e demonstrated the effect of solventmaterial ratio and extraction time on the yields of quercitrin and TF. Increases in quercitrin and TF yields were observed with the increases in extraction time and solventmaterial ratio, which suggests that extraction yields of quercitrin and TF were proportional to extraction time and solventmaterial ratio. The interaction of the two independent variables on extraction of quercitrin and TF from Herba Polygoni Capitati was not statistically significant. Verification experiments In order to verify the predictive capacity of the model, the optimum response variables were tested under the optimized conditions derived from ridge analysis of RSM. As a result, the observed values could be satisfactorily achieved within 95% confidence interval of the predicted values from the model [Table 5].{Table 5} Conclusions By using onefactoratatime approach coupled with CCD, ultrasoundassisted extraction parameters, including solvent composition, solventmaterial ratio and extraction time were optimized for the extraction of quercitrin and TF from Herba Polygoni Capitati. The results showed that the secondorder polynomial model gave a satisfactory description of the extraction data. The optimized conditions are as follows: Extraction method, sonication extraction; solvent, it might be good to round at 65% aqueous ethanol; particle size, 0.450.9 mm; solventmaterial ratio, round to 10:1; extraction time, round to 54 min. This study can provide fundament information and valuable insight into the industrial extraction processes of Herba Polygoni Capitati. Acknowledgments The authors gratefully acknowledge Guizhou Warmen Pharmaceutical Corporation for providing raw materials and the financial supports from Major Science and Technology Special Program of Guizhou Province (No. 2011 601902), Guizhou Provincial Program of Science and Technology Innovation Talent Team on Pharmaceutical Analysis (No. 2011 4008), Guizhou Provincial Special Program on Research and Development of Scientific and Technological Industry about Modernization of TCMs (No. ZY 2011 3013), the Characteristic Key Laboratory of Standardization on Traditional Chinese Medicines and National Medicines and the Characteristic Key Laboratory of Guizhou Province Education Department. (No. KY 2012 005). References


