Ultrasound assisted extraction of bioactive compounds from Nephelium lappaceum L. fruit peel using central composite face centered response surface design

2.1 Plant materials

Fresh fruits of rambutan (N. lappaceum L.) at the commercial mature stage were procured from a commercial fruit market near Chennai, India on May 2012. Fruits were selected based on their uniformity in shape and color. The fruits were washed thoroughly in potable water and then air-dried. Thick layer of skin and thorns were peeled off from the fruits manually. Fruit peels were washed thoroughly in running tap water to get rid of adhered impurities on the surface. The fruit peels were dried in a vacuum oven at 40 °C until to obtain the constant weight. The dried peels were pulverized and sieved through a 40-mesh sieve to obtain the powdered samples. Powder samples (moisture content 12–14%) were stored in dark bags to prevent oxidation of active compounds due to sun light and kept in dry environment prior to the experiments.

2.2 Chemicals and reagents

Cyanidin-3-glucoside (C-3-G), gallic acid (GA) and rutin were purchased from Sigma chemicals, Mumbai. Folin–Ciocalteu reagent, sodium carbonate, Aluminum chloride, sodium nitrite and sodium hydroxide were obtained from Loba chemicals, Mumbai. All the chemicals used in this study are of analytical and HPLC grade.

2.3 UAE of pigment and polyphenols

UAE was performed according to the method described by Ying et al. (2011) in an ultrasonic bath (Power sonic, Korea) equipped with digital sonication power, time and temperature controller with a useful volume of 10 L (internal dimensions: 30 × 24 × 15 cm) to carry out the extraction. About 10 g of ground powder was mixed with an appropriate quantity of distilled water. Triplicate experiments were carried out according to Table 1. After extraction, the extracts were centrifuged at 2600g for 15 min (Remi R-24 Centrifuge, India) and filtered through a filter paper (Whatman No. 1, England). The obtained supernatants (extracts) were collected in a screw-capped dark glass container and stored in the refrigerator until further analysis.

2.4 Total anthocyanin content

The total anthocyanin content (TAC) was estimated by a pH differential method which relies on the structural transformation of the anthocyanin chromophore as a function of pH value (Giusti and Wrolstad, 2000), which can be measured using a spectrophotometer (Shimadzu UV-1800, Kyoto, Japan). Aliquots of extracts were brought to pH 1.0 using potassium chloride (0.025 M) and pH 4.5 using sodium acetate (0.4 M) and allowed to equilibrate for 1 h. The absorbance was recorded at 530 and 700 nm using a spectrophotometer calibrated with distilled water as the blank. The difference in absorbance between pH values and wavelengths was calculated using the formula:

2.5 Total phenolic content

The total phenolic content (TPC) was measured by Folin–Ciocalteu method as described by Vatai et al. (2009) with minor modifications. A known amount of (0.5 ml) diluted sample was mixed with 2.5 ml of Folin–Ciocalteu reagent (0.2 mol/L), and after 5 min, 2.0 ml of sodium carbonate (7.5 g/100 ml) was added. The mixture was kept at room temperature for 2 h after shaken and absorbance of the mixture was measured at 754 nm against a reagent blank (0.5 ml distilled water instead of the sample) with a UV–visible spectrophotometer (Elico, SL 244, India). Results were calculated on the basis of the calibration curve of gallic acid and expressed as gallic acid equivalents (mg GAE/100 g). All samples were analyzed in triplicates and the average values were calculated.

2.6 Total flavonoid content

Aluminum chloride assay method (Zhishen et al., 1999) was used to determine the total flavonoid content (TFC) of the samples. A known volume of extract (0.5 ml) was mixed with distilled water to make 5 ml solution, and 0.3 ml 5% NaNO₂ was added. After 5 min, 0.3 ml 10% AlCl₃ was added. At 6th min, 2 ml 1 M NaOH was added and the total volume was made up to 10 ml with distilled water. The solution was mixed well and the absorbance was measured at 510 nm. Total flavonoid content was calculated on the basis of the calibration curve of rutin and expressed as rutin equivalents (mg RE/100 g). All samples were analyzed in triplicates and the average values were calculated.

2.7 Experimental design

Box-Wilson design, also called central composite design (CCD), is an experimental design used to achieve maximal information about a process from a minimal number of experiments (Yang et al., 2009). In CCD, the central composite face-centered (CCFC) experimental design was used in this study to determine the optimal conditions and study the effect of four variables (extraction temperature (°C), power of ultrasound (W), extraction time (min) and solid–liquid ratio (g/ml)) on three responses (total anthocyanin content (TAC), total phenolic content (TPC) and total flavonoid content (TFC)) of ultrasound-assisted N. lappaceum L. peel extracts. From the preliminary experimental results, process variables and their ranges (extraction temperature (30–50 °C), power of ultrasound (20–40 W), extraction time (10–30 min) and solid–liquid ratio (1:10–1:20 g/ml)) were determined. After selection of independent variables and their ranges, experiments were established based on a CCFC design with four factors at three levels and each independent variable was coded at three levels between −1, 0 and +1. The coding of the variables was done by the following equation (Prakash Maran et al., 2013):

In this study, total number of 30 experiments (consisting of 16 factorial points, eight star points and six replicates at the center points (in order to allow the estimation of pure error)) were carried out and the total number experiments were calculated from the following equation (Azargohar and Dalai, 2005):

The experimental sequence was randomized in order to minimize the effects of unexpected variability in the responses due to extraneous factors. A second-order polynomial equation was used in order to develop an empirical model which correlated the responses to the independent variables. The general form second order polynomial equation is

2.8 Statistical analysis

The statistical analysis was performed using Design Expert Statistical Software package 8.0.7.1 (Stat Ease Inc., Minneapolis, USA). The experimental data were analyzed by multiple regression analysis through the least square method. Two different tests namely sequential sum of squares and model summary statistics were carried out on the experimental data in order to find out adequacy of various models. The regression coefficients of all the terms (linear, quadratic, and interaction) involved in the model and their effect were analyzed by Pareto analysis of variance (ANOVA) and ANOVA tables were generated. All the terms of the model were tested and verified statistically by F-test at probability levels (p ⩽ 0.05). Adequacy of the developed models was tested by performing coefficient of determination (R²), adjusted coefficient of determination ( $R_{\text{adj}}^2$ ) and predicted coefficient of determination ( $R_{\text{pre}}^2$ ). After fitting the models, surfaces and contour plots were constructed to predict the relationship between the independent variables and responses.

2.9 Optimization

A numerical optimization technique (Derringer’s desired function method) was employed for optimizing the various responses simultaneously involved in the UAE process. The desirability function method was applied for generating optimum conditions having some specific desirability value. This optimization technique depends on whether a particular response (Y_i) is to be maximized or minimized or targeted based on the requirement of the process, while the independent variables are kept within the range. The general approach is to first convert each response (Y_i) into a dimensionless individual desirability function (d_i) and it was done with the following equation

In present study, desirability functions were developed for the following responses: maximum total anthocyanin, phenolic and flavonoid content. A weight factor, which defines the shape of the desirability function for each response, is then assigned. Weights must be between 1 and 10, with larger weights corresponding to more important responses. A weight factor of 1 was chosen for all individual desirability’s in this work. The “importance” of a goal can be changed in relation to the other goals. It can range from 1 (least important) to 5 (most important). The default is three representing all goals to be equally important.

2.10 Validation of optimized conditions and predictive models

The suitability of the developed model equations for predicting the optimum response values was verified using the optimal condition. Triplicate experiments were carried out in the optimal condition and the mean experimental values were compared with the predicted values in order to determine the validity of the models.

3.1 CCFC design analysis

The experimental data were analyzed and fitted to the four high degree polynomial models viz., linear, interactive (2FI), quadratic and cubic models. Two different tests namely the sequential model sum of squares and model summary statistics were carried out in this study to decide about the adequacy of models among various models to represent the total anthocyanin content, phenolic content and flavonoid content of the extracts and the results are listed in Table 2.

The adequacy of models tested output indicates that, cubic model was found to be aliased. The linear and quadratic models are found to have lower p-value (<0.0001). But the R², adjusted R² and predicted R² values were found to be low in linear model (Table 2). Thus the linear and cubic models cannot be chosen for further modeling of experimental data. Quadratic model was not only found to have maximum R², adjusted R², predicted R², but also exhibited low p-values. Hence the quadratic model was chosen for further analysis.

3.2 Development of second order polynomial models

Second-order polynomial equation with interaction terms was fitted to the experimental results to develop mathematical model, which will help to predict the extraction efficiency of different sets of combinations of four process variables on the responses. Three empirical models were developed from this study to predict the UAE efficiency of TAC, TPC and TFC from rambutan fruit peels. The final model obtained in terms of coded factors is given below

3.3 Statistical analysis

Multiple regression analysis and Pareto analysis of variance (ANOVA) were used to check the adequacy and fitness of the developed models and results of ANOVA are given in Table 3. Table 3 summarized the effects of the model terms and associated p values for all three responses. At a 95% confidence level, a model was considered significant, if the p value <0.05. The sign and value of the quantitative effect represent tendency and magnitude of the term's influence on the response, respectively. A positive value in the regression equation exhibits an effect that favors the optimization due to synergistic effect, while a negative value indicates an inverse relationship or antagonistic effect between the factor and the response (Verma et al., 2009). From the ANOVA table, it was found that the developed models for all the responses were highly significant at probability level p < 0.0001 and also showed higher Fisher F-value (667.71 for TAC, 1017.95 for TPC and 890.68 for TFC) with significant p-value (p < 0.0001), which indicates that most of the variation in the response could be explained by the developed models.

Determination coefficient (R²), adjusted determination coefficient ( ${\text{R}}_{\text{adj}}^2$ ), predicted determination coefficient ( ${\text{R}}_{\text{pre}}^2$ ) and coefficient of variation (CV%) were calculated to check the goodness and behavior of the developed models with the experimental data. A high R² coefficient ensures a satisfactory adjustment of the quadratic model to the experimental data. The calculated R² (0.998 for TAC, 0.999 for TPC and 0.998 for TFC), ${\text{R}}_{\text{adj}}^2$ (0.996 for TAC, 0.998 for TPC and 0.997 for TFC) and ${\text{R}}_{\text{pre}}^2$ (0.987 for TAC, 0.991 for TPC and 0.992 for TFC) values were high and advocate a high correlation between the observed and the predicted values. The analysis shows that the form of the model chosen to explain the relationship between the factors and the response is well-correlated and proved that the developed models defined exactly the true behavior of the process. The low CV values (1.59, 1.29 and 1.34) clearly indicated that, the deviations between experimental and predicted values are low and also showed a high degree of precision and a good deal of reliability in conducted experiments. Adequate precision is greater than four is desirable (Prakash Maran et al., 2013b) and the ratio was found to be >95, which indicates an adequate signal and confirms that this model can be used to navigate the design space.

3.4 Adequacy of the models

In general, the exploration of fitted response surface models may produce poor or misleading results, unless the model exhibits a good fit, which makes checking of the model adequacy essential (Prakash Maran et al., 2013a). Diagnostic plot such as predicted versus actual values (Fig. 1) is used to evaluate the relationship and model satisfactoriness between experimental data and predicted data obtained from the developed models. From Fig. 1, it was observed that, the data points lie very closely to the straight line. It exhibited high correlation between the experimental data and predicted data obtained from the models.

Close

Figure 1 Diagnostic between experimental and predicted values for TAC (A), TPC (B), and TFC (C).

Export to PPT

Data were also analyzed to check the residuals. The residual gives the difference between the observed value of a response measurement and the value that is fitted under the theorized model. Small residual value indicates that model prediction is accurate (Herbach et al., 2004). By constructing internally studentized residuals plot, a check was made to analyze the experimental data and to find out the satisfactory fit of the developed models and the plots are shown in Fig. 2.

Close

Figure 2 Internally studentized residuals plot for TAC (A), TPC (B), and TFC (C).

Export to PPT

Normal probability plot represents the normal distribution of the residuals and constructing a normal probability plot of the residuals allows the assessment of the normal distribution of the data. The normal probability plots of the residuals are shown in Fig. 3 and the data points on this plot lie reasonably close to a straight line indicating an acceptable fit.

Close

Figure 3 Normal probability plot for TAC (A), TPC (B), and TFC (C).

Export to PPT

3.5 Effect of process variables

In the present study, four factors at three level CCFC design used to study the influence of process variables such as extraction temperature, power of ultrasound, extraction time, and solid–liquid ratio were investigated on the UAE extraction of pigment and polyphenols from rambutan fruit peels. The three dimensional response surface and contour plots were constructed from the developed models. Response surface and contour plots are graphical representations of a regression equation that illustrate the main and interactive effects of independent variables on a response variable. These graphs are drawn by maintaining two factors constant (in turn at its central level) and varying the other two factors in order to understand their main and interactive effects on the dependent variables (Prakash Maran et al., 2013). These graphs are easy to understand and represent the interactions between pairs of independent variables on the responses and also used to locate their optimal levels.

3.5.1

3.5.1 Effect of extraction temperature

The effect of extraction temperature on the yield of anthocyanin and polyphenols was studied. The results showed that, the extraction temperature exhibited a positive linear and quadratic effect (Table 3) on the yield. When the temperature was increased from 30 to 50 °C, the extraction yield of pigment and polyphenols was increased (Fig. 4), which is due to the high number of cavitational nucleus formed during higher extraction temperatures as a result of high cavitation threshold which is responsible for acoustic cavitation. The relative greater force ruptured the cavitational nucleus and disrupted the cell tissues during extraction, which will enhance the mass transfer (Toma et al., 2001). However, higher temperature could enhance the solubility of pigment and phenolic compounds in rambutan fruit peel and decrease the viscosity and density of the extracts (Chen et al., 2012). Reduced viscosity and density facilitate the solvent penetration deeper into the sample matrix which in turn enhances the extraction efficiency by exposing more surface area of the sample to the solvent used.

Close

Figure 4 Response surface plots representing the effect of extraction temperature on the responses.

Export to PPT

3.5.2

3.5.2 Effect of extraction power of ultrasound

The efficiency of extraction power of ultrasound on the anthocyanin and polyphenols was evaluated and the results showed that, the yield increased with the increasing power of ultrasound (Fig. 5). The increase in extraction power facilitates the disruption of the cell walls of the peels, enhances the solubility of the compounds present in the peels and increases the extraction yield (Ying et al., 2011). This enhancement in extraction with ultrasound could be attributed to the ultrasonic effects such as micro jet formation and acoustic streaming (Sivakumar et al., 2009). Moreover, the intensity of ultrasound transmitted to the medium is directly related to the vibration amplitude of sonication, producing greater number of cavitation bubbles and therefore increased extraction efficiency (Dash et al., 2005).

Close

Figure 5 Response surface plots representing the effect of power of ultrasound on the responses.

Export to PPT

3.5.3

3.5.3 Effect of extraction time

The pigment and polyphenols' yield was increased when the duration was maintained from 10 to 20 min but slowly decreased when the duration continued to be extended (Fig. 6). Most of the anthocyanin and polyphenols in broken cells are released at the early period of extraction, because ultrasound enhanced the release of those compounds into the exterior solvent and increased the yield in the first 20 min. However, longer extraction time with ultrasound treatment might induce the degradation of pigment and polyphenols (Tiwari et al., 2009).The number of cavitation micro-bubbles created by ultrasound increased with the duration extended. The asymmetric collapse of micro-bubbles near surfaces was also associated with micro-jets that could scour surfaces and damage substance in solution (Vilkhu et al., 2008). The structure of pigment and polyphenols was destroyed and its stability decreased because of continual collapse of micro-bubbles. The results are in agreement with Rostagno et al. (2007), who clearly indicated that 20 min of sonication time was sufficient enough to extract phenolics from soy beverages.

Close

Figure 6 Response surface plots representing the effect of extraction time on the responses.

Export to PPT

3.5.4

3.5.4 Effect of solid–liquid ratio

As shown in Fig. 7, the yield of anthocyanin and polyphenols was increased gradually when the solid–liquid ratio ranged from 1:10 to 1:20 (g/ml). In general, a higher solvent volume can dissolve target compounds more effectively and result in a better extraction yield. The concentration difference between the interior plant cell and the exterior solvent was caused by the higher solid–liquid ratio, which could dissolve the constituents (anthocyanin and polyphenols) more effectively leading to an enhanced mass transfer rate (Ying et al., 2011) and increase the extraction yield. The efficiency of ultrasonication can be explained by the fact that sonication simultaneously enhances the hydration and fragmentation processes while facilitating mass transfer of solutes to the extraction solvent, without causing significant decomposition of the solvent (Toma et al., 2001).

Close

Figure 7 Response surface plots representing the effect of solid–liquid ratio on the responses.

Export to PPT

3.6 Determination and validation of optimum conditions

An optimum condition for the UAE of total anthocyanin and total polyphenols from rambutan fruit peel was determined to obtain maximum anthocyanin, phenolic and flavonoid content. Second order polynomial models obtained in this study were utilized for each response in order to obtain specified optimum conditions. Simultaneous optimizations of the multiple responses were carried out using Derringer’s desirability function method. This function searches for a combination of factor levels that simultaneously satisfies the requirements for each response in the design. This numerical optimization evaluates a point that maximizes the desirability function. All the factors and responses with the respective high-limit and low-limit experimental region have to satisfy the criteria defined for the optimum operations as stated in Table 4. Applying the methodology of desired function, the optimum level of various parameters was obtained and it indicates that an extraction temperature of 50 °C, ultrasound power of 20 W, extraction time of 20 min and solid–liquid ratio of 1:18.6 g/ml give 10.17 mg/100 g of total anthocyanin content, 546.98 mg GAE/100 g of total phenolic content and 100.93 mg RE/100 g of total flavonoid content respectively with an overall desirability value of 0.927. These optimize conditions could be considered as optimum as well as feasible conditions.

The suitability of the model equations for predicting optimum response values was tested under the conditions (extraction temperature of 50 °C, ultrasound power of 20 W, extraction time of 20 min and solid–liquid ratio of 1:18.6 g/ml) as determined by Derringer’s desired function methodology. Experiments were carried out under the optimal conditions in order to compare the experimental results with the predicted values of the responses using the developed empirical model equation (Eqs. (11)–(13)). The experiments were conducted in triplicates and the average values are reported in Table 5. The mean values of the TAC, TPC and TFC obtained were compared with the predicted values. The experimental values were found to be in agreement with the predicted values and clearly indicated the suitability of the developed quadratic models. The results obtained through confirmation experiments indicate the suitability of the developed quadratic models and it may be noted that these optimal values are valid within the specified range of process parameters.