Optimization of phenolic compounds extraction with antioxidant activity from açaí , blueberry and goji berry using response surface methodology

*Corresponding author: Solange Teresinha Carpes, Department of Chemistry, Federal Technological University of Paraná (UTFPR), Pato Branco, PR, Brazil. E-mail: carpes@utfpr.edu.br Received: 22 June 2017; Accepted: 24 February 2018; R E G U L A R A R T I C L E


Optimization of phenolic compounds extraction with antioxidant activity from açaí, blueberry and goji berry using response surface methodology
Goji berry, a fruit originated from Asian countries such as China and India, has been used for many years in herbal medicine (Carnés et al., 2013;Donno et al., 2015).This fruit is considered a functional food of great importance for the China and has become increasingly common in the Europe and North America (Li et al., 2007, Dong et al., 2009;Carnés et al., 2013).Goji berry has gained prominence in recent years in the scientific community due to its anti-inflammatory (Potterat, 2010;Nardi et al., 2016), antioxidant (Amagase and Farnsworth, 2011;Donno et al., 2015) and antitumor (Wawruszak et al., 2016) activities.In addition, the Gojy berry may be effective in prophylaxis of diseases, such as diabetes, and cardiovascular diseases (Kulczyński and Gramza-Michałowska, 2016).
Given the economic and nutritional importance of these species, some studies regarding the extraction process and optimization of extraction conditions are necessary.Obtaining biologically active compounds involves many factors and the experimental design is an adequate methodology to experimentation, which allows the reduction in the number of the tests without prejudice to the quality of the information.Thus, the objective of this study was to investigate the influence of different factors (time, temperature and nature of solvent) on phenolic compounds extraction present in açaí, blueberry and goji berry fruits, through RSM.The chromatographic profile analysis and antioxidant activity of the best extract condition were aim of this work, too.

Samples
Samples of açaí, blueberry and goji berry were obtained in the Municipal Market of Curitiba -Paraná, Brazil.The fruits were lyophilized (Liotop -L1019, São Paulo, Brazil) and ground in an analytical mill and stored at -12ºC.

Preparation of extract and experimental design
The 2³ factorial design was used to evaluate the effect of solvent, temperature and time over phenolic compounds extraction.The design was composed of eight trials performed in triplicate.Independent variables were: ethanol 800 g/L and pure water as solvent (variable X1), extraction time of 30 min and 60 min (variable X2), and extraction temperature of 30 °C and 60 °C (X3 variable) while dependent variables were antioxidant activity (AA) and total phenolic compounds (TPC).
The extraction of antioxidant compounds were performed according to Ribeiro et al. (2007).Samples containing 3 g of lyophilized fruits were subjected to the extraction process with 30 mL of solvent in a shaker (Solab SL222, Piracicaba, Brazil) according to the experimental design shown in Table 1.The extracts were transferred to Falcon tube and centrifuged at 447 x g for 15 min (Hermle Z 200 A, Wehingen, Germany).After filtrating, the supernatants were stored in a freezer at -12 °C.Each extraction was performed in triplicate amounting to 24 trials for each fruit.

Total phenolic compounds assay
Total phenolic compounds (TPC) were quantified by the Folin-Ciocalteu method described by Singleton et al. (1999) using gallic acid as the standard.

DPPH radical scavenging assay
Measurement of DPPH scavenging activity was performed according to the methodology described by Brand-Williams et al. (1995).The results were expressed as EC 50 (concentration required to obtain a 50% antioxidant effect) and µmol Trolox/g sample.Absorbances were read in spectrophotometer (Bel Photonics 2000, Piracicaba, Brazil) at 517 nm and the tests were performed in triplicate.

ABTS (2,2'-Azino-bis (3-ethylbenzthiazoline-6sulphonic acid)) assay
The antioxidant activity by the ABTS method was performed according to Re et al. (1999), in UV/Vis spectrophotometer (Bel Photonics 2000, Piracicaba, Brazil) at 734 nm.Trolox was used as reference and the results of the antioxidant activity were expressed as μmol of trolox equivalent antioxidant capacity (TEAC)/g of sample.All analyses were carried out in triplicate.

Coupled oxidation of β-carotene and linoleic acid assay
The measure of antioxidant activity was determined by the coupled oxidation of β-carotene and linoleic acid according to Ahn et al. (2004).Emulsion oxidation was spectrometrically monitored (Bel Photonics 2000, Piracicaba, Brazil) and the absorbance was read at 470 nm, at time zero (t = 0) and subsequently after every 20 min, until the characteristic color of β-carotene disappeared in the control reaction (t = 100 min).The antioxidant activity was determined as percent inhibition relative to control sample.

Ferric reducing antioxidant power (FRAP) assay
Antioxidant activity determination using the FRAP method was performed as described by Pulido et al. (2000).The absorbance was measured in UV/Vis spectrophotometer (Bel Photonics 2000, Piracicaba, Brazil) at 593 nm.Aqueous solutions of ferrous sulfate were used for calibration, and the results were expressed as µmol of Fe 2+ /g of the sample.All analyses were carried out in triplicate.

HPLC-DAD-UV-Vis profile
The extracts obtained under optimum conditions as indicated by the factorial design were rotary evaporated (Fisatom® 802, Sao Paulo, Brazil) and lyophilized (Liotop -L1019, São Paulo, Brazil).Aliquots 10 μL at 0.1 g/mL of extract concentration were injected into a chromatograph coupled to a photodiode array (PDA) detector (Varian, 920-LC, Walnut Creek, US) using a C18 RP (250 x 4,6 mm, 5 μm) column.For the fractionation, a binary gradient system was used in which the mobile phase "A" consisted of ultrapure water and phase "B" consisted of acetonitrile, both containing 0.2 mL/L acetic acid.The gradient started with 5 % of B up to 95% of B in 36 min and returned to the initial condition.Total analysis time was 45 min at a flow rate of 1 mL/min and the temperature during the analysis was maintained at 30 °C.Calibration curves and linear regression based on the peak areas were used to identify and quantify peaks corresponding to the phenolic compounds.The identification was performed by comparison of retention times and absorption in ultraviolet at wavelengths of 280 nm and 320 nm.These calibration curves were obtained using external standards of catechin, caffeic acid, ferulic acid, epicatechin, coumaric acid, gallic acid, rutin, and myricetin.All standards were dissolved in phase "B" at the following concentrations: 2 µg/mL; 4 µg/mL; 8 µg/mL; 16 µg/mL; 30 µg/mL.These concentrations were used afterwards for obtaining the limit of quantification (LQ) of 0.35 µg/mL and the detection limit (LD) of the equipment of 0.12 µg/mL employing standards of gallic acid, vanillic acid, caffeic acid, coumaric acid, and ferulic acid according to Oldoni et al. (2015).The content phenolic compounds were expressed for each compound in µg/g of sample.Determination of the phenolic compounds by HPLC was performed in triplicate.

Statistical analysis
The set of data and contents of total phenolic compounds derived from the factorial design were analyzed by response surface methodology (RSM).The anthocyanin content and the antioxidant activity were carried out by principal component analysis (PCA).The data were processed by one-way analysis of variance (ANOVA).The averages were compared by Tukey test, considering the significance level of 95% (p<0.05), using the STATISTICA program 8.0 version (StatSoft, USA).Past Software 3.07 developed by Hammer et al. (2001) expressed the reproducibility of the results as Pooled Standard Deviation (Pooled SD).The global response (GR) was carried out over the data set too.GR was calculated according to equation ( 1): Where: R(x n ) is the response for each dependent variable, MR(x n ) is the maximum value of response for each dependent variable.All experiments were carried out three times.

RESULTS AND DISCUSSION
Total phenolic compounds (TPC) content and antioxidant activity (AA) in fruit extracts after being subjected to  1).
Our results showed lower TPC values than found by Donno et al. (2015) in the goji berry from Alzate di Momo in the Northern Italy, who found phenolic compounds levels ranging from 255.87 to 281.81 mg GAE/g sample.
These differences are usually caused by the great diversity of chemical compounds that can be extracted depending on the methodology used.
The wide range of studies related to conditions and extraction methods of phenolic compounds and antioxidant activity makes it difficult to perform an effective comparison.In addition, the large differences in TPC and AA values observed in these studies can be also attributed to different climactic conditions during cultivation of the fruits due to the different regions where each one fruit was planted.The harvest season is also an important factor for comparison.
The analysis of variance (ANOVA) of the dependent variable; TPC and AA from açaí, blueberry and goji berry subjected to extraction conditions presented by the factorial design is shown in Table 2.The factors time, temperature and solvent nature in the extraction of phenolic compounds and AA were significant (p <0.05) for all fruits.The analysis of variance, F calc for all response variables (TPC and AA) was always greater than F tabulated (4.45), and in some cases, this amount was about 80 times greater than F tabulated (Table 2).The lower ratio found in F calc by F tabulated in the analysis of variance was 5.85.These results showed that the empirical data were adequately adjusted to the proposed models.
The global response (GR) showed that the kind of solvent was the mostly independent variable to extract the phenolic compounds from açaí, blueberry and goji berry.The effect estimated for solvent type was 1.34, while for extraction time and extraction temperature were 0.52 and 0.51, respectively.All effect estimated were significant (p value < 0.0000).Thus the best conditions for phenolic compounds extraction with antioxidant activity from fruits was found when the ethanol was used as extractor solvent in the extraction temperature at 60°C during 60 minutes (Table 3).
The entire data set was adjusted by multiple linear regression and six linear mathematical models were generated established by factorial design and response surface methodology (RSM) (Table 4).The TPC and AA answers were significant at the 5% level for each fruit (Table 4).For each model, we observed a high coefficient correlation varying from 0.81 to 0.98, thus 81 to 98% of data can be explained by the proposed models (Table 4).
Each multiple linear regression model generates a response surface, where the dependent variables (TPC and AA) are shown on the z axis as a function of the independent variables; solvent nature, extraction time and extraction temperature (Figs. 1 to 3).High values TPC from açaí were obtained with pure water as solvent at 60 °C during 60 min (Fig. 1a).On the other hand, ethanol (800 g/L) at a temperature of 60 °C and for 60 min was the best extraction condition for phenolic compounds with high antioxidant activity (Fig. 1b).
The best solvent to extract goji berry TPC was pure water (Fig. 2a), however, this solvent was not the best option for extracting phenolic compounds with antioxidant activity, in this case ethanol 800 g/L obtained better results (Fig. 2b).
The variables time (60 min) and temperature (60 °C) had a positive effect on TPC and AA (Fig. 2

ab).
The best extraction conditions of blueberry phenolic compounds were ethanol 800 g/L at 60 °C during 60 min, showing the most extreme time and temperature conditions led to an increased TPC extraction (Fig. 3a) with greater AA (Fig. 3b).Generally, this same result was observed with the other fruits.Thus, the optimum extraction condition for phenolic compounds with high antioxidant activity was obtained with the ethanol (800 g/L) at 60 °C during 60 minute of extraction.The principal component analysis (PCA) was performed on the data set of anthocyanin and antioxidant activity levels.Two major components were identified, with 99.50% of explained variance, PC1 for 88.16% variation, and PC2 for 11.33% (Fig. 4ab).There was the formation of three groups represented by the analyzed fruits (açaí, blueberry and goji berry) and distributed in the quadrants of the PCA scores chart (Fig. 4a).
The first group is represented by açaí (second quadrant Fig. 4b), which has a high antioxidant activity by ABTS, β-carotene and FRAP.Blueberry extract (third quadrant Fig. 4b) showed high anthocyanin content.The other group, also as sole representative was the goji berry group (quadrants 1 and 4), which showed high EC 50 values.
The two most important dependent variables in groups formation and occurrence were the antioxidant activity by FRAP and EC 50 , taking into account their commonalities.
The least important variable in group classification was anthocyanin content.Groupings of açaí, blueberry and goji berry fruit extracts because of their antioxidant activities and anthocyanin content could be analyzed by PCA.In addition, ANOVA test and PCA confirmed the anthocyanin and antioxidant activity results significantly contributing to improve the knowledge of this three fruits analyzed.

Fruit extracts chemical characterization and antioxidant activity in the best extraction condition
The extracts of the three fruits in the best extraction condition (ethanol at 60 °C and 60 min of extraction), A8 assay, were analyzed for total anthocyanin, AA by four different methods (EC 50 , FRAP, ABTS and β-carotene) and phenolic compounds profile by HPLC/DAD.
Goji berry had the lowest antioxidant activity in all tests compared to açaí and to blueberry (Table 5).
Antioxidant activity of sodium erythorbate was used as positive control and for comparison with the samples  with values of 0.05 ± 0.01 mg/mL, 2871.97 ± 26.97 µmol Fe 2+ /g sample, 4777.93 ± 206.00 µmol Trolox/g sample and 77.27 % ± 0.07, through the EC 50 , FRAP, ABTS and β-carotene method, respectively.These values were higher than those of the analyzed samples (Table 5), once sodium erythorbate -a chemical preservative much used in fruit products -is a pure substance when compared with fruits extract, which contain many other extracted compounds that can interfere with the chemical analysis.The highest antioxidant activity performed through DPPH method and expressed as EC 50 (concentration of extract necessary to decrease the initial concentration of DPPH by 50%) was found in blueberry (0.50 mg/mL), and calculated through the line equation y = 79.141x+ 10.647 (R² = 0.990).
ABTS radical scavenging performed using the line equation y = -11.244x+ 0.689 (R² = 0.997) ranged from 10.675 µmol Trolox/g (goji berry) to 15.285 µmol Trolox/g (açaí) (Table 5).These results were similar to those reported by Santos et al. (2008)  The reducing activity power of Fe 3+ to Fe 2+ (FRAP) of fruit extracts was calculated using the line equation y = 32.991x-0.050 with R² of 0.999 and ranged from 47.54 to 106.35 µmol Fe 2+ /g sample (Table 5).Açaí and goji berry showed the highest and lowest values, respectively (Table 5) and differed statistically (p<0.05).The antioxidant activity by coupled β-carotene/ linoleic acid method at a concentration of 0.01 g/mL for blueberry and açaí samples and at 0.025 g/ml for goji berry, ranged from 74.66 to 47.14%.The highest values was found in açaí extract and smaller values in goji berry extract (Table 5).Donno et al. (2015) found 20.89 µmol Fe 2+ /g in the goji berry from Italy and was lower than those found in this study (Table 5).Pertuzatti et al. (2014) achieved antioxidant activity of 60.9% by the β-carotene/linoleic acid method in the blueberry extracts (0.46 g/mL concentration) with similar results of this study (Table 5).
The quantification of the phenolic compounds by highperformance liquid chromatography in the açaí, blueberry and goji berry extracts was performed in different wavelengths from 277 to 371 nm.Retention times, wavelengths and regression equation data of compounds are given in Table 6.It was possible to visualize the presence of catechin and rutin flavonoids in the three studied fruit extracts (Table 6).Catechin, epicatechin, rutin and myricetin were found in açaí, among them myricetin (0.054 mg/g) and rutin (0.001 mg/g) were present in higher and lower concentration, respectively.No phenolic acids were found in the açaí extracts (Table 6).
Studies for identification and quantification of phenolic compounds from fruits are common (Castrejón et al., 2008;Yuyama et al., 2011;Borges et al., 2016), however, the method used in extraction, equipment, identification and separation conditions are different.Furthermore, the wide species diversity and crops types make it relatively difficult to compare these studies (Rodrigues et al., 2011;Melo et al., 2015).Donno et al. (2015) found caffeic acid, coumaric acid, ferulic acid and epicatechin in goji berry from Italy, while Wang et al. (2010) in their studies found another composition in the same fruit (rutin, quercetin and caffeic acid).The phenolic compound contents found by these authors were higher than those found in this study (Table 6).Gordon et al. (2012) in their studies with açaí had previously identified, compounds like gallic acid, caffeic acid, and vanillic acid.

CONCLUSION
Experimental design and RSM could be used to optimize extraction of phenolic compounds from açaí, blueberry and gojy berry for maximizing the antioxidant capacity.Further, the use of global response was very useful for simplifying and improving the phenolic compounds extraction performance with high antioxidant activity.In the best extraction condition it was possible to extract phenolic compounds such as phenolic acids and flavonoids with high antioxidant activity.The açaí, blueberry and gojy berry extracts obtained in this study can be a potential source of the phenolic compounds for food technology application, as natural antioxidant alternatives.

Fig 1 .
Fig 1. Reponse surface of the TPC (1a) in mg GAE/g of acai and AA (1b) in µmol Trolox/g of acai as a function of time and solvent nature

Fig 2 .
Fig 2. Reponse surface of the TPC (2a) in mg GAE/g of goji berry as a function of time and solvent nature and AA (2b) in µmol Trolox/g of goji berry as a function of temperature and solvent nature

Fig 3 .
Fig 3. Reponse surface of the TPC (3a) in mg GAE/g of blueberry and (3b) in µmol Trolox/g of blueberry as a function of temperature and solvent nature

Fig 4 .
Fig 4. PCA plot Antioxidant activity and total anthocyanin on the acai, blueberry and goji berry a) scores; b) loding plot for the antioxidant actitivty and total anthocyanin on prinicipal 1 and 2

Table 1 : Factor delineation for total phenolic compounds and antioxidant activity in açaí, blueberry and goji berry
The experiments were performed in duplicate and the results expressed as mean (n=2); Runs: assay A1 to A8 -combination of solvent, time and temperature; variable x1: solvent; variable x2: temperature (°C); variable x3: time (min); Values followed by different letters in the same column are significantly different (P<0.05).TPC: Total phenolic compounds; AA: Antioxidant activity (DPPH method).deMoura,etal.varioustreatments (A1 to A8) according to factorial design are summarized in Table1.Extracts obtained a specific and characteristic coloring for each fruit, differing according to the solvent extractor.These variations can be caused by the diversity of compounds each solvent can extract.It is depending on their polarity, thus providing different colors for each respective extract.
Paz et al. (2015))(Table1).In this study, the results of TPC and AA from açaí were lower than those reported byKang et al. (2012)in açaí from Para State, Brazil.The values ranged from 31.2 to 73.0 mg GAE/g sample and 133.40 to 320.30 µmol of trolox/g sample, respectively.However,Paz et al. (2015)showed levels of 1.81 mg GAE/g sample and 1.57 mg of Trolox/g açaí from Brazil's Amazon Forest, for phenolic compounds and antioxidant activity, respectively.the largest AA values for this fruit extract was found in assay A8, with 19.40 µmol of Trolox/g sample (Table

Table 3 : Analysis of variance of global response in fruits subjected to treatments according to a Fractional Design Source of variation Dependent variable: global response SS
* Values in italic refer to significant differences, F 0.05;5;17 =2.81

Table 6 : Phenolic profile determined by HPLC in açaí, blueberry and goji berry extracts
Detection and quantification limits were defined as the concentration of 0.12 µg/mL and 0.35 µg/mL, respectively TR: Retention time; RE: Regression equation;