Joural of ood enginesng 101 (2010) 28-31 ELSEVIER Contents lists available at ScienceDirect sea Journal of Food Engineering journal homepage: www.elsevier.com/locate/ifoodeng Extraction of antioxidant compounds from Jabuticaba (Myrciaria cauliflora) skins: Yield, composition and economical evaluation Diego T. Santos, Priscilla C. Veggi, M. Angela A. Meireles * LASEFVDENTEA (Schoo of ood Engineering UNICAMP (University of Campinas). Monti Lobr, 8, 1308862 Campinas, SP. rai ARTICLE INFO AgSTRACT ‘ie ry Ieceved 28 December 2009 Received in een form 20 May 2010 ‘Accept 5 June 2010 Desi online Jone 2010 Keyword: Amhocyains Cost of manutacrring Excaction methods Juba Meira cutiora Obtaining an extract with high antioxidant activity using environmentally fiendly technologies and low cost raw materials sof grea interest Inthe present work, a combined extraction process developed by ‘ur research group involving ultrasound treatment and aptated solvent extraction was evaluated, This ‘method was compared in cerms of yield, composition, and economical feasibility to traditional extraction methods, including ultrasound assisted, agitated bed and soxhlet extraction with ethanol (aiifed oF ot), The proposed method maximizes the extraction of phenolic compounds with acceptable degrada tion of anthocyanin pigments from an unusual source: Brazilian jabuticaba(Myrckaria calor) skin ‘The use of ultrasonic radiation continuously supporting a main extraction process has demonstrated Increased performance but implies in high consumption of energy and consequently, mon the procedure described in this paper appears tobe a viable option because i diation and results in high antioxidant activity extracts, and the anthocyanin proflecortaborates litera ture data (cyanidin-3-ucoside and delphinidn-3-slucoside), However, ses shorter ulasoni (© 2010 Elsevier Le. All rights reserved 1 Introduction ‘The desire for a healthier diet allied with the increasing concern fof consumers over the use of synthetic additives in food has pushed the food industry to search for new sources of natural pig- -ments (Montes etal. 2005). Anthocyanins are a type of functional pigment responsible for a wide range of colors present in vegeta: bles, flowers, fruits, and derived products. It is known that anthocyanin pigments act as strong antioxidants and are ant inflammatory, with antimutagenic and cancer chemopreventive activities (Kong et al, 2003). These bioactive properties have al- ready been demonstrated in in vitro and in vivo studies (Galvano et al, 2004), and an increase of publications in this ield has been ‘observed in recent years. In a recent review paper, Santos and Meireles (2009) compiled the recent studies on the health-promoting properties of anthocy- nins. This review demonstrated that consumption of dietary phy~ tochemicals, of which anthocyanins represent a considerable patt, ‘may promote several health benefits: reduction in the risk of ear” diovascular diseases, diabetes and cancer; a protective effect against hepatic and gastric damage and collagen degradation; an increase of cognitive performance, et. Grape peels, grape by-products (constituted mainly by peels) and berries are well known for their antioxidant properties due ~F Caresponding autor. Te 055 19352140; fx: 655 1927884007. ‘mall adie! mevleses unkcampbr (M. Angela A Meiees). 0260.8774/5 sce ont mater © 2010 Esevier Ld. Al ights seve, ot 10.1016jfeodeng2010.06005 to the presence of anthocyanins and other phenolic compounds. Many studies have been done to extract and evaluate these com: Pounds on the industrial scale (Santos and Meireles, 2008). In Brazil another source seems promising: jabuticaba (Myrciaria cau- Iilor) is grape-like in appearance and texture, although its skin is thicker and tougher. This fruit has a dark purple to almost black skin color due to a high content of anthocyanins that cover a white {gelatinous flesh inside (Terci, 2004) As the extraction procedure is of great importance for obtaining natural colorants, different research groups have made an effort to develop an efficient extraction procedure. An efficient extraction should maximize anthocyanin recovery with minimal degradation and result in an extract with high antioxidant activity using envi onmentally ftiendly technologies and low-cost raw materials. For this purpose this paper aims to demonstrate a potential tec: nique to extract antioxidant compounds (anthocyanins and other phenolic compounds) from jabuticaba skins: utilizing a short ultra- sonic irradiation as pre-treatment before conventional solvent extraction. This method is proposed based on the fact that ultra- sonic irradiation facilitates the release of extractable compounds and enhances mass transport of the solvent from the continuous Phase into plant cells of target compounds (Lee and Row, 2006), mainly during the initial minutes (Zhang et al, 2009). Ethanol ‘was used in all experiments because it isa GRAS (generally-recog- hized-as-sale) solvent It is well known that many flavonoids and related phenolic, compounds contribute significantly to the total antioxidantPy 11 Somes ea Jura of Food Egiccing 101 (2010)24-31 activity of many fruits and vegetables (Luo et al, 2002). Given that some recent papers have demonstrated thatthe biological (antiox- ‘dant, antiradical, ec.) activities of jabuticaba skin extracts are due to their compositions (Reynertson etal, 2008) the stuly ofthe ef- fect of the extraction method on the extract antioxidant activity was evaluated. ‘To compare the effect ofthe extraction methed on the process n terms of yield, composition and economic feasibility, different extraction methods were carried out for comparison purposes: ultrasound assisted, agitated bed, and soxhlet extraction using eth anol and acidified ethanol as solvent. 2. Materials and methods 2.1. Plant material Jabuticaba fruits (Myrefaria caulfiora) harvested from 2 planta tion in the State of Sao Paulo, Brazil, were acquired from a fruit and vegetable market center (CEASA-Campinas, Brazil). Inmedi- ately after acquisition, the fruits were stored in the dark in a ‘domestic freezer (10°C) (Double Action, Metalfrio, S30 Paulo, Brazil) until sample preparation. Before extraction, the fruits were ‘manually peeted. 22. Extraction procedures 22.1, Ultrasound assisted extraction (UAE) ‘abuticaba skins were added to 50-cm* Erlenmeyer flasks and then mixed with different volumes of ethanol 99.5% (Ecibra, Santo ‘Amaro, Brazil) to give a feed (o solvent ratio of 1:10. immediately after the addition of the solvent, the flasks were sonicated in an tultrasonicator bath with 2 40-KHz frequency (81 W) (model T 1440, Thornton, Sio Paulo, Brazil) at room temperature for 2h. After ultrasound extraction, the solvent was separated from the plant residue by simple filtration and evaporated using a rotary ‘evaporator (Laborota. model 4001. Vertrie’. Germany), with vac~ ‘uum control (Heldolph Instruments Gmbh, Vertrieb, Germany) ‘and a thermostatic bath at 40°C. The extracts were stored (10°C) in the dark until analysis. 22.2, Agitated bed extraction (ABE) ‘The extraction was carried out at 30°C by placing jabuticaba skins into 125-cm? Erlenmeyer flasks containing ethanol (89.5%, Ecibra, Santo Amaro, Brazil) using a feed to solvent ratio of 1:10. Extractions were carried out in a shaker (model MA 420, Piracicab- 2, Brazil) with agitation (150 rpm) for 2h. After extraction, the sol- vent was separated from the plant residue and evaporated, and the ‘extract was stored as described before, 22.3, Combined UAE + ABE Erlenmeyer flasks (125-cm") containing jabuticaba skins and, ethanol 99.5% (Ecibra, Santo Amaro, Brazil (feed to solvent ratio ‘of 1:10) were sonicated for 10min at a frequency of 40 kHz at room temperature; afterwards, the flasks were incubated in a rotary shaker (150 rpm) at 30 °C for 2h. The solvent-plant residue separation, solvent evaporation and extract storage were done as described before. 224, Soxhlet extraction ‘Approximately 25 g of jabuticaba skins and 250 cm? of ethanol or acidified ethanol (acidified to pH 3 with HCI) (99.5% Ecibra, San- o Amaro, Brazil) were used (Ieed to solvent ratio of 1:10). The extraction was done in a soxhlet apparatus for 8 h, and after that the solvent was evaporated, and the extract was stored as de- scribed before. 23. Extract characterization 23.1. Antioxidant activity (4A) ‘The evaluation of antioxidant activity of the extracts was based on the coupled oxidation of f-carotene and linoleic acid. ‘The technique developed by Marco (1968) consisted of measuring the bleaching of f-carotene resulting from oxidation by the deg- radation products of linoleic acid. One milligram of f-carotene (97%, Sigma-Aldrich, St. Louis, USA) was dissolved in 10 cm? of chloroform (89 %, Ecibra, Santo Amaro, Brazil). The absorbance was tested after adding 0.2 cm’ of the solution to 5 cm? of chio- roform, then reading the absorbance of this solution at 470 nm using @ UV-Vis. spectrophotameter (Hitachi, model U-3010, To- kyo, Japan). A reading between 0.6 and 0.9 indicated a workable concentration of frcarotene. One milliliter of f-carotene chloro- form solution was added to a flask that contained 20 mg of lino- leic acid (99% Sigma-Aldrich, St. Louis, USA) and 200 mg, Tween 440 (99%, Sigma-Aldrich, St. Lous, USA). Chloroform was removed using a rotary evaporator (Laborota, model 4001, Vertrieb, Ger- ‘many), with vacuum control (Heidolph Instruments Gmbh, Vert- rieb, Germany) and a thermostatic bath at 40°C: then 50cm? of oxygenated distilled water (oxygenation for 30 min) was added to the flask with vigorous agitation to form an emulsion. Five mil- liters of the emulsion was added to 0.2.cm* of the antioxidant solution (7.5 mg of extract/1 cm? of distilled water) in assay tubes. To the control solution, 02cm? of pure distilled water was added. A blank consisting of 20mg linoleic acid, 200mg ‘Tween 40 and Sem? oxygenated distilled water was used t0 bring the spectrophotometer to zero. Tubes were manually sha- ken, and absorbance measurements made at 470 nm immediately after the addition of the emulsion to the antioxidant solution. The tubes were placed in a water bath (model TE 159, Tecnal, Piraci- caba, Brazil) at 50°C. Absorbance measurements were made at ‘30 min intervals during 3 h. The average deviation of duplicated ‘experiments never exceeded 8%, therefore, no additional statisti- cal analysis was considered necessary. The antioxidant acts (Aas) were calculated by Eq. (1): ‘MA (0) ~ 100 (1 ~APSapan ABS a Acar ~ ADS pera 2322 Total phenolic compounds Total phenolic content was estimated using the Folin-Ciocal- eau method for total phenolics, based on a colorimetric oxida- tion/reduction reaction of phenols (Singleton et al, 1965). Briefly, 1m? of the sample was mixed with 1 en? of Folin and Giocalteu's phenol reagent. After 3 min, 1m’ of saturated so- dium carbonate solution (50% w/w) was added to the mixture, and the volume was adjusted to 10cm? with distilled water. ‘The reaction was kept in the dark for 90 min at room tempera ture, after which the absorbance was read at 725m with a UV-Vis. Spectrophotometer Hitachi, model U-3010 (Tokyo. Ja- an}. For the control sample, 1 cm? of distilled water was taken. ‘The results were calculated on the basis of the calibration curve Of gallic acid (GA) and expressed as milligrams of gallic acid equivalents (GAES)/g dry material. The average deviation of dupli- cated experiments never exceeded 8%; therefore, no additional statistical analysis was considered necessary. 23.3, Total monomeric anthocyanins (TMA) The total monomeric anthocyanin (TMA) content was deter- mined using the pH differential method described by Giusti and Wrolstad (2001), which relies on the structural transforma- tion of the anthocyanin chromophore as a function of pil. A UV~ Vis. spectrophotometer (Hitachi, model U-3010, Tokyo, Japan)1 Sans ea Jura of Food Engine 101 (2010) 24-31 2s was used for spectral measurements at the maximum absor- ance wavelength (approximately 512m) and 700nm, using distilled water as a blank Por this purpose, 20mg of extract ‘was dissolved in 10cm of distilled water. Two dilutions of the sample were prepared: one with hydrochloric acid/potassium Chloride butler pH = 1.0 and the other with sodium acetatefacetic acid buffer pil=45. The pif values of the buffers were measured using a pH-meter (Digimed, model DM-22, Si0 Paulo, Brazil) cal- ‘brated with buffers at pH 401 and 6.86, and they were adjusted. with HCl (98.5% Ecibra, Santo Amaro, Brazil). Aliquots of extract were brought to pH 1.0 and 4.5; 15 min later, the absorbance of each equilibrated solution was measured ‘at the maximum absorption wavelength and 700nm for haze cortection using ‘-cm path length glass cells (1). The dilution factor (DF) was determined (final volume over original sample volume). The di ference in absorbance values at pH 1.0 and 4.5 is directly pro- portional to the TMA concentration. The anthocyanin content was calculated as cyanidin-3-glycoside (MW'= 449.2 g/mol and £2=26,900 L/molcm), and the results were expressed a8 mg cy= Sealycosidelg dry material, The average deviation of duplicated. experiments never exceeded 8%; therefore, no additional statisti cal analysis was considered necessary. The absorbance of the di luted sample (A) and the TMA were calculated with Eqs. (2) and Gr: A= (Ansx~ Arvo (Ane ~ Aro)yes @ TMA (mg/L) — (Ax MW » DF x 1000) (1) 8 234. Thin-layer chromatography (TLC) The extracts were fractionated by thin-layer chromatography (MLO. The TLC was performed using silica plates (20 x 20cm, ‘mm height, Merck, Darmstadt, Germany). To identify which anthocyanins were exiracted in each extraction procedure, accord- ing to Wagner and Bladt (2001). the mobile phase used was com- posed by ethyl acetate (99.5%, Merck, Si0 Paulo, Brazil), glacial acetic acid (99.7%, ECIBRA, Santo Amaro, Brazil), formic acid (GES%, VETEC, Rio de Janeiro, Brazil) and distilled water (100: 111:11:26),and no spray reagent was used, Extracts and anthocyanin standards were diluted in methanol (98%, Synth, Diadema, Brazil) and applied on the plates (TLC). The anthocyanin standards, ‘yanidin-3-glucoside chloride, delphinidin-3-glucoside chloride peonidin-3-ghucoside chloride, malvidin-3-glucoside chloride pelargonidin-3-glucoside chloride and petunidin-3-glucoside chlo- ride were purchased from Extrasynthese (Genay, France). 2.4. Process simulation SuperPro Designer 6:0 was used for process simulation. This software allows the mass and energy balance estimation forall streams of the process, estimates purchase costs, and reports stream and equipment data, as well as capital and manufacturing costs. ‘The conventional extraction denoted in this paper by agitated bed extraction (ABE) was developed using the software in a similar ‘manner to the work of Takeuchi et al. (2009), as shown in Fig. 1 The extraction procedure consists placing a known mass of jabuti- caba skins immersed in a known volume of solvent inside an agitated tank. The equipment consists of two extractors, extract- solution tank, pump, evaporator, condenser, and recycled solvent tank. The presence ofthe second extraction vessel among the equip- ‘ment permits the simulation ofa continuous process: while one of the vessels is under operation, the other one goes through the cleaning and recharging processes. For the simulation of the combined UAE + ABE extraction, in, accordance with a possible setup ofan ultrasound extraction reac- tor described by Vinatoru (2001), it was asstimed that ultrasonic ‘wansducers are bonded to tank external walls. Then, during the first 10 min, the raw material stays immersed in the solvent inside an agitated tank receiving the ultrasound treatment; after that, there i 2h of agitation without ultrasound ieradiation (as in aon ventional extraction by agitation). For simulating UAE, the same process was assumed; however, the ultrasound treatment was applied during the entire process time with an intermittent agitation, just for the purpose of home. geneity. Moreaver, the excessive energy dissipation in the form of heat may lead t0 the degradation of the substrate and hence re- quires cooling; for this reason, the final temperature in the cooling, ‘operation step was 30 °C. To simulate soxhlet extraction, some modifications were made to the process, as shown in Fig. 2. The process also consisted of placing a known mass of jabuticaba skins immersed in a known volume of solvent inside an agitated tank, and the solution is heated to 78°C. The presence of a condenser connected to the extractor simulates the condensation step of solvent extraction, Fig. 1. Fowehurt of Agated Bed Exxaction (ASE) developed in Supetre Designer28 11 Somes ea Jura of Food Egiccing 101 (2010)24-31 ‘and its reflux to the extractor. The eflux operation is exhaustedly repeated, 25. Economical evaluation ‘The estimation ofthe cost of manufacturing (COM) was done for the crude extract, the phenolic compounds fraction and the antho- ‘eyanin-rich fraction for the extracts obtained by UAE, UAE+ ABE, ABE, and soxhlet (acidified solvent or not). ‘The main costs that compose the COM are similar to the ones described by Turton et al (2003). which are given by total capital investment cost and operating cost. The total capital investment cost represents the fixed capital investment (FCI), working, capital ‘and start-up cost. The first one involves expenses with equipment, installation, territorial taxes, engineering, etc., while the second ‘one represents operating liquidity available to a business, and fi- nally, the start-up cost is associated with the beginning af opera- tion ‘and the validation of the process. The operating cost TE Hr represents direct costs that are directly dependent on the produc- tdon rate; itis composed of the cost of raw materials (CRM), the cost of the lost solvent during the process, utilities cost (CUT), which represents the demand for steam and cooling water re- ‘quited for the evaporator and condenser, electricity, and opera- tional labor cost (COL). ‘According to Pereira and Meireles (2007) the estimation of the COM for the phenolic compound fraction and anthocyanin-rich fraction was done by taking into account that the percentage of these fractions in the extracts can affect their specific cost. 26. Scale-up ‘The scale-up procedure assumed that the industrial scale unit hhas the same performance as the laboratorial scale unit when the solvent to feed ratio between the mass of solid and solvent are kept ‘constant (S/F), 28 well the rue density of substrate and operational ‘conditions. The process was designed to run 7920h per year, ae Joh ee ty Fimo rere seat net ue rr rs Fi. 2. Hlowchart of sowlet extraction developed in SperPo Designer” Operational timate COM by softwere for UAE, ABE UAE A, she an owe ‘uracionProcss oF Tenpearure (0) Tine (i) Global visi ‘ABE 10 25 10. 901 Uae + Ant mo 2s 0. 1008 UAE mo B 0, uss Soret mo n 40 982 Souhet ptt 3 mo En 40 35 ay materia cos (USA) 10 ‘Ethanol cost (U/kg) 085 ‘Oparational abo ost (USS) 6 Hlercty east (USS/kWh) 92 CEASA Gin. ea 2 cen, 20s1 Sans ea Jura of Food Engine 101 (2010) 24-31 a” which corresponds to 330 days per year with continuous 24-h per day shift. The technical information used to estimate the COM of the processes is shown in Table 1 This study considered an industrial setup with extractors of (0.05, 0.1 and 03m’. The amount of jabuticaba skins used per industrial batch immersed in ethanol was determined for each capacity. The solventloss was considered to be 10% ofthe total eth anol involved in the process. The true density of the vegetable ‘material was 1450.5 kg/m”. The number of ultrasonic transducers ‘needed for each capacity in the UAE and UAE+ ABE processes and, consequently, their price was estimated using information provided by Unique Group (Indatatuba, Brazil). 3. Results and discussion 3.1. Antioxidant activity Fig. 3 shows the antioxidant activities of the extracts. Although the antioxidant activities ofall extracts were superior to the anti oxidant activity of p-carotene (control) large differences were ob- served among them. Even though the solvent (ethanol) and the process time duration (2h) used in ultrasound assisted and agi tated bed extraction were the same, the antioxidant activities of the extracts were different, demonstrating that the extraction ‘method employed is an important parameter. Similar behavior was observed by Corrales et al. (2008) when using different extraction methods (ultrasound, high hydrostatic pressured and pulsed electric fields assisted extractions), with the best antioxi dant activity attributed to the extract obtained using pulsed elec tic fields, Fig. 3 shows that the soxhlet extraction using ethanol and acid- ified ethanol (pH 3)and the combined UAE ABE process using, ethanol resulted in extracts with the highest antioxidant activities; {heir deviation was less than the experimental error. [can also be noted that recognized synthetic BHT and the natural antioxidant ‘quercetin presented higher antioxidant activities than all extracts In the case of soxhlet extraction, most of the time the extraction time is very long in comparison with all the other techniques. (Wilga et al. 2007), resulting in a high consumption of energy and therefore higher cost. In this work an extraction time of 8h ‘was used for soxhlet extraction, which resulted in an extract with, AA similar to the extract obtained using combined UAE + ABE. To completely understand the results, studying the composition of each extract is necessary because in general diferent extraction. methods introduce different combinations of phenolic compounds and anthocyanins into the extract, resulting in different bioactvi ties (Dai etal. 2009). 2.2. Phenolic compounds Fig. 4 shows the extraction of phenolic compounds using difer- cent extraction methods expressed as milligrams of galic acid equivalents (GAE)/g dry material (let y-axis) and their respective extraction yield (M) expressed as g extract/g wet materials 100 (&) (right y-axis), Reynertson etal. (2008) investigated the total phenolic content of jabuticaba fruit and found the value of 31.6 mg. of GAE/e of dry fruit. Taking into account that the major part of the fruit fs pulp and its water content is very high, by approximation, the previous value becomes little higher than 31.6 mg of GAEjg of dry skin, cor- roborating the results presented in Fig. 4. Because no further effec: tive comparison ean be dane with the results presented and the results obtained by Reynertson et al. (2008) in terms of which phe- nolic compound extraction yields, the best methods, among the ‘ones studied in this paper. were also the three that resulted in ex tracts with the highest antioxidant activities, The correlation between the phenolic compound extraction yields and the antioxidant activities of the extracts was explored. ‘A high and significant (r~0.88) correlation was found between them. This positive correlation indicates that higher values of anti oxidant activities are related to higher yields of phenolic compounds. Dragovié-uzelac et al. (2007) observed the same direct correla- tion between phenolic content and antioxidant capacity of several Fruit extracts. Two different methods were used for measuring the antioxidant capacity, ABTS [2,2-azinobis(3-ethylbenzothiazoline= 6-sulfonic acid) radical cation assay] and ORAC (oxygen radical absorbance capacity), which also resulted in a high and significant Correlation between them, where r=099 and r=084, respectively. It can be observed in Fig. 4 that the combined UAE+ ABE pro- ess resulted in phenolic compound yields higher than that ob- tained by the UAE or ABE process and similar to that obtained by Soxhlet extraction, The extraction yields did not correlate strongly 10 0 Zoo = t—_} Fn z —— zo : — ix z0 Ea 2 ° 20 o 0 120 180 180 Time (ein) URE ABE Combined UAE = ABE “Soxhlet (EMnaro!) = Soxhlet (Ethanol pH) —=BHT —~avercetin ig. 3. Anondant Activity of jabuccaba shin acts obtained using diferent methods (Expressed 25% of hbtion of oxidation).