Antioxidant activity of hydrolysates and peptide fractions of glutelin from cocoa (Theobroma cacao L.) seed

ABSTRACT Glutelin fraction from cocoa almond was hydrolyzed with alcalase for the production of hydrolysates. These were then fractionated by ultrafiltration to obtain peptides with a molecular weight (MW) lower than 3000 Da. The antioxidant activity (AOX) of the hydrolysates and peptides was assessed using 1,1-diphenyl-2-picrylhydrazyl (DPPH•), 2,2´-azinobis-(3-ethylbenzothiazoline-6-sulfonic acid) (ABTS•+), and oxygen radical absorbance capacity (ORAC) assays. Ultrafiltered peptide fraction with the highest AOX was purified by size exclusion chromatography, by which four main peaks were identified with a MW between 535 and 2959 Da. The peptide fraction P1 (MW = 2959 Da) showed the highest capacity to scavenge DPPH• and ABTS•+ radicals with radical scavenging activity DPPH• and ABTS•+ with EC50 values of 237.48 and 19.29 µg/mL, respectively, which were similar to those obtained with glutathione. These results show that enzymatic treatment of cocoa glutelin comprises an attractive bioprocess for the production of peptide fractions with AOX, which could be included in the design of functional foods; moreover, they show an alternative use of cocoa.


Introduction
Free radicals, antioxidants, and oxidative stress are common terms when exploring the mechanisms involved in the origin of a vast number of diseases (chronic degenerative, neurodegenerative, and some types of cancer). Within this context, free radicals are continuously produced in the human organism by means of biochemical reactions, which occur as part of the cell metabolism and which are associated with the generation of reactive oxygen species (ROS) and reactive nitrogen species (RNS); both of them are responsible for oxidative damage to macromolecules, such as DNA, lipids, carbohydrates, and proteins (Valko et al., 2007). Likewise, there is an antioxidant defense system limiting the noxious effects of these radicals; this system can remove reactive species through the use of enzymatic (superoxide dismutase, glutathione (GSH) peroxidase, and catalase) and non-enzymatic antioxidants (antioxidant vitamins, GSH, trace elements, coenzymes, and cofactors) (Cai et al., 2015). Therefore, disequilibrium between the generation of free radicals and antioxidant defenses gives rise to a process denominated oxidative stress, which can induce cellular damage, trigger physiological alterations, and foster pathological processes. Thus, when the endogenous defense system fails to protect the body against reactive radicals, an external supply of antioxidants is needed (Sarmadi, Ismail, & Hamid, 2011).
Cocoa seeds derive from the fruits of the tropical tree Theobroma cacao L. (Sterculiaceae family). Various subspecies including Criollo, Forastero, and Trinitarioare cultivated in Central West Africa, Central America, South America, and Asia (Bertazzo, Comai, Brunato, Zancato, & Costa, 2011). The main component of the cocoa seed is fat (47-55%), which is widely used in the chocolate-manufacturing industry, as well as in pharmaceuticals and cosmetics (Rusconi & Conti, 2010). Likewise, the second most abundant component in the cocoa seed is protein (10-15%), which is constituted by four fractions: albumin (ALB), globulin (GLO), prolamin (PRO), and glutelin (GLUT) (Abecia-Soria, Pezoa-García, & Amaya-Farfan, 2005). In addition, cocoa seed is recognized as a rich source of alkaloids and polyphenols with biofunctional properties (Abbe, Amin, Chong, Muhajir, & Hasbullah, 2008). Sensory and technological properties of chocolate and cocoa-based products depend on the fermentation, drying, and roasting of cocoa seeds (processed cocoa); during these, changes occur in the protein composition. Several studies have reported different proportions of protein fractions after fermentation, with an increase or decrease on the protein fractions' ratio (Amin, Jinap, & Jamilah, 1997;Preza et al., 2010;Voigt & Biehl, 1993); Globulins suffer extensive degradation during fermentation and are responsible for the formation of precursor compounds (oligopeptides and amino acids) producing the aroma and flavor of the final products (Amin, Jinap, Jamilah, Harikrisna, & Biehl, 2002;Biehl, Brunner, Passern, Quesnel, & Adomako, 1985). Moreover, Abecia-Soria et al. (2005) reported that the content of albumins decreases during roasting; hence, fermented cocoa seeds present a particular protein composition. Although abundant information on the chemical composition and biochemical characterization of the proteins of cocoa seed can be found in scientific literature, studies on their antioxidant properties are scarce. Preza et al. (2010) reported that proteins from unfermented and semifermented cocoa seeds showed antioxidant and antitumor activities, while Sarmadi et al. (2011) reported that cocoa autolysates exhibit antioxidant and angiotensin converting enzyme (ACE) inhibitory activities. For the above, the aim of this study was to assess the antioxidant activity (AOX) of protein hydrolysates and peptides obtained from cocoa seed.

Raw material
Cocoa pods (Theobroma cacao L., variety Forastero) harvested in May 2014 by traditional methods were obtained from plantations in the municipality of Comalcalco, State of Tabasco, Mexico. Fruits were depodded and fresh seeds fermented at room temperature in wooden boxes (100 cm 3 ) for 6 days, washed with water to eliminate remaining mucilage and air-dried. Afterwards, the coats were then removed and the cocoa almonds were crushed and stored at −18°C.

Preparation of acetone dry powder (AcDP)
Crushed cocoa almonds (3-4 kg) were defatted with petroleum ether for 24 h and alkaloids partially extracted with chloroform for 8 h in a Soxhlet apparatus. To remove polyphenols, AcDP was prepared according to Kirchhoff, Biehl, and Crone (1989). AcDP obtained was evaporated to dryness and milled using an IKA mill (IKA®A11 basic). Then the flour was sieved through a 150-µm mesh and stored at -18°C until further analysis.

Protein fraction extraction
Extraction of soluble proteins was performed according to the method by Voigt and Biehl (1993), with some modifications. AcDP was extracted successively with 10 mmol L −1 Tris-HCl (pH 7.5 containing 2 mmol L −1 EDTA), 0.5 mol L −1 NaCl (containing 2 mmol L −1 EDTA and 10 mmol L −1 Tris-HCl pH 7.5), 70% (v/v) ethanol, and 0.1 mol L −1 NaOH to obtain ALB, GLO, PRO, and GLUT fractions, respectively. All of these solutions contained sodium ascorbate (5 mmol L −1 ). Suspensions (1:10 w:v) were stirred for 4 h at 4°C and centrifuged at 10,000 × g for 20 min at 4°C. Each supernatant was dialyzed (M W 14 kDa cutoff) against deionized water for 48 h at 4°C, with a change every 24 h. The content of dialysis tubes was centrifuged as previously stated. Finally, solutions were lyophilized and stored at -18°C. Protein concentration was determined according to the method by Lowry, Rosebrough, Farr, and Randall (1951), using bovine serum albumin as standard.
Determination of AOX DPPH • radical scavenging activity (DRSA) DRSA was determined according to the method by Chen, Yang, Sun, Niu, and Liu (2012), with some modifications. A 0.5 mL aliquot of sample solution (600 µg mL −1 ) was added to 0.5 mL of DPPH • radical (0.1 mmol L −1 ) in methanol solution. The mixture was shaken vigorously and incubated for 30 min in the dark at room temperature; absorbance was measured at 515 nm. Methanol, instead of DPPH • , was used for the blank, while deionized water, instead of sample, was used for the control. DRSA was calculated by the following equation: ABTS •+ radical scavenging activity (ARSA) ARSA was determined according to the method by Re et al. (1999), with some modifications. The ABTS •+ radical cation was generated by mixing ABTS stock solution (7 mmol L −1 ) with potassium persulfate (2.45 mmol L −1 ) and the resulting mixture was allowed to rest in the dark at room temperature for 12 h prior to use. The ABTS •+ radical solution was diluted in 0.15 mol L −1 phosphate buffered saline (PBS) pH 7.4 (~1:35) to obtain an absorbance of 0.70 ± 0.02 at 734 nm. About 3 mL of this diluted solution was mixed with 150 µL of sample (600 µg mL −1 ). Absorbance was read from 1 and up to 6 min after the initial mixing, under conditions of darkness. PBS, instead of ABTS •+ , was used for the blank, while deionized water, instead of sample, was used for the control. ARSA was calculated by the following equation: Oxygen radical absorbance capacity (ORAC) assay ORAC was determined according to the method by Dávalos, Gómez-Cordovés, and Bartolome (2004). The reaction was performed at 37°C in 75 mmol L −1 PBS pH 7.4 and the final assay mixture (200 µL) contained fluorescein (FL, 120 nmol L −1 , 120 µL), 2,2´-azobis (2-methylpropionamidine) dihydrochloride (AAPH) (40 mmol L −1 , 60 µL), and sample (600 µg mL −1 , 20 µL). The microplate was automatically shaken before the first reading and fluorescence (λ exc : 485 nm, λ em : 535 nm) was measured every minute for 120 min (Tecan Safire plate reader controlled by Magellan® software). Final results were calculated by monitoring the fluorescence net decay curve of FL. Trolox (0-10 µmol L −1 , final concentration) was employed to generate a standard curve and the Trolox equivalent antioxidant capacity (TEAC, expressed as µmol TE/ mg protein) was obtained for each sample.

Preparation of protein hydrolysates
As depicted in Table 1, ALB, GLO, PRO, and GLUT showed in vitro AOX with significant differences (p < 0.05). Among protein fractions, GLUT showed the highest AOX and was found to be the predominant protein fraction (81.88%). Based on this, GLUT was used to obtain hydrolysates. GLUT was hydrolyzed with alcalase (EC 3.4.21.62, protease from Bacillus licheniformis, specific activity 2.4 Anson units g −1 , Sigma-Aldrich, Co., St. Louis, MO, USA) using an enzyme: substrate ratio (E:S) of 1:10 (w:w) at pH 7.5 and 50°C. Hydrolysis was carried out for 5, 15, 30, 45, 60, 90, and 120 min. At the end of it, the sample was heated to 95°C for 10 min to inactivate the enzyme. All samples were centrifuged at 5000 × g for 10 min at 4°C. Supernatants were denominated GLUT-H, lyophilized, and stored at -18°C.

Degree of hydrolysis (DH)
The DH was determined by the reaction of free amino groups with the 2,4,6-trinitrobenzenesulfonic acid reagent (Adler-Nissen, 1979). L-Leucine (0-2.5 mM) was used to generate a standard curve. DH values were calculated using the following equation: where L t is the amount of free amino groups released after hydrolysis, L 0 is the amount of free amino groups in GLUT, and L max is the total amount of free amino groups in GLUT obtained after total hydrolysis.

Ultrafiltration
For peptide separation, GLUT-H (5-120 min) was fractionated through an ultrafiltration membrane with a M W cutoff (M W CO) of 3000 Da (Millipore, Bedford, MA, USA). For each hydrolysis, ultrafiltered peptide fractions were obtained and denominated GLUT-P (M W < 3000 Da). The AOX of GLUT-H and GLUT-P was assessed under the same experimental conditions previously described (DPPH, ABTS, and ORAC assays).

Polyacrylamide gel electrophoresis (SDS-PAGE)
GLUT and GLUT-H profiles were carried out according to the method by Laemmli (1970), using a 12% separating gel (w/v polyacrylamide) and a stacking gel (4% w/v polyacrylamide). Gels were run in mini-slabs (Bio-Rad, Mini Protean® III Model, Hercules, CA, USA) at 200 V for 45 min.

Size exclusion chromatography
The GLUT-P that showed the highest AOX was separated using fast protein liquid chromatography on a GE Healthcare, FPLC System, equipped with a molecular exclusion Superdex column (G75HR 10/30 GL). The column was equilibrated and eluted with 20 mmol L −1 Tris-HCl, 150 mmol L −1 NaCl, pH 7.8, at a flow of 0.5 mL min −1 , collecting 1 mL fractions. Peptides were detected by their absorbance at 214 nm. The column was calibrated by using the M W -GF-70 gel filtration market kit for protein (Sigma-Aldrich®, Co., St. Louis, MO, USA). Peptide fractions were collected to assess AOX, while GSH was employed as standard. The EC 50 value, defined as the effective concentration of the peptide (µg mL −1 ) required to scavenge 50% of radical activity (DPPH • or ABTS •+ ), was calculated from a linear regression plot of scavenging activity percentage versus peptide concentration. TEAC value (µmol TE mg −1 sample) was also determined for each peptide fraction.

Statistical analysis
Analyses were carried out by triplicate and results were reported as mean ± standard deviations. Significant differences among treatments were determined by analysis of variance and Fisher´s least significant difference (LSD) tests (p < 0.05) using the Statgraphics® Centurion XVI statistical software (StatPoint Technologies Inc., Bedford, MA, USA, 2010).

Protein fractions content in cocoa seed
To avoid irreversible denaturation of proteins by the oxidation of polyphenol products during protein fractioning, as well as an effect on the results, AcDP was obtained from cocoa almond. Employing the qualitative test (with HCl), it was confirmed that AcDP was polyphenol-free. Afterward, ALB, GLO, PRO, and GLUT proteins were obtained at the concentrations depicted in Table 1. Total content of protein was 7.23 g 100 g −1 AcDP, with GLUT being the predominant protein fractionthe latter with a value of 5.92 g 100 g −1 AcDP, significantly higher (p < 0.05) than the values obtained for other proteins (0.14-0.65 g 100 g −1 AcDP)-and with the following percentage distribution: GLUT (81.88 ± 1.28%) > ALB (8.99 ± 0.11%) > GLO (7.19 ± 0.09%) > PRO (1.93 ± 0.06%). Several studies performed in unfermented and fermented cocoa seed have reported a different distribution of protein fractions. Some authors have reported a higher content of ALB and GLO (14-52%) (Lerceteau, Rogers, Petiard, & Crouzillat, 1999;Voigt & Biehl, 1993), while others have reported a higher content of GLUT (30-56%) (Preza et al., 2010;Zak & Keeney, 1976). This difference in the content of the several protein fractions is influenced by numerous factors, such as seed variety, fractionation method, batch size, fermentation, and storage conditions. Regarding fermentation, a degradation of ALB and GLO fractions has been related to the formation of precursor compounds (oligopeptides and amino acids) responsible for the characteristic aroma of cocoa (Amin et al., 2002;Romero-Cortes et al., 2013;Voigt & Biehl, 1993;Zak & Keeney, 1976). In the present study, cocoa almond was used, obtaining a low content (<1.0 g) of ALB and GLO, which was different from that obtained from nonfermented seed (3.23 and 2.36 g for ALB and GLO, respectively). Thus, degradation of these proteins occurred during fermentation. In this case, GLUT was degraded at a lower proportion, with a product of 5.92 g, amounting to 7.76 g in the non-fermented seed (data not shown).  2010). Also, for the three test methods used in the present study, GLUT showed the highest AOX (p < 0.05), as compared with the values obtained in non-fermented and semi-fermented cocoa seed (Preza et al., 2010). According to these results, GLUT is the protein fraction of the cocoa seed (Forastero variety) with the highest antioxidant potential (in vitro). Since GLUT was the predominant protein fraction and the one with the highest AOX, it was chosen for further obtention of enzymatic hydrolysates.

DH and electrophoretic profiles
In this study, GLUT hydrolysis was accomplished with alcalase, which has been frequently used to obtain protein hydrolysates and bioactive peptides, since it is a protease of microbial origin that presents broad specificity and moderate temperature and pH conditions. Additionally, it has been reported that hydrolysis with alcalase generates a higher AOX of hydrolysates and derived peptides (Sarmadi & Ismail, 2010). Figure 1 depicts the DH values obtained for the combinations of time and the E:S ratio assayed for GLUT hydrolysis with alcalase. DH values were 6.31-19.91%, pointing out that in 8 of 9 assayed times, a DH >10% was obtained. The DH reached depends significantly on the reaction time (t), highlighting that in the interval between 5 and 60 min, increases in this variable boost the DH. However, in the range between 60 and 120 min, DH did not increase significantly (p < 0.05). Moreover, maximal DH achieved for GLUT was 19.91% (t = 120 min); it is noteworthy that this was achieved at a relatively low time interval, which may have been due to the E:S ratio employed. Figure 2(a) depicts the electrophoretic SDS-PAGE profiles of GLUT and GLUT-H. On it, the GLUT profile presents two main bands, which correspond to subunits of 200 and 23 kDa; lesser intensity bands corresponding to the 27 and 14.5 kDa subunits were also found. This polypeptide profile was similar to that reported by other authors in the same cocoa species (Preza et al., 2010;Voigt & Biehl, 1993). Additionally, the intensity of the bands in the electrophoretic profiles of GLUT-H decreased, showing that an increase on the enzymatic reaction time generates a progressive decline on the intensity of the subunits' bands; this suggests the release of peptide chains of lower M W . Further, although no significant difference in DH was observed at 60-120 min of hydrolysis, the electrophoretic profiles were different. Hence, it is possible to obtain different electrophoretic profiles with a similar DH. During Tricine-SDS-PAGE electrophoresis, an increasing hydrolysis time generated a progressive degradation of polypeptides, suggesting the formation of peptide chains of lower M W (<3.5 kDa), which cannot be retained in the gel (Figure 2(b)).
The hydrolysis of proteins is a process of simultaneous reactions, during which there is a release of peptide species of different M W . Since bioactive peptides are considered short chains of 2-20 amino acids, separation of the peptides with M W < 3000 Da present in GLUT-H was achieved through ultrafiltration using a 3000-Da M W CO membrane aiming to purify peptides produced by hydrolysis with alcalase.     range (10-90%) of hydrolysates and peptide fractions of barley, walnut, and soy (Bamdad, Wu, & Chen, 2011;Chen et al., 2012;Jiménez-Ruiz et al., 2013).

AOX of GLUT-H and GLUT-P
All GLUT-H presented ABTS •+ scavenging activity, with values 11.05-26.29% ARSA, which were significantly higher (p < 0.05) than those found in GLUT (8.29%). In this case, the hydrolysates obtained at 90 and 120 min depicted a higher scavenging effect, without differences among them (Figure 3  (b)). GLUT-P exhibited a significant increase on ARSA, reaching values 17.38-41.19%; however, the behavior found was not the same as that of GLUT-H, since the highest scavenging activity was observed in fractions corresponding to 15, 60, 90, and 120 min of hydrolysis, without differences among them; nevertheless, as for the DPPH assay, GLUT-P obtained at 15 min presented a higher percentage of ARSA. Additionally, the obtained ARSA values also fell within the reported range (10-85%) for hydrolysates and peptides from other vegetal sources like amaranth and corn (Soriano-Santos & Escalona-Buendía, 2015;Tironi & Añón, 2010;Wang et al., 2014). Figure 3(c) shows that GLUT-H also presented a higher radical absorbance capacity of oxygen than GLUT, highlighting that GLUT-H 15 depicted the highest ORAC value (0.59 µmol TE mg −1 protein), which was significantly higher (p < 0.05) than the one obtained for all other hydrolysates (0.52-0.54 µmol TE mg −1 protein). In this method, GLUT-P also presented an increase in ORAC values (1.0-1.59 µmol TE mg −1 sample), maintaining the same behavior as GLUT-H, being GLUT-P 15 the one with the highest AOX. These results are comparable to those reported for soy, whey, and bovine casein (Del Mar Contreras et al., 2011;Di Pierro et al., 2014;Ranamukhaarachchi, Meissner, & Moresoli, 2013).
In this study, GLUT, GLUT-H, and GLUT-P presented AOX through two different mechanisms: electron transfer (reduction of DPPH • and ABTS •+ radicals) and transfer of hydrogen atoms (through the capture of peroxyl radicals). The ORAC FL assay has been widely used in research as a standard method for the assessment of in vitro AOX; however, methods that assess the reduction of free radicals also represent a chemically valid and important parameter of potential antioxidant compounds (Sarmadi & Ismail, 2010). On this, AOX of GLUT increased when hydrolyzing with alcalase and when performing the peptide chains separation with M W < 3000 Da, the latter showing the highest AOX. In the three assessment methods used, the ultrafiltered peptide fraction obtained at 15 min (GLUT-P 15 ) showed the highest in vitro AOX; hence, it was decided to purify it through size exclusion chromatography.

Purification of antioxidant peptide fractions
GLUT-P 15 was separated in four peptide fractions P1, P2, P3, and P4 with M W of 2959 Da, 1856 Da, 997 Da, and 535 Da, respectively (Figure 4(a)). If 120 g/mol is taken as the mean for the M W of an amino acid, the obtained fractions are constituted by peptide chains of 5-25 amino acids, which agrees with previously reported data showing that in general antioxidant peptides contain 2-20 amino acids (Samaranayaka & Li-Chan, 2011). AOX of all fractions was assessed and compared with GSH peptide. Figure 4(b) shows that fraction P1 had the highest DPPH • and ABTS •+ scavenging capacity (53.33% and 84.47%, respectively). Regarding ORAC value, no significant difference (p < 0.05) was found among the four fractions. Therefore, FPLC purification showed an increase in DPPH • and ABTS •+ scavenging capacity; however, this behavior was not found in the ORAC assay, since even lower ORAC values (1.16-1.27 µmol TE mg −1 ) than that present in GLUT-P 15 (1.58 µmol TE mg −1 ) were found; this shows a higher activity when peptide chains are linked together than when they are separated. For the ORAC assay, all peptide fractions presented a lower AOX than GSH (Table 2). Final concentrations of samples were 300, 30, and 60 µg mL −1 in DRSA, ARSA, and ORAC FL assays, Results are plotted as mean ± standard deviation (n = 3). Different letters in the same parameter indicate significant differences (p < 0.05). Additionally, the antioxidant potential of GLUT-P 15 peptide fractions was determined by the EC 50 value. This parameter is very important since it determines the necessary concentration of an antioxidant to inhibit 50% of the activity of the free radical. Table 2 depicts EC 50 values obtained for GLUT, GLUT-H 15 , GLUT-P 15 , and FPLC-purified peptide fractions. Lower EC 50 values were found in the peptide fractions, where P1 showed the lowest EC 50 values: 237.48 ± 1.75 µg mL −1 and 19.29 ± 0.59 µg mL −1 for DRSA and ARSA, respectively. Furthermore, GLUT presented the highest EC 50 values: 2225.52 ± 19.25 µg mL −1 and 194.54 ± 1.73 µg mL −1 (DRSA and ARSA, respectively); the same behavior was found for both radicals (DPPH • and ABTS •+ ), with a higher concentration of GLUT and GLUT-H required to achieve 50% scavenging; however, the concentration required for GLUT-P 15 and peptide fractions purified by FPLC decreased. Also, it has been reported that low-M W peptides depict higher AOX; in the present study, fractions with M W of 2959 Da (P1) and 1856 Da (P2) were those with the higher antioxidant effect. As previously mentioned, P1 exhibited the highest AOX, reaching EC 50 values similar to those obtained with GSH (Table 2); these results suggest that P1 cocoa peptide fraction may have a high antioxidant potential.

Conclusions
The GLUT represented up to 80% of the total of the protein present in the cocoa seed of the Forastero variety and is a good raw material for the production of hydrolysates and peptide fractions with antioxidant capacity. In particular, the peptide fractions with the highest AOX were obtained in a short time (15 min of digestion with alcalase). Partial purified peptides presented EC 50 values similar to those of glutathione and ORAC values similar to those found in peptides from other food sources.
This study provides an alternative use for the cocoa seed and can be used as a reference for futures studies to improve the efficiency of the process for its potential application in the design of functional foods.

Acknowledgment
English editing and proofreading was kindly performed by Dr Agustin Lugo-Radillo.

Disclosure statement
No potential conflict of interest was reported by the authors.