Prediction and Identification of Antioxidant Peptides in Potato Protein Hydrolysate

Principal component analysis (PCA) was used to cluster the possible amino acid compositions of antioxidant peptides in potato protein hydrolysate (PPH).'e antioxidant peptides exhibiting high ABTS scavenging capacity were isolated with the procedure of ultrafiltration, gel filtration, and preparative RP-HPLC and identified by UPLC-MS/MS. Phe, Tyr, and His were shown to group together with ABTS scavenging capacity in component matrix plot. 'ree prominent peptides, namely, Phe-Tyr, Tyr-Phe-Glu, and Pro-Pro-His-Tyr-Phe, which matched the sequence of patatin and were made up of Phe and Tyr, were identified. 'e peptide Tyr-Phe-Glu demonstrated antioxidant activity against Caco-2 cell oxidation induced by H2O2. 'e results suggested that multivariate analysis could be used to predict the amino acid compositions of antioxidant peptides.


Introduction
Enzymatic hydrolysis has been used to improve the solubility of commercial potato protein with low solubility [1,2]. e radical scavenging and metal chelating capacity [2] of potato protein hydrolysate (PPH) have resulted in its potential role in retarding lipid oxidation of food products. It was demonstrated that PPH was able to suppress the lipid oxidation in cooked beef patties [2], cooked frankfurters [3], and food model emulsions [4]. As protein hydrolysates are a mixture of peptides, exploring the exact structure of antioxidant peptides in the potato protein hydrolysate is of interest. at will help to explain their antioxidant mechanisms in food products and contribute to the relationship between structure and activities of peptides.
Waglay and Karboune have identified a lot of peptides from PPH prepared using Flavourzyme and papain [5].
ose peptides matched the sequence in potato protein. However, their bioactivities were not estimated. Besides, few peptide sequences were related to antioxidant peptides. Several antioxidant peptides have been identified in previous research [6][7][8], while little has been done to predict the amino acid compositions of antioxidant peptides before identification. Bioinformatics methods might be a promising method for predicting the potential sequence of bioactive peptides in potato protein [9][10][11][12]. Unfortunately, release of these bioactive peptides from potato protein using proteases was not achieved.
Since antioxidant activity of peptides is related to the composition and sequence of the amino acids, classifying amino acid composition may be able to predict the sequence of the antioxidant peptides. Multivariate analyses such as principal component analysis (PCA) have been used to differentiate and cluster the samples in food science and technology [13][14][15][16][17] and predict the bioactive peptides presented in food proteins [18]. In silico analysis based on the data from databases and literature was used in these studies. Little has been done to use the data collected from experiments.
In our previous research, we have incorporated PPH fractions with different amino acid compositions into emulsions to retard their lipid oxidation [6,7]. We hypothesized that the chemical action of PPH may be related to antioxidant peptides in these fractions. Furthermore, some special amino acids may make up or contribute a lot to these antioxidant peptides.
erefore, the aim of the present research was to predict possible amino acid compositions in antioxidant peptides presented in PPH using PCA through amino acid compositions of those fractions. en, the individual antioxidant peptides were identified to verify our assumption. e antioxidant capacity of synthesized antioxidant peptide was evaluated by H 2 O 2 induced cell oxidation model.  [2], and the method description partly reproduces their wording. PPH was prepared by hydrolyzing potato protein suspension (40 mg/mL) at pH 8.0 for 60 min at 50°C using Alcalase with enzyme/substrate ratio of 1/100 (w/w). After hydrolysis, the pH of hydrolysate was adjusted to 7.0 with 1 M NaOH, and then the enzyme was inactivated at 80°C for 15 min. e hydrolysate was freeze-dried and stored at 4°C in sealed plastic bags before use.

Peptide Fractionation with Different
Polarity and Molecular Weight. PPH fractions with different polarity were isolated using ammonium sulfate precipitation as in the method described by Cheng et al. [7]. e fractions precipitated with the ammonium sulfate concentration of 30, 50, 70, and 90% were defined as P30, P50, P70, and P90. In addition, PR was the retentate of PPH after ammonium sulfate precipitation at the concentration of 90%. e fractions of peak 1, peak 2, and peak 3 with different molecular weight were separated from PPH using gel filtration as in the method described by Cheng et al. [6]. All the fractions were then freeze-dried and stored at 4°C in sealed bags before use. e amino acid compositions of these fractions were estimated using OPA derivation method [19].

Peptide Fractionation Using Ultrafiltration.
e PPH solution was passed through Millipore Pellicon-2 ultrafiltration system (Millipore Corporation, Billerica, MA, USA) in the order of 10 and 1 kDa molecular weight cut-off (MWCO) ultrafiltration membranes. e retentate from 10 kDa MWCO ultrafiltration membrane (fraction > 10 kDa) and 1 kDa MWCO ultrafiltration membrane (fraction in the range of 10-1 kDa) and the permeate from 1 kDa MWCO ultrafiltration membrane (fraction < 1 kDa) were collected and freeze-dried for further use.

Isolation and Identification of Active Peptides.
e method described by Cheng et al. [6] was followed, and the method description partly reproduces their wording. PPH fraction from 1 kDa MWCO ultrafiltration membrane permeates was isolated using gel filtration (Sephadex G-15), and three fractions were collected. en the PPH fraction that displayed the highest ABTS +• scavenging activity within those three fractions was further purified on a SunFire Prep C18 column (1.9 × 15 cm) with preparative HPLC (Waters Corporation, Milford, MA). e fractions were eluted using methanol as solvent in linear gradient mode (10-95% in 50 min) at the flow rate of 6 mL/min. e elution fractions were collected at the accuracy of 0.01 min. e procedure was run ten times to get sufficient amounts of peptide fractions. e elution fractions were freeze-dried after removing the methanol. e purified peptide fractions that displayed the highest ABTS +• scavenging activity were then subjected to ultraperformance liquid chromatography (Waters Corporation, Milford, MA), followed by Waters Synapt Mass Quadrupole Time-of-Flight Mass Spectrometer (Waters Corporation, Milford, MA) for sequence identification. Peptide sequencing program in the MassLynx software was used to interpret the spectra. Potato protein sequences published in National Center for Biotechnology Information (NCBI) database were used to match the peptide sequences.

Radical Scavenging Activity (RSA).
e ABTS radical cation (ABTS +• ) assay was estimated by the method described by Cheng et al. [6]. In addition, the RSA of peptide samples was expressed as Trolox equivalent (μM). 2+ Chelating Activity. Ferrous metal ion chelating activity was determined according to the method of Wu et al. [20], and the method description partly reproduces their wording. One milliliter of FeCl 2 (20 μM) was mixed with 0.5 mL of PPH fraction, and then 1 mL of ferrozine (500 μM) was added. e absorbance of the mixture was recorded at 562 nm (A s ) after 10 min incubation. Deionized water was used as the control to determine the absorbance of the mixture (A 0 ). e Fe 2+ chelating activity was calculated as

Cell
Culture. Caco-2 cells (1 × 10 5 cells/mL) were cultured in high glucose DMEM with 10% FBS, 1% NEAA, and 50 units/mL of penicillin-streptomycin and incubated in 5% CO 2 at 37°C. Cell passage numbers 20-50 were used in the experiments. e cells were grown in a flask for 6 days with fresh medium replaced every other day.

Induction of Oxidative Stress.
e oxidative stress was induced to confluent cells by adding 0.75 mM H 2 O 2 . e Caco-2 cell was inoculated into 96-well culture plates at a concentration of 4 × 10 3 -6 × 10 3 per well and cultured for 24 h. e cells were then incubated for 4 h with various concentrations of YFE (0.25-5 mg/mL). Cells were washed twice with PBS, followed by the addition of 0.75 mM H 2 O 2 for 4 h. At the end of 8 h, the cell viability was determined using Cell Counting Kit-8. Each plate tested included negative control (NEG, treated with treatment medium only) and positive control (POS, treated with treatment medium and H 2 O 2 ).

Determination of Lipid Peroxidation and Protein Carbonyl Groups of Caco-2 Cell.
e cellular MDA and the protein carbonyl groups were measured to determine the extent of lipid peroxidation and protein oxidation under oxidative stress induced by H 2 O 2 , respectively. MDA and protein carbonyl were determined using kit according to the manufacturer's instructions.

Activity Determination of Antioxidant Enzyme.
CAT, GSH-PX, and SOD activities were determined using kit according to the manufacturer's instructions.

Statistical Analysis.
All assays were carried out in at least three different trials, and data were subjected to analysis of variance (ANOVA). When treatment effects were found significant (p < 0.05), the difference between means was identified by Tukey's test. Amino acid compositions and ABTS radial scavenging capacity of PPH fractions with different molecular weight and polarity were used for principal component analysis (PCA). PCA was performed by SPSS 17.0 (SPSS Inc., Chicago, IL).

Results and Discussion
3.1. Principal Component Analysis. In our previous research, three fractions (peak 1, peak 2, and peak 3) were separated from PPH using Sephadex G-15 gel filtration [6], and five fractions were separated from PPH using ammonium sulfate precipitation [7]. ese fractions were shown to retard lipid oxidation of emulsions. at might be related to their free radical scavenging capacity. Although some antioxidant peptides were identified from these fractions, whether it was possible to predict the sequence of antioxidant peptides by amino acid composition was not clear. In many studies, amino acid composition has been used to assess antioxidant capacity of protein hydrolysates. However, little has been done to use amino acid compositions to predict sequence of antioxidant peptides. To distinguish the effective antioxidant amino acids and predict the antioxidant peptide compositions in PPH, the association of amino acid compositions in those fractions (Table S1) was assessed using principal component analysis (PCA) method. As a result, four components with a cumulative variance of 93.0% were extracted by PCA ( Figure S1). e variance of PC1, PC2, PC3, and PC4 was 47.2%, 20.3%, 16.6%, and 8.9%, respectively. e first two principal components, which were able to explain about 70% of data variance, were used to analyze potential amino acid components of antioxidant peptides in PPH. e component matrix plot of PC1 and PC2 is shown in Figure 1(a). e result showed that the amino acids in different PPH fractions were able to separate into several groups. Within these groups, three amino acids, namely, Phe, Tyr, and His, were found to be close to ABTS.
Actually, Phe, Tyr, and His have proved to be important components of antioxidant peptides from different protein hydrolysates [21,22]. A novel peptide, Pro-Phe-His-Pro-Tyr, which was made up of Phe, Tyr, and His was identified from nest protein hydrolysate [23]. It was suggested that some of the antioxidant peptides in PPH might contain at least one of these three amino acids. Fortunately, \we have identified several antioxidant peptides existing in potato protein in our previous works [6,7], in which Tyr was common amino acid. However, little was known about the existence of antioxidant peptides which were made up of or based on two or three amino acids from Phe, Tyr, and His, such as Tyr-Phe, Phe-Tyr, and His-Tyr-Phe. To make our hypothesis reasonable, we matched the supposed peptide sequences to the sequence of patatin which was the most abundant protein in potato. e result in Figure 1(b) showed that the peptide sequence of YF, FY, YY, FF, FYF, and HYF (bold) was displayed in the amino acid sequence of patatin. To verify the possible existence of peptides, we carried out the common procedure to identify the antioxidant peptides in PPH in the following sections.

Ultrafiltration and Gel Filtration
Fractionation. PPH was first fractionated using ultrafiltration with cut-off molecular weight of 10 and 1 kDa. e ABTS +· scavenging activity and Fe 2+ chelating activity of fractions at the protein concentration of 1 mg/mL are shown in Figure 2. e fraction with the molecular weight less than 1 kDa indicated the highest ABTS +· scavenging activity and Fe 2+ chelating activity. erefore, the fraction with the molecular weight less than 1 kDa was loaded into gel filtration unit. G15 was used to separate the PPH less than 1 kDa, and three fractions (peak 1, peak 2, peak 3) were collected. e fraction peak 3 which had the lowest molecular weight exhibited the highest ABTS +· scavenging activity and Fe 2+ chelating activity (Figure 3). ABTS +· scavenging activity of peak 3 (1 mg/mL) increased by 92.4%, 95.4%, and 124.8% when compared with PPH (1 mg/mL), BHA (0.1 mg/mL), and Vc (0.05 mg/mL), respectively. Meanwhile, Fe 2+ chelating activity of fraction peak 3 (1 mg/mL) increased by 6.4 and 3.0 times when compared with PPH (1 mg/mL) and EDTA (1 mg/mL), respectively.
e results showed that LMW fractions Journal of Food Quality 3 contributed a lot to the free radial scavenging activity and Fe 2+ chelating activity of PPH. In this respect, it has been indicated that many antioxidant peptides were small peptides [24][25][26]. Furthermore, antioxidant peptides identified from PPH were reported to be made up of 2 to 6 amino acids in previous works [6][7][8].

Peptide Purification and Sequence Identification.
As peak 3 from gel filtration was still a mixture of peptides, it was purified using preparative RP-HPLC. Twenty-three fractions were separated from peak 3 (Figure 4(a)) and labeled from P3-F1 to P3-F23, respectively. According to ABTS +· scavenging activity (Figure 4(b)) and peak area of fractions (Figure 4(b)), three fractions, P3-F6, P3-F8, P3-F14, were selected to identify the peptide sequences by UPLC-MS/MS. Ten peptides whose probability of certainty exceeded 95% were identified. e sequence of these peptides, which were made up of 2 to 5 amino acids with the molecular weight range from 280 to 700, is exhibited in Table 1. e results confirmed that antioxidant peptides in PPH were LMW short peptides; nevertheless, different fractionation procedures were used. e peptide sequences identified were then matched to the sequence of patatin in the NCBI database. e existence of Leu-Asp-Ala-Lys, Phe-Tyr, Tyr-Tyr, Tyr-Phe-Glu, and Pro-Pro-His-Tyr-Phe was indicated (Table 1).
Besides the size of peptide, amino acid residues were known as an important factor for the antioxidant activity of protein hydrolysates. Tyr and Phe which were prone to oxidation [27] for their aromatic group were the most abundant amino acids in these peptides. Moreover, His might react with proteins [28] to protect proteins from oxidation. When Pearson's correlation was used to analyze the relation between content of individual amino acid and ABTS +· scavenging activity, content of Tyr and Phe also displayed significantly positive relationship with ABTS +· scavenging activity (p < 0.05). Furthermore, it was interesting that those peptides except Leu-Asp-Ala-Lys fitted well our hypothesis that antioxidant peptides might be made up of or based on two or three amino acids from Phe, Tyr, and His. Some peptides or fragments including FY, YY, YF, and HYF in patatin were successfully predicted and identified. e MS/MS spectra of the prominent peptides including Phe-Tyr, Tyr-Phe-Glu, and Pro-Pro-His-Tyr-Phe are shown in Figure 5. Phe-Tyr which exhibited the highest peak area in chromatography spectrum was the prominent peptide in fraction P3F8. Interestingly, Phe-Tyr was also found in peptide mixture which was separated from the water phase of emulsion prepared using PPH (not published). It seemed that the peptides we identified at the present work were distributed into water phase of emulsions. In our previous work [29], we have identified some antioxidant peptides partitioned at the interface of emulsion, but none of the peptides reported in the present work was the same as them. Our present result could explain the reason why the antioxidant peptides we identified did not match the interfacial peptides. at confirmed our hypothesis in one of the antioxidant mechanisms of PPH for retarding lipid oxidation of emulsion that PPH partitioning in the water phase was able to prevent free radical transferring through the oil droplets by their free radical scavenging capacity [4]. In a word, our result showed that PCA might be an efficient method to predict the sequence of antioxidant peptides based on amino acid compositions.
Although Ser-Ser-Glu-Phe in P3F14 did not match the sequence of patatin, it was shown to exist in the sequence of metallocarboxypeptidase inhibitor which was one of the protease inhibitors in potato [30]. Moreover, Ser-Ser-Glu-Phe is part of the antioxidant peptide Ser-Ser-Glu-Phe-r-Tyr which was identified in the sequence of metallocarboxypeptidase inhibitor [7]. In addition, the remaining part of peptide r-Tyr was also identified as antioxidant peptide in PPH [6]. at might be due to the broad cleavage  site and priority cleavage at the aromatic amino acids of Alcalase [31]. According to the peptides we identified, Alcalase was indicated to hydrolyze the peptide bonds in potato protein on both amine and carboxyl sides of Tyr and Phe; amine side of Ser, r, Leu, and Ile; and carboxyl side of Lys, Glu, Gln, and Met. It was not consisted with the work on casein phosphopeptides released by Alcalase, in which carboxyl side of Glu, Met, Leu, Tyr, Lys, and Gln was the cleavage site [32]. However, the peptides SIDGGGIK (f33-f40), TNKPVIF (f159-f167), and SNLAKSPE (f168-f1175)  which were identified by other researchers from PPH prepared using Alcalase [33] demonstrated the cleavage site in carboxyl side of Lys, Glu, and Phe. Although we did not identify the peptides TNKPVIF (f159-f167) and SNLAKSPE (f168-f1175), the antioxidant peptides SFDIK (f153-f158) and LDAK (f176-179) ( Table 1) next to these two peptides were identified in our previous [29] and present works, respectively. e existence of Tyr-Phe-Glu in PPH which has been identified in our previous work [6] was also demonstrated in the present work again. ose results exhibited useful information for PPH's potential application in food industry and showed that enzymatic hydrolysis of potato protein by Alcalase is stable and could be repeated by different producers.

Effects of YFE on Lipid and Protein Oxidation of Caco-2
Cells. Since YFE was identified as the antioxidant peptide in PPH, the bioactivity of YFE on cellular antioxidation was involved. e cell viability of Caco-2 cells was higher than 85% when the concentration of YFE was 1 mg/mL. Hence, the YFE concentration of 1 mg/mL was used for the cellular oxidation trials. e oxidative stress induced by H 2 O 2 was able to result in oxidation of lipid and protein in cellular membrane. e formation of MDA and carbonyl was used as the biomarker of Caco-2 cell oxidation, as shown in Figure 6.  induced by H 2 O 2 . ese protective effects may be due to the radical scavenging activities of YFE or the activation of the cellular defense system against oxidative stress by YFE.

Effects of YFE on Antioxidant Enzyme
Activities. e role of antioxidant enzymes such as CAT, GSH-PX, and SOD in the cellular antioxidant defense system was critical to elucidate the protective effects of YFE against oxidative stress in the cells. ose enzymes can repair cellular oxidation damage induced by free radical through electrontransfer reactions [34,35]. e effect of YFE on the activities of CAT, GSH-PX, and SOD of Caco-2 cell is shown in Figure 7. H 2 O 2 induced oxidation on Caco-2 cells reduced the activities of CAT, GSH-PX, and SOD of Caco-2 cells. Treating the Caco-2 cells with YFE was able to recover 100%   Journal of Food Quality of the CAT and SOD activities and 66.6% of the GSH-PX activity. e activities of CAT, GSH-PX, and SOD in YFE treated cells increased by 93.5%, 133%, and 114% (p < 0.05) when compared with POS samples. Our results were similar to several previous results reporting that antioxidant peptides were able to protect cells against oxidation stresses induced by exogenous oxidants [36][37][38][39].

Conclusions
His, Phe, and Tyr were shown to be related to ABTS +· scavenging activity of PPH by PCA. Some of the antioxidant peptides including Phe-Tyr, Tyr-Tyr, Tyr-Phe-Glu, and Pro-Pro-His-Tyr-Phe which were based on Phe and Tyr were identified by UPLC-MS/MS after fractionation and purification. e results indicated that multivariate analyses like PCA could be used to predict the possible sequence of antioxidant peptides before identification. e antioxidant peptide Tyr-Phe-Glu was able to suppress Caco-2 cellular oxidation induced by H 2 O 2 through improving the activities of antioxidant enzymes including CAT, GSH-PX, and SOD.

Data Availability
e data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest
e authors have no conflicts of interest to declare.