Introduction of New Computer Softwares for Classification and Prediction Purposes of Bioactive Peptides: Case Study in Antioxidative Tripeptides

Abstract

Antioxidative tripeptides were successfully classified according to their structure by using Sequence Principal Component Similarity (SPCS) software. During SPCS computation, hydrophobicity was the main characteristic of peptide residues in aspect of antioxidative activity to inhibit lipid peroxidation in linoleic acid model system. The prediction made by Homology Similarity Search (HSS)-BIOPEP combined software indicated that C-hordein fraction from barley protein showed the greatest potential to be antioxidative protein source followed by prolamin fraction from rice protein. Moreover, these homology segments of C-hordein were resistant against the digestion of mixed gastrointestinal enzymes in BIOPEP digestive model system.

Keywords:

• Antioxidative peptide
• Hydrophobicity
• Sequence principal component similarity(SPCS)
• Homology similarity search (HSS)
• BIOPEP

INTRODUCTION

Quantitative Structure-Activity Relationships (QSAR or SAR) is the most common concept in chemistry to explain the function of chemical compounds based on their structures. The uniqueness of QSAR is the mathematical expressions of the structure-activity relationships.[1] At first, the application was limited mostly to small molecular compounds such as drug and olfactory stimulants. However, QSAR concept has been extended later to macromolecules such as food proteins and peptides.[2] Nakai and Li-Chan[3] reviewed the advances in QSAR of food proteins in the concept of emulsification, foaming, gelation, and breadmaking. Nakai et al.[4] indicated that QSAR is an important program in the study of the functionality of bioactive peptides.

Sequence Principal Component Similarity (SPCS) is one of several computer software programs, which is based on QSAR concept. It has been established to stand on properties of the entire peptide molecules. This program requires reference peptides with known sequences to compare with peptide chains by initiating from N-terminal. One of six characteristics (hydrophobicity, charge, helix, strand, turn and bulkiness) consisting of SPCS needs to be selected during each calculation. The Principle Component Analysis (PCA) classification is the original algorithm of Principle Component Similarity (PCS). PCS was further modified to SPCS in order to deal with peptide sequences; however, both of them employed the same approach. Vodovotz et al.[5] found that PCS was useful in classifying GC data of mango samples according to cultivar and ripening conditions prior to detailed analysis. Nakai et al.[4] reported that the PCS analysis was successfully used to classify cationic antimicrobial peptides with high contents of arginine residues by using the functional characteristics of their amino acid residues. Also, Nakai et al.[6] applied the PCS analysis to 25 lysozyme sequences using hydrophobicity and charge as the characteristics for classification. The results showed that structure-related properties of side chains in short segments were able to predict functional sites in sequences. The SPCS software is able not only to classify bioactive peptides but also to find a main characteristic, which plays an important role in each functionality according to their sequences. Information on the main characteristic is important because the bioactive mechanism of the functionality can be explained by utilizing it.[4]

Homology Similarity Search (HSS) is also a software based on QSAR concept. It was written for peptide sequence analysis and search for functional sites such as active and binding segments in food proteins.[6] The HSS program along with reference peptide was applied to protein sequences to compare the homology by shifting the search stepwise from N-terminus toward C-terminus. The reference peptide consisted of 3–8 residues or even longer could be used.[7] Rau et al.[8] reported that HSS was successful in defining the active sites and the substrate binding sites of the lysozyme from Steptomyces coelicolor. Nakai et al.[7] introduced the application of HSS to short peptide sequences with fewer than 32 amino acid residues for explaining the underlying mechanism of their emulsifying ability. It has been indicated that pattern similarity between segments in HSS profiles of sample sequences using the motif ELE (Glu-Leu-Glu) in the reference peptide, as a search unit was effective for determining the positions of hydrophobic similarity density within the sample sequences.

Database of protein and bioactive peptide sequences such as UniProtKB/Swiss-Prot, UniProtKB/TrEMBL and BIOPEP collects the information about the sequences, which have been provided from different laboratories, for instance, number of amino acid residues, molecular mass and some additional information such as the role of a protein in a biological system.[9,10] Moreover, the option called “Enzyme action” in BIOPEP was suitable for proteolytic process design in order to obtain bioactive peptides through the systemic digestion. Results would show potential peptides that may be released after the action of the selected enzymes at the location of substrate peptides in the protein chain.[10]

Free radicals generated by exogenous chemicals or endogenous metabolic processes in food systems may cause oxidative damage in biomolecules, thus causing damage to cell membrane, loss of enzyme activity, toxicity in mammalian cell, as well as modification of DNA that causes cancer.[11–14] Recently, there has been increasing interest in antioxidative peptide research, since the antioxidative peptides can protect the human body from free radicals and retard the progress of many chronic diseases.[15,16] Iwaniak et al.[17] applied BIOPEP to screen numerous food proteins as potential sources of peptides with hypertensive activity. They concluded that in silico analysis would be useful in the analysis of bulky data of proteins as the source of biologically active peptides resulting in reducing the cost and time of investigation. Therefore, the objectives of this research were to (i) validate the SPCS software by applying to a homogeneous antioxidative peptide dataset and (ii) predict the antioxidative segments resistant against digestive enzymes by applying the HSS-BIOPEP combined software.

MATERIALS AND METHODS

Preparation of Dataset for Antioxidative Tripeptides

In order to build the homogeneous antioxidative peptide dataset, all antioxidative tripeptides was collected from the experimental data of Saito et al.[18] They constructed two series of combinatorial tripeptide libraries to explore antioxidative activity of peptide measured by inhibition of lipid peroxidation. One was composed of 110 tripeptides containing either two His or Tyr residues in the structure and the other was composed of 108 tripeptides structurally related to Pro-His-His.

Sequence Principal Component Similarity Analysis

SPCS was the combination between PCA and linear regression. After application of PCA to the data set, linear regression analysis was carried out using the PC scores with eigenvalues higher than 1. The principle of the SPCS program has been previously described by Vodovotz et al.[5] A reference sample is selected and the dependent variables for the linear regression are the deviations from the corresponding reference PC scores. The result of SPCS analysis is a plot of slope vs. coefficient of determination (r²) derived from the linear regression analysis.[19] SPCS software package can be downloaded to any PC from ftp://ftp.agsci.ubc.ca/foodsci/SPCS/.

HSS-BIOPEP Combined Software

HSS software requires reference peptides which shows antioxidative activity to search from the N terminus by shifting the search one position at a time toward C terminus.[7] The higher homology sequence present in a protein chain, the greater possibility to be a good antioxidative protein source. Fourteen protein sources were searched in protein database; UniProt KB/ SWISS_PROT and UniProt KB/ TrEMBL for obtaining amino acid sequences of individual protein source. Both protein databases are accessible by http://ca.expasy.org. Pattern similarity constant and the average of property characteristic values were computed in all observed protein sequences. The peptide segments in protein sequences with SimConst > 0.9 and average hydrophobicity (property characteristic) between 0.4–3.0 were counted and calculated the average relative antioxidative activity values.

Protein sequence that showed the highest average antioxidant activity value was selected to apply in BIOPEP digestion model software. A combination of pepsin and trypsin were used in BIOPEP computation to construct in silico digestive model system. The HSS program along with instructions is available as an ftp file (ftp://ftp.agsci.ubc.ca/foodsci/HSA/) downloadable to any PC. The BIOPEP is accessible by http://www.uwm.edu.pl/ biochemia.

RESULTS AND DISCUSSION

SPCS Classification

Bioactive peptides can be released by enzymatic proteolysis from various food protein sources and may act as potential physiological modulators of metabolism during intestinal digestion. Bioactive peptides usually contain 3–20 amino acid residues, and their activity is dependent on their amino acid composition.[20] In this study, all antioxidative tripeptides were collected from the experimental data of Saito et al.[18] They measured antioxidative activity of tripeptide and expressed as inhibition of lipid peroxidation in an aqueous autoxidation system of linoleic acid. The activity was assessed using the ferric thiocyanate method. The antioxidative tripeptide libraries are reproduced in Table 1.

Table 1 Relative antioxidative activity values of antioxidative tripeptides

Number	Peptide	Relative antioxidative activity*	Number	Peptide	Relative antioxidative activity*	Number	Peptide	Relative antioxidative activity*
1	FHH	1.8	18	DHH	1	35	HEH	1.5
2	LHH	2	19	RHH	0	36	HDH	1.5
3	IHH	2	20	HFH	2	37	HRH	0
4	YHH	1.8	21	HLH	1	38	HHF	3.6
5	WHH	1.8	22	HIH	1	39	HHL	0
6	VHH	2	23	HYH	2	40	HHI	0
7	MHH	2	24	HWH	2	41	HHY	3.6
8	CHH	1	25	HVH	1	42	HHW	3.6
9	AHH	2	26	HMH	0.8	43	HHV	0
10	GHH	1	27	HCH	1	44	HHM	0
11	THH	1.3	28	HAH	1	45	HHC	0
12	SHH	1.3	29	HGH	0.8	46	HHA	0
13	KHH	0	30	HTH	0.7	47	HHG	0.3
14	QHH	1	31	HSH	0.7	48	HHT	0
15	NHH	1	32	HKH	0	49	HHS	0
16	HHH	0	33	HQH	0.8	50	HHK	0
17	EHH	1	34	HNH	0.8	51	HHQ	0.3
52	HHN	0.3	69	QYY	5	86	YKY	10
53	HHE	0.2	70	NYY	5	87	YQY	3
54	HHD	0.2	71	HYY	2	88	YNY	3
55	HHR	0	72	EYY	3.2	89	YHY	10
56	FYY	2	73	DYY	3.2	90	YEY	3
57	LYY	3	74	RYY	2	91	YDY	3
58	IYY	3	75	YFY	2	92	YRY	10
59	YYY	2	76	YLY	3.7	93	YYF	4
60	WYY	2	77	YIY	3.7	94	YYL	3
61	VYY	3	78	YWY	2	95	YYI	3
62	MYY	2	79	YVY	3.7	96	YYW	4
63	CYY	0	80	YMY	1.8	97	YYV	3
64	AYY	3	81	YCY	2.8	98	YYM	2
65	GYY	5	82	YAY	3.7	99	YYC	0.3
66	TYY	3	83	YGY	3	100	YYA	3
67	SYY	3	84	YTY	3	101	YYG	2.7
68	KYY	2	85	YSY	3	102	YYT	4
103	YYS	4	120	LHT	5.5	137	PHG	5
104	YYK	5	121	LHS	4	138	PHT	6
105	YYQ	2.7	122	LHK	4	139	PHS	4
106	YYN	2.7	123	LHQ	4	140	PHK	3.5
107	YYH	5	124	LHN	5	141	PHQ	3.5
108	YYE	4	125	LHH	5	142	PHN	6
109	YYD	4	126	LHE	6	143	PHE	6
110	YYR	5	127	LHD	3.7	144	PHD	4.5
111	LHF	3.5	128	LHR	5	145	PHR	4.5
112	LHL	6	129	PHF	4	146	PHH	5.8
113	LHI	6.5	130	PHL	5	147	RHF	3
114	LHY	4	131	PHI	5	148	RHL	3.5
115	LHW	7	132	PHY	4	149	RHI	5
116	LHV	3.5	133	PHW	6	150	RHY	3
117	LHM	5	134	PHV	3	151	RHH	3
118	LHA	4	135	PHM	4	152	RHV	3
119	LHG	6.7	136	PHA	6	153	RHM	3
154	RHA	5	171	LWM	1.5	188	PWV	1
155	RHG	5.5	172	LWA	1.25	189	PWM	0.75
156	RHT	6	173	LWG	1.25	190	PWA	1.5
157	RHS	3	174	LWT	2	191	PWG	2.75
158	RHK	2.5	175	LWS	2	192	PWT	2.5
159	RHQ	3	176	LWK	2	193	PWS	1
160	RHN	5.5	177	LWQ	2.5	194	PWK	0.5
161	RHH	3	178	LWN	1.25	195	PWQ	1.25
162	RHE	5	179	LWH	3.25	196	PHN	2.1
163	RHD	3	180	LWE	1.75	197	PWH	1.25
164	RHR	4	181	LWD	1.75	198	PWE	1
165	LWF	1.5	182	LWR	3	199	PWD	1
166	LWL	0.5	183	PWF	1	200	PWR	2.5
167	LWI	1	184	PWL	1	201	RWF	1
168	LWY	2.25	185	PWI	1.5	202	RWL	3
169	LWW	3.5	186	PWY	1	203	RWI	1.25
170	LWV	1.5	187	PWW	3	204	RWY	0.75
205	RWW	2.5	210	RWT	3.75	215	RWH	1
206	RWV	0.5	211	RWS	0.75	216	RWE	1
207	RWM	0.75	212	RWK	0.5	217	RWD	1
208	RWA	1.25	213	RWQ	0.5	218	RWR	2.5
209	RWG	1.75	214	RWN	2.5
*The relative antioxidative activity was tabulated from the data of Saito et al.[18]

The SPCS program requires reference peptides with known sequences to compute linear regression. There is no reliable rule to select best reference. The rotation of reference for PCS computation should be tried as evidenced in Nakai et al.[21] Therefore, all of 218 tripeptides in Table 1 were rotated to be references one by one. As this research was performed using tripeptides, contribution of helix, strand, turn, bulkiness to the functionality can be mostly ignored. However, from the literature search, there was no evidence indicating an importance of charged amino acid in the aspect of donating protons to free radicals. Thus, charge was also deleted as a property characteristic. Many researchers have proposed that the hydrophobic amino acids play a role in increasing the interaction between peptides and fatty acid.[16,22–24] Hydrophobic peptides can protect oxidation of linoleic acid by donating protons to hydrophobic peroxy radicals.[14,25] In addition, Mendis et al.[26] reported that hydrophobicity of peptides facilitated scavenging free radicals in cellular system by keeping close contact with oxidizing cell membrane lipids. Therefore, only hydrophobicity was chosen as the property characteristic for antioxidative peptide SPCS classification. SPCS computation of antioxidative tripeptides was carried out using the first ten PC scores, which accounted for 93.5% of the total variability; therefore, about 6.5% data was ignored as non-significant as shown in Table 2.

Table 2 Cumulative eigenvalues of antioxidative tripeptides when hydrophobicity was used as the property characteristic

Principal component	Eigenvalue	Variance	Cumulation (%)
1	3.8682	19.3408	19.3408
2	3.1629	15.8145	35.1553
3	2.2616	11.3082	46.4634
4	2.2316	11.3082	57.7716
5	1.7819	8.9094	66.6809
6	1.0917	5.4585	72.1395
7	1.0849	5.4246	77.5641
8	1.0849	5.4246	82.9888
9	1.0849	5.4246	88.4134
10	1.0527	5.2634	93.6768
11	0.9496	4.7478	98.4246
12	0.2197	1.0987	99.5234
13	0.0258	0.1290	99.6524
14	0.0194	0.0972	99.7496
15	0.0154	0.0770	99.8266
16	0.0148	0.0739	99.9005
17	0.0148	0.0739	99.9744
18	0.0037	0.0185	99.9929
19	0.0014	0.0071	100.0000
20	0.0000	0.0000	100.0000

When hydrophobicity was used as the property characteristic, antioxidative peptides were successfully classified in groups I, II, III, IV depending on their different structures which were (I) two histidine-containing peptide, (II) two tyrosine-containing peptide, (III) one histidine-containing peptide, and (IV) one tryptophan-containing peptide (Fig. 1). The digits attached to data-points are means of relative antioxidative activity values from Saito et al.[18] Two histidine-containing peptides were appropriate reference for classification of antioxidative tripeptide as displayed in the similar accuracy of SPCS scatter-gram no matter what residue occupied x position. Other references were not able to classify as clear as two histidine-containing peptides (data not shown).

Figure 1 SPCS scattergram of antioxidative tripeptides. The tripeptides in x position varied from Asp (D), Glu (E), His (H), Lys (K), Arg (R), Ala (A), Ile (I), Leu (L), Val (V), Phe (F), Trp (W),Tyr (Y),Gly (G), Asp (N), Gln (Q), Met (M), Ser (S) and Thr (T). Hydrophobicity was chosen as the property characteristic. Two histidine-containing peptide were used as the references. The digits attached to datapoints in the blanket are means of relative antioxidative activity values from Saito et al.[18]

From SPCS classification results, the importance of hydrophobicity in antioxidative activity of peptides has been confirmed. Antioxidative tripeptide classification by SPCS software was similar to the tripeptide libraries of Saito et al.[18] It might be implied that SPCS software is validated in classifying antioxidative tripeptide, thus, it showed possibility to apply to other bioactive peptides. Nakai et al.[4] successfully classified lactoferricin into four groups by using helix and charge as the property characteristics. They concluded that such characteristics were important in the antimicrobial activity of lactoferricin.

HSS and BIOPEP Combined Software Technology

Table 3 shows major protein fractions in each protein source from the searching of UniProtKB/SWISS-Prot and UniProtKB/TrEMBL. The last column indicates the Entry name and Primary accession name in protein databases. According to the research of Saito et al.[18] as shown in Table 1, YKY (Tyr-Lys-Tyr), YHY (Tyr-His-Tyr) and YRY (Tyr-Arg-Tyr) show the highest relative antioxidative activity of 10 followed by LHG (Leu-His-Gly) and LHW (Leu-His-Gly) with the activity of 6.7 an 7, respectively. Therefore, there should be five tripeptides, which are YKY (no. 86), YHY (no. 89), YRY (no. 92), LHW (no. 115), and LHG (no. 119) to be used as the references in HSS prediction. Nevertheless, YHY, YKY and YRY consist of similar amino acid group at the middle position, which are basic amino acids.[18]

Table 3 Major protein fractions in fourteen protein sources

Protein source	Major protein fractions	Primary Accession number/Protein name
1. Fish	Actin	O42161/Actin, cytoplasmic 1
	Myosin heavy chain	Q2HXU3/Myosin heavy chain
	Myosin light chain	Q06B21/ Myosin light chain
	Tropomyosin	Q91472/Fast myotomal muscle tropomyosin
2. Milk	Alpha S1casein	P02662/Alpha-S1-casein
	Alpha-S2 casein	P02663/ Alpha-S2 casein
	Alpha-lactalbumin	P00711/ Alpha-lactalbumin
	Beta casein	P02662/ Beta casein
	Beta-lactoglobulin	P02754/Beta-lactoglobulin
	Lactoferrin	Q6LC78/ Lactoferrin
3. Soy	Conglycinin alpha chain	P13916/Beta-conglycinin, alpha chain
	Conglycinin beta chain	P25974/Beta-conglycinin, beta chain
	7S globulin	P13917/Basic 7s globulin
	2S albumin	P19594/2S albumin
	Glycinin	P02858/Glycinin G4
4. Cotton	Legumin A	P09802/Legumin A
	Legumin B	P09800/Legumin B
5. Canola	11S globulin	P33525/Cruciferin CRU1
6. Barley	Globulin	Q84NG7/Globulin
	B-hordein	Q40026/ B-hordein
	C-hordein	Q41210/C-hordein
	D-hordein	Q84LE9/D-hordein
7. Buckwheat	13S globulin	O23878/13S globulin seed storage protein
8. Rice	Globulin	P29835/19 kDa globulin
	Glutenin	Q9ZWJ8/Glutelin
	Prolamin	Q9ZWJ7/Prolamin PPROL 17D
9. Rye	Glutenin subunit X	Q93WF0/HMW glutenin subunit X
	Glutenin subunit Y	Q93WM1/HMW glutenin subunit Y
10. Sunflower	Albumin	P15416/2S Albumin storage protein
	11S globulin	P19084/11S Globulin seed storage protein
11. Oat	11S globulin	Q38780/11S globulin
	12S globulin	P12615/12S seed storage globulin1
12. Pea	Convicillin	P13915/Convicillin
	Legumin A	P02857/Lecumin A
	Legumin J	P05692/ Legumin J
	Vicillin	P13918/ Vicillin
13. Wheat	Alpha/beta gliadin	P02863/ Alpha-beta gliadin
	Gamma gliadin	P08453/Gamma-gliadin
	High molecular weight glutenin	Q41516/ High molecular weight glutenin
	Low molecular weight glutenin	Q5MFH2/ Low molecular weight glutenin
14. Egg	Apovitellenin	P02659/ Apovitellenin 1
	Lipovitellin and Phosphovitin	P02845/Vitellogenin 2
	Ovalbumin	P01012/ Ovalbumin
	Ovomucoid	P01005/ Ovomucoid
	Ovotransferrin	P02789/ Ovotransferrin

YHY is selected to be representative since histidine is recognized as the most active amino acid with scavenging singlet oxygen.[27] The antioxidative activity of histidine-containing peptides has been reported to be due to its lipid peroxy-radical trapping ability and/or chelating ability of the imidazole ring in histidine residue.[14,18,22,23,28]

Table 4 exhibits the highest average relative antioxidative activity value of protein fraction presenting in each protein source. Based on the HSS prediction, C-hordein fraction from barley protein has the highest average relative antioxidative activity value of 2.33. It might be implied that barley C-hordein is probably the most potential to be antioxidative protein source. But whether or not these homology segments in C-hordein can resist the digestive tract and pass through small intestine in the body is unknown. Therefore, the resistance of homology segments in C-hordein should be accessed through the digestion of gastrointestinal enzymes of pepsin and trypsin.

Table 4 Number of homology similarity sequence presenting in protein chain

Source of protein	Number of residue	YHY (10) ^a	LHW (6.8) ^a	LHG (6.8) ^a	Average relative antioxidative activity
1. Fish protein (Salmon)
- Actin	375	12	12	30	1.08
- Myosin heavy chain	228	2	16	19	1.13
- Myosin Light chain ^b	186	9	8	16	1.36
- Tropomyosin	284	7	9	24	1.04
2. Milk protein
- Alpha S1 Casein	214	6	10	14	1.04
- Alpha-S2 casein	222	7	8	17	1.08
- Alpha lactalbumin	142	7	10	13	1.59
- Beta casein	214	8	10	21	1.36
- Beta lactoglobulin ^b	178	11	12	16	1.69
- Lactoferrin	117	2	4	13	1.16
3. Soy protein
- Conglycinin alpha chain	605	23	27	52	1.27
- Conglycinin beta chain	439	24	22	40	1.51
- 7S Globulin ^b	427	28	29	41	1.77
- 2S Albumin	158	7	7	13	1.30
- Glycinin	562	27	29	53	1.47
4. Cotton protein
- Legumin A	509	21	21	38	1.20
- Legumin B ^b	516	19	21	45	1.24
5. Canola protein
- 11S globulin ^b	509	22	25	54	1.49
6. Barley protein
- Globulin	224	4	4	14	0.73
- B-hordein	290	14	14	46	1.89
- C-hordein ^b	310	15	15	69	2.33
- D-hordein	757	37	27	63	1.30
7. Buckwheat protein
- 13S Globulin ^b	565	28	24	43	1.30
8. Rice protein
- Globulin	186	6	7	10	0.94
- Glutelin	510	20	25	57	1.49
- Prolamin ^b	156	12	13	19	2.16
9. Rye protein
- Glutenin subunit X ^b	754	11	10	43	0.62
- Glutenin subunit Y	713	7	8	43	0.58
10. Sunflower protein
- Albumin	295	2	2	12	0.39
- 11S Globulin ^b	493	15	19	52	1.36
11. Oat protein
- 11S globulin ^b	527	23	22	58	1.45
- 12S globulin	518	23	20	53	1.34
12. Pea protein
- Convicillin	571	24	25	51	1.34
- Legumin A	517	19	27	54	1.59
- Legumin J	503	23	26	47	1.50
- Vicillin ^b	459	24	23	42	1.46
13. Wheat protein
- Alpha/Beta Gliadin	286	14	16	36	1.73
- Gamma gliadin ^b	327	15	16	50	1.83
- HMW glutenin	405	6	6	29	0.74
- LMW glutenin	388	11	11	55	1.44
14. Egg protein
- Apovitellenin ^b	106	4	10	13	1.85
- Lipovitellin and Phosphovitin	1,850	42	42	67	0.63
- Ovalbumin	386	20	15	30	1.31
- Ovomucoid	210	12	8	13	1.25
- Ovotransferrin	705	28	29	68	1.33
^aThe relative activity was tabulated from the data of Saito et al.[18] ^bProtein fraction in bold showed the highest average relative antioxidative activity in each protein source.

The average relative antioxidative activity value was calculated as the following equation.

(1)

There are different ways of producing biologically active peptide from protein; for example, enzymatic hydrolysis with digestive enzyme either gastro-intestinal enzyme or enzyme from microorganisms, fermentation with proteolytic starter cultures and acid hydrolysis. Apart from these methods, there is useful “BIOPEP” software, which provides computer prediction in silico of proteolytic enzyme digestion, thereby yielding bioactive peptides.[10]

C-hordein fraction yields the highest average relative antioxidative activity when the BIOPEP software is applied. Trypsin and pepsin were applied to C-hordein as shown in Fig. 2. The total homology similarity segments used in gastrointestinal enzyme digestion was 79 segments. The homology segment presented at the same position was counted as one. After applying two enzymes of pepsin and trypsin in the BIOPEP digestive model system, there were 53 homology segments resistant to digestion system. Those peptides are underlined in Fig. 2.

Figure 2 BIOPEP print-out of digested C hordein after a trypsin-pepsin combined action.

Most of resistant sequence in C-hordein is PQQ (Pro-Gln-Gln). It has been indicated that barley protein in C-hordein fraction consists of repeated sequences based on an octapeptide PQQPFPQQ (Pro-Gln-Gln-Pro-Phe-Pro-Gln-Gln).[29] Kawase et al.[30] showed that deamination and fragmentation caused the diminution of antioxidative activity of C-hordein. But the repeated sequence region of C-hordein was not hydrolyzed by trypsin digestion because it has few lysine, arginine and glutamic acid residues. Furthermore, they also hypothesized that this structure is very important to the antioxidative activity of prolamin.

It is possible, however, that proteins containing bioactive peptide segments may have not been included in the database so far reported. This possibility should be confirmed in the future. Nakai et al.[4] stated that since the protein sequences reported in the literature by different researchers from different laboratories were used in this study, the reliability of the observed data should be dependent on their consistency. Thus, the prediction herein made could be an approximation or simply showing a trend.

CONCLUSIONS

SPCS results indicated that hydrophobic amino acids such as isoleucine, leucine, phenylalanine, tryptophan, tyrosine, and valine are required amino acid residues for composing antioxidative peptide segments to inhibit lipid peroxidation. HSS-BIOPEP combined software can predict the most potential antioxidative protein source to be C-hordein in barley.

ACKNOWLEDGMENTS

The authors would like to express sincere gratitude to University of British Columbia, Canada for the financial support via Southeast Asian Regional Center for Graduate Study and Research in Agriculture (SEARCA).