Changes in the enzyme-induced release of bitter peptides from wheat gluten hydrolysates

Abstract

Extraction of wheat gluten hydrolysates prepared using Proteax with isobutyl alcohol has opened a new avenue for exploring the release characteristics of bitter peptides. This report contains the first search for suitable extraction conditions for bitter peptides and the first summation of the changes in the enzyme-induced release of bitter peptides as the hydrolysis progresses. Effects of the amino acid sequence, amino acid composition and molecular weight distribution on the bitterness intensity of the isobutyl alcohol extracts are discussed. The results show that the amino acid composition and molecular weight distribution had a more significant impact on the bitterness intensity. Bitter peptides that had a higher proportion of bitter amino acids were gradually released within the first 2 h of the hydrolysis reaction, then the content of long-chain bitter peptides gradually reduced, so the bitterness intensity increased first and then decreased. It was found that the peptide fraction ranging 500–1000 Da had the strongest bitter taste, which could be further hydrolyzed for scientific production of small peptide powders.

---

Introduction

Currently, more than half of the bioactive peptides¹ described in the literature and found in on-line databases are small peptides (2–6 amino acids). Small peptides have been reported to be more rapidly and evenly absorbed² than large peptides. There is a pragmatic explanation for the literature phenomenon as many industrial processes using enzymatic hydrolysis liberate small peptides. Wheat gluten (WG) is an abundant, relatively inexpensive and safely edible source of natural plant protein.³ The enzymatic hydrolysis of WG to produce small peptides has been a promising area with far-reaching significance in theory and practice. Sequential hydrolysis of WG using endo- and exo-peptidase could not only improve the utilization rate of the protein and the yield of small peptides⁴ but could also reduce the bitter taste of the peptide powder,^5,6 which provided new thinking and a new direction for the exploitation of small peptide powders. Although sequential hydrolysis could provide many of the targeted desirable characteristics, the bitter taste couldn’t be removed completely and was still the main limiting factor for various applications of protein hydrolysates. It is important to investigate the changes in the enzyme-induced release of bitter peptides during the process of sequential hydrolysis.

Proteax is a proteolytic enzyme preparation manufactured through a fermentation process with a selected strain of Aspergillus oryzae. Proteax has provided high endo- and exo-peptidase activities and was suitable for preparing a low-bitterness small peptide powder.⁷ Considering its exploitation value, it is very necessary to optimize the hydrolysis reaction system with Proteax. The hydrolysis reaction system is a complex enzymatic hydrolysis process. The primary task for this report was to summarize the changes in the enzyme-induced release of bitter peptides as the hydrolysis progressed and to expatiate the debittering effect of exo-peptidase, based on the enzymatic properties and knowledge of the structure of bitter peptides. Finally, some useful suggestions are proposed for the preparation of small peptide powders using sequential hydrolysis.

Previous research has mainly focused on the formation mechanism,^8,9 isolation and purification,¹⁰ and removal methods for bitter peptides.¹¹ Until recently, there have been relatively few studies on the release characteristics of bitter peptides in protein hydrolysates. Due to the higher hydrophobicity of bitter peptides, isobutyl alcohol¹² could be a useful solvent for the extraction of bitter peptides. A better understanding of the structural properties of bitter peptides is essential for control of their properties during the processing and application of small peptide powders. For the above purpose, the main objective for the current research was to determine the amino acid composition, molecular weight distribution and amino acid sequence of bitter peptides obtained using the hydrolysis reaction system of Proteax. A combination of sensory analysis and electronic tongue measurements was employed to elucidate the changing trend of the bitterness intensity. The effect of the amino acid composition, molecular weight distribution and amino acid sequence on the bitterness intensity is emphatically discussed. This knowledge will be very helpful for application of the sequential hydrolysis and the scientific production of small peptide powders.

Experimental

Materials and reagents

WG (75% protein, wet base) was provided by the Nan Tong Lian Hai Wei Jing Co., Ltd. (Jiangsu, China). Proteax (1400 U g⁻¹, combined proteases) was donated by the Amano Enzyme Inc (Nagoya, Japan). All the other reagents and chemicals were of analytical grade.

Enzymatic hydrolysis of wheat gluten

WG was dispersed as a 5% (w/v) solution in deionised water and stirred for 1 h at room temperature. The suspension was adjusted to pH 7.0 with 1 mol L⁻¹ sodium hydroxide and kept at 50 °C in a water bath. The reaction was then initiated by addition of Proteax, with an enzyme-to-substrate (E : S) ratio of 1 : 50 (w/w), and continued for up to 6 h. The proteolysis reaction was stopped at intervals of 5, 10, 15, 30, 60, 120, 180, 240, 300 and 360 min, and the enzyme was immediately inactivated through a heat treatment in boiling water for 15 min. Then the resulting hydrolysates were centrifuged at 10 000g for 15 min at 4 °C in a ZOPR-52D refrigerated centrifuge (Hitachi Koki Co. Ltd., Tokyo, Japan). The supernatants were freeze-dried and stored at −20 °C.

Sensory analyses

All experiments were performed in compliance with the relevant laws and institutional guidelines of Jiangnan University (Jiangsu, China). The study was approved by the Ethics Committee of Jiangnan University, and written informed consent was obtained from all volunteers before participation. The sensory evaluation panelists (5 male and 5 female, aged between 20 and 35 years) were trained three times a week with quinine standards for a period of 1 month. The ten trained panelists were chosen to evaluate the bitter taste of the wheat gluten hydrolysates (WGHs) and isobutyl alcohol extracts. The standard quinine solutions were presented in concentrations of 0, 8.0 × 10⁻⁶, 1.6 × 10⁻⁵, 2.4 × 10⁻⁵, 3.2 × 10⁻⁵ and 4.0 × 10⁻⁵ g mL⁻¹, which were scored as 0, 1, 2, 3, 4 and 5, respectively. The standard conditions for sensory evaluation were a room temperature of 25 °C with the WGHs at a concentration of 1% (g mL⁻¹) and pH 6.5. The final values given by the panelists were averaged.

Taste dilution analysis

The bitterness of the WGHs after different enzymatic hydrolysis times (0–6 h) was evaluated using a TDA method,¹³ with some minor changes. Briefly, freeze-dried enzymatic hydrolysates (1 g) were dispersed in 100 mL of deionised water at room temperature. The WGHs were diluted stepwise using a 1 : 1 ratio with deionized water and then presented to a trained sensory panel (5 male and 5 female, aged between 20 and 35 years) in order of increasing solute concentration, and judged in a triangle test. The taste dilution (TD) factor was defined as the dilution at which a taste difference between the diluted sample and two blanks (1% WG solution) could just be detected. The TD value reported is the average value from the evaluation results provided by the inspectors. The differences among the results should not be more than one of the dilution levels.

Isobutyl alcohol extraction

10 g samples of freeze-dried WGHs obtained after different enzymatic hydrolysis times (10 min, 15 min, 30 min, 120 min and 300 min) were dispersed in deionised water at the required concentration (5%, 10%, 15%, 20% and 25% (w/v)). Then the solution was mixed with an equal volume of isobutyl alcohol to give a two-phase system. The mixture was shaken for about 2 min. When the liquid had divided into two distinct layers after 15 min, the isobutyl alcohol phase and aqueous phase were separated, freeze-dried and stored at −20 °C for further analysis. The nitrogen recovery for the two phases was measured separately using the micro-Kjeldahl procedure (N × 5.7).

Analysis of the amino acids

The amino acid composition of the isobutyl alcohol extracts was determined according to a method reported by Fekkes et al.¹⁴ The amino acids in the sample were analyzed using an Agilent liquid chromatograph 1100 with a UV detector operated at 338 nm. An ODS Hypersil (250 × 4.6 mm) column was used at 40 °C with the mobile phase delivered at a flow rate of 1 mL min⁻¹. For the determination of the free amino acids, the sample was diluted to 50 mL in volumetric flasks using trichloroacetic acid (10%, v/v). After incubation for 1 h at room temperature, the solution was filtered through Whatman filter paper no. 4. The filtrate was centrifuged at 7000 × g for 10 min, and the supernatant was stored at 4 °C before injection. For the determination of the total amino acids, the sample was hydrolyzed at 110 °C for 24 h with 6 mol L⁻¹ HCl in evacuated sealed tubes. Then the sample was transferred into a 25 mL volumetric flask, shaken vigorously and then filtered. The filtrate was then transferred into a 10 mL beaker and placed in a vacuum desiccator to remove the hydrochloric acid. The dried sample was then dissolved in 1 mL 0.02 mol L⁻¹ HCl and kept at 4 °C before injection. External standards were used for quantification. The content of the different amino acids, including the free amino acids or the total amino acids, is expressed as mg g⁻¹ of isobutyl alcohol extract.

Determination of the molecular weight distribution

The MW distribution profiles of the isobutyl alcohol extracts were determined using high-performance gel-filtration chromatography (HPGFC) with a TSK gel G2000 SWXL 7.8 × 300 mm column (Tosoh Co., Tokyo, Japan) and a Waters 1525 liquid chromatography system (Waters Co., Milford, MA, USA) according to the method of Liu et al.,¹⁵ with some minor changes. The HPGFC was carried out with the mobile phase (acetonitrile/water/trifluoroacetic acid, 10/90/0.1, v/v/v) at a flow rate of 0.5 mL min⁻¹ and monitored at 220 nm and 30 °C. The standards used from Sigma were cytochrome C (12.5 kDa), aprotinin (6.5 kDa), bacitracin (1450 Da), tetrapeptide GGYR (451 Da), and tripeptide GGG (189 Da).

Electronic tongue measurements

Electronic tongue measurements were carried out with the taste sensing system TS-5000Z (Insent Inc., Japan). This electronic tongue was equipped with five lipid membrane sensors, indicating different taste qualities, and corresponding reference electrodes. These sensors represent the gustatory stimuli umami (SB2AAE), bitterness (SB2C00), saltiness (SB2CT0) and sourness (SB2CA0), and the nociceptive sensation astringency (SB2AE1). The sensor outputs were consequently obtained as values. Each sample was measured four times with the sensor for bitterness. One measurement cycle consisted of a reference solution (V_r), followed by the sample solution (V_s), a short (2 × 3 s) cleaning procedure, the aftertaste detection (V′_r) in the reference solution, and a cleaning procedure for 330 s. The aftertaste was related to the change in the membrane potential caused by substances still adsorbed on the lipid membrane after the short cleaning procedure. The calculation formulae for the sensor output for taste and aftertaste were:

Sensor output for taste = V_r − V_s (1)

Sensor output for aftertaste = V′_r − V_s (2)

Statistical analysis

All of the tests for the WGHs and isobutyl alcohol extracts were performed in triplicate and the values are expressed as mean ± standard deviation (SD). The analysis of the variance (ANOVA) of the data was carried out using SPSS software (version 16.0, SPSS Inc., Chicago, IL, USA). The differences between the mean values were evaluated using Duncan’s multiple range test with a confidence interval of 95%.

Partial least squares regression (PLSR) analysis was used to explore the relationship between the isobutyl alcohol extraction, sensory data, amino acid composition and molecular weight distribution using UNSCRAMBLER ver. 9.7 (CAMO ASA, Oslo, Norway). All variables were centered and standardized (1/Sdev) so that each variable has a unit variance and zero mean before applying the PLSR analysis. By applying PLSR analysis to the standardized data, the importance of the peaks for each attribute could be compared quantitatively based on regression coefficients and loading weights for each predictor or an X variable used in PLSR models.

Results and discussion

Sensory characterization of wheat gluten hydrolysates

Sensory characterization of the wheat gluten hydrolysates was performed using the bitterness intensity score and taste dilution factor respectively. As shown in Fig. 1, the bitterness intensity of the WGHs increased continuously within the first 120 min of the enzymatic hydrolysis reaction. The bitterness intensity of the WGHs reached its highest level at 120 min, for which the bitterness intensity score and taste dilution factor were 3.72 and 32 respectively. The bitterness intensity of the WGHs decreased continuously during 120–300 min, and the bitterness intensity score and taste dilution factor were 1.33 and 16 at 300 min respectively. The bitterness intensity score and taste dilution factor were no longer reduced and began to stabilize after 300 min. The bitterness intensity of the WGHs increased at first and then decreased within 300 min, which indicates that the release characteristics of the bitter peptides obviously changed and the resulting WGHs should be extracted with isobutyl alcohol within the above-mentioned period of time. The bitterness intensity scores and taste dilution factors at 5 min were so low that the release characteristics of the bitter peptides couldn’t be obtained. Along with the prolonging of the reaction time, the taste dilution factor increased after 10 min, 15 min, 30 min and 120 min. The bitterness intensity scores decreased significantly at 300 min. Therefore, in order to study the release characteristics of bitter peptides, it is recommended to extract the bitter peptides from the WGHs at the above time points.

Fig. 1 Bitterness intensity scores and taste dilution factors of the wheat gluten hydrolysates.

Extraction of bitter peptides with isobutyl alcohol

When isobutyl alcohol was used to extract strongly hydrophobic bitter substances from the WGHs, differences in the concentration of the WGHs could affect the nitrogen recovery for the isobutyl alcohol phase and aqueous phase. As shown in Fig. 2, the nitrogen recovery for the isobutyl alcohol phase increased and the nitrogen recovery for the aqueous phase decreased as the concentration of the WGHs increased, which is consistent with previous studies.¹⁶ When the volume of the aqueous solution is kept constant, the dissolving capacity is limited, so increase of the concentration of the WGHs was beneficial for transferring hydrophobic bitter substances from the aqueous phase to the isobutyl alcohol phase. During the process of the transfer, the bitterness intensity of the isobutyl alcohol extracts increased gradually and the bitterness intensity of the aqueous extracts decreased gradually.

Fig. 2 Effect of the concentration of the wheat gluten hydrolysates on the nitrogen recovery ((A) 10 min; (B) 15 min; (C) 30 min; (D) 120 min; (E) 300 min) and bitterness intensity ((a) 10 min; (b) 15 min; (c) 30 min; (d) 120 min; (e) 300 min).

When the concentration of the WGHs reached a certain level, further increase in the concentration could not reduce the bitterness intensity of the aqueous extracts. The most reasonable interpretation is that it was difficult to extract free amino acids using isobutyl alcohol because of the low molecular weight. Some of the bitter-tasting free amino acids remained in the aqueous phase, and the bitter taste could not be completely removed. Bitter-tasting free amino acids have a higher bitter taste threshold^17,18 than bitter peptides under normal circumstances and would not generate a strong bitter taste. When the bitterness intensity score of the aqueous extract was reduced to about 0.2, all the bitter peptides could be considered to have been extracted with isobutyl alcohol. It is worth noting that an excessive increase in the concentration of the WGHs would cause bitterless substances in the aqueous phase to be extracted into the isobutanol phase, which would cause confusion during identification of the structural characteristics of the bitter peptides.

In order to extract bitter peptides efficiently from the WGHs, the amount of WGHs at 10 min, 15 min, 30 min, 120 min and 300 min should be 10%, 15%, 15%, 20% and 15% respectively. When the above conditions were met, the bitterness intensity score of the isobutyl alcohol extracts (abbreviated respectively as Pro-10M, Pro-15M, Pro-30M, Pro-120M and Pro-300M) was 2.74, 3.21, 3.60, 4.18 and 3.39 respectively. The results revealed that the bitterness intensity of the bitter peptides increased first and then decreased during the process of releasing the bitter peptides.

Amino acid sequence of the bitter peptides

The release characteristics of the bitter peptides changed in terms of the amino acid sequence, the proportion of bitter amino acids and the relative molecular weight distribution. In the following sections, the changes are discussed in detail. For the Proteax reaction with WG, the peptide bonds were preferentially cleaved at the carboxyl-terminus of glutamine (Gln), serine (Ser), threonine (Thr) and methionine (Met), followed by cleavage at the carboxyl-terminus of arginine (Arg), and finally at the carboxyl-terminus of lysine (Lys), alanine (Ala), phenylalanine (Phe) and leucine (Leu). According to the primary structures of wheat glutenin and wheat gliadin provided by the National Center for Biotechnology Information, the amino acid sequences of the bitter peptides (the proportion of bitter amino acids¹⁹ was more than 50%) were identified through cleavage of the primary structures and are listed in Table 1.

Table 1 Amino acid sequence of bitter peptides from the wheat gluten hydrolysates

Protein composition	Amino acid sequence of bitter peptide
	Peptide bond hydrolysis (the carboxyl-terminus of glutamine, serine, threonine and methionine)	Peptide bond hydrolysis (the carboxyl-terminus of arginine)	Peptide bond hydrolysis (the carboxyl-terminus of lysine, alanine, phenylalanine and leucine)
Low molecular weight glutenin subunit	HIPGLERPS, PLPPQ, HHHQ, PIQ, FPQ, PPLS, PPFS, PVLPQ, LPPFS, PPFS, PILPQ, PVLLQ, IPFVHPS, ILQ, LNPCKVFLQ, PVAM, LPQ, IPQ, RYEAIRAIVYS, IILQ, FLQ, PHQ, LELM, IALRT, LPT, RVPFGVGT	HIPGLER, PLPPQ, HHHQ, PIQ, FPQ, PPLS, PPFS, PVLPQ, LPPFS, PPFS, PILPQ, PVLLQ, IPFVHPS, ILQ, LNPCKVFLQ, PVAM, LPQ, IPQ, YEAIR, AIVYS, IILQ, FLQ, PHQ, LELM, IALR, LPT, VPFGVGT	HIPGL, PL, PPQ, HHHQ, PIQ, PPL, PPF, PVL, PIL, IPF, VHPS, IL, VF, PVA, IR, IVYS, IIL, PHQ, VPF
High molecular weight glutenin subunit	AKRLVLFGIVVIALVALT, LPWS, PLQ, GYYPS, YYPGQ, GYYPT, GHYPAS, HPQ, PHYPAS, PGHYPAS, GYYIT, PCHVS, PVAQLPT	LVLFGIVVIALVALT, LPWS, PLQ, GYYPS, YYPGQ, GYYPT, GHYPAS, HPQ, PHYPAS, PGHYPAS, GYYIT, PCHVS, PVAQLPT	AVPPK, RL, VL, GIVVIA, PWS, PL, GYYPS, YYPGQ, GYYPT, GHYPA, HPQ, PHYPA, PGHYPA, GYYIT, PCHVS, PVA, QL
Wheat gliadin	FLILALLAIVAT, AVRVPVPQ, VPLVQ, FPPQ, PYPQ, PFPS, PYLQ, PFPQ, PFPPQ, LPYPQ, PPPFS, PIS, ILPQ, LIPCRDVVLQ, VQPLQ, AIHNVVHAIILHQ, YPS, LPQ, LPRMCNVYIPPYCS	FLILALLAIVAT, AVR, VPVPQ, VPLVQ, FPPQ, PYPQ, PFPS, PYLQ, PFPQ, PFPPQ, LPYPQ, PPPFS, PIS, ILPQ, LIPCR, DVVLQ, VQPLQ, YPS, LPQ, AIHNVVHAIILHQ, LPR, MCNVYIPPYCS	FL, IL, VR, VPVPQ, VPL, PF, PPQ, PYPQ, PPPF, PIS, IL, IPCR, DVVL, VQPL, YPS, IHNVVHA, IIL, PR, MCNVYIPPYCS

---

Some of the hydrolysis sites could not be cleaved in practice under the influence of complex molecular configurations. The amino acid sequences for bitter peptides in Table 1 might be the actual sequence, or might be an important part of the real amino acid sequence of the bitter peptides from the WGHs. In either case, the amino acid sequence of the bitter peptide is an important means of exploring the bitter taste characteristics, which would result in a change in the bitterness intensity. During the initial reaction period, bitter peptides with a high proportion of bitter amino acids were released from the WG, and part of them had the particular amino acid sequence (a basic amino acid adjacent to a hydrophobic amino acid,²⁰ basic amino acids located at the N-terminus) that elicits a strong bitter taste. These particular amino acid sequences (HIPGLERPS, RYEAIRAIVYS, IALRT, AVRVPVPQ, RVPFGVGT, AKRLVLFGIVVIALVALT, LIPCRDVVLQ, LPRMCNVYIPPYCS) were further hydrolyzed through cleavage of Arg. The Proteax finally cleaved the peptide bonds at the carboxyl-terminus of Lys, Ala, Phe and Leu. The newly generated peptides (HIPGL, PL, PPL, PPF, PVL, PIL, IPF, IL, VF, IIL, VPF, RL, VL, QL, FL, VPL, PF, PPPF, DVVL, VQPL) with Phe and Leu located at the C-terminus had a small bitter taste threshold. Therefore, the effect of the amino acid sequences produced on the bitterness intensity was to cause a decrease first and then an increase.

Amino acid composition of the peptides

WGHs at different enzymatic hydrolysis times (0–5 h) were extracted with isobutyl alcohol. These extracts showed the different amino acid compositions of the peptides (Table 2). The content of glutamine (Gln), proline (Pro), glycine (Gly) and Leu was relatively high, which is consistent with the amino acid composition of WG. The results indicated that hydrophilic amino acids in hydrophobic peptides were also extracted using isobutyl alcohol, and the content of hydrophobic amino acids could only vary within a certain range.

Table 2 Amino acid composition of the peptides (mmol per 100 g of isobutyl alcohol extract)^a

Peptide-bound amino acids^b	Concentration (mmol per 100 g) (mean ± SD)
	Pro-10M	Pro-15M	Pro-30M	Pro-2H	Pro-5H
a The values in the same row followed by different letters are significantly different (p < 0.05).b Peptide-bound amino acids were calculated by subtracting free amino acids from total amino acids.c Aspartic acid + asparagine.d Glutamic acid + glutamine.e Not determined.f Total hydrophobic amino acids contained Gly, Ala, Val, Leu, Pro, Met, Phe, Trp and Ile.g Total bitter taste amino acids contained Pro, Tyr, Lys, His, Arg, Val, Met, Phe, Ile, Leu, Trp.
Asp^c	20.66 ± 0.82^A	19.28 ± 0.43^B	18.31 ± 0.21^B	17.08 ± 0.58^C	15.30 ± 0.65^D
Glu^d	209.64 ± 6.97^A	198.38 ± 6.31^AB	187.41 ± 7.11^BC	181.25 ± 7.14^C	178.51 ± 6.71^C
Ser	27.05 ± 0.65^A	25.72 ± 0.72^AB	24.36 ± 1.32^BC	23.53 ± 0.70^CD	22.67 ± 0.31^D
His	3.80 ± 0.16^ABC	4.02 ± 0.08^A	3.32 ± 0.50^C	3.46 ± 0.20^BC	3.93 ± 0.04^AB
Gly	41.58 ± 1.89^A	41.03 ± 2.07^A	40.94 ± 1.34^A	39.83 ± 1.53^A	40.72 ± 2.01^A
Thr	5.63 ± 0.28^A	5.34 ± 0.27^A	5.26 ± 0.30^A	3.72 ± 0.18^B	2.78 ± 0.17^C
Arg	15.34 ± 0.19^AB	15.95 ± 0.89^A	14.35 ± 1.05^B	10.92 ± 0.42^C	8.54 ± 0.79^D
Ala	23.02 ± 0.49^AB	23.29 ± 0.39^AB	24.30 ± 1.17^A	22.21 ± 1.37^BC	20.71 ± 0.78^C
Tyr	11.81 ± 0.29^B	12.34 ± 0.45^AB	12.97 ± 0.72^A	12.01 ± 0.34^B	10.84 ± 0.31^C
Cys	1.34 ± 0.12^A	1.22 ± 0.11^AB	1.11 ± 0.16^B	0.70 ± 0.10^C	0.65 ± 0.09^C
Val	24.56 ± 1.02^A	24.81 ± 0.76^A	25.51 ± 1.49^A	23.73 ± 0.95^A	19.60 ± 1.58^B
Met	7.25 ± 0.15^A	7.48 ± 0.35^A	7.43 ± 0.70^A	6.86 ± 0.69^A	4.34 ± 0.57^B
Phe	24.20 ± 1.30^CD	25.86 ± 1.62^BC	27.80 ± 1.43^AB	28.80 ± 0.96^A	22.42 ± 1.68^D
Ile	24.86 ± 1.17^C	26.14 ± 0.86^BC	27.61 ± 0.80^AB	29.54 ± 0.71^A	26.24 ± 1.56^BC
Leu	46.02 ± 2.58^B	46.55 ± 2.93^B	49.72 ± 1.88^AB	51.63 ± 3.36^A	34.60 ± 1.91^C
Lys	4.69 ± 0.27^C	5.60 ± 0.20^B	6.13 ± 0.39^B	7.10 ± 0.60^A	5.59 ± 0.16^B
Pro	95.95 ± 4.90^C	104.14 ± 7.51^BC	109.86 ± 5.63^AB	114.90 ± 5.27^AB	117.52 ± 6.74^A
Trp	n.d^e	n.d^e	n.d^e	n.d^e	n.d^e
Total	587.41 ± 5.25^A	587.16 ± 5.70^A	586.38 ± 8.44^A	577.26 ± 5.82^A	534.98 ± 6.78^B
Total hydrophobic amino acids^f	287.46 ± 10.03^BC	299.30 ± 7.08^B	313.16 ± 3.83^A	317.49 ± 6.41^A	286.15 ± 4.21^C
Total bitter taste amino acids^g	258.49 ± 10.11^C	272.88 ± 7.48^B	284.69 ± 3.36^AB	288.94 ± 5.50^A	253.63 ± 5.32^C

---

The hydrophobic peptides gradually had a higher proportion of hydrophobic amino acids on proceeding from 0–120 min of enzymatic hydrolysis. Two reasons could explain this trend. Firstly, degradation of the high molecular weight peptides generated low molecular weight peptides, some of which had a higher proportion of hydrophobic amino acids. Secondly, the WG had a large number of hydrophobic amino acids and exhibited poor solubility. For the Proteax reaction with WG, the peptide bonds at the carboxyl-terminus of Gln, Ser, Thr and Met were preferentially cleaved. The hydrophilic area of the WG was preferred for participation in hydrolysis reactions. The hydrophobic area of WG gradually participated in the enzyme reaction and constantly released bitter peptides, which contained a higher proportion of hydrophobic amino acids with an increasing degree of hydrolysis. When the proteolysis reaction was stopped, the hydrophobic area of WG form the macromolecular proteins by hydrophobic interaction and disulfide bond, which were removed from WGHs after centrifugation. Therefore, it was reasonable for the content of hydrophobic amino acids (Phe, Ile, Leu, Pro) to increase and the content of hydrophilic amino acids (Asp, Glu, Ser, Thr, Cys) to decrease.

The endo-activity of the Proteax was very low during the 120–300 min section of the enzymatic hydrolysis. Hydrophobic peptides, which had a high proportion of hydrophobic amino acids, were no longer generated. The Proteax had a high exo-activity for cutting free amino acids from the C-terminus of the hydrophobic peptides, and so the content of peptide-bound amino acids decreased from 317.49 mmol g⁻¹ to 286.15 mmol g⁻¹. Arg and Lys have a stimulating unit for bitter taste determinant sites,^21,22 and Tyr, Val, Met, Phe, Ile and Leu have a binding unit for bitter taste determinant sites. A decrease in the content of the above bitter amino acids would produce a debittering effect. Therefore, the effect of the amino acid composition on the bitterness intensity was to cause an increase first and then a decrease.

Molecular weight distribution of isobutyl alcohol extracts

The fraction below 180 Da mainly consisted of free amino acids, which are not the main bitter substances of the isobutyl alcohol extracts. Table 3 shows that the Pro-10M extract had the highest content of peptides ranging from 3000–6000 Da, which was only 4.63%. Moreover, the peptide chains of this peptide fraction are so long that it is difficult for them to contain a high proportion of bitter amino acids, so this peptide fraction also did not contain the main bitter substances. The peptide fraction above 6000 Da (ref. 23) did not have a bitter taste. The main bitter substances of the isobutyl alcohol extracts were concentrated within the peptide fractions ranging from 180–500 Da, 500–1000 Da and 1000–3000 Da. A bitter peptide has a smaller bitter taste threshold as the bitter taste determinant sites for peptides are distributed densely within a certain spatial range of the taste bud. In other words, if a bitter peptide has a higher proportion of bitter amino acids or a larger amount of bitter amino acids, it would have a smaller bitter taste threshold. At the beginning of the hydrolysis reaction, the proportion of bitter amino acids in the long-chain peptides (500–1000 Da and 1000–3000 Da) remained at a low level. Based on unexpected bitterness results from the hydrolysis reaction, some of the short-chain peptides (180–500 Da) that contained more bitter tasting amino acids had a stronger bitter taste. In the later stages of the hydrolysis reaction, the long-chain peptides that had a high proportion of bitter amino acids showed the strongest bitter taste.

Table 3 Molecule weight distribution of isobutyl alcohol extracts^a

Molecular weight (kDa)	Pro-10M (%)	Pro-15M (%)	Pro-30M (%)	Pro-2H (%)	Pro-5H (%)
a The values in the same row followed by different letters are significantly different (p < 0.05).
≥10	4.05 ± 0.22^A	2.66 ± 0.19^B	1.84 ± 0.06^C	1.11 ± 0.05^D	0.75 ± 0.07^E
6–10	9.78 ± 0.52^A	8.79 ± 0.30^B	6.96 ± 0.17^C	3.86 ± 0.19^D	2.14 ± 0.16^E
3–6	4.63 ± 0.18^A	3.73 ± 0.12^B	3.03 ± 0.19^C	1.94 ± 0.03^D	1.58 ± 0.14^E
1–3	17.91 ± 1.43^A	16.23 ± 1.40^AB	14.61 ± 0.52^B	11.44 ± 0.70^C	7.32 ± 0.49^D
0.5–1	25.21 ± 1.52^A	25.15 ± 1.59^A	24.76 ± 1.56^A	23.28 ± 1.40^A	14.92 ± 0.96^B
0.18–0.5	32.03 ± 2.11^D	35.36 ± 2.24^D	39.37 ± 1.82^C	44.52 ± 0.90^B	55.62 ± 2.07^A
≤0.18	6.39 ± 0.38^D	8.08 ± 0.17^C	9.43 ± 0.37^C	13.85 ± 0.93^B	17.67 ± 1.40^A

---

The content of the 500–1000 Da and 1000–3000 Da peptide fractions decreased respectively from 25.21% and 17.91% in the Pro-10M extract to 14.92% and 7.32% in the Pro-5H extract. The content of the 180–500 Da peptide fraction increased from 32.03% in the Pro-10M extract to 55.62% in the Pro-5H extract. The decrease of the relative average molecular weight caused different release characteristics of the bitter peptides, which should be discussed separately. The content of long-chain peptides in the isobutyl alcohol extracts gradually decreased during 0–120 min of the enzymatic hydrolysis, therefore the bitterness intensity of the isobutyl alcohol extracts must have decreased. However, the hydrolysis of the WG generated hydrophobic peptides, which gradually had a higher proportion of hydrophobic amino acids. The main bitter substances had a smaller bitter taste threshold as a whole, so the bitterness intensity of the isobutyl alcohol extracts gradually increased. The endo-activity of the Proteax was very low during the 120–300 min section of the enzymatic hydrolysis. The proportion of hydrophobic amino acids in the hydrophobic peptides no longer increased significantly. The long-chain peptides showed the strongest bitter taste, and decrease of their amount caused a reduction in the bitterness intensity of the Pro-2H extract. Therefore, the effect of the molecular weight distribution on the bitterness intensity was to cause an increase first and then a decrease.

Representation of the sensor response using an electronic tongue

For comparison of the intensity of the sensor output, the initial sensor output for the Pro-10M extract was defined as zero, so as to provide a frame of reference. The initial sensor outputs for the other treatments were converted into new voltage values using a TS-5000Z software system. A greater number represented a higher taste intensity. The intensity of the taste is a measure of the size of the taste response produced by the isobutyl alcohol extracts within a certain period of time. The sensor outputs of bitterness and bitter aftertaste (aftertaste-B) represent the instantaneous maximum intensity and the sustained intensity of the bitter taste, respectively. As shown in Fig. 3, the sensor outputs for bitterness and aftertaste-B for the isobutyl alcohol extracts both increased first and then decreased. The Pro-2H extract had the highest instantaneous maximum intensity (3.727) and sustained intensity (1.583). These results indicate that as the proportion of hydrophobic amino acids gradually increased, the bitter peptides had a longer duration of action at the taste bud and the bitter taste determinant sites of the bitter peptides became denser. This phenomenon stopped after 2 h of hydrolysis reaction, when the content of long-chain bitter peptides began to gradually reduce. The amino acid composition and molecular weight distribution had a more significant impact on the bitterness intensity than the amino acid sequence. In order to find a better way to achieve reduction of the bitter taste, the internal connection between the amino acid composition, molecular weight distribution and bitterness intensity should be further discussed.

Fig. 3 The sensor outputs for the isobutyl alcohol extracts obtained using the electronic tongue.

Relationship between the isobutyl alcohol extract, sensory attributes, amino acid composition and molecular weight distribution

ANOVA-PLSR was used to process the mean data accumulated from the HPLC analysis and the sensor outputs obtained using the electronic tongue. Seventeen kinds of amino acids and seven kinds of peptide fractions were used as variables in the subsequent PLSR analysis. The X-matrix was designed to represent the amino acid composition and molecular weight distribution; the Y-matrix was designed to represent the isobutyl alcohol extract and sensory variables. The calibrated explained variances for this model were PC1 = 81% and PC2 = 18%. PC1 versus 2 (Fig. 4) and PC2 versus 3 were explored. The PC2 versus 3 results are not presented here, as additional information was not gained through their examination. Furthermore, the PCs didn’t provide any predictive improvement in the Y-matrix obtained. Fig. 4 is presented as correlation loadings plot. The inner and outer large circles indicate 50% and 100% explained variances, respectively. Seven Y variables (Y1 (the sensor outputs for bitterness), Y2 (the sensor outputs for aftertaste-B), Pro-10M, Pro-15M, Pro-30M, Pro-2H, and Pro-5H) and twenty-four X variables (the peptide fraction above 10 000 Da, the peptide fraction ranging from 6000–10 000 Da, the peptide fraction ranging from 3000–60 000 Da, the peptide fraction ranging from 1000–3000 Da, the peptide fraction ranging from 500–1000 Da, the peptide fraction ranging from 180–500 Da, the peptide fraction below 180 Da, aspartic acid, glutamic acid, serine, histidine, glycine, threonine, arginine, alanine, tyrosine, cysteine, valine, methionine, phenylalanine, isoleucine, leucine, lysine and proline; arranged in the order listed here and numbered separately using Arabic numerals) were situated between the inner and outer ellipses respectively, indicating that they can be well explained using the PLSR model.

Fig. 4 An overview of the variation found in the mean data from the partial least squares regression (PLSR) correlation loadings plot for the isobutyl alcohol extracts. The model was derived using the amino acid composition and molecular weight distribution as the X-matrix and the extract and sensory attributes as the Y-matrix. Ellipses represent r2 = 0.5 and 1.0, respectively.

Pro-10M and Pro-5H were located to the left of the Y axis, and Pro-30M and Pro-2H were located to the right of the Y axis, so the X axis represents the bitter taste, which increased in intensity from left to right. Pro-10M and Pro-15M were located below the X axis, and Pro-5H and Pro-2H were located above the X axis, so the Y axis represents the average relative molecular weight, which decreased from bottom to top. According to the distribution of the independent variables, two groups of the independent variables were found to have a strong impact on the bitter taste of the isobutyl alcohol extracts. The most influential group consisted of phenylalanine, leucine and tyrosine. The second-most influential group contained valine, methionine, isoleucine, alanine, lysine and the peptide fraction ranging from 500–1000 Da. The proportion of bitter-tasting amino acids had the strongest influence on the bitterness intensity of the bitter peptides. The peptide fraction ranging from 500–1000 Da showed a stronger bitter taste than the other peptide fractions. The most likely explanation is that the peptide fraction ranging from 500–1000 Da provided a larger amount of bitter amino acids, on the premise of it containing a high proportion of bitter amino acids.

Conclusions

The release of the bitter peptides was found to be a dynamic evolution process, which is an existing phenomenon, regulated by the degree of hydrolysis, the enzyme active sites and the amino acid sequence of the protein. Changes in the characteristics of the bitter peptides gave rise to a different bitterness intensity throughout the process. The changes in the characteristics and the scientific principles behind the bitter taste changes were investigated, and then a trend for the enzyme-induced release of bitter peptides was proposed. The amino acid composition and molecular weight distribution had a more significant impact on the bitterness intensity than the amino acid sequence. The peptide fractions ranging from 180–500 Da, 500–1000 Da and 1000–3000 Da gradually had a higher proportion of bitter tasting amino acids, and presented a stronger bitter taste. This phenomenon stopped after 2 h of hydrolysis reaction and the content of long-chain bitter peptides began to gradually reduce, especially for the peptide fraction ranging from 500–1000 Da that had the strongest bitter taste. The bitterness intensity of the bitter peptides increased first and then decreased using the Proteax hydrolysis reaction system.

According to the above findings, optimization of the method could be further explored. For instance, it would be very worthwhile to design a complex enzymatic hydrolysis. Alcalase or Flavourzyme might be useful for further reducing the content of long-chain bitter peptides. In the future, more attention should be focused on scientific analysis of the production of small peptide powders when preparing a low-bitterness peptide powder by sequential hydrolysis with endo- and exo-peptidase.

Acknowledgements

The study was supported by the National High Technology Research and Development Program of China (863 Program, No. 2013AA102201), the Key R & D Programs of Jiangsu Province (BE2015327) and the Jiangsu province “Collaborative Innovation Center for Modern Grain Circulation and Safety” industry development program.

Notes and references

1. S. L. Lahrichi, M. Affolter, I. S. Zolezzi and A. Panchaud, J. Proteomics, 2013, 88, 83–91.
2. K. Horibe, Nihon Rinsho Geka Gakkai Zasshi, 1987, 88, 808–815.
3. L. Day, M. A. Augustin, I. L. Batey and C. W. Wrigley, Trends Food Sci. Technol., 2006, 17, 82–90.
4. A. Villanueva, J. Vioque, R. Sanchez-Vioque, A. Clemente, J. Bautista and F. Millan, Grasas Aceites, 1999, 50, 472–476.
5. L. Li, Z.-Y. Yang, X.-Q. Yang, G.-H. Zhang, S.-Z. Tang and F. Chen, J. Ind. Microbiol. Biotechnol., 2008, 35, 41–47.
6. F. Liu and M. Yasuda, J. Ind. Microbiol. Biotechnol., 2005, 32, 487–489.
7. B.-Y. Liu, K.-X. Zhu, W. Peng, X.-N. Guo and H.-M. Zhou, RSC Adv., 2016, 6, 27659–27668.
8. N. Ishibashi, T. Kubo, M. Chino, H. Fukui, I. Shinoda, E. Kikuchi and H. Okai, Agric. Biol. Chem., 1988, 52, 95–98.
9. N. Ishibashi, K. Sadamori, O. Yamamoto, H. Kanehisa, K. Kouge, E. Kikuchi, H. Okai and S. Fukui, Agric. Biol. Chem., 1987, 51, 3309–3313.
10. X. Liu, D. Jiang and D. G. Peterson, J. Agric. Food Chem., 2014, 62, 5719–5725.
11. B. C. Saha and K. Hayashi, Biotechnol. Adv., 2001, 19, 355–370.
12. G. Lalasidis, Annales De La Nutrition Et De Lalimentation, 1978, 32, 709–723.
13. W. H. Seo, H. G. Lee and H. H. Baek, J. Food Sci., 2008, 73, S41–S46.
14. D. Fekkes, A. Vandalen, M. Edelman and A. Voskuilen, J. Chromatogr. B: Biomed. Sci. Appl., 1995, 669, 177–186.
15. P. Liu, M. Huang, S. Song, K. Hayat, X. Zhang, S. Xia and C. Jia, Food Bioprocess Technol., 2010, 5, 1775–1789.
16. G. Lalasidis and L. B. Sjoberg, J. Agric. Food Chem., 1978, 26, 742–749.
17. T. K. Murray and B. E. Baker, J. Sci. Food Agric., 1952, 3, 470–475.
18. N. Ishibashi, Y. Arita, H. Kanehisa, K. Kouge, H. Okai and S. Fukui, Agric. Biol. Chem., 1987, 51, 2389–2394.
19. L. Liao, C. Y. Qiu, T. X. Liu, M. M. Zhao, J. Y. Ren and H. F. Zhao, J. Cereal Sci., 2010, 52, 395–403.
20. K. X. Zhu, H. M. Zhou and H. F. Qian, Process Biochem., 2006, 41, 1296–1302.
21. N. Ishibashi, I. Ono, K. Kato, T. Shigenaga, I. Shinoda, H. Okai and S. Fukui, Agric. Biol. Chem., 1988, 52, 91–94.
22. N. Ishibashi, K. Kouge, I. Shinoda, H. Kanehisa and H. Okai, Agric. Biol. Chem., 1988, 52, 819–827.
23. K. H. Ney, Eur. Food Res. Tech. Z. Lebensm.Unters. Forsch., 1971, 2, 64.

---