Application of taste sensing system for characterisation of enzymatic

hydrolysates from shrimp processing by-products

Imelda W.Y. Cheung, Eunice C.Y. Li-Chan

The University of British Columbia, Faculty of Land and Food Systems, Food Nutrition and Health Program, 2205 East Mall, Vancouver, British Columbia V6T 1Z4, Canada
a r t i c l e i n f o
Article history:
Received 27 May 2013
Received in revised form 21 August 2013
Accepted 2 September 2013
Available online 11 September 2013
Keywords:
Taste sensors
Hydrolysates
Sensory evaluation
Taste properties
Shrimp processing by-products
a b s t r a c t
The objective of this study was to investigate the potential of an instrumental taste-sensing system to
distinguish between shrimp processing by-products hydrolysates produced using different proteases
and hydrolysis conditions, and the possible association of taste sensor outputs with human gustatory
assessment, salt content, and bioactivity. Principal component analysis of taste sensor output data cate-
gorised samples according to the proteases used for hydrolysis. High umami sensor outputs were char-
acteristic of bromelain- and Flavourzyme-produced hydrolysates, compared to low saltiness and high
bitterness outputs of Alcalase-produced hydrolysates, and high saltiness and low umami outputs of
Protamex-produced hydrolysates. Extensively hydrolysed samples showed higher sourness outputs. Salt-
iness sensor outputs were correlated with conductivity and sodium content, while umami sensor
responses were related to gustatory sweetness, bitterness and umami, as well as angiotensin-I converting
enzyme inhibitory activity. Further research should explore the dose dependence and sensitivity of each
taste sensor to specic amino acids and peptides.
2013 Elsevier Ltd. All rights reserved.
1. Introduction
Extensive research has been performed in recent decades on
developing bioactive peptides and protein hydrolysates for use in
functional foods and nutraceuticals (Udenigwe & Aluko, 2011).
Numerous studies have emerged investigating the use of underval-
ued seafood sources and processing by-products of no or low
market value to produce products with potential bioactivity,
including angiotensin-I converting enzyme (ACE) inhibitory activ-
ity, antioxidant activity, anti-diabetic activity, anticancer activity,
or antimicrobial activity (Harnedy & FitzGerald, 2011; Kim &
Wijesekara, 2010). Unfortunately, enzymatic hydrolysis of proteins
often releases peptides that contain unfavourable taste, limiting
the marketability of the end products (FitzGerald & OCuinn, 2006).
Taste is a criterion that governs the acceptability of a product
in the consumer market, yet there are limited reports on the
taste properties of hydrolysates (Cheison & Wang, 2007; Cheung
& Li-Chan, 2010; Leksrisompong, Mirale, & Drake, 2010; Liao, Qiu,
Liu, Zhao, & Ren, 2010; Schlichtherle-Cerny & Amad, 2002).
Traditionally, taste properties of a new product are assessed
using trained personnel but this process is very subjective,
exhaustive and costly; therefore there is great interest in apply-
ing instrumental analysis as a complementary approach for taste
characterisation. Taste sensing devices, such as the electronic
tongue and taste sensing system, have recently been developed
and were shown to be successful in the taste assessment of tea
(Chang et al., 2010; Chen, Zhao, Guo, & Wang, 2010; Hayashi,
Chen, Ikezaki, & Ujihara, 2008), soy sauce (Ou-Yang, Zhao, Chen,
& Lin, 2011), apple juice (Kovacs et al., 2011), beer (Rudnitskaya
et al., 2009), wine (Rudnitskaya, Rocha, Legin, Pereira, & Marques,
2010), total enteral nutrients (Mukai et al., 2004), bottled nutri-
tive drinks (Kataoka, Miyanaga, Tsuji, & Uchida, 2004; Kataoka
et al., 2005) and medicines (Ishizaka et al., 2004). Moreover, a re-
cent study has used the electronic tongue as a tool to guide the
identication of avour peptides from puffer sh extract (Zhang,
Wang, Liu, & Xu, 2012). Nevertheless, to our knowledge, no work
has been previously reported on using an instrumental taste
sensing device to characterise the taste attributes of hydroly-
sates, which are complex mixtures of peptides and amino acids
that have been postulated to activate different sets of human
taste receptors (Maehashi & Huang, 2009; Temussi, 2011).
Therefore, the objectives of this work were to examine the
applicability of an instrumental taste sensing system to character-
ise shrimp processing by-products hydrolysates produced using
different proteases under various hydrolysis conditions. Further-
more, the potential of these sensors to relate to gustatory sensory
scores as well as other chemical or bioactive properties, such as
salt content or ACE inhibitory activity, were explored and
discussed.
0308-8146/$ - see front matter 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.foodchem.2013.09.004

Corresponding author. Tel.: +1 604 822 6182; fax: +1 604 822 5143.
E-mail address: Eunice.Li-Chan@ubc.ca (E.C.Y. Li-Chan).
Food Chemistry 145 (2014) 10761085
Contents lists available at ScienceDirect
Food Chemistry
j our nal homepage: www. el sevi er . com/ l ocat e/ f oodchem
2. Materials and methods
2.1. Materials
Shrimp processing by-products (including shells, heads and
tails recovered from hand-peeling of cooked shrimp Pandalopsis
dispar) in frozen form were donated by Albion Fisheries Ltd.
(Vancouver, BC, Canada). Two batches of shrimp processing
by-products were obtained, the rst sampled in early spring
(February) and the second in autumn (November). Upon receiving,
samples were kept overnight at 4 C, then distributed into 1000-g
packages and stored at 25 C prior to use.
Food-grade proteases were donated by Neova Technologies Inc.
(Abbotsford, BC, Canada). Alcalase

2.4 l FG (Bacillus licheniformis,

2.4 AU/g), Flavourzyme

1000 l (Aspergillus oryzae, 1000 LAPU/g)

and Protamex

(Bacillus amyloliquefaciens and Bacillus licheniformis,

1.5 AU/g) were products from Novozymes North America Inc.
(Salem, NC) while bromelain (from pineapple stem, 2000 GDU/g)
was manufactured by Bio-Logics, Inc. (Montreal, QC, Canada).
Potassium chloride, L-(+)-tartaric acid, ethanol and potassium
hydroxide were purchased from Kanto Chemical Co., Inc. (Tokyo,
Japan). Food grade sucrose and caffeine were products from Fisher
Scientic (Fairlawn, NJ) while sodiumchloride was purchased from
BDH Inc. (Toronto, ON, Canada) and monosodium glutamate was
from Ajinomoto USA Inc. (Teaneck, NJ).
2.2. Hydrolysate production
Fifteen hydrolysate samples produced under experimental con-
ditions dened by a Taguchis L
16
(4
5
) fractional factorial experi-
mental design were obtained from a concurrent study in our
laboratory, described in Cheung and Li-Chan (2010). Four factors,
each at four levels, were evaluated and they were type of protease
(Alcalase, bromelain, Flavourzyme or Protamex), water-substrate-
ratio (1:1, 1:1.5, 1:2 or 1:2.5), percent enzyme (2%, 4%, 6% or
8% w/w protein contents of shrimp processing by-products) and
time of hydrolysis (1, 4, 8 or 24 h). Eleven additional hydrolysates
(marked with an asterisk in Table 1) were prepared according to
the conditions in Table 1 using a separate batch of shrimp process-
ing by-products obtained in the fall season. Previous research indi-
cated that protease type was the predominating factor inuencing
properties of the resulting hydrolysates (Cheung & Li-Chan, 2010).
Sample codes were represented by a capitalised letter referring to
the type of protease (A = Alcalase, B = bromelain, F = Flavourzyme,
P = Protamex, C = Control with no added proteases) used to pro-
duce the hydrolysates, followed by a number indicating the time
of hydrolysis (1, 4, 8, or 24 h). It should be noted that percent en-
zyme and water-to-substrate ratio also varied for each sample but
were not specied in the sample code. For example, the code A4
stands for hydrolysates produced using 8% Alcalase, 2:1 water-
to-substrate ratio and 4 h of hydrolysis time. The sample A4 was
prepared in triplicate to investigate the reproducibility of the
hydrolysate production process and these replicates were coded
as A4, A4-1 and A4-2.
2.3. pH, Conductivity and sodium content
Sample solutions (10% w/v) were prepared in distilled and
deionised water. Determination of pH was performed using Corn-
ing Pinnacle 530 pH meter (Corning Incorporated, Corning, NY).
Conductivity of the samples was measured using MeterLab

CDM
210 conductivity meter (Radiometer Analytical SAS, Villeurbanne
Cedex, France) while sodium content was determined using an
Accumet

sodium ion selective electrode (Fisher Scientic, Fair-

lawn, NJ).
2.4. Taste analysis of shrimp processing by-products hydrolysates
using taste sensing system
Shrimp hydrolysate samples were analysed using Insent Taste
Sensing System TS-5000Z (Intelligent Sensor Technologies, Inc.,
Kanagawa-Pref., Japan), which utilises articial lipid membranes
to detect potential differences caused by the adsorption of sub-
stances associated with various taste attributes. In addition to taste
attributes represented by the initial responses of the sensors, spe-
cically GL0 for sweetness, CA0 for sourness, AAE for umami, CT0
for saltiness, C00 for bitterness, AN0 for basic bitterness (b-bitter-
ness), and AE1 for astringency, the system also provides informa-
tion on attributes representing aftertaste, such as C00 for
aftertaste bitterness (a-bitterness), AE1 for aftertaste astringency
(a-astringency), and AAE for richness, by measuring the change
of membrane potential caused by adsorption (CPA) (Kobayashi
et al., 2010). These sensors provide potentiometric sensor outputs,
which are recorded and transformed into taste data through the
use of pre-set conversion factors based on the WeberFechner
law (Kobayashi et al., 2010).
To perform sample measurement, the potential (V
r
) was rst re-
corded for the sensor in the reference solution (0.3 mM L-(+)-tar-
taric acid and 30 mM KCl), followed by a measurement (V
s
) in
sample solution (35 ml). After sample measurement, the sensor
was washed gently with the reference solution and the potential
(V
r
0 ) was recorded. Finally the sensor was thoroughly cleaned in
the alcohol solution (100 mM HCl in 30% ethanol for negatively
charged membrane; 10 mM KOH and 100 mM KCl in 30% ethanol
for positively charged membrane) before proceeding to the next
sample. The difference V
s
V
r
represents the initial taste while
V
r
0 V
r
represents the aftertaste.
In this study, hydrolysate samples were analysed in three sepa-
rate groups (Table 1): Group 1 comprised 10 samples (italicised
font) that were prepared as 1% w/v solutions; Group 2 contained
19 samples (bold font) that were analysed at 10% w/v concentra-
tion in 3 sets, using sample A4

as the reference sample; Group 3

consisted of 5 samples (underlined) that were analysed at both
1% and 8% w/v concentrations.
When ten or fewer samples were being analysed, the taste sen-
sor outputs (in millivolts) were reported directly relative to the ref-
erence tartaric acid KCl solution (Groups 1 and 3). However, when
comparing a larger number of samples, a reference sample was
recommended to be included in each set of ten samples and the
sensor output values of all samples were expressed relative to
the reference sample (i.e. A4

in Group 2). This was necessary to

circumvent variability in sensor response across different sets,
and was achieved by subtracting the sensor output values of the
reference hydrolysate from the output of each sample in the same
set. The output value differences were then used for comparison of
samples in the different sets. All taste sensor measurements were
conducted in triplicate.
2.5. Sensory evaluation of shrimp processing by-products hydrolysates
Nine panellists (4 females and 5 males) were recruited and
trained to differentiate and compare the intensity of solutions for
the following taste attributes: sweet, salty, umami and bitter,
according to the method of Cheung and Li-Chan (2010) with some
modications. Panellists were instructed to deposit the 200 lL
solution in the central area of the tongue to note the intensity of
each taste attribute.
During the training session, four standard solutions were
provided for each taste attribute in the following concentrations:
14% sucrose solution for sweetness, 0.41.0% NaCl solution for
saltiness, 0.12% monosodium glutamate solution for umami, and
5005000 ppm caffeine solution for bitterness. Panellists were
I.W.Y. Cheung, E.C.Y. Li-Chan/ Food Chemistry 145 (2014) 10761085 1077
instructed to mark a vertical line on the 15-cm line scale to
represent the respective intensity for each of the standard
solutions. After independently evaluating the standard solutions,
panellists discussed their scores and reached consensus on the
intensity scores for each of the standards. Two shrimp processing
by-products hydrolysate solutions (10% w/v, 1 g solid in 10 ml
ddH
2
O) were then provided and panellists were asked to evaluate
and reach consensus on the intensity of the four taste attributes
relative to the standard solutions. One of the hydrolysate samples
was selected to be provided as a reference (R) at the testing
sessions and the agreed intensity of each taste attribute for R
was marked on the evaluation form.
In the two testing sessions, panellists evaluated duplicate sets
of ve test samples (A4

, B4

, C4

, F4

and P4

). In addition to the
test samples, two standards (the lowest and highest concentrations
used in the training session) for each taste attribute were provided
together with R. The test samples were labelled with 3-digit
numeric codes and the panellists were instructed to taste the
test samples in the randomised order provided on the sensory
evaluation form, and to mark their intensity on the 15-cm line
scales.
2.6. ACE inhibitory activity
Inhibition of ACE activity by the samples was evaluated by a
colorimetric assay with N-hippuryl-L-histidyl-leucine as substrate,
as described by Cheung and Li-Chan (2010).
2.7. Statistical analysis
General linear model analysis of variance (GLM-ANOVA) fol-
lowed by Tukeys comparison test (p < 0.05) as well as principal
component analysis were performed using Minitab software (Ver-
sion 16; Minitab Inc., State College, PA).
Data from sensory evaluation were analysed according to pro-
cedures described by Moon and Li-Chan (2007), in which 3-way
GLM-ANOVA was used for the initial analysis, with sample, panel-
list and replication as factors. The Pearson correlation coefcient
was computed to identify any panellist(s) who might have dif-
culty evaluating particular taste attribute(s). Furthermore, the rep-
lication effect was evaluated using 2-way GLM-ANOVA to detect if
any panellist was inconsistent in assessing the taste attributes for
replicates of the sample. In addition, the panellist effect was re-
moved for particular attributes by performing z-transformation
of the data (Mui, Durance, & Scaman, 2002).
3. Results and discussion
3.1. Taste sensor proles of 1% w/w shrimp processing by-products
hydrolysates
Hydrolysate samples produced by the different protease prepa-
rations, namely Alcalase (A4, A4-1, A24), bromelain (B1, B4), Fla-
vourzyme (F1, F4, F24) and Protamex (P4, P24), were analysed at
1% concentration using the taste sensing system (Table 1). These
Table 1
Conditions for hydrolysis of shrimp processing by-products, extent of hydrolysis of the hydrolysed samples, and the pH, conductivity, sodium content as well as ACE-inhibitory
activity of the hydrolysate solutions.
Sample code
a
Hydrolysis Conditions Hydrolysed sample 10% Hydrolysate solution ACEi (%)
f
Protease W:S
b
E (%)
c
Time (h) EH (meq Leu/g)
d
pH Conductivity (mS/cm) Sodium (ppm)
e
A4 Alcalase 2:1 8 4 2.30 8.50 6.05 350 77
B1 Bromelain 2:1 6 1 0.56 8.50 4.72 710 39
B4 Bromelain 1:1 4 4 0.64 8.57 5.04 710 36
F1 Flavourzyme 2.5:1 8 1 2.31 8.41 5.54 490 49
F4 Flavourzyme 1.5:1 2 4 1.43 8.39 5.93 660 40
F24 Flavourzyme 2:1 4 24 2.69 7.86 9.79 580 33
A4-1 Alcalase 2:1 8 4 2.09 8.50 6.47 400 76
A24 Alcalase 1.5:1 6 24 3.02 7.90 8.00 400 74
P4 Protamex 2.5:1 6 4 1.75 8.30 11.60 2190 79
P24 Protamex 1:1 8 24 2.22 8.00 17.66 4090 75
A1 Alcalase 1:1 2 1 1.97 8.49 5.57 490 65
A4-2 Alcalase 2:1 8 4 2.64 8.56 6.47 430 77
A8 Alcalase 2.5:1 4 8 2.45 8.38 6.90 400 78
B1

Bromelain 2:1 6 1 0.34 8.67 10.59 1840 28

B8

Bromelain 1.5:1 8 8 0.43 8.41 10.73 1770 10

B24

Bromelain 2.5:1 2 24 0.47 8.38 11.92 1770 11

F1

Flavourzyme 2.5:1 8 1 0.98 8.32 10.55 1430 15

F8

Flavourzyme 1:1 6 8 1.05 8.18 9.66 1270 7

F24

Flavourzyme 2:1 4 24 0.96 8.02 13.26 1770 13

P1 Protamex 1.5:1 4 1 1.51 8.42 12.96 2810 79
P8 Protamex 2:1 2 8 1.22 8.31 10.00 1640 78
A4

Alcalase 2:1 8 4 1.13 8.04 8.55 830 62

B4

Bromelain 1:1 4 4 0.36 8.51 12.08 2000 23

C4

Endogenous 2.5:1 0 4 0.35 8.71 16.02 3110 60

F4

Flavourzyme 1.5:1 2 4 0.71 8.40 11.69 1840 15

P4

Protamex 2.5:1 6 4 1.26 8.09 12.65 2080 60

a
Sample codes with an asterisk represent samples re-produced using identical hydrolysis conditions with a separate batch of shrimp processing by-products; A4-1 and A4-
2 were sample replicates of A4. Sample codes that are italicised represent samples analysed in Section 3.1.1; those that are in bold were analysed in Section 3.1.2; sample
codes that are underlined indicate samples analysed in Section 3.2.
b
Water-to-substrate ratio.
c
Percentage of enzyme, calculated based on protein content in the shrimp processing by-products.
d
Extent of hydrolysis, in milliequivalents of L-leucine per gram of dry shrimp by-products. Values for samples coded without an asterisk are from Cheung and Li-Chan
(2010).
e
Sodium content was determined using a sodium ion selective electrode.
f
Angiotensin-I converting enzyme inhibitory activity, based on assay with hydrolysate samples prepared at 0.29 mg/ml.
1078 I.W.Y. Cheung, E.C.Y. Li-Chan/ Food Chemistry 145 (2014) 10761085
samples had been previously assessed to encompass gustatory bit-
terness intensity ranging from 1100 to 2400 ppm caffeine (Cheung
& Li-Chan, 2010).
The characteristic taste sensor proles (in mV) of the samples
are displayed in a cobweb diagram (Fig. 1A). The sensor outputs
for b-bitterness and the aftertaste attributes, a-bitterness and a-
astringency, were too low to be considered distinguishable by hu-
mans, according to WeberFechner law (Kobayashi et al., 2010)
and hence are not displayed. Each hydrolysate sample displayed
a distinctive taste sensor output prole. These results showed that
hydrolysate samples produced using different proteases could be
characterised by the taste sensor system.
Principal component analysis of these taste sensor output data
revealed that 96% of the variability in the data set could be
explained by three principal components, with 58%, 26% and 12%
of the variability explained by PC1, PC2 and PC3, respectively.
Particular taste sensor attributes were associated with different
samples, as shown in the bi-plot of sample scores and taste sensor
loadings on PC1 and PC2 (Fig. 1B). For example, P4 and P24
were characterised by high saltiness while A24, F24 and P24, the
samples that underwent extensive hydrolysis for 24 h, had high
sourness. On the other hand, B1, B4, F1 and F4 all had positive
PC1 scores highly related to umami, astringency and bitterness.
While B4 had a high positive PC2 score (relating to high
sweetness), F1 had negative PC2 score (relating to richness).
3.2. Taste sensor proles of 10% w/w shrimp processing by-products
hydrolysates
The results in Section 3.1 clearly demonstrated the potential of
taste sensors to distinguish amongst the hydrolysate samples
tested at 1% w/v concentration. In this section, 19 hydrolysate sam-
ples (bold in Table 1) were analysed at 10% w/v concentration with
the aim to increase the overall sensor responses for better sample
differentiation.
Fig. 2 shows the sensor output proles for 10% solutions of two
samples, A4

and C4

. Sample A4

was produced using the same

hydrolysis conditions as A4 and A4-1 in Section 3.1 but with a
different batch of shrimp processing by-products as the starting
material, while C4

was the control sample produced without

addition of any proteases. In general C4

gave higher sensor

responses than A4

except for sourness (Fig. 2). Compared to the

sensor outputs for 1% sample solutions shown in Fig. 1A, an
increase in sample concentration to 10% only increased the sensor
outputs for sourness, saltiness and sweetness, which suggests that
some of the taste sensors might have reached the saturation point
Fig. 1. (A) A cobweb diagram depicting the taste sensor outputs (mV) of 1% w/v shrimp processing by-products hydrolysates. (B) A bi-plot displaying scores of the rst two
principal components for taste sensor outputs of hydrolysate solutions at 1% concentration.
I.W.Y. Cheung, E.C.Y. Li-Chan/ Food Chemistry 145 (2014) 10761085 1079
for detection of the higher concentrations, similar to what is com-
monly observed for the human taste receptor in that no further in-
crease in perceived intensity would be obtained beyond the
saturation point (Keast & Breslin, 2002).
Fig. 3 shows the sourness, sweetness, bitterness, saltiness and
umami taste sensor outputs of the 17 samples expressed as differ-
ence values relative to A4

, whose taste prole was displayed in

Fig. 2. The difference values for A4-1 and A4-2 were averaged
and reported as A4
0
in Fig. 3. Signicant differences were observed
among the hydrolysates, and some general trends could be ob-
served. For example, hydrolysis time of 24 h produced hydroly-
sates comprising highest sourness sensor outputs among samples
prepared using the same protease (Fig. 3A). Hydrolysates produced
with shorter hydrolysis times by bromelain and Flavourzyme, spe-
cically B1

, B4

, F1

, F4

, showed high sweetness sensor outputs

(Fig. 3B) while hydrolysates produced with Alcalase had high bit-
terness outputs (Fig. 3C). Samples produced using Protamex had
the highest saltiness sensor outputs (Fig. 3D) whereas higher uma-
mi sensor outputs were observed for samples prepared with bro-
melain and Flavourzyme than with Alcalase and Protamex
(Fig. 3E).
Interestingly, A4
0
showed signicantly lower sweetness
(Fig. 3B) and higher bitterness (Fig. 3C) than A4

, as displayed by
the large difference values. A4
0
was prepared using shrimp process-
ing by-products obtained in early spring while A4

was made using

those acquired in autumn. Previous work had suggested signicant
differences in compositions of shrimp processing by-products
sampled in different seasons (Cheung, Cheung, & Li-Chan, 2012).
The signicant differences observed in the sensor outputs for A4
0
and A4

indicate that the taste sensors were able to detect

variability in hydrolysates produced from starting materials
obtained in different seasons. In fact, the extent of hydrolysis for
A4
0
was 2.38 meq Leu/g while that for A4

was 1.13 meq Leu/g

(Table 1); furthermore, with the exception of the control C4

, all
the samples prepared using shrimp processing by-products
obtained in autumn (i.e. samples coded with an asterisk) had a
lower extent of hydrolysis (Table 1). The lower extent of hydrolysis
could be due to the seasonal variations in the starting materials
that inuenced amino acid composition and altered their suscepti-
bility to protease action (Cheung et al., 2012), thus producing
products with different taste sensor proles.
Aside from some distinct differences in taste sensor outputs of
hydrolysates as a function of the various proteases, it is clear that
other factors also contributed to differences in taste proles. As the
time of hydrolysis increased, sourness sensor outputs also in-
creased signicantly for the bromelain-produced hydrolysates,
whereas the hydrolysates produced using Protamex showed an in-
crease with percent enzyme added (Fig. 3A and Table 1). As for the
sweetness sensor, both bromelain and Flavourzyme produced
hydrolysates yielding high sweetness sensor outputs after 1 and
4 h hydrolysis but these were signicantly lowered as the hydroly-
sis was extended. On the contrary, Alcalase-produced hydrolysates
generally showed lower sweetness outputs, except for the sample
produced after 24 h of hydrolysis (Fig. 3B). The Alcalase-produced
hydrolysates had high bitterness and A1, which had the lowest
bitterness in this group, showed comparable or higher bitterness
than other hydrolysate samples (Fig. 3C). Alcalase has previously
been shown to produce bitter-tasting hydrolysates, owing to the
preferential cleavage sites at hydrophobic amino acid residues
(Cheung & Li-Chan, 2010; Gildberg, Arnesen, & Carlehog, 2002).
Interestingly, both B8

and F8

had much lower bitterness, and

these samples were prepared using a low water-to-substrate ratio
and shorter hydrolysis time (Fig. 3C and Table 1). Hydrolysates
produced using Alcalase had lower saltiness than those prepared
with bromelain and Flavourzyme, which in turn showed lower
saltiness than Protamex-produced hydrolysates (Fig. 3D). Brome-
lain and Flavourzyme-produced hydrolysates had higher umami
than hydrolysates made with Alcalase and Protamex; in general,
a decrease in umami was noted as hydrolysis was extended
(Fig. 3E).
These results collectively suggest that taste sensors have a
strong potential to be used to classify hydrolysate samples based
on hydrolysis conditions. However, analysis of a larger set of sam-
ples should be conducted in the future to conrm this hypothesis.
Fig. 2. A cobweb diagram depicting the taste sensor outputs (mV) of the 10% w/v shrimp processing by-products hydrolysates.
1080 I.W.Y. Cheung, E.C.Y. Li-Chan/ Food Chemistry 145 (2014) 10761085
3.3. Sensor outputs and its relation to pH, sodium content, gustatory
bitterness and angiotensin-I converting enzyme inhibitory activity
3.3.1. pH and salt contents
The pH of the 10% hydrolysate solutions was found to be in the
range 7.868.90, near the pH of the original substrate. This pH
range was slightly above neutral and was not expected to display
sourness since only low pH is associated with sourness in food
products.
Saltiness of the samples could be assessed using other analytical
instruments, such as conductivity meter and sodium ion selective
electrode. The conductivity and sodium ion selective electrode
measurements showed good correlation with saltiness sensor
outputs of 10% hydrolysate solutions (r = 0.8464 and 0.9105,
respectively). It is evident that the saltiness sensor could be
applied to assess salt content of hydrolysate solutions concurrently
with other taste sensors to provide taste proles of hydrolysate
samples.
Fig. 3. Bar graphs displaying (A) sourness, (B) sweetness, (C) bitterness, (D) saltiness and (E) umami taste sensor outputs (mV) of each sample relative to A4

. Samples were
comprised of 10% solutions of shrimp processing by-products hydrolysates produced using Alcalase, bromelain, Flavourzyme and Protamex for 1 ( ), 4 ( ), 8 ( ) and 24 ( )
hours as listed in Table 1. A4
0
denotes average value of A4-1 and A4-2 replicate hydrolysates. Different letters (ak) represent signicant difference (p < 0.05) between
samples for each taste sensor.
I.W.Y. Cheung, E.C.Y. Li-Chan/ Food Chemistry 145 (2014) 10761085 1081
3.3.2. Gustatory bitterness
The bitterness evaluated by panellists in our previous study
(Cheung & Li-Chan, 2010) was found to have a strong negative
correlation with umami sensor outputs of samples measured at
10% concentration (r = 0.9380). Furthermore, gustatory bitter-
ness also negatively related to richness sensor outputs
(r = 0.8407). The results suggested that an increase in umami
sensor outputs might explain the weaker bitterness detected in
the samples by panellists. In fact, it was postulated that moder-
ate concentrations of glutamate would suppress bitterness (Keast
& Breslin, 2002). The umami sensor AAE was designed to mea-
sure monosodium glutamate, an abundant amino acid found in
many foodstuffs (Kobayashi et al., 2010). It is also the only sen-
sor in the taste sensing system used in the current study that
was designed specically to measure amino acids. Since peptides
and amino acids were the major constituents in the hydrolysate
samples, it is possible that the umami sensor could also measure
other amino acids or that the umami sensor responded to amino
acids that produced sweet and umami taste, consequently pro-
viding the negative relationship observed with gustatory bitter-
ness. The signicantly weaker bitterness in hydrolysates
produced using bromelain or Flavourzyme reported in our previ-
ous work (Cheung & Li-Chan, 2010) were well distinguished
from the samples made with Alcalase, which were measured
by the taste sensors to have high bitterness and low umami
(Fig. 3).
On the contrary, sensors, like bitterness, b-bitterness and
astringency, that were expected to associate with gustatory bit-
terness did not show any reasonable correlation. The C00 taste
sensor for bitterness was originally developed to detect acidic
bitter substances such as iso-alpha acid found in beer while
the astringency sensor was constructed to respond to tannic
acid, gallic acid, caffeic acid and epigallocatechin gallate (Kobay-
ashi et al., 2010). These compounds differ in chemical composi-
tion and structure from the bitter substances in hydrolysates,
which are composed mainly of amino acids and peptides with
hydrophobic and bulky residues (Li-Chan & Cheung, 2010), and
therefore may not have stimulated proper responses to the sen-
sors. Recent discoveries have suggested the existence of 25 hu-
man bitter receptors yet very little is known about which and
how each receptor responds to the vast diversity of bitter pep-
tides (Maehashi & Huang, 2009).
Since the umami sensor was the only sensor that generated re-
sponses relating to gustatory bitterness of hydrolysates, potentially
owing to its sensitivity to amino acid compounds, further investi-
gation should examine the selectivity and sensitivity of the umami
sensor to specic amino acids or peptides, to optimise the sensor
for use in determining peptidic bitterness.
3.3.3. Angiotensin-I converting enzyme inhibitory activity
In our previous study, it was shown that ACE inhibitory activity
had a positive correlation with hydrolysate bitterness (Cheung &
Li-Chan, 2010). Moreover, several other studies have proposed
the link between ACE inhibitory activity and bitterness, owing to
the commonality of structural requirements (Pripp & Ardo, 2007;
Wu & Aluko, 2007). Therefore, the potential of using taste sensors
to predict ACE inhibitory activity of hydrolysates was also
examined.
Similar to gustatory bitterness, ACE inhibitory activity was also
found to correlate negatively with the umami sensor outputs
(r = 0.9454). It is evident that the umami taste sensor has the po-
tential to be used as a quick screening tool to provide a surrogate
measure of ACE inhibitory activity of hydrolysates. Nevertheless,
more samples should be investigated to conrm if such a relation-
ship does extend to a wider range of hydrolysates produced using
different substrates and hydrolysis conditions.
3.4. Taste prole analysis of selected shrimp processing by-products
hydrolysates using both taste sensors and human panellists
From the previous sections, it is clear that the type of protease
was the most important factor leading to differences in taste sen-
sor outputs of the resulting hydrolysates, and less difference was
observed between hydrolysates made with the same enzyme but
under different conditions of percent enzyme, water-to-substrate
ratio and time of hydrolysis. Therefore, in this section, ve samples
produced by hydrolysis with different proteases were selected for
further investigation, to examine the relationship between taste
sensor outputs and gustatory responses. Specically the taste sen-
sor responses of 1% and 8% solutions of A4

, B4

, C4

, F4

and P4

were determined and compared to the gustatory scores for these

samples as evaluated by human sensory panellists.
3.4.1. Taste sensor proles of 1% and 8% w/w hydrolysate solutions
The taste sensor outputs of the ve hydrolysate samples pre-
pared at 1% and 8% concentration are shown in Table 2. It can be
noted that sourness, saltiness and richness sensor outputs in-
creased as the concentration of the hydrolysates increased from
1% to 8%; however, the umami sensor gave lower sensor outputs
when measured at 8% compared to 1% and the remaining sensors
produced responses that were comparable for both 1% and 8%
hydrolysate solutions (Table 2). This indicates that different sen-
sors might require different sample concentration to produce sen-
sor outputs that could effectively distinguish among samples.
Therefore, concentration used for the taste sensor analysis should
be optimised for each specic application for best performance.
Similar trends were observed for some samples measured at the
two concentrations. A4

and P4

had higher sourness sensor out-

puts than B4

and F4

, which in turn had higher sourness outputs

than C4

, while saltiness sensor outputs increased with A4

< -
F4

< B4

< P4

< C4

(Table 2). B4

and C4

had high bitterness

and astringency sensor outputs while P4

showed the lowest bit-

terness and A4

and P4

had low astringency. The umami taste sen-

sor outputs for the samples could generally be separated into two
distinct groups: A4

and P4

with low umami sensor outputs and

B4

, C4

and F4

with high umami sensor outputs. Some differences

were noted when the umami sensor outputs of the samples were
ranked and compared between 1% and 8%; these differences could
potentially be due to the fact that the umami taste sensor was de-
signed to detect glutamic acid but might also respond to other ami-
no acids, which were abundant in the hydrolysates. Furthermore,
the umami taste sensor was also shown to respond to nucleotides,
such as disodium 5
0
-inosine monophosphate and disodium 5
0
-gua-
nosine monophosphate (Kobayashi et al., 2010), which are abun-
dant in seafood and could contribute to the sensor responses.
Nonetheless, in this study, the presence of nucleotides was not
measured and the potential effect they might have on the umami
sensor outputs should be taken into consideration. The sensor out-
puts from the sweetness sensor differed greatly when analysed at
1% compared to 8%. The sweetness sensor was specically designed
to detect compounds such as sucrose (Kobayashi et al., 2010) and
has been used successfully in the prediction of sweetness in enteral
nutrients but only 1 of the 9 products was protein-based (Mukai
et al., 2004). Therefore, it is not known whether the sweetness sen-
sor responds to sweet peptides, even though a previous study
showed that an older generation of the sensor measured sweet
amino acids Gly, Ala and Thr (Kikkawa, Toko, Matsuno, & Yamafuji,
1993).
Despite some differences in sensor outputs observed at 1% and
8%, it is still evident from the overall taste sensor proles that the
samples could be characteristically divided into 3 groups: C4

as a
stand-alone sample, B4

and F4

as a group while A4

and P4

formed another, in agreement with results in Figs. 1 and 3.

1082 I.W.Y. Cheung, E.C.Y. Li-Chan/ Food Chemistry 145 (2014) 10761085
3.4.2. Sensory evaluation of 10% w/w hydrolysates solution
The ve hydrolysate samples were sensorially evaluated on
sweet, salty, umami and bitter tastes. The results on saltiness for
replicate samples reported by each panellist as well as those re-
ported by different panellists were inconsistent, suggesting the
complexity of the taste in the hydrolysates made it difcult for
panellists to properly differentiate the saltiness intensity. There-
fore, these data were excluded from further analysis. Signicant
panellist effect was detected for sweet and bitter tastes so z-trans-
formation test as described in Mui et al. (2002) was applied to the
data, as shown in Table 3A.
Sweet, bitter and umami tastes were well differentiated by the
panellists. A4

and P4

had weaker sweet taste than B4

, which in
turn was less sweet than F4

, while the strongest sweetness was

found in C4

(Table 3A). The umami taste was more difcult for

panellists to evaluate due to the taste complexity of the samples.
A4

and P4

had signicantly weaker umami in comparison to

C4

and F4

(Table 3A). B4

, C4

, F4

, samples with stronger sweet

and umami taste, had weakest bitter taste, while A4

had stronger
bitter taste than P4

. In the study performed by Shen, Guo, Dai, and

Zhang (2012), they postulated Collichthys niveatus hydrolysate pro-
duced using Alcalase to have stronger bitter taste than that pre-
pared with Flavourzyme, owing to the fact that Alcalase-
produced hydrolysate had lower amounts of Ala, Asp and Glu
(sweet and umami amino acids) but high amounts of Phe (bitter
amino acid), in agreement with the sensory results shown in the
current study. Furthermore, in another study investigating the
taste of peanut meal hydrolysate, it was also found that hydroly-
sate made using Alcalase had stronger bitter taste than that pre-
pared with Protamex, yet the Protamex-produced hydrolysate
was shown to have stronger umami taste (Su, Ren, Yang, Cui, &
Zhao, 2011); these results were different from what was observed
in our study, where similar umami taste was noted for Alcalase-
and Protamex-produced hydrolysates. In the taste dilution analysis
of soy protein isolate hydrolysate, it was also found that hydroly-
sate prepared using Alcalase had signicantly stronger bitter taste
than those made with bromelain, Flavourzyme and Protamex; nev-
ertheless, hydrolysates prepared using bromelain and Protamex
were found to be more bitter than that with Flavourzyme (Seo,
Lee, & Baek, 2008), unlike the similar bitter taste between B4

and F4

observed in this study. The taste of the hydrolysates was

strongly inuenced by the choice of proteases but the type of sub-
strate also made a large contribution to the composition of pep-
tides and amino acids after hydrolysis, consequently the taste of
the end products. The acceptability rating showed a signicant po-
sitive relationship with sweet taste (r = 0.9804) and umami taste
(r = 0.9730) while the bitterness rating displayed a signicant neg-
ative correlation (r = 0.9658). Bitterness was previously shown to
contribute negatively to palatability in bottled nutritive drinks
(Kataoka et al., 2005) and enteral nutrients (Mukai et al., 2004),
while sweetness displayed a positive correlation in the palatability
of enteral nutrients (Mukai et al., 2004). The bitter sensory scores
of the samples were compared to the bitterness of corresponding
samples evaluated in Cheung and Li-Chan (2010) and a strong po-
sitive relationship (r = 0.9892) was observed, suggesting that the
signicant change in extent of hydrolysis (Table 1) did not alter
the bitter taste differences among the hydrolysates, further con-
rming that the choice of proteases was the main factor that af-
fected the taste of the nal products.
3.4.3. Assessing the correlation between taste sensor outputs and
sensory scores
Many studies have explored the use of taste sensors to measure
taste characteristics of various products (Woertz, Tissen, Kle-
inebudde, & Breitkreutz, 2011). For example, Kataoka et al.
(2005) showed good correlation for sourness and bitterness be-
tween taste sensor outputs and gustatory sensation in several bot-
tled nutritive drinks. In the study of enteral nutrients, gustatory
sweetness, sourness and bitterness were all well predicted by the
corresponding taste sensors (Mukai et al., 2004). Rudnitskaya
et al. (2009) also successfully used an electronic tongue to predict
sensory characteristics of beer. Furthermore, an electronic tongue
was able to classify apple juice samples according to sweet taste,
sour taste and apple taste (Kovacs et al., 2011). Nevertheless, to
our knowledge, there have not been any reports to date on the
use of these taste sensors to evaluate the complex taste character-
istics of hydrolysates.
Table 2
Taste sensor outputs
a
of shrimp processing by-products hydrolysates evaluated as 1% (top) and 8% (bottom) solutions.
Sample Taste sensor output (mV)
Sweetness Sourness Bitterness Astringency Umami Richness Saltiness
(1%)
A4

2.62 b 1.57 c 2.07 b 2.69 a 0.34 a 7.90 b 3.56 a

B4

3.63 c 1.07 b 4.12 d 3.50 d 3.02 c 6.97 a 4.66 c

C4

1.70 a 3.59 a 6.04 e 3.95 e 2.27 b 9.42 c 5.37 e

F4

3.24 bc 2.21 b 3.37 c 3.11 c 2.61 bc 8.51 bc 4.22 b

P4

2.79 b 1.49 c 1.58 a 2.80 b 0.65 a 8.05 b 5.22 d

(8%)
A4

2.80 b 7.88 c 4.10 c 0.93 b 1.90 a 10.56 a 12.43 a

B4

3.01 b 4.07 b 4.78 d 1.95 d 0.35 b 10.95 ab 15.33 c

C4

4.97 d 0.05 a 5.22 e 2.56 e 1.09 c 13.15 c 16.65 e

F4

3.94 c 3.00 b 3.75 b 1.64 c 0.43 b 11.86 b 14.02 b

P4

1.82 a 8.12 c 2.07 a 0.53 a .82 a 11.25 ab 15.93 d

a
Values shown are means of triplicate output measurements. Different letters (ae) within a column indicate signicant difference (p < 0.05) between samples for each
taste sensor output.
Table 3A
Sensory scores
a
for sweet taste, bitter taste, umami taste as well as acceptability
a
of
the shrimp processing by-products hydrolysates.
Sample Sensory attribute Acceptability
Sweet (z)
b
Bitter (z)
b
Umami
A4

1.08 a 1.24 c 4.29 a 3.61 a

B4

0.01 b 0.66 a 6.90 bc 7.30 b

C4

1.33 d 0.87 a 9.80 d 9.52 c

F4

0.68 c 0.66 a 8.72 cd 8.98 c

P4

0.88 a 0.85 b 5.54 ab 4.10 a

a
Values shown are means of replicate tasting data from 7 to 9 panellists. Dif-
ferent letters (a,b,c,d) within a column indicate signicant difference (p < 0.05)
between samples for each sensory attribute.
b
Z-transformed scores were calculated according to Mui et al. (2002).
I.W.Y. Cheung, E.C.Y. Li-Chan/ Food Chemistry 145 (2014) 10761085 1083
In this study, ve hydrolysate samples, specically A4

, B4

, C4

,
F4

and P4

, were evaluated using both instrumental taste sensors

and human sensory panel. Both analyses showed higher umami in
B4

, C4

and F4

than A4

and P4

(Tables 2 and 3A). Although the

bitterness sensor outputs are supposed to represent sample bitter-
ness, it was found that P4

, having the lowest bitterness sensor

output, was evaluated to be quite bitter by the panellists, whereas
C4

, having the highest bitterness sensor output, tasted less bitter

(Tables 2 and 3A). In general, panellists separated the samples into
3 major groups: C4

received the highest acceptability scores,

accompanied by high umami, high sweet and low bitter tastes;
A4

and P4

were rated unacceptable with low umami and sweet

tastes but high bitter taste; B4

and F4

were rated as acceptable,

being less sweet but with relatively high umami and low bitter
tastes (Table 3A). Similarly, the taste sensors also categorised these
samples into 3 main groups: C4

had very high bitterness, astrin-

gency, umami, richness and saltiness sensor outputs; A4

and P4

showed low bitterness, astringency and umami sensor ouputs;

B4

and F4

had sensor outputs lying in between the former two

groups (Table 2). These resemblances in sample grouping evidently
suggest that the two evaluation methods were correlated in certain
ways.
Therefore, the correlation information between taste sensor
outputs and human sensory scores was examined and results were
shown in Table 3B. A positive relationship was observed between
the sensory scores for umami taste and the umami and richness
sensor outputs for the 8% hydrolysate solutions. Nevertheless, the
correlation became weaker when the samples were analysed at
1%. Besides umami taste, sweet taste evaluated by panellists also
showed a positive correlation with the sweetness sensor outputs
of 8% solution; however, poor correlation was noted for sensor out-
puts of 1% solution. As mentioned previously, the sweetness taste
sensor might not measure amino acids and peptides in a predict-
able manner and hence the signicant correlation at 8% might
not be reliable. It is necessary to measure the responses of this
taste sensor to sweet amino acids, to examine whether it interacts
selectively to sweet amino acids and peptides. Interestingly, gusta-
tory sweet taste was also found to correlate well with umami taste
sensor outputs of the 8% solution.
Gustatory bitterness was found to show an inverse relationship
with astringency sensor output of the 8% solution. Furthermore,
bitterness evaluated by the panellists negatively relates to the sen-
sor output for the umami sensor at 1% and 8% solutions, indicating
that an increase in umami detection by the sensor might result in
decreased bitterness in the samples perceived by humans. In fact, it
has been reported that the umami taste could suppress bitterness
(Keast & Breslin, 2002; Rhyu & Kim, 2011). Similar to the negative
correlation observed between ACE inhibitory activity and umami
taste sensor outputs in the previous section, ACE inhibition of the
samples correlated negatively with both gustatory umami taste
(r = 0.9206) and umami taste sensor (r = 0.9803).
The results in this study therefore strongly suggest that umami
taste sensor could be useful in predicting gustatory sweet, umami
and bitter tastes. Furthermore, it is very promising that the umami
sensor could potentially be used to screen hydrolysate samples for
ACE inhibitory activity. However, owing to the inconsistency ob-
served when samples were measured at various concentrations,
it is necessary to study the sensor outputs as a function of increas-
ing concentration to nd the optimal concentration for proper
evaluation using the taste sensors.
4. Conclusion
In the present study, it was demonstrated that the taste sensors
were able to differentiate shrimp processing by-products hydroly-
sates produced using various proteases under different hydrolysis
conditions. The saltiness taste sensor displayed outputs that were
in good agreement with conductivity and sodium content. In addi-
tion, the potential of using the umami sensor for the prediction of
gustatory umami, sweet and bitter tastes in the hydrolysates was
evident. It was also shown that the umami sensor might poten-
tially provide a quick screening tool to predict ACE inhibitory activ-
ity of hydrolysates. However, additional research must be
conducted to examine if the observations made in the current
study could be extrapolated to a larger set of hydrolysate samples
produced using different substrates as well as hydrolysis condi-
tions. Furthermore, it is essential to study the linear detection
range for each sensor membrane, i.e., the range over which the sen-
sor outputs are linearly related to the concentration of the relevant
peptides and amino acids, in order for the device to be widely
applicable for taste evaluation of hydrolysates.
Acknowledgments
This research was funded by the Natural Sciences and Engineer-
ing Research Council of Canada. We thank Insent Technologies Inc.
for providing the technical support of the taste sensor measure-
ments. We also thank Albion Fisheries for supplying the starting
materials used in the study and Neova Technologies Inc. for the
donation of proteases.
References
Chang, K.-H., Chen, R. L. C., Hsieh, B.-C., Chen, P.-C., Hsiao, H.-Y., Nieh, C.-H., & Cheng,
T.-J. (2010). A hand-held electronic tongue based on uorometry for taste
assessment of tea. Biosensors & Bioelectronics, 26, 15071513.
Cheison, S. C., Wang, Z., & Xu, S.-Y. (2007). Use of macroporous adsorption resin for
simultaneous desalting and debittering of whey protein hydrolysates.
International Journal of Food Science & Technology, 43, 12281239.
Chen, Q., Zhao, J., Guo, Z., & Wang, X. (2010). Determination of caffeine content and
main catechins contents in green tea (Camellia sinensis L.) using taste sensor
technique and multivariate calibration. Journal of Food Composition and Analysis,
23, 353358.
Table 3B
Correlation coefcients and p-values for the sensor outputs for the different taste properties measured by the taste sensor at 1%, 8% and 10% of shrimp processing by-products
hydrolysate samples (A4

, B4

, C4

, F4

and P4

) as a function of human sensory scores.

Taste attribute Taste sensor response 1%
a
8%
a
10%
a
Correlation coefcient p-Value Correlation coefcient p-Value Correlation coefcient p-Value
Sweet Sweetness 0.3192 0.601 0.9202 0.027 0.6526 0.347
Umami 0.7669 0.130 0.9609 0.009 0.9356 0.064
Richness 0.6241 0.260 0.8878 0.044 0.9354 0.065
Umami Umami 0.7576 0.138 0.9284 0.023 0.9204 0.080
Richness 0.6344 0.250 0.9003 0.037 0.9106 0.089
Bitter Bitterness 0.8511 0.067 0.6116 0.273 0.4867 0.513
Astringency 0.8610 0.061 0.8920 0.042 0.8408 0.159
Umami 0.9544 0.012 0.9054 0.034 0.9364 0.064
a
Five samples were used to compute the correlation coefcient for 1% and 8% samples while four were used for 10% samples.
1084 I.W.Y. Cheung, E.C.Y. Li-Chan/ Food Chemistry 145 (2014) 10761085
Cheung, L. K. Y., Cheung, I. W. Y., & Li-Chan, E. C. Y. (2012). Effects of production
factors and egg-bearing period on the antioxidant activity of enzymatic
hydrolysates from shrimp (Pandalopsis dispar) processing byproducts. Journal
of Agricultural and Food Chemistry, 60(27), 68236831.
Cheung, I. W. Y., & Li-Chan, E. C. Y. (2010). Angiotensin-I-converting enzyme
inhibitory activity and bitterness of enzymatically-produced hydrolysates of
shrimp (Pandalopsis dispar) processing byproducts investigated by Taguchi
design. Food Chemistry, 122, 10031012.
FitzGerald, R. J., & OCuinn, G. (2006). Enzymatic debittering of food protein
hydrolysates. Biotechnology Advances, 24, 234237.
Gildberg, A., Arnesen, J. A., & Carlehog, M. (2002). Utilisation of cod backbone by
biochemical fractionation. Process Biochemistry, 38, 475480.
Harnedy, P. A., & FitzGerald, R. J. (2011). Bioactive peptides from marine processing
waste and shellsh: A review. Journal of Functional Foods, 4(1), 624.
Hayashi, N., Chen, R., Ikezaki, H., & Ujihara, T. (2008). Evaluation of the umami taste
intensity of green tea by a taste sensor. Journal of Agricultural and Food
Chemistry, 56, 73847387.
Ishizaka, T., Miyanaga, Y., Mukai, J., Asaka, K., Nakai, Y., Tsuji, E., & Uchida, T. (2004).
Bitterness evaluation of medicines for pediatric use by a taste sensor. Chemical
& Pharmaceutical Bulletin, 52, 943948.
Kataoka, M., Miyanaga, Y., Tsuji, E., & Uchida, T. (2004). Evaluation of bottled
nutritive drinks using a taste sensor. nternational Journal of Pharmaceutics, 279,
107114.
Kataoka, M., Yoshida, K., Miyanaga, Y., Tsuji, E., Tokuyama, E., & Uchida, T. (2005).
Evaluation of the taste and smell of bottled nutritive drinks. International
Journal of Pharmaceutics, 305, 1321.
Keast, R. S. J., & Breslin, P. A. S. (2002). An overview of binary taste-taste
interactions. Food Quality and Preference, 14, 111124.
Kikkawa, Y., Toko, K., Matsuno, T., & Yamafuji, K. (1993). Discrimination of taste of
amino acids with a multichannel taste sensor. Japanese Journal of Applied
Physics, 32, 57315736.
Kim, S.-K., & Wijesekara, I. (2010). Development and biological activities of marine-
derived bioactive peptides: A review. Journal of Functional Foods, 2, 19.
Kobayashi, Y., Habara, M., Ikezazki, H., Chen, R., Naito, Y., & Toko, K. (2010).
Advanced taste sensors based on articial lipids with global selectivity to basic
taste qualities and high correlation to sensory scores. Sensors, 10, 34113443.
Kovacs, Z., Sipos, L., Szollosi, D., Kokai, Z., Szekely, G., & Fekete, A. (2011). Electronic
tongue and sensory evaluation for sensing apple juice taste attributes. Sensor
Letters, 9, 12731281.
Leksrisompong, P. P., Mirale, R. E., & Drake, M. A. (2010). Characterization of avor
of whey protein hydrolysates. Journal of Agricultural and Food Chemistry, 58,
63186327.
Liao, L., Qiu, C.-Y., Liu, T.-X., Zhao, M.-M., Ren, J.-Y., & Zhao, H.-F. (2010).
Susceptibility of wheat gluten to enzymatic hydrolysis following deamidation
with acetic acid and sensory characteristics of the resultant hydrolysates.
Journal of Cereal Science, 52, 395403.
Li-Chan, E. C. Y., & Cheung, I. W. Y. (2010). Flavor-active properties of amino acids,
peptides and proteins. In Y. Mine, E. Li-Chan, & B. Jiang (Eds.), Bioactive proteins
and peptides as functional foods and nutraceuticals (pp. 341358). Ames, Iowa:
Wiley Blackwell (IFT Press Series).
Maehashi, K., & Huang, L. (2009). Bitter peptides and bitter taste receptors. Cellular
and Molecular Life Sciences, 66, 16611671.
Moon, S.-Y., & Li-Chan, E. C. Y. (2007). Changes in aroma characteristics of simulated
beef avor by soy protein isolate assessed by descriptive sensory analysis and
gas chromatography. Food Research International, 40, 12391248.
Mui, W. W. Y., Durance, T. D., & Scaman, C. H. (2002). Comparison of headspace
volatiles, sensory characteristics and texture of banana chips dried by
combinations of hot-air, vacuum and microwaves. Journal of Agricultural and
Food Chemistry, 50, 18831889.
Mukai, J., Miyanaga, Y., Ishizaka, T., Asaka, K., Nakai, Y., Tsuji, E., & Uchida, T. (2004).
Quantitative taste evaluation of total enteral nutrients. Chemical &
Pharmaceutical Bulletin, 52, 14161421.
Ou-Yang, Q., Zhao, J.-W., Chen, Q.-S., Lin, H., & Huang, X.-Y. (2011). Study on
classication of soy sauce by electronic tongue technique combined with
articial neural network. Journal of Food Science, 76, S523S527.
Pripp, A. H., & Ardo, Y. (2007). Modelling relationship between angiotensin-(I)-
converting enzyme inhibition and the bitter taste of peptides. Food Chemistry,
102, 880888.
Rhyu, M.-R., & Kim, E.-Y. (2011). Umami taste characteristics of water extract of
Doenjang, a Korean soybean paste: Low-molecular acidic peptides may be a
possible clue to the taste. Food Chemistry, 127, 12101215.
Rudnitskaya, A., Polshin, E., Kirsanov, D., Lammertyn, J., Nicolai, B., Saison, D.,
Delvaux, F. R., Delvaux, F., & Legin, A. (2009). Instrumental measurement of beer
taste attributes using an electronic tongue. Analytical Chimica Acta, 646,
111118.
Rudnitskaya, A., Rocha, S. M., Legin, A., Pereira, V., & Marques, J. C. (2010).
Evaluation of the feasibility of the electronic tongue as a rapid analytical tool for
wine age prediction and quantication of the organic acids and phenolic
compounds. Analytical Chimica Acta, 662, 8289.
Schlichtherle-Cerny, H., & Amad, R. (2002). Analysis of taste-active compounds in
an enzymatic hydrolysate of deamidated wheat gluen. Journal of Agricultural
and Food Chemistry, 50, 15151522.
Seo, W. H., Lee, H. G., & Baek, H. H. (2008). Evaluation of bitterness in enzymatic
hydrolysates of soy protein isolate by taste dilution analysis. Journal of Food
Science, 73(1), S4146.
Shen, Q., Guo, R., Dai, Z., & Zhang, Y. (2012). Investigation of enzymatic hydrolysis
conditions on the properties of protein hydrolysate from sh muscle
(Collichthys niveatus) and evaluation of its functional properties. Journal of
Agricultural and Food Chemistry, 60, 51925198.
Su, G., Ren, J., Yang, B., Cui, C., & Zhao, M. (2011). Comparison of hydrolysis
characteristics on defatted peanut meal proteins between a protease extract
from Aspergillus oryzae and commercial proteases. Food Chemistry, 126,
13061311.
Temussi, P. A. (2011). New insights into the characteristics of sweet and bitter taste
receptors. International Review of Cell and Molecular Biology, 291, 191226.
Udenigwe, C. C., & Aluko, R. E. (2011). Food protein-derived bioactive peptides:
Production, processing, and potential health benets. Journal of Food Science, 71,
R1124.
Woertz, K., Tissen, C., Kleinebudde, P., & Breitkreutz, J. (2011). Taste sensing systems
(electronic tongues) for pharmaceutical applications. International Journal of
Pharmaceutics, 417, 256271.
Wu, J., & Aluko, R. E. (2007). Quantitative structure-activity relationship study of
bitter di- and tri- peptides including relationship with angiotensin I-converting
enzyme inhibitory activity. Journal of Peptide Science, 13, 6369.
Zhang, M.-X., Wang, X.-C., Liu, Y., Xu, X.-L., & Zhou, G.-H. (2012). Isolation and
identication of avour peptides from Puffer sh (Takifugu obscurus) muscle
using an electronic tongue and MALDI-TOF/TOF MS/MS. Food Chemistry, 135,
14631470.
I.W.Y. Cheung, E.C.Y. Li-Chan/ Food Chemistry 145 (2014) 10761085 1085