A novel method to quantify the activity of alcohol acetyltransferase Using a SnO2-based sensor of electronic nose

Article abstract of DOI:10.1016/j.foodchem.2016.02.087

Alcohol acetyltransferase (AATFase) extensively catalyzes the reactions of alcohols to acetic esters in microorganisms and plants. In this work, a novel method has been proposed to quantify the activity of AATFase using a SnO₂-based sensor of electronic nose, which was determined on the basis of its higher sensitivity to the reducing alcohol than the oxidizing ester. The maximum value of the first-derivative of the signals from the SnO₂-based sensor was therein found to be an eigenvalue of isoamyl alcohol concentration. Quadratic polynomial regression perfectly fitted the correlation between the eigenvalue and the isoamyl alcohol concentration. The method was used to determine the AATFase activity in this type of reaction by calculating the conversion rate of isoamyl alcohol. The proposed method has been successfully applied to determine the AATFase activity of a cider yeast strain. Compared with GC-MS, the method shows promises with ideal recovery and low cost.

Full text of DOI:10.1016/j.foodchem.2016.02.087

Food Chemistry 203 (2016) 498–504
Contents lists available at ScienceDirect
Food Chemistry
journal homepage: www.elsevier.com/locate/foodchem
Analytical Methods
A novel method to quantify the activity of alcohol acetyltransferase
Using a SnO -based sensor of electronic nose
2
⇑
Zhongqiu Hu, Xiaojing Li, Huxuan Wang, Chen Niu, Yahong Yuan, Tianli Yue
College of Food Science and Engineering, Northwest A&F University, Yangling, Shaanxi 712100, PR China
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 27 July 2015
Received in revised form 29 January 2016
Accepted 13 February 2016
Available online 15 February 2016
Alcohol acetyltransferase (AATFase) extensively catalyzes the reactions of alcohols to acetic esters in
microorganisms and plants. In this work, a novel method has been proposed to quantify the activity of
AATFase using a SnO
sensitivity to the reducing alcohol than the oxidizing ester. The maximum value of the first-derivative
of the signals from the SnO -based sensor was therein found to be an eigenvalue of isoamyl alcohol con-
2
-based sensor of electronic nose, which was determined on the basis of its higher
2
centration. Quadratic polynomial regression perfectly fitted the correlation between the eigenvalue and
the isoamyl alcohol concentration. The method was used to determine the AATFase activity in this type of
reaction by calculating the conversion rate of isoamyl alcohol. The proposed method has been success-
fully applied to determine the AATFase activity of a cider yeast strain. Compared with GC–MS, the
method shows promises with ideal recovery and low cost.
Keywords:
Electronic nose
Isoamyl alcohol
Acetyltransferase activity
Gas sensor
Feature extraction
Ó 2016 Elsevier Ltd. All rights reserved.
1
. Introduction
MS) (Inoue et al., 1997; Rojas et al., 2002; Yoshioka & Hashimoto,
1
981). The method was often time-consuming and expensive
Alcohol acetyltransferase (AATFase, EC 2.3.1.84) extensively
because of complex steps such as pre-extraction using solid phase
micro-extraction (SPME), complicated oven temperature-lift pro-
gram and higher temperatures together with pure nitrogen or
helium and hydrogen as carrier gas. Furthermore, GC–MS is more
suitable for most of the complex volatile mixtures. A continuing
demand exists for rapid, inexpensive and effective techniques to
determine the activity of AATFase. The development of electronic
nose (e-nose) offers a simple and fast analytical method and the
possibilities of cheap and portable instrument which could be
operated by common testers (Röck, Barsan, & Weimar, 2008). Anal-
yses based on e-nose can be carried out with general filtered-air as
carrier gas (Olguín et al., 2014). E-nose is always equipped with a
set of sensor array, which can respond sensitively and promptly
to different kinds of volatile organic compounds (VOCs) (López
de Lerma, Bellincontro, García-Martínez, Mencarelli, & Moreno,
exists in microorganisms and plants (Kiyohara et al., 2014; Zhang
et al., 2012), which catalyzes the synthesis of various acetate esters
in vivo. One typical acetate ester is isoamyl acetate, which is con-
verted from isoamyl alcohol with the catalysis of AATFase in Sac-
charomyces cerevisiae, as shown in Fig. 1. Isoamyl acetate is
recognized as an important substance for determining the flavor
and quality of sake and other alcoholic beverages and has been
gaining more attention (Inoue et al., 1997). Hence the activity of
AATFase was widely studied to enhance the content of isoamyl
acetate in products or understand its molecular mechanism
(
Galaz, Morales-Quintana, Moya-León, & Herrera, 2013; Pires,
Teixeira, Brányik, & Vicente, 2014; Zhang et al., 2013).
The activity of AATFase was traditionally determined by head
space gas chromatography coupled with mass spectrometry (GC–
2
013). The response outputs of the sensor array often represent
the blending information of all VOCs contained in a sample.
Accordingly, feature extraction to obtain adaptable salient feature
information becomes a critical step. However, it is arduous to prop-
erly interpret the obtained sensor signals corresponding to the
compounds in the mixtures in a practical approach (García-
González, Tena, Aparicio-Ruiz, & Aparicio, 2014; Scott, James, &
Ali, 2006). Some works successfully described the profile of VOCs
using qualitative statistic methods such as principal components
Abbreviations: AATFase, alcohol acetyltransferase; VOCs, volatile organic com-
pounds; G/G
the sensors to the sample gas (G) relative to the carrier gas (G
the first-derivative of the relative conductivity to time; (dG/dt)max, the maximum
value of the first-derivative of the relative conductivity to time; (dG/dt)calibration max
0
, the relative conductivity i.e. the ratio of the conductivity response of
0
) over time; dG/dt,
,
the calibrated (dG/dt)max; dE/dt, enzyme activity expression; ꢀ(dCisoamyl alcohol/dt),
the isoamyl alcohol consumption amounts per unit time; (dCisoamyl alcohol/dt)max, the
maximum value of ꢀ(dCisoamyl alcohol/dt).
⇑
Corresponding author.
E-mail address: yuetl305@nwsuaf.edu.cn (T. Yue).
http://dx.doi.org/10.1016/j.foodchem.2016.02.087
308-8146/Ó 2016 Elsevier Ltd. All rights reserved.
0

Z. Hu et al. / Food Chemistry 203 (2016) 498–504
499
O
in the headspace of exclusive vials were automatically pumped
ꢀ1
AATFase
into the ten-sensor chambers at a speed of 300 mL min . The
operation temperature of the sensors was set at 300 °C. The
response signals of the ten sensors were collected in 58 s. At inter-
vals of every two tests, a stream of clean air was blown into the
measuring chambers to flush the absorbed VOCs from the gas sen-
sors until the signals returned to baseline.
+ Acetyl-CoA
OH
O
Isoamyl alcohol
Isoamyl acetate
Fig. 1. The conversion reaction of isoamyl alcohol to isoamyl acetate as a typical
example of AATFase-catalyzed esterification reactions.
analysis (PCA) (Cheng, Qin, Guo, Hu, & Wu, 2013) and Neural Net-
works (Aleixandre et al., 2004). However, the quantification of
components of VOCs using e-nose is still challenging. Recently,
diethyl cyanophosphonate in different chemical warfare agent
mimics was quantified by e-nose using a partial least squares
2.3. Preparation of AATFase
The preparation of AATFase was performed according to the
previous report with certain modifications (García-González
et al., 2014). Starter cultures of the cider yeast strain 38 were cul-
tured in yeast peptone dextrose (YPD) medium containing 1% yeast
extract, 2% Bactopeptone, and 2% dextrose. The media were auto-
claved at 121 °C for 15 min in advance. Yeast cells were grown at
28 °C for 48 h without shaking, 4–5 g culture was collected by cen-
trifugation at 4500ꢂg for 5 min, then washed twice with 0.85%
sodium chloride, and re-suspended in10 mL ice-chilled potassium
phosphate buffer (10 mM, pH 7.0), and re-centrifuged at 4500ꢂg
for another 5 min. The resulting precipitation was frozen in liquid
nitrogen and extensively ground with 15 mL of glass beads (0.4–
0.5 mm diameter) using a vortex instrument. The cells were sub-
jected to five bursts of 45 s with resting periods of 5 min. The
resulting homogenate was dissolved in 10 mL phosphate buffer
(10 mM, pH 7.0) and centrifuged at 15,000ꢂg for 30 min at 4 °C.
The obtained supernatant containing AATFase was added into a
labeled 50 mL headspace vial with isoamyl alcohol and acetyl-
CoA to quantify AATFase activity by e-nose.
(
PLS) derived mathematical model (Olguín et al., 2014). This pio-
neered the precedence of quantitative analyses using e-nose. In
e-nose, there are three types of sensors used i.e. metal-oxide
(
MOS) gas sensors, conducting polymer (CP) and quartz microbal-
ance (QMB) (Baietto, Wilson, Bassi, & Ferrini, 2010). Therein, MOS
are taking an ever growing part in extensive application areas
(Bayn et al., 2013; Pan, Wu, Peng, Zeng, & Li, 2014; Pan, Zhang,
Zhu, Mao, & Tu, 2014; Tian, Li, Qin, Yu, & Ma, 2014). Their working
mechanism is that the electrical conductivity of sensors varies with
the surrounding atmosphere temperature, species and contents
(Comini et al., 2013; Varpula, Novikov, Haarahiltunen,
&
Kuivalainen, 2011). Among the MOS sensor array, the SnO
2
based
sensor is mostly studied (Comini et al., 2013), which was effective
to identify or classify samples with the appropriate algorithm
(Acevedo, Maldonado, Dominguez, Narvaez, & Lopez, 2007;
Aldao, Schipani, Ponce, Joanni, & Williams, 2014; Kohl, Heinert,
Bock, Hofmann, & Schieberle, 2000). Determination of the activity
of enzymes using the SnO
explored.
2
based sensor still remains to be further
2.4. Choice of the most sensitive sensor
The aim of the present work is to establish a novel quantifica-
tion method to determine the AATFase activity. The quantification
was decided on the depletion of isoamyl alcohol in the enzyme-
The Airsense Electronic Nose System PEN3.5 (Device No. 33086,
Airsense Analytics GmbH, Germany) was used to quantify AATFase
activity, which consists of ten MOS gas sensors in the installed
chambers. According to the reaction progress, five reaction mix-
tures were simulated to contain the different concentrations of iso-
amyl alcohol and isoamyl acetate except necessary enzymes,
which is shown in Table 1. They were scanned by ten sensors.
The sensor whose responses can most obviously differentiate the
five simulated mixtures while can index isoamyl alcohol or isoa-
myl acetate concentrations would be chosen to be the most sensi-
tive sensor. Each simulated mixture was scanned in triplicate for
statistical purposes.
2
catalyzed process using a SnO -based sensor of e-nose PEN3.5.
The first-derivative was found to be the expected feature extrac-
tion for the analysis of isoamyl alcohol concentration. The chemo-
metric model is established by the polynomial regression. The
method showed possibility and accuracy to determine the AATFase
activity of a cider yeast strain.
2
. Materials and methods
2.1. Reagents and materials
2.5. Establishment of chemometric model of isoamyl alcohol
Acetyl-CoA (HPLC grade, P93%), analytical grade isoamyl alco-
hol and isoamyl acetate were the products of Sigma-Aldrich Inc.
The analytically pure anhydrous sodium carbonate, sodium chlo-
ride and 10 mM pH7.0 phosphate buffer were purchased from
Tiancheng Chemical Company (Yangling, China). Bacto-yeast
extract, bacto-peptone, dextrose were purchased from Beijing
Aoboxing Bio-tech Co. Ltd. The cider yeast strain 38 was kept by
our laboratory, which was verified to have ꢁ99% homology with
YOR377W_S288C by Basic Local Alignment Search Tool (BLAST) of
homologous alcohol acetyl-transferase gene (ATF1) and
branched-chain aminotransferase gene 2 (BAT2) sequences.
In order to establish a chemometric model to determine the
concentration of isoamyl alcohol in the AATFase-catalyzed reaction
mixture, two simulation groups were randomly designed accord-
ing to the characteristic of the reaction progress, as shown in
Table 2. In the first group, five decreasing concentrations of isoa-
myl alcohol were in turn mixed with five increasing concentrations
of isoamyl acetate to develop five trials, which were designed in
accordance with the progress of the AATFase-catalyzed reaction.
Table 1
The list of the simulated reaction mixtures screened by e-nose PEN 3.5 sensor array.
2.2. E-nose measurement
Component
Simulated reaction mixtures
The e-nose measurements of all samples were standardized by
1
2
3
4
5
controlling temperature, time and stirring to obtain reproducible
Acetyl-CoA
2.3 mM
responses, viz., 10 mL of sample was maintained at 25 ± 2 °C for
Isoamyl alcohol
AATFase
Isoamyl acetate
0.92 mM
The cell extract of 5 g yeast cells
0 mM 0.23 mM 0.46 mM
0.69 mM
0.46 mM
0.23 mM
0.69 mM
0 mM
3
0 min in a 50 mL headspace vial with cap to obtain headspace
0.92 mM
gas/liquid equilibrium prior to e-nose analysis. The VOCs of sample

5
00
Z. Hu et al. / Food Chemistry 203 (2016) 498–504
The concentration gradients of isoamyl alcohol and isoamyl acetate
in the second group were different from those in the first group.
The final concentrations of isoamyl alcohol and isoamyl acetate
of the trials are listed in Table 2. 10 mL of each mixture was trans-
ferred into a 50-mL headspace vial, 3.3 g anhydrous sodium car-
bonate was immediately added into the vial, and tightly plugged
with a silicon rubber cap. The vials were stored at 25 ± 2 °C to equi-
librate for 30 min prior to e-nose measurement. 10 replicates for
every trial were measured by the e-nose PEN3.5. The response sig-
nals of the most sensitive sensor were extracted out from the data
set obtained by the ten-sensor array to establish chemometric
model to quantify isoamyl alcohol concentrations in the simulated
mixture. In detail, the first-derivative of response signal of each
trial from the sensor was calculated and found thereof an eigen-
value of isoamyl alcohol concentration of the trial. The correlation
between the eigenvalue and the isoamyl alcohol concentration was
fitted by quadratic polynomial regression. Two regressive chemo-
metric equations obtained from the two groups of data were mutu-
ally validated and evaluated through calculating recovery, namely,
the equation of one group is set as a standard model, and the eigen-
value of each trial in the other group is used as a validation set to
calculate the concentration of isoamyl alcohol. The recovery of
each trial is the ratio of the predict value and the set value of iso-
amyl alcohol concentration.
2.6. Verification to determine the activity of AATFase
Cellextractfromthecideryeaststrain38containingAATFasewas
used to verify the resultant models. The preparation protocol of the
AATFase-catalyzed reaction was based on the previous reports
(Inoue et al., 1997; Rojas et al., 2002; Yoshioka & Hashimoto, 1981)
with slight modifications, in which GC–MS was used to determine
the activity of AATFase. Briefly, a 1 mL graduated glass syringe was
used as the enzyme reactor. Reaction mixtures were prepared by
pipetting400
taining 92 mM isoamyl alcohol, 400
phate buffer (pH 7.0) containing 92 mM acetyl-CoA, and 200
l
Lof10 mMpotassiumphosphatebuffer(pH7.0)con-
lL of 10 mM potassium phos-
l
L
cell extract containing AATFase. The total volume was 1 mL in the
syringe. Subsequently, entrapped air in it was removed by pushing
the piston. The reaction mixture was immediately agitated on a sha-
ker at 25 ± 2 °C. At different time intervals (20, 40, 60, 80, 100 min), a
0
.2 mLreactionmixturewastakenfromthesyringeandplacedintoa
5
0-mL exclusive headspace vial, 400 L of saturated KSCN solution
l
was added to stop the reaction. Subsequently, the reaction mixture
was promptly diluted to 10 mL using 10 mM potassium phosphate
buffers (pH 7.0). Then, 3.3 g anhydrous sodium carbonate was
immediately added into the vial for e-nose measurement. The mea-
surement was repeated five times. The obtained measurement data
were processed to calculate the activity of AATFase according to the
proposed method. Meanwhile, GC–MS was also used to determine
the activity of the AATFase according to the reports (Inoue et al.,
1
997; Rojas et al., 2002; Yoshioka & Hashimoto, 1981), its result
was used as the reference standard of that of the proposed method.
2.7. Data analysis
Statistical analyses were processed using SPSS Statistics 18.0
(
SPSS Inc.).
3
. Results and discussion
3.1. Choice of the most sensitive sensor
To quantify the activity of AATFase using e-nose, it is necessary
to find out the most sensitive sensor to the concentration
variations of isoamyl alcohol or isoamyl acetate in the

Z. Hu et al. / Food Chemistry 203 (2016) 498–504
501
Fig. 2. The responses of ten sensors to the simulated mixtures of acetate-forming reaction. The response signals in A are the representative responses of Sensor W3C, W1C
and W5C to the five simulated reaction mixtures. The response signals in B are the representative responses of Sensor W5S, W1S, W2S, W3S, W6S and W2W to the five
simulated reaction mixtures. The response signals in C are the representative responses of W1W to the five simulated reaction mixtures. The detailed recipes of the five
mixtures are displayed in Table 1. Therein, the five mixtures from 1 to 5 successively contain the concentrations of isoamyl alcohol and isoamyl acetate: 1, 0.92 M isoamyl
alcohol; 2, 0.69 M isoamyl alcohol + 0.23 M isoamyl acetate; 3, 0.46 M isoamyl alcohol + 0.46 M isoamyl acetate; 4, 0.23 M isoamyl alcohol + 0.69 M isoamyl acetate; 5, 0.92 M
isoamyl acetate.
AATFase-catalyzed reaction system. Therefore, five simulated
AATFase-catalyzed reaction mixtures were designed according to
the reaction progress, which is shown in Table 1. In the five mix-
tures, the concentrations of isoamyl alcohol continually decreased
and the concentrations of isoamyl acetate simultaneously
increased in an equimolar ratio. The five reaction mixtures were
screened by the ten sensors of e-nose PEN3.5. The representative
conductivity increased with the increase of the concentration of
isoamyl alcohol in the five simulated reaction mixtures. This may
have resulted from the gas exchange between reducing gas like
isoamyl alcohol and pre-adsorbed oxygen on the surface of the
sensor. The result was in accordance with the reported behavior
2
of isopropyl alcohol, ethanol and hexanol vapors on SnO -based
sensors (García-González et al., 2014; Varpula et al., 2011). The
exchange between volatile isoamyl acetate and pre-adsorbed oxy-
gen may rarely happen on W1W sensor when isoamyl alcohol
existed in their mixture gas. In the reaction mixtures, acetyl-CoA
and cell extract were not volatile, they occupied less share in head-
space mixture gas so that their signals were ignored. The screening
results indicated that the W1W sensor was the most sensitive one
to the variation of isoamyl alcohol of the reaction mixtures. There-
fore, the W1W sensor was chosen to quantify the concentration of
isoamyl alcohol in the AATFase-catalyzed reaction system.
responsive signals (G/G
types, which are illustrated in Fig. 2. The representative responses
G/G ) of W3C, W1C and W5C sensors to the five simulated reac-
0
) of the ten sensors are classified into three
(
0
tion mixtures (Fig. 2A) showed that the relative conductivities
increased in the first 3 s and then decreased until ꢁ15 s, which
was induced by the impinging of the headspace gas from reaction
mixture and new equilibrium of the desorption and adsorption. In
the beginning, the transient fall period of 3 s may be due to the
replacement of a small number of oxygen molecules by a target
component of reaction mixture resulting in the decrease of the
sensor resistance and the lift of the relative conductivity at the
moment. Along with the abundant replacement of the occupied
oxygen by the exchanged gas from the reaction mixture, the resis-
tance of sensors rose and the relative conductivity dropped. Such
variation tendency of the relative conductivity may be ascribed
to the gas exchange between oxidizing gas like isoamyl acetate
and pre-adsorbed oxygen on the surface of the MOS sensors like
tungsten trioxide (Vallejos et al., 2008). It also can be seen from
Fig. 2A that the relative conductivity of the sensors was so weak
that the three sensors were not adaptable to determine the con-
centrations of isoamyl alcohol or isoamyl acetate in the five simu-
lated reaction mixtures. The representative responses signals (G/
3.2. Establishment and validation of the chemometric models
3.2.1. Establishment
In order to establish the chemometric model to quantify the
concentration variations of isoamyl alcohol in the AATFase-
catalyzed reaction, two simulation group reaction mixtures con-
taining different concentration gradients of isoamyl alcohol were
randomly designed according to the progress of the AATFase-
catalyzed reaction, as shown in Table 2. These reaction mixtures
were measured by the SnO -based W1W sensor. The responses
2
of W1W sensor to Group 1 and Group 2 are illustrated in
Fig. 3A and D, respectively. It is obvious that the relative conductiv-
ities decreased with the decrease of the concentrations of isoamyl
alcohol in the simulated reaction mixture. But no such positive cor-
relation existed with the increase of the concentrations of isoamyl
acetate in the reaction mixture. Consequently, it became critical to
setup a chemometric model to quantify the concentration of isoa-
myl alcohol using obtained data. The first-derivative curves of the
relative conductivity curves of Group 1 and 2 are illustrated in
Fig. 3B and E, respectively. The first-derivative curves (dG/dt) were
well separated corresponding to various concentrations of isoamyl
G
0
) in Fig. 2B of the W5S, W1S, W2S, W3S, W6S and W2W sensors
show that the relative conductivities slightly decreased in the first
s and then increased until ꢁ15 s, whose responses were opposite
3
to those of W3C, W1C and W5C sensors. It can be seen that the six
responsive curves were overlapped and disordered with the
ordered concentration variations of isoamyl alcohol or isoamyl
acetate. The disorder of the responsive curves may be due to the
gas exchange between isoamyl alcohol or isoamyl acetate and
pre-adsorbed oxygen on the surface of the type sensors, their
molecules may interact with each other in the adsorption–desorp-
tion processes (Vuong, Kim, & Kim, 2015). The result of Fig. 2B sug-
gested that the six sensors were also not suitable to determine the
concentrations of isoamyl alcohol and isoamyl acetate in the five
simulated reaction mixtures. The response signals in Fig. 2C of
alcohol. A maximum value ((dG /dt)max) for each first-derivative
ij
curve was clearly seen, which indicated that the relative conduc-
tivity against time of the SnO -based sensor could successfully
2
describe the responses of W1W sensor to the concentrations of iso-
amyl alcohol in the reaction mixtures. The calculated maximum
values are particularized in Table 2. According to the response
2
W1W sensor (SnO -based sensor) demonstrated that the relative

5
02
Z. Hu et al. / Food Chemistry 203 (2016) 498–504
Fig. 3. The responses of W1W sensor, the first-derivatives of responses and the relational model relationship curve between the (dG/dt)calibration max values and isoamyl
alcohol concentrations. A: the responses of W1W sensor to Group 1; B: the first-derivatives of the responses of Group 1 data; C: the mathematical model relationship curve
between the (dG/dt) calibration max values and the isoamyl alcohol concentrations of Group 1 trials. Similarly, D: the responses of W1W sensor to Group 2; E: the first-
derivatives of the responses of Group 2 data; F: The mathematical model relationship curve between the (dG/dt)calibration max values and the isoamyl alcohol concentrations of
Group 2.
mechanism of e-nose, the (dGij/dt)max values could be the dynamic
balance of the exchange between isoamyl alcohol and oxygen on
Here Y
(mM); X
1
and Y
and X
2
represent the concentration of isoamyl alcohol
2
represent (dG/dt)calibration max value.
1
the gas-sensing film of the SnO
ical significance can be explained to be the kinetic equilibrium con-
stant of the process (Vuong et al., 2015). The maximum value
2
-based W1W sensor, whose phys-
Therefore, the quadratic polynomial regression equations
derived from the first-derivative of sensor responses can perfectly
quantify the concentrations of isoamyl alcohol in the AATFase-
catalyzed reactions.
2
reveals time-domain characteristics of SnO -based sensors put for-
ward by Varpula et al. (2011), which was induced by the concen-
tration variations of reducing gases such as isoamyl alcohol.
2
Although SnO -based W1W sensor is not sensitive to isoamyl acet-
ate, the (dGij/dt)max values might include the noise of isoamyl acet-
ate. To evaluate the effects of pure isoamyl acetate, Trial No. 15 and
3
.2.2. Cross verification of mathematical models
In order to evaluate the reliability of Eqs. (1) and (2), the mutual
cross verification between Eq. (1) of Group 1 and Eq. (2) of Group 2
was thereof carried out (Olguín et al., 2014). The (dG/dt)calibration
max values of Group 1 were put into Eq. (2) in turn to get the the-
oretical values of isoamyl alcohol concentrations of the trials in
Group 1. On the contrary, The (dG/dt)calibration max values of Group
2
5 shown in Table 2 were designed in group 1 and 2 to obtain the
noise values. The corresponding maximum noise value was 0.184
and 0.148, respectively. The noise values of the two groups were
subtracted when quantifying the concentration of isoamyl alcohol,
i.e. the (dGij/dt)max value of Trial No. 15 was separately deducted
from the original (dGij/dt)max values of Trial No.11, 12, 13 and 14.
The obtained difference values were expressed as (dG/dt)calibration
max. Similarly, the (dG/dt)calibration max values of Trial No. 21, 22,
2
were separately put into Eq. (1) to get the theoretical values of
isoamyl alcohol concentrations of the trials in Group 2. The recov-
ery for each trial was obtained by calculating the ratio of the the-
oretical concentration to the set concentration value of isoamyl
alcohol, which are also displayed in Table 2. It can be seen that
the recoveries for all trials were higher than 76.5%. The results sug-
gested that the models were feasible and could be successfully
used to quantify the changes of isoamyl alcohol concentration in
the AATFase catalyzed reaction. As a result, the method can be
potentially used to determine AATFase activity.
2
3 and 24 were obtained by subtracting the (dGij/dt)max value of
Trial No. 25 from the original (dGij/dt)max values. The resulting
dG/dt)calibration max values are listed in Table 2, which were puta-
(
tive to be well correlated to the concentrations of isoamyl alcohol
in the reaction mixture. Subsequently, the chemometric model to
determine the concentration of isoamyl alcohol in groups 1 and 2
was tried to achieve by mathematical regression. The quadratic
polynomial regression equations (Eqs. (1) and (2)) were found to
perfectly describe the relationship between the (dG/dt)calibration
max values and the isoamyl alcohol concentrations (Fig. 3C and F,
3
.3. Application of the proposed models
2
Table 2), due to the goodness of fit (R value was 0.985 and
In order to demonstrate the practical application of the pro-
0
.996, respectively).
posed models, the activity of AATFase derived from an industrial
cider yeast strain 38 was determined by the method. GC–MS was
also used to determine the activity of the AATFase. The measure-
ment results of the reaction mixtures by W1W sensor at different
time intervals (20, 40, 60, 80, 100 min) are shown in Fig. 4A. The
2
Y
1
2
¼ 0:0946X þ 0:471X
1
þ 0:0267
þ 0:0093
ð1Þ
ð2Þ
1
2
2
Y
¼ ꢀ0:0042X þ 0:559X
2

Z. Hu et al. / Food Chemistry 203 (2016) 498–504
503
Fig. 4. The responses of W1W sensor, the first derivatives of the responses and time course of isoamyl alcohol concentrations in the AATFase-catalyzed reaction process of the
cider yeast strain 38 cells. A: The responses of W1W sensor to the reaction mixture at different time points. B: the first derivatives of the responses. C: The time course of
isoamyl alcohol concentrations in the AATFase-catalyzed reaction process of the cider yeast strain 38 cells calculated by Model Eq. (1). D: The time course of isoamyl alcohol
concentrations in the AATFase-catalyzed reaction process of the cider yeast strain 38 cells calculated by Model Eq. (2). Every curve was the average of 5 parallel trials.
first order derivatives of the five responses of Fig. 4A were per-
formed and are displayed in Fig. 4B. The obtained (dG/dt)max values
were subsequently taken into the Model equation Eq. (1) or Eq. (2)
to quantify the real time concentrations of isoamyl alcohol in the
reaction mixtures. The real-time concentrations of isoamyl alcohol
to reaction time course are plotted in Fig. 4C and D. From Fig. 4C, it
can be seen that the relationship of dCisoamyl alcohol/dt to the reac-
tion time presents a negative linear correlation, the slope of it is
4. Conclusion
A novel method to quantify the activity of alcohol acetyltrans-
ferase was developed using e-nose PEN 3.5. The SnO -based sensor
was sensitive to volatile reducing gas among the ten sensors of e-
nose. The first-derivative of the signals from the SnO -basied sen-
2
2
sor was found to well describe the change of the isoamyl alcohol
concentration. The maximum value of the first-derivative reveals
the eigenvalue of the maximum concentration, which was decided
to represent the real concentration of isoamyl alcohol in the reac-
tion mixture. The application of the proposed method to a cider
yeast strain demonstrated that the method is feasible and accurate
for determining the activity of alcohol acetyltransferase in an
esterification reaction system. The method is efficient and could
be used as real-time measurement for isoamyl alcohol concentra-
tion in mixtures.
ꢀ1
ꢀ1
ꢀ
0.356 mM min , i.e. ꢀ21.36 mM h . The value is divided by
average 4512 mg of cell wet weight and its quotient is equal to
ꢀ
1
ꢀ1
ꢀ
4.73
equivalent to ꢀ(dCisoamyl alcohol/dt), hence the activity of AATFase
should be 4.73 M mg . Similarly, the activity value of AAT-
Fase from Fig. 4D was 4.01
obtained by Eqs. (1) and (2) were ꢁ91.0 and ꢁ77.1% of the GC–
lM h mg cells. Because the AATFase activity dE/dt is
ꢀ1
ꢀ1
l
h
ꢀ
1
ꢀ1
l
M mg
h . The two activity values
ꢀ1
ꢀ1
MS result (5.20
lM mg
h ), respectively. These indicated that
the proposed method can perfectly determine the activity of AAT-
Fase of yeast strain, which may be extended to other microorgan-
isms and plants. The proposed method was validated to have
several advantages over head space-gas chromatography coupled
with mass spectrometry (HS-GC–MS) for determining the AATFase
activity (Rojas et al., 2002; Yoshioka & Hashimoto, 1981). The
advantages of the present method are to (1) carry out in 1 min plus
Author contributions
Zhongqiu Hu and Xiaojing Li designed the plan and performed
it, Huxuan wang, Chen Niu and Yahong yuan gave us their arms.
Tianli Yue was the adviser and the financial contributor of the
work.
3
0 min head space time for the whole analysis process, which
saves much time than the reported HS-GC–MS method. HS-GC–
MS method took 13 min for isoamyl acetate and 20 min for isoamyl
alcohol except the same head space time (Bosch-Fusté et al., 2007).
Furthermore, our method only needs 200 s to clean the sensor, but
HS-GC–MS needs tens of min to clean the capillary column. (2)
save resources. HS-GC–MS method always needs helium, hydrogen
or nitrogen as carrier gases and a high temperature range pro-
gramed from 40 °C to 250 °C plus additional SPME fibers and
Compliance with ethical standards
The authors declare that they have no any conflict of interest.
Additional informed consent was obtained from all individual par-
ticipants for whom identifying information is included in this arti-
cle. This article does not contain any studies with human
participants or animals performed by any of the authors.
expensive HS-GC–MS instrument (Comuzzo, Tat, Tonizzo,
&
Acknowledgments
Battistutta, 2006; Styger, Jacobson, & Bauer, 2011). Our method
only needs air as the carrier gas plus cheaper and portable e-
nose. (3) operate simpler than HS-GC/MS method.
The authors are grateful for the financial support from China
State ‘‘12th Five-Year Plan” scientific and technological support

5
04
Z. Hu et al. / Food Chemistry 203 (2016) 498–504
schemes (2012BAD31B01 and 2012BAK17B06) and National natu-
ral science foundation project (31371814).
López de Lerma, M. D. L. N., Bellincontro, A., García-Martínez, T., Mencarelli, F., &
Moreno, J. J. (2013). Feasibility of an electronic nose to differentiate commercial
Spanish wines elaborated from the same grape variety. Food Research
International, 51(2), 790–796.
References
Olguín, C., Laguarda-Miró, N., Pascual, L., García-Breijo, E., Martínez-Mañez, R., &
Soto, J. (2014). An electronic nose for the detection of Sarin, Soman and Tabun
mimics and interfering agents. Sensors and Actuators B: Chemical, 202, 31–37.
Pan, D., Wu, Z., Peng, T., Zeng, X., & Li, H. (2014). Volatile organic compounds profile
during milk fermentation by Lactobacillus pentosus and correlations between
volatiles flavor and carbohydrate metabolism. Journal of Dairy Science, 97(2),
Acevedo, F., Maldonado, S., Dominguez, E., Narvaez, A.,
& Lopez, F. (2007).
Probabilistic support vector machines for multi-class alcohol identification.
Sensors and Actuators B: Chemical, 122(1), 227–235.
Aldao, C. M., Schipani, F., Ponce, M. A., Joanni, E.,
Conductivity in SnO2 polycrystalline thick film gas sensors: Tunneling electron
transport and oxygen diffusion. Sensors and Actuators B: Chemical, 193, 428–433.
Aleixandre, M., Sayago, I., Horrillo, M., Fernandez, M., Ares, L., Garcıa, M., ...
& Williams, F. J. (2014).
624–631.
Pan, L., Zhang, W., Zhu, N., Mao, S., & Tu, K. (2014). Early detection and classification
of pathogenic fungal disease in post-harvest strawberry fruit by electronic nose
and gas chromatography–mass spectrometry. Food Research International, 62,
´
Gutiérrez, J. (2004). Analysis of neural networks and analysis of feature
selection with genetic algorithm to discriminate among pollutant gas. Sensors
and Actuators B: Chemical, 103(1), 122–128.
162–168.
Pires, E. J., Teixeira, J. A., Brányik, T., & Vicente, A. A. (2014). Yeast: the soul of beer’s
aroma-a review of flavour-active esters and higher alcohols produced by the
brewing yeast. Applied microbiology and biotechnology (5), 1–13.
Röck, F., Barsan, N., & Weimar, U. (2008). Electronic nose: current status and future
trends. Chemical Reviews, 108(2), 705–725.
Rojas, V., Gil, J. V., Manzanares, P., Gavara, R., Piñaga, F., & Flors, A. (2002).
Measurement of alcohol acetyltransferase and ester hydrolase activities in yeast
extracts. Enzyme and Microbial Technology, 30(2), 224–230.
Baietto, M., Wilson, A. D., Bassi, D., & Ferrini, F. (2010). Evaluation of three electronic
noses for detecting incipient wood decay. Sensors, 10(2), 1062–1092.
Bayn, A., Nol, P., Tisch, U., Rhyan, J., Ellis, C. K., & Haick, H. (2013). Detection of
volatile organic compounds in Brucella abortus-seropositive bison. Analytical
Chemistry, 85(22), 11146–11152.
Bosch-Fusté, J., Riu-Aumatell, M., Guadayol, J. M., Caixach, J., López-Tamames, E., &
Buxaderas, S. (2007). Volatile profiles of sparkling wines obtained by three
extraction methods and gas chromatography–mass spectrometry (GC–MS)
analysis. Food Chemistry, 105(1), 428–435.
Scott, S. M., James, D., & Ali, Z. (2006). Data analysis for electronic nose systems.
Microchimica Acta, 156(3–4), 183–207.
Styger, G., Jacobson, D., & Bauer, F. (2011). Identifying genes that impact on aroma
profiles produced by Saccharomyces cerevisiae and the production of higher
alcohols. Applied microbiology and biotechnology, 91(3), 713–730.
Tian, H., Li, F., Qin, L., Yu, H., & Ma, X. (2014). Quality evaluation of beef seasonings
using gas chromatography-mass spectrometry and electronic nose: Correlation
with sensory attributes and classification according to grade level. Food
Analytical Methods, 1–13.
Vallejos, S., Khatko, V., Calderer, J., Gracia, I., Cané, C., Llobet, E., & Correig, X. (2008).
Micro-machined WO3-based sensors selective to oxidizing gases. Sensors and
Actuators B: Chemical, 132(1), 209–215.
Cheng, H., Qin, Z., Guo, X., Hu, X., & Wu, J. (2013). Geographical origin identification
of propolis using GC-MS and electronic nose combined with principal
component analysis. Food Research International, 51(2), 813–822.
Comini, E., Baratto, C., Concina, I., Faglia, G., Falasconi, M., Ferroni, M., ... Sberveglieri,
G. (2013). Metal oxide nanoscience and nanotechnology for chemical sensors.
Sensors and Actuators B: Chemical, 179, 3–20.
Comuzzo, P., Tat, L., Tonizzo, A., & Battistutta, F. (2006). Yeast derivatives (extracts
and autolysates) in winemaking: Release of volatile compounds and effects on
wine aroma volatility. Food Chemistry, 99(2), 217–230.
Galaz, S., Morales-Quintana, L., Moya-León, M. A., & Herrera, R. (2013). Structural
analysis of the alcohol acyltransferase protein family from Cucumis melo shows
that enzyme activity depends on an essential solvent channel. FEBS Journal, 280
Varpula, A., Novikov, S., Haarahiltunen, A., & Kuivalainen, P. (2011). Transient
characterization techniques for resistive metal-oxide gas sensors. Sensors and
Actuators B: Chemical, 159(1), 12–26.
(
5), 1344–1357.
Vuong, N. M., Kim, D., & Kim, H. (2015). Surface gas sensing kinetics of a WO
nanowire sensor: Part 1—oxidizing gases. Sensors and Actuators B: Chemical, 220,
32–941.
3
García-González, D. L., Tena, N., Aparicio-Ruiz, R., & Aparicio, R. (2014). Sensor
responses to fat food aroma:
A comprehensive study of dry-cured ham
9
typicality. Talanta, 120, 342–348.
Yoshioka, K., & Hashimoto, N. (1981). Ester formation by alcohol acetyltransferase
from brewers yeast. Agricultural and Biological Chemistry, 45(10), 2183–2190.
Zhang, C. Y., Liu, Y. L., Qi, Y. N., Zhang, J. W., Dai, L. H., Lin, X., & Xiao, D. G. (2013).
Increased esters and decreased higher alcohols production by engineered
brewer’s yeast strains. European Food Research and Technology, 236(6),
Inoue, Y., Trevanichi, S., Fukuda, K., Izawa, S., Wakai, Y., & Kimura, A. (1997). Roles of
esterase and alcohol acetyltransferase on production of isoamyl acetate in
Hansenula mrakii. Journal of Agricultural and Food Chemistry, 45(3), 644–649.
Kiyohara, C., Washio, M., Horiuchi, T., Asami, T., Ide, S., Atsumi, T., ... Tada, Y. (2014).
Modifying effect of N-acetyltransferase 2 genotype on the association between
systemic lupus erythematosus and consumption of alcohol and caffeine-rich
beverages. Arthritis Care & Research, 66(7), 1048–1056.
Kohl, D., Heinert, L., Bock, J., Hofmann, T., & Schieberle, P. (2000). Systematic studies
on responses of metal-oxide sensor surfaces to straight chain alkanes, alcohols,
aldehydes, ketones, acids and esters using the SOMMSA approach. Sensors and
Actuators B: Chemical, 70(1), 43–50.
1009–1014.
Zhang, J., Zhang, C., Dai, L., Dong, J., Liu, Y., Guo, X., & Xiao, D. (2012). Effects of
overexpression of the alcohol acetyltransferase-encoding gene ATF1 and
disruption of the esterase-encoding gene IAH1 on the flavour profiles of
Chinese yellow rice wine. International Journal of Food Science & Technology, 47
(
12), 2590–2596.