912 J. Sep. Sci.

2014, 37, 912–919

Tomas Cajka1 Research Article

Marta Vaclavikova1
Zbynek Dzuman1
Lukas Vaclavik1
Jaroslava Ovesna2
Jana Hajslova1
1 Department of Food Analysis
and Nutrition, Faculty of Food
and Biochemical Technology,
Institute of Chemical
Technology, Prague, Czech
Republic
2 Department of Molecular
Biology, Division of Plant
Genetics, Breeding and Product
Quality, Crop Research Institute,
Prague, Czech Republic
Received November 29, 2013
Revised January 27, 2014
Accepted February 2, 2014
Rapid LC–MS-based metabolomics method
to study the Fusarium infection of barley
Ultra high performance liquid chromatography with quadrupole/time-of-flight mass spec-
trometry was applied to evaluate the potential of nontarget metabolomic fingerprinting
in order to distinguish Fusarium-infected and control barley samples. First, the sample
extraction and instrumental conditions were optimized to obtain the broadest possible rep-
resentation of polar/medium-polar compounds occurring in extracts obtained from barley
grain samples. Next, metabolomic fingerprints of extracts obtained from nine barley vari-
eties were acquired under ESI conditions in both positive and negative mode. Each variety of
barley was tested in two variants: artificially infected by Fusarium culmorum at the beginning
of heading and a control group (no infection). In addition, the dynamics of barley infection
development was monitored using this approach. The experimental data were statistically
evaluated by principal component analysis, hierarchical clustering analysis, and orthogonal
partial least-squares discriminant analysis. The differentiation of barley in response to F.
culmorum infection was feasible using this metabolomics-based method. Analysis in posi-
tive mode provided a higher number of molecular features as compared to that performed
under negative mode setting. However, the analysis in negative mode permitted the de-
tection of deoxynivalenol and deoxynivalenol-3-glucoside considered as resistance-indicator
metabolites in barley.

Keywords: Barley / Fusarium head blight / LC–MS / Metabolomics / Mycotoxins

DOI 10.1002/jssc.201301292

 Additional supporting information may be found in the online version of this article
at the publisher’s web-site

1 Introduction fungi can be relatively rich, deoxynivalenol (DON) is typi-

Fusarium head blight (FHB) represents a major destructive
fungal disease of economically important crops of the Trit-
iceae family, such as wheat, barley, rye, oats, or corn [1]. The
infestation of crops with FHB results in the reduction in yield,
deterioration in grade and end-use quality of commodity ce-
reals. FHB can be caused by several members of filamentous
fungi species of the genus Fusarium; nevertheless, F. gramin-
earum and F. culmorum were shown to be the main agents
of this disease [2]. Fusarium fungi are known for their active
and relatively extensive secondary metabolism within which
more or less toxic mycotoxins are produced [3]. Despite the
fact that the pattern of mycotoxins produced by Fusarium
Correspondence: Dr. Tomas Cajka, Institute of Chemical Technol-
ogy, Technicka 3, 166 28 Prague, Czech Republic
E-mail: tomas.cajka@vscht.cz
Fax: +420-233-339-990
Abbreviations: D3G, deoxynivalenol-3-glucoside; DON, de-
oxynivalenol; FHB, Fusarium head blight; HCA, hierarchi-
cal clustering analysis; OPLS-DA, orthogonal partial least-
squares discriminant analysis; PCA, principal component
analysis; RT, retention time; UHPLC, ultra high performance
liquid chromatography
cally the most abundant toxin [4]. With regard to the health
risks associated with the occurrence of mycotoxins in foods
and feeds, the concentrations of DON and some other Fusar-
ium mycotoxins (zearalenon and fumonisins) are currently
regulated by various national and international authorities
[5, 6].
Since the complete exclusion of mycotoxins from cere-
als and cereal-based products is not possible, there is a con-
tinuous need for procedures that allow the control and re-
duction of FHB and minimization of mycotoxin occurrence
in food chain. Some agricultural practices—such as applica-
tion of fungicides, changing of preceding crops, or tillage—
can to some extent decrease the infestation of crops with
mycotoxins [7]. Nevertheless, the resistance of particular cul-
tivars to FHB is often the determining factor [8], and thus
the reduction of mycotoxin production through breeding of
varieties resistant to FHB disease is nowadays a challeng-
ing task. In general, it is assumed that the contamination
of crops with DON and other trichothecenes correlates with
incidence of FHB and the concentration of DON is often con-
sidered as a marker of the infection [9]. On the other hand,
Colour Online: See the article online to view all Figures in colour
online only.


C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.jss-journal.com
J. Sep. Sci. 2014, 37, 912–919
some studies reported on the high trichothecene amounts in
FHB-resistant wheat varieties with low apparent infection
severity and indicated that the above correlation may not be
always valid [10,11]. More than 100 quantitative trait loci have
been found for FHB resistance in wheat and barley indicating
the multiple mechanisms of the resistance. However, the re-
sistance mechanisms have not yet been fully understood [12].
Lemmens et al. have demonstrated that the quantitative trait
loci, previously reported to be partly responsible for the resis-
tance to FHB, is linked to the ability of plants to metabolize
DON into deoxynivalenol-3-glucoside (D3G) [13]. The forma-
tion of DON–glucose conjugates during phase II of plant
metabolism has a central role in protection of plant from
toxic effects of DON [14].
In the past decade, the discipline of metabolomics has
emerged and found its application in many research fields,
including plant science [15–17]. Metabolomics is a holistic
study of low-molecular-weight compounds (typically <1500
Da) in complex biological systems under a given set of condi-
tions [18]. The data obtained during the metabolomics exper-
iment can provide a deeper insight into biological processes
and support the discovery of various biomarkers [17, 19].
Metabolomics-based studies are facilitated by advanced an-
alytical platforms capable of the analysis of numerous chemi-
cally diverse metabolites at a large range of concentrations. It
is noteworthy that none of the current analytical platforms
is able to detect all of thousands of metabolites typically
present in biological samples. Besides NMR spectroscopy,
GC–MS, and CE–MS, LC–MS enables the most comprehen-
sive metabolite coverage achievable to date [20–25].
Regarding LC, the introduction of ultra high performance
liquid chromatography (UHPLC) operating at very high pres-
sures and using sub-2 ␮m packing columns has allowed a
substantial decrease in analysis time and increase in peak
capacity, sensitivity, as well as reproducibility as compared
to conventional HPLC. Considering these features for non-
target metabolomics of crude plant extracts, the popularity
of coupling the UHPLC with high-resolution MS, especially
with TOF or quadrupole TOF instruments, is steadily grow-
ing [20, 23, 26–28].
Since highly complex data matrices are generated by these
fingerprint/profiling techniques that require further process-
ing, powerful chemometric tools are needed to fully utilize
this comprehensive information [28].
Until now, the metabolomics studies dealing with FHB
have mainly been focused on the selection and identification
of resistance-related metabolites, which could help to eluci-
date the mechanisms behind the plant resistance or could be
used as biomarkers for screening of barley genotypes resis-
tant to FHB. In most cases, the HPLC–MS technique was
used for this purpose [29–31]. The present methodological
study based on nontarget approach describes the develop-
ment and optimization of a UHPLC–Q-TOF-MS procedure
for the rapid metabolomic fingerprinting of infected and con-
trol barley samples followed by multivariate data analysis of
the acquired datasets. To support the results of this nontarget

C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Liquid Chromatography 913
MS-based metabolomics approach we conducted target LC–
MS/MS analysis of DON and D3G, both known as resistance-
indicator metabolites in barley.
2 Materials and methods
2.1 Barley samples
The tests were conducted in field conditions at the Crop Re-
search Institute in Prague–Ruzyně. Each barley genotype was
sown in a 1 m2 plot at a sowing rate of 450 seeds/m. Within
the first set of experiments the following barley varieties were
examined: Aktiv, Gladys, Henrike, Lilly, Radegast, Sebastian,
Signora, Sladar, and Tolar. Each variety was grown and tested
in artificially infected and control variants (sample set no. 1).
Artificially infected samples were inoculated at the beginning
of heading using an isolate mixture from F. culmorum iso-
lates. Spraying of conidial suspension (0.8 × 107 per mL [32])
was applied at once, uniformly with a 1 L hand-sprayer. Both
control and infected barley samples were collected 11 days
after inoculation. In the second phase of the study, the barley
variety Pedant was used for additional experiments. Sam-
ples were infected in the same way as in the first study. A
corresponding control group of samples (no infection) was
collected along with infected samples at days 0, 1, 4, 7, 14, 21,
and 28 (sample set no. 2). All samples were stored at –20⬚C
prior to analysis.
2.2 Chemicals
Methanol and acetonitrile were obtained from Merck (Darm-
stadt, Germany). Deionized water (18 MOhm·cm) was pro-
duced by a Milli-Q system (Millipore; Bedford, MA, USA).
Ammonium acetate (99.99%) and formic acid (95%) for
preparation of LC mobile phase were obtained from Sigma–
Aldrich (Steinheim, Germany). Analytical standards of de-
oxynivalenol, DON (solid compound, purity 99.4 ± 0.6%) and
deoxynivalenol-3-␤-D-glucoside, D3G (solution in acetonitrile
50 ␮g/mL, purity 96.0 ± 4.0%) were purchased from Romer
Labs (Tulln, Austria). A composite standard mixture of these
two mycotoxins was prepared in acetonitrile at a concentra-
tion of 5 ␮g/mL. All standard solutions were stored in the
freezer at −20⬚C.
2.3 Sample preparation: Nontarget metabolomics
approach
An amount of 1 g of barley was weighed into a 50 mL PP cen-
trifugation tube, followed by the addition of 10 mL of a mix-
ture of methanol/water (50:50, v/v). After brief shaking, the
content was homogenized using an Ultraturrax at 5000 rpm
for 3 min. The tube was finally centrifuged (10 000 rpm)
for 10 min. An aliquot of the extract was filtered through a
0.22 ␮m PTFE filter prior to UHPLC–Q-TOF-MS analysis.
www.jss-journal.com
914 T. Cajka et al.
During the sample preparation also solvent blanks were pre-
pared for analysis consisting of all the steps mentioned above
except for the addition of sample.
2.4 Sample preparation: Target analysis of
mycotoxins (DON and D3G)
An amount of 1 g of barley was weighed into a 50 mL PP
centrifugation tube, followed by the addition of 10 mL of a
mixture of acetonitrile/water (84:16, v/v). After brief shaking,
the content was homogenized using an Ultraturrax macera-
tor at 5000 rpm for 3 min. The tube was finally centrifuged
(10 000 rpm) for 10 min. An aliquot of the extract was 100-
times diluted by a mixture of acetonitrile/water (84:16, v/v).
For quantitation of samples the matrix-matched calibration
standards in the concentration range from 1 to 1000 ng/mL
were prepared using blank barley sample, which was pro-
cessed by the identical sample preparation procedure. Diluted
extracts were subsequently passed through a 0.22 ␮m PTFE
filter prior to UHPLC–MS/MS analysis.
2.5 Nontarget metabolomics UHPLC–Q-TOF-MS
analysis
The UHPLC–Q-TOF-MS analyses were performed using an
Acquity Ultra-Performance LC system (Waters; Milford, MA,
USA) equipped with a 10 ␮L sample loop and an Acquity
UPLC BEH C18 column (50 × 2.1 mm id, 1.7 ␮m particle size;
Waters) maintained at 40⬚C. The mobile phases consisted of
(A) 0.1% formic acid in Milli-Q water and (B) 0.1% formic acid
in methanol. The gradient started at 95% (A), linearly changed
to 100% (B) within 5 min. The mobile phase composition was
held at 100% (B) for 1.5 min and subsequently changed to
95% (A) for 2 min to allow reconditioning of the column. A
constant flow of 0.5 mL/min was used within the analysis. A
sample volume of 2 ␮L with the partial loop injection mode
was used. The sample temperature was maintained at 4⬚C.
The Acquity UHPLC system was connected to an or-
thogonal accelerated high-resolution quadrupole/TOF spec-
trometer XEVO (Waters) operated in either ESI(+) or ESI(–)
mode. The parameters of the ESI(+) mode were as follows:
capillary voltage: +3.2 kV; cone voltage: +30 V; source tem-
perature: 120⬚C; desolvation gas temperature: 450⬚C. Nitro-
gen was used as a desolvation and cone gas at a flow rate
of 650 and 50 L/h, respectively. The parameters of ESI(–)
mode were following: capillary voltage: −2.9 kV; cone voltage
−30 V. Temperature of the ion source and desolvation gas
and gas flow rates were the same as in ESI(+) mode. In
each polarity, two functions were set up. The first function
collected the data without collision energy, while the second
function acquired the data with a gradient of collision energy
(30–50 eV). The acquisition time of each function was 0.1 s.
The instrument was tuned using a sodium formate solution
(mass resolving power >7000 fwhm). A leucine–enkephalin
solution was used as a lock mass to correct small mass drifts

C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
J. Sep. Sci. 2014, 37, 912–919
during the measurement. Raw mass spectra were acquired
in the m/z range 50–1000.
The MassLynx 4.1 software (Waters) was used for data
acquisition. The MarkerLynx software (Waters) was used for
data mining and processing with the following parameters:
initial retention time: 0.0 min; final retention time: 6.0 min;
low mass: 50; high mass: 1000; ion extraction chromatogram
window: 0.05 Da; marker intensity threshold: 100 counts;
mass window: 0.05 Da; retention time window: 0.2 min;
deisotope data: yes.
2.6 Target UHPLC–MS/MS analysis of mycotoxins
(DON and D3G)
An Acquity UPLC system (Waters) coupled to a 5500 QTRAP
mass spectrometer (AB SCIEX; Toronto, ON, Canada) was
used for target mycotoxin analysis as described earlier [33].
The Analyst 1.5 software (AB SCIEX) was used for data acqui-
sition and processing. The experimental conditions, perfor-
mance characteristics of the method, and LC–MS/MS chro-
matogram are shown in Supporting Information Tables S1
and S2 and Fig. S1.
2.7 Chemometric analysis
The software SIMCA (v. 13.0, 2011, Umetrics, Umea, Swe-
den) was used for multivariate data analysis. In the first stage
of the data processing, the raw data in the form of absolute
peak intensities (heights) were preprocessed using the con-
stant row sum, that is, each variable was divided by the sum of
all variables for each sample. This procedure transformed all
the data to a uniform range of variability. Subsequently, log
transformation and Pareto scaling (square root of the stan-
dard deviation is used as the scaling factor) were applied for
particular “normalized” data sets prior to multivariate data
analysis.
3 Results and discussion
3.1 Optimization of conditions for nontarget
metabolomic UHPLC–Q-TOF-MS analysis
Within the first phase of this methodological study, we fo-
cused on the sample preparation for the analysis of the
broadest possible spectrum of metabolites/small molecules
isolated from the barley tissue. During the course of our ex-
periments, 1 g of barley grain samples (a pool of control and
infected barley of variety Sebastian) was homogenized using
an Ultraturrax with 10 mL of different solvents or mixtures
of solvents with decreasing polarity. Specifically, (i) deionized
water, (ii) a mixture of methanol/water (50:50, v/v), and (iii) a
mixture of acetonitrile/water (84:16, v/v) were used for extrac-
tion, the last one is frequently being used for target analysis
of mycotoxins from various matrices [34–36]. After the cen-
trifugation step, the obtained extracts were analyzed using
www.jss-journal.com
J. Sep. Sci. 2014, 37, 912–919
UHPLC–Q-TOF-MS in both positive and negative ESI mode.
As Supporting Information Fig. S2 documents, the UHPLC–
Q-TOF-MS fingerprints of metabolites/small molecules dif-
fered significantly depending on the extraction solvent used.
Using ESI(+) mode, the number of features obtained
during the data mining was 1385, 1529, and 1635 for deion-
ized water, a mixture of methanol/water (50:50, v/v), and
a mixture of acetonitrile/water (84:16, v/v), respectively, as
extraction solvents. The number of obtained features in
ESI(–) was 630, 536, and 600 for deionized water, a mix-
ture of methanol/water (50:50, v/v), and a mixture of ace-
tonitrile/water (84:16, v/v), respectively. Considering both
ESI(+) and ESI(–) mode, the highest number of features
was obtained using a mixture of acetonitrile/water (84:16,
v/v) (100%) followed by a mixture of methanol/water (50:50,
v/v) (92%) and deionized water (90%). As a compromise
in terms of sample component representation, a mixture
of methanol/water (50:50, v/v) was used for the subsequent
experiments. This extraction mixture permitted isolation of
both highly polar, early-eluting compounds as well as less
polar, late-eluting metabolites. This was, however, not the
case of the extraction using deionized water that provided the
lower extraction efficiency of less polar compounds. On the
other hand, extraction using a mixture of acetonitrile/water
(84:16, v/v) enhanced the extraction of less polar, late-eluting
metabolites, while more polar compounds were detected at
low extent. Overall, the vast majority [990 in ESI(+) and 430
in ESI(–)] of features were detected regardless the type of
extraction procedure used (Fig. 1).
Along with the sample preparation we also focused on the
separation of metabolome components. Since the separation
of metabolites differing in their polarity was considered, an
analytical column with reversed stationary phase containing
end-capped C18 sorbent was used to enhance the retention of
highly polar components present in the barley metabolome.
The use of a stationary phase with 1.7 ␮m particles signifi-
cantly increased the chromatographic resolution and enabled
relatively rapid analysis with injection-to-injection cycle of
8.5 min with the base width of the detected peaks between
3 and 8 s.
To explore the possible retention time (RT) fluctuation,
RTs of three peaks in ESI(+) [RT 0.77 min, m/z 120.1; RT
3.64 min, m/z 274.3; and RT 5.57 min, m/z 481.4] and in
ESI(–) [RT 0.76 min, m/z 164.1; RT 3.52 min, m/z 329.2;
and RT 5.05 min, m/z 279.2] were monitored during a mea-
surement sequence. Based on these data, typical peak RT

C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Liquid Chromatography 915
Figure 1. Venn diagrams illustrating
shared and uniquely detected fea-
tures in the barley extracts prepared
under the different extraction proce-
dures and analyzed using (A) UHPLC–
ESI(+)-Q-TOF-MS and (B) UHPLC–
ESI(–)-Q-TOF-MS.
variability was found to be below 0.5% (RSD) that can be
considered as an acceptable value. To obtain as low mass
errors as possible, regular external calibration of the Q-TOF-
MS instrument was performed. Additionally, a continuous
correction of small mass drifts was carried out with the use
of mass locking compound delivered to the instrument via
a second ESI sprayer. Under these conditions, the achieved
mass accuracy was below 10 ppm.
3.2 Chemometric analysis and data interpretation
For chemometric analysis, 2493 and 882 features obtained
in ESI(+) and ESI(–), respectively, were used to study the
group of control (n = 9) and infected (n = 9) barley samples
(sample set no. 1). In the case of investigation, the infection
progress of barley variety Pedant (n = 13; sample set no. 2),
in total 2972 and 1242 features in ESI(+) and ESI(–), respec-
tively, were generated by the MarkerLynx software. A slightly
higher number of detected features in the latter case can be
explained by the difference in metabolite composition within
the barley varieties and the collection at different phases of
infection. We should note that the features/peaks extracted
from the UHPLC–MS fingerprints could not be generally
considered as individual metabolites because of the signal re-
dundancy [37] due to formation of multiple signals (various
adduct or fragment ions) for a single metabolite. This redun-
dancy was, to some extent, decreased in the data-mining step
by automatic removal of the peaks of isotope ions (i.e. only
signals with the lowest m/z value within an isotope pattern
were used).
In the first phase, unsupervised pattern recognition tech-
niques were used to observe a potential clustering behav-
ior. For this purpose, principal component analysis (PCA)
and hierarchical clustering analysis (HCA) were employed.
In general, PCA reduces the data dimensionality allowing
their visualization retaining as much as possible from in-
formation present in the original data. On the other hand, in
HCA samples are grouped on the basis of similarities without
taking into account the information about the class member-
ship [28].
Using PCA, both ESI(+) and ESI(–) data of the sample set
no. 1 indicated clustering related to the infection of samples
(Fig. 2A, B). The first two principal components explained for
58 and 56% of the ESI(+) and ESI(–) data variance, respec-
tively. Specifically, the control (noninfected) samples formed
www.jss-journal.com
916 T. Cajka et al. J. Sep. Sci. 2014, 37, 912–919

Figure 2. A two-score plot of PCA (A, B) and HCA (C, D) graphs of UHPLC–ESI(+)-Q-TOF-MS (A, C) and UHPLC–ESI(–)-Q-TOF-MS (B, D)
markers focused on examination of control and infected barley varieties (sample set no. 1). Concentration of DON and D3G in infected
barley varieties is shown in (E).

discrete clusters (with the exception of variety Lilly), while the
infected barley samples showed a higher dispersion in PCA
graph of the first two principal components. This observa-
tion was further confirmed by HCA, where the barley sam-
ples were unambiguously split into two groups (control and
infected) (Fig. 2C, D). In addition, further investigation of
the HCA graph showed that the infected barley samples were
split into two subgroups, one of them containing the bar-
ley samples (variety Sebastian, Gladys, Tolar, and Aktiv) with
the highest concentrations of mycotoxins (DON, D3G) de-
termined by target UHPLC–MS/MS approach (Fig. 2E). The
presence of DON and D3G as resistance-indicator metabo-
lites in barley has recently been reported by Bollina et al., who
used nontarget metabolomic approach to identify metabolites
of this type in barley [30, 31].
The same chemometric approach was also applied to
monitor the dynamics of metabolites in infected barley va-
riety Pedant (Supporting Information Fig. S3) (sample set
no. 2). Using ESI(+) data, PCA plot showed three groups: (i)
control samples at days 0, 1, 4, 7, 14, and infected samples


C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.jss-journal.com
J. Sep. Sci. 2014, 37, 912–919
Figure 3. A two-score plot of OPLS-DA of (A) UHPLC–ESI(+)-Q-TOF-MS and (B) UHPLC–ESI(–)-Q-TOF-MS markers focused on examination
of control and infected barley varieties (sample set no. 1).
at days 1, 4, 7; (ii) control samples at days 21 and 28, and
(iii) infected samples at days 14, 21, and 28. In the case of
ESI(–), this grouping was less obvious. The first two princi-
pal components explained for 70 and 73% of the ESI(+) and
ESI(–) data variance, respectively. Using HCA we observed
two main clusters. The first one contained two subclusters: (i)
control samples at days 21 and 28, and (ii) infected samples at
days 14, 21, and 28. The second cluster contained (i) control
samples at days 0, 1, 4, 7, 14, and (ii) infected samples at days
1, 4, 7. While clustering of control samples at days 0, 1, 4,
7, 14, and infected samples at days 1, 4, 7 can be explained
by the low extent of fungal activity since concentrations of
mycotoxins were <LOQ of the method (“mycotoxins free”),
in the case of subcluster containing infected samples at days
14, 21, 28, high concentrations of both DON and D3G were
observed. On the other hand, the control samples at days 21
and 28 in the subcluster contained no traces of mycotoxins,
but their metabolomic fingerprints largely differed from the
“mycotoxins free” control samples at days 0, 1, 4, 7, 14.
Additionally, for sample set no. 1 orthogonal partial least-
squares discriminant analysis (OPLS-DA) was subsequently
constructed to identify and reveal the differential metabolites
in response to the infection of barley samples [control group
(n = 9) and infected group (n = 9)]. Using OPLS-DA, fur-
ther improvement of separation between control and infected
groups of barley samples was achieved for both ESI(+) and
ESI(–) data (Fig. 3) as compared to PCA clustering (Fig. 2).
Also, within the group of infected barley, samples with the
highest concentrations of target mycotoxins (DON, D3G) de-
termined by the UHPLC–MS/MS approach were separated
in the OPLS-DA plot from those samples with low concentra-
tions of mycotoxins.
3.3 Marker identification
The use of variable importance of the projection of OPLS-DA
allowed us to obtain the most important features. Some of
these features were subsequently tentatively identified on the

C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Liquid Chromatography 917
basis of (i) accurate single MS mass of parent ion; (ii) isotopic
profile of parent ion; (iii) accurate MS/MS fragment ions
(measured in so-called MSE mode) followed by the database
search [(Mass Bank (http://www.massbank.jp/?lang=en),
METLIN (http://metlin.scripps.edu), Plant Metabolic Net-
work (http://pmn.plantcyc.org)], or comparison with recently
published studies [30, 31].
For instance, we were able to identify jasmonic acid (RT
3.57 min; m/z 211.1(+), m/z 209.1(–)), which is biosynthe-
sized from linolenic acid by the octadecanoid pathway. This
metabolite regulates plant responses to abiotic and biotic
stresses as well as plant growth and development [38].
Dihydro-7-hydroxymyoporone (RT 3.2 min; m/z 267.2(–))
represents another metabolite identified in infected barley.
This compound belongs to stress-induced metabolites with a
sesquiterpene structure [39]. Kaempferol 3-O-glucoside 7-O-
rhamnoside (RT 1.7 min; m/z 593.1(–)), another compound
detected in infected barley, has been suggested as a resistance-
related metabolite [31].
4-Methoxycinnamic acid (RT 3.0 min; m/z 177.1(–)) rep-
resents a compound formed in plants under stress condi-
tions [40]. We also observed increased abundance of 7,4 -
dihydroxyflavan (RT 0.7 min; m/z 241.1(–)) and guanosine
(RT 0.5 min; m/z 282.1(–)) in infected barley.
On the other hand, malonic acid (RT 0.3 min; m/z
103.0(–)), (iso)leucine (RT 0.5 min; m/z 130.1(–)), proline
(RT 1.71 min; m/z 116.1(+)), p-coumaric acid (RT 0.3 min;
m/z 163.0(–)), apigenin (RT 3.0 min; m/z 269.0(–)), secoiso-
lariciresinol (RT 4.6 min; m/z 363.2(+)), and kaempferol 3-
O-␣-rhamnoside (RT 2.1 min; m/z 431.1(–)) were observed
at high intensities in control barley samples as compared to
infected ones.
A deeper examination of features showed that the
target mycotoxins were also detected by this nontarget
metabolomics approach in ESI(–): DON as [M+HCOO]− (RT
1.18 min; m/z 341.1) and D3G as [M–H]− and [M+HCOO]−
(RT 1.24 min; m/z 457.2 and m/z 503.2). Additional injec-
tion of matrix-matched standards of DON and D3G con-
firmed their presence. However, their LOQs were ten- and
www.jss-journal.com
918 T. Cajka et al.
twofold higher, respectively, as compared to target UHPLC–
MS/MS method, for which the extracts were 100-times di-
luted prior to analysis. Thus, the UHPLC–Q-TOF-MS tech-
nique would allow the quantification of highly contaminated
samples only.
4 Concluding remarks
An analytical procedure employing a mixture of
methanol/water (50:50, v/v) as the extraction solvent
was developed to isolate polar/medium-polar barley metabo-
lites followed by rapid examination of the extracts by
UHPLC–Q-TOF-MS (8.5 min run) in positive and negative
ESI mode. In general, analysis in ESI(+) provided approx.
2.8 times more features as compared to ESI(–). The use
of advanced chemometric tools allowed a differentiation of
control and Fusarium infected barley as well as monitoring
of the dynamics of barley infection based on acquired
metabolomic fingerprints.
The increase of the intensity of several plant
stress-related metabolites (jasmonic acid, dihydro-
7-hydroxymyoporone, kaempferol 3-O-glucoside 7-O-
rhamnoside, 4-methoxycinnamic acid) in infected samples
correlated positively with increasing concentrations of
DON and D3G as determined by the target UHPLC–
MS/MS method. These two mycotoxins represent important
resistance-indicator metabolites in barley. Although their
presence was confirmed by the nontarget UHPLC–Q-TOF-
MS method in ESI(–) mode, only highly contaminated
samples could be quantified as compared to the target
UHPLC–MS/MS method.
The financial support of the Ministry of Education, Youth
and Sports of the Czech Republic (MSM 6046137305) and the
Ministry of Agriculture of the Czech Republic (project NAZV
QI111B044) is acknowledged.
The authors have declared no conflict of interest.
5 References
[1] Simsek, S., Burgess, K., Whitney, K. L., Gu, Y., Qian, S.
Y., Food Control 2012, 26, 287–292.
[2] Osborne, L. E., Stein, J. M., Int. J. Food Microbiol. 2007,
119, 103–108.
[3] Smedsgaard, J., Nielsen, J., J. Exp. Bot. 2005, 56,
273–286.
[4] Faifer, G. C., De Miguel, M. S., Godoy, H. M., Myco-
pathologia 1990, 109, 165–170.
[5] Capriotti, A. L., Caruso, G., Cavaliere, C., Foglia, P., Sam-
peri, R., Lagana, A., Mass Spectrom. Rev. 2012, 31,
466–503.
[6] Krska, R., Schubert-Ullrich, P., Molinelli, A., Sulyok, M.,
Macdonald, S., Crews, C., Food Addit. Contam. 2008, 25,
152–163.

C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
J. Sep. Sci. 2014, 37, 912–919
[7] Champeil, A., Dore, T., Fourbet, J. F., Plant Sci. 2004, 166,
1389–1415.
[8] Boutigny, A.-L., Richard-Forget, F., Barreau, C., Eur. J.
Plant. Pathol. 2008, 121, 411–423.
[9] Moss, M. O., Mycologist 2002, 16, 158–161.
[10] Mesterhazy, A., Nartok, T., Mirocha, C. G., Komoroczy,
R., Plant Breeding 1999, 118, 97–110.
[11] Snijders, C. H. A., Toxicol. Lett. 2004, 153, 37–46.
[12] Liu, S., Hall, M. D., Griffey, C. A., McKendry, A. L., Crop
Sci. 2009, 49, 1955–1968.
[13] Lemmens, M., Scholz, U., Berthiller, F., Dall’Asta, C., Kout-
nik, A., Schuhmacher, R., Mol. Plant Microbe Interact.
2005, 18, 1318–1324.
[14] Berthiller, F., Crews, C., Dall’Asta, C., De Saeger, S.,
Haesaert, G., Karlovsky, P., Oswald, I. P., Seefelder, W.,
Speijers, G., Stroka, J., Mol. Nutr. Food Res. 2013, 57,
165–186.
[15] Kaddurah-Daouk, R., Krishnan, K. R., Neuropsychophar-
macology 2008, 34, 173–186.
[16] Wishart, D. S., Drugs R D 2008, 9, 307–322.
[17] Wolfender, J.-L., Rudaz, S., Choi, Y. H., Kim, H. K., Curr.
Med. Chem. 2013, 20, 1056–1090.
[18] Nicholson, J. K., Lindon, J. C., Nature 2008, 455,
1054–1056.
[19] Eugster, P. J., Guillarme, D., Rudaz, S., Veuthey, J.-L.,
Carrupt, P.-A., Wolfender, J.-L., J. AOAC Int. 2011, 94,
51–70.
[20] Zhang, A., Sun, H., Wang, P., Han, Y., Wang, X.J., Analyst
2012, 137, 293–300.
[21] Zhou, B., Xiao, J. F., Tuli, L., Ressom, H. W., Mol. Biosyst.
2012, 8, 470–481.
[22] Smolinska, A., Blanchet, L., Buydens, L. M. C., Wijmenga,
S. S., Anal. Chim. Acta 2012, 750, 82–97.
[23] Wang, X., Sun, H., Zhang, A., Wang, P., Han, Y., J. Sep.
Sci. 2011, 34, 3451–3459.
[24] Balmer, D., Flors, V., Glauser, G., Mauch-Mani, B., Front.
Plant Sci. 2013, 4, 82.
[25] Ibanez, C., Simo, C., Garcia-Canas, V., Gomez-Martinez,
A., Ferragut, J. A., Electrophoresis 2012, 33, 2328–2336.
[26] Patti, G.J., J. Sep. Sci. 2011, 34, 3460–3469.
[27] Issaq, H. J., Abbot, E., Veenstra, T. D., J. Sep. Sci. 2011,
31, 1936–1947.
[28] Berrueta, L. A., Alonso-Salces, R. M., Herberger, K., J.
Chromatogr. A 2007, 1158, 196–214.
[29] Kumaraswamy, G. K., Bollina, V., Kushalappa, A. C.,
Choo, T. M., Dion, Y., Rioux, S., Mamer, O., Faubert, D.,
Eur. J. Plant Pathol. 2011, 130, 29–43.
[30] Bollina, V., Kumaraswamy, G. K., Kushalappa, A. C.,
Choo, T. M., Dion, Y., Rioux, S., Faubert, D., Hamze-
hzarghani, H., Mol. Plant Pathol. 2010, 11, 769–782.
[31] Bollina, V., Kushalappa, A. C., Choo, T. M., Dion, Y., Rioux,
S., Plant. Mol. Biol. 2011, 77, 355–370.
[32] Sip, V., Chrpova, J., Stockova, L., Czech J. Genet. Plant
Breed. 2011, 47, 131–139.
[33] Dzuman, Z., Vaclavikova, M., Polisenska, I., Veprikova,
Z., Fenclova, M., Zachariasova, M., Hajslova, J., Anal.
Bioanal. Chem. 2014, 406, 505–514.
www.jss-journal.com
J. Sep. Sci. 2014, 37, 912–919
[34] Mol, H. G. J., Plaza-Bolanos, P., Zomer, P., de Rijk, T. C.,
Stolker, A. A. M., Mulder, P. P. J., Anal. Chem. 2008, 80,
9450–9459.
[35] Sulyok, M., Krska, R., Schuhmacher, R., Food Chem.
2010, 119, 408–416.
[36] Varga, E., Glauner, T., Berthiller, F., Krska, R., Schuh-
macher, R., Sulyok, M., Anal. Bioanal. Chem. 2013, 405,
5087–5104.

C 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Liquid Chromatography 919
[37] Werner, E., Heilier, J. F., Ducruix, C., Ezan, E. R., Junot,
C., Tabet, J. C., J. Chromatogr. B 2008, 871, 143–163.
[38] Delker, C., Stenzel, I., Hause, B., Miersch, O., Feussner,
I., Wasternack, C., Plant Biol. 2006, 8, 297–306.
[39] Keeler, R. F., Tu, A. T., Handbook of Natural Toxins, Vol-
ume 1: Plant and Fungal Compounds, Marcel Dekker,
New York, 1991.
[40] Korkina, L. G., Cell. Mol. Biol. 2007, 53, 15–25.
www.jss-journal.com