Evaluation of liquid chromatographic behavior of lumazinic derivatives, from α-dicarbonyl compounds, in different c18 columns: Application to wine samples using a fused-core column and fluorescence detection

Article abstract of DOI:10.1021/jf404180t

Several C18 columns, packed with totally porous particles of different sizes and shell thicknesses, have been compared for simultaneous determination of α-dicarbonyl compounds, previous derivatization to lumazinic derivatives. Chromatographic conditions for the separation have been optimized for each column, and chromatographic parameters have been calculated and exhaustively compared. A core-shell C18 column provided the best results, and a HPLC method with fluorimetric detection has been proposed. The developed method has been validated in terms of linearity, precision, and sensitivity. Detection and quantification limits obtained were comprised between 0.02 and 0.30 and 0.07 and 1.0 ng mL^-1, respectively, while RSD values obtained were lower than 6% and 5% in intraday and interday repeatability studies, respectively. The method has been applied to analysis of the α-dicarbonyl compounds in different types of wines. The higher levels of the total α-dicarbonyl compounds were found in sweet wines and the lower levels in white wines.

Full text of DOI:10.1021/jf404180t

Article
pubs.acs.org/JAFC
Evaluation of Liquid Chromatographic Behavior of Lumazinic
Derivatives, from α‑Dicarbonyl Compounds, in Different C18
Columns: Application to Wine Samples Using a Fused-Core Column
and Fluorescence Detection
María del Carmen Hurtado-San
́ ́ ́
chez, Anunciacion Espinosa-Mansilla, María Isabel Rodríguez-Caceres,
and Isabel Duran-Meras
́
́
*
Department of Analytical Chemistry, University of Extremadura, 06006 Badajoz, Spain
ABSTRACT: Several C18 columns, packed with totally porous particles of different sizes and shell thicknesses, have been
compared for simultaneous determination of α-dicarbonyl compounds, previous derivatization to lumazinic derivatives.
Chromatographic conditions for the separation have been optimized for each column, and chromatographic parameters have
been calculated and exhaustively compared. A core−shell C18 column provided the best results, and a HPLC method with
fluorimetric detection has been proposed. The developed method has been validated in terms of linearity, precision, and
−1
sensitivity. Detection and quantification limits obtained were comprised between 0.02 and 0.30 and 0.07 and 1.0 ng mL ,
respectively, while RSD values obtained were lower than 6% and 5% in intraday and interday repeatability studies, respectively.
The method has been applied to analysis of the α-dicarbonyl compounds in different types of wines. The higher levels of the total
α-dicarbonyl compounds were found in sweet wines and the lower levels in white wines.
KEYWORDS: α-dicarbonyl compounds, lumazinic derivatives, C18 columns, fluorescence detection, wine samples
1
2
1
. INTRODUCTION
saccharide autoxidation. To date, to our knowledge, no data
about the quantities of these compounds in wines have been
reported.
α-Dicarbonyl compounds are reactive intermediates formed in
physiological systems by lipid peroxidation of polyunsaturated
fatty acids and biological systems through the Maillard
1
,2
A number of methods have been developed for determi-
nation of α-dicarbonyl compounds in several matrices such as
biological samples and processed and fermented foods, and the
prelabeled HPLC method is the most common. The more
frequent derivatization reaction is formation of quinoxaline
derivatives, and different diaminobenzenes have been proposed
as derivatizing reagents: 2,4-dinitrophenylhydrazine, 1,2-
diaminobenzene, and 1,2-diamino-4,5-dimethoxybenzene.
Other reagents used have been 6-hydroxy-1,2,3-triaminopir-
3
,4
reaction by degradation of sugars. In processed foods, these
compounds are formed from carbohydrates during thermal
processing in the course of the Maillard reaction, and they are
also important compounds present in food products obtained
by fermentation processes, such as wine. In this matrix, α-
dicarbonyl compounds are formed as a consequence of the
malolactic fermentation, a process that can occur after or
simultaneously with alcoholic fermentation. During malolactic
fermentation decarboxylation of L-malic acid takes place, and in
consequence, acidity levels are reduced and several compounds
13
14
15,16
1
7,18
19
imidine,
2,3-diaminonaphthalene, or 5,6-diamino-2,4-
20
hydroxypyrimidine.
5
,6
For analysis of these compounds in wine samples, methods
such as gas chromatography with a mass-selective detector
GC-MS) or a thermoionic detector (GC-NPD) have been
proposed as alternative techniques. However, the methodology
more frequently used is based on formation of quinoxaline
derivatives and analysis by RP-LC with photometric detec-
All HPLC methods employ C18 columns with
particles of 5 μm for chromatographic separation of the
corresponding derivatives.
In this point, development of packed columns with smaller
particle diameter to work in reverse phase could improve
analysis in terms of separation efficiency and also in terms of
reduction of the analysis time, which is an important factor to
related with the aroma and flavors are formed. Among the
identified formed compounds, dicarbonyl compounds with a
short chain such as diacetyl (DIA), glyoxal (Gly), methylglioxal
21
22
(
(
MGly), 2,3-pentanedione (2,3-Pen), and phenylglyoxal
5
−7
(PhGly) are included.
These compounds are present in all
8
types of wines, with levels higher in red wines, and these
concentrations are increased during age. They play an
important role in wines because they have a great influence
on flavor and sensory characteristics. Specifically, DIA and 2,3-
Pen are the most important compounds in the aroma of wines,
22−24
9
tion.
25
10
and DIA is responsible for the buttery flavor of certain wines.
In addition, α-dicarbonyl compounds present reactivity with
other components, which contribute to the loss of nutritional
quality and increment production of toxic compounds in the
1
1
wine.
Received: September 17, 2013
Revised: December 4, 2013
Accepted: December 11, 2013
Published: December 11, 2013
Glucosones, such as glucosone (GS) and 3-deoxyglucosone
(
3-DG), are α-dicarbonyl compounds with a C-6 backbone,
which can be also formed from hexoses through mono-
©
2013 American Chemical Society
97
dx.doi.org/10.1021/jf404180t | J. Agric. Food Chem. 2014, 62, 97−106

Journal of Agricultural and Food Chemistry
Article
take into account. In this way, columns packed with core−shell
and totally porous sub-2 μm particles have been most
ammonia (Panreac). Phosphoric acid (85%) was provided by Scharlau
(
(
Barcelona, Spain), acetic acid (95.5%) by Romil Chemical LTD
Cambridge, England), and formic acid (98%) by Fluka (Seelze,
26
frequently employed in qualitative and quantitative analysis.
The efficiency of a packed column is described by the
Germany). Different solutions of pH between 2.9 and 4.0 were
prepared by dilution of an adequate volume of concentrated acid in
ultrapure water. Methanol (MeOH), HPLC grade, was purchased
from Panreac. Ultrapure water was obtained from a Milli-Q water
system (Millipore S.A.S., Molsheim, France). Wines analyzed were
acquired from local markets and kept at 4 °C, avoiding exposure to
27
Deemter’s plate height model, which explains the variation
of this parameter with the linear velocity. The key concept of
porous shell (or fused-core) particles is to increase both
efficiency and separation velocity by reducing the mass transfer
28
(
C term in the van Deemter curve). This is possible because
́
direct light. Specifically, red wines (Monasterio de Tentudia, Ribera
del Duero, and Merlot), white wines (Rioja Comportillo, Vin
Canchal, and Manzanilla), red sweet wines (Vin Santo and Port
Tawny), and white sweet wines (Malaga Dulce and Oremus Takaji)
it has a silica core that allows keeping their diameter large
enough to avoid pressure limitations at high linear velocities.
This type of column is constituted by a solid silica core of 1.7
μm in size with a porous outer layer 0.5 μm thick and a total
particle size of 2.7 μm. On the other hand, packing of columns
with porous sub-2 μm particles consist of derivatized, high-
purity porous-silica microspheres having reproducible bonded
monolayers. The rigidity and extremely narrow particle-size
distribution of totally porous particles allow high-resolution and
fastest analysis since operation at high flow rates is possible.
However, this column has the disadvantage that usually an
ultra-high-pressure liquid chromatography (UPLC) system is
required to achieve the best results, while a fused core column
allows fast separation on a conventional LC system, without
̃
a
́
were purchased. Most of them are from Spain, with the exception of
red sweet wines (from Italy and Portugal) and Oremus Tokaji wine
(from Hungary).
2.2. Instrumentation and Software. Chromatographic studies
were performed on an Agilent model 1100 LC instrument (Agilent
Technologies, Palo Alto, CA, USA), equipped with an online degasser,
quaternary pump, manual six-way injection valve, UV−vis diode-array
detector, rapid scan fluorescence spectrophotometer detector, and the
Chemstation software package to control the instrument, data
acquisition, and data analysis. Chromatographic studies and analytical
separation were carried out in columns purchased from Agilent. The
column temperature was controlled by a coil with recirculating water,
in which the temperature was selected through a thermostatic bath.
The injection volume was set at 20 μL for the Zorbax-Eclipse XDB
C18 column and at 10 μL for columns with minor particle size. The
29
significant loss in efficiency or resolution. In the past decade,
comparative studies about the separation of compounds in
columns with different particle size, such as pharmaceutical
−1
−1
3
0
29
flow rate was 1 mL min for Eclipse columns and 0.5 mL min for
the Poroshell column. Detection was performed with a fluorimetric
detector at 450 nm, exciting at 270, 330, and 350 nm.
A Crison MicropH 501 m (Barcelona, Spain), equipped with a
combined glass/saturated calomel electrode, was used for pH
measurements.
compounds or aflatoxins, among others, have been realized
by several authors, and in all cases, they concluded that
columns with small particle size achieve major separation
efficiency and improve the analytical method developed. On the
other hand, core−shell columns have been compared with sub-
2
μm porous particles columns in pharmaceutical compound
Calibration curves and analytical figures of merit were performed by
3
1
32
analysis, and a higher capacity of the core−shell columns to
accomplish better separation efficiency under the same
operation conditions was demonstrated.
means of the ACOC program, in MATLAB code.
2.3. General Procedure: Calibration Curves. To build the
calibration curves, aliquots of each α-dicarbonyl compound in variable
concentration were placed in 25 mL volumetric flasks, and 0.125 mL
of ammonia/ammonium chloride buffer (0.5 M, pH = 10) and 1.5 mL
of 8.6 mM DDP solution containing 200 mM β-ME were added. After
The present paper is focused in two ways: first, to investigate
the chromatographic behavior, in different packed columns, of
the lumazinic derivatives formed from α-dicarbonyl com-
pounds, in order to explore the advantages of columns with
core−shell and minor particle size in analysis of these
compounds; second, apply the obtained results in development
of a simple, rapid, and competitive method to determine these
compounds in different types of wines.
3
0 min at 60 °C, solutions were cooled in ice water and diluted with
0
.4 mM phosphoric acid solution (pH 3.2) up to the mark. Resulting
solutions were filtered through a 0.22 μm nylon filter, and aliquots of
0 μL were injected in the chromatographic system. Separation was
1
performed with a Poroshell 120 column, employing a solution of 0.4
mM phosphoric acid solution (pH 3.2)/MeOH (95:5, v/v) (eluent A)
and MeOH (eluent B) as mobile phase, with the following gradient
mode: 0−5 min, 0% B; 5−13 min, 30% B; 13−13.5 min, 40% B. These
conditions were maintained until 25 min, and finally, the eluent B
content was decreased to the initial conditions (0% B) and the column
was re-equilibrated for 5 min. The eluate was fluorimetrically
monitored at 450 nm (exciting at 270, 330, and 350 nm), and peak
areas were used as analytical signal. Three replicas of each standard
were used. Temperature was fixed at 25 °C, and a flow rate of 0.5 mL
2. MATERIALS AND METHODS
2.1. Chemicals, Standards, and Samples. DIA, 2,3-Pen, and
PhGly, all of 97% purity, and aqueous solutions of Gly and MGly, 40%
purity, were purchased from Sigma-Aldrich (Madrid, Spain). GS
(
98%) and 3-DG (95%) were obtained from Santa Cruz
Biotechnology (California, USA). Stock standard solutions of 3-DG
(
−1
−1
−1
100 μg mL ), GS (400 μg mL ), and PhGly (135 μg mL ) were
−1
min was employed.
.4. Analysis of α-Dicarbonyl Compounds in Wine Samples.
prepared by dissolving in ultrapure water adequate amounts of the
2
powder presentation of each compound. Stock solutions of DIA (600
−
1
−1
−1
Analysis of the α-dicarbonyl compounds in red, white, and sweet wines
was carried out by the standard addition method. For each wine
sample, adequate volumes were added in a 25 mL volumetric flask.
Increasing volumes of a standard mixture of the dicarbonyl
compounds were added, and the general procedure was followed for
the derivatization step. Separation of derivatives was carried out in a
Poroshell 120 column thermostated at 25 °C. A mobile phase
composed of 0.4 mM phosphoric acid solution (pH 3.2)/MeOH
(95:5, v/v) (eluent A) and MeOH (eluent B) was employed by
following the same gradient mode described in section below.
Fluorescence excitation/emission wavelengths of 330/450 nm were
employed.
μg mL ), 2,3-Pen (150 μg mL ), Gly (50 μg mL ), and MGly (70
−
1
μg mL ) were prepared by weighting and dilution of the adequate
aliquots in ultrapure water. Stock analyte solutions were prepared
separately and stored at 4 °C. Working standard mixture solutions
were daily prepared by suitable dilution of stock analyte solutions with
ultrapure water. 5,6-Diamino-2,4-hydroxypyrimidine sulfate (DDP)
(
95%) was purchased from Sigma-Aldrich, and a 8.6 mM stock
solution was daily prepared by dissolving adequate amounts in
ultrapure water containing 200 mM β-mercaptoethanol (β-ME), also
provided from Sigma-Aldrich. Ammonia/ammonium chloride buffer
was prepared by dissolving ammonium chloride from Panreac
(
Barcelona, Spain) in ultrapure water and fixing the pH at 10.0 with
9
8
dx.doi.org/10.1021/jf404180t | J. Agric. Food Chem. 2014, 62, 97−106

Journal of Agricultural and Food Chemistry
. RESULTS AND DISCUSSION
Article
120.0 ng mL− 2,3-Pen, 500.0 ng mL PhGly, and DDP
solution (7.5 × 10− mM) (concentrations enough so that all
obtained derivatives provide a good signal, easily measurable,
and whose variations are detectable, when the conditions of the
reaction are changed) were placed in 25 mL volumetric flasks at
pH 10 and incubated at 45, 60, and 80 °C during several
heating times in the range 2−60 min. After completing the
reaction times, the flasks were cooled in ice water and made to
volume with 0.4 mM phosphoric acid (pH 3.2), and aliquots of
1
−1
3
2
The first aim of this work is to realize a comparative study of
the chromatographic behavior of the lumazinic derivatives from
α-dicarbonyl compounds using three different C18 columns. In
any case, the focus was to obtain baseline separation of all
compounds in the shortest time and with the lowest solvent
consumption as possible in order to achieve a simple, a fast, and
an environmental friendly method, susceptible of being applied
to analysis of these compounds in wine samples.
1
0 μL of these solutions were injected in the chromatographic
3.1. Formation of Lumazinic Derivatives. The most
system. For all α-dicarbonyl compounds, except for DIA and
PhGly, the peak area of the derivatives increased with
temperature and is maintained constant between 60 and 80
extensively applied methods in wine analysis for control of α-
dicarbonyl compounds levels are based on reaction with 2,3-
diaminobenzene to form quinoxaline derivatives that can be
22
°
C. The peak area of DIA and PhGly derivatives increased up
determined by HPLC with UV detection at 313 nm,
33
to 60 °C, and for higher temperatures, the signal decreases.
Finally, we selected 60 °C as the optimum heating temperature.
With respect to the heating time, the peak area of the seven
lumazinic derivatives increased up to 15 min. However, the
reaction rate for formation of GS and 3-DG lumazinic
derivatives is more dependent on the heating time, and after
60 min, these reactions had not been completed. With the
object of not extending the total time of the derivatization
reactions, a compromise value of 30 min was selected as
optimum.
fluorimetric detection exciting at 350 nm, or GC with a MS
34
or NPD detector. These derivatization reactions need high
temperatures and reaction times close to 3 h. In this paper,
DDP is proposed as derivatizing reagent in order to simplify the
derivatization reaction. Optimization of the conditions for
́
formation of the lumazinic derivatives for Gly and MGly was
20
described in previous work. In this work, use of DDP as
derivatizing reagent was expanded to the seven α-dicarbonyl
compounds that can be present in wine samples. The scheme of
the derivatization reaction is shown in Figure 1.
In the optimized conditions of temperature and reaction
time, the DDP/total aldehyde ratio was studied in a range
between 10:1 and 70:1. For all compounds a 10:1 ratio was
enough to obtain the maxima signal; however, for GS and DIA
a 30:1 ratio was necessary, and this relation was chosen as
optimum.
Studies carried out showed that in the derivatization
conditions DDP is degraded into several products, with the
object of minimizing the appearance of degradation products
that can interfere in analysis of some of the lumazinic
1
4,18
derivatives, and in accordance with previous research
β-
ME is added to the DDP solution. The influence of this reagent
in the stabilization of DDP has been studied by preparing DDP
solutions (0.8 mM) in the presence of different concentrations
of β-ME (200, 400, and 600 mM). Aliquots of each solution
were placed in a 25 mL volumetric flask, heated at 60 °C for 30
min at pH 10, diluted with 0.4 mM phosphoric acid (pH = 3.2)
until the mark, and injected into the chromatographic system. It
can be observed that degradation of DDP was avoided in the
same extension, independent of the concentration of β-ME
employed. Therefore, a solution of DDP was prepared in the
presence in 200 mM of this compound. Figure 2A shows the
chromatograms of a DDP solution in the presence and in the
absence of 200 mM β-ME, and it can be observed that, in the
presence of β-ME, degradation of the reagent is practically
avoided and signal decreases.
Figure 1. Derivatization reaction of α-dicarbonyl compounds with 5,6-
diamino-2,4-dihydropyrimidine (DDP) to yield the corresponding
lumazinic derivatives.
In Figure 2B, a representative chromatogram from lumazinic
derivatives obtained in physicochemical-optimized conditions is
shown. As can be seen, all α-dicarbonyl compounds become a
single lumazinic derivative, except 2,3-Pen and PhGly, which
form two derivatives. For unequivocal identification of the
derivatives, a comparison of the retention time of Gly derivative
with a reference standard of lumazine was done. A standard
solution containing 50 ng mL− of lumazine was injected in the
chromatographic system, in the conditions previously
described. As expected, retention time and excitation spectra
of both peaks are coincident, and this allows confirmation of
formation of the lumazinic derivatives.
The physicochemical variables that influence the yield and
reaction rate, such as pH, temperature, reaction time, and DDP
excess, were studied chromatographically by monitoring the
derivatization reaction. In accordance with previous studies, the
yield of the reaction of α-dicarbonyl compounds with DDP was
optimal at pH 10, and this value was fixed in all solutions with
0.125 mL of ammonia/ammonium chloride buffer (0.5 M).
1
Temperature and reaction time are two very important
parameters that influence in the DDP−dicarbonyl compound
reaction rates. For this reason, a series of standard mixture
−
1
−1
solutions containing 1.0 μg mL GS, 400.0 ng mL 3-DG,
1
00.0 ng mL− Gly, 100.0 ng mL MGly, 100.0 ng mL DIA,
1
−1
−1
9
9
dx.doi.org/10.1021/jf404180t | J. Agric. Food Chem. 2014, 62, 97−106

Journal of Agricultural and Food Chemistry
Article
is more favorable that the other, probably due to steric
impediments.
As a consequence of the different excitation and emission
wavelengths of the lumazinic derivatives, a fluorescence
detector was programed in multiexcitation mode and each
chromatogram was recorded simultaneously exciting at 270,
3
30, and 350 nm. A wavelength of 450 nm was selected as the
emission wavelength as a compromise in order to analyze all
compounds in only one run.
3
.2. Optimization of Separation and Comparison of
Columns Efficiency. With the aim of selecting the more
appropriate column for separation of the seven α-dicarbonyl-
DDP derivatives, we realized a systematic comparison of three
columns assayed. This study was performed in order to develop
a simple, reliable, and robust method for efficient separation
and quantification of these compounds. In this study, reversed-
phase C18 columns with different dimensions and different size
and surface particles were tested. The columns used were
Zorbax-Eclipse XDB C18 (5.6 μm), Poroshell 120 (2.7 μm),
and Zorbax-Eclipse XDB-C18 RRHT (1.8 μm). The most
important parameters that influence chromatographic separa-
tion, such as mobile phase composition, flow rate, and column
temperature, were assayed with each column.
The composition of the mobile phase has a large influence in
complete resolution of the derivatives. Due to the analytes
differing widely in polarity, isocratic elution is unviable and it
was necessary to use a binary gradient. The two solvents
employed in the gradient elution were mixtures of acid:MeOH,
as solvent A, and MeOH 100% as solvent B. The presence of
acid in the mobile phase is essential to complete resolution of
the lumazinic derivatives, so the influence of the nature of the
acid used in solvent A and the pH of the mobile phase was
studied. Acetic, phosphoric, and formic acids were assayed, and
the pH was varied between 2.9 and 4.0 for each acid. The
behavior of the α-dicarbonyl−DDP derivatives was identical
with the three columns assayed, independent of the acid used.
In general, the retention times of DIA, 2,3-Pen, and PhGly
lumazine derivatives are not affected, and the retention times of
the less retained derivatives decreased as the pH increased.
Finally, 0.4 mM phosphoric acid (pH 3.2) was selected as the
aqueous component of solvent A for the three columns.
Below, the composition of solvent A was optimized for each
column by modifying the phosphoric acid (0.4 mM)/MeOH
ratio between 99/1 and 99/5, v/v. Also, different gradients with
methanol as solvent B were assayed to achieve baseline
separation of all derivatives in a time as short as possible.
Optimized conditions for each column are summarized in
Table 1.
Also, the influence of the temperature in the separation has
been studied with the three columns. In general, the capacity
factors decreased with increasing temperature for all com-
pounds. It is noticeable that this effect is greater for the less
retained compounds and when the mobile phase is 100%
aqueous, as can be seen in Figure 3. In gradient mode, as the
percentage of methanol increased and the mobile phase
viscosity was varying, the effect of temperature in the capacity
factors decrease, and the variation in the last four analytes
eluted is lower.
Representative chromatograms obtained in the optimized
conditions with the three columns assayed, XDB-C18,
Poroshell, and XDB-C18 RRHT, are shown in Figure 4. As
can be seen, use of a conventional monomeric C18 column
results in a lack of complete separation between the peak of the
Figure 2. (A) Influence of the presence of β-ME in the degradation
process of DDP in the derivatization reaction conditions. Chromato-
−2
grams correspond to a DDP solution (7.5 × 10 mM) in the absence
of β-ME (continuous line) and in the presence of 200 mM of β-ME
(
dashed line). λ /λ = 330/450 nm. (B) HPLC/FLD chromato-
exc em
grams from a stock standard solution of the α-dicarbonyl compounds
−2
in the presence of DDP solution (7.5 × 10 mM) at pH 10,
derivatized in optimized conditions (heating for 30 min at 60 °C)
−
1
(
(
4
[
5
continuous line), and of a standard solution of lumazine, 50 ng mL
−1
dashed line). λ /λ = 330/450 nm. [GS] = 1.0 μg mL , [3-DG] =
exc em
−1
−1
−1
00.0 ng mL , [Gly] = 100.0 ng mL , [MGly] = 100.0 ng mL ,
−1
−1
DIA] = 100.0 ng mL , [2,3-Pen] = 120.0 ng mL , and [PhGly] =
00.0 ng mL . (C) Excitation and emission spectra of lumazinic
−
1
derivatives GS, 3-DG, Gly, MGly, DIA, and 2,3-Pen (), PhGly 1
(
---), and PhGly 2 (−−−).
Spectral characteristics of all derivatives were studied in order
to achieve higher sensitivity in their analysis, and excitation and
emission spectra were obtained for each peak (Figure 2C). All
of them, except the PhGly derivatives, presented excitation and
emission wavelengths characteristics of the lumazines at 330
and 475 nm, respectively. Also, the two derivatives of 2,3-Pen
have identical spectral characteristics, with excitation and
emission wavelengths at 330 and 475 nm, respectively, and
both are formed in similar extension, as can be seen in the
chromatogram shown in Figure 2B, where the peak areas of the
two derivatives are similar. In the case of the PhGly derivatives,
its spectral characteristics are different. Thus, the lumazinic
derivative with minor elution time (PhGly 1) presents
excitation and emission wavelengths at 270 and 420 nm,
respectively, while the other derivative (PhGly 2) shows
excitation at 355 nm and emission at 410 nm. On the other
hand, in this case formation of one of the derivatives (PhGly 2)
1
00
dx.doi.org/10.1021/jf404180t | J. Agric. Food Chem. 2014, 62, 97−106

Journal of Agricultural and Food Chemistry
Article
Table 1. Column Characteristics and Optimum Chromatographic Parameters for Each Column
column
mobile phase
gradient mode
other parameters
Agilent Zorbax Eclipse XDB-C18
40 mM H PO /MeOH (98/2, v/v)
0−8 min, 0% B;
8−13 min, 20% B;
13−14 min, 35% B;
Tcolumn: 30 °C
Vinjection: 20 μL
t0: 1.4 min
3
4
(150 mm × 4.6 mm, 5.6 μm)
(eluent A)
MeOH (eluent B)
3
5% B until 25 min
P0: 145 bar
−1
flow rate: 1 mL min
Agilent Poroshell 120 EC-C18
150 mm × 3 mm, 2.7 μm)
40 mM H PO /MeOH (95/5, v/v)
0−5 min, 0% B;
5−13 min, 30% B;
Tcolumn: 25 °C
Vinjection: 10 μL
t0: 1.8 min
3
4
(
(eluent A)
MeOH (eluent B)
13−13.5 min, 40% B;
0% B until 25 min
4
P0: 245 bar
−1
flow rate: 0.5 mL min
Agilent Zorbax Eclipse XDB-C18 RRHT
50 mm × 4.6 mm, 1.8 μm)
40 mM H PO /MeOH (98/2, v/v)
0−5 min, 0% B;
Tcolumn: 20 °C
Vinjection: 10 μL
t0: 0.5 min
3
4
(
(eluent A)
MeOH (eluent B)
5−13 min, 35% B;
35% B until 18 min
flow rate: 1 mL min−
1
P0: 180 bar
Figure 3. Variation of capacity factor (k′) with column temperature for the three columns studied: (◆) GS, (●) Gly, (▲) 3-DG, (+) MGly, (×)
DIA, (◇) 2,3-Pen (1), (△) 2,3-Pen (2), (□) PhGly (1), and (○) PhGly (2).
1
01
dx.doi.org/10.1021/jf404180t | J. Agric. Food Chem. 2014, 62, 97−106

Journal of Agricultural and Food Chemistry
Article
columns, getting a total scan time of 18 min, against the 25 min
for the other two columns. Thus, the column length reduction
decreases analysis time, as expected. Retention times of the
lumazinic derivatives in the three columns assayed can be seen
in Table 2.
Peak resolution describes the degree of separation between
two compounds. Usually an R value of 1 is accepted to consider
that a separation is satisfactory, but R must be equal to or
higher than 1.5 to achieve a baseline resolution. As can be seen
in Table 2, better resolution for all compounds is achieved
when the separation was carried out with a Poroshell column. It
is important to highlight that R has been calculated with respect
to the previous peak eluted; thus, in the case of columns XDB-
C18 and Poroshell, MGly resolution is referred to its separation
from the DDP. However, as the order of elution between DDP
and MGly derivative changes in XDB-C18 RRHT, the
resolution of MGly is calculated with regard to the 3-DG
derivative, with a resolution value of DDP-MGly of 1.4 ± 0.1.
In relation to k′ values, in the case of the Poroshell column,
they are comprised between 0.84 and 12.7, which are near the
35
ideal values (1 and 10), while for XDB-C18 RRHT, k′ values
are too high (approximately 35 for the last derivative). This is
in agreement with the fact that superficially porous particles
typically have about one-half to three-quarters the surface area
of totally porous particles, resulting in a smaller capacity factor
value for columns of core−shell particles in size comprised
36
between 2.5 and 2.7 μm. On the other hand, N values,
calculated based on the width of peak at half height, were
greater for the Poroshell column in all cases, which indicates
more efficient chromatographic separation.
Figure 4. Optimum separation of lumazinic derivatives in the three
Other aspects like solvent consumption were taken into
account. In this way, although the analysis time for the XDB-
C18 RRHT column is lower, solvent consumption is higher
than with Poroshell (12.5 against 18 mL per run) due to the
flow rate employed with this column being one-half.
In order to evaluate the precision of analysis with the three
columns, repeatability was assessed by injection of 10 standard
solutions, in the optimized conditions, for each column. The
RSD from the signal of the peak area was between 1.6% and
columns assayed. Dashed lines indicated the gradient program
employed. λ /λ = 330/450 nm.
exc em
MGly−DDP derivative and the peak of the DDP reagent. In
addition, more polar analytes presented peak tailing, and in
general, peak shape was wider than in the other columns. On
the other hand, lower analysis time is obtained with the XDB-
C18 RRHT column. It is important to remark that elution
order of MGly derivative and DDP is inverted in this column.
The Poroshell column achieves complete separation of the
compounds in 25 min with narrow and symmetric peaks.
Once elution conditions were established, chromatographic
parameters (column resolution (R), capacity factor (k′), and
theoretical plate number (N)) were calculated to obtain
information about the efficiency of each column in the
chromatographic separation of the lumazinic derivatives.
These parameters have been calculated following eqs 1−3,
and the obtained values are summarized in Table 2.
7
2
.4% for XDB-C18, 2.4% and 6.3% for XDB-C18 RRHT, and
.4% and 6.1% for Poroshell. Therefore, significant differences
were not found.
In summary, although the XDB-C18 RRHT column has the
fastest elution time, resolution was enhanced with the core−
shell column. Also, the solvent reduction achieved with this
column is a good advantage with the aim of developing a
method of analysis less polluting as possible; thus, the Poroshell
1
20 column was chosen for the following experiments.
.3. Method Validation: Analytical Parameters. For
3
1
.17(tR2 − t )
validation of the method, a calibration curve of each compound
was established employing the peak area as analytical signal and
using the optimized conditions for the selected Poroshell
column (Table 1). The linearity of the method was assessed by
preparing calibration standards with concentrations ranging
R
1
R = (
w1 ) + (w
)
/2 1
1/2 2
(1)
tR − t
0
k′ =
t0
(2)
(3)
−1
−1
from 10.0 to 600 ng mL of GS, from 4.0 to 120.0 ng mL of
−1
N = 5.54x(t /w )2
Gly and 3-DG, from 4.0 to 40.0 ng mL of MGly and DIA,
R
1/2
−1
from 4.0 to 150.0 ng mL of PhGly, and from 4.0 to 60.0 ng
−1
t is the retention time, t is the dead time, and w is the width
mL of 2,3-Pen. The pH 10 of the derivatization reaction was
maintained with ammonia/ammonium chloride buffer, and the
concentration of DDP was 0.86 mM, as previously explained in
section 2.3. Standard solutions containing all α-dicarbonyl
compounds were prepared in triplicate for each concentration
level, and the analytical figures of merit were calculated
R
0
1/2
of the peak at half height.
Retention time is the most important parameter to know the
length of analysis. For all compounds, using the optimized
elution conditions, retention times were shorter with the XDB-
C1 8 RRHT column than with conventional C18 and Poroshell
1
02
dx.doi.org/10.1021/jf404180t | J. Agric. Food Chem. 2014, 62, 97−106

Journal of Agricultural and Food Chemistry
Article
Table 2. Chromatographic Parameters Calculated for Each α-Dicarbonyl Compound in the Three Columns Assayed
a
a
a
a
column
α-dicarbonyl compound
tR ± SD (min)
R ± SD
k′ ± SD
N ± SD
Zorbax Eclipse XDB-C18
GS
2.89 ± 0.05
5.90 ± 0.10
7.80 ± 0.20
12.79 ± 0.07
15.34 ± 0.03
17.20 ± 0.02
17.55 ± 0.02
22.70 ± 0.06
23.51 ± 0.07
6.59 ± 0.06
7.41 ± 0.18
3.03 ± 0.04
1.17 ± 0.07
8.57 ± 0.10
7.07 ± 0.06
1.34 ± 0.01
12.67 ± 0.22
1.40 ± 0.02
1.06 ± 0.03
3.24 ± 0.09
4.57 ± 0.16
8.12 ± 0.03
9.93 ± 0.02
11.27 ± 0.02
11.51 ± 0.02
15.18 ± 0.04
15.76 ± 0.04
1480 ± 77
1984 ± 148
2052 ± 145
27 283 ± 1593
47 071 ± 2150
81 245 ± 1595
72 407 ± 1703
27 549 ± 1303
24 488 ± 754
Gly
3
-DG
MGly
DIA
2
2
,3-Pen (1)
,3-Pen (2)
PhGly (1)
PhGly (2)
Poroshell 120 EC-C18
GS
3.30 ± 0.02
6.25 ± 0.05
7.31 ± 0.06
12.05 ± 0.04
15.04 ± 0.05
18.03 ± 0.04
18.45 ± 0.04
23.70 ± 0.10
24.60 ± 0.10
9.79 ± 0.08
13.04 ± 0.11
4.06 ± 0.03
1.58 ± 0.03
11.68 ± 0.11
12.06 ± 0.12
1.68 ± 0.02
17.63 ± 0.45
2.08 ± 0.09
0.84 ± 0.01
2.49 ± 0.01
3.07 ± 0.01
5.72 ± 0.04
7.39 ± 0.05
9.05 ± 0.06
9.29 ± 0.06
12.28 ± 0.10
12.70 ± 0.11
4348 ± 117
10 018 ± 177
12 178 ± 149
34 227 ± 632
57 992 ± 688
88 847 ± 2298
82 818 ± 2207
74 283 ± 5729
76 963 ± 9321
Gly
3
-DG
MGly
DIA
2
2
,3-Pen (1)
,3-Pen (2)
PhGly (1)
PhGly (2)
Zorbax Eclipse XDB-C18 RRHT
GS
1.30 ± 0.02
2.88 ± 0.03
4.13 ± 0.04
9.37 ± 0.09
12.08 ± 0.03
13.73 ± 0.03
14.00 ± 0.03
16.95 ± 0.06
17.20 ± 0.10
6.34 ± 0.17
7.96 ± 0.16
5.10 ± 0.14
12.13 ± 0.23
8.54 ± 0.16
8.81 ± 0.20
1.45 ± 0.03
10.83 ± 0.44
1.57 ± 0.06
1.65 ± 0.05
4.87 ± 0.13
7.42 ± 0.21
18.10 ± 0.44
23.63 ± 0.55
27.00 ± 0.63
27.55 ± 0.64
33.54 ± 0.75
34.63 ± 0.77
931 ± 52
2605 ± 122
4023 ± 188
Gly
3
-DG
MGly
DIA
3927 ± 186
69 012 ± 7705
86 418 ± 4532
80 582 ± 7418
39 543 ± 3950
43 576 ± 3225
2
2
,3-Pen (1)
,3-Pen (2)
PhGly (1)
PhGly (2)
a
SD: standard deviation.
Table 3. Calibration Data and Validation Parameters
d
precision
a
b
c
α-dicarbonyl
compound
linear range
intercept ±
slope ± SD
LOD
LOQ
intraday
(n = 10)
interday
(n = 6)
(ng mL−
1
)
SD
a
(ng mL )
−1
R2
(ng mL−1)
(ng mL−1)
GS
10−600
4−120
4−125
4−40
4−30
4−150
(−29) ± 17
464 ± 33
70 ± 34
26 ± 26
188 ± 22
(−72) ± 32
(−12) ± 18
(−16) ± 13
2.10 ± 0.02
11.80 ± 0.20
7.74 ± 0.08
32.50 ± 0.40
24.00 ± 0.50
7.80 ± 0.10
8.75 ± 0.06
8.00 ± 0.10
14.30 ± 0.10
0.9989
0.9975
0.9986
0.9978
0.9963
0.9974
0.9994
0.9978
0.9991
0.31
0.05
0.08
0.02
0.03
0.09
0.08
0.16
0.07
1.03
0.18
0.28
0.07
0.10
0.30
0.27
0.52
0.25
2.9
4.5
3.6
4.5
6.0
3.9
2.3
4.7
5.7
5.0
4.9
4.3
2.1
2.3
3.4
3.2
5.0
3.3
Gly
3-DG
MGly
DIA
2
,3-Pen (1)
,3-Pen (2)
2
PhGly (1)
PhGly (2)
4−60
32 ± 15
a
b
c
d
SD: standard deviation. Limit of detection. Limit of quantification. Expressed as relative standard deviation (% RSD). Standard solution
−1
−1
−1
−1
−1
employed containing 300.0 ng mL GS, 60.0 ng mL Gly, 3-DG, 20.0 ng mL MGly, DIA, 75.0 ng mL 2,3-Pen, and 30.0 ng mL PhGly.
employing the peak areas as analytical signal. The analytical
figures of merit obtained for each α-dicarbonyl compound are
shown in Table 3. Evaluation of the precision of the optimized
method was done by analyzing standard solutions of the α-
dicarbonyl compounds in the same day (intraday precision, n =
corresponding to 3 and 10 times the standard deviation of
the signal from baseline from chromatograms obtained in
intraday precision study. Low values are obtained that are
indicative of the high sensitivity of the developed method.
Results obtained are summarized in Table 3.
1
0) and in consecutive days (interday precision, n = 6).
3.4. Quantitative Measurement of α-Dicarbonyl
Compounds in Wines. After evaluation and selection of the
optimum column for separation of seven α-dicarbonyl
compounds, the proposed method was applied to analysis of
different red, white, and sweets wines. To carry out
quantification of the α-dicarbonyl compounds in wine samples,
the influence of the matrix effect over chromatographic
Intraday and interday repeatability values, expressed as relative
standard deviation (RSD), are lower than 6.0% and 5.0%,
respectively, so they may be considered as a guarantee of the
goodness of the proposed method, and use of an internal
standard is not required. The limits of detection (LOD) and
quantification (LOQ) were calculated as concentrations
1
03
dx.doi.org/10.1021/jf404180t | J. Agric. Food Chem. 2014, 62, 97−106

Journal of Agricultural and Food Chemistry
Article
Table 4. Concentrations of the Six α-Dicarbonyl Compounds Detected in Commercial Wine Samples
a
a
a
a
a
a
a
type of wine
[GS]
[Gly]
[3-DG]
[MGly]
[DIA]
[2,3-Pen]
total α-DC
b
red wines
Monasterio Tentudia
Spain)
́
, 2008 (Extremadura,
4.50 ± 0.20
1.85 ± 0.08
13.50 ± 0.30
1.58 ± 0.05
3.10 ± 0.60
3.00 ± 0.04
27.53 ± 0.71
Ribera Duero, 2010 (Spain)
0.50 ± 0.20
1.70 ± 0.20
1.35 ± 0.03
7.10 ± 0.20
7.70 ± 0.40
10.40 ± 0.50
0.68 ± 0.01
0.42 ± 0.02
8.80 ± 0.20
1.22 ± 0.05
1.45 ± 0.02
5.60 ± 0.03
20.48 ± 0.49
26.46 ± 0.58
Merlot, 2009 (Extremadura, Spain)
c
white wines
Rioja Comportillo, 2011 (D.O. Rioja, Spain)
0.70 ± 0.10
0.40 ± 0.10
5.20 ± 0.20
1.53 ± 0.08
1.12 ± 0.05
1.10 ± 0.10
4.90 ± 0.20
7.90 ± 0.30
6.10 ± 0.20
0.73 ± 0.03
0.21 ± 0.01
0.28 ± 0.01
0.29 ± 0.01
0.63 ± 0.02
0.64 ± 0.02
2.22 ± 0.03
1.17 ± 0.03
0.90 ± 0.04
10.37 ± 0.24
11.43 ± 0.32
14.22 ± 0.30
̃
Vina Canchal, 2011 (Extremadura, Spain)
́
Manzanilla, dry wine (D.O. SanLucar de
Barrameda, Spain)
d
red sweet wines
Vin Santo (Toscana, Italy)
138 ± 11
89 ± 13
4.50 ± 0.30
8.90 ± 0.40
133 ± 5
39 ± 4
1.68 ± 0.08
0.95 ± 0.05
1.02 ± 0.06
1.50 ± 0.09
2.30 ± 0.10
1.90 ± 0.10
280 ± 12
141 ± 14
Port Tawny (Porto, Portugal)
d
white sweet wines
Mal
́
aga Dulce (Mal
́
aga, Spain)
261 ± 7
37 ± 1
8.40 ± 0.20
5.50 ± 0.20
126 ± 5
75 ± 2
0.90 ± 0.05
3.21 ± 0.06
2.80 ± 0.10
0.60 ± 0.07
2.61 ± 0.08
4.60 ± 0.10
402 ± 9
126 ± 2
Oremus Tokaji, 2008 (Hungary)
a
Concentration expressed as μg mL−1 ± SD (standard deviation) Aliquot of wine samples: 500 μL for Gly, MGly, DIA and 2,3-Pen, and 120 μL for
b
c
d
3
-DG and DIA wine/water (1/50, v/v) for GS and 3-DG. Aliquot of wine samples: 1500 μL wine/water (1/50, v/v) for GS and 3-DG. Aliquot of
wine samples: 120 μL for Gly, MGly, DIA and 2,3-Pen, and 50 μL of a solution wine/water (1/50, v/v) for GS and 3-DG.
−1
Figure 5. Chromatograms obtained from each type of wine analyzed (continuous line) and from wine samples spiked with 125.0 ng mL GS, 60.0
−
1
−1
−1
−1
−1
ng mL 3-DG, 40.0 ng mL Gly, 6.0 ng mL MGly, 6.0 ng mL DIA, and 40.0 ng mL 2,3-Pen (dashed line). (A1) Analysis of GS, Gly, MGly,
́
́
and 2,3-Pen in red wine (Monasterio Tentudia, 0.50 mL of sample). (A2) Analysis of 3-DG and DIA in red wine (Monasterio Tentudia, 0.12 mL of
sample. (B1) Analysis of Gly, MGly, DIA, and 2,3-Pen in sweet wine (Vin Santo, 0.12 mL of sample). (B2) Analysis of GS and 3-DG in sweet wine
(
Vin Santo, 0.50 mL of a solution wine/water (1/50 v/v)). (C) Analysis of six α-dicarbonyl compounds in white wine (Manzanilla, 1.50 mL of
sample). λ /λ = 330/450 nm.
exc em
separation was first evaluated. For this standard addition
calibration curves, from each of the α-dicarbonyl compounds in
wine samples, were established, and calibration slopes were
compared with the corresponding slopes of the external
standard calibration plots of each analyte at the 95% confidence
level. In all cases, considerable differences were obtained, and
these results suggest that there is a matrix effect. As a
consequence, calibration curves were constructed using the
standard addition method for each of the wines analyzed and
used to quantify each analyte in duplicate.
adequate sample dilutions were optimized for each type of wine
analyzed. The volume chosen for each wine is shown in Table
4. The low volumes of samples needed for analysis, in
comparison with other published methods, highlight the
sensitivity of the proposed method.
Ten wines were analyzed following the methodology
described in section 2.4: 3 red wines, 3 white wines, 2 sweet
red wines, and 2 sweet white wines. Figure 5 shows an example
of chromatograms obtained from each type of wine analyzed,
and the results obtained for each α-dicarbonyl are shown in
22
It is important to bear in mind that levels of α-dicarbonyl
compounds vary with the type of wine, and for this reason,
Table 4. Contrary to the literature, PhGly was not detected in
any of the different wines analyzed in this study. Also, in all
1
04
dx.doi.org/10.1021/jf404180t | J. Agric. Food Chem. 2014, 62, 97−106

Journal of Agricultural and Food Chemistry
Article
samples, only one lumazinic derivative from 2,3-Pen was
observed.
particle size. By comparing the chromatographic behavior of
lumazinic derivatives in small particle size columns, it can be
observed that although the XDB-C18 RRHT column provides
the lowest retention times the best separation efficiency and
lowest solvent consumption is achieved with the Poroshell
column. For this reason, the comparison study carried out
showed that the Poroshell column was most appropriate for
analysis of α-dicarbonyl compounds and previous derivatization
to lumazines, and this column was selected for their analysis in
wine samples. Derivatization reaction time, temperature, and
sample volume employed in analysis have considerably
decreased with respect to the previous published method. On
the other hand, a very sensitive and reproducible method to
analyze these compounds in different commercial wine samples
has been developed, and no pretreatment of sample is required.
The sensitivity of the method is presented as a useful alternative
to other methods, even with MS detection, whose instrumen-
tation is more costly and complicated and not available in many
laboratories. A comparison between red and white wines shows
that the total α-dicarbonyl concentration is significantly higher
To date, the dicarbonyl compound more relevant in wines is
DIA, since it has a pronounced butter odor and its presence in
8
wines at high concentrations is undesirable. Depending on the
3
7,38
type of wine, the contents of DIA
ranged from 0.2 to 12 mg
−1
L . As it can be seen in Table 4, in the samples analyzed in this
−1
paper the content of DIA varies between 0.29 and 8.8 mg L .
It can be observed that in red wines the content is higher than
in white wines, in accordance with previous research, and no
significant differences were found between not sweet and sweet
wines, with the exception of Ribera Duero wine.
With respect to the other α-dicarbonyl compounds with a
short chain, concentrations found for Gly and MGly were
33
similar to those previously reported. Values ranged from 1.1
to 1.85 mg L− for Gly and 0.28 to 1.68 mg L for MGly, with
1
−1
the exception of Merlot wine, whose concentration of Gly
−1
found was 7.1 mg L . Thus, the levels found in Merlot wine
are close to the levels found for sweet wines, in which the
quantities of Gly and MGly found were higher (between 4.5
and 8.9 mg L− and 0.9 and 3.2 mg L for Gly and MGly,
respectively). In all cases the relation Gly/MGly is greater than
unity. With respect to the levels of 2,3-Pen, these are similar in
1
−1
−1
in red ones. In sweet wines, total amounts up to 400 mg mL
have been found.
−1
all wines analyzed ranging between 0.9 and 5.60 mg L , and no
appreciable differences were found between sweet and not
sweet wines.
AUTHOR INFORMATION
■
*
Corresponding Author
Tel.: +34924289375. Fax: +34924274244. E-mail: iduran@
unex.es.
On the other hand and with respect to the α-dicarbonyl
sugars, 3-DG and GS have been detected in all wines analyzed.
In red and white wines, the content of GS varies between 0.4
Funding
−1
The authors gratefully acknowledge funding from the Junta de
Extremadura (Project GR10033 of Research Group FQM003)
and 5.2 mg L and the 3-DG content between 4.9 and 13.5 mg
−1
L . In sweet wines, the concentration of both glucosones
increases considerably, independent of whether they are red or
white, and their concentrations vary between 37 and 261 mg
́
and the Ministerio de Economia y Competitividad of Spain
Project CTQ2011-25388), both cofinanced by the European
(
́
FEDER Funds. M.C.H.S. is grateful to the Consejeria de
−1
L .
Due to the high concentration of glucosones obtained in
́
Economia, Comercio e Innovacion
a fellowship (DOE 04/01/2011).
́
of Junta de Extremadura for
sweet wines, it was important to check if formation of the α-
dicarbonyl compounds from sugars was not favored by the high
temperature of the derivatization reaction conditions. There-
fore, in sweet wines, the derivatization reaction was performed
at room temperature for 2 h and 30 min and aliquots of these
samples were analyzed. A comparison of the chromatograms,
obtained at room temperature and higher temperature, does
not show significant differences. This fact allows us to conclude
that the elevated concentrations of glucosones found in sweet
wines are related to its high levels of sugars.
Notes
The authors declare no competing financial interest.
ABBREVIATIONS USED
■
GS, D-glucosone; 3-DG, 3-deoxyglucosone; Gly, glyoxal; MGly,
methylglyoxal; DIA, diacetyl; 2,3-Pen, 2,3-pentanedione;
PhGly, phenylglyoxal; DDP, 5,6-diamino-2,4-hydroxypyrimi-
dine; β-ME, β-mercaptoethanol
Finally, if analyzing the total α-dicarbonyl concentration, it
can be observed that no significant variations were found in
each type of red or white wine analyzed. However, if we made a
comparison between red and white wines, we can confirm that
the total α-dicarbonyl concentration is significantly higher in
red wines. On the other hand, in sweet wines, a great difference
has been found between the wines analyzed, and values of total
α-dicarbonyl concentration amounting to 402 mg L− have
been found. In addition, Tokaji wine, which it is a sweet white
wine produced with botrytized grapes, contains higher levels of
MGly. In spite of this, the total α-dicarbonyl concentration
found is lower compared to the other sweet wines analyzed.
In conclusion, all types of columns tested were able to separate
the lumazinic derivatives with sufficient resolution and peak
symmetry in the optimum chromatographic conditions,
although they differed in analysis time and separation efficiency
of the derivatives analyzed. The conventional C18 column had
the lowest resolution and peaks wider than columns with minor
REFERENCES
■
(
1) Thornalley, P. J. Advanced glycation and development of diabetic
complications: Unifying the involvement of glucose, methylglyoxal and
oxidative stress. Endocrinol. Metab. 1996, 3, 149−166.
2) Uchida, K.; Kanematsu, M.; Sakai, K.; Matsuda, T.; Hattori, N.;
(
Suzuki, D.; Miyata, T.; Noguchi, N.; Niki, E.; Osawa, T. Protein-bound
acrolein: Potential markers for oxidative stress. Proc. Natl. Acad. Sci.
1998, 95, 4882−4887.
1
(3) Van Boekel, M. A. J. S. Formation of flavour compounds in the
Maillard reaction. Biotechnol. Adv. 2006, 24, 230−233.
(
4) Chen, X. M.; Kitts, D. D. Identification and quantification of a-
́
dicarbonyl compounds produced in different sugar-amino acid
Maillard reaction model systems. Food Res. Int. 2011, 4, 2775−2782.
5) Flamini, R.; De Luca, G.; Di Stefano, R. Changes in carbonyl
compounds in Chardonnay and Cavernet Sauvignon wines as a
consequence of a malolactic fermentation. Vitis 2002, 41, 107−112.
6) Flamini, R.; Dalla Vedova, A. Glyoxal/Glycolaldehyde: A redox
system involved in malolactic fermentation of wine. J. Agric. Food
Chem. 2003, 51, 2300−2303.
(
(
1
05
dx.doi.org/10.1021/jf404180t | J. Agric. Food Chem. 2014, 62, 97−106

Journal of Agricultural and Food Chemistry
Article
(
7) De Revel, G.; Bertrand, A. A method for the detection of
(26) Fekete, S.; Olah
́
, E.; Fekete, J. Fast liquid chromatography: the
carbonyl compounds in wine: glyoxal and methylglyoxal. J. Sci. Food
Agric. 1993, 61, 267−272.
(
domination of core-shell and very fine particles. J. Chromatogr. A 2012,
1228, 57−71.
(27) Neue, U. D. HPLC Columns: Theory, Technology, and Practice;
Wiley-VCH: New York, 1997.
8) Bartowsky, E. J.; Henschke, P. A. The buttery attribute of wine
-
diacetyl- desirability spoilage and beyond. Int. J. Food Microbiol. 2004,
6, 235−252.
9) Oliveira, C. M.; Silva-Ferreira, A. C.; De Freitas, V.; Silva, A. M. S.
Oxidation mechanisms ocurring in wines. Food Res. Int. 2011, 44,
115−1126.
10) Davis, C. R.; Wibowo, D.; Eschenbruch, R.; Lee, T. H.; Fleet, G.
9
(
(28) McCalley, D. V. Instrumental considerations for the effective
operation of short, highly efficient fused-core columns. Investigation of
performance at high flow rates and elevated temperatures. J.
Chromatogr. A 2010, 1217, 4561−4567.
1
(
(29) Abrahim, A.; Al-Sayah, M.; Skrdla, P.; Bereznitski, Y.; Chen, Y.;
Wu, N. Practical comparison of 2.7 im fused-core silica particles and
̀
H. Practical implications of malolactic fermentation: A review. Am. J.
Enol. Viticult. 1985, 36, 290−301.
(
Valente, I. M.; Rodrigues, J. A.; Arauj
free and total diacetyl in wine by HPLC−UV using gas-diffusion
microextraction and pre-column derivatization. Food Control 2012, 24,
̀
porous sub-2 im particles for fast separations in pharmaceutical process
11) Ramos, R. M.; Grosso Pacheco, J.; Moreira Gonca
̧
lves, L.;
development. J. Pharm. Biomed. 2010, 51, 131−137.
́
o Barros, A. Determination of
́ ́
(30) Novakova, L.; Solich, P. A comparison of performance of
various analytical columns in pharmaceutical analysis: conventional
C18 and high throughput C18 zorbax columns. J. Chromatogr. A 2005,
1088, 24−31.
2
(
20−224.
12) Thornalley, P. J. Monosaccharide autoxidation in health and
disease. Environ. Health Perspect. 1985, 64, 297−307.
13) Deng, Y.; Yu, P. H. Simultaneous determination of form-
(31) Medina, A.; Magan, N. Comparison of three different C18
HPLC columns with different particle sizes for the optimization of
aflatoxins analysis. J. Chromatogr. B 2012, 889−890, 138−143.
(
(
̃ ̃
́ ́
32) Espinosa-Mansilla, A.; Munoz de la Pena, A.; Gonzalez-Gomez,
aldehyde and methylglyoxal in urine: involvement of semicarbazide-
sensitive amine oxidase-mediated deamination in diabetic complica-
tions. J. Chromatogr. Sci. 1999, 37, 317−322.
D. Using univariate linear regression calibration software in the
MATLAB Environment, application to chemistry laboratory practices.
Chem. Educ. 2005, 10, 337−345.
(
14) Cordeiro, C.; Ponces-Freire, A. Methylglyoxal assay in cells as 2-
(33) Da Silva-Ferreira, A. C.; Reis, S.; Rodrigues, C.; Oliveira, C.;
methylquinoxaline using 1,2-diaminobenzene as derivatizing reagent.
Anal. Biochem. 1996, 234, 221−224.
(
Guedes de Pinho, P. Simultaneous determination of ketoacids and
dicarbonyl compounds, key Maillard intermediates on the generation
of aged wine aroma. J. Food Sci. 2007, 72, S314−S318.
15) McLellan, A. C.; Phillips, S.; Thornalley, P. J. The assay of
methylglyoxal in biological systems by derivatization with 1,2-diamino-
,5-dimethoxybenzene. Anal. Biochem. 1992, 206, 17−23.
16) Akira, K.; Matsumoto, Y.; Hashimoto, T. Determination of
(34) Compendium of international analysis of methods: a-dicarbonyl
4
(
compounds by GC after derivatization Method for the determination
of a-dicarbonyl compounds of wine by GC after derivatization by 1,2-
diaminobenzene (OIV-Oeno 386B-2010) Method OIV-MA-AS315−
urinary glyoxal and methylglyoxal by high-performance liquid
chromatography. Clin. Chem. Lab. Med. 2004, 42, 147−153.
2
1 , 2010.
(
́ ́
17) Espinosa-Mansilla, A.; Duran-Meras, I.; Salinas, F. High-
(35) Murphy, I. B.; Stutte, C. A. Analysis for substituted benzoic and
performance liquid chromatographic-fluorometric determination of
glyoxal, methylglyoxal, and diacetyl in urine by prederivatization to
pteridinic rings. Anal. Biochem. 1998, 255, 263−273.
cinnamic acids using high-pressure liquid chromatography. Anal.
Biochem. 1978, 86, 220−228.
(36) DeStefano, J. J.; Schuster, S. A.; Lawhorn, J. M.; Kirkland, J. J.
(
18) Hurtado-San
́
chez, M. C.; Espinosa-Mansilla, A.; Rodríguez-
-Meras, I. Development of a
Performance characteristics of new superficially porous particles. J.
Chromatogr. A 2012, 1258, 76−83.
Caceres, M. I.; Martín-Tornero, E.; Duran
́
́
́
method for the determination of advanced glycation end products
precursors by liquid chromatography and its application in human
urine samples. J. Sep. Sci. 2012, 35, 2575−2584.
(37) Guillou, J.; Bertrand, A.; De Revel, G.; Barbe, J. C. Occurrence
of hydroxypropanedial in certain musts and wines. J. Agric. Food Chem.
1
(
997, 45, 3382−3386.
(
19) Yamada, H.; Miyata, S.; Igaki, N.; Yatabe, H.; Miyauchi, Y.;
Ohara, T.; Sakai, M.; Shoda, H.; Oimomi, M.; Kasuga, M. Increase in
-deoxyglucosone levels in diabetic rat plasma. Specific in vivo
38) Peynaud, E. Le Gout du Vin; Dunod: Paris, 1980; pp 44−45.
3
determination of intermediate in advanced Maillard reaction. J. Biol.
Chem. 1994, 269, 20275−20280.
(
20) Espinosa-Mansilla, A.; Duran
́ ́
̃ ̃
-Meras, I.; Canada-Canada, F.;
Prado-Mar
́
quez, M. P. High-performance liquid chromatographic
determination of glyoxal and methylglyoxal in urine by prederivatiza-
tion to lumazinic rings using in serial fast scan fluorimetric and diode
array detectors. Anal. Biochem. 2007, 371, 82−91.
(
21) Hayasaka, Y.; Bartowsky, E. J. Analysis of diacetyl in wine using
solid-phase microextraction combined with gas chromatography-mass
spectrometry. J. Agric. Food Chem. 1999, 47, 612−617.
(
22) De Revel, G.; Pripis Nicolau, L.; Barbe, J. C.; Bertrand, A. The
detection of alpha-dicarbonyl compounds in wine by formation of
quinoxaline derivatives. J. Sci. Food Agric. 2000, 80, 102−108.
(
23) Barros, A.; Rodrigues, J. A.; Almeida, P. J.; Oliva-Teles, T.
Determination of glyoxal, methylglyoxal, and diacetyl in selected beer
and wine, by HPLC with UV spectrophotometric detection, after
derivatization with o-phenylenediamine. J. Liq. Chromatogr. Relat.
Technol. 1999, 22, 2061−2069.
(
24) Ramos, R. M.; Pacheco, J. G.; Gonc a̧ lves, L. M.; Valente, I. M.;
Rodrigues, J. A.; Barros, A. Determination of free and total diacetyl in
wine by HPLC-UV using gas-diffusion microextraction and pre-
column derivation. Food Control 2012, 24, 220−224.
(
25) Shibamoto, T. Analytical methods for trace levels of reactive
carbonyl compounds formed in lipid peroxidation systems. J. Pharm.
Biomed. 2006, 41, 12−25.
1
06
dx.doi.org/10.1021/jf404180t | J. Agric. Food Chem. 2014, 62, 97−106