Modification of casein by the lipid oxidation product malondialdehyde

Article abstract of DOI:10.1021/jf072385b

The reaction of malondialdehyde with casein was studied in aqueous solution to evaluate the impact of this lipid oxidation product on food protein modification. By using multiresponse modeling, a kinetic model was developed for this reaction. The influence of temperature and pH on protein browning and malondialdehyde degradation was evaluated. The hypothesis that one malondialdehyde unit leads to the cross-linking of two casein-bound lysine residues was in accordance with the data. At higher malondialdehyde concentrations, a different reaction mechanism was operative, probably involving a dihydropyridine cross-link. The results obtained were compared with the reaction of casein with 2-oxopropanal, a well-studied α-dicarbonyl compound. The reaction of casein with 2-oxopropanal followed a different reaction pathway. Comparison of the degree of browning of casein by reaction with malondialdehyde and 2-oxopropanal showed a considerably higher degree of browning induced by malondialdehyde. This research has shown that kinetic modeling is a useful tool to unravel reaction mechanisms. Clearly, the contribution of lipid oxidation products, such as malondialdehyde, to protein modification (both in food and in vivo) can be substantial and needs to be taken into account in future studies.

Full text of DOI:10.1021/jf072385b

J. Agric. Food Chem. 2008, 56, 1713–1719 1713
Modification of Casein by the Lipid Oxidation
Product Malondialdehyde
,
†
†
§
AN ADAMS,* NORBERT DE KIMPE, AND MARTINUS A. J. S. VAN BOEKEL
Department of Organic Chemistry, Faculty of Bioscience Engineering, Ghent University, Coupure
links 653, B-9000 Ghent, Belgium, and Department of Agrotechnology and Food Sciences, Product
Design and Quality Management, Wageningen University, P.O. Box 8129, 6700 EV Wageningen,
The Netherlands
The reaction of malondialdehyde with casein was studied in aqueous solution to evaluate the impact
of this lipid oxidation product on food protein modification. By using multiresponse modeling, a kinetic
model was developed for this reaction. The influence of temperature and pH on protein browning
and malondialdehyde degradation was evaluated. The hypothesis that one malondialdehyde unit
leads to the cross-linking of two casein-bound lysine residues was in accordance with the data. At
higher malondialdehyde concentrations, a different reaction mechanism was operative, probably
involving a dihydropyridine cross-link. The results obtained were compared with the reaction of casein
with 2-oxopropanal, a well-studied R-dicarbonyl compound. The reaction of casein with 2-oxopropanal
followed a different reaction pathway. Comparison of the degree of browning of casein by reaction
with malondialdehyde and 2-oxopropanal showed a considerably higher degree of browning induced
by malondialdehyde. This research has shown that kinetic modeling is a useful tool to unravel reaction
mechanisms. Clearly, the contribution of lipid oxidation products, such as malondialdehyde, to protein
modification (both in food and in vivo) can be substantial and needs to be taken into account in
future studies.
KEYWORDS: Malondialdehyde; protein browning; kinetic modeling; Maillard reaction
INTRODUCTION
marker compound to measure the extent of oxidative deteriora-
tion of lipids in food and biological systems (2, 3).
Lipid oxidation and the Maillard reaction are probably the
two most important chemical reactions in food. They are largely
responsible for the formation of flavor, color, and texture and
for the resulting nutritional value of processed foods. In complex
food products, composed of lipids and carbohydrates as well
as proteins, both reactions are closely related and involve
common intermediates and polymerization pathways. In addi-
tion, lipid oxidation products as well as carbohydrates or their
degradation products can induce protein browning in a com-
parable manner (1). However, whereas many publications
describe browning as a result of the Maillard reaction, much
less is known on the impact of lipid oxidation products on
protein browning.
The reactivity of various proteins with malondialdehyde under
physiological conditions has been shown, leading to the
identification of various amino acid modifications in model
reactions or in vivo modifications (Figure 1). Reaction of
malondialdehyde with primary amino groups leads to the
formation of resonance-stabilized imines, that is, 3-amino-2-
propenal derivatives or enaminals 1. Formation of these
compounds is of little nutritional concern because they are most
Lipid oxidation yields a large variety of reactive carbonyl
compounds. Most of these possess only one functional group,
whereas malondialdehyde possesses two functionalities and
therefore a higher reactivity. Malondialdehyde is a secondary
product of lipid peroxidation and has been widely used as a
*
Author to whom correspondence should be addressed (telephone
0 32 9 264 59 64; fax 00 32 9 264 62 43; e-mail An.Adams@
UGent.be).
0
†
Ghent University.
Figure 1. Malondialdehyde-induced modifications of amino acid residues
identified in various proteins and model systems.
§
Wageningen University.
1
0.1021/jf072385b CCC: $40.75  2008 American Chemical Society
Published on Web 02/14/2008

1714 J. Agric. Food Chem., Vol. 56, No. 5, 2008
Adams et al.
probably hydrolyzed at the acidic pH of the stomach. Nucleo-
philic addition of a second free amino functionality leads to
the formation of a 3-amino-1-iminopropene cross-link 2.
Another type of structure containing bound malondialdehyde
concerns 4-substituted 1,4-dihydropyridine-3,5-dicarbaldehyde
derivatives (e.g., 3). These are formed when malondialdehyde
reacts with an amino compound in the presence of aldehydes
solutions; ꢀ ) 31500). Aliquots of this stock solution were used to
prepare model mixtures of various concentrations. After MDA addition,
the pH was adjusted with 1 M NaOH when necessary.
Heating of Samples. Ten milliliter aqueous solutions of 3% (w/w)
casein and various concentrations of MDA or 2-oxopropanol in 0.1 M
phosphate buffer were heated in screw-capped pressure-resistant metal
tubes (1.35 cm × 11.0 cm) in an aluminum heating block for a certain
period of time and temperature. After heating, the samples were rapidly
cooled in an ice bath. The samples were transferred to plastic 10 mL
tubes (Greiner) and stored at -18 °C. Samples of 0.5 mL were diluted
in 1.5 mL of 16% sodium dodecyl sulfate (SDS) and stored overnight
at 4 °C. All samples were heated and analyzed in duplicate.
Determination of Lysine Residues. Available lysine residues were
determined by derivatization with o-phthaldialdehyde (OPA). Fresh
OPA reagent was prepared daily by dissolving 40 mg of o-phthaldial-
dehyde in 1 mL of ethanol, adding 25 mL of 0.1 M sodium tetraborate
buffer (pH 9.8), 2.5 mL of 20% SDS, and 0.1 mL of 2-mercaptoethanol
in demineralized water, and adjusting the volume to 50 mL. The SDS-
diluted reaction mixtures were diluted 25 times in 0.1 M sodium
tetraborate buffer, and to 80 µL of diluted sample was added 2.4 mL
of OPA reagent. The sample was vortexed for 10 s, and the fluorescence
was measured after 2 min (or as close to 2 min as practically
achievable). A fluorescence spectrophotometer (LS50B, Perkin-Elmer,
Beaconsfield, U.K.) was used at excitation and emission wavelengths
of 340 and 430 nm, respectively, with slit widths of 2.5 µm. The OPA
reagent was used as the blank.
(
4). Because of the presence of two additional formyl groups
in 3, these lysine modifications can induce further cross-linking.
Synthesis and bioavailability studies of N,N′-di-(4-methyl-1,4-
dihydropyridine-3,5-dicarbaldehyde)lysine showed that this
compound represents a form of unavailable lysine, because it
is not metabolized to free lysine and cannot be absorbed from
the gut (5). From model reactions of N-acetylglycyl-L-lysine
methyl ester with malondialdehyde, an N-substituted 1,4-
dihydropyridine-3,5-dicarbaldehyde substituted at the 4-position
with a pyridinium unit 4 was identified (6). A modification of
collagenous arginine with the formation of 2-aminopyrimidine
derivative 5 has also been described (7). Other investigations
showed the possibility of an imidazole cross-link formed in the
reaction of malondialdehyde with lysine and arginine 6.
Modifications of the imidazole functionality of histidine as a
result of the reaction with malondialdehyde have also been
observed (7 and 8) (8). Malondialdehyde has been shown to
cross-link bovine serum albumin, forming dimers, and to modify
RNase, crystallin, and hemoglobin (3). Malondialdehyde can
also react with DNA and is as such both mutagenic and
carcinogenic (3). Requena et al. developed an analytical method
for the analysis of lysine-malondialdehyde adducts and proved
them to be useful biomarkers of lipid peroxidative protein
modification and of oxidative stress in vivo and in vitro (9).
Investigations of the antioxidant activity of the soluble reaction
products of beef sarcoplasmic proteins with malondialdehyde
showed that light brown pigments were produced with antioxi-
dant properties, in analogy with nonenzymatic browning (10).
Determination of Malondialdehyde. Concentrations of malondi-
aldehyde were determined by using the thiobarbituric acid (TBA) assay.
Reaction mixtures were diluted 10 times with demineralized water, and
5
0 µL of this dilution was added to 3 mL of TCA solution (7.5%
trichloroacetic acid, 0.1% propyl gallate, and 0.1% EDTA in water).
To these samples was added 3 mL of 0.02 M TBA, and the solutions
were heated for 30 min in a boiling water bath. After cooling of the
samples to room temperature, the amount of the TBA-MDA complex
was determined by measuring the absorbance at 532 nm, reduced by
the absorbance at 600 nm to correct for turbidity.
Separation of High and Low Molecular Weight Compounds. To
determine the absorbance of the high molecular weight fraction and
thus the melanoidin concentration, the low molecular weight compounds
were separated from the protein via Sephadex G25 disposable columns
Because of the importance of malondialdehyde as a lipid
oxidation product in vivo, most studies have been carried out
on the reaction of this ꢀ-dicarbonyl compound with human
proteins in physiological conditions. However, the reaction of
malondialdehyde with food proteins during food processing is
also of interest. In this respect, it has been shown that the
decrease of free lysine in milk as a result of heating can be due
in part to the reaction of proteins with malondialdehyde, and,
as a result, the determination of heat-induced damage in milk-
based products by furosine measurements can lead to an
underestimation (11). Therefore, this research was undertaken
to study the browning of casein with malondialdehyde in
comparison with the browning of casein induced by the well-
known R-dicarbonyl compound 2-oxopropanal (often referred
to as methylglyoxal), resulting from sugar degradation as well
as from lipid peroxidation (12). Kinetic modeling was used to
quantitatively describe the changes observed during the reaction.
(
PD-10, Sigma-Aldrich). A sample of 2.5 mL was brought on the
column and was eluted with 3.5 mL of water to collect the protein
fraction. With another 7.0 mL of water, the low molecular weight
fraction was eluted.
Analysis of Brown Compounds. The browning intensity of the
heated reaction mixtures was determined by measuring the absorbance
of the SDS-diluted samples at 420 nm with a spectrophotometer (Cary
5
0, Varian). The browning of the protein-free fraction was measured
without dilution and subtracted from the browning of the complete
reaction mixture to obtain protein browning.
Analysis of 2-Oxopropanal by HPLC. 2-Oxopropanal was deriva-
tized with o-phenylenediamine (OPD), according to a known procedure
(13), by adding 1 mL of Milli-Q water and 2 mL of 0.1 M
OPD/methanol to 1 mL of sample (80 mM samples were first diluted
10 times). This reaction was kept at room temperature overnight after
adjustment of the pH to 6.5. The samples were analyzed by HPLC
using a Lichrospher column (60 RP-Select B; 5 µm; 60 A; 250 × 4
mm) and a solvent gradient (A, water; B, acetonitrile; start 80/20 to
4
0/60 in 45 min). Standards of 2-methylquinoxaline (10–500 µg/mL)
MATERIALS AND METHODS
were analyzed to make the calibration curve (t ) 9.5 min).
R
Chemicals. All chemicals were of analytical grade. Sodium caseinate
91% protein, 4% water, 0.1% lactose, 3.8% ash, and 1.1% fat) was
obtained from DMV (Veghel, The Netherlands). 2-Oxopropanal (me-
thylglyoxal) was obtained as a 40% solution in water (Sigma-Aldrich,
Bornem, Belgium). A 100 mM solution of malondialdehyde (MDA)
was prepared by dissolving 1100 mg of 1,1,3,3-tetraethoxypropane
Kinetic Modeling. On the basis of the proposed reaction pathways
a kinetic model was built and translated into a mathematical model by
deriving coupled differential equations for each reaction step, as
explained previously (14). These equations contain rate constants as
parameters. For the studies done at various temperatures, the rate
constants were coupled to the reparametrized Arrhenius equation.
Reparametrization is necessary for statistical reasons. The advantage
of coupling the rate constants directly to the Arrhenius equation is that
all data at all temperatures studied are used at once to estimate activation
energies, resulting in much more precise parameter estimates (14). The
software package Athena Visual Workbench (www.athenavisual.com)
(
(
Acros Organics, Geel, Belgium) in 50 mL of 0.1 M HCl. This solution
was stirred and heated for 60 min at 50 °C in a water bath, upon which
a dark yellow color developed. The concentration of malondialdehyde
was determined by spectrophotometric measurements of the dilution
-
3
1
0
at 245 nm (acid solutions; ꢀ ) 13700) or at 267 nm (basic

Modification of Casein by Malondialdehyde
J. Agric. Food Chem., Vol. 56, No. 5, 2008 1715
was used for numerical integration of the coupled differential equations
as well as for parameter estimation. The model parameters, that is, the
rate constants and activation energies, were estimated by nonlinear
regression using minimization of the so-called determinant criterion.
This criterion replaces the commonly used least-squares minimization,
to comply with the statistical demands typical for multiresponse
modeling (14).
RESULTS AND DISCUSSION
The reaction of casein with different concentrations of MDA
was studied in a phosphate buffer at various conditions of pH
(
6, 7, 8) and temperature (90, 110, 120, 130 °C). Samples were
taken at regular time intervals (0–120 min). In the reaction
mixtures, the remaining free lysine residues (after derivatization
with o-phthaldialdehyde), the remaining malondialdehyde (mea-
sured as thiobarbituric acid reactive compounds) and protein
browning were determined. To quantify protein browning, an
approach used for melanoidin quantification was used. The
concentration of brown high molecular weight Maillard reaction
products, named melanoidins, can be quantified by spectropho-
tometrically measured browning. Brands et al. (15) have shown
that, by means of the molar extinction coefficient, the browning
of the melanoidins resulting from heated glucose-casein and
fructose-casein model systems could be converted into the
concentration of sugar incorporated. The molar extinction
coefficient was calculated by measuring the concentration of
Figure 2. Kinetic models used for the reaction of lysine residues of casein
(Cas-Lys) with malondialdehyde (MDA).
characterized or quantified. In the literature, it has been reported
that malondialdehyde slowly undergoes self-condensation or
cleavage in aqueous solution at room temperature with the
formation of various reaction products, such as 2,4-bishy-
droxymethylene-3-methylpentanedial (17).
Influence of Temperature. The reaction of casein with
malondialdehyde in the described reaction conditions occurred
quite rapidly, and the results and model fits were well in line
with the hypothesis that 1 mol of malondialdehyde reacts with
2 mol of lysine residues. This hypothesis was not disproved at
the different temperatures investigated (90, 110, 130 °C). In
Figure 3, the experimental data and the model fits are shown.
The model was fitted to all of the data at the same time, taking
into account the temperature dependency by including the
reparametrized Arrhenius equation at the four heating temper-
atures simultaneously. As can be seen in Figure 3B, the decrease
of malondialdehyde with time was described well by the model
and showed a temperature dependency as expected. In contrast,
the courses of the decrease of lysine were quite similar at 110,
120, and 130 °C (Figure 3A). Also, the browning of casein
was not significantly different for these higher temperatures
(Figure 3C). The fact that the same behavior was noted in both
cases is reassuring because in the kinetic model the two are
directly linked. One remark to be made is that also at t0 some
protein browning was measured, which could not be excluded
but developed in the time period between the mixing of the
reagents and the analytical measurements (storage at 0 °C). In
Table 1, the different rate constants for this set of reactions are
shown. The degradation reaction of malondialdehyde showed
a temperature dependency (k2) that is in accordance with normal
chemical reactions. The value of the activation energy found
for lysine decrease and casein browning (32 kJ/mol) is, however,
quite low for a chemical reaction, reflecting that the reactants
lysine and malondialdehyde react readily already at lower
temperatures. The higher temperature dependency of the second
reaction, the degradation of malondialdehyde, must thus be due
to reactions other than with casein. It can be deduced that the
estimate of rate constant k1 is much more precise than the
estimate of k2, which is due to the fact that the concentration of
the degradation products was not experimentally measured.
The browned casein displayed fluorescence (λexc ) 377 nm;
λem ) 459 nm) that was in accordance with the values reported
for 3-amino-1-iminopropene cross-links (λexc ) 370-400 nm;
λem ) 450-470 nm), which supports the hypothesis of model
A in Figure 2 (6).
1
4
C-labeled sugar incorporated into the browned protein and
remained constant in time. This experimental value was not
significantly different for the melanoidins resulting from the
reaction of glucose or fructose with casein and was determined
as 500 L/(mol·cm) (λ ) 420 nm). Calculation of the extinction
coefficient of melanoidins derived from glucose with various
amino acids showed that this extinction coefficient is dependent
on the amino acid involved (16) and varied roughly between
500 and 1000 L/(mol ·cm). The browned protein resulting from
the reaction of casein with malondialdehyde is a so-called
advanced lipoxidation end product (ALE) but can to some extent
be considered as melanoidin-like. Therefore, the quantification
approach described was applied for the quantification of casein
browning after reaction of casein with malondialdehyde or
2-oxopropanal.Amolarextinctioncoefficientof1.000L/(mol ·cm)
was applied, supposing that for one C6-sugar (glucose or
fructose), two C3-compounds can be incorporated into the casein
structure. Solutions of 3% of caseinate (corresponding to 15
mM of lysine residues) in a phosphate buffer (pH 7) were
reacted with 8 mM of malondialdehyde at 120 °C. In such a
simple model system of casein and MDA it can be assumed
that 1 mol of malondialdehyde (containing two reactive carbonyl
functions) reacts with 2 mol of lysine residues and as such leads
to a cross-linking of the protein (2). In addition, malondialde-
hyde can also undergo self-condensation by aldol reactions.
Therefore, a degradation reaction was included in the model to
account for additional losses of malondialdehyde (model A,
Figure 2). In the headspace of the reaction mixtures and in
extracts of control MDA heating experiments, various conden-
sation products of MDA were detected such as benzene,
benzaldehyde, traces of acetaldehyde, and some unidentified
compounds. To be able to determine browning of the high
molecular weight fraction, a separation of high and low
molecular weight compounds was carried out. Because the
absorbance at 420 nm of the low molecular weight fraction did
not change during the course of the reaction, browning could
not be used to monitor the formation of low molecular weight
reaction products. Because of their multitude and low amounts,
the malondialdehyde degradation products were not further

1716 J. Agric. Food Chem., Vol. 56, No. 5, 2008
Adams et al.
Figure 3. Model fits (drawn lines) based on kinetic model A (Figure 2) (pred ) predicted) and experimental data (obs ) observed) for the model
reaction of 3% casein with 8 mM malondialdehyde at different temperatures and pH 7.
Table 1. Rate Constants of the Reaction of 8 mM Malondialdehyde with
% Casein as a Function of Temperature and pH, As Calculated with
malondialdehyde degradation reaction was not significantly
different for the three pH values tested. Here as well, a high
imprecision of the estimate of the rate constant k2 is noted.
Influence of Concentration of Malondialdehyde. To evalu-
ate the kinetic model at different malondialdehyde concentra-
tions, the reaction between malondialdehyde was performed at
3
Kinetic Model A (Figure 2)
a
rate constants
k1 (L mmol min-1 k2 (min-1
)
2
-2
reaction conditions
)
-
-
4
-3
9
1
1
1
0 °C, pH 7
10 °C, pH 7
20 °C, pH 7
30 °C, pH 7
(0.57 ( 0.08) × 10
(0.99 ( 0.07) × 10
(0.86 ( 0.67) × 10
1
20 °C, pH 7, and 3% of caseinate with malondialdehyde
4
-3
(4.1 ( 1.4) × 10
concentrations of 1, 8, and 15 mM. At 1 mM malondialdehyde
the model fitted the data quite well, considering that 1 mol of
malondialdehyde reacted with 2 mol of lysine residues, as was
the case for 8 mM malondialdehyde. However, at higher
malondialdehyde concentration (15 mM), this assumption was
no longer valid, as the experimentally measured final concentra-
tion of browned protein was higher than the maximum
concentration possible as simulated by model A (being half the
initial lysine concentration or 7.5 mM). At elevated malondi-
aldehyde concentrations, it is more likely that 1 mol of lysine
residues reacts preferably with 1 mol of malondialdehyde
(formation of structure 1), rather than the cross-linking reaction
occurring at lower malondialdehyde concentrations. However,
the experimental concentrations of browned casein exceeded
even the initial lysine concentration, indicating that other
reactions, for example, with other amino acid residues such as
arginine or histidine, may contribute to protein browning.
Incorporating these assumptions in the model still did not result
in a good fit, because the experimental lysine concentration
decreased very quickly, whereas the experimental melanoidin
concentration increased much more slowly. This might indicate
the formation of an intermediate, which is nevertheless difficult
to explain theoretically, and also did not improve the model fit.
Such high malondialdehyde concentrations are in fact not
realistic, neither physiologically nor related to food systems.
As many other reactive carbonyl compounds are present in real
food systems, these will compete with malondialdehyde for the
reaction with nucleophilic amino groups. In this respect it has
been shown that the reaction of malondialdehyde and alkanals
with proteins proceeds synergistically with the formation of a
dihydropyridine cross-link. Such a dihydropyridine cross-link,
7, can be formed from the reaction of 1 mol of lysine residues,
2 mol of malondialdehyde, and 1 mol of acetaldehyde, being a
hydrolytic degradation product of malondialdehyde (4). To
-
-
4
4
-3
(1.3 ( 0.1) × 10
(1.6 ( 0.2) × 10
32 ( 5.9
(8.5 ( 1.5) × 10
-3
(17 ( 3.2) × 10
Ea (kJ/mol)
91 ( 26
-
4
-3
pH 6, 120 °C
pH 7, 120 °C
pH 8, 120 °C
(4.3 ( 0.57) × 10
(6.8 ( 4.1) × 10
-
4
-3
(1.3 ( 0.1) × 10
(1.7 ( 0.49) × 10
(8.5 ( 1.5) × 10
-4
-3
(12 ( 6.1) × 10
a
The precisions indicated are 95% higher posterior densities (HPD), the
Bayesian equivalents of 95% confidence intervals.
Influence of pH. Study of the reaction between 8 mM
malondialdehyde and 3% casein at 120 °C at different pH values
showed that the reaction proceeded more rapidly at lower pH
(
pH 6), but also to a lesser extent at higher pH values (pH 8);
the pH values given are the values at room temperature, not at
the heating temperature (Figure 4). The nucleophilic addition
of the ꢀ-amino group of lysine toward the MDA carbonyl
function can be acid as well as base catalyzed. At pH 8,
however, the model did not fit the observed data points very
well. Study of the residuals showed that not all of the
information which the data points contain was explained by the
model. A possible explanation can be related to the preferential
tautomerization at higher pH of compound 1, resulting from
the first step of the reaction of lysine with malondialdehyde,
which reduces the electrophilicity of the second carbonyl
function and thus the susceptibility for a second nucleophilic
attack of lysine. However, an alternative model, in which 1 mol
of malondialdehyde was supposed to react with 1 mol of lysine
residues, did not result in a better fit of the data. Also, a
reversible formation of malondialdehyde degradation products
and browned protein did not improve the fit of the model to
the experimental data at pH 8.
Table 1 shows the values of the rate constants at the different
pH values. This table shows that the rate constant k2 of the

Modification of Casein by Malondialdehyde
J. Agric. Food Chem., Vol. 56, No. 5, 2008 1717
Figure 4. Model fits (drawn lines) based on kinetic model A (Figure 2) (pred ) predicted) and experimental data (obs ) observed) for the reaction of
% casein with 8 mM malondialdehyde at different pH values and 120 °C.
3
Figure 5. Model fits (drawn lines) based on kinetic model B (Figure 2) (pred ) predicted) and experimental data (obs ) observed) for the reaction of
% casein with different concentrations of malondialdehyde at 120 °C and pH 7.
3
Table 2. Rate Constants of the Reaction of Malondialdehyde with
Caseinate at 120 °C, pH 7, as a Function of MDA Concentration,
Calculated with Kinetic Model B (Figure 2)
model this reaction the kinetic model was extended with a third
reaction (model B, Figure 2). Applying this model to the data
obtained from the reaction of casein with 15 mM malondial-
dehyde showed a considerably improved fit (Figure 5). The
model fit deviated still slightly from the observed protein
browning. When kinetic model B was applied to the reaction
of casein with 8 mM malondialdehyde, the quality of the fit
and the values of k1 and k2 did not differ significantly from
rate constants
MDA
mM) k1 (L mmol min-1
2
-2
k2 (min-1
k3 (L mmol min-1
3
-3
(
)
)
)
-5
-3
-3
-3
1
8
15
(3.2 ( 1.0) × 10
(11 ( 2.1) × 10
(4.8 ( 1.7) × 10
(9.6 ( 4.0) × 10
(8.2 ( 4.0) × 10
(0.2 ( 5.9) × 10
-5
-9
-6
6.8 × 10 ( 5.0 × 10
-9
3
-5
-5
model A, and a very low value of k3 was found (7 × 10
L
(1.3 ( 0.5) × 10
-3
-1
mmol min ) with a confidence interval that was much higher
than the estimate itself. This means that this parameter is
redundant for the data obtained at 1 and 8 mM. It makes sense
only for the results obtained with 15 mM. The degradation rate
constant k2 was not significantly different for the three malon-
dialdehyde concentrations (Table 2).

1718 J. Agric. Food Chem., Vol. 56, No. 5, 2008
Adams et al.
Figure 6. Comparison of the experimental data for the reaction of 3% casein with malondialdehyde (8 mM MDA) and 2-oxopropanal (7 mM 2-OP) at
120 °C and pH 7.
For all of these experiments, translation of protein browning
mechanism. It is well-known that 2-oxopropanal reacts prefer-
ably with arginine residues as compared to lysine residues (19),
but the concentration of arginine residues was not determined
in this study. This reaction may be partly responsible for casein
browning at 120 °C (where browning was more significant).
The fluorescence recorded for the 2-oxopropanal modified casein
into concentrations of C3-compounds incorporated by means
of the molar extinction coefficient was used to estimate molar
concentrations, which are needed for kinetic modeling. Earlier
investigations have shown the usefulness of this approach,
calculating extinction coefficients between 500 and 1000
L/(mol·cm) (λ ) 420 nm) for various model systems. Appar-
ently, the chromophores formed in the nondialyzable melanoi-
dins in the early stages of the Maillard reaction are similar to
those at the later stages, and their formation is not very sensitive
to the reaction conditions (18). However, in the course of this
study of the casein-malondialdehyde condensation, two dif-
ferent chromophores, namely, the 3-amino-1-iminopropene
cross-link 2 and the dihydropyridine cross-link 3, were hypoth-
esized, resulting from the condensation of one lysine residue
with one and three malondialdehyde units, respectively. On the
other hand, the second cross-link becomes significant only at
concentrations above 8 mM. This is not completely consistent
with the background of using the molar extinction coefficient,
but enabled a successful prediction of casein browning induced
by malondialdehyde. Although the value of the extinction
coefficient that was used to transform the browning of casein
into number of moles of product remains an uncertain element,
this was the best approach available to tackle the formation of
brown color quantitatively in a kinetic model. Modifying the
numerical value of the extinction coefficient resulted in differ-
ences in the numerical values of the rate constants, but not in
the general trend. In other words, the general conclusions drawn
remained valid.
(
λexc ) 318 nm; λem ) 389 nm) was in agreement with
fluorescence values reported for the formation of the so-called
argpyrimidine” cross-link, resulting from the modification of
“
arginine residues by the condensation product of two 2-oxo-
propanal molecules (λexc ) 320 nm; λem ) 382 nm) (20).
Comparison of the reaction course of casein with malondi-
aldehyde and 2-oxopropanal at 120 °C (pH 7) showed that the
decreases of lysine were the same in both systems (Figure 6).
The decrease in 2-oxopropanal concentration, however, was
considerably faster than the decrease of malondialdehyde. This
can be explained by the stabilization of the ꢀ-dicarbonyl
structure by resonance at a pH higher than the pKa (4.46),
making malondialdehyde less prone to self-condensation and
other degradative reactions than is the case for 2-oxopropanal.
Casein browning, on the other hand, occurred considerably more
quickly with malondialdehyde than with 2-oxopropanal. This
illustrates the importance of this lipid oxidation product in the
induction of protein browning as compared to the better known
Maillard browning caused by reactive sugar degradation prod-
ucts such as 2-oxopropanal.
Considering that the characteristic fluorescence of advanced
glycation endproducts (AGEs) (λexc ) 370 nm; λem ) 440 nm)
has been widely used as an indicator of the level of AGE-
modified proteins (21), the contribution of malondialdehyde to
these fluorescence values cannot be neglected. Whereas the
fluorescence resulting from the modification of casein by
Reaction of 2-Oxopropanal with Casein. In comparison
with the reaction of casein with malondialdehyde, the reaction
of casein with 2-oxopropanal was studied, applying similar
reaction conditions. At 120 °C, the concentration of 2-oxopro-
panal was depleted almost immediately (data not shown).
Therefore, the reaction between casein and 2-oxopropanal was
studied at 60 °C. At this temperature, the concentration of lysine
residues decreased only slightly (about 2 mM), whereas the
concentration of 2-oxopropanal (initial concentrations of 10 and
2
-oxopropanal in this research was substantially different from
the typical values reported for AGEs, the fluorescence of the
MDA-modified casein (ALE) contributes to the so-called AGE
browning. Casein-bound fluorescence reported as a result of the
advanced Maillard reaction of lactose/caseinate or glucose/
caseinate was characterized by maximum excitation and emis-
sion wavelengths of 347 and 415 nm, respectively (22).
2
0 mM) decreased steadily and continuously until 120 min.
Protein browning was also very low at 60 °C and corresponded
to a concentration of 2 mM of C3-compound incorporated.
These results suggest that an important part of 2-oxopropopanal
is lost in the formation of low molecular weight degradation
products.
The different models evaluated for the reaction of casein with
malondialdehyde could not be applied to the model reaction of
casein with 2-oxopropanal, which followed a different reaction
In conclusion, this study has shown that kinetic modeling is
a useful tool to unravel reaction mechanisms. Clearly, the
contribution of lipid oxidation products, such as malondialde-
hyde, to protein browning (both in food and in vivo) can be
substantial and needs to be taken into account in future studies.

Modification of Casein by Malondialdehyde
J. Agric. Food Chem., Vol. 56, No. 5, 2008 1719
LITERATURE CITED
(13) Hofmann, T. Quantitative studies on the role of browning
precursors in the Maillard reaction of pentoses and hexoses with
L-alanine. Eur. Food Res. Technol. 1999, 209, 113–121.
(14) Van Boekel, M. A. J. S. Statistical aspects of kinetic modelling
for food science problems. J. Food Sci. 1996, 61, 477–485.
(15) Brands, C. M. J.; Wedzicha, B. L.; van Boekel, M. A. J. S.
Quantification of melanoidin concentration in sugar-casein sys-
tems. J. Agric. Food Chem. 2002, 50, 1178–1183.
(16) Leong, L. P. Modelling of the Maillard reaction involving more
than one amino acid. Ph.D. Thesis, Procter Department of Food
Science, University of Leeds, U.K., 1999.
(17) Gomezsanchez, A.; Hermosin, I.; Lassaletta, J. M.; Maya, I.
Cleavage and oligomerization of malondialdehyde. Tetrahedron
1993, 49, 1237–1250.
(
1) Zamora, R.; Hidalgo, F. J. Coordinate contribution of lipid
oxidation and Maillard reaction to the nonenzymatic food brown-
ing. Crit. ReV. Food Sci. Nutr. 2005, 45, 49–59.
(
2) Gray, J. I. Measurement of lipid oxidationsreview. J. Am. Oil
Chem. Soc. 1978, 55, 539–546.
(
3) Slatter, D. A.; Bolton, C. H.; Bailey, A. J. The importance of
lipid-derived malondialdehyde in diabetes mellitus. Diabetologia
2
000, 43, 550–557.
(
(
4) Slatter, D. A.; Murray, M.; Bailey, A. J. Formation of a
dihydropyridine cross-link derived from malondialdehyde in
physiological systems. FEBS Lett. 1998, 421, 180–184.
5) Girón-Calle, J.; Alaiz, M.; Millán, F.; Ruiz-Gutierrez, V.; Vioque,
E. Bound malondialdehyde in foods: bioavailability of N,N′-di-
(
18) Martins, S.I.F.S; van Boekel, M. A. J. S. Melanoidins extinction
coefficient in the glucose/glycine Maillard reaction. Food Chem.
(4-methyl-1,4-dihydropyridine-3,5-dicarbaldehyde)lysine. J. Agric.
2
003, 83, 135–142.
Food Chem. 2003, 51, 4799–4803.
(
19) Lederer, M. O.; Klaiber, R. G. Cross-linking of proteins by
Maillard processes: characterization and detection of lysine-arginine
cross-links derived from glyoxal and methylglyoxal. Bioorg. Med.
Chem. 1999, 7, 2499–2507.
(
6) Itakura, K.; Uchida, K.; Osawa, T. A novel fluorescent malon-
dialdehyde-lysine adduct. Chem. Phys. Lipids 1996, 84, 75–79.
7) Slatter, D. A.; Paul, R. G.; Murray, M.; Bailey, A. J. Reactions
of lipid-derived malondialdehyde with collagen. J. Biol. Chem.
(
(
20) Shipanova, I. N.; Glomb, M.A.; Nagaraj, R. H. Protein modifica-
tion by methylglyoxal: chemical nature and synthetic mechanism
of a major fluorescent adduct. Arch. Biochem. Biophys. 1997, 344,
1
999, 274, 19661–19669.
(
8) Slatter, D. A.; Avery, N. C.; Bailey, A. J. Identification of a new
cross-link and unique histidine adduct from bovine serum albumin
incubated with malondialdehyde. J. Biol. Chem. 2004, 279, 61–
2
9–36.
(
21) Obayashi, H.; Nakano, K.; Shigeta, H.; Yamaguchi, M.; Yoshi-
mori, K.; Fukui, M.; Fujii, M.; Kitagawa, Y.; Nakamura, N.;
Nakamura, K.; Nakazawa, Y.; Ienaga, K.; Ohta, M.; Nishimura,
M.; Fukui, I.; Kondo, M. Formation of crossline as a fluorescent
advanced glycation end product in Vitro and in ViVo. Biochem.
Biophys. Res. Commun. 1996, 226, 37–41.
6
9.
(
9) Requena, J. R.; Fu, M. X.; Ahmed, M. U.; Jenkins, A. J.; Lyons,
T. J.; Baynes, J. W.; Thorpe, S. R. Quantification of malondial-
dehyde and 4-hydroxynonenal adducts to lysine residues in native
and oxidized human low-density lipoprotein. Biochem. J. 1997,
3
22, 317–325.
(
22) Morales, F. J.; van Boekel, M. A. J. S. A study on advanced
Maillard reaction in heated casein/sugar solutions: fluorescence
accumulation. Int. Dairy J. 1997, 7, 675–683.
(
10) Romero, A. R.; Doval, M. M.; Sturla, M. A.; Judis, M. Antioxidant
behaviour of products resulting from beef sarcoplasmic proteins-
malondialdehyde reaction. Eur. J. Lipid Sci. Technol. 2005, 107,
9
03–911.
Received for review August 8, 2007. Revised manuscript received
January 7, 2008. Accepted January 9, 2008. We are indebted to the
Research Foundation Flanders (FWO-Vlaanderen) for financial support
and for a postdoctoral fellowship of A.A.
(
11) Morales, F. J.; Jiménez-Pérez, S. Effect of malondialdehyde on
the determination of furosine in milk-based products. J. Agric.
Food Chem. 2000, 48, 680–684.
12) Miakar, A.; Spiteller, G. Reinvestigation of lipid peroxidation of
linolenic acid. Biochim. Biophys. Acta 1994, 1214, 209–220.
(
JF072385B