Formation of radicals during heating lysine and glucose in solution with an intermediate water activity

Article abstract of DOI:10.3109/10715762.2013.812206

Heating glucose with lysine under alkaline conditions (pH 7.0-10.0) was found to take place with consumption of oxygen together with formation of brown-colored compounds. Highly reactive intermediary radicals were detected when lysine and glucose were heated at intermediate water activity at pH 7.0 and 8.0. The detection was based on initial trapping of highly reactive radicals by ethanol followed by spin trapping of 1-hydroxyethylradicals with α-(4-pyridyl N-oxide)-N-tert-butylnitrone (POBN) and Electron Spin Resonance (ESR) spectroscopy. The generation of reactive intermediary radicals from the Maillard reactions was favored by enhancing alkaline conditions (pH 8.0) and stimulated by presence of the transition metal ion Fe²⁺. The stability of the nitrone spin traps, N-tert-butyl-α-phenylnitrone and POBN was examined in buffered aqueous solutions within the pH range 1-12, and found to be less temperature dependent at acidic pH compared to alkaline conditions. A low rate (kobs) of hydrolysis of POBN was found at the used experimental conditions of 70°C and pH 7.0 and 8.0, which made this spin trap method suitable for the detection of radicals in the Maillard reaction system.

Full text of DOI:10.3109/10715762.2013.812206

Free Radical Research, August 2013; 47(8): 643–650
©
2013 Informa UK, Ltd.
ISSN 1071-5762 print/ISSN 1029-2470 online
DOI: 10.3109/10715762.2013.812206
ORIGINAL ARTICLE
Formation of radicals during heating lysine and glucose in solution with an
intermediate water activity
J. Yin, M. L. Andersen, M. K. Thomsen, L. H. Skibsted & R. V. Hedegaard
Department of Food Science, Food Chemistry, University of Copenhagen, Frederiksberg C, Denmark
Abstract
Heating glucose with lysine under alkaline conditions (pH 7.0–10.0) was found to take place with consumption of oxygen together with
formation of brown-colored compounds. Highly reactive intermediary radicals were detected when lysine and glucose were heated at
intermediate water activity at pH 7.0 and 8.0. The detection was based on initial trapping of highly reactive radicals by ethanol followed
by spin trapping of 1-hydroxyethylradicals with α-(4-pyridyl N-oxide)-N-tert-butylnitrone (POBN) and Electron Spin Resonance (ESR)
spectroscopy. The generation of reactive intermediary radicals from the Maillard reactions was favored by enhancing alkaline conditions
2
ϩ
(
pH 8.0) and stimulated by presence of the transition metal ion Fe . The stability of the nitrone spin traps, N-tert-butyl-α-phenylnitrone
and POBN was examined in buffered aqueous solutions within the pH range 1–12, and found to be less temperature dependent at acidic pH
compared to alkaline conditions. A low rate (k ) of hydrolysis of POBN was found at the used experimental conditions of 70°C and pH 7.0
obs
and 8.0, which made this spin trap method suitable for the detection of radicals in the Maillard reaction system.
Keywords: Maillard reactions, intermediary radicals, ESR, spin trapping, oxidation, stability of spin traps
Introduction
have been detected by electron spin resonance (ESR)
spectroscopy formed as part of the MR. The radicals
were found to be generated prior to the Amadori rear-
rangement by the so-called Namiki pathway probably
originating from degradation of the Schiff base [5,6].
Formation of these dialkylpyrazinium radical cations is
pH dependent with low levels formed at neutral pH and
with increasing levels up to pH 11 [7]. However, other
highly reactive intermediary radicals derived from oxi-
dative pathways involving metal-catalyzed degradation
of central intermediates also seem to be involved in pro-
longation of MR [8,9]. Less detailed information exists
about their nature and formation, due to difficulties
related with detection of short-lived and highly reactive
radicals in complex systems. The spin trapping technique
can be used to study short-lived radicals. It is a tech-
nique, where short-lived radicals are trapped and
converted to stable spin adducts with sufficient life-
time for detection by ESR [10]. The majority of spin
traps are nitrones such as α-(4-pyridyl N-oxide)-N-tert-
butylnitrone (POBN) and α-(4-pyridyl N-oxide)-N-tert-
butylnitrone (PBN), which can react with a broad range
of radicals, that is, C-, N-, O-, and S-centered radicals,
and thereby produce long-lived nitroxyl radicals as
spin adducts. Most studies involving the spin trapping
technique have been performed under physiological
conditions, that is, pH 7.0–7.4, and in the temperature
range 20–40°C [11]. However, detection of radicals as
intermediates during oxidative degradations or MR in
The Maillard reactions (MR) generate aroma, flavor com-
pounds, and brown-colored polymers responsible for the
characteristic taste and appearance of heat-processed
foods. However, compounds known to induce a risk to
human health are also formed concurrently. This includes
so-called Advanced Glycation End products (AGEs),
which are also formed at slower rates and at lower tem-
peratures in the body. The AGEs formed either endoge-
nously or derived from the diet have been associated with
development of diabetes and cardiovascular diseases [1,2].
Rate of formation and final content of AGEs in food are
influenced by many factors such as pH, heating tempera-
ture, time, water activity (a ), as well as availability of
w
precursors, which most often are reducing sugars, free
amino acids, or proteins.
The pathways of the MR in food have been described
and studied extensively since its discovery more than 100
years ago by the French chemist Luis Camille Maillard
[
3]. However, many mechanistic details remain unre-
solved due to analytical difficulties related to the com-
plex combinations of intermediates and products, and
the involvement of high temperatures used during heat
processing of foods. At low pH (pH 5.0) the MR is
mainly taking place by polar reactions (ionic pathways),
while at alkaline conditions oxidative conditions are
increasingly favored thereby promoting formation of
intermediary radicals [4]. Stable intermediary radicals
Correspondence: Rikke V. Hedegaard, Department of Food Science, Food Chemistry, University of Copenhagen, Rolighedsvej 30, 1958
Frederiksberg C, Denmark. Tel: ϩ45 35333268. Fax: ϩ45 35333344. E-mail: rvh@life.ku.dk
(
Received date: 13 March 2013; Accepted date: 3 June 2013; Published online: 28 June 2013)

644 J. Yin et al.
heat-processed foods and beverages is becoming rele-
vant, and the high temperatures together with the fact
that food systems have pH values ranging from 2.0 to 8.0
makes it necessary to test the stability of spin traps under
these conditions. Hydrolysis of nitrone spin traps is likely
to be enhanced under such conditions leading to the for-
mation of an aldehyde and a hydroxylamine (reaction 1).
hydrolysis (pH 1.0–5.0) was also studied at 20 and 60°C.
Whereas the alkaline degradation of POBN also was fol-
lowed at pH 10.0–12.0 at two temperatures: 60 and 70°C.
Experiments running for less than 24 h were started
by adding 15 μL of stock solutions of POBN or PBN
(6.0 mM in ethanol) to 2.25 mL preheated buffer in a
cuvette placed in a thermostated cell holder in a Hewlett-
Packard HP 8453 diode array spectrophotometer (Palo
Alto, CA, USA). Experiments running for more than 24 h
were carried out by adding 200 μL of stock solutions of
POBN or PBN (6.0 mM in ethanol) to 25 mL preheated
buffer in capped 25 ml blue-cap bottles placed in a ther-
mostated water bath. Samples (2.5 mL) were taken at
varying intervals and the UV-spectra were recorded. The
decay of PBN was followed by measuring the absorbance
at 287 nm, while formation of benzaldehyde was followed
at 250 nm. Similarly, the decay of POBN was followed by
measuring the absorbance at 330 nm, while formation of
the aldehyde was followed at 262 nm.
ArCHϭN(O)RϩH O→ArCHOϩHN(OH)R
(1)
2
The aim of this study was to examine formation of unsta-
ble intermediary radicals during the reactions of glucose
with lysine at an elevated temperature [12]. This was used
as a model of a low molecular weight (LMW) food-related
MR system, in order to explore the effects of pH, water
activity, a , and presence of transition metals like iron on
w
radical pathways of the MR. Furthermore, the thermal
hydrolysis of the spin traps PBN and POBN at both acidic
and alkaline conditions were characterized with the aim
of establishing their stabilities under the used reaction
conditions.
Development of a model system with Fenton reaction
A model system appropriate for detection of radicals
Material and methods
formed during heating at intermediary a conditions was
w
developed by testing different solvents. MES buffer
Chemicals
(
(
(
5 mM) adjusted to pH 7.0 or 8.0 was mixed with glycerol
30% and 60%), ethylene glycol (30% and 60%) or ethanol
5%, 30%, and 60%). Radicals were generated in the
4
-(2-Hydroxyethyl)-1-piperazineethanesulfonic
acid
(
(
(
HEPES) (99.5%), 2-(N-morpholino) ethanesulfonic acid
MES) (99.5%), POBN (95%), PBN (purity 95%), L-lysine
98%), and D-(ϩ)-glucose (99.5%) was from Sigma-
model systems by the Fenton reaction after addition of
FeSO ⋅7H O (0.1 mM) and H O (0.3 mM) and detected
4
2
2
2
by ESR as spin adducts after reaction with POBN, which
was added to a final concentration of 10 mM. Samples
Aldrich (St. Louis, MO, USA). 5-(Diethoxyphosphoryl)-
5
-methyl-1-pyrroline-N-oxide] (DEPMPO) (99%) was
(
100 μL) were transferred into a 2 mL Eppendorf tube
from Alexis Biochemicals (Lausen, Switzerland). Ethanol
and heated in a glycerol bath at 70Ϯ1.0°C for 10, 20, 30,
0, and 50 min prior to analysis.
(
96%) was from Kemetyl A/S (Køge, Denmark). 3-
4
Deoxyglucosone (CAS 4084-27-9) (95%) was from Santa
Cruz Biotechnology (Santa Cruz, CA, USA). Iron(II) sul-
fate heptahydrate (FeSO ⋅7H O) and hydrogen peroxide
Preparation of LMW MR model system
4
2
(
30%) were from Merck (Darmstadt, Germany). Water
LMW MR model systems were prepared by dissolving
glucose (0.1 M), lysine (0.1 M), or the mixture of glucose
was purified through a MilliQ purification train (Milli-
pore, Bedford, MA, USA). The concentration of hydro-
gen peroxide was determined spectrophotometrically
(
0.1 M) with lysine (0.1 M) in HEPES buffer adjusted
to pH 7.0 and pH 8.0 with 30% ethanol (a ϭ0.88)
Ϫ1
Ϫ1
w
(
ε
ϭ39.4 M ⋅cm ). MES buffer (5 mM) and HEPES
240
together with POBN (10 mM). The model systems with
buffer (200 mM) adjusted to ionic strengthϭ0.16 were
prepared from the corresponding analytical grade chemi-
cals using MilliQ water and adjusted to pH 7.0 and 8.0.
3
-deoxyglucosone (0.04 M) were prepared by dissolving
in HEPES buffer adjusted to pH 7.0 and 8.0 with 30%
ethanol (a ϭ0.88) together with POBN (10 mM). Iron
w
was added as FeSO ⋅7H O (0.1 mM) to the model sys-
4
2
Kinetic studies of spin trap hydrolysis
tems. Samples (100 μL) were transferred into a 2 mL
Eppendorf tube and placed in a glycerol bath at 70Ϯ1.0°C
for 10, 20, 30, 40, and 50 min prior to analysis.
Kinetic experiments of thermal and acid/alkaline hydroly-
sis of POBN and PBN were performed by dissolving
4
0
0–50 μM of the spin traps in buffers (concentration
Water activity (a ) measurements
w
.10–0.60 M) made from either phosphoric acid, dichlo-
roacetic acid, formic acid, citric acid, and acetic acid
for acidic conditions, and sodium hydroxide and potas-
sium hydroxide, for alkaline conditions. The ionic
strength of all buffers was adjusted to 1.0 with potassium
chloride. The hydrolysis of POBN and PBN were followed
at the pH ranging from 1.0 to 12.4 at 40°C. The acidic
The a was measured at room temperature by Aqua Lab
w
from ADAB Analytical Devices Ab (Stockholm, Sweden)
or calculated by the Raoult’s Law [13] for the solutions
with 30% and 60% ethanol: a ϭn/(nϩn′), where
w
nϭmoles of water and n′ϭmole of solute ethanol. All
measurements were performed in duplicates.

Reactive intermediary radicals in the Maillard reactions 645
pH measurements
during heating solutions containing either glucose, lysine,
or glucose mixed with lysine. No substantial oxygen
depletion was observed during heating of glucose alone.
Oxygen was consumed during heating of the lysine solu-
tion and the rate of oxygen consumption had a maximal
rate at pH 8.0 (Figure 1). The rate of oxygen consump-
tion in the mixed glucose and lysine solution increased
when enhancing pH from 7 to 10 (Figure 1). During
heating of the mixed lysine glucose system brown-
colored products were also generated that absorbed light
at 420 nm and this formation of brown-colored MR
products increased with pH with a maximum at pH 9
pH was measured in the samples with a Metrohm
6
.0234.100 combination glass electrode from Hamilton
Bonaduz AG (Bonaduz, Switzerland) connected to a
Metrohm 713 pH-meter from Metrohm (Herisau,
Switzerland), which was calibrated at the relevant tem-
peratures with standard buffers of pH 4.01 and 7.00. All
measurements were performed in duplicates.
ESR spectroscopy
Samples were transferred to 50 μL ESR micropipettes
from BLAUBRAND IntraMark (Wertheim, Germany)
and placed in the resonator of the Miniscope MS 200 ESR
spectrometer from Magnettech GmbH (Berlin, Germany).
The following ESR parameters were used: center field,
(
Figure 1). Heating either glucose or lysine alone did
not generate colored compounds under the same condi-
tions. The results suggest the MR between glucose and
lysine is coupled to oxidation reactions, since the high
rates of oxygen consumption also occurred at pH 9 and
3
357 G; sweep width, 100 G; sweep time, 30 s; microwave
1
0 where the formation of brown compounds was most
power, 10 mW; and modulation amplitude, 1000 mG. The
middle duplicate peak-to-peak amplitude of the ESR sig-
nal of the spin adducts was measured by the Analysis 2.02
software program (ESR applications, Berlin, Germany).
All measurements were performed in duplicates.
pronounced.
Stability of spin traps
Studying the possible radical formation at high tempera-
tures in acidic or alkaline solutions requires knowledge
about the stability of spin traps under these conditions.
The stability of the two spin traps PBN and POBN at
varying temperatures and pH was studied by following
Oxygen consumption assay
The oxygen consumption was carried out with lysine
(
1.0 M), glucose (1.0 M), or glucose (1.0 M) with lysine
(
1.0 M) dissolved in MilliQ water during heating at 45°C.
0
0
0
0
0
0
.10
.08
.06
.04
.02
.00
pH was adjusted with HCl or NaOH to pH 7.0, 8.0, 9.0 or
1
0.0 using the buffer capacity of lysine itself. A HQ30d
electrode connected to HQd meter (HACH LANGE,
Germany) was used for oxygen measurements. The elec-
trode was calibrated before use by air-saturated MilliQ
water at 25°C. The measurement was started immediately
after placing the electrode in 20 mL of sample. Measure-
ment time was adjusted according to pH due to a
pH dependent difference in oxygen consumption. All
measurements were performed in duplicates.
Measurement of brown polymers
0
0
0
0
.0008
.0006
.0004
.0002
The change of absorbance at 420 nm was used as a mea-
sure for development of brown products from lysine
(
1.0 M), glucose (1.0 M) or glucose (1.0 M) with lysine
1.0 M) dissolved in MilliQ water. pH was adjusted with
(
HCl or NaOH to pH 7.0, 8.0, 9.0, or 10.0 using the
buffer capacity of lysine. The samples were heated at 45°C
and the absorbance measured in the spectral range
2
50–500 nm by an HP8453 UV/VIS diode array spectro-
photometer from Hewlett-Packard Co. (Palo Alto, CA).
All measurements were performed in duplicates.
0.0000
7
8
9
10
pH
Figure 1. Above: Rate of oxygen consumption during heating
Results
(at 45°C) in a closed system with lysine (1.0 M) (■), or glucose
(
1.0 M) with lysine (1.0 M) (●) dissolved in MilliQ water and
pH adjusted with either HCl or NaOH. Bottom: Rate of formation
of brown-colored products as determined by absorbance at 420 nm
during heating (at 45°C) of lysine (1.0 M), (■) and glucose (1.0 M)
with lysine (1.0 M) (●) dissolved in MilliQ water and pH adjusted
with HCl or NaOH.
Reactions of glucose and lysine
The role of oxidative reactions in the Maillard reactions was
initially examined by monitoring the oxygen consumption

646 J. Yin et al.
the time-dependent changes of UV-absorption in buff-
ered solutions. Generally, isosbestic points were observed
in the UV-spectra during the decay of the spin traps
indicating a reaction, where formation of products
takes place without accumulation of intermediates
(
Figure 2A). The decay of spin traps and the formation
of products were both found to follow first-order kinetics
with identical rate constants at a given pH and tempera-
ture (Figure 2B). The UV-spectra observed at the end
of the reactions were identical to the spectra of the alde-
hydes expected to be formed by hydrolysis of the nitrone
spin traps, that is, benzaldehyde in the case of PBN and
Figure 3. Hydrolysis of spin traps.
4
-pyridinecarboxaldehyde N-oxide in the case of POBN
(
Figure 3). In both cases tert-butylhydroxylamine is also
indicating specific acid and base catalysis (Figure 4). The
hydrolysis of POBN was fastest under basic conditions,
while the hydrolysis of PBN was fastest under acidic
conditions.
formed, however, it is not expected to absorb light with
wavelengths longer than 200 nm and does not contribute
to the UV-spectra.
In acidic and alkaline solutions, the decay of the spin
traps could be followed for more than two or three half-
lifes, which allowed reliable determination of the observed
For hydrolysis of PBN the pH dependence of the
observed first-order rate constants, k , could be satisfac-
obs
torily fitted to equation (2) (Figure 4 and Table I).
This equation is based on the assumption that the hydro-
lysis is taking place by three competing pathways, where
first-order rate constants, k . However, in neutral solu-
obs
tions the decays of both spin traps were extremely slow
and could only be followed during a time period that
k is the first-order rate constant for a noncatalyzed hydro-
1
was shorter than one half-life. Under these conditions, k
obs
lysis, k is the second-order rate constant for a specific
H
was determined assuming a first-order decay. The observed
acid-catalyzed hydrolysis, and k
is the second-order
OH
first-order rate constants, k , for the hydrolysis of
obs
rate constant for a specific base-catalyzed hydrolysis. K
w
POBN and PBN increased under both acidic (pH 1–7) and
is the ionization constant for water.
alkaline (pH 10–12) conditions during heating at 40°C
ϩ
ϩ
kobs ϭk ϩk ⋅[H ]ϩk ⋅K /[H ]
(2)
1
H
OH
w
(A)
For hydrolysis of POBN a truncated version of equation
(
2), where k was omitted, gave good fits to k
1 obs
(
Figure 4). The V-shape of the logarithmic plot of k as
obs
a function of pH suggests that log(k ) must be lower than
1
Ϫ8.5 for POBN at 40°C. The hydrolysis of POBN
and PBN in acidic solutions (pHϽ5) was also studied at
2
0 and 60°C (data not shown). In the acidic solutions, k
H
could be determined using the simplified equation, k
ϭ
obs
ϩ
k ⋅[H ], which adequately described the pH-dependence
H
(B)
Figure 4. Logarithmic plot of observed first-order rate constants
for hydrolysis of PBN ( ), and POBN (●) as a function of pH at
4
0°C and ionic strength 1.0. The dashed line shows the best fit to
ϩ ϩ
Figure 2. Spectra (A) and time traces (B) recorded during the
hydrolysis of PBN (40 μM) in a phosphoric acid buffer (pHϭ1.68)
at 20°C.
the expression k ϭk ϩk ⋅[H ] ϩk ⋅K /[H ] for the PBN
obs 1 H OH w
ϩ
data, and the full line shows the best fit of k ϭk ⋅[H ] ϩk ⋅K /
obs
H
OH
w
ϩ
[H ] for the POBN data.

Reactive intermediary radicals in the Maillard reactions 647
Table I. Kinetic data for the hydrolysis of PBN and POBN at 40°C
and Iϭ1.0.
Table II. Water activity (a ) of model systems.
w
Ethanol (%)
MES buffer (%)
aw
Spin traps
Ethanol model systems
5
95
70
40
0.98
∗
Rate constant
PBN
POBN
3
0
0.88∗
Ϫ9 Ϫ1
k
k
k
3.7ϫ1 0
s
–
1
H
60
0.68∗
Ϫ1 Ϫ1
Ϫ1 Ϫ1
0.24 M ⋅s
0.10 M ⋅s
MES buffer Ethanol
Ϫ1 8
M⋅s Ϫ1
Ϫ1 7
M⋅s Ϫ1
OH⋅Kw
1.0ϫ1 0
2.4ϫ1 0
Glycerol (%)
(%)
(%)
aw
Glycerol model systems
3
4
6
0
5
5
65
50
30
5
5
5
0.93Ϯ0.01
0.83Ϯ0.02
0.66Ϯ0.01
of k . Based on an Arrhenius plot of k , the activation
obs
H
energies, E , for the acid catalyzed hydrolysis were deter-
a
mined to be 16.8 kcal/mol for POBN and 16.7 kcal/mol
MES buffer Ethanol
for PBN.
Ethylene glycol (%)
(%)
(%)
aw
The observed rate constants, k , of thermal alkaline
obs
Ethylene glycol system
hydrolysis of POBN at 70°C were determined at pH 10,
6
5
30
5
0.89Ϯ0.02
1
1, and 12. Based on a linear extrapolation of these rate
constants k at pH 7.0 and pH 8.0 were estimated to
∗Calculated by Raoult’s Law.
obs
Ϫ6 Ϫ1
Ϫ5 Ϫ1
be 1.2ϫ1 0
s
and 1.6ϫ1 0
s
, respectively. Based
on these rate constants, it was estimated that more than
shown). Radicals were only observed at room temperature
in the presence of 30% glycerol, while increasing the
concentration of glycerol to 45 and 65% prevented detec-
tion of radicals. Glycerol therefore seemed to quench
radicals in a concentration dependent manner. No clear
pattern was found between the amounts of radicals detected
9
5% of POBN will be present after 50 min heating at
pH 8.0, which makes it suitable to use POBN as spin
trap under the conditions used in the present study of the
Maillard reactions.
Model system with intermediate water activity
and a .
w
In model systems with only ethanol, Raoult’s Law was
The Maillard reactions are known to proceed at a high rate
in systems with intermediate water activities. A MES-
used to calculate a according to equation, where n is
w
moles of water and n′ is moles of ethanol (Table II) [13].
buffered model system with an intermediate a was
w
therefore developed to obtain optimal conditions for radi-
cal formation during the MR between glucose and lysine.
Different solvents glycerol, ethylene glycol, and ethanol
aw ϭn/(nϩn′)
(7)
The rate of formation of radicals in the MES-buffered
ethanol model systems was also found to be concentra-
tion dependent, and radicals formed by the Fenton reac-
tion were detected even at room temperature (Figure 5).
A significant higher amount of radicals was detected in
the model system with 30% ethanol compared to the
model systems with 5% and 60% ethanol. The observed
ESR spectra (see Figure 6) were triplets of doublets with
coupling constants a ϭ15.2 gauss and a ϭ2.5 gauss,
were tested according to their ability to decrease a
w
(
Table II) and the ability to detect highly reactive radicals.
The latter was tested by generating radicals by the Fenton
reaction by addition of FeSO (0.1 mM) and H O
4
2
2
(
0.3 mM) (reactions 3–5), and evaluating the intensity of
the ESR signals of spin adducts obtained by trapping
radicals with the spin trap POBN in combination with
added ethanol (reactions 5 and 6). All systems contained
at least 5% ethanol to achieve an improved sensitivity
toward trapping reactive radicals such as alkoxyl and
hydroxyl radicals [20]. Hydroxyl radicals formed by the
Fenton reaction abstract hydrogen atoms from ethanol
leading to 1-hydroxyethyl radicals, which after being
trapped by POBN is converted into spin adducts that are
detectable by ESR [14].
N
H
typical for spin adducts of 1-hydroxyethyl radicals with
POBN [14]. In the model system with 5% ethanol, the
concentration of ethanol seemed to be too low to react
with all the radicals formed, leading to fewer radicals
detected compared to the model system with 30% etha-
nol. The same were seen at high ethanol concentration
(60%), where less radicals were detected compared to
3
0% ethanol, probably due to an overload of radicals in
2
ϩ
3ϩ
Ϫ
Fe ϩH O →Fe ϩ˙ OHϩOHϩ
(3)
(4)
(5)
(6)
2
2
the model system leading to reactions between spin
adducts and radicals, which reduce concentration of
ESR-detectable paramagnetic spin adducts. POBN is
known to trap R˙ and RO˙ radicals, whereas the spin trap
DEPMPO has an affinity to broader range of radicals,
3
ϩ
2ϩ
Fe ϩH O →Fe ϩ˙ OOHϩH
2
2
˙
OHϩCH CH OH→H OϩCH C⋅HOH
3 2 2 3
CH C⋅HOHϩPOBN→spin adduct
3
Ϫ
Model systems based on glycerol and ethylene glycol were
first tested. Addition of glycerol and ethylene glycol to
such as ˙OH, O ˙ , RO ˙, R˙, and sulfur radicals, and it
2
2
was therefore also tested in the model system. However,
detection of spin adducts from DEPMPO decreased with
increasing heating time (data not shown), probably due
to a shorter lifetime of the spin adducts, and this spin
the MES buffer resulted in a lowering of a , however, the
w
amount of detected radicals decreased with increasing
concentrations of glycerol or ethylene glycol (data not

648 J. Yin et al.
(
A)
25000
20000
15000
10000
(B)
5000
(C)
0
0
10
20
30
40
Heating time (min)
(D)
Figure 5. Intensity of spin adducts generated by mixing FeSO
4
(
(
(
0.1 mM) and H O (0.3 mM) and trapping radicals with POBN
2 2
10 mM) during heating at 70°C in mixtures of MES buffer
3340
3360
3380
3400
pH 7.0) and varying amounts of ethanol: (■) 5% ethanol, (●)
Magnetic field (G)
3
0% ethanol, and (▲) 60% ethanol.
Figure 6. Electron spin resonance (ESR) spectrum of spin
adducts generated by trapping radicals with POBN (10 mM) in 30%
trap was therefore not found suitable for experiments
involving heat treatment.
EtOH/MES buffer (pH 7.0) containing: A) FeSO (0.1 mM) and
4
H O (0.3 mM) B) Without addition. C) Lysine (0.1 M). D) Glucose
2
2
(
0.1 M) and lysine (0.1 M). All systems were heated at 70°C for
3
0 min before recording the ESR spectra. The peak-to-peak
Detection of radicals in glucose and lysine reactions
amplitude of the second doublet peak was used for quantification.
Based on the pre-trials a 30% ethanol MES-buffered model
system with POBN as spin trap was selected as an exam-
ple of a LMW MR model system for studies of formation
of intermediary radicals from lysine and glucose during
heating. Formation of radicals was detected during heating
of the model system containing glucose and lysine and
the amounts of spin adducts increased with increasing
pH (Figure 7). Spin adducts were only detected in very
low levels by heating glucose alone under the given condi-
tions. Autoxidation of monosaccharides has previously
been suggested to occur through an enediol tautomer involv-
ing enediol oxy radicals, when examined in phosphate
model system at pH 7.4 and 37°C using 5,5-dimethyl-1-
pyrroline N-oxide (DMPO) as spin trap and detection by
ESR. The consequence is loss of monosaccharide, oxygen
With Fe2+
1
1
60000 pH 7.0
20000
pH 7.0
8
4
0000
0000
0
pH 8.0
pH 8.0
With Fe2+
1
1
60000
20000
8
4
0000
0000
0
0
10
20
30
40
50
0
10
20
30
40
50
Heating time (min)
Figure 7. Formation of spin adducts in 30% ethanol/HEPES model system at pH 7.0 (above) or in 30% ethanol/HEPES model system at
pH 8.0 (below) with and without addition of FeSO (0.1 mM). (■) Control with POBN alone. (●) With added lysine (0.1 M). (▲) With
4
added glucose (0.1 M) and lysine (0.1 M). All systems contained POBN (10 mM) and were heated at 70°C for 30 min before recording the
ESR spectra.

Reactive intermediary radicals in the Maillard reactions 649
consumption, and formation of α-oxo-aldehyde, and the
formation of hydroxyl and hydroxyalkyl radicals appear
also to be involved [11].
The pathways of the Maillard reactions have been
attempted explained by Hodge involving the Amadori
rearrangement [18], but alternative routes involving the
cleavage of sugars with liberation of glycolaldehyde
imine have been proposed [19]. Dimerization and oxida-
tion of glycolaldehyde imine lead to formation of 1,4-
dialkylpyrazinium radical cations, which are stable
enough to be detected directly by ESR, when alanine is
heated at 95°C with pentoses or hexoses [20]. Therefore
LMW MR model systems without a spin trap were heated
under the same conditions to test if 1,4-dialkylpyrazinium
radical cations were formed. However, no signal were
observed, and the amount of 1,4-dialkylpyrazinium radi-
cal cations from reacting glucose with lysine at these
conditions of high ethanol content and slightly decreased
Spin adducts derived from 1-hydroxyethyl radicals
were detected by heating lysine alone due to heat-in-
duced oxidative degradation of lysine, and similar spin
adducts were also formed by heating lysine together
with glucose (Figure 6). The levels of spin adducts
increased when heating lysine together with glucose as
compared to heating lysine alone (Figure 7). More rad-
icals were detected during heating glucose with lysine
at pH 7.0 and 8.0 for 50 min at 70°C, as compared to
heating lysine alone under similar conditions (68% and
5
6%, respectively). A catalytic effect of the transition
2
ϩ
metal Fe
was observed on formation of the 1-
hydroxyethyl radical spin adducts both from lysine
alone and lysine mixed with glucose (Figure 7). The
presence of transition metals like iron induces the for-
mation of radicals as part of the Maillard reactions as
has previously been found for copper induced autoxida-
tion of Amadori products or monosaccharaides [11,15].
a was below the detection limit of the instrument.
w
Discussion
Intermediary radicals with high reactivity and short-
lived existence are formed as part of the Maillard reac-
tions, but little is known about the mechanism of
formation, since their direct detection by ESR is difficult.
Spin traps were used in the present study to trap unstable
reactive radicals during the heating lysine and glucose
leading to formation of stable spin adducts detected
directly by ESR. The applied model system contains
ethanol, which both reduces the water activity of the sys-
tem, but also acts as an initial radical trapping agent,
thereby enhancing the detection of radical intermediates
[14]. The exact identity of the formed radicals is
unknown, since they are detected as spin adducts of
POBN and 1-hydroxyethyl radicals. However, the radi-
cals formed in the LMW MR model systems were able
to abstract hydrogen atoms from ethanol prior to detec-
tion, and must therefore be of highly reactive nature such
as hydroxyl radicals or alkoxyl radicals. These radicals
formed in the LMW MR model systems can possibly be
derived from several pathways involving autoxidation of
lysine or MR intermediates, which both are catalyzed by
iron. Oxidative fragmentation of sugars into C2-, C3-,
and C4-sugar fragments involving radical mechanisms is
also expected to occur in the LMW MR model system
Addition of FeSO increased the formation of radicals,
4
where 48% and 70% more spin adducts were formed in
the glucose/lysine system compared to heating lysine
2
ϩ
alone at pH 7.0 and pH 8.0. Fe catalyzes the oxida-
2
ϩ
tion of lysine more efficiently at pH 7.0, whereas Fe
has a higher catalytic effect on formation of radicals
from glucose and lysine at pH 8.0.
1
-Hydroxyethyl spin adducts were also found to
be formed by heating the central MR intermediate, 3-
deoxyglucosone, alone at pH 8.0 (Figure 8). Addition of
FeSO also accelerated the radical formation. Heat-
4
induced oxidative degradation of 3-deoxyglucosone prob-
ably leads to formation of glucosone and carboxylic acids,
but may also involve a radical mechanism [16]. Radical
containing sugar fragments formed from a C3–C4
cleavage of 3-deoxyglucosone has previously been found
to be generated in solution at room temperature in
dimethyl sulfoxide [17].
1
1
1
1
6000
4000
2000
0000
[
17] as well as degradation of deoxyxones leading to
the formation of H O and hydroxyl radicals [9] could
2
2
lead to highly reactive radicals. Reactive oxygen species
ROS) formed by autoxidation of Maillard reaction prod-
8
6
4
2
000
000
000
000
0
(
ucts such as Amadori compounds have also previously
been suggested [9]. However, the 1,4-dialkylpyrazinium
radical cations which have been suggested as initial
intermediates in the MR, is not expected be able to
abstract hydrogen atoms from ethanol, due to the
delocalized nature of the unpaired spins, and they will
therefore not give rise to formation of spin adducts in
the model systems in the present study [10].
0
10
20
30
40
50
Heating time (min)
Figure 8. Spin adducts formed by POBN (10 mM) during heating
at 70°C in 30% ethanol/70% HEPES model system at pH 8.0 with
2ϩ
The results obtained using the model systems in this
study indicate that radicals will be formed as part of the
MR that take place during heat processing of foods and
and without the addition of Fe ; (▲) control with POBN alone,
(
♦) with added FeSO (0.1 mM), (■) 3-deoxyglucosone (0.04 M),
4
and (●) 3-deoxyglucosone (0.04 M) with FeSO (0.1 mM).
4

650 J. Yin et al.
that heating time and pH conditions will be important
factors. However, the presence of other compounds in
the food matrix would be important for the involvement
of highly reactive radicals in the MR, such as the avail-
ability of transition metals like iron, which for example
can be found in high concentrations in products like meat.
More experiments are needed exploring the full effect
of temperature and pH on both formation of highly reac-
tive radicals in food as well as involvement of other com-
pounds with consequences for formation. Knowledge
of the participation of radical processes in the MR can
be useful in controlling the reaction in relation to for-
mation of beneficial compounds like color and flavor, in
contrast to harmful compounds like AGEs.
[4] Cammerer B, Kroh LW. Investigation of the contribution of
radicals to the mechanism of the early stage of the Maillard
reaction. Food Chem 1996;57:217–221.
[
5] Hayashi T, Ohta Y, Namiki M. Electron-spin resonance
spectral study on structure of novel free-radical products
formed by reactions of sugars with amino-acids Or amines.
J Agric Food Chem 1977;25:1282–1287.
[
6] Hofmann T, Bors W, Stettmaier K. Radical-assisted melanoidin
formation during thermal processing of foods as well as
under physiological conditions. J Agric Food Chem 1999;
4
7:391–396.
[7] LiedkeR,EichnerK.RadicalinducedformationofD-glucosone
from Amadori compounds. In: Morello MJ, Shahidi F,
Ho CT (eds.). Free Radicals in Foods Chemistry, Nutrition
and Health. Washington, DC: American Chemical Society;
2
002. pp. 69–83.
[
8] Rizzi GP. Free radicals in the Maillard reaction. Food Rev Int
2003;19:375–395.
In conclusion, slightly alkaline conditions were found
to increase oxygen consumption and formation of brown-
colored products, as well as formation of highly reactive
intermediary radicals in the Maillard reactions, when
glucose and lysine is heated together at elevated
temperatures.
[9] W o l ff SP, Crabbe MJC, Thornalley PJ. The autoxidation of
glyceraldehyde and other simple monosaccharides. Experientia
1
984;40:244–246.
[
10] Davies MJ, Hawkins CL. EPR spin trapping of protein
radicals. Free Radic Biolo Med 2004;36:1072–1086.
[11] Thornalley PJ, Wolff SP, Crabbe MJC, Stern A. The oxidation
of oxyhemoglobin by glyceraldehyde and other simple
monosaccharides. Biochem J 1984;217:615–622.
[
12] Sato T, Iwaki M, Shimogaito N, Wu XG, Yamagishi S,
Takeuchi M. TAGE (toxic AGEs) theory in diabetic
complications. CurrMol Med 2006;6:351–358.
Acknowledgment
This work is carried out as a part of the research program
of the UNIK: Food, Fitness & Pharma for Health and
Disease (see www.foodfitnesspharma.ku.dk). The UNIK
project is supported by the Danish Ministry of Science,
Technology and Innovation.
[13] Kalathenos P, Russel NJ. Ethanol as a Food Preservative.
New York: Kluwer Academic/Plenum Publishers; 2003.
[
14] Pou S, Ramos CL, Gladwell T, Renks E, Centra M, Young D,
et al. A kinetic approach to the selection of A sensitive spin-
trapping system for the detection of hydroxyl radical. Anal
Biochem 1994;217:76–83.
[
15] Kawakishi S, Tsunehiro J, Uchida K. Autoxidative degradation
of amadori compounds in the presence of copper-ion.
Carbohydr Res 1991;211:167–171.
Declaration of interest
[
[
16] Belitz HD, Grosch W, Schieberle P. Food Chemistry. Berlin:
Springer; 2009.
The authors report no conflicts of interest. The authors
alone are responsible for the content and writing of the
paper.
17] Calle P, Sanchez A, Sieiro C. Studies of the alkaline-
degradation of carbohydrates in methyl sulfoxide by Esr
Spectroscopy .3. Esr study of the alkaline-degradation of
disaccharides in methyl sulfoxide. Carbohydr Res 1992;224:
1
9–27.
References
[
18] Hodge JE, Rist CE. The amadori rearrangement under new
conditions and its significance for non-enzymatic browning
reactions. J Am Chem Soc 1953;75:316–322.
[
[
[
1] Brownlee M. Advanced protein glycosylation in diabetes and
aging. Annu Rev Med 1995;46:223–234.
[19] Namiki M, Hayashi T. A new mechanism of the Maillard-
reaction involving sugar fragmentation and free-radical
formation. Acs Symp Ser 1983;215:21–46.
2] Vlassara H, Striker GE. AGE restriction in diabetes mellitus:
a paradigm shift. Nat Rev Endocrinol 2011;7:526–539.
3] Maillard LC. The action of amino acids on sugar; The
formation of melanoidin by a methodic route. Comptes Rendus
Hebdomadaires des Seances de l Academie des Sciences
[20] Hofmann T, Bors W, Stettmaier K. Studies on radical
intermediates in the early stage of the nonenzymatic browning
reaction of carbohydrates and amino acids. J Agric Food
Chem 1999;47:379–390.
1
912;154:66–68.