Structural and functional characterization of pronyl-lysine, a novel protein modification in bread crust melanoidins showing in vitro antioxidative and phase I/II enzyme modulating activity

Article abstract of DOI:10.1021/jf020618n

Application of an in vitro antioxidant assay to solvent fractions isolated from bread crust, bread crumb, and flour, respectively, revealed the highest antioxidative potential for the dark brown, ethanol solubles of the crust, whereas corresponding crumb and flour fractions showed only minor activities. To investigate whether these browning products may also act as antioxidants in biological systems, their modulating activity on detoxification enzymes was investigated as a functional parameter in intestinal Caco-2 cells. The bread crust and, in particular, the intensely brown, ethanolic crust fraction induced a significantly elevated glutathione S-transferase (GST) activity and a decreased phase I NADPH-cytochrome c reductase (CCR) activity compared to crumb-exposed cells. Antioxidant screening of Maillard-type model mixtures, followed by structure determination, revealed the pyrrolinone reductones 1 and 2 as the key antioxidants formed from the hexose-derived acetylformoin and Nα-acetyl-L-lysine methyl ester or glycine methyl ester, chosen as model substances to mimic nonenzymatic browning reactions with the lysine side chain or the N terminus of proteins, respectively. Quantitation of protein-bound pyrrolinone reductonyl-lysine, abbreviated pronyl-lysine, revealed high amounts in the bread crust (62.2 mg/kg), low amounts in the crumb (8.0 mg/kg), and the absence of this compound in untreated flour. Exposing Caco-2 cells for 48 h to either synthetically pronylated albumin or purified pronyl-glycine (3) significantly increased phase II GST activity by 12 or 34%, respectively, thus demonstrating for the first time that "pronylated" proteins as part of bread crust melanoidins act as monofunctional inducers of GST, serving as a functional parameter of an antioxidant, chemopreventive activity in vitro.

Full text of DOI:10.1021/jf020618n

J. Agric. Food Chem. 2002, 50, 6997−7006 6997
Structural and Functional Characterization of Pronyl-lysine, a
Novel Protein Modification in Bread Crust Melanoidins Showing
in Vitro Antioxidative and Phase I/II Enzyme Modulating Activity
MICHAEL LINDENMEIER, VERONIKA FAIST,‡ AND THOMAS HOFMANN*,#
†
Deutsche Forschungsanstalt f u¨ r Lebensmittelchemie, Lichtenbergstrasse 4,
D-85748 Garching, Germany, Institut f u¨ r Humanern a¨ hrung und Lebensmittelkunde,
Duesternbrooker Weg 17, D-24105 Kiel, Germany, and Institut f u¨ r Lebensmittelchemie,
Corrensstrasse 45, D-48149 M u¨ nster, Germany
Application of an in vitro antioxidant assay to solvent fractions isolated from bread crust, bread crumb,
and flour, respectively, revealed the highest antioxidative potential for the dark brown, ethanol solubles
of the crust, whereas corresponding crumb and flour fractions showed only minor activities. To
investigate whether these browning products may also act as antioxidants in biological systems, their
modulating activity on detoxification enzymes was investigated as a functional parameter in intestinal
Caco-2 cells. The bread crust and, in particular, the intensely brown, ethanolic crust fraction induced
a significantly elevated glutathione S-transferase (GST) activity and a decreased phase I NADPH-
cytochrome c reductase (CCR) activity compared to crumb-exposed cells. Antioxidant screening of
Maillard-type model mixtures, followed by structure determination, revealed the pyrrolinone reductones
1
R
and 2 as the key antioxidants formed from the hexose-derived acetylformoin and N -acetyl-L-lysine
methyl ester or glycine methyl ester, chosen as model substances to mimic nonenzymatic browning
reactions with the lysine side chain or the N terminus of proteins, respectively. Quantitation of protein-
bound pyrrolinone reductonyl-lysine, abbreviated pronyl-lysine, revealed high amounts in the bread
crust (62.2 mg/kg), low amounts in the crumb (8.0 mg/kg), and the absence of this compound in
untreated flour. Exposing Caco-2 cells for 48 h to either synthetically pronylated albumin or purified
pronyl-glycine (3) significantly increased phase II GST activity by 12 or 34%, respectively, thus
demonstrating for the first time that “pronylated” proteins as part of bread crust melanoidins act as
monofunctional inducers of GST, serving as a functional parameter of an antioxidant, chemopreventive
activity in vitro.
KEYWORDS: Antioxidants; detoxication enzymes NADPH-cytochrome c reductase; glutathione S-
transferase; melanoidins; pronyl-lysine; pronyl-glycine
INTRODUCTION
Although numerous attempts have been undertaken to isolate
and purify melanoidins from foods, for example, from coffee
The Maillard reaction between reducing carbohydrates and
amino acids or proteins is chiefly responsible for the develop-
ment of the brown color that occurs during baking, roasting, or
frying of foods. This browning of thermally processed foods,
for example, the crust of a freshly baked bread, is highly
desirable and is intimately associated in consumers’ minds with
a delicious, high-grade product. Depending on their molecular
weight, the colored components affecting this nonenzymatic
browning may be divided into two classes, namely, the low
molecular weight colored compounds and the melanoidins,
which are assumed to be nitrogen-containing, high molecular
weight colored compounds with masses up to 100000 Da (1).
(2-4), soy sauce (5), malt (6), and dark beer (7), it has as yet
not been possible to isolate and characterize partial structures
of complex food melanoidins. Very recently, EPR and LC/MS
spectroscopy as well as carefully planned synthetic and quan-
titative model experiments have given first insights into mel-
anoidin formation in toasted wheat bread crust and roasted
coffee and demonstrated the previously unknown protein-bound
1
,4-bis(5-amino-5-carboxy-1-pentyl)pyrazinium radical cation
(CROSSPY), showing protein cross-linking ability and high
browning potential, as a key intermediate in roasting-induced
melanoidin genesis (8). These findings suggested that the lysine
side chains of proteins are the active sites in the formation of
melanoproteins as part of food melanoidins (9, 10).
*
Author to whom correspondence should be addressed (telephone +49-
8
9-289-14170; fax +49-89-289-14183; e-mail thomas.hofmann@lrz.tum.de).
Besides the sensory impact of the browning products, little
is known so far about the physiological relevance of these
Maillard-type melanoidins. Maillard reaction products with
†
Deutsche Forschungsanstalt f u¨ r Lebensmittelchemie.
‡
Institut f u¨ r Humanern a¨ hrung und Lebensmittelkunde.
#
Institut f u¨ r Lebensmittelchemie.
1
0.1021/jf020618n CCC: $22.00 © 2002 American Chemical Society
Published on Web 10/16/2002

6998 J. Agric. Food Chem., Vol. 50, No. 24, 2002
Lindenmeier et al.
3
(2H)-furanone structures have been reported to possess DNA-
Table 1. Fractionation of Flour, Bread Crumb, and Crust by Sequential
Solvent Extraction
a
damaging properties (11, 12). However, potentially beneficial
effects were also reported, for example, the intensely colored
yield, g/100 g of
3-hydroxy-4-[(E)-(2-furyl)methylidene]methyl-3-cyclopentene-
fractiona
flour
bread crumb
bread crust
1
,2-dione was found to potently inhibit the growth of human
tumor cells in vitro by effectively shutting down the activity of
the MAP kinase cascade (13). By far the most studies in the
past three decades, however, addressed the antioxidant function
of Maillard products. With a few exceptions, most investigations
were focused on carbohydrate/amino acid model mixtures only
I
II
III
IV
26.2
2.8
0.7
28.6
2.1
0.4
20.7
2.0
0.2
70.3
69.0
77.2
a
Fractions were obtained by extracting the material with water (fraction I) or
ethanol (fraction II), followed by 2-propanol (fraction III) and removing the solvent
in vacuo or by freeze-drying. Fraction IV contains the nonsoluble materials.
(14, 15), rather than on performing systematic studies on
authentic food melanoidins. Iwansky and Franzke (16) were the
first to show that fat spoilage in cookies could be significantly
inhibited by the addition of melanoidins to the dough. Although
total antioxidative activity was measured several times in
browned foods such as roasted coffee (17, 18), roasted malts
The following compounds were synthesized as reported recently:
-deoxy-1-hexosulose (25), acetylformoin (26), and 1-deoxy-2,3-
hexodiulose (unpublished results).
Preparation of Rye/Wheat Mixed Bread. A preferment was
prepared from rye flour (type 1150, 108 g), starter cultures (B o¨ cker
mother, 12 g), and water (96 g), which were kneaded and incubated
for 16-20 h at 26 °C. A sourdough (1500 g) was prepared by incubating
a dough of rye flour (type 1150, 674 g), preferment (75 g), and water
(750 g) for 16 h at 26 °C. For preparation of the rye/wheat mixed bread,
rye flour (type 1150, 1117 g), wheat flour (type 812, 745 g), water
3
(19), Asian soy sauce (20), or tomato powder (17), most
investigations provided only cumulative information on the
physiological effects of the particular browned food, rather than
trying to correlate these effects on a molecular level with distinct
chemical structures of the antioxidatively active sites present
in these melanoidins.
In addition, it is as yet not clear how nonenzymatic browning
products influence xenobiotic enzymes in the human body. Most
chemopreventive, nonendogenously formed agents act through
enzyme systems by modulating phase I and phase II enzymes.
Phase I metabolic transformations include reduction, oxidation,
and hydrolytic reactions, whereas phase II transformations
generally act through conjugation reactions of the parent
xenobiotics or of phase I metabolites. The conjugation reactions
facilitate transport and enhance elimination of the inactive
compounds via the renal and biliary routes. Therefore, the main
determinant of whether exposure to xenobiotics will result in
toxicity is the balance between the activities of phase I and phase
II enzymes (21). Although phase II enzymes are hypothesized
to facilitate the metabolic transit of food-derived Maillard
reaction products formed in heat-treated proteins (22, 23), it is
still an open question whether nonenzymatic browning products
formed in foods require specific detoxifying mechanisms or
contribute to the chemopreventive potential of the organism via
induction of phase II enzymes. Specifically, induction of the
phase II glutathione S-transferase by antioxidants is proposed
as a promising strategy for cancer prevention (24).
The objectives of the present study were, therefore, (i) to
determine the in vitro antioxidant activity of fractions isolated
from bread crust chosen as an example of an intensely browned
food, (ii) to identify the chemical structure of an antioxidant
site of bread crust melanoidins, and (iii) to investigate whether
this structural domain of the crust melanoidins showing anti-
oxidant activity in vitro may also modulate detoxification
enzyme activities in biological systems using a human intestinal
cell culture model.
(
1650 g), sourdough (1241 g), baker’s yeast (43 g), and NaCl (43 g)
were kneaded (dough weight ) 4839 g), filled into two baking tins (5
cm high), and baked using the following procedure: dough temperature
(28-29 °C), dough resting (15 min), dough fermentation (45 min),
water vaporization (10 s before and 5 s after the bread was introduced
into the oven; after 1 min, the water vapor was removed through the
discharge pipe for 1 min); baking was done for 24 min at 260 °C,
followed by 186 min at 220 °C.
Fractionation of Bread Crumb and Crust. The brown bread crust
was carefully separated from the bread crumb with a kitchen knife;
crust and crumb were frozen in liquid nitrogen and then ground in a
mill. The powder obtained from crust and crumb (300 g each) as well
as the flour (300 g) were defatted by stirring with chloroform (3 ×
3
00 mL), and, after filtration, solvent residues were removed in vacuo.
The defatted powders were extracted twice with tap water (800 mL)
upon stirring for 3 h at 50 °C. After filtration, the aqueous filtrates
were combined to give fraction I. The nonsoluble residue was then
extracted twice for 3 h at room temperature with aqueous ethanol
(60% EtOH in water, 800 mL), yielding fraction II upon filtration and
solvent evaporation and a residue, which was finally extracted for 24
h with aqueous 2-propanol (50% 2-propanol in water, 1000 mL) to
give fraction III after centrifugation and solvent evaporation. Trace
amounts of solvents were removed from fractions I-III as well as the
nonsoluble materials (fraction IV) in high vacuum. After freeze-drying,
the yields of the individual fractions I-IV were determined by weight
(Table 1).
Measurement of Antioxidative Activity in Vitro. Following a
procedure reported in the literature (19) with some modifications, the
antioxidative activity of melanoidin fractions was determined in vitro
by measuring their inhibitory effect on linoleic acid peroxidation. First,
the linoleic acid substrate was prepared by dropwise adding linoleic
acid (0.125 mL) to a mixture of oxygen-free borate buffer (2.5 mL; 50
mmol/L; pH 9.0) and Tween 20 (0.125 mL). Then aqueous sodium
hydroxide solution (1 mol/L) was added until a clear solution was
obtained, and the mixture was diluted with oxygen-free borate buffer
(50 mmol/L; pH 9.0) to 25 mL and made up to 50 mL with oxygen-
free, distilled water. An aliquot (60 µL) of a solution of the melanoidin
sample (10 mg/mL in 50% aqueous ethanol) or a purified compound
(1 mmol/mL in 50% aqueous ethanol) was added to a solution of
oxygen-saturated phosphate buffer (3 mL; 0.2 mol/L; pH 6.75),
hydrogen peroxide (100 µL; 16 mmol/L), iron(II) sulfate (100 µL; 16
mmol/L; containing 15 mM EDTA), and linolenic acid substrate (1.0
mL). After incubation of the mixture for 10 min at room temperature,
an aliquot (1.0 mL) of that solution was pipetted into disposable cuvettes
(1 cm i.d.) containing a solution (2.0 mL) of the color reagent consisting
of dimethylformamide (8%), Triton X-100 (1.4%), hemoglobin (56
MATERIALS AND METHODS
Chemicals. The following compounds were obtained commer-
cially: starch, glucose, hydrogen peroxide, iron(II) sulfate, EDTA,
dimethylformamide, ethanol, 2-propanol, and 1-chloro-2,4-dinitro-
benzene (Merck, Darmstadt, Germany); glycine methyl ester and
R
N -acetyl-L-lysine methyl ester (Nova Biochem, Schwabach/Ts., Ger-
many); albumin bovine fraction V and Tween 20 (Serva, Heidelberg,
Germany); linolenic acid, Triton X-100, hemoglobin, fetal bovine
serum, L-glutamine, penicillin, streptomycin, porcine trypsin, and
dithiothreitol (Sigma, Deisenhofen, Germany); benzoyl leucomethylene
blue and N-methylpyridone (Aldrich, Deisenhofen, Germany); Trolox
(Fluka, Deisenhofen, Germany); and cytochrome c (Boehringer, Man-
nheim, Germany).

Pronyl-lysine as an Antioxidant Site of Bread Crust Melanoidins
J. Agric. Food Chem., Vol. 50, No. 24, 2002 6999
Table 2. Yields of Molecular Weight Fractions Obtained by
Subfractionation of the Ethanolic Fraction II Using Multistep
Ultrafiltration
Chart 1. Structures of 2,4-Dihydroxy-2,5-dimethyl-1-(5-acetamino-
5-methoxycarbonyl-pentyl)-3-oxo-2H-pyrrole (1) and 2,4-Dihydroxy-
2,5-dimethyl-1-methoxycarbonylmethyl-3-oxo-2H-pyrrole (2) Isolated as
the Key Antioxidants from Heated Solutions of Acetylformoin and
N -Acetyl-L-lysine Methyl Ester or Glycine Methyl Ester, Respectively,
R
and Structure of 2,4-Dihydroxy-2,5-dimethyl-1-carboxymethyl-
yield (%)
fraction
MW (kDa)
flour
bread crumb
bread crust
II/5
II/4
II/3
II/2
II/1
g100
g30
g10
g1
<1
26.3
16.1
6.6
5.5
40.3
94.8
4.6
2.5
4.5
14.8
73.6
100.0
1.8
2.5
1.9
21.2
68.9
96.3
3-oxo-2H-pyrrole (3) Liberated from 2 by Enzyme Treatment
Σ (I/1−VII/5)
mg/L), and benzoyl leucomethylene blue (130 µM) in phosphate buffer
0.2 mol/L; pH 5.0). After an additional incubation time of 30 min,
(
the absorbance was measured at λ ) 666 nm against the buffer/color
reagent blank. The results were related to the absorption of an aqueous
standard solution of Trolox (1 mmol/L) and were expressed as Trolox
equivalents (TE values). Each of the experiments was performed in
triplicate.
Ultrafiltration. A solution of fraction II (400 mg in 100 mL 50%
EtOH) was fractionated sequentially by multistep ultrafiltration (Ami-
con, Witten, Germany) starting with a filter having a cutoff of 100000
Da (Diaflo YM 100), followed by 30000 Da (Diaflo YM 30), 10 kDa
1
) δ 1.34 [s, 3H, H-C(1)], 1.36-1.86 [3 × m, 6H, H-C(8-10)], 1.98
[
[
s, 3H, H-C(15)], 2.22 [s, 3H, H-C(6)], 3.31 [m, 2H, H-C(7)], 3.72
s, 3H, H-C(13)]; 4.4 [m, 1H, H-C(11)].
Isolation of 2,4-Dihydroxy-2,5-dimethyl-1-methoxycarbonyl-
(
Diaflo YM 10), and, finally, 1 kDa (Diaflo YM 1). The separation
was performed under a nitrogen pressure of 1.5 bar for the filter YM
00 or under 4.0 bar for the filters YM 1, YM 10, and YM 30. The
methyl-3-oxo-2H-pyrrole (2) from a Model Mixture of Acetylfor-
moin and Glycine Methylester. A solution of acetylformoin (20 mmol)
and glycine methyl ester (20 mmol) in phosphate buffer (120 mL; 0.1
mol/L; pH 5.5) was heated at 100 °C for 25 min. After cooling to
room temperature, a small aliquot of the reaction mixture was separated
by RP-HPLC. When the effluent was monitored at λ ) 360 nm, a peak
was detected and collected between 15 and 16 min. For a preparative
isolation of the target compound, the major aliquot of the reaction
mixture was extracted with methylene chloride (5 × 50 mL); the
aqueous phase was mixed with silica gel (20 g) and then freeze-dried.
The residue was applied onto the top of a column filled with a slurry
of silica gel (150 g) in ethyl acetate. Chromatography was performed
with ethyl acetate (300 mL), followed by ethyl acetate/methanol
1
single fractions were freeze-dried, and the residues were weighed and
stored in a desiccator. The weight content of each fraction from the
total fraction II was calculated in percent (Table 2).
Antioxidant Screening of Maillard Reaction Mixtures Prepared
from N
solutions of N
r
-Acetyl-L-lysine Methyl Ester and Carbohydrates. Binary
-acetyl-L-lysine methyl ester (1 mmol) and starch (162
R
mg), glucose (1 mmol), 3-deoxy-1-hexosulose (1 mmol), 1-deoxy-2,3-
diulose (1 mmol), or acetylformoin (1 mmol), respectively, were heated
in phosphate buffer (2.5 mL; 0.1 mol/L; pH 5.5) for 25 min at 100 °C.
After cooling to room temperature, the reaction mixtures were freeze-
dried, the residues were taken up in aqueous ethanol (4 mL, 50% EtOH
in water), and aliquots (60 µL) were then used to measure the
antioxidative potential using the in vitro assay reported above. As the
control, the antioxidative potential of the corresponding nonheated
solutions was determined.
(
90:10, v/v; 300 mL), affording an orange fraction. Thin-layer chro-
matography on RP-18 material using water/methanol (70:30) as the
eluent revealed a yellow, fluorescent compound at R 0.61. The solvent
f
was removed in vacuo and the residue taken up in methanol (3 mL).
Upon addition of ethyl acetate, the target compound was obtained as
an yellow powder. Recrystallization from methanol/ethyl acetate (80:
Antioxidant Screening of Maillard Reaction Products Formed
from Acetylformoin and Glycine Methyl Ester or N
Methyl Ester, Respectively. Solutions of acetylformoin (1 mmol) and
glycine methyl ester (1 mmol) or N -acetyl-L-lysine methyl ester (1
r
-Acetyl-L-lysine
2
0, v/v) afforded compound 2 as yellow crystals (4.5 mmol; 15%
yield): UV-vis (H
2
8
MeOD-d
2
2
O) λmax 363 nm; MS(EI), m/z 43 (100), 56 (89),
CO), 84 (55), 149 (53), 115 (52),
R
+
15 (78, M ), 172 (86, M + -CH
3
mmol), respectively, in phosphate buffer (5 mL; 0.1 mol/L; pH 5.5)
were heated at 100 °C for 25 min. After cooling to room temperature,
the reaction mixtures were separated by RP-HPLC, and the effluent
was separated into 26 fractions, which were separately collected in glass
vials. The corresponding fractions obtained from six HPLC runs were
collected, combined, and freeze-dried. The residues obtained from these
pooled HPLC fractions were taken up in water/etanol (4 mL; 1:1, v/v)
and were then used to measure the antioxidative potential using the in
vitro assay reported above.
1
6 (47), 156 (35), 140 (32), 144 (25), 199 (20); H NMR (360 MHz,
; arbitrary numbering of the carbon atoms refers to structure
in Chart 1) δ 1.26 [s, 3H, H-C(1)], 2.16 [s, 3H, H-C(6)], 3.73 [s,
3
2
3
H, H-C(9)], 4.15 [d, 1H, J7a,7b ) 18.13 Hz, H
a
-C(7)], 4.23 [d, 1H,
2
13
J
7a,7b ) 18.13 Hz, H
a
-C(7)]; C NMR (360 MHz, MeOD-d , HMQC,
3
HMBC, DEPT; arbitrary numbering of the carbon atoms refers to
structure 2 in Chart 1) δ 11.5 [CH , C(1)], 22.0 [CH , C(6)], 43.3
, C(7)], 88.2 [C, C(2)], 128.2 [C, C(5)], 169.1
3
3
3 2
[CH , C(9)], 53.4 [CH
[C, C(4)], 172.7 [C, C(8)], 193.1 [C, C(3)].
Isolation of 2,4-Dihydroxy-2,5-dimethyl-1-(5-acetamino-5-meth-
oxycarbonyl-pentyl)-3-oxo-2H-pyrrole (1) from a Model Mixture
Isolation of 2,4-Dihydroxy-2,5-dimethyl-1-carboxymethyl-3-oxo-
of Acetylformoin and N
of acetylformoin (4 mmol) and N
r
-Acetyl-L-lysine Methyl Ester. A solution
-acetyl-L-lysine methyl ester (4 mmol)
2H-pyrrole (Pronyl-glycine, 3). Following a procedure reported
recently (26) with some modifications, a solution of 2,4-dihydroxy-
2,5-dimethyl-1-methoxycarbonylmethyl-3-oxo-2H-pyrrole (2; 1.0 mmol)
and porcine liver esterase (2000 units) in phosphate buffer (10 mL;
0.2 mol/L; pH 7.5) was stored overnight at 37 °C. The mixture was
concentrated by partial freeze-drying to 1 mL, and the target compound
was purified by chromatography on RP-18 material (15.0 g; Lichroprep
25-40 µm, Merck, Darmstadt, Germany) using a mixture (20:80, v/v)
of methanol and trifluoracetic acid (0.1% TFA in water) as the mobile
phase. After application of the crude material, chromatography with
the same eluent afforded compound 3 in the effluent >80 mL. The
combined eluates were freeze-dried, yielding the colored compound
R
in phosphate buffer (20 mL; 0.1 mol/L; pH 5.5) was heated at 100 °C
for 25 min. After cooling to room temperature, the reaction mixtures
were separated by semipreparative RP-HPLC. When the effluent was
monitored at λ 360 nm, a peak was detected and collected between 19
and 20 min. After freeze-drying, the target compound was obtained as
1
a pale yellow powder. On the basis of UV-vis, LC/MS, and H NMR
data the lysine derivative was identified as 2,4-dihydroxy-2,5-dimethyl-
1-(5-acetamino-5-methoxycarbonylpentyl)-3-oxo-2H-pyrrole (1 in Chart
+
1
): UV-vis (water) λmax 363 nm; LC/MS(APCI ), m/z 311 (100, [M
+
+
1
+
1 - H
2
3
O] ), 329 (86, [M + 1] ); H NMR (400 MHz, MeOD-d ;
-
arbitrary numbering of the carbon atoms refers to structure 1 in Chart
as an orange powder (0.35 mmol, ∼35% yield): LC/MS(APCI ), m/z

7000 J. Agric. Food Chem., Vol. 50, No. 24, 2002
Lindenmeier et al.
-
1
2
00 (100, [M + 1 - H
2
] ); H NMR (360 MHz, MeOD-d
3
; arbitrary
were maintained at 37 °C in Dulbecco’s modified Eagle’s medium,
containing 10% fetal bovine serum, 2% L-glutamine (200 mM), and
2% penicillin-streptomycin (5000 units of penicillin and 5 mg of
numbering of the carbon atoms refers to structure 3 in Chart 1) δ
2
1
1
.28 [s, 3H, H-C(1)], 2.17 [s, 3H, H-C(6)], 4.17 [d, 1H, J7a,7b )
2
8.2 Hz, H
Synthesis of 5-Acetyl-4-hydroxy-1,3-dimethylpyrazole. A solution
of methyl hydrazine (1.5 mmol) in ethanol (5 mL) was adjusted to pH
with concentrated hydrochloric acid, pronyl-glycine (3, 1 mmol) was
a
-C(7)], 4.25 [d, 1H, J7a,7b ) 18.2 Hz, H
a
-C(7)].
2
streptomycin/mL of 0.9% NaCl) in an atmosphere of CO /air (1:20,
2
v/v). Cells grown in 75 cm culture flasks were supplied on culture
medium (30 mL), which was exchanged (50%) twice a week. Cells
2
4
were seeded at a density of 20000 cells/cm to achieve confluency at
added, and the mixture was refluxed for 5 min. The solution was
neutralized with an aqueous sodium hydroxide solution (1 mmol/L),
concentrated, and then separated by semipreparative HPLC using
RP-18 as the stationary phase. Two peaks were detected in a ratio of
day 6 after seeding. The fractions isolated from bread crust, BSA, and
the synthesized model compounds pronyl-BSA and pronyl-glycine (3)
were dissolved in the medium and were exposed to the cells for 48 h.
Each of the experiments was performed in triplicate. After exposure
to the different bread fractions, the cells were washed with Dulbecco’s
phosphate-buffered saline (DPBS), harvested with trypsin-EDTA (0.5
g of porcine trypsin and 0.2 g of NaEDTA/L of Hank’s balanced salt
solution; incubation time ) 10 min), centrifuged (5 min, 1000g), washed
(DPBS), and centrifuged (5 min, 1000g) again. Cells were diluted 1:5
in Tris-HCl buffer (20 mM, pH 7.4) containing sucrose (0.25 M) and
dithiothreitol (1.4 mM), subsequently homogenized in a glass/Teflon
homogenizer (27), and centrifuged (60 min, 105000g). The cytosolic
105000g supernatant was used for the GST analysis. The pellet
containing the microsomal fraction was resuspended in 5 mL of 0.9%
NaCl. Microsomal NADPH-cytochrome c reductase activity was
determined in the remaining pellet according to a method reported in
the literature (28). GST activity was determined using 1-chloro-2,4-
dinitrobenzene as the substrate as described in the literature (29), and
protein content was measured as reported earlier (30).
Data obtained from triplicate experiments are given as means and
standard deviations in relation to the basal activity of nonexposed
control cells (basal activity analyzed for CCR and GST was 3.85 (
0.40 nmol of cytochrome c/mg of protein/min and 350 ( 42 nmol of
CDNB/mg of protein/min, respectively). Means of each treatment were
compared with untreated control cells by Student’s t test. The level of
significance was set at P < 0.05 (*).
High-Performance Liquid Chromatography (HPLC). The HPLC
apparatus (BIO-TEK Instruments, Eching, Germany) consisted of two
pumps (type 422), a gradient mixer (M 800), a Rheodyne injector (100
µL loop), and a diode array detector (DAD type 540+) monitoring the
effluent in a wavelength range between 210 and 500 nm. Separations
were performed on a stainless steel column packed with RP-18 (C-18
Nucleosil 300 nm, 5 µm) either in an analytical (4.6 × 240 mm, 1.6
mL/min) or in a semipreparative scale (10 × 250 mm, 3.0 mL/min).
For the analytical runs, a gradient was used starting with a 100%
aqueous TFA (0.1% TFA in water) and increasing the methanol content
to 100% within 40 min. For the semipreparative runs, a gradient was
used starting with a 100% aqueous formic acid (0.1% in water) and
increasing the methanol content to 62.5% within 25 min and then to
100% within 28 min.
High-Resolution Gas Chromatography/Mass Spectrometry
(HRGC/MS). HRGC was performed with a GC 3800 gas chromato-
graph (Varian, Darmstadt, Germany) equipped with a 30 m × 0.32
mm i.d., 0.25 µm, fused silica capillary CP SIL 19CB (Chrompack,
Frankfurt, Germany) by on-column injection at 40 °C. After 1 min,
the temperature of the oven was raised at 15 °C/min to 100 °C, then
raised at 6 °C/min to 160 °C, and finally raised at 10 °C/min to 230
°C and held for 5 min. The flow of the carrier gas, helium, was 2.5
mL/min. MS analysis was performed with a Saturn GC MS/MS 2000
(Varian) in tandem with the HRGC. Mass chromatography in the
electron impact mode (MS/EI) was performed at 70 eV and in the
chemical ionization mode (MS/CI) at 115 eV with methanol as the
reactant gas.
Liquid Chromatography/Mass Spectrometry (LC/MS). An ana-
lytical HPLC column (Nucleosil 100-5C18, Macherey and Nagel,
D u¨ rren, Germany) was coupled to an LCQ-MS (Finnigan MAT GmbH,
Bremen, Germany) using electrospray ionization (ESI). After injection
of the sample (2-20 µL), analysis was performed using a gradient
starting with 100% aqueous TFA (0.1% TFA in water) and increasing
the methanol content to 100% within 40 min.
3
4
:1; the effluents were collected and freeze-dried affording 5-acetyl-
-hydroxy-1,3-dimethylpyrazole (0.48 mmol, 48% yield) as the major
reaction product and 3-acetyl-4-hydroxy-1,5-dimethylpyrazole as a trace
compound in a purity of >99%. Spectroscopic data of 5-acetyl-4-
hydroxy-1,3-dimethyl-pyrazole: GC/MS(EI), m/z 154 (78), 139 (54),
1
1
11 (5), 97 (5), 56 (100), 42 (46); H NMR (400 MHz, CDCl
3
, DQF-
), 3.78 (s, 3H,
, HMQC, HMBC) δ 8.2 (CH ),
), 123.5 (NdC), 135.5 (NsCd), 143.2
C-OH), 198.1 (CO). Spectroscopic data of 3-acetyl-4-hydroxy-1,5-
dimethylpyrazole: GC/MS(EI), m/z 154 (100), 139 (39), 85 (34), 70
COSY) δ 2.20 (s, 3H, CH
N-CH
2
3
), 2.52 (s, 3H, CO-CH
); C NMR (360 MHz, CDCl
5.5 (CH ), 37.6 (N-CH
3
1
3
3
3
3
3
3
(
1
(
2
21), 56 (10), 42 (88); H NMR (400 MHz, CDCl
.20 (s, 3H, CH ), 2.56 (s, 3H, CO-CH ), 4.04 (s, 3H, N-CH
Quantitation of Pronylation Degree of Proteins. Ground bread
crust, crumb, or flour (each 50 g) was defatted with chloroform (3 ×
00 mL) and suspended in water (250 mL); after addition of methyl
3
, DQF-COSY) δ
3
3
3
).
1
hydrazine (5 g), the pH was adjusted to 4.0 using concentrated
hydrochloric acid. After the mixture had been incubated for 45 min at
8
0 °C, it was cooled to room temperature, the pH was adjusted to 7
using aqueous sodium hydroxide (1 mmol/L), and the solution was
extracted with methylene chloride (3 × 200 mL). The combined organic
layers were extracted with aqueous sodium hydroxide solution (0.1
mmol/L, 150 mL), and the aqueous phase was adjusted to pH 3.0 with
concentrated hydrochloric acid and was then again extracted with
methylene chloride (3 × 50 mL). A defined amount of 1-methylpyr-
rolidone in methanol was added as the internal standard, and, after
concentration, the extract was analyzed by HRGC/MS(CI). The amount
of pyrrolinone reductone was calculated from a calibration curve
determined from aqueous solutions containing defined amounts of
pronyl-glycine. Each of the experiments was performed in triplicate.
Preparation of Model Melanoidins. Gluten/starch and gluten/
glucose melanoidins were prepared by heating doughs prepared from
wheat gluten (40 g), water (150 mL), and starch (160 g) or glucose
(160 g), respectively, in a baking oven for 1 h at 220 °C. Gluten/
acetylformoin melanoidins were prepared by heating a mixture of
acetylformoin (500 mg), gluten (500 mg), and water (1.5 mL) for 30
min at 150 °C in an baking oven. For isolation of the high molecular
weight melanoidins, the thermally treated mixtures were suspended in
water (40 mL) and were freed from low molecular weight compounds
by ultrafiltration (Amicon, Witten, Germany) with a molecular weight
cutoff of 10 kDa (Diaflo YM 10). The low molecular weight fraction
was freeze-dried and then used for the quantitation experiments.
Preparation of Pronylated Bovine Serum Albumin (Pronyl-BSA).
A suspension of BSA (10.24 g) and acetylformoin (5.76 g) in phosphate
buffer (100 mL; 0.1 mol/L; pH 6.5) was incubated for 1 h at 80 °C,
with stirring. After cooling to room temperature, the mixture was placed
in a dialysis tubing (Sigma, Deisenhofen, Germany) with a molecular
weight cutoff of 12 kDa, and distilled water was added until the height
of the solution in the tube was ∼50 cm. The tubing was closed and
submerged in distilled water (4 L) at 4 °C. After 12 h, the water
surrounding the dialysis tubing was replaced with fresh water (4 L)
and dialysis continued for another 12 h at 4 °C. After this step had
been repeated twice, the content of the tubing was freeze-dried, yielding
pronyl-BSA (13.8 g) as a yellow powder, which was stored in a
desiccator. Quantitation of the content of BSA-linked pyrrolinone
reductone revealed an amount of 270 µg/g calculated as pronyl-lysine
corresponding to a lysine substitution of 1.2 nmol/mmol.
Nuclear Magnetic Resonance Spectroscopy (NMR). 1H, 13C,
DEPT-135, DQF-COSY, HMQC, and HMBC spectroscopies were
performed on AMD 360 and AMX 400 spectrometers (Bruker,
Rheinstetten, Germany), respectively.
Cell Culture Experiments. Caco-2 cells were obtained from the
German Collection of Microorganisms and Cell Cultures (DMSZ,
Braunschweig, Germany). Caco-2 cells (passages 11, 12, 15, and 16)

Pronyl-lysine as an Antioxidant Site of Bread Crust Melanoidins
J. Agric. Food Chem., Vol. 50, No. 24, 2002 7001
Figure 2. Effect of bread crumb and different fractions of bread crust on
the enzyme activity of NADPH−cytochrome c reductase (CCR) and
glutathione S-transferase (GST) in Caco-2 cells relative to nonexposed
controls ()100%). Treated Caco-2 cells were exposed to 0.5 g of the
individual bread crust or bread crust fraction per 100 mL of cell culture
medium for 48 h. GST*/CCR#: Enzyme activity of cells exposed to bread
crust fractions was significantly different from that of nonexposed control
cells (p < 0.05).
Figure 1. In vitro antioxidative activity of soluble fractions I−III isolated
from flour, bread crumb, and bread crust.
RESULTS AND DISCUSSION
To gain first insights into the antioxidative potential of bread
crust, a sourdough bread was freshly baked to a pleasant dark
brown crust, the crust was then carefully separated from the
crumb, and both materials were frozen in liquid nitrogen and
ground in a kitchen mill. The powder obtained from crumb and
crust, respectively, was then defatted by chloroform extraction
and then sequentially extracted with water, 60% aqueous
ethanol, and 50% aqueous 2-propanol, affording the correspond-
ing solvent fractions I-III. These fractions containing the
soluble compounds as well as the nonsoluble materials (fraction
IV) were freed from solvent in vacuo and then freeze-dried. In
comparison, the wheat/rye flour used for bread preparation was
fractionated following this procedure. The yields obtained for
the fractions I-IV of flour, bread crumb, and crust are given
in Table 1.
Antioxidant Activity of Bread Crust Fractions. To compare
their antioxidant potentials, the efficiencies of the soluble
fractions I-III, isolated from flour, crumb, and crust, in
inhibiting the peroxidation of linolenic acid were measured
following the procedure reported recently (19). Using Trolox
as the reference for a highly active antioxidant, the antioxidative
potential was calculated as Trolox equivalents (TE values). As
given in Figure 1, the antioxidative potential of all the fractions
isolated from bread crust exceeded the activities measured for
the flour and crumb isolates. In particular, the dark brown
ethanol solubles (fraction II) of the crust showed the highest
activity; for example, 6-fold higher TE values were measured
in comparison to the corresponding fraction isolated from flour
and the bread crumb, respectively. Also, the crust fraction III
showed a high antioxidative activity of ∼1.0 TE value. In com-
parison, the pale fractions isolated from the bread crumb did
not show higher activities that those isolated from the flour
(GST), have been identified and associated with increased risk
for colon cancer (21). On the other hand, induction of phase II
enzymes by antioxidants represents a promising strategy for
cancer prevention. Although the molecular mechanism by which
antioxidants bind to an antioxidant response element resulting
in the specific, monofunctional induction of phase II enzymes
has been intensively studied so far (24), it is still an open
question as to whether a compound showing an antioxidant
activity in vitro may function as a phase II inducer in biological
systems. Among all of the biological systems suitable for in
vitro studies, the intestinal Caco-2 cell line is widely used to
investigate the effects of dietary compounds on xenobiotic
enzymes as the colon is clearly one of the most likely sites for
the development of different types of dietary-induced cancers
(27, 31).
In the present study, intestinal Caco-2 cells exposed to bread
crust and the respective fractions I-III showed a significantly
elevated GST activity compared to nonexposed control cells
and compared to those cells exposed to bread crumb. In contrast,
phase I NADPH-cytochrome c reductase (CCR) activity was
inhibited by all of the fractions that activated the GST enzyme
(Figure 2). Thus, CCR and GST activities in intestinal Caco-2
cells exposed to bread crust and its isolated fractions were
modulated in such a way as to be interpreted as a protective
functional parameter of an antioxidant activity in vitro.
To gain further insights into the molecular weight of those
crust compounds exhibiting the highest antioxidative potential,
the most active, ethanolic fraction II was further fractionated
using multistep ultrafiltration. Using membrane filters with
stepwise decreasing molecular weight cutoffs, five fractions were
obtained containing ethanol-soluble crust compounds with
molecular weights of <1 kDa (fraction II/1), 1-10 kDa (fraction
II/2), 10-30 kDa (fraction II/3), 30-100 kDa (fraction II/4),
and >100 kDa (fraction II/5). From fraction II/1 to II/5 a
significant increase in browning intensity was observable, thus
demonstrating that the browning is due to high molecular weight
melanoidins. As a reference, the corresponding fraction II
isolated from the flour and showing only a very low antioxi-
dative potential (Figure 1) was fractionated by ultrafiltration.
As summarized in Table 2, nearly 69% of the ethanol-soluble
crust compounds were obtained in fraction II/1 and exhibited
molecular weights of <1 kDa. In addition, 21.2% of these
compounds showed molecular weights between 1 and 10 kDa,
(Table 1). These data clearly indicate that during bread baking,
nonenzymatic browning reactions, which are obviously favored
in the bread crust, lead to the formation of reaction products
with antioxidative potential in vitro.
To investigate whether these reaction products showing a
significant antioxidant activity in vitro may also act as antioxi-
dants in biological systems, their enzyme modulating activity
on detoxification enzymes was investigated as a functional
parameter in intestinal Caco-2 cells. In general, detoxification
enzymes protect cells from a wide variety of nonendogenously
formed xenobiotics and endogenous toxins. Current data suggest
that the balance between the phase I carcinogen-activating
enzymes and the phase II detoxifying enzymes is critical to
determining an individual’s risk for cancer. Human deficiencies
in phase II enzyme activity, specifically glutathione S-transferase

7002 J. Agric. Food Chem., Vol. 50, No. 24, 2002
Lindenmeier et al.
Figure 4. In vitro antioxidative activity of binary mixtures of carbohydrate
R
precursors and N -acetyl-L-lysine methyl ester prior to and after thermal
treatment, respectively.
Figure 3. In vitro antioxidative activity of molecular weight fractions
obtained by multistep ultrafiltration of fraction II isolated from flour, bread
crumb, and crust, respectively.
with NR-acetyl-L-lysine methyl ester, chosen as a suitable model
substance for the ꢀ-amino group of protein-bound lysine. The
individual binary reaction mixtures were analyzed for antioxi-
dative activity in vitro prior to and after thermal treatment. The
results, summarized in Figure 4, show that under the mild
reaction conditions applied, neither starch nor glucose is able
to produce high amounts of antioxidants. Experiments with
whereas ∼6.2% consisted of macromolecules with molecular
weights of >10 kDa. Also, the crumb fraction II consisted
mainly of low molecular weight compounds; for example, 73.6%
was present in fraction II/1. Comparing these data with those
obtained from the flour fraction II clearly showed major
differences (Table 2); for example, only 40.3% of the flour
components showed molecular weights of <1 kDa, whereas
3
-deoxyosone as well as 1-deoxyosone revealed that, in
particular, the 1-deoxyosone-derived reaction products showed
high antioxidative potential; for example,e.g. 5.4 TE values were
measured for the experiment including 1-deoxyosone (Figure
4
2.4% consisted of macromolecules with molecular weights of
30 kDa. These differences clearly demonstrate that the baking
>
process induced major changes in the chemical composition of
the flour.
4
). In comparison, 3-deoxyosone showed only 25% of the
antioxidative potential found for 1-deoxyosone. The highest
activity was, however, found for the reaction mixture containing
acetylformoin, a major dehydration product formed from
These fractions obtained by ultrafiltration were then used to
estimate the molecular weights of the antioxidants in bread crust,
crumb, and flour (Figure 3). Crust fraction II/4, containing
compounds with a molecular weight between 30 and 100 kDa,
was found to be most efficient in inhibiting linolenic acid
peroxidation, closely followed by fractions II/3 and II/5.
However, the low molecular weight fraction II/1 showed only
low activity (Figure 3). In comparison, the highest antioxidative
activity in the crumb was detectable in fractions II/2 and II/3
containing compounds with a molecular weight between 1 and
1-deoxyosone. Measuring the TE values of these mixtures prior
to thermal treatment clearly showed that all of the hexose-
derived intermediates show antioxidative potential per se, among
which 1-deoxyosone and acetylformoin were the most active
due to their reductone-type structures. Comparing these data
with those found after heating, however, clearly showed that
the reaction of, in particular, 1-deoxyosone and acetylformoin
with the lysine derivative led to the formation of reaction
products with high antioxidative potential.
On the basis of these findings, the most active precursor
mixture, consisting of acetylformoin and NR-acetyl-L-lysine
methyl ester, was selected to elucidate the active key compounds
by a screening procedure. To achieve this, the reaction mixture
was separated by RP-HPLC, and the effluent was monitored
using either a diode array detector (DAD) operating at wave-
lengths between 210 and 500 nm or an LC/MS. The effluent
was separated into 26 fractions, which were separately collected
in glass vials (Figure 5A). After freeze-drying, the residues
obtained from these HPLC fractions were taken up in the same
amount of water and were then used to measure the antioxidative
activity using the in vitro assay (Figure 5B). The results clearly
showed by far the highest antioxidative activity for the fraction
14; for example, 4.5 TE values were determined. With the
exception of fraction 7, the antioxidative potential of all the
other HPLC fractions did not exceed the activity of 1.0 TE value,
thus indicating the active principle in fraction 14 to be the key
compound responsible for the high antioxidative potential of
the heated Maillard mixture.
3
0 kDa. The crumb fractions II/4 and II/5 were significantly
less active than the corresponding crust fractions containing the
dark brown melanoidins. None of the molecular weight fractions
isolated from the ethanolic flour extract showed any significant
antioxidative potential. These results indicate that macromol-
ecules exhibiting antioxidant potential are formed nonenzy-
matically from inactive precursors during the baking process.
Structure of an Antioxidant Site of Melanoidins. Because
brown melanoidins were shown to be formed upon carbohydrate-
induced modifications and/or cross-linking of reactive amino
acid side chains in food proteins (9, 10) and acidic hydrolysis
of the crust fractions II/3-II/5 revealed all of the proteinogenic
amino acids, it was hypothesized that the antioxidatively active
sites of the crust macromolecules might be due to Maillard-
type modifications of flour proteins. It is well-known from the
literature that, besides arginine residues, in particular, the
ꢀ-amino group of lysine side chains as well as the amino group
of the N terminus are the primary targets for nonenzymatic
browning reactions with carbohydrate degradation products (9).
To gain first insights into the efficiency of carbohydrates and
carbohydrate degradation products in transforming the ꢀ-amino
group of protein-bound lysine into reactions products with
antioxidative activity, starch, glucose, and the hexose degrada-
tion products 3-deoxy-2-hexosulose and 1-deoxy-2,3-hexo-
diulose, as well as acetylformoin, a well-known dehydration
product formed from 1-deoxy-2,3-hexodiulose, were reacted
The following identification experiments were, therefore,
focused on the main reaction product present in fraction 14.
After chromatographic isolation, the determination of its
1
chemical structure (1 in Chart 1) was performed by H NMR,
LC/MS, and UV-vis spectroscopy. LC/MS measurements of

Pronyl-lysine as an Antioxidant Site of Bread Crust Melanoidins
J. Agric. Food Chem., Vol. 50, No. 24, 2002 7003
Figure 6. HPLC chromatogram (A) and in vitro antioxidative potential of
fractions collected (B) from a heated solution of acetylformoin and glycine
methyl ester.
Figure 5. HPLC chromatogram (A) and in vitro antioxidative potential of
fractions collected (B) from a heated solution of acetylformoin and
R
N -acetyl-L-lysine methyl ester.
absorption maximum at 363 nm and the molecular weight of
1
215 Da as well as the resonance signals in the H NMR spectrum
the purified compound, exhibiting an absorption maximum at
were very well in line with the structure of 2,4-dihydroxy-2,5-
dimethyl-1-methoxycarbonylmethyl-3-oxo-2H-pyrrole (2), shown
in Chart 1. The structure of the proposed pyrrolinone ring was
+
363 nm, gave an intense [M + 1] ion at m/z 329 (86%) with
a loss of 18 to the base peak at m/z 311 (100%), most
likely due to the cleavage of one molecule of water. In addition,
LC-MS/MS experiments revealed the loss of 42 from m/z 311
to 269 and, in addition, the loss of 59 to m/z 251 corresponding
most likely to the elimination of ketene from the acetylamino
group and of the methoxycarbonyl group from the ester function,
respectively. These data corroborate the N-acetyllysine methyl
ester moiety in the proposed structure of 2,4-dihydroxy-2,5-
dimethyl-1-(5-acetamino-5-methoxycarbonylpentyl)-3-oxo-2H-
pyrrole (1) outlined in Chart 1.
13
unequivocally confirmed by C NMR and heteronuclear
chemical shift experiments. Although this compound was
reported earlier as a Maillard reaction product formed from the
Amadori product 1-deoxy-1-piperidino-D-fructose and glycine
esters (32), its function as an antioxidant has as yet not been
reported.
To confirm the antioxidative potential of these purified
pyrrolinone reductones, compounds 1 and 2, and, in comparison,
ascorbic acid, were analyzed using the in vitro antioxidant assay.
Both pyrrolinone reductones, 1 and 2, showed high antioxidative
activities of 0.53 and 0.49 TE, respectively, whereas the ascorbic
acid showed a lower activity by a factor of 5.
Quantitation of Protein-Bound Pyrrolinone Reductones.
Neither acidic hydrolysis nor enzymatic hydrolysis revealed high
recoveries of pyrrolinone reductones (data not shown). Because
the pyrrolinone reductonyl moiety, which we abbreviate “pro-
nyl”, is fixed to the protein backbone via the lysine side chain
as pronyl-lysine and via the N terminus, respectively, this
carbohydrate modification was cleaved from the protein and
subsequently transformed into 5-acetyl-4-hydroxy-1,3-dimeth-
ylpyrazole upon hydrazinolyses using methyl hydrazine (Figure
7). After this derivatization, solvent extraction, and simple
cleanup, the amount of pyrrolinone reductones could be quanti-
fied as the stable 5-acetyl-4-hydroxy-1,3-dimethylpyrazole by
means of HRGC/MS(CI). In control expertiments, it could be
shown that heating carbohydrates in the presence of methyl
hydrazine does not generate 5-acetyl-4-hydroxy-1,3-dimethyl-
pyrazole (data not shown). Thus, this method offered the possi-
bility to quantitate the total degree of pronylation of the N termi-
nus as well as the lysine side chains, calculated as pronyl-lysine.
To gain insights into the efficiency of carbohydrates and
carbohydrate degradation products in pyrrolinone reductone
formation during bread baking, binary model mixtures contain-
ing wheat gluten and either starch, glucose, or acetylformoin,
respectively, were thermally treated in an baking oven, the high
molecular weight melanoidins were isolated by ultrafiltration,
and the amounts of pronyl-lysine formed were determined. The
data, summarized in Table 3, showed that acetylformoin was
most efficient in generating high mounts of pyrrolinone reduc-
1
The H NMR spectrum of compound 1 showed nine reso-
nance signals. Double-quantum-filtered δ,δ-correlation spec-
troscopy (DQF-COSY) revealed strong H,H-shift correlations
between the hydrogens H-C(7)/H-C(8), H-C(8)/H-C(9),
H-C(9)/H-C(10), and H-C(10)/H-C(11) and, in addition, the
two singlets at 1.98 and 3.72 ppm, corresponding to an acet-
amide group and a methoxy group, respectively, thus confirming
the presence of the NR-acetyl-L-lysine methyl ester moiety in
compound 1. In addition, two singlets were detected resonating
at 1.34 and 2.22 ppm and were assigned as the methyl groups
C(1) and C(6) in structure 1. Due to the instability of compound
1
during concentration, it was, however, not possible to isolate
13
the target compound in sufficient amounts to perform C NMR
studies for the final structure confirmation.
With the expectation that probably another amino acid
derivative might be more stable for doing C NMR studies and
to further confirm the proposed reductone structure of compound
1
3
1
, an additional set of experiments was performed in which the
lysine derivative was substituted by glycine methyl ester as a
model for the N terminus of a protein. An equimolar mixture
of acetylformoin and glycine methyl ester was thermally treated,
cooled to room temperature, and then screened for the most
active antioxidant by using the HPLC screening assay reported
above (Figure 6). From the 26 fractions isolated, fraction 9 was
found to have by far the highest antioxidative potential. For
this fraction, an activity of nearly 4 TE values were determined,
whereas all of the other fractions did not exceed an activity of
1
.5 TE values. Isolation and purification afforded the target
compound as yellow crystals, which were stable enough to
perform UV-vis, GC/MS, and H and C NMR studies. The
1
13

7004 J. Agric. Food Chem., Vol. 50, No. 24, 2002
Lindenmeier et al.
Figure 8. Concentrations of pyrrolinone reductones (calculated as pronyl-
lysine) in individual bread crust fractions.
Figure 7. Cleavage of pronylated proteins by hydrazinolysis using methyl
hydrazine and formation of 5-acetyl-4-hydroxy-1,3-dimethylpyrazole.
Table 3. Amounts of Protein-Bound Pyrrolinone Reductones
(Calculated as Pronyl-lysine) in Model Melanoidins Generated by
Heating Wheat Gluten in the Presence of Carbohydrate Sources under
Baking Conditions
a
carbohydrate precursor
starchb
concn of pronyl-L-lysine (mg/kg)
4.0 (± 3.4−4.6)
18.9 (± 15.2−22.6)
nd
b
glucose
glucose
c
d
acetylformoin
7100.0 (± 6035.0−8165.0)
Figure 9. Influence of the molecular weight on the concentrations of
pyrrolinone reductones (calculated as pronyl-lysine) in crust fraction II.
Molecular weights of fractions II/1−II/5 are given in Table 2.
a
Glucose (160 g) was heated in the absence of wheat gluten. nd, not detectable.
b
A mixture of acetylformoin (500 mg), wheat gluten (500 mg), and water (1.5 mL)
was heated for 30 min at 150 °C in an baking oven. The concentration is calculated
as milligram per kilogram of the total reaction mixture. A mixture of wheat gluten
c
d
To gain further insights into the distribution of these protein-
bound antioxidants in the crust fractions, the amount of
pyrrolinone reductones, calculated as pronyl-lysine, was deter-
mined in fractions I-IV (Figure 8). All of the fractions analyzed
contained the pyrrolinone reductone; however, by far the highest
amounts were found in the ethanolic fraction II. The high
amount of pronyl-lysine, determined to be 169.3 mg/kg in
fraction II, is well in line with the high antioxidative activity
of this fraction (Figure 1). In comparison, fractions III and I
contained only 42.9 and 30.6 mg/kg pronyl-lysine, respectively,
thus corresponding well with the lower antioxidative activity
of the crust compounds extracted by water or 2-propanol. It is
interesting to note that the nonsoluble crust materials (fraction
IV) also contain rather high amounts of pyrrolinone reductones,
for example, 88.5 mg/kg pronyl-lysine. As the insolubility of
this fraction does not allow the measurement of the antioxidative
activity in vitro, the quantitation of pronyl-lysine as an indicator
for thermally generated, protein-linked antioxidants might be a
suitable tool to gain some insights into their antioxidative status.
(
40 g), water (150 mL), and starch (160 g) or glucose (160 g), respectively, was
heated in a baking oven for 1 h at 220 °C.
Table 4. Amounts of Protein-Bound Pyrrolinone Reductone (Calculated
as Pronyl-lysine) in Flour, Bread Crumb, and Crust
concn of pronyl-
L-lysine (mg/kg)
lysine substitution
(nmol/mmol)
a
sample
flour
bread crumb
bread crust
nd
nd
1.40
11.24
8.0 (± 6.8−9.2)
62.2 (± 52.9−71.5)
a
Degree of pronylation was calculated as pronyl-L-lysine. nd, not detectable.
tones under baking conditions, for example, 7100 mg/kg pronyl-
lysine was formed. Also, glucose and starch were precursor
active in the presence of gluten but produced significantly lower
amounts of pronyl-lysine when compared to acetylformoin. A
control experiment, in which glucose was heated in the absence
of gluten, did not show any amounts of the pronyl-lysine. These
data clearly indicate that the reaction of lysine side chains and
the N termini of flour proteins, respectively, with starch or
hexoses is efficient in generating protein-bound pyrrolinone
reductones under baking conditions via acetylformoin as the
penultimate precursor.
To gain insights into the degree of protein pronylation in the
flour proteins, the pyrrolinone reductone, calculated as pronyl-
lysine, was determined in flour, bread crumb, and bread crust.
The results given in Table 4 reveal by far the highest amounts
in the dark bread crust; for example, ∼62.2 mg/kg pronyl-lysine
was present in bread crust, corresponding to a lysine substitution
of 11.24 nmol/mmol. In comparison, the pale bread crumb
contained these pyrrolinone reductones in 8-fold lower amounts,
whereas the nonthermally treated flour did not show any
significant amounts of pronylated proteins (Table 4).
To study the influence of the molecular weight of the ethanol-
soluble crust browning products on their amount of pyrrolinone
reductone, pronyl-lysine was quantitated in the molecular weight
fractions II/1-II/5 obtained by ultrafiltration of the ethanolic
crust extract. The quantitative data, given in Figure 9, clearly
show that the highest amounts of pyrrolinone reductones were
present in the macromolecular browning products; for example,
52.3 and 35.2 mg/kg pronyl-lysine were determined in fractions
II/4 and II/5, respectively. In comparison, the low molecular
weight fraction II/1 contained only very low concentrations.
Comparison of the amounts of pronyl-lysine (Figure 9) and
the antioxidative activities (Figure 3) of these molecular weight
fractions showed a close correlation between the antioxidative
potential and the concentration of the pronyl-lysine, again
demonstrating pronyl-lysine as a suitable indicator substance

Pronyl-lysine as an Antioxidant Site of Bread Crust Melanoidins
J. Agric. Food Chem., Vol. 50, No. 24, 2002 7005
by using ultracentrifugation and color dilution techniques. J.
Agric. Food Chem. 1998, 46, 3891-3895.
2) Maier, H. G.; Buttle, H. Isolation and characterization of brown
roast compounds of coffee (in German). Z. Lebensm. Unters.
Forsch. 1973, 150, 331-334.
(
(
3) Steinhart, H.; M o¨ ller, A.; Kletschkus, H. New aspects in the
analysis of melanoidins in coffee with liquid chromatography.
1
3th Colloque Scientifice Internationale sur le Cafe: ASIC:
Paris, France, 1989.
(4) Hofmann, T.; Schieberle, P. Chemical interactions between odor-
active thiols and melanoidins involved in the aroma staling of
coffee beverages. J. Agric. Food Chem. 2002, 50, 319-326.
(5) Hashiba, H. Non-enzymatic browning of soy saucesUse of an
ion-exchange resin to identify types of compounds involved in
oxidative browning. Agric. Biol. Chem. 1973, 37, 871-877.
(6) Obretenov, T. D.; Kuntcheva, M. J.; Mantchev, S. C.; Valkova,
G. D. Isolation and characterization of melanoidins from malt
and malt roots. J. Food Biochem. 1991, 15, 279-294.
Figure 10. Effect of pronyl-BSA and pronyl-glycine (3) on the enzyme
activity of NADPH−cytochrome c reductase (CCR) and glutathione
S-transferase (GST) in Caco-2 cells relative to nonexposed controls
()100%). During the exposure time of 48 h, Caco-2 cells were exposed
to the different compounds dissolved into the cell culture medium at final
concentrations corresponding to the content of 0.03 mg of pyrrolinone
reductone (calculated as pronyl-lysine) per 0.5 g of bread crust: 0.11 g
of pronyl-BSA/100 mL, 0.11 g of BSA/100 mL, and 0.03 mg of pronyl-
glycine/100 mL of medium for 48 h. *: Enzyme activity of exposed cells
was significantly different from that of nonexposed control cells (p < 0.05).
(7) Kuntcheva, M. J.; Obretenov, T. D. Isolation and characterization
of melanoidins from beer. Z. Lebensm. Unters. Forsch. 1996,
202, 238-243.
(
8) Hofmann, T.; Bors, W.; Stettmaier, K. On the radical-assisted
melanoidin formation during thermal processing of foods as well
as under physiological conditions. J. Agric. Food Chem. 1999,
4
7, 391-396.
(
9) Hofmann, T. Structure, color and formation of low and high
molecular weight products formed by food-related Maillard-type
reactions. In Chemistry and Physiology of Food Colors; Ames,
J., Hofmann, T., Eds.; ACS Symposium Series 775; American
Chemical Society: Washington, DC, 2001.
to quantitatively monitor the thermal generation of antioxidants
during baking processes.
Degree of pronylation is, however, tightly connected to the
higher molecular weight crust melanoidins. To get closer to the
real food system, “pronylated” protein, namely, BSA, was
prepared by reacting acetylformoin and purified BSA under
mild, aqueous conditions. Model experiments showed that under
these mild conditions used, the lysine side chain was the only
amino acid rapidly reacting with acetylformoin to give a lysine
substitution of 1.2 nmol/mmol.
(
10) Hofmann, T.; Bors, W.; Stettmeier, K. CROSSPYsA radical
intermediate of melanoidin formation in roasted coffee. In Free
Radicals in FoodsChemistry, Nutrition and Health Effects;
Morello, M. J., Sahidi, F., Ho, C.-T., Eds.; ACS Sympsoium
Series 807; American Chemical Society: Washington, DC, 2002.
11) Hiramoto, K.; Aso-o, R.; Ni-iyama, H.; Hikage, S.; Kato, T.;
Kikugawa, K. DNA strand break by 2,5-dimethyl-4-hydroxy-
(
To study the effect of pronylated amino acids on the activity
of detoxification enzymes, in comparison to the results obtained
for the bread crust fractions (Figure 1), Caco-2 cells were
exposed to either pronyl-BSA or purified pronyl-glycine (3 in
Chart 1), liberated from the corresponding methyl ester 2 by
esterase treatment. After an exposure time of 48 h, phase II
GST activity of cells exposed to pronyl-BSA and to pronyl-
glycine was increased by 12 and 34%, respectively (Figure 10).
A significant decrease in the CCR enzyme activity of ∼12%
compared to controls was analyzed for those cells exposed to
pronyl-glycine. These results clearly show that the most effective
compound in modulating the activity of phase II GST enzyme
is the pronylated amino acid; its protein-linked form, as present
in crust melanoidins, also shows significant activity. In intestinal
Caco-2 cells at least, pronylated proteins act as a monofunctional
inducer of GST, serving as a functional parameter of an
antioxidant, chemopreventive activity in vitro.
3
(2H)-furanone, a fragrant compound in various foodstuffs.
Mutat. Res. 1996, 359, 17-24.
(
12) Hiramoto, K.; Ishihara, A.; Sakui, N.; Daishima, S.; Kikugawa,
K. Formation of hydroxyfuranone and hydroxypyranone deriva-
tives with DNA-breaking activity in the Maillard reaction of
glucose and albumin under physiological conditions. Biol. Pharm.
Bull. 1998, 21, 102-104.
13) Marko, D.; Kem e´ ny, M.; Bernady, E.; Habermeyer, M.; Weyand,
U.; Meiers, S.; Frank, O.; Hofmann, T. Studies on the inhibition
of tumor cell growth and microtubule assembly by 3-hydroxy-
4-[(E)-(2-furyl)methylidene]methyl-3-cyclopentene-1,2-dione, an
intensively coloured Maillard reaction product. Food Chem.
Toxicol. 2002, 40, 9-18.
(
(
14) Eichner, K. Antioxidative effect of Maillard reaction intermedi-
ates. Prog. Food Nutr. Sci. 1981, 5, 441-451.
(
15) Lingnert, H.; Eriksson, C. E.; Waller, G. R. Stability of
antioxidants formed from histidine and glucose by the Maillard
reaction. J. Agric. Food Chem. 1983, 31, 27-30.
The question of whether dietary pronylated proteins present
in bread crust indeed are physiologically active antioxidants,
being absorbed in the intestine and metabolized endogenously,
is currently under investigation in an animal feeding study.
(
16) Ivainsky, H.; Franzke, Cl. On the antioxidant activity of
melanoidins (in German). Dtsch. Lebensm. Rundsch. 1956, 52,
1
29-133.
(17) Nicoli, M. C.; Anese, M.; Parpinel, M. T.; Franceschi, S.; Lerici,
C. R. Loss and/or formation of antioxidants during food
processing and storage. Cancer Lett. 1997, 114, 71-74.
ACKNOWLEDGMENT
(
18) Nicoli, M. C.; Anese, M.; Manzocco, L.; Lerici, C. R. Antioxidant
properties of coffee brews in relation to the roasting degree.
Lebensm. Wiss. Technol. 1997, 30, 292-298.
We thank K. Kahlenberg, K. Harnack, and K. Krome for skillful
technical assistance.
(19) Bright, D.; Stewart, G. G.; Patino, H. A novel assay for the in
antioxidant potential of speciality malts. J. Am. Soc. Brew. Chem.
LITERATURE CITED
1
999, 57, 133-137.
(
1) Hofmann, T. Studies on the relationship between molecular
weight and the color potency of fractions obtained by thermal
treatment of glucose/amino acid- and glucose/protein-solutions
(20) Cheigh, H.-S.; Lee, J. S.; Lee, C. Y. Antioxidant characteristics
of melanoidin related products fractionated from fermented
soybean sauce. J. Korean Soc. Food Nutr. 1993, 22, 570-575.

7006 J. Agric. Food Chem., Vol. 50, No. 24, 2002
Lindenmeier et al.
(
21) Wilkinson, J.; Clapper, M. L. Detoxication enzymes and chemo-
prevention. Proc. Soc. Exp. Biol. Med. 1997, 216, 192-200.
22) Wenzel, E.; Tasto, S.; Erbersdobler, H. F.; Faist, V. Effect of
heat treated proteins on selected parameters of the biotransfor-
mation system in the rat. Ann. Nutr. Metab. 2002, 46, 9-16.
23) Faist, V.; Erbersdobler, H. F. Metabolic transit and in vivo effects
of melanoidins and precursor compounds deriving from the
Maillard reaction. Ann. Nutr. Metab. 2001, 45, 1-12.
(29) Habig, W. H.; Pabst, M. J.; Jakoby, W. B. Glutathione-S-
transferases: the first enzymatic step in mercapturic acid
formation. J. Biol. Chem. 1974, 249, 7130-7139.
(
(
(
(
(
30) Lowry, O. H.; Rosebrough, N. J.; Farr, A. L.; Randall, R. J.
Protein measurement with the Folin phenol reagent. J. Biol.
Chem. 1951, 193, 265-275.
(
31) Schmiedlin-Ren, P.; Thummel, K. E.; Fisher, J. M.; Paine, M.
F.; Lown, K. S.; Watkins, P. B. Expression of enzymatically
active CYP3A4 by Caco-2 cells grown on extracellular matrix-
coated permeable supports in the presence of 1R,25-dihydrox-
yvitamin D3. Mol. Pharmacol. 1997, 51, 741-754.
24) Prochaska, H. J.; Talalay, P. Regulatory mechanisms of mono-
functional and bifunctional anticarcinogenic enzyme inducers in
murine liver. Cancer Res. 1988, 48, 4776-4782.
25) Hofmann, T.; Schieberle, P. Quantitative model studies on the
effectivity of different precursor systems in the formation of the
intense food odorants 2-furfurylthiol and 2-methyl-3-furanthiol.
J. Agric. Food Chem. 1998, 46, 235-241.
(
32) Ledl, F.; Fritsch, G. Formation of pyrrolinone reductones upon
heating of hexoses with amino acids (in German). Z. Lebensm.
Unters.-Forsch. 1984, 178, 41-44.
(
(
(
26) Hofmann, T. Acetylformoinsa chemical switch in the formation
of colored Maillard reaction products from hexoses and primary
and secondary amino acids. J. Agric. Food Chem. 1998, 46,
Received for review June 3, 2002. Revised manuscript received August
22, 2002. Accepted August 22, 2002. This research project was
supported by the FEI (Forschungskreis der Ern a1 hrungsindustrie e.V.,
Bonn, Germany), the AiF, the Ministry of Economics and Technology
3
918-3928.
27) Peters, W. H. N.; Roelofs, H. M. J. Time-dependent activity and
expression of glutathione S-transferases in the human colon
adenocarcinoma cell line Caco-2. Biochem. J. 1989, 264, 613-
(Project AIF-FV 12403 N), and the Th oud der Chemischen Industrie.
6
16.
We also thank the European Commission for supporting the research
in the frame of the European Union COST Action 919 “Melanoidins
in Food and Health”.
28) Masters, B. S. S.; Williams, C. H.; Konnin, H. The preparation
and properties of microsomal TPNH cytochrome c reductase
from pig liver. In Methods in Enzymology; Eastabroook, R. W.,
Pullman, M. E., Eds.; Academic Press: New York, 1967; Vol.
1
0.
JF020618N