Nδ-(5-Hydroxy-4,6-dimethylpyrimidine-2-yl)-L-ornithine, a novel methylglyoxal - Arginine modification in beer

Article abstract of DOI:10.1021/jf000493r

N^δ-(5-Hydroxy-4,6-dimethylpyrimidine-2-yl)-L-ornithine, or Argpyrimidine, was identified and quantified in beer by high-performance liquid chromatography (HPLC) and coupled gas chromatography-mass spectrometry (HRGC-MS). This novel fluorescent

Full text of DOI:10.1021/jf000493r

366
J. Agric. Food Chem. 2001, 49, 366−372
δ
N -(5-Hyd r oxy-4,6-d im eth ylp yr im id in e-2-yl)-L-or n ith in e, a Novel
Meth ylglyoxa l-Ar gin in e Mod ifica tion in Beer
,
†
†
‡
Marcus A. Glomb,* Daniel R o¨ sch, and Ramanakoppa H. Nagaraj
Institute of Food Chemistry, Technical University of Berlin, Gustav-Meyer-Allee 25, TIB4/3-1,
1
3355 Berlin, Germany, and Center for Vision Research, Department of Ophthalmology,
University Hospitals of Cleveland, Cleveland, Ohio
δ
N -(5-Hydroxy-4,6-dimethylpyrimidine-2-yl)-L-ornithine, or Argpyrimidine, was identified and
quantified in beer by high-performance liquid chromatography (HPLC) and coupled gas chroma-
tography-mass spectrometry (HRGC-MS). This novel fluorescent arginine Maillard modification
represents the first amino acid modification reported in beer retaining the full backbone of the
original amino acid. Two mechanisms of formation could be verified: the major pathway via
methylglyoxal and the minor pathway via 5-deoxypentoses. Argpyrimidine concentrations, deter-
mined in 35 lager-type beer varieties, reached up to 27 nmol/L and could be positively correlated to
beer color and wort content. Within this context, 5-deoxy-D-ribose was identified as a novel
intermediate of the Maillard reaction of maltose by HRGC-MS and independent synthesis.
δ
Keyw or d s: N -(5-Hydroxy-4,6-dimethylpyrimidine-2-yl)-L-ornithine; Argpyrimidine; 5-deoxypentose;
5
-deoxy-D-ribose; Maillard reaction; beer; amino acid modification
INTRODUCTION
to oligosaccharides, maltose, and glucose. In addition,
the reaction is promoted by the high temperatures
applied during kilning, mashing, and wort boiling. As
proline is the dominating amino acid in beer, research
on modifications has been exclusively focused on proline-
derived structures. Although countless molecules were
identified in model reactions of proline with maltose and
glucose, only six could be verified in malt, wort, and
beer, representing pyrrolizines, pyridines, azepines, and
oxazines (16-21). Molecules published still incorporate
the intact pyrrolidine ring, but lost the carboxylic
function of the original amino acid skeleton. The notion
that methylglyoxal is formed in beer and second, that
beer also contains substantial amounts of arginine,
led us to investigate the formation of Argpyrimidine in
this beverage. This paper describes the identification
and quantitation of this novel arginine-derived structure
in beer and verification of an alternative synthesis
pathway.
Methylglyoxal is a known R-dicarbonyl within the
Maillard reaction of reducing sugars such as maltose
(1). In general, this reaction leads initially to the
formation of an aminoketose or Amadori product, which
is further degraded to yield a very complex reaction
mixture having substantial influence on aroma, color,
and amine modifications of foods. It is generally
accepted that the Maillard reaction proceeds via
highly reactive intermediates with R-dicarbonyl moiety
like methylglyoxal. Several suggestions have been
reported for the synthesis of methylglyoxal within
the Maillard reaction (2-5), but the final mechanism
remains unclear. As a consequence of the Maillard
reaction, methylglyoxal has been identified and quanti-
fied in thermally treated foods such as caramel, coffee,
and cocoa, but also in fermented foods such as wine
(
6-8). The high reactivity of methylglyoxal leads to
modification of functional groups from free or protein-
Ma ter ia ls. Chemicals of the highest quality available
were obtained from Aldrich (Steinheim, Germany) and
Fluka (Neu-Ulm, Germany), unless otherwise indicated.
Beer samples were purchased from local retail stores.
Ch r om atogr aph y. Thin-layer chromatography (TLC)
was performed on silica gel 60 F254 plates (Merk,
Darmstadt, Germany). Visualization of separated mate-
rial was achieved with the detection reagent given.
Preparative column chromatography was performed on
silica gel 60, 63-200 µm (Merk), unless otherwise noted.
Solvents were all chromatographic grade. From the
individual chromatographic fractions solvents were
removed under reduced pressure.
bound amino acids, especially lysine and arginine
δ
(
9-12). Recently, we succeeded in identifying N -(5-
Hydroxy-4,6-dimethylpyrimidine-2-yl)-L-ornithine, or
Argpyrimidine, in physiological reaction mixtures of
methylglyoxal with proteinbound arginine and in vivo
(13). Within this context the formation of an imidazolone
has been proposed (14). However, ongoing studies prove
this structure to be a misinterpretation of spectroscopic
data and to be identical to Argpyrimidine (15; Glomb
and Nagaraj, unpublished).
Beer is traditionally manufactured from water, hops,
malt, and yeast. During the technology of beer brewing
all essential substrates for the Maillard reaction are
provided. Proteins are enzymatically broken down to
smaller peptides and amino acids, starch is broken down
Ma gn etic Reson a n ce Sp ectr oscop y (NMR). NMR
spectra were recorded on a Bruker AC 200 instrument
(Rheinstetten, Germany). Chemical shifts are relative
to residual nondeuterated solvent as the internal refer-
ence.
H igh -R esolu t ion Ga s Ch r om a t ogr a p h y-Ma ss
Sp ect r om et r y (H R GC-MS). HRGC-MS was per-
*
To whom correspondence should be addressed (e-mail
glom1530@mailszrz.zrz.tu-berlin.de; fax ++49-30-314-72812).
†
Technical University of Berlin.
University Hospitals of Cleveland.
‡
1
0.1021/jf000493r CCC: $20.00 © 2001 American Chemical Society
Published on Web 12/06/2000

Maillard Reactions in Beer
J. Agric. Food Chem., Vol. 49, No. 1, 2001 367
formed on a HRGC 6160 Mega Series (Fisons Instru-
ments, Mainz, Germany); quartz capillary column (30
m, inner diameter 0.25 mm, DB-5, 0.25 µm, He, 37.3
cm/s, J &W Scientific, Cologne, Germany); injection port,
2,3-O-Isopropylidene-5-deoxy-D-ribono-1,4-lactone 6.
2.54 g (2.54 mmol) of calcium carbonate and Raney
nickel catalyst were added to a solution of 0.76 g of (2.56
mmol) 5 in 12 mL of anhydrous ethanol. Completeness
of reaction was checked by HRGC-MS (after about 1
h). After the reaction mixture was filtered, solvents were
evaporated, and the residue was subjected to column
chromatography (4:1 hexanes/EtAc). Fractions with 6
(HRGC-MS) were combined, and the solvents were
evaporated to yield a colorless oil (0.34 g, 77%).
2
70 °C; temperature program, after injecting the samples
at 100 °C the temperature of the oven was raised at 5
-
1
-1
°
°
C‚min to 200 °C, then raised at 10 °C‚min to 270
C, and held for 10 min. For MS analysis the HRGC
was connected to a MAT ITD 700 (Finnigan, Bremen,
Germany); transfer line, 280 °C; EI at 70 eV.
High -Resolu tion Ma ss Sp ectr om etr y (HRMS).
HRMS was applied on a VG 7070 (VG, Manchester, UK)
with heptacosane as the internal standard.
+
•
HRGC-MS: tR 7.2 min; m/z 172 (M , 2%), 157 (85),
113 (10), 85 (50), 70 (100), 69 (30).
HRMS: m/z 157.0501, found; 157.0508, calcd for
Syn th eses. Argpyrimidine. Argpyrimidine was pre-
pared mainly according to Shipanova et al. (13).
C7H9O4, M-15.
1
H NMR (CDCl ): δ 1.38 (s, 3H), 1.47 (s, 3H), 4.33
3
5
-Deoxy-D-ribose was prepared in a 5-step synthesis
as described below.
,3-O-Isopropylidene-5-O-toluol-p-sulfonyl-D-ribono-
,4-lactone 4. To a solution of 1.0 g (5.3 mmol) 2,3-O-
(d, 3H, J ) 6.9 Hz), 4.50 (d, 1H, J ) 5.4 Hz), 4.69 (q,
1H, J ) 6.9 Hz), 4.79 (d, 1H, J ) 5.5 Hz).
1
3
2
C NMR (CDCl3): δ 19.5, 25.5, 26.6, 74.6, 78.9, 80.3,
1
113.7, 173.7.
isopropylidene-D-ribono-1,4-lactone in 20 mL of anhy-
drous pyridine, 2.03 g (10.6 mmol) p-toluenesulfonyl
chloride was added at -15 °C, and the solution was
stirred for 10 h at -15 °C and then 10 h at ambient
temperature. After addition of 1 mL of ice water, the
reaction mixture was poured into 600 mL of ice water
under vigorous stirring. The resulting precipitate was
filtered, dried, and subjected to column chromatography
2,3-O-Isopropylidene-5-deoxy-D-ribose 7. To a solution
of 178 mg (1.04 mmol) of 6 in 12 mL of anhydrous THF,
4.2 mL (4.2 mmol) of a 1 M diisobutylaluminumhydride
solution in toluene was added slowly. After 30 min, the
reaction was quenched by 15 mL of a saturated aqueous
solution of Rochele’s salt, and the temperature was
allowed to rise until two layers became visible. The
reaction mixture was extracted with diethyl ether, the
organic layer was dried over calcium sulfate, and the
solvents were evaporated after filtration. The residue
was subjected to column chromatography (5:1 hexanes/
(
2
CH2Cl2). Fractions with material having a Rf 0.27 (TLC,
:1 hexanes/EtAc, FeCl3/hydroxylamine-reagent) were
combined, and the solvents were evaporated to yield
colorless crystals (1.1 g, 60%).
EtAc). Fractions with material having a R 0.39 (TLC,
f
HRGC-MS: tR 31.5 min; m/z 327 (M-15, 100%), 155
2:1 hexanes/EtAc, alkaline methanolic triphenyltetra-
zolium chloride solution) were combined, and the sol-
vents were evaporated to yield a colorless oil (125 mg,
(
50), 127 (30), 91 (90), 85 (80), 68 (45).
HRMS (m/z): 327.0538, found; 327.0533, calcd for
C14H15O7S, M-15.
1H NMR (CDCl3): δ 1.37 (s, 3H), 1.45 (s, 3H), 2.46 (s,
6
9%).
HRGC-MS: tR 5.8; m/z 159 (M-15, 30%), 157 (20),
13 (10), 99 (15), 85 (10), 71 (30), 59 (100).
1
3
1
H), 4.16 (dd, 1H, J ) 11.2 Hz, J ) 2.4 Hz), 4.33 (dd,
H, J ) 11.1 Hz, J ) 1.9 Hz), 4.68 (m, 1H), 4.73 (d, 1H,
HRMS: m/z 159.0657, found; 157.0660, calcd for
J ) 5.6 Hz), 4.77 (d, 1H, J ) 5.6 Hz), 7.37 (d, 2H, J )
C7H11O4, M-15.
1
8
.4 Hz), 7.74 (d, 2H, J ) 8.3 Hz).
H NMR (CDCl ): δ 1.32 (s, 3H), 1.35 (d, 3H), 1.47
3
1
3
C NMR (CDCl3): δ 21.6, 25.4, 26.5, 68.2, 74.9, 77.3,
9.0, 113.7, 127.8, 130.2, 131.5, 145.8, 173.0.
,3-O-Isopropylidene-5-iodo-D-ribono-1,4-lactone 5. To
(s, 3H), 2.80 (d, 1H, J ) 2.1 Hz), 4.36 (q, 1H, J ) 7.1
Hz), 4.55 (d, 1H, J 5) .9 Hz), 4.67 (d, 1H, J ) 5.9 Hz),
7
5
.43 (d, 1H, J ) 2.4 Hz).
2
13C NMR (CDCl3): δ 21.6, 24.8, 26.4, 83.2, 85.4, 86.3,
a solution of 1.10 g (3.31 mmol) 4 in 10 mL of anhydrous
acetone, 2.13 g (12.8 mmol) of potassium iodide was
added, and the mixture was refluxed for about 20 h.
Completeness of reaction was monitored by TLC. In-
soluble material was filtered off, the solvents were
evaporated, and the residue was subjected to column
chromatography (EtAc). Fractions with material having
a Rf 0.61 (TLC, 2:1 hexanes/EtAc, detection as 4) were
combined, the solvents were evaporated, and the result-
ing yellow crystals again subjected to column chroma-
tography (3:1 hexanes/EtAc). Fractions with 5 were
combined, and solvents were evaporated to yield color-
less crystals (0.89 g, 93%).
103.2.
The NMR data showed also small amounts of the
R-anomer.
5
-Deoxy-D-ribose. A solution of 25 mg (0.14 mmol)of 7
in 1 mL of 10% aqueous HAc was refluxed for about 2
h. Completeness of the reaction was monitored with
TLC (EtAc, Rf 0.55, detection as 7). Solvents were
evaporated, and the residue was dried under high
vacuum to yield a colorless oil (18 mg, 93%).
HRGC-MS: tR (after oximation with hydroxyl-
ammonium chloride and trimethylsilylation) 16.0 min
+•
(
m/z 437 (M , 2%), 320 (25), 219 (45), 191 (10), 147 (20),
HRGC-MS: tR 18.0 min; m/z 283 (M-15, 100%), 223
5), 141 (5), 127 (10), 85 (20), 69 (20), 59 (15).
1
31 (20), 129 (30), 117 (20), 100 (5), 73 (100). 16.2 min
(
(m/z 422 (M-15, 5%), 320 (20), 219 (40), 191 (5), 147 (20),
HRMS: m/z 282.9470, found; 282.9468, calcd for
131 (20), 129 (30), 117 (15), 101 (5), 73 (100). Peak ratio
C7H8IO4, M-15.
tR 16.0:16.2 ) 1:3.
HRMS: m/z 422.2034, found; 422.2030, calcd for
C16H40NO4Si4, M-15.
1H NMR (CDCl3): δ 1.40 (s, 3H), 1.47 (s, 3H), 3.38
(
1
1
dd, 1H, J ) 11.2 Hz, J ) 5.2 Hz), 3.45 (dd, 1H, J )
1.2 Hz, J ) 3.5 Hz), 4.61 (d, 1H, J ) 6 Hz), 4.63 (dd,
1H NMR (CDCl3): δ 1.15 (d, 3H, J ) 6.4 Hz, H5-R),
H, J ) 5.3 Hz, J ) 3.5 Hz), 4.98 (d, 1H, J ) 6 Hz).
1
.24 (d, 3H, J ) 5.9 Hz, H5-â), 3.72 (t, 1H, J ) 5.6 Hz,
1
3
C NMR (CDCl3): δ 5.6, 25.4, 26.4, 75.2, 80.2, 80.7,
14.0, 172.9.
H3-R), 3.90 (m, 3H, H2,3,4-â), 4.05 (m, 2H, H2,4-R), 5.10
(d, 1H, J ) 1.2 Hz, H1-â), 5.26 (d, 1H, J ) 4.2 Hz, H1-R).
1

368 J. Agric. Food Chem., Vol. 49, No. 1, 2001
Glomb et al.
13C NMR (CDCl3): δ 19.2 (C5-R), 20.4 (C5-â), 71.8
C3-R), 76.2 (C2-R), 76.5 (C3-â), 76.8 (C2-â), 79.4 (C4-R),
rimidine was added to 1 mL of beer prior to workup.
Values plotted are means of two independent analyses
for each sample.
(
7
9.6 (C4-â), 97.1 (C1-R), 102.2 (C1-â).
1
H- and 13C NMR identified the final material as a
In Model Reactions. After sterile filtration, a solution
R
1
:2.2 mixture of the R- and â-anomer.
of 50 mM N -t-BOC-arginine hydrochloride and 50 mM
sugar (methylglyoxal, 5-deoxy-D-ribose, D-ribose, D-
glucose, D-maltose) in 0.2 M phosphate buffer (pH 7.4)
was incubated at 55 °C. Aliquots of 50 µL were drawn
at 24, 48, 72, 96, and 168 h and 200 µL of 3 N HCl was
added. After 30 min, the solvents were evaporated
H igh -P er for m a n ce Liq u id Ch r om a t ogr a p h y
HP LC). Analytical Systems. A J asco (Groâ-Umstadt,
(
Germany) ternary gradient unit 980-PU-ND, with de-
gasser, autosampler 851-AS, column oven CO-200 set
at 25 °C, and fluorescence detector 920-FP was used.
The effluent was monitored at 320 nm for the excitation
and 398 nm for the emission. Chromatographic separa-
tions were performed in stainless steel columns filled
with RP-18 silica gel (VYDAC 218TP54, 250 × 4.6 mm,
(Speed Vac) and, after reconstitution with water, the
residue worked up as described for beer samples.
Analysis was performed only on the HPLC-VYDAC/
MeOH system. Values given were obtained by analysis
of duplicate incubations.
5
4
µm, Hesperia, CA; and Knauer Eurospher 100, 250 ×
.6 mm, 5µm, Berlin, Germany) using a flow rate of 1.0
Id en t ifica t ion of Ar gp yr im id in e in Beer b y
HRGC-MS. Decarbonated beer (1 L) was subjected to
ion exchange chromatography (70 g Dowex 50WX4-400,
-
1
mL‚min . The mobile phase used with the VYDAC
column was water (solvent A) and MeOH/water (7:3, v/v;
solvent B). To both solvents (A and B), 1.2 mL/L of
heptafluorobutyric acid was added. Samples were in-
jected at 20% B, the gradient changed to 35% B in 35
min, then changed to 70% B in 5 min, and was held at
+
H -form). After rinsing with 500 mL of H2O, elution was
started with 250 mL of 1N NH4OH solution. The first
50 mL of NH4OH solution was discarded, the rest was
combined and the solvents were evaporated. The residue
was subjected first to column chromatography (RP18,
40-63 µm, Merk, Darmstadt, Germany, 8:2 methanol/
H2O), then to preparative Lobar-HPLC, and finally to
preparative VYDAC-HPLC. Fractions having material
eluting at the same tR as Argpyrimidine (analytical
HPLC-VYDAC/MeOH system) were combined, solvents
were evaporated, and the residue was subjected to the
next chromatographic step. The final resulting residue
was dissolved in 500 µL of trifluoroacetic anhydride;
after 1 h the solvents were evaporated, and 500 µL of a
solution of diazomethane in diethyl ether (Aldrich,
Milwaukee, WI) was added. After 10 min the solvents
were removed under nitrogen, and the residue was
subjected to column chromatography (EtAc). Fractions
were monitored by HRGC-MS. For comparison, 1 mg
of authentic Argpyrimidine standard was derivatized
the same way using 100 µL of trifluoroacetic anhydride
and 100 µL of diazomethane solution, but no column
chromatography.
7
0% B for 10 min. The mobile phase used with the
Knauer column was water with 1.13 g NaH2PO4 x 2
H2O/L (adjusted to pH 6.5; solvent A) and propanol/
water (6:4, v/v) with 1.0 g NaH2PO4 x 2 H2O/L (adjusted
to pH 7.0; solvent B). To both solvents A and B, 3 g
sodium dodecyl sulfate/L was added. Samples were
injected at 2% B, the solvent ratio held at 2% B for 15
min, the gradient then changed to 100% B in 5 min,
and held at 100% B for 10 min.
Preparative Systems. A Waters Chromatography gra-
dient system (Waters, Milford, MA) with two pumps
(
model 6000A) and a solvent programmer (model 660)
-1
was used at a flow rate of 7 mL‚min . Chromatographic
separations were performed on glass columns (Lobar
Merk Lichroprep, 310 × 25 mm, RP18, 40-63 µm) and
stainless steel columns (VYDAC 218TP1022, 250 × 25
mm, RP18, 10 µm). The mobile phase used with both
systems was identical to the analytical VYDAC/MeOH
system. On the Lobar system samples were injected at
5
6
% B. This ratio was held for 30 min, then changed to
0% B in 40 min, and held for 50 min. Finally the
Id en tifica tion of 5-Deoxy-D-r ibose. In Incubations
of Maltose and Lysine. Maltose monohydrate (211 mg,
0.58 mmol) and 53 mg (0.29 mmol) of lysine hydro-
chloride in 2 mL of phosphate buffer (0.2 M, pH 8.5)
were heated at 80 °C for 3 h. The mixture was passed
through protonated Dowex 50WX4-400, washed with
20 mL of water, and the solvents were evaporated. The
resulting residue was subjected to column chroma-
tography (9:1, CH2Cl2/MeOH). Eluate from 90 to 180
mL was combined, solvents were evaporated, and the
residue was dissolved in 100 µL of anhydrous pyridine.
After addition of 2 mg of hydroxylamine hydrochloride,
the solution was stirred for 20 h at ambient tempera-
ture. N,O-bis(trimethylsilyl)acetamide (100 µL) was
added to the reaction mixture and, after 1 h, analysis
was performed on HRGC-MS.
gradient was changed to 100% B in 5 min, and held for
0 min. On the preparative VYDAC system samples
were injected at 10% B, the gradient then changed to
0% B, this ratio held for 30 min, and the gradient then
2
4
changed to 70% B, and held for 25 min. The gradient
was then changed to 100% B in 20 min and held for
3
0 min.
Qu a n tita tion of Ar gp yr im id in e. In Beer. Decar-
bonated beer (1 mL) was applied to ion exchange
chromatography (Dowex 50WX4-400, H -form, 0.5 mL
+
in a pasteur pipet). After rinsing with 5 mL of H2O,
elution was started with 1N NH4OH solution. The
second mL of NH4OH solution was collected, solvents
were removed in a Savant Speed Vac Plus SC 110A (Life
Science International, Frankfurt, Germany), and the
residue was reconstituted in 200 µL of H2O. The
reconstitution (100 µL) was injected on the HPLC-
VYDAC/MeOH system and the eluate was collected
between tR of 24 and 30 min. For determination of the
exact time window, an authentic standard was injected
first. Solvents were evaporated (Speed Vac), and the
residue was reconstituted in 200 µL of H2O, and 50 µL
was injected on the HPLC-Knauer/propanol system for
quantification. For recovery studies, 19.9 pmol Argpy-
In Incubations of Glycolaldehyde and Hydroxyacetone.
Glycolaldehyde dimere (60 mg, 0.50 mmol) and 74 mg
(
1.00 mmol) of hydroxyacetone in 10 mL of phosphate
buffer (0.2 M, pH 8.5) were incubated at 37 °C for 12 h.
Solvents were evaporated, and the resulting residue was
directly subjected to column chromatography and worked
up as for maltose reaction mixtures.
Mea su r em en t of Beer Color a n d Wor t. Methods
of the European Brewery Convention (EBC) were
used: EBC 9.6 and EBC 9.4.

Maillard Reactions in Beer
J. Agric. Food Chem., Vol. 49, No. 1, 2001 369
F igu r e 1. Mechanisms of Argpyrimidine formation from maltose.
RESULTS AND DISCUSSION
pyrimidine can be detected only after enzymatic hy-
drolysis, as the molecule is labile to acid- and base-
catalyzed protein hydrolysis. Beer contains significant
amounts of free amino acids, most importantly proline,
but also arginine, in concentrations up to about 500 and
Syn th esis of Ar gp yr im d in e fr om Meth ylglyoxa l.
δ
We first identified N -(5-hydroxy-4,6-dimethylpyrimi-
dine-2-yl)-L-ornithine or Argpyrimidine as the major
fluorescent compound in incubation mixtures of N -t-
BOC-arginine with methylglyoxal and various higher
sugars under physiological conditions. The structure
R
1
00 ppm, respectively. In addition, methylglyoxal has
been quantified in beer within the range of 0.08-0.2
ppm. We therefore developed a method to detect free
Argpyrimidine in this beverage. As the molecule was
expected to be present only as a minor constituent, a
very efficient workup was developed. Beer (1 mL) was
first passed through a protonated cation exchanger and
washed with water to remove carbohydrates and non-
basic material. Material containing amines was then
eluted with diluted ammonia in a very defined small
fraction. Removal of solvents resulted in a still volumi-
nous brown residue, which showed that substantial
parts of the beer matrix had not been removed. Injected
on high-performance liquid chromatography (HPLC),
this resulted in a huge background fluorescence present
in the upper trace chromatogram of Figure 2 A and
omitted direct measurement of Argpyrimidine. The first-
step HPLC was performed on a reversed-phase C-18
column with methanol/water eluent and heptafluoro-
butyric acid as the ion pair reagent. By addition of
authentic standard (lower trace, Figure 2A) a six
minutes retention time window (CUT) was defined to
collect material, which was reinjected on a second HPLC
system. The signal in the upper trace, coeluting at about
28 min with the standard, does not correlate with
Argpyrimidine. In addition, to follow the elution pattern
on the methanol/water HPLC system, the sensitivity of
the detector had to be lowered by one magnitude
compared to that of chromatogram B. The second HPLC
system also consisted of a C-18 column, but with a
propanol/water eluent and sodium dodecyl sulfate as the
ion pair reagent. Figure B (lower trace) showes a
baseline separated signal for Argpyrimidine, which
1
13
was unequivocally confirmed by H NMR, C NMR,
high-resolution FAB-MS, and independent synthesis
from 3-O-acetyl-2,4-pentanedione 1 (Figure 1). In the
reaction scheme shown, a reducing sugar like maltose
generates methylglyoxal through interaction with an
amine. For the reaction of two molecules of methyl-
glyoxal we proposed an intermolecular disproportion-
ation to form 3-hydroxy-2,4-pentanedione 2. This step
would entail the generation of formic acid, a common
Maillard reaction product. Also, this kind of chemistry,
basically representing the cleavage of an R-dicarbonyl
moiety, can be found as a known reaction pattern with-
in the context of the Maillard reaction, e.g. leading
to the formation of imidazolysine/MOLD (9, 10). The
intermediate reductone 2 can then condense with the
guanidino group of arginine to form Argpyrimidine.
The formation of 2 could also be explained under
participation of the guanidino nitrogens without chang-
ing the overall mechanism. As reductones are extremely
sensitive to oxidation and fragmentation reactions, we
synthesized 3-hydroxy-2,4-diethoxy-1,4-pentanediene 3
as a stable derivative to substantiate our hypothesis.
The vinyl ether-protected carbonyl functions can be set
free under mild acid conditions. Incubations with the
such-generated reductone 2 yielded eight times more
Argpyrimidine than when incubations were started from
methylglyoxal. Thus, the formation of Argpyrimidine
involves methylglyoxal with the generation of a 5-carbon
backbone reductone as the crucial intermediate.
Id en tifica tion a n d Qu a n tita tion of Ar gp yr im i-
d in e in Beer . In sugar-protein reaction mixtures Arg-

370 J. Agric. Food Chem., Vol. 49, No. 1, 2001
Glomb et al.
F igu r e 3. Identification of Argpyrimidine in beer by mass
spectrometry after derivatization: (A) beer workup, (B) au-
thentic standard.
HPLC with selective fluorescence detection proved to
be the superior method for quantitation. Taken together,
the results of the HPLC and HRGC-MS data unequivo-
cally establish the presence of Argpyrimidine in beer.
3
5 lager-type beer varieties were analyzed for Arg-
pyrimidine following the procedure described above.
According to their declaration, they all complied with
the German Purity Law, which means they were brewed
traditionally using only water, hops, barley malt, and
yeast, but no extra ingredients. Also, beer varieties are
to be manufactured and classified within certain ranges
of wort content (Schankbier 7-8%, Vollbier 11-14%,
Starkbier g16%). As depicted in Figure 4 A, Argpyri-
midine concentration correlated positively with wort
content. Obviously, there is considerable scattering
because of many variables in the individual brewery
recipes such as temperature control and malt selection.
The results suggest that higher concentrations of fer-
mentable carbohydrates led to more methylglyoxal
formation during mashing and wort boiling, and thus,
to more amino acid modifications. Darker beers (closed
circles) showed higher amounts of Argpyrimidine levels
than comparable lighter varieties within the same wort
range. Therefore, considerable modification must also
occur during malting, leading probably to both Argpy-
rimidine itself and methylglyoxal or some methylglyoxal
precursor like the Amadori product. This can also been
seen in Figure 4B, plotting Argpyrimidine concentration
against beer color measured as EBC units. On the other
hand, among darker beer varieties, Figure 4B elucidates
that darker color does not result in more Argpyrimidine.
One explanation might be the use of small quantities
of dark specialty malt to achieve the desired final tone
of an otherwise regularly brewed product. These malt
varieties are often manufactured under extreme con-
ditions that do not lead to amino acids released to the
final beverage. Finally, considerable amounts of methyl-
glyoxal should also be released by yeast metabolism
during fermentation, but, because of the relatively low
temperatures, impact on Argpyrimidine formation must
be considered to be small.
F igu r e 2. Detection of Argpyrimidine in beer by a 2-step
HPLC system. After workup by ion exchange chromatography,
samples were injected on HPLC system A, eluate containing
Argpyrimidine was collected (CUT) and reinjected on HPLC
system B for quantitation. Addition of standard is marked by
arrows.
clearly increased after standard addition. The peak in
this sample injection represents the absolute amount
of 450 fmol Argpyrimidine. By addition of standard, the
recovery rate of the total analysis procedure, starting
from 1 mL of beer, could be verified to be 67% with a
relative standard deviation of 5% (n ) 15).
To further verify the identity of the material detected
by HPLC, coupled gas chromatography-mass spectrom-
etry (HRGC-MS) was performed. As the sensitivity of
this method is much lower than that of fluorescence
detection, 1 L of beer had to be worked up. First,
material was subjected to ion-exchange chromatography
to remove carbohydrates, and then it was fractionated
by repeated chromatography on reversed phase columns
with consecutively smaller particle size. Finally, the
resulting residue was derivatized in two steps with
trifluoroacetic anhydride and diazomethane. The frag-
mentation patterns of mass spectra obtained with
electron impact ionization (EI) were virtually identical
to the one of the derivatized standard (Figure 3 A
and B). With chemical ionization, a pseudo molecular
ion of m/z 379 was obtained, i.e., Argpyrimidine was
Taken together, the detection of Argpyrimidine proves
that Maillard modifications of intact amino acids are
present in beer. Qualitative and quantitative studies are
now needed to identify the impact of the various stages
of brewing technology on the formation of such struc-
tures. The results could be used to understand and
correlate macroscopic changes on a molecular basis and,
thus, to define the final quality of beer.
R
derivatized to the corresponding N -trifluoroacetyl-
δ
N -(5-methoxy-4,6-dimethylpyrimidine-2-yl)-L-ornithine
methylester (data not shown). EI in Figure 3B led to a
base peak of m/z 347 representing the expected loss of
an O-CH3 fragment. As Argpyrimidine was only a
minor constituent even in the fractionated beer sample,
Alter n a tive Syn th esis P a th w a y for Ar gp yr im i-
d in e fr om Glycola ld eh yd e a n d 2-H yd r oxyp r o-
p a n a l. Glycolaldehyde and 2-hydroxypropanal are, like

Maillard Reactions in Beer
J. Agric. Food Chem., Vol. 49, No. 1, 2001 371
F igu r e 5. Identification of 5-deoxy-D-ribose in maltose reac-
tion mixtures by HRGC-MS after derivatization: (TIC) total
ion chromatogram, (A) mass spectra of sample workup, and
(B) authentic standard.
r
Ta ble 1. Ar gp yr im id in e in In cu ba tion s of N -t-BOC-
Ar gin in e w ith Va r iou s Su ga r sa
incubation 5-deoxy-
methyl-
time (d)
D-ribose D-ribose D-glucose D-maltose glyoxal
Argpyrimidine (µmol/mol arginine)
1
2
3
4
7
26
44
116
116
214
9
29
52
73
94
n.d.b
n.d.
n.d.
n.d.
17
n.d.
n.d.
n.d.
n.d.
12
33100
42800
47500
54200
49800
a
Both reactants 50 mM at 55 °C in 0.2 M phosphate buffer,
pH 7.4. b n.d., not detectable.
F igu r e 4. Correlation of Argpyrimidine concentration in beer
to wort content (A) and beer color expressed as EBC units (B).
Closed circles indicate samples with EBC units of more than
oximation with hydroxylamine and finally trimethyl-
silylation. Figure 5 shows the detection of 5-deoxy-D-
ribose in incubations of maltose and lysine at 80 °C. The
mass spectrum obtained at about 16.1 min of the total
ion chromatogram (TIC) is virtually identical to the one
from authentic 5-deoxyribose. The fragment at m/z 422
represents the characteristic loss of a methyl group from
the molecular ion. Although only trace amounts were
detectable in the TIC, these data unequivocally establish
the formation of 5-deoxypentoses as novel Maillard
intermediates of maltose.
2
8.
methylglyoxal, known Maillard reaction intermediates
of higher sugars (22 and 23). Both carbonyl structures
could undergo aldol condensation to form 5-deoxypen-
toses, which, after enolization and elimination of water,
would result in 3-hydroxy-2,4-pentanedione 2 (Figure
1
). As in the case of methylglyoxal, reaction of this
precursor with the guanidino group of arginine should
give Argpyrimidine. However, up to now there are no
reports on the formation of 5-deoxypentoses within the
context of the Maillard reaction. To establish these
structures as novel sugar degradation intermediates
and to test for the proposed alternative reaction path-
way leading to Argpyrimidine, we synthesized 5-deoxy-
D-ribose as an authentic reference. Two procedures have
been published using methyl-2,3-O-isopropylidene-â-D-
ribofuranoside (24) and 2,3-O-isopropylidene-D-ribono-
Argpyrimidine formation was compared in incuba-
R
tions of N -t-BOC-arginine with methylglyoxal, 5-deoxy-
D-ribose, and various other sugars at 55 °C (Table 1).
The small quantities derived from 5-deoxy-D-ribose
indeed verify the proposed alternative synthesis path-
way, but, contrary to expectations, prove it to be of
only minor importance. Obviously, methylglyoxal is by
far the better precursor of Argpyrimidine within the
reaction schemes shown in Figure 1.
1
,4-lactone (25) as the educts, respectively. As 2,3-O-
isopropylidene-D-ribono-1,4-lactone is commercially
readily available, we followed the latter strategy with
major modifications.
The formation of 5-deoxy-D-ribose was established in
reaction mixtures of glycolaldehyde and hydroxyacetone,
thus verifying the proposed aldol condensation reaction.
Because of the expected reactivity of 5-deoxypentoses,
only minor quantities were detected after column chro-
matography and derivatization for HRGC-MS by first
LITERATURE CITED
(
1) Ledl, F.; Schleicher, E. New aspects of the Maillard
reaction in food and in the human body. Angew. Chem.,
Int. Ed. Engl. 1990, 29, 565-594.
(
2) Hayashi, T.; Mase, S.; Namiki, M. Formation of a three-
carbon sugar fragment at an early stage of the browning
reaction of sugar with amines or amino acids. Agric.
Biol. Chem. 1986, 50, 1959-1964.

372 J. Agric. Food Chem., Vol. 49, No. 1, 2001
Glomb et al.
(
(
(
3) Velisek, J .; Davidek, T.; Davidek, J . 1,3-Disubstituted
imidazoles in the glucose-glycine Maillard reaction.
Lebensm. Wiss. Technol. 1992, 25, 74-76.
(16) Helak, B.; Spengler, K.; Tressl, R.; Rewicki, D. Forma-
tion of 7H-cyclopent[b]pyridin-7-ones as proline-specific
Maillard products. J . Agric. Food Chem. 1989, 37, 400-
404.
(17) Pabst, H. M. E.; Ledl, F.; Belitz, H.-D. Bitterstoffe
beim Erhitzen von Saccharose, Maltose und Prolin. Z.
Lebensm.-Unters. Forsch. 1985, 181, 386-390.
(18) Tressl, R.; Gr u¨ newald, K. G.; Helak, B. Formation of
flavor compounds from proline and hydroxyproline with
glucose and maltose and their importance to food flavor.
In Flavour 81; Schreier, P., Ed.; Walter de Gruyter:
Berlin/New York, 1981; pp 397-416.
(19) Tressl, R.; Gr u¨ newald, K. G.; Silwar, R.; Helak, B.
Formation of compounds with bread-like aroma char-
acter in malt and beer. In Proceedings of the 18th EBC
Congress; European Brewing Convention: Copenhagen,
1981; pp 391-403.
(20) Tressl, R.; Helak, B.; Rewicki, D. Maltoxazin, eine
tricyclische Verbindung aus Gerstenmalz. Helv. Chim.
Acta 1982, 65, 483-489.
(21) Tressl, R.; Helak, B.; K o¨ ppler, H.; Rewicki, D. Formation
of 2-(1-pyrrolidinyl)-2-cyclopentenones and cyclopent(b)-
azepin-8(1H)-ones as proline specific Maillard products.
J . Agric. Food Chem. 1985, 33, 1132-1137.
(22) Glomb, M. A.; Monnier, V. M. Mechanism of protein
modification by glyoxal and glycolaldehyde, reactive
intermediates of the Maillard reaction. J . Biol. Chem.
1995, 270, 10017-10026.
(23) Kim, M.-O.; Baltes, W. On the role of 2,3-dihydro-3,5-
dihydroxy-6-methyl-4(H)-pyran-4-one in the Maillard
reaction. J . Agric. Food Chem. 1996, 44, 282-289.
(24) Snyder, J . R.; Serianni, A. S. Synthesis and NMR-
spectral analysis of unenriched and 1-13C-enriched
5-deoxypentoses and 5-O-methylpentoses. Carbohydr.
Res. 1987, 163, 169-188.
(25) Escandar G. M.; Sala, L. F. A facile synthesis of 5-deoxy-
D-ribonic acid lactone and of 5-deoxy-D-ribose. Org. Prep.
Proc. Int. 1990, 22, 623-654.
4) O¨ rsi, F.; Homoki, P.; Phong, V. N. Bestimmung von
Methylglyoxal in Karamel-Proben. Die Nahrung 1995,
3
9, 90-97.
5) Hughues-Despointes, A.; Yaylayan, K. A. Retro-aldol
and redox reactions of Amadori compounds: mechanistic
1
3
studies with variously labeled D-( C)glucose. J . Agric.
Food Chem. 1996, 44, 672-681.
(
(
(
(
6) Moree-Testa, P.; Saint-J alm, Y. Determination of R-di-
carbonyl compounds in cigarette smoke. J . Chromatogr.
1
981, 217, 197-208.
7) Hayashi, T.; Shibamoto, T. Analysis of methylglyoxal
in foods and beverages. J . Agric. Food Chem. 1985, 33,
1
090-1093.
8) Revel, G.; Bertrand, A. A method for the detection of
carbonyl compounds in wine: glyoxal and methylgly-
oxal. J . Sci. Food Agric. 1993, 61, 267-272.
9) Ahmed, M. U.; Brinkmann-Frye, E.; Degenhardt, T. P.;
ꢀ
Thorpe, S. R.; Baynes, J . W. N -(Carboxyethyl)lysine, a
product of the chemical modification of protein by
methylglyoxal, increases with age in human lens pro-
teins. Biochem. J . 1997, 324, 565-570.
(
10) Nagaraj, R. H.; Shipanova, I. N.; Faust, F. M. Protein
cross-linking by the Maillard reaction. J . Biol. Chem.
1
996, 271, 19338-19345.
(
11) Henle, T.; Walter, A. W.; Haessner, R.; Klostermeyer,
H. Detection and identification of a protein-bound imi-
dazolone resulting from the reaction of arginine residues
and methylglyoxal. Z. Lebensm.-Unters. Forsch. 1994,
1
99, 55-58.
(
(
(
(
12) Lederer, M. O.; Klaiber, R. G. Cross-linking of proteins
by Maillard processes: characterization and detection
of lysine-arginine cross-links derived from glyoxal and
methylglyoxal. Bioorg. Med. Chem. 1999, 7, 2499-2507.
13) Shipanova, I. N.; Glomb, M. A.; Nagaraj, R. H. Protein
modification by methylglyoxal: chemical nature and
synthetic mechanism of a major fluorescent adduct.
Arch. Biochem. Biophys. 1997, 344, 29-36.
14) Lo, T. W. C.; Westwood, M. E.; McLellan, A. C.; Selwood,
T.; Thornalley, P. J . Binding and modification of proteins
by methylglyoxal under physiological conditions. J . Biol.
Chem. 1994, 269, 32299-32305.
15) Oya, T.; Hattori, N.; Mizuno, Y.; Miyata, S.; Maeda, S.;
Osawa, T.; Uchida, K. Methylglyoxal modifications of
protein. J . Biol. Chem. 1999, 274, 18492-18502.
Received for review April 19, 2000. Revised manuscript
received October 13, 2000. Accepted October 16, 2000. This
work was supported by a grant of the Deutsche Forschungs-
gemeinschaft, Bonn, Germany (to M. A. G.).
J F000493R