Characterization of precursors and elucidation of the reaction pathway leading to a novel coloured 2H,7H,8aH-pyrano[2,3-b]pyran-3-one from pentoses by quantitative studies and application of 13C-labelling experiments

Article abstract of DOI:10.1016/S0008-6215(98)00280-8

The intensely coloured (1R,8aR)- and (1S,8aR)-4-(2-furyl)-7-[(2-furyl)methylidene]-2-hydroxy-2H,7H,8aH-pyrano[2,3-b]pyran-3-one (1a/1b) have recently been identified among the main coloured compounds formed in the presence of pentoses. To clarify its formation pathway, quantitative studies on the effectivity of certain carbohydrate degradation products as precursors of 1a/1b were performed indicating the 3-deoxypentose-2-ulose, furan-2-carboxaldehyde and hydroxyacetaldehyde as the penultimate precursors of the colourant. In addition, a labelling experiment with [¹³C₁]-xylose was performed to elucidate how these precursors are transformed into 1a/1b. Copyright (C) 1998 Elsevier Science Ltd.

Full text of DOI:10.1016/S0008-6215(98)00280-8

Carbohydrate Research 313 (1998) 215–224
Characterization of precursors and elucidation of the
reaction pathway leading to a novel coloured
2
H,7H,8aH-pyrano[2,3-b]pyran-3-one from pentoses by
quantitative studies and application of C-labelling
experiments
13
Thomas Hofmann *
Deutsche Forschungsanstalt f u¨ r Lebensmittelchemie, Lichtenbergstr. 4, D-85748 Garching, Germany
Received 5 May 1998; accepted 14 October 1998
Abstract
The intensely coloured (1R,8aR)- and (1S,8aR)-4-(2-furyl)-7-[(2-furyl)methylidene]-2-hydroxy-2H,7H,8aH-
pyrano[2,3-b]pyran-3-one (1a/1b) have recently been identified among the main coloured compounds formed in the
presence of pentoses. To clarify its formation pathway, quantitative studies on the effectivity of certain carbohydrate
degradation products as precursors of 1a/1b were performed indicating the 3-deoxypentose-2-ulose, furan-2-carbox-
aldehyde and hydroxyacetaldehyde as the penultimate precursors of the colourant. In addition, a labelling experiment
1
3
with [ C ]-xylose was performed to elucidate how these precursors are transformed into 1a/1b. © 1998 Elsevier
1
Science Ltd. All rights reserved.
13
Keywords: Maillard reaction; Coloured compounds; Non-enzymatic browning; Pentoses; C-labelling experiment; Glyoxal;
Glycolaldehyde; Deoxyaldosuloses
1
. Introduction
7
- [(2 - furyl)methylidene] - 2 - hydroxy - 2H,7H,
aH-pyrano[2,3-b]pyran-3-one (1a and 1b in
8
Extensive studies have been performed in
Fig. 1) were identified among the main
coloured compounds formed in an aqueous
the last 30 years to obtain more detailed infor-
mation on the structures of the chromophores
involved in non-enzymatic browning reac-
tions; however, only a very limited number of
chromophoric structures could be character-
ized [1–7].
solution of xylose and
thermally treated in the presence the pentose-
degradation product furan-2-carboxaldehyde
L-alanine, which was
[
9,10]. The precursors and the formation route
leading to this type of chromophore are, how-
ever, as yet not clear. To disentangle the com-
plex network of non-enzymatic browning
reactions, it is a necessary prerequisite to clar-
ify the precursors, and to elucidate the reac-
tion pathway, as the carbohydrate is
By application of colour dilution analysis
[
8], the intensely orange coloured, previously
unknown (1R,8aR)- and (1S,8aR)-4-(2-furyl)-
*
Tel.: +49-89-28914181; fax: +49-89-28914183; e-mail:
hofmann@dfa.leb.chemie.tu-muenchen.de
0008-6215/98/$ - see front matter © 1998 Elsevier Science Ltd. All rights reserved.
PII: S0008-6215(98)00280-8

216
T. Hofmann / Carbohydrate Research 313 (1998) 215–224
Fig. 1. Structures of (1R,8aR)- (1a) and (1S, 8aR)-4-(2-furyl)-7-[(2-furyl)methylidene]-2-hydroxy-2H,7H,8aH-pyrano[2,3-b]pyran-
-one (1b).
3
converted into the 2H,7H,8aH-pyrano[2,3-
b]pyran-3-one chromophore in the course of
the Maillard reaction.
spectrum, therefore, definitely indicates an in-
corporation of the label into the carbon posi-
tion under investigation. This technique was
very recently successfully used to clarify the
formation pathway of red coloured 3(2H)-
pyrrolinone chromophores formed from the
Maillard reaction of furan-2-carboxaldehyde
To locate precursors of this chromophore, it
might be a promising approach to react cer-
tain Maillard intermediates, which are most
likely involved in its formation, with each
other and to determine the effectiveness of
these intermediates in generating 1a/1b by
means of quantitative measurements. This
technique was successfully used to identify the
precursors of several key-odorants ruling the
overall aroma of heated carbohydrate/amino
acid mixtures such as, e.g., 2-methyl-3-
furanthiol [10], 2-furfurylthiol [10] and 2-
acetyl-2-thiazoline [11].
and -alanine [12].
L
The objectives of the present investigation
were, therefore, to identify the precursors of
1a/1b by quantitative studies on certain Mail-
lard reaction intermediates and, to clarify the
reaction pathways governing the formation of
the colourants 1a/1b from pentoses by appli-
1
3
cation of a C-labelling experiment.
When the precursors are then identified, it is
a necessary further step to elucidate the mech-
anisms by which these precursors are con-
verted into 1a/1b. To meet this demand, the
most promising technique is the use of la-
belled precursors offering the possibility to
follow how the labelled atoms are shifted into
the compound under investigation. Because of
2. Results and discussion
It is obvious from the structure 1a/1b (Fig.
1) that the furan rings of (1R,8aR)- and
(1S,8aR)-4-(2-furyl)-7-[(2-furyl)methylidene]-
2 - hydroxy - 2H,7H,8aH - pyrano[2,3 - b]pyran-
3-one correspond to two molecules of furan-2-
carboxaldehyde and that, besides five carbon
atoms of the pentose skeleton, two further
carbon atoms have to be incorporated in the
colourant. This encouraged us to study,
whether these carbon atoms originate from
the C-2 compound hydroxyacetaldehyde or
glyoxal, well-known reaction products derived
from cleavage of the carbohydrate skeleton
[13,14].
2
1
H/ H exchange in aqueous Maillard systems,
deuterated precursors cannot be used and the
labelling experiments have to be performed
1
3
with C isotopes. To detect the labelled posi-
tions in 1a/1b, mass spectrometry might not
definitely assign the labelled carbon atoms.
13
C NMR spectrosocopy, however, allows an
unequivocal assignment of each labelled car-
bon site. Due to the low natural abundance
1
3
(
1.1%) of the C nucleus, the site specific
In the hope of shedding more light onto the
precursors of the colourant, the xylose/ -ala-
nine/furan-2-carboxaldehyde mixture was
thermally treated in the presence of either
hydroxyaldehyde or glyoxal, and the amounts
1
3
C-enrichment of a specific carbon site has
L
13
the advantage of increasing the C NMR
signal of the corresponding resonance up to
13
9
0-fold. A signal increase in the C NMR

T. Hofmann / Carbohydrate Research 313 (1998) 215–224
217
Table 1
the literature [13], this encouraged us to study
which of the two precursors is predominantly
liberated from the pentose. The formation of
glyoxal and hydroxyacetaldehyde in a heated
Influence of C-2 carbonyl compounds on the amounts of
1
a/1b generated from xylose/L-alanine in the presence of
a
furan-2-carboxaldehyde
C-2 compound
Amount of 1a/1b
mg)
aqueous solution of xylose and -alanine was,
L
therefore, followed by quantitative measure-
ments during a time period of 60 min. The
results, given in Fig. 2, showed that the
amounts of glyoxal rapidly increased as soon
as the mixture was heated. After running
through a maximum of about 58 mg/mmol
xylose at 3 min, the amount of glyoxal de-
creased again to about 10% after 10 min.
Prolonging the reaction time led only to a
slight decrease of the glyoxal amounts approx-
imating to a concentration of about 3 mg/
mmol xylose after 60 min. The concentrations
of hydroxyaldehyde increased only slowly dur-
ing the first 10 min and, after reaching a
maximum at 15 min, decreased again to about
5 mg/mmol xylose after 60 min. These data
revealed glyoxal as the primary reaction
product from the pentose and demonstrated
that the decrease of glyoxal after 3 min of
heating runs in parallel with the increase of
glycolaldehyde. It would, therefore, appear
that the former intermediate might be reduced
to the latter C-2 compound. This is well in line
with the recent finding that, after an induction
period of some minutes, the concentration of
reducing substances formed in aqueous carbo-
(
(%)
(
No additive)
0.5
3.1
5.6
0.01
0.06
0.11
Glyoxal
Hydroxyacetaldehyde
a
A mixture of xylose (16.5 mmol) and
L
-alanine (4 mmol)
was heated in phosphate buffer (20 mL; 1 mmol/L, pH 7.0)
under reflux for 10 min, then, the C-2 compound (0.75 mmol)
and furan-2-carboxaldehyde (25 mmol) was added and heat-
ing was continued for another 60 min.
of 1a/1b formed were compared with those
generated in a control experiment, in which
the mixture was heated without addition of a
C-2 compound. The results, given in Table 1,
showed that, in comparison with the control
experiment, 1a/1b were produced in the Mail-
lard mixtures containing hydroxyaldehyde or
glyoxal, by factors of 11 or 6 higher amounts,
respectively. The favoured production of 1a/
1
b in the presence of hydroxyacetaldehyde
documented this C-2 compound as a penulti-
mate precursor in the formation of the
chromophore.
Because a redox equilibrium between hy-
droxyacetaldehyde and glyoxal is discussed in
Fig. 2. Time course of the formation of glyoxal (- - -) and hydroxyacetaldehyde (—) in a heated neutral aqueous solution of xylose
and -alanine.
L

218
T. Hofmann / Carbohydrate Research 313 (1998) 215–224
Fig. 3. Time course of the formation of 1-deoxypentose-2-ulose (- - -) and 3-deoxypentose-2-ulose (—) in a heated neutral aqueous
solution of xylose and -alanine.
L
hydrate/
14]. The fact that an additional reduction step
is a necessary prerequisite for the formation of
a/1b from glyoxal is reflected by the higher
precursor effectivity of hydroxyacetaldehyde in
comparison with glyoxal.
Besides furan-2-carboxaldehyde and hydrox-
yacetaldehyde, the carbon skeleton of the pen-
tose is most likely involved in the formation of
L
-alanine solutions increased rapidly
The high concentrations of the 3-deoxypen-
tose-2-ulose in the heated xylose/ -alanine so-
lution prompted us to study, whether the
pentose skeleton is transformed into 1a/1b via
the 3-deoxyosone pathway. Aqueous solutions
[
L
1
of xylose/
Amadori rearrangement product N-(1-deoxy-
-xylulos-1-yl)- -alanine and 3-deoxypentose-
L
-alanine, the corresponding
D
L
2-ulose, respectively, were therefore heated in
the presence of furan-2-carboxaldehyde as
well as hydroxyaldehyde, and the amounts of
1a/1b generated were determined. The results,
given in Table 2, showed that, compared with
the Amadori rearrangement product or the
1
a/1b. Because 3-deoxypentose-2-ulose and 1-
deoxy-2,3-pentodiulose are known as major
pentose dehydration products with intact car-
bon chain, we followed their formation in the
aqueous xylose/ -alanine solution over a time
L
period of 120 min. The data, given in Fig. 3,
demonstrated that the formation of 3-de-
oxypentose-2-ulose and 1-deoxy-2,3-pentodiu-
lose increased rapidly in the first 10 min and,
after running through a maximum, decreased
again to about 50 and 30% after 120 min,
respectively. Independent from the reaction
time, the actual concentrations of 3-deoxypen-
tose-2-ulose were a factor of two to three
higher than those of the 1-deoxy-2,3-pentodiu-
lose. Comparing the time course of the de-
oxyosone formation (Fig. 3) with the time
course of the glyoxal formation (Fig. 2) showed
that the glyoxal is liberated from the pentose
skeleton at a very early stage of the Maillard
reaction, prior to deoxyosone formation, fitting
well with data found recently for hexoses [14].
xylose/
L
-alanine mixture, the formation of 1a/
Table 2
a
Influence of the C-5 skeleton on the generation of 1a/1b
Precursors
Amount of 1a/1b
(
mg)
(%)
Xylose (16.5 mmol)/
L
-alanine (4 mmol) 5.6
0.11
0.14
N-(1-Deoxy-
D
-xylulos-1-yl)-
L
-alanine
7.1
(
16.5 mmol)
3-Deoxypentose-2-ulose (16.5 mmol)
14.8
0.29
a
The precursor mixtures were heated in phosphate buffer
(
20 mL; 1 mmol/L, pH 7.0) under reflux for 10 min, then, a
mixture of glycolaldehyde (0.75 mmol) and furan-2-carbox-
aldehyde (25 mmol) was added and heating was continued for
another 60 min.

T. Hofmann / Carbohydrate Research 313 (1998) 215–224
219
nate from the anomeric carbon atom of the
pentose.
On the basis of the quantitative studies and
1
3
the results of the C-labelling experiments,
the reaction pathway, displayed in Fig. 5, was
proposed for the formation of 1a and 1b. To
1
3
follow the fate of the C label of the pentose
throughout its incorporation into the
1
3
colourant, the C-enriched atom is marked in
the structures outlined in Fig. 5. The reaction
between the pentose and the amino acid re-
sults in the Schiff base (I) as an intermediate
in the very beginning of the Maillard reaction.
In course of the Amadori rearrangement this
intermediate is converted into the N-(1-deoxy-
D-xylulos-1-yl)-amino acid (II), which, upon
dehydration at the 3-position and hydrolysis
of the amino acid, reveals the 3-deoxypentose-
2-ulose (IIa), found as a penultimate precursor
of 1a/1b. Quantitative studies revealed that
glyoxal can be formed at a very early stage of
the Maillard reaction, prior to deoxyosone
formation. Because Maillard-derived a-
aminoketones and a-iminoalcohols were
shown to be easily oxidized into the corre-
sponding 1,2-iminoketones [11,15] and glyoxal
formation was found to be strongly oxygen-
dependend [14], glyoxal (IIIa) might be
formed by an oxidative breakdown of the
Schiff base I. With increasing formation of
reducing substances in the course of the Mail-
lard reaction, hydroxyacetaldehyde (IIIb) is
then formed from IIIa upon reduction [14].
Condensation between the methylene-active
hydroxyacetaldehyde (III) and furan-2-car-
boxaldehyde (IIb), formed by cyclization and
dehydration of 3-deoxy-2-pentosulose (IIa),
yields the 2-(2-furyl)methylidene-2-hydroxyac-
etaldehyde (IV) showing strong nucleophilic
characteristics at the methylidene group. Con-
densation with the cyclic hemi acetal of the
3-deoxypentose-2-ulose then gives rise to a
a-ketoaldehyde intermediate (V). Due to the
strong electron-withdrawing effect of the alde-
hyde function, this intermediate subsequently
forms the 4-(2-furyl)-2,6-dihydroxy-2,5,6,7,8a-
1
3
Fig. 4. Excerpt of C NMR spectrum (900 MHz; Me SO-d )
2
6
13
of 1a/1b formed from (A) C -xylose and (B) xylose with
natural C abundance, respectively.
1
13
1
b was significantly favoured from the 3-de-
oxypentose-2-ulose, by a factor of 2.6 or 0.8,
respectively. These quantitative data indicated
the incorporation of the pentose skeleton into
the colourant via the 3-deoxypentose-2-ulose
as the key intermediate.
In the hope of shedding more light on how
the 3-deoxypentose-2-ulose is incorporated
into 1a/1b, a mixture of [ C]-labelled xylose
and -alanine was heated in the presence of
the precursors furan-2-carboxaldehyde and
hydroxyacetaldehyde in order to follow the
fate of the anomeric carbon atom of the pen-
tose during its incorporation into the
colourant. After its isolation, the site specific
13
L
13
C-enriched isotopomer of the colourant (e-
1
1
a/e-1b) was analysed by H broad band de-
1
3
coupled C NMR spectroscopy (Fig. 4).
1
3
Comparing the C NMR spectrum of e-1a/e-
b (A in Fig. 4) with the spectrum of non-la-
belled 1a/1b (B in Fig. 4), shows the signal at
1
1
3
7
1.4 ppm to be C-enriched. As reported by
Hofmann [8], heteronuclear correlation exper-
iments (HMQC, HMBC, HMQC-DEPT) led
to the assignment of the increased signal as
C-8 in structure 1a/1b. The position of the
C-enriched carbon atom at C-8 in e-1a/e-1b
clearly demonstrates that the aldehyde func-
tion in the hemiacetal group C-1 cannot origi-
pentahydropyrano[2,3-b]pyran-3-one
(VI)
upon ring closure between the hydroxyl group
of the hemiacetal and the aldehyde. Viny-
logous keto–enol tautomerism, followed by
water elimination leads to the 4-(2-furyl)-2,6-
13

220
T. Hofmann / Carbohydrate Research 313 (1998) 215–224
13
Fig. 5. Formation pathway leading to 1a/1b from pentoses and amino acids (“, C-labelled carbon atom).
dihydroxy - 2H,7H,8aH - pyrano[2,3 - b]pyran-
-one (VII), which, upon condensation with
3. Experimental
3
General methods.—Preparative thin-layer
chromatography (TLC) was performed on silica
gel (20×20 cm; 0.5 mm; E. Merck, Darmstadt,
Germany) using the following solvent systems
as the mobile phase: (A) MeCN–water, 9:1; (B)
EtOAc; (C) toluene–EtOAc, 1:1.
Flash chromatography (16×200 mm) was
performed using a RP-18 stationary phase (15.0
g; Lichroprep 25–40 mm, E. Merck, Darmstadt,
Germany) using MeOH–water (7:3) as the
mobile phase.
The high-performance liquid chromatogra-
phy (HPLC) apparatus (Kontron, Eching, Ger-
many) consisted of two pumps (type 422), a
gradient mixer (M 800), a Rheodyne injector
(100 mL loop) and a diode array detector
an additional molecule of furan-2-carboxalde-
hyde then gives rise to the (1R,8aR)- and
(
2
3
1S,8aR)-4-(2-furyl)-7-[(2-furyl)methylidene]-
- hydroxy - 2H,7H,8aH - pyrano[2,3 - b]pyran-
-one (1a/1b).
These data demonstrate that the identifica-
tion of precursors by quantitative studies, fol-
lowed by the elucidation of reaction pathways
1
3
using C-labelling experiments, offers the pos-
sibility to disentangle the complex network of
non-enzymatic browning reactions. Details,
obtained thereof, will help to construct a
route map of reactions leading to non-enzy-
matic browning during thermal processing of
foods.

T. Hofmann / Carbohydrate Research 313 (1998) 215–224
221
1
3
(
DAD type 440) monitoring the effluent in a
Methods.—
D
-Xylose, [ C-1]-xylose,
L
-ala-
wavelength range between 220 and 500 nm.
Separations were performed on a stainless
steel column packed with RP-18 (ODS-Hyper-
sil, 5 mm, 10 nm, Shandon, Frankfurt, Ger-
many) in an analytical scale (4.6×250 mm,
flow rate 0.8 mL/min). The following solvent
gradients were used: (A) starting with a mix-
ture (1:9) of MeCN and aqueous trifluoracetic
acid (0.1% TFA in water), the MeCN content
was raised to 60% after 50 min; (B) starting
with a mixture (1:9) of MeCN and aqueous
ammonium formiate buffer (pH 3.5; 20 mmol/
L), the MeCN content was increased to 30%
within 50 min.
nine, glyoxal (30% solution in water), hydrox-
yacetaldehyde (glycolaldehyde), 1-hydroxy-
cyclo-hexan-2-one, ethoxamine hydrochloride,
1,2-diaminobenzene, furan-2-carboxaldehyde,
trifluoracetic acid (TFA) were obtained from
Aldrich (Steinheim, Germany). Furan-2-car-
boxaldehyde was freshly distilled prior to use.
1,2-Diaminobenzene was recrystallized twice
from MeOH prior to use. Me SO-d was from
2
6
Isochom (Landshut, Germany). Solvents were
HPLC grade (Aldrich, Steinheim, Germany).
The following compounds were synthesized
as described in the literature: (1S,8aR)- and
(1R,8aR)-4-(2-furyl)-7-[(2-furyl)methylidene]-
2-hydroxy-2H,7H,8aH-pyrano[2,3-b] pyran-3-
one (1a/1b in Fig. 1) [8], 3-deoxypentos-2-
An analytical HPLC column (Nucleosil
00-5C18, Macherey and Nagel, D u¨ rren, Ger-
1
1
3
many) was coupled to an LCQ-MS (Finnigan
MAT GmbH, Bremen, Germany) using elec-
trospray ionization (ESI). After injection of
the sample (2.0 mL) analysis was performed
using a gradient starting with a mixture (9:1)
of MeCN and water and increasing the MeCN
content to 100% within 15 min.
ulose [10], [ C ]-butane-2,3-dione [16].
4
Syntheses.—N-(1-Deoxy-
alanine. A mixture of powdered anhydrous
-xylose (300 mmol) and -alanine (500
D
-xylulos-1-yl)-
L
-
D
L
mmol) was refluxed in anhydrous MeOH
(400 mL), whilst stirring. After heating for 3
h, malonic acid (90 mmol) was added and
the solution was refluxed for additional 2 h.
The mixture was cooled to room temperature
and concentrated to about 200 mL under
Gas chromatography–mass spectroscopy
GC–MS) was performed with a type 5160
(
gas chromatograph (Fisons Instruments,
Mainz, FRG) using an SE-54 capillary (30
m×0.32 mm, 0.25 mm; J&W Scientific, Fisons
Instruments, Mainz, FRG) coupled with an
MD-800 mass spectrometer (Fisons Instru-
ments, Mainz, FRG) running in chemical ion-
ization (CI) mode with sample application (0.5
mL) by on-column injection at 40 °C. After 2
min, the temperature was raised at 40 °C/min
to 50 min, held for 1 min isothermally, and
then raised at 6 °C/min to 230 °C and held for
vacuum. Unreacted -alanine was filtered off
L
and the filtrate was then cooled to −20 °C.
Dropwise addition of acetone yielded the
Amadori rearrangement product as a brown
hygroscopic solid. Concentration of the
mother liquor, recooling and addition of ace-
tone yielded the additional raw product. Both
crops were combined, dissolved in a mixture
of water (80 mL) and EtOH (200 mL) and
applied to cation-exchange chromatography.
Non-reacted carbohydrate was eluted with a
mixture of water (120 mL) and EtOH (280
mL), followed by water (100 mL). After elu-
tion with an aqueous ammonium hydroxide
solution (0.2 mol/L) and freeze-drying, the
1
0 min.
1
13
H broad-band decoupled C NMR spec-
troscopy was performed on a Bruker-AM-360
spectrometer using the following aquisition
parameters: transmitter frequency 90.56 MHz,
spectral width 25,000 Hz, pulse length 10.4 ms,
recorded with 64 K data points, repetition
time 2.5 s, 1 Hz line broadening, 20,000 scans;
processing was done by multiplication with a
Lorentz–Gaussian function prior to transfor-
N-(1-deoxy-
D
-xylulos-1-yl)- -alanine was ob-
L
tained as a white powder with a purity of
about 90% (24 mmol; yield: 8%). LC–MS:
+
222 (100, [M+1] ), 204 (25, [M+1−
+
1
H O] ); H NMR (360 MHz in D O; DQF-
2
2
3
mation. The sample was dissolved in Me SO-
COSY): l 1.50 (d, 3H, J=7.1 Hz, –CH ),
3.10 (d, 1H, J=10.0 Hz, –CH H –N), 3.14
(d, 1H, J=10.0 Hz, –CH H –N), 3.70–4.50
2
3
2
d6, chemical shifts expressed in ppm were
measured from residual Me SO-d (39.5 ppm).
a
b
2
2
6
a
b

222
T. Hofmann / Carbohydrate Research 313 (1998) 215–224
(
m, 4H, 2× –CH(OH)–, –CH –O), 3.80 (d,
revealed the target compound in a band at
f
2
3
1
H, J=7.1 Hz, N–CH(COOH)–CH ).
R =0.3, which was scraped off and suspended
3
2
-(2,3-Dihydroxypropyl) quinoxaline.
A
in MeOH. Filtration and concentration af-
forded the target compound as white crystals
(6.4 mmol; 33% in yield). LC–MS: 205 (100;
mixture of 3-deoxypentose-2-ulose (2 mmol)
and 1,2-diaminobenzene (2.2 mmol) in phos-
phate buffer (5 mL; 0.2 mol/L; pH 7.0) was
kept under argon in the dark at 40 °C for 3 h,
and was then extracted with CH Cl (5×10
mL). The combined organic layers were dried
over Na SO , concentrated under vacuum to
+
+
[M+1] ), 187 (18, [M+1−H O] ), 237
2
+
1
(19; [M+Na] ); H NMR (360 MHz;
CD OD): l 2.80 (s, 3H, –CH ), 3.86 (m, 1H,
2
2
3
3
–CH H –OH), 4.04 (m, 1H, –CH H –OH),
a
b
a
b
5.12 (m, 1H; ꢀC–CH(OH)–), 7.72 (m, 2H,
2
4
1
3
about 1 mL and the target compound was
then isolated by preparative TLC on silica gel
using solvent system A as the mobile phase.
2×ꢀCH–), 8.01 (m, 2H, 2×ꢀCH–);
C
NMR (360 MHz; CD OD): l 21.9 (–CH ),
65.8 (–CH –OH), 71.0 (–CH(OH)–), 128.3
3
3
2
The band with R =0.4–0.5 was scraped off
(ꢀCH–), 128.4 (ꢀCH–), 129.4 (ꢀCH–), 130.0
(ꢀCH–), 139.3 (=C(C)-), 141.7 (ꢀC(C)–),
151.9 (ꢀC(C)–), 152.9 (ꢀC(C)–).
f
and suspended in MeOH. After filtration and
concentration, the 2-(2,3-dihydroxypropyl)
quinoxaline (118 mg; yield: 58%) was obtained
as a colourless oil. LC–MS: 205 (100; [M+
Quantification of (1R,8aR)- and (1S,8aR)-
4 - (2 - furyl) - 7 - [(2 - furyl)methylidene] - 2-
hydroxy - 2H,7H,8aH - pyrano[2,3 - b]pyran-
3-one (1a/1b) in precursor mixtures.—After
cooling the reaction mixture, detailed in Table
1, the aqueous solution was extracted with
EtOAc (5×10 mL), the combined organic
layers were dried over Na SO and then dis-
+
+
1
] ), 187 (21, [M+1−H O] ), 237 (12;
2
+ 1
[
M+Na] ); H NMR (360 MHz; CD OD): l
3
2
3
3
.08 (dd, 1H, J=15.04, J=4.87, Hz, –
2
CH H –C(OH)–), 3.12 (dd, 1H, J=15.04
a
b
3
Hz, J=7.96 Hz, –CH H –C(OH)–), 3.62–
a
b
3
.74 (m, 2H, –CH –OH), 4.27 (m, 1H, –
2
2
4
CH(OH)–), 7.59 (m, 2H, 2× =CH–), 7.84
tilled under high vacuum (0.04 mbar) at 35 °C.
The residue was taken up in MeCN, mem-
brane filtered and then analysed by RP-18
HPLC using the solvent system A. Monitoring
the effluent at umax=460 nm, colourant 1a/1b
(
1
m, 1H, ꢀCH–), 7.92 (m, 1H, ꢀCH–), 8.69 (s,
1
3
H, NꢀCH–); C NMR (360 MHz; CD OD):
3
l 38.8 (–CH ), 65.9 (–CH –OH), 71.2 (–
2
2
CH(OH)–), 128.3 (ꢀCH–), 128.9 (ꢀCH–),
1
29.3 (ꢀCH–), 130.1 (ꢀCH–), 140.9 (ꢀC(C)–),
41.3 (ꢀC(C)–), 146.2 (ꢀC(C)–), 154.7
(R =21.6 min) was quantified by using the
t
1
pure reference compound as external stan-
dard. The results given in Tables 1 and 2 are
the mean of duplicates.
(
ꢀC(C)–).
2
-Methyl-3-(1,2-dihydroxyethyl)-quinoxal-
ine. A mixture of -xylose (19.6 mmol), -ala-
D
L
Quantification of hydroxyacetaldehyde after
deri6atization with ethoxamine.—Mixtures of
nine (18.3 mmol), 1,2-diaminobenzene (14.8
mmol) in phosphate buffer (80 mL; 0.5 mol/L,
pH 6.8) was refluxed for 12 h. After cooling to
room temperature, the reaction mixture was
extracted with CH Cl (5×20 mL), the com-
xylose (6 mmol) and -alanine (6 mmol) were
L
refluxed in phosphate buffer (4 mL). At the
reaction times, given in Fig. 2, the mixtures
were rapidly cooled to room temperature, the
internal standard 1-hydroxycyclohexan-2-one,
dissolved in water, and, ethoxamine hy-
drochloride (5 mmol) were added. The pH
was adjusted to 7.5 with aq sodium hydroxide
(0.1 mol/L) and the mixtures were maintained
for 3 h at 30 °C. The pH was then adjusted to
6.0 with hydrochloric acid (0.1 mol/L) and the
mixtures were extracted with diethyl ether
(3×15 mL). After drying over Na SO , the
2
2
bined organic layers were dried over Na SO ,
2
4
and, after concentration, separated by column
chromatography (35×400 mm) on silica gel
(
150 g, Silica Gel 60, E. Merck, Darmstadt,
Germany), which was conditioned with
EtOAc. After application of the crude mate-
rial onto the column, chromatography was
performed with EtOAc (400 mL), followed by
EtOAc–MeOH (1:1; 500 mL) affording the
target compound as a crude product. Further
fractionation by preparative TLC on silica gel
using solvent system B as the mobile phase
2
4
organic layer was concentrated to about 3 mL
and applied to GC–MS analyses scanning the
mass traces of the ethoxamine derivatives of

T. Hofmann / Carbohydrate Research 313 (1998) 215–224
223
hydroxyacetaldehyde (m/z 104) and 1-hy-
droxy-cyclohexan-2-one (m/z 158). For the de-
termination of the response factor (0.5),
solutions containing defined amounts of hy-
droxyacetaldehyde and 1-hydroxycyclohexan-
mL). After addition of 1,2-diaminobenzene (2
mmol), dissolved in MeOH, the mixture was
maintained at 30 °C for 3 h. The derivatized
reaction mixture was then diluted with water
and analyzed by RP-HPLC using the solvent
gradient B. Monitoring the effluent at u=320
nm, 2-methyl-3-(1,2-dihydroxyethyl)-quinoxa-
line, the derivative of 1-deoxy-2,3-pentodiu-
lose, and 2-(2,3-dihydroxypropyl)-quinoxaline,
the derivative of 3-deoxypentose-2-ulose were
detected at the retention times of 14.3 and
15.9 min, respectively. Quantification of these
derivatives was performed by using the syn-
thetic compounds as external standards.
2
-one in a ratio of 3:1 to 1:3 were treated as
described above for the Maillard mixtures.
GC–MS–CI of O-ethyl hydroxyacetaldehyde
oxime (R =6.09/6.50 min): 104 (100, [M+
t
+
1
] ); GC–MS–CI of O-ethyl 1-hydroxycyclo-
hexan-2-one oxime (R =15.25 min): 158 (100,
t
+
[
M+1] ).
Quantification of glyoxal after deri6atization
with 1,2-diaminobenzene.—Mixtures of xylose
6 mmol) and -alanine (6 mmol) were
1
3
(
L
C stable isotope-labelling experiment.—A
1
3
refluxed in phosphate buffer (4 mL). At the
reaction times, given in Fig. 2, a quarter of the
reaction mixture was withdrawn, rapidly
cooled to room temperature and dissolved
with water (5 mL). The solution was adjusted
to pH 6.5 with hydrochloric acid (0.1 mol/L),
the internal standard [13C-4]-butane-2,3-
dione, dissolved in diethyl ether, and 1,2-di-
aminobenzene (2 mmol), dissolved in MeOH,
was added. After maintaining the mixture for
solution of [ C ]-
D
-xylose (2.0 mmol) and -
L
1
alanine (0.48 mmol) in phosphate buffer (2.4
mL; 1 mmol/L, pH 7.0) was refluxed for 10
min in a closed vial. A mixture of hydroxyac-
etaldehyde (0.1 mmol) and furan-2-carbox-
aldehyde (3.0 mmol) in water (0.4 mL) was
added and heating was continued for another
50 min.
1
3
Isolation of C-enriched (1R,8aR)- and
(1S,8aR)-4-(2-furyl)-7-[(2-furyl)methylidene]-
2 - hydroxy - 2H,7H,8aH - pyrano[2,3 - b]pyran-
3-one (e-1a/e-1b).—The reacted mixture was
cooled to room temperature, the pH was ad-
justed to 5.0 and the aqueous solution was
extracted with diethyl ether (5×4 mL), the
organic layer was dried over Na SO and the
3
h at 30 °C, the pH was adjusted to 5.0 with
hydrochloric acid (0.1 mol/L) and the mix-
tures were extracted with diethyl ether (3×15
mL). After drying over Na SO , the organic
2
4
layer was concentrated to about 3 mL and
was applied to GC–MS analyses scanning the
mass trace of m/z 131 for the derivative of the
glyoxal and m/z 163 for the derivative of the
internal standard. For the determination of
the response factor (0.95), solutions contain-
ing definite amounts of glyoxal and [13C-4]-
butane-2,3-dione in a ratio of 1:3 to 3:1 were
treated as described above for the Maillard
2
4
orange colourant was isolated by preparative
thin layer chromatography using solvent sys-
tem C. An orange coloured band at R =0.4
f
was scraped off and dissolved in MeOH (20
mL). After filtration and concentration, the
coloured compound was further purified by
RP-18 flash chromatography. The coloured
fraction was collected and freed from MeOH
under vacuum. The aqueous phase was ex-
tracted with EtOAc (3×20 mL) and, after
drying over Na SO , the solvent was removed
mixtures. GC–MS–CI of quinoxaline (R =
t
+
1
5.38 min): 131 (100, [M+1] ); GC–MS–CI
of [ C ]-2,3-dimethylchinoxaline (R =19.70
1
3
4
t
+
min): 163 (100, [M+1] ).
2
4
under vacuum yielding a red residue, which
was dissolved in Me SO-d and then analyzed
2
6
1
13
by H broad band decoupled C NMR spec-
troscopy and LC–MS. The comparison of the
LC–MS data with those of natural C abun-
1
3
dant 1a/1b [8] demonstrates the incorporation
1
3
of one C atom in 1a/1b. LC–MS: 296 (100,
+
+
[M+1−H O] ), 314 (29, [M+1] ).
2

224
T. Hofmann / Carbohydrate Research 313 (1998) 215–224
[
[
3] F. Ledl, Th. Severin, Z. Lebensm. Unters. Forsch., 175
1982) 262–265.
4] A. Arnoldi, E.A. Corain, L. Scaglioni, J.M. Ames, J.
Agric. Food Chem., 45 (1997) 650–655.
Acknowledgements
(
I wish to thank the Institute of Organic
Chemistry, Technical University of Munich,
for the NMR measurements and the Institute
of Food Chemistry, Technical University of
Munich, for the LC–MS analyses. I am grate-
ful to the Deutsche Forschungsgemeinschaft
[5] T. Hofmann, J. Agric. Food Chem., 46 (1998) 932–940.
[6] T. Hofmann, Hel6. Chim. Acta., 80 (1997) 1843–1856.
[7] T. Hofmann, Z. Lebensm. Unters. Forsch., 206 (1998)
251–258.
[
[
8] T. Hofmann, Carbohydr. Res., 313 (1998) 203–213.
9] F. Ledl, E. Schleicher, Angew. Chem., 102 (1990) 597–
7
34.
[10] T. Hofmann, P. Schieberle, J. Agric. Food Chem., 46
1998) 235–241.
11] T. Hofmann, P. Schieberle, J. Agric. Food Chem., 43
1995) 2946–2950.
(
DFG) for partly funding this work and to
(
Mr. J. Stein for his excellent technical assis-
tance.
[
(
[
[
12] T. Hofmann, J. Agric. Food Chem., 46 (1998) 914–945.
13] T. Hayashi, M. Namiki, Agric. Biol. Chem., 44 (1980)
2575–2580.
References
[
14] T. Hofmann, K. Stettmeier, W. Bors, J. Agric. Food
Chem., 47 (1999) in press.
[
[
1] Th. Severin, U. Kr o¨ nig, Chem. Mikrobiol. Technol.
Lebensm., 1 (1972) 156–157.
2] F. Ledl, Th. Severin, Z. Lebensm. Unters. Forsch., 167
[15] T. Hofmann, P. Schieberle, J. Agric. Food Chem., 46
(1998) 616–619.
[16] T. Hofmann, P. Schieberle, J. Agric. Food Chem., 45
(1997) 227–232.
(
1978) 410–413.
.