Heterocyclic volatiles formed by heating cysteine or hydrogen sulfide with 4-hydroxy-5-methyl-3(2H)-furanone at pH 6.5

Article abstract of DOI:10.1021/jf0008644

The reaction of 4-hydroxy-5-methyl-3(2H)-furanone (HMF) with cysteine or hydrogen sulfide at pH 6.5 for 60 min at 140°C produced complex mixtures of volatile compounds, the majority of these containing either sulfur or nitrogen. Of the 68 compounds detect

Full text of DOI:10.1021/jf0008644

816
J. Agric. Food Chem. 2001, 49, 816−822
Heter ocyclic Vola tiles F or m ed by Hea tin g Cystein e or Hyd r ogen
Su lfid e w ith 4-Hyd r oxy-5-m eth yl-3(2H)-fu r a n on e a t p H 6.5
Frank B. Whitfield† and Donald S. Mottram*,‡
Food Science Australia, P.O. Box 52, North Ryde, NSW 1670, Australia, and School of Food Biosciences,
The University of Reading, Whiteknights, Reading RG6 6AP, United Kingdom
The reaction of 4-hydroxy-5-methyl-3(2H)-furanone (HMF) with cysteine or hydrogen sulfide at pH
6.5 for 60 min at 140 °C produced complex mixtures of volatile compounds, the majority of these
containing either sulfur or nitrogen. Of the 68 compounds detected, 63 were identified, some
tentatively, by GC-MS. Among the identified compounds were thiophenes (10), thiophenones (6),
thienothiophenes (5), thiazoles (5), trithiolanes (4), pyrazines (6), and oxazoles (4). More compounds
were produced in the reaction of HMF with cysteine (63) than were formed in the reaction with
hydrogen sulfide (33). In both systems, thiophenones were major reaction products, accounting for
25-36% of the total volatiles formed. Possible reasons for the differences in the composition of the
two systems are discussed. The contributions of these reactions, and their products, to the flavor of
heated foods are considered.
Keyw or d s: Maillard reaction; aroma; volatiles; GC-MS analyses; amino acids; pentoses
INTRODUCTION
Phosphate buffer (0.5 M, pH 6.5) was prepared in distilled
water. A saturated solution of hydrogen sulfide (∼0.2 M) was
prepared by passing hydrogen sulfide through phosphate
buffer (pH 6.5) at 0 °C as described previously (8). Authentic
samples of reference compounds were either purchased from
a range of laboratory chemical suppliers or obtained as gifts
from flavor laboratories.
Rea ction Mixtu r es. Mixtures containing 11.4 mg of HMF
and 12.1 mg of cysteine in phosphate buffer (2 mL) were
prepared in 5 mL glass ampules. Other mixtures were
prepared by mixing 1 mL of a solution containing 12.1 mg of
cysteine, in buffer, and 1 mL of the saturated solution of
hydrogen sulfide (∼6.6 mg). The ampules were flame sealed
and then heated in an oven, controlled at 140 °C, for 60 min.
Isola tion of Vola tile Rea ction P r od u cts. After cooling,
the reaction mixtures were diluted with 20 mL of 0.5 M
phosphate buffer (pH 6.5), and volatiles were collected on glass-
lined stainless steel traps containing Tenax GC (Scientific
Glass Engineering Pty. Ltd., Melbourne, Australia) as descibed
previously (8). Prior to volatile collection, methyl decanoate
(100 µg in 0.1 mL of ethanol) was added to the reaction mixture
as an internal standard. During the collection of the volatile
components, the mixture was stirred slowly and maintained
at 60 °C in a water bath.
Ga s Ch r om a togr a p h y)Ma ss Sp ectr om etr y (GC-MS).
A Varian 1440 gas chromatograph fitted with a “Unijector”
(Scientific Glass Engineering Pty. Ltd.), and a fused-silica
capillary column (50 m × 0.32 mm i.d.) coated with 5% phenyl
methylsilicone, BP5 (Scientific Glass Engineering Pty Ltd.),
was used for all analyses. The GC was coupled directly to a
Varian-MAT 311A double-focusing mass spectrometer con-
trolled by a Finnigan-MAT Incos 2200 data system. The
trapped volatile components were thermally desorbed onto the
GC column by heating the trap at 260 °C for 5 min and were
cyrofocused at the front of the GC column. Full details of the
GC-MS conditions have been published previously (8). C6-C20
n-alkanes (100 ng of each in 1 µL of diethyl ether added to a
Tenax trap) were used as external standards for the calculation
of linear retention indices of the volatile reaction products.
Compounds were tentatively identified by comparing their
mass spectra with those contained in the NIST/EPA/NIH and
Wiley mass spectral databases, in collections of mass spectra
of flavor compounds (10), and in previously published litera-
4-Hydroxy-5-methyl-3(2H)-furanone (HMF) is formed
in Maillard reactions involving pentose sugars such as
ribose (1). Both HMF and its 2,5-dimethyl homologue
have been isolated from beef broth (2) and are consid-
ered to be likely precursors of meaty flavors through
their reaction with either hydrogen sulfide or sulfur-
containing amino acids (3, 4). In studies of the reaction
of cysteine and ribose at pH 5.6 it was shown that a
wide range of compounds were formed containing one
or more sulfur atoms (5, 6). As a consequence, it was
suggested that HMF was a possible intermediate in the
formation of some of these compounds (7). To further
the understanding of the role of HMF in such reactions,
this compound was reacted with either cysteine or
hydrogen sulfide at pH 4.5 (8) and was shown to yield
many of the sulfur-containing compounds previously
identified in reaction mixtures involving ribose. These
compounds included disulfides (26), thiols (7), dithiol-
anones (6), thiophenones (4), dithianones (3), and
thienothiophenes (6). Of these compounds the disulfides
and their thiol precursors were particularly interesting
because of the association of this type of compound with
meatlike aromas (9). However, it is known that pH can
influence the products formed in such reactions (4). As
a consequence, we have repeated the reactions of HMF
with cysteine and hydrogen sulfide at pH 6.5. Although
a similar number of compounds were identified in these
reaction mixtures as were found at pH 4.5, the number
of sulfur-containing compounds found at pH 6.5 (40) was
far fewer than found at pH 4.5 (64).
MATERIALS AND METHODS
Ma ter ia ls. Details of reagents used are given in a previous
paper (8). HMF was obtained as a gift from a flavor company.
* Author to whom correspondence should be addressed (fax
+ 44 118 931 0080; e-mail D.S.Mottram@reading.ac.uk).
† Food Science Australia, a joint venture of CSIRO and Afisc.
‡ The University of Reading.
10.1021/jf0008644 CCC: $20.00 © 2001 American Chemical Society
Published on Web 01/30/2001

Reaction of Cysteine or H S with 4-Hydroxy-5-methyl-3(2H)-furanone
J. Agric. Food Chem., Vol. 49, No. 2, 2001 817
2
ture. When possible, confirmations of identifications were
carried out by comparing linear retention indices (LRI) with
those of authentic compounds. Masslib Software (MSP Friedli
and Co., Ko¨niz, Germany) was used to establish possible
structures for some compounds with no reference mass spectra
in the literature.
To obtain estimates of the relative quantities of the identi-
fied components, the approximate concentrations of selected
compounds were determined by comparing their GC-MS
chromatogram peak areas with the area of the internal
standard, methyl decanoate, which was taken as 100 µg, and
assuming all response factors were 1. The concentrations of
these compounds are reported as micrograms per 10 mg of
HMF used in the reaction. The detection limit for individual
compounds present in the reaction mixture was estimated to
be 0.1 µg/10 mg of HMF, based on 3 times background noise.
Compounds described as “trace” components were present in
concentrations between 0.1 and 1 µg/10 mg of HMF.
involved in their formation were promoted by acidic
conditions. In addition, dithiolanones and dithianones,
major compounds formed in both systems at pH 4.5,
either were not formed or were formed in greatly
reduced quantities at the higher pH (Table 2). At pH
6.5, the main sulfur compounds in both systems were
thiophenones, whereas thiophenes were also major
products of the reaction of HMF with hydrogen sulfide.
Alkyl-substituted trithiolanes and trithianes were also
found at pH 6.5 but were absent from the reaction
performed at pH 4.5 (Table 2).
Expressed as a percentage of total volatiles found, the
major compounds obtained from the reaction involving
cysteine at pH 6.5 were 4,5-dihydro-2-methylthiophen-
3(2H)-one (28), 11%; (E and Z) 4,5-dihydro-2,5-dimeth-
ylthiophen-3(2H)-one (33 and 34), 10%; unknown N
compound A (40), 9%; unknown N compound C (58),
7-8%; 2,5-dimethylthiophene (10), 2-6%; 2,4,5-tri-
methyloxazole (9), 4-5%; 2,5(or 2,6)-dimethylpyrazine
(18), 3-4%; and 4,5-dimethylthiazole (23), 3-4%. By
comparison, the major compounds formed in the reac-
tion between HMF and hydrogen sulfide at pH 6.5 were
2-acetylthiophene (41), 15-18%; 4,5-dihydro-2-meth-
ylthiophen-3(2H)-one (28), 13-18%; (E and Z) 4,5-
dihydro-2,5-dimethylthiophen-3(2H)-one (33 and 34),
6-8%; 3-ethyl-2-formylthiophene (57), 5-6%; 2,5-di-
methylthiophene (10), 1-7%; 4,5-dihydro-2-ethylthio-
phen-3(2H)-one (39), 5%; and 4,5-dihydro-2-methyl-
3(2H)-furanone (6), 2-4%. Of these 13 compounds only
the dihydrothiophenones 28, 33, and 34 and the
thiophene 10 were major components in both reaction
systems. Although individually present in minor con-
centrations, the thienothiophenes (63-67) and trithio-
lanes (48, 49, 59, and 60) represented 4-6% and 4-7%
of the total volatiles found in the reactions involving
cysteine and 1-4% and 2-3% of those found in the
reactions with hydrogen sulfide, respectively. Accord-
ingly, as classes the thienothiophenes and trithiolanes
make significant contributions to the volatile content
of these reaction systems. However, major differences
exist between the quantitative and qualitative composi-
tions of the volatile components generated in the two
systems. As a consequence, the remainder of this paper
will concentrate on providing reasons for these differ-
ences.
RESULTS AND DISCUSSION
A total of 68 compounds, including 5 not identified,
were found in the headspace volatiles of reaction
mixtures containing HMF and cysteine or hydrogen
sulfide (Table 1). These included thiophenes (10), thiophe-
nones (6), thienothiophenes (5), thiazoles (5), trithio-
lanes (4), pyrazines (6), and oxazoles (4). Whenever
possible, identities were confirmed by comparison of the
mass spectra and LRI with those of authentic com-
pounds. For components for which no reference com-
pounds were available, tentative identities were deter-
mined either by comparison with published data or by
interpretation of their mass spectrum and comparison
with related compounds. Except for the oxazoles, most
of the other classes of these compounds had been
previously identified in cysteine-ribose reaction mix-
tures (5, 11-15). This was to be expected as HMF is a
major product of the degradation of pentoses in Maillard
and caramelization reactions (1).
The aromas of the reaction mixtures were evaluated
by three assessors who worked in the laboratory. The
aroma of the reaction between HMF and cysteine was
described as caramel, roasted, and fried bacon rind,
whereas that between HMF and hydrogen sulfide had
an aroma described as caramel, metallic, and medicinal.
Neither system had the meatlike aromas previously
observed when these mixtures were reacted at pH 4.5
(8).
Th iop h en on es. Reaction between HMF and cysteine
gave six thiophenones (27, 28, 33, 34, 39, and 43) that
accounted for 25-27% of the total volatiles formed,
whereas in the reaction between HMF and hydrogen
sulfide they accounted for 29-36% of the volatiles
recovered. All six compounds were formed in both
systems. As such, they are major products of these
reactions. However, with the exception of the stereo-
isomers 33 and 34 the other thiophenones were all
formed in greater quantities in the reaction between
HMF and hydrogen sulfide. Previously it had been
proposed (8) that in comparable reactions performed at
pH 4.5 compounds 27, 28, 33, and 34 were formed from
acetaldehyde (derived from cysteine) and pyruvaldehyde
or 2,3-butanedione (both from HMF). To account for the
current results, we now suggest alternative possible
reaction sequences involving hydroxyacetaldehyde (in
place of acetaldehyde), the above dicarbonyl compounds,
and 2,3-pentanedione (all four from HMF, see ref 8) for
the formation of 27, 28, 33, 34, 39, and 43. A similar
reaction sequence could account for the formation of 4,5-
dihydro-2-ethylthiophen-3(2H)-one (39) with formalde-
Of the 63 identified and tentatively identified com-
pounds reported in Table 1, 58 were found in the
volatiles formed in the reaction between HMF and
cysteine, and 32 were found in the volatiles formed in
the reaction between HMF and hydrogen sulfide. Both
reaction mixtures gave similar quantities of total vola-
tiles; those formed in the reaction with cysteine were
369 and 354 µg/10 mg of HMF for the duplicated
reactions, whereas those formed in the reaction with
hydrogen sulfide were 367 and 417 µg/10 mg of HMF.
There were major differences between the products
of these reaction systems and those found previously
in similar reactions carried out at pH 4.5 (8). The
dominant compounds produced at pH 4.5, both quali-
tatively and quantitatively, were sulfur compounds,
whereas at pH 6.5 far fewer sulfur compounds were
produced in both systems. In the reaction of HMF with
cysteine, nitrogen compounds accounted for 41% of the
total volatiles produced (Table 2). The major classes of
sulfur compounds not formed at pH 6.5 were mercap-
toketones, furanthiols, thiophenethiols, and their re-
lated disulfides, which suggests that the reactions

818 J. Agric. Food Chem., Vol. 49, No. 2, 2001
Whitfield and Mottram
Ta ble 1. Vola tile Com p ou n d s Obta in ed fr om Rea ction s betw een 4-Hyd r oxy-5-m eth yl-3(2H)-fu r a n on e a n d Cystein e or
Hyd r ogen Su lfid e
approximate concna
mass spectral data, m/z
no.
compound
cysteine
H2S
method of IDb LRIc
(rel intensity or reference)d
1
2
3
4
5
6
7
8
9
2,3-pentanedione
3-penten-2-one
4,5-dimethyloxazole
2-hexanone
7
tr
1
1
3
6
2
3
20
6
nd
6
tr
tr
1
2
4
16
1
2
1
2
13
1
3
1
6
39
7
8
11 16
nd
nd
nd
29
nd
8
nd
nd
nd
4
MS + LRI
MS + LRI
MS + LRI
MS + LRI
MS + LRI
MS
680
755
771
792
800
806 (10)
816 (20)
825
846
867
875
876
880
883
887
902 (20)
905
910
915
919
919
925
933
944
952 (10)
959
982 (12)
990
997
997
1002
1008
1016 (12)
1027 (12)
1040
1047 (21)
1078
1071 (22)
1082 (10)
nd
2
1
8
nd
4
3,4-hexanedione
3
6
4,5-dihydro-2-methyl-3(2H)-furanone
3-mercaptobutan-2-one
methylpyrazine
8
5
15
1
MS + LRI
MS + LRI
MS + LRI
MS + LRI
MS + LRI
MS + LRI
MS + LRI
MS + LRI
MS + LRI
MS + LRI
MS + LRI
MS + LRI
MS + LRI
MS + LRI
MS
2
nd
2,4,5-trimethyloxazole
14 nd
22 24
10 2,5-dimethylthiophene
11 2,4-dimethylthiophene
12 2-methyl-2-thiazoline
13 2,4-hexanedione
14 2,4-dimethylthiazole
15 3,4-dimethylthiophene
16 3-mercaptopentan-2-one
17 2-methyl-2-cyclopenten-1-one
18 2,5(or 2,6)-dimethylpyrazine
19 2,5-dimethylthiazole
nd
4
nd
9
nd
9
11
nd
7
nd nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
17
20
75
nd
nd
nd
nd
29
6
4
4
3
1
nd
6
11 nd
nd nd
nd nd
nd nd
20 2,3-dimethylpyrazine
21 4-ethyl-2,5-dimethyloxazole
22 2-ethyl-4,5-dimethyloxazole
23 4,5-dimethylthiazole
1
nd
MS
11 nd
MS + LRI
MS + LRI
MS
MS + LRI
MS
MS + LRI
MS + LRI
MS + LRI
MS + LRI
MS + LRI
MS
MS
MS + LRI
MS
MS + LRI
MS
MS
24 2-ethyl-1H-pyrrole
2
4
4
6
nd
nd
9
25 1-(2-furyl)-2-propanone
26 2-ethyl-5-methylthiophene
27 4,5-dihydro-5-methylthiophen-3(2H)-one
28 4,5-dihydro-2-methylthiophen-3(2H)-one
29 2,4,5-trimethylthiazole
30 2-ethyl-(5 or 6)-methylpyrazine
31 2,3,5-trimethylpyrazine
32 1-(2-furyl)-1-propanone
15
40 48
7
7
7
3
nd
nd
nd
4
10
3
33 4,5-dihydro-2,5-dimethylthiophen-3(2H)-one (E or Z)e 31
28 17
34 4,5-dihydro-2,5-dimethylthiophen-3(2H)-one (E or Z)e
35 2,3-dimethyl-2-cyclopenten-1-one
36 1-(5-methyl-2-furyl)-2-propanone
37 2-ethyl-3,6-dimethylpyrazine
5
6
2
3
nd
9
8
4
4
7
10
7
nd
nd
19
nd
2
6
2
nd
8
nd
6
17
38 3-methyl-1,2-dithiolan-4-one
39 4,5-dihydro-2-ethylthiophen-3(2H)-one
40 nitrogen compound A
34
31 nd
ms
1089 137 (100), 43 (82), 94 (62),
95 (58), 39 (57), 67 (20),
109 (18)
41 2-acetylthiophene
42 3,5-dimethyl-1,2-dithiolan-4-one (E or Z)
43 2-ethyl-5-methyl-4,5-dihydrothiophen-3(2H)-one
nd
1
4
nd 55
76
25
nd
MS + LRI
MS
ms
1092
1098 (12)
1104 74 (100), 41 (38), 144 (37),
nd
4
7
7
45 (11), 39 (9), 42 (8), 59 (6)
44 2-thiazolyl-1-propanone
45 2-formyl-5-methylthiophene
46 3,4,5-trimethyl-2-furfural
2
3
2
1
7
7
5
1
3
2
4
2
nd
nd
nd
nd
nd
nd
nd
nd
nd
15
nd
nd
MS
MS + LRI
MS
1119
1124
1130
1134 (12)
1138
1144
1151 (12)
1157
47 1-(3-thienyl)-2-propanone
nd nd
nd
14 nd
MS
48 3,5-dimethyl-1,2,4-trithiolane (E or Z)
49 3,5-dimethyl-1,2,4-trithiolane (E or Z)
50 1-(dimethyl-2-furyl)-2-propanone
51 2-acetyl-5-methylthiophene
52 nitrogen compound B
8
MS + LRI
MS + LRI
MS
MS + LRI
ms
5
nd
2
6
6
nd
1167 151 (100), 108 (65), 43 (60),
109 (48), 53 (30), 80 (20),
123 (13)
53 3-ethyl-1,2-dithiolan-4-one
54 sulfur compound MW 142
nd
1
nd
nd
1
9
nd
19
MS
ms
1169 (12)
1171 142 (100), 60 (64), 99 (34),
59 (27), 127 (21), 54 (20),
113 (13), 83 (13)
1183
1199 (10)
1206 (12)
1219 151 (100), 108 (70), 43 (52),
109 (43), 53 (35), 80 (20),
136 (17)
55 1-(3-thienyl)-1-propanone
56 2,3-dihydro-6-methylthieno[2,3c]furan
57 3-ethyl-2-formylthiophene
58 nitrogen compound C
2
1
nd
30
nd nd
nd nd
nd 22
23 nd
nd
nd
21
nd
MS + LRI
MS + LRI
MS
ms
59 3-ethyl-5-methyl-1,2,4-trithiolane (E or Z)
60 3-ethyl-5-methyl-1,2,4-trithiolane (E or Z)
61 3-methyl-1,2,4-trithiane
1
1
3
4
nd
2
3
4
5
nd
nd
7
7
nd
nd
MS
MS
MS + LRI
ms
1242 (12)
1250 (12)
1254
1314 165 (100), 122 (63), 123 (33),
43 (32), 137 (17), 67 (16),
150 (12)
62 nitrogen compound D
2
63 a dihydrothienothiophene
8
5
nd
nd
MS
1319 (12)

Reaction of Cysteine or H S with 4-Hydroxy-5-methyl-3(2H)-furanone
J. Agric. Food Chem., Vol. 49, No. 2, 2001 819
mass spectral data, m/z
2
Ta ble 1 (Con tin u ed )
approximate concna
no.
compound
cysteine
H2S
method of IDb
LRIc
(rel intensity or reference)d
64
65
66
67
68
a methylthienothiophene
4
3
nd
2
3
1
9
4
nd
nd
nd
5
nd
nd
nd
nd
MS
MS
MS
MS
1357
1378
1409
1418
1501
(12)
(8)
(8)
(8)
(20)
a dihydromethylthienothiophene
a dihydromethylthienothiophene
a dihydromethylthienothiophene
3-(2-methyl-3-furyldithio)-2-butanone
1
4
7
1
MS + LRI
total
369
354
367
417
a
Concentrations (µg/10 mg of HMF obtained by comparing GC-MS peak area with that from 100 µg of methyl decanoate internal
standard added to the HMF solution before volatile collection); duplicate analyses are shown; nd, not detected (limit of detection ∼0.1
b
µg/10 mg of HMF); tr, between 0.1 and 1 µg/10 mg of HMF. MS + LRI, identified by comparison of mass spectrum and LRI with those
of authentic compound; MS, tentative identification by comparison with mass spectrum reported in the literature; ms, tentative identification
d
by interpretation of mass spectrum. c Linear retention index. Where neither mass spectrum nor a reference is given, the reference spectrum
can be found in the NIST/EPA/NIH mass spectral database. e Previously incorrectly reported as the 2,4-isomer (8).
Ta ble 2. Com p a r ison of Cla sses of Com p ou n d s F ou n d in
Rea ction s betw een HMF a n d Cystein e or Hyd r ogen
Su lfid e a t p H 4.5 a n d 6.5
In previous studies of the reaction of HMF with
cysteine or hydrogen sulfide at pH 4.5 (8), 3-ethyl-2-
formylthiophene (57) was obtained in relatively high
yield (17-57 µg/10 mg of HMF) in reactions containing
approximate concna
cysteine but was not detected in the corresponding
reactions with hydrogen sulfide. As a consequence, it
was proposed that acetaldehyde from cysteine was a key
precursor in the formation of 57. However, in the
current reactions at pH 6.5 the reverse was observed
with significant concentrations of 57 (21-22 µg/10 mg
of HMF) found in the volatiles obtained from the
reaction of HMF with hydrogen sulfide, but 57 was not
detected in those reactions involving cysteine (Table 1).
Accordingly, an alternative reaction sequence involving
precursors derived only from HMF (1-deoxypentosone
and hydroxyacetaldehyde) is now proposed for the
formation of 57 at pH 6.5.
cysteine
H2S
compound class
pH 4.5b pH 6.5 pH 4.5b pH 6.5
dithiolanones and dithianones
trithiolanes and trithianes
thiophenes
41
nd
94
94
28
1
23
25
94
22
7
443
nd
21
39
39
44
18
12
nd
nd
nd
616
20
12
128
129
51
1
nd
31
nd
nd
nd
372
thiophenones
aliphatic and alicyclic ketones
thiols and mercaptoketones
disulfides
furan derivatives
thiazoles and oxazoles
pyrazines
294
24
1
5
23
42
36
71
345
nd
nd
nd
580
other nitrogen compounds
total
Of the remaining thiophenes, 11 was formed only in
the reaction with hydrogen sulfide, whereas 45 was
formed only in small quantities in the reaction with
cysteine (Table 1). The remaining four thiophenes (15,
47, 51, and 55) were formed in only minor quantities
in either one or other of the reaction systems (Table 1).
Tr ith iola n es a n d Dith iola n on es. The four trithio-
lanes (48, 49, 59, and 60) accounted for 4-7% of the
total volatiles formed in the reaction with cysteine and
2-3% of those formed in the system with hydrogen
sulfide. The trithiolanes 48 and 49 were formed only in
the reaction with cysteine, whereas 59 and 60 were
preferentially formed in the reaction with hydrogen
sulfide (Table 1). It has been suggested (16) that 48 and
49 can be formed by the reaction of two molecules of
acetaldehyde with hydrogen sulfide. In this sequence,
acetaldehyde is derived from the decomposition of
cysteine (17). The absence of a ready source of acetal-
dehyde in the reaction of HMF with hydrogen sulfide
would account for the absence of 48 and 49 from the
subsequent reaction products (Table 1). The trithiolanes
59 and 60 could also be formed by the same reaction
sequence (16) by the reaction of propanal and acetal-
dehyde with hydrogen sulfide. However, 59 and 60 are
formed in greater quantities in the reaction between
HMF and hydrogen sulfide (a system with no ready
source of acetaldehyde) than in the reaction of HMF
with cysteine (Table 1). It is feasible that acetaldehyde
can be formed by the breakdown of HMF, although the
failure to detect the trithiolanes 48 and 49 as products
of the reaction of HMF with hydrogen sulfide (Table 1)
would indicate that acetaldehyde is not formed in
significant quantities by this process. Furthermore,
none of these trithiolanes were found when the reaction
was carried out at pH 4.5 (8). The formation of 59 and
a
Mean concentrations (µg/10 mg of HMF) from duplicate
b
analyses; nd, not detected. From ref 8.
hyde replacing hydroxyacetaldehyde in its reaction with
2,3-pentanedione. The presence of the six dihydrothiophe-
nones in comparable concentrations in both reaction
systems (Table 1) strongly supports the proposal that
these compounds are principally derived from break-
down products of HMF. The compound tentatively
identified as 2-ethyl-5-methyl-4,5-dihydrothiophen-
3(2H)-one (43) has a simple mass spectrum similar to
that of the 2-ethyl derivative (39). Compound 43 has a
base peak m/z 74 (C2H2OS+) corresponding to the loss
of the netural fragments C2H4 and C3H6 (Figure 1). It
also has weak ions at m/z 116 (C5H8OS+) corresponding
to the loss of C2H4 and m/z 102 (C4H6OS+) corresponding
to the loss of C3H6.
Th iop h en es. A total of 10 thiophenes (10, 11, 15, 26,
41, 45, 47, 51, 55, and 57) were formed in the two
reaction systems. These thiophenes accounted for 4-10%
of the total volatiles produced in the system with
cysteine but 31-34% of those formed in the system with
hydrogen sulfide. The major compounds formed in the
reaction between HMF and hydrogen sulfide were
2-acetylthiophene (41), 15-18%; 3-ethyl-2-formylthio-
phene (57), 5-6%; 2,5-dimethylthiophene (10), 1-7%;
and 2-ethyl-5-methylthiophene (26), 2-4%. The thio-
phenes 41 and 57 were not detected in the system with
cysteine, whereas the other two compounds were present
in either the same concentrations or slightly less (Table
1). Compounds 10 and 41 can be formed from the same
pair of HMF retro-aldol condensation products (8),
hydroxyacetaldehyde and 2,3-butandione, and 26 from
hydroxyacetaldehyde and 2,3-pentanedione.

820 J. Agric. Food Chem., Vol. 49, No. 2, 2001
Whitfield and Mottram
F igu r e 1. Proposed mass spectral fragmentation of 2-ethyl-5-methyl-4,5-dihydrothiophen-3(2H)-one.
F igu r e 2. Proposed mass spectral fragmentation of acetylalkylpyridinones tentatively identified in the reaction of HMF with
cysteine.
60 in the reaction between HMF and hydrogen sulfide
warrants further study.
28 with mercaptoacetaldehyde have been described (8).
The same reaction sequences can be used to account for
the formation of the dihydrothienothiophene 63 from
the reaction of tetrahydrothiophen-3(2H)-one and mer-
captoacetaldehyde and of the methylthienothiophene 64
from the reaction of (2 or 5)-methylthiophen-3(2H)-one
with mercaptoacetaldehyde. However, further work
would be necessary to assign structures to these com-
pounds. Of interest, although most of these thienothio-
phenes were formed in higher yields in the reaction of
HMF with cysteine at pH 4.5 (8), the dihydro-
thienothiophene 63 was found in significant quantities
only when this reaction was performed at pH 6.5.
P yr a zin es, Th ia zoles, Oxa zoles, a n d Oth er Ni-
tr ogen Heter ocyclic Com p ou n d s. Six pyrazines (8,
18, 20, 30, 31, and 37), five thiazoles (14, 19, 23, 29,
and 44), four oxazoles (3, 9, 21, and 22), and four
unidentified nitrogen compounds (40, 52, 58, and 62)
were formed in the reaction of HMF with cysteine (Table
1). The pyrazines account for 8-11% of the total
volatiles, thiazoles, 6%; oxazoles, 5-7%; and the uni-
dentified compounds, 16-19%.
The three dithiolanones (38, 42, and 53) were found
only in trace quantities in the reaction of HMF with
cysteine but accounted for 4-6% of the total volatiles
formed in the reaction of HMF with hydrogen sulfide.
Significantly, these compounds were major products
(>70%) of the reaction of HMF with hydrogen sulfide
at pH 4.5 (8). The difference in yields of the dithiol-
anones 38, 42, and 53 from the reactions carried out at
pH 4.5 and 6.5 indicates that pH has a profound effect
on the course of the reactions leading to these com-
pounds. Possible pathways for the formation of 38, 42,
and 53 from the breakdown products of HMF have been
described (8).
Th ien oth iop h en es. Five thienothiophenes (63-67),
accounting for 4-7% of the total volatiles, were formed
in the reaction of HMF with cysteine. However, only two
of these, 64 and 65, accounting for 1-4% of the volatiles,
were formed in the reaction of HMF with hydrogen
sulfide (Table 1). Possible pathways to the three dihy-
dromethylthienothiophenes 65-67 involving the reac-
tion of the dihydromethylthiophen-3(2H)-ones 27 and
Five of the six pyrazines have a feature common to

Reaction of Cysteine or H S with 4-Hydroxy-5-methyl-3(2H)-furanone
J. Agric. Food Chem., Vol. 49, No. 2, 2001 821
2
each of their structures. They have a methyl adjacent
to an unsubstituted ring carbon atom on one side of the
molecule. This would suggest that these pyrazines have
a common precursor, possibly hydroxyacetone formed
by the reduction of the HMF hydrolysis product, pyru-
valdehyde. Other compounds that could be involved in
the formation of these pyrazines would be glyoxal,
pyruvaldehyde, 1,2-butanedione, 2,3-butanedione, and
2,3-pentanedione, all feasible hydrolysis products of
HMF (8). The key steps in the formation of these
pyrazines are the reaction of ammonia with 2-hydroxy-
propanal (from the reduction of pyruvaldehyde) to form
an aminoacetone and the condensation of this compound
with appropriate R-dicarbonyl compounds (18).
With the exception of the thiazoles 23 and 44 and the
oxazoles 3 and 22, all other thiazoles and oxazoles
formed in the reaction of HMF with cysteine have a
methyl substituent in the 2-position (Table 1). Accord-
ingly, it is likely that one of the precursors to all of these
compounds is acetaldehyde, derived from the decompo-
sition of cysteine (17). Other compounds likely to be
involved in the formation of such 2-methylthiazoles and
oxazoles would be the HMF hydrolysis products pyru-
valdehyde (for compounds 14 and 19), 2,3-butanedione
(for 9 and 29), and 2,3-pentanedione (for 21). The key
steps in the reaction sequences to these compounds are
the reaction of ammonia with acetaldehyde to form
1-aminoethanol and the condensation of this compound
with appropriate R-dicarbonyl compounds. Subsequent
cyclization of one of these condensation products fol-
lowed by dehydration would lead to the oxazoles 9 and
21. The thiazoles 14, 19, and 29 could be formed from
the same condensation products by reaction with hy-
drogen sulfide followed by cyclization and dehydration
(19).
role of pH in determining the course of the later stages
of the Maillard reaction. In particular, at pH 4.5
ammonia from the decomposition of cysteine is es-
sentially not available for reaction with the HMF
hydrolysis products to form nitrogen compounds such
as pyrazines, thiazoles, and oxazoles. However, lower
pH values are essential for the reaction of hydrogen
sulfide with such HMF products to form mercaptoke-
tones, furan- and thiophenethiols, and dithiolanones.
These compounds make important contributions to the
meaty characteristics of cooked meats. Accordingly, the
reactions described in this, and in our earlier work (8),
may explain not only the routes to some important
aroma compounds in heated foods but also the condi-
tions necessary for their formation. Future studies
should therefore be directed toward establishing whether
such reaction pathways and reaction conditions are of
similar importance in the formation of key cooked flavor
compounds in natural systems.
ACKNOWLEDGMENT
We gratefully acknowledge the assistance of Kevin
Shaw in carrying out the GC-MS analyses and Franz
Hammerschmidt (DRAGOCO, Holzminden, Germany)
for the Masslib Software interpretations of unidentified
mass spectra.
LITERATURE CITED
(1) Feather, M. S. Amine-assisted sugar dehydration reac-
tions. In Maillard Reactions in Food; Eriksson, C., Ed.;
Pergamon Press: Oxford, U.K., 1981; pp 37-45.
(2) Tonsbeek, C. H. T.; Plancken, A. J .; Weerdhof, T. v. d.
Components contributing to beef flavor. Isolation of
4-hydroxy-5-methyl-3(2H)-furanone and its 2,5-dimethyl
homolog from beef broth. J . Agric. Food Chem. 1968,
16, 1016-1021.
The four unidentified nitrogen compounds (40, 52, 58,
and 62) are major volatile products (16-19%) of the
reaction of HMF with cysteine. Only the thiophenones
(25-27%) were produced in greater quantities in this
system. All four compounds gave mass spectra with the
molecular ions as base peaks, and all had fragment ions
at M - 15 (M - CH3), M - 28 (M - CO), M - 42 (M -
CH2dCdO), M - 43 (M - CH3CO), M - 42 - 28 (M -
CH2dCdO-CO), and M - 43 - 28 (M - CH3CO-CO).
Interpretation of their mass spectra using Masslib
Software indicated that 40 could be 2-acetyl-3(2H)-
pyridinone and that 52, 58, and 62 could also be
2-acetyl-3(2H)-pyridinones with alkyl substituents in
the 4-, 5-, or 6-position. The fragmentation patterns of
these four components can be rationalized as shown in
Figure 2. However, a search of the literature has shown
that these compounds have not been synthesized. Ac-
cordingly, studies are currently in progress to estab-
lish the structures of the major reaction products 40 and
58.
In summary, the reactions of HMF with cysteine or
hydrogen sulfide in heated aqueous solutions at either
pH 4.5 or 6.5 produced complex mixtures of volatiles.
However, the compositions of the mixtures formed at
different pH values were significantly different. At pH
4.5 the dominant compounds produced, both qualita-
tively and quantitatively, were sulfur compounds (8),
whereas at pH 6.5 41% of the total volatiles formed in
the reaction of HMF with cysteine were nitrogen
compounds and far fewer sulfur compounds were pro-
duced in both the cysteine and hydrogen sulfide sys-
tems. These results further demonstrate the important
(3) van den Ouweland, G. A. M.; Peer, H. G. Components
contributing to beef flavor. Volatile compounds produced
by the reaction of 4-hydroxy-5-methyl-3(2H)-furanone
and its thio analog with hydrogen sulfide. J . Agric. Food
Chem. 1975, 23, 501-505.
(4) Shu, C.-K.; Ho, C.-T. Effect of pH on the volatile
formation from the reaction between cysteine and 2,5-
dimethyl-4-hydroxy-3(2H)-furanone. J . Agric. Food Chem.
1988, 36, 801-803.
(5) Mottram, D. S.; Whitfield, F. B. Volatile compounds from
the reaction of cysteine, ribose and phospholipid in low
moisture systems. J . Agric. Food Chem. 1995, 43, 984-
988.
(6) Mottram, D. S.; Whitfield, F. B. Maillard-lipid interac-
tions in nonaqueous systems: volatiles from the reaction
of cysteine and ribose with phosphatidylcholine. J . Agric.
Food Chem. 1995, 43, 1302-1306.
(7) Whitfield, F. B.; Mottram, D. S. Maillard lipid interac-
tions in low moisture systems. In The Contribution of
Low- and Nonvolatile Materials to the Flavor of Foods;
Pickenhagen, W., Ho, C.-T., Spanier, A. M., Eds.; Allured
Publishing: Carol Stream, IL, 1996; pp 149-181.
(8) Whitfield, F. B.; Mottram, D. S. Investigation of the
reaction between 4-hydroxy-5-methyl-3(2H)-furanone
and cysteine or hydrogen sulfide at pH 4.5. J . Agric.
Food Chem. 1999, 47, 1626-1634.
(9) Mottram, D. S. Meat. In Volatile Compounds in Foods
and Beverages; Maarse, H., Ed.; Dekker: New York,
1991; pp 107-177.
(10) ten Noever de Brauw, M. C.; Bouwman, J .; Tas, A. C.;
La Vos, G. F. In Compilation of Mass Spectra of Volatile
Compounds in Food; Institute for Nutrition and Food
Research TNO: Ziest, The Netherlands, 1979.
(11) Mulders, E. J . Volatile components from the non-
enzymic browning reaction of cysteine/cystine-ribose

822 J. Agric. Food Chem., Vol. 49, No. 2, 2001
Whitfield and Mottram
system. Z. Lebensm. Unters. Forsch. 1973, 152, 193-
201.
(12) Farmer, L. J .; Mottram, D. S.; Whitfield, F. B. Volatile
compounds produced in Maillard reactions involving
cysteine, ribose and phospholipid. J . Sci. Food Agric.
1989, 49, 347-368.
(17) Obata, Y.; Tanaka, H. Studies on the photolysis of sulfur
containing compounds in foods. Agric. Biol. Chem. 1965,
29, 196-199.
(18) Shibamoto, T.; Bernhard, R. A. Investigation of pyrazine
formation pathways in sugar-ammonia model systems.
J . Agric. Food Chem. 1977, 25, 609-614.
(19) Vernin, G.; Pa´rka´nyi, C. Mechanisms of formation of
heterocyclic compounds in Maillard and pyrolysis reac-
tions. In Chemistry of Heterocyclic Compounds in Fla-
vours and Aromas; Vernin, G., Ed.; Ellis Horwood:
Chichester, U.K., 1982; pp 151-207.
(20) Mottram, D. S.; Madruga, M. S.; Whitfield, F. B. Some
novel meatlike aroma compounds from the reactions of
alkanediones with hydrogen sulfide and furanthiols. J .
Agric. Food Chem. 1995, 43, 189-193.
(21) Stoll, M.; Winter, M.; Gautschi, F.; Flament, I.; Will-
halm, B. Recherches sur les aromes. Sur l’aroˆme de cafe´.
Helv. Chim. Acta 1967, 50, 628-694.
(22) Shu, C.-K.; Hagedorn, M. L.; Mookherjee, B. D.; Ho, C.-
T. pH effect on the volatile components in the thermal
degradation of cysteine. J . Agric. Food Chem. 1985, 33,
442-446.
(13) Farmer, L. J .; Mottram, D. S. Interaction of lipid in the
Maillard reaction between cysteine and ribose: effect
of a triglyceride and three phospholipids on the volatile
products. J . Sci. Food Agric. 1990, 53, 505-525.
(14) Gu¨ntert, M.; Bru¨ning, J .; Emberger, R.; Kopsel, M.;
Kuhn, W.; Thielmann, T.; Werkoff, P. Identification and
formation of some selected sulfur-containing flavor
compounds in various meat model systems. J . Agric.
Food Chem. 1990, 38, 2027-2041.
(15) Hofmann, T.; Schieberle, P. Evaluation of the key
odorants in a thermally treated solution of ribose and
cysteine by aroma extract dilution techniques. J . Agric.
Food Chem. 1995, 43, 2187-2194.
(16) Takken, H. J .; van der Linde, L. M.; de Valois, P. J .;
van Dort, H. M.; Boelens, M. Reaction products of
R-dicarbonyl compounds, aldehydes, hydrogen sulfide,
and ammonia. In Phenolic, Sulfur and Nitrogen Com-
pounds in Food Flavors; Charalambous, G., Katz, I.,
Eds.; American Chemical Society: Washington, DC,
1976: pp 114-121.
Received for review J uly 11, 2000. Revised manuscript received
December 5, 2000. Accepted December 5, 2000.
J F0008644