2-Oxopropanal, Hydroxy-2-propanone, and 1-Pyrroline - Important Intermediates in the Generation of the Roast-Smelling Food Flavor Compounds 2-Acetyl-1-pyrroline and 2-Acetyltetrahydropyridine

Article abstract of DOI:10.1021/jf970990g

On the basis of labeling experiments with [¹³C]₆-glucose/unlabeled proline as well as quantitative data obtained in model studies using stable isotope dilution assays, 1-pyrroline and hydroxy-2-propanone were identified as effective intermediates in generating the roast-smelling food odorant 2-acetyltetrahydropyridine (ATHP; two tautomers). Synthesis of the key precursor compound, 2-(1-hydroxy-2-oxo-propyl)pyrrolidine, and studies on its degradation confirmed the important role of this intermediate in ATHP formation. Boiling of the intermediate for 30 min in aqueous solution generated >30% of ATHP on a molar basis. 1-Pyrroline and 2-oxopropanal were confirmed as important intermediates in the generation of the further roast food odorant, 2-acetyl-1-pyrroline (AP). On the basis of results of labeling experiments with [¹³C]₆-glucose/unlabeled proline, two different mechanisms could be proposed. One leads to AP via 2-(1,2-dioxopropyl)pyrrolidine as the precursor with elimination of the aldehyde group in 2-oxopropanal as carbon dioxide. The other one suggests elimination of carbon-2 of the pyrroline ring. The latter mechanism was further established by a result showing that from the reaction of 2-methyl- 1-pyrroline with 2-oxopropanal AP was also generated.

Full text of DOI:10.1021/jf970990g

2270
J. Agric. Food Chem. 1998, 46, 2270−2277
2-Oxop r op a n a l, Hyd r oxy-2-p r op a n on e, a n d 1-P yr r olin esIm p or ta n t
In ter m ed ia tes in th e Gen er a tion of th e Roa st-Sm ellin g F ood F la vor
Com p ou n d s 2-Acetyl-1-p yr r olin e a n d 2-Acetyltetr a h yd r op yr id in e
Thomas Hofmann and Peter Schieberle*
Institut fu¨r Lebensmittelchemie und Deutsche Forschungsanstalt fu¨r Lebensmittelchemie,
Lichtenbergstrasse 4, D-85748 Garching, Germany
On the basis of labeling experiments with [13C]6-glucose/unlabeled proline as well as quantitative
data obtained in model studies using stable isotope dilution assays, 1-pyrroline and hydroxy-2-
propanone were identified as effective intermediates in generating the roast-smelling food odorant
2-acetyltetrahydropyridine (ATHP; two tautomers). Synthesis of the key precursor compound, 2-(1-
hydroxy-2-oxo-propyl)pyrrolidine, and studies on its degradation confirmed the important role of
this intermediate in ATHP formation. Boiling of the intermediate for 30 min in aqueous solution
generated >30% of ATHP on a molar basis. 1-Pyrroline and 2-oxopropanal were confirmed as
important intermediates in the generation of the further roast food odorant, 2-acetyl-1-pyrroline
(AP). On the basis of results of labeling experiments with [13C]6-glucose/unlabeled proline, two
different mechanisms could be proposed. One leads to AP via 2-(1,2-dioxopropyl)pyrrolidine as the
precursor with elimination of the aldehyde group in 2-oxopropanal as carbon dioxide. The other
one suggests elimination of carbon-2 of the pyrroline ring. The latter mechanism was further
established by a result showing that from the reaction of 2-methyl-1-pyrroline with 2-oxopropanal
AP was also generated.
Keyw or d s: 1-Pyrroline; 2-oxopropanal; hydroxy-2-propanone; 2-acetyl-1-pyrroline; 2-acetyltetrahy-
dropyridine; 2-acetyl-3-methyl-3,4,5,6-tetrahydropyridine; 2-(1-hydroxy-2-oxo-1-propyl)pyrrolidine
Ta ble 1. Con cen tr a tion s a n d Od or Activity Va lu es
(OAVs) of AP a n d ATHP in P r ocessed F ood sa
INTRODUCTION
On the basis of high odor activity values (ratio of
concentration to odor threshold), the intense roast
odorants 2-acetyl-1-pyrroline (AP) and 2-acetyltetrahy-
dropyridine (ATHP) have been established as key
contributors to the roasty, popcorn-like odor of several
processed foods (Table 1). It has long been suggested
that the Maillard reaction between amino acids and
carbohydrates, or degradation products thereof, gener-
ates both odorants in foods, and several investigations
have been performed to clarify the intermediates and
reaction pathways leading to AP and ATHP during food
processing.
Hodge et al. (1972) were the first to postulate the
amino acid proline and the carbohydrate degradation
product 2-oxopropanal as key intermediates in the
formation of ATHP. They speculated that a hypotheti-
cal intermediate, N-acetonyl-4-aminobutanal (I in Fig-
ure 1), should give the ATHP simply by elimination of
water. However, recent results by de Kimpe et al.
(1994) questioned this mechanism, because they were
unable to generate the ATHP from the synthesized
intermediate.
AP
ATHP
concn
(µg/kg)
concn
(µg/kg)
food
OAVb
OAVb
wheat bread crust
popcorn
toasted wheat bread
roasted sesame
basmati rice
19
24
2602
3288
1205
4110
83516
6027
53
981
8092
28
437
1.5
ndc
ndc
ndc
8.8
30
610d
4d
cooked sweet corn
a
Based on results reported by Schieberle and Grosch (1987),
Schieberle (1991), and Rychlik and Grosch (1996). OAVs were
calculated by dividing the concentrations by the odor thresholds
b
in wheat starch: AP, 0.0073 µg/kg; ATHP, 0.054 µg/kg (Rychlik
and Grosch, 1996). c nd, not determined. Data from Buttery et
d
al. (1994).
berle, 1989, 1990) established the important role of the
amino acid proline as precursor of both odorants and
indicated that the AP is preferentially formed when
proline is reacted with trioses such as glycerine alde-
hyde, 1,3-dihydroxyacetone, or their common dehydra-
tion product 2-oxopropanal. In contrast, the ATHP was
shown to be formed in higher amounts when proline was
reacted with hexoses. Further findings (Schieberle,
1990) indicated the amino acid ornithine as another, but
more effective, precursor of AP. However, ATHP was
not generated from this amino acid.
Tressl et al. (1985) found trace amounts of AP and
ATHP when proline was reacted in the presence of
several carbohydrates. Quantitative investigations per-
formed by using stable isotope dilution assays (Schie-
1-Pyrroline, which may be formed either from proline
via a Strecker degradation initiated by R-dicarbonyls,
such as 2-oxopropanal (Figure 1), or, alternatively, from
ornithine via 4-aminobutanal (Schieberle, 1990) is
* Author to whom correspondence should be addressed
(telephone +49-89-289 141 70; fax +49-89-289 141 83; e-mail
lebchem.schieberle@lrz.tu-muenchen.de).
S0021-8561(97)00990-4 CCC: $15.00 © 1998 American Chemical Society
Published on Web 05/13/1998

Intermediates in Roast Odorant Formation
J. Agric. Food Chem., Vol. 46, No. 6, 1998 2271
F igu r e 1. Formation of ATHP, 1-pyrroline, and hydroxy-2-propanone from proline and 2-oxopropanal [according to Hodge et al.
(1972)].
generally accepted as a key intermediate in the genera-
tion of AP (Schieberle, 1989, 1990; Tressl et al., 1993a,b;
Rewicki et al., 1993).
To gain more detailed insight into AP formation,
model studies using labeled glucose have been per-
formed in the literature. In AP formed from U-13C-
labeled glucose in the presence of proline, the main
isotopomer (80%) contained two labeled carbon atoms
(Schieberle, 1989). Depending on this result, an “acyl-
ation” of 1-pyrroline by 2-oxopropanal with elimination
of formaldehyde was suggested as the key step in AP
formation (Schieberle, 1990, 1995). Tressl et al. (1993a,b)
and Rewicki et al. (1993) detected a 1:1 mixture of either
unlabeled or 13C1-labeled AP when proline was reacted
with 1-13C-labeled glucose. They proposed an acylation
of 1-pyrroline by acetylformoine leading to 2-acetyl-
pyrrolidine, which should subsequently be oxidized into
the AP. However, this key reaction step has not yet
been proven by systematic quantitative studies.
The literature survey indicates that, first, the inter-
mediates and reaction pathways leading to the forma-
tion of ATHP are still unclear and, second, it is not
known in detail how the AP is formed from 1-pyrroline.
The following investigations were, therefore, under-
taken to further the understanding of the mechan-
isms governing the formation of these two very im-
portant food odorants during thermal processing of
foods.
F igu r e 2. Mass spectrum (MS/EI) of AMTHP.
decarboxylation of L-proline (Yoshikawa et al., 1965; Schie-
berle, 1990). [2H]-2-Acetyl-1-pyrroline and [2H]-2-acetyltetra-
hydropyridine were synthesized as described by Schieberle and
Grosch (1987) and Schieberle (1995).
2-Acetyl-3-methyl-3,4,5,6-tetrahydropyridine. 2-Methyl-1-
pyrroline (5 mmol) and hydroxy-2-propanone (50 mmol) were
dissolved in phosphate buffer (500 mL, 0.5 mol/L; pH 7.0) and
the solution was refluxed for 2 h. The aqueous mixture was
extracted eight times with diethyl ether (total volume ) 250
mL); the etheral solution was treated with a sodium bicarbon-
ate solution (50 mL, 0.5 mol/L) and finally dried over Na2SO4.
After concentration to 2 mL, the target compound was isolated
by flash chromatography on a Diol-phase (column, 20 cm ×
1.9 cm; J . T. Baker BV, Deventen, The Netherlands) using an
n-pentane/diethyl ether gradient. HRGC separation on a
Silicone SE-54 column afforded two peaks showing nearly
identical mass spectra (Figure 2). One major peak (∼95%) was
eluted at a retention index of 1068 (relative to n-alkanes) and
a minor one (5%) at a retention index of 1166. The major
compound was isolated by preparative GLC using the equip-
ment described previously (Schieberle, 1991). Its 1H NMR
data (Table 2) confirmed the structure of the target compound
as 2-acetyl-3-methyl-3,4,5,6-tetrahydropyridine. The minor
compound was assigned as 2-acetyl-3-methyl-1,4,5,6-tetrahy-
dropyridine.
EXPERIMENTAL PROCEDURES
Ch em ica ls. L-Proline, pipecolinic acid, 4-aminobutanal (as
diethyl acetal), 3-benzyl-5-(2-hydroxyethyl)-4-methylthiazol-
ium chloride, 2-methyl-1-pyrroline, hydroxy-2-propanone, 1-hy-
droxy-2-butanone, 2-oxopropanal (40% solution in water), and
2-(hydroxymethyl)pyrrolidine were from Aldrich (Steinheim,
Germany). 2-[(tert-Butoxycarbonyl)oximino]-2-phenylaceto-
nitrile was supplied by Lancaster (Mu¨hlheim, Germany).
[13C]6-Glucose was from Sigma (Munich, Germany).
Syn th eses. AP, 2-acetylpyrrolidine, ATHP, and 2-acetyl-
piperidine were prepared as described recently (Hofmann and
Schieberle, 1998). 1-Pyrroline was generated by an oxidative
2-(1-Hydroxy-2-oxo-1-propyl)pyrrolidine (IV in Figure 3).
The target compound was synthesized using the four-step

2272 J. Agric. Food Chem., Vol. 46, No. 6, 1998
Hofmann and Schieberle
Ta ble 2. 1H NMR Da ta Obta in ed for AMTHP
the organic phase was separated and dried over Na2SO4, and
the solvent was distilled off. The remaining material was
purified by flash chromatography as described above. The
target compound (1.1 g, )49% of theory) was obtained and
characterized by mass spectral and 1H NMR measurements:
MS/EI [m/z (%)] 114 (100), 70 (96), 41 (92), 57 (82), 126 (77),
170 (77), 82 (52); MS/CI [m/z (%)] 144 (100), 100 (95), 114 (49),
126 (45), 128 (26), 170 (12), 172 (9), 200 (7), 184 (6).
hydrogen
at carbona
δ (ppm)
1.04
1.48-1.63
multiplicity J (Hz)
TOCSYb
9
4
d, 3H
7.63
H at C-9/C-5
H at C-4/C-5
H at C-4/C-3
H at C-3/C-4
H at C-3/C-2
m, 2H
3
1.61-1.75
m, 2H
1
The H NMR signals of II were very similar to those of the
8
5
2.03
2.8-2.9
s, 3H
m, 1H
corresponding alcohol (I in Figure 3) with the exception of the
signal at δ 9.5 (d; 1H; C-8) for the aldehyde hydrogen.
N-(tert-Butoxycarbonyl)-2-(1-hydroxy-2-oxo-1-propyl)pyr-
rolidine (III in Figure 3). To a mixture of II (5.5 mmol), freshly
distilled acetaldehyde (17.3 mmol), and 3-benzyl-5-(2-hydroxy-
ethyl)-4-methylthiazolium chloride (1.5 mmol) was added
dropwise dry triethylamine (9 mmol), and the mixture was
heated for 90 min at 80 °C under pure argon with vigorous
stirring. After cooling, water (50 mL) was added and the pH
adjusted to 5.0 by using hydrochloric acid (1 mol/L). Extraction
with diethyl ether, drying over Na2SO4, and distilling off the
solvent afforded a crude mixture from which III was purified
by flash chromatography (yield ) 0.28 g, )21% of theory). III
was purified from the coeluting isomer N-(tert-butoxycarbonyl)-
2-(2-hydroxy-1-propanoyl)pyrrolidine by preparative HPLC on
reversed phase material (C18; Shandon Hypersil, 5 µ; East-
more, U.K.) and by using an acetonitrile/water gradient from
20+80 to 80+20 by vol within 30 min. HRGC/mass spectral
measurements of the purified compound gave the following
signals: MS/EI [m/z (%)] 43 (100), 170 (95), 114 (92), 57 (83),
74 (72), 70 (71), 100 (70), 144 (63), 126 (36); MS/CI [m/z (%)]
114 (100), 188 (51), 70 (34), 144 (31), 170 (21), 142 (13), 126
(8), 186 (5), 244 (2).
H at C-5/C-9
H′ at C-5/C-4
H at C-2/C-3
H at C-2/C-3
2
3.6-3.7
3.82-3.93
m, 1H
m, 1H
a
Numbering refers to the structure displayed in Figure 2.
Total correlated (1H, 1H) shift spectroscopy.
b
The structure of III was confirmed by 1H NMR measure-
ments (360 MHz; CDCl3): δ 1.43 (s, 9H; C-1; cf. Figure 3),
1.62-1.88 (m; 3H; C5/C6), 1.90-2.06 (m; 1H; C-5), 2.26 (s, 3H;
C-10), 3.28-3.37 (m; 1H; C-4), 3.38-3.57 (m, 1H; C-4), 4.00-
4.4 (m, 2H; C-7/C-8).
2-(1-Hydroxy-2-oxo-1-propyl)pyrrolidine (IV in Figure 3). III
(1 mmol) was dissolved in dichloromethane (2 mL) and, after
addition of trifluoroacetic acid (60 mg), stirred for 3 h at room
temperature. The mixture was dried at 20 °C in a stream of
pure nitrogen, the residue was taken up in water (100 mL),
and aliquots (1 mL) of this solution were used in the model
experiments without further purification.
La belin g Exp er im en ts. L-Proline (2 mmol) and [13C]6-D-
glucose (1 mmol) were intimately mixed with silica gel (2.7 g
containing 300 µL of a phosphate buffer (0.1 mol/L; pH 7.0)
and heated for 10 min at 160 °C in closed glass vials (total
volume ) 10 mL). The mixture was extracted three times with
diethyl ether (total volume ) 50 mL) and concentrated to 200
µL by distilling off the solvent at 35 °C using a Vigreux column
(60 cm × 1 cm) followed by microdistillation (Schieberle, 1991).
Mod el Rea ction s. The composition of the model systems
and the reaction parameters applied are detailed in the
respective tables. The amounts of AP and ATHP formed were
determined from the volatile fractions obtained by sublimation
in vacuo by means of stable isotope dilution assays and by
using [2H]-AP and [2H]-ATHP as internal standards (Schie-
berle, 1995).
High -Resolu tion Ga s Ch r om a togr a p h y (HRGC)/Ma ss
Sp ectr om etr y (MS). HRGC was performed with a type 5160
gas chromatograph (Fisons Instruments, Mainz, Germany) by
using the following capillaries: FFAP (30 m × 0.32 mm fused
silica capillary, free fatty acid phase, 0.25 µm; J &W Scientific,
Fisons Instruments, Mainz, Germany) and SE-54 (30 m × 0.32
mm fused silica capillary DB-5; 0.25 µm; J &W Scientific,
Fisons Instruments). The samples were applied by the on-
column injection technique at 40 °C. After 2 min, the tem-
perature of the oven was quickly raised at 40 °C/min to 50 °C
(SE-54) or 60 °C (FFAP), respectively, held for 5 min isother-
mally, then raised at 6 °C/min to 230 °C, and held for 15 min.
The flow of the carrier gas helium was 2.5 mL/min. Linear
retention indices (RI) of the compounds were calculated from
the retention times of n-alkanes by using a computer program
[cf. Schieberle (1991)]. MS analysis was performed with an
F igu r e 3. Synthetic route used for the preparation of HOP.
synthesis detailed in Figure 3 starting from 2-(hydroxymethyl)-
pyrrolidine.
N-(tert-Butoxycarbonyl)-2-(hydroxymethyl)pyrrolidine (I in
Figure 3). 2-Hydroxymethyl pyrrolidine (15 mmol), triethyl-
amine (22.5 mmol), and 2-[(tert-butyoxycarbonyl)oximino]-2-
phenylacetonitrile (16.5 mmol) were stirred for 2 h at room
temperature in dioxane/water (20 mL; 1+1 by vol). The excess
of solvent was then distilled off at 20 °C in vacuo, and, after
addition of water (50 mL) and cooling, the oily phase obtained
was purified by flash chromatography on a Diol-phase sus-
pended in n-pentane (column, 15 × 1.9 cm; J . T. Baker BV).
After flushing with n-pentane (150 mL), the target compound
was eluted with n-pentane/diethyl ether (150 mL; 9+1 by vol)
followed by n-pentane/diethyl ether (150 mL; 8+2 by vol). A
colorless oil (2.3 g, )75% of theory) was obtained after distilling
off the solvent. The target compound was characterized by
HRGC/mass spectrometry: MS/EI [m/z (%)] 114 (100), 41 (95),
170 (84), 70 (82), 57 (78), 128 (77); MS/CI [m/z (%)] 146 (100),
102 (83), 128 (49), 114 (31), 130 (15), 170 (13), 202 (12), 84 (9).
By 1H NMR measurements the following signals were
obtained (δ; multiplicity): δ 1.47 (s, 9H; C-1 in Figure 3), 1.70-
2.12 (m; 4H; C-5 and C-6), 3.29-3.72 (m; 2H; C-4), 3.65-3.83
(m; 1H; C-7), 3.95-4.05 (m; 1H; C-8), 4.12-4.21 (m; 1H; C-8).
N-(tert-Butoxycarbonyl)-2-formylpyrrolidine (II in Figure 3).
A solution of I (11.3 mmol) and pyridinium chlorochromate
(17 mmol) in dichloromethane (25 mL) was stirred for 150 min
at room temperature. After addition of diethyl ether (40 mL),

Intermediates in Roast Odorant Formation
J. Agric. Food Chem., Vol. 46, No. 6, 1998 2273
Ta ble 3. Ca r bon Isotop e Ra tio (P er cen t) in ATHP
Gen er a ted fr om P r olin e in th e P r esen ce of Glu cose (Glc)
or [13C]6-Glu cose ([13C]6-Glc), Resp ectivelya
Ta ble 4. In flu en ce of th e p H Va lu e on th e F or m a tion of
ATHP fr om 1-P yr r olin e a n d Hyd r oxy-2-p r op a n on ea
ATHP
ATHP
ATHP
(m/z)
ATHP
(m/z)
pH
µg
%
pH
µg
%
Glc
[
13C]6-Glc
Glc
[
13C]6-Glc
3.0
5.0
<0.1
7.0
9.0
10.8
38.4
0.9
3.1
123
124
125
126
0.1
4.9
88.3
6.4
<0.1
<0.1
<0.1
0.7
127
128
129
130
0.3
<0.1
<0.1
<0.1
5.2
89.2
4.7
0.9
0.1
a
1-Pyrroline (10 µmol) and hydroxy-2-propanone (10 µmol) were
0.2
reacted for 30 min at 100 °C in phosphate buffer (5 mL, 0.5 mol/
L).
The data would be in good agreement with the
reaction mechanism proposed by Hodge et al. (1972)
suggesting 2-oxopropanal (a carbon-3 compound) as the
key intermediate in ATHP formation (Figure 1). How-
ever, as mentioned in the Introduction, recent studies
had shown that the intermediate N-acetonyl-4-amino-
butanal (Figure 1) did not yield ATHP (de Kimpe et al.,
1994).
A
A Strecker degradation is most probable if an amino
acid is reacted with an R-dicarbonyl compound, such as
2-oxopropanal. From proline, the reaction products are
1-pyrroline and hydroxy-2-propanone (Figure 1). Both
compounds are reactive intermediates and, therefore,
further reactions can be proposed. To study their
efficiency as precursors of ATHP, 1-pyrroline and hy-
droxy-2-propanone were reacted by boiling for 30 min
in an aqueous buffer. Under weak acidic conditions (pH
3.0; Table 4), no ATHP was formed. However, increas-
ing the pH to 9.0 led to a significant formation of the
odorant in a 3.1% molar yield.
B
In Figure 5, a reaction pathway leading from 1-pyr-
roline and hydroxy-2-propanone to ATHP is proposed.
From the result that higher pH values favor the
formation of the odorant (Table 4), it can be assumed
that the key step in the reaction is the attack of carbon-1
in the enolized hydroxy-2-propanone at carbon-2 of
1-pyrroline. Once formed, the 2-(1-hydroxy-2-oxopro-
pyl)pyrrolidine (HOP) intermediate may undergo a ring
opening leading to 5,6-dioxoheptylamine, which should
finally give ATHP by a Schiff reaction.
To prove that a ring enlargement takes place in the
reaction, 2-methyl-1-pyrroline instead of 1-pyrroline was
reacted with hydroxy-2-propanone. 2-Acetyl-3-methyl-
3,4,5,6-tetrahydropyridine (AMTHP; Figure 2) was
formed as the main volatile reaction product (3 mol %;
data not shown). Because the methyl group at carbon-2
of the 1-pyrroline was shifted into position 3 in the
AMTHP, this result unambiguously confirms a ring
enlargement reaction during the formation of ATHP
(Figure 5).
It is interesting to note that, compared to ATHP, the
AMTHP mainly occurred in one tautomeric form during
HRGC, namely the 3,4,5,6-tetrahydro compound. The
corresponding 1,4,5,6-tautomer was present only in a
more than 30-fold lower amount (data not shown).
Obviously the electron-donating property of the methyl
group stabilizes the imine tautomer. As recently shown
(Hofmann et al., 1995), the reverse was true for the thio
homologue 5-acetyl-2,3-dihydro-1,4-thiazine identified
in a thermally treated cysteine/ribose mixture. In the
latter molecule, obviously the electron-withdrawing
sulfur atom stabilized predominantly the enamine
structure.
F igu r e 4. Mass spectra of ATHP (A) and [13C]3-ATHP (B).
MS 95 S (Finnigan, Bremen, Germany) in tandem with the
capillaries described above. Mass spectra in the electron
impact mode (MS/EI) were generated at 70 eV and in the
chemical ionization mode (MS/CI) at 115 eV with isobutane
as reactant gas.
RESULTS AND DISCUSSION
F or m a tion of ATHP . To gain a first insight into
the formation of ATHP from glucose and L-proline,
U-13C-labeled glucose was reacted with the amino acid.
On the basis of their molecular ions monitored by MS/
EI, the relative amounts of the isotopomers in ATHP
formed were determined. The main isomer (m/z 128;
Table 3) formed from [13C]6-glucose showed a clear
upshift of 3 mass units compared to the ATHP gener-
ated in the same experiment with unlabeled glucose
(m/z 125). This result indicates the presence of three
carbon-13 atoms in the labeled ATHP. The positions
of the labels were derived from a comparison of the mass
spectra of [13C]3-ATHP and ATHP (Figure 4). A shift
of 2 mass units in the fragment representing the acetyl
group (m/z 45 vs m/z 43) indicates the presence of two
carbon-13 labels (Figure 4B). In addition, the fragment
m/z 83 of the tetrahydropyridine ring (Figure 4B)
indicated one labeled carbon atom in the ring (m/z 83
vs m/z 82).
AMTHP showed the same popcorn-like odor at the
same low odor threshold (0.06 ng/L in air) as ATHP
(Schieberle, 1991).
As a final proof for the mechanism suggested in
Figure 5, the proposed intermediate HOP was synthe-

2274 J. Agric. Food Chem., Vol. 46, No. 6, 1998
Hofmann and Schieberle
F igu r e 5. Hypothetical pathway leading from 1-pyrroline and hydroxy-2-propanone to ATHP.
Ta ble 5. F or m a tion of ATHP fr om HOP a t Differ en t p H
Va lu esa
Ta ble 7. In flu en ce of th e Con cen tr a tion of
2-Oxop r op a n a l on th e F or m a tion of AP fr om 1-P yr r olin ea
ATHP
mol %
ATHP
mol %
AP
2-oxopropanal
expt
(µmol)
µg
%
pH
µg
pH
µg
1
2
3
4
50
319.1
58.5
3.7
28.7
5.3
0.3
3.0
5.0
2.1
14.3
1.7
11.4
7.0
9.0
29.3
43.7
23.4
35.0
10
10b
10c
a
The N-shielded HOP (1 µmol) was treated with trifluoroacetic
acid (5 µmol in 100 µL of dichloromethane) and, after addition of
phosphate buffer (5 mL, 0.5 mol/L), heated for 30 min at 100 °C
in a closed vessel.
1.1
0.1
a
1-Pyrroline (10 µmol, 690 µg) and 2-oxopropanal were reacted
for 30 min at 100 °C in phosphate buffer (5 mL, 0.5 mol/L; pH
b
7.0). 50 µmol of 1-pyrroline was reacted. c 1-Pyrroline (10 µmol)
and 2-oxopropanal (10 µmol) were mixed with silica gel (2.7 g
containing 300 µL of phosphate buffer (0.1 mol/L; pH 7.0) and
reacted for 5 min at 180 °C.
Ta ble 6. Ca r bon Isotop e Ra tio in AP Gen er a ted fr om
P r olin e a n d D-Glu cose (Glc) or [13C]6-D-Glu cose
([13C]6-Glc), Resp ectivelya
AP (m/z)
Glc
[
13C]6-Glc
AP (m/z)
Glc
[
13C]6-Glc
formed compared to that formed in aqueous conditions
(cf. expt 2 and 4; Table 7).
111
112
113
93.9
5.9
0.2
0.2
3.1
76.8
114
115
116
<0.1
<0.1
<0.1
19.1
0.7
<0.1
On the basis of the results of the labeling experiments
(Table 6) and assuming 1-pyrroline and 2-oxopropanal
as the intermediates in AP formation from glucose and
proline, one carbon atom has to be eliminated. We had
recently proposed a mechanism of this reaction (Schie-
berle, 1995) suggesting formaldehyde as the leaving
group. However, this was based on the assumption that
AP is formed during the reaction without any oxidation
step. In a recent publication (Hofmann and Schieberle,
1998) we could, however, show that 2-acetylpyrrolidine
is easily oxidized with high yields into AP. On the basis
of this result, we assume that the reaction sequence
leading from 1-pyrroline and 2-oxopropanal to AP goes
via 2-acetylpyrrolidine as the key intermediate.
a
L-Proline (2 mmol) and glucose (1 mmol) or [13C]6-glucose,
respectively, were mixed with silica gel (3 g) and reacted for 10
min at 160 °C.
sized (cf. Figure 3). To elucidate its effectiveness in
generating ATHP, aqueous solutions of HOP were boiled
for 30 min at different pH values and the ATHP formed
was quantified. The odorant was very effectively gener-
ated with increasing pH values increasing the amounts
of the ATHP (Table 5). At pH 9.0, 35 mol % was formed,
thereby confirming the reaction mechanism displayed
in Figure 5.
F or m a t ion of AP . In a previous investigation
(Schieberle, 1989), one of us had found that in AP
generated from proline and [13C]6-glucose, the main
isotopomer was the [13C]2-AP. Reinvestigation of the
isotope distribution established our previous data (Table
6) and indicated a main isotopomer having two labeled
carbon atoms (m/z 113 vs m/z 111). Further results had
established the reaction of 1-pyrroline with 2-oxopro-
panal as an effective system to generate AP (Schieberle,
1995). Therefore, in the present study, the experiments
on 1-pyrroline and 2-oxopropanal were extended.
Heating of 1-pyrroline and 2-oxopropanal gave high
yields of AP, especially when the concentrations of
2-oxopropanal much exceeded those of 1-pyrroline (expt
1; Table 7). However, changing the ratio of 2-oxopro-
panal to 1-pyrroline from 5:1 to 1:5 (cf. expt 1 and 3)
led to a drastic decrease in the formation of AP.
Furthermore, under dry-heating conditions and at
higher temperatures, a 50-fold lower amount of AP was
Because higher amounts of AP are formed under
aqueous conditions (Table 7), the hydrated 2-oxopro-
panal, which is easily formed in the presence of water
(Buettner et al., 1996), might be the reactive species.
Therefore, the reaction pathway might be suggested as
follows (Figure 6): hydrated 2-oxopropanal, as the
nucleophile, attacks 1-pyrroline at carbon-2. Due to its
N-analogous reductone structure, the 2-(1,2-dioxopro-
pyl)pyrrolidine formed is very susceptible to air oxida-
tion and will easily be oxidized into 2-(1,2-dioxopropyl)-
pyrroline. As found for triketones (Dao et al., 1974),
this N-analogous triketone will then be hydrated and
should undergo a rearrangement leading to 2-acetyl-2-
carboxypyrrolidine. This â-keto acid easily loses water
to yield 2-acetylpyrrolidine. The latter is in turn
oxidized into AP (Hofmann and Schieberle, 1998).
As suggested in Figure 7, finally the aldehyde group
of 2-oxopropanal is lost by oxidation into carbon dioxide.
However, this has to be verified by an experiment using

Intermediates in Roast Odorant Formation
J. Agric. Food Chem., Vol. 46, No. 6, 1998 2275
F igu r e 6. Formation of AP from 1-pyrroline and 2-oxopropanal hydrate.
F igu r e 7. Alternative reaction pathway leading from 1-pyrroline and 2-oxopropanal to AP.
[13C]1-2-oxopropanal and by monitoring the carbon-13
labeling in the carbon dioxide eliminated.
Ta ble 8. Am ou n ts of AP Gen er a ted fr om 1-P yr r olin e or
2-Meth yl-1-p yr r olin e, Resp ectively, in th e P r esen ce of
2-Oxop r op a n a la
Referring to Table 6, a minor amount of AP formed
from labeled glucose had incorporated three labeled
carbon atoms (m/z 114 vs m/z 111). Assuming also
1-pyrroline and 2-oxopropanal as the key intermediates
in the formation of [13C]3-AP, it has to be concluded that
one carbon of the 1-pyrroline (which stems from the
unlabeled proline) is lost in the course of the reaction.
To gain an insight into this alternative pathway,
2-methyl-1-pyrroline was reacted with 2-oxopropanal.
As indicated in Table 8, somewhat lower, but significant,
amounts of AP were formed from the homologous
2-methyl-1-pyrroline. The data establish that carbon-2
of both pyrrolines is lost during the reaction with
2-oxopropanal. In Figure 7, a reaction pathway is given
explaining these results. It is assumed that the tauto-
meric 2-pyrroline, as a nucleophile, attacks carbon-1 in
nonhydrated 2-oxopropanal. Hydration of the imine
reactant
1-pyrroline
2-methyl-1-pyrroline
AP (µg)
yield (%)
53.5
39.5
6.5
4.8
a
The reactants (10 µmol; 690 µg or 830 µg, respectively) were
dissolved in phosphate buffer (5 mL, 0.5 mol/L; pH 7.0) and reacted
for 30 min at 100 °C.
formed, followed by a ring opening and subsequent
hydrolysis of the formamide generated, yields 4-oxo-5-
hydroxyhexylamine. This intermediate will then cyclize
into 2-acetylpyrrolidine, the precursor of AP (Hofmann
and Schieberle, 1998).
Gen er a l Con sid er a tion s. During food processing,
the amino acid proline should react first with, for
example, a deoxyosone generated from a carbohydrate.
As shown in Figure 8, the Maillard reaction of the

2276 J. Agric. Food Chem., Vol. 46, No. 6, 1998
Hofmann and Schieberle
F igu r e 8. Generation of 1-pyrroline, glycerine aldehyde, 2-oxopropanal and hydroxy-2-propanone from proline and 1-deoxy-
glucosone.
Ta ble 9. In flu en ce of th e Con cen tr a tion of
1-deoxyglucosone with proline will yield 1-pyrroline,
hydroxy-2-propanone, and glycerinaldehyde as the first
reaction products. The latter aldehyde might subse-
quently lose water to yield 2-oxopropanal. Because the
2-Oxop r op a n a l on th e F or m a tion of AP a n d ATHP fr om
P r olin ea
2-oxopropanal
(mmol)
AP (µg)
ATHP (µg)
results presented here suggest 1-pyrroline as the key
intermediate in the formation of both AP and ATHP,
the relative concentrations of these odorants formed, for
example, during thermal processing of foods, should be
dependent on the relative amounts of the three carbo-
hydrate cleavage product present in a food. Reacting
the amino acid proline with increasing amounts of
2-oxopropanal resulted in the preferential formation of
AP, whereas ATHP was preferentially formed from
proline in the presence of lower amounts of 2-oxopro-
panal (Table 9). This result implies that a lower proline/
2-oxopropanal ratio will preferentially generate the
ATHP precursors 1-pyrroline and hydroxy-2-propanone
by Strecker degradation (Figure 1). However, at higher
ratios of proline/2-oxopropanal, the excess of 2-oxopro-
panal will immediately react with the 1-pyrroline
formed, yielding preferentially AP.
0.02
0.2
2.0
12.8
27.8
39.8
89.4
61.6
3.5
a
Proline (2 mmol) was reacted in the presence of increasing
concentrations of 2-oxopropanal for 30 min at 100 °C in phosphate
buffer (5 mL; pH 7.0; 0.5 mol/L).
ATHP is predominantly formed, as in popcorn (Schie-
berle, 1991), or AP is preferentially generated, as in
bread crust (Schieberle and Grosch, 1987), depends
significantly on the carbohydrate cleavage products
present. If high amounts of 2-oxopropanal are present,
AP will be formed, whereas in the presence of its
reduction product hydroxy-2-propanone, formation of
ATHP is favored. This is because in the presence of
high amounts of free amino acids, the Strecker reaction
of 2-oxopropanal will be favored, hydroxy-2-propanone
will predominate, and, consequently, generation of
ATHP should be favored.
Con clu sion s. The results suggest 1-pyrroline as the
key intermediate in the formation of both food odorants
AP and ATHP from the amino acid proline. Whether

Intermediates in Roast Odorant Formation
J. Agric. Food Chem., Vol. 46, No. 6, 1998 2277
LITERATURE CITED
Thermal Generation of Aromas; ACS Symposium Series 409;
American Chemical Society: Washington, DC, 1989; pp
268-275.
Schieberle, P. The role of free amino acids present in yeast as
precursors of the odorants 2-acetyl-1-pyrroline and
2-acetyltetrahydropyridine in wheat bread crust. Z. Leb-
ensm. Unters. Forsch. 1990, 191, 206-209.
Schieberle, P. Primary odorants of popcorn. J . Agric. Food
Chem. 1991, 39, 1141-1144.
Schieberle, P. Quantitation of important roast-smelling odor-
ants in popcorn by stable isotope dilution assays and model
studies on flavor formation during popping. J . Agric. Food
Chem. 1995, 43, 2442-2448.
Schieberle, P.; Grosch, W. Quantitative analysis of aroma
compounds in wheat and rye bread crusts using a stable
isotope dilution assay. J . Agric. Food Chem. 1987, 35, 252-
257.
Tressl, R.; Helak, B.; Martin, N. Formation of Flavor Com-
pounds from L-proline. In Topics in Flavour Research;
Berger, R. G., Nitz, S., Schreier, P., Eds.; Eichhorn: Marz-
ling-Hagenham, Germany, 1985; pp 139-159.
Tressl, R.; Helak, B.; Kersten, E.; Rewicki, D. Formation of
proline- and hydroxyproline-specific Maillard products from
[1-13C]glucose. J . Agric. Food Chem. 1993a , 41, 547-553.
Tressl, R.; Helak, B.; Kersten, E.; Nittka, C. Formation of
flavor compounds by Maillard reaction. In Recent Develop-
ments in Flavor and Fragrance Chemistry; Hopp, R., Mori,
K., Eds.; VCH: Weinheim, Germany, 1993b; pp 167-181.
Buettner, U.; Ochs, S.; Severin, Th. Formation of amino acids
by reaction of glucose and xylose with primary amines.
Carbohydr. Res. 1996, 291, 175-181.
Buttery, R. G.; Stern, D. J .; Ling, L. C. Studies on the flavor
volatiles of some sweet corn products. J . Agric. Food Chem.
1994, 42, 791-795.
Dao, H. L.; Dayer, F.; Duc, L.; Rode´-Gowal, H.; Dahn, H.
Alkaline cleavage and benzilic acid rearrangement of 1,2,3-
triketones (in German). Helv. Chim. Acta 1974, 57, 2215-
2223.
De Kimpe, N. G.; Dhoge, W. S.; Shi, Y.; Keppens, M. A.;
Boelens, M. M. On the Hodge mechanism of the formation
of the bread flavor component 6-acetyl-1,2,3,4-tetrahydro-
pyridine from proline and sugars. J . Agric. Food Chem.
1994, 42, 1739-1742.
Hodge, J . E.; Mills, F. D.; Fisher, B. E. Compounds derived
from browned flavors. Cereal Sci. Today 1972, 17, 34-40.
Hofmann, T.; Schieberle, P. New and convenient syntheses of
the important roasty, popcorn-like smelling food aroma
compounds 2-acetyl-1-pyrroline and 2-acetyltetrahydro-
pyridine from their corresponding cyclic R-amino acids. J .
Agric. Food Chem. 1998, 46, 616-619.
Hofmann, T.; Haessner, R.; Schieberle, P. Determination of
the chemical structure of the intense roasty popcorn-like
odorant 5-acetyl-2,3-dihydro-1,4-thiazine. J . Agric. Food
Chem. 1995, 43, 2195-2198.
Rewicki, D.; Tressl, R.; Ellerbeck, U.; Kersten, E.; Burgert, W.;
Gorzynski, M.; Hauck, R. S.; Helak, B. Formation and
synthesis of some Maillard generated aroma compounds. In
Progress in Flavor Precursor Studies; Schreier, P., Winter-
halter, P., Eds.; Allured Publishing: Carol Stream, IL, 1993;
pp 301-314.
Yoshikawa, K.; Libbey, L. M.; Cobb, W. J .; Day, E. A.
1-Pyrroline: the odor component of Strecker-degraded pro-
line and ornithine. Food Sci. 1965, 30, 991-994.
Rychlik, M.; Grosch, W. Identification and quantification of
potent odorants formed by toasting of wheat bread. Leb-
ensm. Wiss. Technol. 1996, 29, 515-525.
Schieberle, P. Formation of 2-acetyl-1-pyrroline and other
important flavor compounds in wheat bread crust. In
Received for review November 21, 1997. Revised manuscript
received March 18, 1998. Accepted March 23, 1998.
J F970990G