Formation of 3-hydroxy-4,5-dimethyl-2(5H)-furanone (sotolone) from 4-hydroxy-L-isoleucine and 3-amino-4,5-dimethyl-3,4-dihydro-2(5H)-furanone

Article abstract of DOI:10.1021/jf9506702

The proposed formation of 3-hydroxy-4,5-dimethyl-2(5H)-furanone (sotolone) from 4-hydroxy-L-isoleucine (1) and the corresponding lactone 3-amino-4,5-dimethyl-3,4-dihydro-2(5H)-furanone (2) by thermally induced oxidative deamination was corroborated. The formation of sotolone was studied in model systems by reacting 1 or 2 with different carbonyl compounds in a phosphate buffer at pH 5 at 100 °C for 1 h. The amount of sotolone was quantified by stable isotope dilution assays using ¹³C₂-labeled sotolone as internal standard and GC-MS operating in the selected ion monitoring mode. In general, a-ketoaldehydes were found to be more reactive than a-diketones. Methylglyoxal gave rise to about 64 μg sotolone per mg 1 (7.4 mol %) compared to less than 1μg (<0.1 mol %) when reacted with 2,3-pentanedione. Using 2 as the starting material, the yields were increased to 274 μg (35.9 mol %) and 5.4 μg (0.7 mol %), respectively. The optimum pH of the reaction with HIL was 5, representing the best compromise between the lactonization step and the amino-carbonyl reaction. Significant amounts of sotolone were generated only at temperatures higher than 70 °C. The yield increased over a period of 10 h to about 210 μg/mg 1 (23.8 mol %). The Strecker degradation of 1, resulting in 3-hydroxy-2-methylbutanal, was a competitive reaction to the formation of sotolone.

Full text of DOI:10.1021/jf9506702

J. Agric. Food Chem. 1996, 44, 1851
−1856
1851
F or m a tion of 3-Hyd r oxy-4,5-d im eth yl-2(5H)-fu r a n on e (Sotolon e)
fr om 4-Hyd r oxy-L-isoleu cin e a n d
3-Am in o-4,5-d im eth yl-3,4-d ih yd r o-2(5H)-fu r a n on e
Imre Blank,* J ianming Lin, Rene´ Fumeaux, Dieter H. Welti, and Laurent B. Fay
Nestec Ltd., Nestle´ Research Centre, Vers-chez-les-Blanc, P.O. Box 44, 1000-Lausanne 26, Switzerland
The proposed formation of 3-hydroxy-4,5-dimethyl-2(5H)-furanone (sotolone) from 4-hydroxy-L-
isoleucine (1) and the corresponding lactone 3-amino-4,5-dimethyl-3,4-dihydro-2(5H)-furanone (2)
by thermally induced oxidative deamination was corroborated. The formation of sotolone was studied
in model systems by reacting 1 or 2 with different carbonyl compounds in a phosphate buffer at pH
5 at 100 °C for 1 h. The amount of sotolone was quantified by stable isotope dilution assays using
13C2-labeled sotolone as internal standard and GC-MS operating in the selected ion monitoring mode.
In general, R-ketoaldehydes were found to be more reactive than R-diketones. Methylglyoxal gave
rise to about 64 µg sotolone per mg 1 (7.4 mol %) compared to less than 1 µg (<0.1 mol %) when
reacted with 2,3-pentanedione. Using 2 as the starting material, the yields were increased to 274
µg (35.9 mol %) and 5.4 µg (0.7 mol %), respectively. The optimum pH of the reaction with HIL was
5, representing the best compromise between the lactonization step and the amino-carbonyl reaction.
Significant amounts of sotolone were generated only at temperatures higher than 70 °C. The yield
increased over a period of 10 h to about 210 µg/mg 1 (23.8 mol %). The Strecker degradation of 1,
resulting in 3-hydroxy-2-methylbutanal, was a competitive reaction to the formation of sotolone.
Keyw or d s: 3-Hydroxy-4,5-dimethyl-2(5H)-furanone (sotolone); 4-hydroxy-L-isoleucine; oxidative
deamination; Strecker degradation; quantification by isotope dilution assay; model system studies;
synthesis
INTRODUCTION
authors proposed sotolone as a character-impact com-
pound of fenugreek (Trigonella foenum-graecum L.).
Girardon et al. (1986) confirmed the presence of sotolone
in fenugreek on the basis of MS data and synthesis.
Blank et al. (1993) reported 4-25 mg/kg of sotolone in
fenugreek seeds from different geographical origins.
Girardon et al. (1986) have pointed out the structural
similarity between sotolone and 4-hydroxy-L-isoleucine
(HIL) which is the most abundant free amino acid in
fenugreek seeds (Fowden et al., 1973; Sauvaire et al.,
1984). The authors postulated that this unusual amino
acid could be the precursor of sotolone in fenugreek. This
hypothesis has recently been supported by the fact that
only the 4S enantiomer of sotolone is present in
fenugreek (Sauvaire et al., 1993). This is in good
agreement with the stereochemistry of HIL isolated
from fenugreek which was shown to be 2S,3R,4S (Alcock
et al., 1989).
3-Hydroxy-4,5-dimethyl-2(5H)-furanone (sotolone) is
a powerful flavor compound found in several foods and
spices. It contributes significantly to the burnt/sweet
note of cane sugar (Tokitomo et al., 1980), aged sake
(Takahashi et al., 1976), and coffee (Blank et al., 1992),
to the spicy/curry note of fenugreek (Girardon et al.,
1986), lovage (Blank and Schieberle, 1993), and condi-
ments (Blank et al., 1993), as well as to the typical
nutty/sweet flavor of botrytized wines (Masuda et al.,
1984) and flor-sherry wines (Dubois et al., 1976; Martin
et al., 1992).
The flavoring potential of sotolone is due to its low
threshold value of 0.02 ng/L air (Blank et al., 1992). The
following threshold values were reported in water:
detection/nasal 0.3 µg/L (Blank et al., 1993); detection/
retronasal 0.01 µg/L (Tokitomo, 1980), 0.036 µg/L (Wild,
1988), 1-5 µg/L (Sulser et al., 1972); recognition/nasal
70 µg/L (Ro¨del and Hempel, 1974), 30 µg/L (Blank et
al., unpublished results).
In sugar manufacturing, the formation of sotolone was
explained via an amino-carbonyl reaction of pyruvate
and R-ketoglutarate, the latter originating from glu-
tamate (Kobayashi, 1989). In aged sake, sotolone was
produced by condensation of R-ketobutyrate and acetal-
dehyde, both being acid decomposition products of
threonine (Takahashi et al., 1976). The concentration
of acetaldehyde and sotolone was positively correlated
to the nutty/sweet flavor of flor-sherry wines (Martin
et al., 1992).
Blank et al. (1993) have recently formulated a path-
way for the formation of sotolone via thermally induced
oxidative deamination of HIL (1). As shown in the
simplified scheme, acid-catalyzed cyclization of HIL
leads to the corresponding lactone (2) which reacts with
an R-dicarbonyl. Rearrangement of the resulting Schiff
The presence of sotolone in fenugreek oleoresin was
first reported by Rijkens and Boelens (1975). The
base and subsequent hydrolysis gives rise to sotolone.
The aim of this study was to verify this hypothesis and
to study the parameters influencing the formation of
sotolone from 1 and 2.
* Author to whom correspondence should be ad-
dressed. Telephone: +21/785-8607. Fax: +21/785-
8554. E-mail: imre.blank@chlsnr.nestrd.ch.
S0021-8561(95)00670-4 CCC: $12.00
© 1996 American Chemical Society
+
+

1852 J. Agric. Food Chem., Vol. 44, No. 7, 1996
Blank et al.
EXPERIMENTAL PROCEDURES
1. Ma ter ia ls a n d Rea gen ts. The following chemicals and
materials were obtained commercially: sotolone (Aldrich, Neu-
Ulm, Germany), diethyl 2-methyl-3-oxobutanedioate, L-isoleu-
cine, Dowex 50WX8 (20-50 mesh, Na+ form), methylglyoxal
(40% in water), phenylglyoxal, propionaldehyde, phenylacetal-
dehyde, 2,3-butanedione, and 2,3-pentanedione (Fluka, Buchs,
Switzerland); [1,2-13C]-acetaldehyde (99% purity, Cambridge
Isotope Lab., Andover, MD); TLC plate silica gel 60 F254
(Merck, Darmstadt, Germany). The rotation perforator was
from Normag (Weinheim, Germany). The deuterated solvents
(D2O, DMSO-d2, CD3OD) were from Dr. Glaser AG (Basel,
Switzerland). Other solvents and chemicals were of analytical
grade from Merck.
F igu r e 1. Electron impact (EI) mass spectrum of [5,6-13C]-
3-hydroxy-4,5-dimethyl-2(5H)-furanone (13C2-sotolone). The
MS-EI spectrum of the corresponding unlabeled sotolone is m/z
(% relative abundance) 83 (100), 55 (95), 128 (60, M+), 43 (55),
57 (45), 29 (40), 27 (30), 39 (25), 72 (25), 85 (25).
2. Syn th esis. 3-Amino-4,5-dimethyl-3,4-dihydro-2(5H)-
furanone hydrochloride (2‚HCl). Compound 2‚HCl was syn-
thesized by photochemical chlorination of L-isoleucine in
concentrated HCl and subsequent hydrolysis according to
Faulstich et al. (1973) and Hasan (1986). A solution of
L-isoleucine (21 g) in concentrated HCl (400 mL) was cooled
to 0 °C. Chlorine gas was then bubbled through the solution
for 6 h while it was irradiated in a photochemical reactor using
a high-pressure Hg vapor lamp (380 nm). The reaction
mixture was heated under reflux for 4 h and then concentrated
to dryness. Water (50 mL) was added to the residue, and the
pH was adjusted to 8.7 with Na2CO3 (28 g) at 5 °C. After
dilution with water (200 mL), the sample was continuously
extracted with Et2O (100 mL) for 24 h using a rotation
perforator. The organic phase was dried over anhydrous
Na2SO4. Dry HCl gas was passed into the solution until
complete precipitation was achieved. The precipitate was
filtered and dried over KOH. The brown product (4.8 g) was
dissolved in methanol and precipitated with Et2O to give 3.1
g of the compound 2‚HCl as a diastereomeric mixture of the
isomers (3S,4R,5R) and (3S,4R,5S) (overall yield 12%).
Elemental analysis: Found C (43.25), H (7.53), N (8.43), Cl
(21.66); C6H12NO2Cl requires C (43.51), H (7.30), N (8.46), Cl
(21.41).
1H-NMR (D2O/internal TSP). Two subspectra representing
two isomers in a mole ratio of about 59:41 were again obtained,
the minor of which was nearly identical with the data
published for the (2S,3R,4S) isomer (Alcock et al., 1989): δ
3.908 (1H, d, J ) 4.4 Hz, H-2), 3.865 (1H, d q, J ) 8.0, 6.4 Hz,
H-4), 1.933 (1H, m, 16 lines, H-3), 1.256 (3H, d, J ) 6.3 Hz,
CH3-5), 0.970 (3H, d, J ) 7.1 Hz, CH3-6). The major compo-
nent was therefore attributed to the (2S,3R,4R) isomer:
δ
4.070 (1H, q d, J ) 6.4, 2.8 Hz, H-4), 3.825 (1H, d, J ) 4.0 Hz,
H-2), 2.144 (1H, m, 16 lines, H-3), 1.220 (3H, d, J ) 6.4 Hz,
CH3-5), 1.074 (3H, d, J ) 7.3 Hz, CH3-6).
[5,6-13C]-3-Hydroxy-4,5-dimethyl-2(5H)-furanone (13C2-So-
tolone). 13C2-Sotolone was synthesized by condensation of
diethyl 2-methyl-3-oxobutanedioate and [1,2-13C]-acetaldehyde
followed by lactonization and subsequent decarboxylation
under strongly acidic conditions as described for the corre-
sponding unlabeled compound (Ro¨del and Hempel, 1974). The
reaction was carried out on a micro-scale using a Sovirel test
tube (20 mL) which was modified with a double jacket to cool
the upper part. Diethyl 2-methyl-3-oxobutanedioate (22 mmol)
was placed in the tube, to which labeled acetaldehyde (21.7
mmol) and dry pyridine (40 mmol) were added at 0 °C. The
mixture was heated in an oil bath at 120 °C for 2 h, while the
upper part of the tube was kept at +5 °C using a cryostat.
After adding HCl (37%, 7 mL), acetic acid (99%, 6 mL), and
water (9 mL), the solution was refluxed for 4 h. The solution
was diluted with water (300 mL), and the pH was adjusted to
8 with NaOH (10 mol/L). The target compound was extracted
with Et2O (100 mL) overnight using a rotation perforator. The
solvent was removed by distillation, and the residue was
dissolved in methanol (100 mL). The amount of labeled
sotolone was determined by GC-MS measuring a 1:1 mixture
of the labeled and unlabeled sotolone used as internal standard
resulting in 323 mg of labeled sotolone (12% yield). The MS
spectrum of the compound shown in Figure 1 indicates the
incorporation of two 13C atoms. The position of the 13C atoms
was verified by NMR by comparison with the proton and
carbon spectra of unlabeled sotolone.
MS-EI, m/z (rel int): 129 (9, M+), 85 (29), 70 (100), 57 (31),
56 (29).
1H-NMR (DMSO-d6, internal TMS): Apart from a D2O
exchangeable slightly broadened signal at 9.01 ppm represent-
ing about 3 (total) protons, subspectra of two isomers in an
approximate mole ratio of 59:41 were obtained. Major compo-
nent: δ 4.766 (1H, q d, J ) 6.3, 4.6 Hz, H-5), 4.591 (1H, d, J
) 7.4 Hz, H-3), 2.800 (1H, m, g10 lines, H-4), 1.271 (3H, d, J
) 6.5 Hz, CH3-6), 0.866 (3H, d, J ) 7.3 Hz, CH3-7). Minor
component: δ 4.524 (1H, d, J ) 8.3 Hz, H-3), 4.491 (1H, q d,
J ) 6.5, 3.0 Hz, H-5), ca. 2.523 (1H, m, mostly under solvent
signal, H-4), 1.367 (3H, d, J ) 6.5 Hz, CH3-6), 1.076 (3H, d, J
) 7.2 Hz, CH3-7). When the unavoidable milieu-induced shift
effects are taken into account, these data are in good agree-
ment with those of Hasan et al. (1976) and Hasan (1986) for
the (3S,4R,5R) and the (3S,4R,5S) isomers, respectively.
4-Hydroxy-L-isoleucine (HIL, 1). HIL (1) was synthesized
by the base-catalyzed opening of the lactone function of 2‚
HCl on a cation exchange resin (Hasan et al., 1976; Alcock et
al., 1989). The column (24 × 200 mm) was packed with Dowex
50WX8 (150 g), converted to its NH4+ form with NH4Cl (2 mol/
L), and washed with water to remove the chloride ions.
Compound 2‚HCl (1.5 g) dissolved in water (30 mL) was added
layered to the column. Chloride ions were removed by washing
with water (500 mL). Compound 1 was eluted with ammonia
(0.5 mol/L) to obtain six fractions each of 500 mL. The
presence of the amino acid was checked by TLC using butanol/
acetic acid/water in the ratio 4/1/1 (v/v/v) as the mobile phase
and detected with ninhydrin. The ninhydrin-positive fractions
were concentrated at 37 °C and lyophilized to obtain 1.2 g of
a colored product. Recrystallization from ethanol (90%, 50 mL)
and hot filtration gave 700 mg of a white powder (4.8 mmol,
53% yield).
1H-NMR (CD3OD, internal TMS; moderate resolution en-
hancement after zero filling was required to extract the
coupling constants (0.089 Hz/point resolution), the italicized
data are due to coupling with the two 13C atoms); δ 4.835 (d q
d q, J ) 152.1, 6.6, 3.2, 1.2 Hz, H-5), 1.870 (d d d, J ) 4.3, 1.2,
0.7 Hz, CH3-7), 1.362 (d d d, J ) 128.5, 6.6, 4.6 Hz, CH3-6).
13C-NMR (CD3OD, internal TMS, proton decoupled, spectral
width 20 000 Hz, FID 64 K data points, zero filling, resolution
0.305 Hz/data point); dominating signals from the 13C sub-
stituted sites: δ 78.94 (d, J ) 38.7 Hz, C-5), 18.98 (d, J ) 38.7
Hz, C-6). Small signals from unsubstituted sites: 171.81 (d,
J ) 1.8 Hz, C-2), 138.87 (d d, J ) 4.7, 1.5 Hz, C-3), 134.37 (d
d, J ) 41.6, 1.3 Hz, C-4), 9.08 (d d, J ) 3.7, 0.6 Hz, C-7); all
shifts were within 0.1 ppm of those of natural abundance
sotolone (D. H. Welti and F. Arce Vera, unpublished data).
3. In str u m en ta l An a lysis. Preparative Gas Chromatog-
raphy. Labeled sotolone was purified for NMR analysis by
preparative GC using a MCS Gerstel gas chromatograph
Elemental analysis: Found C (48.74), H (9.00), N (9.47);
C6H13NO3 requires C (48.97), H (8.90), N (9.52). MS-EI, m/z
(rel int): 132 (2), 129 (10), 102 (63), 74 (65), 58 (100); MS-PCI
(NH3) m/z (rel int): 147 (100, M+), 148 (49), 165 (6).
+
+

Formation of 3-Hydroxy-4,5-dimethyl-2(5H)furanone
J. Agric. Food Chem., Vol. 44, No. 7, 1996 1853
Ta ble 1. F or m a tion of Sotolon e fr om
4-Hyd r oxy-L-isoleu cin e (HIL, 1)a
carbonyl
sotoloneb (µg/mg HIL)
yield (mol %)
2,3-butanedione
2,3-pentanedione
methylglyoxalc
phenylglyoxal
propionaldehyde
phenylacetaldehyde
0.34 ( 0.03
0.30 ( 0.03
64.2 ( 0.3
<0.1
<0.1
7.4
2.5
0.1
22.2 ( 0.3
1.00 ( 0.06
0.24 ( 0.03
<0.1
a
HIL control experiment (without carbonyl) yielded 0.04 µg
b
sotolone/mg HIL (<0.01 mol %). Data are the means of at least
two experiments, each of them injected twice. c Methylglyoxal
control experiment (reaction without HIL) yielded 0.07 µg of
sotolone.
Ta ble 2. F or m a tion of Sotolon e fr om
3-Am in o-4,5-d im eth yl-2-oxotetr a h yd r ofu r a n (2)a
R-dicarbonyl
sotolone (µg/mg 2)
yield (mol %)
F igu r e 2. Quantification of sotolone by stable isotope dilution
assays using 13C2-sotolone as internal standard. The mass
chromatograms were recorded by GC-MS operating in the
selected ion monitoring (SIM) mode to measure the molecular
ions m/z 128 of sotolone and m/z 130 of the doubly-labeled
sotolone standard.
methylglyoxal
2,3-pentanedione
274.4 ( 3.4
5.4 ( 0.3
35.9
0.7
a
Control experiment (reaction without carbonyl) yielded 0.03
mol % sotolone.
the procedure described by Guth and Grosch (1990). A good
linearity was found in the concentration range 3-150 µg/mL
(r2 ) 0.999). Samples for establishing the calibration curve
and for quantifying sotolone were injected twice.
(Mu¨lheim, Germany). Two DB-Wax widebore capillaries were
employed (J &W Scientific), both 15 m × 0.53 mm with a film
thickness of 1 µm. Helium was used as carrier gas (13 psi).
The oven was programmed as follows: 60 °C (1 min), 10 °C/
min up to 200 °C (3 min). Nine injections resulted in about 3
mg of sotolone trapped in a glass tube at 50 °C.
Nuclear Magnetic Resonance (NMR) Spectroscopy. All NMR
spectra were acquired on a Bruker AM-360 spectrometer
equipped with a quadrinuclear 5 mm probe head (QNP), at
360.13 MHz for 1H and at 90.56 MHz for 13C under standard
conditions. The probe temperature was 21 °C for the proton
spectra and slightly higher for the carbon spectra due to
heteronuclear composite pulse decoupling. The frequency
resolution for the proton spectra was 0.231 Hz/point. The
shifts are cited in parts per million from the respective internal
standards.
4. P r ep a r a tion of Sa m p les for Qu a n tita tive An a lysis.
HIL (1, 2-10 mg) and 2‚HCl (2-10 mg) were dissolved in a
phosphate buffer (2-10 mL, 0.1 mol/L, pH 5.0). The carbonyl
reactant was added so that the molar ratio of precursor to
carbonyl was 1:10. The solution was boiled for 1 h in a glass
tube closed with a screw cap. After rapidly cooling down the
reaction mixture, water (50 mL) and then the internal
standard (13C2-sotolone, 25.8-64.5 µg) were added. The
sample was saturated with NaCl, and the pH was adjusted to
4 with HCl (1 mol/L). Sotolone was continuously extracted
with Et2O for 8 h using a rotation perforator. The extract was
dried over Na2SO4 and concentrated to about 1 mL on a
Vigreux column (1 m × 1 cm). All experiments were performed
in duplicate.
Mass Spectrometry (MS). a. Qualitative Analysis. Electron
impact (EI) and positive chemical ionization (PCI) mass
spectra of the synthesized compounds 1 and 2 were obtained
at a resolution of 1000 on a Finnigan MAT 8430 mass
spectrometer. The samples were directly introduced into the
ion source heated at 180 °C. The acquired mass range was
20-500 Da for both EI and PCI experiments. Ammonia was
used as reagent gas for chemical ionization. Labeled sotolone
and 3-hydroxy-2-methylbutanal (9) were introduced via a
Hewlett-Packard HP-5890 gas chromatograph using the cold
on-column injection technique. The samples were analyzed
on the FFAP capillary column (J &W Scientific, 30 m × 0.25
mm, film thickness 0.25 µm) using helium as carrier gas (10
psi). Temperature program: 50 °C (2 min), 4 °C/min to 180
°C, 10 °C/min to 240 °C (10 min).
b. Quantitative Analysis. Sotolone was quantified by stable
isotope dilution assays using 13C2-sotolone as internal standard
(Blank et al., 1993). Quantification experiments were per-
formed with a Hewlett-Packard HP-5971 GC-MS equipped
with a DB-Wax capillary column (J &W Scientific): 30 m ×
0.25 mm, film thickness 0.25 µm. Helium was used as carrier
gas (10 psi). The sample (2 µL) was introduced via splitless
injection at 250 °C. Temperature program: 20 °C (0.5 min),
30 °C/min to 100 °C, 4 °C/min to 145 °C, 70 °C/min to 220 °C
(10 min). The compounds were recorded by electron impact
ionization at 70 eV. The temperature of the ion source was
190 °C. Sotolone and the labeled internal standard (13C2-
sotolone) were detected by selected ion monitoring (SIM) of
their molecular ions m/z 128 and 130, respectively. The traces
of the ions were recorded as shown in Figure 2, and the
concentration of sotolone was calculated from the peak areas
using a calibration factor of 1.1. The calibration curve was
established with standard mixtures containing defined amounts
of labeled and unlabeled sotolone in different ratios following
RESULTS AND DISCUSSION
1. Ver ifica tion of th e Hyp oth esis. 4-Hydroxy-L-
isoleucine (HIL, 1) as Precursor. The formation of
sotolone from HIL was studied by reacting HIL with
different mono- and R-dicarbonyl compounds in a phos-
phate-buffered model system (0.1 mol/L, pH 5.0) at 100
°C for 1 h. The results summarized in Table 1 show
that both 2,3-butanedione and 2,3-pentanedione formed
only low amounts of sotolone. Higher yields were
achieved when using methylglyoxal (7.4 mol %) and
phenylglyoxal (2.5 mol %) as reactant, producing about
70-200 times more sotolone than the corresponding
reaction with 2,3-butanedione. The yields obtained with
propionaldehyde and phenylacetaldehyde were low (e0.1
mol %), i.e. about 60-90 times less compared to the
corresponding R-ketoaldehydes.
3-Amino-4,5-dimethyl-3,4-dihydro-2(5H)-furanone (2)
as Precursor. The efficiency of the lactone of HIL was
studied by reacting 2‚HCl with R-dicarbonyls. As
shown in Table 2, significantly higher amounts of
sotolone were generated from 2‚HCl as compared to 1.
Using methylglyoxal, the yields were increased by a
factor of about 5 to 35.9 mol %. The lactone 2 was found
to be a better precursor than the amino acid.
Formation of Sotolone (Figure 3). The data reported
above confirm the hypothesis of the formation of so-
tolone by thermally induced oxidative deamination of
HIL (1) (Blank et al., 1993). Acid-catalyzed cyclization
of HIL leads to the corresponding lactone (2) which
+
+

1854 J. Agric. Food Chem., Vol. 44, No. 7, 1996
Blank et al.
F igu r e 4. Mass spectra of 3-hydroxy-2-methylbutanal (9)
eluted as the second peak of a diastereomeric mixture on an
FFAP (retention index: 1560): (A) electron impact spectrum
and (B) spectrum by chemical ionization using ammonia as
reagent gas.
Ta ble 3. F or m a tion of Sotolon e fr om HIL (1) a n d th e
Cor r esp on d in g La cton e 2 As Affected by th e p Ha
from HIL (1) yield
from 2
yield
R-dicarbonyl pH
(µg/mg)
(mol %)
(µg/mg)
(mol %)
F igu r e 3. Formation of sotolone (6) and 3-hydroxy-2-meth-
ylbutanal (9) from 4-hydroxy-L-isoleucine (1) and 3-amino-4,5-
dimethyl-3,4-dihydro-2(5H)-furanone (2) using methylglyoxal
(3) as the carbonyl reactant (more details in the text).
methylglyoxal
methylglyoxal
methylglyoxal
methylglyoxal
3
5
6
7
42.8 ( 3.2
64.2 ( 0.3
16.5 ( 1.3
2.3 ( 0.3
5.0
7.4b
1.9
0.2
178.2 ( 8.5
274.4 ( 3.4
311.2 ( 12.5
48.6 ( 2.3
23.0
35.9
40.2
6.3
a
reacts with an R-dicarbonyl (3) to form the Schiff base
4. Rearrangement of 4 and subsequent hydrolysis of 5
give rise to sotolone (6). The data show that R-dicar-
bonyls are capable of generating sotolone from both HIL
(1) and the lactone 2. However, the latter is more
efficient in producing sotolone.
The lactonization step (1 f 2) is apparently an
important parameter of the reaction and is favored
under acidic conditions. The lactonization yields were
about 50% at pH 3 and 10% at pH 5 (Blank et al.,
unpublished results, 1996). Furthermore, R-ketoalde-
hydes (3) are much more efficient than R-diketones,
indicating that the reactivity of the carbonyl compounds
is another crucial parameter, particularly for the forma-
tion of the Schiff base 4. This is in agreement with the
data obtained with R-keto acids and ascorbic acid which
generate only low amounts of sotolone from HIL (Blank
et al., 1995).
The fact that Strecker-inactive carbonyl compounds,
such as propionaldehyde and phenylacetaldehyde, are
also able to transform HIL into sotolone indicates an
alternative formation pathway. Aldol condensation
products, e.g. 2,4-dimethyl-2,4-heptadienal, may func-
tion as active components in the transamination of HIL.
Similar reactions, i.e. reductive amination of conjugated
carbonyls, have been reported by Rizzi (1976). Hence,
condensation products of methylglyoxal may also be
involved in the formation of sotolone.
Data are the means of at leas two experiments, each of them
b
injected twice. The reaction performed at pH 5 without using
phosphate resulted in 2.8 mol % sotolone.
in the presence of an active R-dicarbonyl, i.e. methyl-
glyoxal (3). As shown in Figure 3, the amino-carbonyl
reaction of 1 and 3 results in the Schiff base 7 which
may cyclize to 4 or decompose to 8 via decarboxylation.
The Strecker aldehyde of HIL is released from 8 by
hydrolysis, i.e. 3-hydroxy-2-methylbutanal (9). Indeed,
this compound was tentatively identified by GC-MS
(Figure 4) as a diastereomeric mixture. Both isomers
showed nearly identical EI and CI spectra, respectively.
In the sample based on HIL and methylglyoxal, the
ratio of formed Strecker aldehyde (9) to sotolone was
about 1:2 (Blank et al., unpublished results, 1996).
Hence, the Strecker degradation of HIL is a competitive
reaction to the formation of sotolone. In contrast, only
traces of 9 were detected in the sample containing the
lactone 2, i.e., about 50 times less than in the reaction
with HIL. The formation of sotolone from 2 is the
favored reaction, most likely due to the blocked carboxyl
group.
2. P a r a m eter s Affectin g th e F or m a tion of So-
tolon e fr om HIL (1) a n d th e Cor r esp on d in g La c-
ton e (2). Influence of pH (Table 3). It is well-known
that both lactonization and the formation of the Schiff
base are strongly dependent on the pH of the reaction
medium. To find the optimum pH, methylglyoxal was
reacted with HIL and the lactone 2‚HCl, respectively.
The lower yields obtained with HIL can be explained
by a partial Strecker degradation of the amino acid (1)
+
+

Formation of 3-Hydroxy-4,5-dimethyl-2(5H)furanone
J. Agric. Food Chem., Vol. 44, No. 7, 1996 1855
CONCLUSIONS
The proposed formation of sotolone from 4-hydroxy-
L-isoleucine (HIL, 1) via thermally induced oxidative
deamination was substantiated. The lactone of HIL,
3-amino-4,5-dimethyl-3,4-dihydro-2(5H)-furanone (2),
was found to be a better precursor compared to the
amino acid. R-Ketoaldehydes were more effective in
generating sotolone from both HIL and the correspond-
ing lactone 2 than R-diketones. The reactivity of the
dicarbonyl and the lactonization step are important
parameters, particularly for the formation of the Schiff
base. The transformation yields from HIL into sotolone
greatly depend on the reaction conditions, such as
temperature, time, pH, and the amount of the dicarbo-
nyl. Best results were obtained by boiling methylgly-
oxal and HIL for 10 h at pH 5 (about 24 mol %).
Alternatively, due to the inhibition of the Strecker
degradation of the amino acid by the lactone form, even
better results (about 40 mol %) were achieved in a
shorter time (1 h) at pH 6 using 2 as precursor.
F igu r e 5. Formation of sotolone from 4-hydroxy-L-isoleucine
(HIL, 1) as a function of the reaction temperature (reaction
with methylglyoxal in a phosphate buffer of pH 5.0 for 1 h).
ACKNOWLEDGMENT
We are grateful to Ms. S. Devaud, Mrs. S. Metairon,
Mrs. F. Arce Vera, and Mr. Y. Krebs for expert technical
assistance. We also thank Dr. G. Philippossian for
helpful suggestions and Dr. E. Prior for critically
reading the manuscript.
F igu r e 6. Formation of sotolone from 4-hydroxy-L-isoleucine
(HIL, 1) depending on the reaction time (reaction with meth-
ylglyoxal in a phosphate buffer of pH 5.0 at 100 °C).
Ta ble 4. F or m a tion of Sotolon e fr om HIL As Affected by
th e Con cen tr a tion of Meth ylglyoxa l
LITERATURE CITED
Alcock, N. W.; Crout, D. H.; Gregorio, M. V. M.; Lee, E.; Pike,
G.; Samuel, C. J . Stereochemistry of the 4-hydroxyisoleucine
from Trigonella foenum-graecum. Phytochemistry 1989, 28,
1835-1841.
Blank, I.; Schieberle, P. Analysis of the seasoning-like flavour
substances of a commercial lovage extract (Levisticum
officinale Koch.). Flavour Fragrance J . 1993, 8, 191-195.
Blank, I.; Sen, A.; Grosch, W. Potent odorants of the roasted
powder and brew of Arabica coffee. Z. Lebensm. Unters.
Forsch. 1992, 195, 239-245.
molar ratio
HIL/methylglyoxal
sotolone
(µg/mg HIL)
yield
(mol %)
1:1
1:2
1:5
1:10
1:20
24.0 ( 0.3
28.8 ( 0.9
41.3 ( 2.1
64.2 ( 0.3
62.9 ( 1.1
2.7
3.3
4.7
7.4
7.2
The reaction of methylglyoxal and HIL was favored at
pH 5 which apparently is the best compromise between
the lactonization step and the reactivity of the amino
group to form the Schiff base (Figure 3). Once the
lactone 2 is formed, the amino-carbonyl reaction was
favored at pH 5-6. In general, the yields obtained with
the lactone 2 were significantly higher compared to the
amino acid, particularly at pH 6 (40.2 mol %). The
reaction of HIL and methylglyoxal at pH 5 performed
in water yielded 2.8 mol % sotolone compared to 7.4 mol
% when using the phosphate-buffered system. This
suggests a catalytic effect of phosphate on the formation
of sotolone from HIL.
Blank, I.; Schieberle, P.; Grosch, W. Quantification of the
flavour compounds 3-hydroxy-4,5-dimethyl-2(5H)-furanone
and 5-ethyl-3-hydroxy-4-methyl-2(5H)-furanone by a stable
isotope dilution assay. In Progress in Flavour and Precursor
Studies; Schreier, P., Winterhalter, P., Eds.; Allur. Publ.:
Wheaton, 1993; pp 103-109.
Blank, I.; Lin, J .; Fay, L. B.; Fumeaux, R. Formation of
3-hydroxy-4,5-dimethyl-2(5H)-furanone (sotolon) from 4-hy-
droxy-L-isoleucine. In Bioflavour 95. Analysis - Precursor
Studies - Biotechnology; EÄ tie´vant, P.; Schreier, P., Eds.; Les
Colloques no. 75, INRA: Paris, 1995; pp 385-388.
Dubois, P.; Rigaud, J .; Dekimpe, J . Identification of 4,5-
dimethyltetrahydrofuran-2,3-dione in a Flor sherry wine.
Lebensm. Wiss. Technol. 1976, 9, 366-368 (in French;
Chem. Abstr. 1977, 86, 137884m).
Influence of Reaction Temperature and Time. The
sotolone yields from HIL and methylglyoxal were strongly
dependent on both the reaction time and temperature.
Significant amounts were generated only above 70 °C
(Figure 5). At a constant temperature of 100 °C, the
yield of sotolone continuously increased over a period
of 10 h and flattened off thereafter (Figure 6). About
23-210 µg sotolone was generated from 30 min to 10 h
which corresponds to 2.7 and 23.8 mol %, respectively.
Amount of R-Dicarbonyl (Table 4). The efficiency of
transforming HIL into sotolone can be enhanced by
using higher amounts of the carbonyl reactant. In-
creasing the concentration of methylglyoxal 5-fold al-
most doubled the amount of sotolone produced. How-
ever, the concentration of sotolone did not increase
linearly with the amount of the R-dicarbonyl.
Faulstich, H.; Do¨lling, J .; Michl, K.; Wieland, T. Synthesis of
R-amino-γ-hydroxy acids by photochlorination. Liebigs Ann.
Chem. 1973, 560-565 (in German; Chem. Abstr. 1973, 79,
66769v).
Fowden, L.; Pratt, H. M.; Smith, A. 4-Hydroxyisoleucine from
seeds of Trigonella foenum-graecum. Phytochemistry 1973,
12, 1701-1711.
Girardon, P.; Sauvaire, Y.; Baccou, J .-C.; Bessiere, J .-M.
Identification of 3-hydroxy-4,5-dimethyl-2(5H)-furanone in
aroma of fenugreek seeds (Trigonella foenum-graecum L.).
Lebensm. Wiss. Technol. 1986, 19, 44-46 (in French; Chem.
Abstr. 1986, 105, 132459c).
Guth, H.; Grosch, W. Deterioration of soya-bean oil: Quanti-
fication of primary flavour compounds using a stable isotope
dilution assay. Lebensm. Wiss. Technol. 1990, 23, 513-522.
Hasan, M. The diastereomeric γ-hydroxy-isoleucines. In New
Trends in Natural Products Chemistry; Rahman, A., Le
+
+

1856 J. Agric. Food Chem., Vol. 44, No. 7, 1996
Blank et al.
Quesne, P. W., Eds.; Studies in Organic Chemistry, Vol. 26;
Elsevier Sci. Publ.: Amsterdam, The Netherlands, 1986; pp
123-141.
Sauvaire, Y.; Girardon, P.; Baccou, J . C.; Risterucci, A. M.
Changes in growth, proteins, and free amino acids of
developing seeds and pod of fenugreek. Phytochemistry
1984, 23, 479-486.
Sauvaire, Y.; Brenac, P.; Guichard, E.; Fournier, N. Relation
entre la composition en acides amine´s libres des graines de
fenugrec et la qualite´ aromatique. In Fourth International
Workshop on Seeds; Coˆme, D., Corbineau, F., Eds., Volume
1, ASFIS: Paris, France, 1993; pp 201-206.
Sulser, H.; Habegger, M.; Bu¨chi, W. Synthesis and sensory
evaluation of 3,4-disubstituted 2-hydroxy-2-buten-1,4-olides.
Z. Lebensm. Unters. Forsch. 1972, 148, 215-221 (in Ger-
man; Chem. Abstr. 1972, 77, 48112e).
Takahashi, K.; Tadenuma, M.; Sato, S. 3-Hydroxy-4,5-di-
methyl-2(5H)-furanone, a burnt flavoring compound from
aged sake. Agric. Biol. Chem. 1976, 40, 325-330.
Hasan, M. Georgopoulos, D.; Wieland, T. The four diastereo-
meric lactones of γ-hydroxyisoleucine. Liebigs Ann. Chem.
1976, 781-787 (in German; Chem. Abstr. 1976, 85, 63319m).
Kobayashi, A. Sotolon. Identification, formation and effect on
flavor. In Flavor Chemistry Trends and Developments;
Teranishi, R., Buttery, R. G., Shahidi, F., Eds.; ACS
Symposium Series 388; American Chemical Society: Wash-
ington, DC, 1989; pp 49-59.
Martin, B.; Etievant, P. X.; Le Quere, J . L.; Schlich, P. More
clues about sensory impact of sotolon in some flor sherry
wines. J . Agric. Food Chem. 1992, 40, 475-478.
Masuda, M.; Okawa, E.; Nishimura, K.; Yunome, H. Identi-
fication of 4,5-dimethyl-3-hydroxy-2(5H)-furanone (Sotolon)
and ethyl 9-hydroxynonanoate in botrytised wine and evalu-
ation of the roles of compounds characteristic. Agric. Biol.
Chem. 1984, 48, 2707-2710.
Rijkens, F.; Boelens, H. The future of aroma research. In Proc.
Int. Symp.Aroma Research; Maarse, H., Groenen, P. J ., Eds.;
Pudoc: Wageningen, The Netherlands, 1975; pp 203-220.
Rizzi, G. P. Non-enzymatic transamination of unsaturated
carbonyls: A general source of nitrogenous flavor com-
pounds in foods. In Phenolic, Sulfur, and Nitrogen Com-
pounds in Food Flavors; Charalambous, G., Katz, I., Eds.;
ACS Symposium Series 26; American Chemical Society:
Washington, DC, 1976; pp 122-132.
Tokitomo, Y.; Kobayashi, A.; Yamanishi, T.; Muraki, S. Studies
on the “sugary flavor” of raw cane sugar. III Key compound
of the sugary flavor. Proc. J pn. Acad. 1980, 56B, 457-462.
Wild, H. Modern synthetic routes towards important flavour
raw materials. Chem. Ind. 1988, 580-586.
Received for review October 12, 1995. Revised manuscript
received May 2, 1996. Accepted May 3, 1996.X
J F9506702
Ro¨del, W.; Hempel, U. Synthesis and sensory characteristics
of alkylsubstituted 2-hydroxybutenolides and of some com-
pounds of similar structure. Nahrung 1974, 18, 133-141
(in German; Chem. Abstr. 1974, 81, 118804r).
X Abstract published in Advance ACS Abstracts, J une
15, 1996.
+
+