Generation of (E)-1-(2,3,6-trimethylphenyl)buta-1,3-diene from C 13-norisoprenoid precursors

Article abstract of DOI:10.1021/jf051039w

Three C₁₃-norisoprenoid compounds, 3,6,9-trihydroxymegastigma-4, 7-diene (6), 3,4,9-trihydroxymegastigma-5,7-diene (4), and the actinidols (8), have all been synthesized and subjected to acid hydrolysis. All three were shown to generate (E)-1-(2,3,6-trimethylphenyl)buta-1,3-diene (1) under wine conservation conditions. At 45°C, approximately 4000-5000 ng/L of 1 was formed from 1.0 mg/L of precursor, after 173 days, while at 25°C more wine-like amounts (200-600 ng/L) were observed. A glucoside, 4,5-dihydrovomifoliol-C₉-β-D-glucopyranoside (9b), was isolated from grapevine leaves by multilayer coil countercurrent chromatography (MLCCC), and its stereochemistry was deduced as being (5R, 6S, 9R) by NMR and CD spectroscopy. Hydrolysis of this glucoside produced 1, but in quantities insufficient to account for the levels observed in wine.

Full text of DOI:10.1021/jf051039w

J. Agric. Food Chem. 2005, 53, 6777−6783 6777
Generation of (E)-1-(2,3,6-Trimethylphenyl)buta-1,3-diene from
C13-Norisoprenoid Precursors
AGNIESZKA COX,†,‡,§ GEORGE K. SKOUROUMOUNIS,‡,§ GORDON M. ELSEY,*,†,‡,§
‡,§
MICHAEL V. PERKINS,†,§ AND MARK A. SEFTON
School of Chemistry, Physics and Earth Sciences, Flinders University, Post Office Box 2100,
Adelaide, South Australia, 5001, Australia, The Australian Wine Research Institute, Post Office Box
197, Glen Osmond, South Australia, 5064, Australia, and Cooperative Research Centre for Viticulture,
Post Office Box 154, Glen Osmond, South Australia 5064, Australia
Three C13-norisoprenoid compounds, 3,6,9-trihydroxymegastigma-4,7-diene (6), 3,4,9-trihydroxyme-
gastigma-5,7-diene (4), and the actinidols (8), have all been synthesized and subjected to acid
hydrolysis. All three were shown to generate (E)-1-(2,3,6-trimethylphenyl)buta-1,3-diene (1) under
wine conservation conditions. At 45 °C, approximately 4000-5000 ng/L of 1 was formed from 1.0
mg/L of precursor, after 173 days, while at 25 °C more wine-like amounts (200-600 ng/L) were
observed. A glucoside, 4,5-dihydrovomifoliol-C9-â-D-glucopyranoside (9b), was isolated from grapevine
leaves by multilayer coil countercurrent chromatography (MLCCC), and its stereochemistry was
deduced as being (5R, 6S, 9R) by NMR and CD spectroscopy. Hydrolysis of this glucoside produced
1, but in quantities insufficient to account for the levels observed in wine.
KEYWORDS: TPB; SIDA; polyphenols; grapes; precursors; C13-norisoprenoids; MLCCC.
INTRODUCTION
halter and colleagues (5) have shown that both TDN and
damascenone are generated from multiple precursor forms which
appeared to be glycoconjugates involving different conjugating
moieties.
(E)-1-(2,3,6-Trimethylphenyl)buta-1,3-diene (1, Figure 1) is
a wine component which we recently identified in white wines
(1). The aroma detection threshold of this compound was 40
ng/L in white wine, which places it among the most potent of
wine odorants. It was generated by the heating of crude
glycosidic extracts of both red and white varieties under acidic
conditions (1). While 1 was found in all the crude glycoside
hydrolysates, it was found only in white wine (at concentrations
up to ∼250 ng/L), suggesting that there could be a chemical
basis for its presence in the glycosidic fractions but not in red
wine itself (2). We have subsequently shown that 1 reacts rapidly
and immediately with wine polyphenols, which could well
account for its absence in red wine. Studies on the occurrence
and stability of 1 revealed that it is probably formed during
wine aging, very like its C13-norisoprenoid isomer 1,1,6-
trimethyl-1,2-dihydronaphthalene (TDN, 2) (2).
It has previously been established that 3,4,9-trihydroxyme-
gastigma-5,7-diene (4) as well as the dihydro analogue 5 are
present in grapes, most probably as glycoconjugates (6).
However, the rearranged compounds 6 and 7 have yet to be
identified in grapes. Plucheoside B (triol 4 with â-D-glucose at
C3) was identified in the aerial part of Pluchea indica (7). Two
monosaccharides of 4 [plucheoside B and alangionoside C (â-
D-glucose at C4)] as well as a disaccharide [ebracteatosie A
(apiose and â-D-glucose at C9)] were identified by Kancha-
napoom et al. (8) in a Thai medicinal plant, Acanthus ebrac-
teatus. Strauss et al. (6) investigated the hydrolytic behavior of
both 4 and 6 to try and rationalize the formation of several C13-
norisoprenoids (Figure 2). Interestingly, one of the hydrolysis
products they observed was an unknown (referred to by them
as “compound 20”) which had a similar mass spectrum to that
of 1. Mechanistically it would be quite straightforward for 1 to
form from the triols 4 and 6. The triols contain three allylic (or
potentially allylic) alcohols that would be amenable to acid-
catalyzed processes such as dehydration. Hydrolytically, triol
6 was shown to produce a complex mixture of products,
including triol 4, TDN (2), the actinidols (8), and several minor
compounds, including the aforementioned “unknown 20”. The
actinidols (8) have been observed as volatiles in bottle-aged
white wines, brandies, and heated muscat grape juices and in
grapes (9), as well as in the acid hydrolysates of grape and wine
glycoside extracts (10-12). They were originally isolated from
C13-Norisoprenoids are believed to arise from degradation
of carotenoid precursors, and commonly accumulate in grapes
as glycoconjugates (3). These glycoconjugates can then undergo
acid-catalyzed transformations to produce other C13-noriso-
prenoids. For example, Humpf et al. (4) have shown that TDN
is generated from the glycoside 3 by heating at pH 2.5 under
simultaneous distillation-extraction (SDE) conditions. Winter-
* Corresponding author (telephone +61 8 8201 3071; fax +61 8 8201
2905; e-mail Gordon.Elsey@flinders.edu.au.
† Flinders University.
‡ The Australian Wine Research Institute.
§ Cooperative Research Centre for Viticulture.
10.1021/jf051039w CCC: $30.25
© 2005 American Chemical Society
Published on Web 07/23/2005

6778 J. Agric. Food Chem., Vol. 53, No. 17, 2005
Cox et al.
Figure 1. Structures (and numbering scheme) of compounds referred to in the text.
Figure 2. Hydrolysis products of 6, as reported by Strauss et al. (6).
distilled from sodium/benzophenone immediately prior to use. All
organic solutions were dried over anhydrous sodium sulfate prior to
filtration. Unlabeled (E)-1-(2,3,6-trimethylphenyl)buta-1,3-diene (1) was
prepared as described by Janusz et al. (1). Crude glycosidic extracts
(from both fruit and leaves) were prepared as described earlier (1).
GC-MS-O analysis of hydrolysates was conducted as described
earlier. (1). GC-MS analysis of acid hydrolysates and quantification
of 1 were conducted as described elsewhere (2). Buffered model wine
solutions were prepared by saturating a 10% solution of ethanol in water
with potassium hydrogen tartrate and adjusting the pH to the desired
value with 10% aqueous tartaric acid solution.
Isolation of Glycoside 9b by Multilayer Coil Countercurrent
Chromatography (MLCCC). Separation 1: Gradient Chloroform/
Methanol/Water (7/13/8 to 7/1/8). The MLCCC apparatus was a Quattro
(Analytical and Environmental Consultancy Services (AECS), Bridgend,
UK) equipped with 4 coils on 2 holders (bobbins of 190 mm diameter).
Each bobbin was composed of 2 coils (1.6 mm i.d. Teflon tubing) with
volume capacities of 100 mL and 250 mL, respectively. The temperature
of the cabinet containing the different coils was controlled by a
thermostated cooling system (Ratek circulated cooling device), (Adelab,
Australia). The whole system was equipped with a back-pressure
regulator (250 psi) (Upchurch Scientific, New Zealand) set upstream
of the bobbins. The individual separations were performed using one
the leaves of Actinidia polygama by Sakan et al. (13) who also
synthesized these compounds (14), although they did not
determine the stereochemistry of either the natural or synthetic
products. Subsequently, Dimitriadis et al. (9) used the same
synthetic approach and reported that four isomeric actinidols
were formed in a ratio of 40:53:3:4. They subsequently
confirmed that the two major, naturally occurring, grape-derived
isomers of the actinidols had a trans relationship between the
C5-methyl and the C8-hydroxyethyl (9). Each of the three
compounds 4, 6, and 8 could be considered potential precursors
to 1. This study was undertaken to determine the role of these
three in generating 1 under wine conservation conditions.
MATERIALS AND METHODS
General. NMR spectra were recorded as solutions in chloroform-d
and were obtained on a Varian Gemini spectrometer operating at either
300 MHz (1H) or 75.5 MHz (13C). Circular dichroism (CD) spectra
were recorded on an Applied Photophysics π*-180 CD spectrometer,
and specific rotations were recorded on a PolAAr 21 polarimeter. All
reagents were purchased from Sigma-Aldrich. All solvents were of the
highest commercial grade available. Diethyl ether and THF were

Generation of (E)-1-(2,3,6-Trimethylphenyl)buta-1,3-diene
J. Agric. Food Chem., Vol. 53, No. 17, 2005 6779
100 mL coil only. The revolution speed was set at 800 rpm, the
temperature was kept at 25 °C, and the flow rate was delivered at 1.5
mL/min by a Waters 600E pump. The elution was monitored by a GBC
LC1210 UV-vis detector (320 nm) and UPC-900 detector (254 nm)
(Amersham Pharmacia Biotech, Sweden). Chromatograms were re-
corded using Agilent Chemstation software. The Riesling glycosidic
extract (approximately 300-400 mg) was dissolved in a (50/50 v/v)
mixture of the upper and lower phases of the solvents, filtered through
a 0.45 µm filter (Millipore) prior to injection through a (2 mL)
Rheodyne injection loop. A methanol-gradient solvent system consisting
of chloroform/methanol/water (7/x/8) was used. The proportion of
methanol was reduced, at 30 min intervals, from 13 parts to,
successively, 10, 7, 4, and one part (i.e., from x ) 13 to x ) 1, via x
) 10, x ) 7, and x ) 4). The solvents were sonicated for 30 min
before use. The elution mode of the mobile phase was tail-to-head,
with the less dense layer (aqueous) being the mobile phase. Eighty
fractions (3 mL) were collected with a fraction collector. Fractions were
also screened by enzyme hydrolysis, followed by GC-MS analysis.
In total, 9 runs of approximately 300-400 mg material per run were
performed and combined based on their 254 nm detection. The average
stationary phase retention was 88% before injection and 50% after
injection.
Separation 2: Isocratic Chloroform/Methanol/Water (7/13/8). Frac-
tions 11-25 (approximately 800-900 mg) from separation 1 were
further separated on a 200 mL coil (2 × 100 mL coils joined in tandem)
as per separation 1, but an isocratic solvent system was used with 13
parts methanol. The stationary phase retention was 50% before injection,
and 24% after injection. A total of 120 fractions (3 mL) were collected.
Separation 3: Isocratic Butanol/Ethyl Acetate/Water (2/3/5). Frac-
tions 51-60 from separation 2 (approximately 200-300 mg) were
further purified on a 200 mL analytical column coil (2 × 100 mL coils
joined in tandem) as per separation 1. However, an isocratic butanol/
ethyl acetate/water (2/3/5) solvent system was used. The stationary
phase retention before injection was 89% and 82% after injection. A
total of 110 fractions (5 mL) were collected. The above 3 separations
resulted in a clear oil for fractions 56-63 (31 mg, >95% purity) which
was found to be 4,5-dihydrovomifoliol-â-D-glucopyranoside (9b). The
spectroscopic data for 9b was consistent with that reported by Inada et
al. (18).
Enzyme Hydrolysis. The glycosidic extract (10-15 mg) was
dissolved in pH 5 phosphate-citrate buffer (0.5 mL) and an equivalent
mass of AR 2000 glycosidase enzyme preparation was added. The vial
was capped under nitrogen, and the solutions were incubated for 16 h
at 40 °C. The aglycones were then extracted with dichloromethane:
pentane (1:2; 5 × 1 mL) and diethyl ether:dichloromethane (3:1; 2 ×
1 mL) and filtered, and the organic portion was dried through a plug
of cotton wool containing Na2SO4. Internal standard (25 µL of 600
mg/L 3,5-dimethylphenol) was added, and the extract was then analyzed
directly by GC-MS.
4,5-Dihydrovomifoliol (9a). The glycoside (9b) (10 mg) was treated
with AR 2000 as described above. δH (CDCl3): 5.84 (1H, dd, J )
15.8, 5.6, H8), 5.70 (1H, dd, J ) 15.8, 1.1, H7), 4.44 (1H, app quin, J
) 6.1, H9), 2.84 (1H, d, J ) 13.5, H2a), 2.15-2.50 (3H, m, H4a,4b,5),
1.92 (1H, dd, J ) 13.5, 2.2, H2b), 1.33 (3H, d, J ) 6.5, H10), 0.97, 0.94
(6H, 2xs, H11,12), 0.88 (1H, d, J ) 6.5, H13). MS m/z (%): 208 ((M-
H2O)+, 9), 193 (2), 169 (7), 165 (13), 152 (9), 141 (31), 129 (27), 124
(33), 111 (31), 109 (51), 97 (33), 95 (42), 85 (100), 71 (71), 55 (35),
43 (71).
m, H3,9), 2.80-2.15 (3H, br s, 3OH), 1.88-1.44 (5H, m, H2,13), 1.26
(3H, d, J ) 6.5, H10), 0.98, 0.89 (6H, 2s, H11,12). δH (d6-acetone):
(major isomer) 5.84 (1H, ddd, J ) 15.6, 5.9, 2.3, H8), 5.61 (1H, dd, J
) 15.5, 1.0, H7), 5.48 (1H, m, H4), 4.44-4.08 (2H, m, H3,9), 3.88-
3.28 (3H, br s, 3OH), 1.80-1.50 (5H, m, H2,13), 1.22 (3H, d, J ) 6.5,
H10), 0.99, 0.88 (6H, 2 s, H11,12).
3,4,9-Trihydroxymegastigma-5,7-diene (4). The triols (6) (111 mg,
0.49 mmol) were dissolved in pH 3.0 aqueous buffer solution (14 mL)
and heated on a boiling water bath in a stoppered round-bottom flask
for 15 min. The reaction mixture was cooled and extracted with
dichloromethane (2 × 10 mL) and chloroform (12 × 10 mL). The
organic layers were combined, dried, and concentrated in vacuo to yield
a yellow oil (93 mg). The crude reaction mixture was subjected to 2
silica chromatography columns (column 1 eluant, methanol/dichlo-
romethane gradient; column 2 eluant, ethyl acetate) to give the triols
(4) as a mixture of diastereomers (39 mg, 35%). δH (CDCl3): (major
isomer) 6.00 (1H, br d, J ) 16.0, H7), 5.53 (1H, dd, J ) 16.0, 6.3, H8),
4.37 (1H, app quin, J ) 6.3, H9), 4.00-3.70 (2H, m, H3,4), 2.34 (3H,
br, 3 OH), 1.82 (3H, br s, H13), 1.76-1.40 (2H, m, H2a,2b), 1.30 (3H,
d, J ) 6.3, H10), 1.04, 1.01 (6H, 2s, H11,12). δH (d4-MeOH): (major
isomer) 6.07 (1H, d, J ) 16.1, H7), 5.40 (1H, ddd, J ) 16.1, 6.2, 1.7,
H8), 4.05 (1H, app quin, J ) 6.3, H9), 3.87 (1H, br d, J ) 3.3, H4),
3.79 (1H, m, H3), 1.86 (3H, br s, H13), 1.78 (1H, m, H2a), 1.45 (1H,
ddd, J ) 12.3, 3.6, 1.3, H2b), 1.29 (3H, d, J ) 6.3, H10), 1.07, 1.10
(6H, 2s, H11,12).
A second fraction (31 mg) was also isolated with the hydroxyketone
12 as the major product, and the diastereomeric actinidols (8a/b) were
isolated as a minor component. The spectroscopic data for 12 was in
agreement with that of Strauss et al. (6).
Actinidols (8). The triols (6) (70 mg, 0.31 mmol) were dissolved in
pH 3.0 aqueous buffer solution (7 mL) and heated under reflux for 30
min. The reaction mixture was then cooled to room temperature and
extracted with diethyl ether (2 × 5 mL). The hydrolysis and extraction
step was then repeated 7 times. The organic layers were combined,
dried, and concentrated in vacuo to yield the crude product (60 mg).
This was purified by two silica chromatography columns (column 1
elutant, ethyl acetate; column 2 elutant, 30% ethyl acetate in pentane)
to give the actinidols (8a/b) as a mixture of diastereomers (8 mg, 13%),
in approximately a 1.3 to 1 ratio. The hydroxyketone (12) (18 mg,
31%) and diastereomeric 3,4,9-trihydroxymegastigma-5,7-dienes (4) (25
mg, 36%) were also isolated. δH (CDCl3): (major isomer of 8, consistent
with isomer (Kovats index (KI) 1200) reported by Dimitriadis et al.
(9)) 5.78 (1H, m, H4), 5.62 (1H, dt, J ) 10.3, 3.2, H3), 5.30 (1H, br s,
H7), 4.60 (1H, dd, J ) 4.9, 1.1, H8), 3.60 (1H, qd, J ) 6.3, 4.9, H9),
2.01 (2H, dd, J ) 3.2, 1.4, H2), 1.44 (3H, s, H13), 1.22 (3H, s, H11),
1.17 (3H, d, J ) 6.3, H10), 1.14 (3H, s, H12). δC (CDCl3): 154.0, 132.3,
126.2, 117.5, 86.6, 86.2, 69.9, 43.3, 34.1, 28.7, 26.7, 26.4, 19.2. MS
m/z (%): 193(2), 175(1), 164(14), 163(100), 157(1), 149 (12), 145-
(13), 135(5), 131(5), 121(14), 119(7), 107(11), 105(12), 93(10), 91-
(12), 79(10), 77(10), 55(6), 43(23).
δH (CDCl3): (minor isomer of 8, consistent with isomer (KI 1211)
reported by Dimitriadis et al. (9)) 5.79 (1H, m, H4), 5.59 (1H, dt, J )
10.0, 3.2, H3), 5.37 (1H, br s, H7), 4.75 (1H, dd, J ) 3.0, 1.1, H8), 3.81
(1H, qd, J ) 6.6, 3.1, H9), 2.01 (2H, dd, J ) 3.2, 1.4, H2), 1.45 (3H,
s, H13), 1.23 (3H, s, H11), 1.15 (3H, d, J ) 6.6, H10), 1.14 (3H, s, H12).
δC (CDCl3): 154.5, 132.0, 126.7, 115.5, 86.3, 86.0, 68.4, 43.3, 34.2,
28.6, 26.6, 26.2, 17.5. MS m/z (%): 193(3), 175(1), 164(15), 163-
(100), 157(1), 149 (12), 145(13), 135(4), 131(4), 121(13), 119(7), 107-
(9), 105(12), 93(9), 91(10), 79(6), 77(7), 55(3), 43(13).
Generation of 1 by Acid Hydrolysis. Solutions of each of the
compounds, 4, 6, and 8, as well as glycoside 9b, were prepared by
dissolving the particular compound (1.0 mg/L) in pH 3.2 model wine.
The solutions were transferred into glass ampules and sealed under
nitrogen. The ampules were heated at 25 °C, 45 °C or 100 °C. Duplicate
ampules were removed at selected times, opened, and the concentration
of 1 determined (2).
3,6,9-Trihydroxymegastigma-4,7-diene (6). Dehydroionone (10)
(0.404 g, 2.13 mmol) was converted into 3,6-dihydroxymegastigma-
4,7-dien-9-one (11) (299 mg, 63%) as described by Strauss et al. (6).
A portion of this product (150 mg, 0.67 mmol) was dissolved in
methanol (10 mL) and CeCl3‚7H2O (93.5 mg, 0.25 mmol) and NaBH4
(18 mg, 0.48 mmol) were added. After 10 min water (10 mL) was
added slowly and the solution was saturated with NaCl. The reaction
mixture was extracted with dichloromethane and diethyl ether, the
extracts were combined, dried with Na2SO4 and concentrated in vacuo
to yield a pale yellow oil (91 mg). The crude reaction mixture was
purified by two silica chromatography columns (eluant, ethyl acetate)
to yield the dienes (6) as a mixture of diastereomers (26 mg, 17%). δH
(CDCl3): (major isomer) 6.00-5.25 (3H, m, H4,7,8), 4.50-4.00 (2H,
RESULTS AND DISCUSSION
Isolation of Pure 4,5-Dihydrovomifoliol-C9-â-D-glucopy-
ranoside. MLCCC was employed for the isolation of individual

6780 J. Agric. Food Chem., Vol. 53, No. 17, 2005
Cox et al.
Figure 3. Synthesis of triol 6.
glycoside 9b from a crude glycosidic mixture (15). GC-MS
analysis (after enzyme hydrolysis) revealed that aglycone 9a
was a major component of all the crude glycosidic extracts
studies, comprising both red and white varieties. As this
particular component was most abundant in a 2001 South
Australian Riesling leaf extract, this particular extract was
chosen for MLCCC separation. The solvents and conditions used
were adapted from those of Skouroumounis and Winterhalter
(16).
(as the tetraacetate) than they are in the synthetic aglycone, as
expected for a 1,3-diaxial interaction with the C6-hydroxyl in
the former.
Hydrolysis of Glycoside 9b. Having isolated a pure fraction
of the glycoside 9b, we were keen to conduct some hydrolysis
experiments to determine if it produced 1, or indeed any other
aromas worth pursuing. Preliminary hydrolysis of the isolated
glycoside 9b (at approximately 50 mg/L) gave 1 as one of the
products identified by gas chromatograph-olfactometry (GC-
O), with this compound being one of the strongest aromas in
the hydrolysate. However, when this glycoside was hydrolyzed
at a more wine-like concentration of 1 mg/L in model wine
(pH 3.2) at 25 °C, 45 °C, or 100 °C, only low levels (∼20
ng/L) of 1 were observed. Therefore, although 9b was initially
considered a possible precursor to 1, these results suggest that
unrealistically high amounts of this glycoside would have to
be present in wine to account for the levels observed in some
white wines. (11,12). Thus, other more significant precursors
to 1 must be present in glycosidic extracts and at least some
wines.
Synthesis of C13-Norisoprenoids 4, 6, and 8. The 3,6,9-
triols (6) were synthesized (as a mixture of diastereomers) in
two steps (Figure 3) from 3,4-dehydro-â-ionone (10) using the
procedure of Strauss et al. (6). It was found not to be practical
to attempt separation of the individual diastereomers, and the
diastereomeric mixture was used in the hydrolytic study. In
keeping with the observations of Strauss et al., when 6 was
heated briefly to 100 °C at pH 3.0 the major products were as
follows: unreacted 6, rearranged triol 4 (as a mixture of
diastereomers), the actinidols (8), and the rearranged ketone 12.
Pure fractions of both 4 and 8 were obtained as diasteromeric
mixtures, after column chromatography. As with 6, the subse-
quent hydrolysis experiments were performed on these mixtures.
Hydrolysis of Aglycones 4, 6, and 8. Compounds 4, 6, and
8, (each at 1 mg/L and pH 3.2) were separately subjected to
acid hydrolysis in model wine at 25 °C and 45 °C, and the
product mixtures were examined by GC-MS (2); all 3
norisoprenoids (hydrolyzed independently of each other) pro-
duced hydrolysates which had the same overall aroma, with
descriptors including ”cut-grass”, ”green”, ”metallic”, ”fly
spray”, and ”fruity”. Furthermore, all three hydrolysates
were found to contain 1. The two dienes 4 and 6 can be formed
from each other under acidic conditions (6, 9), and it is possible
that that they can also form from the actinidols (8), if formation
of the latter is reversible (Figure 4). It is possible, therefore,
to see similar product profiles irrespective of the chosen
precursor.
The identity of the aglycone, 4,5-dihydrovomifoliol (9a), was
confirmed by NMR and GC-MS analysis of the product after
enzyme hydrolysis, and the latter was in broad agreement with
that of Sefton et al. (17). The sugar unit in 9b was assigned as
1
â-D-glucopyranoside based on comparison of the 13C and H
NMR data with the literature (18,19). Inada et al. (18)
established, with the aid of NOESY NMR experiments, that
the â-D-glucopyranose was attached at C9. As our NMR data
were identical with those reported, the sugar was also determined
to be attached at the C9 position.
The circular dichroism (CD) spectrum of the glycoside 9b
showed a positive Cotton effect at 290 nm (methanol) as was
observed by both Inada et al. (18) and Otsuka et al. (19), as
well as Tamaki et al. (20) for the analogous disaccharide. On
the basis of application of the octant rule (21, 22), they
determined the absolute stereochemistry at C5 and C6 to be R
and S, respectively. Hence, the absolute stereochemistry of the
glycoside 9b isolated in the present study is also assigned as
(5R,6S). Tamaki et al. (20) deduced the stereochemistry at C9
as R based on the fact that the 13C NMR data for the ring carbons
were superimposable on those of alangionoside G and alang-
ionoside I (23) which both had R stereochemistry at C9.
Meanwhile, Shen and Terazawa (24) deduced the stereochem-
istry as R by the Helmchen method.
Inada et al. (18) reported a relatively large specific rotation
([R]D -35.1, c 0.11, methanol) for 9b, whereas Otsuka et al.
(19) reported a value ([R]24D ca. 0, c 0.71, methanol) closer to
the one determined here ([R]D -5.4, c 1.0, methanol). This
supports Otsuka and colleagues’ claim (19) that the optical
rotation obtained by Inada et al. is not accurate, based on their
calculations of the optical rotation using the Klyne rule from
an analogous disaccharide, which they determined to be
approximately -7°.
Sefton et al. (17) previously identified the glycoside 9b
(isolated as the tetraacetate) as a grape-derived product but were
unable to determine C5/C6 relative stereochemistry, which
differed between the aglycone obtained from the natural
glycoside and a synthetic isomer. The C5-secondary methyls of
both the synthetic aglycone and the glucoside obtained in this
study are equatorial. This suggests that the stereochemistry of
the synthetic aglycone differs from that of the natural gluco-
side in that the C6-side chain is equatorial in the latter but axial
Quantitative studies were carried out for each of the three
suspected precursors, and all three generated 1 over time under
acidic conditions at both temperatures employed. Trace levels
(<100 ng 1 per mg starting material) of 1 were present in all
zero time samples; these amounts are likely to have been
generated during sample preparation as there was a small time
delay (several hours) between preparation of the samples in
model wine and their analysis. The amounts of 1 present in the
1
in the former. This assignment is supported by the H NMR
spectra of these compounds (the glucoside as its tetraacetate)
in CDCl3 and C6D6. For both solvents, the H2 and H4-axial
hydrogens are significantly further downfield in the glucoside

Generation of (E)-1-(2,3,6-Trimethylphenyl)buta-1,3-diene
J. Agric. Food Chem., Vol. 53, No. 17, 2005 6781
Figure 4. Proposed mechanism for the formation of 1 from 4, 6, and 8.
Figure 5. Production of 1 from 4, 6, and 8 (at 1 mg/L) at 45 °C and 25 °C after 6 months. All data are an average of 2 replicates.
hydrolysates after 6 months are shown in Figure 5. After 6
months at 45 °C, 4000-5000 ng/L of 1 was present, irrespective
of the precursor. All three precursors also generated 1 at the
lower temperature of 25 °C. Under these more typical maturation
conditions, the precursors (1 mg/L) formed much more wine-
like amounts (200-600 ng/L) of 1 (1,2).
TPB Generation from Crude Glycosidic Extracts. Com-
pound 1 was first identified in the acid hydrolysates of both
red and white crude glycosidic extracts (1). These hydrolysates
were examined qualitatively by GC-MS-O, which indicated
the presence of a powerful odorant. We have now analyzed
hydrolysates of such extracts using the stable isotope dilution
assay (SIDA) method developed specifically for 1 (2), to assess
their potential as a precursor pool to this compound. Both grape
and leaf glycosidic extracts were studied, with the results being
shown in Table 1.
Similar amounts of 1 were produced from the grape extracts
of all varieties, approximately 5000-10000 ng of 1 per gram
of crude glycosidic material after heating for 3 months at 45
°C. However, the true amount of 1 generated is likely to be

6782 J. Agric. Food Chem., Vol. 53, No. 17, 2005
Cox et al.
LITERATURE CITED
Table 1. Amounts of 1 Generated by Acid Hydrolysis of Crude Fruit
and Leaf Glycosidic Extracts
(1) Janusz, A.; Capone, D. L.; Puglisi, C. J.; Perkins, M. V.; Elsey,
G. M.; Sefton, M. A. (E)-1-(2,3,6-Trimethylphenyl)buta-1,3-
diene: A Potent Grape-Derived Odorant in Wine. J. Agric. Food
Chem. 2003, 51, 7759-7763.
fruit
leaf
samplea
ng/gb,c
ng/kgd
ng/Le
ng/gb,c
ng/kgf
2000 S
2001 S
2000 CS
2001 CS
2001 R
2001 M
2001 T
7733
5267
7308
8983
7100
9715
7592
1036
1448
782
1249
ndg
2378
3092
1554
2429
ndg
958
2217
1127
1285
1169
9450
5827
9415
13905
14234
20537
19067
29077
30301
(2) Janusz, A. J.; Capone, D. L.; Elsey, G. M.; Perkins, M. V.;
Sefton, M. A. Quantitative Analysis, Occurrence and Stability
of (E)-1-(2,3,6-Trimethylphenyl)buta-1,3-diene (TPB) in Wine.
J. Agric. Food Chem. 2005, 53, 3584-3591.
(3) Gerdes, S. M.; Winterhalter, P.; Ebeler, S. E. Effect of Sunlight
Exposure on Norisoprenoid Formation in White Riesling Grapes.
In Carotenoid-DeriVed Aroma Compounds; Winterhalter, P.,
Rouseff, R. Eds.; ACS Symposium Series 802; American
Chemical Society: Washington, DC, 2002; pp 262-272.
(4) Humpf, H.-U.; Winterhalter, P.; Schreier, P. 3,4-Dihydroxy-7,8-
dihydro-â-ionone-â-D-glucopyranoside: Natural Precursor of
2,2,6,8-Tetramethyl-7,11-dioxatricyclo[6.2.1.01,6]undec-4-ene
(Riesling Acetal) and 1,1,6-Trimethyl-1,2-dihydronaphthalene in
Red Currant (Ribes rubrum L.) Leaves. J. Agric. Food Chem.
1991, 39, 1833-1835.
ndg
ndg
ndg
ndg
a S (Shiraz), CS (Cabernet Sauvignon), R (Riesling), M (Muscat Rose), T
(Traminer). b Hydrolyzed approximately 0.1 mg extract per milliliter of model wine
(pH 3.2) at 45 °
C for 3 months; average of 2 replicates. c Mass of 1 produced per
gram of crude glycosidic extract. d Mass of 1 produced per kg of fruit. e Mass of
1 produced per liter of juice. f Mass of 1 produced per kilogran of leaves. g Not
determined.
(5) Winterhalter, P.; Sefton, M. A.; Williams, P. J. Volatile C13-
Norisoprenoid Compounds in Riesling Wine are Generated from
Multiple Precursors. Am. J. Enol. Vitic. 1990, 41, 277-283.
(6) Strauss, C. R.; Dimitriadis, E.; Wilson, B.; Williams, P. J. Studies
on the Hydrolysis of Two Megastigma-3,6,9-triols Rationalising
the Origins of Some Volatile C13-Norisoprenoids in Vitis Vinifera
Grapes. J. Agric. Food Chem. 1986, 34, 145-149.
(7) Uchiyama, T.; Miyase, T.; Ueno, A.; Usmanghani, K. Terpenic
Glycosides from Pluchea indica. Phytochemistry 1989, 28,
3369-3372.
(8) Kanchanapoom, T.; Kasai, R.; Picheansoonthon, C.; Yamasaki,
K. Megastigmane, Aliphatic Alcohol and Benzoxazinoid Gly-
cosides from Acanthus ebracteatus. Phytochemistry 2001, 58,
811-817.
(9) Dimitriadis, E.; Strauss, C. R.; Wilson, B.; Williams, P. J. The
Actinidols: Nor-isoprenoid Compounds in Grapes, Wines and
Spirits. Phytochemistry 1985, 24, 767-770.
(10) Sefton, M. A.; Francis, I. L.; Williams, P. J. Free and Bound
Volatile Secondary Metabolites of Vitis Vinifera Grape cv.
Sauvignon Blanc. J. Food Sci. 1994, 59, 142-147.
(11) Sefton, M. A. Hydrolytically Released Volatile Secondary
Metabolites from a Juice Sample of Vitis Vinifera Grape cvs
Merlot and Cabernet Sauvignon. Aust. J. Grape Wine Res. 1998,
4, 30-38.
(12) Sefton, M. A.; Francis, I. L.; Williams, P. J. The Volatile
Composition of Chardonnay Juices: A Study by Flavor Precursor
Analysis. Am. J. Enol. Vitic. 1993, 44, 359-370.
(13) Sakan, T.; Isoe, S.; Hyeon, S. B. Structure of Actinidiolide,
Dihydroactinidiolide, and Actinidol. Tetrahedron Lett. 1967, 8,
1623.
(14) Isoe, S.; Hyeon, S. B.; Ichikawa, H.; Katsumura, S.; Sakan, T.
Synthesis of Actinidiolide, Dihydroactinidiolide, and Actinidol.
Tetrahedron Lett. 1968, 9, 5561.
(15) Janusz, A. Separation with Multilayer Coil Countercurrent
Chromatography (MLCCC). Aust. J. Chem. 2002, 55, 745.
(16) Skouroumounis, G. K.; Winterhalter, P. Glycosidically Bound
Norisoprenoids from Vitis Vinifera cV. Riesling Leaves. J. Agric.
Food Chem. 1994, 42, 1068-1072.
substantially higher than the amounts actually measured, as 1
is both generated and degraded simultaneously (2). In keeping
with earlier reports which suggest that grapevine leaves are a
particularly rich source of glycosides (16), the yield of crude
glycosidic extract from the leaf samples was much higher
(approximately 2 orders of magnitude) than from the corre-
sponding fruit samples. In terms of the actual amount of 1
produced hydrolytically, when ”standardised” against the origi-
nal masses employed, the leaf samples demonstrate a greater
potential (approximately 10-fold) for generating 1 than their
corresponding fruit counterparts.
Although the purified glucoside 9b generated 1 on acid
hydrolysis, and although 1 undoubtedly contributed to the overall
aroma of the hydrolysates of 9b, the amounts formed were small,
and this compound cannot account for the amounts of 1 seen
in the acid hydrolysates of crude glycosidic fractions obtained
from grapes and vine leaves (Table 1).
Three C13-norisoprenoid compounds, 3,6,9-trihydroxyme-
gastigma-4,7-dienes (6), 3,4,9-trihydroxymegastigma-5,7-dienes
(4), and the actinidols (8), all generated 1 at wine pH, in
significant quantities. All three formed 1 over time at both
ambient and elevated temperatures. Glycoconjugates of these
compounds could account for the amounts of 1 found in the
hydrolysates of the crude grape glycosidic extracts (Table 1).
Such glycoconjugates may be the only important precursors,
or there may be others, as yet unidentified. Even if all potential
precursors were to be identified, it would be difficult, if not
impossible, to assess the amount of 1 which can potentially form
in wine by examining grape extracts. The processes of depletion
of 1 are concomitant with its generation and cannot as yet be
accurately mimicked experimentally. Diene 1 was only one of
several interesting compounds that have been investigated in
the acid hydrolysates of crude glycosidic extracts from Shiraz
and Cabernet Sauvignon, and there are other aromas in these
hydrolysates requiring further investigation.
(17) Sefton, M. A.; Winterhalter, P.; Williams, P. J. Free and Bound
6,9-Dihydroxymegastigm-7-en-3-one in Vitis Vinifera Grapes and
Wine. Phytochemistry 1992, 13, 1813-1815.
(18) Inada, A.; Nakamura, Y.; Konishi, M.; Murata, H.; Kitamura,
F.; Toya, H.; Nakanishi, T. A New Ionone Glucoside and a New
Phenylpropanoid Rhamnoside from Stems of Ampelopsis breVi-
pedunculata (Maxim.) Trautv. Chem. Pharm. Bull. 1991, 39,
2437-2439.
(19) Otsuka, H.; Zhong, X.-N.; Hirata, E.; Shinzato, T.; Takeda, Y.;
Myrsinionosides A-E. Megastigmane Glycosides from the Leaves
of Myrsine seguinii Lev. Chem. Pharm. Bull. 2001, 49, 1093-
1097.
ACKNOWLEDGMENT
The work was conducted by The Australian Wine Research
Institute and Flinders University as part of the research program
of the Cooperative Research Centre for Viticulture. The authors
would like to thank Dimitra L. Capone for assistance with GC-
MS, as well as Isak S. Pretorius, Markus J. Herderich, and
Elizabeth J. Waters for their support and feedback.

Generation of (E)-1-(2,3,6-Trimethylphenyl)buta-1,3-diene
J. Agric. Food Chem., Vol. 53, No. 17, 2005 6783
(20) Tamaki, A.; Otsuka, H.; Ide, T.; Platanionosides A-C. Megastig-
mane Diglycosides from the Leaves of Alangium platanifolium.
J. Nat. Prod. 1999, 62, 1074-1076.
(21) Klyne, W.; Kirk, D. N. The Carbonyl Chromophore, Saturated
Ketones. In Fundamental Aspects and Recent DeVelopments in
Optical Rotatory Dispersion and Circular Dichroism; Ciardelli,
F., Salvadori, P., Eds.; Heyden & Sons Ltd.: London, 1973; pp
89-107.
(22) Crabbe, P. Optical Rotatory Dispersion and Circular Dichroism
in Organic Chemistry; Holden-Day, Inc.: San Francisco, 1965.
(23) Otsuka, H.; Yao, M.; Kamada, K.; Takeda, Y.; Alangionosides
G-M. Glycosides of Megastigmane Derivatives from the Leaves
of Alangium premnifolium. Chem. Pharm. Bull. 1995, 43, 754-
759.
(24) Shen, Y.; Terazawa, M. Dihydroroseoside, a New Cyclohexanone
Glucoside, from the Leaves of Shirakamba (Betula platyphylla
Sukatchev var. japonica Hara). J. Wood Sci. 2001, 47, 145-
148.
Received for review May 5, 2005. Revised manuscript received June
17, 2005. Accepted June 20, 2005. This project is supported by
Australia’s grapegrowers and winemakers through their investment
body, the Grape and Wine Research and Development Corporation,
with matching funds from the Australian Government, and by the
Commonwealth Cooperative Research Centres Program.
JF051039W