Thermal oxidation of 9′-cis-neoxanthin in a model system containing peroxyacetic acid leads to the potent odorant β-damascenone

Article abstract of DOI:10.1021/jf051330b

The potent odorant β-damascenone was formed directly from 9′-cis-neoxanthin in a model system by peroxyacetic acid oxidation and two-phase thermal degradation without the involvement of enzymatic activity. β-Damascenone formation was heavily dependent on pH (optimum at 5.0) and temperature, occurring over the two sequential phases. The first was incubation with peroxyacetic acid at 60°C for 90 min, and the second was at above 90°C for 20 min. Only traces of β-damascenone were formed on application of only one of the two phases. Formate and citrate solutions produced a much better environment for β-damascenone formation than acetate and phosphate. About 7 μg/L β-damascenone was formed from 5.8 mg/L 9′-cis-neoxanthin under optimal experimental condition. The detailed pathway by which β-damascenone is formed remains to be elucidated.

Full text of DOI:10.1021/jf051330b

J. Agric. Food Chem. 2005, 53, 9199−9206 9199
Thermal Oxidation of 9
Containing Peroxyacetic Acid Leads to the Potent Odorant
-Damascenone
′-cis-Neoxanthin in a Model System
â
YAIR BEZMAN,† ITZHAK BILKIS,† PETER WINTERHALTER,‡ PETER FLEISCHMANN,‡
RUSSELL L. ROUSEFF,§ SUSANNE BALDERMANN,‡ AND MICHAEL NAIM*,†
Institute of Biochemistry, Food Science and Nutrition, Faculty of Agricultural,
Food and Environmental Quality Sciences, The Hebrew University of Jerusalem, PO Box 12,
Rehovot 76-100, Israel, Institute of Food Chemistry, Technical University Braunschweig,
Braunschweig, 38106 Germany, and CREC, University of Florida, Lake Alfred, Florida 33850
The potent odorant â-damascenone was formed directly from 9′-cis-neoxanthin in a model system
by peroxyacetic acid oxidation and two-phase thermal degradation without the involvement of
enzymatic activity. â-Damascenone formation was heavily dependent on pH (optimum at 5.0) and
temperature, occurring over the two sequential phases. The first was incubation with peroxyacetic
acid at 60 °C for 90 min, and the second was at above 90 °C for 20 min. Only traces of â-damascenone
were formed on application of only one of the two phases. Formate and citrate solutions produced a
much better environment for â-damascenone formation than acetate and phosphate. About 7 µg/L
â-damascenone was formed from 5.8 mg/L 9′-cis-neoxanthin under optimal experimental condition.
The detailed pathway by which â-damascenone is formed remains to be elucidated.
KEYWORDS: â-Damascenone; 9′-cis-neoxanthin; peroxyacetic acid; epoxide; thermolysis
INTRODUCTION
â-Damascenone [1-(2,6,6-trimethyl-1,3-cyclohexadien-1-yl)-
2-buten-1-one] is a potent odorant having an odor threshold of
2 ppt in water (1). It was first reported in the oil of the Bulgarian
rose (Rosa damascena) (2) and later in many other fresh and
processed foods and natural products such as coffee, black tea,
tobacco, tomatoes, beer, apple products, grapes, wines, tropical
fruits such as star fruit and passion fruit, and more (3-16).
Isoe et al. (17) suggested a possible mechanism for the bio-
generation of â-damascenone from allenic carotenoids such as
neoxanthin, via oxidative cleavage, resulting in the grasshopper
ketone (1) as an intermediate which must then be reduced by
an enzyme to the allenic triol (2) (Figure 1). The latter then
may undergo the Meyer-Schuster rearrangement (18) with
subsequent dehydrations under acidic conditions to yield
â-damascenone. Evidence supporting such a biochemical path-
way came from the discovery of both glycoconjugate and
nonglycosidic polar precursors of â-damascenone in white wines
and fruit juices (14, 19, 20) and later in a variety of food
products (6, 10, 13, 21, 22). The work of Skouroumounis et al.
(19), who studied the possible hydrolytic pathway of these
precursors, suggested that compound 2 is a key intermediate in
the formation of â-damascenone. These authors found 3-hy-
Figure 1. Proposed pathway for the biogeneration of
from neoxanthin, (1) grasshopper ketone, (2) allenic triol, and (3) 3-hydroxy-
-damascone, according to Isoe et al. (17).
â-damascenone
â
droxy-â-damascone (3) to be the major transformation product
of 2, and only about 5% yield of â-damascenone at elevated
temperatures.
A significant elevation in the content of â-damascenone
occurred when fresh food products are subjected to thermal
processing (1, 6, 10, 13); for example in roasting green Arabica
coffee content may increase by 100-fold (4). The increased
â-damascenone content was usually proposed to involve hy-
drolysis and transformations of glycoconjugates and polar
precursors at the elevated temperatures during processing and
under the acidic conditions occurring in many food systems.
Interestingly, roasted coffee and beer wort (made of dark malt),
* To whom correspondence should be addressed. Tel.: 972-8-9489276.
Fax: 972-8-9476189. E-mail: naim@agri.huji.ac.il.
† The Hebrew University of Jerusalem.
‡ Technical University Braunschweig.
§ University of Florida.
10.1021/jf051330b CCC: $30.25
© 2005 American Chemical Society
Published on Web 10/22/2005

9200 J. Agric. Food Chem., Vol. 53, No. 23, 2005
Bezman et al.
13C, was used for NMR experiments; chemical shifts are expressed in
although possess pH range of 5 or above, contain â-dama-
scenone at levels between 255 and 450 ppb (4, 10) (among the
highest level in food products) under conditions where acid
hydrolysis is not favored. On the other hand, these two products
(green coffee and some barley malts) are subjected to roasting
temperatures (e.g., 200 °C in the case of coffee) (23). In coffee,
degradation of carotenoids to â-damascenone due to heating,
oxidation, and isomerization was proposed (24) but no evidence
for its formation from direct chemical degradation of neoxanthin
was provided.
In this study we used a model system to explore conditions
in which neoxanthin may be degraded to â-damascenone by
thermal oxidation without involvement of enzymatic activity.
Our working hypothesis (see Discussion for a hypothetical
pathway) included the use of moderate temperatures (defined
as phase 1) prior to applying high temperatures (e.g., phase 2).
δ (parts per million) referring to the solvent peaks δH 3.34 and δX 49.0
1
for CD3OD; coupling constants, J, are in hertz. DEPT, 13C, H-1H
1
COSY, H-13C HMQC, and HMBC NMR experiments were carried
out using the conventional pulse sequences as described in the literature.
9′-cis-Neoxanthin: UV (EtOH) λmax nm 324, 392 (s), 414, 437, 466;
% (III/II) ratio ) 86%; NMR parameters presented in Tables S1 and
S2 (see Supporting Information).
High Performance Liquid Chromatography (HPLC). Analyses
were performed on a Merck-Hitachi L-7100 pump equipped with
L-7455 UV-vis photodiode array detector (set to scan from 350 to
550 nm at 2 nm spectral bandwidth) and D-7100 interface module.
The data were recorded and analyzed with Merck-Hitachi D-7000
chromatography data-station software. An RP-18 column (ODS 3 µ,
100 × 4.6 mm) (Hypersil, Runcorn, U.K.) was used for carotenoids
separation using a tertiary solvent gradient elution adopted from Rouseff
and Raley (26) with the following modifications: Initial solvent
composition consisted of 80% methanol, 19% water, and 1% MTBE
(v/v) and then changed to 80% methanol, 12% water, and 8% MTBE
in 15 min. Next, the proportion was modified to 85% methanol and
15% MTBE in 5 min. All solvent changes were made in a linear fashion.
Initial conditions were reestablished in 5 min, and the column was
reequilibrated for 10 min before the next injection. Flow rate was 0.85
mL/min, and injection volume was 20 µL.
Neoxanthin Oxidation in Aqueous Solutions. A 60 µL volume of
5% (v/v) Tween-80 in ethanol and 58 µg of neoxanthin were mixed
with 1 mL of dichloromethane in a 4 mL glass vial (Fisherbrand, 15 ×
45 mm). The solvent was evaporated under a nitrogen stream, and the
residue (prereaction mixture of neoxanthin, PRMN) was dissolved in
1 mL of sodium formate solution [0.4 M, pH 5.0, 0.005% EDTA
(w/v)] while gently mixing to the point where the solution was clear
(final concentration of neoxanthin approximately 10 µM). Next, the
oxidant was introduced [75 µL of 0.78% (v/v) peroxyacetic acid in
water] to a final concentration of 9 mM in the reaction mixture, mixed,
and sealed with a Teflon screw-cap (National Scientific Co.). The
reaction mixture was then treated as described below. All experiments
involving carotenoids were handled at dark or at dim light conditions,
and treatments were executed in duplicates. Treated samples were stored
at -80 °C before analyzed.
MATERIALS AND METHODS
Chemicals. Tween-80 and EDTA were purchased from Sigma (St.
Louis, MO). Ammonia solution (25%), formic acid, citric acid, sodium
acetate, sodium dihydrogen phosphate, and disodium hydrogen phos-
phate were from Merck (Darmstadt, Germany). Methanol (HPLC
grade), diethyl ether, and dichloromethane were from J. T. Baker
(Deventer, Holland). Potassium hydroxide, sodium hydroxide, sodium
chloride, sodium sulfate, acetic acid, ethanol, and petroleum ether
40-60 °C were purchased from Frutarom (Haifa, Israel). Methyl tert-
butyl ether (MTBE) was from Aldrich (Milwaukee, WI), and peroxy-
acetic acid (39% in acetic acid) was from Fluka (Steinheim, Germany).
â-Damascenone was kindly donated by Dr. Gary Takeoka of the USDA
(WRRC, Albany, CA), neoxanthin marker was a gift of Prof. Naoharu
Watanabe of Shizuoka University (Shizuoka, Japan), and 3-OH-â-
damascone was a gift of Dr. George Skouroumounis of the Australian
Wine Research Institute (Adelaide, Australia). Deionized water (Barn-
stead, Dubuque, IA) was used throughout the study.
Neoxanthin Isolation. Isolation followed the procedure of Baumeler
et al. (25) with modifications. Briefly: 1 kg of fresh spinach leafs
(purchased at a local supermarket) was washed with water, blanched
in a microwave oven for 3 min at maximum power (1000 W), and
manually squeezed to remove water as much as possible. Subsequently,
the spinach was extracted with 4 L of methanol/diethyl ether/
concentrated ammonia solution (8:2:0.25 v/v) for 24 h at ambient
temperature. Following extraction, the solvent was paper filtered and
evaporated under reduced pressure by a rotary evaporator (40 °C). The
residue was dissolved in 280 mL of 15% (w/v) KOH in methanol and
140 mL of diethyl ether and left to stand overnight at dark. Following
saponification, the carotenoid fraction was extracted in a separatory
funnel (100 mL aliquots of the saponified extract, 100 mL of 10%
NaCl solution, and 100 mL of diethyl ether), and the alkalinity was
removed by repeatedly washing with water. Finally, the combined
etheral phase was dried over anhydrous sodium sulfate and concentrated
under reduced pressure. Column chromatography [2 × 40 cm, 60 g
alumina (6% moisture) filled in as a slurry with petroleum ether
(40-60 °C)] was used to separate the carotenoids. The column was
first conditioned with 100 mL of 75/25 (v/v) petroleum ether 40-60
°C/diethyl ether prior to loading. Then step-gradient elution was applied
with 100 mL of each: 75/25 petroleum ether 40-60 °C/diethyl ether,
50/50 petroleum ether 40-60 °C/diethyl ether, 25/75 petroleum ether
40-60 °C/diethyl ether, 100% diethyl ether, 5% ethanol in diethyl ether,
and 20% ethanol in diethyl ether (all v/v). The neoxanthin fraction
was eluted as the last band at 20% ethanol in diethyl ether. After
removal of the solvent under reduced pressure (40 °C), the residue
was redissolved in 3 mL of methanol and stored in a glass vial under
nitrogen at -20 °C. The yield was calculated spectrophotometrically
[EtOH, 437 nm, ꢀ ) 139 250 (25)] to be approximately 25 mg of
neoxanthin. This fraction was used for identification and purity analyses
as well as for the experiments throughout this study.
Oxidation of Neoxanthin at Different pH Levels. PRMN was
dissolved in 1 mL of formic acid [0.4 M, 0.005% (w/v) EDTA] which
was preadjusted to pH 2.0, 3.0, 3.5, 4.0, 4.5, 5.0, 6.0, and 7.0 by NaOH
granules, and the oxidant was introduced as described above. Vials
were then incubated in an oven set to 60 °C for 90 min (e.g., moderated
conditions, phase 1), and then the temperature rose to 110 °C for
additional 20 min (phase 2).
Varying Incubation Periods during Phase 1. Aqueous solutions
containing neoxanthin (0.4 M formate, pH 5.0, 0.005% EDTA, 9 mM
peroxyacetic acid) were incubated at 60 °C for 1, 3, 20, 90, and 300
min. Then the temperature was immediately elevated to 110 °C (rate
) 50 °C/min) for an additional 20 min.
Effect of Temperatures during Phase 1. Aqueous solutions
containing neoxanthin (0.4 M formate, pH 5.0, 0.005% EDTA, 9 mM
peroxyacetic acid) were incubated for 90 min at 4, 20, 40, 60, 75, and
85 °C. Then each treatment was further heated to 110 °C for an
additional 20 min.
Effect of Temperatures during Phase 2. Solutions containing
neoxanthin (0.4 M formate, pH 5.0, 0.005% EDTA, 9 mM peroxyacetic
acid) were incubated for 90 min at 60 °C (i.e. phase 1). Then
temperature was set to 60, 80, 90, 100, 110, 120, or 130 °C for an
additional 20 min in the oven (i.e., phase 2).
Excluding Molecular Oxygen. Formate solution (0.4 M, pH 5.0,
0.005% EDTA) was bubbled with argon for 10 min in a septum-sealed
bottle. The oxidant solution [(0.78% (v/v) peroxyacetic acid in water)]
was also treated with argon in the same manner. A 1 mL volume of
the argon-treated formate solution and 75 µL of the treated oxidant
solution were added to the vial containing PRMN while simultaneously
purging argon and sealed immediately. Control samples were prepared
in the presence of air with peroxyacetic acid or with air as the only
oxidant.
Spectroscopic Data. UV: a Unicam UV1 spectrophotometer (Uni-
cam LTD, Cambridge, U.K.) was used. NMR: a Bruker DRX-500
NMR spectrometer, operating at 500 MHz for 1H and at 125 MHz for

Direct Degradation of Neoxanthin to
â
-Damascenone
J. Agric. Food Chem., Vol. 53, No. 23, 2005 9201
Neoxanthin Oxidation in Different Buffer Solutions. A 1 mL
volume of formate, citrate, acetate, or phosphate buffer (0.4 M, pH
5.0, 0.005% EDTA) and 75 µL of 0.78% peroxyacetic acid solution
were added to vials containing PRMN. The vials were incubated at 60
°C for 90 min, and then the temperature was increased to 110 °C for
an additional 20 min.
Solid-Phase Microextraction (SPME). Into each vial containing
the incubated samples were added 0.25 g of NaCl and a small stirring
bar, and vials were sealed with a Teflon septum cap. The vial content
was continuously agitated in a water bath at 50 ( 0.5 °C. Samples
were allowed to equilibrate for 5 min prior to SPME and maintained
at 50 °C throughout the 95-min assay. SPME of the sample’s volatiles
was conducted by inserting a 2-cm fiber coated with 50/30 µm DVB/
Carboxen/PDMS (Supelco, Bellefonte, PA). The full length of the
coated fiber was exposed to the headspace for 90 min. The fiber was
then removed from the headspace and immediately inserted into the
gas-chromatograph (GC) injector.
Quantitative Determination Using SPME. Stock solution of
â-damascenone (10 000 mg/L) was prepared in methanol. Five calibra-
tion solutions, ranging from 0.3 to 8 µg/L â-damascenone, were made
by serial dilutions of the stock in formate solution [0.4 M, pH 5.0,
0.005% (v/v) EDTA]. The desired concentration was prepared in a vial
containing dried Tween-80 as previously mentioned (final volume 1
mL). Quantification was performed by gas-chromatography mass
spectrometry (GC-MS) according to peak areas of ion m/z 121 on the
basis of the calibration curve of the same mass fragment.
Figure 2. HPLC chromatogram at 440 nm (A) and UV−vis spectrum (B)
for the fraction eluted from the alumina column with 20% ethanol in diethyl
ether. (See Methods.)
RESULTS AND DISCUSSION
GC-MS. â-Damascenone was identified and monitored using a
Saturn 2000 mass spectrometer (Varian, Walnut Creek, CA) connected
to a Varian 3800 gas chromatograph, equipped with a SPB-5 capillary
column (30 m × 0.25 mm i.d., df ) 0.25 µm) (Supelco, Bellefonte,
PA). Following the aforementioned SPME procedure, samples were
inserted into the injector (held at 250 °C) for 5 min, in splitless mode.
The oven program was set to 40 °C for 3 min and then 4°/min to 185
°C and held for 15 min. Helium was used as the carrier gas with a
flow rate of 1.5 mL/min. A 0.75-mm i.d. SPME injector liner was used.
The transfer line was held at 250 °C, and the source, at 180 °C. Mass
spectra in the electron impact (EI) mode were generated at 70 eV. A
series of n-alkanes was used to generate a linear retention index (RI)
with the above conditions. Chromatograms and spectra were recorded
with Saturn GC/MS Workstation software, version 5.41 (Varian).
Neoxanthin Isolation. Since this xanthophyll is one of the
principal carotenoids of green leafs, spinach was chosen as the
plant material source for its extraction according to the relatively
simple isolation procedure of Baumeler et al. (25). Ap-
proximately 25 mg of the supposed 9′-cis-neoxanthin was pre-
pared from 1 kg (fresh weight) of spinach (80% pure, determined
by HPLC with comparison to a marker and spectral data from
literature (28)). A sample chromatogram and UV-visible spec-
trum of the isolated fraction is shown in Figure 2. The struc-
ture of the isolated compound was confirmed in an un-
equivocal manner by NMR technique (see Supporting Informa-
tion). The 13C NMR spectrum of 9′-cis-neoxanthin showed 38
different signals, 17 of which belong to the polyene moiety (C14
and C14′ with the same chemical shift), 12 belong to the
cyclohexane moieties, and 9 others to methyl groups (C17′ and
C18 with the same chemical shift) (Table S1; see Supporting
Two specific control experiments were performed regarding the use
of SPME technology in these experiments. The first was designed to
explore the possibility that â-damascenone was formed from a putative
precursor such as compound 3 (3-OH-â-damascone) during SPME
desorption at the GC injection port. For this purpose, a standard of
compound 3 (over 95% purity) was exposed to a SPME fiber and then
inserted into the GC injector for GC-MS analysis under the same
conditions described above. A large peak of compound 3 (RI 1620)
and traces of â-damascenone (RI 1390) (less than 0.02% of that of
compound 3) were observed in the GC-MS chromatogram (not shown).
A GC-MS chromatogram obtained from the incubated PRMN showed
the presence of â-damascenone peak (RI 1390) without any traces of
compound 3 (expected RI 1620) under the experimental conditions (not
shown). Therefore, as previously suggested (27) conversion of 3-OH-
â-damascone to â-damascenone is very unlikely and certainly did not
occur under the experimental conditions described here. The second
control experiment was aimed at exploring if varying the pH of the
PRMN solution may affect the volatility of â-damascenone (matrix
effect) and thus the resulting quantitative determination by SPME. A
20 ppb concentration of the â-damascenone standard was inserted into
each of six vials containing formate solutions (as in PRMN) adjusted
to pHs of either 2, 3, 4, 5, 6, or 7. SPME was conducted, and
â-damascenone content was determined by GC-MS under conditions
described above. There were no differences (three replicates for each
pH) in â-damascenone content across pHs, indicating no pH matrix
effect under the experimental conditions (data not shown).
1
Information). The H NMR spectrum can be divided in two
regions. One includes signals from protons that belong to the
polyene moiety. This part of the spectrum is well resolved and
could be easily analyzed. The second region includes signals
from the two cyclohexane rings and methyl groups. This part
is characterized by substantial overlap between signals from
different protons.
The full assignment of the signals in 13C and 1H NMR spectra
was achieved taking into account the data obtained from the
DEPT and HMQC spectra on the C atoms bonded to H atoms
and on the number of the hydrogen atoms bonded to the same
carbon. The splitting patterns of signals in the 1H NMR
1
spectrum, H-1H COSY, and HMBC spectra (Table S2; see
Supporting Information) allowed understanding of the con-
nectivity of carbon atoms. The attribution of the hydrogen atoms
bonded to C2 (C2′) or to C4 (C4′) to pseudoequatorial or
1
pseudoaxial positions is based on the values of J: H4eq-1H3ax
,
1H4′eq-1H3′ax
,
1H2eq-1H3ax, and H2′eq-1H3′ax hyperfine in-
1
1
teractions are relatively small, J ∼ 5 Hz, and H4ax-1H3ax
,
1H4′ax-1H3′ax, 1H2ax-1H3ax, and 1H2′ax-1H3′ax hyperfine interac-
tions are twice as large, J ∼ 10 Hz. The spectral data are in full
agreement with the suggestion that the isolated compound has
the structure 9′-cis-neoxanthin and fit well with the previously
published data for this compound (25, 29, 30).
Data Processing. Statistical analysis (two-way ANOVA) of the
resulting data was performed using JMP statistical software (SAS
Institute Inc., Cary, NC).

9202 J. Agric. Food Chem., Vol. 53, No. 23, 2005
Bezman et al.
Figure 4. Effect of pH on
â
-damascenone formation. Samples adjusted
to different pH values were incubated for 90 min at 60
then for an additional 20 min at 110 C (phase 2). Results are of GC-MS
analyses monitored at m/z 121. Values are the means SEM of duplicates
of an experiment performed at least twice. Values not sharing the same
°C (phase 1) and
°
±
superscript letter are different (p < 0.05) in
different pH treatments.
â-damascenone yield across
extract â-damascenone from the aqueous medium, the use of
SPME-headspace was selected as a rapid and efficient extraction
tool.
Effect of pH. There was a strong pH dependency of
â-damascenone formation, with a maximal level of about 4 µg/L
at pH 5.0 (Figure 4). This optimal pH is somewhat sur-
prising, since lower pH values (i.e. 3.0-3.5) would be ex-
pected to favor acid-catalyzed conversion of compound 2
(Figure 1) via the Meyer-Schuster rearrangement, with sub-
sequent dehydrations (19). This is also true for acid hydroly-
sis of glycoconjugate precursors of â-damascenone in some
natural products. Perhaps a vicinal diol (see discussion later on),
though its presence was not evidenced, was involved as an
intermediate during neoxanthin degradation. If occurs, it would
be expected to be sensitive to low pH values, since under such
conditions it may undergo the Pinacol rearrangement (31), thus
eliminating the possibility to generate 2. Nevertheless, slightly
acidic conditions are needed for the transformations of 2 into
â-damascenone (17), which seems to be in line with the
tendency for a decrease in the yield of â-damascenone at pH’s
higher than 5.
Time and Temperature during Phase 1. The objectives here
were to study the effects of time and temperature on â-dama-
scenone formation during the incubation period of the carotenoid
with the oxidant (i.e, phase 1). Time effect was studied by
varying incubation periods up to 300 min, under constant
temperature of 60 °C, and with subsequent heating to 110 °C
for an additional 20 min (i.e., phase 2). An optimal incubation
period of 90 min at 60 °C yielding approximately 4 µg/L
â-damascenone was found for phase 1 (Figure 5A). An
interesting observation is the rather moderate decrease in
â-damascenone formation during the period from 90 to 300 min.
The effect of the temperature on the yield of â-damascenone
during 90 min incubation is illustrated in Figure 5B. As
mentioned above, the temperature was then elevated to 110 °C
for additional 20 min. The optimal temperature for phase 1 under
these experimental conditions was found to be 60 °C yielding
approximately 4 µg/L of â-damascenone; even 4 °C during this
phase facilitated the formation of â-damascenone. Heating to
70 and 85 °C adversely affected the yield, which might suggest
that the effect of temperature may affect the degradation of the
already formed intermediates.
Figure 3. MS of a putative
1390) (A) derived from GC-MS chromatogram (m/z 121) of
neoxanthin oxidation treatment (B).
â
-damascenone peak at RT
)
25.6 (RI
)
9′
-cis-
The relatively scarce amounts of neoxanthin in plants made
it hard to obtain large quantities of this pigment. Hence, a
microscale experimental approach was used throughout the
study.
Formation of â-Damascenone from Neoxanthin. This study
provides for the first time the evidence that thermal oxidation
of neoxanthin without the presence of an enzyme leads to
formation of the potent odorant â-damascenone and indicates
the experimental conditions for its occurrence. This is in line
with the notion that for some food products the formation of
â-damascenone during food processing cannot be explained
alone by acid hydrolysis of glycoconjugates of this odorant. Our
initial experiments investigating neoxanthin oxidation at pH
3.5, using citrate buffer and peroxyacetic acid as the oxidant,
resulted in a clear detection of a â-damascenone odor signal
in GC-olfactometer aromagram as evidenced by 3 trained
panelists (data are not shown). This odor signal matched that
of a â-damascenone marker in the GC-FID chromatogram, but
only traces were detected in the FID and the GC-MS chro-
matograms of the putative â-damascenone extracted from the
incubated sample. Subsequent optimization (in particular eleva-
tion of the pH; see below) clearly indicated direct chemi-
cal formation of â-damascenone (RI 1390) from neoxanthin by
GC-MS analysis (Figure 3). The presented chromatogram is
only a fraction of the full analysis range, monitored at m/z 121
(the base mass fragment of â-damascenone). Spectral data and
blanks (oxidation system without the carotenoids) of the largest
peak in this section (left to â-damascenone) suggest that it is
part of the oxidation system and not a degradation product of
9′-cis-neoxanthin. Maximal levels were approximately 7 µg/L
â-damascenone under the experimental conditions, which cor-
responds to a molar yield of about 0.04%. This yield indicates
that the pathway is not chemically preferred. Nevertheless, even
very low yields (ppb range) of â-damascenone produce strong
olfactory sensation in the human nose due to the very low odor
threshold of this odorant (1) and may be highly significant if
occurs during food processing. Since the main focus was to
Temperature Effect during Phase 2. All samples were
treated equally during phase 1 (incubation for 90 min at 60 °C)
but subsequently were treated at elevated temperatures (see

Direct Degradation of Neoxanthin to
â
-Damascenone
J. Agric. Food Chem., Vol. 53, No. 23, 2005 9203
Figure 7. Molecular oxygen (air) not needed for peroxyacetic acid-induced
neoxanthin degradation to
â-damascenone. Peroxyacetic acid (PAA) was
used as the oxidant. The involvement of molecular oxygen in
â
-dama-
scenone formation was tested by the presence of air or after its removal
by argon. Phase 1 was conducted by samples incubation for 90 min at
60
°
C and then for an additional 20 min at 110
SEM of duplicates.
°C (phase 2). Values are
the means
±
Figure 5. Effect of modifications in time (A) and temperature (B) during
phase 1 incubation of 9 -cis-neoxanthin on -damascenone formation.
After phase 1 was completed, the temperature was immediately elevated
to 110 C (rate 50 C/min) for an additional 20 min (phase 2). Values
are the means SEM of duplicates of an experiment performed at least
twice. Values not sharing the same superscript letter are different (p <
0.05) in -damascenone yields across different time (A) or among different
′
â
°
)
±
°
Figure 8.
incubation for 90 min at 60
20 min at 110 C) or when the two phases were sequentially conducted.
â
-Damascenone formation during either phase 1 only (sample
â
°
C) or phase 2 only (sample incubation for
temperature (B) treatments.
°
Values are the means
at least twice.
± SEM of duplicates of an experiment performed
It was technically impractical to use temperatures higher
than 130 °C with this system, since the glass vials might burst
under the steam pressure. However, it should be borne in mind
that some food products, especially those containing low
moisture, such as certain grains (e.g coffee beans and roasted
malts), may be exposed to roasting temperatures of 180 °C and
above (23).
Role of Molecular Oxygen. Samples containing or lacking
oxygen (air) were studied to explore its involvement in the
oxidation of neoxanthin to â-damascenone. Argon was intro-
duced into the samples to ensure the removal of oxygen out of
the solutions and the headspace of the treated samples. Re-
sults presented in Figure 7 indicate no effect of oxygen pres-
ence on the yield of â-damascenone. Furthermore, when
peroxyacetic acid was excluded and molecular oxygen was the
only oxidant in the system, â-damascenone was not detected.
Therefore, molecular oxygen did not play a role in the oxidation-
induced â-damascenone formation from the degraded neoxan-
thin.
Two-Phase Pathway. Perhaps a direct way to verify the need
for the proposed sequential two phases for â-damascenone
formation from neoxanthin is to conduct experiments in which
either phase 1 or phase 2 is omitted. When only phase 1 was
conducted, no â-damascenone was formed during the 90 min
incubation at 60 °C (see Figure 8, left). To study the role of
phase 2 alone, we omitted phase 1 by preheating the neoxan-
thin solution to 100 °C prior to the addition of the oxidant,
and then the oxidant was added and the temperature was
Figure 6. Temperature effect on
2. Phase 1 was conducted for 90 min at 60
â
-damascenone formation during phase
C. Then samples were
incubated for an additional 20 min (phase 2) at the indicated temperatures.
Values are the means SEM of duplicates of an experiment performed
at least twice. Values not sharing the same superscript letter are different
(p 0.05) in -damascenone yields across different temperature
treatments.
°
±
<
â
Materials and Methods) for 20 min. As shown in Figure 6, a
prominent effect of ascending temperatures on â-damascenone
formation occurred, reaching about 7 µg/L â-damascenone at
130 °C. A threshold of 90 °C for phase 2 was evident, with
only trace mass fragments detected by the MS at this temper-
ature. These results strengthen the suggestion of a biphasic
mechanism, since the lower temperatures such as those applied
in phase 1 (i.e. 60 and 80 °C, Figure 5B) alone would not yield
â-damascenone unless a subsequent step employing higher
energy is applied.

9204 J. Agric. Food Chem., Vol. 53, No. 23, 2005
Bezman et al.
Figure 9.
different buffers. All buffer solutions were 0.4 M and adjusted to pH 5.0.
Phase 1 was conducted by sample incubation for 90 min at 60 C and
then for an additional 20 min at 110 C (phase 2). Values are the means
SEM of duplicates of an experiment performed at least twice. Values
not sharing the same superscript letter are different (p 0.05) in
-damascenone yields across different buffer treatments.
â-Damascenone formation during incubations conducted in
°
°
±
<
â
immediately increased to 110 °C for 20 min. Results of
these two series of experiments compared with that in-
cluding both phases in a sequence are shown in Figure 8. A
clear synergism of â-damascenone formation when both
phases are conducted in a sequence is evident with phase 1
conducted at moderate temperatures and phase 2 at high
temperatures.
Figure 10. Hypothetical scheme indicating possible involvement of phase
1 and phase 2 in the thermal oxidation of neoxanthin to â-damascenone.
Phase 1: direct oxidation of neoxanthin by hydroperoxide, leading to the
formation of an epoxide (4), which hydrolyzed to a vicinal diol (5). Phase
2: at temperatures above 90
subsequent disproportionation to yield the allenic triol (2), which is further
converted to -damascenone and (3) under mild acidic conditions.
Neoxanthin and -damascenone were identified; dotted arrows indicate
°C, the C9−C10 bond is thermolyzed with
Buffer Effect. Formic acid was chosen as the medium for
the experiments since it is a customary medium for alkene
epoxidation in the presence of hydroperoxides (31). In addition,
formic acid (as well as other natural products such as mannitol,
glucose, and ethanol) is known to be an efficient hydroxyl-
radical scavenger (32) and might help creating milder oxidation
conditions. Nevertheless, we were interested to reveal whether
the reaction still works under similar pH, in the presence of a
more characteristic food systems buffers, e.g., citrate. Acetate
and phosphate were also tested. Interestingly, â-damascenone
was not formed in media containing acetate and phosphate
buffers (Figure 9). We assume that the yield of this odorant in
each buffer system was affected by the buffer ability to scavenge
free radical products of the hydroperoxide such as hydroxyl and
peroxyl radicals, thus inhibiting a more accelerated oxidative
degradation of the carotenoids. Support for this notion comes
from the bleaching degree of the treated samples: when acetate
and phosphate media were used, a complete bleaching of the
carotenoids was evident, whereas formate and citrate media
retained much of the initial color intensity even at the end of
the experiment (data not shown).
Hypothetical Pathway. The studies of Skouroumounis et al.
(19, 33) suggested that compound 2 is the key intermediate in
the formation of â-damascenone, since it will spontaneously
convert to the latter under acidic environment such as fruit juices
and wines (Figure 1). Yet, oxidative cleavage of neoxanthin at
the C9dC10 double bond, by radicals or carotenases, will only
result in ketone 1 (34, 35) rather than alcohol 2, and the former
is unlikely to be reduced in food systems without the involve-
ment of an enzyme or some special reagents (e.g. metal
hydrides). Although the detailed mechanism in which â-dama-
scenone was produced from neoxanthin in the current study
remains to be elucidated, the experimental approved two
sequential phases allowing the following hypothesis (Figure
10). During phase 1, sporadic oxidation of the polyene back-
bone by hydroperoxides or peroxy acids may result in structures
such as epoxide 4, which might be further hydrolyzed by water
â
â
hypothetical stages along the pathway.
under the slight acidic environment to form the vicinal diol 5.
Phase 2 is carried out at relatively high temperatures to facilitate
the thermal cleavage of the rather weak C9-C10 bond of 5,
with subsequent disproportionation, resulting in the key-
compound 2 and the carbonyl residue 6 (of course, additionally
expected products are ketone 1 and the respective alcohol of
6). Finally, triol 2 may form â-damascenone as demonstrated
by Skouroumounis et al. (19). Evidently, in view of the low
yield of â-damascenone formation, the proposed mechanism is
not a major pathway of neoxanthin degradation under these
conditions. Peroxyacetic acid was selected as the oxidant since
it is a common reagent for obtaining epoxides from alkenes
and polyenes (36, 37) and it is readily soluble in aqueous
solutions compared with other peroxy acids such as m-
chloroperbenzoic acid (31). Preliminary experiments indicated
that hydrogen peroxide was inferior as 0.3 M of this hydro-
peroxide yielded approximately 1.0 µg/L â-damascenone (data
not shown). Since such concentrations of hydroperoxides are
unlikely to be found in any food product, and because of the
low yields which were produced, hydrogen peroxide was not
used further in this study. A possible source of hydroperoxides
in food systems are the hydroperoxides of fatty acids resulting
from lipoxygenase activity, mainly activated upon the first steps
of food processing, e.g. maceration or grinding of the intact
crop (38).
Conclusion. Since the original suggestion by Isoe et al. (17),
this study provides the evidence for the first time that 9′-cis-
neoxanthin can be oxidized and thermally degraded to â-dama-
scenone. Furthermore, this degradation occurred under rather
moderate pH and temperature values during phase 1 and a high
temperature in phase 2, without the involvement of enzymatic
activity under conditions which may occur in certain food
systems.

Direct Degradation of Neoxanthin to
â
-Damascenone
J. Agric. Food Chem., Vol. 53, No. 23, 2005 9205
ABBREVIATIONS USED
and the identification of a precursor of damascenone in these
fruits. Aust. J. Chem. 1989, 42, 2071-2084.
(15) Skouroumounis, G. K.; Winterhalter, P. Glycosidically bound
norisoprenoids from Vitis Vinifera cv. Riesling leaves. J. Agric.
Food Chem. 1994, 42, 1068-1072.
PRMN, prereaction mixture of neoxanthin; PAA, peroxyacetic
acid.
(16) Herderich, M.; Neubert, C.; Winterhalter, P.; Schreier, P.;
Skouroumounis, G. K. Identification of C13 norisoprenoid flavour
precursors in starfruit (AVerrhoa carambola L.). FlaVour Fra-
grance J. 1992, 7, 179-185.
(17) Isoe, S.; Katsumura, S.; Sakan, T. The synthesis of damascenone
and â-damascone and the possible mechanism of their forma-
tion from carotenoids. HelV. Chim. Acta 1973, 56, 1514-
1516.
ACKNOWLEDGMENT
We thank N. Watanabe for a gift of 9′-cis-neoxanthin marker,
G. Takeoka for a gift of â-damascenone marker, and G.
Skouroumounis for a gift of 3-OH-â-damascone marker. We
also wish to thank S. Kurasov for advising us on GC-MS
analyses, H. Keisar for assisting in the isolation of neoxanthin,
I. Peri for technical advice, and O. Tirosh for helpful sugges-
tions. The careful editing of the manuscript by C. Balcke is
highly appreciated.
(18) Edens, M.; Boerner, D.; Chase, C. R.; Nass, D.; Schiavelli, M.
D. The mechanism of the Meyer-Schuster rearrangement. J.
Org. Chem. 1977, 42, 3403-3408.
(19) Skouroumounis, G. K.; Massy-Westropp, R. A.; Sefton, M. A.;
Williams, P. J. Precursors of damascenone in fruit juices.
Tetrahedron Lett. 1992, 33, 3533-3536.
(20) Skouroumounis, G. K.; Massy-Westropp, R. A.; Sefton, M. A.;
Williams, P. J. â-Damascenone formation in juices and wines.
In Progress in FlaVour Precursor Studies; Schreier, P., Win-
terhalter, P., Eds.; Allured Publishing: Carol Stream, IL, 1993;
pp 275-278.
(21) Straubinger, M.; Knapp, H.; Oka, N.; Watanabe, N.; Winterhalter,
P. Isolation of a glucosidic â-damascenone precursor from rose
petals. J. Agric. Food Chem. 1997, 45, 4053-4056.
(22) Naiker, M. â-Damascenone-yielding precursor(s) from Cabernet
Sauvignon grapes. South Pac. J. Nat. Sci. 2001, 19, 11-17.
(23) Radtke, R.; Mohr, W.; Springer, R. Zur Kenntnis des Kaffee-
Aromas und seiner Vera¨nderungen im Laufe der Lagerung von
Ro¨stkaffee. III. Untersuchung der schwerflu¨chtigen Aroma-
verbindungen und der flu¨chtigen Sa¨uren. (On the knowledge of
coffee flavor and its changes during the storage of roasted coffee.
III. Analysis of the low-volatile aroma compounds and the
volatile acids.) Z. Lebensm. Unters. Forsch. 1966, 129, 349-
359.
(24) Holscher, W.; Steinhart, H. Formation pathways for primary
roasted coffee aroma compounds. In Thermally Generated
FlaVors: Maillard, MicrowaVe, and Extrusion Processes; Par-
liment, T. H., Morello, M. J., McGorrin, R. J., Eds.; ACS
Symposium Series 543; American Chemical Society: Washing-
ton, DC, 1994; pp 206-217.
(25) Baumeler, A.; Zerbe, O.; Kunz, R.; Eugster, C. H. (6R,9′Z)-
Neoxanthin: Synthese, Schmelzverhalten, Spektren und Kon-
formationsberechnungen. ((6R,9′Z)-Neoxanthin: Synthesis, melt-
ing conduct, spectra and calculation of conformations.) HelV.
Chim. Acta 1994, 77, 909-930.
(26) Rouseff, R.; Raley, L. Application of diode array detection with
a C-30 reversed phase column for the separation and identifica-
tion of saponified orange juice carotenoids. J. Agric. Food Chem.
1996, 44, 2176-2181.
(27) Ohloff, G.; Rautenstrauch, V.; Schulte-Elte, K. H. Model
reactions in the biosynthesis of compounds in the damascone
series and their preparative application. HelV. Chim. Acta 1973,
56, 1503-1513.
(28) Davies, B. H. Carotenoids. In Chemistry and Biochemistry of
Plant Pigments; Goodwin, T. W., Ed.; Academic Press: New
York, 1976; pp 38-155.
Supporting Information Available: Structure and NMR data
for 9′-cis-neoxanthin presented in Figure S1 and Tables S1 and
S2. This material is available free of charge via the Internet at
http://pubs.acs.org.
LITERATURE CITED
(1) Buttery, R. G.; Teranishi, R.; Ling, L. C.; Turnbaugh, J. G.
Quantitative and sensory studies on tomato paste volatiles. J.
Agric. Food Chem. 1990, 38, 336-340.
(2) Demole, E.; Enggist, P.; Sa¨uberli, U.; Stoll, M.; Kova`ts, E. S.
Structure and synthesis of damascenone (trimethyl-2,6,6-trans-
crotonoyl-1-cyclo-hexadiene-1, 3), odoriferous constituent of
Bulgarian rose oil (Rosa damascena). HelV. Chim. Acta 1970,
53, 541-551.
(3) Sen, A.; Laskawy, G.; Schieberle, P.; Grosch, W. Quantitative
determination of â-damascenone in foods using a stable iso-
tope dilution assay. J. Agric. Food Chem. 1991, 39, 757-
759.
(4) Czerny, M.; Grosch, W. Potent odorants of raw arabica coffee.
Their changes during roasting. J. Agric. Food Chem. 2000, 48,
868-872.
(5) Renold, W.; Na¨f-Mu¨ller, R.; Keller, U.; Wilhelm, B.; Ohloff,
G. Investigation of tea aroma. Part 1. New volatile black tea
constituents. HelV. Chim. Acta 1974, 57, 1301-1308.
(6) Kumazawa, K.; Masuda, H. Change in the flavor of black tea
drink during heat processing. J. Agric. Food Chem. 2001, 49,
3304-3309.
(7) Wahlberg, I.; Karlsson, K.; Austin, D. J.; Junker, N.; Roeraade,
J.; Enzell, C. R.; Johnson, W. H. Effects of flue-curing and aging
on the volatile, neutral and acidic constituents of Virginia
tobacco. Phytochemistry 1977, 16, 1217-1231.
(8) Buttery, R. G.; Teranishi, R.; Ling, L. C. Identification of
damascenone in tomato volatiles. Chem. Ind. 1988, 238.
(9) Buttery, R. G.; Teranishi, R.; Ling, L. C. Fresh tomato aroma
volatiles: A quantitative study. J. Agric. Food Chem. 1987, 35,
540-544.
(10) Chevance, F.; Guyot-Declerck, C.; Dupont, J.; Collin, S.
Investigation of the â-damascenone level in fresh and aged
commercial beers. J. Agric. Food Chem. 2002, 50, 3818-3821.
(11) Guido, L. F.; Carneiro, J. R.; Santos, J. R.; Almeida, P. J.;
Rodrigues, J. A.; Barros, A. A. Simultaneous determination of
E-2-nonenal and â-damascenone in beer by reversed-phase liquid
chromatography with UV detection. J. Chromatogr., A 2004,
1032, 17-22.
(29) Strand, A.; Kvernberg, K.; Karlsen, A. M.; Liaaen-Jensen, S.
Geometrical E/Z isomers of (6R)- and (6S)-neoxanthin and
biological implications. Biochem. Syst. Ecol. 2000, 28, 443-
456.
(12) Schreier, P.; Drawert, F.; Schmid, M. Changes in the composition
of neutral volatile components during the production of apple
brandy. J. Sci. Food Agric. 1978, 29, 728-736.
(30) Strand, A.; Liaaen-Jensen, S. First preparation and characteriza-
tion of the alleniccarotenoids (all-E,6S)- and (9′Z,6S)-neoxanthin.
J. Chem. Soc., Perkin Trans. 1 2000, 4, 595-598.
(31) Smith, M. B.; March, J. March’s AdVanced Organic Chemistry.
Reactions, Mechanisms, and Structure, 5th ed.; John Wiley &
Sons, Inc.: New York, 2001; pp 1396-1398.
(13) Roberts, D.; Morehai, A. P.; Acree, T. E. Detection and partial
characterization of eight â-damascenone precursors in apples
(Malus domestica Borkh. Cv. Empire). J. Agric. Food Chem.
1994, 42, 345-349.
(14) Sefton, M. A.; Skouroumounis, G. K.; Massy-Westropp, R. A.;
Williams, P. J. Norisoprenoids in Vitis Vinifera white wine grapes

9206 J. Agric. Food Chem., Vol. 53, No. 23, 2005
Bezman et al.
(32) Shen, B.; Jensen, R. G.; Bohnert, H. J. Mannitol protects against
oxidation by hydroxyl radicals. Plant Physiol. 1997, 115, 527-
532.
(33) Skouroumounis, J. K.; Sefton, M. A. Acid-catalyzed hydrolysis
of alcohols and their â-D-glucopyranosides. J. Agric. Food
Chem. 2000, 48, 2033-2039.
(37) Khursan, S. L.; Antonovsky, V. L. Molecular structure, confor-
mational mobility, vibrational spectra, and thermochemistry of
peroxyacetic acid: an ab initio and density functional study.
Russ. Chem. Bull. Int. Ed. 2003, 52, 1908-1919.
(38) Belitz, H. D.; Grosch, W. Food Chemistry, 2nd ed.; Berlin,
Springer Heidelberg, Germany, 1999; pp 152-236.
(34) Schwartz, S. H.; Qin, X.; Zeevaart, J. A. D. Characterization of
a novel carotenoid cleavage dioxygenase from plants. J. Biol.
Chem. 2001, 276, 25208-28211.
(35) Fleischmann, P.; Lutz-Ro¨der, A.; Winterhalter, P.; Watanabe,
N. Carotenoid cleavage enzymes in animals and plants. In
Carotenoid-DeriVed Aroma Compounds; Winterhalter, P., Rous-
eff, R. L., Eds.; ACS Symposium Series 802; American Chemical
Society: Washington, DC, 2002; pp 76-88.
Received for review June 6, 2005. Revised manuscript received
September 15, 2005. Accepted September 18, 2005. This study was
supported by the State of Lower-Saxony and the Volkswagen Founda-
tion, Hannover, Germany, and, in part, by Grant US-2914 from
BARD, The U.S.-Israel Binational Agricultural Research and De-
velopment Fund. Preliminary results of this study were presented at
the 229th ACS National Meeting, March 2005, San Diego, CA,
Abstract 76.
(36) Okieimen, F. E.; Bakare, O. I.; Okieimen, C. O. Studies on the
epoxidation of rubber seed oil. Ind. Crops Prod. 2002, 15, 139-
144.
JF051330B