Characterization of (E,E,Z)-2,4,6-nonatrienal as a character impact aroma compound of oat flakes

Article abstract of DOI:10.1021/jf051601i

To identify the compounds evoking the characteristic cereal-like, sweet aroma of oat flakes, an aroma extract dilution analysis (AEDA) was applied to a distillate prepared by solvent extraction/vacuum distillation from commercial oat flakes. Among the nine aroma-active compounds detected by gas chromatography-olfactometry and AEDA in the flavor dilution (FD) factor range of 4-1024, eight odorants, for example, (E)-β-damascenone, (Z)-3-hexenal, and butanoic acid, showed only low FD factors. However, one odorant eliciting the typical cereal, sweet aroma of the flakes was detected with the highest FD factor of 1024. By mass spectrometry and nuclear magnetic resonance measurements followed by a synthesis, (E,E,Z)-2,4,6-nonatrienal, exhibiting an intense oat flake-like odor at the extremely low odor threshold of 0.0002 ng/L in air, was identified as the key odorant of the flakes. By means of a newly developed stable isotope dilution analysis using synthesized, carbon-13-labeled nonatrienal as the internal standard, a concentration of 13 μg of (E,E,Z)-2,4,6-nonatrienal per kilogram of the flakes was measured. Model studies suggested linolenic acid as the precursor of nonatrienal in oats.

Full text of DOI:10.1021/jf051601i

J. Agric. Food Chem. 2005, 53, 8699−8705 8699
Characterization of (E,E,Z)-2,4,6-Nonatrienal as a Character
Impact Aroma Compound of Oat Flakes
CHRISTIAN SCHUH AND PETER SCHIEBERLE*
Deutsche Forschungsanstalt f u¨ r Lebensmittelchemie, Lichtenbergstrasse 4,
D-85748 Garching, Germany
To identify the compounds evoking the characteristic cereal-like, sweet aroma of oat flakes, an aroma
extract dilution analysis (AEDA) was applied to a distillate prepared by solvent extraction/vacuum
distillation from commercial oat flakes. Among the nine aroma-active compounds detected by gas
chromatography-olfactometry and AEDA in the flavor dilution (FD) factor range of 4-1024, eight
odorants, for example, (E)-â-damascenone, (Z)-3-hexenal, and butanoic acid, showed only low FD
factors. However, one odorant eliciting the typical cereal, sweet aroma of the flakes was detected
with the highest FD factor of 1024. By mass spectrometry and nuclear magnetic resonance
measurements followed by a synthesis, (E,E,Z)-2,4,6-nonatrienal, exhibiting an intense oat flake-like
odor at the extremely low odor threshold of 0.0002 ng/L in air, was identified as the key odorant of
the flakes. By means of a newly developed stable isotope dilution analysis using synthesized, carbon-
13-labeled nonatrienal as the internal standard, a concentration of 13 µg of (E,E,Z)-2,4,6-nonatrienal
per kilogram of the flakes was measured. Model studies suggested linolenic acid as the precursor of
nonatrienal in oats.
KEYWORDS: (E,E,Z)-2,4,6-Nonatrienal; (E,Z,E)-2,4,6-nonatrienal; precursor; linolenic acid; oats; stable
isotope dilution analysis; aroma extract dilution analysis
INTRODUCTION
food aroma (11-13). Such aroma-active components can be
separated from the bulk of odorless volatiles by screening
methods, such as the aroma extract dilution analysis (AEDA)
Cereal grains constitute the major part of human diets
throughout the world. Although the world production of oats is
minor as compared to that of wheat or rice, there is growing
interest in oats because of the higher content of, for example,
unsaturated lipids or vitamin B1 (1). However, because of the
comparatively higher activity of lipases and lipoxygenases in
oats, the kernels have to be thermally treated during manufac-
turing to prevent an enzymatic lipid peroxidation.
Heydanek and McGorrin (2) were among the first to
characterize volatiles inherent in dried oat groats prior to
processing. On the basis of vacuum distillation of 8 kg of
hydrated oat groats, the authors reported the identification of
about 100 volatile compounds. Further studies were focused
on the volatile composition of oils isolated from crude and
roasted oats (3, 4), oatmeal porridge (5), extrusion cooked oats
(13). However, applications of such dilution to odor threshold
methods on oat volatiles are scarcely available. In a study aimed
at identifying aroma compounds causing an off-odor in extruded
oat flour after storage, Guth and Grosch (14) previously reported
on vanillin, trans-4,5-epoxy-(E)-2-decenal, and (E,E)-2,4-deca-
dienal followed by (E,E)-2,4-nonadienal, trans-2,3-epoxyoctanal,
octanal, and hexanal as the most odor-active compounds among
the 25 aroma compounds detected in the freshly extruded
product. During storage of the dried extrudate, in particular,
hexanal and hexanoic acid were much increased and, therefore,
the authors suggested these compounds to mainly cause the
green, rancid off-note that had developed in the stored product.
Because, to our knowledge, no data are available on the aroma
compounds causing the characteristic, sweet, nutty, cereal-like
aroma of fresh oat flakes, the purpose of this study was to
identify these on the basis of an application of the AEDA.
(6-8), or rolled oats (9). Besides inactivating the enzymes, the
heat treatment is also considered to be necessary to develop
the typical nutty aroma of oat products (5), and Heydanek and
McGorrin (10) found an increase in typical Maillard-derived
compounds, such as pyrazines and furanones, after the heat
treatment of oats.
MATERIALS AND METHODS
It is widely accepted in modern aroma analysis that only those
volatiles present above their odor thresholds contribute to a given
Material. Three different batches of oat flakes (K o¨ lln, Elmshorn,
Germany) were purchased from a local supermarket.
Chemicals. Samples of the following pure reference chemicals were
obtained from the commercial sources given in parentheses: acetic acid,
(E,E)-2,4-heptadienal, 3-methylbutanoic acid, methyl octanoate, vanillin,
*
Corresponding author (telephone +49 89 289 141 70; fax +49 89 289
1
41 83; e-mail Peter.Schieberle@Lrz.tum.de).
1
0.1021/jf051601i CCC: $30.25
© 2005 American Chemical Society
Published on Web 09/28/2005

8700 J. Agric. Food Chem., Vol. 53, No. 22, 2005
Schuh and Schieberle
(E,E,Z)-2,4,6-Nonatrienal. (Z)-2-Penten-1-ol (1.5 mmol, 130 mg) and
Dess-Martin periodinane (2 mmol, 880 mg) were dissolved in
methylene chloride (10 mL), and the solution was stirred for 12 h at
room temperature. Then, the solvent was evaporated at 35 °C, the
residue was dissolved in pentane (2 mL), and the remaining residue
was filtered off. The solution containing the (Z)-2-pentenal was used
without further purification.
The further reaction steps were performed as described above for
(E,E,E)- and (E,Z,E)-2,4,6-nonatrienal (cf. Figure 1).
Preparative Isolation of the 2,4,6-Nonatrienal Isomers. The
mixture of isomers obtained in both synthetic routes was first purified
by means of column chromatography. For this purpose, the solution (1
mL) containing the nonatrienals was placed on the top of a glass column
(
30 × 1.5 cm) filled with a slurry of silica (modified according to the
method of ref 18) in n-pentane. After the column had been flushed
with n-pentane (100 mL), the target compounds were eluted with
n-pentane/diethyl ether (200 mL, 4:1, v/v).
The isomers were separated by means of preparative thin-layer
chromatography on silica gel (60 mesh, 20 × 20 cm, fluorescence
indicator F254, focusing zone of 20 × 2.5 cm, thickness ) 0.5 mm;
Merck, Darmstadt, Germany) using a mixture of n-pentane and diethyl
ether (9:1, v/v) as the mobile phase. The isomers were detected by UV
adsorption and collected separately by scraping off the respective areas
from the plate, followed by extraction with diethyl ether.
Figure 1. Synthetic route used in the synthesis of (E,E,E)- and (E,Z,E)-
,4,6-nonatrienal (DM, Dess Martin periodinane).
2
−
1
3
(
2
E)-2-pentenal, (Z)-2-penten-1-ol, (E,E)-2,4-heptadienal, [ C ]acet-
aldehyde, and (E,Z)-2,6-nonadienal (Aldrich, Steinheim, Germany);
butanoic acid and linolenic acid (Fluka, Neu-Ulm, Germany); 1-octen-
Quantitation of each isomer, collected in a volumetric flask of defined
volume, was performed by GC-FID using methyl octanoate as the
internal standard. The results were corrected by an FID response factor
obtained for a mixture of methyl octanoate and (E,Z)-2,6-nonadienal.
3-one, tris[2-(2-methoxyethoxy)ethyl]amine, methyl-4-(triphenylphos-
phonium)crotonate bromide, and Dess-Martin periodinane (Lancaster,
M u¨ hlheim, Germany). (E)-â-Damascenone was a gift from Symrise
(
Holzminden, Germany). All other chemicals were of analytical grade.
Z)-3-Hexenal was synthesized as previously described (15).
Syntheses. (E,E,E)-2,4,6-Nonatrienal. The synthesis was performed
[13
13
C
2
]-(E,E,E)-2,4,6-Nonatrienal. Synthesis of [ C
2
]-(E,E,E)-2,4,6-
13
(
nonatrienal ([ C
2
]-NT) was performed following closely the procedure
described for the unlabeled compound (16).
via an aldol condensation of (E,E)-2,4-heptadienal and acetaldehyde
following closely the method described by Buttery (16).
(E,E)-2,4-Heptadienal (0.6 mg, 6 mmol) was cooled at -5 °C in an
13
ice bath. After the addition of [ C ]acetaldehyde (0.55 mg, 12 mmol)
2
(
E,Z,E)-2,4,6-Nonatrienal. The synthetic procedure used for the
and potassium hydroxide (0.5 mL; 50%, v/v), the mixture was stirred
at -5 °C for 15 min and then for 1 h at room temperature. Diethyl
ether (100 mL) was added, and the mixture was washed with
hydrochloric acid (50 mL, 3 mol/L), followed by aqueous saturated
sodium bicarbonate solution (50 mL). Finally, the organic layer was
dried over anhydrous sodium sulfate. The concentration was determined
using methyl octanoate as the internal standard using the response factor
determined for the unlabeled compound.
Isolation of Volatiles from Oat Flakes. Oat flakes (50 g) were
powdered in liquid nitrogen using a Waring Blendor. The powder was
repeatedly extracted with diethyl ether (total extraction time ) 2 h;
total volume ) 300 mL). The combined extracts were filtered, dried
over anhydrous sodium sulfate, and concentrated to 100 mL by distilling
off the solvent at 40 °C using a Vigreux column (50 × 1 cm). To
remove the nonvolatile material, the extract was distilled at 40 °C using
the solvent-assisted flavor evaporation (SAFE) technique (19). Finally,
the distillate was concentrated to 0.5 mL at 40 °C by means of a Vigreux
column (50 × 1 cm), followed by microdistillation.
preparation of this nonatrienal isomer is displayed in Figure 1.
In a first step, a mixture of methyl (E,E,E)- and (E,Z,E)-2,4,6-
nonatrienoate was synthesized using a modified Wittig reaction and
following a procedure described for cinnamic acid derivatives (17).
Tris[2-(2-methoxyethoxy)ethyl]amine (TDA-1; 323 mg; 1 mmol), 80
mL of aqueous saturated sodium bicarbonate, and 80 mL of methylene
chloride were combined and intensely stirred for several minutes. Then,
methyl-4-(triphenylphosphonium)crotonate bromide (1.5 mmol, 662 mg)
and (E)-2-pentenal (1 mmol, 84 mg) were added, and the mixture was
stirred for 12 h at room temperature in the dark. The organic layer was
separated, and the aqueous phase was extracted twice with methylene
chloride (total volume ) 100 mL). The combined organic phases were
acidified using hydrochloric acid (75 mL, 10%), and the solution was
vigorously stirred for 12 h at room temperature in the dark. The organic
layer was separated, and the aqueous phase was extracted twice with
methylene chloride (100 mL). The combined organic layers were finally
dried over anhydrous sodium sulfate and then concentrated to 2 mL.
The solution was placed on the top of a water-cooled glass column
High-Resolution Gas Chromatography-Olfactometry (HRGC-
O); AEDA. HRGC was performed by means of a gas chromatograph
type 5160 (Fisons Instruments, Mainz, Germany) using the following
fused silica capillary columns: DB-5 (30 m × 0.32 mm, film thickness
) 0.25 µm) (J&W Scientific, Folsom, CA), DB-FFAP (30 m × 0.32
mm, film thickness ) 0.25 µm) (Phenomenex, Folsom, CA). The
samples were injected at 40 °C using the cold on-column technique.
The oven was held at 40 °C for 2 min, and the temperature was then
raised to 150 °C at 6 °C/min and then to 230 °C at 20 °C/min. For
HRGC-O, the effluent was split evenly between a flame ionization
detector (FID) and a sniffing port. Retention indices (RI) were calculated
from the retention times of n-alkanes by linear interpolation. For AEDA
(13, 14), extracts were diluted stepwise with diethyl ether (1:1, v/v)
and each dilution was analyzed by HRGC-O (injection volume ) 0.5
µL).
(
30 × 1.5 cm i.d.) filled with silica, modified according to the method
of Esterbauer (18), in n-pentane. After the column had been flushed
with n-pentane (50 mL), the target compound was eluted with
n-pentane/diethyl ether (8 + 2 by vol; 200 mL). The solvent was
evaporated at 35 °C to obtain a final volume of 10 mL.
The methyl esters obtained were then reduced into the corresponding
alcohols (E,E,E)- and (E,Z,E)-2,4,6-nonatrien-1-ol, using LiAlH
h of stirring at room temperature in the dark, a few drops of water
2
were added with continuous stirring. When the H production stopped,
sulfuric acid (10%, w/w) was added until the precipitate formed was
dissolved. The organic layer, containing the alcohols, was separated
and dried over anhydrous sodium sulfate.
In the third step (Figure 1), the alcohols were oxidized into a mixture
of (E,E,E)- and (E,Z,E)-2,4,6-nonatrienal by means of Dess-Martin
periodinane (1 mmol, 440 mg). After the addition of the periodinane,
the solution was stirred for 12 h at room temperature in the dark. The
solution was concentrated to 200 µL at 35 °C, the suspension was
extracted with n-pentane (2 mL) and filtered.
4
. After
1
Mass Spectrometry (MS). For compound identification, the effluent
of the GC column was introduced by means of the open split mode
into the mass spectrometer, and mass spectra were generated using the
MAT 95 S (Finnigan, Bremen, Germany) at 70 eV in the electronic

Aroma Compounds of Oat Flakes
J. Agric. Food Chem., Vol. 53, No. 22, 2005 8701
ionization mode (MS/EI) and at 115 eV in the chemical ionization mode
Table 1. Important Aroma Compounds (FD
Extract of Oatmeal Flakes
g
4) Identified in an
(MS/CI; reagent gas was isobutane).
Quantitation of 2,4,6-Nonatrienal Isomers in Oat Flakes. Quan-
titation was performed by means of a newly developed stable isotope
RI on
FFAP
13
13
odoranta
c
FDd
dilution assay. For this purpose, [ C
2
]-(E,E,E)-2,4,6-nonatrienal ([ C
2
]-
no.
odor quality
NT) was synthesized and, first an MS response factor of 0.99 was
determined from five different model solutions containing known, but
differing, amounts of the unlabeled and labeled nonatrienal. The ions
1
2
3
4
5
6
7
8
9
(Z)-3-hexenal
1-octen-3-one
acetic acid
butanoic acid
3-methylbutanoic acid
green, grassy
mushroom-like
sour
sweaty, rancid
sweaty
1148
1298
1430
1625
1663
1815
1869
1877
2605
8
4
8
8
4
13
2
m/z 137 (for 2,4,6-nonatrienal) and m/z 139 (for [ C ]-2,4,6-nonatrienal)
were used.
For quantitation, [ C
13
]NT (5 µg) was added to powdered oat flakes
200 g), and the volatiles and the labeled standard were isolated by
(E)-â
-damascenone
cooked apple
16
8
1024
4
2
unknown
unknown
vanillin
oatmeal-like, sweet
oatmeal-like, sweet
vanilla-like
(
extraction and SAFE distillation as described above. For two-
dimensional GC-MS a Mega 2 series gas chromatograph (Fisons
Instruments, Mainz-Kastel, Germany) was coupled to a Mega series
a
The compound was identified by comparing the mass spectra (MS/EI and
MS/CI) and retention indices on three columns of different polarities as well as the
odor quality and odor threshold with the respective data obtained from the reference
5
160 gas chromatograph (Carlo Erba, Hofheim, Germany) via the
MCCS system (Fisons Instruments). MS analysis was performed by
means of an ITD-800 mass spectrometer (Finnigan MAT; Bremen,
Germany), which was operated in the chemical ionization mode using
methanol as reactant gas. The separation of the distillate was ac-
complished by using a DB-5 (first oven) in combination with an FFAP
column (second oven), following the GC oven program outlined above.
The cut time intervals were determined by injection of the reference
compound prior to the analysis of the samples.
b
compound. Odor quality of the food constituent and the respective reference
compound as perceived at the sniffing port. Retention index. Flavor dilution factor.
c
d
Model Experiment on 2,4,6-Nonatrienal Formation from Lino-
lenic Acid. Linolenic acid (100 mg) was dissolved in 10 mL of water
using 5 mL of an aqueous solution of the emulsifier Tween 80 (0.001%)
and sodium hydroxide (1 mol/L; 0.1 mL). The solution was poured
into phosphate buffer (185 mL, 0.1 mol/L, pH 5.5), flushed with pure
oxygen for 5 min, and stirred for 14 h at 40 °C in a closed vessel.
After incubation, the volatiles were isolated by extracting the solution
twice with methylene chloride (50 mL) followed by SAFE distillation
of the dried extract.
Additionally, a blank sample using the same mixture, but without
incubation, was analyzed. The quantitation of nonatrienal was performed
as described above for oat flakes.
Proton Magnetic Resonance Spectrometry. NMR spectra were
3
recorded in CDCl solution by means of a Bruker AMX 400-III (Bruker,
Figure 2. HRGC separation of column fraction II isolated from a distillate
of oat flakes. Arrows indicate the retention indices of compounds 7 and
8
as well as of 8a showing identical mass spectra.
Rheinstetten, Germany; 400 MHz) spectrometer using tetramethylsilane
as the internal standard.
as odor-active constituents in oat flakes (Table 1). However,
for the two compounds eliciting oatmeal-like aromas (7 and
RESULTS AND DISCUSSION
8
), no reference compound was available and, also, no clear
Odorants in Oat Flakes. First, an overall sensory evaluation
of a distillate prepared by extraction of oat flakes with diethyl
ether followed by SAFE distillation was performed by 10
experienced panelists. On the basis of results obtained by
smelling an aliquot of a distillate on a strip of filter paper, they
characterized the aroma of the extract to be very close to the
characteristic cereal, nutty aroma of the flakes.
To characterize the volatile constituents responsible for the
overall odor impression, the distillate was separated by high-
resolution gas chromatography and the eluate was sniffed by
four assessors using an odor port (GC-O). The “free fatty acid
phase” (FFAP) was used to enable a separation of acidic and
neutral/basic volatiles in one GC run. GC-O revealed a total of
only nine odor-active areas; surprisingly, two of these clearly
revealed on odor quality similar to the overall aroma of the oat
flakes. By application of the AEDA, one of the two aroma
compounds was found to reveal by far the highest FD factor
among all odor-active areas detected.
A comparison of retention indices and odor qualities of the
odor-active areas detected with those of an in-house database
prepared by using reference odorants (20) suggested the
chemical structures for seven of the nine odor-active areas.
Finally, a comparison of mass spectrometric data obtained for
the oat constituent and the respective reference compound
confirmed (E)-â-damascenone, (Z)-3-hexenal, acetic acid, bu-
tanoic acid, 1-octen-3-one, vanillin, and 3-methylbutanoic acid
mass spectrum could be obtained in the extract prepared from
50 g of oat flakes, suggesting both compounds to be very odor-
active.
Thus, to obtain an unequivocal mass spectrum, the volatiles
from 1 kg of oat flakes were isolated, both odorants were
enriched by column chromatography on silica gel using an
n-pentane/diethyl ether gradient, and the odor-active region in
the gas chromatogram of the respective fraction was ac-
cumulated in a cooled trap by two-dimensional HRGC and by
repeatedly injecting the extract. In Figure 2, the gas chromato-
gram containing the oat-like smelling compound is shown. By
GC-O, compounds 7 and 8 were clearly located at retention
indices of 1869 and 1877. In both areas, the same mass spectrum
(MS/EI) was obtained (Figure 3A). In addition, a nearly
identical mass spectrum (Figure 3B) was obtained for a third
peak (compound 8a) eluting at RI 1900 (Figure 2), which,
however, did not have an aroma. In all three compounds, the
ions m/z 136 were confirmed as the molecular mass by MS/CI.
High-resolution mass spectrometry revealed the elemental
composition C9H12O, which indicated the presence of four
double-bond equivalents.
A thorough review of the literature identified a study by
Buttery (16) on volatiles of dry beans, in which the author
reported on (E,E,E)-2,4,6-nonatrienal as the first identification
of this volatile compound as a food constituent. The mass
spectrum published by Buttery (16) was in good agreement with

8702 J. Agric. Food Chem., Vol. 53, No. 22, 2005
Schuh and Schieberle
Figure 4. HRGC separation of (A) two isomers of 2,4,6-nonatrienal
prepared by starting from (E)-2-pentenal (cf. Figure 1) and (B) nonatrienal
isomers obtained in a synthesis starting from (Z)-2-pentenal instead of
(
E)-2-pentenal (cf. Figure 1).
Figure 3. Mass spectra (MS/EI) of (A) compounds 7 and 8 and (B)
compound 8a.
formation of a mixture of methyl (E,E,E)- and (E,Z,E)-2,4,6-
nonatrienoate, which, after reduction into the corresponding
alcohols followed by oxidation, should yield two isomers of
nonatrienal, namely, the E,E,E and E,Z,E isomers.
Synthesis following this sequence resulted in two peaks in a
1:1 ratio (Figure 4A). Whereas peak II agreed with (E,E,E)-
2,4,6-nonatrienal in its mass spectrum and retention index, peak
I showed the same retention index and mass spectrum as odorant
7 in oat flakes (Table 1), thus suggesting its structure to be
(E,Z,E)-2,4,6-nonatrienal.
the spectrum shown in Figure 3B. However, no data on the
aroma quality of the nonatrienal nor on its GC retention index
were published (16). Further to this study, another six publica-
tions had previously reported on the presence of 2,4,6-
nonatrienal in different foods, such as endive (21), lamb’s lettuce
(22), curuba fruit (23), orange juice (24), anchovies (25), and
spinach (26).
However, also in these studies no data on the correct structure
of the respective isomers were published, for example, by
performing a synthesis or NMR measurements, nor have any
data on the respective odor attributes been published yet.
Therefore, first, the (E,E,E)-2,4,6-nonatrienal was synthesized
following closely the procedure described by Buttery (16). The
compound obtained showed the same mass spectrum displayed
in Figure 3B, and a retention index of 1900 was determined
on the DB-FFAP column.
Because compounds 7 and 8 were eluted at retention indices
of 1869 and 1877, respectively, on this column (Figure 2), it
was shown that (E,E,E)-2,4,6-nonatrienal was present in oats,
but it could be excluded that this isomer contributed to the aroma
of the flakes. On the other hand, the data suggested that 7 and
For NMR measurements, both isomers were isolated by thin-
layer chromatography and were directly taken up in the solvent
used for NMR measurements. In a first experiment, an H,H-
homonuclear correlation spectroscopy was done with peak II
to establish the correct correlation of H atoms and the chemical
1
shifts. Then, H NMR spectra were measured, in particular, to
determine the coupling constants in the area between C-2 and
C-6.
1
In Table 2, the H NMR data obtained for peaks I and II are
1
contrasted. The H NMR data of peak II, the E,E,E isomer
(Table 2), showed coupling constants of 15.2 Hz (J2-3) and
14.8 Hz (J4-5), corroborating the trans configuration at the
respective double bonds at C-2 and C-4. In peak II, coupling
constants of 15.1 Hz (J2-3) and 11.1 Hz (J4-5) were measured,
allowing the conclusion that the double bond at C-2 is trans-
configured and that at C-4 cis-configured in this isomer. On
the basis of these data, peak I was assigned as (E,Z,E)-2,4,6-
nonatrienal. Because all data of peak I were in agreement with
those of compound 7, this was identified as (E,Z,E)-2,4,6-
nonatrienal. The most odor-active isomer, 8 (Table 1 and Figure
2), however, was still unknown.
Following the idea that 8 might be the E,E,Z isomer, (E)-2-
pentenal was substituted by (Z)-2-pentenal in the reaction
scheme shown in Figure 1. The synthetic route should then
result in a mixture of (E,E,Z)- and (E,Z,Z)-2,4,6-nonatrienal.
However, separation of the isomeric mixture of 2,4,6-
nonatrienals obtained showed four peaks with the (E,Z,E)- and
8
are possibly geometrical isomers of the E,E,E isomer, but
having at least one Z double bond in the molecule, because of
the lower retention index.
In total, eight isomers of 2,4,6-nonatrienal may occur.
Because, except for the E,E,E isomer, no data are available in
the literature on the other isomers, synthetic experiments were
performed. Because an E configuration in the 2-position might
be most probable due to the higher thermal stability of, for
example, (E)-2-alkenals, first, the synthesis of (E,Z,E)-2,4,6-
nonatrienal was attempted on the basis of the reaction sequence
shown in Figure 1.
As shown, a Wittig-type reaction of (E)-2-pentenal and
methyl-4-(triphenylphosphonium) crotonate should result in the

Aroma Compounds of Oat Flakes
J. Agric. Food Chem., Vol. 53, No. 22, 2005 8703
Table 2. 1H NMR Data of Synthesized Peaks I, II, and IV (cf. Figure
Table 3. Retention Indices, Odor Thresholds, and Odor Qualities of
2,4,6-Nonatrienal Isomers
4)
retention index on
odor thresholdb
isomera
FFAP
DB-5
OV-1701
(ng/L in air)
odor quality
E,Z,E
E,E,Z
E,E,E
1869
1877
1900
1271
1276
1286
1433
1440
1452
0.1
0.0002
4.4
oatmeal-like, sweet
oatmeal-like, sweet
oatmeal-like, sweet
a
Isomer of 2,4,6-nonatrienal. b Odor thresholds were determined as described
in ref 14.
chemical shift (coupling constants)
H at Ca
Sb
N (H)c
peak II
9.58 (8.0)
peak I
peak IV
9.55 (7.7)
1
2
3
4
5
8
9
(2)
d
1
3
1
1
1
2
3
9.65 (7.9)
, 6, 7
(2/4)
(3/5)
(4/6)
(7/9)
(8)
m
6.06 6.26
−
6.05
−6.25
6.03 6.23
−
dd
dd
dd
quint
t
7.13 (15.2/11.2)
6.38 (11.1/14.8)
6.68 (14.8/10.5)
2.22 (7.1/7.4)
1.08 (7.4)
7.60 (15.1/10.9)
6.43 (10.9/11.1)
6.66 (10.2/14.1)
2.28 (7.2/7.5)
1.15 (7.5)
7.17 (15.2/11.1)
6.43 (11.2/14.8)
6.95 (14.8/11.5)
2.30 (7.6/7.6)
1.05 (7.4)
a
Hydrogen at the numbered carbon atom of 2,4,6-nonatrienal; coupling constants
b
(
hertz) are given in parentheses. Multiplicity (d
double doublet, quint
)
doublet, m
)
multiplet, dd
)
c
)
quintet, t
)
triplet). Number of hydrogen atoms at the
respective carbon atom.
(E,E,E)-2,4,6-nonatrienal as the main constituents (Figure 4B).
Figure 5. Mass spectrum (MS/EI) of [13
2
C ]-(E,E,E)-2,4,6-nonatrienal.
In addition, two further isomers, characterized by the same mass
spectrum, were eluted at RI 1851 (peak III) and RI 1877 (peak
IV), the latter being identical in its retention index with the most
odor-active compound, 8, in oat flakes (Table 1) and, also,
showing the most intense odor among the four isomers during
GC-O analysis of the synthetic mixture. Because Daubresse et
al. (17) had already reported that the basic conditions used in
the Wittig reaction converted a major part of the Z isomers into
the more stable E isomers, this result was to be expected.
used suggested peak III in Figure 4 to be the (E,Z,Z)-2,4,6-
nonatrienal, the amounts available were not sufficient to
1
unequivocally confirm its structure by H NMR.
In Table 3, the retention indices and the odor thresholds
determined for the three isomers identified in oat flakes are
summarized. Although all three isomers elicited the same oat-
like, sweet aroma as evaluated by a panel of 10 trained panelists,
the odor thresholds determined in air showed significant
differences. Although already the odor threshold of the (E,E,E)-
By application of thin-layer chromatography, the (E,Z,E)-
2,4,6-nonatrienal (peak I in Figure 4B) as well as peak III could
2
,4,6-nonatrienal was quite low (4.4 ng/L in air), the threshold
of the E,Z,E isomer was by a factor of 40 lower. The (E,E,Z)-
,4,6-nonatrienal, however, showed the lowest threshold among
be separated from peak IV. However, it was impossible to
separate the (E,E,E)-2,4,6-nonatrienal (peak II) from peak IV.
In addition, also attempts to separate the isomers of the methyl
esters or the alcohols prior to the oxidation step (cf. Figure 1)
were unsuccessful (data not shown).
2
the three isomers investigated, being by a factor of 20 000 lower
as compared to the E,E,E isomer. Thus, this compound currently
belongs to the most odor-active aroma compounds found in food
so far. In higher concentrations, the odor attribute changed to a
fatty, deep-fried odor similar to that of (E,E)-2,4-decadienal.
Thus, the pleasant oat-flakes-like aroma might be overlooked
at the high concentrations present in the air during synthesis.
Quantitation of (E,E,Z)- and (E,E,E)-2,4,6-Nonatrienal in
Oat Flakes. Stable isotope dilution assays are an exact tool to
quantify trace odorants in foods (13). However, the isotopically
labeled internal standards are often commercially not available.
To develop a quantitation procedure, first, (E,E)-2,4-heptadienal
1
For this reason, the H NMR measurements were performed
on a mixture of peak IV and (E,E,E)-2,4,6-nonatrienal. The
signals obtained after subtracting the signals for the E,E,E isomer
are summarized in Table 2. The data confirmed a trans
configuration at C-2 and C-4, because coupling constants of
1
5.2 Hz (J3-2) and 14.8 Hz (J4-5) were measured for peak IV
(Table 2). Unfortunately, as for all isomers, it was not possible
to determine the coupling constant J6-7, because a multiplet
was always obtained in this area. However, because the chemical
shift of the H atom at C-5 differed significantly from the
chemical shift of the H atom at C-5 in the E,E,E isomer (Table
1
3
was reacted with [ C ]acetaldehyde in an Aldol-type reaction
2
13
to generate [ C ]-(E,E,E)-2,4,6-nonatrienal. The mass spectrum
2
2
), the double bond at C-6 must, consequently, have a cis
(Figure 5) confirms the incorporation of two carbon-13 atoms
configuration in peak IV. The same upfield shift was found for
the proton at carbon 3, in the (E,Z,E)-2,4,6-nonatrienal.
into the target compound. The isotopomer was purified by
chromatography, and a defined mixture of the unlabeled analyte
and the labeled standard were analyzed by two-dimensional
HRGC-MS as described above. From the mixtures, an MS
response factor of 0.99 was calculated.
By applying the newly developed stable isotope dilution assay
on three different batches of oat flakes, mean concentrations of
28 µg/kg of the (E,E,E)-2,4,6-nonatrienal and 13 µg/kg of the
Thus, compound 8 in oat flakes, agreeing in its retention index
with peak IV, was identified as (E,E,Z)-2,4,6-nonatrienal. The
elution order of the three isomers (Figure 2) on an unpolar
stationary GC phase, such as DB-5, clearly followed the well-
known rule for R,γ-unsaturated aldehydes with the E,E isomer
eluting later than the E,Z isomer. Although the synthetic pathway

8704 J. Agric. Food Chem., Vol. 53, No. 22, 2005
Schuh and Schieberle
Figure 6. Pathway suggested for nonatrienal formation from a 9,16-dihydroperoxide of linolenic acid.
A possible formation pathway may start from dihydroperoxy
fatty acids (Figure 6). Such peroxides have already been
reported by Neff et al. (30) and Grechkin et al. (31) using either
a photosentized peroxidation of methyl linolenate acid or an
enzymatic peroxidation of the free acid with potato lipoxyge-
nase. The formation could thus start from the 9,16-dihydro-
peroxy-10,12,14-octadecatrienoate (Figure 6). A proton attack
at the 9-hydroperoxy group, followed by water elimination in
a Hock-type reaction. may lead to an oxa-cation, which in turn
rearranges into a carbonium ion. Addition of water leads to a
semiketal, which is split into a triene-hydroperoxide. Enolization
and elimination of hydrogen peroxide finally result in 2,4,6-
nonatrienal.
Table 4. Amounts of (E,E,E)- and (E,E,Z)-2,4,6-Nonatrienal (NT)
Formed during Autoxidation of Linolenic Acid (LnA)
concn (
µ
g/100 mg of LnA)
(E,E,Z)-NT
expt
(E,E,E)-NT
1
2
3
4
54
83
0.3
0.2
30
52
<0.1
<0.1
a
a
b
b
a
Unstored, blank sample. b Below detection limit of 0.1
µg/100 mg of LnA.
(E,E,Z)-2,4,6-nonatrienal were determined. The analysis of both
compounds in three different batches of oat flakes did not differ
more than 15%. Other isomers were generally below the
detection limit (0.5 µg/kg, if 200 g of oat flakes is used for
extraction).
Because the E,E,Z isomer elicited the lowest odor threshold
among all isomers occurring in oats, the quantitative data
confirmed this isomer to be the character impact odorant of the
flakes.
Because the homolytic cleavage of hydroperoxides is also
discussed for hydroperoxide lyase reactions, it can thus be
explained that the nonatrienal can be formed either by an
autoxidation leading to dihydroperoxides followed by a hetero-
lytic cleavage of one hydroperoxy group or, alternatively, by
enzymatic degradation pathways.
Very recently, we were also able to identify (E,E,Z)-2,4,6-
nonatrienal as an important odorant in black tea (Schuh and
Schieberle, unpublished results), suggesting that this odorant
may occur ubiquitously in foods containing linolenic acid.
Studies on the Origin of 2,4,6-Nonatrienal. Buttery (16)
had already suggested linolenic acid as a precursor of (E,E,E)-
2
,4,6-nonatrienal, but he had indicated that common pathways
leading to such aldehydes, that is, a simple R-cleavage of
monohydroperoxides (27), would not be able to explain a third
double bond. In a study on the oxidation of linolenic acid by
soybean lipoxygenase (28) or, later, on potato and bean
lipoxygenases (29), Grosch and Laskawy reported the presence
of 2,4,6-nonatrienal using derivatization of the aldehyde with
LITERATURE CITED
(
(
(
1) Lokhart, H. B.; Hurt, H. D. Nutrition of oats. In Oats: Chemistry
and Technology; Webster, F. W., Ed.; AACC: St. Paul, MN,
1
986; pp 297-308.
2
,4-dinitrophenylhydrazine.
2) Heydanek, M. G.; McGorrin, R. J. Gas chromatography-mass
spectrometry investigations on the flavor chemistry of oat groats.
J. Agric. Food Chem. 1981, 29, 950-954.
3) Fors, S. M.; Eriksson, C. E. Characterization of oils extracted
from oats by supercritical carbon dioxide. Lebensm. Wiss.
Technol. 1990, 23, 390-395.
To investigate whether the nonatrienal isomers might also
be formed by an autoxidation of linolenic acid, the unsaturated
fatty acid was suspended in aqueous buffer at pH 5.5 and stirred
for 14 h at 40 °C. The isomers formed were first analyzed by
GC-O and then quantified by means of the stable isotope dilution
analysis. The quantitative results obtained (Table 4) clearly
indicated a significant increase in (E,E,E)- and (E,E,Z)-2,4,6-
nonatrienal after 14 h as compared to the very low amounts
already present in the linolenic acid before storage. As found
for oat flakes, also in the model experiment the E,E,E isomer
was higher in its concentration as compared to the E,E,Z isomer.
These data proved that nonatrienal can be formed not only by
an enzymatic degradation of linolenic acid but also by an
autoxidation.
Unsaturated aldehydes, such as 2,4-decadienal, are suggested
to be formed from unsaturated fatty acids via a cleavage of
alkoxy radicals formed by a homolytic cleavage of monohy-
droperoxides (27). However, up to now, no pathway has been
described leading to a conjugated triene system, as in 2,4,6-
nonatrienal.
(4) Fors, S. M.; Schlich, P. Flavor composition of oil obtained from
crude and roasted oats. In Thermal Generation of Aromas;
Parliament, T. H., McGorrin, R. J., Ho, C.-T., Eds.; American
Chemical Society: Washington, DC, 1989; pp 121-131.
(
5) Morello, M. Influence of preparation on the aroma compounds
of oatmeal porridge. In Food FlaVors: Formation, Analysis and
Packaging Influences; Coutis, E. T., Ho, C.-T., Mussinan, C. J.,
Parliament, T. H., Shahidi, F., Spanier, A. M., Eds.; Elsevier:
Amsterdam, The Netherlands, 1998; pp 415-422.
(
6) Sjoevall, O.; Lapuetelainen, A.; Johansson, A.; Kallio, H.
Analysis of volatiles formed during oxidation of extruded oats.
J. Agric. Food Chem. 1997, 45, 4452-4455.
(
7) Parker, J. K.; Hassell, G. M. E.; Mottram, D. S.; Guy, R. C. E.
Sensory and instrumental analysis of volatiles generated during
the extrusion cooking of oat flours. J. Agric. Food Chem. 2000,
48, 3497-3506.

Aroma Compounds of Oat Flakes
J. Agric. Food Chem., Vol. 53, No. 22, 2005 8705
(
8) Sides, A.; Robards, K.; Helliwell, S.; An, M. Changes in the
volatile profile of oats induced by processing. J. Agric. Food
Chem. 2001, 49, 2125-2130.
(21) G o¨ tz-Schmidt, E.; Schreier, P. Neutral volatiles from blended
endive (Cichorium endiVia L.). J. Agric. Food Chem. 1986, 34,
212-215.
(22) G o¨ tz-Schmidt, E.; Schreier, P. Volatile constituents of Valeri-
anella locusta. Phytochemistry 1988, 27, 845-848.
(23) Fr o¨ hlich, O.; Duque, C.; Schreier, P. Volatile constituents of
Curuba (Passiflora mollissima) fruit. J. Agric. Food Chem. 1989,
37, 195-201.
(24) N a¨ f, R.; Velluz, A. Volatile constitutents of blood and blond
orange juices: a comparison. J. Essent. Oil Res. 1996, 8, 587-
595.
(25) Triqui, R.; Guth, H. Determination of potent odorants in ripened
anchovy (Engraulis encrasicholus L.) by aroma extract dilution
analysis and by gas chromatography-olfactometry of headspace
samples. In FlaVor and Lipid Chemistry of Seafoods; Shahidi,
F., Cadwallader, K. R., Eds.; ACS Symposium Series 674;
American Chemical Society: Washington, DC, 1997; pp 31-
38.
(26) N a¨ f, R.; Velluz, A. The volatile constituents of extracts of cooked
spinach leaves (Spinacia oleracea L.). FlaVour Fragrance J.
2000, 15, 329-334.
(27) Belitz, H.-D.; Grosch, W.; Schieberle, P. Food Chem. 2004.
(28) Grosch, W.; Laskawy, G. Differences in the amount and range
of volatile carbonyl compounds formed by lipoxygenase isoen-
zymes from soybeans. J. Agric. Food Chem. 1975, 23, 791-
794.
(29) Grosch, W.; Laskawy, G.; Weber, F. Formation of volatile
carbonyl compounds and cooxidation of â-carotene by lipoxy-
genase from wheat, potato, flax and beans. J. Agric. Food Chem.
1976, 24, 456-459.
(30) Neff, W. E.; Frankel, E. N.; Weisleder, D. Photosensitized
oxidation of methyl linolenatessecondary products. Lipids 1982,
17, 780-790.
(
9) Zhou, M.; Robards, K.; Glennie-Holmes, M.; Helliwell, S.
Contribution of volatiles to the flavour of oatmeal. J. Sci. Food
Agric. 2000, 80, 247-254.
(
10) Heydanek, M. G.; McGorrin, R. J. Gas chromatography-mass
spectroscopy identification of volatiles from rancid oat groats.
J. Agric. Food Chem. 1981, 29, 1093-1095.
(
11) Acree, T. A. Bioasssays for flavor. In FlaVor SciencesSensible
Principles and Techniques; ACS Professional Reference Book;
Acree, T. A., Teranishi, R., Eds.; American Chemical Society:
Washington, DC, 1993; pp 1-20.
(
12) Grosch, W. Aroma compounds. In Handbook of Food Analysis,
2
nd ed., reVised and expanded, Physical Characterization and
Nutrient Analysis, Food Science and Technology; Nollet, L. M.
L., Ed.; 2004; pp 717-746.
(
13) Schieberle, P. New developments in methods for analysis of
volatile compounds and their precurors. In Characterization of
Food: Emerging Methods; Goankar, A. G., Ed.; Elsevier
Science: Amsterdam, The Netherlands, 1995; pp 403-431.
14) Guth, H.; Grosch, W. Odorants of extruded oat floursschanges
during storage. Z. Lebensm. Unters. Forsch. 1993, 196, 22-28
(
(
(
in German).
2
15) Fielder, S.; Rowan, D. D. The synthesis of 3,4- H
2
-3-Z-hexenal
-Z-hexenal. J. Labelled Compd. Radiopharm. 1995,
6, 465-470.
2
and 666- H
3
3
(
16) Buttery, R. G. Nona-2,4,6-trienal, an unusual component of
blended dry beans. J. Agric. Food Chem. 1975, 23, 1003-1004.
17) Daubresse, N.; Francesch, C. A mild synthesis of coumaryl,
conferyl, sinapyl aldehydes and alcohols. Synthesis 1994, 369-
(
3
71.
(
18) Esterbauer, H. On the autoxidation of methyl linoleate in water.
III. Isolation of the water soluble reaction products. Fette Seifen
Anstrichm. 1968, 70, 1-4 (in German).
(31) Grechkin, A. N.; Kuramshin, R. A.; Safonova, E. Y.; Yefremov,
Y. J.; Latypov, S. K.; Ilyasov, A. V.; Tarchevsky, I. A. Double
hydroperoxidation of R-linolenic acid by potato tuber lipoxy-
genase. Biochim. Biophys. Acta 1991, 1081, 79-84.
(
19) Engel, W.; Bahr, W.; Schieberle, P. Solvent assisted flavour
evaporationsa new and versatile technique for the carful and
direct isolation of aroma compounds from complex food
matrices. Eur. Food Res. Technol. 1999, 209, 237-241.
20) Rychlik, M.; Schieberle, P.; Grosch, W. Compilation of Odor
Thresholds, Odor Qualities and Retention Indices of Key Food
Odorants; Deutsche Forschungsanstalt f u¨ r Lebensmittelchemie,
Institut f u¨ r Lebensmittelchemie der Technischen Universit a¨ t
M u¨ nchen: Munich, Germany, 1998; ISBN 3-9803426-5-4.
(
Received for review July 6, 2005. Revised manuscript received August
9, 2005. Accepted August 22, 2005.
1
JF051601I