Antioxidation mechanism studies of caffeic acid: Identification of antioxidation products of methyl caffeate from lipid oxidation

Article abstract of DOI:10.1021/jf800781b

As a part of a research project on the elucidation of the chain-breaking antioxidation mechanism of natural phenols in food components, caffeic acid, a polyphenolic acid widely distributed in edible plants, was investigated. The identification and time course analysis of the antioxidation reaction products from methyl caffeate were carried out in the ethyl linoleate oxidation system. The antioxidation reaction produced a quinone derivative of methyl caffeate as an antioxidation product during the initial stage, which was identified by ¹³C NMR. The quinone, however, was not the final product, and a further reaction occurred to produce several new peroxides. The isolation and structure determination of the peroxides revealed that they had tricyclic structures, which consisted of ethyl linoleate, methyl caffeate, and molecular oxygen. On the basis of the formation pathway of these products, an antioxidation reaction mechanism of methyl caffeate, including the redox reaction of the caffeate and Diels-Alder reaction of the produced peroxides, was proposed.

Full text of DOI:10.1021/jf800781b

J. Agric. Food Chem. 2008, 56, 5947–5952 5947
Antioxidation Mechanism Studies of Caffeic Acid:
Identification of Antioxidation Products of Methyl
Caffeate from Lipid Oxidation
TOSHIYA MASUDA,*,† KAZUKI YAMADA,† JUN AKIYAMA,† TATSUSHI SOMEYA,†
YUKA ODAKA,† YOSHIO TAKEDA,† MOTOO TORI,‡ KATSUYUKI NAKASHIMA,‡
§
TOMOMI MAEKAWA,§ AND YOSHIAKI SONE
Faculty of Integrated Arts and Sciences, University of Tokushima, Tokushima 770-8502, Japan;
Faculty of Pharmaceutical Sciences, Tokushima Bunri University, Tokushima 770-8514, Japan; and
Graduate School of Human Life Science, Osaka City University, Osaka 558-8585, Japan
As a part of a research project on the elucidation of the chain-breaking antioxidation mechanism of
natural phenols in food components, caffeic acid, a polyphenolic acid widely distributed in edible
plants, was investigated. The identification and time course analysis of the antioxidation reaction
products from methyl caffeate were carried out in the ethyl linoleate oxidation system. The antioxidation
reaction produced a quinone derivative of methyl caffeate as an antioxidation product during the
initial stage, which was identified by 13C NMR. The quinone, however, was not the final product, and
a further reaction occurred to produce several new peroxides. The isolation and structure determination
of the peroxides revealed that they had tricyclic structures, which consisted of ethyl linoleate, methyl
caffeate, and molecular oxygen. On the basis of the formation pathway of these products, an
antioxidation reaction mechanism of methyl caffeate, including the redox reaction of the caffeate and
Diels-Alder reaction of the produced peroxides, was proposed.
KEYWORDS: Antioxidation mechanism; caffeic acid; methyl caffeate; o-quinone; Diels-Alder reaction
INTRODUCTION
2, radical termination stage
A• + X• f nonradical materials
Phenolic acids have received much attention as powerful
antioxidants to protect against the oxidative deterioration of food
components, such as polyunsaturated fatty acids (1). Recently,
some phenolic acids were found to be absorbed by the human
body through an active transport system in the intestine (2),
and then they act as antioxidants to prevent oxidative stress.
Caffeic acid is one of the most potently antioxidative phenolic
acids (3). It has o-diphenol and hydroxycinnamoyl structures,
both of which are responsible for its potent activity (4, 5). The
antioxidation mechanism of polyphenol is briefly understood
as a quinone formation from the dihydroxybenzene structure,
because dihydroxylbenzene is much more oxidizable than
biological materials (6). However, detailed insight about the
chain-breaking antioxidation, which polyphenols may demon-
strate, clarified that the antioxidation reaction is divided into
two stages (7):
(2)
S is the substance for oxidation, SsOO• is the peroxyl radical
of S, AH is the antioxidant, A• is the antioxidant radical, and
X• is another radical species including the same species as the
A radical. Although the first stage is a reversible process, the
second stage is irreversible and must produce stable radical
termination products. Structural information about these non-
radical products would afford important contributions to the
detailed antioxidation reaction mechanism (8–12). It should be
noted that most of the caffeic acid in edible plants exists as
esters of other plant constituents (13). For example, chlorogenic
acid in various fruits, coffee beans, and potatoes and rosmarinic
acid in the Labiatae herbs are famous antioxidative caffeic
esters (14–17). Therefore, methyl caffeate (1) (Figure 1) was
employed as the simplest caffeic ester, and the antioxidation
reaction mechanism of the methyl caffeate was investigated.
1, radical trapping stage
MATERIALS AND METHODS
SsOO• + AH f SsOOH + A•
(1)
Chemicals and Instruments. Methyl caffeate was synthesized from
commercially available caffeic acid (Nacalai Tesque, Kyoto, Japan).
2,2′-Azobis(2,4-dimethylvaleronitrile) (AMVN) was obtained from
Wako Pure Chemicals (Osaka, Japan). Ethyl linoleate was obtained
from Kanto Chemicals (Tokyo, Japan) and used after purification by
silica gel 60 (Merck, Darmstadt, Germany) chromatography developed
with 2.5% ethyl acetate in hexane. All solvents and other reagents were
* Author to whom correspondence should be addressed (fax +81-
88-656-7244; e-mail masuda@ias.tokushima-u.ac.jp).
† University of Tokushima.
‡ Tokushima Bunri University.
§ Osaka City University.
10.1021/jf800781b CCC: $40.75  2008 American Chemical Society
Published on Web 06/18/2008

5948 J. Agric. Food Chem., Vol. 56, No. 14, 2008
Masuda et al.
Figure 1. Chemical structures of methyl caffeate (1) and antioxidation reaction products 2, 3A, 3B, 4A, and 4B. (Tentative position numbering is given
on the basis of the numbering system of the starting caffeate and linoleate.)
also prepared in a similar manner with the addition of 48 µL of CH3CN
instead of the methyl caffeate solution. Both solutions were well stirred
and then incubated at 37 °C in air for 6 h with shaking (100 times/
min) by a water bath shaker. A 20 µL aliquot was removed from the
solution at 1 h intervals and diluted with 380 µL of methanol. Ten
microliters of the diluted solution was injected into the HPLC to analyze
the ethyl linoleate hydroperoxides using a 150 × 4.6 mm i.d. ODS-A
column (YMC, Tokyo, Japan) eluted with CH3CN/H2O (9:1, v/v) at
1.0 mL/min with detection at 234 nm. At the same intervals, an
additional 10 µL aliquot was removed from the reaction mixture and
injected into the HPLC to analyze the reaction products from the methyl
caffeate using a 150 × 4.6 mm i.d. Cosmosil 5SL-II column (Nacalai
Tesque) eluted with n-hexane (solvent A) and 1% AcOH in ethyl acetate
(solvent B), using linear gradient from 15% of solvent B to 50% of
solvent B for 25 min at 1.0 mL/min flow rate with detection at 280
nm.
Preparation and Identification of a Quinone Derivative of Methyl
Caffeate (2). Methyl caffeate (10 mg) and DPPH (41 mg) were
dissolved in acetone-d6 (0.6 mL). After mixing well, the 13C NMR of
the solution was measured. The 13C NMR (125 MHz, CDCl3) δ 181.0
Figure 2. Antioxidant activity of methyl caffeate against AMVN-induced
ethyl linoleate oxidation. (Concentration of reactants: 0.25 mM of methyl
caffeate, 50 mM of ethyl linoleate, and 15 mM of AMVN.)
(quinone CO), 180.6 (quinone CO), 166.4 (ester CO), 52.3 (CH3O),
and the other signals could not be assigned due to interference of the
signals of diphenylpicrylhydrazine. A small part of the solution was
withdrawn, diluted with CH3CN, and injected into the HPLC under
the analytical conditions mentioned above to identify the production
of peak 2 [UV (EtOAc/hexane) λmax 380, 295 nm].
obtained from Nacalai Tesque. The 1D NMR spectra were measured
with an EX-400 spectrometer (JEOL, Tokyo, Japan). The 2D NMR
spectra were measured with a Unity Plus 500 spectrometer (Varian,
Palo Alto, CA) using the manufacturer-supplied pulse sequences [1H,
13C, correlated spectroscopy (COSY), heteronuclear multiple quantum
coherence (HMQC), heteronuclear multiple bond correlation (HMBC),
and total correlation spectroscopy (TOCSY)]. Variable-temperature
NMR spectra were obtained by an ESP-400 spectrometer (JEOL). The
mass spectra were measured with an LCT Premier ESI-TOFMS
spectrometer (Waters Japan, Tokyo, Japan) in the positive ESI mode.
A PU-980 high-pressure gradient system (Jasco, Tokyo, Japan) equipped
with a photodiode array detector (SPD-M10AVP, Shimadzu, Kyoto,
Japan) was used for the analytical HPLC. A PU-980 pump equipped
with a UV-975 detector (Jasco) was used for the analysis of lipid
hydroperoxides. An LC-6AD recycle system (Shimadzu) equipped with
a UV-970 detector (JASCO) was used for preparative HPLC.
Isolation Procedure for the Coupling Products (3A, 3B, 4A, and
4B). To a 100 mL straight vial (40 mm, diameter) were added 480 µL
of 40 mM methyl caffeate (CH3CN solution), 400 µL of 0.3 M AMVN
(CH3CN solution), 560 µL of ethyl linoleate, and 6.6 mL of CH3CN.
The vial was incubated at 60 °C in the dark with shaking (100 times/
min) for 2 h. The combined reaction solution from 80 vials was cooled
at -30 °C to freeze the ethyl linoleate, and the produced supernatant
was collected. The precipitate was dissolved again in CH3CN (60 mL)
and then cooled at the same temperature to collect the next supernatant.
This procedure was repeated once more, and all supernatants were
combined and evaporated in vacuo. The residue was subjected to
column chromatography on BW-350 silica gel (Fuji Silysia Chemical,
Kasugai, Japan) developed with hexane and then hexane/ethyl acetate
(4:1) to remove the residual lipid and AMVN. The fraction, which
contained the coupling products, was evaporated and next purified by
preparative HPLC on a 250 × 10 mm i.d. Cosmosil 5SL-II column
eluted with hexane/ethyl acetate (85:15) at 6 mL/min with detection at
HPLC Analysis of Antioxidation Products from Methyl Caffeate
and Linoleate Hydroperoxide in the Oxidation System. To 35 µL
of ethyl linoleate in a 10 mL screw-capped tube were successively added
48 µL of a 10 mM methyl caffeate solution in CH3CN, 100 µL of 0.3
M AMVN in CH3CN, and 1.8 mL of CH3CN. The control tube was

Antioxidation Mechanism of Methyl Caffeate
J. Agric. Food Chem., Vol. 56, No. 14, 2008 5949
Figure 3. HPLC analytical data of antioxidation reaction mixture at 2 and 4 h.
280 nm to give peak compounds 3 (38 mg) and 4 (63 mg). The HPLC
purification using a reversed phase column of peak 3 was carried out
on a 250 × 20 mm i.d. Cosmosil 5C18-AR-II eluted with CH3CN/
H2O (75:25) at 5.5 mL/min, with detection at 280 nm to give
compounds 3A and 3B in 13.9 and 13.5 mg yields, respectively. Peak
4 was also purified by HPLC under the same conditions to give
compounds 4A and 4B in 7.3 and 6.3 mg yields, respectively.
3B: ESI-TOFMS (m/z) [M + H]+ calcd for C30H45O8, 533.3114;
found, 533.3103; IR (film) ν cm-1 3368 (OH), 2928 (CH), 1719 (CO);
1H NMR (500 MHz, CDCl3) δ 6.15 (d, J ) 16.0 Hz, 1H, H-2), 7.39
(d, J ) 16.0 Hz, 1H, H-3), 3.24 (dd, J ) 3.1 and 2.0 Hz, 1H, H-2′),
3.38 (dd, J ) 6.7 and 3.1 Hz, 1H, H-5′), 6.44 (dd, J ) 6.7 and 2.0 Hz,
1H, H-6′), 2.28 (t, J ) 7.5 Hz, 2H, H-2′′), 1.60 (br q, J ) 6.9 Hz, 2H,
H-3′′), 1.26-1.35 (complex, 10H, H-4′′, H-5′′, H-6′′, H-16′′, and
H-17′′), 1.40 (complex, 2H, H-7′′), 1.54 (m, 2H, H-8′′), 4.36 (br t, J )
6.3 Hz, 1H, H-9′′), 2.00 (dt, J ) 3.1 and 1.0 Hz, 1H, H-10′′), 2.90 (dt,
J ) 8.3 and 3.1 Hz, 1H, H-11′′), 5.22 (br dd, J)15.0 and 8.3 Hz, 1H,
H-12′′), 5.48 (dt, J ) 15.0 and 7.0 Hz, H-13′′), 1.95 (br q, J ) 7.0 Hz,
1H, H-14′′), 1.40 (complex, 1H, H-14′′), 1.20 (m, 2H, H-15′′), 0.88
(br t, J ) 6.9 Hz, 3H, H-18′′), 4.12 (q, J ) 6.9 Hz, 2H, H-1′′′), 1.25
(t, J ) 6.9 Hz, 3H, H-2′′′), 3.77 (s, 3H, 1-OCH3); 13C NMR (125 MHz,
CDCl3) δ 167.2 (C-1), 118.5 (C-2), 141.1 (C-3 or C-1′), 141.0 (C-1′
or C-3), 42.8 (C-2′), 92.7 (C-3′), 204.6 (C-4′), 54.6 (C-5′), 132.3 (C-
6′), 173.9 (C-1′′), 34.3 (C-2′′), 24.9 (C-3′′), 29.2, 29.0, 28.9, 28.9 (C-
4′′, C-5′′, C-6′′, and C-15′′), 25.7 (C-7′′), 29.9 (C-8′′), 85.0 (C-9′′),
41.3 (C-10′′), 40.5 (C-11′′), 130.3 (C-12′′), 132.8 (C-13′′), 32.3 (C-
14′′), 31.3 (C-16′′), 22.4 (C-17′′), 14.1 (C-18′′), 60.2 (C-1′′′), 14.2 (C-
1′′′), 51.9 (1-OCH3).
Analytical Data for the Coupling Products (3A, 3B, 4A, and 4B).
3A: ESI-TOFMS (m/z) [M + H]+ calcd for C30H45O8, 533.3114; found,
533.3088; IR (film) ν cm-1 3379 (OH), 2928 (CH), 1718 (CO); 1H
NMR (500 MHz, CDCl3) δ 6.15(d, J ) 15.8 Hz, 1H, H-2), 7.40 (d, J
) 15.8 Hz, 1H, H-3), 3.24 (dd, J ) 3.3 and 2.0 Hz, 1H, H-2′), 3.37
(dd, J ) 6.5 and 3.3 Hz, 1H, H-5′), 6.44 (dd, J ) 6.5 and 2.0 Hz, 1H,
H-6′), 2.28 (t, J ) 7.5 Hz, 2H, H-2′′), 1.61 (br q, J ) 7.5 Hz, 2H,
H-3′′), 1.24-1.34 (complex, 14H, H-4′′, H-5′′, H-6′′, H-7′′, H-15′′,
H-16′′, and H-17′′), 1.95 (br q, J ) 7.0 Hz, 2H, H-8′′), 5.48 (dt, J )
15.0 and 7.0 Hz, 1H, H-9′′), 5.22 (br dd, J ) 15.0 and 9.0 Hz, 1H,
H-10′′), 2.90 (dt, J ) 9.0 and 3.3 Hz, 1H, H-11′′), 2.01 (dt, J ) 3.3
and 1.0 Hz, 1H, H-12′′), 4.37 (br t, J ) 6.0 Hz, 1H, H-13′′), 1.40
(complex, 1H, H-14′′), 1.53 (m, 1H, H-14′′), 0.88 (br t, J ) 7.0 Hz,
3H, H-18′′), 4.12 (q, J ) 7.5 Hz, 2H, H-1′′′), 1.25 (t, J ) 7.5 Hz, 3H,
H-2′′′), 3.78 (s, 3H, 1-OCH3); 13C NMR (125 MHz, CDCl3 δ 167.2
(C-1), 118.5 (C-2), 141.0 (C-3 or C-1′), 141.1 (C-3 or C-1′), 42.8 (C-
2′), 92.7 (C-3′), 204.6 (C-4′), 54.6 (C-5′), 132.3 (C-6′), 173.9 (C-1′′),
34.3 (C-2′′), 24.9 (C-3′′), 29.1, 29.1, 29.0, 28.9 (C-4′′, C-5′′, C-6′′,
and C-7′′), 32.3 (C-8′′), 132.6 (C-9′′), 130.4 (C-10′′), 40.5 (C-11′′),
41.4 (C-12′′), 85.1 (C-13′′), 29.8 (C-14′′), 25.4 (C-15′′), 31.6 (C-16′′),
22.4 (C-17′′), 14.0 (C-18′′), 60.2 (C-1′′′), 14.3 (C-1′′′), 51.9 (1-OCH3).
4A: ESI-TOFMS (m/z) [M + H]+ calcd for C30H45O8, 533.3114;
found, 533.3165; IR (film) ν cm-1 3356 (OH), 2928 (CH), 1732 (CO);
1H NMR (500 MHz, CDCl3) δ 6.16 (d, J ) 15.8 Hz, 1H, H-2), 7.38
(d, J ) 15.8 Hz, 1H, H-3), 3.45 (br t, J ) 2.5 Hz, 1H, H-2′), 3.42 (dd,
J ) 6.5 and 2.5 Hz, 1H, H-5′), 6.47 (dd, J ) 6.5 and 1.5 Hz, 1H,
H-6′), 2.28 (t, J ) 7.5 Hz, 2H, H-2′′), 1.61 (q, J ) 7.5 Hz, 2H, H-3′′),
1.24-1.38 (complex, 16H, H-4′′, H-5′′, H-6′′, H-7′′, H-14′′, H-15′′,

5950 J. Agric. Food Chem., Vol. 56, No. 14, 2008
Masuda et al.
Figure 4. Time course analytical data of methyl caffeate and peaks 2-4 from the antioxidation reaction for 5 h.
J ) 7.0 Hz, 2H, H-1′′′), 1.25 (t, J ) 6.9 Hz, 2H, H-2′′′), 3.78 (s, 3H,
1-OCH3); 13C NMR (125 MHz, CDCl3) δ 167.1 (C-1), 118.7 (C-2),
141.9 (C-3 or C-1′), 140.0 (C-1′ or C-3), 37.2 (C-2′), 94.0 (C-3′), 54.9
(C-5′), 132.7 (C-6′), 173.9 (C-1′′), 34.3 (C-2′′), 24.9 (C-3′′), 29.2, 29.1,
29.0, 28.8 (C-4′′, C-5′′, C-6′′, and C-15′′), 25.5 (C-7′′), 30.3 (C-8′′),
85.8 (C-9′′), 42.8 (C-10′′), 47.7 (C-11′′), 130.0 (C-12′′), 132.9 (C-13′′),
32.3 (C-14′′), 31.3 (C-16′′), 22.5 (C-17′′), 14.1 (C-18′′), 60.2 (C-1′′′),
14.3 (C-2′′′), 51.9 (1-OCH3).
RESULTS AND DISCUSSION
Antioxidation Reaction of Methyl Caffeate. The antioxidant
reaction conditions were designed by using freshly purified ethyl
linoleate (50 mM) as the oxidation substrate and AMVN (15
mM) as the radical oxidation initiator. The antioxidant activity
of 0.25 mM methyl caffeate is shown in Figure 2. Methyl
caffeate shows a very strong antioxidant activty for the initial
2 h; then the activity potency gradually decreased, and from
4 h, no activity was observed. Figure 3 shows the HPLC
analysis of the reaction mixture at 2 and 4 h. For the 2 h data,
new peaks at retention times of 16 min (peak 2) and 8 min
(peak 3) were observed besides the peak of methyl caffeate (1)
at 15 min. On the other hand, the methyl caffeate peak and
peak 2 disappeared and peak 4 was observed in the 4 h data. A
time course analysis of peaks 2-4 and the methyl caffeate peak
(1) was carried out, and these data are shown in Figure 4. The
concentration of methyl caffeate continuously decreased during
the 4 h antioxidation period. Peaks 3 and 4 continuously
increased during this period; however, peak 2 increased for the
initial 1 h and then decreased up to 4 h, suggesting that the
product corresponding to peak 2 was an intermediate for
the other antioxidation products and the peak 3 and 4 compounds
were stable antioxidation reaction products.
Figure 5. 13C NMR data of (A) methyl caffeate and (B) DPPH reaction
with methyl caffeate.
H-16′′, and H-17′′), 1.95 (br q, J ) 7.0 Hz, 1H, H-8′′), 5.49 (dt, J )
15.0 and 7.0 Hz, 1H, H-9′′), 5.16 (dd, J ) 15.0 and 8.5 Hz, 1H, H-10′′),
2.89 (br d, J ) 8.5 Hz, 1H, H-11′′), 1.83 (br s, 1H, H-12′′), 0.90 (t, J
) 6.5 Hz, 3H, H-18′′), 4.12 (q, J ) 7.0 Hz, 2H, H-1′′′), 1.25 (t, J )
7.0 Hz, 3H, H-2′′′), 3.79 (s, 3H, 1-OCH3); 13C NMR (125 MHz, CDCl3)
δ 167.1 (C-1), 118.7 (C-2), 141.0 (C-3 or C-1′), 140.9 (C-1′ or C-3),
37.1 (C-2′), 94.3 (C-3′), 54.9 (C-5′), 132.6 (C-6′), 173.9 (C-1′′), 34.3
(C-2′′), 24.9 (C-3′′), 29.2, 29.1, 29.0, 28.8 (C-4′′, C-5′′, C-6′′, and C-7′′),
32.3 (C-8′′),132.8 (C-9′′), 130.2 (C-10′′), 47.7 (C-11′′), 42.8 (C-12′′),
86.0 (C-13′′), 30.2 (C-14′′), 25.3 (C-15′′), 31.5 (C-16′′), 22.5 (C-17′′),
14.0 (C-18′′), 60.2 (C-1′′′), 14.3 (C-2′′′), 51.9 (1-OCH3).
4B: ESI-TOFMS (m/z) [M + H]+ calcd for C30H45O8, 533.3114;
found, 533.3145; IR (film) ν cm-1 3363 (OH), 2927 (CH), 1734 (CO);
1H NMR (500 MHz, CDCl3) δ 6.15 (d, J ) 15.8 Hz, 1H, H-2), 7.38
(d, J ) 15.8 Hz, 1H, H-3), 3.44 (br t, J ) 2.5 Hz, 1H, H-2′), 3.42 (dd,
J ) 6.5 and 2.5 Hz, 1H, H-5′), 6.46 (dd, J ) 6.5 and 2.5 Hz, 1H,
H-6′), 2.29 (t, J ) 7.5 Hz, 2H, H-2′′), 1.62 (br t, J ) 7.0 Hz, 2H,
H-3′′), 1.21-1.31 (complex, 12H, H-4′′, H-5′′, H-6′′, H-15′′, H-16′′,
and H-17′′), 1.32 (m, 2H, H-7′′), 1.53 (m, 2H, H-8′′), 1.83 (br s, 1H,
H-10′′), 2.89 (br d, J ) 8.8 Hz, 1H, H-11′′), 5.16 (dd, J ) 15.0 and
8.8 Hz, 1H, H-12′′), 5.49 (dt, J ) 15.0 and 7.0 Hz, 1H, H-13′′), 1.95
(q, J ) 7.0 Hz, 2H, H-14′′), 0.88 (t, J ) 6.9 Hz, 3H, H-18′′), 4.12 (q,
Identification of Peak 2 Compound. To identify the
produced peak compounds, we attempted the isolation of each
peak compound from the reaction mixture. Unfortunately, the
isolation of the peak 2 compound failed due to its instability.
The photodiode array detection of peak 2 gave a bathochromic
shifted UV spectrum (λmax 380 nm), which indicated that 2 was

Antioxidation Mechanism of Methyl Caffeate
J. Agric. Food Chem., Vol. 56, No. 14, 2008 5951
Figure 6. Stereostructures around dioxacyclohexane ring of 3A and 3B. (Dashed arrows indicate NOE correlations observed in NOESY.)
Figure 7. Scheme for the proposed mechanism of antioxidation reaction of methyl caffeate in ethyl linoleate.
a quinone derivative of methyl caffeate. Sawai and co-workers
(18, 19) reported the identification of the quinone derivatives
of several polyphenols, which were produced by DPPH oxida-
tion, using 13C NMR without purification. The 13C NMR
spectrum of the DPPH-treated methyl caffeate, which was
obtained by Sawai’s procedure, revealed two quinone carbonyl
signals at 181.0 and 180.6 ppm (Figure 5). This DPPH oxidation
product was identical to the peak 2 compound by HPLC
analysis. Therefore, the peak 2 compound should be the
o-quinone of methyl caffeate (2) as depicted in Figure 1.
Identification of Peak Compounds 3 and 4. To clarify the
peak 3 and 4 compounds, we isolated them and determined their
chemical structures. Peaks 3 and 4 were not observed in the
reaction without the lipid; therefore, these compounds might
be coupling products with ethyl linoleate, not dimeric products
previously observed during the oxidation of caffeic acid (20, 21).
The antioxidation reaction of methyl caffeate in the presence
of a large amount of ethyl linoleate was carried out, and the
peak 3 and 4 compounds were isolated from the reaction
mixture. The final purification of peaks 3 and 4 using reversed
phase HPLC afforded compounds 3A and 3B from peak 3 and
compounds 4A and 4B from peak 4.
part of ethyl linoleate had disappeared, indicating that the double
bond of ethyl linoleate reacted with the aromatic part of methyl
caffeate. These NMR data were very similar to those of the
tricyclic coupling product of methyl ferulate and ethyl linoleate
peroxide, which was previously reported as an antioxidation
product from methyl ferulate (12). The structures of 3A and
3B were confirmed by the 2D NMR techniques (COSY,
NOESY, HMQC, and HMBC). The distinction between the
structures of 3A and 3B was carried out by TOCSY correlations
(H-13′′-H-18′′ and H-2′′-H-8′′ in 3A, H-2′′-H-9′′ and
H-18′′-H-14′′ in 3B) to determine the structures 3A and 3B as
depicted in Figure 1.
Compound 4A was also isolated as a colorless oil, and its
molecular formula was determined to be C30H44O8 from the ESI-
MS. All of the spectroscopic data were very similar to those of
3A, indicating that 4A was one of the stereoisomers of 3A. Very
interestingly, a carbon signal (C-4′) and a proton signal (H-
13′′) disappeared and four carbon signals (C-3′, C-12′′, C-13′′,
and C-14′′) were broadened in the NMR spectra. We presumed
that various conformational isomers existed in 4A and exchange
between the conformers hid and broadened the NMR signals.
The NMR measurement of 4A at elevated temperature failed
because 4A decomposed at 80 °C; however, in the low-
temperature experiment (-50 °C), several conformers of 4A
Compounds 3A and 3B were isolated as colorless oils. Their
molecular formulas were estimated to both be C30H44O8 from
the ESI-MS results. The molecular formula indicated that 3A
and 3B were oxidative coupling products consisting of methyl
caffeate, ethyl linoleate, and molecular oxygen. In the 1H NMR
of 3A and 3B, both signal sets due to the original methyl caffeate
and ethyl linoleate were observed; however, several signals
assignable to the aromatic part of methyl caffeate and the olefinic
1
were observed in its H NMR and the carbonyl signal at the
4′-position of one of the conformers was observed at 205.5 ppm
in its 13C NMR. The NOE correlations (H-2′-H-14′′), which
were observed in the NOESY of 4A, indicated that a bridge-
head proton (H-2′) and an alkyl group at the 13′′-position should
have the same orientation as illustrated on the right side of

5952 J. Agric. Food Chem., Vol. 56, No. 14, 2008
Masuda et al.
(4) Chen, J. H.; Ho, C.-T. Antioxidant activities of caffeic acid and
its related hydroxycinnamic acid compounds. J. Agric. Food
Chem. 1997, 45, 2374–2378.
(5) Moon, J.-H.; Terao, J. Antioxidant activity of caffeic acid and
dihydrocaffeic acid in lard and human low-density lipoprotein. J.
Agric. Food Chem. 1998, 46, 5062–5065.
Figure 6. As shown in Figure 6, the axial alkyl group at the
13′′-position labilized the chair conformation of the dioxacy-
clohexane ring and generated several conformers that interfered
with the NMR signal assignment of 4A. Compound 4B has the
same molecular formula as 4A and spectroscopic data very
similar to those of 4A. Only the TOCSY data were different
from those of 4A. In the TOCSY spectra of 4A and 4B,
correlations were observed from H-2′′ to H-8′′ and H-9′′ in 4A
and from H-2′′ to H-8′′ and H-10′′ in 4B. These data indicated
that 4A and 4B have the structures 4A and 4B as depicted in
Figure 1.
(6) Shahidi, F.; Wanasundara, P. K. J. P. D. Phenolic antioxidants.
Crit. ReV. Food Sci. Nutr. 1992, 32, 67–103.
(7) Frankel, N. F. Lipid Oxidation, 2nd ed.; The Oily Press:
Bridgwater, U.K., 2005; pp 209-212.
(8) Masuda, T.; Hidaka, K.; Shinohara, A.; Maekawa, T.; Takeda,
Y.; Yamaguchi, H. Chemical studies on antioxidant mechanism
of curcuminoids: analysis of radical reaction products from
curcumin. J. Agric. Food Chem. 1999, 47, 71–77.
(9) Masuda, T.; Inaba, Y.; Takeda, Y. Antioxidant mechanism of
carnosic acid: structural identification of two oxidation products.
J. Agric. Food Chem. 2001, 49, 5560–5565.
(10) Masuda, T.; Maekawa, T.; Hidaka, K.; Bando, H.; Takada, Y.;
Yamaguchi, H. Chemical studies on antioxidant mechanism of
curcumin: analysis of oxidative coupling produces from curcumin
and linoleate. J. Agric. Food Chem. 2001, 49, 2539–2547.
(11) Masuda, T.; Yamada, K.; Maekawa, T.; Takeda, Y.; Yamaguchi,
H. Antioxidant mechanism studies on ferulic acid, isolation and
structure identification of the main antioxidation product from
methyl ferulate. Food Sci. Technol. Res. 2006, 12, 178–182.
(12) Masuda, T.; Yamada, K.; Maekawa, T.; Takeda, Y.; Yamaguchi,
H. Antioxidant mechanism studies on ferulic acid, identification
of oxidative coupling products from methyl ferulate and linoleate.
J. Agric. Food Chem. 2006, 54, 6069–6074.
(13) Hermann, K. Occurrence and content of hydroxycinnamic and
hydroxybenzoic acid compounds in foods. Crit. ReV. Food Sci.
Nutr. 1989, 28, 315–347.
(14) Manach, C.; Scalbert, A.; Morand, C.; Remesy, C.; Jimenez, L.
Polyphenols: food sources and bioavailability. Am. J. Clin. Nutr.
2004, 79, 727–747.
(15) Ky, C.-K.; Noirot, M.; Hamon, S. Comparison of five purification
methods for chlorogenic acids in green coffee beans (Coffea sp.).
J. Agric. Food Chem. 1997, 45, 786–790.
(16) Friedman, M. Chemistry, biochemistry, and dietary role of potato
polyphenols. A review. J. Agric. Food Chem. 1997, 45, 1523–
1540.
(17) Nakatani, N. Antioxidants from spices and herbs. In Natural
Antioxidants Chemistry, Health Effects, and Applications; Shahidi,
F. Ed.; AOCS Press: Champaign, IL, 1996; pp 64-75.
(18) Sawai, Y.; Sakata, K. NMR analytical approach to clarify the
antioxidative molecular mechanism of catechins using 1,1-
diphenyl-2-picrylhydrazyl. J. Agric. Food Chem. 1998, 46, 111–
114.
(19) Sawai, Y.; Moon, J.-H.; Sakata, K.; Watanabe, N. Effect of
structures on radical-scavenging abilities and antioxidative activi-
ties of tea polyphenols: NMR analytical approach using 1,1-
diphenyl-2-picrylhydrazyl radicals. J. Agric. Food Chem. 2005,
53, 3598–3604.
(20) Cilliers, J. J. L.; Singleton, V. L. Characterization of the products
of nonenzymic autoxidative phenolic reactions in a caffeic acid
model system. J. Agric. Food Chem. 1991, 39, 1298–1303.
(21) Tazaki, H.; Kawabata, J.; Fujita, T. Novel oxidative dimer from
caffeic acid. Biosci., Biotechnol., Biochem. 2003, 67, 1185–1187.
(22) Tazaki, H.; Taguchi, D.; Hayashida, T.; Nabeta, K. Stable isotope-
labeling studies on the oxidative coupling of caffeic acid via
o-quinone. Biosci., Biotechnol., Biochem. 2001, 65, 2613–2621.
Proposed Antioxidation Reaction Mechanism of Methyl
Caffeate in Linoleate. Potent antioxidatively active polyphe-
nols, which bear the ortho- or para-substituted diphenol structure,
have been believed to afford their quinone derivatives as the
antioxidation products because the diphenol structure is easily
oxidizable to a quinone. In this investigation, we found that
the quinone is not the final antioxidation product of methyl
caffeate. From our elucidation of the chemical structures of the
five compounds including the o-quinone of methyl caffeate, we
propose the antioxidation reaction mechanism of methyl caffeate
in ethyl linoleate as illustrated in Figure 7. As shown in Figure
7, methyl caffeate traps two radicals at both the phenolic groups
to produce the o-quinone (2). Tazaki et al. (22) revealed that
the o-quinone of caffeic acid underwent a redox reaction with
caffeic acid. Therefore, the o-quinone (2) is unstable in the
presence of methyl caffeate to produce a semiquinone radical
(5) in the next step. The radical 5 reacts with two types of
peroxyl radicals of the ethyl linoleate (6 and 7) at the 3′-position,
affording four coupling products through a peroxyl linkage. This
coupling is the important radical-scavenging step of the anti-
oxidation of the methyl caffeate. The coupling products are not
very stable because the aromatic stability of the original benzene
ring is disrupted by the coupling reaction. Thus, the subsequent
Diels-Alder reaction occurs in the coupling products. In our
previous antioxidation mechanism studies, curcumin and methyl
ferulate also provided this type of stabilized cyclic peroxide
(10, 12) in the presence of a large amount of linoleate. These
Diels-Alder products of peroxides would be common stable
antioxidation products of the antioxidant bearing a hydroxy-
cinnmamoyl structure.
ACKNOWLEDGMENT
We thank the Central Instrument Center of the Faculty of
Pharmaceutical Science of the University of Tokushima for MS
measurements and Department of Chemical Science and Tech-
nology of the University of Tokushima for opportunities to use
400 MHz NMR.
Supporting Information Available: Four tables of correlation
data of COSY, NOESY, HMBC, and TOCSY of products 3A,
3B, 4A, and 4B. This material is available free of charge via
the Internet at http://pubs.acs.org.
LITERATURE CITED
(1) Cuppet, S.; Schnepf, M.; Hall, C., III. Natural antioxidantssare
they a reality? In Natural Antioxidants Chemistry, Health Effects,
and Applications; Shahidi, F. Ed.; AOCS Press: Champaign, IL,
1996; pp 12-24.
(2) Konishi, Y.; Hitomi, Y.; Yoshida, M.; Yoshioka, E. Pharmaco-
kinetic study of caffeic and rosmarinic acids in rats after oral
administration. J. Agric. Food Chem. 2005, 53, 4740–4746.
(3) Curveier, M.-E.; Richard, H.; Berset, C. Comparison of the
antioxidative activity of some acid-phenols: structure-activity
relationship. Biosci., Biotechnol., Biochem. 1992, 56, 324–325.
Received for review March 13, 2008. Revised manuscript received May
9, 2008. Accepted May 9, 2008. This study was financially supported
by a grant from the Iijima Memorial Foundation for the Promotion of
Food Science and Technology (Ichikawa, Japan).
JF800781B