Chemical studies on antioxidant mechanism of curcuminoid: Analysis of radical reaction products from curcumin

Article abstract of DOI:10.1021/jf9805348

In the course of studies on the antioxidant mechanism of curcumin, its radical reaction was investigated. Curcumin was reacted with radical species, which were generated from the pyrolysis of 2,2'-azobis(isobutyronitrile) under an oxygen atmosphere, and the reaction products from curcumin were followed by HPLC. The reaction at 70 °C gave several products, three of which were structurally identified to be vanillin, ferulic acid, and a dimer of curcumin after their isolation. The dimer was a newly identified compound bearing a dihydrofuran moiety, and its chemical structure was elucidated using spectroscopic analyses, especially 2D NMR techniques. A mechanism for the dimer production is proposed and its relation to curcumin's antioxidant activity discussed. The time course and gel permeation chromatography studies of the reaction were also investigated, and the results indicate that the dimer is a radical-terminated product in the initial stage.

Full text of DOI:10.1021/jf9805348

J. Agric. Food Chem. 1999, 47, 71−77
71
Ch em ica l Stu d ies on An tioxid a n t Mech a n ism of Cu r cu m in oid :
An a lysis of Ra d ica l Rea ction P r od u cts fr om Cu r cu m in
Toshiya Masuda,*,† Kayo Hidaka,‡ Ayumi Shinohara,† Tomomi Maekawa,‡ Yoshio Takeda,† and
Hidemasa Yamaguchi‡
Faculty of Integrated Arts and Sciences, University of Tokushima, Tokushima 770-8502, J apan, and
Faculty of Human Life Science, Osaka City University, Osaka 558, J apan
In the course of studies on the antioxidant mechanism of curcumin, its radical reaction was
investigated. Curcumin was reacted with radical species, which were generated from the pyrolysis
of 2,2′-azobis(isobutyronitrile) under an oxygen atmosphere, and the reaction products from curcumin
were followed by HPLC. The reaction at 70 °C gave several products, three of which were structurally
identified to be vanillin, ferulic acid, and a dimer of curcumin after their isolation. The dimer was
a newly identified compound bearing a dihydrofuran moiety, and its chemical structure was
elucidated using spectroscopic analyses, especially 2D NMR techniques. A mechanism for the dimer
production is proposed and its relation to curcumin’s antioxidant activity discussed. The time course
and gel permeation chromatography studies of the reaction were also investigated, and the results
indicate that the dimer is a radical-terminated product in the initial stage.
Keyw or d s: Curcumin; antioxidant; radical termination; dimer; antioxidant mechanism; HPLC
analysis; GPC analysis
INTRODUCTION
would afford important contributions to the antioxidant
mechanism studies. Recently, a phenolic antioxidant
bearing ortho-dihydroxyl groups was intensively inves-
tigated as an efficient antioxidant, and the produced
stable o-quinone material plays a key role in its anti-
oxidation (Hall and Cuppett, 1997). However, studies
of the other types of phenolic antioxidants are very
limited.
Some phenolic compounds in edible plants have
received much attention as powerful antioxidants not
only to protect against the oxidative deterioration of food
but also to reduce oxidation-induced diseases by their
ingestion. A main mechanism for a phenolic antioxidant
in food is the trapping and stabilizing of radical species,
such as the lipid peroxyl radical, which is generated
from radical chain oxidation of food components. This
property is also important in biological systems for
scavenging harmful oxygen radicals. The antioxidation
process is thought to be divided into two stages as shown
in following schemes:
Curcumin (1) is the main yellow phenolic material of
turmeric and has been widely used as a food coloring
reagent (Govindarajan, 1980). Various curcumin-related
phenols (curcuminoids) have also been found in edible
plants, especially Zingiberaceae plants (Toda et al.,
1985; Uehara et al., 1987; Masuda et al., 1992, 1993,
1994, 1995). Curcumin has a strong antioxidant activity
not only in food systems but also in biological systems
(Huang and Ferraro, 1992). Recently, its antioxidant
activity in biological systems has received attention as
it prevents some peroxidation-related diseases (Ruby et
al., 1995). A study of the clinical application of curcumin
as a chemopreventive anticancer drug was intensively
carried out at the National Cancer Institute (Kelloff et
al., 1996). Structurally, curcumin consists of two ortho-
methoxylated phenols and a â-diketone moiety, and they
are all conjugated. Radical trapping would first occur
at a phenol position as in other phenolic antioxidants;
however, the position of the radical termination has not
been defined so far because of the complexity of its
highly conjugated structure. The objectives of this study
were to clarify the stable radical termination products
from curcumin and to examine the antioxidation process
of curcumin on the basis of the products (Figure 1).
(1) radical trapping stage
S-OO• (or S•) + AH f S-OOH (or SH) + A•
(2) radical termination stage
A• f nonradical materials
S is the substance for oxidation, AH is the antioxidant,
and A• is the antioxidant radical.
Recent mechanistic studies of a plant phenolic anti-
oxidant have been focused on the trapping stage using
a kinetic approach or a structure-activity relationship
approach [for recent reviews see: Aruoma et al. (1997)
and Hall and Cuppett (1997)]. The antioxidant radical,
which forms in the trapping stage, should be converted
to a nonradical material in the termination stage
without any production of further radical species.
Structural information about these nonradical products
MATERIALS AND METHODS
* Author to whom correspondence should be addressed (fax
+81-886-56-7244; e-mail masuda@ias.tokushima-u.ac.jp).
† University of Tokushima.
Ch em ica ls a n d In str u m en ts. Curcumin was synthesized
using a previously reported method (Pabon, 1964). 2,2′-Azobis-
(isobutyronitrile) (AIBN) was obtained from Tokyo Kasei Co.,
‡ Osaka City University.
10.1021/jf9805348 CCC: $18.00 © 1999 American Chemical Society
Published on Web 12/22/1998

72 J. Agric. Food Chem., Vol. 47, No. 1, 1999
Masuda et al.
(4.6 × 250 mm, Nomura Chemical, Seto, J apan); solvent
system, 2% AcOH in n-hexane (solvent A) and 2% AcOH in
ethyl acetate (solvent B); elution, linear gradient from 14%
solvent B to 71% solvent B for 40 min (after the gradient, 100%
of solvent B was eluted for an additional 10 min); flow rate, 1
mL/min; detection, 420, 360, and 300 nm. The UV spectra from
280 to 480 nm for each observed peak were obtained.
For the time course studies, at 1 h intervals, 10 µL of the
reaction solution was withdrawn and analyzed according to
the same method. Peak area was calculated using a Shimadzu
M-10A data analyzing system.
P u r ifica tion P r oced u r e for P r od u cts 2-5. A CH3CN
solution (25 mL) of curcumin (0.5 g) and AIBN (0.5 g) was
degassed and heated under an oxygen atmosphere at 70 °C
for 1 h. The solution was evaporated in vacuo to dryness and
immediately filtered with methanol to remove most of the
curcumin and AIBN. The filtrate was evaporated and the
residue (∼0.5 g) subjected to silica gel column chromatography
eluted with 1-10% methanol in CH2Cl2 in the stepwise
gradient mode, giving six fractions [fraction 1 (193 mg),
fraction 2 (20 mg), fraction 3(175 mg), fraction 4 (44 mg),
fraction 5 (27 mg), fraction 6 (25 mg)]. Fraction 2 was purified
with silica gel TLC (5% methanol in CH2Cl2) to afford 5 (1.5
mg). From fraction 4, 2 (1.9 mg), 3 (3.2 mg), and 4 (1.3 mg)
were obtained by repeated silica gel TLCs (5% methanol in
CH2Cl2). Each isolated product was assigned to the corre-
sponding peak in the HPLC data in Figure 2 by comparison
of their retention times.
Isola tion P r oced u r e for P r od u t 2 Usin g P r ep a r a tive
HP LC. A CH3CN solution (30 mL) of curcumin (0.6 g) and
AIBN (0.6 g) was heated for 2 h at 70 °C in air. The solution
was evaporated in vacuo to dryness and immediately filtered
with methanol. After evaporation, the residue was dissolved
in CH2Cl2 (3 mL). A 0.5 mL sample of the solution was injected
into a preparative HPLC equipped with an LOP CN 24S
column (Nomura Chemicals) using the following conditions:
flow rate, 10 mL/min; gradient solvent system, 35% ethyl
acetate in hexane containing 0.2% AcOH to 100% ethyl acetate
containing 0.2% AcOH for 1 h; detection, 420 nm. A peak
eluted at 26 min was collected. After the collected solution was
evaporated, the residue was purified by silica gel TLC (3%
methanol in CH2Cl2) to give 2 (11 mg) as an orange amorphous
solid.
F igu r e 1. Chemical structures of curcumin (1) and the radical
reaction products (2, 4, and 5).
Ltd. (Tokyo, J apan). All solvents were obtained from Wako
Pure Chemical Industries, Ltd. (Osaka, J apan). Silica gel for
column chromatography (NAM-300) was obtained from NAM
Laboratory (Motosu, J apan). Silica gel TLC plates (silica gel
60 F254 PLC plates, No. 1.05744) were obtained from Merck
(Darmstadt, Germany). NMR spectra were measured with a
Varian Unity Plus 500 spectrometer. Mass spectra were
measured with a Hitachi M-200 or a J EOL SX-102A spec-
trometer. A Shimadzu LC 6A system equipped with a photo-
diode array detector (Shimadzu, SPD-M10AP) was used for
the analytical HPLC and a J ASCO Gulliver Gradient LC
system was used for the preparative HPLC.
An a lysis of Ra d ica l Rea ction P r od u cts. Curcumin (100
mg) and AIBN (100 mg) were dissolved in CH3CN (5 mL) in a
flask fitted with an oxygen balloon through a three-neck valve.
After degassing and substitution with oxygen, the solution was
heated at 70 °C. One hour later, 10 µL of the solution was
withdrawn and diluted with 100 µL of CH3CN. Ten microliters
of the diluted solution was injected into the analytical HPLC
system under the following conditions: column, Develosil CN-5
F igu r e 2. HPLC chromatograms of the radical reaction mixture of curcumin at 1 h.

Radical Reaction Products from Curcumin
J. Agric. Food Chem., Vol. 47, No. 1, 1999 73
F igu r e 3. Comparison with each UV absorption pattern
corresponding to product peaks 1-5.
GP C An a lysis of Rea ction Mixtu r e. The reaction was
carried out using the previously described protocol. At 1 h
intervals, 100 µL of sample solution was withdrawn and
concentrated to dryness in an N2 stream. The residue was
dissolved in 500 µL of THF, and 1.5 µL of the solution was
then injected into an analytical HPLC equipped with the GPC
column (GPC KF-803, Showa Denko, Tokyo, J apan) under the
following conditions: flow rate, 1 mL/min; solvent 2% AcOH
in THF; detection, 300 and 420 nm.
RESULTS
HP LC An a lysis of Ra d ica l Rea ction P r od u cts
fr om Cu r cu m in . Organic azo compounds are easily
handled reagents because they are less explosive com-
pared with organic peroxides. Azo compounds produce
radicals at a constant rate by pyrolysis. Recent studies
on the rate analysis of food oxidation employed them
(Niki, 1987). We also employed AIBN as an azo com-
pound to produce the curcumin radical. Pure curcumin,
which was synthetically obtained, and AIBN were
incubated in CH3CN in an oxygen atmosphere, and the
reaction products from curcumin were observed by the
HPLC method. Although no product was observed in
the reaction below 70 °C for 1 h, many products were
observed in the reaction mixture at 70 °C for 1 h. The
analytical data for the reaction mixture at 1 h are shown
in Figure 2. With 420 nm detection, two peaks are
mainly observed with retention times at 24.9 (product
2) and 28.5 min (product 3) along with curcumin’s (1)
peak at 16.9 min. On the other hand, with 300 nm
detection, several additional peaks are observed from 6
to 13 min in addition to the 420 nm detected peaks.
Curcumin has an absorption maxima at 420 nm because
of its conjugated diaryl heptanoid chromophore. The two
peaks at 24.9 and 28.5 min retain an absorption maxima
around 400 nm as shown in Figure 3, indicating that
there was no damage to the curcumin chromophore.
However, the 300 nm detected peaks at 7.3 (product 4)
and 8.4 min (product 5) do not have an absorption at
420 nm, which could be due to the damaged conjugation
system of curcumin’s structure.
F igu r e 4. Selected carbon-proton long-range connectivities
observed in HMBC (arrows in A) and NOE correlations
observed in NOESY (arrows in B). Tentative position num-
berings are given on the basis of curcumin’s numbering.
gave a molecular ion peak at m/z 152. The 1H NMR
spectrum of 4 showed an aldehyde signal at 9.87 ppm,
three substituted benzene protons at 7.26 (1H, d, J ) 8
Hz) and 7.50 (2H, m) ppm, and one phenolic methoxyl
signal at 4.00 ppm. These data indicated that 4 was
vanillin. A co-injection experiment of the authentic
sample confirmed the results. The MS of 5 showed a
molecular ion peak at m/z 194, and its 1H NMR
spectrum revealed the signals due to three substituted
benzene protons [6.92 (1H, d, J ) 8 Hz), 7.18 (1H, dd,
J ) 8 and 2 Hz), and 7.36 (1H, d, J ) 2 Hz) ppm], a low
field shifted trans double bond [6.42 (1H, d, J ) 16 Hz)
and 7.63 (1H, d, J ) 16 Hz) ppm], and a phenolic
methoxyl group [3.98 (3H, s) ppm]. These data indicated
that 5 was ferulic acid. The structure was confirmed
with the authentic sample using HPLC. Product 2 was
obtained as an orange amorphous solid using repeated
chromatography. Although the EI and positive SI mass
spectrometries gave no molecular ion peaks, the nega-
tive FAB mass spectrometry using magic bullet matrix
afforded a pseudo molecular ion peak at m/z 733 [M -
H]-. The high-resolution mode of the FAB MS revealed
that the molecular formula of 2 was C42H38O12 (calcd
for C42H37O12, 733.2285; found, 733.2297), which indi-
cated that 2 was a dimeric compound of curcumin. The
fine structure of 2 was elucidated using two-dimensional
Isola tion a n d Str u ctu r e Id en tifica tion of Rea c-
tion P r od u cts. To confirm the chemical structures of
the products, an isolation study was carried out. The
protocol used and yield for each product are described
under Materials and Methods. Structures of the isolated
products were analyzed by a spectroscopic method, and
the data are summarized in Tables 1-3. The MS of 4

74 J. Agric. Food Chem., Vol. 47, No. 1, 1999
Masuda et al.
Ta ble 1. Sp ectr oscop ic An a lysis Da ta for P r od u cts 2, 4, a n d 5
2
4
5
MS
negative HR-FAB mode
m/z 733.2297 [M - H]-
see Table 2
EI mode
EI mode
m/z 152 [M]+
m/z 194 [M]+
1H NMR
δ (acetone-d6, 500 MHz)
4.00 (3H, s)
δ(acetone-d6, 500 MHz)
3.98 (3H, s)
7.26 (1H, d, J ) 8 Hz)
7.50 (2H, m)
9.87 (1H, s)
6.42 (1H, d, J ) 16 Hz)
6.92 (1H, d, J ) 8 Hz)
7.18 (1H, dd, J ) 8 and 2 Hz)
7.36 (1H, d, J ) 2 Hz)
7.63 (1H, d, J ) 16 Hz)
13C NMR
see Table 3
Ta ble 2. 1H NMR Da ta a n d NOE Cor r ela ted P r oton s
Obser ved in a P h a se Sen sitive NOESY of P r od u ct 2
[δ in Aceton e-d 6 (500 MHz)]
Ta ble 3. 13C NMR Da ta a n d Cor r ela ted P r oton s
in a n HMBC Sp ectr u m of P r od u ct 2 [δ in Aceton e-d 6
(125 MHz)]
correlated protons
correlated protons
H-positiona
chemical shiftb
5.70, d, (5.9)
4.42, d, (5.9)
5.98, s
in NOESY
C-positiona
chemical shift
87.69
in HMBC
1
2
4
6
7
2′
5′
6′
2′′
5′′
6′′
1′′′
2′′′
6′′′
7′′′
2′′′′
H-2, H-4,H-2′,H-6′
H-1, H-4, H-2′, H-6′, H-6′′′
H-1, H-2, H-6
H-2′′, H-6′′, H-4
H-2′′, H-6′′
H-1, H-2,3′-OCH3
H-6′
H-5′, H-1, H-2
H-6, H-7,3′′-OCH3
H-6′′
H-6, H-7, H-5′′
H-2′′′′, H-6′′′′
H-6′′′, H-2′′′′, H-6′′′′
H-2, H-2′′′, H-2′′′′′
H-2′′′′′, H-6′′′′′
H-1′′′, H-2′′′,3′′′′-OCH3
H-6′′′′′
H-1′′′, H-2′′′
H-6′′′, H-7′′′,3′′′′′-OCH3
H-6′′′′′
1
2
3
4
5
6
7
1′
2′
3′
4′
5′
6′
1′′
2′′
3′′
H-2′, H-6′
63.02
200.35
100.17
179.84
120.28
142.53
133.05
110.30
148.57b
147.92
115.98c
119.58
127.90
111.14
148.61b
150.13
116.10c
124.08d
140.03
114.71
165.92
114.13
185.38
123.61
141.84
128.78
111.08
148.73b
149.61
116.10c
123.11
128.10
111.20
148.74b
149.94
116.23c
124.16d
56.27, 56.70 (×2),
56.16 (×2)
H-1, H-2, H-4
H-2, H-6
H-4, H-6, H-7
H-4, H-7
H-2′′, H-6′′
H-1, H-2, H-2′, H-6′
H-1, H-6′
overlapped
overlapped
overlapped
H-1, H-2′
H-6, H-7, H-2′′, H-6′′
overlapped
overlapped
overlapped
overlapped
H-7, H-2′′
H-2′′′′, H-6′′′′
H-1′′′
H-1, H-2, H-1′′′, H-2′′′
H-1, H-2
6.65, d, (15.6)
7.54, d, (15.6)
7.08, d,(1.6)
6.86, d, (7.8)
6.92, dd, (7.8, 1.6)
7.30, d, (1.6)
6.80, d, (7.6)c
7.11, dd, (7.6, 1.6)
7.40, d, (15.6)
7.78, d, (15.6)
7.00, d, (15.5)
7.49, d, (15.5)
7.26, d, (1.6)
6.87, d, (7.8)
7.15, dd, (7.8, 1.6)
7.22, d, (1.6)
6.83, d, (7.6)c
7.11, dd, (7.6, 1.6)
3.82, s
5′′′′
6′′′′
4′′
5′′
6′′
2′′′′′
5′′′′′
6′′′′′
3′-OCH3
3′′-OCH3
3′′′′-OCH3
1′′′
2′′′
3′′′
4′′′
5′′′
6′′′
7′′′
1′′′′
2′′′′
3′′′′
4′′′′
5′′′′
6′′′′
1′′′′′
2′′′′′
3′′′′′
4′′′′′
5′′′′′
6′′′′′
OCH3 (×4)
H-6′′′, H-7′′′, H-5′′′′′
H-2′
H-2′′
H-2′′′′
H-2′′′′′
3.85, s
3.88, s
H-6′′′, H-7′′′
3′′′′′-OCH3 3.79, s
OH
7.90, 8.22, 8.22, 8.30,
each br s
H-2′′′′′, H-6′′′′′
H-1′′′, H-2′′′, H-2′′′′
overlapped
overlapped
overlapped
a
H-position is tentatively numbered as shown in Figure 4.
b Coupling constants (J in hertz) in parentheses. c Assignment may
be interchangeable.
overlapped
H-1′′′, H-2′′′′
H-7′′′, H-2′′′′′, H-6′′′′′
overlapped
overlapped
overlapped
overlapped
H-7′′′, H-2′′′′′
overlapped
NMR techniques. The H-H COSY spectrum of 2 gave
the proton-proton coupling connectivities due to four
sets of three substituted benzene rings (ring A-1, H2′-
H5′-H6′; ring A-2, H2′′-H5′′-H6′′; ring B-1, H2′′′′-H5′′′′-
H6′′′′; and ring B-2, H2′′′′′-H5′′′′′-H6′′′′′; these num-
bered positions and proton chemical shifts are shown
in Figure 4 and Table 2, respectively), three sets of trans
olefins (H6-H7, H1′′′-H2′′′, H6′′′-H7′′′), and two pro-
tons at 5.70 and 4.42 ppm, which were coupled by 5.9
Hz. In addition to these proton connections, the signals
due to four phenolic methoxyl groups (3.79-3.88 ppm)
and one isolated proton (H4) at 5.98 ppm were observed
in the spectrum. The 13C NMR spectrum of 2 gave 42
carbon signals, which consisted of 38 separated signals
and two overlapped signals due to four phenolic meth-
oxyl groups. All carbons except the quaternary carbons
were assigned by the heteronuclear multiple-quantum
coherence experiment (HMQC) spectrum, with the data
being summarized in Table 3. The carbon-proton
networks were deduced with the carbon-proton long-
range couplings obtained from the heteronuclear mul-
tiple-bond coherence experiment (HMBC) spectral data.
Figure 4A shows important carbon-proton correlations,
which clarify all carbon-carbon linkages of structure 2
a
C-position is tentatively numbered as shown in Figure 4.
Assignments may be interchangeable.
b-d
except for the C4′′′-C5′′′ bond. The C4′′′-C5′′′ linkage
in 2 was determined by NOE correlations between H2-
H6′′′and H2′′′-H6′′′ in the phase sensitive nuclear
Overhauser effect spectroscopy (NOESY) spectrum (Fig-
ure 4B). The substituted positions of the four methoxyl
groups were determined by NOEs with ortho-protons
in the benzene rings. Thus, the structure of 2 was
determined as shown in Figure 1. The structure has
stereocenters at the 1- and 2-positions. The relative
stereochemistry of the protons on the dihydrofuran ring
is difficult to determine. When protons on the dihydro-
furan have appropriate NOEs resulting from the fixed
conformation of the furan, the NOE data are useful for
stereochemical analysis (Li et al., 1997). However, H1

Radical Reaction Products from Curcumin
J. Agric. Food Chem., Vol. 47, No. 1, 1999 75
F igu r e 5. Time course of accumulation of the radical reaction
products 2-5. Each peak area was calculated from the 420
nm peak for 2 and 3, the 300 nm peak for 4, and the 360 nm
peak for 5.
showed NOEs with H2 and H4, and H2 also showed
NOEs with H1, H2′, and H6′′′ in both the NOE differ-
ence spectrum and the phase sensitive NOESY spec-
trum of 2. These results revealed that 2 had some
flexible conformation in its dihydrofuran ring, whereas
the stereochemical relationship of H1 and H2 remained
unclear. The 1H NMR data and the estimated molecular
size (∼750) of product 3, which was obtained by GPC
analysis, were similar to those of 2, suggesting that 3
was also a dimer of curcumin. Unfortunately, the
structure could not be obtained at this stage because of
its instability and complexity.
Tim e Cou r se Stu d ies of th e P r od u ct F or m a tion .
The time course of accumulation of the radical reaction
products was investigated using HPLC analytical meth-
ods. The reaction was followed for 6 h, and the results
are shown in Figure 5, which shows that products 2 and
3 were immediately formed after start of the reaction
and that, after 1 h, the amounts of both decreased,
whereas products 4 and 5 independently increased
during the investigation. These results indicated that
products 2 and 3 were the initial products derived from
curcumin in its antioxidation process, and they were
much less stable than products 4 and 5. After the peaks
of 2 and 3 disappeared, no other new peaks were
observed by HPLC (data not shown).
Gel P er m ea tion Ch r om a togr a p h y (GP C) An a ly-
sis of Rea ction Mixtu r e. GPC is one of the easiest
methods to estimate the molecular size of a compound.
We analyzed the reaction mixtures at 1 and 6 h (Figure
6). Figure 6A shows that curcumin’s peak is observed
at 10.3 min and the curcumin dimer 2 at 9.84 min. From
these retention times, both molecular sizes were esti-
mated to be ∼350 and ∼700, which were in good
agreement with the MS analysis. In the 6 h data (Figure
6B), a peak observed at 420 nm showed a reading shape
to the retention time of 7 min. Two peaks at 10.8 and
11.2 min were also clearly observed at the 300 nm
detection (Figure 6B). For the GPC analytical condi-
tions, a higher molecular sized material is eluted earlier.
Thus, the obvious reading peak indicated that polym-
erization of curcumin would occur in the reaction.
F igu r e 6. GPC analytical data [(A) analysis for the reaction
mixture at 1 h, (B) analysis for the reaction mixture at 6 h]
with an estimated molucular weight scale.
other two compounds, one of which was structurally
elucidated, were complicated compounds such as a
dimer. Our proposed mechanism for their production
from curcumin radical is illustrated in Figure 7. In the
mechanism shown, curcumin is first converted to the
curcumin radical at the phenolic position (structure I).
The radical can move to appropriate positions as
represented by structures II and III. When two types
of curcumin radicals (II and III) are coupled between
both radical positions, the radical reaction is terminated
by producing this nonradical material. The terminated
product is not very stable because it has two unstable
quinone methide structures. Thus, the hydroxyl group
on the upside curcumin part immediately cyclizes at the
benzyl position of the other curcumin part, forming a
thermodynamically stable furan ring. Finally, a hydro-
gen on the furan ring transfers to a phenolic position
to give compound 2, which has a more stable dihydro-
furan ring. The radical termination by dimerization of
the antioxidant has been also observed in vitamin E
(Winterle et al., 1984; Yamauchi et al., 1989) and BHT
(Zhang et al., 1997). This radicalic dimerization should
play an important role in curcumin’s antioxidant activ-
ity. On the other hand, when curcumin radicals II and
III independently react with molecular oxygen, two
types of peroxyl radicals are produced. As a possible
mechanism, peroxyl radicals are cyclized at adjacent
benzylic or ketone positions and then decomposed to
form compounds 4 and 5, respectively. It should be noted
DISCUSSION
The isolation work of radical reaction products from
curcumin gave four compounds. Two of them were
structurally fragmented compounds from curcumin. The

76 J. Agric. Food Chem., Vol. 47, No. 1, 1999
Masuda et al.
F igu r e 7. Proposed mechanism for formation of products 2, 4, and 5 in the radical reaction.
that for this pathway of 4 and 5, the radical could not
have a radical termination stage but instead required
an abstraction of the hydrogen atom from the other
compound. Degradation of curcumin under various
conditions such as alkaline or photoirradiated conditions
was examined, and compounds 4 and 5 are always
observed in the degraded materials as stable compounds
(Tønnesen and Karsen, 1985; Tønnesen et al., 1986;
Wang et al., 1997). Antioxidant studies of the phenolic
compounds also reveal that the radical trapping activity
of vanillin and ferulic acid is lower than that of cur-
cumin (Sharma, 1976; Toda et al., 1985; Brand-Williams
et al., 1995), indicating that these compounds are more
stable than curcumin under radical conditions and
would not degrade in the presence of curcumin. Thus,
4 and 5 were probably accumulated as stable compounds
under the stated conditions. The time course results
indicate that the curcumin dimer is not the final
terminated compound. It has an additional radical
trapping ability because the reactive phenolic hydroxyl
group remained. GPC analysis of the reaction mixture
clearly showed the production of higher molecular
weight materials than the dimer with longer reaction
time. The materials would be oligomeric substances of
curcumin from the analysis data and would be produced
through the dimer or directly from curcumin. These
results indicate that curcumin acts as a chain-breaking
antioxidant due to its radical termination ability by its
dimerization and following polymerization process.
ACKNOWLEDGMENT
We thank Ms. Yoshioka, Faculty of Pharmaceutical
Science, University of Tokushima, for FAB MS mea-
surement.
LITERATURE CITED
Aruoma, O. I.; Halliewell, B.; Williamson, G. In vitro methods
for characterizing potential prooxidant and antioxidant

Radical Reaction Products from Curcumin
J. Agric. Food Chem., Vol. 47, No. 1, 1999 77
actions of nonnutritive substances in plant foods. In Anti-
oxidant Methodology in Vivo and in Vitro Concepts; Aruoma,
O. I., Cuppett, S. L., Eds.; AOCS Press: Champaign, IL,
1997; pp 173-204.
Brand-Williams, W.; Cuvelier, M. E.; Berset, C. Use of a free
radical method to evaluate antioxidant activity. Lebensm-
Wiss. -Technol. 1995, 28, 25-30.
Niki, E. Antioxidants in relation to lipid peroxidation. Chem.
Phys. Lipids 1987, 44, 227-253.
Pabon, H. J . J . A synthesis of curcumin and related com-
pounds. Recl. Trav. Chim. Pays-Bas 1964, 83, 379-385.
Ruby, A. J .; Kuttan, G.; Babu, K. D.; Rajasekharan, K. N.;
Kuttan R. Anti-tumour and antioxidant activity of natural
curcuminoids. Cancer Lett. 1995, 94, 79-83.
Govindarajan, V. S. Turmeric-chemistry, technology, and
quality. CRC Crit. Rev. Food Sci. Nutr. 1980, 12, 199-246.
Hall, III, C. A.; Cuppett, S. L. Structure-Activities of Natural
Antioxidants. In Antioxidant Methodology in Vivo and in
Vitro Concepts; Aruoma, O. I., Cuppett, S. L., Eds.; AOCS
Press: Champaign, IL, 1997; pp 141-172.
Sharma, O. P. Antioxidant activity of curcumin and related
compounds. Biochem. Pharmacol. 1976, 25, 1811-1812.
Toda, S.; Miyase, T.; Arichi, H.; Tanizawa, H.; Takino, Y.
Natural antioxidants III, antioxidative compornents isolated
from rhizomes of Curcuma longa. Chem. Pharm. Bull. 1985,
33, 1725-1728.
Huang, M. T.; Ferraro, T. Phenolic compounds in food and
cancer prevention. In Phenolic Compounds in Food and
Their Effects on Health II. Antioxidants and Cancer Preven-
tion; ACS Symposium Series 507; Houng, M. T., Ho, C. T.,
Lee, C. Y., Eds.; American Chemical Society: Washington,
DC, 1992; pp 8-34.
Kelloff, G. J .; Crowell, J . A.; Hawk, E. T.; Steele, V. E.; Lubet,
R. A.; Boone, C. W.; Covey, J . M.; Doody, L. A.; Omenn, G.
S.; Greenwald, P.; Hong, W. K.; Parkinson, D. R.; Bagheri,
D.; Baxter, G. T.; Blunden, M.; Doelts, M. K.; Eisenhauer,
K. M.; J ohnson, K.; Knapp, G. G.; Longfellow, D. G.; Malone,
W. F.; Nayfield, S. G.; Sefried, H. E.; Swall, L. M.; Sigman,
C. C. Strategy and planning for chemopreventive drug
development: clinical development plans II. J . Cell. Bio-
chem. 1996, 26S, 54-71.
Li, S.; Iliefski, T.; Lundquist, K.; Wallis, A. F. A. Reassignment
of relative stereochemistry at C-7 and C-8 in arylcoumarin
neolignan. Phytochemistry 1997, 46, 927-934.
Masuda, T.; J itoe, A. Antioxidative and antiinflammatory
compounds from tropical gingers: isolation, structure de-
termination, and activities of cassumunins A, B, and C, new
complex curcuminoids from Zingiber cassumunar. J . Agric.
Food Chem. 1994, 42, 1850-1856.
Masuda, T.; Isobe, J .; J itoe, A.; Nakatani, N. Antioxidative
curcuminoids from rhizomes of Curcuma xanthorrhiza.
Phytochemistry 1992, 31, 3645-3647.
Masuda, T.; J itoe, A.; Isobe, J .; Nakatani, N. Antioxidative and
antiinflammatory curcumin-related phenolics from rhizomes
of Curcuma domestica. Phytochemistry 1993, 32, 1557-
1560.
Masuda, T.; J itoe, A.; Mabry, T. J . Isolation and structure
determination of cassumunarins A, B, and C, new antiin-
flammatory antioxidants from a tropical ginger, Zingiber
cassumunar. J . Am. Oil Chem. Soc. 1995, 72, 1053-1057.
Tønnesen, H. H.; Karlsen, J . Studies on curcumin and cur-
cuminoids VI, kinetics of curcumin degradation in aqueous
solution. Z. Lebensm Unters. Forsch. 1985, 180, 402-404.
Tønnesen, H. H.; Karlsen, J .; van Henegouwen, G. B. Studies
on curcumin and curcuminoids VIII, photochemical stability
of curcumin. Z. Lebensm Unters. Forsch. 1986, 183, 116-
122.
Uehara, S.; Yasuda, I.; Akiyama, K.; Morita, H.; Takeya, K.;
Itokawa, H. Diarylheptanoids from the rhizomes of Cur-
cuma xanthorrhiza and Alpinia officinaraum. Chem. Pharm.
Bull. 1987, 35, 3298-3304.
Wang, Y.-J .; Pan M.-H.; Cheng, A.-L.; Lin L.-I.; Ho, Y.-S.;
Hsieh, C.-Y.; Lin, J .-K. Stability of curcumin in buffer
solutions and characterization of its degradation products.
J . Pharm. Biomed. Anal. 1997, 15, 1867-1876.
Winterle, J .; Dulin, D.; Mill, T. Products and Stoichiometry of
reaction of vitamin E with alkylperoxy radicals. J . Org.
Chem. 1984, 49, 491-495.
Yamauchi, R.; Matsui, T.; Kato, K.; Ueno, Y. Reaction of
R-tocopherol with 2,2′-azobis(2,4-dimethylvaleronitrile) in
benzene. Agric. Biol. Chem. 1989, 53, 3257-3262.
Zhang, N.; Kawakami, S.; Higaki, M.; Wee, V. T. New
oxidation pathway of 3,5,-di-tert-butyl-4-hydroxytoluene: an
ion spray tandem mass spectrometric and gass chromato-
graphic/mass spectrometric study. J . Am. Oil Chem. Soc.
1997, 74, 781-786.
Received for review May 20, 1998. Revised manuscript received
October 20, 1998. Accepted October 27, 1998. This work was
financially supported in part by a grant from Uragami
Foundation, Tokyo, J apan.
J F9805348