Photochemistry and Photobiology. Vol. 59, No. 3, pp. 295-302, 1994 0031-8655/93 $05.00+0.

00
Printed in the United States. All rights reserved 0 1994 American Society for Photobiology

SPECTRAL AND PHOTOCHEMICAL PROPERTIES OF CURCUMIN

J. RESZKA',ANN G. MOTTEN',
COLINF. CHIGNELL*',PIOTRB ~ r s ~ uKRZYSZTOF
l,
ROBERT H. SIK' a n d THOMASA. D A H L ~
ILaboratory of Molecular Biophysics, National Institute of Environmental Health Sciences, P.O. Box 12233,
Research Triangle Park, NC 27709, USA and ,Department of Pharmacology, School of Veterinary Medicine,
Tufts University, Boston, MA 02 111, USA
(Received 1 June 1993; accepted 12 November 1993)

Abstract -Curcumin, bis(4-hydroxy-3-methoxyphenyl)-1,6-heptadiene-3,5-dione, is a natural yellow-orange dye de-

rived from the rhizome of Curcuma longu, an East Indian plant. In order to understand the photobiology of curcumin
better we have studied the spectral and photochemical properties of both curcumin and 4-(4-hydroxy-3-methoxy-
phenyl)-3-buten-2-one (hC, half curcumin) in different solvents. In toluene, the absorption spectrum of curcumin
contains some structure, which disappears in more polar solvents, e.g. ethanol, acetonitrile. Curcumin fluorescence
is a broad band in acetonitrile (A, = 524 nm), ethanol (A, = 549 nm) or micellar solution (A, = 557 nm) but
has some structure in toluene (A, = 460, 488 nm). The fluorescence quantum yield of curcumin is low in sodium
dodecyl sulfate (SDS) solution (+ = 0.01 1) but higher in acetonitrile (+ = 0.104). Curcumin produced singlet oxygen
upon irradiation (A > 400 nm) in toluene or acetonitrile (6 = 0.1 1 for 50 fiuM curcumin); in acetonitrile curcumin
also quenched '0,(k, = 7 x 106M-I s-I). Singlet oxygen production was about 10 times lower in alcohols and was
hardly detectable when curcumin was solubilized in a D,O micellar solution of Triton X-100. In SDS micelles
containing curcumin no singlet oxygen phosphorescence could be observed. Curcumin photogenerates superoxide in
toluene and ethanol, which was detected using the electron paramagnetic resonance/spin-trapping technique with
5,s-dimethyl-pyrroline-N-oxide as a trapping agent. Unidentified carbon-centered radicals were also detected. These
findings indicate that the spectral and photochemical properties of curcumin are strongly influenced by solvent. In
biological systems, singlet oxygen, superoxide and products of photodegradation may all participate in curcumin
phototoxicity depending on the environment of the dye.

INTRODUCTION offer additional photochemical pathways for interaction with

C u r c u m i n , t bis(4-hydroxy-3-methoxyphenyl)-1,6-hepta- oxygen not available t o the isolated chromophore. In addi-
diene-3,s-dione (C), is a yellow-orange dye derived from the tion, the photochemistry could be affected by the configu-
rhizome o f Curcuma longa. T h e powdered root of this East ration o f the feruloyl chromophores-keto or enol. In the
Indian plant, turmeric, is often used as a component in curry cell, curcumin's photochemistry, including its reactions with
a n d is widely employed in India a s a food coloring, dye a n d oxygen, could also depend o n the specific microenvironment
food preservative. Curcumin exhibits a variety of biological of t h e molecule: polar or nonpolar, protic or aprotic, a n d
a n d photochemical activity, including phototoxicity t o with individual curcumin molecules isolated or present in
bacteria',2 a n d t o rat basophilic leukemia cells in culture.' concentrations high enough t o react bimolecularly with each
The photosensitizing mechanism of curcumin is not known other.
but has been shown t o require It has been dem- Our goal i n this study was t o elucidate the dependence of
onstrated that stable photoproducts such a s vanillin a n d feru- curcumin's oxygen-related photochemistry o n structural
lic acid are not responsible for curcumin c y t o t o ~ i c i t yMost
.~ properties a n d molecular environment. W e have character-
speculation has centered on photolytically produced active ized the spectral a n d photochemical properties of curcumin,
oxygen species such as singlet oxygen,' hydroxyl especially those related t o the production of active oxygen
superoxide or hydrogen peroxide.2 species: singlet oxygen, superoxide, hydroxyl radical a n d hy-
The curcumin molecule is composed of two aryl buten-2- drogen peroxide. In order to correlate these properties with
one (feruloyl) chromophores joined by a methylene group. structural features we have compared them with those of
It is n o t known whether the photochemical a n d cytotoxic half-curcumin, 4-(4-hydroxy-3-methoxyphenyl)-3-buten-2-
properties are mediated primarily by the ortho-methoxy- o n e (hC), a close analog of ferulic acid. We have also ex-
phenol groups, by the larger feruloyl chromophores acting in amined the effects of different solvents, representing different
isolation o r by the molecule as a whole: if the two chro- cellular microenvironments, o n the photochemical and pho-
mophores are connected by conjugation, the connection may tophysical properties of curcumin a n d hC. Such a charac-
terization m a y ultimately be useful i n elucidating the mech-
anism o f biological photoactivity for curcumin.
*To whom correspondence should be addressed.
tAbbreviations: curcumin, bis(4-hydroxy-3-methoxyphenyl)-1,6- MATERIALS AND METHODS
heptadiene-3,s-dione; DMF, N,N-dimethylformamide; DMPO,
5,5-dimethyl- 1-pyrroline-N-oxide; DMSO, dimethyl sulfoxide; Curcumin, hC, meso-tetraphenyl-porphinezinc complex (ZnTPP)
EPR, electron paramagnetic resonance; hC, 4-(4-hydroxy-3-me- and strychnine were purchased from Sigma Chemical Co. (St. Louis,
thoxyphenyl)-3-buten-2-one,(half-curcumin); RB, rose bengal; MO) and used as received. To verify that the batch of curcumin we
SDS, sodium dodecyl sulfate; ZnTPP, tetraphenylporphyrin zinc used was not contaminated, we used thin-layer chromatography and
complex. fluorescenceexcitation spectra; the unpurified curcumin behaved the
296 COLMF. CHIGNELL
ct a[.

same as a curcumin sample that had been purified by high-perfor- to exist i n the enol configuration, while h C is known t o as-
mance liquid chromatography. (In other selected experiments such s u m e t h e ketone form. W e confirmed, using nuclear magnetic
as low-temperature phosphorescence, spin trapping or '02produc-
resonance, that unirradiated curcumin and h C occurred es-
tion, both purified and nonpurified curcumin batches gave the same
results, data not shown.) Quinine sulfate monohydrate (Aldrich sentially entirely i n t h e enol a n d keto forms, respectivel)
Chemical Co., Milwaukee, WI) was recrystallized three times from (data not shown).
water. The spin trap 5,Sdimethyl- 1-pyrroline-N-oxide (DMPO) (Al-
drich Chemical Co.) was vacuum distilled and stored in a freezer. Absorption, fluorescence and low-temperature
Sodium dodecyl sulfate (SDS) was obtained from SchwadMann phosphorescence
Biotech. (Cleveland, OH), Triton X- 100 from Aldrich Chemical Co.
and catalase from Boehringer Mannheim (Indianapolis, IN). All sol- T h e spectral properties o f curcumin a n d h C were system-
vents were spectroscopic grade and other chemicals were reagent
atically investigated i n organic solvents a n d i n aqueous mi-
grade or better.
Absorption spectra were measured using an HP Diode Array 845 1 cellar solutions. The primary purpose of this study was to
Spectrophotorneter (Hewlett Packard Co., Palo Alto, CA). Fluores- examine t h e influence o f solvent polarity and proticity on
cence and phosphorescence spectra were recorded on an SLM SPC spectral properties and then use t h e information t o explain
823-SMC 220 spectrofluorometer (SLM Instruments, Urbana, IL). differences i n photoreactivity between curcumin a n d hC.
Emission spectra were corrected for PMT/monochromator response
using correction factors obtained with an incandescent lamp. Exci- Absorption spectra of curcumin consisted o f broad struc-
tation spectra were corrected using a series of photon counters. Flu- tureless bands i n acetonitrile and ethanol (Fig. 1A) and in
orescence spectra were also normalized for absorption at A,, using SDS a n d DMF (Fig. 1B). Absorption spectra o f h C i n water
the Beer-Lambert law. Refractive indices of solutions were measured a n d i n ethanol contained s o m e structure (Fig. 2), as did cur-
with a Bausch & Lomb refractometer. The quantum yield of fluo-
rescence, b(F), was obtained using quinine sulfate in 1 N H2S0,
solution as a standard according to described procedures.'* For +(F)
calculations, the integrated fluorescence intensity (over X) was used; l ' I ' I ' , , I ' , . I I

the estimated error for +(F) evaluation is about 5% for curcumin h

and 10% for hC. Phosphorescence lifetimes were measured using a 0.8 =!
Hitachi 9L902 chopper connected to a transient recorder interfaced 9
with a PC computer. Decay kinetics of phosphorescence and de-
convolution were analyzed using the Peakfit nonlinear curve-fitting 5 0.6
r
x
.-
c.
cn
C
software by Jandel Scientific(San Rafael, CA). Preliminary transient II
U a,
c
absorption measurements were performed using a previously de- .-C
scribed conventional flash photolysis ~ y s t e m . ~ ai
The phosphorescence was recorded on a steady-state '02 phos- 2 0.4 a,
0
m t
phorescence spectrometer as previously described.'O The quantum
yield of lo2formation in toluene was evaluated relative to ZnTPP
e0 8ln
in toluene (b('0,)= 0.73 in benzene") by comparing the integrated $ 0.2 2
0
steady-state intensities of '02phosphorescence of curcumin or hC
corrected for the number of absorbed photons to the standard. The
-3
LL
experimental compounds and the standard were usually compared
0.0
in the same solvent; both phosphorescence intensities and number
of absorbed photons were integrated over the appropriate wave-
length. Wavelength (nm)
Electron paramagnetic resonance (EPR) measurements were per-
formed on a Varian E-Line Century series EPR spectrometer op-
erating at 9.5 GHz. Samples in a quartz EPR flat cell (0.3 mm
lightpath) were illuminated directly inside the microwave cavity of
the spectrometer using a 1 kW Xe arc lamp and a Schoeffel grating
monochromator (Kratos, Inc., Westwood, NJ).
o.81
0.7 A" 5a
%
.-
c
ln
Photobleaching experiments were performed using radiation from - 3
t
a,
a 100 W tungsten-halogen lamp filtered through a 360 nm cutoff CI

glass filter combination (Oriel Corp., Stratford, CT). Irradiations .-t

were camed out in a 1 cm spectrophotometric cell at room tem-
-2 c 8
perature, and the solution was stirred with a magnetic bar. The
relative number of absorbed photons was calculated as a double 8
ln
integral of absorbance over wavelength and irradiation time. 2!
Oxygen consumption was measured using the assembly for oxygen 0
3
photoconsumption previously described,'-' with 2 mL of curcumin ii
solution (0.5 mM) in SDS (0.1 M ) irradiated through a 400 nm cutoff
glass filter at 25°C. 350 400 450 500 550 600
Wavelength (nm)
RESULTS Figure 1. (A): Absorption and fluorescence spectra of curcumin.
Curcumin was soluble i n both polar a n d nonpolar organic Absorption spectra in acetonitrile, toluene and ethanol; lines la, 2a
and 3a, respectively; [curcumin] = 15 pM, 1 cm pathlength. Cor-
solvents but was practically insoluble in water a t neutral a n d rected and normalized fluorescence spectra in acetonitrile, toluene
acidic p H values. I t was more soluble i n alkaline solutions, and ethanol; lines lf, 2f and 3f, respectively; A,, = 375 nm. (B):
most likely due t o ionization o f t h e phenolic groups. In strongly Absorption and fluorescence spectra of curcumin. Absorption spec-
alkaline solutions the dye was less stable and underwent deg- tra in SDS, Triton X-100 micelles and in N,N-dimethylformamide
(DMF); lines 4a, 5a and 6a, respectively; [curcumin] = 15 FM, 1 cm
r a d a t i ~ n . 'T
~ h e dye's solubility i n water increased i n the
pathlength; surfactant, 0.1 M. Corrected and normalized fluores-
presence of detergent micelles. cence spectra in SDS, Triton X- 100 micelles and in D M F lines 4f,
Curcumin. bv analoev with other B-diketones. is extxcted 5f and 6f. resaectivelv: A--- = 375 nm.
 Curcumin photochemistry 297

Table 1. Spectral and photochemical properties of curcumin and hC in different solvents*

~~~ ~

XA,,,~~ x 10-4 XFmax hPhmax s(Ph) at 77 K

Compoundholvent nm M-lcm-l nm b(F) nm ms $J(Q2)*

Curcumin/toluene 420 4.276 460,488 0.067 73Oe/'P 11,5ei'P 0.11

Curcumin/EtOH 430 2.747 549 0.063 7.2 <0.01
Curcumin/CH,CN 422 524 0.104 0.llt
CurcumidSDS micelle/
D2O 432 5.1 17 557 0.01 1 0
Curcumin/Triton X-100
micelle/D20 428 5.207 500 0.033 <0.005
CurcumidDMF 430 540 0.027
hC/aprotic solvent 546, 595 8.2; 35.Se/'P 0.097'01
0.1 1'01
hC/protic solvent 340" 450e1 <0.0006 547, 595" 34.1; 57.7" O W
335" 500'"

*XAmax, position of absorption maximum; c, molar extinction coefficient; hF,,,, position of fluorescence maximum; @(F), quantum yield of
fluorescence; XPbmax, position of low-temperature phosphorescence maximum; s(Ph), lifetimes of phosphorescence in low temperature
glasses; $J(IO2),quantum yield of singlet oxygen formation. nd, Not detectable; et, ethanol; w, water; e/ip, ethyl ethedisopentane 1: 1; tol,
toluene.
$Yield was not corrected for different lo2lifetimes and radiative rate constants in toluene and acetonitrile.
*Measured using RB (ETOH, CH,CN, D20)or ZnTPP in toluene as a reference, we assumed that the yield of lo2formation by ZnTPP in
toluene is the same as in benzene.

cumin in toluene and in Triton X-100 micelles (Figs. 1 and Curcumin and hC both phosphoresce at 77 K in aprotic
2). Dissolved curcumin obeyed the Lambert-Beer law in the glasses (Fig. 4). In ethanol, the phosphorescence spectrum
WM-mM concentration range in organic solvents and mi- of hC shows distinct structure (Fig. 4); a similar spectrum
cellar solutions, suggesting that under these conditions ag- was observed in Et,O/zso-pentane glass (not shown). Kinet-
gregation of dye molecules is negligible. In ethanol the ab- ically, the hC phosphorescence decay in frozen ethanol was
sorption maximum of curcumin is red-shifted by ca 70 nm best fitted to a biexponential equation whose two compo-
in comparison to that of hC (Figs. 1 and 2). Similar shifts nents showed different lifetimes (Table 1) with roughly equal
were observed in other solvents as well (Table 1). The nature contributions to the decay. Such a biexponential decay is
of the solvent affected the absorption spectrum of curcumin often found for carbonyl compounds that have both n l r *
only slightly, producing a small red shift on going from tol- and T-T* transitions"; alternatively, in the case of the hC
uene to more polar solvents (Fig. l). molecule the biexponential decay could arise from the pres-
Curcumin fluoresced strongly in toluene. Both the fluo- ence of two phosphorescent isomers.*
rescence intensity and the position of the most intense band In contrast to the fluorescence, the intensity of the phos-
of the dye were much more sensitive to the nature of the phorescence was much higher for hC than for curcumin itself.
solvent than were its absorption maxima. A large spectral The intensity for curcumin phosphorescence also varied more
red shift in the fluorescence maximum was observed on going with the excitation wavelength than would be expected from
from toluene to alcohol (Ax 6 1 nm) and SDS micelles (Ax 69 the change in extinction coefficient. The phosphorescence of
nm) (Fig. 1). The fluorescence quantum yield of curcumin curcumin in an etherliso-pentane glass (Aex = 280 nm, Fig.
was highest in acetonitrile; in SDS aqueous solution it was 4) decayed in milliseconds (Table 1). Excitation at 430 nm
10 times lower (Table 1). Particularly striking was the ob- (or at other wavelengths in the longwave absorption band)
servation that curcumin fluorescence was about two orders produced a very noisy, barely detectable spectrum somewhat
of magnitude stronger than that of hC (Fig. 2) in all solvents blue-shifted compared to the spectrum in Fig. 4 (not shown).
tested. In toluene the fluorescence spectrum of curcumin con-
tained some structure and could be resolved into several
Gaussian bands. However, in more polar solvents the spectra
were broad and structureless (Fig. 3). The fluorescence emis-
*Cis-trans isomerization of hC may result in the nonoverlapping of
sion of hC in water and ethanol was also more red-shifted phosphorescence excitation spectra recorded using different emis-
than its absorption (Fig. 2). sion wavelengths,At,,,. We found that the excitationspectrum mon-
Even though curcumin in its enol configuration may exist itored at k,,, = 540 nm differed somewhat from the spectra re-
in several cis-trans isomeric forms, no wavelength depen- corded at 600 and 660 nm (not shown). This seems to suggest that
dence was observed in its fluorescence excitation spectra. In at least two species (isomers?) may be contributing to the total
phosphorescence (the commercial compound was a mixture of
toluene, the excitation spectra recorded at five different emis- isomers, predominantly trans). Another reason for a complex de-
sion wavelengths were identical and overlapped well with cay may be aggregation upon cooling. Multiple transitions may
the absorption spectrum (Fig. 3). If curcumin exists in so- arise from the many types of chromophores in the hC molecule,
lution as a mixture of isomers then this observation suggests
that either all isomers produce the same fluorescent excited
-
including cis-trans and keto-enol isomers. Dual phosphorescence
has been observed for other keto enol ta~tomers.'~ Although a
detailed examination was beyond the scope of our present inves-
state or that all excited isomers, even ifdifferent, have similar tigation,time-resolved techniques might resolve these states better
fluorescence DroDerties in steadv-state exDeriments. than our steadv-state techniaues.
298 COLINF. CHIGNELL
et al.

? 0.8

300 340 380 420 4M) 5w 540 580

WAVELENGTH (nm)
Figure 2. Absorption and fluorescence spectra of hC. Absorption
spectra in water and ethanol, lines 1 and 2, respectively, sanitizer
concentration was 10 pM, 1 cm pathlength. Corrected and normal- Wavelength (nm)
ized fluorescence spectra of hC in water and ethanol; lines 1 and 2e,
respectively; A,, = 360 nm. The fluorescence intensity in Fig. 2 is Figure 4. Phosphorescence spectra of hC (Aex = 360 nm) (1) and
about two orders of magnitude lower than that of curcumin in Figs. curcumin (Aex 282 nm) (2) in low-temperature glass (77 K). The
1 and 2. difference in the phosphorescence intensities is arbitrary: the cur-
cumin phosphorescence was much weaker than that for hC, and also
depended on Acx.
It is not clear whether the difference was due t o detection
inaccuracy caused by the extreme weakness of the phospho- implies that lo2formation i n aerobic solution is limited, not
rescence o r reflected a real contribution(s) from different elec- by quenching kinetics, but rather by the yield o f excited
tronic transition(s) or impurities. What is clear is that the curcumin triplet.
curcumin phosphorescence intensity strongly depends o n the
energy o f exciting photons. Interaction with a n d wirh excited RB

Singlet oxygen generation It is reasonable t o ask whether curcumin, by virtue o f its

phenolic functional groups, may be able t o quench lo, by a
In toluene, benzene o r acetonitrile, curcumin generated O2 '
(Table I ) . However i n ethanol, iso-propanol a n d SDS a n d
Triton X- 100 micelles i n D20, the yields of singlet oxygen
were more than 10 times lower (Table 1). In toluene h C also
1 - 2.4
produced singlet oxygen with a yield similar to that of cur- B
cumin.
T h e yield of ' 0 2generated by curcumin depended only
slightly o n oxygen concentration. In toluene saturated with - 2.1
oxygen, t h e intensity of lo2phosphorescence was only about IJl
6% higher than in toluene saturated with a nitrogen-xygen 1,/1
mixture containing only 14% O2 (not shown). T h i s result
- 1.8
1
> . I . I . I ' I

10 20 30 40

1.5

1.2

Wavelength (nm) 0.0 0.2 0.4 0.6 0.8 1.0 1.2

Figure 3. Deconvolution of excitation (A) and fluorescence(B) spec- Curcumin (mM)
tra of curcumin in toluene. Excitation spectra of curcumin (1 5 p M )
in toluene were monitored at 460, 480, 500, 520 and 540 nm and Figure 5. (A): Stern-Volmer quenching of RB fluorescence by hC
they overlapped well with absorption spectrum. For comparison, the in acetonitrile; RB, 25 p M , A,, = 530 nm. (B): Stern-Volmer quench.
spectra were normalized to the same scale. Position (nm) of peaks ing of lo2phosphorescence by curcumin in air (0-0-)-and oxyger
after deconvolution was 459,485, 5 18 and 536 for fluorescence and (-.-.-)-saturated acetonitrile; RB, 25 p M ; RB was irradiated using
379, 398, 421 and 445 for excitation. a glass 550 nm cutoff filter.
 Curcumin photochemistry 299

small. (If curcumin did not quench ‘0,at all, but only RB
triplet, the difference in slopes would be expected to be a
factor of five, reflecting the difference in oxygen concentra-
tions in the two solutions.) Using these observations we es-
timated an upper limit for lo, quenching by curcumin to be
ca 7 x lo6M - ’ s-l in 0,-saturated solution, taking 58.3 ps
as the lifetime of in a~etonitri1e.I~
The absorbance of curcumin did not change during the
course of this experiment. Apparently, like other phenolic
quenchers, curcumin quenches lo2predominantly by a phys-
ical mechanism.

Radical reactions
We used EPR spin-trapping techniques to investigate the
free radical oxygen photoproducts of curcumin and hC. We
found that in benzene, curcumin generated the 0, - radical
efficiently when irradiated at 420 nm. Figure 6 shows the
Figure 6. (A): An EPR spectrum obtained from irradiatingcurcumin EPR spectrum of the D M P 0 / 0 2 - - adductZo(aN = 12.4, aBH
(ca 0.5 mM) in benzene at 420 nm in the presence of 80 M D M P O . = 10.2 and a?,, = 1.2 G). A carbon-centered radical was also
The sample was deaerated briefly prior to light exposure to reduce
the oxygen-broadening effect. (B): Computer-simulatedEPR spec- trapped (aN = 13.9 G, aHB = 22.3 G), which is most likely
trum of a DMP0/02.- adduct (aN= 12.4 G, aBH= 10.2 G and aTH derived from curcumin. Half-curcumin also generated the
= 1.2 G) superimposed on the spectrum of a carbon-centered radical DMPO/O,- - adduct in air-saturated benzene when irradiated
(identified with asterisk; aN= 13.9 G and auH= 22.3 G). at 342 nm (not shown).
The D M P 0 / 0 2 - signal was not significantly decreased in
the presence of the singlet oxygen quencher strychnine, sug-
mechanism similar to lo2quenching by other (po1y)phenolic
gesting that superoxide production does not involve lo2,but
substances,16or by enolic tautomers of @-ketones,” thus low-
rather is caused by triplet self-quenching via electron transfer
ering net lo2production. Using rose bengal (RB) to produce
followed by oxidation of the curcumin radical anion by 0,.
lo2,we found that the addition of curcumin decreases the This suggestion is supported by the observation that the
singlet oxygen phosophorescence emission in 0,-saturated
D M P 0 / 0 2 -- signal depended strongly on curcumin concen-
acetonitrile (Fig. 5A). To estimate a rate constant for this
tration.
reaction it was also necessary to find out whether curcumin
The D M P 0 / 0 2 - - adduct was also produced by curcumin
can quench RB excited states, singlet and triplet, because
in ethanol, dimethyl sulfoxide (DMSO), acetoaitrile, acetone
such quenching might also decrease the efficacy of singlet
and toluene. In all solvents the DMPO/O, adduct was iden-
~

oxygen production.
tified by comparing the photoinduced ESR spectra with those
Deactivation of the RB singlet state is a fast process even
generated by 18-crown-6-ether-solubilized KO, in the same
in organic solvents; in acetonitrile the radiative lifetime of
solvents.2’ Irradiation of hC in aerated ethanol also generated
RB is 2.38 ns and its nonradiative lifetime is 3.60 ns, while
the D M P 0 / 0 2 ’ - adduct (not shown).
in water the measured lifetime decreases to 90 PS.’~To obtain
The very low solubility of curcumin in water made it im-
measurable quenching ofthis short-lived excited singlet state,
possible to study the generation of superoxide in aqueous
it was necessary to use a high concentration of quencher.
solution. Solubility was higher in a micellar system, but no
Curcumin is not sufficiently soluble in acetonitrile to attain
02. was produced, probably because intermolecular triplet
-
the required concentration; therefore, we examined the more
quenching could not occur when the curcumin molecules
soluble hC as a potential model for curcumin’s behavior.
were isolated in the micelles.
When the RB fluorescence quenching data for acetonitrile
We found no evidence for hydroxyl radical (HO-) for-
are analyzed (Fig. 5A), the Stern-Volmer slope of 88.9 M-I
mation in aqueous micellar solution using DMPO, a sensitive
shows that hC quenching of RB fluorescence is diffusion
detector of this radical, as a spin-trapping agent. No other
controlled. It is reasonable to expect that curcumin itselfwill
long-lived free radicals were observed when curcumin was
also effectively quench RB fluorescence. However, RB singlet
irradiated in the absence of a spin trap in the systems we
quenching by curcumin cannot contribute much to the ap-
used. Thus, the direct generation of H O radical by curcumin
parent lo2quenching rate because at the 2 m M curcumin
to produce photocytotoxicity is not very likely, although in
concentration required for a 50% decrease in lo2phospho-
a biological environment curcumin could enhance the pro-
rescence, RB fluorescence quenching was negligible.
duction of HO- radical in the Fenton r e a ~ t i o n . ~
We noticed that in the presence of curcumin, the lo2phos-
phorescence intensity in acetonitrile depended somewhat on
Photodegradation
whether the RB solution was saturated with air or oxygen
(Fig. 5B). This observation suggests that the RB triplet was When irradiated in organic or micellar solution, the ab-
quenched not only by oxygen, but also to some extent by sorbance of curcumin decreased, indicating that the dye was
curcumin itself. However, based on the small difference in photobleached (Fig. 7). The nonlinear bleaching rate points
slopes in Fig. 5B it is safe to assume that the contribution of to the accumulation of a product(s) of photodegradation thai
curcumin to 3RB*auenchina in 0,-saturated solution is auite absorbs in the same wavelength range as curcumin but is
300 COLIN F. CHIGNELL ef a/.

300 350 400 450 500

- 0.8 0.15 5
2 0.6
0.10 zaa
0.05 GN
8
C
al
5: 0.4
0.00
2 1200 1250 1300 135C
0
L Wavelength (nm)
n 0.2
v)
0
c
"
n
,
0.0
7
v
1230 1260 1290 1320 1350 1380
. .
I I I I I 0.0 Wavelength (nm)
300 350 400 450 500 Figure 8. Singlet oxygen phosphorescence during curcumin irradi-
Wavelength (nrn) ation in toluene; duration of one scan was 1 rnin. Phosphorescence
spectra over time of curcumin (A) and hC (B) during irradiation in
Figure 7. Photobleaching of curcumin in micellar and nonmicellar toluene through 320 nm and 400 nm cutoff filters for hC and cur-
solutions. (A): Absorption spectra during bleaching in toluene; spec- cumin, respectively. Both sets of spectra cover the same irradiation
tra 1 4 taken after 10,20,30 and 40 min ofirradiation, respectively. total time of about 15 min. The top spectrum in each time series
(B): [Inset] Rate of bleaching in toluene (line 1) and in toluene con- (1A and 1B) is the first (t = 0) one of the sequence.
taining 20 mM strychnine (line 2). (C): Absorption spectra during
bleaching in SDS micelles; spectra 1-4 taken after 10, 20, 30 and
40 min of irradiation respectively. (D): [Inset]Rate of bleaching in
SDS (line 3) and Triton X-100 (line 2) micellar solutions. All so- in the ground state one of the two ketone groups in curcumin
lutions contained l mMcurcumin; micellar solutions contained 0. l should exist as an enol (Scheme 1); the enolic form has been
M surfactant. Number of photons is in arbitrary units; bleached pictured as forming an intramolecular hydrogen bond be-
curcumin concentration units are nanornolar. tween the enolic hydrogen and the second ketone.24The enol
can be formed from either ketone. The remaining unenolized
part will then have the same keto structure as hC (Scheme
apparently more photostable (Fig. 7A). In toluene, curcumin 1). If the two ends of the chromophore can communicate via
photobleaching was accompanied by a decrease in '02pro- resonance structures in the enol forms, intramolecular con-
duction (Fig. 8 4 . The decrease was fast initially but slower jugation of the *-electrons may couple the two feruloyl chro-
during longer irradiation. Thus, the products of curcumin mophores into one extended ?r-electron system. Upon exci-
photodegradation must be weaker '02 sensitizers than cur- tation, beta diketones can undergo k e t o n i ~ a t i o n , ~ which
',~~
cumin itself. Several of the known photoproducts of would leave both sides with the hC structure. Again, these
curcumin22contain the same chromophore as found in hC. two chromophores could act independently or be coupled.
We observed, however, that the lo2yield from hC in toluene
also decreased with continued irradiation (Fig. 8B), although
bleaching appeared to be less than that observed for curcu-
min. Ho7& c=c.H,
hC
In air-saturated toluene solutions of curcumin, the pho-
tobleaching was inhibited by the singlet oxygen quencher
strychnine. However, photobleaching was also observed in
deoxygenated toluene solution, and this photobleaching was
also inhibited by strychnine (not shown). Thus, in contrast
to Tannesen and coworkers,2z we conclude that the photo-
degradation of curcumin is not mediated solely by loz.
In SDS micellar solution, photobleaching of curcumin re-
sulted in oxygen consumption. After 3 h of irradiation, about
35% of the oxygen initially present in the sample was con-
sumed; 5% of the consumed oxygen was recovered by catalase
(120 U/mL) added after irradiation. (Similar experiments
0
I
B
'H

C
carried out in H,O/EtOH 1:1 vol mixture containing 0. 1 M
-*.. H /
of curcumin revealed similar oxygen photoconsumption and
release.) This recovery suggests that small amounts of per- Scheme 1
oxide(s) or H 2 0 2are produced along with the other oxidationThus any similarities in spectral properties of curcumin
products detected previously.2z
and hC must be attributed to either the ortho-methoxyphenyl
groups or to independent action of the side of the curcumin
DISCUSSION
molecule that retains the ketone group. Differences may be
Curcumin contains two feruloyl moieties joined by a meth- attributed to either the difference in keto-enol configuration
ylene group (Scheme 1). By analogy to other P-diketone~,~' or to the existence of extended conjugation in the curcumin
 Curcumin photochemistry 30 1

molecule, or both. In particular, curcumin's yellow-orange lutions was observed only in the presence of Triton X-100
color would seem to be accounted for better by extended micelles,§ where it was very weak. In aprotic solvents, the
conjugation, since hC is colorless. ability to sensitize lo2may be associated with the keto groups,
Intramolecular hydrogen bonding usually provides a chan- as ketones are known to form long-lived excited states rel-
nel for fast radiationless deactivation of excited states via atively easily, with long lifetimes in aprotic solvents. In aque-
excited-state hydrogen atom transfer.25 However, in aprotic ous solutions, where intermolecular hydrogen bonding pro-
media curcumin fluoresces intensely (Table 1) and generates vides channels for efficient deactivation of ketone triplet states,
lo2, implying low nonradiative singlet deactivation to the no lo2was observed in water or D,O.
ground state and the formation of a relatively long-lived While the ketone character may contribute to lo2sensi-
triplet state in solution. It then follows that any intramolec- tization by curcumin, the reported antioxidant activity of the
ular hydrogen bond must disappear when curcumin is ex- dye can probably be attributed to the phenolic and enolic
cited. If the excited state undergoes ketonization, any hy- O H groups. Such properties usually originate in the mole-
drogen bond will of necessity be broken. If the excited cule's electron-donor function. Indeed, hC is a strong quench-
configuration remains enolic, isomerization to the trans form er of RB fluorescence, which probably occurs via electron
upon excitation would also preclude hydrogen bonding in transfer as in other phenols.26 Curcumin is also a moderate
the excited state. In frozen glasses, where the enolic trans lo2quencher.
isomerization must proceed quite slowly if at all and hydro- Curcumin and hC appear to be potent generators of the
gen bonding may remain mostly intact in excited states, cur- superoxide radical in nonaqueous solution. Unlike lo2gen-
cumin phosphorescence is dramatically decreased compared eration, this production is not restricted to entirely aprotic
to hC. As would be expected, the phosphorescence intensity environments, and 02'- is formed in ethanol. Superoxide
then begins to depend on the energy of the exciting photons. production was not observed in aqueous solution. It may
In protic solvents such as ethanol, the fluorescence of cur- involve a bimolecular process requiring higher dye concen-
cumin is strongly red-shifted compared to hC (Figs. l and 2; trations, which could not be achieved due to curcumin's low
Table 1). The magnitude of the red shift (- 100 nm) implies solubility in water.
that both parts of the curcumin molecule are involved in the In contrast, curcumin photodegradation appears to be a
excited singlet state. The large red shift observed in the cur- monomolecular process independent of solvent or micelli-
cumin fluorescence spectra on going from nonpolar (toluene) zation and thus may proceed from the singlet state. Photo-
to polar solvents (EtOH, H 2 0 ) also indicates that the singlet degradation does not produce a better lo2sensitizer than
state must be very polar. Interaction between the two chro- curcumin itself but leads to oxygen consumption that may
mophores is undoubtedly also responsible for the higher flu- be mediated by the carbon-centered radicals.
orescence yield (by cu two orders of magnitude) of curcumin In conclusion, our spectral observations indicate that the
compared to hC (Table 1). two feruloyl moieties in the curcumin molecule probably d o
The phosphorescence of curcumin is also red-shifted when not act as independent chromophores but interact strongly
compared to hC. Thus, all emission properties are consistent with each other, due either to the conjugation of their
with strong interaction between the two feruloyl chromo- ?r-electrons in the enolic configurations or to the electron-
phores and with the absence of intramolecular hydrogen donating properties of one end of the molecule. Certainly in
bonding in excited curcumin. The absorption and emission the enolic configuration that predominates in unirradiated
spectra of curcumin exhibit some structure in toluene or curcumin, the feruloyl chromophore on the ketone side (the
Triton X- I00 micelles; the structure vanishes in polar sol- hC analog) cannot be solely responsible for the luminescence
vents such as acetonitrile or water. The structure is also the properties. It is the interaction of the two chromophores that
most striking feature of all hC spectra in aprotic and nonpolar allows the photochemical reactions of curcumin to be initi-
solvents. At least four gaussian bands are present in the ab- ated by visible light.
sorption and fluorescence spectra of hC; similar structure is The photochemical properties of both curcumin and hC
also seen in the phosphorescence spectrum at 77 K (Fig. 4). depend strongly on their environment. Depending on the
This structure may be vibronic or may originate in different localization of curcumin in biological systems, singlet oxy-
electronic transitions from the different functionalities pres- gen, superoxide and products of photodegradation/oxidation
ent in both hC and in the curcumin molecule. Solvent pro- may all participate either individually or collectively in the
tocity may strongly influence the transitions that originate biological photoactivity of the dye. In the accompanying
from the contribution of the ketone groups, which may be paper3 an attempt is made to delineate the relative impor-
important for the photochemical activity of curcumin. tance of these reactive species in curcumin photosensitization
Curcumin and hC both require an aprotic environment to of cultured rat basophilic leukemia cells.
photosensitize lo2formation. Singlet oxygen phosphores-
cence was strongly reduced in alcohols and in aqueous so- REFERENCES

1. Lutomski, V. J., B. Kedzia and W. Debska (1974) Wirkung

des Aethanolextraktesund aktiver Substanzen aus Curcumalonga
§The absorption spectrum of curcumin in Triton x-100 micelles auf Bakterien und Pilze. Planta Med. 26, 9-19.
exhibits some structure, similar to that seen in toluene (Fig. l), 2. Dahl, T. A., W. M. McGowan, M. A. Shand and V. S. Srinivasan
which suggests that the dye may be located deep in the hydro- (1989) Photokilling of bacteria by the natural dye curcumin.
phobic micellar region. In biological formationssuch as liposomes, Arch. Microbiol. 151, 183-185.
the localization of curcumin in the bilayer may be even deeper 3. Dahl, T. A., P. Bilski, K. Reszka and C. F. Chignell (1994)
than in Triton X-100micelles, which would be relevant to the Photocytotoxicity of curcumin. Photochem. Photobiol. 59,290-
mechanism of photobiological activity. 294.
302 a at.
COLINF. CHIGNELL

4. Tannesen, H. H., H. de Vries, J. Karlsen and G. B. van He- 16. Scurlock, R., M. Rougee and R. V. Bensasson (1990) Redox
negouwen (1987) Studies on curcumin and curcuminoids IX: properties of phenols, their relationship to singlet oxygen
investigation of the photobiological activity of curcumin using quenching and their inhibitory effects on benzo(a)pyrene-in-
bacterial indicator systems. J . Pharm. Sci. 76, 37 1-373. duced neoplasia. Free Rad. Res. Commun. 8, 25 1-288.
5. Kunchandy, E. and M. N. A. Rao (1989) Effect of curcumin 17. Yoshioka, M., T. Nishioka and T. Hasegawa (1991) Reaction
on hydroxyl radical generation through Fenton reaction. Int. J. of singlet oxygen with enolic tautomers of P-dicarbonyls, a-hy-
Pharm. 57, 173-176. droxylation of 2-acyl- and 2-carboalkoxy-3,4-dihydronaphta-
6. Kunchandy, E. and M. N. A. Rao (1990) Oxygen radical scav- len-I(2H)-ones. Tetrahedron Lett. 32, 1471-1474.
enging activity of curcumin. Int. J. Pharm. 58, 237-240. 18. Cramer, L. E. and K. G. Spears (1 978) Hydrogen bond strengths
7 . Melhuish, W. H. (1961) Quantum efficiencies of fluorescence from solvent-dependent lifetimes of rose bengal dye. J. Am.
of organic substances: effect of solvent and concentration of the Chem. Soc. 100, 221-227.
fluorescent solute. J. Phys. Chem. 65, 229-235. 19. Rodgers, M. A. J. (1983) Solvent-induced deactivation ofsin-
8. Meech, S. R. and D. Phillips (1983) Photophysics of some glet oxygen: additivity relationships in nonaromatic solvents. J.
common fluorescence standards. J . Photochem. 23, 193-2 17. Am. Chem. SOC.105, 6201-6205.
9. Bilski, P., R. Dabestani and C. F. Chignell (1991) Influence 20. Reszka, K. and C . F. Chignell (1991) Spin-trapping of the
of cationic surfactant on the photoprocesses of eosine and rose superoxide radical in aprotic solvents. Free Rad. Res. Commun.
bengal in aqueous solution. J. Phys. Chem. 95, 5784-579 1. 14, 97-106.
10. Hall, R. D. and C. F. Chignell (1 987) Steady-state near infrared 21. Reszka, K., P. Bilski, R. H. Sik and C . F. Chignell (1992)
detection of singlet molecular oxygen. A Stern-Volmer quench- Photosensitized generation of superoxide radical in aprotic sol-
ing experiment with sodium azide. Photochem. Photobiol. 45, vents: an EPR and spin trapping study. Free Rad. Res. Commun.
459464. 19, S33-S44.
1 I . Rossbroich, G., N. A. Garcia and S. E. Braslavsky (1985) Ther- 22. Tannesen, H. H., J. Karlsen and G. B. van Henegouwen (1986)
mal-lensing measurements of singlet molecular oxygen (IAJ: Studies on curcumin and curcuminoids VIII. Photochemical
quantum yields of formation and lifetimes. J. Photochem. 31, stability of curcumin. Z. Lebensm. Unfers. Forsch. 183, 116-
37-47. 122.
12. Bilski, P., A. S. W. Li and C. F. Chignell (1991) The photo- 23. Weedon, A. C. (1990) Photochemical reactions involving enols.
oxidation of N,N-diethylhydroxylamine by rose bengal in ace- In The Chemistry of Enols. (Edited by Z. Rapport), pp. 591-
tonitrile and water. Photochem. Photobiol. 54, 345-352. 638. Wiley & Sons, Chichester.
13. Tannesen, H. H. and J. Karlsen (1985) Studies on curcumin 24. Emsley, J. (1984) The composition, structure and hydrogen
and curcuminoids. VI. Kinetics of crucumin degradation in bonding of the 6-diketones. Struct. Bond. 57, 147-1 9 1.
aqueous solution. Z. Lebensm. Untersuch. Forsch. 180, 402- 25. Barbara, P. F., P. K. Walsh and L. E. Brus (1989) Picosecond
404. kinetic and vibrationally resolved spectroscopic studies of in-
14. Suter, G. W., U. P. Wild and K. Schaffner (1986) Selective tramolecular excited-state hydrogen transfer. J . Phys. Chem. 93,
photobleaching of the long-lived phosphorescence of 29-34.
I-indanones. J. Phys. Chem. 90, 2358-2361. 26. Jones, G. I1 and S. Chatterjee (1988) Steric control of distance
15. Eisenberger, H., B. Nickel, A. A. Ruth, W. Al-Soufi, K. H. parameters and the yield of charge camers in photochemical
Grellmann, and M. Novo ( 1 99 1) Keto-enol tautomenzation electron transfer. The quenching of eosine Y by hindered phe-
of 2-(2’-hydrohyphenyl)benzaxazoleand 2-(2‘-hydroxy-4’-me- nols. J. Phys. Chem. 92, 6862-6864.
toxyphenyl)benzoxazole in the triplet state: hydrogen tunneling
and isotope effects. 2. Dual phosphorescence kinetics. J. Phys.
Chem. 95, 10509-10518.