Quenching of triplet-excited flavins by flavonoids. Structural assessment of antioxidative activity

Article abstract of DOI:10.1021/jo901301c

(Figure Presented) The mechanism of flavin-mediated photooxidation of flavonoids was investigated for aqueous solutions. Interaction of triplet-excited flavin mononucleotide with phenols, as determined by laser flash photolysis, occurred at nearly diffusion-controlled rates (k～1.6x10 ⁹ Lmol^-1 s^-1 for phenol at pH 7, 293 K), but protection of the phenolic function by methylation inhibited reaction. Still, electron transfer was proposed as the dominating mechanism due to the lack of primary kinetic hydrogen/ deuterium isotope effect and the low activation enthalpy (<20 kJ mol^-1) for photooxidation. Activation entropy worked compensating in a series of phenolic derivatives, supporting a common oxidation mechanism. Anortho-hydroxymethoxy pattern was equally reactive (k～2.3x10⁹Lmol^-1 s^-1 for guaiacol at pH 7) as compounds with ortho-dihydroxy substitution (k～2.4x10⁹ L mol^-1 s^-1 for catechol at pH 7), which are generally referred to as good antioxidants. This refutes the common belief that stabilization of incipient phenoxyl radicals through intramolecular hydrogen bonding is the driving force behind the reducing activity of catechol-like compounds. Instead, such bonding improves ionization characteristics of the substrates, hence the differences in reactivity with (photo)oxidation of isolated phenols. Despite the similar reactivity, radicals from ortho-dihydroxy compounds are detected in high steady-state concentrations by electron paramagnetic resonance (EPR) spectroscopy, while those resulting from oxidation of ortho-hydroxymethoxy (or isolated phenolic) patterns were too reactive to be observed. The ability to deprotonate and form the corresponding radical anions at neutral pH was proposed as the decisive factor for stabilization and, consequently, for antioxidative action. Thus, substituting other ionizable functions for the ortho- or para-hydroxyl in phenolic compounds resulted in stable radical anion formation, as demonstrated for para-hydroxybenzoic acid, in contrast to its methyl ester. 2009 American Chemical Society.

Full text of DOI:10.1021/jo901301c

pubs.acs.org/joc
Quenching of Triplet-Excited Flavins by Flavonoids. Structural
Assessment of Antioxidative Activity
Kevin Huvaere, Karsten Olsen, and Leif H. Skibsted*
Food Chemistry, Department of Food Science, Faculty of Life Sciences, University of Copenhagen,
Rolighedsvej 30, DK-1958 Frederiksberg C, Denmark
ls@life.ku.dk
Received June 17, 2009
The mechanism of flavin-mediated photooxidation of flavonoids was investigated for aqueous
solutions. Interaction of triplet-excited flavin mononucleotide with phenols, as determined by laser
flash photolysis, occurred at nearly diffusion-controlled rates (k ∼ 1.6 ꢀ 109 L mol-1 s-1 for phenol at
pH 7, 293 K), but protection of the phenolic function by methylation inhibited reaction. Still, electron
transfer was proposed as the dominating mechanism due to the lack of primary kinetic hydrogen/
deuterium isotope effect and the low activation enthalpy (<20 kJ mol-1) for photooxidation.
Activation entropy worked compensating in a series of phenolic derivatives, supporting a common
oxidation mechanism. An ortho-hydroxymethoxy pattern was equally reactive (k ∼ 2.3 ꢀ 109 L mol-1
s-1 for guaiacol at pH 7) as compounds with ortho-dihydroxy substitution (k ∼ 2.4 ꢀ 109 L mol-1 s-1
for catechol at pH 7), which are generally referred to as good antioxidants. This refutes the common
belief that stabilization of incipient phenoxyl radicals through intramolecular hydrogen bonding is
the driving force behind the reducing activity of catechol-like compounds. Instead, such bonding
improves ionization characteristics of the substrates, hence the differences in reactivity with
(photo)oxidation of isolated phenols. Despite the similar reactivity, radicals from ortho-dihydroxy
compounds are detected in high steady-state concentrations by electron paramagnetic resonance
(EPR) spectroscopy, while those resulting from oxidation of ortho-hydroxymethoxy (or isolated
phenolic) patterns were too reactive to be observed. The ability to deprotonate and form the
corresponding radical anions at neutral pH was proposed as the decisive factor for stabilization and,
consequently, for antioxidative action. Thus, substituting other ionizable functions for the ortho- or
para-hydroxyl in phenolic compounds resulted in stable radical anion formation, as demonstrated
for para-hydroxybenzoic acid, in contrast to its methyl ester.
Introduction
(particularly red wine) induces beneficial health effects.1,2
Antioxidant activity of polyphenols has been attributed to a
pivotal role herein, particularly since various pathologies
have been associated with metabolic disorders that provoke
excessive formation of oxidizing species (oxidative stress).
Polyphenols, as secondary metabolites, are abundant in
the plant kingdom, and their structural diversity is virtually
infinite. Epidemiological studies suggest that high intake of
polyphenol-rich foods (fruits and vegetables) and beverages
ꢀ
ꢀ
ꢀ
(1) Scalbert, A.; Manach, C.; Morand, C.; Remesy, C.; Jimenez, L. Crit.
*To whom correspondence should be addressed. E-mail: ls@life.ku.dk.
Phone: þ 45 35 33 32 21. Fax: þ 45 35 33 33 44.
Rev. Food Sci. Nutr. 2005, 45, 287–306.
(2) Renaud, S.; de Lorgeril, M. Lancet 1992, 339, 1523–1526.
DOI: 10.1021/jo901301c
r
Published on Web 09/08/2009
J. Org. Chem. 2009, 74, 7283–7293 7283
2009 American Chemical Society

Huvaere et al.
JOCArticle
CHART 1. Structures of Riboflavin and Its Congeners, Flavin
Mononucleotide and Flavin Adenine Dinucleotide
light,11-13 but damage could be inhibited by vitamin C.14
Likewise, antioxidative properties of vitamin E and related
model compounds (Trolox) were reported to deactivate
pivotal triplet-excited flavins.15-17 Still, physiological con-
centrations of vitamins C and E are limited, but the intake of
supplementary antioxidants such as flavonoids (an impor-
tant subclass of the polyphenol series) through particular
diets possibly assists in inhibiting biological damage caused
by photogenerated oxidizing species.18-21 A similar mecha-
nism was accordingly suggested for protection of light-
exposed foods and beverages.22,23
Since reactive triplet-excited flavin was considered to be a
pivotal intermediate in both type I and type II photooxida-
tion reactions, the possibility of its quenching by flavonoids
was investigated. Insight in the quenching mechanism was
the aim, and therefore, a systematic set of relevant flavonoids
was subjected to a series of kinetic experiments involving
laser flash photolysis. To determine whether hydrogen ab-
straction or electron transfer accounted for quenching, the
occurrence of a kinetic isotope effect was investigated and
activation parameters were calculated. Radical intermedi-
ates were identified by electron paramagnetic resonance
(EPR) spectroscopy, but only after generation of detectable
steady-state concentrations. Eventually, results were com-
bined to propose a molecular explanation that contributes to
the general understanding of antioxidative activity and
particularly to the protection against light-induced damage.
On the other hand, prevalence of polyphenols protects the
food matrix itself against oxidative instability, thus reducing
intake of toxic oxidation products (such as peroxides) and
provoking an indirect health benefit.3 Putative molecular
culprits in the mentioned oxidation reactions are, among
others, superoxide (O2•-), as precursor for the reactive
hydroperoxyl radical (HOO•), hydroxyl radicals (HO•) pro-
duced by Fenton chemistry, and lipid-derived peroxyl radi-
cals (ROO•). Radical attack accounts for breakdown of
biomolecules, but also singlet oxygen (1O2), a nonradical
reactive oxygen species, induces harmful peroxide (ROOH)
formation.4-6 Singlet oxygen is generated from various
sources, of which photosensitization (so-called type II
photooxidation) has been extensively studied. Suitable sen-
sitizers, such as riboflavin (vitamin B2), flavin mononucleo-
tide (FMN), and flavin adenine dinucleotide (FAD)
(Chart 1), are endogenous, as they prevail in the cellular
respiration cycle and act as important cofactors or are
present as essential nutritients (as in dairy products). Their
photoreactivity is due to the isoalloxazine moiety, which
accounts for absorption maxima around 375 and 445 nm.7
Exposure to UV-A (from ∼320 to ∼ 400 nm) or to blue light
yields a singlet-excited state, which undergoes efficient inter-
system crossing to triplet-excited flavin. Besides transferring
energy to molecular oxygen to produce 1O2, the triplet state
acts as a powerful oxidant (E ∼ 1.7 V) capable of oxidizing
various organic compounds (type I photooxidation).8 Cel-
lular structures have been found to be sensitive to light
exposure, and reaction products of flavin-mediated photo-
reactions of tyrosine and tryptophan residues were found to
be lethal to human and mammalian cells.9,10 Moreover,
promutagenic DNA lesions were detected in the genomic
sequence after exposing flavin-enriched mammalian cells to
Results
Kinetic Analyses of Flavonoid Photooxidation. Reactivity
of flavonoids toward short-lived triplet-excited flavin mole-
cules was investigated in a set of kinetic analyses based on
laser flash photolysis coupled to transient absorption spec-
troscopy. Flavin mononucleotide was preferred as light-
absorbing species, as the phosphate moiety increased solu-
bility without compromising photochemical properties.
However, due to poor solubility of various substrates, buffer
solutions were mixed with acetonitrile. The significant absorp-
tion of flavon-3-ol compounds (for structures, see Chart 2) in the
near-UV-A region led to the choice of a 440 nm laser pulse as
excitation source. Subsequent transient absorption spectroscopy
produced time-resolved spectra in which the band around
720 nm was exclusively attributed to triplet-excited flavin mono-
nucleotide (3FMN*) (Figure 1). Its lifetime, monitored at
the respective wavelength, followed an exponential decay
and was significantly reduced in the presence of the selected
substrates (Figure 2). The observed pseudo-first-order rate
(3) Aruoma, O. I. Food Chem. Toxicol. 1994, 32, 671–683.
(4) Davies, M. J. Biochem. Biophys. Res. Commun. 2003, 305, 761–770.
(5) Wright, A.; Hawkins, C. L.; Davies, M. J. Free Radical Biol. Med.
2003, 34, 637–647.
(6) Girotti, A. W. J. Lipid Res. 1998, 39, 1529–1542.
(7) Grodowski, M. S.; Veyret, B.; Weiss, K. Photochem. Photobiol. 1977,
26, 341–352.
(14) Pflaum, M.; Kielbassa, C.; Garmyn, M.; Epe, B. Mutat. Res. 1998,
408, 137–146.
(15) McVean, M.; Liebler, D. C. Mol. Carcinog. 1999, 24, 169–176.
(16) Cardoso, D. R.; Olsen, K.; Skibsted, L. H. J. Agric. Food Chem.
2007, 55, 6285–6291.
ꢀ
(17) Gutierrez, I.; Criado, S.; Bertolotti, S.; Garcıa, N. A. J. Photochem.
´
Photobiol. B 2001, 62, 133–139.
(8) Sun, M.; Moore, T. A.; Song, P.-S. J. Am. Chem. Soc. 1972, 94, 1730–
1740.
(18) Kullavanijaya, P.; Lim, H. W. J. Am. Acad. Dermatol. 2005, 52, 937–
958.
(9) Nixon, B. T.; Wang, R. J. Photochem. Photobiol. 1977, 26, 589–593.
(10) Stoien, J. D.; Wang, R. J. Proc. Natl. Acad. Sci. U.S.A. 1974, 71,
3961–3965.
(19) Sies, H.; Stahl, W. Annu. Rev. Nutr. 2004, 24, 173–200.
(20) Swindells, K.; Rhodes, L. E. Photodermatol. Photoimmunol. Photo-
med. 2004, 20, 297–304.
(11) Besaratinia, A.; Kim, S.-I.; Bates, S. E.; Pfeifer, G. P. Proc. Natl.
Acad. Sci. U.S.A. 2007, 104, 5953–5958.
(21) F’guyer, S.; Afaq, F; Mukhtar, H. Photodermatol. Photoimmunol.
Photomed. 2003, 19, 56–72.
^
(12) Drobetsky, E. A.; Turcotte, J.; Chateauneuf, A. Proc. Natl. Acad.
(22) Becker, E. M.; Cardoso, D. R.; Skibsted, L. H. Eur. Food Res.
Technol. 2005, 221, 382–386.
(23) Cardoso, D. R.; Olsen, K.; Moller, J. K. S.; Skibsted, L. H. J. Agric.
Food. Chem. 2006, 54, 5630–5636.
Sci. U.S.A. 1995, 92, 2350–2354.
(13) Hiraku, Y.; Ito, K.; Hirakawa, K.; Kawanishi, S. Photochem.
Photobiol. 2007, 83, 205–212.
7284 J. Org. Chem. Vol. 74, No. 19, 2009

Huvaere et al.
JOCArticle
TABLE 1. Second-Order Rate Constants k (109 L mol-1 s-1)a for the
Interaction of Flavonoids with Triplet-Excited Flavin Mononucleotide in
Aqueous Buffer Mixed with Acetonitrile
pH
compound
4.2
5.4
c
7.0
8.5
10.0
flavanone
chrysin
naringenin
apigenin
kaempferol
luteolin
nqb
1.0
1.2
1.1
1.5
2.0
1.9
1.4
1.8
/
nq
/
nq
0.9
1.1
/
1.0
1.1
/
/
/
2.0
/
1.6
0.8
1.0
1.1
1.3
2.0
1.9
1.2
1.6
1.0
1.1
/
/
/
1.2
/
1.3
xd
/
quercetin
rutine
x
x
1.6
catechine
FIGURE 1. Transient absorption spectra of triplet-excited flavin
recorded 0.1 μs (a), 0.5 μs (b), 1 μs (c), 5 μs (d), 10 μs (e), and 100 μs
(f) after the excitation pulse.
aExperimental error on reported values is e10%. bNo quenching
observed. cNot determined. dData collection hampered by instability of
e
compounds at high pH. 1.0 ꢀ 109 and 1.4 ꢀ 109 L mol-1 s-1 for the
interaction of rutin and catechin with triplet-excited riboflavin, respec-
tively, in a mixture of acetonitrile and pH 6.8 buffer (see ref 22).
TABLE 2. Second-Order Rate Constants k (109 L mol-1 s-1)a for
the Interaction of Selected Phenol Derivatives with Triplet-Excited
Flavin Mononucleotide in Aqueous Buffer Mixed with Acetonitrile
(Oxidation Peak Potentials, Ep (Expressed in V vs SHE), Determined
in Acetonitrile)
pH
compound
4.2
5.4
7.0
8.5
10.0
Ep
phenol derivatives
phenolb
anisole
catechol
guaiacole
20,40-DHPf
5-MRg
1.7
/
1.7
1.6
nqd
2.4
2.3
0.8
1.8
nq
1.4
/
2.1
/
/
/
/
1.6
/
1.67
2.01
1.23
1.39
/
c
/
2.3
2.8
2.5
0.8
1.9
nq
2.0
2.6
1.4
1.8
nq
/
/
/
/
FIGURE 2. Transient absorption traces of triplet-excited flavin
mononucleotide (3FMN*), observed at 720 nm, in the presence of
varying concentrations of catechin, (a) 0 μM, (b) 125 μM, (c) 250
μM, (d) 375 μM, (e) 500 μM, at pH 7.0. Inset: Linear plot of the
observed pseudo-first-order rate constants for the decay of 3FMN*
as a function of the concentration of catechin. The nonzero intercept
corresponds to the triplet lifetime in the absence of quenchers.
/
/
chromanone
aExperimental error on reported values is e10%. b0.5 ꢀ 109 L mol-1
s-1 for the interaction with triplet-excited riboflavin in methanol (see
ref 24). cNot determined. dNo quenching observed. e2.6 ꢀ 109 L mol-1 s-1
for the interaction with triplet-excited riboflavin in a mixture of acet-
onitrile and pH 4.6 buffer (see ref 23). f20,40-Dihydroxypropiophenone.
g5-Methoxyresorcinol.
constants for quenching (kobs) increased proportional to the
polyphenol concentration, thus corresponding second-order rate
constants were obtained from the slope of the resulting plot
(Figure 2, inset). Except for flavanone, reaction rates were high
(corresponding to rate constants ∼109 L mol-1 s-1) and not
strongly affected by pH change (Table 1). However, kinetic
monitoring of flavon-3-ol reactions at high pH was hindered,
probably as a result of autoxidation reactions. Oxidation in the
presence of an ortho-dihydroxy or an ortho-hydroxymethoxy
pattern was generally faster than for isolated phenols, which was
confirmed by the analysis of model systems (Table 2; for
structures, see Chart 3). Furthermore, the lack of phenolic
function, as in chromanone and anisole, failed to provoke
quenching and concurred with the findings for flavanone.
Hydrogen Abstraction versus Electron Transfer. Oxidation
reactions of (poly)phenolic compounds have been previously
associated with hydrogen atom abstraction from a phenolic
function or with electron transfer followed by proton migra-
tion from the resulting (poly)phenol radical cation.
Although the net result of both mechanisms is identical,
the reaction pathways are determined by different thermo-
dynamic and kinetic properties. One-step hydrogen atom
abstraction mainly depends on the bond dissociation energy
(BDE) and is hampered by hydrogen bonding. On the
contrary, electron transfer, determined by the substrate’s
ionization potential (IP), is favored in protic solvents due to
stabilization of charged intermediates. In an attempt to
identify the major pathway in flavin-mediated photooxida-
tion of phenolic compounds, the occurrence of a primary
kinetic isotopic effect (kH/kD) was investigated. Photooxida-
tion rate constants obtained for phenol and phenol-d6 (1.5 ꢀ
109 and 1.1 ꢀ 109 L mol-1 s-1, respectively) resulted in a
small kH/kD of 1.4. Hydrogen transfer is thus refuted, a result
that concurs with the information extracted from activation
parameters that were calculated according to the Eyring
equation (eq 1). Activation enthalpy (ΔHq) of phenol and
derivatives (Table 3 and Figure 3) was relatively low and in
agreement with electron transfer rather than a process of
bond breaking and bond formation in the transition state.
Slightly higher values were observed for catechol-like struc-
tures, most likely because oxidation involves breaking of an
intramolecular hydrogen bond. Negative activation entro-
pies (ΔSq) accommodate major solvent rearrangements in
´
(24) Haggi, E.; Bertolotti, S.; Garcıa, N. A. Chemosphere 2004, 55, 1501–
1507.
J. Org. Chem. Vol. 74, No. 19, 2009 7285

Huvaere et al.
JOCArticle
TABLE 3. Activation Parameters Calculated for the Interaction of a
Series of Phenolic Compounds with Triplet-Excited Flavin Mononucleo-
tide in Phosphate Buffer (pH 7) Mixed with Acetonitrile (1:1, v/v)
compound
ΔHq (kJ mol-1
)
ΔSq (J K-1 mol-1
)
phenol
catechol
chrysin
naringenin
catechin
quercetin
8.7 ( 0.5
16.1 ( 1.4
12.3 ( 2.0
10.5 ( 3.3
11.6 ( 3.6
13.8 ( 3.3
-39 ( 2
-11 ( 5
-31 ( 7
-35 ( 11
-29 ( 12
-20 ( 11
FIGURE 4. Formation and decay of the quercetin radical, mea-
sured at 485 nm, after flash photolysis of a mixture of flavin
mononucleotide and quercetin in buffer (pH 4.2) mixed with
acetonitrile (1:1, v/v).
FIGURE 5. EPR trace measured in time sweep mode at static
magnetic field during irradiation of a mixture of flavin mono-
nucleotide and rutin in buffer (pH 7.0) mixed with acetonitrile
(1:1, v/v).
FIGURE 3. Top panel: Eyring plot for the oxidation of phenol by
triplet-excited flavin mononucleotide (3FMN*) in a mixture of
phosphate buffer (pH 7.0) and acetonitrile (1:1, v/v). Bottom panel:
Isokinetic plot for the deactivation of 3FMN* by various phenolic
compounds.
stationary 440 nm irradiation source for the laser flash
produced detectable steady-state concentrations of radicals,
as shown by changes in signal intensity (at fixed magnetic
field) during irradiation of rutin at pH 7 (Figure 5). Except
for quercetin, which produced weak and unresolved spectra
(Figure 6), intense spectra were observed for catechol-con-
taining flavonoids (Figure 7). Corresponding coupling con-
stants, presented in Table 4, were obtained from appropriate
simulations. Assignment was based on literature data, but
the transition state and corroborate solvent-assisted electron
transfer followed by proton exchange. Entropy control was
more pronounced in substrates lacking a catechol unit,
resulting in a compensation effect (Figure 3). The latter
supports a common oxidation mechanism for the investi-
gated compounds.
0
0
coupling to hydrogens on C5 and C6 (or C5 and C6 for
caffeic acid) as proposed in Kuhnle’s seminal work26 has
been questioned. According to van Acker et al., the highest
ꢀ
ꢁ
k ꢀ h
ΔSq ΔHq
ln
¼
-
ð1Þ
0
0
spin density prevailed at C5 , and consequently, C5 -H was
T ꢀ kB
R
RT
associated with the largest coupling constant.27 However,
0
the radical from a catechin derivative that has C6 occupied
Polyphenol-Derived Radicals. Electron transfer eventually
produced polyphenol-derived radical intermediates, but the
detection window for transient absorption spectroscopy was
narrowed by the accumulated absorption of the flavin triplet
state and its reduced radical, FMNH• (pKa ∼ 8.3),25 in the
500-730 nm region and flavin photobleaching around
440 nm (ground state absorption) (Figure 1). Still, radical
formation from flavon-3-ol derivatives such as quercetin and
rutin, with maximum absorption around 485 nm, was readily
observed (Figure 4). It should be noted that detection of
radicals by electron paramagnetic resonance spectroscopy is
more informative, but response time is generally too slow to
detect short-lived intermediates. However, substitution of a
only produced two small coupling constants (2 ꢀ 1.1 G)
28
0
0
assigned to C2 -H and C5 -H. Hence, the original
0
identification according to Kuhnle, with C6 -H accounting
for major coupling, was followed.
(26) Kuhnle, J. A.; Windle, J. J.; Waiss, A. C. J. Chem. Soc. B 1969, 613–
616.
(27) van Acker, S. A. B. E.; de Groot, M. J.; van den Berg, D.-J.; Tromp,
ꢀ
M. N. J. L.; Donne-Op den Kelder, G.; van der Vijgh, W. J. F.; Bast, A.
Chem. Res. Toxicol. 1996, 9, 1305–1312.
(28) Fukuhara, K.; Miyata, N.; Ikota, N.; Fukuzumi, S.; Nakanishi, I.;
Shimada, T.; Ohkubo, K.; Miyazaki, K.; Hakamata, W.; Urano, S.; Ozawa,
T.; Okuda, H. Chem. Res. Toxicol. 2003, 16, 81–86.
(29) Tomkiewicz, M.; Goren, A.; Cocivera, M. J. Am. Chem. Soc. 1971,
93, 7102–7103.
(25) Heelis, P. F.; Parsons, B. J.; Phillips, G. O.; McKellar, J. F. Photo-
chem. Photobiol. 1978, 28, 169–173.
(30) Adams, M.; Blois, M. S.; Sands, R. H. J. Chem. Phys. 1958, 28, 774–
776.
7286 J. Org. Chem. Vol. 74, No. 19, 2009

Huvaere et al.
JOCArticle
CHART 2. Structures of Flavonoids Investigated in This Study
Contrary to quercetin, its C3-glycoside rutin generated
intense signals which included coupling to the three remain-
ing B-ring hydrogens. Although invisible in this spectrum,
spin density was delocalized along the C2-C3 unsaturation,
as demonstrated by the C3-H coupling in the luteolin-
derived radical. On the other hand, flavanones and flavan-
3-ols showed an extra coupling to C2-H on the C-ring, the
magnitude of which was strongly variable. The trans-dis-
position of C3-OH (in reference to C2-aryl) in catechin and
taxifolin resulted in a relatively small coupling compared to
epicatechin (cis-OH). The dihedral angle θ, determined
between the C2-H bond and the radical pz-orbital (which
is perpendicular to the plane of the B-ring), is directly related
to the magnitude of the respective coupling constant via the
McConnell equation (aH ∼ cos2 θ).39 The smaller θ in the
eriodictyol radical, as concluded from comparing its opti-
mized geometry with that of the taxifolin radical, corrobo-
rates the higher coupling constant (Figure 8).
For compounds that lack a catechol ring, apigenin pro-
duced detectable radicals, but no spectra were observed
for naringenin and kaempferol. Still, the interpretation of
the apigenin spectrum was confusing, as the spectrum was
strikingly similar to that of luteolin. Furthermore, two
radical species were detected for chrysin (aH ∼ 1.8 and 1.1
G and aH ∼ 2.0 and 3.3 G, respectively) which, obviously,
could not be ascribed to radical formation in the B-ring, but
they were also considerably different from the chrysin radical
FIGURE 6. (A) EPR spectrum measured during irradiation of a
mixture of flavin mononucleotide and quercetin in buffer (pH 7.0)
with acetonitrile (1:1, v/v). (B,C) EPR spectra measured prior to (B)
and during (C) irradiation of a mixture of flavin mononucleotide
and quercetin in buffer (pH 10.0) with acetonitrile (1:1, v/v).
(31) Dixon, W. T.; Murphy, D. J. Chem. Soc., Faraday Trans. 2 1976, 72,
1221–1230.
(32) Felix, C. C.; Sealy, R. C. J. Am. Chem. Soc. 1982, 104, 1555–1560.
(33) Bors, W.; Michel, C.; Stettmaier, K.; Lu, Y.; Foo, L. Y. Biochim.
Biophys. Acta 2003, 1620, 97–107.
(34) Cotelle, N.; Bernier, J.-L.; Catteau, J.-P.; Pommery, J.; Wallet, J.-C.;
Gaydou, E. M. Free Radical Biol. Med. 1996, 20, 35–43.
(35) Bors, W.; Michel, C.; Stettmaier, K.; Heller, W. In Natural Antiox-
idants. Chemistry, Health Effects, and Applications; Shahidi, F., Ed.; AOCS
Press: Champaign, IL, 1996; pp 346-357.
(36) Le Nest, G.; Caille, O.; Woudstra, M.; Roche, S.; Burlat, B.; Belle,
V.; Guigliarelli, B.; Lexa, D. Inorg. Chim. Acta 2004, 357, 2027–2037.
(37) Bors, W.; Michel, C.; Stettmaier, K. Arch. Biochem. Biophys. 2000,
374, 347–355.
(38) Jensen, O. N.; Pedersen, J. A. Tetrahedron 1983, 39, 1609–1615.
(39) Babcock, G. T.; El-Deeb, M. K.; Sandusky, P. O.; Whittaker, M. M.;
Whittaker, J. W. J. Am. Chem. Soc. 1992, 114, 3727–3734.
FIGURE 7. EPR spectra (top traces) and corresponding simula-
tions (bottom traces) obtained during irradiation of polyphenols in
the presence of flavin mononucleotide: (A) luteolin, (B) rutin, (C)
catechin, (D) epicatechin, (E) eriodictyol, (F) taxifolin.
J. Org. Chem. Vol. 74, No. 19, 2009 7287

Huvaere et al.
JOCArticle
TABLE 4. Coupling Constants (in Gauss) and Resonance Parameters of Radicals Generated by Irradiation of Phenolic Model Compounds and
Polyphenols in the Presence of Flavin Mononucleotide Compared to Reference Values
b
aCoupling with hydrogen attached to the respective carbon atom. Values from literature sources were given at 0.1 G accuracy. g value (x = not
given). cSpectral intensity, as referred to the intensity of the spectrum of an aqueous 2.0 μM TEMPO solution. Catechin and epicatechin intensities were
only compared mutually due to nonlinearity between signal intensity and modulation amplitude. dModulation amplitude in Gauss (var. = variable, x =
not given). References [in order of appearance; tw: this work, i.e., flavin-mediated photooxidation in a mixture of phosphate buffer (pH 7) and
e
acetonitrile (1:1, v/v); 29: photolysis of alkaline solution; 30: quinone reduction with metallic zinc in alkaline solution; 31: cerium(IV) oxidation; 32:
photolysis of aqueous solution; 33: alkaline autoxidation at pH 9.5; 26: autoxidation in DMSO/aqueous sodium hydroxide (4:1); 27: alkaline
autoxidation at pH 13; 34: alkaline autoxidation by addition of NaOH; 35: enzymatic oxidation (horse radish peroxidase) at pH 8.5; 36: in situ
electrolysis in aqueous Tris buffer/DMSO (1:1, v/v) in the presence of zinc acetate; 37: enzymatic oxidation (horse radish peroxidase) at pH 9.0; 38:
alkaline autoxidation at pH 12.7 in aqueous DMSO]. fFor the sake of uniformity, numbering of carbon atoms in model systems was arbitrarily set.
Numbering in polyphenols is similar as used in Chart 2. g4-Hydroxybenzoic acid. h3,4-Dihydroxybenzoic acid. iCoupling constants with the olefinic
protons are listed in the first column. jCoupling to C5 according to Kuhnle26 was assigned to C6 (and vice versa) by van Acker et al.27 (see text).
0
0
detected after oxidation by cerium(IV).40 In the absence of
phenolic functions, like in flavanone, EPR measurements
remained silent and concurred with the lack of reactivity seen
in the kinetic analyses.
In an attempt to obtain practicable spectra from quercetin,
so-called spin stabilization was applied.41 Thus, incipient
ortho-semiquinone radicals were complexated by addition
of Zn2þ, but no stabilizing effect was observed. Spectral
intensity remained invariably low for quercetin, while pre-
ceding intense signals, for example, from taxifolin, practi-
cally disappeared. Alternatively, irradiations were carried
out at pH 10.0, which was reported to increase radical
stability.38 Thus, rutin photooxidation resulted in a very
intense EPR response, although minor radical formation
was already detected in the nonexposed control sample.
Probably residual oxygen accounted for autoxidation, a
similar problem that afflicted flash photolysis experiments.
Under these conditions, radicals were also detected in un-
exposed quercetin solutions, but only a single line emerged
during irradiation (Figure 6). From the latter experiment, it
was assumed that quercetin radicals readily disappeared
through interaction with 3FMN*, rather than by dispropor-
tionation.
In an attempt to characterize and confirm radical struc-
tures, relevant model compounds were investigated under
identical reactions conditions (Figure 9). Phenol produced
an identical spectrum as hydroquinone, but intensity of the
latter was ∼3 times higher. The observed coupling to four
equivalent hydrogens indicated higher symmetry than the
2 ꢀ 2 coupling pattern for the regioisomer catechol. Blocking
(40) Dixon, W. T.; Moghimi, M.; Murphy, D. J. Chem. Soc., Perkin
Trans. 2 1975, 101–103.
(41) Stegmann, H. B.; Bergler, H. U.; Scheffler, K. Angew. Chem., Int. Ed.
Engl. 1981, 20, 389–390.
7288 J. Org. Chem. Vol. 74, No. 19, 2009

Huvaere et al.
JOCArticle
CHART 3. Structures of Relevant Phenolic Models
0
FIGURE 8. Newman projection of the C2-C1 bond in the erio-
dictyol radical and the taxifolin radical. The smaller dihedral angle θ
between C2-H and the pz-orbital with unpaired spin (perpendicular
to the plane of the B-ring) accounts for a higher coupling constant.
from polyphenols.42-46 On the other hand, characterization
of incipient radicals via EPR spectroscopy provides most
structural information, but the slow response of continuous-
wave spectrometers is inconsistent with the fast decay of
radical intermediates. Therefore, particular techniques were
developed to produce steady-state concentrations of poly-
phenol radicals at levels detectable on EPR time scale
(enzymatic oxidation, autoxidation at alkaline pH, and
electrochemical oxidation),27,36-38 although sometimes dif-
ferent species than those generated by pulse radiolysis were
detected.47 The present work demonstrates that such high
steady-state radical concentrations are also produced upon
stationary irradiation in the presence of FMN. Moreover,
since excitation of FMN by laser flash photolysis allows
monitoring rise and decay of the reactive 3FMN*, reactivity
of various substrates can be investigated under the same
conditions. Thus, it was found that fast quenching of 3FMN*
by flavonoids was slightly slower than the scavenging of
hydroxyl or azidyl radicals48 but faster than the interaction
with bromide anion radicals (Br2•-) or superoxide.45
a phenolic function, as in p-hydroxyanisole (methylated
hydroquinone) or in guaiacol (methylated catechol), com-
promised radical stability, and steady-state concentrations
became too low to detect. Similarly, cresol (4-methylphenol),
which is an appropriate model for the B-ring in naringenin,
showed only traces of radical formation during irradiation.
On the other hand, a carboxyl moiety instead of a methyl
group, as in 4-hydroxybenzoic acid, produced an intense
EPR signal (in contrast to the methyl ester, which was EPR
silent). A doubled doublet was expected when considering
symmetry, but values slightly differed due to C2-H or C6-H
interaction with the carboyxlate moiety. Symmetry was lost
by the extra hydroxyl in 3,4-dihydroxybenzoic acid, and
three different coupling constants were observed. Basically,
a similar pattern was observed for the vinylogous caffeic
acid, although coupling to two conjugated methine protons
introduced extra lines and raised complexity.
Spectral intensities from irradiations of 2,4-dihydroxypro-
piophenone or 5-methoxyresorcinol were too low to perform
accurate simulations. The meta-substitution pattern, like in
the A-ring of various flavonoids, produced highly reactive
radicals that failed to build up to detectable concentrations.
Chromanone, similar as flavanone, lacked reactivity due to
the absence of free phenolic functions and failed to produce
EPR signals.
Efficient quenching of 3FMN* (E ∼ 1.7 V) by quercetin, a
powerful antioxidant found in dietary sources such as onion
and apple skin, was in agreement with its low reduction
potential at physiological pH (E7.4 ∼ 0.3-0.4 V).44,49 The
number and favorable disposition of the hydroxyl groups,
particularly the ortho-configuration on the B-ring, seem
essential, while other structural features such as a C2-C3
unsaturation and, to a lesser extent, a C4-oxo moiety also
favored antioxidative activity.50,51 Like quercetin, the
flavone luteolin complied with above-mentioned struc-
tural requirements and its low reduction potential (E7.4
∼
0.41 V)49 accounted for a large thermodynamic driving
force. The resulting rate constants were invariably higher
than the analogous apigenin (E7.4 ∼ 0.71 V), which lacks the
ortho-dihydroxy substitution in the B-ring. Rutin, a major
(45) Jovanovic, S. V.; Steenken, S.; Tosic, M.; Marjanovic, B.; Simic,
M. G. J. Am. Chem. Soc. 1994, 116, 4846–4851.
(46) Jovanovic, S. V.; Hara, Y.; Steenken, S.; Simic, M. G. J. Am. Chem.
Soc. 1995, 117, 9881–9888.
Discussion
(47) Bors, W.; Michel, C.; Stettmaier, K. Methods Enzymol. 2001, 335,
166–180.
Pulse radiolysis, with optical spectroscopy on a sub-
second time scale, has proven to be a valuable technique
for obtaining mechanistic insights in behavior of radicals
(48) Bors, W.; Heller, W.; Michel, C.; Saran, M. In Free Radicals and the
Liver; Csomos, G., Feher, J., Eds.; Springer Verlag: Berlin, Germany, 1992;
pp 77-95.
(49) Jorgensen, L. V.; Skibsted, L. H. Free Radical Res. 1998, 28, 335–351.
(50) Rice-Evans, C. A.; Miller, N. J.; Paganga, G. Free Radical Biol. Med.
1996, 20, 933–956.
(51) van Acker, S. A. B. E.; van den Berg, D.-J.; Tromp, M. N. J. L.;
Griffioen, D. H.; van Bennekom, W. P.; van der Vijgh, W. J. F.; Bast, A. Free
Radical Biol. Med. 1996, 20, 331–342.
(42) Bors, W.; Michel, C. Free Radical Biol. Med. 1999, 27, 1413–1426.
(43) Bors, W.; Michel, C.; Schikora, S. Free Radical Biol. Med. 1995, 19,
45–52.
(44) Jovanovic, S. V.; Steenken, S.; Hara, Y.; Simic, M. G. J. Chem. Soc.,
Perkin Trans. 2 1996, 2497–2504.
J. Org. Chem. Vol. 74, No. 19, 2009 7289

Huvaere et al.
JOCArticle
SCHEME 1. Hydrogen Bonding in Phenols with ortho-Dihy-
droxy or ortho-Hydroxymethoxy Substitution Pattern Affects
Ionization Properties and Facilitates Electron Transfer toward
Triplet-Excited Flavinsa
FIGURE 9. EPR spectra (top traces) and corresponding simula-
tions (bottom traces) obtained during irradiation of phenolic model
compounds in the presence of flavin mononucleotide: (A) hydro-
quinone, (B) catechol, (C) 3,4-dihydroxybenzoic acid, (D) caffeic
acid.
aStabilization of phenoxyl radicals (resulting from proton loss from
incipient radical cations) to corresponding radical anions is reserved to
phenols with ortho-dihydroxy substitution.
quercetin glycoside (at C3) with the same reduction potential
as luteolin,49 reacted slower, possibly due to reduced diffu-
sion and steric effects from the bulky sugar moiety. Despite
the lack of a C2-C3 unsaturation and C4-oxo moiety,
reactivity of catechin approached that of quercetin and
luteolin. The importance of the catechol function in the
quenching mechanism, as thus demonstrated, is commonly
attributed to stabilization of ensuing radicals by intramole-
cular hydrogen bonding. However, such a stabilization path-
way fails to explain the similar reactivity observed for the
model compounds catechol and guaiacol, and moreover,
catechol-derived radicals are readily deprotonated to the
corresponding radical anion (pKr ∼ 4-5).45,52 Accordingly,
interference of intramolecular hydrogen bonding in resulting
radical intermediates is refuted, but it is more likely involved
in increasing electron density on the phenolic oxygen
(Scheme 1). The resulting lower ionization potential of
catechol and derivatives concurs with a more efficient elec-
the B-ring phenol, which, as a consequence, promoted
participation of the A-ring in electron transfer. Previous
observations for galangin, in which free hydroxyls on the
B-ring are absent, led to the conclusion that multiple phe-
nolic functions are involved in kinetically controlled oxida-
tion reactions,44,46 followed by intramolecular hydrogen
transfer to yield the thermodynamically favored radical
(mostly on the B-ring). However, the mechanism remained
open to question since the decay of A-ring radicals could not
be correlated to B-ring radical growth.55
Mechanistic Interpretation of Photooxidation. The nearly
diffusion-controlled rate constants, in combination with the
lack of kinetic isotopic effect, were strong indications of an
electron transfer toward 3FMN*. As a comparison, interac-
tion of phenol derivatives with peroxyl radicals, which are
3
3
weaker oxidants (E ∼ 1.0 V56 vs E ∼ 1.7 V for FMN*),
tron transfer toward FMN* (vide infra) and supports the
occurred by hydrogen atom abstraction (k ∼ 105-106 L
mol-1 s-1) and resulted in a kH/kD g 4.57,58 In line with the
proposed electron transfer in flavin-mediated photooxida-
tions, the rate constant for anisole was expected to be similar
to that of phenol. However, O-methylation inhibited oxida-
tion, similar to that seen for the interaction of 3FMN* with
tyrosine, TyrOH (k ∼ 109 L mol-1 s-1), and TyrOMe (no
quenching).59 The lack of labile hydrogens in the methylated
substrates is obvious, but most likely inhibition followed
from changes in ionization properties. The significant shift
in peak potential for anisole, as indicated by cyclic voltam-
metry experiments in aprotic solvent (Table 2), supports
this view. Most likely, incipient phenolic radical cations
barely tolerate positive charge and instantly deproto-
nate after electron transfer, hence the absence of such
superior reducing activity of these compounds. A compar-
able, albeit intermolecular, interaction (i.e., with solvent
molecules) was suggested to accelerate reduction of a 2,2-
diphenyl-1-picrylhydrazyl radical by phenolic compounds in
alcohol compared to aprotic solvents.53 The advantage of
ortho-substitution was demonstrated after comparison with
oxidation of meta-substituted 5-methoxyresorcinol, which
occurred at rates similar to those of unsubstituted phenol.
The presence of an exocyclic oxo moiety, as in 2,4-dihydroxy-
propiophenone, impaired oxidation, as the electron-with-
drawing effect resulted in higher reduction potentials (E7 =
0.89 vs 0.84 V for 5-methoxyresorcinol).44,46 It was therefore
concluded that reactivity of the A-ring in most flavo-
noids was low, and interaction preferably occurred via the
B-ring.45,54 Nevertheless, comparable rates of 3FMN*
quenching were observed for chrysin (lacks B-ring
hydroxyls) and apigenin, which assumed the feasibility of
mixed mechanisms. Possibly, conjugation to the electron-
withdrawing C4-oxo moiety in apigenin lowered reactivity of
ꢀ
(55) Cren-Olive, C.; Hapiot, P.; Pinson, J.; Rolando, C. J. Am. Chem. Soc.
2002, 124, 14027–14038.
(56) Buettner, G. R. Arch. Biochem. Biophys. 1993, 300, 535–543.
(57) Burton, G. W.; Doba, T.; Gabe, E.; Hughes, L.; Lee, F. L.; Prasad,
L.; Ingold, K. U. J. Am. Chem. Soc. 1985, 107, 7053–7065.
(58) Burton, G. W.; Le Page, Y.; Gabe, E. J.; Ingold, K. U. J. Am. Chem.
Soc. 1980, 102, 7791–7792.
(52) Steenken, S.; Neta, P. J. Phys. Chem. 1979, 83, 1134–1137.
(53) Litwinienko, G.; Ingold, K. U. J. Org. Chem. 2003, 68, 3433–3438.
(54) Pannala, A. S.; Chan, T. S.; O’Brien, P. J.; Rice-Evans, C. A.
Biochem. Biophys. Res. Commun. 2001, 282, 1161–1168.
(59) Tsentalovich, Y. P.; Lopez, J. J.; Hore, P. J.; Sagdeev, R. Z. Spectro-
chim. Acta A 2002, 58, 2043–2050.
7290 J. Org. Chem. Vol. 74, No. 19, 2009

Huvaere et al.
JOCArticle
SCHEME 2. Fast Disappearance of Quercetin Radicals is Ascribed to Efficient Disproportionation (minor) or Consecutive
(Photo)oxidation Steps (major)a
aThe role of C3-OH is pivotal because luteolin produces stable, detectable radical anions.
stabilization mechanism in anisole is suspected to inhibit
(photo)oxidation.
transfer from quercetin to 3FMN*) with regeneration of
FMN. In addition, autoxidation, which is particularly effec-
tive due to the lower reduction potential of quercetin at high
pH,27 produces superoxide, as well.
Structural Implications for Radical Stability. Stationary
irradiation at neutral pH (mixed with organic solvent) was
effective in generating steady-state concentrations of poly-
phenolic radicals. Despite the thermodynamic control,
which not necessarily yields incipient species, radical struc-
tures concurred with oxidation mechanisms suggested from
kinetic data. Besides, signal intensity, as observed in EPR
spectroscopy, was used for estimating radical stability since
this largely determines antioxidative properties. Indeed,
unstable, reactive polyphenolic radicals are difficult to detect
and probably behave as chain-carrying species which pro-
pagate oxidation. On the other hand, high steady-state
concentrations, as found for radicals from catechol-contain-
ing compounds, suggest excellent chain-breaking activity.
Quercetin was a remarkable exception since, despite its
known powerful chain-breaking properties, radicals readily
disappeared from the reaction mixture. Still, quercetin ana-
logues like luteolin (no C3-OH) and rutin (C3-OH pro-
tected as glycoside) furnished relatively long-lived species
(for example, x in Scheme 2), hence disappearance of quer-
cetin radicals was ascribed to the presence of C3-OH. The
latter probably interfered in a disproportionation step60 or,
more likely, in a consecutive electron transfer which led to
From the above-mentioned observations, the chemistry
behind the favorable antioxidative properties of compounds
with ortho- or para-dihydroxy substitution was assigned to fast
deprotonation of incipient phenoxyl radicals to form stable
radical anions at neutral or physiological pH. Model com-
pounds like catechol and hydroquinone nicely illustrate this,
particularly when compared to their mono-O-methylated
derivatives (Scheme 1). Indeed, neutral phenoxyl radicals from
guaiacol (v) or p-hydroxyanisole were too unstable to detect
and are probably incapable of breaking the radical chain.62
It is worth noting that the foregoing compromises in vivo
antioxidative activity of polyphenols, as methylation, accom-
modated by phase II enzymes like catechol-O-methyl transfer-
ase, is an important biotransformation during metaboli-
zation.63 The importance of the dihydroxy pattern for stabiliz-
ing radical intermediates was seemingly contradictory to the
intense signals observed during phenol irradiation. However,
the resemblance to the spectrum of the p-benzosemiquinone
radical anion (xiv) was striking and assumed preceding hydro-
quinone formation (Scheme 3), similar to that found for
photolysis of phenoxide anions in alkaline solution.29 The
latter involved nucleophilic addition of a hydroxide anion
to phenol, but such a mechanism was unlikely under the
current conditions (mixed buffer at pH 7). Therefore, water
addition to an intermediate phenoxonium (xii) was pro-
posed, whereby the cation results from disproportionation or
formation of EPR-silent reaction products (Scheme 2). At
•-
high pH, the single line corresponds to superoxide (O2
)
formation and not to a quercetin radical as previously
thought.61 A feasible mechanism for its formation involves
reduction of residual oxygen by FMN•- (formed by electron
ꢀ
(62) Lemanska, K.; van der Woude, H.; Szymusiak, H.; Boersma, M. G.;
(60) Dangles, O.; Fargeix, G.; Dufour, C. J. Chem. Soc., Perkin Trans. 2
1999, 1387–1395.
(61) Miura, T.; Muraoka, S.; Fujimoto, Y. Food Chem. Toxicol. 2003, 41,
ꢀ
ꢀ
Gliszczynska-Swigzo, A.; Rietjens, I. M. C. M.; Tyrakowska, B. Free Radical
Res. 2004, 38, 639–647.
(63) Zhu, B. T.; Ezell, E. L.; Liehr, J. G. J. Biol. Chem. 1994, 269, 292–299.
759–765.
J. Org. Chem. Vol. 74, No. 19, 2009 7291

Huvaere et al.
JOCArticle
SCHEME 3. Photooxidative Conversion of Phenol to Hydro-
antioxidative mechanism, particularly since control of con-
ditions was superior to alkaline autoxidation reactions.
Using this approach, it was concluded that radical anion
formation from incipient neutral radicals was decisive for
chain-breaking antioxidant activity. An ortho- or para-dihy-
droxy-substituted benzene as substructure complied and
allowed formation of resonance-stabilized radical anions.
Interestingly, replacing a hydroxyl by a carboxylate function
maintained stabilizing activity, as concluded from the high
steady-state concentration of the resulting radical anion.
This provides important new insights in the chemistry behind
phenolic antioxidants, but correlation with physiological
benefits, such as inhibition of pathological conditions, re-
mains difficult to predict.1 Particularly, the limited uptake,
due to fast metabolization and excretion, only results in
transient polyphenol concentrations in human plasma and
compromises in vivo activity.64 Their role in indirect health
effects, as through preserving sensitive foods from forming
toxic oxidation products, is less controversial and supports a
functional role for polyphenols in future food systems.
quinone, Followed by Stable Radical Anion Formationa
Experimental Section
Substrates (polyphenols and model compounds) of high
purity (>98%) were used as such. Acetonitrile was of spectro-
photometric grade, while deuterium oxide was 99.96% pure.
Aqueous solutions were prepared using a Milli-Q purification
aUnlike for hydroquinone, stable radical anion formation from 4-
hydroxybenzoic acid is due to the ionized carboxylate at neutral pH.
device (Millipore, Bedford, MA) (R = 18 MΩ cm). Buffers at
3
0.10 M were freshly prepared from analytical grade chemicals,
that is, acetic acid and sodium acetate (pH 4.2), citric acid and
sodium citrate (pH 5.6), potassium dihydrogenphosphate and
sodium hydroxide (pH 7.0), boric acid and sodium hydro-
xide (pH 8.5), and sodium bicarbonate and sodium hydroxide
(pH 10.0).
a second photooxidation step. In this respect, the similarity
between luteolin and apigenin radicals was suspected to arise
from an analogous hydroxylation reaction. On the contrary,
radicals from 4-hydroxybenzoic acid did not result from
intermediate hydroxylation reactions (Scheme 3). The high
steady-state concentrations were ascribed to formation of
stable radical anions (xv T xvi) instead of reactive neutral
radicals (as with methyl 4-hydroxybenzoate). The conclusion
that stabilization is maintained by substituting another ioni-
zable group for one of the phenolic functions in the ortho- or
para-dihydroxy pattern possibly contributes to the design of
advanced antioxidants.
Laser Flash Photolysis Spectroscopy. A dye laser (Spectron
Laser Systems, Rugby, U.K.) containing a coumarin 440
(Exciton, Dayton, OH) solution in methanol was pumped by
the third harmonic (355 nm) of a pulsed Q-switch Nd:YAG
laser. The resulting 440 nm laser flash (∼2.6 mJ per pulse) was
guided into the cell cavity, which housed a quartz cuvette
containing a nitrogen-purged solution of flavin mononucleotide
(125 μM) and selected compounds (concentrations varying from
125 μM up to 5 mM) in a buffer-acetonitrile mixture (1:1, v/v).
Specific reaction conditions were applied for the experiments
involving kinetic isotope effects as buffer molecules were
omitted from the reaction mixture, while deuterium oxide was
substituted for water as solvent for photooxidations of deuter-
ated compounds.
The analytical light beam, generated by a pulsed xenon
lamp (Applied Photophysics, Leatherhead, U.K.), was sent
through an optical filter (cutoff wavelength = ca. 610 nm)
before reaching the sample. The analyzing light beam then
passed a monochromator, and transient absorption at 720 nm
was eventually measured by a R928 photomultiplier tube
(Hamamatsu Photonics, Hamamatsu City, Japan). Formation
and decay traces of radicals, derived from quercetin and rutin,
were monitored at 485 nm (after installation of filters with
appropriate cutoff wavelength). Pseudo-first-order rate con-
stants were determined (at 293 K in a mixture of buffer pH 7.0
and acetonitrile (1:1, v/v)) for varying concentrations of poly-
phenols and model compounds and subsequently transferred
into second-order rate constants. Temperature dependence (for
calculation of activation parameters) was investigated between
293 and 313 K. Absorption spectra of excited flavin transients in
Conclusions
Studies on photochemical properties of flavins are numer-
ous and cover a broad scientific area. Flavin photochemistry
was previously found to be essential to blue-light-mediated
signaling pathways in the plant kingdom, thus triggering
electron transfer and radical cation formation in phototro-
pins and cryptochromes. Furthermore, endogenous sensiti-
zers in cellular matrices were activated upon exposure and
resulted in cytotoxicity or damage to biomolecules. Similar
processes affect food products and dietary supplements rich
in vitamin B2, leading to nutritional loss and deprived
quality. The intermediate triplet-excited flavin plays a pivo-
tal role in these phenomena due to its sensitizing properties
(leading to type II photooxidation) and its high redox
potential (accounting for type I oxidation). However, effi-
cient quenching of this transient oxidant by natural antiox-
idants such as polyphenols allows inhibition of adverse
effects associated with light exposure. Beside this, irradiation
was found to be an efficient mechanistic tool for characteriz-
ing oxidized intermediates and to determine their role in the
(64) Lotito, S. B.; Frei, B. Free Radical Biol. Med. 2006, 41, 1727–1746.
7292 J. Org. Chem. Vol. 74, No. 19, 2009

Huvaere et al.
JOCArticle
the wavelength region of 400-760 nm were analyzed by an
optical multichannel analyzer (OMA II, EG&G, Princeton
Instruments, Princeton, NJ).
(Bioanalytical Systems (BAS), West Lafayette, IN) using a
glassy carbon electrode (BAS MF-2012) as working electrode
and a platinum wire as auxiliary electrode (BAS MW-1032).
Substrates (2 mM) were dissolved in a solution of 0.10 M
Bu4NBF4 in acetonitrile, which was purified over aluminum
oxide prior to use. Potentials were referenced to a nonaqueous
pseudoreference electrode (BAS MF-2026), which was checked
daily against the standard potential of the ferrocene/ferroce-
nium couple, Fc/Fcþ. Potentials, measured at a fixed sweep rate
of 100 mV s-1, are reported against the standard hydrogen
electrode (SHE) by using E° = þ650 mV vs SHE for the Fc/Fcþ
couple in acetonitrile.66 The working electrode was polished and
rinsed before each measurement, while solutions were thorough-
ly purged with nitrogen prior to analysis.
Electron Paramagnetic Resonance Spectroscopy. Solutions
were prepared by dissolving phenolic compounds (10 mM)
and flavin mononucleotide (250 μM) in a 1:1 mixture (v/v) of
acetonitrile and aqueous buffer at pH 7.0 or at pH 10.0.
Extensive degassing by purging with nitrogen was carried out
before transferring the reaction mixture into a flat quartz cell.
Irradiation was carried out inside the EPR resonator cavity
using a focused light beam (wavelength = 440 nm, power = ∼2
mW) generated by a Photonics Polychrome II unit (TILL
€
Photonics, Grafelfing, Germany).
An ECS 106 spectrometer (Bruker, Karlsruhe, Germany) was
used for EPR spectroscopy. The following settings were applied:
center field = 3475 G (gauss); sweep width = 40 G; microwave
power = 10 mW; modulation frequency = 100 kHz; modula-
tion amplitude was optimized according to the intensity of the
signal and ranged from 0.25 to 1.0 G; conversion time = 20.48
ms; time constant = 10.24 ms; sweep time = 21 s. Spectra of
radicals were the result of a four-scan accumulation. Subsequent
simulation of EPR spectra, in order to determine hyperfine
coupling constants, was performed by the PEST WinSIM
program.65 Radical intensities were compared to the intensity
of a 2 μM 2,2,6,6-tetramethylpiperidine 1-oxyl (TEMPO) radi-
cal solution after double integration of the respective signals
(only intensities recorded at similar modulation amplitudes were
compared). Rise and decay of the rutinyl radical was measured
at static field (3476 G) with modulation amplitude of 1.0 G and
sweep time of 168 s.
Quantum Chemical Calculations. Geometries of radicals,
0
resulting from hydrogen abstraction at C4 -OH, were opti-
mized for minimal energy with the semiempirical AM1 method
in the MOPAC interface in the ChemOffice package
(Cambridgesoft, Cambridge, MA). Thus, estimates of torsion
angles between C2-H and the pz-orbital, perpendicular to the
B-ring of the respective flavonoid radical, were made.
Acknowledgment. Financial support by Arla Foods, the
Danish Dairy Research Foundation, and the Directorate for
Food, Fisheries, and Agricultural Business was gratefully
accepted.
Supporting Information Available: Atom coordinates
and geometries of taxifolin and eriodictyol radicals. This mate-
rial is available free of charge via the Internet at http://
pubs.acs.org.
Cyclic Voltammetry. Cyclic voltammetry experiments were
carried out on
a BAS CV-50W voltammetric analyzer
(66) Daasbjerg, K.; Pedersen, S. U.; Lund, H. General Aspects of the
Chemistry of Radicals; John Wiley & Sons: New York, 1999; pp 385-427.
(65) Duling, D. R. J. Magn. Reson., Ser. B 1994, 104, 105–110.
J. Org. Chem. Vol. 74, No. 19, 2009 7293