Content uploaded by Amy Rand
Author content
All content in this area was uploaded by Amy Rand on Nov 30, 2017
Content may be subject to copyright.
Journal of Photochemistry and Photobiology A: Chemistry 208 (2009) 79–90
Contents lists available at ScienceDirect
Journal of Photochemistry and Photobiology A:
Chemistry
journal homepage: www.elsevier.com/locate/jphotochem
Photo-oxidations initiated by UV radiation of urocanic acid and its methyl ester
in solution, micelles, and lipid bilayers: TYPE I (free radical) or
TYPE II (singlet oxygen) mechanisms depend on the medium
Amelia A. Rand, L. Ross C. Barclay ∗
Department of Chemistry, Mount Allison University, Sackville, New Brunswick, Canada E4L 1G8
article info
Article history:
Received 14 July 2009
Received in revised form 19 August 2009
Accepted 27 August 2009
Available online 2 September 2009
Keywords:
Free radical peroxidation
Singlet oxygen
Micelles
Lipid bilayers
Mechanism
abstract
Photo-oxidative reactions on methyl linoleate (ML) initiated by UV radiation of urocanic acid (UCA) or
methyl urocanate (UCAME) in a non-polar solvent, aqueous sodium dodecyl sulfate (SDS) micelles or ML
in lipid bilayers of dimyristoylphosphatidylcholine (DMPC) or of dilinoleoylphosphatidylcholine (DLPC)
are investigated to determine medium effects on reaction mechanisms. Experiments include: kinetics of
oxygen uptake, effects of radical trapping antioxidants or singlet oxygen quenchers and product analysis
of cis,trans to trans,trans (c,t//t,t) isomer ratios of oxidized linoleate to distinguish free radical, TYPE I
and singlet oxygen TYPE II reactions. Irradiation of the system toluene/ML/UCAME reaction occurred by
TYPE I according to kinetics studies and inhibiting effects of 2,6-di-t-butyl-4-methoxyphenol (DBHA)
and the four typical c,t//t,t isomer ratios formed. Similarly irradiation of the system SDS/ML/UCAME (or
UCA) resulted in the TYPE I reaction. The latter system was used to evaluate the antioxidant activities,
kinh, showing the relative activities as: Trolox ≥2,2,5,7,8-pentamethylhydroxychroman (PMHC) > DBHA.
The low kinh values of these antioxidants compared to those in a non-polar solvent are interpreted by
quantitative methods of “solvent effects”. Photo-oxidation of the systems DMPC/ML/UCA or DLPC/UCA
occurred by singlet oxygen (TYPE II) in contrast to those in solution or micelles according to effects of a
singlet oxygen quencher, sodium azide, and lack of inhibition by antioxidants. This contrast is explained
by the structure of lipid bilayers where the excited 3n,* of UCA produces singlet oxygen in the aqueous
phase which diffuses into the bilayer phase to cause TYPE II reactions.
© 2009 Elsevier B.V. All rights reserved.
1. Introduction
Urocanic acid (UCA) occurs naturally, for example in human
skin, as a natural metabolite of the amino acid, histidine. UCA
receives a lot of attention because of evidence that its UV excitation
in air affects the immune system possibly through the formation
of reactive oxygen species (ROS) that are implicated in damaging
skin, resulting in skin cancer [1–7]. Concurrent to its biological
significance, interest grew on the nature of the excited state(s)
and photochemical mechanisms involving UCA [8–11]; such as the
wavelength dependence of the initial trans to cis isomerization
[4]. Reports on the formation of singlet oxygen from electronically
excited UCA are of particular interest because of the high reactiv-
ity of singlet oxygen with sensitive biological systems. Evidence
for singlet oxygen formation included product studies [1,12], and
observation of luminescence at 1270 nm [5,13]. Theoretical meth-
ods have also been applied to the photosensitization mechanisms
∗Corresponding author. Tel.: +1 506 364 2369; fax: +1 506 364 2313.
E-mail address: rbarclay@mta.ca (L.R.C. Barclay).
and photoisomerization of UCA [14–16]. While the sensitized pro-
duction of singlet oxygen and subsequent TYPE II reactions appear
to be a possible or even usual reaction pathway on substrates from
UV-excited UCA, these reactions may not prevail in all cases. The
cholesterol hydroperoxide assay test for singlet oxygen formed on
photoexcitation of UCA in air yielded the 7␣-ChOOH isomer as well
as 5␣-ChOOH, the former indicative of a free radical, TYPE I pro-
cess [9]. Also the inhibition by Trolox of the photo-peroxidation of
methyl linoleate (ML) in sodium dodecyl sulfate (SDS) micelles is
indicative of a TYPE I process [17]. The possible reactions become
complicated due to direct oxidation reactions of UCA itself. Direct
oxidation of UCA by the riboflavin triplet rather than singlet oxygen
accounted for the inactivation of glucose 6-phosphate dehydroge-
nase [18] and UCA isomers were reported to be “good” hydroxyl
radical scavengers [19,20] and to trap peroxyl radicals, although
much more slowly than does Trolox [21].
On considering some simple chemical systems to clarify the
possible photo-oxidation reactions due to photo-excited UCA, we
were reminded by the late C.S. Foote that the “significant compe-
tition which determines whether TYPE I or TYPE II reaction occurs
is thus between substrate and oxygen for triplet sensitizer” [22].In
1010-6030/$ – see front matter © 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.jphotochem.2009.08.008
80 A.A. Rand, L.R.C. Barclay / Journal of Photochemistry and Photobiology A: Chemistry 208 (2009) 79–90
addition to this “competition”, it is possible that the nature of
the system employed may dramatically affect the pathway of the
main reactions. For example, electronically excited benzophenone,
which is well known to be an excellent H-atom abstractor, switches
to give energy transfer to oxygen and TYPE II products when the
donor substrate and benzophenone are “phase-separated” by a nar-
row air gap [23]. In biological systems, analogous separations, such
as between aqueous and lipid phases, could have a profound effect
on reactions initiated by photo-excited states. Consequently the
main objective of this investigation was to determine the effect, if
any, of different media on the mechanisms of photo-oxidation ini-
tiated by UCA and its methyl ester. Accordingly we now report on
experiments employing UCA or its methyl ester, UCAME, in several
different systems including homogeneous solution and systems
often used to mimic the natural molecular environment such as
heterogeneous aqueous-lipid media of micelles and lipid bilayers.
It was first necessary to determine if UCAME exhibited hydrogen
atom transfer (HAT) activity in its ground state since it contains
a free N–H group and might act as a HAT donor towards peroxyl
radicals similar to the behavior of some pyrroles [24]. An unsat-
urated lipid, the linoleate chain, was selected as a substrate for a
series of photo-initiated experiments since this lipid is known to
give specific kinetics of peroxidation [25–27] and known products
[23,28–30] typical of a free radical (TYPE I) or of a singlet oxy-
gen reaction (TYPE II) depending on which pathway predominates.
The photo-initiated reactions were carried out in homogeneous
solution, in micelles of SDS, in lipid bilayers of dimyristoylphos-
phatidylcholine (DMPC) containing methyl linoleate as substrate
and in bilayers of dilinoleoylphosphadtidylcholine (DLPC).
The methods used to investigate the pathways of the peroxi-
dations included kinetic studies of oxygen uptake along with the
effects of antioxidants or singlet oxygen quenchers, as appropriate,
and product analyses of linoleate hydroperoxides to test proposed
reaction pathways.
2. Materials and methods
2.1. General
Solvents used, common chemicals and antioxidants were of
highest quality from Aldrich. Antioxidants were re-crystallized
from methanol before use and stored at −20 ◦C. Sodium dodecyl
sulfate (SDS) was electrophoresis purity obtained from BIO-RAD.
The phospholipids, dimyristoyl- and dilinolylphosphatidylcholine
(DMPC and DLPC) were obtained form Avanti Polar Lipids in sealed
vials on “dry ice” and stored below −20 ◦C. Methyl linoleate (ML)
was obtained from NuCheck Prep in sealed vials under nitrogen. It
was determined to be hydroperoxide free just before use by TLC
analysis on silica gel developed with heptane/ethyl acetate (8/2,
v/v). For this purpose parallel TLC analyses were run on the new ML
compared to a sample partly oxidized. The hydroperoxides were
detected with N,N-dimethyl-1,4-phenylenediamine dihydrochlo-
ride spray (in methanol/water/acetic acid, 100/25/1, v/v/v) and
compared to the observed TLC position of the peroxide-free ML
detected by a spray of 1.5% ceric ammonium nitrate in 10% sul-
furic acid. This latter sprayed TLC detected both ML and oxidized
ML. NMR spectra were determined on a JOEL 270 MHz Spectrome-
ter, mass spectra on a HP 5988A instrument using a GLC interface
and 12m capillary column and UV/vis spectra on a Varian Cary
Bio 100 Spectrometer. HPLC analyses were carried out on a Var-
ian 9050/9012 system with an auto-sampler. An adsorption phase
column, 25 cm ×5 mm, generally 10 m silica, was used with a flow
rate of 1.5–2.0 ml/min unless other wise indicated. The oxidation
products from ML were identified by reduction with triphenylphos-
phine followed by HPLC analysis of their corresponding hydroxyl
compounds using the solvent mixture hexane/2-propanol//acetone
(992/4/4, v/v/v) and confirmed by comparison of their HPLC with a
known sample from an independent laboratory as described earlier
[26].
Phosphate buffer was prepared from de-ionized distilled water
containing 0.01 phosphate buffer saline from SIGMA passed
through a column of Chelex 100 resin (BIO-RAD) that was
conditioned with the buffer to pH 7.4–7.6. Diethylenetriaminopen-
taacetic acid, 1.4 ×10−4M, was then added to complex traces of
heavy metal ions. This buffer will be referred to as PBS.
2.2. Trans-urocanic (UCA) and methyl urocanate (UCAME)
Trans-UCA (Aldrich), re-crystallized from methanol/water
(30/10, v/v) gave m.p. 225 ◦C (rapid dec., lit. m.p. 223–225 ◦C dec.
[3]). Conversion to UCAME was carried out by acid catalysis in
methanol by the published procedure [12]. The crude product was
re-crystallized from ethyl acetate and further purified by sublima-
tion in vacuo to yield a colorless solid, m.p. 100–101 ◦C (lit. m.p.
100–101 ◦C[31]. The mass spectrum showed the parent ion at M+
152 (rel. inten. 45) and major fragment ions at m/e121 (rel. inten.
100) and 93 (rel. inten. 71). The 1H NMR spectrum was consistent
with that reported [12].
2.3. Autoxidation and photo-initiated procedures
The oxygen uptake studies were carried out at 30 or 37 ◦Cat
760 Torr in air using a sensitive, calibrated, dual-channel apparatus
that is described elsewhere [32]. Photo-initiated reactions were ini-
tiated through Pyrex using a 200 W super pressure mercury lamp,
the relative intensity of which was monitored throughout each
run by fiber optics sampling of the light beam with a phototube
detector. A series of neutral density filters of known transmittance
mounted on a filter wheel were used to change light intensities.
2.4. Preparations of aqueous/lipid dispersions
Solutions of ML, PMHC and DBHA of known concentrations in
0.50 M SDS/PBS were prepared by vortex stirring under nitrogen to
give clear solutions just before use. Trolox was prepared and used
in PBS. Multilamellar vesicules (MLV) of DMPC or DLPC containing
known amounts of additives as required (ML, UCAME) were pre-
pared by co-evaporation to films followed by vortex stirring and 10
freeze–thaw cycles in liquid nitrogen as reported earlier [27]. Unil-
amellar vesicles (ULV) were prepared in Tham buffer (pH 7.0) by
extrusion of MLV through micron filters under nitrogen (five times
with 0.40 m, twice through two 0.10 m and finally four times
through 0.05 m filters). The 31 P NMR spectra of the vesicles were
determined on a JEOL 270 MHz instrument at 109.25 MHz by the
pulse sequence reported earlier [33]. The MLV vesicles exhibited
the typical very broad anisotropic 31P NMR signal (50 ppm) while
the ULV showed a simple narrow signal [33].Supporting Material
includes traces of the 31P spectra of the MLV and ULV vesicles and
1H NMR spectra of UCA and UCAME.
3. Results
3.1. Quantitative kinetic studies and product analysis to
distinguish free radical (TYPE I) and singlet oxygen (TYPE II)
pathways of photo-oxidations
The kinetics and mechanism of free radical autoxidation have
been discussed in several reviews (e.g. [34,35]). In general, these
most common reactions of organic substrates usually involve either
initiation by peroxyl radicals in thermal reactions or excited triplet
states in photo-initiated reactions and take place in a sequence of
A.A. Rand, L.R.C. Barclay / Journal of Photochemistry and Photobiology A: Chemistry 208 (2009) 79–90 81
initiation, propagation, and termination steps. For uninhibited reac-
tions, the general kinetic expression of oxygen uptake which applies
to these reactions is given by Eq. (1),
−d[O2]
dt=kp
(2kt)1/2×[RsH]×Ri1/2(1)
where kpand 2ktare the rate constants for radical chain propaga-
tion and termination respectively, RsHthe substrate, and Riis the
rate of free radical initiation. These reactions are generally inhibited
by antioxidants (typically substituted phenols) which terminate
the reaction of peroxyl radicals by reactions (2) and (3).
ROO•+ArOH kp
−→ ROOH +ArO•(2)
ROO•+ArO•→nonradicalproducts (3)
Antioxidants suppress the oxygen uptake for a length of time, ,
which is related to the Riby Eq. (4), where n, the stoichiometric fac-
tor, is the number of radicals trapped per molecule of antioxidant,
generally 2.0 for phenolic antioxidants [34].
Ri=n[ArOH]
(4)
If the stoichiometric factor is unknown, it can be determined by
measuring the Riwith a known antioxidant in a separate experi-
ment and the ndetermined for the unknown by the use of Eq. (4).
During inhibited oxidations, the oxygen uptake is given by Eq. (5).
−d[O2]
dtinh
=kp
kinh
×[RsH]×Ri
n[ArOH] (5)
Integrating Eq. (5) provides Eq. (6) for the incremental oxygen
uptake.
−[O2]t=kp
kinh
×[RsH]ln1−t
(6)
For reactions which obey these equations, a plot of −[O2]tver-
sus ln(1 −t/) will give a linear slope of kp[RsH]/kinh from which the
absolute rate constant for inhibition, kinh, is obtained provided the
propagation rate constant, kp, for the substrate is known or can be
determined. Alternately, where there is sufficient oxygen uptake at
the beginning of the induction period, the initial −d[O2]/dtinh can
be estimated and the kinh calculated using the differential equation
(5).
In photo-initiated oxidations the rate of oxygen consumption is
determined by the light intensity and, assuming there is no “dark
reaction”, this kinetic order of the rate should be proportional to
the half power of the light intensity since this controls the Riin Eq.
(1). This relationship has been found to apply in several systems,
including those relevant to the current investigation in micelles
[26,29] and lipid bilayers [27]. In addition, reactions propagated by
oxygen-centered radicals are expected to be inhibited by phenolic
antioxidants. Consequently the reaction kinetic order in light inten-
sity and the effect of antioxidants are both used to test the role of
the TYPE I mechanism (see Section 3.3).
Reactions initiated by singlet oxygen are not expected to exhibit
a one-half reaction kinetic order in light intensity. As postulated
earlier [23], the rate of such photo-oxidation reactions are expected
to be directly related to the steady state concentration of singlet
oxygen and therefore show a first order relationship for oxygen
uptake with light intensity. This simple kinetic relationship has
been found to hold for several systems and combined with the
effects of a singlet oxygen quencher, such as sodium azide, pro-
vided evidence for the role of singlet oxygen in photo-initiated lipid
peroxidation [17,30].
Analysis of oxidation products from the natural lipid linoleate
is a well-known method to distinguish the roles of TYPE I versus
TYPE II reaction pathways [23,28–30] bearing in mind that sec-
ondary photo-reactions on initially formed hydroperoxides [36] or
quenching of the photo-reactor by additives may complicate the
results [37]. The general Scheme 1 outlines the different reaction
pathways for TYPE I and TYPE II oxidations on the linoleate chain.
A more complete mechanism of TYPE I reactions is given in [38].
TYPE I reactions on linoleate are initiated by hydrogen atom trans-
fer (HAT) from the reactive methylene at position 11 followed by
rapid reaction with oxygen at the carbon-centered radical forming
peroxyl radicals and by chain-propagating HAT on the substrate and
subsequent rearrangements of peroxyl radicals leading to mixtures
of cis,trans and trans,trans 9- and 13-hydroperoxides in which the
ratios (c,t to t,t;kinetic to thermodynamic product ratios) depend on
the H-atom donating ability of the medium(e.g. substrate concen-
tration or added antioxidants). In contrast, TYPE II singlet oxygen
reactions produce hydroperoxides directly in kinetically controlled
reactions. Reaction by 1O2at the more reactive bis-allylic position
11 produces an excess of cis,trans conjugated hydroperoxide iso-
mers [39] (Scheme 1,1and 2) and, since the more reactive singlet
oxygen is not selective, attack also occurs at the allylic Hs at posi-
tions 8 and 14 leading to non-conjugated hydroperoxides (5and 6).
So overall, singlet oxygen leads to a more complex reaction mixture
of at least six hydroperoxides. In the absence of a powerful antiox-
idant, a negligible amount of the non-conjugated isomer, 7, forms
[38]. In both kinds of reactions, the products can be separated and
quantified by HPLC, usually after reduction to their corresponding
hydroxyl derivatives. Some examples of product analyses are given
in Section 3.3.
3.2. Urocanic acid, UCA, or its methyl ester, UCAME, as
antioxidants
As indicated in Section 1, it was necessary to re-examine the
possibility that UCA or UCAME exhibits antioxidant activity as
implied in several reports [19–21]. Since these compounds are
used in this investigation as potential hydrogen atom acceptors or
energy transfer agents in their excited states, any powerful hydro-
gen atom transfer (HAT) reactions from their ground states might
mask or complicate the study. To check for possible HAT activity
in UCAME we selected the classical thermal initiation by the azo-
initiator, AIBN, in the hydrocarbon cumene as substrate because
in this system of very low kp[40] even very weak (HAT) antiox-
idants (e.g. 2,6-di-tert-butyl-4-methoxy phenol DBHA) [34] give
well-defined inhibition periods [41]. An example of such an exper-
iment, shown in Fig. 1, indicates very clearly that UCAME does not
possess any detectable inhibition of the oxygen uptake whereas the
weak antioxidant, DBHA, stopped the oxygen uptake almost com-
pletely. However the literature examples employed UCA, not an
ester, and this may explain some of the differences observed (see
Section 4).
Fig. 1. Oxygen uptake profile for the oxidation of cumene, 5.36 M, in chloroben-
zene initiated by AIBN, 0.0198 M, at 30◦C. (A) Uninhibited reaction; (B) inhibited
by 2,6-di-tert-butyl-4-methoxyphenol (DBHA), 2.25 ×10−6M; (C) effect of methyl
urocanate (UCAME), 2.69 ×10−5M.
82 A.A. Rand, L.R.C. Barclay / Journal of Photochemistry and Photobiology A: Chemistry 208 (2009) 79–90
Scheme 1.
3.3. Photo-initiated peroxidation of methyl linoleate in solution
3.3.1. Evidence for TYPE I reactions
Kinetic studies of photo-irradiation using methyl linoleate (ML)
as substrate in non-protic solvents provided clear evidence that
UCAME readily initiates TYPE I photo-oxidation. A typical example
of these results is shown in Fig. 2A and B. Toluene was selected as
the solvent for solubility purposes and it was necessary to show
that this solvent was inert to UV irradiation in the presence of
UCAME and air as indicated in Fig. 2A. Then on the addition of ML
there is rapid oxygen uptake which was efficiently inhibited by
antioxidants, even by the comparatively weak inhibitor, DBHA, as
shown in Fig. 2B. Not surprisingly, singlet oxygen quenchers such
as -carotene did not reduce the oxidation, Fig. 2A, trace B (dotted
line) and a relatively high concentration of the quencher, 1,4-
diazabicyclo[2.2.2]octane, DABCO, 2.0 ×10−4M, similarly showed
no effect on the oxygen uptake.
The oxygen uptake rates were followed for seven different
light intensities. The resulting linear plot shown in Fig. 3 results
in a kinetic order in oxygen uptake versus light intensity of
approximately one-half. This reaction kinetic order is expected
of free radical oxidation and together with the above results
establishes a classical, TYPE I, free radical peroxidation path-
way.
Product studies were carried out to complete the results for
UCAME-sensitized peroxidation of ML in toluene solution. The
hydroperoxides were reduced with triphenyl phosphine imme-
diately after the reaction and the products analyzed as their
Fig. 2. (A) Oxygen uptake profile for the peroxidation of methyl linoleate (ML),
0.45 M, in toluene at 30◦C, photo-initiated initiated by UCAME, 2.30 mM. (A) UV
light on without ML, ML added at 20 min.; (B) effect of -carotene 4.07 ×10−5M.
DABCO, 2.0 ×10−4M, has no effect on the oxygen uptake (not shown). (B) Inhibited
by DBHA, 8.94 ×10−6M.
A.A. Rand, L.R.C. Barclay / Journal of Photochemistry and Photobiology A: Chemistry 208 (2009) 79–90 83
Fig. 3. Kinetic order plot for the effect of UV light intensity on the peroxidation
of ML, 0.45 M, photo-initiated by UCAME, 2.30mM in toluene at 30 ◦C. The kinetic
order = 0.47 with R2= 0.99.
corresponding hydroxyl derivatives. A comparable experiment was
carried out employing benzophenone as the photo-initiator, a well-
known H-atom abstractor via its triplet state [26], and comparative
chromatograms using benzophenone and UCAME are shown in
Fig. 4A and B. The results showing the four main products, the two
cis,trans and two trans,trans hydroxy-substituted isomers are typi-
cal of the TYPE I radical peroxidation of linoleate. The slightly higher
c,t to t,t isomer ratio found with ucame-sensitized photo-initiation
(0.39) compared to that with benzophenone (0.30) is expected
because the higher ML concentration (0.695 M versus 0.483 M) is
expected to increase this kinetic to thermodynamic product ratio
[26,42].
Since our objective was to evaluate the role of free radical perox-
idation versus singlet oxygen reactions on linoleate, it was desirable
to carry out some kinetic experiments and product studies involv-
ing singlet oxygen for comparison purposes. For this purpose we
used dye-sensitized oxidation in solution, a well-known method
to generate singlet oxygen [23,28]. A kinetic order experiment of
oxygen uptake versus light intensity for methylene blue-sensitized
oxidation of methyl linoleate in solution is shown in Supporting
Fig. 4. HPLC analyses of the oxidation products of ML, as the hydroxyl deriva-
tives; 2, 13-hydroxy-cis,trans,3, 13-hydroxy-trans,trans,1, 9-hydroxy-trans,cis
and 4, 9-hydroxy-trans,trans isomers in order of elution. (A) Reaction initi-
ated by UV irradiation of benzophenone 4.99 mM in heptane on ML 0.483M.
Cis,trans//trans,trans isomeric ratio = 0.30 ±0.03 at 37 ◦C. (4) Reaction initiated by
UV irradiation of UCAME, 10.8 mM in toluene on ML 0.695M. Cis,trans//trans,trans
isomeric ratio = 0.39 ±0.05.
Material (SM), Fig. S3, where the kinetic order of approximately
unity is typical of a singlet oxygen reaction [23].
Earlier product studies from methylene blue-sensitized oxida-
tion of ML gave a product profile where the two cis,trans isomers 1
and 2(Scheme 1) dominated the products formed resulting in a c,t
to t,t ratio of 14 [23]. A similar experiment was repeated herein by
photo-initiated oxidation of ML with methylene blue in chloroben-
zene (not shown) which gave a comparable result with a c,t to t,t
isomer product ratio of 16.
3.4. Photo-initiated peroxidation of ML in sodium dodecyl sulfate
(SDS) micelles by UCAME and by UCA: antioxidant activities in
micelles
Aqueous micelles such as SDS have provided a useful model for
experiments on thermal or photo-initiated peroxidations in het-
erogeneous aqueous/lipid systems [26,43,44]. Such systems can be
used to initiate oxidation from the aqueous phase or from the lipid
phase. The ester, UCAME, was sparingly soluble in organic solvents
but dissolved readily in 0.50 M SDS, suggesting that it partitions
into the micellar/lipid phase as does ML [26]. Therefore a photo-
oxidation of the combination SDS/ML/UCAME was carried out using
different light intensities and the result of such an experiment is
shown in SM, Fig. S4. The observed kinetic order of one-half for
oxygen uptake versus light intensity is similar to that found earlier
with photo-initiation of linoleic acid peroxidation by benzophe-
none in a TYPE I reaction [26,42]. These results from kinetic studies
with UCAME are thus very similar to those found from benzophe-
none photo-initiated peroxidation of linoleic acid in SDS micelles
[26]. For comparison purposes, it was of interest to examine the
effect of light intensity on the oxidation of ML on the combina-
tion SDS/ML/UCA prepared by adding a buffer solution of UCA to a
ML/SDS solution. The resulting kinetic order, approximately one-
half, is shown in SM, Fig. S5 and indicates that both excited UCAME
and UCA initiate TYPE I reactions.
Since the results from photo-initiation using UCA as well as
UCAME in SDS micelles support the TYPE I mechanism, we investi-
gated more fully the use of the water-soluble UCA as photo-initiator
to compare the antioxidant capacities of typical phenolic antiox-
idants. The use of this form of the compound provided more
flexibility in designing experiments since it can be added from
aqueous buffer in varying amounts to provide measurable oxy-
gen uptake during the UV radiation. By this method we obtained
results to compare the inhibiting effects of three common pheno-
lic antioxidants on the inhibition of lipid peroxidation: PMHC, a
powerful antioxidant of the hydroxyl chromanol (vitamin E) class
[45], DBHA, a hindered phenol and commercial antioxidant, both
added from known concentrations prepared in SDS micelles, and
Trolox, a water-soluble compound of the vitamin E class. Trolox
has the advantage of water-solubility and consequently has been
frequently used in clinical trials and to compare the efficacy of
other antioxidants, so that it is not surprising to find that SciFinder
Scholar currently gives at least 1000 references to “Trolox as antiox-
idant”. The structures of these antioxidants are shown in Fig. 5.
A typical experiment illustrating oxygen uptake during pho-
tolysis of the SDS/ML/UCA combination and inhibition by PMHC
is illustrated in Fig. 6 showing the uninhibited rate, curve A, the
inhibiting effect of PMHC, curve B, and an inset, C, a plot of −[O2]
versus ln(1 −t/). This linear plot shows that the inhibition profile
follows the classical kinetics of Eq. (6). The antioxidants DBHA and
Trolox also exhibited antioxidant activity during photo-initiation
by UCA in aqueous SDS and the effects of the three antioxidants on
the inhibited oxygen uptake are compared qualitatively in Fig. 7.
There is significant oxygen uptake during the induction periods
for all three antioxidants compared to typical results in non-
protic solvents (see Figs. 1 and 2) which qualitatively indicates low
84 A.A. Rand, L.R.C. Barclay / Journal of Photochemistry and Photobiology A: Chemistry 208 (2009) 79–90
Fig. 5. Structures of the antioxidants used: PMHC, Trolox, and DBHA.
Fig. 6. Oxygen uptake profiles for the peroxidation of ML, 0.60 M, in SDS, phosphate
buffer, pH 7.4, photo-initiated by UCA, 0.492 ×10−4molat30
◦C. (A) Uninhibited
reaction; (B) inhibited with PMHC, 61 M (calcd. for SDS phase); (C) inset of the
data from B plotted according to the linear equation (6).
antioxidant activities in the aqueous SDS medium. By repetition
of these experiments with the three antioxidants, their antioxi-
dant activities were determined by measurement of their absolute
rate constants for inhibition, kinh, and these are summarized in
Table 1. Their stoichiometric factors, n, were calculated relative to
n= 2 for PMHC [34] by determination of the Riby measurement
of the induction period using PMHC under the same conditions
and ncalculated using Eq. (4). Additional details of these experi-
ments are given in SM, Table S1. The relative antioxidant activities
are, Trolox ≥PMHC > DBHA. The stoichiometric factors, n, for PMHC
Fig. 7. A qualitative comparison of the effect of antioxidants on the peroxidation of
ML, 0.60 M, photo-initiated by UCA, 0.492×10−4mol in SDS, phosphate buffer, pH
7.4 at 30 ◦C. (A) Uninhibited reaction; (B) inhibited with DBHA, 21M; (C) inhib-
ited with PMHC, 20 M; (D) inhibited with Trolox, 21M (assuming that Trolox is
completely partitioned into the SDS micellar phase).
Table 1
Antioxidant activities, kinh, and stoichiometric factors, n, of phenolic antioxidants
during inhibited peroxidation of methyl linoleate (ML)aphotoinitated by urocanic
acid (UCA)bin 0.50 M sodium dodecyl sulfate (SDS) at 30◦C in saline phosphate
buffer, pH 7.4.
Antioxidantckinh (M−1s−1×10−3)dne
PMHC 50.4 ±5.0 2.0
DBHA 31.7 ±0.9 1.9e
Trolox 60.3 ±7.8 1.7 ±0.1
aThe concentration of ML was 0.60 M in the SDS phase where the micellar volume
was calculated to be 2.50 ×10−4l per 2.00ml of 0.50 M SDS [26].
bExperiments generally used 50 M UCA, except one run with Trolox used 25 M.
cConcentrations of antioxidants in the micellar phase ranged for PMHC,
25.2–136.4 M, DBHA, 20.6–61.7 M, Trolox, 20.9–54.5M.
dThe rate constants from at least three experiments were calculated using Eq.
(5) from the initial rate of oxygen uptake, −d[O2]/dtinh =kp/kinh ×[ML]Ri/n[ArOH],
using kp=36M
−1s−1[26] and the Rimeasured for each experiment using PMHC;
Ri=2×[PMHC]/. The Riand induction period, , for each experiment are given in
Supporting Material (SM). Error limits were determined at 95% confidence interval
for kinh.
eFrom two measurements.
and DBHA both are approximately 2, whereas the value for Trolox,
which is introduced from the aqueous phase, is about 15% less. It
distributes only partially into 0.10 M SDS (f= 0.19 [26]) and pre-
sumably partially into 0.50 M SDS as used herein, therefore some
of the Trolox is probably wasted due to competing radical reactions
in the aqueous phase (see Section 4). Overall the antioxidant activ-
ities of these typical phenols are dramatically different in aqueous
micellar media compared to known values in a non-protic solvent
[45]. Briefly, their kinh values are both leveled and marked reduced
in aqueous SDS. For example, in styrene the value for PMHC in
styrene is 75 times and that for DBHA 3.5 times of that shown here
in aqueous/SDS. The values in aqueous micelles are not only greatly
reduced but are similar (Table 1) between these three antioxidants
in contrast to the values in styrene [45]. Several effects including
solvation by water and the unique effects of a micellar environment
itself appear to be operating to cause these reductions in observed
antioxidant activities in aqueous micelles (see Section 4).
3.5. Peroxidation of lipid bilayers photo-initiated by UCA from the
aqueous phase
Micelles are known to be dynamic species which undergo
rapid breaking and reforming which allows for rapid exchange of
monomers [46]. In contrast, lipid bilayers form more stable aggre-
gates in water and phosphatidylcholine bilayers are often selected
as models of natural membranes since they mimic much of the
structural type and physical properties of natural membranes. Con-
sequently it was anticipated that such a boundary between the
aqueous and lipid phases might provide a system that would exhibit
competition between TYPE I and TYPE II reactions.
3.5.1. Photo-initiated peroxidation of ML in
dimyristoylphosphatidylcholine (DMPC) bilayers by UCA
The system selected for these experiments, was methyl linoleate
(ML), the oxidizable substrate, contained in multilamellar vesicles
(MLV) consisting of saturated lipid chains from DMPC. This is an
appropriate system to compare/contrast with the kinetic results
obtained with the combination of ML in SDS micelles. Also the
concentration of the oxidizable lipid, ML, can be controlled in this
mixture and some kinetic evidence is available on this system from
earlier thermal-initiated peroxidation studies [25].
The results from photo-initiation of the combination ML/DMPC
in bilayers by UCA contrasted with those for the combination of
photo-initiation by UCA in ML/SDS micelles or in homogeneous
solution by UCAME. The contrasts are shown in Fig. 8 for the bilayer
A.A. Rand, L.R.C. Barclay / Journal of Photochemistry and Photobiology A: Chemistry 208 (2009) 79–90 85
system compared, for example, to Fig.6or7in micelles. In micelles,
antioxidants were effective inhibitors but the water-soluble sin-
glet oxygen quencher, sodium azide, exhibited no effect on oxygen
uptake, whereas in ML/DMPC initiation with UCA was not affected
by Trolox, the better antioxidant in the SDS system, but sodium
azide proved to be effective at reducing oxidation. This effect was
also found to be dependent on the concentration of sodium azide
added as shown in the experiment summarized in Fig. 8, Inset. The
profile of reduced oxidation was non-linear in this system showing
a greater reduction during the first additions of quencher.
The ML/DMPC dispersions used were quite turbid so that filtered
light at the lower intensities did not provide reproducible rate mea-
surements, however kinetic results were obtained for four different
intensities. A plot of such data shown in SM, Fig. S6 indicates a first
order in rate of oxidation with light intensity. Consequently the
kinetic evidence overall supports a TYPE II mechanism. It is quite
clear that a change in the type of heterogeneous aqueous/lipid dis-
persion from micellar to bilayer AND initiation with water-soluble
UCA has resulted in a change of mechanism from TYPE I to TYPE II.
Since UCA is reported to undergo photoinduced decomposition
in oxygen as well as isomerization [1,11,15] and UCAME read-
ily adds generated singlet oxygen [12], it is somewhat surprising
that the kinetic studies give constant uptakes of oxygen for our
kinetic order and/or inhibition experiments where the irradiation
times were often prolonged to several hours. Therefore a study
was made of the UCA concentrations at the beginning and dur-
ing several of these experiments by scanning the UV/vis through
the range 200–400 nm. All samples were diluted by methanol by
the identical amount to keep absorbance in the range 0.200–0.500
where they showed the general absorption typical of UCA with max
277–278 nm. The reference side of the spectrometer contained a
reference sample without UCA prepared from the same concen-
tration of ML/SDS (or ML with DMPC) diluted with methanol to
the same extent. The initial analysis were made by removing sam-
ples starting before radiation (T-0) and at various stages during
the run. However it was soon found that there was no change in
the UCA absorbance during the run. The remaining samples were
taken from the final solution after the run and compared to the T-
0 sample. In the experiments using the ML/SDS system, 93–97% of
the UCA remained after the photo-initiation, while a typical sample
from the ML/DMPC bilayer system, 99% of the UCA remained after
the run. A control experiment carried out on UV/vis irradiation of
UCA in SDS but without ML showed that 93% of UCA remained after
30 min irradiation and after 4 h this dropped to 84% of the UCA at
T-0. The persistence of UCA through a run was fortuitous since it
Fig. 8. Oxygen uptake profiles for the peroxidation of ML, 1.20 ×10−4mol, in ini-
tiated by UCA, 5.80 ×10−5mol, in phosphate buffer, pH 7.4 at 37◦C. (A) UV light
on dimyristoylphosphatidylcholine (DMPC) multilamellar bilayers, 2.60 ×10−4mol,
photo with DMPC and UCA without ML; (B) in the presence of ML; the arrow
shows the addition of Trolox, 6.54 ×10−9mol; (C) the effect of sodium azide
4.55 ×10−6mol. The inset shows the effect of sodium azide concentration on the
oxygen uptake.
Fig. 9. Plot of the effect of sodium azide concentration on the photo-oxidation of
dilinoleoylphosphatidylcholine (DLPC) multilamellar bilayers, 7.45 ×10−5mol, ini-
tiated by UCA, 12.5 ×10−5mol, in phosphate buffer, pH 7.4, at 37◦C.
permitted the completion of the various kinetic experiments with-
out loss of the rate of oxygen uptake due to loss of the initiator.
This is similar to previous results found on using benzophenone or
azaaromatics as photo-initiators [26,29] where the initiators were
regenerated during the reaction (see Section 4).
3.5.2. Photo-peroxidation of dilinoleoylphosphatidylcholine
(DLPC) by UCA
Evidence given in the previous section indicated that photo-
excited UCA in the aqueous phase resulted in reaction by singlet
oxygen on ML encapsulated into DMPC liposomes as shown by the
effect of sodium azide quencher and the kinetic order in light inten-
sity (Fig. 8 and SM, S6). It was desirable to repeat experiments on
the model membrane prepared from DLPC where the lipid chains
are bonded in the bilayer structure. The DLPC bilayer is readily
oxidized and used quite frequently as a model for an unsaturated
membrane [34]. In one respect, the result using DLPC bilayers was
very similar to that with the DMPC/ML/UCA combination. That is
the oxygen uptake was reduced in much the same way with sodium
azide (Fig. 9) and the oxidation was not inhibited by Trolox. How-
ever the kinetic order in light intensity was not actually first order,
but 0.83 as shown in SM, Fig. S7. This unusual kinetic order may be
due to other reactions that complicate the kinetics of multilamellar
DLPC (see Section 4).
3.6. Photo-initiated peroxidation of
dilinoleoylphosphatidylcholine (DLPC) by UCAME within the
bilayer phase
Section 3.5 shows that in lipid bilayers, in contrast to micelles,
photo-initiation with water-soluble UCA resulted in lipid oxida-
tion by singlet oxygen; that is, by the TYPE II mechanism. This
important result could be a consequence of the initiation and
the reaction being carried out in the separate aqueous and lipid
phases of the aqueous bilayers. To test this idea, we carried out
an experiment to compare this result with one where the peroxi-
datiom of the linoleate chain is carried out with the initiator and
substrate in the same phase. This was done by incorporating the
lipid-soluble ester, UCAME, within the lipid phase of DLPC by co-
evaporating the two from methanol, a common procedure used in
such cases [27]. The results from a typical experiment of this type
is shown in Fig. 10. In this case, Trolox exhibits antioxidant activ-
ity as shown earlier for photo-initiation within the DLPC bilayer by
benzophenone [27]. The oxygen uptake during the Trolox induc-
tion period is appreciable and a plot of −[O2] versus ln(1 −t/)
(see inset) of this inhibition according to the linear equation (6)
appears to be somewhat curved compared, for example, to such a
plot from the combination SDS/ML/UCA (Fig. 6, inset C). The kinh
of 18.8 ×103M−1s−1for Trolox, calculated from the initial inhib-
ited oxygen uptake using Eq. (5) and the kpof 36.1 for DLPC [27],is
86 A.A. Rand, L.R.C. Barclay / Journal of Photochemistry and Photobiology A: Chemistry 208 (2009) 79–90
Fig. 10. Oxygen uptake profiles for the photo-oxidation of DLPC bilayers,
5.11 ×10−5mol, photo-initiated at 37 ◦C by UCAME, 3.42 ×10−5mol, by co-
evaporation form methanol; (A) uninhibited reaction; (B) inhibited by Trolox,
6.32 ×10−9mol; (C) inset of a the data from B plotted according to the linear equation
(6).
markedly lower than found in solution or even in micelles (Table 1)
and demonstrates the effect that this heterogeneous aqueous/lipid
system has on the inhibition reaction (see Section 4).
The oxidation rates in MLV-DLPC initiated by UCAME were
relatively slow in part due to the high turbidity of the MLV sys-
tem. However smaller unilamellar vesicles (ULV) are known to
give greater rates of oxidation [27]. The 31P NMR spectrum of the
MLV-DLPC showed a very broad signal, chemical shift anisotropy
of 40 p.p.m., typical of such MLV particles ([56] and references
therein). After several extrusions of this DLPC containing UCAME
by co-evaporation the 31P spectrum changed to a single line typ-
ical of smaller ULV-DLPC (see SM, Fig. S8A and B).This ULV-DMPC
gave much more rapid oxidation which proved useful in efficient
isolation of sufficient amounts of linoleate hydroperoxides for anal-
yses as corresponding hydroxyl derivatives. Trans methylation and
reduction by TPP of the isolated products [27] gave sufficient mixed
hydroxy methyl linoleates for HPLC analyses and the four isomers
formed were those typical of a TYPE I peroxidation (see SM, Fig. S9).
These mixed isomers showed a high c,t/t,t isomer ratio, approxi-
mately 1.54 (average of two determinations, 1.51, 1.57), compared
to our results in solution and even in SDS micelles. This ratio is
well known to depend on the concentration of the linoleate and
the temperature of the oxidation [42]. In addition the isomer ratio
appears to depend on the type of bilayer particle (see Section 4).
However the products formed during peroxidation are entirely con-
sistent with the kinetic data; both results indicate a classical TYPE
I reaction photo-initiated with UCAME inside the lipid phase.
4. Discussion
4.1. Thermal and photochemical results: UCA and UCAME in
homogeneous solution
On the question of antioxidant activity of UCA (or UCAME), the
observation that UCA reacts readily with hydroxyl radicals [19,20]
does not signify that it is a radical chain-breaking antioxidant. As
emphasized earlier, the vast majority of organic compounds react
very rapidly with this very reactive radical [47]. The first real test of
UCA as a trap of the main chain-carrying radicals in autoxidation,
namely peroxyl radicals, detected some low antioxidant activity of
UCA, only 400 times less efficient than Trolox [48]. We could not
detect HAT activity from UCAME in cumene solution even com-
pared to the weak antioxidant, DBHA (Fig. 1). It is quite possible
that the small antioxidant effect observed for UCA in a polar, aque-
ous medium [48] could be due to single electron transfer (SET) or
proton-coupled electron transfer (PCET) [24]. However we did not
detect any antioxidant activity by UCA in aqueous-phosphate buffer
solutions used in our experiments.
The ester, UCAME, dissolves in non-protic solvents and was con-
sidered to be an appropriate compound for evaluating the possible
HAT activity in its excited state. All experiments showed clearly
that photo-excited UCAME initiated a TYPE I free radical peroxi-
dation reaction on methyl linoleate (ML) under these conditions.
This process is shown by the inhibiting effect of an antioxidant
(Fig. 2A and B) and the one-half kinetic order in oxygen uptake
(Fig. 3) in contrast with the kinetic order of unity for a reaction initi-
ated by singlet oxygen (SM, Fig. S3). In addition the TYPE I pathway
with UCAME was confirmed by the typical product analysis pro-
file (Fig. 4B) where the cis,trans to trans,trans (c,t//t,t) isomer ratio
contrasted with known results from reaction with singlet oxygen
where the analysis is dominated by the cis,trans isomers 1and 2
[23].
4.2. Photo-peroxidation of ML by UCAME and by UCA in
aqueous-SDS micelles: factors controlling antioxidant activities in
aqueous micelles
The ester, UCAME, dissolved readily in 0.50 M SDS by vortex stir-
ring and presumably is incorporated into the micellar phase along
with ML [26]. Irradiation of these solutions with different light
intensities gave a kinetic order of oxygen uptake with light intensity
of one-half (SM, Fig. S4) indicative of the HAT-TYPE I mechanism in
the micellar phase. The comparable result in kinetic order of oxy-
gen uptake from photo-oxidation of the combination SDS/ML/UCA
by simply injecting UCA into the aqueous phase (SM, Fig. S5) indi-
cates a TYPE I pathway for both methods and this was supported
by inhibition studies (vide infra).
Photo-peroxidation of ML in SDS was readily initiated by
injection of UCA into the aqueous phase and these reactions
were inhibited by three typical antioxidants (Figs. 6 and 7). The
photochemical method has certain advantages over the more
conventional method employing thermal initiators; namely: (i)
corrections to the measured oxygen uptake required for thermal
azo-initiators due to nitrogen evolution are not needed, (ii) conve-
nient control of the oxygen uptake rates and the rate of radical
initiation chain, Ri, are possible by changing the light intensity
and (iii) any thermal reaction can be checked with the light off.
Consequently reliable results were obtained for the absolute rate
constants of antioxidant activities, kinh, and stoichiometric factors,
n, for the typical antioxidants PMHC, Trolox and DBHA as summa-
rized in Table 1.
Aqueous micelles have been used frequently as heterogeneous
aqueous-lipid phases to mimic natural biochemical systems for
measurements of antioxidant activities [26,43,44]. The measured
antioxidant activities are well known to be markedly reduced and
leveled in aqueous media compared to those in non-polar solvents
and this solvent effect has been extensively investigated ([49], and
references therein). In a homogeneous protic solvent, the solvent
effect can be attributed to hydrogen bonding by solvent at the phe-
nolic hydroxyl group preventing attack by peroxyl radicals in the
rate-determining step (Scheme 2)[49].
It has been recognized for some time that the reduction of phe-
nolic antioxidant activities in aqueous-lipid heterogeneous lipid
Scheme 2.
A.A. Rand, L.R.C. Barclay / Journal of Photochemistry and Photobiology A: Chemistry 208 (2009) 79–90 87
systems cannot be explained entirely by hydrogen bonding [50,51].
The reductions in antioxidant activities could be due to the nature
of the particular lipid system itself which would cause rate-limiting
diffusion [50] or result in physical inaccessibility between the
peroxyl radical and the antioxidant [51]. However the effect of H-
bonding can be evaluated separately since quantitative studies by
Ingold and co-workers has provided a relationship which permits
actual calculations of rate constants for H-atom abstraction by rad-
icals in protic solvents [52]. Their Eq. (7), permits calculation of the
HAT rate constant, ks, in any H-bonding accepting (HBA) solvent
from the rate constant in a non-H-bonding solvent, k0, and known
thermodynamic parameters [53]; namely, the H-bonding donat-
ing ability of the donor antioxidant, ˛H
2, and the H-bond accepting
ability of the solvent, ˇH
2.
log(ks/M−1s−1)=log(k0/M−1s−1)−8.3˛H
2ˇH
2(7)
It is of interest to apply this equation to the current kinetic data
and evaluate the relative effects of H-bonding by the water and
specific micellar effects on the HAT activities of the antioxidants.
For this purpose some relevant data is given in Table 2.
This shows: (i) required literature rate constants for PMHC,
Trolox and DBHA in a non-H-bonding solvent, k0in styrene; (ii) the
calculated and experimental results, kcalc
MeOH and kexp
MeOH, in methanol,
a common H-bonding solvent, to test the method used here and
(iii) kcalc
H2O,SDS and kexp
H2O,SDS for our current results on PMHC and
Trolox. The method is not applicable to highly hindered phenols
like DBHA since the ˛and ˇparameters are not determined. To
calculate kcalc
MeOH for PMHC in methanol we used the ˛H
2available
for ␣-tocopherol, 0.370 [52], since the chromanol structures are
the same and their antioxidant activities are very similar, within
16%, in homogeneous solution [45]. For calculations in aqueous
micelles, we used the ˇvalue, of 0.31, determined for water since
water is expected to be the main HBA to the phenolic hydroxyl in
aqueous micelles [51]. The calculated kcalc
MeOH is in reasonable agree-
ment with the experimental value. However the measured value
in aqueous SDS, 5.0 ×104M−1s−1, is markedly different by more
than eight times lower than the calculated value, 42 ×104M−1s−1.
This means that only a very small part of the reduction in antiox-
idant activity can be attributed to hydrogen bonding by water in
this system, much of the effect must be due to the effect of the
micellar system which can reduce the rate either by restricted dif-
fusion between the lipid peroxyl radicals and antioxidant and/or
compartmentalization of the lipid peroxyl radical relative to the
antioxidant. The ˛value for Trolox is not known, however it is
very probable that the large reduction in kinh for Trolox in aque-
ous micelles is also due mainly to the micellar environment rather
than hydrogen bonding. Compared to other vitamin E class of phe-
nolic antioxidants Trolox exhibits much lower antioxidant activity
in non-polar solvents (3.5 times lower than PMHC, Table 2) yet it at
least matches this activity in SDS micelles at pH 7.4. This switch
Table 2
Rate constants, kinh (×10−4M−1s−1) for PMHC, Trolox and DBHA in a non-polar sol-
vent, k0, compared to calculated and experimental (kcalc and kexp) values in methanol
and, in water/SDS.
Antioxidant
PMHC Trolox DBHA
k0
styrenea380 110 11
kcalc
MeOH 47b
kexp
MeOH 57b
kcalc
H2O,SDS 42
kexp
H2O,SDS 5.0 6.0
aRate constants are taken from Ref. [45].
bFrom Ref. [24].
is attributed to the presence of an electron-attracting carboxyl
group in the non-polar solvent (Fig. 5, Trolox A) which destabi-
lizes the intermediate phenoxyl radical of the rate-determining
step (reaction (2)). In contrast this group will be a carboxylic
anion (Fig. 5, Trolox B) at pH 7.4 which is electron supplying to
stabilize the intermediate. The Swain field effect parameter pro-
vides a measure of these opposite effects, for COOH, F= 0.552
and for COO−,F=−0.221) [54], where a negative value indicates
an electron-supplying effect while a positive value an electron-
attracting effect.
4.3. Photo-peroxidation of linoleate in lipid bilayers by UCA and
UCAME: TYPE II versus TYPE I reactions
As indicated in Section 3, the system consisting of ML
sequestered in bilayers of DMPC containing saturated lipid chains
was used for photo-initiated experiments to compare with the clas-
sical results for earlier thermal free radical peroxidation [25] and
in particular with results herein for photo-peroxidation of ML by
UCA in SDS micelles. The results from photo-peroxidation from
the combination DMPC/ML/UCA stand in sharp contrast with both
the free radical peroxidation in solution and the photo-initiated
reactions in the system SDS/ML/UCA in micelles. The reactions
clearly proceeded in the DMPC bilayers by attack with singlet
oxygen since they were not inhibited by the antioxidant, Trolox,
which is generally very effective to inhibit peroxyl radical attack
in this system [25], and the oxidation was effectively reduced by
sodium azide (Fig. 8). In addition a TYPE II process is supported by
a kinetic order of unity in oxygen uptake with light intensity (SM,
Fig. S6).
It is somewhat fortuitous that UCA is not consumed by the active
oxygen species (AOS) in these reactions in micelles or bilayers.
In this regard the results resemble the earlier result when ben-
zophenone was used as photo-initiator [26,29]. So from analogy we
propose that there is a pathway to regenerate the UCA during the
reaction. This is illustrated in Scheme 3. In this scheme the reactions
are initiated from the 3n,*state of UCA [11] from which reaction
can take two pathways: (i) energy transfer may form singlet oxy-
gen which diffuses into bilayers to cause singlet oxygen products
directly whereas (ii), in solution or micelles the prominent reaction
is hydrogen transfer from a active C–H bond of linoleate starting a
TYPE I free radical process and generating a reactive UCA-H•. The
latter is expected to react readily with either lipid peroxyls to form
observed LOOH products or oxygen (or both) to release HOO•rad-
icals into the aqueous phase. These could react with Trolox and
also account for the lower stoichiometric factor observed for Trolox
compared to that for PMHC (Table 1).
A very important question remains. Why do the different media,
like aqueous/lipid bilayers and aqueous/lipid micelles result in such
markedly different reaction pathways? To explain these different
results, we consider the essential differences between homoge-
neous solutions and micelles on the one hand compared to lipid
bilayers on the other. In solution it is obvious that the UCA (UCAME)
excited state will collide freely with ML the give a HAT reaction. In
micelles, which are also continuously providing rapid exchange at
least of small molecules like UCA (but some restricted diffusion
of hydrophobic molecules, e.g. ML), there is also sufficient molec-
ular contact between excited UCA and ML to generate mainly a
HAT process but with lower efficiency compared to homogeneous
solution. The difference in bilayers is attributed to the very signifi-
cant difference with lipid bilayers where there is a boundary/barrier
between the aqueous and lipid phases. This is shown schematically
in Scheme 4. When UCA is excited in the aqueous phase, molecu-
lar contact between excited UCA and ML/bilayer is prevented but
the small reactive oxygen species, singlet oxygen, like oxygen, can
diffuse sufficiently into the bilayer to cause a TYPE II reaction. This
88 A.A. Rand, L.R.C. Barclay / Journal of Photochemistry and Photobiology A: Chemistry 208 (2009) 79–90
Scheme 3.
is analogous to the earlier observation where light-irradiation of
bilayers containing methylene blue in the aqueous phase produced
singlet oxygen products by diffusion of singlet oxygen into the lipid
phase [30].
Scheme 4.
The simple model of Scheme 4 should also account for the recent
observation that UCA can distribute partially into bilayers, at least
into egg lecithin [55]. It is proposed that UCA, which is in the form
of the charged anion, is probably not incorporated deep into the
hydrophobic region of the bilayers but is likely bound near the polar
surface.
Assuming that the ML lipid is situated as shown in Scheme 4
with the polar ester group near the polar surface, the photo-excited
UCA will not reach the unsaturated region of ML to initiate HAT at
the reactive methylene site. To test this hypothesis, it was help-
ful to carry out the opposite experiment where the photo-initiator
is located within the hydrophobic region of the bilayer. Photo-
oxidation of this system was inhibited by Trolox (Fig. 10) indicative
of a HAT-type reaction.
Some product analysis results were in agreement with this oxy-
gen uptake-inhibition result. The significant result from product
analysis from the ULV-DLPC/UCAME photo-oxidation is formation
of the four c,t//t,t isomer combination typical of the TYPE I (HAT)
reaction. The high c,t//t,t ratio, 1.54, found is somewhat higher than
expected since at 37 ◦C this ratio from DLPC was in the range of
1.2–1.3 for radical peroxidations initiated by either a lipid-soluble
[42] or water-soluble thermal initiator [56]. The complete char-
acterization of the extruded DLPC is beyond the scope of this
investigation. However the higher c,t//t,t but it is probably due to
the smaller particle size of the extruded DLPC (presumably at least
partly ULV according to the 31P NMR spectrum). Several studies
have concluded that the “packing” of the lipid chains is tighter in
the inner mono-layer [57–59]. In addition the “arm-to-arm” prop-
agation observed for DLPC [60] is expected to be significant in
closely packed chains. The result of these two factors would be to
favor H-atom abstraction by the initially formed peroxyl radical
leading to the t,c (c,t) isomers (Scheme 2,1and 2) in competition
with -scission leading to the t,t-isomers 3and 4[42]. So the two
contrasting results, photo-oxidation of the bilayer combinations
(1) DMPC/ML/UCA (or DLPC/UCA) and conversely (2) DLPC/UCAME
illustrate the very important effect that different media such as lipid
A.A. Rand, L.R.C. Barclay / Journal of Photochemistry and Photobiology A: Chemistry 208 (2009) 79–90 89
bilayers, compared to homogeneous solution or micelles, have in
controlling the photo-oxidation reaction pathways in such aque-
ous/lipid systems.
Various observed non-ideal results of kinetic behavior of MLV
bilayers such as the kinetic order observed for the system DLPC/UCA
(SM, Fig. S7) and the non-linear plot according to Eq. (6) for the
system DLPC/UCAME/Trolox (Fig. 10) can be attributed to the
heterogeneity within the bilayer. In the latter example, the kinh
(18.8 ×104M−1s−1) extracted from the early part of the induction
period is not reliable probably due to photo-oxidation occurring
deep within the MLV system before Trolox has distributed uni-
formly throughout the system. Reactions other than propagation
by peroxyl radicals and their self-termination could complicate the
results. For example, the recent report of hydroxyl radical genera-
tion during azo-initiated peroxidation of methyl linoleate leading
to a lipid ketone could affect both kinetic and product studies, espe-
cially if this reaction “is more probable in membranes” as suggested
[61].
5. Conclusions
The formation of singlet oxygen by UV irradiation of urocanic
acid (UCA) is very well documented and the potential damag-
ing effect on human skin has been a major concern. However
the detection of singlet oxygen from electronically excited UCA
by energy transfer to oxygen does not necessarily mean that
the electronically excited oxygen results in TYPE II singlet oxy-
gen reactions under all conditions. In general competition can
exist between a target molecule and oxygen for the excited state
with a resulting competition between TYPE I, free radical pro-
cesses, and TYPE II by singlet oxygen. It is clear from the results
herein that urocanic acid in its 3n,* excited state is capable of
exhibiting hydrogen atom abstraction from organic substrates con-
taining weak C–H bonds such as unsaturated lipids in systems
where direct contact between UCA and the substrate is possible
such as in homogeneous solution or in dynamic aqueous/lipid
mixtures like micelles. In the presence of oxygen in these fluid
systems, free radical reactions of the TYPE I process are initi-
ated. In contrast, when the UCA is isolated from direct contact
with the organic substrate; for example, when UCA is irradiated
in the aqueous phase of lipid bilayers, energy transfer to oxygen
forms singlet oxygen and this competing reaction can dominate
leading to typical TYPE II reactions rather than free radical reac-
tions.
Kinetic methods and product studies of lipid hydroperoxides
provide useful methods to distinguish between these two main
reaction pathways initiated photochemically in solution and in
aqueous/lipid mixtures. Antioxidants are commonly applied to
studies in polar solvents and in heterogeneous lipid/aqueous sys-
tems wherein hydrogen bonding by solvents greatly reduce their
antioxidant activity. Such effects can be predicted by quantitative
methods of “solvent effects”. However in hydrophobic regions of
micelles and of lipid bilayers other factors, such as diffusion limiting
behavior, may dominate antioxidant activities and the reduction
of antioxidant activities are much greater than can be predicted.
In order to determine the intrinsic antioxidant activity of antioxi-
dants, it is advisable to first determine their activities in a non-polar
solvent where these effects are absent.
The potential dual effects of UV-excited UCA are of particular
significance for the protection of human skin against damaging
UV radiation. The simple model systems used in this study may
have significant implications when considering the processes in
more complex natural biological media. In real biological systems,
a knowledge of the singlet oxygen and of the free radical reactions
should both be considered in a strategy for protection against UV
radiation [62].
Supporting material
Supporting Material available includes: (1) Table S1, with addi-
tional details of the kinetic data for the peroxidation of methyl
linoleteate in SDS micelles inhibited by phenolic antioxidants;
(2) 1H NMR spectra of urocanic acid and methyl urocanate,
Figs. S1 and S2; (3) kinetic order plots of oxygen uptake with UV
light intensity during photo-oxidation of methyl linoleate under
different conditions, Figs. S3–S5; (4) kinetic order plots of oxy-
gen uptake with UV light intensity during peroxidation of methyl
linoleate in DMPC bilayers photo-initiated by urocanic acid, Fig. S6,
and of DLPC bilayers photo-initiated by methyl urocanate, Fig. S7;
(5) 31P NMR spectra of multilamellar and of unilamellar, DLPC
bilayers Fig. S8; (6) HPLC trace of the product isomers from photo-
peroxidation of DPLC by methyl urocanate, Fig. S9.
Acknowledgements
We thank the staff of our Department who provided techni-
cal help in this investigation, Danny Durant, Philip Cormier, Roger
Smith, and Patt Wilson. L.R.C.B. acknowledges receipt of a Discov-
ery Grant in support of this research from the Natural Sciences and
Engineering Council of Canada (NSERC) and A.A.R. acknowledges
receipt of an Undergraduate Student Award from NSERC.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.jphotochem.2009.08.008.
References
[1] E.L. Menon, H. Wasserman, Formation of singlet oxygen by urocanic acid by UVA
irradiation and some consequences thereof, Photochem. Photobiol. 75 (2002)
565–569.
[2] A. Kammeyer, S. Pavel, S.S. Asghar, J.D. Bos, M.B.M. Teunissen, Prolonged
increase of cis-urocanic acid in human skin and urine after single total-body
ultraviolet exposures, Photochem. Photobiol. 65 (1997) 593–598.
[3] T. Mohammad, H. Wasserman, H. HogenEsch, Urocanic acid photochemistry
and photobiology, Photochem. Photobiol. 69 (1999) 115–135.
[4] K.M. Hanson, B. Li, J.D. Simon, A spectroscopic study of the epidermal ultra-
violet chromophore trans-urocanic acid, J. Am. Chem. Soc. 119 (1997) 2715–
2721.
[5] K.M. Hanson, J.D. Simon, Epidermal trans-urocanic acid and the UV-A-induced
photoaging of the skin, Proc. Natl. Acad. Sci. U.S.A. 95 (1998) 10576–10578.
[6] J.D. Simon, Spectroscopic and dynamic studies of the epidermal chromophores
trans-urocanic acid and eumelanin, Acc. Chem. Res. 33 (2000) 307–313.
[7] K.M. Hanson, R.M. Clegg, Observation and quantification of ultraviolet-induced
reactive oxygen species in ex vivo human skin, Photochem. Photobiol. 76 (2002)
57–63.
[8] B. Li, K.M. Hanson, J.D. Simon, Primary processes of the electronic excited states
of trans-urocanic acid, J. Phys. Chem. 101 (1997) 969–972.
[9] N. Haralampus-Grynaviski, C. Ransom, T. Ye, M. Rô ˙
zanowska, M. Wrona, T.
Sarna, J.D. Simon, Photogeneration and quenching of reactive oxygen species
by urocanic acid, J. Am. Chem. 124 (2002) 3461–3468.
[10] W.L. Ryan, D.H. Levy, Electronic spectroscopy and photoisomerization of
trans-urocanic acid in a supersonic jet, J. Am. Chem. 123 (2001) 961–966.
[11] T. Mohammad, Laser-induced invitro isomerization of urocanic acid in UV
A region and the origin of excited triplet state, Tetrahedron Lett. 43 (2002)
8897–8900.
[12] R. Roa, K.E. O’Shea, Reactions of urocanic acid (UCA) methyl esters with sin-
glet oxygen and 4-methyl-1,2,4-triazoline-3,5-dione (MTAD), Tetrahedron 62
(2006) 10700–10708.
[13] J. Baier, T. Maisch, M. Maier, E. Engel, M. Landthaler, W. Bäumler, Singlet oxygen
generation by UVA light exposure of endogenous photosensitizers, Biophys. J.
91 (2006) 1452–1459.
[14] L. Shen, H.-F. Ji, Theoretical investigation of the photosensitization mechanisms
of urocanic acid, J. Photochem. Photobiol. B: Biol. 91 (2008) 96–98.
[15] O. Dmitrenko, W. Reischl, R.D. Bach, J. Spanget–Larsen, TD-DFT Computational
insight into the origin of wavelength-dependent E/Z photoisomerization of
urocanic acid, J. Phys. Chem. 108 (2004) 5662–5669.
[16] J. Danielsson, J. Uliˇ
cn ´
y, A. Laaksonen, A TD-DFT study of the photochemistry of
urocanic acid in biologically relevant ionic, rotameric, and photometric forms,
J. Am. Chem. Soc. 123 (2001) 9817–9821.
[17] L.R.C. Barclay, M.C. Basque, V.C. Stephenson, M.R. Vinqvist, Photoxidations
initiated or sensitized by biological molecules: singlet oxygen versus radical
90 A.A. Rand, L.R.C. Barclay / Journal of Photochemistry and Photobiology A: Chemistry 208 (2009) 79–90
peroxidation in micelles and human blood plasma, Photochem. Photobiol. 78
(2003) 248–255.
[18] E. Silva, L. Herrera, A.M. Edwards, J. de la Fuenta, E. Lissi, Enhancement of
riboflavin-mediated photo-oxidation of glucose 6-phosphate dehydrogenase
by urocanic acid, Photochem. Photobiol. 81 (2005) 206–211.
[19] A. Kammeyer, T.A. Eggelte, J.D. Bos, M.B.M. Teunissen, Urocanic acid isomers
are good hydroxyl radical scavengers: a comparative study with structural
analogues and with uric acid, Biochim. Biophys. Acta 1428 (1999) 117–120.
[20] A. Kammeyer, T.A. Eggelte, H. Overmars, A. Bootsma, J.D. Bos, M.B.M. Teunissen,
Oxidative breakdown of urocanic acid isomers by hydroxyl radical generating
systems, Biochim. Biophys. Acta 1526 (2001) 277–285.
[21] C. López-Alarcón, A. Aspée, C. Henríquez, A.M. Campos, E.A. Lissi, Interaction
and reactivity of urocanic acid towards peroxyl radicals, Redox Report 10 (2005)
227–234.
[22] C.S. Foote, Photosensitized oxidation and singlet oxygen: consequences in bio-
logical systems, in: W.A. Pryor (Ed.), Free Radicals in Biology, vol. II, Academic
Press, New York, 1976, pp. 85–133.
[23] L.R.C. Barclay, M.-C. Basque, M.R. Vinqvist, Singlet-oxygen reactions sensitized
on solid surfaces of lignin or titanium dioxide: product studies from hin-
dered secondary amines and from lipid peroxidation, Can. J. Chem. 81 (2003)
457–467.
[24] P.D. Maclean, E.R. Chapman, S.L. Dobrowolski, A. Thompson, L.R.C. Barclay,
Pyrroles as antioxidants: solvent effects and the nature of the attacking radical
on antioxidant activities and mechanisms of pyrroles, dipyrrinones, and bile
pigments, J. Org. Chem. 73 (2008) 6623–6635.
[25] L.R.C. Barclay, S.L. Locke, J.M. MacNeil, J. VanKessel, Quantitative studies of
the autoxidation of linoleate monomers sequestered in phosphatidylcholine
bilayers. Absolute rate constants in bilayers, Can. J. Chem. 63 (1985) 2633–2638.
[26] L.R.C. Barclay, K.A. Baskin, S.J. Locke, T.D. Schaefer, Benzophe-
none-photosensitized autoxidation of linoleate in solution and sodium
dodecyl sulfate micelles, Can. J. Chem. 65 (1887) 2529–2540.
[27] L.R.C. Barclay, K.A. Baskin, K.A. Dakin, S.J. Locke, M.R. Vinqvist, The antioxidant
activities of phenolic antioxidants in free radical peroxidation of phospholipid
membranes, Can. J. Chem. 68 (1990) 2258–2269.
[28] M.J. Thomas, W.A. Pryor, Singlet oxygen oxidation of methyl linoleate: iso-
lation and characterization of the NaBH4-reduced products, Lipids 15 (1980)
544–548.
[29] L.R.C. Barclay, E. Crowe, C.D. Edwards, Photo-initiated peroxidation of lipids in
micelles by azaaromatics, Lipids 32 (1997) 237–245.
[30] L.R.C. Barclay, J.K. Grandy, H.D. MacKinnon, H.C. Nicol, M.R. Vinqvist, Peroxi-
dations initiated by lignin model compounds: investigating the role of singlet
oxygen in photo-yellowing, Can. J. Chem. 76 (1998) 1805–1816.
[31] T. Mohammad, H. Morrison, A convenient synthesis of cis-urocanic acid, an
endogenous immunosuppressant, OPPI Briefs 32 (2000) 581–584.
[32] D.D.M. Wayner, G.W. Burton, Handbook of Free Radicals and Antioxidants in
Biomedicine, CRC Press, Boca Raton, FL, 1989, pp. 223–232.
[33] L.R.C. Barclay, A.M.H. Bailey, D. Kong, The antioxidant activity of ␣-tocopherol-
bovine serum albumin complex in micellar and liposome autoxidations, J. Biol.
Chem. 260 (1985) 15809–15814.
[34] L.R.C. Barclay, M.R. Vinquist, Phenols as antioxidants, in: Z. Rappoport (Ed.), The
Chemistry of Phenols, John Wiley and Sons, Ltd., Chichester, England, 2003, pp.
839–908.
[35] G.W. Burton, K.U. Ingold, Vitamin E: applications of the principles of physical
organic chemistry to the exploration of its structure and function, Acc. Chem.
Res. 19 (1986) 194–201.
[36] C. Tanielian, R. Mechin, R. Seghrouchi, C. Schweitzer, Mechanistic and kinetic
aspects of photosensitzation in the presence of oxygen, Photochem. Photobiol.
71 (2000) 12–19.
[37] F. Boscá, M.A. Miranda, I.M. Morea, A. Samadi, Involvement of type I and type
II mechanisms in the linoleic acid peroxidation photosensitized by tiaprofenic
acid, J. Photochem. Photobiol. B: Biol. 58 (2000) 1–5.
[38] B. Roschek, K.A. Tallman, C.L. Rector, J.G. Gillmore, D.A. Pratt, C. Punta, N.A.
Porter, Peroxyl radical clocks, J. Org. Chem. 71 (2006) 3527–3532.
[39] C. Tanielian, R. Mechin, Reaction and quenching of singlet molecular oxygen
with esters of polyunsaturated fatty acids, Photochem. Photobiol. 59 (1994)
263–268.
[40] J.A. Howard, K.U. Ingold, Absolute rate constants for hydrocarbon
oxidation XII. Rate constants for secondary peroxy radicals, Can. J. Chem. 46
(1968) 2661–2666.
[41] D.V. Gardner, J.A. Howard, K.U. Ingold, The inhibition of autoxidation by 2,4,6-
tri-tert-butyl substituted phenol, aniline, and thiophenol, Can. J. Chem. 42
(1964) 2847–2851.
[42] N.A. Porter, B.A. Weber, H. Weenen, J.A. Khan, Autoxidation of polyunsaturated
lipids. Factors controlling the stereochemistry of product hydroperoxides, J.
Am. Chem. Soc. 102 (1980) 5597–5601.
[43] W.A. Pryor, J.A. Cornicelli, L.J. Devall, B. Tait, B.K. Trivedi, D.T. Witiak, M. Wu, A
rapid screening to determine the antioxidant potencies of natural and synthetic
antioxidants, J. Org. Chem. 58 (1993) 3521–3532.
[44] L.R.C. Barclay, C.D. Edwards, K. Mukai, Y. Egawa, T. Nishi, Chain-
breaking naphtholic antioxidants: antioxidant activities of polyalkylben-
zochromanol, polyalkylbenzochromenol, and 2,3-dihydro-5-hydroxy-2,2,4-
trimethylnaphto[1,2-b]furan compared to an ␣-tocopherol model in sodium
dodecyl sulphate micelles, J. Org. Chem. 60 (1995) 2739–2744.
[45] G.W. Burton, T. Doba, E.J. Gabe, L. Hughes, F.L. Lee, L. Prasad, K.U. Ingold,
Autoxidation of biological molecules. 4. Maximizing the antioxidant activity
of phenols, J. Am. Chem. Soc. 107 (1985) 7053–7065.
[46] M. Novo, S. Felekyan, C.A.M. Seidel, W. Al-Soufi, Dye-exchange dynamics in
micellar solutions studied by fluorescence correlation spectroscopy, J. Phys.
Chem. B 111 (2007) 3614–3624.
[47] F. Antunes, L.R.C. Barclay, K.U. Ingold, M. King, J.Q. Norris, J.C. Scaiano, F. Xi,
On the antioxidant activity of melatonin, Free Radic. Biol. Med. 26 (1999)
117–128.
[48] C. López-Alarcón, A. Aspée, C. Henriquez, A.M. Campos, E.A. Lissi, Interaction
and reactivity of urocanic acid towards peroxyl radicals, Redox Report 10 (2005)
227–234.
[49] G. Litwinienko, K.U. Ingold, Solvent effects on the rates and mechanisms
of reaction of phenols with free radicals, Acc. Chem. Res. 40 (2007)
222–230.
[50] L. Castle, M.J. Perkins, Inhibition kinetics of chain-breaking phenolic antiox-
idants in SDS micelles. Evidence that intermicellar diffusion rates may be
rate-limiting for hydrophobic inhibitors such as ␣-tocopherol, J. Am. Chem.
Soc. 108 (1986) 6381–6382.
[51] L. Valgimigli, K.U. Ingold, J. Lusztyk, Antioxidant activities of vitamin E ana-
logues in water and a Kamlet-Taft ˇ-value for water, J. Am. Chem. Soc. 118
(1996) 3545–3549.
[52] D.W. Snelgrove, J. Lusztyk, J.T. Banks, P. Mulder, K.U. Ingold, Kinetic solvent
effects on hydrogen-atom abstractions: reliable quantitative predictions via a
single empirical equation, J. Am. Chem. Soc. 123 (2001) 469–477.
[53] M.H. Abraham, P.L. Grellier, D.V. Prior, P.P. Duce, Hydrogen bonding. Part 7.
A scale of solute hydrogen-bond activity based on log K values for com-
plexation in tetrachloromethane, J. Chem. Soc. Perkin Trans. II (1989) 699–
711.
[54] C.G. Swain, E.C. Lupton Jr., Field and resonance components of substituent
effects, J. Am. Chem. Soc. 90 (1968) 4328–4337.
[55] A.M. Campos, C. Cárcamo, E. Silva, S. García, E. Lemp, E. Alarcón, A.M. Edwards,
G. Günther, E. Lissi, Distribution of urocanic acid isomers between aqueous
solutions and n-octanol, liposomes or bovine serum albumin, J. Photochem.
Photobiol. B: Biol. 90 (2008) 41–46.
[56] L.R.C. Barclay, K.A. Baskin, D. Kong, S.J. Locke, Autoxidation of model mem-
branes. The kinetics and mechanism of autoxidation of mixed phospholipid
bilayers, Can. J. Chem. 65 (1987) 2541–2550.
[57] S.J. Marrink, A.E. Mark, Molecular dynamics simulation of the formation, struc-
ture, and dynamics of small phospholipid vesicles, J. Am. Chem. Soc. 125 (2003)
15233–15242.
[58] C.G. Brouillette, J.P. Segrest, T.C. Ng, J.L. Jones, Minimal size phosphatidylcholine
vesicles: effects of radius of curvature on head group packing and conformation,
Biochemistry 21 (1982) 4569–4575.
[59] J.P. Douliez, A.M. Bellocq, E.J. Dufourc, Effect of size, polydispersity and mul-
tilayering on solid state 31P- and 2H-NMR spectra, J. Chim. Phys. 91 (1994)
874–880.
[60] F. Antunes, L.R.C. Barclay, M.R. Vinqvist, R.E. Pinto, Determination of propagat-
ing and termination rate constants by using an extension to the rotating-sector
method: application to PLPC and DLPC bilayers, Int. J. Chem. Kinet. 30 (1998)
753–767.
[61] M. Frenette, J.C. Scaiano, Evidence for hydroxyl radical generation during lipid
(Linoleate) peroxidation, J. Am. Chem. Soc. 130 (2008) 9634–9635.
[62] J.L. McCullough, K.M. Kelly, Prevention and treatment of skin aging, Ann. N.Y.
Acad. Sci. 1067 (2006) 323–331.