Free radical oxidation of coriolic acid (13-(S)-hydroxy-9Z,11E- octadecadienoic acid)

Article abstract of DOI:10.1016/j.chemphyslip.2005.01.005

The reaction of (13S,9Z,11E)-13-hydroxy-9,11-octadecadienoic acid (1a), one of the major peroxidation products of linoleic acid and an important physiological mediator, with the Fenton reagent (Fe²⁺/EDTA/H ₂O₂) was investigated. In phosphate buffer, pH 7.4, the reaction proceeded with >80% substrate consumption after 4 h to give a defined pattern of products, the major of which were isolated as methyl esters and were subjected to complete spectral characterization. The less polar product was identified as (9Z,11E)-13-oxo-9,11-octadecadienoate (2) methyl ester (40% yield). Based on 2D NMR analysis the other two major products were formulated as (11E)-9,10-epoxy-13-hydroxy-11-octadecenoate (3) methyl ester (15% yield) and (10E)-9-hydroxy-13-oxo-10-octadecenoate (4) methyl ester (10% yield). Mechanistic experiments, including deuterium labeling, were consistent with a free radical oxidation pathway involving as the primary event H-atom abstraction at C-13, as inferred from loss of the original S configuration in the reaction products. Overall, these results provide the first insight into the products formed by oxidation of 1a with the Fenton reagent, and hint at novel formation pathways of the hydroxyepoxide 3 and hydroxyketone 4 of potential (patho)physiological relevance in settings of oxidative stress.

Full text of DOI:10.1016/j.chemphyslip.2005.01.005

Chemistry and Physics of Lipids 134 (2005) 161–171
Free radical oxidation of coriolic acid
(13-(S)-hydroxy-9Z,11E-octadecadienoic Acid)
P. Maninia, E. Camerab, M. Picardob, A. Napolitanoa,∗, M. d’Ischiaa
a
Department of Organic Chemistry and Biochemistry, University of Naples “Federico II”,
Via Cinthia 4, I-80126 Naples, Italy
Istituto Dermatologico S. Gallicano, Via S. Gallicano 25a, 00153 Rome, Italy
b
Received 25 June 2004; received in revised form 12 January 2005; accepted 12 January 2005
Abstract
The reaction of (13S,9Z,11E)-13-hydroxy-9,11-octadecadienoic acid (1a), one of the major peroxidation products of linoleic
acid and an important physiological mediator, with the Fenton reagent (Fe2+/EDTA/H2O2) was investigated. In phosphate buffer,
pH 7.4, the reaction proceeded with >80% substrate consumption after 4 h to give a defined pattern of products, the major
of which were isolated as methyl esters and were subjected to complete spectral characterization. The less polar product was
identified as (9Z,11E)-13-oxo-9,11-octadecadienoate (2) methyl ester (40% yield). Based on 2D NMR analysis the other two
major products were formulated as (11E)-9,10-epoxy-13-hydroxy-11-octadecenoate (3) methyl ester (15% yield) and (10E)-9-
hydroxy-13-oxo-10-octadecenoate (4) methyl ester (10% yield). Mechanistic experiments, including deuterium labeling, were
consistent with a free radical oxidation pathway involving as the primary event H-atom abstraction at C-13, as inferred from loss
of the original S configuration in the reaction products. Overall, these results provide the first insight into the products formed
by oxidation of 1a with the Fenton reagent, and hint at novel formation pathways of the hydroxyepoxide 3 and hydroxyketone
4 of potential (patho)physiological relevance in settings of oxidative stress.
© 2005 Elsevier Ireland Ltd. All rights reserved.
Keywords: Linoleic acid; Fenton reagent; Deuterium labeling; (1/2)D NMR; Lipid peroxidation
Abbreviations: 13-(S)-HODE, (13S,9Z,11E)-13-hydroxy-9,11-octadecadienoic acid; 13-(S)-HPODE, (13S,9Z,11E)-13-hydroperoxy-9,11-
octadecadienoic acid; LDL, low-density lipoprotein; PPAR-␥, peroxysome proliferator-activated receptor; BSTFA, bis(trimethylsilyl)-
trifluoroacetamide; EDTA, ethylendiaminotetraacetic acid; ABAP, 2,2ꢀ-azobis(2-amidinopropane); HRP, horseradish peroxidase; COSY, cor-
relation spectroscopy; HMQC, heteronuclear multiple quantum coherence; HMBC, heteronuclear multiple bond correlation; TOCSY, total
correlation spectroscopy; PICI/MS, positive chemical ionization/mass spectrometry; tetramethylsilane, TMS
∗
Corresponding author. Tel.: +39 081 674 133; fax: +39 081 674 393.
E-mail address: alesnapo@unina.it (A. Napolitano).
0009-3084/$ – see front matter © 2005 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.chemphyslip.2005.01.005

162
P. Manini et al. / Chemistry and Physics of Lipids 134 (2005) 161–171
1. Introduction
Although the (patho)physiological routes leading
to 1a production in settings of oxidative stress now
appear to be well established, much less is known
about the effects of reactive oxygen species on the
ultimate fate and biological activity of this important
lipid mediator. Knowledge of the oxidative chemistry
of 1a is currently limited to formation of (9Z,11E)-
13-oxo-9,11-octadecadienoic acid (2) by the action
of a NAD(+)-dependent dehydrogenase (13-HODE
dehydrogenase) (Bronstein and Bull, 1997), and no
other oxidation product has been isolated in pure form
and chemically characterized. Notably enough, 1a and
its congeners have been reported to be stable to various
oxidizing systems, including air, air/Fe2+/ascorbate,
air/Fe2+, air/Fe2+/H2O2, air/Fe3+, and have therefore
gained the reputation of excellent markers of lipid
peroxidation (Spiteller and Spiteller, 1997).
In the present paper, we have investigated the reac-
tion of 1a with Fe2+/EDTA/H2O2 (the Fenton reagent),
which is widely used to model non-enzymatic oxida-
tive processes, as well as with other oxidizing systems
of physiological relevance. Aim of the work was to
assess whether 1a displays patterns of reactivity other
than conversion to the keto compound 2, and to provide
a detailed structural characterization of the oxidation
products
(13S,9Z,11E)-13-Hydroxy-9,11-octadecadienoic
acid (13-(S)-HODE, 1a), commonly referred to as
coriolic acid, is a major product of lipid metabolism
which arises by 15-lipoxygenase-catalyzed oxidation
of linoleic acid followed by reduction of the resulting
hydroperoxide ((13S,9Z,11E)-13-hydroperoxy-9,11-
¨
octadecadienoic acid, 13-(S)-HPODE) (Kuhn, 1996).
Additional routes to 1a involve cycloxygenase-
catalyzed oxidation of linoleic acid (Hamberg and
Samuelsson, 1980) as well as non-enzymatic peroxi-
dation (Nemann and Khenkin, 1997). In these latter
cases, however, the R enantiomer is also formed to
comparable extents, along with the 9-hydroxy isomers,
due to the lack of stereo- and regioselectivity of these
reactions.
BothS-andracemic13-HODEareproducedbyava-
riety of cell types, including polymorphonuclear leuko-
cytes (Soberman et al., 1985), eosinophils (Engels et
al., 1996), breast carcinoma cells (Reddy et al., 1997),
and human dermal fibroblasts (Godessart et al., 1996).
Racemic 13-HODE has been identified as a component
of oxidized low-density lipoprotein (LDL) (Jira and
Spiteller, 1996; Jira et al., 1997) and is the chief
hydroxylated fatty acid in human psoriatic skin scales,
where it occurs in levels of 115 ng/mg (Baer et al.,
1990).
In mammalian cells 1a acts as a physiological medi-
ator, being involved in signal transduction and gene ex-
pression. It potentiates the mitogenic signal produced
by epidermal growth factor in human breast carcinoma
cells (Reddy et al., 1997), displays chemotactic proper-
ties toward polymorphonuclear leukocytes (Henricks
et al., 1991), modifies inflammatory cell activity
and is a ligand to peroxysome proliferator-activated
receptor (PPAR-␥) (Nagy et al., 1998; Marx et al.,
1999).
In the skin, 1a has been ascribed antiinflamma-
tory and antiproliferative properties, whereas in the
vascular system it acts as a chemorepellant, reducing
the adhesion of platelets (Buchanan et al., 1985; Haas
et al., 1988). It also causes relaxation of coronary
arteries (De Meyer et al., 1992) and may have both
pro- and anti-atherogenic effects, depending on the
rate of generation by the 15-lipoxygenase expressed
in the macrophages recruited at sites of atherosclerotic
lesions (Simon et al., 1989).
2. Materials and methods
2.1. Materials
Linoleic acid (99%), hydrogen peroxide (water
solution, 33%), d-mannitol, 2-iodobenzoic acid, oxone
(2KHSO5–KHSO4–K2SO4), sodium borohydride,
sodium borodeuteride and bis(trimethylsilyl)tri-
fluoroacetamide (BSTFA) were from Aldrich
Chemie; Fe(NH4)2(SO4)2·6H2O was from Carlo
Erba; ethylenediaminetetraacetic acid (EDTA) was
from Fluka; 2,2ꢀ-azobis(2-amidinepropane) (ABAP)
chlorhydrate was from Polysciences. Horseradish per-
oxidase (HRP) (H2O2 oxidoreductase; E.C. 1.11.1.7)
type II, catalase (H2O2:H2O2 oxidoreductase; EC

P. Manini et al. / Chemistry and Physics of Lipids 134 (2005) 161–171
163
1.11.1.6) and soybean lipoxidase (linoleate oxygen
reductase, E.C. 1.13.11.12) type IB were from Sigma.
Organic solvents were HPLC quality; phosphate buffer
(0.1 M, pH 7.4) and borate buffer (0.1 M, pH 9.0) were
treated with Chelex-100 resin before use to remove
transition metal contaminants.
(13S,9Z,11E)-13-Hydroxy-9,11-octadecadienoic
acid (1a, 13-(S)-HODE) and (13S,9Z,11E)
-13-hydroperoxy-9,11-octadecadienoic acid (13-
(S)-HPODE) were synthesized and purified as
described (Napolitano et al., 2002). 1a methyl ester
was obtained by treatment of the free acid with
diazomethane. o-Iodoxybenzoic acid was freshly
prepared from 2-iodobenzoic acid according to a
reported procedure (Frigerio et al., 1999).
chemical ionization/mass spectrometry (PICI/MS)
measurements were carried out using methane as the
reagent gas. Data were processed using G1701AA
data analysis software. The following analytical con-
ditions were used: 30 m cross-bond 5% diphenyl-95%
dimethylpolysiloxane column (0.25 mm i.d., 0.25 ␮m
d.f.). Temperature program: 40 ◦C, hold time 1 min;
up to 280 ◦C, rate 5 ◦C/min. The inlet and detector
were taken at 180 and 250 ◦C, respectively. The
acquisition started 5 min after the injection (solvent
delay 5 min), and was set in scan 50, and sampling was
1.6 scans/s.
Prior to GC/MS analysis, all the samples were
treated with 200 ␮L of ethereal diazomethane solu-
tion to give the corresponding methylester derivative.
When required, the samples were treated with 50 mL of
a 50% solution of BSTFA in pyridine and kept at 50 ◦C
for 30 min to give the corresponding trimethylsilyl
derivative.
Diazomethane was prepared by reaction of N-
methyl-N-nitroso-p-toluenesulfonamide in ethanolic
KOH and collected in peroxide-free ether in a dry
ice/acetone bath. Caution! Diazomethane is explosive
and must be kept at −20 ◦C.
UV and IR spectra were obtained using a Beckman
DU 640 spechtrophotometer and a FT-IR Perkin Elmer
1760-X spectrophotometer, respectively. NMR spectra
were recorded in CDCl3 with a Bruker DRX-400 MHz
2.2. Synthesis of (9Z,11E)-13-oxo-9,11-
octadecadienoic acid (2)
1a (120 mg, 0.41 mmol) was dissolved in ethyl
acetate (75 mL), treated with o-iodoxybenzoic acid
(1.05 g, 3.7 mmol) and kept under reflux at 85 ◦C. Af-
ter 90 min the mixture was filtered to remove unreacted
o-iodoxybenzoic acid as a white solid. The filtrate was
dried and purified on preparative TLC (eluant B) to
give pure 2 (105 mg, 88% yield). FT-IR, 1H NMR and
13C NMR for compound 2 were in accord with those
reported (Kuklev et al., 1997).
1
instrument. H and 13C NMR spectra were recorded
at 400.1 and 100.6 MHz, respectively, using TMS
as the internal standard. 1H–1H COSY, 1H–13C
1
HMQC, H–13C HMBC and TOCSY NMR experi-
ments were run at 400.1 MHz using standard pulse
programs.
For all oxidation products of 1a isolated as methyl
esters, resonances due to OCH3, COOCH3, C-2
(CH2), C-3/C-4/C-5/C-6/C-7/C-15/C-16/C-17 (CH2)
and C-18 (CH3) groups appear in the H/13C NMR
2.3. Synthesis of (13-2H,9Z,11E)-13-hydroxy-
1
spectra at δ (CDCl3) 3.66 (3H, s)/52.2, 175.0, 2.31
(2H, t J = 7.6 Hz)/34.8, 1.2–1.7 (16H, m)/22–38, 0.89
(3H, m)/14.8 in the order. Analytical and preparative
TLC analyses were performed on F254 silica gel
plates (0.25 and 0.5 mm, respectively) using 70:30
(v/v) cyclohexane–ethyl acetate (eluant A), 70:30
(v/v) cyclohexane–ethyl acetate plus 2% acetic acid
(eluant B). Ce(SO4)2 (0.05 M in 10% H2SO4), and
iodine were used for product detection on TLC plates.
Column chromatography was performed on silica gel
(70–270 mesh).
9,11-octadecadienoic acid (1b)
2 (100 mg, 0.34 mmol) dissolved in CHCl3 (1 mL)
was added to 0.1 M borate buffer, pH 9.0 (50 mL) and
treated under vigorous stirring at room temperature
with sodium borodeuteride (20 mg, 0.48 mmol). After
5 min the mixture was acidified to pH 3 with HCl and
extracted with CHCl3 (3 × 30 mL). The organic layers
were combined, treated with anhydrous sodium sul-
fate, and dried under reduced pressure to afford pure
1b (70 mg, 70% yield).
GC/MS was carried out on a GC instrument
coupled with a quadrupole mass spectrometer. Helium
was the carrier gas with a 1 mL/min flow rate. Positive
1b: 1H NMR (400 MHz, CDCl3), δ (ppm):
1.40–1.55 (4H, m, H-8, H-14), 5.43 (1H, dt J = 10.8,
7.6 Hz, H-9), 5.65 (1H, d J = 15.2 Hz, H-12), 5.96 (1H,

164
P. Manini et al. / Chemistry and Physics of Lipids 134 (2005) 161–171
t J = 10.8 Hz, H-10), 6.47 (1H, dd J = 15.2, 10.8 Hz, H-
11).
stirring and at 60 ◦C with ABAP (5.4 mg, 0.02 mmol).
After 4 h the reaction mixture was worked up and
analyzed as above.
2.4. Oxidation of 1a with the Fe2+/EDTA/H2O2
system
2.7. Isolation of compounds 2–4 methyl ester
1a (250 mg, 0.84 mmol) was dissolved in CHCl3
(1 mL) and emulsified in 0.1 M phosphate buffer,
pH 7.4 (170 mL). The mixture was treated un-
der vigorous stirring at room temperature with
Fe(NH4)2(SO4)2·6H2O (164 mg, 0.42 mmol) and
EDTA (156 mg, 0.42 mmol). The oxidation was
started by addition of four aliquots of 1 M hydrogen
peroxide (420 ␮L, 0.56 mmol each) over a period
of 45 min. After 4 h working up as above afforded
a yellow oily residue which was fractionated by
column chromatography (2.0 cm × 40 cm) using a
cyclohexane/ethyl acetate gradient (starting from pure
cyclohexane up to cycloesane/ethyl acetate 6:4 (v/v))
as the eluant to give, besides unreacted 1a methyl ester
(Rf = 0.80, eluant A, 50 mg, 20% yield), 2 methyl ester
(Rf = 0.82, eluant A, 103 mg, 40% yield), 3 methyl
ester (Rf = 0.48, eluant A, 41 mg, 15% yield) and 4
methyl ester (Rf = 0.35, eluant A, 27 mg, 10% yield).
In other experiments the prepative scale reaction was
carried out using 1b as the substrate.
The reaction was carried out under conditions
substantially similar to those reported (Spiteller and
Spiteller, 1997) with some modifications. In brief, 1a
(15 mg, 0.05 mmol) was dissolved in CHCl3 (100 ␮L)
and added to 0.1 M phosphate buffer, pH 7.4 (22 mL)
containing 0.2 M KCl (45 mL). The mixture was
treated under vigorous stirring at room temperature
withFe(NH4)2(SO4)2·6H2O(19.8 mg, 0.05 mmol)and
EDTA(18.6 mg, 0.05 mmol). Theoxidationwasstarted
by addition of four aliquots of 1 M hydrogen peroxide
(12.5 ␮L, 0.013 mmol each) over a period of 45 min.
After 4 h the reaction mixture was extracted with ethyl
acetate (2 × 10 mL), and the organic layers were com-
bined, dried over sodium sulfate and treated with an
ethereal solution of diazomethane (2 mL). After 15 min
the mixture was dried under reduced pressure and the
residue was analyzed by TLC (eluant A).
Similarexperimentswerecarriedout(i)asabovebut
omitting EDTA; (ii) treating the reaction mixture with
d-mannitol (3.2 g, 16.8 mmol), or catalase (3100 U);
(iii) under an oxygen depleted atmosphere, purging all
the reagents with a flux of argon.
In other experiments the reaction was carried out
under the general conditions described above but using
13-(S)-HPODE as the substrate.
3 methyl ester: FT-IR (CHCl3) νmax 3547–3296,
1732, 968, 883 cm−1; [␣]D25 = 0.0 (c 2.7, CHCl3);
1H NMR (400 MHz, CDCl3),
δ
(ppm): see
Table 1; 13C NMR (100 MHz, CDCl3) δ (ppm):
see Table 1; GC/MS (tetramethylsilane (TMS)
derivative) tR 40.1 min; EI m/z (relative inten-
sity) 398 (M+, 3), 383 (M+ − CH3, 15), 327
(M+ − C5H11, 6), 241 (M+ − (CH2)7CO2CH3, 8), 199
(M+ − C8H14OSi(CH3)3 and M+ − C9H16OCO2CH3,
100) (see Fig. 1); PICI m/z 399 ([M + H]+).
2.5. Oxidation of 1a with the HRP/H2O2 system
1a (25 mg, 0.084 mmol) was dissolved in CHCl3
(100 ␮L) and added to 0.1 M phosphate buffer, pH 7.4
(17 mL). The mixture was treated under vigorous stir-
ring at room temperature with HRP (52 U). One molar
of hydrogen peroxide was then added in seven aliquots
(84.3 ␮L, 84.3 mmol each) over a period of 90 min. Af-
ter4 hthereactionmixturewasworkedupandanalyzed
as above.
4 methyl ester: FT-IR (CHCl3) νmax 3547–3296,
1732, 1715 cm−1; [␣]D25 = 0.0 (c 2.0, CHCl3); 1H NMR
(400 MHz, CDCl3), δ (ppm): see Table 1; 13C NMR
(100 MHz, CDCl3) δ (ppm): see Table 1; GC/MS (TMS
derivative) tR 41.0 min; EI m/z (relative intensity) 398
(M+, 2), 383 (M+ − CH3, 3), 367 (M+ − OCH3, 2),
308 (M+ − (CH3)3SiOH, 3), 299 (M+ − C5H11CO, 4),
285 (M+ − C5H11COCH2, 5), 259 (M+ − C9H15O, 6),
241 (M+ − C9H17O2, 100), 179 (308 − C7H13O2, 6),
157 (M+ − C13H25O2Si, 4), 129 (•C7H13O2, 3), 99
(•C5H11CO, 78), 71 (•C5H11, 18) (see Fig. 1); PICI
m/z 399 ([M + H]+).
2.6. Oxidation of 1a with ABAP
1a (15 mg, 0.05 mmol) was dissolved in CHCl3
(100 ␮L) and added to 0.1 M phosphate buffer, pH 7.4
(13.5 mL). The mixture was treated under vigorous

P. Manini et al. / Chemistry and Physics of Lipids 134 (2005) 161–171
165
Table 1
Selected spectral data of compounds 3 and 4 methyl esters
Carbon
3
4
δC
32.7
61.4
58.7
128.9
138.8
72.9
δH (mult. J (Hz))
δC
37.9
73.4
138.2
124.0
46.9
δH (mult. J (Hz))
8
9
1.53–1.59 (2H, m)
2.81 (1H, dt 5.6, 2.4)
3.09 (1H, dd 7.6, 2.4)
5.41 (1H, dd 15.6, 7.6)
5.92 (1H, dd 15.6, 6.4)
4.13 (1H, m)
1.53 (2H, m)
4.10 (1H, m)
5.57 (1H, dd 15.2, 6.4)
5.76 (1H, dt 15.2, 6.8)
3.15 (2H, d 6.8)
–
10
11
12
13
14
210.0
43.5
37.9
1.53–1.59 (2H, m)
2.42 (2H, t 7.2)
3. Results
Compound A (103 mg, about 40% yield) was iden-
tified as (9Z,11E)-13-oxo-9,11-octadecadienoate (2)
methyl ester by straightforward spectral analysis.
The 1H NMR spectrum of product B (41 mg, 15%
yield) displayed as most significant features two sig-
nals at δ 5.41 (1H, dd J = 15.6, 7.6 Hz) and 5.92 (1H,
dd J = 15.6, 6.4 Hz), the latter showing coupling with
a signal at δ 4.13 (1H, m), suggesting an allylic al-
cohol moiety with an E configuration at the double
The reaction of 1a with the Fenton reagent
(Fe2+/EDTA/H2O2, 1 molar equivalent) was typically
carried out with the lipid emulsified in 0.1 M phosphate
buffer, pH 7.4 (Spiteller and Spiteller, 1997). After 4h,
HPLC analysis showed ca. 80% substrate consump-
tion and the formation of a defined pattern of products.
TLC analysis of the mixture after methylation with di-
azomethane revealed the presence of three main prod-
ucts at Rf = 0.88 (A, λ = 254 nm), 0.48 (B, positive to
Ce(SO4)2), and 0.35 (C, weak UV absorption, positive
to Ce(SO4)2). Column chromatography (eluant cyclo-
hexane/ethyl acetate gradient) afforded the products in
pure form.
1
bond. The H–1H COSY spectrum revealed correla-
tion of the olefin proton at δ 5.41 with a resonance
at δ 3.09 which in turn gave a cross peak with a sig-
nal at δ 2.81. These latter two signals were strongly
suggestive of protons on a trans epoxide function-
ality.
The 1H–13C HMQC spectrum showed cross peaks
between the epoxide proton resonances at δ 2.81 and
3.09 and carbon resonances at δ 61.4 and 58.7, re-
spectively; between the olefin proton signals at δ 5.92
and 5.41 and two carbon signals at δ 138.8 and 128.9;
and between the multiplet at δ 4.13 and a carbon sig-
nal at δ 72.9. The 1H–13C HMBC spectrum displayed
well apparent correlations between the proton at δ 4.13
and carbon resonances at δ 37.9 (CH2), 138.8 and
128.9, corroborating the allylic alcohol functionality.
Additional correlations between the olefin proton res-
onance at δ 5.41 and a carbon signal a δ 58.7; and
between the epoxide proton signal at δ 3.09 and the
carbon resonance at δ 61.4 indicated that the epox-
ide ring was close to the carbon–carbon double bond
(see Table 1).
Finally, a TOCSY experiment, showing a well ap-
parent correlation between the terminal methyl group
at δ 0.89 and the proton alpha to the OH group at δ 4.13,
provided conclusive evidence that the position of the
OH was unchanged with respect to 1a.
Fig. 1. Origin of salient fragmentation peaks in the EI/MS spectrum
of the O-TMS derivative of compounds 3 (a) and 4 (b) methyl esters.

166
P. Manini et al. / Chemistry and Physics of Lipids 134 (2005) 161–171
On the basis of these data, the product was as-
signed the structure of (11E)-9,10-epoxy-13-hydroxy-
11-octadecenoate (3) methyl ester.
displayed cross peaks in the 1H–13C HMBC spectrum
with the methylene resonances at δ 2.42 and 3.15. Fur-
thermore, the TOCSY spectrum indicated correlation
between the terminal methyl group at δ 0.89 and the
methylene protons at δ 2.42, adjacent to the carbonyl
functionality, as well as between the resonance of the
proton alpha to the OH group at δ 4.10 and the signal
at δ 1.53 and the chain resonances in the region at δ
1.2–1.8. These latter data clearly denoted that the al-
lylic OH group was on the 9-position and, hence, that
the former alcoholic functionality at the 13-position
was replaced by the carbonyl group.
On this basis, the compound was assigned the struc-
ture of the hydroxyketone 4 methyl ester. A closely
similar proton spectrum was reported for a mixture
of positional isomers (13-/9-oxo-9-/13-hydroxy-
trans-10/11-octadecenoic acids) obtained by Fe(II)-
catalyzeddecompositionof13-HPODE(Gardneretal.,
1974).
Consistent with proposed structure 4 the EI/MS
spectrum of the O-TMS derivative showed a molecular
ion peak at m/z 398, and the PICI spectrum a pseudo-
molecular ion peak at m/z 399. The diagnostic EI frag-
mentation pattern (see Fig. 1) confirmed the structural
formulation proposed for 4 and its positional isomers
identified as constituents of complex mixtures obtained
under different conditions (Sessa et al., 1977; Spiteller
and Spiteller, 1998).
In accord with this conclusion, the FT-IR spectrum
displayed intense bands in the range 3547–3296 cm−1
,
consistent with a hydroxyl group; at 968 cm−1 (C H
bending of trans olefins) and at 883 cm−1 typical
for C H stretching of trans epoxides (Gardner and
Kleiman, 1981). The EI/MS spectrum of the O-TMS
derivative gave a molecular ion peak at m/z 398, in line
with the PICI (positive ion chemical ionization) spec-
trum showing pseudo-molecular ion peak at m/z 399.
Diagnostic fragmentation peaks in the EI/MS spectrum
corroborated the structural conclusions derived from
NMR analysis (see Fig. 1). The proton and mass spec-
trum of the compound bore considerable resemblance
to those reported for fractions containing methyl 13-
/9-hydroxy-9,10/12,13-epoxyoctadec-11/10-enoates
obtained by decomposition of 13-(S)-HPODE (Frankel
et al., 1977; Neff et al., 1978; Blee et al., 1993; Gardner
and Kleiman, 1981; Spiteller and Spiteller, 1998;
Spiteller, 1998). Interestingly, the product was opti-
cally inactive, suggesting that the stereogenic centre
was affected during the oxidation process.
Overall yields of products 2–4 methyl esters ac-
counted for about 80% of reacted 1a.
Other minor components of the reaction mixture
could not be obtained in sufficient amounts for struc-
tural characterization. Thus, for example, a fraction
(Rf = 0.70–0.80, eluant A) which resisted at all attempts
of purification of preparative value, and on NMR analy-
sis proved to consist of the starting material and a minor
constituent featuring an allylic alcoholic group, a trans
double bond and a trans epoxide functionality. Based
on the mapping of the spin system as deduced by the
1H–1H COSY spectrum of the gross fraction, the com-
pound was tentatively formulated as a diastereomer
of 3.
Depletion of oxygen by running the reaction un-
der an argon atmosphere markedly slowed down the
rate of substrate consumption and product formation,
whereas omission of H2O2 or EDTA resulted in com-
plete prevention of the reaction, in line with previous
observations (Spiteller and Spiteller, 1997).
The most polar product (C) exhibited a diagnos-
tic band in the FT-IR spectrum at 1715 cm−1, indicat-
1
ing a C O group. The H NMR spectrum (Gardner
et al., 1974) displayed a characteristic pattern of res-
onances at δ 5.76 (1H, dt, J = 15.0, 6.8 Hz), 5.57 (1H,
dd, J = 15.0, 6.4 Hz), and 4.10 (1H, m), suggestive of
an allylic alcohol moiety with E configuration. The sig-
nal at δ 5.76 was moreover coupled with a doublet at δ
3.15 ascribed to a dieshielded CH2 group. Two triplets
for additional methylene groups at δ 2.42 and 2.31 (2H
each) were also observed.
Noticeable feature of the 13C NMR spectrum was
a signal at δ 210.0 ascribed to a keto group, which

P. Manini et al. / Chemistry and Physics of Lipids 134 (2005) 161–171
167
In separate experiments it was found that products
2–4 were also formed by HRP/H2O2 oxidation of 1a
as well as by vigorously stirring 1a in air-equilibrated
phosphate buffer at pH 7.4 in the presence of the free
radical initiator ABAP. Under the latter conditions the
reaction rate was much slower than that observed in
the Fenton reaction (40% consumption at 4 h reaction
time) and the overall product yield was about 70% with
respect to reacted 1a.
For comparative purposes, the hydroperoxide 13-
(S)-HPODE was allowed to react with the Fenton
reagent under the typical reaction conditions used for
1a. After 4 h substrate consumption was >90% and
a much more complex product pattern was observed
compared to the case of 1a. Following methylation,
compounds 2–4 methyl esters could be identified by
TLC analysis under different elution conditions and
NMR analysis of the reaction mixture.
Oxidation of 1a by the Fenton reagent
(Fe2+/EDTA/H2O2) conceivably involves the gen-
eration of the OH radical as the species triggering
substrate conversion. The role of this radical was
supported by experiments run in the presence of OH
radical scavengers, such as d-mannitol and DMSO,
which inhibited both substrate consumption and
product formation. Typically, a >95% recovery of
1a was obtained using 100-fold molar excess of
d-mannitol at 4 h reaction time.
To inquire into the mechanism of the Fenton-
induced oxidation, a competition experiment was
devised in which an equimolar mixture of 1a and
(13-2H,9Z,11E)-13-hydroxy-9,11-octadecadienoic
acid (1b) was allowed to react with the Fenton reagent
as described above. Analysis of unreacted 1a/b at
about 50% consumption indicated a virtually 1:1
ratio, indicating a comparable extent of oxidation. In
other experiments 1b was oxidized with the Fenton
reagent and the reaction products were isolated and
analyzed for deuterium content at about 50% substrate
consumption. No detectable signal for the proton
alpha to the OH group was observed in the spectrum
of the recovered substrate. However, isolated product
Fig. 2. Selected region of the 1H NMR spectrum of compound 3
methyl ester obtained from oxidation of 1a (top plot) and 1b (bottom
plot).
provided additional evidence that the isolated product
was indeed a mixture of 13-1H/2H 3 methyl ester.
4. Discussion
Oxidative modifications of polyunsaturated fatty
acids and their derivatives represent important post-
synthetic pathways of lipid metabolism and transfor-
mation, and have been the focus of extensive chemical
and biochemical literature (Dix and Marnett, 1985;
Aikens and Dix, 1993; Schneider et al., 2001). In this
frame, the isolation and complete spectral character-
ization of the oxidation products of 1a reported in this
1
3 methyl ester featured in the H NMR spectrum a
well discernible resonance at δ 4.13 with an integrated
area of ca. 0.15H taking the protons of the double
bond as reference (see Fig. 2, bottom panel). The
splitting of the proton signal at δ 5.92, due to the
1
2
presence of either H or H on the vicinal carbon,

168
P. Manini et al. / Chemistry and Physics of Lipids 134 (2005) 161–171
paper offers an improved basis to look at the biological
fate and roles of 1a under oxidative stress conditions.
Compound 3 and structurally related epoxides
were described in the reaction of linoleic and linolenic
acid or their hydroperoxides with lipoxygenases from
different sources (Hamberg and Hamberg, 1990;
Hamberg, 1989). In these processes the epoxidation is
the result of an epoxidase activity of the enzymes that
use hydroperoxides as oxygen donor, as demonstrated
by 18O labeling experiments (Blee et al., 1993). On
the other hand, the ketone 4 was identified among the
constituents of soybean phosphatidylcholines based
on MS analysis (Sessa et al., 1977). Compounds 3
and 4 were also identified among the products formed
in the decomposition of 13-(S)-HPODE as promoted
by Fe2+ (Spiteller and Spiteller, 1998; Gardner et
al., 1974) or the cysteine–FeCl3 catalyst (Gardner
and Jursinic, 1981) or in the autooxidation of methyl
linoleate (Frankel et al., 1977; Neff et al., 1978).
In all these studies, however, compounds 3 and 4
could not be distinguished from their regioisomers,
i.e. 12,13-epoxy-9-hydroxy-10-octadecenoate and
9-oxo-13-hydroxy-11-octadecenoate.
Product 1a was postulated as one of the products
or an intermediate in the iron(II) catalyzed decomposi-
tion of the corresponding hydroperoxide (Mlakar and
Spiteller, 1996; Spiteller and Spiteller, 1998). Further
transformation of this species under such reaction con-
ditionsto givehydroxy/hydroperoxidesderivatives was
also proposed (Mlakar and Spiteller, 1996). Yet, in no
case 1a was considered as a direct precursor of prod-
ucts 3 and 4. Moreover, the specific conditions of the
Fenton system including EDTA–chelated iron(II) and
hydrogen peroxide have never been examined as the
oxidizing agent toward either 1a or 13-(S)-HPODE.
Based on the mechanistic experiments including
deuterium labeling, it could be argued that formation
of products 2–4 by free radical oxidation of 1a pro-
ceeds by an oxygen-dependent mechanism outlined in
Scheme 1.
Fig. 3. Possible pentadienyl radicals generated by H-atom abstrac-
tion from 1a.
(Mlakar and Spiteller, 1996). The comparable rate
of consumption of 1a and 1b indicated that H-atom
abstraction at C-13 was not rate-determining in the
oxidation/oxygenation process.
The proposed route would also be consistent with
the observed oxidation or loss of configuration at the
C-13 stereogenic centre in products 2–4 as indicated
in the case of 3 by ORD measurements. Whereas with
the Fenton reagent the critical H-atom abstraction step
would be brought about by the OH radical, in the case
of the HRP/H2O2 system this could be done by the en-
zyme forms in which iron is at higher oxidation states,
in particular compound I. This is two oxidizing equiv-
alents above the resting ferrous state and consists of
a haem derived ferryl intermediate (FeIV O) and por-
phyrin ␲ cation radical which make it competent to
oxidize a broad spectrum of organic substrates by se-
quential one-electron transfer mechanisms. Alkylper-
oxyl radicals generated in the aerobic decomposition
of the free radical initiator ABAP can likewise be en-
visaged to account for the initial H-atom abstraction.
Subsequent coupling of radical I with oxygen at the
9- or 13-positions would give rise to peroxyl radicals
III and IV. Radical III could be engaged with two main
reaction routes. The first involves a rearrangement step
with loss of hydroperoxyl radical HO2• to give ketone
2. Similar mechanisms are usually reported to account
for the generation of ␣,␤-unsaturated ketones by con-
version of hydroxy- and hydroperoxy derivatives pro-
duced during lipid peroxidation (Spiteller, 1998, 2001).
The other route involves conversion of III to the hy-
droperoxide V which in the presence of a hydrogen
donor may be reduced by Fe2+ to give the alkoxyl rad-
ical VI, by a mechanism similar to that of the Fenton
reaction. Radical VI then would be converted to 2 by
H-atom abstraction from a donor and dehydration of
the resulting geminal diol. Formation of peroxyl radi-
cals of type III and their subsequent decomposition to
According to this scheme, the primary event would
be H-atom abstraction from the 13-position to give
the resonance-stabilized pentadienyl radical I (Fig. 3).
Formation of this radical would be favored over the al-
ternative H-atom abstraction route from the 8-position,
leading to the pentadienyl radical II, because of the
lower energy of the C H bond adjacent to the OH
group, as reported in the case of alcohols and polyols

P. Manini et al. / Chemistry and Physics of Lipids 134 (2005) 161–171
169
Scheme 1. Proposed mechanism of formation of compounds 2–4.
alkoxy radicals was proposed in the aerobic oxidation
of alcohols to carbonyl compounds mediated by tran-
sition metal salts (Iwahama et al., 2000; Minisci et al.,
2003, 2004).
Through an analogous reaction pathway, radical IV
could evolve to give the alkoxyl radical VII. This latter
could be converted to the corresponding alcohol, and
then to 4 by tautomerization, or undergo intramolecular
cyclization to 3. The mechanistic route to 3 involves in
the final stage an exchange at the C-13 with a H-atom
donor. It is likely that the starting alcohol substrate
could primarily serve this function at least in the initial
stages of the reaction. Thus, the observed partial (15%)
gain of hydrogen at the C-13 site in product 3 obtained
from 1b could be interpreted as due to the contribution
to this latter step of H-atom donors other than the

170
P. Manini et al. / Chemistry and Physics of Lipids 134 (2005) 161–171
starting material. Plausible candidates are hydroper-
oxide intermediates e.g. that deriving from peroxyl
radical IV. It is clear however that the OH bearing
C-13 site remains the main source of 1H/2H atom.
lar vescicles (liposomes): comparative and mechanistic studies.
Arch. Biochem. Biophys. 305, 516–525.
Baer, A.N., Costello, P.B., Green, F.A., 1990. Free and esteri-
fied 13(R,S)-hydroxyoctadecadienoic acids: principal oxygenase
products in psoriatic skin scales. J. Lipid Res. 31, 125–130.
Blee, E., Wilcox, A.L., Marnett, L.J., Schuber, F., 1993. Mechanism
of reaction of fatty acid hydroperoxides with soybean peroxyge-
nase. J. Biol. Chem. 268, 1708–1715.
5. Conclusions
Bronstein, J.C., Bull, A.W., 1997. Substrate specificity and character-
ization of partially purified rat liver 13-hydroxyoctadecadienoic
acid (13-HODE) dehydrogenase. Arch. Biochem. Biophys. 348,
219–225.
Buchanan, M.R., Haas, T.A., Lagarde, M., Guichardant, M., 1985.
13-Hydroxyoctadecadienoic acid is the vessel wall chemorepel-
lant factor, LOX. J. Biol. Chem. 260, 16056–16059.
De Meyer, G.R., Bult, Y.H., Verbeuren, T.J., Herman, A.G., 1992.
The role of endothelial cells in the relaxations induced by 13-
hydroxy- and 13-hydroperoxylinoleic acid in canine arteries. Br.
J. Pharmacol. 107, 597–603.
Dix, T.A., Marnett, L.J., 1985. Conversion of linoleic acid hydroper-
oxide to hydroxy, keto, epoxyhydroxy, and trihydroxy fatty acids
by hematin. J. Biol. Chem. 260, 5351–5357.
Engels, F., Kessels, G.C., Henricks, P.A., Nijkamp, F.P., 1996. Prefer-
ential formation of 13-hydroxylinoleic acid by human peripheral
blood eosinophils. Prostaglandins 52, 117–124.
Frankel, E.N., Neff, W.E., Rohwedder, W.K., Khambay, B.P.S., Gar-
wood, R.F., Weedon, B.C.L., 1977. Analysis of autoxidized fats
by gas chromatography-mass spectrometry. II. Methyl linoleate.
Lipids 12, 908–913.
Frigerio, M., Santagostino, M., Sputore, S., 1999. A user-friendly
entry to 2-iodoxybenzoic acid (IBX). J. Org. Chem. 64, 4537–
4538.
Gardner, H.W., Kleiman, R., Weisleder, D., 1974. Homolytic decom-
position of linoleic acid hydroperoxide: identification of fatty
acid products. Lipids 9, 696–706.
This study represents the first investigation of the
reactivity of 1a, one of the major metabolic product of
linoleic acid peroxidation, toward the Fenton reagent
under biomimetic conditions. Although generally
regarded as a convenient model for free radical oxi-
dations of (patho)physiological relevance, to the best
of our knowledge the reactivity of the Fenton system
toward polyunsaturated lipids and products thereof has
remained so far uncharted. Highlights of the present
study include: (a) the first isolation and complete
structural characterization of products 3 and 4 previ-
ously described among the constituents of the mixtures
generated by decomposition of 13-(S)-HPODE or
obtained by lipoxygenase-catalyzed oxygenation of
linoleic acid, but not unambigously discriminated from
other regioisomers; (b) provide the first experimental
evidence for a prominent reactivity of 1a toward
oxidation as promoted by OH radicals in contrast with
previous reports. Overall the results presented hint
at novel (patho)physiologically relevant pathways of
linoleic acid oxidation in oxidative stress settings.
Gardner, H.W., Jursinic, P.A., 1981. Degradation of linoleic acid
hydroperoxidebyacysteine–FeCl3 catalystasamodelforsimilar
biochemical reactions. I. Study of oxygen requirement, catalyst
and effect of pH. Biochim. Biophys. Acta 665, 100–112.
Gardner, H.W., Kleiman, R., 1981. Degradation of linoleic acid hy-
droperoxides by a cysteine–FeCl3 catalyst as a model for similar
biochemical reactions. II. Specificity in formation of fatty acid
epoxides. Biochim. Biophys. Acta 665, 113–125.
Godessart, N., Camacho, M., Lopez-Belmonte, J., Anton, R., Garcia,
M., de Moragas, J.M., Vila, L., 1996. Prostaglandin H-synthase-2
is the main enzyme involved in the biosynthesis of octadecanoids
from linoleic acid in human dermal fibroblasts stimulated with
interleukin-1 beta. J. Invest. Dermatol. 107, 726–732.
Haas, T.A., Bastida, E., Nakamura, K., Hullin, F., Admirall, L.,
Buchanan, M.R., 1988. Binding of 13-HODE and 5-, 12- and
15-HETE to endothelial cells and subsequent platelet, neutrophil
and tumor cell adhesion. Biochim. Biophys. Acta 961, 153–159.
Hamberg, M., 1989. Fatty acid hydroperoxide isomerase in Saproleg-
nia parasitica: structural studies of epoxy alcohols formed from
isomeric hydroperoxyoctadecadienoates. Lipids 24, 249–255.
Hamberg, M., Samuelsson, B., 1980. Stereochemistry in the forma-
tion of 9-hydroxy-10,12-octadecadienoic acid and 13-hydroxy-
Acknowledgments
This study was carried out in the frame of
MIUR (“Sostanze naturali ed analoghi sintetici
`
ad attivita antitumorale”, PRIN 2003) and Min-
istry of Health (“Studio comparativo sullo stress
ossidativo nell’invecchiamento cutaneo spontaneo
e foto-indotto”) projects. We thank the “Centro
Interdipartimentale di Metodologie Chimico-Fisiche”
(CIMCF, University of Naples Federico II) for NMR
and mass facilities. We thank Mrs. Silvana Corsani for
technical assistance.
References
Aikens, J., Dix, T.A., 1993. Hydrodioxyl (perhydroxyl), peroxyl, and
hydroxyl radical-initiated lipid peroxidation of large unilamel-

P. Manini et al. / Chemistry and Physics of Lipids 134 (2005) 161–171
171
9,11-octadecadienoic acid from linoleic acid by fatty acid cy-
clooxygenase. Biochim. Biophys. Acta 617, 545–547.
Hamberg, M., Hamberg, G., 1990. Hydroperoxide-dependent epox-
idation of unsaturated fatty acids in the broad bean (Vicia faba
L.). Arch. Biochem. Biophys. 283, 409–416.
Henricks, P.A., Engels, F., van der Vliet, H., Nijkamp, F.P., 1991.
9- and 13-Hydroxylinoleic acid possess chemotactic activity
for bovine polymorphonuclear leukocytes. Prostaglandins 41,
21–27.
Iwahama, T., Yoshino, Y., Keitoku, T., Sakaguchi, S., Ishii, Y.,
2000. Efficient oxidation of alcohols to carbonyl compounds with
molecularoxygencatalyzedbyN-hydroxyphthalimidecombined
with a Co species. J. Org. Chem. 65, 6502–6507.
Napolitano, A., Camera, E., Picardo, M., d’Ischia, M., 2002. Re-
actions of hydro(pero)xy derivatives of polyunsaturated fatty
acids/esters with nitrite ions under acidic conditions. Unusual ni-
trosative breakdown of methyl 13-hydro(pero)xyoctadeca-9,11-
dienoate to a novel 4-nitro-2-oximinoalk-3-enal product. J. Org.
Chem. 67, 1125–1132.
Neff, W.E., Frankel, E.N., Scholfield, C.R., Weisleder, D.,
1978. High-pressure liquid chromatography of autoxidized
lipids. I. Methyl oleate and linoleate. Lipids 13, 414–
421.
Nemann, R., Khenkin, A.M., 1997. In: Patai, S., Rapport, Z. (Eds.),
The Chemistry of Dienes and Polyenes, vol. 1, John Wiley and
Sons, New York, pp. 1916–1918.
Jira, W., Spiteller, G., 1996. Increase in hydroxy fatty acids in hu-
man low density lipoproteins with age. Chem. Phys. Lipids 84,
165–173.
Reddy, N., Everhart, A., Eling, T., Glasgow, W.C., 1997. Character-
ization of a 15-lipoxygenase in human breast carcinoma BT-20
cells: stimulation of 13-HODE formation by Tg␣/EGF. Biochem.
Biophys. Res. Commun. 231, 111–116.
Schneider, C., Tallman, K.A., Porter, N.A., Brash, A.R., 2001. Two
distinct pathways of formation of 4-hydroxynonenal. Mech-
anisms of nonenzymatic transformation of the 9- and 13-
hydroperoxides of linoleic acid to 4-hydroxyalkenals. J. Biol.
Chem. 276, 20831–20838.
Sessa, D.J., Gardner, H.W., Kleiman, R., Weisleder, D., 1977. Oxy-
genated fatty acid constituents of soybean phosphatidylcholines.
Lipids 12, 613–619.
Simon, T.C., Makheja, A.N., Bailey, J.M., 1989. The induced lipoxy-
genase in atherosclerotic aorta converts linoleic acid to the
platelet chemorepellant factor, 13-HODE. Thromb. Res. 55,
171–178.
Soberman, R.J., Harper, T.W., Betteridge, D., Lewis, R.A.,
Austin, K.F., 1985. Characterization and separation of
the arachidonic acid 5-lipoxygenase and linoleic acid 6-
lipoxygenase (arachidonic acid 15-lipoxygenase) of human
polymorphonuclear leukocytes. J. Biol. Chem. 260, 4508–
4515.
Spiteller, G., 1998. Linoleic acid peroxidation—the dominant lipid
peroxidation process in low density lipoprotein-and its rela-
tionship to chronic diseases. Chem. Phys. Lipids 95, 105–
162.
Jira, W., Spiteller, G., Richter, A., 1997. Increased levels of lipid ox-
idation products in low density lipoproteins of patients suffering
from rheumatoid arthritis. Chem. Phys. Lipids 87, 81–89.
Ku¨hn, H., 1996. Biosynthesis, metabolization and biological im-
portance of the primary 15-lipoxygenase metabolites 15-
hydro(pero)xy-5Z,8Z,11Z,13E-eicosatetraenoic acid and 13-
hydro(pero)xy-9Z,11E-octadecadienoic acid. Prog. Lipid Res.
35, 203–226.
Kuklev, D.V., Christie, W.W., Durand, T., Rossi, J.C., Vidal, J.P.,
Kasyanov, S.P., Akulin, V.N., Bezuglov, V.V., 1997. Synthesis of
keto- and hydroxydienoic compounds from linoleic acid. Chem.
Phys. Lipids 85, 125–134.
Marx, N., Bourcier, T., Sukhova, G.K., Libby, P., Plutzky, J., 1999.
PPAR␥ activation in human endothelial cells increases plasmino-
gen activator inhibitor type-1 expression-PPAR␥ as a potential
mediator in vascular disease. Arterioscler. Thromb. Vasc. Biol.
19, 546–551.
Minisci, F., Recupero, F., Cecchetto, A., Gambarotti, C., Punta, C.,
Faletti, R., Paganelli, R., Pedulli, G.F., 2004. Mechanisms of the
aerobic oxidation of alcohols to aldehydes and ketones, catalysed
under mild conditions by persistent and non-persistent nitroxyl
radicals and transition metal salts—polar, enthalpic, and captoda-
tive effects. Eur. J. Org. Chem., 109–119.
Minisci, F., Recupero, F., Pedulli, G.F., Lucarelli, M., 2003. Transi-
tion metal salts catalysis in the aerobic oxidation of organic com-
pounds. Thermochemical and kinetic aspects and new synthetic
developments in the presence of N-hydroxy-derivative catalysts.
J. Mol. Catal. A 204/205, 63–90.
Mlakar, A., Spiteller, G., 1996. Dihydroxy and hydroxyoxo fatty
acids as products of nonenzymic lipid peroxidation of polyun-
saturated fatty acids. Chem. Phys. Lipids 82, 25–32.
Nagy, L., Tontonoz, P., Alvarez, J.G.A., Chen, H., Evans, R.M., 1998.
Oxidized LDL regulates macrophage gene expression through
ligand activation of PPAR␥. Cell 93, 229–240.
Spiteller, G., 2001. Peroxidation of linoleic acid and its relation
to aging and age dependent diseases. Mech. Ageing Dev. 122,
617–657.
Spiteller, P., Spiteller, G., 1997. 9-Hydroxy-10,12-octadecadienoic
acid (9-HODE) and 13-hydroxy-9,11-octadecadienoic acid (13-
HODE): excellent markers for lipid peroxidation. Chem. Phys.
Lipids 89, 131–139.
Spiteller, P., Spiteller, G., 1998. Strong dependence of the lipid
peroxidation product spectrum whether Fe2+/O2 or Fe3+/O2
is used as oxidant. Biochim. Biophys. Acta 1392, 23–
40.