Regio- and stereoselective oxidation of linoleic acid bound to serum albumin: Identification by ESI-mass spectrometry and NMR of the oxidation products

Article abstract of DOI:10.1016/j.chemphyslip.2005.08.003

An efficient RP-HPLC method was developed for the detection of the oxidation products derived from the AAPH-initiated peroxidation of linoleic acid bound to human serum albumin. Diode array UV-detection allowed the quantification at 234 nm of four regioisomeric hydroperoxyoctadecadienoic acids (HPODE) and four hydroxyoctadecadienoic acids (HODE) while at 280 nm four oxooctadecadienoic acid isomers (KODE) were detected. Full identification of the different underivatized HODE, HPODE and KODE isomers was achieved by negative ESI-mass spectrometry outlining common fragmentation pathways for 9- and 13-regioisomers. Chemical synthesis of 9-(E,Z)-, 9-(E,E)-, 13-(Z,E)- and 13-(E,E)-KODE helped to their structural characterization by ¹H NMR. Lipid peroxidation in the presence of albumin proved to be regioselective with a larger accumulation of 13-HPODE and 9-KODE isomers. Thermodynamically more stable E,E-stereoisomers were also favored by albumin for both HPODE and KODE.

Full text of DOI:10.1016/j.chemphyslip.2005.08.003

Chemistry and Physics of Lipids 138 (2005) 60–68
Regio- and stereoselective oxidation of linoleic acid bound to
serum albumin: identification by ESI–mass spectrometry
and NMR of the oxidation products
∗
Claire Dufour , Mich e` le Loonis
Institut National de la Recherche Agronomique, UMR A408, Safety and Quality of Plant Products,
Agroparc, 84914 Avignon Cedex 9, France
Received 24 May 2005; received in revised form 5 August 2005; accepted 17 August 2005
Available online 12 September 2005
Abstract
An efficient RP-HPLC method was developed for the detection of the oxidation products derived from the AAPH-initiated
peroxidation of linoleic acid bound to human serum albumin. Diode array UV-detection allowed the quantification at 234 nm of
four regioisomeric hydroperoxyoctadecadienoic acids (HPODE) and four hydroxyoctadecadienoic acids (HODE) while at 280 nm
four oxooctadecadienoic acid isomers (KODE) were detected. Full identification of the different underivatized HODE, HPODE
and KODE isomers was achieved by negative ESI–mass spectrometry outlining common fragmentation pathways for 9- and 13-
regioisomers. Chemical synthesis of 9-(E,Z)-, 9-(E,E)-, 13-(Z,E)- and 13-(E,E)-KODE helped to their structural characterization by
1
H NMR. Lipid peroxidation in the presence of albumin proved to be regioselective with a larger accumulation of 13-HPODE and
9
©
-KODE isomers. Thermodynamically more stable E,E-stereoisomers were also favored by albumin for both HPODE and KODE.
2005 Elsevier Ireland Ltd. All rights reserved.
Keywords: Hydroperoxyoctadecadienoic acid; Oxooctadecadienoic acid; Hydroxyoctadecadienoic acid; Lipid peroxidation; Electrospray
ionization–mass spectrometry; NMR
1
. Introduction
involvement in eicosanoid synthesis or cell maturation
Kuhn and Borchert, 2002). Lipoxygenase-catalyzed
(
For a long time, lipid peroxidation has been consid-
lipid peroxidation which can be up- and downregulated
is well adapted to the above biological requirements
whereas lipid autoxidation resulting from pathological
events may escape metabolic control leading thus to
harmful side effects.
ered as a deleterious process leading to atherogenic LDL
or membrane disruption. However, when cell-mediated,
lipid peroxidation is a normal physiological process with
Although both reactions with linoleic acid lead
to hydroperoxyoctadecadienoic acids as primary oxi-
dation products, only lipoxygenases (LOX) display
positional and stereospecificities. For example, soy-
bean LOX catalyzes the conversion of linoleic acid
into 13(S)-hydroperoxy-(9Z,11E)-octadecadienoic acid
Abbreviations: HODE, hydroxyoctadecadienoic acid; HPODE,
hydroperoxyoctadecadienoic acid; KODE, oxooctadecadienoic acid;
LH, linoleic acid
Corresponding author. Tel.: +33 4 32 72 25 15;
fax: +33 4 32 72 24 92.
∗
E-mail address: cdufour@avignon.inra.fr (C. Dufour).
0
009-3084/$ – see front matter © 2005 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.chemphyslip.2005.08.003

C. Dufour, M. Loonis / Chemistry and Physics of Lipids 138 (2005) 60–68
61
while tomato or corn seed LOX preferentially pro-
duce 9(S)-hydroperoxy-(10E,12Z)-octadecadienoic acid
Hamberg, 1971; Matthew et al., 1997; Gardner and
2. Experimental
(
2.1. Chemicals
Grove, 2001). In autoxidation, however, both EZ and EE
stereoisomers are formed with the double bond closer to
the oxygenated center being E (Chan and Levett, 1977).
Usually, these unstable hydroperoxides (HPODE) decay
to secondary oxidation products, identified mainly as
hydroxy- (HODE), epoxyhydroxy- and oxo-derivatives
Fatty acid-free human serum albumin (HSA, A
1887), linoleic acid (LH), 13(S)-hydroperoxy-(9Z,11E)-
octadecadienoic
acid,
13(S)-hydroxy-(9Z,11E)-
ꢀ
octadecadienoic acid, 2,2 -azobis(2-amidinopropane)
hydrochloride (AAPH) and sodium dihydrogenphos-
phate were used as received from Sigma–Aldrich (St.
Quentin Fallavier, France). Phosphate buffer (pH 7.4,
50 mM NaH2PO4, 100 mM NaCl) was prepared in
Millipore Q-Plus water and stirred with a chelating
resin (Chelex 100, 0.4 mequiv./mM, BioRad). MeOH
and CH3CN were of HPLC grade.
(KODE) and dependent on the presence of factors such
as metal ions or reducing agents (Spiteller and Spiteller,
1
998; Dix and Marnett, 1985).
The separation of the linoleic acid oxidation prod-
ucts is usually challenged by chromatographic means.
Normal-phase HPLC has been widely used for the ten-
tative separation of either native HODE, HPODE and
KODE (Wu et al., 1995) or the corresponding methyl
esters derivatives (Hilbers et al., 1996). Reverse-phase
HPLC may appear less efficient (P e´ rez Gilabert and Gar-
cia Carmona, 2002; Banni et al., 1996).
2.2. Lipid peroxidation
Linoleic acid (100 ␮L, 200 mM in MeOH) was
added via syringe to albumin in buffer (10 mL to
33 g/L) followed by solid AAPH (82 mg). The reac-
Structural informations on the peroxidation prod-
ucts such as mass data and position of the oxy-
genated center are gained by mass spectrometry. Gas
chromatography–mass spectrometry (GC–MS) usually
requires derivatization of the polar lipid derivatives
before injection (Spiteller and Spiteller, 2000). Further-
more, analysis of intact hydroperoxides by GC–MS is
impeded by their thermal instability (P e´ rez Gilabert and
Garcia Carmona, 2002). Development of soft ioniza-
tion techniques such as electrospray has allowed the
detection of molecular ions of native oxidation prod-
ucts (Iwase et al., 1998). The state of the art in this area
is brought by tandem mass spectrometry as shown by
applications to fatty acid oxygenation products (Bylund
et al., 1998; Kerwin and Torvik, 1996; MacMillan and
Murphy, 1995; Schneider et al., 1997). However, such
an equipment is not routinely available in laboratories
supporting the development of simpler but as efficient
methods in LC–MS.
Our interest in the inhibition of plasma lipid perox-
idation by bioavailable secondary metabolites led us to
study a biologically-relevant plasma model chiefly con-
stituted by linoleic acid and albumin. A reverse-phase
HPLC method was thus developed to quantify simul-
taneously the resulting lipid oxidation products and the
antioxidants. Full identification of the different underiva-
tized HODE, HPODE and KODE isomers was achieved
by mass spectrometry in combination with electrospray
ionization. Assignment was further assisted by chemi-
cal synthesis and NMR analysis of the different KODE.
The regio- and stereoselectivities for the peroxidation of
HSA-bound linoleic acid were also established.
◦
tion mixture was stirred in an oven at 37 C. Aliquots
(0.5 mL) were removed, acidified with an equal vol-
ume of 0.05 N HCl and extracted three times with
1 mL of ethyl acetate/hexane (1:1). The organic phase
was washed with a saturated NaCl solution, dried over
Na2SO4, concentrated under N2 and taken up in ace-
tone before freezing. The extraction yields were eval-
uated by spiking directly albumin with linoleic acid
(66%), HODE and HPODE (84%) and KODE (88%)
(n = 3).
2.3. HPLC and HPLC/MS analyses of lipid
peroxidation products
Lipid peroxidation products were separated by
RP-HPLC coupled to Diode Array Detection (Hewlett-
Packard 1100) using a Alltima C18 column (5 ␮m,
150 mm × 4.6 mm, Alltech) equipped with a Alltima
C18 guard column (5 ␮m) (flow rate = 1 mL/min,
◦
T = 35 C, volume injected = 10 ␮L). The solvent
system was a gradient of A (0.05% HCOOH in H2O)
and B (CH3CN) with 50% B from 0 to 5 min, 60%
B at 15 min, 73% B at 30 min and 100% B at 40 min.
HPODE and HODE were detected at 234 nm, KODE
at 280 nm and LH at 210 nm. Structure determina-
tion was performed by HPLC–MS (Plateform LCZ,
Micromass, coupled to a HP 1050, see HPLC analysis
for the conditions of elution) in the negative electro-
spray ionization mode (cone voltage = 30 and 50 V,
nitrogen flow = 3.2 L/h, source block and desolvation
◦
temperatures = 150 and 250 C, respectively, flow rate

6
2
C. Dufour, M. Loonis / Chemistry and Physics of Lipids 138 (2005) 60–68
into the source = 50 ␮L/min). Authentic standards of
1
1
3(S)-hydroperoxy-(9Z,11E)-octadecadienoic acid and
3(S)-hydroxy-(9Z,11E)-octadecadienoic acid as well
as synthetic KODE samples were used for co-injection
and calibration curves (five point measurements)
of total, respectively, HPODE, HODE and KODE.
Micromolar solutions gave areas of 11 for HPODE and
HODE, 2 for KODE and 2 for LH (n = 3 and r2 > 0.999).
2
.4. Ketodiene synthesis and NMR analysis
From HODE. Neat linoleic acid (1.19 g) was left
to autoxidize at room temperature for a week. The
mixture containing LH and the HPODE in ethanol
Fig. 1. HPLC chromatogram after 7 h of peroxidation for linoleic acid
bound to human serum albumin.
(
35 mL) was reduced with sodium borohydride (106 mg)
◦
3. Results
at 0 C, hydrolyzed with water (35 mL), acidified to
pH 3 with 1 N HCl and extracted three times with
hexane/ethyl acetate (1:1). The organic extracts were
washed with sat. NaCl, dried over Na2SO4 and con-
centrated in vacuo. Silica gel chromatography (eluent:
ethyl acetate/hexane 2:8) yielded four fractions enriched
in the different HODE (156 mg, 12.5% overall yield)
and used for mass analysis. Oxidation of the respective
fractions by pyridinium dichromate (PDC) helped to the
chromatographic assignment of the HODE. For exam-
ple, HODE3 (28 mg) was oxidized with PDC (50 mg)
The HPLC chromatogram obtained for LH autoxida-
tion in the 18–24 min area is qualitatively similar to the
one obtained for chemical oxidation of HSA-bound LH
except that HPODE are major compounds in autoxida-
tion (Fig. 1). Chemical synthesis of HODE and KODE
from HPODE produced by autoxidation was used to
assess the structures of the oxidation products obtained
in the presence of HSA. Solvent and gradient condi-
tions in HPLC were refined until well-separated peaks
were obtained for each product. Each oxidation prod-
uct is numbered according to the LC peak from which it
originates.
◦
in chloroform (4 mL) at 0 C. PDC was precipitated
with Et2O (8 mL) and the solution eluted over sil-
ica gel (ethyl acetate/Et2O 6:4) to yield quantitatively
KODE4.
From HPODE. Pure HPODE were obtained by
autoxidation of neat linoleic acid followed by gel
chromatography on silica gel (ethyl acetate/hexane,
3.1. Structures of the HPODE products
Mass analysis was conducted at low and high voltages
in order to get either the molecular ion or full fragmen-
tation for structural assignment. Product ions and abun-
dances are reported for the latter conditions in Table 1.
Although no sodium-based chemical was used for
HPODE purification and HPLC analysis, ions detected
at m/z 333 and 315 suggest the presence of sodium in
5
:95). The HPODE (25 mg) were then treated with
pyridine (1.75 mL) and acetyl chloride (63.5 ␮L) to
yield directly the corresponding KODE (16.7 mg, 71%
yield) (Porter and Wujek, 1987). The fraction con-
taining the KODE was analyzed in HPLC/MS and
next purified by semi-preparative HPLC (Waters 600)
coupled to a UV detection (Waters 486) using a
Uptisphere ODB column (10 ␮m, 250 mm × 21.2 mm)
−
−
the parent ions [M + Na − 2H] and the dehydrated par-
ent ions [M + Na − H2O − 2H] . Abundant ions at m/z
equipped with a Uptisphere ODB guard column (5 ␮m,
293 resulted from the loss of water and proved to be
major for HPODE2 and 4. The expected loss of H2O2
did not appear as a degradation pathway and decarboxy-
lation of the fatty acids was the next step. Other cleav-
ages were characteristic of the 9- and 13-oxygenation
of the primary oxidation products. Similar patterns in
the product ion spectra were displayed for HPODE1 and
3. The product ion at m/z 223 resulted from cleavage
between C13 andC14. Subsequent losses of CO andCO2
yielded ions at m/z 195 and 179 while the most abundant
ion was detected at m/z 113. HPODE1 and 3 are thus
◦
3
3 mm × 21.2 mm) (flow rate = 18 mL/min, T = 35 C,
volume injected = 500 ␮L). The solvent system was
a gradient of A (0.05% HCOOH in H2O) and B
(CH3CN) with 65% B from 0 to 5 min, 70% B at
1
3
0 min, 73% B from 15 to 25 min and 80% B at
5 min. Four fractions were obtained containing each
1
a major KODE (purity 85–95%). H NMR spectra were
obtained with a cryoprobe on a Bruker DRX-500 in
CDCl3 (300 K) with CHCl3 as the internal reference at
7
.26 ppm.

C. Dufour, M. Loonis / Chemistry and Physics of Lipids 138 (2005) 60–68
63
Table 1
Product ions (m/z) and proposed structures for fragments issued from hydroperoxyoctadecadienoic acids (HPODE) in ESI–MSa
HPODE1
3-Z,E-HPODE
HPODE2
9-E,Z-HPODE
HPODE3
13-E,E-HPODE
HPODE4
9-E,E-HPODE
1
−
[
[
[
[
[
[
[
[
[
[
M + Na − 2H]
333 (2)
315 (8)
311 (1)
293 (55)
249 (8)
233 (4)
223 (5)
195 (22)
179 (11)
167 (14)
333 (4)
315 (12)
311 (3)
293 (100)
249 (9)
233 (3)
333 (5)
315 (12)
311 (3)
293 (52)
249 (3)
233 (2)
223 (39)
195 (44)
179 (18)
167 (15)
139 (14)
113 (100)
333 (5)
315 (8)
311 (2)
293 (100)
249 (7)
233 (2)
−
M + Na − H2O − 2H]
−
M − H]
−
M − H2O − H]
−
M − H2O − CO2 − H]
−
M − H2O2 − CO2 − H]
M − CH3(CH2)2CHCH2 − H2O − H]
−
−
−
M − CH3(CH2)2CHCH2 − H2O − CO − H]
M − CH3(CH2)2CHCH2 − H2O − CO2 − H]
M − CH3CH2(CH2)5COOH − H]−
167 (7)
139 (6)
167 (14)
139 (5)
1
1
39 (14)
13 (100)
1
1
97 (22)
85 (75)
197 (25)
185 (72)
171 (7)
149 (14)
141 (11)
125 (49)
123 (15)
[
M − CH3(CH2)4CHCHCCH − H2O − H]−
171 (8)
49 (17)
1
[
[
[
M − CH3(CH2)4CHCHCCH − H2O − H2CO − H]−
M − CH3(CH2)4CHCHCCH − H2O − HCO2H − H]
CH3(CH2)4CHCHCHCH2 − H]−
141 (15)
125 (51)
123 (20)
−
a
Abundances for cone voltage −50 V are given in parentheses.
1
3-regioisomers. Their maximal absorption wavelengths
in diode array detection were found to be, respectively,
36 and 232 nm. Last, the co-elution of HPODE1 with
octadecadienoic acid while HPODE4 (λmax 232 nm) is
9-hydroperoxy-(10E,12E)-octadecadienoic acid.
2
an authentic standard of 13(S)-hydroperoxy-(9Z,11E)-
3.2. Structures of the HODE products
octadecadienoic acid allowed to assign HPODE3 as
1
3-hydroperoxy-(9E,11E)-octadecadienoic acid. In the
The separation of the 4 HODE could not be fully
achieved in the selected chromatographic system. Peaks
1 and 3 could be assigned by mass spectrometry
(Table 2). In addition, peak 1 co-eluted with an authen-
tic standard of 13(S)-hydroxy-(9Z,11E)-octadecadienoic
acid. Mass spectra of peaks 1 and 3 revealed parent ions
at m/z 295 for [M − H] and 317 for [M + Na − 2H] .
Dehydration led to m/z 277. Characteristic fragments at
m/z 171 and 123 were obtained from scission between
case of HPODE2 and 4, characteristic product ions were
detected at m/z 197, 185 and 171. The latter ion resulted
from the fission between C9 and C10 and subsequent
elimination of water. Decarbonylation and decarboxyla-
tion led to the production of typical ions at m/z 141 and
−
−
1
25. These 9-regioisomers were further assigned based
on their UV maximal absorption wavelengths. HPODE2
λmax 236 nm) appears as 9-hydroperoxy-(10E,12Z)-
(
Table 2
a
Product ions (m/z) and proposed structures for fragments issued from hydroxyoctadecadienoic acids (HODE) in ESI–MS
HODE1
HODE3
13-Z,E-HODE
9-E,E-HODE
−
[
[
[
[
[
[
[
[
[
M + Na − 2H]
317 (2)
295 (100)
277 (42)
251 (2)
233 (1)
195 (77)
179 (6)
317 (1)
−
M − H]
295 (56)
277 (100)
251 (1)
−
−
M − H2O − H]
M − CO2 − H]
M − H2O − CO2 − H]
−
233 (2)
−
M − CH3(CH2)4CHO − H]
−
M − CH3(CH2)2CHCH2 − HCO2H − H]
−
M − CH3(CH2)4CHCHCHCH2 − H]
171 (100)
123 (10)
M − HCO(CH2)7CO2H − H]−
113 (13)
a
Abundances for cone voltage −50 V are given in parentheses.

6
4
C. Dufour, M. Loonis / Chemistry and Physics of Lipids 138 (2005) 60–68
Table 3
1
H NMR spectral data for regioisomeric 9- and 13-oxooctadecadienoic acidsa
Compound
H␣
H␤
H␦
H␥
CH2–CO
CH2–C C
1
1
9
9
3-Z,E-KODE 6.19 d (15.2) 7.50 dd (11.5, 15.2)
6.12 dd (11.5, 10.4) 5.91 dt (8.3, 10.4) 2.54 t (7.2), 2.35 t (7.5) 2.30 m
3-E,E-KODE 6.08 d (15.4) 7.13 dm (15.4)
6.14–6.18 m 6.14–6.18 m 2.53 t (7.4), 2.35 t (7.5) 2.17 m
6.16 d (15.3) 7.49 ddd (0.7, 11.4, 15.3) 6.12 dd (11.4, 10.9) 5.91 dt (8.0, 10.9) 2.55 t (7.4), 2.34 t (7.5) 2.30 m
6.07 d (15.5) 7.13 dm (15.5) 6.15–6.19 m 6.15–6.19 m 2.53 t (7.4), 2.33 t (7.5) 2.17 m
-E,Z-KODE
-E,E-KODE
aChemical shift in ppm, multiplicity (coupling constant in Hz). Numbering is as follows: CO–CH␣ CH␤–CH␦ CH␥. Common chemical shifts as
follows: 1.63 (m, 4H), 1.43 (m, 2H), 1.32 (m, 10H), 0.88 (t, 3H, 7.0 Hz).
C9 and C10 for peak 3. Presence of a maximal absorption
at 232 nm led to the assignment of peak 3 as 9-hydroxy-
comparedtoKODE1. Chemicalshiftsareconsistentwith
those published previously (Bull and Bronstein, 1990).
KODE3 presented a UV max at 280 nm. Four well
(
10E,12E)-octadecadienoic acid. Typical ions displayed
by 13-hydroxy-(9Z,11E)-octadecadienoic acid (λmax
34 nm) are m/z 195, 179 and 113. Ion at m/z 195 may
1
resolved signals were observed in the H NMR ethylenic
2
region. The most deshielded proton H11 appeared as a
ddd at 7.49 ppm with a large J value of 15.3 Hz indicating
a trans configuration for the double bond. At 5.91 ppm,
a dt (J = 8.0, 10.9 Hz) was expected for H13 pointing
to a cis geometry for the second unsaturation. A d at
6.16 ppm (J = 15.3 Hz, H10) and a dd (J = 11.4, 10.9 Hz,
H12) confirmed the assignment of KODE3 as 9-oxo-
(10E,12Z)-octadecadienoic acid (Iacazio, 2003; Kuklev
et al., 1997).
result from bond cleavage between C12 and C13 while
cleavage between C13 and C14 followed by a deformy-
lation may lead to m/z 179.
Peak 2 is indeed a mixture of 13-hydroxy-(9E,11E)-
octadecadienoic acid (λmax 232 nm) and 9-hydroxy-
(10E,12Z)-octadecadienoic acid (λmax 234 nm) as
revealed from UV and mass probing.
3
.3. Structures of the KODE products
KODE1 exhibited a UV max at 280 nm. H NMR
The ethylenic region for KODE4 exhibited three dis-
tinct patterns: a dm at 7.13 ppm, part of an ABMX
system (J = 15.5 Hz) and assigned to H11, a multiplet
at 6.15–6.19 ppm for H12 and H13 and a d (J = 15.5 Hz)
at 6.07 ppm for H10. Finally, a COSY revealed strong
correlations between H10 and H11, H11 and the multi-
plet common to H12 and H13 but no correlation between
H10andbothH12andH13. Additionalcorrelationswere
observed between the allylic methylene at 2.17 ppm and
both H12 and H13 and more weakly with H11. Last,
correlations were displayed between methylenes ␣ (2.33
and 2.53 ppm) and ␤ (1.63 ppm) to the different carbonyl
groups. With a maximal absorption at 276 nm, KODE4
corresponds to 9-oxo-(10E,12E)-octadecadienoic acid
(Kawagishi et al., 2002). Moreover, chemical shifts
reported in pyridine for the ethylenic protons of the four
KODE show a downfield move compared to those in
chloroform (Watanebe et al., 1999). Last, equilibration
of a mixture of the four KODE in the presence of iodine
led to isomerization to all-trans KODE2 and 4.
1
of the ethylenic region displayed a dd at 7.50 ppm with
J = 11.5, 15.2 Hz assigned to H11 (Table 3). At 5.91 ppm,
H9 appeared as a dt (J = 8.3, 10.4 Hz). At, respec-
tively, 6.19 and 6.12 ppm the splitting patterns were a
d (J = 15.2 Hz, H12) and a dd (J = 11.5, 10.4 Hz, H10).
Coupling constants of 10 and 15 Hz pointed to cis and
trans geometries for the two ethylenic moieties suggest-
ing that KODE1 is 13-oxo-(9Z,11E)-octadecadienoic
acid. These results are in agreement with chemical
shifts (Napolitano et al., 2002) and coupling con-
stants obtained for 13-oxo-(9Z,11E)-octadecadienoic
acid(Iacazio, 2003)orforthecorrespondingmethylester
(Kuklev et al., 1997; Hidalgo et al., 1992). Discrepan-
cies are observed with other literature results (Chudinova
et al., 1995; Dong et al., 2000). In particular, the lack of
use of multiplicity led to inversion in proton assignments
(Bull and Bronstein, 1990; Blackburn et al., 1997).
With KODE2, the most downfield proton at 7.13 ppm
All KODE presented common fragments in their
negative ESI mass spectra (Table 4). Parent ions were
detected at m/z 293. A low abundance of sodium adducts
was also observed at m/z 315. Decarboxylation led
to m/z 249. Furthermore, 13-regioisomers displayed
characteristic fragments at m/z 195 and 179 resulting
from bond scission between C13 and C14 followed by
either decarbonylation or decarboxylation. Successive
is part of an ABMX spin system from which a coupling
constant of 15.4 Hz could be extracted. It was assigned
to H11. A d (J = 15.4 Hz) was observed for the other
ethylenic proton assigned as H12. The configuration of
the double bond alpha to the carbonyl group is thus trans.
KODE2isascribedto13-oxo-(9E,11E)-octadecadienoic
acid owing to a shorter absorption maximum (276 nm)

C. Dufour, M. Loonis / Chemistry and Physics of Lipids 138 (2005) 60–68
65
Table 4
Product ions (m/z) and proposed structures for fragments issued from oxooctadecadienoic acids (KODE) in ESI–MSa
KODE1
3-Z,E-KODE
KODE2
13-E,E-KODE
KODE3
9-E,Z-KODE
KODE4
9-E,E-KODE
1
−
[
[
[
[
M − H]
293 (100)
249 (8)
293 (100)
249 (8)
293 (100)
249 (3)
197 (11)
293 (100)
249 (3)
197 (9)
−
M − CO2 − H]
−
M − CH3(CH2)4CHCH2 − H]
M − CH3(CH2)2CHCH2 − CO − H]−
195 (16)
179 (13)
167 (13)
139 (6)
195 (10)
179 (5)
167 (12)
139 (3)
1
1
85 (37)
49 (5)
185 (37)
149 (5)
141 (6)
[
M − CH3(CH2)2CHCH2 − CO2 − H]−
−
[
[
[
[
[
M − CH3(CH2)2CHCH2 − CO − CH2CH2 − H]
−
M − CH3(CH2)4(CHCH)2CHO − H]
141 (5)
−
M − CH3(CH2)2CHCH2 − CO − 2xCH2CH2 − H]
−
M − CH3(CH2)4CHCHCHCH2 − CO2 − H]
125 (20)
123 (5)
125 (18)
123 (4)
CH3(CH2)4CHCHCHCH2 − H]−
113 (82)
113 (76)
113 (76)
113 (82)
a
Abundances for cone voltage −50 V are given in parentheses.
fragmentations of ion 195 led to characteristic ions at
m/z 167, 139 and 113 resulting from the loss of ethylene
or CO. Typical ions for 9-ketodienes were detected at
m/z 197, 185, 149, 141, 125 and 123. Cleavage between
C11 and C12 may result in m/z 197 while cleavage
between C9 and C10 may produce m/z 123.
rates appeared to differ appreciably for the various iso-
mers with a predominant production of 9-(E,E)-KODE
(Fig. 4). In addition, higher levels in E,E-stereoisomers
were formed compared to the Z,E-ones. At the end of
the reaction the following order of accumulation was
observed: 9-(E,E)-KODE > 13-(E,E)-KODE > 13-(Z,E)-
KODE > 9-(E,Z)-KODE. With HODE, this analysis is
partly prevented owing to the lack of product separation.
The peak for 9-(E,E)-HODE was however integrated as
twice the area of the shoulder on the left side of the
peak for 13-(Z,E)-HPODE. Differences are again out-
lined: 13-(Z,E)-HODE was slightly formed whereas 9-
3
.4. Peroxidation of HSA-bound linoleic acid
The course of the peroxidation of HSA-bound linoleic
acid was followed by RP-HPLC (Fig. 1). The peroxida-
tion products HPODE, HODE and KODE as well as
remaining LH were quantified by peak area according to
the calibrations with 13(S)-(10Z,12E)-HPODE, 13(S)-
(E,Z)-HODE and 13-(E,E)-HODE accumulated almost
linearly (Fig. 5). The trend for 9-(E,E)-HODE should be
viewed with care because of a high similitude with that
observed for 13-(Z,E)-HPODE.
(10Z,12E)-HODE, a synthetic sample of KODE and LH.
During the first 7 h of reaction, HSA-bound linoleic acid
was mainly converted into the primary oxidation product
HPODE and secondary oxidation products HODE and
KODE (Fig. 2). Levels in KODE were found to be sub-
stantial whereas HODE appeared as minor degradation
products in the albumin system. Additionally, the mass
balance is consistent with HPLC chromatograms show-
ing no further degradation products during this period.
The values for the extraction yields and calibration fac-
tors are thus confirmed.
Major differences were displayed in the accumulation
of the four HPODE isomers (Fig. 3). Similar concentra-
tions in 13- and 9-(E,E)-HPODE were observed after
7
h of reaction whereas 13-(Z,E)- and 9-(E,Z)-HPODE
accumulated to a lesser extent. The trend observed
at 7 h was confirmed by the data collected between
2
4 and 30 h of reaction. The most intriguing behav-
Fig. 2. Accumulation of HPODE (ꢀ), KODE (ꢁ), HODE (ꢂ) during
ior was displayed by 9-(E,Z)-HPODE whose concen-
◦
the course of the peroxidation of HSA-bound linoleic acid (᭹) at 37 C.
tration peaked at 6 h and then decreased. For KODE,
Total lipids (ꢃ) are the sum of oxidized lipids and LH (n = 4).

6
6
C. Dufour, M. Loonis / Chemistry and Physics of Lipids 138 (2005) 60–68
Fig. 3. Formation of the different HPODE regioisomers during the
course of the peroxidation of HSA-bound linoleic acid: 13-(E,E)-
HPODE (ꢂ), 13-(Z,E)-HPODE (ꢁ), 9-(E,E)-HPODE (᭹) and 9-(E,Z)-
HPODE (ꢀ).
Fig. 5. Formation of the different HODE regioisomers during the
course of the peroxidation of HSA-bound linoleic acid: 9-(E,Z)-
HODE + 13-(E,E)-HODE (ꢀ), 9-(E,E)-HODE (ꢂ) and 13-(Z,E)-
HODE (᭹).
for all 13-regioisomers while 9-regioisomers share a
fragment ion at m/z 123. Furthermore, almost identical
spectra to those obtained in this work were obtained by
collision-induced decomposition of the molecular ions
for 9- and 13-HPODE in the negative mode (MacMillan
andMurphy, 1995). Thisemphasizesthatroutinelyavail-
able LC–ESI–MS can be used for the characterization of
theseoxygenatedproducts. Characteristicionsatm/z113
from 13-regioisomers and m/z 185 from 9-regioisomers
appear to result from an apparent cleavage of the double
bond adjacent to the oxygenated center. A mechanism
involving a 1,5 sigmatropic proton shift was suggested
for the formation of these fragments (MacMillan and
Murphy, 1995).
Attempts in positive ESI–MS proved to be less suc-
cessful. At low voltage, no single ion prevails. The
parent ions mainly appear as sodium and/or acetoni-
trile adducts coexisting with the dehydrated forms,
Fig. 4. Formation of the different KODE regioisomers during the
course of the peroxidation of HSA-bound linoleic acid. 9-(E,E)-KODE
(
(
ꢁ), 13-(E,E)-KODE (ꢀ), 13-(Z,E)-KODE (᭹) and 9-(E,Z)-KODE
ꢂ).
+
respectively, [M − H2O + H] for HODE and KODE and
+
4
. Discussion
[M − H2O2 + H] for HPODE. At higher voltages, an
elevated proportion of triply unsaturated aliphatic frag-
ments was detected indicating easy cleavages of the ions
in these conditions. The use of ammonium acetate in
the HPLC solvent system did not improve the response
in the KODE mass. Characteristic ions of the 9- or
13-regiochemistry were also found although in a lower
abundance than in the negative mode (Schneider et al.,
1997). As a matter of fact, the negative mode in ESI–MS
is the most valuable as it provides clean molecular ions at
low voltages as well as several characteristic fragments
of the 9- and 13-oxygenated products at higher voltages.
Negative ESI–MS provided sufficient information for
identification of classes of oxidation products in the per-
oxidation of HSA-bound linoleic acid. At −20 V, parent
ions were mainly obtained for KODE and HODE while
HPODEdisplayedadehydratedformoftheparention. In
addition, negative ESI–MS proved to be a valuable tool
for determining the position of oxygenation in the 9- and
1
3-regioisomers of the oxidation products derived from
linoleic acid. At the higher voltage of −50 V, common
characteristic ions at m/z 195, 179 and 113 are revealed

C. Dufour, M. Loonis / Chemistry and Physics of Lipids 138 (2005) 60–68
67
We also report a full characterization of the differ-
ent KODE isomers in the same H NMR conditions.
ages in the accumulated primary or secondary oxidation
products.
1
Identical spectra were obtained for both E,E-isomers, on
one hand, and Z,E-isomers on the other hand, reflecting
the isolation of the structural moiety in the hydrocarbon
chain. The main differences were linked to the stere-
ochemistry of the double bond ␦ to the keto group.
Chemical shifts for olefinic and allylic protons were
thus found to be characteristic of the E,E- and Z,E-
stereochemistries.
In a homogeneous system, lipid peroxidation yields
a 13-HPODE/9-HPODE ratio of one. In the presence
of albumin, this ratio was found to increase steadily
to end up at 1.8 after 30 h of reaction. The 13- to 9-
regioselectivity induced by HSA probably results from
the location of the linoleic acid molecules in their tunnel-
like binding sites. ESR studies have shown that the
tightest restriction in the binding channel is between C5
and C13–15 (Peters, 1996). Because the dienyl moiety
is located in this area, steric hindrance may favor the
addition of oxygen on the 13-position rather than the 9-
position of the pentadienyl radical issued from linoleic
acid (L + O2 → LOO ). In contrast, a 9-regioselectivity
was observed for the KODE formation in the presence
of HSA. The lower stability of the 9-HPODE may result
in an accelerated conversion to the 9-KODE regioiso-
mers. The presence of HSA during lipid peroxidation
favors the formation of thermodynamically more sta-
ble E,E-stereoisomers. The reversible addition of oxy-
gen onto the pentadienyl radical may be controlled
by HSA, thus favoring the E,E-stereoisomers. More-
over, the conversion of HPODE to KODE (formally
a dehydration) may involve a double-bond rearrange-
ment to produce the thermodynamically more stable
E,E-KODE.
Long chain fatty acids (LCFA, C16–C20) are cru-
cial intermediates in the lipid metabolism. The non-
esterified forms of fatty acids are not soluble in plasma
and are thus transported by albumin (Peters, 1996). Up
to seven binding sites for LCFA have been located in
•
•
7
8
−1
human albumin with affinities in the 10 to 10 M
range (Petitpas et al., 2001). Because the total concen-
tration of LCFA is just below 1 mM in plasma and that
of albumin is about 0.6 mM, circulating albumin typ-
ically carries 1–2 LCFA molecules. This number may
rise to about 4 LCFA per HSA after strenuous exer-
cise or other adrenergic stimulation (Peters, 1996). In
our study, HSA was bound with four molecules of
linoleic acid and placed next in oxidizing conditions.
The thermal decomposition of lipid peroxidation initia-
tor AAPH delivers peroxyl radicals at a constant rate
(Niki, 1990). The accumulation in time of the peroxi-
dation products as well as the consumption of LH was
followed by reverse-phase HPLC. Although the full
separation of the primary and secondary products of
linoleic acid cannot be achieved in this chromatographic
system, their quantification remains feasible. Absorp-
tion of hydroxy- and hydroperoxyoctadecadienoic acids
extends from 210 to 260 nm while that of the correspond-
ing ketodiene is in the range 250–310 nm. Quantitative
analysis was performed for concentrations 1–250 ␮M
in a single oxidation product without any undesirable
overlap for detection at 234 and 280 nm. During the
first hour of reaction, the levels in peroxidation products
remain low suggesting an antioxidant activity of albu-
min. When the protection of HSA-bound linoleic acid is
over, HPODE, HODE and KODE accumulate at a nearly
linear rate. After 7 h of reaction, measured concentra-
tions were found to be 251 ␮M for HPODE, 84 ␮M for
KODE and 37 ␮M for HODE. In the presence of albu-
min, hydroperoxides appear as the major peroxidation
products followed by ketodienes and finally hydroxides.
Beyond this time, the rate in HPODE levels off while the
secondary oxidation products keep accumulating sup-
porting the conversion of HPODE in HODE and KODE.
Furthermore, the HPLC profile displays the formation of
compoundswithshorterretentiontimesindicatingcleav-
5. Conclusion
An efficient RP-HPLC method was developed for the
detection of primary and secondary peroxidation prod-
ucts issued from oxidation of linoleic acid bound to
human serum albumin. Diode array detection allowed
the quantification of all the regioisomeric KODE and
HPODE as single compounds while the four HODE
isomers were resolved as three peaks. Full characteri-
zation by ESI–MS and NMR was conducted revealing a
13- versus 9-regioselectivity for the peroxidation step
in this plasma model. Furthermore, unstable 9-(E,Z)-
HPODE decayed rapidly to give major KODE isomer, 9-
(E,E)-KODE. The favored formation of thermodynam-
ically controlled E,E-stereoisomers was also observed
for both HPODE and KODE. Mechanisms involved in
the regio- and stereoselectivities of the peroxidation of
HSA-bound linoleic acid will be discussed more deeply
elsewhere.
Acknowledgement
We thank Dr. Robert Faure (Spectropole, Marseille,
France) for help in NMR determination.

6
8
C. Dufour, M. Loonis / Chemistry and Physics of Lipids 138 (2005) 60–68
References
Kawagishi, H., Miyazawa, T., Kume, H., Arimoto, Y., Inakuma, T.,
002. Aldehyde dehydrogenase inhibitors from the mushroom cli-
2
Banni, S., Contini, M.S., Angioni, E., Deiana, M., Dessi, M.A., Melis,
M.P., Carta, G., Corongiu, F.P., 1996. A novel approach to study
linoleic acid autoxidation: importance of simultaneous detection
of the substrate and its derivative oxidation products. Free Radic.
Res. 25, 43–53.
tocybe clavipes. J. Nat. Prod. 65, 1712–1714.
Kerwin, J.L., Torvik, J.J., 1996. Identification of monohydroxy fatty
acids by electrospray mass spectrometry and tandem mass spec-
trometry. Anal. Biochem. 237, 56–64.
Kuhn, H., Borchert, A., 2002. Regulation of enzymatic lipid peroxida-
tion: the interplay of peroxidizing and peroxide reducing enzymes.
Free Radic. Biol. Med. 33, 154–172.
Blackburn, M.L., Ketterer, B., Meyer, D.J., Juett, A.M., Bull, A.W.,
1
997. Characterization of the enzymatic and nonenzymatic reac-
tion of 13-oxooctadecadienoic acid with gluthatione. Chem. Res.
Toxicol. 10, 1364–1371.
Kuklev, D.V., Christie, W.W., Durand, T., Rossi, J.C., Vidal, J.P.,
Kasyanov, S.P., Akulin, V.N., Bezuglov, V.B., 1997. Synthesis of
keto- and hydroxydienoic compounds from linoleic. Chem. Phys.
Lipids 85, 125–134.
MacMillan, K.D., Murphy, R.C., 1995. Analysis of lipid hydroper-
oxides and long-chain conjugated keto acids by negative ion
electrospray mass spectrometry. J. Am. Soc. Mass Spectrom. 6,
1190–1201.
Matthew, M.A., Chan, H.W., Galliard, T., 1997. A simple method for
the preparation of pure 9-d-hydroperoxide of linoleic acid and
methyl linoleate based on the positional specificity of lipoxyge-
nase in tomato fruit. Lipids 12, 324–326.
Napolitano, A., Crescenzi, O., Camera, E., Giudicianni, I., Picardo, M.,
d’Ischia, M., 2002. The acid-promoted reaction of ethyl linoleate
with nitrite. New insights from N-15-labelling and peculiar reac-
tivity of a model skipped diene. Tetrahedron 58, 5061–5067.
Niki, E., 1990. Freeradicalinitiatorsassourceofwater-orlipid-soluble
peroxyl radicals. Meth. Enzymol. 186, 100–108.
Bull, A.W., Bronstein, J.C., 1990. Production of unsaturated car-
bonyl compounds during metabolism of hydroperoxy fatty acids
by colonic homogenates. Carcinogenesis 11, 1699–1704.
Bylund, J., Ericsson, J., Oliw, E.H., 1998. Analysis of cytochrome
P450 metabolites of arachidonic and linoleic acids by liquid
chromatography–mass spectrometry with ion trap MS2. Anal.
Biochem. 265, 55–68.
Chan, H.W.S., Levett, G., 1977. Autoxidation of methyl linoleate.
Separation and analysis of isomeric mixtures of methyl linoleate
hydroperoxides and methyl hydroxylinoleates. Lipids 12, 99–104.
Chudinova, V.V., Chudinov, M.V., Eremin, S.V., Alekseev, S.M., 1995.
1
Preparative separation and H NMR identification of products of
linoleic acid autoxidation. Bioorg. Khim. 21, 545–553.
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.
Dong, M., Oda, Y., Hirota, M., 2000. (10E,12Z,15Z)-9-
hydroxy-10,12,15-octadecatrienoic acid methyl ester as an
anti-inflammatory compound from Ehretia dicksonii. Biosci.
Biotechnol. Biochem. 64, 882–886.
Gardner, H.W., Grove, M.J., 2001. Method to produce 9(S)-
hydroperoxides of linoleic and linolenic acids by maize lipoxy-
genase. Lipids 36, 529–533.
P e´ rez Gilabert, M., Garcia Carmona, F., 2002. Chromatographic anal-
ysis of lipoxygenase products. Anal. Chim. Acta 465, 319–335.
Peters, T., 1996. All About Albumin. Academic Press, San Diego.
Petitpas, I., Gr u¨ ne, T., Bhattacharya, A.A., Curry, S., 2001. Crystal
structures of human serum albumin complexed with monoun-
saturated and polyunsaturated fatty acids. J. Mol. Biol. 314,
955–960.
Hamberg, M., 1971. Steric analysis of hydroperoxydes formed by
lipoxygenase oxygenation of linoleic acid. Anal. Biochem. 43,
Porter, N.A., Wujek, J.S., 1987. Allylic hydroperoxide rearrangement:
␤-scission or concerted pathway. J. Org. Chem. 52, 5085–5089.
Schneider, C., Schreier, P., Herderich, M., 1997. Analysis of
lipoxygenase-derived fatty acid hydroperoxides by electrospray
ionization tandem mass spectrometry. Lipids 32, 331–336.
Spiteller, P., Spiteller, G., 1998. Strong dependence of the lipid perox-
idation product spectrum whether Fe /O2 or Fe /O2 is used as
oxidant. Biochim. Biophys. Acta 1292, 23–40.
Spiteller, D., Spiteller, G., 2000. Oxidation of linoleic acid in LDL:
an important event in atherogenesis. Angew. Chem. Int. Ed. 39,
585–589.
5
15–526.
Hidalgo, F.J., Zamora, R., Vioque, E., 1992. Syntheses and reactions of
methyl (Z)-9,10-epoxy-13-oxo-(E)-11-octadecenoate and methyl
(
E)-9,10-epoxy-13-oxo-(Z)-11-octadecenoate. Chem. Phys. Lipids
0, 225–233.
Hilbers, M.P., Finazzi-Agr o` , A., Veldink, G.A., Vliegenthart, J.F.G.,
996. Purification and characterization of a lentil seedling lipoxy-
2
+
3+
6
1
genase expressed in E. coli: implications for the mechanism of
oxodiene formation by lipoxygenases. Int. J. Biochem. Cell Biol.
2
8, 751–760.
Watanebe, J., Kawabata, J., Kasai, T., 1999. 9-Oxooctadeca-10,12-
dienoic acids as acetyl-CoA carboxylase inhibitors from red pep-
per (Capsicum annuum L.). Biosci. Biotechnol. Biochem. 63,
489–493.
Iacazio, G., 2003. Easy access to various natural keto polyunsaturated
fatty acids and their corresponding racemic alcohols. Chem. Phys.
Lipids 125, 115–121.
Iwase, H., Takatori, T., Nagao, M., Nijima, H., Iwadate, K., Matsuda,
Y., Kobayashi, M., 1998. Formation of keto and hydroxy com-
pounds of linoleic acid in submitochondrial particles of bovine
heart. Free Radic. Biol. Med. 24, 1492–1503.
Wu, Z., Robinson, D.S., Domoney, C., Casey, R., 1995. High-
performance liquid chromatography analysis of the products of
linoleic acid oxidation catalyzed by pea (Pisum sativum) seed
lipoxygenases. J. Agric. Food Chem. 43, 337–342.