Lysophosphatidylethanolamine is - in contrast to - choline - generated under in vivo conditions exclusively by phospholipase A2 but not by hypochlorous acid

Article abstract of DOI:10.1016/j.bioorg.2009.09.002

Inflammatory liver diseases are associated with oxidative stress mediated particularly by neutrophilic granulocytes. At inflammatory loci, hypochlorous acid (HOCl) is generated by myeloperoxidase. HOCl reacts with a large variety of molecules and induces

Full text of DOI:10.1016/j.bioorg.2009.09.002

Bioorganic Chemistry 37 (2009) 202–210
Contents lists available at ScienceDirect
Bioorganic Chemistry
journal homepage: www.elsevier.com/locate/bioorg
Lysophosphatidylethanolamine is – in contrast to – choline – generated under
in vivo conditions exclusively by phospholipase A2 but not by hypochlorous acid
Celestina Schober a, Jürgen Schiller a, Franziska Pinker a, Jan G. Hengstler b, Beate Fuchs a,
*
a University of Leipzig, Medical Faculty, Institute of Medical Physics and Biophysics, Germany
b Leibniz Research Centre for Working Environment and Human Factors (IfADo), Dortmund, Germany
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 13 May 2009
Available online 12 September 2009
Inflammatory liver diseases are associated with oxidative stress mediated particularly by neutrophilic
granulocytes. At inflammatory loci, hypochlorous acid (HOCl) is generated by myeloperoxidase. HOCl
reacts with a large variety of molecules and induces (among other reactions) the formation of lysophos-
phatidylcholine (LPC) from polyunsaturated phosphatidylcholines (PC).
As liver tissue contains huge amounts of polyunsaturated PC species enhanced LPC concentrations are
detectable under these conditions. However, human liver contains also major amounts of polyunsatu-
rated phosphatidylethanolamine (PE). It is so far widely unknown, if PE oxidation by HOCl leads to the
generation of LPE in a similar way as observed in the case of PC. Using MALDI-TOF mass spectrometry
(MS) and 31P NMR spectroscopy, LPC generation from unsaturated PC could be verified in the presence
of HOCl. In contrast, unsaturated PE led exclusively to chlorohydrins and other oxidation products but
not to LPE.
Although these data were obtained with a quite simple model system, it is obvious that LPC is a much
more suitable biomarker of oxidative stress than LPE: LPC is more readily generated and also more sen-
sitively detectable by means of mass spectrometry and other spectroscopic methods. Nevertheless, it will
also be shown that the nitrile of LPE is also generated. However, this compound is exclusively detectable
as negative ion.
Keywords:
31P NMR spectroscopy
MALDI-TOF mass spectrometry
Hypochlorous acid
Phosphatidylcholine
Phosphatidylethanolamine
Lysophosphatidylcholine
Lysophosphatidylethanolamine
Inflammation
Liver tissue
Oxidative stress
Ó 2009 Elsevier Inc. All rights reserved.
1. Introduction
protein mass) and in much smaller amounts in macrophages. As
the number of neutrophilic granulocytes increases massively un-
It is commonly accepted that the concentration of lysophospho-
lipids (LPL) – in particular that of lysophosphatidylcholine (LPC) –
increases under inflammatory conditions. This has been shown, for
instance, for patients suffering from rheumatoid arthritis [1], ath-
erosclerosis [2] and lung infection [3]. The prevalent opinion is that
LPLs are generated under in vivo conditions by the release and/or
activation of the enzyme phospholipase A2 (PLA2) that is present,
for instance, in neutrophilic granulocytes, important ‘‘cellular”
mediators of inflammation. However, neutrophils do not only se-
crete PLA2 upon activation, but are also capable of generating reac-
tive oxygen species (ROS) [4]. Among other ROS, hypochlorous acid
(HOCl) seems primarily responsible for the increased levels of LPC
under pathological conditions [5]. HOCl is generated under in vivo
conditions from H2O2 and Clꢀ ions under the catalytic influence of
the enzyme myeloperoxidase (MPO) [6] that is predominantly
found in neutrophils (where it constitutes about 5% of the total
der inflammatory conditions, the role of MPO and its products
are obvious [7].
The generation of LPC from phosphatidylcholine (PC) under
in vivo conditions is normally discussed by an increased activity
of PLA2 in the presence of HOCl [8]. However, it could be shown
on a model level that LPC is also generated in the absence of
PLA2, i.e. under the influence of HOCl alone: Using PC vesicles with
different fatty acyl compositions it could be proven by mass spec-
trometry that LPC is an abundant reaction product if PC reacts with
either HOCl alone or with the products of the complete MPO/H2O2/
Clꢀ system [9]. Although chlorohydrins (addition products of HOCl
to the double bonds of the unsaturated fatty acyl residues) are the
first detectable products, LPC is also generated in dependence on
the number of double bonds in the fatty acyl residues [9,10]. Satu-
rated LPC species are exclusively generated by this reaction indi-
cating that the unsaturated oxidatively modified, but not the
saturated fatty acyl residue is fragmented from the lipid backbone:
The yield of LPCs increases if the saturation degree of the applied
PC decreases. A mechanism to explain this behavior has recently
been proposed [10]: Oxygen and chlorine are rather electronega-
tive elements and weaken the ester bond of lipids by the
* Corresponding author. Address: Universität Leipzig, Medizinische Fakultät,
Institut für Medizinische Physik und Biophysik, Härtelstr. 16-18, D-04107 Leipzig,
Germany. Fax: +49 341 9715709.
E-mail address: Beate.Fuchs@medizin.uni-leipzig.de (B. Fuchs).
0045-2068/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.bioorg.2009.09.002

C. Schober et al. / Bioorganic Chemistry 37 (2009) 202–210
203
withdrawal of electrons (ꢀI effect). Therefore, the acyl residues in
the chlorohydrins are more sensitive to hydrolysis than the
unmodified acyl residues of the original lipids [10].
the Institute of Pathology, University Clinics of Leipzig, Germany
(Prof. C. Wittekind), in agreement with national laws and ethical
guidelines. Tissue portions were fixed in formaldehyde and
Chlorohydrins and LPCs were also obtained in significant yields
if e.g. lipoproteins from human blood were incubated with HOCl
[11]. However, due to the high protein content of lipoproteins
and the considerable reactivity of the thiol and amino groups of
proteins with HOCl, a considerable excess of HOCl was necessary
to induce alterations of the lipid constituents [12].
embedded in paraffin, and 6 lm thick sections were stained with
Hematoxylin and Eosin (HE). The severity of the inflammatory
state of the liver sections was estimated by light microscopy.
2.2. Incubation of isolated phospholipids with hypochlorous acid
Surprisingly, LPL were not obtained if phosphatidylethanol-
amine (PE) or phosphatidylserine (PS) vesicles were treated with
HOCl, even if a significant excess of HOCl over the PL was used
[13]. This is a clear indication that the above-mentioned mecha-
nism does not represent the complete truth (LPL generation would
be expected independent of the head group of the PL as soon as
oxidative modification of the fatty acyl residues occurs) and that
further factors (maybe also the structure of the PL head group)
determine the LPL yield. Obviously, amino groups possess a much
higher reactivity with HOCl than the double bonds of the fatty acyl
residues [14]. Unfortunately, the prime reaction products that are
generated under these conditions (mono- and di-chloramines)
are not completely stable and decompose in a time-dependent
manner complicating the evaluation of the products. Additionally,
side reactions between the chloramines and the unsaturated fatty
acyl residues cannot be completely ruled out.
An aliquot of each PL dissolved in chloroform was evaporated to
dryness. Vesicles (1.5 mmol/l phospholipid in total) were prepared
by dissolving the phospholipid film in 100 mmol/l phosphate buf-
fer (pH 7.4) and vortexing vigorously for 30 s. In order to mimic
physiological conditions and to generate defined vesicles, a 1:1
mixture between PC 16:0/18:2 and PE 16:0/18:2 was additionally
reacted with HOCl.
A stock solution of NaOCl was kept in the dark at 4 °C. Its con-
centration was determined at pH 12 using e
290 = 350 Mꢀ1cmꢀ1 for
ꢀOCl [20]. NaOCl was diluted with phosphate buffered saline
(100 mmol/l) immediately prior to use. Liposomes were incubated
with varying concentrations of sodium hypochlorite for 1 h at pH
7.4 at 37 °C. In order to extract the lipids the aqueous suspension
was treated with the double volume of a chloroform/methanol
mixture (2:1, v/v) [21]. The organic layer was used for further
analysis.
Finally, the mass spectrometric detection as well as the supra-
molecular structure of PEs is much more difficult. This might be
an additional reason why PEs were so far less intensively investi-
gated than PCs [10].
For comparative purposes, PLs were also digested with PLA2
according to the method essentially described in [22].
2.3. Phospholipid analysis by MALDI-TOF MS
Besides some other diseases, different chronic liver diseases are
also characterized by persistent inflammatory processes and, thus,
associated with increased oxidative stress [15]. As human liver tis-
sue contains considerable amounts of polyunsaturated PC and PE
species the LPLs derived thereof could represent important bio-
markers of oxidative stress and inflammation. Therefore, this paper
has two aims: (a) to investigate inflamed liver tissues towards the
LPL content and (b) to determine the extent of LPL generation if
unsaturated PC and PE species react with HOCl. Due to the com-
plexity of the in vivo system, matrix-assisted laser desorption
and ionization time of flight (MALDI-TOF) mass spectrometry
(MS) [16–18] and 31P NMR spectroscopy [19] were simultaneously
applied as they are completely independent methods.
Positive and negative ion MALDI-TOF mass spectra were ac-
quired on a Bruker Daltonics Autoflex workstation (Bruker, Ger-
many). The system utilizes a pulsed nitrogen laser emitting at
337 nm. The extraction voltage was 20 kV, the ‘‘low mass gate”
was turned on at m/z = 400 and 128 single laser shots were aver-
aged for each mass spectrum. In order to enhance the spectral res-
olution, spectra were recorded in the reflector mode under
‘‘delayed extraction” conditions [16,17].
The organic lipid extracts were normally applied onto the MALDI
target using DHB as matrix for the positive ion mode and PNA for the
negative ion mode [23]. Selected experiments were also performed
in the presence of 9-AA that was recently suggested as a matrix with
superior properties for the analysis of PLs [24]. All spectra were
processed using the software ‘‘Flex Analysis” version 2.2 (Bruker
Daltonics, Germany).
2. Materials and methods
2.1. Chemicals and liver samples
2.4. Liver extraction and liver phospholipid analysis by 31P NMR
spectroscopy
All chemicals for NMR spectroscopy (sodium cholate, EDTA, and
deuterated water with an isotopic purity of 99.6%), buffer prepara-
tion (NaH2PO4 ꢁ H2O, Na2HPO4 ꢁ 2 H2O and trishydroxymethy-
laminomethane, TRIS), and matrix preparation (p-nitroaniline,
PNA, 9-aminoacridine, 9-AA and dihydroxy benzoic acid, DHB) as
well as all solvents (chloroform and methanol) were obtained in
highest commercially available purity from Fluka Feinchemikalien
GmbH (Taufkirchen, Germany). Phospholipase A2 (PLA2) (from por-
cine pancreas) was also purchased from Fluka.
The liver lipid extraction was performed according to Folch [25]
by suspending the liver tissue (sliced in approximately
5 ꢁ 5 ꢁ 5 mm pieces) in 0.9% NaCl solution and the subsequent
addition of chloroform/methanol (2:1:1, chloroform:metha-
nol:water, final volume ratio). After separation of the chloroform
and the aqueous phases, the chloroform layer was evaporated to
dryness in a centrifugal evaporator (Jouan, Germany).
1-palmitoyl-2-linoleoyl-sn-glycero-3-phosphocholine
(PLPC)
The dried organic residues were dissolved in 50 mmol/l TRIS
(pH 7.65) containing 200 mmol/l sodium cholate and 5 mmol/l
EDTA [26]. After intense vortexing of the 0.5 ml samples 31P NMR
spectra were recorded in 5 mm NMR tubes on a Bruker AVANCE-
600 spectrometer operating at 242.88 MHz for 31P. All measure-
ments were performed using a direct 31P/1H NMR probe at 30 °C
(303 K) with composite pulse decoupling (Waltz-16) to eliminate
31P-1H coupling. 31P NMR spectra were additionally recorded at
60 °C (333 K). At this temperature the phosphatidylglycerol (PG)
and 1-palmitoyl-2-linoleoyl-sn-glycero-3-phosphoethanolamine
(PLPE) were purchased as 10 mg/ml solutions in CHCl3 from Avanti
Polar Lipids (Alabaster, MA, USA) and used without further
purification.
Human liver samples from autopsies of normal controls (n = 4)
and patients with inflammatory liver disease (n = 3) were kindly
provided by the Leibniz Research Centre for Working Environment
and Human Factors, Dortmund, Germany (Prof. J.G. Hengstler) and

204
C. Schober et al. / Bioorganic Chemistry 37 (2009) 202–210
resonance shifts to lower ppm values and the lysophosphatidyleth-
anolamine (LPE) resonance can be clearly resolved [27].
Other NMR parameters were as follows: experiment time: 2–
4 h, data size: 8 K, 60° pulse (6.7 ls), pulse delay 2 s, and a line
broadening of 1 Hz. All peak assignments were confirmed by com-
parison with the shift of commercially available reference com-
pounds. Spectra were processed using the software ‘‘1D WIN
NMR” version 6.2Ò (Bruker Analytische Messtechnik GmbH, Rhein-
stetten, Germany).
position of human liver [28]. Although details of the fatty acyl com-
positions and linkage types (diacyl, alkyl–acyl and alkenyl–acyl)
cannot be completely resolved by 31P NMR, splittings of the PE
and PC resonances clearly indicate a rather complex fatty acyl com-
position [29].
In comparison to the spectrum of the lipid extract of the healthy
human liver three lipid extracts of different inflamed human livers
(cf. Fig. 1) are also shown in traces (b), (c) and (d) of Fig. 2. It is evi-
dent that a new, but small resonance arises at d = 0.15 ppm corre-
sponding to lysophosphatidylcholine (LPC). The formation of LPC
under inflammatory conditions can be explained, for instance, by
the hydrolysis of PC under the catalysis of phospholipase A2
(PLA2) [30]. In most cases PLA2 is not specific for PC but is also ac-
tive toward PE (as well as other PL, but the monitoring of the cor-
responding LPL species is difficult due to their rather small
concentrations). This explains the relative PE decrease in inflam-
matory livers (traces (b) and (d)). The reason why in trace (c) no
decrease of the PE resonance intensity occurs is unfortunately un-
known so far. Although there are also effects regarding the PI res-
onance, changes of this resonance will not be considered because
the focus of this study will be on changes of the PE and PC. Unfor-
tunately, the resonance of PG overlaps with the peak of lysophos-
phatidylethanolamine (LPE) released by PLA2 activity from PE if the
31P NMR spectra are recorded at 303 K. Therefore, all 31P NMR
spectra were recorded a second time at 333 K as well (Fig. 2 right
hand side). At 333 K the LPE peak (d = 0.40 ppm), can be easily dif-
ferentiated from the PG resonance (d = 0.35 ppm) (trace (b), (c) and
(d) right hand side). From the spectra recorded at 333 K it is evi-
dent that the LPE resonance is observed only in the case of the in-
flamed livers but not the healthy human liver (trace (a) right hand
side).
Fig. 3 compares the positive ion MALDI-TOF mass spectra of the
organic extracts of the healthy (trace a) and one inflammatory
(trace b) human liver tissue (patient Y; cf. Figs. 1 and 2). All m/z val-
ues are given on the top of the corresponding peaks and all peak
assignments are given in Table 1. The positive ion MALDI-TOF mass
spectra exhibit primarily signals of SM 16:0, different PCs (thus,
particularly PLs with quaternary ammonia groups) and triacylglyc-
erol (TAG) species are detectable. One of the most abundant PC
species present in the human healthy liver extract is PC 16:0/
18:2 (m/z = 758.6) with two double bonds at the sn-2 position of
the glycerol backbone. Of course, unambiguous assignments of
the fatty acyl residues to the sn-1 and sn-2 positions cannot be
3. Results
In Fig. 1 representative histological changes of three different li-
ver specimens affected by inflammatory diseases in comparison to
healthy human liver (trace a) are depicted. Different disease pat-
terns, namely incipient micronodular cirrhotic disease probably
due to chronic alcoholism, minor fatty degeneration and cholesta-
sis in a patient with Lupus erythematodes, and hemorrhagic liver
necrosis in a graft failure seven days after liver transplantation
were included in this study. Under all three conditions, the inflam-
mation was largely confined to the portal fields and of mixed, pre-
dominantly lymphocytic nature. Locally more dense infiltrates
were detectable in the first specimen (trace b; patient X), loose
but more generalized infiltrates in the third (trace d; patient Z),
and very sparse infiltrates with an admixture of eosinophils in
the second (trace c; patient Y). Acute inflammatory, purulent or
granulomatous infiltrates were not present.
Fig. 2 shows the 31P NMR spectra of the organic extracts of dif-
ferently inflamed liver biopsies from the different patients re-
corded at 303 K (left hand side) and 333 K (right hand side).
These different temperatures were chosen in order to demonstrate
the significant NMR spectroscopic resolution increase under
slightly different conditions. Differences in the signal-to-noise ra-
tios of the individual 31P NMR spectra are stemming from the dif-
ferent number of scans averaged for each spectrum. In trace (a) on
the left hand side a representative spectrum of the lipid extract of a
healthy human liver is shown as reference. This spectrum exhibits
six resonances of strongly varying intensities representing phos-
phatidylcholine (PC; d = 0.60 and 0.62 ppm), phosphatidylinositol
(PI; d = 0.48 ppm), phosphatidylserine (PS; d = 0.26 ppm), phospha-
tidylethanolamine (PE; d = 0.06 and 0.04 ppm), sphingomyelin
(SM; d = 0.01 ppm) and phosphatidylglycerol (PG; d = 0.38 ppm)
and is in good agreement with the known phospholipid (PL) com-
Fig. 1. Light microscopic images of autopsy liver samples, showing mild to moderate chronic inflammatory infiltrates in the portal spaces, in addition to other changes under
three different conditions: (a) healthy liver; (b) micronodular cirrhosis due to chronic alcoholism, HE ꢁ 50; (c) minor fatty degeneration and cholestasis in Lupus
erythematodes, HE ꢁ 400; (d) liver necrosis in graft failure seven days after transplantation.

C. Schober et al. / Bioorganic Chemistry 37 (2009) 202–210
205
333 K
303 K
PC
PC
PE
PE
SM
a) healthy
SM
PS
PS
PI
PI
PG
PG
b) X
c) Y
d) Z
LPE+PG
LPE
PG
LPC
LPC
LPC
PS
LPC
LPE
oxidation
products?
oxidation
products?
LPE
0.3
0.1
-0.1
-0.3
-0.5
-0.7
-0.9
0.3
0.1
-0.1
-0.3
-0.5
-0.7
-0.9
31P chemical shift [ppm]
31P chemical shift [ppm]
Fig. 2. 242.88 MHz 31P NMR spectra of the organic extracts of human liver representing the phospholipid composition of human healthy liver (a) and three different human
inflammatory livers (patients X, Y, Z; traces b–d). Spectra were scaled according to the intensity of the most intense resonance. Spectra at the left were recorded at 303 K and
at the right at 333 K in order to clearly differentiate PG and LPE. Abbreviations: PC, phosphatidylcholine; PE, phosphatidylethanolamine; LPC, lysophosphatidylcholine; LPE,
lysophosphatidylethanolamine; PG, phosphatidylglycerol; PI, phosphatidylinositol; PS, phosphatidylserine; SM, sphingomyelin.
758.6 782.6
a) healthy
784.6
788.6
577.5 603.5
650.5
806.6
810.6
855.8
551.5
672.4
881.8
760.6 780.6
LPE
LPC
16:0+18:0 16:0,18:2,18:0
482.3
520.3
786.6
790.6
808.6 832.6
804.6
b) Y
577.6 603.6
454.3
518.3
853.8 879.8
907.8
551.5
703.6 725.6 756.6
734.6
471.3
496.3 504.3 524.4
450
500
550
600
650
700
750
800
850
900
m/z[Th]
Fig. 3. Positive ion MALDI-TOF mass spectra of the organic extracts of healthy (a) and inflammatory (patient Y; b) human livers, while DHB served as matrix. m/z Values are
indicated, for m/z assignments please see Table 1.
made by recording simple mass spectra, but assignments can be
easily confirmed by recording post source decay (PSD) mass spec-
tra [31]. Please also note that the peaks at m/z = 551.5, 577.5 and
603.5 are caused by the loss of one fatty acyl residue from the dif-
ferent triacylglycerol as well as PE species although PE as such is
not detectable in the positive ion spectra in the presence of PC
[17 and references cited therein] – at least if complex mixtures
are investigated. Please also note that the signals at m/z = 650.5
and 672.4 are caused by oxidation of the oleoyl residue in PC
16:0/18:1 under generation of the aldehyde [32]. In contrast to
the healthy, the inflammatory liver extract (trace b) exhibits addi-
tional LPL species (LPC and LPE) in the mass region between
m/z = 450 and 530. Concomitantly the relative contents of the high-
er unsaturated PCs decrease in the inflammatory liver extracts (cf.
PC 16:0/18:1; m/z = 760.6 and PC 16:0/18:2; m/z = 758.6). It is
remarkable that in addition to saturated LPCs there is also a peak
corresponding to LPC 18:2 (m/z = 520.3 for the H+ adduct). Nor-
mally, PLs contain in sn-1 position the saturated and in sn-2 posi-
tion the unsaturated fatty acyl residue [5] and this should lead to
the exclusive generation of saturated LPCs under the influence of
PLA2 and ROS. As we do not yet have convincing explanation of
LPC 18:2 generation, this compound will not be further discussed.

206
C. Schober et al. / Bioorganic Chemistry 37 (2009) 202–210
Table 1
of PE upon hypochlorous acid (HOCl) treatment was compared
with PC in model experiments using isolated PL species. In both
cases, the unsaturated PL species with two double bonds at the
sn-2 position of the glycerol backbone, namely 1-palmitoyl-2-lin-
oleoyl-PL (16:0/18:2) – as one of the main PC and PE species in
the human liver – were chosen in order to simplify the complexity
of the oxidation products. PE and PC species are exclusively consid-
ered in this paper because they are most abundant in the human
liver, while all other PL species contribute only to a much smaller
extent. After incubation of the individual PL with different HOCl
concentrations the reaction products were investigated by
MALDI-TOF MS. 31P NMR spectra gave only broadened resonances
and were, thus, not really helpful regarding detailed product anal-
ysis (data not shown).
Selected examples of positive ion MALDI-TOF mass spectra of
the extracts of the reaction mixtures of PLPC and HOCl are shown
in Fig. 4. All m/z values are given on the top of the peaks and were
assigned according to Table 2. The spectrum of untreated PLPC is
shown in trace (a) as control. PLPC gives two major peaks at
m/z = 758.6 and 780.6 corresponding to the H+ and Na+ adduct,
respectively. The smaller peaks are most probably caused by impu-
rities but could not be assigned. Both PLPC peaks decrease upon
treatment with 2.5 and 5 mmol/l HOCl (trace b and c) and disap-
peared completely upon treatment with 10 mmol/l HOCl (trace
d). In trace (b) the expected oxidation products – epoxides, chlo-
rides [33] and chlorohydrins – appear as new peaks in the higher
mass range. Furthermore, in the lower mass region new peaks arise
at m/z = 496.3 and 518.3 corresponding to the H+ and Na+ adduct of
LPC 16:0. With increasing HOCl concentrations (trace c) the inten-
sities of all peaks of the oxidation products increase and further
peaks of oxidation products (particularly dichlorides) arise. After
a further increase of the HOCl concentration (trace d) the peaks
at m/z = 844.6 and 866.6 (dichloride/epoxide) and m/z = 862.6
and 884.6 (bis-chlorohydrin) represent the main oxidation prod-
ucts beside other minor products and the peaks corresponding to
the degradation product, LPC 16:0, reach their maxima. All these
products are not surprising and in agreement with previous stud-
ies. Therefore, the generation of LPC in the diseased liver could be
explained – beside PLA2-induced cleavage of PC – also by HOCl-in-
duced oxidation.
Assignments of the positive ion signals obtained from the MALDI-TOF mass spectra of
the organic extracts of human livers.
Mass (m/z)
Peak assignment
454.3
471.3
482.3
496.3
504.3
518.3
520.3
524.3
551.0
577.5
603.5
650.5
672.4
703.6
725.6
734.6
756.6
758.6
760.6
780.6
782.6
784.6
786.6
788.6
790.6
804.6
806.6
808.6
810.6
832.6
853.8
855.8
879.7
881.7
907.8
[LPE 16:0 + H]+
Not identified
[LPE 18:0 + H]+
[LPC 16:0 + H]+
[LPE 18:0 + Na]+
[LPC 16:0 + Na]+
[LPC 18:2 + H]+
[LPC 18:0 + H]+
DHB matrix peak
TAG/PE fragment
TAG/PE fragment
Aldehyde from PC 16:0/18:1
Aldehyde from PC 16:0/18:1
[SM 16:0 + H]+
[SM 16:0 + Na]+
[PC 16:0/16:0 + H]+
[PC 16:0/16:0 + Na]+
[PC 16:0/18:2 + H]+
[PC 16:0/18:1 + H]+
[PC 16:0/18:2 + Na]+
[PC 16:0/20:4 + H]+
[PC 18:0/18:3 + H]+
[PC 18:0/18:2 + H]+
[PC 18:0/18:1 + H]+
[PC 18:0/18:0 + H]+
[PC 16:0/20:4 + Na]+
[PC 18:0/18:3 + Na]+
[PC 18:0/20:5 + H]+
[PC 18:0/20:4 + H]+
[PC 18:0/20:4 + Na]+
[TAG (50:2) + Na]+
[TAG (50:1) + Na]+
[TAG (52:3) + Na]+
[TAG (52:2) + Na]+
[TAG (54:3) + Na]+
Nevertheless, the positive ion MALDI-TOF mass spectra basically
confirm the results obtained by 31P NMR: the LPC and LPE contents
increase significantly under inflammatory conditions in the human
liver tissue.
Additionally, negative ion MALDI-TOF mass spectra of the or-
ganic extracts of healthy and inflammatory human liver tissues
were recorded (data not shown). The negative ion MALDI-TOF
mass spectra exhibit signals of different diacyl-GPE, alkenyl-linked
GPE, PG, PS and PI species confirming the results obtained by 31P
NMR. One of the most abundant PE species present in the liver ex-
tracts from the healthy control is PE 18:0/18:2 with two double
bonds at the sn-2 position of the glycerol backbone. Furthermore,
the negative ion spectra of the inflammatory liver extracts do not
show other LPL peaks beside LPE 16:0 and 18:0 (cf. Fig. 3). Concom-
itantly the relative content of the stronger unsaturated PEs de-
creases in the inflammatory liver extracts. However, if the LPLs
would be generated exclusively by the action of PLA2 it must be
anticipated that this enzyme is specific towards the PC and PE spe-
cies because no lysolipids of other PL classes are detectable:
Although their contributions are expected to be much lower in
comparison to LPE and LPC due to the lower concentration of the
original PLs, LPLs such as lysophosphatidylglycerol or serine would
be more sensitively detectable as negative ions due to their in-
creased negative charge.
In Fig. 5 the positive ion MALDI-TOF mass spectra of the extracts
of the reaction mixtures of PLPE reacted with different amounts of
HOCl are shown. In contrast to PC, PE contains a ‘‘free” amino
group exhibiting high reactivity with HOCl. Thus, additional prod-
ucts in comparison to PC (discussed above) can be expected. All
m/z values are given on the top of the peaks and were assigned
according to Table 3. The spectrum of untreated PLPE is shown in
trace (a). PLPE gives three peaks at m/z = 716.5, 738.5 and 760.5
corresponding to the H+, Na+ and H+ + 2Na+ adduct, respectively.
Furthermore, the untreated PLPE gives two different fragmentation
products at m/z = 575.5 corresponding to the loss of the head group
and m/z = 601.5 that is likely to represent a product of a rearrange-
ment [31]. The presence of these fragments (in the gas phase) is a
clear difference in comparison to PC. The intensities of the PLPE
peaks decreased upon treatment with 2.5 mmol/l HOCl (trace b)
and disappeared completely upon treatment with
5 and
10 mmol/l HOCl (trace c and d). While in trace (b) no peaks of cer-
tain products appear in trace (c) the oxidation products – chlorides,
chlorohydrins, dichlorides, epoxides – arise as new peaks in the
higher mass range. All oxidation products are also obvious subse-
quent to fragmentation of the head group (cf. Table 3). After a
further increase of the HOCl concentration (trace d) no additional
peaks of further oxidation products appear but the peaks at
m/z = 802.5 and 824.5 (dichloride/epoxide), m/z = 820.5 and 842.5
(bis-chlorohydrin) and their fragmentation products at m/z = 661.5
and 679.5 (dichloride/epoxide and bis-chlorohydrin) represent
However, there is one additional mechanism by which at least
LPC may be generated from PC at inflammatory loci [7]. The hypo-
chlorous acid-mediated oxidation of PC at inflammatory loci and
the subsequent release of LPC from unsaturated PC are well-docu-
mented [9]. Nevertheless, it is not yet known if other unsaturated
PLs behave similarly under inflammatory conditions and generate
lysophospholipids (LPL) as well. Therefore, the chemical behavior

C. Schober et al. / Bioorganic Chemistry 37 (2009) 202–210
207
758.6
780.6
a)
b)
c)
d)
733.1
631.2
612.4
LPC
16:0
810.6
824.6
846.6
862.5
496.3
663.4
LPC
832.6
844.5
828.6
16:0
884.6
792.6
496.3
518.3
518.3
LPC
16:0
496.3
500
550
600
650
700
750
800
850
900
m/z [Th]
Fig. 4. Positive ion MALDI-TOF mass spectra of the organic extracts of PLPC (1.5 mmol/l each) after incubation with: (a) buffer, (b) 2.5, (c) 5 and (d) 10 mmol/l HOCl for 1 h at
37 °C. m/z Values are indicated, for m/z assignments please see Table 2. DHB was used as matrix. For details see text.
tional experiment was performed, where a 1:1 mixture between
Table 2
PLPC and PLPE was used. A relatively high moiety of PE was used
because the liver contains high amounts of PEs [28]. This system
was also used to illustrate the effects of PLA2. In Fig. 6 some se-
lected MALDI-TOF mass spectra subsequent to PLA2 digestion (a)
and reaction with HOCl (b–d) are shown and the used matrices
as well as the polarities of the measurements are given in the indi-
vidual traces.
In trace (6a) the effect of PLA2 on the PC/PE mixture is shown. In
addition to the peaks of PE 16:0/18:2 and PC 16:0/18:2 there are
intense signals of both expected LPLs. Although the PC and LPC sig-
nals are stronger in comparison to the PE and LPE peaks, it is obvi-
ous that both species are detectable in the positive ion mass
spectra of the mixture. Of course, the ratio between the PLs and
the LPLs can be easily altered by the applied PLA2 activity or the
incubation time. The peaks of both LPLs are marked by vertical dot-
ted lines in order to emphasize their presence or absence in the
individual traces.
Assignments of the positive ion signals obtained from the MALDI-TOF mass spectra of
the organic extracts of PLPC after incubation with HOCl.
Mass (m/z)
Peak assignment
496.3
518.3
612.4
631.2
663.4
733.1
758.6
780.6
792.6
810.6
824.6
828.6
832.6
844.6
846.6
862.6
884.6
[LPC16:0 + H]+
[LPC16:0 + Na]+
Unidentified
Unidentified
Impurity from pipette tip
Unidentified
[PC 16:0/18:2 + H]+
[PC 16:0/18:2 + Na]+
[PC 16:0/18:2 + chlorine + H]+
[PC 16:0/18:2 + chlorohydrin + H]+
[PC 16:0/18:2 + chlorine + bis-epoxid + H]+
[PC 16:0/18:2 + dichloride + H]+
[PC 16:0/18:2 + chlorohydrin + Na]+
[PC 16:0/18:2 + dichloride + epoxid + H]+
[PC 16:0/18:2 + chlorine + bis-epoxid + Na]+
[PC 16:0/18:2 + bis-chlorohydrin + H]+
[PC 16:0/18:2 + bis-chlorohydrin + Na]+
Traces (b) and (c) represent the same sample but were recorded
in the presence of different matrix compounds. As a significant ex-
cess of HOCl (1:10 in comparison to the PLs) was used, the ex-
pected product, i.e. the bis-chlorohydrine of PLPC is easily
detectable in the positive ion spectra. This is particularly evident
if the spectra are recorded in the presence of 9-AA (6c) because un-
der these conditions primarily the H+ (m/z = 862.5) but not the Na+
adducts are detectable. In the presence of DHB (6b) the same prod-
uct cationized by H+ as well as Na+ (m/z = 884.5) are detectable.
However, in addition to the bis-chlorohydrin of PLPC, the bis-chlo-
rohydrin of PLPE is also detectable. It is not surprising that the PE is
exclusively detectable in the positive ion spectrum in the presence
of DHB because DHB is a stronger acid in comparison to 9-AA. Nev-
ertheless, in both cases, the expected LPC but not even small
amounts of the LPE are detectable (marked by the dotted vertical
line) although a considerable excess of HOCl was used.
the main oxidation products beside other minor products. It was,
however, not the aim of this work to evaluate all – even minor
products – in more detail. Obviously (and in clear contrast to PC)
the HOCl oxidation of unsaturated PLPE does not lead to a signifi-
cant formation of LPE 16:0. LPE 16:0 would be expected at
m/z = 454.3, 476.3 and 498.3 corresponding to the H+, Na+ and
H+ + 2Na+ adducts, respectively.
Unfortunately, it is well-known that the tendency of PC and PE
to generate defined supramolecular structures is different as the PE
forms lamellar phases [34]. This is not a major problem in the case
of physiological PL mixtures because there is normally an excess of
PC that forms defined vesicles, where the PEs are entrapped, but is
a potential problem if pure PE is investigated. Therefore, an addi-

208
C. Schober et al. / Bioorganic Chemistry 37 (2009) 202–210
716.5
738.5
575.5
601.8
760.5
a)
b)
c)
d)
679.5
663.5
627.5
no LPE
842.5
16:0
790.5
661.5
645.5
864.5
876.5
824.5
591.5
573.5
772.5
806.5
898.5
551.0
467.4
449.3
609.5
no LPE
820.5
16:0
802.5
858.5
730.0
706.1
450
500
550
600
650
700
750
800
850
900
m/z [Th]
Fig. 5. Positive ion MALDI-TOF mass spectra of the organic extracts of PLPE (1.5 mmol/l each) after incubation with: (a) buffer, (b) 2.5, (c) 5 and (d) 10 mmol/l HOCl for 1 h at
37 °C. m/z values are indicated, for m/z assignments please see Table 3. The same conditions as indicated in Figs. 3 and 4 were used.
Table 3
bis-chlorohydrine of PE 16:0/18:2. It is not astonishing that this
compound is particularly well detectable in the negative ion mode
spectrum because the nitrile has lost the positively charged quater-
nary ammonia group but retained the negatively charged phos-
phate residue. Therefore, the nitrile is not detectable in the
positive ion spectra. Accordingly, the nitrile of LPE 16:0 is detect-
able at m/z = 448.3.Please also note that the peak marked by ‘‘x”
is stemming from PC that is detectable in the presence of 9-AA
as negative ion subsequent to loss of a methyl group. For further
information see [35].
Thus, the positive ion MALDI-TOF mass spectra of PC 16:0/18:2
and PE 16:0/18:2 indicate a comparable behavior upon HOCl treat-
ment and the same addition products of HOCl to the two double
bonds at the sn-2 position of the glycerol backbone are obvious. It
does obviously not matter whether both PLs are independently
investigated or in a mixture. A pronounced difference is, however,
the LPC formation from PC upon HOCl treatment. Due to the pres-
ence of the additional amino group that exhibits high reactivity with
HOCl, the generation of head group-modified oxidation products of
PE is most likely [13,36], but their analysis was not the aim of this
study because it is much more difficult and depends considerably
on the applied matrix. Nevertheless, the generation of LPE in the po-
sitive ion mode can be certainly excluded under the used experi-
mental conditions. The LPE is not detectable as negative ion, too.
However, the corresponding nitrile of the LPE is detectable in the
negative ion mode (cf. 6d) due to its excessive negative charge.
Assignments of the positive ion signals obtained from the MALDI-TOF mass spectra of
the organic extracts of PLPE after incubation with HOCl.
Mass (m/z)
Peak assignment
551.0
575.5
591.5
609.5
627.5
645.5
661.5
663.4
679.5
685.4
716.5
738.5
760.5
772.5
790.5
802.5
806.5
820.5
824.5
842.5
858.5
864.5
876.5
898.5
DHB matrix peak
Fragment-head group
[Fragment from epoxid + H]+
[Fragment from glycol + H]+
[Fragment from chlorohydrin + H]+
[Fragment from dichloride + H]+
[Fragment from dichloride + epoxid + H]+
Impurity from pipette tip + H+
[Fragment from 2 ꢁ chlorohydrin + H]+
Impurity from pipette tip + Na+
[PE 16:0/18:2 + H]+
[PE 16:0/18:2 + Na]+
[PE 16:0/18:2-H+ + 2Na]+
[PE 16:0/18:2 + chlorine + Na]+
[PE 16:0/18:2 + chlorohydrin + Na]+
[PE 16:0/18:2 + dichloride + epoxid + H]+
[PE 16:0/18:2 + bis-chlorine + Na]+
[PE 16:0/18:2 + bis-chlorohydrin + H]+
[PE 16:0/18:2 + dichloride + epoxid + Na]+
[PE 16:0/18:2 + bis-chlorohydrin + Na]+
[PE 16:0/18:2 dichloride + chlorine + epoxid + Na]+
[PE 16:0/18:2 + bis-chlorohydrin – H+ + 2Na]+
[PE 16:0/18:2 dichloride + epoxid + chlorohydrin + Na]+
[PE 16:0/18:2 dichloride + epoxid + chlorohydrin – H+ + 2Na]+
9-AA is a highly suitable matrix to record negative ion spectra
and the corresponding spectrum is shown in trace (6d). It is obvi-
ous that the most intense peak (m/z = 814.5) does not correspond
to the expected negative ion of the bis-chlorohydrin of PLPE (ex-
pected at m/z = 818.5) but differs for 4 amu. Therefore, as already
previously shown [13], this peak corresponds to the nitrile of the
4. Discussion and conclusions
LPC and LPE formation could be detected by 31P NMR spectros-
copy and MALDI-TOF MS under in vivo conditions in different hu-
man inflammatory liver tissues. Both species may be generated

C. Schober et al. / Bioorganic Chemistry 37 (2009) 202–210
209
LPC
16:0
COOH
LPE
16:0
PC 16:0/18:2
PE 16:0/18:2
OH
HO
(a)
(b)
(c)
PLPE
862.5
+ 2 HOCl
COOH
OH
(820.5/842.5)
884.5
HO
*
862.5
PLPC
NH2
N
+ 2 HOCl
-HCl
-H2O
796.5
814.5
PLPE (Nitrile)
+ 2 HOCl
796.5
PLPE (Nitrile)
+ 2 HOCl
-H2O
PLPE (Nitrile)
+ 2 HOCl
-HCl
448.3
Nitrile from
LPE 16:0
NH2
N
X
(d)
400
450
500
550
600
650
700
750
800
850
900
m/z [Th]
Fig. 6. Positive and negative ion MALDI-TOF mass spectra of the organic extracts of a PLPE/PLPC mixture (1:1) after incubation with PLA2 (a) and a tenfold excess of HOCl (b-
d). The used matrices and the polarities of the individual spectra are given directly in the figure. The assignments of the most abundant products are also given directly in the
figure. The vertical dotted lines indicate the H+ and Na+ adducts of LPE 16:0 and LPC 16:0. For further details see the text.
by the enzyme PLA2. It is, however, also known that HOCl is capa-
ble to generate LPC from unsaturated PC under inflammatory con-
ditions. The LPC formation from unsaturated PCs has been already
well documented [10]. Therefore, MALDI-TOF MS analysis has been
used to clarify in more detail the potential LPL generation from PE
by HOCl. Both PL classes – PC and PE – have been compared to
clarify which PL class leads to LPL generation under the oxidative
influence of HOCl. Only the diunsaturated species (linoleoyl resi-
dues) – as one of the main PC and PE species in the human liver
– have been chosen for all incubations with HOCl in order to reduce
the resulting complexity of the products and because monounsat-
urated species (with e.g. oleoyl residues) do not yield major
amounts of LPLs [10]. Relatively large amounts of HOCl have been
used to mimic the continuous generation of HOCl by MPO under
inflammatory conditions [4].
the PE head groups may account for their rigidity [38,39]. Thus,
this existence of a network of noncovalent bonds and the conse-
quential rigidity of the head group may account on the one hand
for the typical fragmentation behavior of PE and may stabilize
the negative inductive effect caused by the chlorohydrin formation
in the aliphatic hydrocarbon chain on the other hand.
Summarizing, it is obvious that the LPC formation from unsatu-
rated PC and the accumulation in inflammatory liver tissue can be
caused by both, PLA2-activity and oxidation by hypochlorous acid
whereas LPE formation and accumulation results exclusively from
PLA2-activity. It is, thus, concluded that LPC represents a useful
biomarker of oxidative stress that can be easily determined by
MS – even in complex mixtures and without the need of previous
separation. Finally, however, it is also obvious that under condi-
tions of negative ion detection, the nitrile of LPE is detectable. As
this compound is not detectable in the positive ion spectra, a quan-
titative assessment of this compound is rather difficult and it could
not be the aim of this study to clarify the contribution of the nitrile
with reference to the LPE. Nevertheless, clarifying this problem is a
current research topic in our laboratory.
It could be shown that both PL classes give the expected chlori-
nation/oxidation of the double bonds in the unsaturated fatty acyl
residue under chlorohydrin, dichloride and epoxide formation. In
contrast, PC (but not PE) is also characterized by subsequent LPL
formation upon treatment with HOCl.
Therefore, it is assumed that the PL head group exhibits a major
influence on the tendency of the PL to LPL formation upon HOCl
oxidation. In this respect it is also a striking result that PE gives in-
tense fragmentation indicating head group losses during positive
ion MALDI-TOF MS measurements. Concomitantly, PE does not
form LPE in the presence of HOCl. Browning [37–39] found that
the PC head group is more flexible than the PE head group and dis-
cussed the differences in head group rigidity in terms of the capac-
ity of the various head groups to bind noncovalently to their
neighbors. It was assumed that PC cannot form a proton transfer
complex (—POꢀ2 ꢂ ꢂ ꢂ H—þNH2—) with neighboring phosphates [40]
and can only interact electrostatically that appears to be quite
weak and accounts for the more flexible nature of the choline head
group [37]. In contrast, the proton transfer complex described for
A correlation of liver lipidomics and histological changes has
also not been attempted so far. Although we could show exempl-
arily clear differences between diseased and healthy liver samples,
more stringent statements with potential diagnostic relevance
must await investigations of larger patient series.
Acknowledgments
This work was supported by the Deutsche Forschungsgemeins-
chaft (DFG Schi 476/5-1 and FU 771/1-1) and the German Ministry
of Education and Research (BMBF Grant 0313836). We gratefully
thank Prof. C. Wittekind (Institute of Pathology, University Clinics
of Leipzig) for the human liver samples from autopsies of patients
with inflammatory liver diseases.

210
C. Schober et al. / Bioorganic Chemistry 37 (2009) 202–210
[19] J. Schiller, M. Müller, B. Fuchs, K. Arnold, D. Huster, Curr. Anal. Chem. 3 (2007)
References
283–301.
[20] J.C. Morris, J. Phys. Chem. 70 (1966) 133–140.
[21] J. Leßig, J. Schiller, J. Arnhold, B. Fuchs, J. Lipid, Res. 48 (2007) 1316–1324.
[22] J. Schiller, K. Müller, R. Süß, J. Arnhold, C. Gey, A. Herrmann, J. Leßig, K. Arnold,
P. Müller, Chem. Phys. Lipids 126 (2003) 85–94.
[23] J. Schiller, J. Arnhold, S. Benard, M. Müller, S. Reichl, K. Arnold, Anal. Biochem.
267 (1999) 46–56.
[24] G. Sun, K. Yang, Z. Zhao, S. Guan, X. Han, R.W. Gross, Anal. Chem. 80 (2008)
7576–7585.
[25] J. Folch, M. Lees, G.H. Sloane, Stanley, J. Biol. Chem. 226 (1957) 497–509.
[26] E. London, G.W. Feigenson, J. Lipid, Res. 20 (1979) 408–412.
[27] A. Puppato, D.B. DuPré, N. Stolowich, M.C. Yappert, Chem. Phys. Lipids 150
(2007) 176–185.
[28] P. Puri, R.A. Baillie, M.M. Wiest, F. Mirshahi, J. Choudhury, O. Cheung, C.
Sargeant, M.J. Contos, A.J. Sanyal, Hepatology 46 (2007) 1081–1090.
[29] J.M. Pearce, R.A. Komoroski, R.E. Mrak, Magn. Reson. Med. 61 (2009) 28–34.
[30] M. Menschikowski, A. Hagelgans, G. Siegert, Prostag. Lipid Mediat. 79 (2006)
1–33.
[31] B. Fuchs, C. Schober, G. Richter, R. Süß, J. Schiller, J. Biochem, Biophys. Methods
70 (2007) 689–692.
[32] B. Fuchs, J. Schiller, R. Süß, A. Nimptsch, M. Schürenberg, D. Suckau, J. Planar,
Chromatogr. 22 (2009) 35–42.
[33] H. Spalteholz, K. Wenske, O.M. Panasenko, J. Schiller, J. Arnhold, Chem. Phys.
Lipids 129 (2004) 85–96.
[34] A.C. Carr, J.J. van den Berg, C.C. Winterbourn, Biochim. Biophys. Acta 1392
(1998) 254–264.
[35] B. Fuchs, A. Bischoff, R. Süß, K. Teuber, M. Schürenberg, D. Suckau, J. Schiller,
Anal. Bioanal. Chem., in press.
[36] T. Jaskolla, B. Fuchs, M. Karas, J. Schiller, J. Am, Soc. Mass Spectrom. 20 (2009)
867–874.
[37] J.L. Browning, Biochemistry 20 (1981) 7123–7133.
[38] J.L. Browning, Biochemistry 20 (1981) 7133–7143.
[39] J.L. Browning, Biochemistry 20 (1981) 7144–7151.
[40] J.M. Boggs, Biochim. Biophys. Acta 906 (1987) 353–404.
[1] B. Fuchs, J. Schiller, U. Wagner, H. Häntzschel, K. Arnold, Clin. Biochem. 38
(2005) 925–933.
[2] T. Matsumoto, T. Kobayashi, K. Kamata, Curr. Med. Chem. 14 (2007) 3209–
3220.
[3] S. Hammerschmidt, N. Büchler, H. Wahn, Chest 121 (2002) 573–581.
[4] J. Schiller, B. Fuchs, J. Arnhold, K. Arnold, Curr. Med. Chem. 10 (2003) 2123–
2145.
[5] B. Fuchs, C. Schober, G. Richter, A. Nimptsch, R. Süß, J. Schiller, Mini-Rev. Org.
Chem. 5 (2008) 254–261.
[6] E. Malle, P.G. Furtmüller, W. Sattler, C. Obinger, Brit. J. Pharmacol. 152 (2007)
838–854.
[7] B. Halliwell, Am. J. Med. 91 (1991) 14S–22S.
[8] P. Kougias, H. Chai, P.H. Lin, A.B. Lumsden, Q. Yao, C. Chen, Med. Sci. Monit. 12
(2006) RA5–16.
[9] J. Arnhold, A.N. Osipov, H. Spalteholz, O.M. Panasenko, J. Schiller, Free Radical
Biol. Med. 31 (2001) 1111–1119.
[10] J. Arnhold, A.N. Osipov, H. Spalteholz, O.M. Panasenko, J. Schiller, Biochim.
Biophys. Acta 1572 (2002) 91–100.
[11] J. Schiller, O. Zschörnig, M. Petkovic´, M. Müller, J. Arnhold, K. Arnold, J. Lipid,
Res. 42 (2001) 1501–1508.
[12] O. Zschörnig, C. Bergmeier, R. Süß, K. Arnold, J. Schiller, Lett. Org. Chem. 1
(2004) 381–390.
[13] G. Richter, C. Schober, R. Süß, B. Fuchs, C. Birkemeyer, J. Schiller, Anal. Biochem.
376 (2008) 157–159.
[14] D.I. Pattison, C.L. Hawkins, M.J. Davies, Chem. Res. Toxicol. 16 (2003) 439–449.
[15] R.F. Schwabe, D.A. Brenner, Am. J. Physiol. Gastrointest. Liver Physiol. 290
(2006) G583–G589.
[16] B. Fuchs, K. Arnold, J. Schiller, in: R.A. Meyers (Ed.), Encyclopedia of Analytical
Chemistry, Wiley, Chichester, 2008, pp. 1–39.
[17] J. Schiller, R. Süß, J. Arnhold, B. Fuchs, J. Leßig, M. Müller, M. Petkovic´, H.
Spalteholz, O. Zschörnig, K. Arnold, Prog. Lipid Res. 43 (2004) 443–478.
[18] J. Schiller, R. Süß, B. Fuchs, M. Müller, O. Zschörnig, K. Arnold, Front. Biosci. 12
(2007) 2568–2579.