Available online at www.sciencedirect.com
Progress in
Lipid Research
Progress in Lipid Research 47 (2008) 15–36
www.elsevier.com/locate/plipres
Review
Lipidomics: Practical aspects and applications
Claude Wolf
a
a,*
, Peter J. Quinn
b
UMRS 538, UMPC Faculté de Medecine Pierre et Marie Curie, 27 Rue Chaligny, 75012 Paris, France
b
Department of Biochemistry, King’s College London, 150 Stamford Street, London SE1 9NH, UK
Received 11 July 2007; received in revised form 7 September 2007; accepted 7 September 2007
Abstract
Lipidomics is the characterization of the molecular species of lipids in biological samples. The polar lipids that comprise the bilayer
matrix of the constituent cell membranes of living tissues are highly complex and number many hundreds of distinct lipid species. These
differ in the nature of the polar group representing the different classes of lipid. Each class consists of a range of molecular species
depending on the length, position of attachment and number of unsaturated double bonds in the associated fatty acids. The origin of
this complexity is described and the biochemical processes responsible for homeostasis of the lipid composition of each morphologically-distinct membrane is considered. The practical steps that have been developed for the isolation of membranes and the lipids there
from, their storage, separation, detection and identification by liquid chromatography coupled to mass spectrometry are described.
Application of lipidomic analyses and examples where clinical screening for lipidoses in collaboration with mass spectrometry facilities
are considered from the user point of view.
Ó 2007 Elsevier Ltd. All rights reserved.
Keywords: Membrane preparation; Phospholipid; HPLC separation; Tandem mass spectrometry
Contents
1.
2.
3.
4.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Structure–function relationships . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.1. Barrier functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2. Interaction with membrane proteins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Physical properties of membrane lipids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.1. Polar groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2. Properties of the hydrocarbon groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Biochemical homeostasis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.1. Phospholipases A2 (PLA2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2. Acyltransferases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.3. Acylation–deacylation cycle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16
17
17
18
18
18
19
20
20
21
21
Abbreviations: PC, 1,2-diacyl-sn-3-glycerophosphocholine; PE, 1,2-diacyl-sn-3-glycerophosphoethanolamine; PS, 1,2-diacyl-sn-3-glycerophosphoserine; PI, 1,2-diacyl-sn-3-glycerophosphoinositol; PA, 1,2-diacyl-sn-3-glycerophosphatidic acid; PG, 1,2-diacyl-sn-3-glycerophosphoglycerol; IP3, inositol3,4,5-trisphosphate; MS, mass spectrometry; LCMS2, liquid chromatography coupled to tandem mass spectrometry; SPE, solid phase extraction; BHT,
2,6-di-tert-butyl-p-cresol; ESI, electro-spray-ionisation source; APCI, atmospheric pressure chemical ionisation source; MRM, multiple reaction monitoring; CTEP, cholesterol triglyceride exchange protein; LCAT, lecithin cholesterol acyl transferase; EFA, essential fatty acid; CID, collision induced
dissociation (fragmentation).
*
Corresponding author.
E-mail addresses: [email protected] (C. Wolf), [email protected] (P.J. Quinn).
0163-7827/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.plipres.2007.09.001
16
5.
6.
7.
8.
9.
C. Wolf, P.J. Quinn / Progress in Lipid Research 47 (2008) 15–36
4.4. Phospholipases C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.5. Phospholipases D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.6. Sphingomyelinases. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Sample preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.1. Homogenization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.2. Membrane fractionation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.3. Lipid extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.4. Sample storage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Factors underlying a lipidomic analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.1. Sample size. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.2. Sample choice and handling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Standards for authentication and quantitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Advances in analytical techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.1. HPLC separation of complex mixtures is usually required prior to MS detection . . . . . . . . . . . . . . . . . . . . . . . . . .
8.2. Electrospray ionization procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1. Introduction
Lipodomics is the methodology, analogous to proteomics in the characterization of cellular proteins, able to
produce an extensive listing of lipid classes and their distinctive molecular species that are present in a biological
sample. The constituents of complex lipids are extremely
diverse and the combinations and permutations in which
they are assembled in the constitution of individual molecular species of lipid further amplify this diversity.
A priori the characterization of complex lipids, which
represent one of the primary structural components of cell
membranes, presents a considerable challenge. The underlying strategy involves firstly, the isolation of morphologically-distinct membranes or subfractions there from and,
secondly, the extraction of these lipids free from the membrane proteins and other components. The identity of lipids responsible for anchoring and lipids non-covalently
attached to a specific group of proteins in membranes is
another aspect of lipidomics. Once extracted, the lipids
must then be fractionated, usually requiring a multi-step
chromatography process, to allow identification and quantitation of the individual molecular species. The intricacy of
these operations may be appreciated by the fact that there
are 17 or so long-chain fatty acids found associated in complex lipids in human cells which are distributed in pairs in
five major classes of glycerophospholipids and there are
also six or so very long-chain fatty acids in the various
sphingolipids.
The complexity of the task is exemplified by the human
red cell which contains only a plasma membrane [1]. This
membrane is known to contain in the order of 300 molecular
species of glycerophospholipids formed by different longchain (C14–C22) fatty acids each present at a level exceeding 2% of the total fatty acids. The fatty acids are linked
by ester bonds with the major phosphatidyl moieties found
in PC, PE, PS, PI, PA classes. The role of each molecular
species resulting from a particular combination of a polar
21
22
22
23
23
23
25
26
27
27
28
29
29
29
32
34
34
34
headgroup and two acyl chains, located at specific carbon
atoms of the glycerol, of particular length and number
and position of unsaturated bonds, on the properties of
the membrane is poorly understood. In addition to glycerophospholipids, a variety of sphingolipid classes is also found
amidified by many minor fatty acid species such as
branched- or hydroxylated-fatty acids (<1 mol%). The profiling of fatty acids in red cell membranes has been recognized for some time as a diagnostic indicator of essential
fatty acid deficiency in children and has become a routine
test long before the role of the molecular diversity of membrane lipids had been fully appreciated. The same tentative
approach is inferred for the developing lipidomics field at a
time when not all bioactive lipids have yet been discovered.
Lipids considered initially as the building material of membranes or as a fuel for bioenergetics are now frequently
regarded as a potential reservoir for precursors of signaling
second messengers. This could represent a clue for understanding the molecular diversity observed in membrane
phospholipids. Subtle biophysical properties are also
another possible explanation especially with reference to
the emerging field of heterogeneous membrane domains.
Other applications of lipidomic technology include tracing of lipid metabolites such as detection of a particular
lipid present in low proportions and subject to rapid turnover, for example, inositol-1,4,5-trisphosphate (IP3).
Another example is the production of eicosanoids from
arachidonic acid derived from precursor arachidonic acid.
Lipidomics can also be applied to trace the interconversions that operate between lipid classes such as the transformation of PE to PC by successive methylation of the
amino group of PE or the attachment of sugar residues
in glycosphingolipids. These applications can be categorized into a separate group referred to as ‘‘Metabolomics’’
which utilizes the general methodology applied to establish
metabolic pathways. A notable difference from conventional methods, however, is that recent methods of lipidomics do not necessarily employ tracers or probes like
C. Wolf, P.J. Quinn / Progress in Lipid Research 47 (2008) 15–36
fluorescent or radioisotopic derivatives such as radioactive
fatty acids, 32P or fluorescent group labeling. Instead, the
fate of the naturally-occurring lipids is followed. This has
a distinct advantage in that it is not necessary to assume
that a particular molecular species which incorporates the
tracer is representative for the entire lipid class. For
instance, similarities of molecular species profiles between
PE and PC demonstrate common biosynthetic pathways
from a single precursor (phosphatidic acid) but in
actively-synthesizing tissues there may be a superposition
of an active methylation of PE to PC. With the aim of
establishing the similarities, minor molecular species may
be more significant than the similarity between abundant
species as probed by tracer methods. It is also possible to
quantify the incorporation of 2H and 13C isotopic tracer
to measure lipid turnover in cells and tissues. Turnover in
this context differs from the transformation of one lipid
into another in that it involves enzymic degradation and
de novo biosynthesis from non-lipid components. It is confidently anticipated that the lipidomics approach using
lipid profiling will help to assess lipid remodeling of molecular species, interconversions between lipid classes and
rates of de novo synthesis.
There is a clear expectation that the application of proteomic technology supplemented by databank-mining will
correlate the proteome with particular cellular responses
or adaptation sought by the cell biologist or pathologist
to explain the molecular mechanisms underlying cellular
processes. Thus, strict rules have been devised to determine
how domain sequences of polypeptide will arrange in the
three-dimensional structure of proteins and functional consequences (enzyme activity) can be assessed directly in biochemical function. In the case of lipidomics the link
between the proportions of molecular species of lipids present in different cell membranes and the functional properties they endow is not so apparent. The properties of
membranes that depend on the lipid component are subject
to how the lipids are arranged in the structure. It is well
known that ATP dependent sorting processes are responsible for the generation and preservation of lipid asymmetry
across the two leaflets of the membrane lipid bilayer. The
forces that give rise to lateral domain segregation in the
plane of the membrane, however, are as yet ill defined.
However, it is anticipated that the arrangement of lipids
in membranes might not be at their lowest free energy level
but activated by energy-derived relocation of lipids. Likewise, the biochemical mechanisms that sense the molecular
species composition of each cell membrane and preserve
the ensemble within relatively narrow and clearly defined
limits (lipid homeostasis) is poorly understood.
What is clear is that a complete molecular species composition of lipids in a cell or tissue will not lead directly to a
description of membrane function based on current knowledge of the structures that would result from such an
ensemble. To eventually understand the significance of
the lipidome further knowledge of the factors that govern
the phase separation of the lipids of biological membranes
17
and how this influences the sorting and activation of the
membrane proteins is required. However, simulations of
self-assembly in model lipids mixtures have been tentatively
calculated.
For the present, this work attempts to review current
practices in lipidomics in the expectation that from a better
knowledge of the molecular species of membrane lipids and
the changes therein that accompany physiological
responses a greater understanding of the role of lipids in
membranes will be achieved.
2. Structure–function relationships
The complexity of molecular species of membrane structural lipids may a priori be attributed to the functions they
are required to perform. These are to create selectively permeable barriers between the living cell and its environment
and, within eukaryotic cells, to compartmentalize the cell
into separate functional organelles. They act to support
the various membrane proteins and often modulate the
biochemical properties of these proteins. Membrane lipids
are known to undergo structural rearrangement during
the process of membrane fusion which is a fundamental
event in membrane biogenesis and cell division. Certain
membrane lipids also act as receptors on the cell surface
such as the ABO blood group antigens of the human erythrocyte and as reservoirs for precursors used in the biosynthesis of eicosenoids such as prostaglandins, leucotrienes,
etc. Despite these widely varying functions it is not easy
to explain why such diversity of molecular species of membrane lipids is observed in living organisms.
2.1. Barrier functions
A characteristic feature of all membrane lipids is that
they are weak surfactants. They are comprised of chemical
substituents that are hydrophobic and others that are
hydrophilic. The combination of these two affinities located
in separate domains of the molecule confers on it the property of amphiphilicity, literally ‘‘two loves’’. Being weak
surfactants they have critical micellar concentrations in
the order of nM and thus tend to form aggregates in physiological salt solutions. The structure of these aggregates is
either a bilayer or hexagonal-II arrangement at temperatures that prevail during cell growth. All biological membranes are comprised of a mixture of these two types of
structure-forming molecular species of polar lipid.
When total polar lipid extracts of biological membranes
are dispersed in aqueous solutions under physiological
conditions both bilayer and non-bilayer structures are
observed. Since the molecular species of membrane lipid
that form non-bilayer structures are found only in bilayer
configuration in membranes it is assumed that their interaction with other membrane constituents imposes a
bilayer arrangement upon them. Because the amphipathic
balance in non-bilayer forming lipids favours hydrophobic
affinity it has been argued that the bulky hydrocarbon
18
C. Wolf, P.J. Quinn / Progress in Lipid Research 47 (2008) 15–36
chains interface with the irregular surface of intrinsic
membrane proteins to seal the membrane against the passage of solutes.
In order for lipid bilayers to serve as a barrier against
the movement of water soluble solutes and ions the minimal requirements can be achieved by simple amphiphiles.
It is therefore not obvious why such a wide variety of
molecular species of membrane lipids have been evolved
to perform simple barrier functions.
2.2. Interaction with membrane proteins
Extrinsic membrane proteins interact with the ionic
groups of polar lipids via electrostatic interactions. This
is evident from the definition of such proteins in that they
can be extracted from the membrane simply by disturbance
of the ionic interactions between them. The charged groups
of membrane lipids can be negative in the form of phosphate, sulphate or carboxyl groups or positive in the form
of amino groups. The net charge, however, is almost invariably negative or zwitterionic. This means that extrinsic proteins interact with membrane lipid bilayers via their basic
amino acid residues.
3. Physical properties of membrane lipids
Because the lipids of biological origin have such diverse
composition and structure there is no systematic method of
nomenclature that would embrace all members of this
group. Lipids are therefore classified operationally as compounds that are soluble in solvents of low polarity. The solubility in organic solvents nevertheless varies greatly but
most are virtually insoluble in water.
A general classification, and one that recognizes differences in solubility, is into either neutral or polar lipids.
The neutral lipids comprise the waxes, fats and oils of
plants and animals. Polar lipids are often found in structural elements of cells and, in particular, the subcellular
membranes.
Polar lipids are amphipathic (or amphiphilic). The molecules are divided into non-polar domains consisting predominantly of hydrocarbon and polar domains hydrated
with water. Their amphipathic property is required to fulfill
their primary function in membranes. They can be
regarded as weak surfactants able to solubilize membrane
proteins in a two-dimensional structure. The surface activity can be assessed from their relative solubility in water
expressed in terms of their critical micelle concentration.
Most polar lipids found in cell membranes have critical
micelle concentrations in the nM range [2]. This compares
with conventional detergents that have critical micelle concentrations in the mM range. Thus the amphipathic balance in polar lipids is weighted heavily in favour of the
non-polar domain of the molecule. This can be seen by
the consequences of removing one of the acyl chains from
a phospholipid; the critical micelle concentration of lysophospholipids are in the mM range [3].
The amphipathic balance in polar lipids found in biological membranes is finely poised such that under physiological conditions they form either a bilayer or hexagonal-II
structure. The bilayer is stabilized by exposure of the
hydrated polar group on the surface and sequestering the
hydrocarbon constituents within the interior of the structure [4]. Hexagonal-II structure consists of tubes of water
formed by the lipid polar groups and the tubes are packed
into a two-dimensional hexagonal array. The amphipathic
balance favours the hydrocarbon affinity and typically the
polar group is less hydrated than bilayer-forming lipids. In
living cells the membranes are comprised of representatives
of both types of polar lipid but the lipid matrix of the membrane is invariably a bilayer arrangement of the lipids. An
outline of the biochemical mechanisms in place to preserve
the molecular species of polar lipids within relatively narrow limits are presented in the next section.
3.1. Polar groups
The polar lipids are classified on the basis of the type of
polar group. The simplest structures of the polar lipids and
also the most abundant are the galactosylglycerolipids.
These are the dominant polar lipids of photosynthetic thylakoid membranes and consist of diglyceride to which one
(monogalactosyldiacylglycerols) or two (digalactosyldiacylglycerols) galactose sugars are attached. Other glycosylglycerides are also known including sulphonated sugars as
in the case of sulphoquinovosyldiacylglycerol which is
found abundantly in higher plant photosynthetic membranes. The sugar residues represent the polar groups of
these lipids and the extent of this interaction is manifest
in the fact that with one sugar residue the membrane lipids
form hexagonal-II structure while the presence of two
sugar residues is conducive to formation of bilayer structures under physiological conditions.
The phospholipids are the most common membrane lipids. The glycerol backbone is acylated with long-chain fatty
acids at carbon-1 and carbon-2 and in some molecular species the fatty acids may be attached by ether or vinyl ether
bonds. The polar groups consist of a phosphate esterified
to the carbon-3 position of the glycerol to which a base
(choline, ethanolamine), polyol (glycerol, inositol) or
amino acid (serine) is attached. All classes of membrane
phospholipids tend to form bilayer structures under physiological conditions except the polyunsaturated phosphatidylethanolamines which form hexagonal-II structure.
Some classes possess a net negative charge at pH 7 such
as the inositol, serine, glycerol and ethanolamine phosphatides while the choline phosphatides are zwitterionic.
Another large group of polar membrane lipids are
based on ceramide which is comprised of a long-chain
base, sphingosine, to which a long-chain fatty acid is
attached by a peptide bond. The most common are the
sphingomyelins which are characterised by the presence
of a phosphorylcholine polar group. Most representatives
of sphingolipids have sugar residues attached to the
C. Wolf, P.J. Quinn / Progress in Lipid Research 47 (2008) 15–36
hydroxyl group of the ceramide. The attachment of phosphorylinositol to ceramide represents the sphingolipid
analogue of phosphatidylinositol and it is a major component of sphingolipids in yeasts where it occurs with two
further inositol-containing sphingophospholipids, mannosylinositolphosphorylceramide and mannosyldiinositolphosphorylceramide. The cerebrosides have one hexose
sugar group, most commonly glucose or galactose, and
when a sulphur containing group is esterified to the sugar
to form a negatively-charged polar group the lipids are
referred to as sulphatides. Sphingolipids with complex oligosaccharides and sialic residues form the gangliosides
which are classified according to the number and arrangement of sugar residues constituting the polar group. They
are oligoglycosylceramides containing N-acetyl-neuraminic acid (sialic acid) or less commonly, N-glycolyl-neuraminic acid, joined via glycosidic linkages to one or
more of the monosaccharides or to another sialic acid residue. The sialic residues of gangliosides carry a net negative charge at neutral pH. For a more comprehensive
description of these and more exotic polar lipids found
in living organisms the reader is referred to the Lipid
Library maintained by W.W. Christie (http://www.lipidlibrary.co.uk/lipids.html).
3.2. Properties of the hydrocarbon groups
The long-chain fatty acids associated with membrane
polar lipids are generally unbranched, even carbon numbered (C14–C26) and possess cis-unsaturated double bonds
located at specific positions in the hydrocarbon chain. The
chemical convention is to locate the position of the unsaturated bond with respect to the carboxyl functional group
(D classification) or to the terminal methyl group (x classification). The biochemical steps involved in desaturation of
the hydrocarbon chain involves insertion of the first double
bond in the middle of a C-18 chain (D/x9), however, subsequent desaturations take place sequentially in x6 and
then x3 positions of the chain in plants and phytoplanktons. Rotation about these cis-unsaturated bonds is
restricted introducing a kink into the structure and preventing close packing of the adjacent hydrocarbon chains.
The physical consequence of the presence of cis-unsaturated bonds in the chain is to reduce the melting point of
the lipid and this is manifest in lipid assemblies such as
bilayers as a phase transition. The cis-unsaturated bond
hinders the close hexagonal packing of the chains associated with crystalline or gel phases and favours the liquidcrystalline or disordered fluid state.
With the notable exception of lipids in the purple membrane of Halobacterium, which adopt more or less a crystal
structure, the lipid matrix of biological membranes is said to
be fluid. The term fluidity is not a precise quantitative
parameter but it implies that the molecules of the structure
exhibit motion with respect to one another. This motion has
been investigated using a number of spectroscopic and
other techniques and the general picture that has emerged
19
is that lipid molecules are relatively mobile within the membrane and constraints on this motion occur by interactions
with the proteins and other membrane constituents.
Studies using membrane probes investigated both lateral
diffusion of lipids in the plane of biological membranes and
movement of lipids from one leaflet of the membrane lipid
bilayer to the other. Fluorescent derivatives of phospholipids and cholesterol have been introduced into human erythrocyte ghost membranes and living cells to measure lateral
diffusion rates in biological membranes and to compare
these with rates of diffusion in phospholipid bilayer membranes. Lateral diffusion rates have been obtained using a
spot fluorescence photobleaching recovery method. It was
found that fluorescent derivatives of cholesterol and phosphatidylethanolamine diffused rapidly with a diffusion
coefficient >1 lm2 s 1 in lipid dispersions at temperatures
greater than the gel to liquid-crystalline phase transition
temperature but the diffusion rate decreased dramatically
for probes in gel phase lipid (<0.01 lm2 s 1). Intermediate
values of diffusion coefficient were observed in erythrocyte
ghost membranes and in the plasma membrane of live melanoma cells. In the latter case only approximately half the
probe molecules were identified in the mobile fraction indicating that motion of the probe is restricted by interaction
with other membrane components. The order of diffusion
coefficients recorded for these probes is similar to that of
other fluorescent lipid probes in mammalian plasma membranes. Translational motion of lipid can also be determined from bimolecular collision frequencies that can be
measured by perturbation of the signals derived from
probe molecules. Such methods include fluorescence
quenching or excimer formation for luminescent probes
and spin–spin interactions between spin-labelled probes.
It is clear from these observations that the diffusion of
lipids does not take place through a homogeneous lipid
bilayer matrix and there is evidence from photo-activable
fluorescent probes of a non-homogeneous distribution of
lipids within the plane of the membrane. Studies of lateral
distribution of lipids in the bacterium Micrococcus luteus,
for example, have been undertaken using a reversible photo
cross-linker, anthracene phospholipid analogue [5]. The
results showed that the two major polar lipid components,
phosphatidylglycerol and dimannosyldiacylglycerol, are
not homogeneously distributed in the plane of the membrane. The lateral diffusion coefficient of about 0.1–
0.2 lm2 s 1 is in line with the above measurements.
In addition to translational motion, membrane lipids
undergo rotational motion about their long axes perpendicular to the plane of the membrane as well as motion
within the molecule. The latter involves rotational motion
of the polar group and trans-gauche isomerisation of the
hydrocarbon chains. Rotational diffusion rates can be
determined from decay of polarisation of fluorescence of
probe molecules excited by plane polarised light. The lifetime of the excited fluorescent state is in the order of a
few ns which is appropriate to the rotational relaxations
of lipids in biological membranes. Both time-resolved and
20
C. Wolf, P.J. Quinn / Progress in Lipid Research 47 (2008) 15–36
steady-state depolarisation methods have been used in
these measurements.
A novel fluorescence method has been reported recently
[6] in which fluorescent probes are located on the cell surface or within the membrane and their absorption anisotropy used to provide information on orientational
constraints within their local environment. The method is
able to detect changes in lipid organization in cell membranes on time scales faster than 1 s. One of the results
obtained showed that the presence of cholesterol in the cell
membrane exerts a considerable ordering effect on the surrounding lipids. The depletion of cholesterol from the
membrane had little effect on the orientation of the molecules in the surface region but was associated with a
marked transition to a more disordered environment in
the hydrocarbon domain.
The spontaneous movement of polar lipids from one leaflet of the bilayer to the other is a relatively slow event and
takes place on a timescale of hours or days. This is because
of the energy required to move the hydrated polar head
group from an aqueous environment into the hydrocarbon
domain in the center of the bilayer. This factor is responsible
for maintaining the asymmetric distribution of polar lipids
across biological membranes. The creation of phospholipid
asymmetry is an enzymic, energy-requiring process and the
dissipation of the asymmetry mediated by scramblases is
associated with a number of physiological processes such
as apoptosis. Active translocation of phospholipids across
the plasma membrane has been demonstrated both from
the inner to the outer leaflet and from the outer to the inner
leaflet of the plasma membrane. The translocation processes
specifically transport phosphatidylserine and phosphatidylethanolamine from the cytoplasmic to the outer surface of
the membrane while choline phosphatides are transported
from the outer to the cytoplasmic surface. The rate of translocation, in general, is greater for the amino phospholipids
compared with the choline phospholipids.
4. Biochemical homeostasis
Complex lipids found in subcellular structures are frequently assembled with ester bonds linking the individual
polar and non-polar components. These bonds are subject
to enzyme hydrolysis and the regulation of these enzymes
underlies the homeostatic mechanisms. Furthermore, the
specific activation of particular hydrolytic enzymes is a
common strategy for the conduct of a number of physiological processes such as signal transduction in membranes.
Since an understanding of the way lipids are metabolically
turned over is pertinent to interpreting analyses of the lipidome this section reviews current knowledge of hydrolases
in complex lipid homeostasis.
4.1. Phospholipases A2 (PLA2)
Hydrolases that attack the acyl ester bonds linking fatty
acids to the sn-1 and sn-2 position of the glycerol backbone
of phospholipid molecules are categorized by their positional specificity as A1 and A2, respectively. Membranebound and soluble forms of phospholipase A2 have been
identified and their interaction with phospholipid substrates characterised [7]. The notion that these enzymes
are involved in signal transduction by mediation of metabolic turnover of membrane phospholipids has been less
favoured by the discovery of cytosolic species of PLA2
whose action appears more appropriate to this function
[8]. It was also realized that phospholipids of the membrane matrix were relatively resistant to enzyme hydrolysis.
A number of mechanisms are responsible for the protection
of phospholipids against PLA2 attack. These include:
(a) limitation of the penetration of the enzyme into the
substrate due to the tight packing of the lipid
molecules;
(b) restricted access of the enzyme to its preferred substrate. For instance, exogenous secretory type-II
PLA2 outside cells cannot gain access to phosphatidylethanolamine substrate located in the inner leaflet
of plasma membranes;
(c) dilution of susceptible substrates within non-substrate membrane lipids such as sphingolipids and
cholesterol.
These limitations serve to restrict enzyme activity to a
relatively small proportion of the lipids forming the membrane lipid matrix leading to the conclusion that the hydrolysis of such a small number of molecules would be unlikely
to result in any physiological consequences.
The demonstration that cPLA2 is the particular phospholipase involved in the release of arachidonic acid from
the sn-2 position of phosphatidylcholine under conditions
where the cascade of reactions leading to eisosanoid biosynthesis is triggered has been fundamental to understanding the role of PLA2 in this process [8]. By contrast,
secretory type-II PLA2 is known to be calcium-dependent
with an optimum calcium concentration greater than
1 mM which is consistent with its preferred extra-cellular
activity.
Secretory type-II phospholipase A2 (sPLA2) is inhibited
by sphingomyelin [9]. Cholesterol, either mixed with the
model glycerophospholipid substrate or added to the assay
medium in separate liposomes, effectively counteracts this
inhibition. The inhibition of fatty acid release is also
observed when sphingomyelin is added to erythrocyte
membranes as substrate. Interestingly, it turns out that
the specificity for the release of polyunsaturated fatty acids,
mostly C20:4 for cPLA2, is a property that can also be
acquired by other types of PLA2 when the ratio sphingomyelin/cholesterol is manipulated. When the ratio is
decreased from 10 to 1 in the lipid mixture serving as
the substrate the release by sPLA2 of C20:4 relative to
D9-C18:1 increases from 1.5 to 2.1. Such evidence serves
to exemplify how the manner of presentation of substrate
to the enzymes is able to modulate hydrolytic activity.
C. Wolf, P.J. Quinn / Progress in Lipid Research 47 (2008) 15–36
4.2. Acyltransferases
Acyltransferases are enzymes that catalyse the exchange
of esterified fatty acids from one lipid to another [10]. An
acyltransferase activity able to reverse the action of PLA2
was initially described in membranes of the endoplasmic
reticulum [11] and they are largely confined to this membrane in eukaryotes. Activity of the enzyme is regulated
by sphingomyelin [12], whereas in prokaryotes such as
Escherichia coli, modulation occurs by the lipid environment of the substrate [13]. Acyltransferases are wide ranging with respect to their dependence on intact membrane
phospholipids. In the case of 1-acylglycerophosphate and
1-acylglycerylphosphorylcholine acyltransferase systems
the enzymes do not require specific phospholipids such as
phosphatidylcholine, phosphatidylethanolamine, and phosphatidylinositol for their catalytic activities, but the diacyl
phospholipids must be intact for the proper functioning
and stability of the enzymes.
The activity of acyltranferase in the plasma membrane of
liver cells has been suggested to play a major role in the
secretion of bile lecithins [14]. Phospholipid acyltransferase
activities of plasma membranes have been investigated with
various acyl-CoA thioesters (palmitoyl, stearoyl, oleoyl, linoleoyl and arachidonoyl) with and without added lysoderivatives. Different patterns of incorporation were observed for
each acyl-CoA into endogenous phosphatidylcholine and
phosphatidylethanolamine. The turnover rates calculated
with trace amounts (10 lM) of acyl-CoA thioesters were
five times faster for the polyunsaturated than for the
saturated acyl molecular species of phosphatidylethanolamine and phosphatidylcholine. Arachidonoyl-CoA was
the most efficient acyl donor at low concentrations and maximal turnover rate was observed at about 25 lM. No saturation was observed at concentrations at up to 100 lM
linoleoyl-CoA. Linoleoyl-CoA transacylase acylated the
lyso-compounds in the following order: lysophosphatidylcholine > lysophosphatidylserine = lysophosphatidylinositol. Lysophosphatidylethanolamine was found to inhibit
linoleate incorporation into the phosphatidylethanolamine.
No satisfactory explanation was given for this effect so it
remains uncertain as to whether it is connected with an
interference of translocators for lysoderivatives and phosphatidylethanolamine. Linoleoyl-CoA transacylation was
not affected by the fatty acyl moiety at the 1-position of
the lysophosphatidylcholine.
21
cycle is via the acyltransferase reaction rather than the
phospholipase in resting tissue. In stimulated tissues such
as granulocytes treated with phorbol myristate acetate
there is a selective incorporation of arachidonic acid into
phosphatidylinositol which is thought to reflect the degradation that produces the eicosanoid precursor. The high
specificity of the recipient lysoderivative has also been
examined in blood platelets [16]. The transfer of arachidonate to 1-alkyl-2-lyso-sn-glycero-3-phosphocholine is of
importance in the termination step for platelet activating
factor (1-alkyl-2-acetyl-sn-glycero-3-phosphocholine) activity, whereby 1-alkyl-2-arachidonoyl-sn-glycerol-3-phosphocholine (a stored precursor of both platelet activating
factor and arachidonic acid metabolites) is restored.
In alveolar macrophages the transacylation system was
shown to exhibit a complex selectivity according to distinct donor and acceptor and CoA dependency [17]. It
was shown that acylation can be a specific and active
pathway for polyunsaturated fatty acids cleaved from
the sn-2 position of phospholipids serving as substrate
during activation of the cell. Thus in activated neutrophils
the circulation of arachidonate between alkyl and alkenylderivatives participate in the generation of the lyso-derivative precursor of platelet activating factor [17]. The
circulation of polyunsaturated fatty acids between
alkylPC and lysoPAF via alkenyPE has been demonstrated in a variety of tissues [18,19]. The transfer of acyl
groups by the transacylase appeared to be equally effective for either arachidonic or docosapentaenoic (x6) fatty
acids, whereas linoleic and oleic acids are not readily
transferred. PAF-related transacylase is widely distributed
among tissues and, although highly selective for polyunsaturated acyl groups it does not discriminate selectively
among the polyunsaturates.
A particular role has been ascribed to the deacylation/
reacylation of lysoderivatives on the inner monolayer of
the erythrocyte membrane. This is the maintenance of the
highly asymmetric distribution of phosphatidylethanolamine (PE) in ruminant membrane and the relatively low
content of phosphatidylcholine (PC) [20]. The acylation
of PE is by far the most important biosynthetic event in
this cell following deacylation by phospholipase A2. The
selective reacylation of lyso-PE on the cytoplasmic side of
the membrane can account for the asymmetry of PE distribution. Moreover, the removal of lyso-PC extracted by
serum albumin can account for the low content of PC in
bovine erythrocyte membranes.
4.3. Acylation–deacylation cycle
4.4. Phospholipases C
Few details have been published on the relative activities
of PLA2 and acyltransferase in relation to the homeostatic
regulation of membrane lipids. Regulation of phosphorylation/dephosphorylation of key enzymes was suggested to
be a key factor from studies of the incorporation of 14Cpalmitoyl CoA into membrane phospholipids via the deacylation/acylation cycle conducted in rat liver microsomes
[15]. It was found that regulation of deacylation–acylation
Phospholipases C are enzymes that hydrolyse diacylglycerophospholipids to diacylglycerols and a water-soluble
phosphorylated product. There are four main families of
mammalian PLC. These include PLC-b, -c, -d and -e all
of which are characterised by their structural organization
and mechanism of regulation [21,22]. All families are regulated by numerous cell regulators but some are specific to
22
C. Wolf, P.J. Quinn / Progress in Lipid Research 47 (2008) 15–36
particular phospholipases. The PLCb family are uniquely
regulated by heterotrimeric G-proteins; PLCc enzymes
are regulated by both receptor and non-receptor tyrosine
kinases; PLCd are modulated by agents which include
RhoGAP and aH; PLCe enzymes contain a GTP exchange
factor and are regulated by RAS that interacts via two
RAS-binding domains [23,24].
Phospholipase C are enzymes that specifically hydrolyse
phosphatidylinositol 4,5-bisphosphate (PIP2) to form diacylglycerol and inositol-1,4,5-trisphosphate. The enzymes
are activated to differing extents by G-protein aq subunits
and by G-protein bc dimers [21]. It is known that the C-terminal region of PLCb is essential for stimulation by aq but
the phospholipase can still be activated by Rho GTPases
and G-protein bc subunits that bind to different regions
of the enzyme [25,26]. Thus the catalytic subdomains of
PLCb2 are all that is required for efficient stimulation by
bc dimers, whereas additionally the putative pleckstrin
homology domain is required for stimulation by Rho GTPases. Amongst the Rho GTPases, Rac 1 and Rac 2 were
found to be more important stimulators than Cdc42 and
all are implicated in receptor-mediated stimulation of
PLCb2 activity. The molecular mechanisms of stimulation
by either heterotrimeric G-proteins or Rho GTPases is
presently unknown but targeting of the enzyme to the substrate and allosteric regulation of the enzyme are both factors [27]. The role of pleckstrin homology domains serve to
modulate the activity of their catalytic sites upon either
interaction with the substrate or G-protein activators [28].
The PLCc family, like PLCb acts on PIP2. Two isoforms
have been identified, PLCc1 which is widely distributed in
mammalian tissues, and PLCc2 the expression of which is
restricted mainly to haematopoieic cells [29]. PLCc1 has a
function essential to growth and development [30]. The
molecular features that distinguish PLCc from other PLC
isotypes is the presence of two Src homology domains
located within a pleckstrin homology domain which are
responsible for localization at the membrane substrate
and an activation by phosphorylation of multiple tyrosyl
residues on the enzyme mediated by tyrosine kinases at
sites that are system-dependent. The regulatory process
controlling activation of PLCe is mediated by two Ras/
Rap-1 associating domains located at the C-terminus of
the molecule and a CDC25 homology domain near the
N-terminus [24,31].
4.5. Phospholipases D
Phospholipase D hydrolyses phospholipids yielding
phosphatidic acid and water-soluble base such as choline
[32]. Two genes coding for phospholipase D have been
identified in mammals, PLD1 and PLD2. Both genes have
been cloned and overexpressed in different cell lines [33].
PLD1 is located in association with intracellular membranes and is known to be active in living cells. PLD2,
by contrast, is associated with plasma membranes but has
low resting activity that can be activated by a variety of
factors including protein kinase C, family members of
ADP-ribosylation factor and Rho and the lipid PIP2
[32,34]. Targetting of PLD1 to membranes may involve
palmitoylation of cysteine residues at a domain near the
C-terminus of the protein [35]. The binding of substrate
phosphatidic acid to modulator proteins like Raf-1, a serine/threonine kinase [36], c-AMP-specific phosphodiesterase [37], the mammalian target of rapamycin, mTOR
[38], protein phosphatase-1 [39] and Src homology 2domain containing protein tyrosine phosphatase [40] are
thought to control enzyme activity. Recently, a solid-phase
adsorption system has been described to identify trafficrelated phosphatidic acid binding proteins and proteins
that may be implicated in phospholipase D dependent
pathways and a number of specific proteins have been characterised [41].
4.6. Sphingomyelinases
Sphingomyelinases are enzymes that cleave the phosphorylcholine moiety from sphingomyelin to yield lipophilic
ceramide. Five different categories of sphingomyelinase
have been characterised on the basis of their optimal conditions for catalytic activity [42] although a different sphingomyelinase has been identified in bacteria [43].
Evidence from model membrane studies has indicated
that the action of sphingomyelinase in generating ceramide
from sphingomyelin can induce lamellar to non-lamellar
phase transitions leading to membrane fusion [44] and to
lateral phase separations of ceramide-enriched domains
which exist in gel phase in fluid bilayers of phospholipid
[45]. Ceramide, but not dihydroceramide, is also able to
induce the formation of pores in phospholipid bilayers, a
property that has been attributed to the extensive hydrogen-bonding capacity of ceramide [46].
There is evidence that the action of neutral sphingomyelinase causes clustering of L-selectin in lymphocytes [47].
The action of a Zn-dependent acid sphingomyelinase in
response to interleukin-1b treatment of human fibroblasts
has been found to be associated with depletion of sphingomyelin and a corresponding increase of ceramide in the
caveolae compartment [48]. Similar studies of the action
of nerve growth factor signalling have emphasized the
importance of intact caveolae for sphingomyelinase action
[49].
Ceramide generation and creation of rafts has been
shown to be essential for optimal Fas signalling and induction of apoptosis in both B- and T-lymphocytes [50]. On
the basis of these studies a model has been proposed for
the action of ceramide in signalling processes associated
with apoptosis [51]. Essentially, the engagement of Fas triggers translocation of acid sphingomyelinase to the plasma
membrane where it acts on its substrate segregated into
sphingomyelin-cholesterol rafts. The formation of ceramide induces coalescence of the rafts into large domains
in which oligomerisation of downstream effectors such as
C. Wolf, P.J. Quinn / Progress in Lipid Research 47 (2008) 15–36
FADD/MORT-1 and pro-caspase-8 can take place leading
to the Fas death signal.
5. Sample preparation
Interpretation of the biochemical and biophysical properties of lipids in membranes in respect of their cellular
functions relies on the preparation of highly pure, morphologically-distinct membranes, or increasingly relevant, subfractions of membranes. This requires destruction of the
integrity of organs, tissues and constituent cells to release
the subcellular membranes in a form they can be subsequently fractionated. The objectives that must be met in
order to achieve satisfactory yields are that the destructive
step should be sufficiently rigorous to break down the cells
into their subcellular organelles but mild enough to preserve the integrity of vesicular organelles that encapsulate
hydrolases or other enzymes capable of degrading the
membranes. In the main these organelles are the lysosomes
that contain lipid hydrolases whose action on membrane
organelles would alter their biochemical composition.
Other strategies commonly used to prevent biochemical
changes in membrane fractions are to perform the operations quickly, to work at temperatures close to freezing
and to add inhibitors of hydrolytic enzymes and
antioxidants.
5.1. Homogenization
The, so-called homogenization step differs from one tissue to another and has been established in each case by
operational criteria, namely, that method which optimally
achieves the above criteria is refined and used. The method
of choice is influenced by such factors as the presence of a
tough extracellular matrix or a cell wall of cellulose or similar material. In such tissues, pretreatment with collagenase
or cellulase as appropriate may render the tissue more susceptible to milder homogenization than would otherwise be
the case. One unavoidable factor in preparing membrane
fractions is their dilution in an excess of buffer. The solute
environment is also drastically altered from the environment experienced by the membrane in the living cell. To
ameliorate these changes buffers are used to mimic the
pH and solute composition that are most compatible with
preservation of the membranes as they exist in vivo. Ultimately, the homogenization step must result in a high relative yield of a specific organelle with its composition
and functional integrity intact.
A variety of homogenization methods have been devised
based on mechanical (Potter-Elvehjem, rotating blade
devices such as Waring blendor and Ultraturax, glass
beads) or other shear forces (pressure/cavitation, freeze–
thaw). Most are capable of causing artefactual reshuffling
of membrane subfractions or rupturing organelles and
must be employed with due regard to these effects. As a
result of chimerical fusion of naturally separated domains
in membranes ‘‘mixed’’ preparations may be obtained. It
23
is thought that mixing can only be partially circumvented
but, hopefully, the excessive mechanical stress on tissues
and cells producing an artefactual membrane subfraction
will be eventually detected applying ion beam- or
MALDI-TOF mass spectrometry [52,53] in vivo.
Some typical problems are evident with homogenization
using a loosely fitted Potter homogenizer when a low yield
results from formation of clumps representing associations
of different organelle membranes observed as extended
membrane leaflets together with small vesicles. Such
clumping can be avoided by using buffers of high ionic
strength and/or the presence of chelating agents like EDTA
but the effect of such buffers is likely to cause desorption of
peripheral proteins that are normally associated with the
membrane.
Another example of problems associated with use of
inappropriate buffer composition for isolating membrane
subfractions is the profound influence of the ionic strength
and divalent cations on lipid composition of detergent
resistant subfraction of membranes (DRM) prepared from
rat brain [54]. Ions including divalent cations appear to
participate in the creation of a delicate scaffold network
of lipids held together by binding and bridging polar headgroups. In turn this modulates the activity of the detergent
and prevents the micellization of the cytosolic membrane
leaflet. Furthermore, the temperature at which the detergent treatment is performed changes the compactness of
the lipid phase and thereby influences the penetration of
the detergent into the hydrophobic bilayer core. Various
detergents with distinct hydrophilic/hydrophobic balance
also produce different effects that cannot readily be predicted from standard physicochemical parameters. For
example, detergent treatment of membranes with detergents of different critical micellar concentrations such as
Triton X-100 and Brij98 produce detergent resistant membrane fractions differing in terms of lipids represented in
the inner and outer leaflets of the original membrane [55].
5.2. Membrane fractionation
The differential and isopycnic centrifugation steps are
applied sequentially. Flotation of membrane fractions at
the interface of layers with distinct density allows the recovery of intermediate density fractions in a small volume. The
isolation of pure membrane fractions is the objective but in
practice an enrichment as detected by marker enzyme
assays is the best that can be achieved. Marker enzymes
are used because they are unique to particular morphologically distinct membrane. They are intrinsic membrane proteins with the exception of enzymes located within
lysosomes and peroxysomes. They are robust to the operations performed in membrane isolation and can be assayed
conveniently. The value of biochemical assays is that the
extent and origin of contamination of membrane isolates
can be established accurately. Lipidomics can also be helpful to characterize membrane fractions, for example, to
24
C. Wolf, P.J. Quinn / Progress in Lipid Research 47 (2008) 15–36
establish the cholesterol/sphingomyelin ratio in detergent
resistant membrane fractions [56,57].
Affinity chromatographic and polymer phase separation
methods have been described for separation of subcellular
membrane fractions. In these systems, when two structurally different polymers in aqueous media such as polyethylene glycol and dextran are mixed two phases are formed,
an upper layer enriched in polyethylene glycol and a lower
layer enriched in dextran [58]. Membranes of different subcellular origin then partition between the two phases
according to their surface charge or hydrophilicity. Plasma
membranes, for example, have a higher affinity for the
upper layer and can be highly enriched compared with
microsomal membranes [59]. Polymer partition has been
described in the separation of plasma membranes from
rat liver [60] and crude rat brain microsomal membrane
preparations [61]. A scheme for the polymer partition purification of membranes is illustrated in Fig. 1. In the example of brain plasma membrane the N-acetyl-D-glucosamine
and sialic acid residues located exclusively on the outer surface of plasma membrane, but not of other microsomal
membranes, were strongly bound to wheat-germ agglutinin. This results in a selective partition of the right side
out plasma membrane vesicles into the wheat-germ agglutinin-dextran-enriched lower phase. In contrast, all other
membranes remain in the polyethylene glycol-enriched
upper phase. The separation of an enriched inner membrane fraction from a crude membrane mixture obtained
from Escherichia coli has been reported using two-phase
partitioning in tandem with affinity to agarose beads
coated with nickel-nitrilotriacetic acid [62]. It was shown
that an interaction between the beads and an intrinsic protein exposed on the membrane surface cause the adherent
membranes to selectively partition to the lower phase of
a polymer/polymer aqueous two-phase system consisting
of polyethylene glycol and dextran.
Immunoisolation of membrane fractions have been
reported that employ antibodies specific for membrane-
Tissues and cells
Homogenization of tissues or cells
/ membrane preparation
Addition of antioxidant
Solvent extraction of
lipids
Addition of Internal
Standards
Storage/Concentration
of extracts
Direct Infusion
in EPI/APCI source
LC separation of polar
lipid classes
Separation of molecular species
on reverse stationary phase
MS
(single stage)
MS survey of lipid classes
fragment ion or neutral loss
MS/MS determination
of molecular species
structure
Fig. 1. A flow diagram of the overall lipidomics procedure.
C. Wolf, P.J. Quinn / Progress in Lipid Research 47 (2008) 15–36
specific antigens which are bound to beads to select the
membrane fraction. The method has been used to isolate
peroxisomes [63,64], tubulovesicles [65], Golgi membranes
[66], microsomal membranes [67], plasma membranes
[68,69] and detergent-resistant membrane fractions from
brain [70]. The method for isolating plasma membrane
from mouse liver microsomes involved an initial purification by sucrose density gradient centrifugation followed
by immunoprecipitation on magnetic Dynabeads coated
with anti-flotillin and anti-Na+–K+–ATPase, both plasma
membrane-specific proteins [71]. Such purification resulted
in a threefold enrichment of plasma membrane and a twofold reduction in contamination from mitochondria compared with the density gradient step and there were
undetectable levels of endoplasmic reticulum or nuclear
proteins.
5.3. Lipid extraction
Lipidomics studies often refer to studies of crude lipid
extracts inferring that there is no requirement for the complete separation of protein, sugar and lipid classes in the
sample subject to analysis. Tandem mass spectrometry
has considerably simplified the pre-analytical steps as compared to relatively ‘‘low tech’’ procedures such as thin layer
chromatography. This is explained by the high specificity
of MS and weak influence of the matrix for monitoring
the characteristic transition of a parent molecular ion to
a fragment product (tandem MS). Electrospray (ESI) with
intrasource separation has led to strategies using 2D mass
spectrometry in, so called, shotgun lipidomics and quantitation of cellular lipidomes directly from ‘‘crude’’ extracts
of biological samples [72]. The intrasource separation procedure is based on the activity of an external electric field to
induce separation of cations from negative ions in the infusate while different ionization of molecular species that possess differential electrical propensities can be induced in
either the positive- or negative-ion mode during the electrospray ionization process. Intrasource separation and selective ionization are expected to simplify lipid purification
prior to MS and result in greater accuracy of the analysis
[73]. Pre-analytical separation steps and enrichment of lipid
sample, however, are still required in many cases prior lipidomic final analyses.
Ion suppression, for example, is a major complication
with phospholipids. It can be viewed as the competition
for ionization of molecules introduced simultaneously in
the ESI source. Phospholipids have a high propensity to
suppress ionization of coeluting molecules such as drugs
[74] or gangliosides [75]. In biological samples, abundant
molecular species also compete with less abundant species
and examination of diluted samples is recommended for
characterization of minor species of phospholipid. Moreover, the ionization is dependent on the saturation and acyl
chain length of phospholipid molecular species. This
dependence is a function of the total lipid concentration
[76]. Inclusion in the infusate of ammonia or solvent-solu-
25
ble trimethyl-, diisopropyl-amine or piperidine enhances
phospholipid ionization in the negative mode and displaces
Na+ and K+ counterions in the positive mode which simplifies quantitation. To maintain equipment performance
during multiple sample analyses it is recommended that
extracts are clarified. Indeed, insoluble lipid precipitates
are frequently observed in solvent mixtures used for chromatography or infusion where high concentration and limited solubility occurs in the solvent system required for
HPLC separation. Precipitation can be reduced by heating
the sample vial, connection tubes and column with heating
tape and in a thermostated oven at around 45 °C. It should
be noted that temperature may profoundly alter the performances of chromatography systems and modifies the retention times of components eluted from columns. On a
reverse stationary phase grafted with octadecyl (C18) acyl
chains the temperature reduces the troublesome ionic interactions and helps to maintain peak symmetry, so reducing
the need for ionic pairing [77]. On the other hand on polar
stationary phases (e.g. diol-silica), retention is delayed and
useful degrees of resolution can be achieved at the expense
of analysis time.
In biological materials such as membranes and lipoproteins lipids and proteins are associated by hydrophobic and
polar bonds. At the lipid/water interface hydrogen- and
ionic-binding between headgroups are prevalent. Inside
the bilayer core hydrophobic binding is responsible for
the compactness and ordering of acyl chains and sterol
rings. Therefore both polar and non-polar solvents should
be used for extraction of lipids bound to proteins. Direct
treatment by non water-miscible solvent (isohexane) and
heating is used for oil-seed processing. However, rapid protein precipitation and extensive trapping of the lipids occur
applying this method to small tissue samples initially contained as an aqueous suspension.
Quantitative extraction requires a gradual procedure to
separate the lipids from proteins by a method that avoids
the rapid denaturation of the proteins. The clumping into
aggregates of denaturated protein reduces lipid extractability into non-polar solvents. A comparison of solvent activities and a description of the most common lipid extraction
procedures can be found in the Cyberlipid website:
(www.cyberlipid.org). Briefly, an aqueous suspension of
lipid-containing biological material is firstly mixed with a
polar solvent such as methanol (or isopropanol or ether).
In a second step, the less-polar solvent (chloroform, hexane) is added in a proportion which forms a single phase.
In this environment proteins form a fine precipitate which
does not tend to clump and hinder lipid extractability. Lipidomics studies by MS may be conducted directly from this
solvent mixture. However, a subsequent step is usually the
partitioning of the previous extract mixture into two
immiscible layers [78,79] after addition of an extra volume
of aqueous buffer. This partition helps to separate watersoluble constituents acting as an ESI suppressor (proteins,
sugar and salts) into a phase separated from the solvent
containing the water-insoluble lipids. Ionic strength and
26
C. Wolf, P.J. Quinn / Progress in Lipid Research 47 (2008) 15–36
acidic buffer pH favour cleavage of anionic lipids (e.g.
PIPx, gangliosides) bound to basic protein residues. Anionic lipids, however, tend to partition into the aqueous-alcoholic layer at pH greater than 4. Excessive acidification
results in the cleavage of vinylether bond of plasmalogens.
In tissues where both acid-resistant and plasmenyl lipids
are to be examined two separate extracts should be prepared. In solvent mixtures with chloroform most of the lipids are extracted into a lower dense layer which separates
from an aqueous-alcoholic upper layer. Proteins accumulate at the aqueous/chloroform layer interface with
sequences oriented according to their hydrophilic/hydrophobic affinities. Indeed a turbid suspension is observed
if abundant proteins or amphiphiles of the biological preparation stabilize the interface of chloroform droplets.
When emulsions are formed (e.g. in some lipid extracts
from brain or from plants) the separation of the upper/
lower phases is long lasting and difficult. Protocols derived
from the Folch procedure [78] have been devised to speed
up the separation of layers such as centrifugation, cooling,
alteration of chloroform/methanol ratio, or addition of saline buffer. When the lower chloroform layer is turbid it is
recommended that the extract is percolated through a
dehydrating agent to remove water. Barium oxide or anhydrous sodium sulphate is often employed for this purpose.
Contamination by water jeopardizes the integrity of lipid
extracts after solvent evaporation because a residue of
acidic water remains which decomposes vinyl ether bonds.
Lipid oxidation is also enhanced at the air interface
because evaporation of water takes a long time.
To minimize degradation and set up automatic pre-analytical facilities for a simultaneous treatment of numerous
samples, methods have been described using an automat
robot with solid phase extraction (SPE) protocols in which
there is no need for partition of the solvent/water mixture.
Extraction by solid phase extraction reduces the volume of
solvent required. The principles of lipid SPE have been
reviewed elsewhere [80]. Stationary phases are designed
for quantitative extraction of phospholipids on reversed
phase, or separation of neutral/phospholipid on straight
phase [81]. The recovery of anionic lipids has to be assessed
against standards for polar stationary phases.
5.4. Sample storage
LC–MS analysis has a high sample throughput relative
to the methods described previously for the lipid extraction. Samples are pooled after extraction and stored before
analysis as a series. The storage of lipids in organic solvent
is critical if not protected against chemical alteration.
Intrinsic protective mechanisms and association with natural compounds act in living organism to defend the vulnerable double bonds of vinyl ether and polyunsaturated fatty
acids against attack by reactive oxygen species. The
removal of these lines of defense by isolation and extraction of lipids requires addition of antioxidants. It is therefore usual to add a radical scavenger such as butylated
hydroxy-toluene (BHT, 2,6-di-tert-butyl-p-cresol) and/or
tocopherol to the extract. Being soluble in solvents of low
polarity they can be easily added to the extract and their
oxidation can act as a potential index of damage to the biological lipids. Nevertheless, it is necessary to ensure that the
concentration of such agents does not exceed a threshold of
0.00001 (w/w) when antioxidants like BHT have been
shown to act as prooxidants at high concentrations [82].
Oxidation can also be exacerbated by contamination with
heavy metals such as copper or/and iron which catalyse
production of free radicals in Fenton-type reactions. Contamination can be avoided by using single-use vessels with
PTFE fittings and it is recommended for all pre-analytical
steps in Lipidomics. Exposure of lipid extracts to UV light,
for example, when performing operations on a UV-illuminated sterile bench, can be particularly deleterious. Elevated temperatures also favour double-bond migration
and diene conjugation. The isomerization proceeds by an
initial abstraction of an electron from an intermediary –
CH2 methylene group between double-bonds and induces
an extensive peroxidation especially in concentrated solutions. It is ideally recommended that lipid extracts are
stored at low temperatures ( 20 °C) in a diluted solvent
solution contained in an opaque vessel purged with argon
gas and sealed by PTFE tape. Evaporation of the solvent
prior to analysis is performed under a stream of oxygenfree dry nitrogen at a temperature not exceeding 40 °C.
SpeedVac is an effective alternative to this method since
numerous samples can be evaporated simultaneously in
vials which can be subsequently transferred to an automatic sampler module of an HPLC system. Up to 40 samples can be evaporated simultaneously before loading into
the cooled module. Cooling of the injection module prevents evaporation of solvent after the septum is perforated
by the sampling needle allowing re-injection of sample for
analysis by different data acquisition modes. Caution
should be exercised, however, as temperatures lower than
about 10 °C may result in precipitation of the fraction of
the most insoluble lipids concentrated in the type of solvent
system used for HPLC. This results in the splitting of elution peaks into micellar and monomeric compounds.
Application of solubilized lipids in efficient solvents like
chloroform/methanol mixtures (1/1 or 2/1) can interfere
with retention on the silica gel and the metal tubing of
the equipment may deteriorate due to exposure to HCl
formed from chloroform.
Using double layered TPFE-sealed vials and extensive
purging with argon samples of highly polyunsaturated
glycerophospholipids were found to remain unchanged
for up to 6 months storage at subzero temperatures. Sphingolipids are expected to be stable for much longer periods
of time since they are generally more saturated. Sterols, steroids and bile acids are also innately resistant to chemical
alteration as indicated by analyses of paleofecal specimens
[83,84]. By contrast, compounds such as 7-dehydrocholesterol or ergosterol with conjugated double bonds in the
B-ring require more cautious handling amongst which is
C. Wolf, P.J. Quinn / Progress in Lipid Research 47 (2008) 15–36
recommended low illumination, inert atmosphere and analysis as soon as possible after isolation.
6. Factors underlying a lipidomic analysis
Before the advent of tandem MS methods of lipid analysis relatively large samples were required to achieve a
fairly simple analysis. By way of comparison with these
earlier methods, which employed a separation of lipid classes by thin layer chromatography, phosphorus assay and
determination of fatty acids by gas chromatography performed on samples scraped from the plate, approximately
100 lg of each lipid class was required for structure confirmatory analysis. An alternative method was the TLC separation of classes, the cleavage of the headgroup by
phospholipase C followed by the derivatization of the diacylglycerol by a fluorochrome or a benzoyl group. This
tedious procedure required multiple steps before HPLC
separation. However, the method was successfully applied
for metabolic studies using radioactive tracer incorporation
monitored by liquid scintillation counting simultaneously
and by fluorescence or absorption methods coupled in
series.
Clearly, the accuracy and scope of these methods are
limited by the multiple handling steps that are required
and the amount of sample needed to perform other than
a cursory structural analysis. Such issues have been
addressed in current lipidomic analysis.
6.1. Sample size
To judge the power of current lipidomic techniques, a
conventional tissue culture flask containing in the order
of 106 cells yields about 100 nmol of lipid in a solvent
extraction. This amount of sample is generally sufficient
for a lipidomic analysis in which detection and structure
assignment can be achieved in the picomole concentration
range with a combination of enhanced ionization of parent ions in the electro-spray-ion source (ESI source) and
fragmentation by collision induced dissociation. Chemical
derivatization of lipids can be also combined. Derivatization of amino-phospholipids PE and PS by trinitrobenzene sulphonic acid (TNBS) is a promising strategy to
increase sensitivity of the detection down to picomole
amounts. The conditions of the complete derivatization
are well-defined. For example, identification of very small
amounts of molecular species of PE exposed on the outer
leaflet of the plasma membrane of living cells are defined
under conditions where the reagent is impermeant to the
cell membrane [85,86]. For ‘‘neutral’’ lipids such as sterols
chemical derivatization may be used to obtain enhanced
yield for ionization and high sensitivity with the ESI
source [87] without recourse to the atmospheric-pressurechemical-ionization (APCI) source. Addition of amine
base to the phospholipid sample also enhances ionization
in the negative ionization mode by inducing an efficient
release and transfer of slightly acidic protons. Under such
27
conditions selective monitoring of a transition from a parent ion to a specific product ion (a fatty acid) becomes a
reliable method of quantitation in the picomole range.
Indeed, multiple transitions are simultaneously monitored
by multiple reaction monitoring (MRM) acquisition methods and hundreds of distinct species can be quantitated in
this way during a single chromatographic run. To achieve
a sensitive detection, distinct groups of parent > product
ion pairs are separated in the successive periods of time
corresponding to elution of the different lipid classes from
the chromatography step. Indeed, each distinct parent
ion > product ion pair is acquired by the tandem MS during a short period, the dwell time, during which the fraction is being supplied by infusion or liquid chromatograph
and separated by the spectrometer. Using a dwell time of
30–50 ms per transition and the survey of 300 distinct
pairs, around 6–4 data points can be acquired per minute
per transition. A chromatographic peak with a width of
3 min will be reconstructed by the software from 18–12
experimental data points. The range 30–50 ms per transition is considered in recent equipment as the minimum
sampling frequency consistent with reliable fits of the
data. Multiple periods are programmed along the elution
profile to improve the reliability of quantitation by
increasing the number of data points (decreasing the number of transitions to be monitored) and keeping the dwell
time sufficiently long to collect counts per second (cps)
with high signal/noise ratio. For assignment of chemical
structures by the characterization of fatty acids, fragment
ions (FA-carboxylate with m/z 200–400) are recorded in
the negative ionization mode. Commonly a ‘‘full’’ spectrum (m/z scanned from 150 to 900) is recorded which
comprises in the negative mode the parent ion, the lysoderivatives, the FA-carboxylates and a characteristic fragment indexing the headgroup ( 196 for PE, 241 for PI,
153 for PA, PG or PS). The negatively-charged fragment
ion with m/z = 153 corresponds to a negative fragment
ion of glycerophosphate and is common to all the glycerophospholipids but the fragment ion is more conspicuous
in the mass spectra of PA, PG and PS. It is especially
abundant in spectra of PS where it provides a more sensitive method of detection which is required in mixtures of
biological lipids where the proportion of PS is relatively
small. The corresponding neutral loss 87 fragment can
be used to confirm the assignment of PS. This procedure
reveals that a combination of 2 or 4 FAs may be explained
by co-elution of distinct molecular species of phospholipids from HPLC. ‘‘Full scan’’ acquisition requires a ‘‘relatively’’ long scan time (m/z 150–900 takes 1 s at medium
resolution for an ion trap MS). This means that if the
‘‘slow’’ acquisition method is coupled to MRM the high
sampling frequency which is required to optimize the latter mode of assay deteriorates. In another strategy, reinjection and separate acquisition methods of the
phospholipid sample may be run to reconcile the quantitative approach (MRM) with the structure determination
(molecular species resolution by full scan).
28
C. Wolf, P.J. Quinn / Progress in Lipid Research 47 (2008) 15–36
The determination of lipid structure as compared to
MRM assay requires injection of larger amounts of sample
to obtain sufficient signal/noise ratios for ‘‘diagnostic’’
fragment ions with a low abundance such as lyso-derivative
fragments (parent ion–Fasn1orsn2–H2O). Approximately a
nanomole of lipid is required if structural confirmatory
studies are to be undertaken. If quantitative measurements
by MRM for assay and chemical structure determination
by product ion analysis are sought the methods may be
run subsequently. As an example, for a biological lipid
sample comprising a mixture of 300 different molecular
species, each being detected with nanomole sensitivity, five
successive runs require a starting amount of total lipids of
around 125 lg. As noted above, previous analytical methods involving separation of lipids by thin layer chromatography, phosphorus assay and determination of fatty acids
by gas chromatography require approximately 100 lg of
each lipid class.
6.2. Sample choice and handling
Factors that need to be considered when undertaking a
lipidomic investigation can be exemplified in a clinical context by monitoring the essential fatty acid status of a
patient group. A first step is to decide the most appropriate
tissue that is to be subject to analysis. The alternative is tissue biopsy, fluid (e.g. serum) or blood cells. Blood tissues
are the most convenient for analysis because obtaining
samples is less invasive and samples can be taken at frequent intervals for monitoring purposes.
The physiological state of the subject must be standardized because blood lipids vary accordingly. Lipids are
found in the membranes of blood cells and serum lipoproteins. Lipids found in the various lipoproteins in the form
of phospholipids, triglycerides and cholesterol esters have
a distinct origin. Following a meal containing fats or after
an infusion of a fat emulsion supplement, chylomicrons
and remnants thereof are present in the blood. The complete clearance of chylomicrons and remnants requires a
minimum of about 10 h, e.g. fasting overnight in adults.
This requirement obviously cannot be satisfied in normal
or sick new-born babies receiving milk every 3–4 h or
supplements.
Lipids from the three major lipoprotein classes,
VLDL, LDL and HDL, can be prepared from fasting
subject serum using sequential flotation methods. The origin of lipoprotein lipids is different; VLDL secreted by
the liver carry triglycerides packed around with mostly
choline-containing phospholipids that serve as surfactant
vectors. LDL enriched in cholesterol is produced after
degradation of VLDL by lipoprotein lipase in adipose
and muscle. By contrast nascent HDL contains amounts
of choline-containing phospholipids to take up and convey cholesterol from the peripheral tissues back to the
liver. The lipids transported in the serum are circulating
continuously between lipoprotein classes as well as to
and from the peripheral tissues. For instance, triglycerides
and cholesterol esters are exchanged in the serum by cholesterol triglyceride exchange protein (CTEP) between
HDL and VLDL. It means that phospholipid profiling
to differentiate lipoprotein classes is a particularly difficult
task. Phospholipids are also altered by circulating
enzymes such as secreted phospholipases A2 (type II
and V) and acyltransferase such as lecithin cholesterol
acyl transferase (LCAT). Finally, the phospholipid composition of lipoproteins is highly dynamic and usually
not in steady-state equilibrium as a function of time.
Therefore rapid changes of lipid are observed. Because
of these physiological and temporal dependencies of the
lipids present in serum lipoproteins it is difficult to characterize the essential fatty acid (EFA) status from these
serum components.
The phospholipids of the erythrocyte membrane, on the
other hand, are maintained close to steady state equilibrium and there is a relatively slow turnover of molecular
species. Therefore the erythrocyte fatty acids found in
phospholipids give a reliable retrospective view of the
EFA status of the patient. The life-time of circulating
erythrocytes is 4 months except if a regenerative anemia
increases the turnover. Within the erythrocyte membrane
PE serves as a reservoir class of EFA. Its composition is
thought to accurately reflect the availability of EFA at
the time of membrane assembly. Other phospholipid classes are regarded as more stable compartments with a strict
homeostatic maintenance of EFA composition and with a
slower remodeling rate.
The next question is what volume of blood needs to be
collected and under what conditions to preserve the integrity of blood cells? Drawing of a 1 ml sample of blood is
likely to be sufficient for lipidomic analysis. This volume
can be reduced in new born babies without compromising
the analysis. It is advisable to adapt the blood volume and
the tube capacity to minimize the air in contact with the
sample. Collection of the sample into a vacuum container
containing metal cations addition of a chelator like EDTA
is recommended. This prevents the chance of reactive oxygen species oxidizing unsaturated molecular species of
lipid. The blood sample should not be shaken vigorously
to minimize the damage to the erythrocytes; vesicles are
known to be released from stored erythrocyte membranes
after prolonged shaking and can be recovered in the plasma
after centrifugation. Because the lipid composition of the
released membranes is very different from the parent erythrocyte the plasma composition is corrupted by minor contaminations due to the presence of fragmented membrane
vesicles.
The storage conditions before analysis are important.
Freezing whole blood causes extensive membrane fragmentation and the resulting plasma contamination. Freezing
after separation of plasma from the blood cells is recommended in order to avoid exchange protein and enzyme
activities. The activity of these proteins is slow in EDTAcontaining samples and a time of 2–4 h for transportation
after drawing the sample is reasonable.
C. Wolf, P.J. Quinn / Progress in Lipid Research 47 (2008) 15–36
If samples are to be stored for longer periods of time this
should be at 20 °C rather than 70°/ 80 °C. Unlike storage of samples for proteomic studies there is no advantage
in storage at the lower temperature in maintaining lipid
integrity. The major cause of lipid deterioration is peroxidation and oxygen gas is more soluble at low temperatures.
Also oxygen derived radicals are more stable at low temperatures so very cold temperatures are of no advantage.
The oxygen should be chased from the sample by gassing
with a dense inert gas like argon; nitrogen is a less satisfactory substitute because, being less dense, it is less effective
in displacing oxygen.
7. Standards for authentication and quantitation
A variety of lipid standards are now available commercially with Lipidomics studies being undertaken in a number of laboratories. Rigorous quantitation using MS
requires internal standards enriched in one or more stable
isotopes [88,89]. Stable isotope (2H or 13C) labeled sphingolipids, phospholipids and sterols can be used to dope samples and act as internal standards. However, these
relatively expensive lipids are not useful for authenticating
lipid class or molecular species of phospholipids achieved
with high reliability by fragmentography in tandem mass
spectrometers. The fatty acid composition associated with
phospholipids retained for distinct specified times on a
polar stationary phase (such as a silica-diol HPLC column)
can be obtained after CID in the negative mode of ionization. Authentication is reliable but internal standards are
needed for quantitation. Because the number and complexity of distinct molecular species of lipids in biological
extracts is considerable it is not feasible to provide internal
standards for each. It is, however, acceptable to use synthetic di-saturated synthetic phospholipids (di-C14, diC15 or di-C17) or preferably mixed odd carbon number
phospholipids (C13-C15, C15-C17) as weighed standards
for each phospholipid class. Mixed odd carbon phospholipid standards offer the advantage that sn-1 and sn-2 fatty
acid fragmentation can be quantitated separately. Weighed
mixtures of sphingolipids and glycerophospholipids are
also commercially available which are assayed as an external calibrator.
Internal standards should be used to normalize the different response factor of the particular lipid class. However, because these internal standards have saturated
chains at the 1- and 2-positions of the glycerol they cannot
serve to quantitate the distinct molecular species of a given
phospholipid class since different fatty acids are detected
with a variable sensitivity as a function of the position,
length and unsaturation of the chain. A practical solution
is to use a formulated mixture of species (e.g. egg and/or
soy lecithin) assayed by the method of HPLC for benzoyl
derivatives to calculate the response coefficient of the distinct molecular class as an external calibrator. Detection
of benzoyl derivatives by absorbance is unaffected by the
particular lipid molecular species but only species which
29
are completely separated with no co-elution can be used
for quantitation.
8. Advances in analytical techniques
Major advances in lipidomic technology have come
from combining powerful separation techniques with
sophisticated detection methods. Methods in series with
HPLC have been developed with specific detection of
phospholipid head-groups. These sensitive methods have
limitations such as the requirement of derivatization with
a fluorescent label or a chromophore residue. Methods
have been developed for monitoring single lipid classes
such as choline-containing phospholipids by immobilized
choline-oxidase electrochemical detector [90,91]. These
methods, which are appropriate for routine analysis in an
industrial environment, require extensive setup for the
completion of derivatization reaction according to each
particular matrix. The present development of robust and
affordable equipments of MS has limited their
applicability.
8.1. HPLC separation of complex mixtures is usually
required prior to MS detection
Many important and pioneering lipidomic studies have
already been published without any preliminary separation
of lipid classes by HPLC. The whole lipid extract was
infused by a syringe directly in the ESI source. The infusion
syringe equipment is configured on most equipment to provide the slow infusion of standards for resolution and optimization (usually PEG polymer solution) for setting up the
spectrometer. It can be used to infuse a lipid extract at the
flow of few lL/min (or nanoL/min into ESI nano-source).
Data acquisition protocols allow repeated accumulation of
mass spectra. Using this technique, the signal/noise ratio of
minor molecular species can be significantly enhanced by
adjusting the setup of the equipment during the infusion.
A major disadvantage is many compounds, including
prominent molecular species of phospholipids, and counter-ions infused simultaneously in the ESI source counteract for ionization. This competition phenomenon known
as ‘‘signal suppression’’ can result in failure to detect minor
species present in the sample.
To avoid the co-infusion of molecules acting as suppression agents, a preliminary separation by HPLC is
required off line or in line. Connection of HPLC into
MS has been recently facilitated by the design of ESI
sources which can handle solvent flows of up to 1 mL/
min from conventional HPLC systems. Chromatographic
solvent elution mixtures usually contain alcohols (methanol or isopropanol), acetonitrile, hexane, chloroform,
water. Acetic or formic acid and an amino-base (ammonia, methylamine, diisopropylamine and piperidine) are
also added for ion-pairing, to displace counter-ions and
to assist the negative or positive ionization modes. Flow
of these various liquids into the ion source may interfere
30
C. Wolf, P.J. Quinn / Progress in Lipid Research 47 (2008) 15–36
with the spraying. Electro-nebulization in the ESI source
is assisted by a flow of heated inert gas to assist evaporation with the final objective to obtain the formation of a
stable spray of very fine droplets from which parent ions
can appropriately be attracted and collected for MS. In
modern ESI-sources contamination of the mass filtering
equipment is prevented by spraying the uncharged compounds and larger droplets out of the ion trajectory in
the initial quadrupole stage of the MS. A curtain of nitrogen gas flowing in front of the electrostatic lens at the
entrance also diverts contamination of RF quadrupole
or hexapole transferring the ion into the active part of
the mass filter (quadrupole or ion trap). The curtain is
created by a continuous flow (a few litres per minute at
atmospheric pressure) of nitrogen gas flowing in front of
the entrance of the quadrupole. The entrance lens to the
quadrupole is set at a positive potential for attracting negatively charged ions formed in the source and repelling
the positively charged ions. Neutral (non-ionized) molecules and droplets of unevaporated spray are diverted
from the quadrupole entrance by the nitrogen curtain
gas at the entrance to the first electrostatic lens. Therefore
only negatively charged ions penetrate the quadrupole
analyser for separation as a function of m/z. In the positive mode of MS detection the entrance lens is set to a
negative potential. Cleaning of quadrupole pre-filters (a
part of the quadrupole) is a delicate maintenance operation which is largely avoided by the use of the curtain gas.
These improvements have been incorporated into robust
mass spectrometers following development of highly efficient turbo-mechanical vacuum pumps. Lipid extracts representing a clean matrix lipidomics system dedicated
equipment requires relatively little maintenance (a ‘‘source’’
cleaning including vent/vacuum takes around 2 h, 6 h
vacuum pumping to recover maximum specifications).
Initial separation of lipids by liquid chromatography
allows the resolved compounds to sequentially enter the
tandem MS. This results in less suppression and high ionization yield and an increased sensitivity for minor molecular species.
Multiple reaction monitoring can then be used to quantitate a series of lipids relative to an internal standard. The
assay exploits the high specificity of the parent > product
ions pair characterizing a molecular species of lipid.
Because of the short time-delay between successive measurements of ion pairs (<5 ms) multiple pairs can be monitored simultaneously. For instance up to 300 pairs with
20 ms sampling time can be acquired during the elution
from the HPLC column. Fitting of chromatography peaks
by data point interpolation can be accurately performed by
recording more than 20 points using a slow elution regime
(flow of 300 lL/min gives half-width around 2 min). The
method allows the accurate quantification of multiple lipids
within a single run.
It is not clear at this point in time how useful it is to separate lipid classes on a straight polar stationary phase and
molecular species on a reverse stationary phase ahead of
the MS detector (2D chromatography). It might be that
2D tandem liquid chromatography (off line or on line) will
be a used [92] for future applications requiring a very high
quantitative accuracy and exclusion of any matrix interference and suppression. The method is already available on
platforms currently used for Proteomics as an alternative
to gel 2D electrophoresis. Conditions for lipid ionization
in the ESI source are operationally defined to keep interference to a minimum depending on the matrix. Addition into
the chromatography solvent mixture of formic and acetic
acids forming an ion pair with choline-containing phospholipids is also helpful to avoid chlorine interference.
Addition of organic bases miscible in the chromatography
solvent helps the dissociation and transfer of mobile protons in the ion source. Ammonia, methylamine, trimethylamine, diisopropylamine, piperidine have been tested Figs.
2 and 3.
A promising technique has been recently setup which
retains most of the main advantages (avoidance of ion suppression and increased sensitivity by time accumulation) of
both of the gradual elution of lipids through HPLC and of
time accumulated signal during slow infusion. The technique takes advantage of a quantitative treatment of the
m/z versus retention-time contour plot created during
LCMS acquisition (Fig. 4). A software module called
‘‘SECD’’ for ‘‘Spectrum Extraction from Chromatographic
Data’’ has been made available from Uphoff [93,94]. SECD
tool is designed for analysis of LC–MS datasets to provide
contour display of MS chromatograms as two-dimensional
‘‘maps’’ for visual inspection of the data. This allows the
user to select and extract mass spectra from arbitrary
regions. Due to selective and load/capacity biased adsorption of HPLC columns the user selection is commonly not
a simple square area but an irregular parallelogram. Reliable analysis of complex lipidome data after separation
of lipid classes by HPLC is obtained with improved signal-to-noise ratio owing to less background signal integration and to the gradual injection of lipids in the source as a
function of time. The method compares favourably with
standard time-range averaged spectra obtained by infusion.
The software is distributed freely (Department of Biochemistry, University of Helsinki). It requires previous reformatting of acquisition files into the standard NETCDF format.
Most proprietary softwares from LCMS supplier comprise
the appropriate translator to the Standard Specification for
Analytical Data Interchange format NETCDF (UCAR
Unidata Program). A complementary LIMSA tool
designed by Haimi [93,94] finds and integrates peaks in
the mass spectra exported from the SECD selected regions.
LIMSA matches the peaks with expected lipids. The
‘‘library’’ is user-supplied and can be easily adapted to
the specificity of biological specimen. Adaptation of the
library usually requires the resources of tandem MS to
assign appropriate structures of molecular species to mass
spectrum peaks. LIMSA corrects for overlap in their isotopic patterns and quantifies the identified lipid species
according to internal standards. Three different algorithms
C. Wolf, P.J. Quinn / Progress in Lipid Research 47 (2008) 15–36
31
Fig. 2. Method for enrichment of plasma membrane from brain microsomes by affinity partitioning. Microsomes obtained from brain homogenate by
differential centrifugation are partitioned in a polyethylene glycol–dextran system and the plasma membranes are recovered from the upper layer in several
steps. The combined upper phases are then applied to an affinity lower phase coupled with wheat-germ agglutinin which combines with the glycolipids and
proteins on the surface of the plasma membrane to remove them from the upper layer. After washing with fresh upper layer the membranes are released by
dilution of lower phase with a solution of N-acetyl-D-glucosamine.
Fig. 3. Schematic view of a tandem mass spectrometer. Tandem MS2 represents the central equipment of the lipidomics platform. Lipids are introduced
via syringe infusion or HPLC into the electro-spray- (ESI) or atmospheric-pressure chemical-ionization (APCI) ionization source. Molecular ions (Parent
ions, positively or negatively charged) resulting of the soft ionization procedure are ‘‘screened’’ through the first mass filter m/z (a quadrupole, Q1).
Selected Parent ions are channeled into the collision cell (a quadrupole, Q2) filled with an inert gas wherein ions are accelerated and fragmented. Resulting
fragment ions (product ions have retained the Parent ion charge) are filtered in the second mass filter, either a quadrupole (Q3), or a time-of-flight (TOF),
or an ion trap as a function of the required mass resolution. Ion pairs (parent and product ion) serve for lipid structure assignment and quantification.
32
C. Wolf, P.J. Quinn / Progress in Lipid Research 47 (2008) 15–36
for isotopic correction are comprised in LIMSA including
an efficient lineshape deconvolution algorithm.
8.2. Electrospray ionization procedures
The soft ionization procedure developed for macromolecules was the long-waited technical development required
for lipid analysis. Soft ionization methods are able to preserve the lipid molecular structure after ionization by minimizing the fragmentation associated with the excess of
potential energy added by impact. This goal was difficult
to achieve with fast atom bombardment and direct chemical ionization methods used hitherto in which the spectra
were replete with fragment ions. Fragmentation was especially high if molecular ions were enclosed in a matrix
heated at a temperature where the excess potential energy
could not be dissipated quickly. APPI-source (atmospheric
pressure chemical ionization) may be regarded as a variation of the ESI-source. This method incorporates a sparking electric discharge device which produces the reagent gas
at atmospheric pressure. The activated gas transfers
charges onto lipids and forms charged adducts similar to
the traditional chemical ionization system designed for
GC–MS. The procedure is applicable for neutral lipids
such as triglycerides and sterols devoid of any ionizable
groups. Combined ESI-APCI sources are now available
commercially which can be switched into one or other
mode.
Collision-induced dissociation of parent molecular ions
or fragmentation by an auxiliary pulsed voltage in the
ion trap (MSn) produces fragment ions corresponding to
the fatty acid moieties required for assigning the particular
mass peaks with phospholipids. Fragment ions corresponding to carboxylate anions of fatty acids are specifically
obtained in the negative ionization mode of ESI after a collision-induced dissociation. Deciphering the combination
of fatty acids or ether-linked hydrocarbon chains is critical
for lipidomic studies because many isobaric molecules correspond to a single mass. For instance, 13 combinations of
diacyl-PE containing fatty acids totaling 36 carbon atoms
and 2 double bonds can be represented in spectra by m/
z = 742.53 and 2 ether-PE molecules with 37 carbons correspond to m/z = 742.57. With tandem MS (triple quadrupole (TQ) or Q-Trap or Q-TOF) the parent ion sorted by
the first quadrupole is obtained in a gas collision cell filled
with nitrogen or argon (P = 10 4 Torr). Collision cells are
m/z
989
PE-TNB
750
PE
734
21.61
32.43
Retention time (min)
Fig. 4. Measurement of exposure of PE on the external leaflet of the plasma membrane. Red blood cells are withdrawn by phlebotomy and rapidly
incubated with trinitrobenzenesulphonic acid (TNBS) under conditions in which the reagent is known to be impermeant to the erythrocyte membranes.
Amino groups of phosphatidylethanolamine (PE) exposed to TNBS on the external surface of the membrane are derivatized. The molecular species of PE
which are labeled are detected as PE-TNBS (4A). The mass spectrum corresponding to the area for PE-TNBS is recorded with an enhanced resolution of
the ion trap at 0.02 amu. This is shown in the insert labeled PE-TNBS in (A). The contour plot shows also the underivatized molecular species of PE
present abundantly on the cytoplasmic surface of the membrane (lower panel). The mass spectrum corresponding to the unlabeled PE on the inner surface
of erythrocytes is shown in the insert PE in (A). (B). The elution time is centered on 21.6 min for PE-TNBS (PE exposed on the external surface of the cell
and derivatized by TNBS) and 32.4 min for non-derivatized PE after chromatographic separation on a silica-diol (a straight phase) column. In this
experiment HPLC is in line with the Q-Trap spectrometer (Applied Biosystems). The HPLC system separates plasmalenyl-PE which has a vinyl-ether acyl
chain at the sn-1 position in front of molecular species of diacyl-PE (dashed lines indicate the corresponding series). The m/z versus time contour plot is
obtained using ‘‘Spectrum Extracted from Chromatographic Data’’ (SECD) software kindly provided by Dr. A. Uphoff. Spectra including isotope
correction for quantitation and the tentative assignment of mass peaks are obtained using ‘‘Lipid Mass Spectrum Analysis’’ (LIMSA) software kindly
provided by Dr. P. Haimi [93,94]. Mass peak assignment for internal and external side PE and PE-TNBS, respectively, is obtained by comparison with a
library prepared as an Excel add-in for the human red blood cell PE molecular species. Tandem MS analysis is used to prepare the corresponding library.
(A) contour plot of m/z versus LC retention time. (B) PE molecular species of the cytofacial erythrocyte membrane side. (C) PE molecular species of the
external erythrocyte membrane side exposed to TNBS reagent.
C. Wolf, P.J. Quinn / Progress in Lipid Research 47 (2008) 15–36
B
33
450000000
400000000
Original Spectrum
Found Peaks
Fitted Peak
663.3248
683.3248
703.3248
723.3248
PE40:04alkenyl
778.6, 119223422
PE38:05
764.5, 4618221
743.3248
PE40:04alkyl
780.6, 9949470
PE40:06alkenyl
774.5, 1456325
PE38:04
766.5, 67989975
PE38:03
768.6, 378171
PE38:05alkenyl
748.5, 27637810
PE40:06alkyl
776.6, 14025466
PE38:04alkenyl
750.5, 207282411
PE36:00
746.6, 8156332
643.3248
PE38:06alkyl
748.5, 28733284
PE36:01alkenyl
728.6, 1157279
50000000
0
623.3248
PE36:03
740.5, 19149730
100000000
PE36:04
738.5, 33519153
PE28:00
634.4, 62184205
150000000
PE36:04alkenyl
722.5, 23660089
PE34:02
714.5, 46114702
200000000
PE36:02
742.5, 11467181
PE34:01
716.5, 46757758
250000000
PE36:01
744.6, 1991092
PE34:00
718.5, 53638022
300000000
763.3248
783.3248
300000000
TNP-PE38:04alkeny
961.5, 169123486
Original Spectrum
Found Peaks
Fitted Peaks
0
900.413
920.413
940.413
960.413
TNP-PE40:04alkeny
989.6, 32566928
TNP-PE40:05alkeny
987.5, 11027023
TNP-PE38:03
979.5, 278363
TNP-PE40:06alkeny
985.5, 16525634
TNP-PE38:06alky
959.5, 50125959
TNP-PE38:05
975.5, 9610955
50000000
TNP-PE36:04alkeny
933.5, 17967273
100000000
TNP-PE34:02
925.5, 38034675
150000000
TNP-PE36:04
949.5, 62992519
TNP-PE34:01
927.5, 37688878
200000000
TNP-PE36:03
951.5, 45498274
TNP-PE34:00
929.5, 117282725
250000000
TNP-PE38:04
977.5, 10602420
C
PE38:03alkenyl
752.6, 1075145
350000000
980.413
1000.413
Fig. 4 (continued)
now designed for short residency times for ions of only a
few milliseconds, an acceleration which serves to avoid
cross-contamination of successive parent ions. Indeed fragmentation of phospholipid molecules may occur prematurely in the ionization source at atmospheric pressure
where cleavage of the headgroup occurs frequently in the
form of a neutral loss. The careful modulation of temperature and declustering potential applied in the ESI-source is
necessary to maximise sensitivity on collision-induced dissociation transitions.
Monitoring of parent and product ions can be used to
trigger supplementary analyses using criteria such as abundance or a preset inclusion list of m/z values. For instance,
the analysis may be a triggered ‘‘full-scan’’. Enhanced resolution spectra can also be triggered in hybrid tandem MS
incorporating ion trap or time of flight as a second high
resolution m/z filter. Preselection of the most interesting
molecular species can be made on the criteria of mass
and abundance or the criteria can be calculated during
acquisition. For instance software routines can be set up
to include or exclude particular parent ion mass. Dynamic
exclusion of m/z corresponding to the 13C-isotope peaks
(M + 1, M + 2, . . .) is an example for such a routine of
directed data acquisition. In biological lipids differing by
one double bond the criteria for selection is not easily
handled by the acquisition software so that exclusion of
34
C. Wolf, P.J. Quinn / Progress in Lipid Research 47 (2008) 15–36
isotope contributions (de-isotopisation) is usually performed after recording the experimental mass spectra by
a specialized softwares [93,94].
Interpretation of the vast amount of data produced by
the platforms dedicated to Lipidomics is labor-intensive
and requires sophisticated data analysis protocols. Quantification of hundreds of MRM transitions with commercially available software is notoriously inefficient since it
was designed initially for the time integration of few metabolites. More sophisticated software is required for the practical handling of large collections of lipidomic data. The
procedure involves smoothing, time integration, validation
of data quality, normalization with internal standards and
testing of a priori hypotheses using standard statistical
tests. Multivariate testing not depending on data reduction
such as principal component analysis, partial least squares
and various clustering techniques can also be applied to
discover unexpected linkages with disease (‘‘biomakers’’)
[95]. Lipid fingerprinting programs for biomarkers have
recently been implemented by suppliers for lipid platform.
Computational lipidomics [96] couples mass spectrometry
with statistical algorithms to facilitate the comprehensive
analysis of hundreds of lipid species from cellular extracts.
As a result, lipid arrays are generated to identify qualitatively changes in lipid composition between experimental
or disease states.
9. Conclusion
The present review has aimed to provide a practical view
of current lipidomic studies. A number of recent methodological aspects are detailed in several reviews published
during the first 6 months of 2007 [53,97–104]. During
2006, four reviews of tandem mass spectrometry were published [105–108] showing a growing interest in the field.
Whether the procedures described in the above review are
accessible to non specialized laboratories is tentatively
answered in the present review. Though lipidomics was initially designed for specialized laboratories it has now
evolved into a method which can be applied in general biology laboratories or in cooperative approach with a MS unit
as long as the particular precautions for lipid sampling and
data treatment are followed. The practical procedures are
presented in this review so that the particularities of lipids
are made understandable to a more general audience.
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
Acknowledgements
Financial support was provided by the Human Frontier
Science Programme. The assistance of Dr. P. Haimi (Helsinski University) for the software LIMSA is gratefully
acknowledged.
References
[1] Koumanov KS, Tessier C, Momchilova AB, Rainteau D, Wolf C,
Quinn PJ. Comparative lipid analysis and structure of detergent-
[18]
[19]
[20]
resistant membrane raft fractions isolated from human and ruminant erythrocytes. Arch Biochem Biophys 2005;434:150–8.
Israelachvili J. Relationship between chain protrusions and surfactant critical micelle concentration values. Langmuir 1992;8:1501.
Heerklotz H, Epand RM. The enthalpy of acyl chain packing and
the apparent water-accessible apolar surface area of phospholipids.
Biophys J 2001;80:271–9.
Tieleman DP, Marrink SJ. Lipids out of equilibrium: energetics of
desorption and pore mediated flip-flop. J Am Chem Soc
2006;128:12462–7.
Debony J, Lopez A, Gilleron M, Welby M, Laneelle G, Rousseau B,
et al. Transverse and lateral distribution of phospholipids and
glycolipids in the membrane of the bacterium Micrococcus luteus.
Biochemistry 1989;28:3728–37.
Benninger RKP, Onfelt B, Neil MAA, Davis DM, French PMW.
Fluorescence imaging of two-photon linear dichroism: Cholesterol
depletion disrupts molecular orientation in cell membranes. Biophys
J 2005;88:609–22.
He Q, Li JB. Hydrolysis characterization of phospholipid monolayers catalyzed by different phospholipases at the air-water interface. Adv Coll Interf Sci 2007;131:91–8.
Clark JD, Lin LL, Kriz RW, Ramesha CS, Sultzman LA, Lin AY,
et al. A novel arachidonic acid-selective cytosolic PLA2 contains a
Ca(2+)- dependent translocation domain with homology to PKC
and GAP. Cell 1991;65:1043–51.
Koumanov KS, Quinn PJ, Bereziat G, Wolf C. Cholesterol relieves
the inhibitory effect of sphingomyelin on type II secretory phospholipase A2. Biochem J 1998;336(Pt 3):625–30.
Kerkhoff C, Trumbach B, Gehring L, Habben K, Schmitz G,
Kaever V. Solubilization, partial purification and photolabeling of
the integral membrane protein lysophospholipid:acyl-CoA acyltransferase (LAT). Eur J Biochem 2000;267:6339–45.
Magargal WW, Dickinson ES, Slakey LL. Distribution of membrane marker enzymes in cultured arterial endothelial and smooth
muscle cells. The subcellular location of oleoyl- CoA:1-acyl-snglycero-3-phosphocholine acyltransferase. J Biol Chem 1978;253:
8311–8.
Momchilova A, Markovska T, Pankov R. Arachidonoyl-CoA:lysophosphatidylcholine acyltransferase activity in ras-transformed NIH
3T3 fibroblasts depends on the membrane composition. Biochem
Mol Biol Int 1999;47:555–61.
Scheideler MA, Bell RM. Efficiency of reconstitution of the
membrane-associated sn-glycerol 3- phosphate acyltransferase of
Escherichia coli. J Biol Chem 1986;261:10990–5.
Colard O, Bard D, Bereziat G, Polonovski J. Acylation of
endogenous phospholipids and added lysoderivatives by rat liver
plasma membranes. Biochim Biophys Acta 1980;618:88–97.
Hutson JL, Higgins JA. Reversible activation/inactivation of the
deacylation/acylation cycle in rat liver microsomes. Biosci Rep
1987;7:73–80.
McKean ML, Silver MJ. Phospholipid biosynthesis in human
platelets. The acylation of lyso-platelet-activating factor. Biochem J
1985;225:723–9.
Sugiura T, Masuzawa Y, Nakagawa Y, Waku K. Transacylation of
lyso platelet-activating factor and other lysophospholipids by
macrophage microsomes. Distinct donor and acceptor selectivities.
J Biol Chem 1987;262:1199–205.
Blank ML, Smith ZL, Fitzgerald V, Snyder F. The CoA-independent transacylase in PAF biosynthesis: tissue distribution and
molecular species selectivity. Biochim Biophys Acta 1995;1254:
295–301.
Nixon AB, Greene DG, Wykle RL. Comparison of acceptor and
donor substrates in the CoA-independent transacylase reaction in
human neutrophils. Biochim Biophys Acta 1996;1300:187–96.
Florin-Christensen J, Suarez C, Florin-Christensen M, Wainszelbaum M, Brown W, McElwain T, et al. A unique phospholipid
organization in bovine erythrocyte membranes. PNAS 2001;98:
7736–41.
C. Wolf, P.J. Quinn / Progress in Lipid Research 47 (2008) 15–36
[21] Rhee SG. Regulation of phosphoinositide-specific phospholipase C.
Ann Rev Biochem 2001;70:281–312.
[22] Rebecchi MJ, Pentyala SN. Structure, function, and control of phosphoinositide-specific phospholipase C. Physiol Rev 2000;80:1291–335.
[23] Lopez I, Mak EC, Ding JR, Hamm HE, Lomasney JW. A novel
bifunctional phospholipase C that is regulated by G alpha(12) and
stimulates the Ras/mitogen-activated protein kinase pathway. J Biol
Chem 2001;276:2758–65.
[24] Kelley GG, Reks SE, Ondrako JM, Smrcka AV. Phospholipase C
epsilon: a novel Ras effector. Embo J 2001;20:743–54.
[25] Illenberger D, Walliser C, Nurnberg B, Lorente MD, Gierschik P.
Specificity and structural requirements of phospholipase C-beta
stimulation by Rho GTPases versus G protein beta gamma dimers. J
Biol Chem 2003;278:3006–14.
[26] Illenberger D, Walliser C, Strobel J, Gutman O, Niv H, Gaidzik V,
et al. Rac2 regulation of phospholipase C-beta(2) activity and mode
of membrane interactions in intact cells. J Biol Chem 2003;278:
8645–52.
[27] McLaughlin S, Wang JY, Gambhir A, Murray D. PIP2 AND
proteins: Interactions, organization, and information flow. Ann Rev
Biophys Biomol Struct 2002;31:151–75.
[28] Philip F, Guo YJ, Scarlata S. Multiple roles of pleckstrin homology
domains in phospholipase C beta function. FEBS Lett 2002;531:
28–32.
[29] Wilde JI, Watson SP. Regulation of phospholipase C gamma
isoforms in haematopoietic cells – why one, not the other? Cell
Signal 2001;13:691–701.
[30] Wang DM, Feng J, Wen RR, Marine JC, Sangster MY, Parganas E,
et al. Phospholipase C gamma 2 is essential in the functions of B cell
and several Fc receptors. Immunity 2000;13:25–35.
[31] Song CH, Satoh T, Edamatsu H, Wu DM, Tadano M, Gao XL,
et al. Differential roles of Ras and Rap1 in growth factor-dependent
activation of phospholipase C epsilon. Oncogene 2002;21:8105–13.
[32] Singer WD, Brown HA, Sternweis PC. Regulation of eukaryotic
phosphatidylinositol-specific phospholipase C and phospholipase D.
Ann Rev Biochem 1997;66:475–509.
[33] Frohman MA, Sung TC, Morris AJ. Mammalian phospholipase D
structure and regulation. Biochim Biophy Acta 1999;1439:175–86.
[34] Exton JH. Regulation of phospholipase D. FEBS Lett 2002;531:
58–61.
[35] Sugars JM, Cellek S, Manifava M, Coadwell J, Ktistakis NT.
Hierarchy of membrane-targeting signals of phospholipase D1
involving lipid modification of a pleckstrin homology domain. J
Biol Chem 2002;277:29152–61.
[36] Rizzo MA, Shome K, Watkins SC, Romero G. The recruitment of
Raf-1 to membranes is mediated by direct interaction with
phosphatidic acid and is independent of association with Ras. J
Biol Chem 2000;275:23911–8.
[37] Grange M, Sette C, Cuomo M, Conti M, Lagarde M, Prigent AF,
et al. The cAMP-specific phosphodiesterase PDE4D3 is regulated
by phosphatidic acid binding – consequences for cAMP signaling
pathway and characterization of a phosphatidic acid binding site. J
Biol Chem 2000;275:33379–87.
[38] Fang YM, Vilella-Bach M, Bachmann R, Flanigan A, Chen J.
Phosphatidic acid-mediated mitogenic activation of mTOR signaling. Science 2001;294:1942–5.
[39] Jones JA, Hannun YA. Tight binding inhibition of protein
phosphatase-1 by phosphatidic acid – specificity of inhibition by
the phospholipid. J Biol Chem 2002;277:15530–8.
[40] Frank C, Keilhack H, Opitz F, Zschornig O, Bohmer FD. Binding
of phosphatidic acid to the protein-tyrosine phosphatase SHP-1 as a
basis for activity modulation. Biochemistry 1999;38:11993–2002.
[41] Ktistakis NT, Delon C, Manifava M, Wood E, Ganley I, Sugars
JM. Phospholipase D1 and potential targets of its hydrolysis
product, phosphatidic acid. Biochem Soc Trans 2003;31:94–7.
[42] Samet D, Barenholz Y. Characterization of acidic and neutral
sphingomyelinase activities in crude extracts of HL-60 cells. Chem
Phys Lipids 1999;102:65–77.
35
[43] Goni FM, Alonso A. Sphingomyelinases: enzymology and membrane activity. FEBS Lett 2002;531:38–46.
[44] Kinnunen PK, Holopainen JM. Mechanisms of initiation of
membrane fusion: role of lipids. Biosci Rep 2000;20:465–82.
[45] Huang HW, Goldberg EM, Zidovetzki R. Ceramide induces
structural defects into phosphatidylcholine bilayers and activates
phospholipase A(2). Biochem Biophys Res Commun 1996;220:
834–8.
[46] Siskind LJ, Kolesnick RN, Colombini M. Ceramide channels
increase the permeability of the mitochondrial outer membrane to
small proteins. J Biol Chem 2002;277:26796–803.
[47] Junge S, Brenner B, Lepple-Wienhues A, Nilius B, Lang F,
Linderkamp O, et al. Intracellular mechanisms of L-selectin
induced capping. Cell Signal 1999;11:301–8.
[48] Liu PS, Anderson RGW. Compartmentalized production of ceramide at the cell-surface. J Biol Chem 1995;270:27179–85.
[49] Dobrowsky RT. Sphingolipid signalling domains – floating on rafts
or buried in caves? Cell Signal 2000;12:81–90.
[50] Kirschnek S, Paris F, Weller M, Grassme H, Ferlinz K, Riehle A,
et al. CD95-mediated apoptosis in vivo involves acid sphingomyelinase. J Biol Chem 2000;275:27316–23.
[51] Cremesti AE, Goni FM, Kolesnick R. Role of sphingomyelinase
and ceramide in modulating rafts: do biophysical properties
determine biologic outcome? FEBS Lett 2002;531:47–53.
[52] Jackson SN, Wang HY, Woods AS. Direct profiling of lipid
distribution in brain tissue using MALDI-TOFMS. Anal Chem
2005;77:4523–7.
[53] Schiller J, Suss R, Fuchs B, Muller M, Zschornig O, Arnold K.
MALDI-TOF MS in lipidomics. Front Biosci 2007;12:2568–79.
[54] Chen X, Morris R, Lawrence MJ, Quinn PJ. The isolation and
structure of membrane lipid rafts from rat brain. Biochimie
2007;89:192–6.
[55] Pike LJ, Han X, Gross RW. Epidermal growth factor receptors are
localized to lipid rafts that contain a balance of inner and outer
leaflet lipids: a shotgun lipidomics study. J Biol Chem 2005;280:
26796–804.
[56] Koumanov K, Tessier C, Momchilova A, Rainteau D, Wolf C,
Quinn P. Comparative lipid analysis and structure of detergentresistant membrane rafts isolated from human and ruminant
erythrocytes. Arch Biochem Biophys 2005;434:150–8.
[57] Koumanov KS, Wolf C, Quinn PJ. Lipid composition of membrane
domains. Subcell Biochem 2004;37:153–63.
[58] Persson A, Jergil B. The purification of membranes by affinity
partitioning. Faseb J 1995;9:1304–10.
[59] AP A. In: AP A, editor. Partition of cell particles and macromolecules. New York: Wiley-Interscience; 1971. p. 233–42.
[60] Cao R, Li XW, Liu Z, Peng X, Hu WJ, Wang XC, et al.
Integration of a two-phase partition method into proteomics
research on rat liver plasma membrane proteins. J Proteome Res
2006;5:634–42.
[61] Schindler J, Lewandrowski U, Sickmann A, Friauf E, Nothwang
HG. Proteomic analysis of brain plasma membranes isolated by
affinity two-phase partitioning. Mol Cell Proteom 2006;5:390–400.
[62] Everberg H, Clough J, Henderson P, Jergil B, Tjerneld F, Ramirez
IB. Isolation of Escherichia coli inner membranes by metal affinity
two-phase partitioning. J Chromatogr A 2006;1118:244–52.
[63] Luers GH, Hartig R, Mohr H, Hausmann M, Fahimi HD, Cremer
C, et al. Immuno-isolation of highly purified peroxisomes using
magnetic beads and continuous immunomagnetic sorting. Electrophoresis 1998;19:1205–10.
[64] Kikuchi M, Hatano N, Yokota S, Shimozawa N, Imanaka T,
Taniguchi H. Proteomic analysis of rat liver peroxisome – presence
of peroxisome-specific isozyme of Lon protease. J Biol Chem
2004;279:421–8.
[65] Calhoun BC, Goldenring JR. Two Rab proteins, vesicle-associated
membrane protein 2 (VAMP-2) and secretory carrier membrane
proteins (SCAMPs), are present on immunoisolated parietal cell
tubulovesicles. Biochem J 1997;325:559–64.
36
C. Wolf, P.J. Quinn / Progress in Lipid Research 47 (2008) 15–36
[66] Henley JR, McNiven MA. Association of a dynamin-like protein
with the golgi apparatus in mammalian cells. J Cell Biol
1996;133:761–75.
[67] Srinivasan S, Traini M, Herbert B, Sexton D, Harry J, Alexander H,
et al. Proteomic analysis of a developmentally regulated secretory
vesicle. Proteomics 2001;1:1119–27.
[68] Morciano M, Burre J, Corvey C, Karas M, Zimmermann H,
Volknandt W. Immunoisolation of two synaptic vesicle pools from
synaptosomes: a proteomics analysis. J Neurochem 2005;95:
1732–45.
[69] Lawson EL, Clifton JG, Huang FL, Li XS, Hixson DC, Josic D.
Use of magnetic beads with immobilized monoclonal antibodies for
isolation of highly pure plasma membranes. Electrophoresis
2006;27:2747–58.
[70] Madore N, Smith KL, Graham CH, Jen A, Brady K, Hall S, et al.
Functionally different GPI proteins are organized in different
domains on the neuronal surface. Embo J 1999;18:6917–26.
[71] Zhang LJ, Wang X, Peng X, Wei YJ, Cao R, Liu Z, et al.
Immunoaffinity purification of plasma membrane with secondary
antibody superparamagnetic beads for proteomic analysis. J Proteome Res 2007;6:34–43.
[72] Han X, Gross RW. Shotgun lipidomics: multidimensional MS
analysis of cellular lipidomes. Expert Rev Proteomics 2005;2:
253–64.
[73] Adibhatla RM, Hatcher JF, Dempsey RJ. Lipids and lipidomics in
brain injury and diseases. Aaps J 2006;8:E314–21.
[74] Shen JX, Motyka RJ, Roach JP, Hayes RN. Minimization of ion
suppression in LC–MS/MS analysis through the application of
strong cation exchange solid-phase extraction (SCX-SPE). J Pharm
Biomed Anal 2005;37:359–67.
[75] Tsui ZC, Chen QR, Thomas MJ, Samuel M, Cui Z. A method for
profiling gangliosides in animal tissues using electrospray ionizationtandem mass spectrometry. Anal Biochem 2005;341:251–8.
[76] Koivusalo M, Haimi P, Heikinheimo L, Kostiainen R, Somerharju
P. Quantitative determination of phospholipid compositions by ESIMS: effects of acyl chain length, unsaturation, and lipid concentration on instrument response. J Lipid Res 2001;42:663–72.
[77] Christie WW. Lipid analysis: isolation, separation, identification
and structural analysis of lipids. 3rd ed. Bridgwater: Oily Press;
2003.
[78] Folch J, Lees M, Stanley GHS. A simple method for the isolation
and purification of total lipides from animal tissues. J Biol Chem
1957;226:497–509.
[79] Bligh EG, Dyer WJ. A rapid method of total lipid extraction and
purification. Can J Biochem Physiol 1959;37:911–7.
[80] Ruiz-Gutierrez V, Perez-Camino MC. Update on solid-phase
extraction for the analysis of lipid classes and related compounds.
J Chromatogr A 2000;885:321–41.
[81] Persson E, Lofgren L, Hansson G, Abrahamsson B, Lennernas H,
Nilsson R. Simultaneous assessment of lipid classes and bile acids in
human intestinal fluid by solid-phase extraction and HPLC methods. J Lipid Res 2007;48:242–51.
[82] Christie WW. Gas chromatography and lipids: a practical
guide. Ayr: The Oily Press; 1989.
[83] Sobolik KD, Gremillion KJ, Whitten PL, Watson PJ. Sex determination of prehistoric human paleofeces. Am J Phys Anthropol
1996;101:283–90.
[84] Lin DS, Connor WE. Fecal steroids of the coprolite of a Greenland
Eskimo mummy, AD 1475: a clue to dietary sterol intake. Am J Clin
Nutr 2001;74:44–9.
[85] Hullin F, Kim HY, Salem Jr N. Analysis of aminophospholipid
molecular species by high performance liquid chromatography. J
Lipid Res 1989;30:1963–75.
[86] Schwartz R, Olson J, Raventos-Suarez C, Yee M, Heath R, Lubin
B, et al. Altered plasma membrane organization in Plasmodium
falciparum-infected human erythrocytes. Blood 1987;69:401–7.
[87] Schneiter R, Brugger B, Sandhoff R, Zellnig G, Leber A, Lampl M,
et al. Electrospray ionization tandem mass spectrometry (ESI-MS/
[88]
[89]
[90]
[91]
[92]
[93]
[94]
[95]
[96]
[97]
[98]
[99]
[100]
[101]
[102]
[103]
[104]
[105]
[106]
[107]
[108]
MS) analysis of the lipid molecular species composition of yeast
subcellular membranes reveals acyl chain-based sorting/remodeling
of distinct molecular species en route to the plasma membrane. J
Cell Biol 1999;146:741–54.
Hunt AN, Postle AD. Mass spectrometry determination of endonuclear phospholipid composition and dynamics. Methods
2006;39:104–11.
Sun D, Cree MG, Zhang XJ, Boersheim E, Wolfe RR. Measurement of stable isotopic enrichment and concentration of long-chain
fatty acyl-carnitines in tissue by HPLC–MS. J Lipid Res
2006;47:431–9.
Damsma G, Westerink BH, Horn AS. A simple, sensitive, and
economic assay for choline and acetylcholine using HPLC, an
enzyme reactor, and an electrochemical detector. J Neurochem
1985;45:1649–52.
Shaikh NA. Characterization and analysis of the subclasses and
molecular species of choline phosphoglycerides from porcine heart
by successive chemical hydrolyses and reverse phase high-performance liquid chromatography. Mol Cell Biochem 1990;96:43–55.
Han X, Gross RW. Global analyses of cellular lipidomes directly
from crude extracts of biological samples by ESI mass spectrometry:
a bridge to lipidomics. J Lipid Res 2003;44:1071–9.
Haimi P, Uphoff A, Hermansson M, Somerharju P. Software tools
for analysis of mass spectrometric lipidome data. Anal Chem
2006;78:8324–31.
Rockwood AL, Haimi P. Efficient calculation of accurate masses of
isotopic peaks. J Am Soc Mass Spectrom 2006;17:415–9.
Wiest MM, Watkins SM. Biomarker discovery using high-dimensional lipid analysis. Curr Opin Lipidol 2007;18:181–6.
Forrester JS, Milne SB, Ivanova PT, Brown HA. Computational
lipidomics: a multiplexed analysis of dynamic changes in membrane
lipid composition during signal transduction. Mol Pharmacol
2004;65:813–21.
Albert CJ, Anbukumar DS, Monda JK, Eckelkamp JT, Ford DA.
Myocardial lipidomics. Developments in myocardial nuclear lipidomics. Front Biosci 2007;12:2750–60.
Berliner JA, Zimman A. Future of toxicology-lipidomics, an
important emerging area for toxicologists: focus on lipid oxidation
products. Chem Res Toxicol 2007;20:849–53.
Gaspar ML, Aregullin MA, Jesch SA, Nunez LR, Villa-Garcia M,
Henry SA. The emergence of yeast lipidomics. Biochim Biophys
Acta 2007;1771:241–54.
German JB, Gillies LA, Smilowitz JT, Zivkovic AM, Watkins SM.
Lipidomics and lipid profiling in metabolomics. Curr Opin Lipidol
2007;18:66–71.
Kuksis A. Lipidomics in triacylglycerol and cholesteryl ester
oxidation. Front Biosci 2007;12:3203–46.
Rainville PD, Stumpf CL, Shockcor JP, Plumb RS, Nicholson JK.
Novel application of reversed-phase UPLC-oaTOF-MS for lipid
analysis in complex biological mixtures: a new tool for lipidomics. J
Proteome Res 2007;6:552–8.
Welti R, Shah J, Li W, Li M, Chen J, Burke JJ, et al. Plant
lipidomics: discerning biological function by profiling plant complex
lipids using mass spectrometry. Front Biosci 2007;12:2494–506.
Yetukuri L, Katajamaa M, Medina-Gomez G, Seppanen-Laakso T,
Vidal-Puig A, Oresic M. Bioinformatics strategies for lipidomics
analysis: characterization of obesity related hepatic steatosis. BMC
Syst Biol 2007;1:12.
Hunt AN. Dynamic lipidomics of the nucleus. J Cell Biochem
2006;97:244–51.
Milne S, Ivanova P, Forrester J, Alex Brown H. Lipidomics: an
analysis of cellular lipids by ESI-MS. Methods 2006;39:92–103.
Watson AD. Thematic review series: systems biology approaches to
metabolic and cardiovascular disorders. Lipidomics: a global
approach to lipid analysis in biological systems. J Lipid Res
2006;47:2101–11.
Wenk MR. Lipidomics of host–pathogen interactions. FEBS Lett
2006;580:5541–51.