Detergent-resistant membrane fractions contribute to the total
1
H NMR-visible lipid signal in cells
Lesley C. Wright
1
, Julianne T. Djordjevic
2
, Stephen D. Schibeci
2
, Uwe Himmelreich
5
, Nick Muljadi
3,4
,
Peter Williamson
2
and Garry W. Lynch
3,4
1
Centre for Infectious Diseases and Microbiology, Institute of Clinical Pathology & Medical Research;
2
Institute of Immunology &
Allergy Research;
3
Centre for Virus Research, Westmead Millennium Institute, the Cellular and Molecular Pathology Research Unit,
Department of Oral Pathology and Oral Medicine, Westmead Hospital Dental School and
4
National Centre for HIV Virology
Research, University of Sydney, Westmead Hospital, NSW, Australia;
5
Institute for Magnetic Resonance Research,

the NMR-visible mobile lipid of whole cells between intra-
cellular lipid droplets, where most of this lipid resides, and
detergent-resistant plasma membrane domains.
Keywords: lipid; membrane; domain; NMR; Triton X-100.
The origin of prominent
1
H NMR signals from lipids in
spectra from many different cell types has been the subject
of controversy for almost two decades. Currently, two
sources for the
1
H NMR-visible lipid have been suggested;
these are the mobile acyl chains of triacylglycerol and/or
cholesterol ester) localized to either membranes, or to
EM-visible intracellular lipid droplets [1]. Ferretti et al. [2]
concluded that these NMR signals originate from both
cytoplasmic lipid droplets and intramembrane amorphous
lipid vesicles.
Highly intense lipid resonances have been associated with
activation or proliferation of lymphocytes, macrophages
and neutrophils [3–5], as well as T cell lines, many cancer
cells, and cancer tissue both ex vivo and in vivo [6]. Other
cellular conditions linked with the appearance of NMR-
visible lipid include the antiproliferative effects of tetra-
phenylphosphonium chloride on a transformed breast cell
line [7], unstimulated human neutrophils in the presence
of high levels of free fatty acids [8], treatment of thymic
lymphocytes with anti-CD3 antibody [4], and the induction
of apoptosis or activation in Jurkat T-lymphoblasts [9]. The
conclusion to be drawn from these and many other studies

Eur. J. Biochem. 270, 2091–2100 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03586.x
phase, DRM structure is more ordered giving it character-
istics of being in the liquid-ordered (lo) phase [10]. DRMs
are rich in cholesterol and sphingolipids (including GM1
and GM3) and selectively retain a number of proteins
including CD4, GPI-anchored proteins such as CD48
[11,12], and proteins associated with T cell receptor (TCR)
signalling such as the src-family tyrosine kinase Lck, and
LAT [10]. In addition, TCR engagement with antigen or by
anti-CD3 antibody crosslinking leads to an increased
partitioning of TCR subunits and their associated signalling
molecules into DRMs implicating their central role as a
focal point for T cell activation. Interestingly, DRM-
associated proteins, such as caveolins and GPI-anchored
proteins, have been identified in the surface monolayer of
lipid droplets [13], indicating a possible link between lipid
droplets and DRMs [14].
ThesizeofDRMshasbeensuggestedbysometo
be > 100 nm [15], and by others to be around 26 nm
[16]. The smaller domain size coincides with that suggested
for NMR-visible microdomains [6]. In addition, parallels
can be drawn between the diversity of functions associated
with DRMs and the appearance of NMR-visible lipid in
cells. These factors, as well as the presence in lipid droplets
of known components of rafts and caveoli, such as
cholesterol and caveolin, prompted us to investigate
whether DRMs contain NMR-visible lipid. In this study
we show that DRMs containing NMR-visible lipid can be
isolated from several cell lines under conditions where
mobile lipid is evident in the

Cell culture and stimulation
All cell lines were maintained in RPMI/10% fetal bovine
serum. For Jurkat cell stimulation, cells were seeded into
serum-free RPMI/0.1% BSA (10
6
ÆmL
)1
) and incubated for
24 h (viability > 80%). Harvested cells were then seeded
into RPMI/10% fetal bovine serum and were either left
unstimulated or were stimulated with PMA (30 ngÆmL
)1
)
and ionomycin (300 ngÆmL
)1
) for 24 h.
Preparation and fractionation of plasma membranes
Jurkat and CEM-T4 cell membranes and cytosol, prepared
from 1 and 5 · 10
8
cells, respectively, were solubilized with
1% Triton X-100 (v/v) and fractionated by 5–40% sucrose
gradient centrifugation as described previously [17]. THP-1
cells were used for quantitative NMR and lipid analytical
work. Membranes from THP-1 cells (1.56 · 10
9
)were
separated from the cytosol by ultracentrifugation at
105 000 g for 60 min. The membrane-containing pellet
was then solubilized with 1% Triton X-100 and fractionated

software (Becton Dickinson).
1
H NMR analysis of whole cells, cell fractions
and sucrose gradient fractions
Sample Preparation. Cells (5 · 10
7
) were washed three
times in NaCl/P
i
(–) containing 0.1 mgÆmL
)1
BSA, then
washed and resuspended in 400 lLNaCl/P
i
(–) made up in
2
H
2
O. Homogenates, supernatants, floating lipid droplets
and gradient fractions were dialyzed against NaCl/P
i
(–), and
NaCl/P
i
(–)
2
H
2
O added to a final concentration of 10%.
Membrane pellets were washed three times in NaCl/P

From the biochemical fatty acid analysis it was estimated that
the average number of such CH
2
groups per fatty acid residue
was 9.1. This was used for calculation of the fatty acid residue
concentrations based on the NMR integrals. For most
fractions, only trace amounts of valine, leucine, threonine
and isoleucine contributed also to the 1.3 p.p.m. resonance.
However, for the THP-1 supernatant fraction, a substantial
contribution from these amino acid residues was noted,
resulting in overestimation of the fatty acid concentration by
up to twofold. Some contribution from amino acids was also
noted in the membrane and DSM fractions from THP-1 cells.
NMR spectroscopy. For both cells and gradient fractions,
NMR spectra were acquired at 37 °C with the sample
spinning at 20 Hz, using a Bruker Avance 360 MHz
spectrometer; parameters for NMR spectroscopy of the
cells were essentially as in [8]. 1D
1
HNMR spectra of
gradient fractions were run using a selective excitation field
gradient method of water suppression [18], a spectral width
of 4000 Hz, 256 accumulations, and total acquisition time
per transient of 3.14 s. 2D NMR spectroscopy of sucrose
gradient fractions was carried out by acquiring
1
H,
1
H
COSY NMR spectra in magnitude mode. Remaining water

with iodine vapour and also stained with sterol spray
reagent and Coomassie Blue [8]. Their identities were
confirmed by comparison with authentic standards.
Lipid analyses
Cholesterol and triacylgycerol estimations were performed
using the Roche Modular Analytical System with CHOD-
PAP methodology for cholesterol and GPO/PAP metho-
dology for triacylgycerols (Roche). Fatty acid analyses were
conducted on saponified lipid extracts converted to fatty
acid methyl esters by acid methanolysis. These were
separated on an HP Series II 5890 gas chromatograph with
an Agilent Ultra 2 capillary column 19091B-102, using the
method of MIDI Inc. (Delaware, USA). Results are
expressed as percentages of the sum of the areas of all
peaks identified.
Results
Examination of stimulated Jurkat cells by NMR
spectroscopy
Stimulation of Jurkat T cells with PMA/ionomycin was
assessed using flow cytometry as shown in Fig. 1A.
Augmentation of the fluorescence levels attributable to
surface expression of the T-cell activation markers, IL-2
receptor alpha chain (CD25) and CD69, was observed
following stimulation.
Fig. 1. Examination of stimulated Jurkat T cells by flow cytometry and
1
H NMR spectroscopy. PMA and ionomycin-stimulated and non-
stimulated Jurkat cells were incubated with anti-IgG
1
(isotype control),

ratio increased from 1.44 to 3.02 after stimulation, with a
concomitant increase in the CH
2
/choline peak height ratio
from 0.44 to 2.82, due to a decrease in choline. These trends
have also been observed previously [4,9].
Examination of plasma membrane fractions
from stimulated Jurkat cells by NMR spectroscopy
Next we examined the plasma membrane distribution of
mobile lipids by fractionating plasma membranes on the
basis of Triton X-100 solubility and buoyant density using
sucrose gradient ultracentrifugation. The separation of lipid
domains to the lighter, detergent-insoluble gradient frac-
tions (DRMs, 3–6) was indicated indirectly by detecting the
DRM-resident proteins Lck and CD48, and directly by
detecting the glycosphingolipid GM1 with cholera toxin
(Fig. 2A–C). The DRM protein markers were also detected
in the high density fractions (9–10) which contain cytosolic
material and Triton-soluble membrane (DSM) components.
Lck and CD48 were, however, preferentially associated with
DRMs, which contained less than 2% of the total protein
found within the DSM fractions (determined by densito-
metric scanning of Coomassie Blue-stained gels).
When the DRM and DSM fractions were analysed for
mobile lipid content by
1
H NMR, lipid CH
2
and CH
3

resonance was much more prominent (Fig. 3A). No
increase in mobile lipid was observed upon stimulation
(data not shown). NMR-visible lipid and carbohydrate (but
not choline) were again present in the DRM fractions (4–5)
isolated from these cells (Fig. 3A), but this lipid was not
observed in the DSM fractions (not shown). The distribution
of CD4 protein is shown in Fig. 3B and demonstrates that a
portion of CD4 is present in the DRM fractions containing
the NMR-visible lipid. The DRM fractions contained about
2% of the total membrane protein content.
THP-1 cells and membranes. NMR spectra of intact cells
of the human monocytoid cell line THP-1, were clearly
dominated by protons arising from lipid (Fig. 4A). The
CH
2
/CH
3
peak height ratio was calculated to be 5.2, which
was much higher than the equivalent signal observed in
stimulated Jurkat cells. In contrast, the intensity of the
choline resonance was very low, compared with the
unstimulated Jurkat or CEM-T4 cells. The THP-1 cells
did not increase their NMR-visible lipid when stimulated
(data not shown).
As with CEM-T4 DRM fractions, THP-1 DRM fractions
4–6, which contain less than 2% of the total membrane pro-
tein, were found to localize specifically the DRM-associating
proteins CD4 and Hck, without contamination by either of
the abundant endoplasmic reticulum (protein disulphide
isomerase) [19] or cytoskeletal (tubulin) representative

H correlation spectroscopy
(COSY) confirmed that the resonance at 1.3 p.p.m. was
indeed derived mainly from lipid (Fig. 6). The crosspeaks
labelled A–G¢ in Fig. 6 indicate spin-spin coupling between
protons on adjacent carbon atoms. Crosspeaks A–F have
previously been assigned to resonances from acyl chain
protons found in triacylglycerol and/or cholesterol esters
[20,21]. These could also arise from phospholipids. G¢
is derived from the glycerol backbone of triacylglycerol
[20]. The intensities of crosspeaks C and D (– CH
2
-CH
2
-
CH¼CH- and ¼CH-CH
2
-CH¼CH-, respectively) indicate
that relatively large amounts of unsaturated fatty acid
residues are present. The absence of resonances from
lactate, threonine, valine, leucine and isoleucine was
confirmed by COSY and spin-echo experiments
(TE ¼ 135 ms, data not shown).
Distinguishing the NMR-visible lipids of DRMs
and cytosolic lipid droplets
Most of the cytosolic lipid droplets floated as a visible, milky
layer on the top of the supernatant formed when the plasma
membrane fraction was sedimented from the cellular
homogenate. The remainder of the lipid droplets were
2094 L. C. Wright et al.(Eur. J. Biochem. 270) Ó FEBS 2003
observed floating on top of fraction 1 of the sucrose gradient

We calculated that approximately 12.4% of the total
Fig. 2. Identification and examination of Jurkat DRMs by
1
HNMR
spectroscopy. Proteins in sucrose gradient fractions from nonstimu-
lated Jurkat cells were fractionated by SDS/PAGE, electrotransferred
to nitrocellulose and immunoblotted with antibodies to DRM protein
markerssuchasLck(A)orCD48(B)(1lgÆmL
)1
) followed by ECL
detection as in the Materials and methods. Sucrose gradient fractions
were also spotted onto nitrocellulose and probed with CT-B
(2 lgÆmL
)1
) followed by ECL detection of GM1, a DRM lipid marker
(C). The
1
H NMR spectra of dialysed DRMs from 10
8
stimulated and
nonstimulated Jurkat cells are compared in D, indicating the presence
of NMR-visible CH
2
resonances at 1.3 p.p.m. in the former, and low
amounts of these resonances (sometimes below the limits of detection,
as shown here) in the latter.
Fig. 3. Examination of CEM-T4 cells and DRMs by
1
HNMR
spectroscopy. The

differed considerably in their content of these two neutral
lipids (Table 1). As expected, the DRM fractions were
enriched in total cholesterol relative to triacylglycerol
(cholesterol/triacylglycerol ratio of 3.0) when compared
with the plasma membrane fraction from which they were
derived (ratio of 1.5), and the DSM fractions (ratio 1.0). The
DRM fraction contained about 35% of the total cellular
cholesterol and 23% of the total cellular triacylglycerol.
Conversely, the lipid droplet fraction was enriched in
triacylglycerol (cholesterol/triacylglycerol ratio 0.6), as was
the lipid droplet fraction floating on the top of the sucrose
gradient (ratio 0.5, not shown). The lipid droplet fraction
contained 15.4% of the total cellular triacylglycerol, and this
fraction plus the supernatant fraction (which presumably
also contains lipid droplets) contained 40.3% of the cellular
triacylgycerols.
The major neutral and polar lipid components of DRM
and DSM fractions from CEM-T4, THP-1 and Jurkat cells
were examined qualitatively by TLC. The dominant neutral
lipid in the Jurkat DRM fractions (from both stimulated
and nonstimulated cells) was free cholesterol, with a small
amount of triacylglycerol and free fatty acid. CEM-T4
DRM fractions contained mainly free cholesterol and
cholesterol ester, whereas THP-1 DRMs, by contrast,
contained predominantly free cholesterol, triacylglycerol,
Fig. 5. Examination of THP-1 gradient fractions by
1
HNMRspectro-
scopy. Spectra of dialysed DRM fractions (4–6) isolated from
1.56 · 10

membranes, proteins in each sucrose gradient fraction were subjected
to SDS/PAGE, electrotransferred to nitrocellulose and immunoblot-
tedwithantibodiestoDRMmarkers,CD4andHck,andantibodiesto
protein disulphide isomerase (PDI, endoplasmic reticulum) and tub-
ulin (cytoskeleton), followed by ECL detection, shown in B. Lysates
were also loaded as controls.
2096 L. C. Wright et al.(Eur. J. Biochem. 270) Ó FEBS 2003
and small amounts of cholesterol ester, ether-linked triacyl-
glycerol and free fatty acids. Phosphatidylcholine (PtdCho),
sphingomyelin and other phospholipids were present in the
polar lipid component of all DRMs.
DSM fractions from CEM-T4 cells contained mainly free
cholesterol and PtdCho, and those from THP-1 cells
contained mainly cholesterol ester, triacylglycerol and
PtdCho. Small amounts of other phospholipids were
detected also. Apart from GM1, the glycolipid content of
DRM and DSM fractions was not determined.
The fatty acid composition of the total lipids extracted
from the various fractions of the THP-1 cells is shown in
Table 2. Surprisingly, the DRM fractions were not enriched
in saturated fatty acids, relative to the cell homogenate and
the total membrane fraction, except for a very small increase
in myristic acid (14:0). Rather there was a small increase in
palmitoleic acid (16:1), 18:2 + 18:3, and arachidonic acid
(20:4) at the expense of palmitic, stearic and oleic acids (16:0,
18:0, 18:1, respectively). The DSM fractions contained
higher levels of 18:0, 18:2 + 18:3 and 18:1 at the expense of
16:0, 14:0 and 16:1, whereas there was a marked increase in
the amount of 16:0 and a decrease in the amount of 16:1 and
17:1 in the lipid droplet fraction (Table 2). The fatty acid

1.3 p.p.m. in the lipid droplets, but were obviously less prominent in
the membrane-containing pellet (B) obtained after removal of the lipid
droplets and supernatant and resuspended in NaCl/P
i
(–)
2
H
2
O. PABA
(20 lLofa10-m
M
solution) was added as internal standard to both
samples.
Table 1. Cholesterol, triglyceride and NMR-visible lipid content of
THP-1 cell fractions. Results are expressed as lmol of CH
2
equivalents
(NMR-visible lipid) or lipid species (cholesterol and triacylglycerols)
per 1.56 · 10
9
cells. The total cellular content is the sum of the top four
fractions. Percentage compositions are shown in brackets. The method
for calculation of the amount of NMR-visible lipid is described in the
Materials and methods section.
Fraction
NMR-visible
lipid Cholesterol Triacylgycerols
Low speed pellet 2.47(13.7) 0.90(10.4) 0.42(9.7)
Membranes 4.30(23.8) 3.41(39.4) 2.17(50.0)
Supernatant 8.46(46.8) 3.93(45.5) 1.08(24.9)

can now visualize by 1D and 2D
1
H NMR spectroscopy the
mobile lipid component of such domains, which we have
found to have a protein and lipid composition characteristic
of DRMs or rafts, as described by others. While the
membrane fractionation procedure may have altered the
physical state of the lipids from that in the whole cells,
the resonances appearing in DRM spectra are probably
derived from the lipid acyl chains of triacylgycerols and
cholesterol esters, as well as small amounts of free fatty
acids. Some contribution could also come from the acyl
chains of PtdCho or sphingomyelin; however, neither the
choline headgroups of PtdCho and sphingomyelin (except
for a small peak in THP-1) nor the sphingosine chain of
sphingomyelin were visible in the NMR spectra of DRMs.
In bilayer environments, the choline headgroups are relat-
ively immobile and therefore of low visibility in NMR
spectra [7]. This suggests that headgroup mobility might
also be restricted in DRMs from some cells.
Although DRMs from three different cell types produced
NMR spectra dominated by resonances from lipid acyl
chains, we found differences in the composition of the
Fig. 8. Analysis of THP-1 DRMs and lipid droplets for GM1 and total
protein. (A) THP-1 DRMs were isolated from total membranes by
sucrose gradient fractionation as described in Materials and methods.
Lipid droplets, prepared as in the legend to Fig. 7, were subjected to
sucrose density gradient fractionation to remove any cytoplasmic
proteins and were collected from the top layer (fraction 1). Pooled
DRMs (fractions 4–8) and lipid droplets (0.5 lL of each) were spotted

Jurkat, were comprised mainly of cholesterol ester and free
cholesterol/triacylgycerols, respectively. This indicates that
just as there are cell-specific differences in the proteins found
in DRMs, with cell-specific functions, there is cell-specific
variability in the neutral lipid content of DRMs. While no
quantitation was performed on the DRM lipids from
stimulated and nonstimulated Jurkat cells, it appeared that
the same components were present, therefore either an
increase in the amount of DRMs present, or in the amount of
lipid in the DRMs from the stimulated cells may explain their
more intense lipid spectra. The fact that DRMs were difficult
to detect in both stimulated and nonstimulated Jurkat cells
may be due to the qualitatitive observation that little neutral
lipid was present in this cell line and its DRMs, and also the
use of much smaller cell numbers used to prepare gradients
(10
8
), compared with the THP-1 cells (1.56 · 10
9
). While
some neutral lipids, as well as phospholipids, were present in
DSM fractions, their high protein to lipid ratio may account
for the observation of very little NMR-visible lipid.
Although triacylgycerols and cholesterol esters have been
observed in intact cells and pure plasma membrane fractions
by
1
H NMR spectroscopy and chemical analysis [5,8,20,21],
triacylgycerols, as described for the THP-1 membrane
fractions, have not previously been identified as a compo-

DRM lipids (12.4%) to the THP-1 cell spectrum is a
minimum percentage. Although no nonlipid contribution to
the lipid signal at 1.3 p.p.m. could be detected in the low
speed pellet, lipid droplets and DRM fractions, there was
(as expected) a large contribution from nonlipid compo-
nents in the supernatant fraction (see Materials and
methods). The supernatant contains only 25% of the
cellular triacylgycerol, but contributes 46.8% towards the
mobile ÔlipidÕ signal (Table 1), therefore this latter figure
would be a considerable overestimate, leading to an
overestimation of the total cellular lipid signal. Because in
THP-1 cells much of the lipid signal would derive from
triacylgycerol, a better estimate may be obtained from the
triacylgycerol content of the DRMs, namely 23% of the
total, and for the supernatant plus lipid droplet fraction a
figure of 40% might be more accurate (Table 1). Thus we
would estimate the contribution of the DRMs to the cellular
NMR-visible lipid in the range of 12–23%, with lipid
droplet contribution at around 40%. Interestingly, the
contribution of the total membrane fraction to the NMR-
visible lipid is around 24%, and as the amount of
triacylgycerol in this fraction is 50% of the total, almost
all of the NMR-visible triacylgycerol in membranes must
reside in the DRM fraction.
It could be argued that the NMR signal detected in
DRMs is merely contamination from cytoplasmic lipid
droplets. This is unlikely, for the following reasons. Firstly,
during method development, a lipid droplet preparation
from THP-1 cells (rich in NMR-visible lipid) was incubated
with membranes from CEM-T4 cells (poor in NMR-visible

found, with growing evidence of direct physical continuities
between lipid droplets and bilayer membranes [25,27]. This
deserves further investigation, especially the possibility of
translocation of lipid from droplets to DRM domains
during stimulation of some cells (e.g. Jurkats), with
subsequent effects on cell function.
Ó FEBS 2003
1
H NMR of cells/Triton-insoluble membrane fractions (Eur. J. Biochem. 270) 2099
Acknowledgements
We would like to thank Ms Leanne Hicks, Department of Infectious
Diseases, for the fatty acid analyses, and the Department of
Biochemistry (Core Pathology), Westmead Hospital for the neutral
lipid analyses. At the time of re-submission of this manuscript, we
discovered that A. Ferretti et al. have quite independently and
simultaneously come to similar conclusions, namely that NMR-visible
lipid is present in DRMs. Their work is now available in the European
Biophysical Journal online, DOI 10.1007/s00249–002-0273-8 as of
January 2003.
References
1. Hakuma
¨
ki, J.M. & Kauppinen, R.A. (2000)
1
HNMRvisible
lipids in the life and death of cells. TIBS 25, 357–362.
2. Ferretti,A.,Knijn,A.,Iorio,E.,Pulciani,S.,Giambenedetti,M.,
Molinari, A., Meshini, S., Stringaro, A., Calcabrini, A., Freitas, I.,
Strom, R., Arancia, G. & Podo, F. (1999) Biophysical and
structural characterization of

1
H NMR-visible tri-
acylglycerol in human neutrophils. High fatty acid environments
result in preferential sequestration of palmitic acid into plasma
membrane triacylglycerol. Eur. J. Biochem. 267, 68–78.
9. Di Vito, M., Lenti, L., Knijn, A., Iorio, E., D’Agostino, F.,
Molinari, A., Calcabrini, A., Stringaro, A., Meschini, S., Arancia,
G.,Bozzi,A.,Strom,R.&Podo,F.(2001)
1
H NMR-visible
mobile lipid domains correlate with cytoplasmic lipid bodies in
apoptotic T-lymphoblastoid cells. Biochim. Biophys. Acta 1530,
47–66.
10. Brown, D.A. & London, E. (1998) Functions of lipid rafts in
biological membranes. Ann. Rev. Cell Dev. Biol. 14, 111–136.
11. Parolini,I.,Topa,S.,Sorice,M.,Pace,A.,Ceddia,P.,Montesoro,
E., Pavan, A., Lisanti, M.P., Peschle, C. & Sargiacomo, M. (1999)
Phorbol ester-induced disruption of the CD4-Lck complex occurs
within a detergent-resistant microdomain of the plasma mem-
brane. Involvement of the translocation of activated protein
kinase C isoforms. J. Biol. Chem. 274, 14176–14187.
12. Stefanova,I.,Horejsi,V.,Ansotegui,I.J.,Knapp,W.&Stock-
inger, H. (1991) GPI-anchored cell-surface molecules complexed
to protein tyrosine kinases. Science 254, 1016–1091.
13. Pol,A.,Luetterforst,R.,Lindsay,M.,Heino,S.,Ikonen,E.&
Parton, R.G. (2001) A caveolin dominant negative mutant
associates with lipid bodies and induces intracellular cholesterol
imbalance. J. Cell Biol. 152, 1057–1070.
14. Prattes, S., Ho
¨

of Chinese hamster ovary cells. Eur. J. Biochem. 205, 827–839.
22. Davis, R.A. (1991) Lipoprotein structure and secretion. In:
Biochemistry of Lipids, Lipoproteins and Membranes (Vance, D.E.
& Vance, J., eds), Vol. 20 p. 419. Elsevier, Amsterdam.
23. Hamilton, J.A. (1989) Interactions of triglycerides with phos-
pholipids: incorporation into the bilayer structure and formation
of emulsions. Biochemistry 28, 2514–2520.
24. Gorrissen, H., Tulloch, A.P. & Cushley, R.J. (1982) Deuterium
magnetic resonance of triacylglycerols in phospholipid bilayers.
Chem. Phys. Lipids 31, 245–255.
25. Murphy, D.J. (2001) The biogenesis and functions of lipid bodies
in animals, plants and microorganisms. Prog. Lipid Res. 40, 325–
438.
26. Ostermeyer,A.G.,Paci,J.M.,Zeng,Y.,Lublin,D.M.,Munro,S.
& Brown, D.A. (2001) Accumulation of caveolin in the
endoplasmic reticulum redirects the protein to lipid storage drop-
lets. J. Cell Biol. 152, 1071–1078.
27. Murphy, D.J. & Vance, J. (1999) Mechanisms of lipid-body for-
mation. TIBS 24, 109–115.
2100 L. C. Wright et al.(Eur. J. Biochem. 270) Ó FEBS 2003