| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Biophys J, September 2002, p. 1443-1454, Vol. 83, No. 3
and
*Institut de Biologie Physico-Chimique, Unité Mixte
de Recherche Centre National de la Recherche Scientifique 7099, Paris
75005 France; and Institut Curie, Section de Recherche,
Unité Mixte de Recherche-Centre National de la Recherche
Scientifique 168 and Laboratoire de Recherche
Correspondant-Commissariat à l'Energie Nucléaire 8, 75231 Paris cedex, France
 |
ABSTRACT |
|---|
 TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
The 31P-nuclear magnetic resonance chemical shift of phosphatidic acid in a membrane is sensitive to the lipid head group packing and can report qualitatively on membrane lateral compression near the aqueous interface. We have used high-resolution 31P-nuclear magnetic resonance to evaluate the lateral compression on each side of asymmetrical lipid vesicles. When monooleoylphosphatidylcholine was added to the external monolayer of sonicated vesicles containing dioleoylphosphatidylcholine and dioleoylphosphatidic acid, the variation of 31P chemical shift of phosphatidic acid indicated a lateral compression in the external monolayer. Simultaneously, a slight dilation was observed in the inner monolayer. In large unilamellar vesicles on the other hand the lateral pressure increased in both monolayers after asymmetrical insertion of monooleoylphosphatidylcholine. This can be explained by assuming that when monooleoylphosphatidylcholine is added to large unilamellar vesicles, the membrane bends until the strain is the same in both monolayers. In the case of sonicated vesicles, a change of curvature is not possible, and therefore differential packing in the two layers remains. We infer that a variation of lipid asymmetry by generating a lateral strain in the membrane can be a physiological way of modulating the conformation of membrane proteins.
| |
INTRODUCTION |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
Lipid asymmetry in biomembranes is often viewed
solely as a difference in chemical composition between the two
monolayers. For example, it is well known that the plasma membrane of
eukaryotic cells contains essentially phosphatidylcholine and
sphingomyelin in the outer monolayer, whereas aminophospholipids are
predominantly located in the inner monolayer (Devaux, 1991
). However,
lipid asymmetry can also mean a difference in surface density resulting from a difference in the number of lipids between the two coupled monolayers. Such a difference in lipid surface density can be achieved
artificially by the addition of lysophosphatidylcholine (LPC) to the
external surface of a lipid vesicle or of a biological membrane.
Because of the very slow flip-flop of lipids, the asymmetrical surface
density is generally stable. However, the mismatch causes a difference
in lateral pressure or tension between the two leaflets that often, but
not always, leads to vesicle shape change. In biological membranes, the
lateral pressure formed by an asymmetrical lipid distribution could
affect the structure and function of intrinsic membrane proteins.
We have used nuclear magnetic resonance (NMR) to probe the local
variations of the lateral pressure in various types of unilamellar vesicles. High-resolution 31P-NMR spectra of
phospholipid vesicles containing a mixture of dioleoylphosphatidylcholine (DOPC) and dioleoylphosphatidic acid (DOPA)
were recorded and the chemical shift of DOPA was monitored. The
potentiality of 31P-NMR spectroscopy for the
assessment of lateral compressibility in a membrane was brought to
light by Swairjo et al. (1994a). They showed that the
31P chemical shift of DOPA is different in the
inner and outer monolayers of sonicated unilamellar vesicles (SUVs).
They inferred that the difference in chemical shifts was due to the
difference in local curvature of the two monolayers, which is certainly
true in sonicated vesicles. In fact, via its sensitivity to the local
pKa, the chemical shift effectively reports on
the packing of the lipid head groups and can be considered as a
qualitative indicator of any change of surface tension.
NMR spectroscopy can be used to monitor qualitatively the surface
tension of vesicles of various sizes as long as they contain phosphatidic acid and that high-resolution NMR measurements are possible. We have investigated successively asymmetrical SUVs, LUVs
(large unilamellar vesicle) and FTVs (freeze-thaw vesicles). FTVs are
unilamellar vesicles obtained by repetitive cycles of freeze thawing, a
process that generates unilamellar vesicles with a distribution of
sizes but with a significant contribution of vesicles with a diameter
above 200 nm (Traïkia et al., 2000). With giant unilamellar
vesicles (GUVs) the lipid concentration would be too low to allow NMR
experiments to be performed. In the case of LUVs and FTVs magic angle
spinning was necessary to achieve high resolution. Shape changes were
visualized by cryoelectron microscopy.
In agreement with a former theoretical article (Farge et al., 1990), we
have found that the addition of lysophosphatidylcholine to one side of
LUVs or FTVs triggers an overall change of curvature that is
accompanied by an increased lateral compression in both leaflets and
not only on the external one. However, with sonicated vesicles, which
are spheres with the highest curvature that can withstand a
phospholipid bilayer, there is no possibility of changing the curvature
and the surface pressure increases solely in the outer monolayer.
| |
MATERIALS AND METHODS |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
Chemicals
Lipids (DOPC, DOPA, and monooleoylphosphatidylcholine (MOPC)) and chemicals (HEPES, KCl, and EDTA) were purchased from Sigma Chemical Co. (Saint Quentin Fallavier, France) and were used without further purification. The purity of all lipids was verified by thin layer chromatography and by high resolution NMR in chloroform.
Vesicles preparation
All samples were prepared as an initial dry lipidic film of
DOPC/DOPA (80:20 mol%) and the buffer was: 0.1 M HEPES, 0.1 M KCl, and
0.005 M EDTA at pH 8.0. SUVs (30 mg/mL) were prepared by sonication of
a lipid dispersion in buffer under a stream of argon using a probe type
sonicator (model VC50, Bioblock Scientific, Paris, France) at 40 W in
an ice bath until a clear solution was obtained. The sample was
afterwards centrifuged at 11,000 × g for 15 min to
ensure the removal of metallic particles. Unless otherwise mentioned,
LUVs were prepared by the reverse phase evaporation technique followed
by filtration of the preparation (50 mg/mL) through a polycarbonate
membrane, successively at 1, 0.4, and at 0.2 µm (Szoka et al., 1980;
Traïkia et al., 2000). The LUVs were concentrated to 100 mg/mL
by ultracentrifugation at 395,000 × g for at least
2 h (Beckman, TLC100). FTVs were prepared at 100 mg/mL.
Characterization of their lamellarity and size distribution was
described in Traïkia et al. (2000). Stock solution of MOPC was
prepared at 10 mg/mL in the buffer. Because of the large amount of
lipids necessary (several hundred microliters of 100 mM phospholipids in one sample), a rather accurate determination of the ratio of DOPC/DOPA or DOPC/DOPA/MOPC could be achieved from the weight of the
dry lipids and was always controlled by integration of the high
resolution NMR spectra recorded with magic angle spinning (MAS). In a
few instances, for SUVs or LUVs, the ratio of DOPC/DOPA was also
checked by NMR in organic solvent but not for all samples.
NMR
NMR experiments were performed and processed on a Bruker AVANCE
DMX400-WB NMR spectrometer (1H resonance at 400 MHz, 31P resonance at 162 MHz) using a Bruker
4-mm MAS probe with an external lock for the LUVs, FTVs, and MLVs
(multilamellar vesicles). A few experiments were carried out with an
insert-containing rotor, which allows better field homogeneity but
because of the low volume contained in such rotor longer accumulations
were necessary. The spinning speed of the 4 mm
ZrO2 MAS rotor was controlled to within 5 Hz at 8 kHz and the temperature (fixed at 8°C unless otherwise specified) was
calibrated as described in Traïkia et al. (1997). A 10-mm
broadband liquid state NMR probe was used for the SUVs experiments. The
31P-90° pulse lengths were 4.6 and 16 µs for
the MAS and the liquid state probes, respectively. The recycle delay
was 2 s and typical 31P spectral width was
20 ppm (3.24 kHz). 1H decoupling was not applied
during acquisition of the 31P magnetization. In
all experiments, 4096 complex points were acquired. Before Fourier
transformation, the data were zero filled to 8192 points, exponentially
multiplied with 10 Hz line broadening and treated with automatic
baseline correction. The theoretical frequency reference, 0 ppm,
corresponds in MAS spectra to phosphoric acid. As reported by Swairjo
et al. (1994a), we found that DOPC resonance at 8°C consist of a
single peak at 
0.64 ppm that can be used as an internal reference.
For broadband spectra, 0 ppm corresponds to the isotropic peak of DOPC
micelles. Other conditions for broadband 31P-NMR
spectroscopy were those of Traïkia et al. (2000).
Cryoelectron microscopy
A few microliters of the lipid mixtures were applied to a holey
carbon film on a copper grid that was held by tweezers mounted on a
shaft above a liquid ethane bath cooled by liquid nitrogen (Leica EM
CPC). The sample was blotted with filter paper and immediately plunged
into the liquid ethane (Dubochet et al., 1988). The vitreous specimens were transferred under liquid nitrogen to the
cryo-transmission electron microscope cold stage (model 626 Gatan),
which was inserted into the Philipps CM 120 electron microscope and
maintained at 170°C throughout specimen observation. Specimens were
imaged at 120 kV by low dose technique, and micrographs were recorded on Kodak SO163 films with a magnification of 45,000× and a 1-µm defocus.
| |
RESULTS |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
Sonicated unilamellar vesicles
Fig. 1 shows
31P-NMR spectra of DOPC/DOPA (80:20 mol%) SUVs
with and without MOPC added externally. The top spectra in Fig. 1 were
recorded at 45°C. At this temperature DOPC gives rise to two
resonance peaks partially overlapping. When at 45°C MOPC is added,
the lines slightly broaden and the two phosphatidylcholine peaks
collapse in agreement with the results of Kumar et al. (1989), who
found that 31P-T2 of POPC
decreases upon addition of LPC to SUVs. As reported previously by
Swairjo et al. (1994a), DOPA in SUVs gives rise to a doublet with two
well-separated peaks, the low field line corresponds to the phosphorus
on the external side of the vesicles and the high field line to the
inner leaflet molecules. We have carried out NMR experiments (not
shown) in the presence of 10 mM Pr3+ ions, which
selectively shifted the low field peak of DOPA and allowed us to
confirm the conclusion of Swairjo et al. (1994a). Integration of DOPA
peaks indicated roughly a ratio of 2 to 1, which was consistent with
the proportion of outer to inner surfaces of SUVs.
|
The two DOPA peaks are still well separated in the presence of MOPC.
The chemical shift of DOPA has been used to investigate the influence
of MOPC on the surface pressure of both leaflets. Because the position
and splitting between the two DOPA peaks are sensitive not only to the
presence of MOPC but also to the temperature, the pH and the ratio of
DOPA to DOPC (Swairjo et al., 1994a), it was important to maintain the
latter parameters fixed in all further experiments. A relatively low
temperature (8°C) was adopted to allow comparison with experiments
involving LUVs that required a low temperature for reasons that will be explained below.
Fig. 2 shows the effect on the
31P-NMR spectra of increasing concentration of
MOPC at 8°C. The two chemical shifts associated with DOPA molecules
are modified: 
PA(ext) moves up-field whereas PA(int) moves slightly down-field. At this
temperature there is no obvious change in the line width of the DOPC
peak and no obvious redistribution between the spectral intensities of
the two DOPA peaks. This suggests that addition of MOPC neither induced SUVs fusion nor caused their solubilization. Changes in
PA would simply reveal changes in lateral
pressure of each monolayer without important overall shape changes.
|
Large unilamellar vesicles
By cryoelectron microscopy, we found that the presence of an
important percentage of DOPA (20 mol%) in the vesicles made it difficult to obtain an homogeneous population of LUVs when prepared by
direct lipid extrusion under pressure according to the protocol designed by Cullis and collaborators (Hope et al., 1985). Fig. 3 shows some examples of extruded
DOPC/DOPA vesicles with odd shapes. Furthermore, we found that the
ratio of DOPA/DOPC was generally modified with a loss of DOPA. On the
other hand, when LUVs were prepared by the phase reverse technique
(Szoka etal. 1980), we obtained more homogeneous populations of
unilamellar vesicles with many spherical vesicles and almost no tubular
ones (Fig. 4). Besides, the proportion of
DOPC/DOPA was not modified. Addition of MOPC to the latter vesicles
provoked the formation of more elongated vesicles (Fig.
5, A and B).
Budding was observed but required the addition of 5% to 10% of MOPC.
|
|
|
Because glycerol was used in some NMR experiments described below, we have attempted to determine LUVs shapes with and without lysophosphatidylcholine addition in the presence of glycerol. Unfortunately when glycerol is present in samples examined by cryoelectron microscopy, the contrast is weak. Nevertheless, we found that all vesicles seemed to be more spherical in the presence of glycerol than without and that the addition of MOPC tended on the contrary to give homogeneous populations of perfectly spherical vesicles instead of generating elongated or eight-shapes vesicles (see above). Fig. 5 C was obtained with 25% of glycerol and 10% MOPC added externally. With 50% glycerol the vesicles were very difficult to see but seemed to be perfect spheres (data not shown).
When 31P-NMR spectra of LUVs or MLVs are recorded
without sample spinning, the resolution is poor and does not allow one
to separate the various phospholipid head groups, even if the spectrum is recorded at 45°C. Consequently, we have performed all further 31P-NMR experiments with 8 kHz sample spinning at
the magic angle (MAS). The 31P-NMR spectrum of
MLVs containing DOPA/DOPC exhibits a single narrow peak for DOPA at 2.2 ppm (Fig. 6 a). This is
consistent with the lipid surface density being the same in both
leaflets for lipid vesicles with large diameters. In the case of LUVs
with a diameter of ~200 nm, vesicle tumbling and phospholipid lateral diffusion on the surface of LUVs create a nonnegligible incoherent averaging that, in practice, diminishes the efficiency of magic angle
spinning (Traïkia et al., 1997). In that case, it is preferable to reduce the averaging due to thermal motion and thus, paradoxically, to reduce the temperature to obtain narrow lines (Fig. 6 b).
In fact, the best 31P-NMR spectra with LUVs were
obtained at 8°C after addition of 50% glycerol to minimize vesicle
tumbling (Fig. 6 c). However, besides the resolution
enhancement reported above, glycerol can modify the interface between
phospholipids and the aqueous environment due to the well known
dehydration effect of glycerol (Fenske and Cullis, 1993). This may
explain why, for example in the presence of 50% glycerol,
PA is moved up field by ~0.5 ppm in the
absence of any MOPC. For that reason, we have recorded spectra of
DOPA/DOPC-LUVs with and without glycerol despite the fact that in the
latter case the resolution was poorer and required the use of an
insert-containing rotor. In all instances, we obtained a single line
associated with DOPA (Fig. 6, b and c). When MOPC
was added, the DOPA peak was shifted up field as shown by the dotted
line spectra in Fig. 6, b and c.
|
Freeze-thawed vesicles
Cycles of freeze-thawing on lipid dispersions generate a
population of unilamellar vesicles (MacDonald et al., 1994;
Traïkia et al., 2000). Although these FTVs are not homogeneous
in size, the contribution of the larger vesicles (200-500 nm) is
dominant in the NMR spectrum (Traïkia et al., 2000). We have
recorded 31P-MAS-NMR spectra of FTVs obtained
after 10 freeze-thaw cycles of a DOPC/DOPA mixture (80:20 mol%) with
or without MOPC. In the absence of MOPC a single line was assigned to
DOPA (dotted curve in Fig. 6 a), indicating that the packing
of the outer and inner monolayers were very similar, thereby confirming
that the contribution of very small vesicles (with a diameter analogous
to that of sonicated vesicles) was not dominant in the spectra. When
MOPC was added externally to preformed FTVs, the phosphorus peak of
DOPA moved up field and there was still no indication of a line
splitting, which would correspond to DOPA located, respectively, in the
inner and the outer leaflets (Fig. 7).
Thus, as in the case of 200-nm LUVs, the chemical shift of DOPA present
in the inner leaflet of FTVs is modified by addition of MOPC in the
outer leaflet.
|
Finally we have carried out experiments with FTVs for which MOPC was mixed with DOPC/DOPA before vesicle formation. In that case, MOPC is likely to be present in both leaflets. These "symmetrical" vesicles, as expected, give rise to a single DOPA peak (not shown). The position of this peak moved slightly up field when MOPC concentration was increased. However, the variation of DOPA peak position with MOPC concentration in symmetrical vesicles was much smaller than in asymmetrical vesicles as shown in Fig. 8. The obvious difference between the influence of MOPC on the chemical shift of DOPA in asymmetrical versus symmetrical vesicles confirms that MOPC, when added externally, does not equilibrate rapidly between both leaflets. Nevertheless, the change in chemical shift observed at high MOPC concentration in symmetrical vesicles is indicative of a lateral strain in both leaflets. This requires some comments that will be given in the Discussion section.
|
Broadband spectra of LUVs in the presence of MOPC
The quantitative interpretation of the data summarized in Fig. 8
would require a precise knowledge of the fraction of MOPC added that
effectively is localized within the outer monolayer of each type of
vesicles. Because of the limited expansitivity of a lipid monolayer,
the addition of MOPC to only one monolayer in a lipid vesicle is likely
to be limited. To determine what proportion of MOPC added actually
resides in the membrane, we have examined broadband
31P-NMR spectra of DOPC/DOPA-LUVs in the presence
of increasing concentrations of MOPC (Fig.
9). If a fraction of MOPC remains in
micelles, it can be identified in broadband NMR spectroscopy by a
narrow peak in the central part of the spectrum. Ideally, such
experiments should be done with unilamellar vesicles with a diameter of
several hundred nanometers to avoid narrowing of the NMR spectrum by
vesicle tumbling. Because the 31P-NMR spectra of
LUVs with a diameter in the 100- to 200-nm range is partially narrowed
(Traïkia et al., 2000), we have used for these experiments LUVs
prepared by phase reversion and filtered with 400 nm pores. Fig. 9,
a to e, are broadband
31P-NMR spectra of 400-nm DOPC/DOPA-LUVs recorded
in the presence of increasing concentrations of MOPC added externally.
Fig. 9 f is the spectrum of MOPC in buffer recorded under
the same conditions and corresponds to micelles of MOPC. Fig. 9
a was recorded in the absence of MOPC. It is not the typical
spectrum of a homogeneous population of very large liposomes. Indeed,
the central part of the spectrum reveals the existence of a
distribution of sizes (Traïkia et al., 2000). This confirms
that the extrusion of liposomes with pore sizes of 400 nm does not lead
to a homogeneous population of unilamellar vesicles. In fact, these
vesicles probably contain bilamellar vesicles as reported by Mayer et
al. (1986). Nevertheless, such vesicles permit an easier evaluation of
the micellar contribution in a composite NMR spectrum than 200-nm LUVs
do.
|
Fig. 9 (b to e) shows that the addition of MOPC
gives rise to a central narrow peak, the intensity of which grows with
the proportion of MOPC added. At or above ~5% MOPC, the contribution of this peak is nonnegligible, which suggests that a significant fraction of MOPC does not incorporate in the membrane. Fig. 9, b, c, d, and e, cannot be
simulated by a linear combination of spectra a (spectrum of
liposomes) and f (spectrum of micelles) because the peak at
0 ppm is too broad (for example, see Fig. 9 e). Most likely,
the broadening is due to rapid exchange of MOPC between micelles and
membranes. By subtracting a fraction of spectrum a from
spectra b, c, d, and e
respectively, we have estimated the contribution of MOPC
nonincorporated into the vesicles and deduced the percentage of MOPC,
which was effectively in the membrane (Fig. 9 B). For sake
of simplicity the hypothesis used to draw the solid curve in Fig. 9
B was that all vesicles were unilamellar. If these
"LUVs" actually contain a nonnegligible proportion of bilamellar
vesicles (Mayer et al., 1986), then the proportion of MOPC per
phospholipid in the external monolayer is higher than indicated in Fig.
9 B. Hence, the curve corresponds to a lower limit of MOPC
percentage that is effectively incorporated into unilamellar vesicles.
| |
DISCUSSION |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
Asymmetrical insertion of lysophosphatidylcholine
MOPC was used here to create asymmetrical membranes. Long chain
lysophosphatidylcholine molecules have a critical micellar concentration in water in the micromolar range (Kumar and Baumann, 1991; Needham et al., 1997). Adding lysophosphatidylcholine to a very
concentrated suspension of unilamellar vesicles (several millimolar of
phospholipids) allows one to rapidly intercalate new molecules in the
outer surface. NMR experiments shown in Fig. 9 indicate
nevertheless that a fraction of MOPC is rapidly exchanging between
micelles and membranes, when the percentage of MOPC compared with total
phospholipids present is above 2% to 3%. At least 50% of MOPC added
penetrate the vesicle surface. There was no saturation under our
experimental conditions (Fig. 9 B).
NMR data (Fig. 9 A) and electron microscope
micrographs (Fig. 5) obtained with LUVs show that LUVs are not
destroyed, neither do they form smaller vesicles when 5% MOPC is
added. This is a strong indication that MOPC has no lytic activity.
Actually, more than 30% lysophosphatidylcholine can be mixed with
phosphatidylcholine without preventing bilayer formation (van Echteld
et al., 1981; Bhamidipati and Hamilton, 1995).
There are several direct reports in the literature about LPC flip-flop
rate in liposomes and in biological membranes showing that the
transmembrane diffusion of long chain LPC is extremely slow, even
slower than that of phosphatidylcholine. De Kruyff et al. (1977), van
den Besselaar et al. (1977), and Bhamidipati and Hamilton (1995)
reported that the rate of LPC flip-flop is so small in SUVs that it is
barely measurable. A half time of LPC flip-flop at 37°C above
100 h was estimated for SUVs and 46 h for hand-shaken
liposomes (van den Besselaar et al., 1977). Experiments showing shape
changes induced by the addition of LPC to erythrocytes (Sheetz and
Singer, 1974) or to unilamellar vesicles (Farge and Devaux, 1992;
Mathivet et al., 1996; Mui et al., 1995; this paper) can be understood
in the framework of the bilayer couple hypothesis if, and only if, LPC
does not equilibrate rapidly between the two monolayers. Needham and
Zhelev (1995) have reported that LPC flips rapidly with a half time of
a few minutes. However, they investigated the reorientation of LPC in
giant vesicles that were stretched by aspiration in a micropipette.
This constraint may accelerate lipid flip-flop (Raphael and Waugh,
1996).
Vesicles shape and surface tension
The shapes of giant unilamellar vesicles with an asymmetrical
lipid distribution has been documented in previous articles (Farge and
Devaux, 1992; Mui et al., 1995; Mathivet et al., 1996). Complete phase
diagrams of vesicles morphologies have been calculated by bending
energy minimization as proposed originally by Helfrich (1973; Svetina
and Zeks, 1989; Berndl et al., 1990; Seifert et al., 1991). A more
recent and elaborate model takes into account the lateral elasticity of
each lipid monolayers (Miao et al., 1994). Whereas theoretical shapes
can be confronted easily to experiments, the surface tension that is
responsible for shape changes is difficult if not impossible to measure directly.
If one assumes that the surface tension or lateral pressure of a lipid
bilayer in the resting state is zero, lateral pressure within each
monolayer manifests itself directly only in the lateral compressibility
when the membrane is subject to a lateral stress (Marsh, 1996). Each
layer of a lipid bilayer can be described as a two-dimensional elastic
surface with its own elastic moduli (Kin and
Kout, respectively). Typical values
for the area compressibility modulus of fluid bilayers are
KA 
0.14 N/m (Kwok and Evans,
1981). In the case of asymmetrical constraints applied to the two
leaflets, bilayer bending allows a membrane to partially relax the
compression or dilation of each leaflet. An elastic membrane should
continue to bend as long as the forces exerted on the two membrane
halves form a nonzero torque. Thus, at equilibrium, the stress of each layer should be the same but not necessarily zero. The mechanical response of a bilayer to a net addition of surface area due to the
insertion of N lipids in one monolayer comprising
initially N0 lipids has been
previously analyzed (Farge and Devaux, 1993; Mui et al., 1995). If
h is the thickness of the bilayer,
R0 the average radius of the vesicle
budding out, the areal strain for a mechanically symmetrical membrane
is approximately:
|
R0.
When MOPC is added to a membrane formed essentially of DOPC, this
formula applies if one assumes as a first approximation that the
cross-section of a MOPC molecule is identical to that of a DOPC
molecule. h/R0 appears as a
scaling parameter that modulates the influence of lipid asymmetry on
the lateral tension of a vesicle. As noted also by Mui et al. (1995), it may have a dramatic effect on the tension caused by the insertion of
new molecules. In the case of bud visible by optical microscopy (R 2 µm) on a giant liposome,
h/R0 is of the order of
2.5 × 10
3. Vesicle budding in
GUVs is triggered when
N/N0 is of the order of
102 (Farge and Devaux, 1992) or less (Berndl et
al., 1990). Then 
is ~2.5 × 105 and the surface tension due to the
compression of the outer monolayer is of the order of:
T = KA 0.35 × 105 N/m, which is negligible. On
the other hand, in the case of LUVs with a diameter of ~200 nm, the
bud may have a radius of ~25 nm (see Fig. 5 B), and
h/R0 0.2. In the latter
case, to induce such shape change by addition of MOPC,
N/N0 has to be of the order of 5% to 10%. Then 2 × 102 and T 3 × 103 N/m. A tension of that order of magnitude
is not far to the tension that can produce vesicle lysis by dilation of
the membrane (Kwok and Evans, 1981). It is sufficient to inhibit
surface undulations (Farge and Devaux, 1993). In other words, an
asymmetrical lipid distribution can generate an important surface
tension in LUVs. In those vesicles, tension builds up before shape
change takes place.
The extreme situation is that of SUVs, which cannot undergo shape change because their curvature is the highest possible for a lipid bilayer. Indeed, it is difficult to decrease or increase, by whatever means, the average area per lipid molecules in a single leaflet by more than 2% to 4% because of the limited range of elastic deformations. The inner layer of sonicated vesicles is likely to be compressed to the limit of what is possible before a change of phase or membrane collapse.
DOPA as a lateral pressure sensor
A basic assumption for the interpretation of the NMR experiments
is that the phosphorus chemical shift of phosphatidic acid allows one
to detect a lateral compression of the lipids near the aqueous
interface and not only membrane bending. A water/lipid interface
represents an interphase between a bulk aqueous phase of high
dielectric constant (
80) and the hydrophobic phase of a
membrane of very low dielectric constant. One consequence of the lower
dielectric constant is that the adsorption of cations and anions to the
ionizable groups of lipids is facilitated (Tocanne and Teissié,
1990). However, the actual ionization state of phosphatidic acid (PA) at the aqueous interface is also strongly dependent upon the
phosphate accessibility to the buffer. Close packing by reducing the
area available for the head-group phosphate allows the monoanionic form
of PA to be more favorable than the dianionic form. Titration curves at
various pH with sonicated dispersions of phosphatidylcholine/PA
mixtures allowed Swairjo et al. (1994a) to show that the difference in
chemical shift of PA in the two monolayers of SUVs can be explained by
a difference in the apparent pKa of the phosphate
moiety on the two sides of SUVs. They calculated a
pKa value around 12 in the inner leaflet instead
of 7 for a free phosphate. The pKa shift in the
inner leaflet would arise from the tight packing associated with the
inner leaflet curvature. On the other hand, the PA head-group in the
outer-leaflet would be in a more relaxed geometry that allows the
phosphorus to sense the buffer pH.
A priori, the expansion (or contraction) of each monolayer
is associated with a stress profile through the membrane (Gruner and
Shyamsunder, 1991; Cantor, 1997). Lipids labeled with
13C or 2H could be used to
investigate by NMR the entire stress profile (Chiu and Wu, 1990; Koenig
et al., 1997). 31P, on the other hand, is a
natural probe that enables one to explore easily at least the two
interfaces between phospholipids and water. Although it appears
difficult to relate theoretically in a quantitative and rigorous manner
the variation of DOPA chemical shift (PA) to
the lateral pressure or surface tension, the modification of PA is a convenient indicator of lateral stress
modifications. Providing the two membrane leaflets have the same lipid
composition and are facing the same buffer, identical values of
PA for the inner and outer leaflets reveal
identical stress. The actual modification of the stress profile in a
given experiment depends on what causes the membrane to bend. An
external constraint, for example the aspiration of the membrane in a
micropipette or actin polymerization pushing the membrane are processes
that are likely to stretch one monolayer and compress the other
monolayer (Marsh, 1996). But when a new lipid distribution, caused by
the insertion of drugs or by a flippase activity, is sole responsible
for the bending, the constraint is internal, i.e., the deformation must
reach a final equilibrium with zero torque as long as the membrane
remains elastic.
Our NMR experiments and those of Kumar et al. (1989) indicate that the
addition of LPC to SUVs has practically no influence on the line width,
hence on vesicles size and finally on the average curvature. Yet,
PA varied considerably in the outer leaflet
upon addition of MOPC, whereas it remained almost constant in the inner one. Because it is hard to imagine how the curvature could change on
one side of a membrane and remain constant on the opposite side, we
suggest that a change of PA should not be
associated systematically with a change of membrane curvature but
rather to compression/dilation near each interface accessible to the buffer.
Interpretation of data obtained with sonicated unilamellar vesicles
Compression of SUVs outer monolayer after the addition of MOPC is
indicated by the up-field shift of PA(ext).
This is not surprising; inserting MOPC in the outer monolayer should
allow less area per lipid (Fig. 10
A). Perhaps more surprising is the slight downfield shift of
PA(int), which suggests membrane dilation, hence a small increase of area per head group in the inner monolayer. However, because SUVs are spherical, there is no alternative to a
slight expansion of the outer surface upon insertion of new molecules
(elastic response). Because the inner monolayer adheres to the outer
monolayer, it can expand slightly and release some pressure in the
head-group region, which is probed by the phosphate chemical shift.
|
A similar observation was made by Swairjo et al. (1994b). The authors
reported that the addition of annexin V to SUVs resulted in the same
modification as the one reported here. They proposed a protein-induced
change in vesicle morphology that corresponds to reduced curvature.
Their data could be explained as well by implying a slight expansion of
the outer leaflet due to the interaction of annexin V with the lipid
head groups, involving perhaps a partial protein intercalation within
the outer monolayer. Simultaneously the inner leaflet would be slightly
relaxed as indicated by the sign of the change of
PA(int) reported by these authors. Kumar et
al. (1989) concluded from
31P-NMR-T2 measurements
that the addition of LPC to sonicated POPC vesicles tightens
phospholipid packing but only in the outer monolayer. In the present
study, we see a small variation of PA(int), it may simply tell that T2 and
PA, which are measurements of two different
magnetic resonance properties, are not simply correlated and may have
different sensitivity domains.
Large unilamellar vesicles and freeze-thawed vesicles
Comparison between Figs. 4 and 5 reveal that the addition of MOPC
to the outer leaflet of LUVs, triggers an overall shape change.
Although we have not seen the formation of protrusions and pseudopods
as reported by Mui et al. (1995), we saw the formation of elongated
vesicles and even eight shaped vesicles, at least in the absence of
glycerol (Fig. 5, A and B). The vesicles shapes shown in Fig. 5, A and B resemble giant vesicles
with an asymmetrical lipid distribution (Farge and Devaux, 1992), it
should be stressed that it was necessary to add 5% LPC to achieve
these transformations, whereas less than 1% mismatch suffice to obtain
these shapes with GUVs. These observations are consistent with the
scaling factor discussed above.
When LUVs were formed in the presence of 25% glycerol, their shapes
before addition of MOPC were rather spherical. Upon addition of MOPC,
there was no significant elongation. On the contrary they formed
homogeneous populations of perfect spheres (Fig. 5 C). Such
shape change implies an average change of volume: perhaps glycerol that
has a disordering effect on lipid acyl chains (Fenske and Cullis, 1993)
makes the bilayer more permeable to water.
In any case, when MOPC is added to large vesicles, it can be inferred
from the high resolution 31P-NMR spectra that
there is an increased surface pressure not only in the outer leaflet
but also in the inner leaflet because the signal corresponding to DOPA
remains a single peak and moves up-field. The high spectral resolution
in Fig. 6 excludes the possibility that
PA(ext) could move up-field, whereas
PA(int) would move down-field or even remain
constant. Thus, for LUVs (and FTVs) the insertion of MOPC in the outer
leaflet triggers both bending and surface tension. The important point
we want to emphasize here is that the lateral pressure not only
increases in the outer leaflet but also in the inner leaflet. Thus,
without necessitating a transfer of molecules the two monolayers are
subjected to a lateral stress. If the membrane was a limited sheet
without boundary conditions (Fig. 10 B), it would simply
bend until both layers were fully relaxed. However, the equilibrium
conformation for a closed membrane is that for which the strains in
both layers are identical and correspond to the same lateral pressure.
Symmetrical membranes containing MOPC
Fig. 8 shows that if MOPC is included in both monolayers of FTVs,
the variation of DOPA resonance position is very small at least as long
as the percentage of MOPC included in each monolayer is less than 10%.
Thus, symmetrical and asymmetrical distributions of MOPC have very
different effects. Clearly PA variation after addition of MOPC to the external monolayer cannot be explained by the
dilution of the DOPA head group. For symmetrical membranes, there is no
curvature expected from MOPC addition. On the other hand, adhesion of
the two monolayers, which have opposite spontaneous curvatures,
generates a "frustrated" flat bilayer with a constrained packing
near the phospholipid head groups of both leaflets (Fig. 10
C) (Charvolin and Sadoc, 1988; Marsh, 1996).
Biological relevance
We have shown that an excess of lipids as small as 2% to 3% on one side of a vesicles can generates surface tension. Such situation can be triggered in a biological membrane by the synthesis and release of new lipids on one side of a membrane or by the activity of a phospholipid translocase. For small organelles, with a diameter ~200 nm, a surface tension should appear even before a significant shape change (budding) takes place.
The shape change caused by the aminophospholipid translocase activity
in human erythrocytes (Seigneuret and Devaux, 1984) has led to propose
that the accumulation of phospholipids on the inner leaflet of the
plasma membrane of eukaryotic cells by the aminophospholipid
translocase could be involved at the early stage of endocytosis
(Devaux, 1991; Müller et al., 1994; Farge et al., 1999; Devaux,
2000; Rauch and Farge, 2000). The hypothesis that modulation of lipid
asymmetry could also modulate surface pressure and, by this process,
conformation of membrane protein has also been suggested (Bogdanov et
al., 1993). There are experimental observations actually demonstrating
the role of surface pressure on the modulation of membrane protein
function. Martinac et al. (1990) have shown that the addition of
amphiphiles such as lysolecithin or chlorpromazine to giant
Escherichia coli spheroplasts resulted in the opening of
mechanosensitive ion channels. The work of Martinac et al. (1990) shows
that compounds intercalated in the outer or in the inner leaflet have
the same physiological effect. This is a strong indication that the
mechanism involved is not the local curvature per se but rather the
lateral compression. The influence of lateral pressure on phospholipase
activity has been illustrated in monolayers (Bougulavsky et al., 1994)
and in large unilamellar vesicles (Lehtonen and Kinnunen, 1995). The
lytic activity of melittin is also strongly influenced by the tension of vesicles under osmotic stress (Benachir and Lafleur, 1996). Using
osmotic stress, Goulian et al. (1998) have shown that the tension of
giant vesicles can modulate the dimerization of gramicidin and hence
the opening of the channel.
Finally, we might speculate that the influence of the flippase family
on the lateral pressure of a membrane could be a way to regulate the
activity of transmembrane proteins and not only of mechanosensitive
channels. The fact that in reconstituted vesicles with a diameter of
~200 nm the P-glycoprotein activity was found quasi null (Rothnie et
al., 2001) or very small (Romsicki and Sharom, 2001) is perhaps the
consequence of the lateral pressure generated by the unidirectional
transport of lipids.
| |
ACKNOWLEDGMENTS |
|---|
This work was supported by grants from the Center National de la Recherche Scientifique (UMR 7099 and UMR 168), the Center d'Etude Atomique (LRC 8), and from the European Community (HPRN-CT-2000-00077).
| |
FOOTNOTES |
|---|
Address reprint requests to Philippe F. Devaux, Institut de Biologie Physico-Chimique, UMR CNRS 7099, 13 rue Pierre et Marie Curie, F75005 Paris, France. Tel.: 33-1-58-41-51-05; Fax: 33-1-58-41-50-24; E-mail: Philippe.Devaux{at}ibpc.fr.
Submitted April 17, 2001, and accepted for publication April 26, 2002.
| |
REFERENCES |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
Biophys J, September 2002, p. 1443-1454, Vol. 83, No. 3
© 2002 by the Biophysical Society 0006-3495/02/09/1443/12 $2.00
This article has been cited by other articles:
 |
|
 |
|
  I. Lopez-Montero, N. Rodriguez, S. Cribier, A. Pohl, M. Velez, and P. F. Devaux Rapid Transbilayer Movement of Ceramides in Phospholipid Vesicles and in Human Erythrocytes J. Biol. Chem., July 8, 2005; 280(27): 25811 - 25819. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 F.-X. Contreras, G. Basanez, A. Alonso, A. Herrmann, and F. M. Goni Asymmetric Addition of Ceramides but not Dihydroceramides Promotes Transbilayer (Flip-Flop) Lipid Motion in Membranes Biophys. J., January 1, 2005; 88(1): 348 - 359. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. M. Hanczyc, S. M. Fujikawa, and J. W. Szostak Experimental Models of Primitive Cellular Compartments: Encapsulation, Growth, and Division Science, October 24, 2003; 302(5645): 618 - 622. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||
) and inner (
) leaflets, respectively. The
uncertainty of the mol% of MOPC was estimated from lipid analysis in
organic solvent.
) LUVs with MOPC added externally and with
50% glycerol; (
) FTVs with MOPC added externally, no glycerol;
(
) FTVs with MOPC added externally, 25% glycerol; (
