Traffic 2000 1: 605–613
Munksgaard International Publishers
Review
Greasing Membrane Fusion and Fission Machineries
Koert N.J. Burger
Department of Molecular Cell Biology, Institute of
Biomembranes, Utrecht University, Padualaan 8, 3584 CH
Utrecht, The Netherlands
[email protected]
Biological membrane fusion is a local-point event, extremely fast, and under strict control. Proteins are responsible for the mutual recognition of the fusion
partners and for the initiation of biomembrane fusion,
and thus determine where and when fusion occurs.
However, the central event during membrane fusion is
the merger of two membranes, which requires a transient reorganization of membrane lipids into highly
curved fusion intermediates. This review focuses on the
potential role of lipids in the generation of membrane
curvature, and thus in the regulation of membrane fusion and fission.
Key words: Fission, fusion, lipid polymorphism,
lysophosphatidic acid, membrane curvature, phosphatidic acid, type-II lipids
Received 15 May 2000, revised and accepted for publication 22 May 2000
Membrane fusion is a universal event in the functioning of a
living organism. It is crucial for the formation of the mature
muscle fiber, for exocytosis and endocytosis, and for intracellular membrane traffic, and is used by enveloped viruses to
enter and infect cells (1–3). Membrane fusion is energetically
unfavorable and does not occur spontaneously. Instead, specific molecules are responsible for the induction of biomembrane fusion in vivo. Proteins determine the specificity of
biomembrane fusion, and there is little doubt that proteins
are also directly involved in the induction of membrane fusion. However, it has proven extremely difficult to discriminate fusogenic factors directly involved in the induction of
membrane fusion from secondary factors only involved in
processes that precede or follow actual membrane coalescence. Only in the case of virus–membrane fusion, have the
fusogenic proteins directly responsible for the induction of
membrane fusion been identified unequivocally (4).
Important clues on the general mechanism of biomembrane
fusion were obtained using morphological and electrophysiological methods, which are among the few techniques in
membrane fusion research that can be applied to the intact
biological system. In 1979, new electron microscopy (EM)
techniques became available in which the structure of a
biological specimen is stabilized by fast freezing avoiding the
changes in sample ultrastructure induced by the chemical
pretreatments used during conventional EM techniques (5).
The large-scale removal of membrane proteins from the
contact area and the large single-bilayer diaphragms observed during conventional EM analysis of biomembrane
fusion proved to be artifacts (6). Using fast freezing, biomembrane fusion is consistently observed as a local-point event
that involves an area of the interacting membranes of less
than 20 nm in diameter [reviewed in (7)]. This local-point
fusion concept of biomembrane fusion was later confirmed
by patch-clamp analysis, showing that fusion starts with the
formation of a minute fusion pore with a diameter of only
1 – 3 nm (8). Importantly, the local-point fusion concept also
applies to pure lipid model membrane fusion, where EM
analysis revealed the presence of very small fusion intermediates, ‘lipidic particles’, suggesting a common lipidic intermediate in model and biological membrane fusion (9,10). The
fact that membrane fusion is a local-point event, implies that
fusion only requires a local restructuring of the interacting
lipid bilayers, and that relatively few lipid molecules (hundred
to a few thousand) are directly involved in the fusion process.
Since there is no large-scale removal of proteins from the
contact area, proteins can be directly involved in biomembrane destabilization and the induction of membrane fusion.
The regulation of biomembrane fusion is dominated by
proteins, but the central event during membrane fusion is the
merger of two membranes, which requires a transient reorganization of membrane lipids into highly curved fusion intermediates. The potential role of lipids in the generation of
these intermediates and thus in the regulation of membrane
fusion and fission is discussed below.
Molecular Models of Membrane Fusion
In the end, membrane fusion requires close apposition and
coalescence of a small area of the lipid bilayers. A progressive force is required to drive two bilayers closer than 3 nm
together (11). Two forces dominate, a repulsive hydration
force arising from water tightly bound to the lipid headgroups, and an attractive hydrophobic force between the
hydrocarbon interiors of the membranes. At a pressure of
about 100 atm, the attractive hydrophobic force locally bypasses the repulsive hydration force and the bilayers semifuse. The introduction of membrane defects exposes more
of the membrane interior and substantially reduces the pressure required for semifusion. This is likely to be a general
theme in all membrane fusion events. During biomembrane
fusion, defects in membrane lipid packing could arise from
extreme membrane curvature, local changes in lipid composition, or the insertion or de-insertion of a stretch of hydrophobic amino acids (1,2,12).
605
Burger
With few exceptions [see (13,14)], membrane fusion is envisaged as a ‘non-leaky’ process involving three distinct
stages: membrane adhesion, semifusion, and pore formation
(or vice versa, during membrane budding and fission). At the
semifusion stage of membrane fusion, the outer (cis ) lipid
monolayers of the two membranes are continuous but the
inner (trans ) lipid monolayers are still separated. Experimental support for the existence of a semifusion intermediate in
both model and biological membrane fusion has been obtained (15–18). Two main molecular models have been proposed for the semifusion intermediate. In both models, the
semifusion intermediate has a strong net negative (concave)
monolayer curvature. In the first model, two membranes join
to form an inverted lipid micelle at the semifusion stage (19),
while the second, and currently prevailing model, predicts
the formation of a so-called stalk (20). In a modification of the
stalk model (21,22), the stalk develops into a trans monolayer
contact, which is short-lived and ruptures resulting in the
formation of a small pore (Figure 1). It is important to note
that the evaluation of the different models of membrane
fusion is dominated by theoretical considerations, and that
direct data on the structure and composition of the fusion
intermediates (in the end in biomembrane fusion) are absolutely required before a definite answer concerning the
molecular nature of the semifusion intermediate can be
given. Nonetheless, the stalk mechanism is compatible with
a wide range of observations on model membrane fusion,
and there is increasing evidence that similar intermediates,
with a net negative monolayer curvature, mediate biomembrane fusion (23).
Figure 1: Stalk model for the intermediate stages during
membrane fusion (20, 21): Adhesion (A), semifusion (B), and
pore-formation (C). Semifusion starts with the formation of a
stalk (B-1), which develops into a trans monolayer contact (B-2).
Hydrophobic interstices are shaded.
606
The Role of Lipids in Biomembrane Fusion
During patch-clamp analysis of biomembrane fusion, the
early fusion pore is often seen to reversibly open and close
several times, observed as capacitance flickering, before
expanding irreversibly. A key question concerns the nature of
the initial, transient pore. Is the pore formed by a multimeric
protein complex, or is the pore a mixed protein-lipid or an
entirely lipidic structure (8,24)? Because direct data on the
structure and composition of the fusion pore are lacking, a
definitive answer to this question cannot be given. However,
lipids are likely to play a key role in the induction of biomembrane fusion for several reasons. First, proteins are not an
absolute prerequisite for membrane fusion; pure lipid model
membranes can be induced to fuse in the absence of added
proteins. Second, important characteristics of biomembrane
fusion such as capacitance flickering are also observed in
pure lipid model systems, and the activation energies of the
molecular rearrangements involved appear to be very similar
(25). Finally, low (sublytic) concentrations of LPC not only
inhibit model membrane fusion but also biomembrane fusion
in an efficient and reversible manner, and in biological systems as different as the low pH-induced fusion of virus-infected cells, and exocytosis in patch-clamped mast cells or
isolated sea urchin cortices (23). The latter suggests that
model and biological membrane fusion both involve a lipidic
intermediate with a net negative monolayer curvature [for a
critical discussion on the effects of lysolecithin (LPC) on
membrane fusion [see (23), and Stegmann, this issue of
Traffic ]. Together, these findings support the idea that the
membrane lipid composition is an important determinant of
the willingness of a membrane to fuse, and suggest that
specific lipid species may play an active role in the local
destabilization and fusion of membranes.
Biomembranes typically contain some 100 different lipid
molecules, which, upon isolation, show dramatic differences
in physicochemical properties. Early studies suggested that
LPC, a micelle-forming (‘type-I’) lipid, was involved in
biomembrane fusion (26). However, as outlined above, recent data have shown that lysophospholipids do not stimulate but inhibit model and biological membrane fusion. Acidic
phospholipids, especially phosphatidic acid (PA) and phosphatidylserine (PS), have also been proposed to play a central
role in the induction of biomembrane fusion, in conjunction
with Ca2 + (27). Extensive model membrane studies have
shown that divalent cations can induce membrane fusion of
acidic phospholipid-containing membranes, probably by a
combination of charge neutralization, cross-linking of membranes, local dehydration, and the induction of local defects
in lipid packing. However, in view of the high concentration
of Ca2 + required for direct membrane destabilization, a general role for acidic phospholipids and Ca2 + in biological membrane fusion seems unlikely. Instead, Ca2 + probably exerts
its effect by inducing a conformational change in Ca2 + -binding proteins such as annexins (28). A possible exception is
the fusion of fully primed and docked secretory vesicles with
the plasma membrane during regulated exocytosis where
fusion occurs in the immediate proximity of calcium influx,
Traffic 2000: 1: 605 – 613
Lipid-Involvement in Membrane Fusion and Fission
bilayer structures (35). On the other hand, the concept
should not be oversimplified. In particular, it should be realized that the phase-preference or spontaneous curvature not
only depends on the molecular structure of the lipid in question, but also on its dynamic properties, the hydration of the
lipid headgroup, as well as on intra- and intermolecular interactions. Thus, the curvature preference of lipids is greatly
influenced by factors such as temperature, pH, salt and
cation concentrations. For example, LPC is organized in bilayers at temperatures below the gel-to-liquid crystalline phase
transition temperature.
Figure 2: Shape-structure concept of lipid polymorphism
(33,34). Note that unsaturated fatty acids, ceramide, diacylglycerol, and cholesterol will themselves not form an HII-phase, but
promote HII-phase formation in mixed-lipid systems. For details
see text.
and the fusion machinery is transiently exposed to high local
calcium concentrations (29). Finally, inverted nonbilayer-preferring lipids (‘type-II lipids’) have been proposed to play a
key role in biomembrane fusion (9,30). In contrast to type-I
lipids, which are virtually absent from biomembranes, type-II
lipids are present in significant amounts in almost every
biomembrane (31,32). One of the nonbilayer lipid phases
formed by these lipids is the inverted hexagonal-II (HII) phase.
Almost any biological membrane contains lipids that, upon
isolation, will form an HII-phase under physiological conditions of ionic strength, pH and temperature. Examples are
the unsaturated phosphatidylethanolamines (PEs) present in
all eukaryotic membranes (in the plasma membrane, typically
25 mol% of total phospholipid, of which 80% is present in
the cytosolic leaflet). The lipid molecules in the HII-phase are
organized in hexagonally arranged cylinders, with the polar
headgroups lining a central aqueous channel (Figure 2). The
tendency of type-II lipids to form inverted nonbilayer lipid
structures can be explained by considering their overall dynamic shape (33,34). Type-II lipids are cone-shaped, they
have a small headgroup area as compared with the crosssectional area of the acyl chains. In contrast, type-I lipids,
such as most lysophospholipids, have an inverted coneshape, while bilayer-preferring lipids are best described as
having the shape of a cylinder. It is important to note that the
terms inverted cone- and cone-shaped have not been used
consistently in literature, and that confusion can be avoided
by using the terms type-I and type-II for lipids with a positiveand negative-curvature preference, respectively. Despite its
simplicity, the shape–structure concept of lipid polymorphism has proven to be very powerful in predicting the phase
behavior of many membrane lipids and their mixtures. This is
best illustrated by the fact that an equimolar mixture of a
type-I and a type-II lipid, LPC and unsaturated PE, forms
Traffic 2000: 1: 605– 613
A biomembrane consists of a mixture of bilayer-preferring
and type-II lipids, but despite the presence of type-II lipids, its
basic structure is a lipid bilayer. The type-II lipids are forced
to remain in the lipid bilayer, a situation which is often
referred to as ‘frustration’. In prokaryotes where the lipid
composition of the membrane can be easily manipulated, the
ratio of bilayer to nonbilayer preferring lipids was shown to
be under strict metabolic control, indicating that the propensity of the membrane to form type II structures is of great
importance for cellular functions (36,37). The ratio is an
important determinant of the interfacial properties of the
bilayer as well as membrane stability, and thus influences the
activity of integral and peripheral membrane proteins (38) and
determines the propensity of the membrane to undergo
fusion. Model membrane studies show that physiologically
relevant factors such as diacylglycerol (DG), unsaturated fatty
acids, and hydrophobic proteins can induce a bilayer to HIIphase transition in physiologically relevant mixtures of bilayer-preferring and type-II lipids (31,39). The transition of a
bilayer to an HII-phase occurs through membrane fusion, and
the same nonbilayer semifusion intermediates (Figure 1) are
most likely involved in fusion and HII-phase formation (22).
The total lipid extract of several biological membranes forms
an HII-phase upon hydration, suggesting that, in vivo, these
membranes will be eager to form nonbilayer lipid structures
such as those involved in membrane fusion. Thus, the willingness of a biomembrane to fuse may, at least in part, be
determined by the ratio of bilayer to nonbilayer-preferring
lipids present in the membrane.
Because biomembranes contain large amounts of type-II
lipids, on one hand, and a correlation exists between their
presence and membrane fusion activity in model systems,
on the other hand; type-II lipids are attractive candidates for
a direct involvement in biomembrane fusion. By virtue of
their low headgroup hydration, type-II lipids would be expected to enhance membrane adhesion by reducing the
hydration repulsion, while their negative curvature preference should facilitate the formation of highly curved concave
semifusion intermediates and stimulate membrane fusion.
The idea that biomembrane fusion may take advantage of the
phase-preference of type-II lipids is supported by studies on
the lipid dependency of biomembrane fusion. External addition or local generation of cis -unsaturated fatty acids, oleic
acid and arachidonic acid, stimulates chromaffin granule fusion (40,41), ER-derived microsome fusion (42), and virus-
607
Burger
induced cell–cell fusion (43). This stimulatory effect can be
prevented by co-incubating with LPC, and is fully reversible
by albumin extraction of the fatty acids (23). Second, inhibitors of phospholipase-A2 impede endosome fusion both
in permeabilized and intact cells, while in permeabilized cells,
fusion activity can be partially restored by the addition of
arachidonic acid [(44); also see (45)]. In addition to lipid
breakdown products, fusogenic proteins may promote membrane fusion by enhancing the negative curvature tendency
of the lipid system; synthetic viral fusion peptides promote
the formation of type-II lipid structures and model membrane
fusion (46). The same mechanism may be used in vivo by the
fusion peptide of viral fusion proteins, or a hydrophobic entity
of a fusogenic protein in general. Unfortunately, the best
characterized biomembrane fusion protein, the hemagglutinin glycoprotein (HA) of influenza virus, mediates exoplasmic fusion with the endosomal membrane; the initial
interaction occurs between exoplasmic membrane leaflets
containing only small amounts of type-II lipids. Although
HA-mediated membrane fusion appears to involve a stalk-like
semifusion intermediate, the formation of this intermediate
may depend primarily upon specific lipid–protein interactions
(14,47). Future experiments will have to show whether a
fusion machinery acting on membrane leaflets rich in type-II
lipids, for example during synaptic vesicle exocytosis may be
designed to gain more profit from the presence of type-II
lipids in the membrane.
(51 – 53). Recently, the formation of COPI-coated vesicles
was reconstituted in a liposomal system and, although
coatomer was absolutely required, vesiculation was remarkably insensitive to the lipid composition of the liposomes
suggesting that lipids are not part of the general core mechanism for budding of COPI-coated vesicles (54). These studies show that membrane coats are essential for Golgi
membrane budding in vitro, but do not formally prove that
coat assembly is the direct driving force for bud formation
and membrane fission in vivo (see below). Moreover, local
high concentrations of specific lipids are likely to facilitate
(protein-mediated) membrane constriction and fission in vivo.
The potential role of lipids in membrane bending and the
induction of membrane fission is discussed below.
Membrane Domains and Transbilayer Area
Asymmetry
There are ample indications for the existence of membrane
microdomains with a specific lipid and protein composition in
vivo (55,56), and model membrane studies have shown that
microdomain formation can drive membrane deformation
and fission (57 – 59). In this domain-induced budding concept
(Figure 3A-1), the edge of a membrane domain has an energy
proportional to the length of the edge, and membrane bending is a spontaneous process governed by a competition
between a decrease in edge energy and an increase in
bending energy.
The Role of Lipids in Biomembrane Fission
Although both fission and fusion require close membrane
contact, in the case of fission this requires strong membrane
bending and the formation of a highly constricted neck. An
additional complication in exoplasmic membrane fission concerns cytosolic factors involved in the regulation of fission,
which now act on the membrane leaflets opposite to the
membrane leaflets that initially interact and semifuse. Thus,
a key question concerns the mechanism by which the extreme curvature in the neck region is generated, and the
contacting membrane leaflets are destabilized and induced to
(semi) fuse.
In model membrane systems, budding and fission can occur
in the absence of added proteins, and some intracellular
membrane budding events in vivo might not require a
supramolecular complex of coat proteins. For example, endocytosis occurs via a clathrin- and dynamin-independent pathway involving buds and vesicles that lack a visible coat (48).
However, coat proteins appear to be crucial in most biomembrane budding/fission events, although the precise role of
membrane coats in shaping highly constricted membrane
buds and, even more so, in semifusion and fission, is not
completely understood. Membrane coats are probably multifunctional, playing a role in the formation of the vesicle,
determining its size, selecting and concentrating its contents,
and preventing fusion prior to the completion of budding
(49,50). In a number of elegant in vitro studies, coat proteins
turned out to be sufficient for efficient vesicle formation
608
An alternative driving force for membrane bending originates
from the fact that membranes are very flexible but can hardly
be stretched. Thus, a difference in lateral pressure between
the two membrane leaflets immediately results in membrane
deformation [bilayer couple hypothesis; (60)]. Both in erythrocytes and in pure lipid model systems, the insertion of
additional lipid (equivalent to 0.1 – 1% of total lipid) in one of
the membrane leaflets induces dramatic shape changes
(61,62). Thus, energy-driven lipid translocases may induce
transbilayer area asymmetry and regulate membrane bending in vivo (Figure 3A-2). Indeed, inward translocation of
short-chain PS and PE by the ATP-driven aminophospholipid
translocase in the plasma membrane of human erythroleukemia cells, stimulates endocytosis (63). Moreover,
deletion of the DRS2 gene, which encodes a putative
aminophospholipid translocator present in late Golgi of yeast,
results in a decrease in clathrin-coated vesicles budding from
the Golgi complex (64). Transbilayer area asymmetry can also
be induced by lipid metabolism when this results in a change
in the molecular area or biophysical properties of the membrane lipid. ATP depletion of human erythrocytes is accompanied by breakdown of phosphatidylinositol-4,5-bisphosphate
shrinking the inner membrane leaflet by 0.6%, and probably explaining the spiny appearance of human erythrocytes
after ATP depletion (65). In contrast to phospholipids, the
spontaneous transbilayer movement (flipflop) of several
phospholipid breakdown products, such as ceramide and DG,
is very rapid. If lipid metabolism is confined to one of the
membrane leaflets, equilibration of the breakdown products
Traffic 2000: 1: 605 – 613
Lipid-Involvement in Membrane Fusion and Fission
will induce a relative increase in the lateral pressure in the
opposite membrane leaflet and induce membranes to bend
(Figure 3A-3). Microinjection of sphingomyelinase into sphingomyelin-containing giant liposomes results in the asymmetric production of ceramide and the outward budding of
vesicles (66), while the treatment of macrophages with externally added sphingomyelinase results in ATP-independent
endocytosis through the formation of large endocytic vesi-
Figure 3: Speculative models for lipid involvement in membrane fission. Protein-mediated changes in membrane composition may change membrane curvature and lead to membrane
fission. Membrane microdomains driving membrane deformation and fission (A-1). Membrane bending/fission as a result of
transbilayer area asymmetry induced by an energy-driven lipidtranslocase (A-2), a phospholipase (A-3), or a lipid transfer protein
in combination with a phospholipase (A-4). Membrane bending
induced by lipid exchange and driven by transbilayer curvature
asymmetry (A-5). Models for the role of LPA acyltransferases in
membrane budding/fission (B). LPA acyltransferases may
change the spontaneous monolayer curvature (B-1), induce
transbilayer area asymmetry (B-2), result in a fusogenic lipid in
the lumenal membrane leaflet (B-3), or provide insertion sites for
fissiogenic proteins in the cytoplasmic membrane leaflet (B-4).
For details see text. Note that the curvature of the constricted
neck region is different in the x–y plane (shown) as compared
with the x – z plane.
Traffic 2000: 1: 605– 613
cles lacking discernable coats [(67); note that these treatments also affect the spontaneous monolayer curvature, see
below).
Both yeast and mammalian cells contain cytosolic proteins
capable of exchanging one lipid species for another between
membranes in vitro, and these proteins are likely to affect,
directly or indirectly, the (local) membrane lipid composition
in vivo and play a role in the regulation of membrane bending
and fission (Figure 3A-4). The yeast Sec14 gene product
preferentially binds phosphatidylinositol (PI) (68). It is structurally and functionally related, but not homologous, to mammalian PI-transfer proteins (PITP). Sec14 yeast mutants are
defective in protein transport from the Golgi to the plasma
membrane, a defect which can be partially rescued by expressing the mammalian PITP. In a recent in vitro study, PITP
was found to be essential for the formation of COPI-coated
vesicles from the trans -Golgi network isolated from dog
kidney cells [(69); also see (70)]. Only the PI-loaded form of
Sec14p could substitute for mammalian PITP in the budding
reaction, suggesting that the delivery of PI to the budding
sites is essential for vesicle formation. In the absence of
cytosol, PI-loaded PITP vesiculates uncoated Golgi membranes in an uncontrolled fashion, but in its presence, the
action of PITP is restricted to the sites where COPI-coated
buds are formed. A local change in lipid composition in the
neck region due to the recruitment of PI and its conversion
into DG (or PA, see below) could be directly responsible for
membrane fission in vitro. Translocation of DG to the exoplasmic leaflet of the Golgi membrane could drive the final
constriction of the neck region while the type-II character of
DG would be expected to facilitate semifusion and fission.
The same mechanism of membrane fission may also be
operational in vivo : studies performed in yeast indicate that
Sec14p regulates the Golgi DG pool, although PI binding and
transfer appear to be dispensable for this function (71). Finally, transbilayer area asymmetry and membrane bending
may result from asymmetric insertion of proteins into the
membrane (see below).
Note that local membrane bending is not necessarily due to
a local difference in lateral pressure between the membrane
leaflets; a global difference could drive local membrane bending with membrane microdomains acting as the nucleation
sites. Coat proteins may very well regulate membrane microdomain formation, and could thus indirectly control membrane invagination and budding, and determine the rate of
vesiculation as well as the chemical composition and size of
the vesicles.
Lipid Acyltransferases and Transbilayer
Curvature Asymmetry
A difference in spontaneous monolayer curvature between
the two membrane leaflets, transbilayer curvature asymmetry, can also drive membrane bending. This is best illustrated
by the shape changes that occur in erythrocytes upon exchange of natural, mono-unsaturated, phosphatidylcholine
609
Burger
(PC) present in the exoplasmic leaflet for synthetic di-saturated or di-unsaturated PC using a PC-specific exchange
protein (Figure 3A-5). The PC-exchange reaction hardly affects the surface area of the exoplasmic leaflet, yet results in
dramatic shape changes: the di-saturated PC increases the
spontaneous curvature of the exoplasmic leaflet and results
in outward membrane bending, while the opposite is observed for di-unsaturated PC (72). Lipid transfer proteins, lipid
translocases, and lipid metabolic enzymes, may all influence
the spontaneous curvature of the membrane leaflets by
affecting membrane lipid composition and asymmetry, and
thus regulate membrane bending in vivo.
The possibility that local lipid metabolism and a change in
transbilayer curvature asymmetry is used in vivo to regulate
membrane bending and fission is suggested by the recent
discovery that phospholipid acyltransferases are part of the
fission machineries involved in endocytic vesicle formation
and Golgi membrane fragmentation (73,74). Strikingly, the
two cytosolic proteins involved, endophilin I in endocytic
vesicle formation and CtBP/BARS in Golgi membrane fission,
are functionally related but show no sequence homology.
Both proteins are capable of converting, in vitro, lyso-PA
(LPA) into PA using oleoyl- or arachidonoyl-CoA as the acyl
donor, and in both cases this activity appears to be essential
for membrane fission. Because acylation of LPA selectively
increases the cross-sectional area of the tail region, the role
of the acyltransferases may be to reduce the spontaneous
curvature of the cytosolic membrane leaflet and facilitate the
formation of negative (concave) membrane curvature during
constriction of the neck region (Figure 3B-1). Originally, neck
constriction was proposed to result from the conversion of
inverted-cone-shaped LPA into cone-shaped PA (73). However, the polymorphic phase behavior of LPA has not been
studied so far and, although a conversion of LPA into PA
probably reduces the spontaneous monolayer curvature, unsaturated (di-oleoyl) PA behaves as a cylindrical, bilayer-preferring, lipid at the neutral pH of the cytosol [(75,76); K.N.J.
Burger, V. Chupin, and B. de Kruijff, unpublished results].
The idea that LPA acyltransferases regulate membrane bending and fission by changing the transbilayer curvature asymmetry is appealing, yet there are many uncertainties and
alternative possibilities. First, LPA and PA can only exert their
effect if they localize to specific regions of the developing
bud and do not mix or diffuse out into the surrounding
membrane. One possibility is that local membrane curvature
drives selection of lipid molecules based on their molecular
shape [e.g. see (77)], for example restricting LPA to sites
with positive monolayer curvature. However, direct experimental evidence to support this notion in an experimental
system that allows free diffusion of lipids between areas of
different curvature has so far not been obtained [see (78)].
An alternative possibility is the formation of PA-rich microdomains. PA can act both as hydrogen bond donor and acceptor
and has a high tendency to cluster (79); domains could be
fixed at the neck region by protein–lipid interaction (see
below). Second, the origin of the LPA used in the acylation
610
reaction is unknown. LPA is a highly water-soluble lipid, and
if LPA would be recruited from the cytosol, its acylation
increases the area of the cytosolic membrane leaflet (Figure
3B-2). This would facilitate positive instead of negative membrane curvature, and might explain the fact that disruption of
endophilin function in stimulated synapses blocks invagination of clathrin-coated pits (80). On the other hand, if LPA is
generated in the membrane e.g. by the action of a phospholipase A2 (81), the effect on membrane curvature will depend
on the fate of the fatty acids generated. Third, the extent to
which LPA and PA undergo transbilayer movement, in particular in the Golgi complex, is unknown, but the neutralized
form of PA undergoes rapid transbilayer movement in model
membranes (82). If PA reaches the lumenal leaflet of the
Golgi membrane, the mildly acidic pH and high Ca2 + concentration in the Golgi lumen trigger type-II behavior which may
well lead to membrane destabilization and fission (Figure
3B-3). Fourth, PA may be further metabolized and, for example, be converted into DG, a type-II lipid capable of rapid
transbilayer movement. Finally, PA or one of its metabolites
may activate downstream effectors involved in membrane
bending or fission. PA is unique among anionic lipids because
of its small and highly charged headgroup very close to the
glycerol backbone. We recently observed that dynamin, a
large GTPase regulating membrane constriction and fission
during receptor-mediated endocytosis (83), is a membraneactive molecule capable of penetrating into the acyl chain
region of membrane lipids, and that lipid penetration is
strongly stimulated by PA (K.N.J. Burger, R.A. Demel, S.L.
Schmid, and B. de Kruijff, unpublished observations). Thus,
instead of promoting negative membrane curvature, the PA
generated by dynamin-bound endophilin may induce deep
penetration of dynamin into the membrane of the neck
region, followed by membrane destabilization and fission
(Figure 3B-4). Other membrane fission events may involve
the generation of PA by enzymes such as CtBP/BARS (74) or
phospholipase D (84,85), followed by membrane insertion of
a fissiogenic protein.
It is important to note that the principles discussed above
may operate simultaneously, and together determine the
bending response of a membrane in vivo. For example, lipid
metabolism may result in membrane domain formation, induce transbilayer area asymmetry, as well as influence the
spontaneous monolayer curvature and protein – lipid interactions. Although many factors involved in the regulation of
endocytosis and intracellular membrane budding have been
identified, the regulation of the central step, semifusion and
fission, remains elusive. One might expect the mechanism
to be similar to that of vesicle – membrane fusion, and involve
a local perturbation of the contacting membrane leaflet by a
protein factor. The function of coat proteins would then be to
bring the interacting membrane surfaces close enough together to allow the fusogenic protein to induce membrane
fission. However, this mechanism seems unlikely in view of
the fact that in numerous genetic screens, a candidate fusogenic protein that might fulfill this function has not been
found. Alternatively, the cooperative low-energy interactions
Traffic 2000: 1: 605 – 613
Lipid-Involvement in Membrane Fusion and Fission
of coat proteins or their capacity to induce membrane microdomains may result in the extreme membrane curvature
in the neck region, which in itself could be sufficient to allow
spontaneous semifusion of the contacting membrane leaflet.
In this scheme, membrane deformation and semifusion/
fission are expected to be facilitated by specific lipid species
or, possibly, by deep penetration of a protein in the non-contacting membrane leaflet.
8.
9.
10.
Concluding Remarks
11.
Lipid bilayers can fuse in the complete absence of proteins,
but membrane fusion and fission in vivo are regulated by
specialized proteins. There is strong experimental and theoretical support for the involvement of largely lipidic fusion
intermediates and a semifusion intermediate with a net negative monolayer curvature in both model and biological membrane fusion. The most likely role of specialized proteins in
biomembrane fusion is to reduce the energy barriers to
fusion by facilitating the formation of these highly curved
fusion intermediates. In principle, two mechanisms can be
envisaged: a protein that acts as a fusogen by itself, or a
protein that, after activation, produces a fusogen, for example a lipid breakdown product. Conclusive evidence for the
former mechanism has been obtained: the fusogenic
proteins directly responsible for virus–membrane fusion
have been identified, and fusion is induced by hydrophobic
peptides perturbing the lipid packing of the interacting membranes. This mechanism is likely to apply to other biomembrane fusion events as well, but it may not apply to
membrane fission. Although, lipids may be merely greasing
membrane fusion machineries, the recent discovery that
phospholipid acyltransferases are part of two otherwise unrelated fission machineries suggests that lipids could be the
key players during membrane bending and fission.
12.
Acknowledgments
I am grateful to Thomas Pomorski, Arie Verkleij, Antoinette Killian,
and Alberto Luini for helpful discussions and critical reading of the
manuscript, and to FEI/Philips Electron Optics (Eindhoven, NL) for
financial support. I apologize to those investigators whose contributions were not included in this review due to space limitations.
References
1.
2.
3.
4.
5.
6.
7.
Blumenthal R, Düzgünes N, Hoekstra D, Villar, E (Eds). Membrane
fusion. Mol Membr Biol. 1998;16 (special issue): 1 – 140.
White JM. Membrane fusion. Science 1992;258: 917 – 924.
Jahn R, Südhof TC. Membrane fusion and exocytosis. Annu Rev
Biochem 1999;68: 863 – 911.
Stegmann T, Doms RW, Helenius A. Protein-mediated membrane
fusion. Annu Rev Biophys Chem 1989;18: 187 – 211.
Heuser JE, Reese TS, Dennis MJ, Jan Y, Jan L, Evans L. Synaptic
vesicle exocytosis captured by quick freezing and correlated with
quantal transmitter release. J Cell Biol 1979;81: 275 – 300.
Chandler DE. Exocytosis and endocytosis: membrane fusion events
captured in rapidly frozen cells. Curr Top Membr Transp 1988;32:
169–202.
Burger KNJ. Morphology of membrane fusion. Curr Top Mem
1998;44: 403–445.
Traffic 2000: 1: 605– 613
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
Monck JR, Fernandez JM. The fusion pore and mechanisms of
biological membrane fusion. Curr Opin Cell Biol 1996;8: 524 – 533.
Verkleij AJ, Mombers C, Gerritsen WJ, Leunissen-Bijvelt L, Cullis
PR. Fusion of phospholipid vesicles in association with the appearance of lipidic particles as visualized by freeze fracturing. Biochim
Biophys Acta 1979;555: 358 – 361.
Verkleij AJ. Role of lipids during fusion of model and biological
membranes. In Horrocks LA, Freysz L, Toffano G (Eds.), Phospholipid Research and the Nervous System, Biochemical and Molecular
Pharmacology, Fidia Research Series, vol. 4. Padova: Liviana Press,
1986: 207 – 216.
Helm CA, Israelachvili JN. Forces between phospholipid bilayers
and relationship to membrane fusion. Methods Enzymol 1993;220:
130 – 143.
Bentz J. Minimal aggregate size and minimal fusion unit for the first
fusion pore of influenza hemagglutinin-mediated membrane fusion.
Biophys J 2000;78: 227 – 245.
Shangguan T, Alford D, Bentz J. Influenza-virus-liposome lipid mixing is leaky and largely insensitive to the material properties of the
target membrane. Biochemistry 1996;35: 4956 – 4965.
Bonnafous P, Stegmann T. Membrane perturbation and fusion pore
formation in influenza hemagglutinin-mediated membrane fusion – a
new model for fusion. J Biol Chem 2000;275: 6160 – 6166.
Lentz BR, Lee JK. Poly(ethylene glycol) (PEG)-mediated fusion between pure lipid bilayers: a mechanism in common with viral fusion
and secretory vesicle release? Mol Mem Biol 1999;16: 279 – 296.
Kemble GW, Danieli T, White JM. Lipid-anchored influenza hemagglutinin promotes hemifusion, not complete fusion. Cell 1994;76:
383 – 391.
Razinkov VI, Melikyan GB, Cohen FS. Hemifusion between cells
expressing hemagglutinin of influenza virus and planar membranes
can precede the formation of fusion pores that subsequently fully
enlarge. Biophys J 1999;77: 3144 – 3151.
Chanturiya A, Chernomordik LV, Zimmerberg J. Flickering fusion
pores comparable with initial exocytotic pores occur in protein-free
phospholipid bilayers. Proc Natl Acad Sci USA 1997;94: 14423–
14428.
Verkleij AJ. Lipidic intramembranous particles. Biochim Biophys
Acta 1984;779: 43 – 63.
Markin VS, Kozlov JV, Borovyagin VL. On the theory of membrane
fusion. The stalk mechanism. Gen Physio Biophys 1984;5: 361 – 377.
Siegel DP. Energetics of intermediates in membrane fusion – comparison of stalk and inverted micellar intermediate mechanisms.
Biophys J 1993;65: 2124 – 2140.
Siegel DP. The modified stalk mechanism of lamellar/inverted phase
transitions and its implications for membrane fusion. Biophys J
1999;76: 291 – 313.
Chernomordik L. Non-bilayer lipids and biological fusion intermediates. Chem Phys Lipids 1996;81: 203 – 213.
Lindau M, Almers W. Structure and function of fusion pores in
exocytosis and ectoplasmic membrane fusion. Curr Opin Cell Biol
1995;7: 509 – 517.
Lee JK, Lentz BR. Secretory and viral fusion may share mechanistic
events with fusion between curved lipid bilayers. Proc Natl Acad Sci
USA 1998;95: 9274 – 9279.
Lucy JA. Mechanisms of chemically induced cell fusion. Cell Surf
Rev 1978;5: 267 – 304.
Papahadjopoulos D, Nir S, Düzgünes N. Molecular mechanisms of
calcium-induced membrane fusion. J Bioenerg Biomembr 1990;22:
157 – 179.
Creutz CE. The annexins and exocytosis. Science 1992;258: 924–
931.
Coorssen JR, Blank PS, Tahara M, Zimmerberg J. Biochemical and
functional studies of cortical vesicle fusion: the SNARE complex and
Ca2 + sensitivity. J Cell Biol 1998;143: 1845 – 1857.
Cullis PR, Hope MJ. Effects of fusogenic agent on membrane
structure of erythrocyte ghosts and the mechanism of membrane
fusion. Nature 1978;271: 672 – 674.
de Kruijff B, Cullis PR, Verkleij AJ, Hope MJ, Van Echteld CJA,
Taraschi TF. Lipid polymorphism and membrane function. In Martonosi AN (Ed.), The Enzymes of Biological Membranes. New York:
Plenum Press, 1985: 131 – 204.
611
Burger
32. Epand RM (Ed). Lipid polymorphism and membrane properties. Curr
Top Mem 1998; 44: 1 – 568.
33. Israelachvili JN, Marcelja S, Horn RG. Physical principles of membrane organization. Q Rev Biophys 1980;13: 121 – 200.
34. Cullis PR, Hope MJ, Tilcock CP. Lipid polymorphism and the roles
of lipids in membranes. Chem Phys Lipids 1986;40: 127 – 144.
35. Madden TD, Cullis PR. Stabilization of bilayer structure for unsaturated phosphatidylethanolamines by detergens. Biochim Biophys
Acta 1982;684: 149–153.
36. Osterberg F, Rilfors L, Wieslander A, Lindblom G, Gruner SM. Lipid
extracts from membranes of Acholeplasma laidlawii A grown with
different fatty acids have a nearly constant spontaneous curvature.
Biochim Biophys Acta 1995;1257: 18 – 24.
37. Rietveld AG, Killian JA, Dowhan W, de Kruijff B. Polymorphic regulation of membrane phospholipid composition in Escherichia coli. J
Biol Chem 1993;268: 12427 – 12433.
38. Epand RM. Lipid polymorphism and protein – lipid interactions.
Biochim Biophys Acta 1998;1376: 353 – 368.
39. Bentz J, Ellens H. Membrane fusion: kinetics and mechanisms. Coll
Surf 1988;30: 65–112.
40. Creutz CE. Cis -unsaturated fatty acids induce the fusion of chromaffin granules aggregated by synexin. J Cell Biol 1981;91: 247 –
256.
41. Nagao T, Kubo T, Fujimoto R, Nishio H, Takeuchi T, Hata F.
Ca2 + -independent fusion of secretory granules with phospholipase
A(2)-treated plasma membranes in vitro. Biochem J 1995;307: 563 –
569.
42. Paiement J, Lavoie C, Gavino GR, Gavino VC. Modulation of GTPdependent fusion by linoleic and arachidonic acid in derivatives of
rough endoplasmic reticulum from rat liver. Biochim Biophys Acta
1994;1190: 199–212.
43. Chernomordik L, Leikina E, Cho MS, Zimmerberg J. Control of
baculovirus gp64-induced syncytium formation by membrane lipid
composition. J Virol 1995;69: 3049 – 3058.
44. Mayorga LS, Colombo MI, Lennartz M, Brown EJ, Rahman KH,
Weiss R, Lennon PJ, Stahl PD. Inhibition of endosome fusion by
phospholipase-A(2) (PLA2) inhibitors points to a role for PLA2 in
endocytosis. Proc Natl Acad Sci USA 1993;90: 10255 – 10259.
45. Vidal M, Hoekstra D. In vitro fusion of reticulocyte endocytic vesicles
with liposomes. J Biol Chem 1995;270: 17823 – 17829.
46. Colotto A, Epand RM. Structural study of the relationship between
the rate of membrane fusion and the ability of the fusion peptide of
influenza virus to perturb bilayers. Biochemistry 1997;36: 7644 –
7651.
47. Alford D, Ellens H, Bentz J. Fusion of influenza virus with sialic
acid-bearing target membranes. Biochemistry 1994;33: 1977 – 1987.
48. Lamaze C, Schmid SL. The emergence of clathrin-independent
pinocytic pathways. Curr Opin Cell Biol 1995;7: 573 – 580.
49. Schekman R, Orci L. Coat proteins and vesicle budding. Science
1996;271: 1526–1533.
50. Rothman JE, Wieland FT. Protein sorting by transport vesicles.
Science 1996;272: 227 – 234.
51. Wieland F, Harter C. Mechanisms of vesicle formation: insights from
the COP system. Curr Opin Cell Biol 1999;11: 440 – 446.
52. Takei K, Slepnev VI, Haucke V, De Camilli P. Functional partnership
between amphiphysin and dynamin in clathrin-mediated endocytosis. Nature Cell Biol 1999;1: 33 – 39.
53. Sweitzer SM, Hinshaw JE. Dynamin undergoes a GTP-dependent
conformational change causing vesiculation. Cell 1998;93: 1021 –
1029.
54. Bremser M, Nickel W, Schweikert M, Ravazzola M, Amherdt M,
Hughes CA, Sollner TH, Rothman JE, Wieland FT. Coupling of coat
assembly and vesicle budding to packaging of putative cargo receptors. Cell 1999;96: 495 – 506.
55. Rietveld A, Simons K. The differential miscibility of lipids as the
basis for the formation of functional membrane rafts. Biochim Biophys Acta 1998;1376: 467 – 479.
56. Pralle A, Keller P, Florin EL, Simons K, Horber JKH. Sphingolipidcholesterol rafts diffuse as small entities in the plasma membrane of
mammalian cells. J Cell Biol 2000;148: 997 – 1007.
57. Lipowsky R. Domain-induced budding of fluid membranes. Biophys
J 1993;64: 1133–1138.
612
58. Sackmann E, Feder T. Budding, fission and domain formation in
mixed lipid vesicles induced by lateral phase separation and macromolecular condensation. Mol Membr Biol 1995;12: 21 – 28.
59. Döbereiner HG, Käs J, Noppl D, Sprenger I, Sackmann E. Budding
and fission of vesicles. Biophys J 1993;65: 1396 – 1403.
60. Sheetz MP, Singer SJ. Biological membranes as bilayer couples. A
molecular mechanism of drug – erythrocyte interactions. Proc Natl
Acad Sci USA 1974;71: 4457 – 4461.
61. Farge E, Devaux PF. Shape changes of giant liposomes induced by
an asymmetric transmembrane distribution of phospholipids. Biophys J 1992;61: 347 – 357.
62. Mui BLS, Döbereiner HG, Madden TD, Cullis PR. Influence of
transbilayer area asymmetry on the morphology of large unilamellar
vesicles. Biophys J 1995;69: 930 – 941.
63. Farge E, Ojcius DM, Subtil A, Dautry-Varsat A. Enhancement of
endocytosis due to aminophospholipid transport across the plasma
membrane of living cells. Am J Physiol 1999;276: C725 – 733.
64. Chen CY, Ingram MF, Rosal PH, Graham TR. Role for Drs2p, a
P-type ATPase and potential aminophospholipid translocase, in
yeast late Golgi function. J Cell Biol 1999;147: 1223 – 1236.
65. Ferrell JE Jr, Huestis WH. Phosphoinositide metabolism and the
morphology of human erythrocytes. J Cell Biol 1984;98: 1992 – 1998.
66. Holopainen JM, Angelova MI, Kinnunen PKJ. Vectorial budding of
vesicles by asymmetrical enzymatic formation of ceramide in giant
liposomes. Biophys J 2000;78: 830 – 838.
67. Zha X, Pierini LM, Leopold PL, Skiba PJ, Tabas I, Maxfield FR.
Sphingomyelinase treatment induces ATP-independent endocytosis. J Cell Biol 1998;140: 39 – 47.
68. Kearns BG, Alb JG Jr, Bankaitis V. Phosphatidylinositol transfer
proteins: the long and winding road to physiological function.
Trends Cell Biol 1998;8: 276 – 282.
69. Simon JP, Morimoto T, Bankaitis VA, Gottlieb TA, Ivanov IE, Adesnik
M, Sabatini DD. An essential role for the phosphatidylinositol transfer protein in the scission of coatomer-coated vesicles from the
trans -Golgi network. Proc Natl Acad Sci USA 1998;95: 11181–
11186.
70. Ohashi M, Devries KJ, Frank R, Snoek G, Bankaitis V, Wirtz K,
Huttner WB. A role for phosphatidylinositol transfer protein in secretory vesicle formation. Nature 1995;377: 544 – 547.
71. Phillips SE, Sha B, Topalof L, Xie Z, Alb JG, Klenchin VA, Swigart P,
Cockcroft S, Martin TFJ, Luo M, Bankaitis VA. Yeast Sec14p deficient in phosphatidylinositol transfer activity is functional in vivo. Mol
Cell 1999;4: 187 – 197.
72. Christiansson A, Kuypers FA, Roelofsen B, Op den Kamp JAF, van
Deenen LLM. Lipid molecular shape affects erythrocyte morphology:
a study involving replacement of native phosphatidylcholine with
different species followed by treatment of cells with sphingomyelinase C or phospholipase A2. J Cell Biol 1985;101: 1455 – 1462.
73. Schmidt A, Wolde M, Thiele C, Fest W, Kratzin H, Podtelejnikov AV,
Witke W, Huttner WB, Söling HD. Endophilin I mediates synaptic
vesicle formation by transfer of arachidonate to lysophosphatidic
acid. Nature 1999;401: 133 – 141.
74. Weigert R, Silletta MG, Spano S, Turacchio G, Cericola C, Colanzi
A, Senatore S, Mancini R, Polishchuk EV, Salmona M, Facchiano F,
Burger KNJ, Mironov A, Luini A, Corda D. CtBP/BARS induces
fission of Golgi membranes by acylating lysophosphatidic acid.
Nature 1999;402: 429 – 433.
75. Farren SB, Hope MJ, Cullis PR. Polymorphic phase preferences of
phosphatidic acid: a 31P and 2H NMR study. Biochem Biophys Res
Commun 1983;111: 675 – 682.
76. Verkleij AJ, De Maagd R, Leunissen-Bijvelt J, de Kruijff B. Divalent
cations and chlorpromazine can induce non-bilayer structures in
phosphatidic acid-containing model membranes. Biochim Biophys
Acta 1982;684: 255 – 262.
77. Mukherjee S, Maxfield FR. Role of membrane organization and
membrane domains in endocytic lipid trafficking. Traffic 2000;1:
203 – 211.
78. Moran L, Janes N. Tracking phospholipid populations in polymorphism by sideband analyses of 31P magic angle spinning NMR.
Biophys J 1998;75: 867 – 879.
79. Boggs JM. Lipid intermolecular hydrogen bonding: influence on
structural organization and membrane function. Biochim Biophys
Acta 1987;906: 353 – 404.
Traffic 2000: 1: 605 – 613
Lipid-Involvement in Membrane Fusion and Fission
80. Ringstad N, Gad H, Löw P, Di Paolo G, Brodin L, Shupliakov O, De
Camilli P. Endophilin/SH3p4 is required for the transition from early
to late stages in clathrin-mediated synaptic vesicle endocytosis.
Neuron 1999;24: 143– 154.
81. De Figueiredo P, Drecktrah D, Katzenellenbogen JA, Strang M,
Brown WJ. Evidence that phospholipase A2 activity is required for
Golgi complex and trans Golgi network membrane tubulation. Proc
Natl Acad Sci USA 1998;95: 8642 – 8647.
82. Eastman SJ, Hope MJ, Cullis PR. Transbilayer transport of phosphatidic acid in response to transmembrane pH gradients. Biochemistry 1991;30: 1740 – 1745.
83. Sever S, Damke H, Schmid SL. Garrotes, springs, ratchets, and
whips: putting dynamin models to the test. Traffic 2000;1: 385 – 392.
84. Cockcroft S. ARF-regulated phospholipase D: a potential role in
membrane traffic. Chem Phys Lipids 1996;80: 59 – 80.
85. Exton JH. Regulation of phospholipase D. Biochim Biophys Acta
1999;1439: 121 – 133.
.
Traffic 2000: 1: 605– 613
613