LAdvanced Drug Delivery Reviews 38 (1999) 207â232 Lipid vesicles and membrane fusion *Gregor Cevc , Holger Richardsen ¨ ¨Medizinische Biophysik, Technische Universitat Munchen, Klinikum r.d.I., Ismaningerstrasse 22, D-81675 Munich, Germany Abstract Membrane fusion is essential for cell survival and has attracted a great deal of both theoretical and experimental interest. Fluorescence (de)quenching measurements were designed to distinguish between bilayermerging and vesicle-mixing. Theoretical studies and various microscopic and diffraction methods have elucidated the mechanism of membrane fusion. These have revealed that membrane proximity and high defect density in the adjacent bilayers are the only prerequisites for fusion. Intermediates, such as stalk or inverse micellar structures can, but need not, be involved in vesicle fusion. Nonlamellar phase creation is accompanied by massive membrane fusion although it is not a requirement for bilayer merging. Propensity for membrane fusion is increased by increasing the local membrane disorder as well by performing manipulations that bring bilayers closer together. Membrane rigidification and enlarged bilayer separation opposes this trend. Membrane fusion is promoted by defects created in the bilayer due to the vicinity of lipid phase transition, lateral phase separation or domain generation, high local membrane curvature, osmotic or electric stress in or on the membrane; the addition of amphiphats or macromolecules which insert themselves into the membrane, freezing or other mechanical membrane perturbation have similar effects. Lowering the water activity by the addition of water soluble polymers or by partial system dehydration invokes membrane aggregation and hence facilitates fusion; as does the membrane charge neutralization after proton or other ion binding to the lipids and intermembrane scaffolding by proteins or other macromolecules. The alignment of defect rich domains and polypeptides or protein binding is pluripotent: not only does it increase the number of proximal defects in the bilayers, it triggers the vesicle aggregation and is fusogenic. Exceptions are the bound molecules that create steric or electrical barriers between the membranes which prevent fusion. Membrane fusion can be non-leaky but it is very common to lose material from the vesicle interior during the later stages of membrane unification, that is, after a few hundred microseconds following the induction of fusion. à 1999 Published by Elsevier Science B.V. All rights reserved. Keywords: Liposomes; Lipid vesicles; Membrane fusion; Drug delivery Contents 1. Introduction ............................................................................................................................................................................ 208 2. Mechanism of fusion ............................................................................................................................................................... 211 2.1. Vesicle aggregation .......................................................................................................................................................... 211 2.2. Membrane merging .......................................................................................................................................................... 211 2.2.1. Temperature controlled fusion of liposomes with the cells in vitro ............................................................................. 213 2.2.2. Temperature controlled vesicle fusion in vivo .......................................................................................................... 213 3. Methods of detection ............................................................................................................................................................... 215 4. Physical induction ................................................................................................................................................................... 216 *Corresponding author. 0169-409X/99/$ â see front matter à 1999 Published by Elsevier Science B.V. All rights reserved. PI I : S0169-409X( 99 )00030-7 208 G. Cevc, H. Richardsen / Advanced Drug Delivery Reviews 38 (1999) 207 â232 4.1. Catalysis via defects ......................................................................................................................................................... 216 4.1.1. Phase separation..................................................................................................................................................... 217 4.1.2. Thermal induction .................................................................................................................................................. 217 4.2. Electrical induction .......................................................................................................................................................... 218 4.3. Curvature effects.............................................................................................................................................................. 218 4.4. Surface tension ................................................................................................................................................................ 218 5. Chemical induction ................................................................................................................................................................. 219 5.1. Charge neutralization ....................................................................................................................................................... 219 5.2. (De)Hydration ................................................................................................................................................................. 219 5.2.1. Polyethyleneglycol ................................................................................................................................................. 219 5.3. (De)Protonation ............................................................................................................................................................... 220 5.4. Ion binding ...................................................................................................................................................................... 221 5.5. Induction by fusogenic lipids ............................................................................................................................................ 221 5.5.1. Fatty acids ............................................................................................................................................................. 221 5.5.1.1. Chainlength dependence ............................................................................................................................ 222 5.5.1.2. Chain unsaturation..................................................................................................................................... 223 5.5.2. Fatty alcohols ........................................................................................................................................................ 223 5.5.3. Other amphiphiles .................................................................................................................................................. 224 6. Biological fusion ..................................................................................................................................................................... 224 6.1. Protein effect ................................................................................................................................................................... 225 7. Pore hypothesis / stalk intermediate ........................................................................................................................................... 226 8. Fusion kinetics ........................................................................................................................................................................ 227 9. Final remarks .......................................................................................................................................................................... 227 References .................................................................................................................................................................................. 228 1. Introduction Molecular factors that control the aggregation and fusion of phospholipid vesicles are surveyed in Ref. Membrane fusion is of paramount importance for [3]. The effects of membrane composition, vesicle the functioning and, indeed, the existence of all size, cations, membrane phase state, dehydrating living systems. The multiplication of even the sim- agents and proteins, especially in the bile vesicle plest, primordial cells relies on the fusion of cellular processing are covered. Review [4] puts the em- envelopes; the intra-cellular trafficking of more phasis on the surface chemistry of phospholipid sophisticated organisms involves continuous merging vesicles and their relevance for membrane aggrega- of the intra-cellular organelle membranes as well. tion and fusion [4]. Simultaneously, extensive exo- and endocytosis In this article we attempt to provide an up-to-date takes place at the level of plasma membrane. A picture of the lipid vesicle fusion, with emphasis on macrophage, for example, renews its cell surface the properties of the participating lipid bilayers. The every 45 min through the processes of small scale aspect of protein-induced fusion is only covered to pinocytosis and the larger scale phagocytosis, which the extent that is important to understand the restruc- together comprise endocytosis. turing of bilayers during the merger. At first, the Membrane fusion has therefore attracted scientists question of non-lamellar phase participation is ad- for a long time. A large number of review articles dressed, since this has occupied the minds of re- has dealt with the topic, of which only some are searchers in the field for at least the past 10 years. mentioned here. The calcium-induced fusion of lipid An appreciable proportion of the intra-cellular bilayer membranes and the role of synexin and other material transport involves lipids in the non-lamellar calcium-binding proteins (annexins) in membrane state. Such transport may include different membra- fusion are reviewed in Ref. [1]. The roles of cations, neous intra-cellular compartments. In the past, it was lipid phase transitions, long chain fatty acids and thought [5] that transportation relies on the domains other fusogenic molecules are also discussed. in the inverted hexagonal (H , H ) and, occasion-II a The mechanisms of biological membrane fusion ally, in the inverted cubic phases (Q ). However,a and the role of fusion pores is reviewed in Ref. [2]. now it seems that the latter phases prevail and that G. Cevc, H. Richardsen / Advanced Drug Delivery Reviews 38 (1999) 207 â232 209 identifications of the hexagonal phase were largely branes in the form of small inverted micellar bodies based on the insufficient experimental resolution or (Q , [7]) or elongated non-lamellar contact struc-a on the non-deliberate errors in data analysis [6]. The tures (H , [8]. The upper half of Fig. 2 gives aII inverted hexagonal and the most widely distributed schematic view of the corresponding sequence of types of cubic phases are schematically presented in steps which can lead to the final membrane unifica- Fig. 1. tion. Inverted phases were postulated to be an integral We are convinced that the requirement for the part of membrane fusion in general. It was suggested formation of non-lamellar phases during membrane that they appear at the contact sites of two mem- fusion is too strong to be met in general. While the Fig. 1. Schematic representation of various non-lamellar phases thought to be involved in at least some of the membrane fusion processes: upper left: inverted hexagonal phase: H ; upper right: the simplest inverted cubic (micellar) phase; lower left: bicontinuous cubic phase ofII type In3m (from Ref. [130]); lower right: bicontinuous cubic phase of type Pn3m (from Ref. [130]). 210 G. Cevc, H. Richardsen / Advanced Drug Delivery Reviews 38 (1999) 207 â232 Fig. 2. Basic sequences of steps leading to membrane unification, which all involve initial membrane approach. Inverted phases may appear at the sites of membranes contact, in the form of small inverted micellar bodies or elongated non-lamellar contact structures (upper part), both result in subsequent membrane merging. Membrane restructuring and merging can also rely on local membrane disordering and defect formation, due to the hydrophobic interaction (lower part). G. Cevc, H. Richardsen / Advanced Drug Delivery Reviews 38 (1999) 207 â232 211 formation of inverted non-bilayer (non-lamellar) opposite effect. The former changes therefore pro- phases always involves fusion, the restructuring, and mote the aggregation of vesicles. ultimately the merger, of membranes is also feasible on a less ordered basis. We therefore advocate an 2.2. Membrane merging additional, simpler scenario that is energetically as favourable as the formation of non-lamellar fusion In our opinion, two adjacent lipid bilayers will intermediates, at least for small scale fusion. Further- fuse when at least two regions of transient and of more, we suggest that such simple fusion should be sufficient bilayer disorder temporarily come into considered as an occasional alternative to the more close contact. The mutual membrane proximity and conventional fusion process involving the non-lamel- the exposure of the hydrophobic regions in the lar membrane phase. The lower half of Fig. 2 bilayers then will permit the membranes to reorgan- presents the sequence of membrane fusion steps ize themselves. This should happen to the extent which does not rely on participation of such non- necessary for the completion of fusion: relocation of lamellar phases. individual lipid molecules or of lipid clusters be- tween the adjacent lipid layers. We are convinced that the merging of lipid membranes is normally a 2. Mechanism of fusion âmessyâ and very site-restricted process which does not require new phase generation but always in- The steps believed to be involved in membrane volves defects. Defect accumulation in the contact fusion are schematically presented in Fig. 2. In order regions therefore promotes membrane fusion. This to unite, two membranes must first approach and may explain part of the action of fusogenic proteins. then combine into a single bilayer, which is referred A less biological example is the extensive fusion of to as âhemifusionâ. The starting phase is the close surface confined, deformed vesicles at the airâwater apposition of two lipid bilayers, which typically interface, which is due to localized defects in the leads to vesicle aggregation. vesicles [10]. Our minimalistic picture of membrane fusion 2.1. Vesicle aggregation differs only quantitatively from a localized bilayer restructuring in one membrane. Membrane fusion In general, the origin of vesicle aggregation and therefore resembles, in certain aspects, the lipid membrane approach is Van der Waals attraction. This translocation process or the insertion of proteins into force always tends to bring two membranes together. a lipid membrane. In a somewhat artificial picture, Any vesicle suspension would collapse into a dry the fusion of a lipid vesicle with another lipid (cell-) lipid crystal if such a process was not prevented by membrane may be considered as an insertion of the the repulsion that acts in the opposite direction. vesicle membrane into the âacceptor membraneâ. Normally, the electrostatic, steric and hydration This is not to say that the non-lamellar fusion forces create a sufficiently strong repulsive barrier to intermediates do not exist. Doubtlessly, the inverted prevent membranes coming into a close contact. micellar bodies and other non-lamellar structures These forces, therefore, are believed to maintain the abound in the biological world. But equally surely, colloidal stability of the vesicles suspension under two small lipid vesicles, e.g. with a diameter of 45 4physiologically relevant conditions [9]. nm, which comprise , 10 molecules, do not have Consequently, to initiate vesicle aggregation one sufficient material to form a non-lamellar phase can diminish the inter-membrane repulsion. Lower- without completely disintegrating; and yet, such ing the number of charges or decreasing the number vesicles are known to fuse rapidly and extensively of water binding sites on membranes, or else con- [11]. densing their interfaces, lowers the lipid bilayer In the following, some experimental support for repulsion and facilitates vesicle aggregation. Increas- the above mentioned picture is given, as collected in ing the bilayer surface charge density or polarity and our laboratory. The more general principles of making the polarâapolar interface thicker has the membrane fusion are then discussed in greater detail. 212 G. Cevc, H. Richardsen / Advanced Drug Delivery Reviews 38 (1999) 207 â232 To test the sensitivity of membrane fusion to the fects is seen from the time dependence of the rate of presence of defects in membranes, we have com- fusion between DPPCâELA-COOH (1:2) vesicles. pared the interactions between the freshly prepared We have measured this rate after the lipid vesicles (defect-rich) and aged (and thus defect-poor) were prepared by the means of ultrasound (see Fig. fusogenic DPPCâPAc (1:2) mixtures, as a function 4). The average fusion efficacy over a period of 100 of temperature (Fig. 3). Near the chain melting phase h was found to decrease to 62% of the initial value transition temperature, T , of phospholipid com- (see Fig. 4) whilst the rate of vesicle fusion simul-m ponent, at 41.758C, the extent of membrane fusion in taneously declined 2.4-fold, also in a mono-exponen- the freshly made suspension was found to be much tial manner. The change in characteristic decay time greater than for the aged formulation. The inverted for both formulations was approximately 7 h. This is hexagonal phase creation (L fi H ) followed the quite similar to the half-time of defects annealing ina II opposite trend. This suggests that in the former, phosphatidylcholine membranes [12]. Data shown in defect-rich suspension, the bilayer density fluctua- Figs. 3 and 4 document the importance, or even tions near the T of the DPPC rich domains, increase occasional dominance, of membrane defects in them the propensity for membrane fusion. On the other fusion process. hand, under comparable experimental conditions, the To demonstrate the relevance of bilayer defects in formation of the H -phase is partly suppressed the fusion with biological membranes we haveII because of the same reasons. induced imperfections (fluctuations) in the bilayer by More directly, the importance of membrane de- raising the temperature. The temperature-controlled Fig. 3. Significance of membrane defects for the efficacy of temperature-dependent fusion between the mixed lipid membranes consisting of DPPCâPAc (1:2). For data discussion see the text. G. Cevc, H. Richardsen / Advanced Drug Delivery Reviews 38 (1999) 207 â232 213 21Fig. 4. Fusion probability (i.e. inverse half-time of fusion: t and relative fusion efficacy (as determined fluorimetrically) for the1 / 2 DPPCâELA-COOH (1:2) mixed vesicles as a function of time after the liposome preparation by means of ultrasound. vesicle fusion with the cells in vitro and in vivo was DPPCâELA-COOH liposomes (see Fig. 6) and the then followed. change of liposome fluorescence in the contact with cells (see Fig. 5) also supports the concept of the 2.2.1. Temperature controlled fusion of liposomes temperature-induced liposomeâcell fusion (see fur- with the cells in vitro ther discussion). We studied first the temperature-induced fusion between the thermolabile DPPCâELA-COOH (1:2) 2.2.2. Temperature controlled vesicle fusion in vesicles and the murine bladder cancer cells in vitro. vivo This was done with the fluorescence-based fusion Experiments similar to those described in Section assay developed by Hoekstra et al. (see Section 3). 2.2.1 were reproduced in vivo. To this end, the Pure DPPC liposomes served as controls. targeting efficacy of several types of liposome to the The results pertaining to the thermolabile tumour tissue with partially leaky capillaries was fusogenic vesicles, described in greater detail later in studied at 37 and 438C in mice (see Fig. 7 and Ref. the text, are shown in Fig. 5. The characteristic [13]). After 30 min of tumour hyperthermia (T5 change in membrane fluorescence suggests that the 438C) the intra-tumour accumulation of the label ambient heat induces liposomeâcell fusion. No such derived from DPPCâELA-COOH (1:2) liposomes fluorescence change, and thence no fusion, is ob- was approximately 4-fold higher than the label served for the simple DPPC liposomes (not shown). concentration in the normothermic tumour. The use The constancy of the results measured with DPPC of slightly different, less pH-sensitive vesicles suggest that the results given in Fig. 5 are not due to (DPPCâELAâELA-COOH, 1:1:1) resulted in 3.7- the adsorption of lipid vesicles on the cell surface but fold relative vesicle accumulation in the tumour. rather are diagnostic of the vesicleâcell fusion. The Prolonging the tumour hyperthermia to 60 min good correlation between the phase behaviour of improved the result to the factor of 7.8. The accumu- 214 G. Cevc, H. Richardsen / Advanced Drug Delivery Reviews 38 (1999) 207 â232 Fig. 5. Effect of temperature on the fusion between bladder tumour cells (MBT) and thermolabile fusogenic liposomes consisting of a DPPCâELA-COOH (1:2) mixture. (At t560 min, 1 million cells were added to the 1 mM suspension of fluorescently labelled lipids. The change in signal from the membrane-associated labels is roughly proportional to the extent of vesicleâcell fusion.) Fig. 6. Temperature dependence of aggregation (upper panel) and fusion (lower panel) between the DPPCâELA-COOH (1:2) mixed lipid membranes. In contrast to the situations illustrated in Fig. 3, the T and T values for this lipid mixture nearly coincide. This results in ratherm h efficient membrane aggregation as well as fusion between 41 and 458C. G. Cevc, H. Richardsen / Advanced Drug Delivery Reviews 38 (1999) 207 â232 215 Fig. 7. Effect of vesicle composition and tissue type on the efficacy of hyperthermia-dependent accumulation of the tritium labeled DPPC in the heated tissue when delivered by means of composite lipid vesicles with the titratable (DPPCâELA; DPPCâELAâEL-OH) or untitratable (DPPC; DPPCâDPPGâEL-OH) surfaces in the physiologic pH region. For data discussion see the text. lation of DPPC-derived radioactivity was much (NBD) and rhodamine (Rh) labelled phospholipids lower and fully accountable to the increase in the the are most commonly used for such a purpose. Alter- blood vessel volume and flow in the region of natively, only one kind of label can be used. For hyperthermia. example, the diminution of self-quenching of NBD- Taken together the data described in previous 6-phosphatidylethanolamine (NBD-PE) fluorescence, paragraphs document the controllability of vesicleâ which occurs when the intramembrane concentration cell fusion under the biologically relevant conditions of fluorescent lipids falls below approximately 5â and confirm the role of membrane defects in the 10%, is indicative of membrane fusion. The de- process. quenching of the fluorescent 1-oleoyl-2-[12-[(7- nitro-2, 1, 3-benzoxadiazol-4-yl)amino]dodecanoyl]- phosphatidylcholine provides similar circumstantial 3. Methods of detection evidence for fusion [16]. To monitor the mixing of vesicle contents, lipo- To detect vesicle fusion, the merging of lipid somes that contain the coencapsulated N,N9- bilayers or the mixing of vesicle contents are as- p-xylylene-bis(-pyridinium bromide) (DPX) sessed, typically with some suitable flourescence and disodium-8-aminonaphthalene-1,3,6-trisulfonate technique (see e.g. Ref. [14] for a review). (ANTS) are mixed with the empty vesicles. Due to The fusion of lipid bilayers is most frequently the dilution of vesicle contents after fusion with the assayed by resonance energy transfer [15]: unifica- unloaded vesicles, or after fluorophore leakage, the tion of two membranes containing the fluorescent fluorescence intensity is enhanced, as fluorescence is lipophilic photon donors and acceptors, originally dequenched [17]. present in separate bilayers, increases the propensity In a different fluorescence-based fusion assay, a for the resonance energy transfer between the fluoro- calcium-sensitive metallochromic dye, arsenazo III phores. This enhances the intensity of emitted (AIII), is partially saturated with calcium (AIII-Ca) fluorescent light. 7-Nitro-2,1,3-benzoxadiazol-4-yl and trapped into one population of liposomes; ethyl- 216 G. Cevc, H. Richardsen / Advanced Drug Delivery Reviews 38 (1999) 207 â232 ene glycol-bis(b-aminoethyl ether)-N,N9-tetraacetate samples, care must be taken not to misinterpret the (EGTA) is incorporated in another vesicle popula- data. Conventional EM images also do not give tion. When the two are mixed and fuse the inter- always representative impression, owing to the bias action between EGTA and AIII-Ca gives rise to a and numerous potential artifacts. The more modern large, and easily detectable, color shift from blue to cryo-EM is better and more reliable, but also more red; this is often measured by monitoring the de- demanding on the experimentalist and not as readily creased absorbance at 660 nm. available. 31 31The luminescent lanthanides, Tb , Ce and The release of secretory products, which often 31Eu , are also potentially useful for the detection of accompanies transient fusion events, can be moni- cation-induced events involving phospholipid mem- tored by micro-voltammetry [21]. brane fusion. For example, terbium can be encapsulated as the 62Tb(citrate) chelation complex in one population of 4. Physical induction3 vesicles (donors). Dipicolinic acid (DPA) is enclosed in the acceptor vesicles. After the fusion between the Vesicle fusion can be induced by changing the donor and acceptor vesicles the fluorescent physical system parameters, such as temperature, 32Tb(DPA) chelation complexes are formed [18]. curvature, lateral tension, etc.. Temperature serves as3 21The presence of EDTA (0.1 mM) and Ca ($1 an indirect fusion promotor, first, by increasing the mM) prevents the formation of TbâDPA complexes defect density in the membranes (e.g. in the phase in the external medium [19]. transition region) and, second, under some condi- 31 31The spectroscopic properties of Tb , Ce and tions, by decreasing the inter-membrane separation 31Eu are complementary. They were used, for exam- (e.g. in the low-temperature, lamellar gel phases). ple, to define the three different forms of dye Normally, however, the temperature-induced defect complex with the phosphatidic acid vesicles [20]. generation only promotes membrane fusion at rela- 31Ce is particularly useful for studying the dilute tively low water contents, or when bilayers are cationâlipid complexes, owing to its strong excita- pushed together by an external osmotic pressure. The tion bands in the near ultraviolet. In addition to reason for this is that temperature driven membrane providing means for detecting the chemically distinct undulations typically enlarge the membrane sepa- forms of lanthanideâlipid complexes, the lumines- ration and thus lower fusion tendency. cence dyes can be used to monitor the cation-induced lateral segregation [20]. 4.1. Catalysis via defects The dynamic light scattering is useful for directly assaying the vesicle fusion. The problem with the Several examples for the catalysis and/or the method is that it is extremely sensitive to strong promotion of membrane fusion by the defects are scatterers, such as dust particles or lipid aggregates, given in previous section. It is also likely that in and is consequently often misleading. Extreme care numerous other reported cases membrane imperfec- is therefore required to avoid sample contamination tions have played an important, but overlooked, role. and false positive assignments. The inability of the This may be due to the difficulty of directly asses- method to distinguish between particles with a size sing the defects in bilayers by the methods other than ratio less than 2 and the failure of many commercial measurements of membrane leakiness, which is not correlators to cover the size range greater than two applicable when fusion also occurs. Some indirect orders of magnitude adds further serious problems to hints are found in the literature, however. this kind of assay. Freezing and thawing promotes membrane fusion, Qualitatively, freeze-fracture electron microscopy probably because of the defects generated in the (EM) and fluorescence microscopy may provide membranes during the process. Membrane perfor- morphological insight into the fusion process. How- ation by ice crystals and thermotropic phase transi- ever, when cryo-protectants, membrane dehydration tions in the lipidâice mixture, as well as effectively agents, or other additives are used to prepare the increased lipid concentration in the partly frozen G. Cevc, H. Richardsen / Advanced Drug Delivery Reviews 38 (1999) 207 â232 217 suspension, all contribute to this. Specifically, the From the lipid point of view, the membrane phase point defects, which act as fusion intermediates, were transition temperature is most affected by the hydro- inferred to arise in the bilayers composed of mixed carbon chains. Extending aliphatic chains on the egg phosphatidylcholine and soybean phosphatidyl- lipid molecules, as a rule, increases the temperature ethanolamine [22]. of all transitions into a state of higher headgroup Increasing the lipid immiscibility enlarges the disorder, for example, the chain melting phase 9 9average width of defect lines, or fractures, within the transition temperature (L , L or P fi L transitionb b b a bilayer [23]. This promotes transbilayer solute diffu- at T ). On the other hand, prolonging the hydro-m sion and membrane fusion. carbon chain length decreases the T value of allm Gangliosides inhibit the phospholipase C-pro- phases in which the interfacial region is condensed moted fusion of lipidic vesicles in a multiple fashion: by heating, that is, the fluid bilayer- to non-bilayer they lower the activity of phospholipase C [24], (L or L or L fi H or fi Q at T ) transitions. Ifb b a a a h stabilize the lamellar lipid phase [24] and increase the transition into a non-bilayer phase occurs directly the repulsion between lipid bilayers [25]. from the ordered (gel, crystal) phase, the T vs.m chain length dependence is also positive. 4.1.1. Phase separation Chain unsaturation, which effectively shortens the Lateral phase separation in the mixed lipid vesi- part of the chain that is involved in collective phase cles favours bilayer fusion. Such separation is more transitions [29], decreases the T value. The chemi-m likely upon close membrane approach or where the cal structure of the lipid and its stereoisomerism buckling of membrane surface occurs [26]. Phase may, but need not, affect the chain-melting phase separation is also far more probable in the systems transition temperature; in the dry state, most phos- where at least one of the membrane components, pholipids with comparable chains melt at a similar when used as a pure substance, is in the ordered, gel temperature. phase [27]. The effect of lipid headgroup type on the mem- The importance of domain boundaries is supported brane phase transition increases with the amount of by the observation that the addition of saturated the membrane associated water. The variability of phospholipids to the unsaturated dioleoylphos- chain melting phase transition temperature with the phatidylethanolamine (DOPE) results in membrane changing lipid headgroup properties, consequently, is fusion below the chain melting phase transition the highest for the strongly polar and/or hydrated temperature of DOPE. This effect is eliminated by systems. the addition of membrane-stiffening cholesterol, The above mentioned considerations provide which also eliminates well-defined phase boundaries rationale for selecting thermolabile systems with a [28]. propensity for fusion. The additional requests are Defects, as well as phase separation effects on the that the low temperature phase must be acceptably fusion, can be demonstrated nicely by studying the stable and that the high temperature phase must have temperature-induced membrane unification. a sufficiently high aggregation capability. Several examples of the temperature-induced vesi- 4.1.2. Thermal induction cle fusion are given in Section 2.2. They all rely on Efficient and robust temperature-dependent control the use of relatively unpolar lipid mixtures with of the lipid membrane properties is possible when sufficiently high intrinsic tendencies for fusion to the thermal leverage factor is high. This means that respond to even a minor temperature change with a the membranes must be near a phase transition dramatic bilayer reorganization. Other such examples temperature for the induction of fusion, as only then are selected mixtures of dioleoylphosphatidyl- a small temperature shift will result in sufficiently ethanolamine, dioleoylphosphatidylcholine and large variations in the lipid bilayer characteristics. cholesterol. Heating a suspension of corresponding Chemical analogue for such a requirement is the vesicles reportedly results in membrane fusion [30], choice of a buffer with the pK value close to the possibly owing to the changes in lipid polymorphism working pH for optimum control. [31]. 218 G. Cevc, H. Richardsen / Advanced Drug Delivery Reviews 38 (1999) 207 â232 4.2. Electrical induction generation of a surface-confined stack of bilayers under the monolayer at 298C. The evolution of oligo- Any electric transmembrane field creates a pres- lamellae persisted for more than 15 h and generated sure across the hydrophobic membrane core [32]. As a stack with the bilayer repeat distance corre- long as the intermembrane field is independent of the sponding to that of the fully hydrated DMPC position, this pressure is proportional to the dielectric bilayers. Partial lipid dehydration was inferred from membrane constant and to the square of electrostatic the shift in the chain-melting temperature of the potential gradient across the membrane; however, it surface-attached multi-bilayers. The surface-induced is inversely proportional to the lipid bilayer thickness vesicle dehydration and deformation must also have [33]. been associated with the very high local curvature of Mechanical compression of membranes under the the deformed, pancake-like liposomes. This enforced influence of transmembrane pressure results in an vesicle fusion near the airâwater interface, probably elastic strain. If the latter is sufficiently high the at the sites 11 of contact between the defect rich bilayer collapses in a process called dielectric mem- edges. As the result, one layer of adhered vesicles brane breakdown [32]. Membrane fusion is often formed two surface-confined bilayer sheets until at observed in the cases in which several lipid vesicles least four fused vesicle layers were observed. are electrically aligned and the adjacent bilayer Similar processes were inferred, based on circum- regions are experiencing simultaneous membrane stantial data, to be ongoing on the surface after breakdown. non-occlusive administration of liposomes on the Hemifusion might precede the electrically-induced skin [40]. cell fusion [34]. It was reported that the fusion-inhibiting peptides may stabilize the membrane structures with a large 4.3. Curvature effects radius of curvature and thus diminish the probability for the formation of small, defect rich and thus The rate of vesicle fusion and the role of bilayer fusogenic domains [41]. curvature was investigated a long time ago [35,36]. It is now agreed that the high membrane curvature increases the likelihood for membrane unification. 4.4. Surface tension The stress and the strain in the bilayer are the probable reasons for this, as they both induce Surface tension is yet another example of physical membrane defects and provide the energetic source âforceâ that can create stress accompanied by defects for fusion. It is therefore difficult, if not impossible, in the membrane and thus promote membrane fusion. to distinguish the effects of increasing membrane In an experiment designed to clarify the role of curvature from the consequences of growing mem- surface tension in the fusion process, monolayers brane defect density. were prepared from the same lipids as were used A combination of acyl chain unsaturation with the separately in the vesicle fusion experiments. The high membrane curvature was postulated to favor monolayer surface tension was measured as a func- fusion of pure phosphatidylcholine membranes [37]. tion of the divalent cation concentration in the A similar conclusion was also reached for the fusion supporting aqueous medium. Changing the divalent between different vesicles [28]. ion concentrations from zero to a threshold con- Vesicles formed from the bisubstituted, bola lipids centration needed to induce fusion (¯8 mN/m) initially do not fuse, irrespective of the employed increased the surface tension [42]. The latter was 21fusogen, Ca or poly(ethylene glycol) [38]. At later hence inferred to play a dominant role in the fusion times, however, non-leaky fusion is observed [39]. process. A similar message could also be cast in The airâwater interface above a suspension of different words, however, using the fact that surface sonicated, and thus defect-rich, DMPC vesicles was tension corresponds to the surface density of the studied by means of X-ray reflectivity as a function membrane surface free energy; it includes the contri- of temperature and time [10]. This revealed the butions from surface electrostatics, hydration, steric G. Cevc, H. Richardsen / Advanced Drug Delivery Reviews 38 (1999) 207 â232 219 interactions, etc., which have been argued to play a dependent repulsion [45â47]. Bilayers must conse- role in the membrane fusion process. quently first be dehydrated partly in order to be brought close together. Moreover, osmotic swelling of the vesicles contacting the planar membrane puts 5. Chemical induction bilayers under stress and may lead to their fusion [48]. Membrane fusion is induced or promoted by many For example, high molecular, hydrophilic poly- molecules with several features in common. Most mers compete with the vesicles for the water mole- often, such molecules either interact with the lipid cules and thus exert an osmotic effect on the headgroup or else insert themselves into the lipidâ membranes. Partial bilayer dehydration is a conse- water interface. In either case the forces between the quence of this. Membrane fusion during bilayer lipid bilayer, and between the lipid molecules them- dehydration is believed to result from the close selves, are modified. This provokes changes in the contacts near the highly curved edges [49]. membrane structure and may result in membrane Osmotic membrane dehydration and perturbation fusion. by ions synergistically promotes vesicle fusion [50]. More direct membrane dehydration, combined with 5.1. Charge neutralization the high membrane tension or local curvature, also results in a spontaneous bilayer fusion [51]. The Membrane surface charge density affects the latter is consequently observed when the osmotically propensity for bilayer fusion. The net surface charge active molecules such as polyethyleneglycols or of equal sign lowers the fusion probability owing to other hydrophilic polymers are added to a suspension the long range Coulombic repulsion between the of lipid vesicles. membranes [43]. The presence of screening, or even complexing ions, diminishes such repulsion or elimi- 5.2.1. Polyethyleneglycol nates it completely. The presence of charged lipids in In contrast to the fatty alcohols, which are dealt only one population of vesicles is not detrimental to with in the following section, polyethyleneglycols the fusion with the other uncharged or oppositely (PEG) have negligible membrane solubilities. PEG charged liposomes. Indeed, the latter fuse more polymers or their derivatives hence perturb the avidly because of the electrostatic attraction between surface more than the hydrophobic core of the lipid the membranes. This was shown by studying unifica- bilayers [52]. This was concluded from the large tion between the large unilamellar cationic vesicles effect of the polymer on the headgroup or on the composed of dioleoylphosphatidylethanolamine membraneâsolution interface of DPPC bilayers, as (DOPE) and N,N-dimethyl-N,N-di-9-cis-octa- assessed spectroscopically. In the presence of PEG- decenylammonium chloride (DODAC) (1:1) with the ,fluorescent anisotropy of an interfacial probing dye target liposomes of varying composition [28]. Fur- [52] and interfacial tension [53] both increase where- ther evidence for the importance of charge neutrali- as the interfacial dielectric constant decreases [53]. zation in the fusion process will be given in Section This supports the above mentioned concept of PEG- 5.4. induced membrane dehydration and subsequent membrane fusion. 5.2. (De)Hydration PEG-induced membrane dehydration is well corre- lated with the extent of membrane fusion. It also At short distances membranes repel each other leads to an asymmetry in the lipid packing pressure owing to the hydration-dependent steric barrier. The in the two leaflets of the membrane bilayer. A single range as well as the magnitude of such repulsion, bilayer septum is formed at the point of closest depend on the effective surface polarity [44] and on apposition of the two membranes. Such single the interfacial thickness and softness; the wider and bilayer septum then decays during the formation of softer the transition region between the membrane the initial fusion pore [54] or a defect cluster. core and water sub-phase, the stronger the hydration In the presence of high-molecular-weight PEG, 220 G. Cevc, H. Richardsen / Advanced Drug Delivery Reviews 38 (1999) 207 â232 membranes preferentially fuse at the sites of the ly interact with the membranes, and affect the bilayer dehydration-induced discontinuities in the adjacent fusion, in the same manner. According to the results 1bilayers [55]. This was explained within the frame- of H-NMR, poly(styrene sulfonic acid) and poly- work of mechanical stress model using so-called (ethylene glycol) both slow down the choline methyl âosmophobic associationâ theory [56â58]. A similar group motion [64]. On the other hand, poly(L- explanation was proposed for the fusion of two glutamic acid) and poly(N-vinyl-2- pyrrolidone) adjacent cells in the presence of PEG [55]. The affect not only the choline methyl group but also the addition of polymers increases the permeability of hydrophobic methylene and terminal methyl groups artificial as well as natural (hen erythrocyte) mem- [64]. The propensity for DPPC vesicle fusion is branes [55], probably via the osmotic induction of simultaneously decreased by the latter two kinds of the defects in membranes. polymer. Complexation of the polymer with dipal- Small unilamellar vesicles prepared from egg mitoylphosphatidylcholine vesicles may be respon- phosphatidylcholine fuse readily [37], when the sible for such stabilization [64]. bilayer surface is sufficiently dehydrated with in- creasing amounts of PEG. Enforced cell fusion 5.3. (De)Protonation shows shows a similar dependence on PEG con- centration [59,60]. The protonation of lipid headgroups diminishes Increasing PEG concentration reportedly increases membrane propensity for water binding and typically the efficiency of erythrocyte fusion, but the fusion of brings the vesicle suspension to a collapse [44]. This whole cells must be interpreted with caution [16]. is true even when the final membrane charge differs Studies with the lipid vesicles are more reliable in from zero, as long as the final membrane hydro- this respect. For example, PEG solution (3.8â40 philicity (polarity) is low; acidic solutions often wt.%), brought into contact with the large unilamel- exhibit such a behaviour. Stripping protons from the lar DPPC vesicles formed by a rapid extrusion lipid membrane often gives rise to a net membrane technique, resulted in lipid mixing between the charge and generates electrostatic repulsion between vesicles but not in fusion [61]. In contrast to this, the bilayers. Membrane deprotonation also increases DOPC vesicles with a small amount of an am- lipid hydrophilicity and hampers the inter-lipid hy- phipathic impurity have been observed to fuse in the drogen bonding; both loosen the lipid headgroup presence of PEG at room temperature [61]. The packing and minimize the likelihood for lipid bilayer requirement for a (small) membrane perturbation, as fusion. a prerequisite for the vesicle fusion, was confirmed The phenomena mentioned in the previous para- for dipalmitoylphosphatidylcholine and cardiolipinâ graph are explicable by the relative weakness of dioleoylphosphatidylcholine (1:10, mol /mol) mixed electrostatic repulsion near a membrane: the charges membranes [50]: PEG induced the fusion of such on a bilayer contribute to the stability of the suffi- 21 vesicles only after the addition of Ca and phos- ciently hydrophilic and well hydrated lipid layers. pholipase A2 [50]. The opposite effect was also However, lipid ionization is of little importance observed: in the presence of PEG, sub-threshold when the polarity of membrane ingredients is so low 21 21 concentrations of Ca (0.5 mM) or Mg (2 mM) that only poorly hydrated bilayers are formed. In the provoke an extensive fusion of phosphatidylserine limited residual inter-membrane space the hydration containing vesicles [62]. and steric forces then prevail [65,33]. The susceptibility of various membranes to PEG- Acidification of the mixed lipid suspensions con- induced fusion appears to be inversely correlated taining PCâFA (1:2) complexes therefore induces with susceptibility to virus-induced fusion; increas- massive vesicle aggregation, which is the prereq- ing the degree of chain unsaturation in the membrane uisite of successful intervesicle fusion. The pH-sen- decreases the susceptibility of simple lipid vesicles to sitivity of some lipid vesicles thus provides the PEG-induced fusion but makes the cells more sensi- means for devising agent carriers that are directed tive to the similarly conducted viral infection [63]. chemically, via the local acidification or acidity, to Different water-soluble polymers do not necessari- the designated site. G. Cevc, H. Richardsen / Advanced Drug Delivery Reviews 38 (1999) 207 â232 221 Protons binding to a membrane not only affect the precipitation of phosphatidylserine vesicles but no colloidal stability of the suspension but also in- significant fusion or phase transition [71]. 21fluence the membrane polymorphism. Moreover, pK The fusogenicity of Ca excels over that of the values [66], and even ion binding constants [67] of other divalent cations [72], probably owing to the phospholipids depend on the physicochemical state convenient stereochemical properties of calcium for 21 of the membrane. The degree of membrane protona- the binding to biological targets. The influx of Ca tion can therefore be affected without changing the into vesicles is not a prerequisite for the highly bulk pH value. This, together with the accompanying effective induction of fusion [73]. For example, the modifications in the lipid headgroup packing, offer secretory vesicles from rat liver adhere in the 21 means for controlling the aggregation of lipid vesi- presence of micromolar concentration of Ca , other 21 21 21 cles by physical methods, such as hyperthermia, divalent cations, such as Ba , Sr , and Mg 21described in previous section. being ineffective in this respect [72]. Mg even 21inhibits the Ca -induced fusion of rat liver secret- 5.4. Ion binding ory vesicles [72]. When added to a suspension of 21phosphatidylserine vesicles Mg induces fusion, 21Ions not only screen the electric charges on a however, albeit at higher concentrations than Ca ; 21 membrane but also modify the surface polarity and Mg -induced phase separation is not detected [62]. thus the hydration-dependent inter-membrane repul- Ions and other fusion-relevant factors often act sion [43]. The inorganic di- and trivalent cations synergistically: the threshold concentration for the 21typically lower and the corresponding organic ions Ca induced fusion between the defect-rich small most often increase such repulsive force. However, and the defect-poor large unilamellar phosphatidyl- anions, especially halides, tend to increase the mem- serine vesicles is 1.2 and 2.4 mM, respectively. 21brane swelling (with the exception of fluorine). However, at saturating Ca concentration (.10 The chief explanation for the effect of simple mM), the rate of small unilamellar vesicle fusion is inorganic cations is the decrease of the electrostatic only slightly higher than that of large vesicles under interaction length in the electrolyte. This reason does identical external conditions [19]. 21 not hold in the case of protons and lithium. These In the presence of Ca , fusion of phosphatidyl- ions directly decrease the surface charge density of serine vesicles proceeds more rapidly (,5 s) than the (negatively charged) membranes and also lower the process of phase separation (1 min) [62]. the surface hydration, due to their high affinity for Membrane sensitivity to the induction of fusion by the bilayer surface [68â70]. ions increases with the surface charge density. Other ions with the comparable ability normally Liposomes containing polyphosphoinositide, with 21 carry several charges. Alkali earth ions (Mg , several negative charges, or phosphatidic acid, with 21 31 31Ca , Sr , Ba ) are particularly well explored, 1.5 negative charges, respond with extensive fusion 21 21 and relevant, examples for this [42]. Further exam- to the addition of Ca or Mg . Similar vesicles ples are referred to in Table 1. comprising only mono-charged phosphatidylserine or In the presence of lithium [69], calcium [71] or phosphatidylinositol, or the zwitterionic phospha- several other multiply charged ions [53], suspensions tidylcholine, are less responsive [74]. of charged (e.g. phosphatidylserine) lipid vesicles Table 1 gives a survey of selected literature coalesce. The ion-induced fusion between the acidic dealing with such different systems. and neutral phospholipid vesicles is also accom- panied by a decrease in the surface dielectric con- 5.5. Induction by fusogenic lipids stant and surface-tension [53]. Quite often, the hydrocarbon lipid chains simultaneously undergo an 5.5.1. Fatty acids isothermal phase transition from a fluid into a Fatty acids, added to the phospholipids in a molar crystalline state [29,71]. Both processes are sequen- ratio of a few percent, induce isothermal phase tial but the former is independent of the latter: 2.0â transitions between the various lamellar phases and 215.0 mM Ca solution induces aggregation and promote vesicle fusion [85â87]. Greater amounts of 222 G. Cevc, H. Richardsen / Advanced Drug Delivery Reviews 38 (1999) 207 â232 Table 1 5.5.1.1. Chainlength dependence We have based the search for the best thermolabilePS PG PI PC 21 fusogenic vesicles on the principles of rationalCa [35,75] [76,77] [78] [79â82,75] 21 membrane design [92]. Fig. 8 illustrates the rationaleMg [75â77] [35,71] [76,77] [75,79,81â84] 21Mn [76,75] [76,77] 2 [81,75] for this and also elucidates the effect of hydrocarbon 31Sr [79] [77] 2 [81,84] chain length on the colloidal and phase properties of the mixed PCâFA systems. Some of the corre- sponding experimental data are given in Fig. 9: the such chemical fusogens, typically at the phos- upper panels pertain to the vesicle aggregation and pholipidâfatty acid ratio of 1:2, lead to the formation the lower panels to the fusion between lipid mem- of non-lamellar fluid phases [88] which are preceded branes consisting of stoichiometric PCâFA mixtures by the total collapse of colloidal vesicle stability and with different chain lengths. The temperature at massive membrane unification [89,90,13]. which lipid chains melt and at which the transition Several fusogenic lipid mixtures have been iden- into the inverted hexagonal phase occurs are given as tified to date. Some of these offer the advantage of T and T , respectively.m h the transition temperature in the physiological range In the upper panel of Fig. 8 the chain-melting or close to it (37â428C). Such mixtures qualify for phase transition temperature of pure fatty acids is the controlled fusion with the living cells. Combina- given to clarify the correlation between the thermo- tions of phosphatidylcholines (PCs) and fatty acids dynamic properties of the single- chain component (FAs) are the most affordable and the least toxic vs. the mixed lipid system. In the lower half of Fig. amongst these mixtures [13, 89,91]. 8, the corresponding data for the pure PCs (thin line), Fig. 8. Chain melting phase transition temperature (T ) of dry fatty acids (upper panel; X-COOH) and of the fully hydratedm phosphatidylcholines (DXPC) or DXPCâhomologous fatty acid (1:2) mixtures. Upon the transition into a non-lamellar phase at T 5 T ,h massive membrane aggregation and fusion occurs. Chain-melting of the phospholipid component (at T5T ) may catalyze fusion out of itsm own strength, however (see Fig. 3). Horizontal lines give the temperature of therapeutically interesting hyperthermia. For optimal performance, the curves of rapid aggregation and maximum fusion should be close together. G. Cevc, H. Richardsen / Advanced Drug Delivery Reviews 38 (1999) 207 â232 223 Fig. 9. Effect of temperature on the aggregation (upper panels) and fusion (lower panels) between the freshly made phosphatidylcholineâ homologous fatty acid (1:2) mixtures at pH 5 5. Note that the aggregation rates (C ) given in the upper panels are plotted logarithmically. Ta m and T give the positions of chain-melting and lamellar-to non-lamellar phase transitions, respectively. The extent of fusion above T ish h underestimated, owing to the lipid sticking to the cuvette walls. for the homologous PCâsaturated-FA mixtures (1:2; (DPPC fi DSPC) raises the phase transition tempera- thick curve with dots) and for the DPPCâalkenoic- ture only by a few degrees, the T value then beingh FA 1:2 mixtures (line with crosses) are shown. The ¯658C in the case of DSPCâPA (1:2), for example. chains of the corresponding asymmetric DPPCâMA All these changes are reflected in the temperature- and DPPCâSA mixtures melt at 54 and 668C, dependence of the corresponding mixed lipid mem- respectively. brane fusion, which always has a maximum near the chain-melting phase transition temperature. 5.5.1.2. Chain unsaturation Of all the PCâFA mixtures explored to date, the Substitution of the free palmitic acid with 9-trans- 1:2 molar mixture of DPPCâELA-COOH comes octadecenoic acid (C , elaidic acid; ELA-COOH: closest to the ideal requirements for the temperature18:1t T 5518C) in the DPPCâFA complexes shifts the controlled vesicle fusion. Comparison of the datam chain-melting phase transition temperature from shown in Figs. 6 and 9 supports such a conclusion. ¯62 to 458C (see Fig. 8). Palmitic acid substitution The DPPCâELA-COOH mixture from Figs. 8 and 6 with 11-trans-octadecenoic acid (C transvaccenic is seen to have the transition into the inverted18:1 acid: T 5398C) brings the transition temperature hexagonal phase at ¯458C and the chain meltingm value from 62 to 468C at pH 5 9 (open disk in Fig. phase transition at ¯438C. While the former maxi- 8). (Scanning calorimetry in acetate buffer (pH 4.9) mizes the degree of vesicle aggregation, the latter gives 418C for the latter PCâFA complex.) Fully makes sure that the fusogenic membrane defects will hydrated DPPCâoleic acid (1:2) complexes melt arise at the preferred temperature of 438C. near 408C. Palmitic acid replacement with the 9-cis- hexadecenoic acid (C , palmitoleic acid: T 50 5.5.2. Fatty alcohols16:1t m 58C) in the DPPCâFAs complexes lowers the T Fatty alcohols with a chain length greater than onem value to ¯368C. In contrast to this, an extension of half of the phospholipid chain exert very similar the phospholipid chains by two methylene groups effects on the lipid bilayers as do fatty acids. They 224 G. Cevc, H. Richardsen / Advanced Drug Delivery Reviews 38 (1999) 207 â232 induce non-lamellar, inverted fluid phases and pro- layers of two membranes. This is called the fusion mote vesicle aggregation and fusion. Short chain stalk [101â103] (see Fig. 10). alcohols induce hydrocarbon interdigitation in the gel The membraneâsolution interphase plays an im- phase and lipid bilayer solubilization in the fluid portant role in the process of fusion. If the plasma phase [93,94]. Before the solubilization, however, membrane is covered with a thick protein layer or massive vesicle fusion takes place, which leads to glycocalix, fusion is difficult to achieve. On the other increased turbidity of the suspension. In this aspect, hand, a superficial macromolecular scaffold of pro- fatty alcohols resemble lysophospholipids, as they teins that brings two plasma membranes close to- first induce fusion and then solubilization of lipid membranes [95]. 5.5.3. Other amphiphiles Lysophospholipids or other biological detergents, such as deoxycholate, partition avidly into lipid bilayers. In the low relative concentration range such molecules consequently facilitate vesicle fusion [49]. The chief reason for this is that the detergent-like molecules create defects in the membrane, which catalyze membrane fusion [96]. At higher concen- trations such molecules suppress the fusion [97]. This is partly owing to the increased repulsion between the membranes. Another possible source of suppression is the inhibition of membrane curving in the right direction by the âinverted cone shapedâ lysophosphatidylcholine [97]. For example, when lysophosphatidylcholine or retinol are added to phosphatidylcholine liposomes at 378C for 5â24 h, fusion efficiency of up to 25% is measured [98]. Sodium cholate and other bile salts also induce vesicle fusion and growth, in a con- centration dependent manner [99]. Hydrocortisol, which stabilizes the membrane, as do some other sterol derivatives, decreases the fusion probability [98]. 6. Biological fusion Biological membrane fusion is a localized event: extremely fast, non-leaky, and well controlled. Very often, such fusion originates from the protein-modu- lated membrane destabilization. Many researchers believe that the transition involves some kind of Fig. 10. Stalk mechanism of membrane fusion: two closelynon-bilayer structure [100]. Indeed, it has been approached bilayers (A) form a stalk following close apposition ofpostulated that any biological fusion, irrespective of the interfaces (B). Under certain conditions [102] the monolayersits trigger, involves the formation of a highly bent undergo a continuous deformation toward a trans-monolayer intermediate between the interacting membranes â a contact, TMC (C). This strongly curved dimple ruptures under neck-like local connection between contacting mono- stress to form a fusion pore (D). G. Cevc, H. Richardsen / Advanced Drug Delivery Reviews 38 (1999) 207 â232 225 gether, creates the right architecture for fusion [51]. near an extended (plasma) membrane finally triggers This happens with the secretory granule, for exam- exocytosis [48]. ple. A dimple in the plasma membrane is then Sarcoplasmic reticulum vesicles fuse during formed that overcomes the interbilayer repulsion and anisodiametric dehydration. They then form a sheet- permits the two membranes to âjumpâ into intimate like structure with a mean extension of 500 nm [49]. contact. As a result, a single hemifused bilayer Furthermore, it was hypothesized that the phase- emerges [104]. separated domains of cholesterol in the bile vesicle Another possible biological mechanism of vesicle bilayer mediate the close approach of the vesicles docking and fusion is formulated in the SNARE [3]. Molecular packing defects at the domain hypothesis: vesicle SNAREs (protein synaptobrevin boundaries were suggested to facilitate the hydro- and homologues) mediate docking by binding to phobic interaction between and fusion of apposed target SNAREs (protein syntaxin-SNAP-25 and membranes. Other mediators of aggregation and homologues), whereupon SNAPs and NSF bind to fusion are likely to be proteins that interact hydro- elicit membrane fusion [105]. SNAREs (SNAP phobically with the bile vesicle and diacylglycerol receptors), rab3 and Gai3 proteins are compartment- formed in the vesicles by phospholipase C action [3]. specific. They form a putative scaffold of proteins on Influenza hemagglutinin also induces transient do- the membrane, thus causing bilayer fusion during mains during membrane fusion [113]. exocytosis [21,106]. Lysolipids added in appropriate concentration 21NSFâSNAP fusion machinery, with the NSF, range reversibly inhibit the Ca -, GTP- and pH- SNAP and SNAP receptors, assembles to bridge dependent fusion of biological membranes [101], partner membranes selectively in a complex that partly through their action on proteins and partly by contains elements of both vesicle and target mem- increasing the intermembrane repulsion. Used in branes. Similar fusion machines drive both constitu- greater quantity similar lipids provoke membrane tive fusion (ER fi Golgi fi surface and endocytosis) morphology changes, including budding and finally and regulated exocytosis [107]. Exocytosis may also membrane lysis [114]. involve hydrophobic bridges that depend on the presence of synexin [108]. 6.1. Protein effect Enzymatic modification of the polyphos- phoinositide content of intra-cellular membranes can Proteins are responsible for the mutual recognition contribute to the regulation of biological fusion [74]. of the fusion partners. Moreover, they appear to be The contact-induced formation of nonbilayer lipid involved in the initiation of biomembrane fusion, by structures triggered by phosphatidylinositol turnover locally producing or activating fusogens, or by acting is an example of how the biochemical and required as fusogens in their own right [100]. physical changes may be coupled [109]. While the specificity and timing of membrane Fusion of human erythrocytes induced by Sendai fusion, including virus cell fusion, is governed by virus appears to be triggered indirectly through proteins, fusion always involves merging of the conditioning and modification of the erythrocyte adjacent lipid bilayer membranes. Protein-induced membranes by viral membrane components and/or fusion was therefore suggested to begin with the the generation of membrane connecting tubuli formation of a bent, lipid involving, stalk inter- [110,111]. mediate [97]. Another possibility is that such pro- Simple physical phenomena are also involved in teins simply induce enough disorder in two mem- the process of biological fusion, in addition to the branes, which they have brought close together, to above mentioned mechanisms. For example, the fuse in a less ordered fashion. secretory granule membrane is under tension; this Peptides carbobenzoxyâL-PheâL-Phe (ZâL-PheâL- results in a net transfer of lipids from the plasma Phe), ZâL-Phe, ZâD-Phe and ZâGlyâL-PheâL-Ph membrane to the secretory granule, while the two inhibit the fusion between N-methyl DOPE large participating bilayers are connected by the fusion vesicles. The effect on Sendai virusâvesicle fusion is pore [112]. Bilayer stress in the secretory vesicle similar [41]. 226 G. Cevc, H. Richardsen / Advanced Drug Delivery Reviews 38 (1999) 207 â232 Phospholipase A2 digestion, which generates Phospholipase C activity, which generates dia- lysophospholipids, facilitates vesicle coalescence cylglycerol from various phospholipids, induces and, finally, fusion [49]. The fusogenic activity of massive aggregation and fusion of multilamellar and the snake venom neurotoxin b-bungarotoxin is also large unilamellar lipid vesicles and leads to the correlated with the known phospholipase A2 activity formation of âsealedâ lipid aggregates [118]. It was 21 of the toxin [96]: both functions require Ca at speculated that the resulting membrane fusion in- almost stoichiometrical concentrations, much below volves inverted phases in which the locally produced the threshold values found for unspecific divalent diacylglycerol is accumulated [118]. cation-induced vesicle fusion. Similarly, the presence of phosphatidic acid in the target membrane is essential for both fusogenic and enzymatic activity of b-bungarotoxin [96]. Molecular mechanism of fusion 7. Pore hypothesis / stalk intermediate involving snake venom neurotoxin binding to the negatively charged groups on the membrane surface Highly bent (net negative curvature) stalk-type was hypothesized to depend on the local formation intermediates [119] or other transient lipidic con- of lysophospholipids which, in certain concentration nections between the fusing membranes are typically range, promote fusion (see previous discussion). involved in membrane unification. The former inter- Synexin, an adrenal medullary protein, facilitates mediates can be studied well in artificial systems, 21 21the Ca - but not the Mg - mediated fusion which permit separate manipulation of the outer and of phosphatidateâphosphatidylethanolamine (1:3) inner monolayers. and phosphatidateâphosphatidylserineâphosphatidyl- In one particular study, amphipaths inserted into ethanolamineâcholesterol (1:2:3:2) vesicles [115]. the outer membrane leaflet did not promote cell Synexin, in the mg/ml range, lowers the threshold fusion, independently of whether the agent induced 21 concentration for Ca dependent fusion to approxi- monolayers to bend outward (conferring positive mately 10 mM. The effect is drastically reduced by spontaneous monolayer curvature) or inward (nega- including 25 mol% PC in the vesicle membrane. tive curvature) [120]. In contrast, when created by Gramicidin, with detergent-like properties, induces agent incorporation into the inner leaflet, positive aggregation and fusion of large unilamellar curvature resulted in extensive fusion. This suggests dioleoylphosphatidylcholine vesicles when peptide to that fusion is completed when a lipidic fusion pore, lipid molar ratio exceeds 1:100. Increased membrane with the net positive curvature, is formed by the permeability to small solutes, as well as contents inner leaflets composing a hemifusion diaphragm. mixing, occur subsequent to the vesicle aggregation After the ectodomain induces hemifusion, the trans- and lipid mixing. Although addition of gramicidin at membrane domain forms a pore by conferring posi- peptideâlipid molar ratio exceeding 1:50 eventually tive spontaneous curvature to the leaflets of the leads to the inverted hexagonal H phase formation, hemifusion diaphragm [120].II leakage during fusion (1â2 min) is not the result of Lysophosphatidylcholine added to the contacting H phase formation but rather the result of local monolayers of adjacent membranes inhibit hemifu-II changes in the lipid structure caused by precursors of sion. The following experimental finding supports this phase [116]. The peptide ability to induce H this idea: when inserted into the distal monolayer ofII phase formation and its ability to induce vesicle the planar membrane, lysophosphatidylcholine pro- fusion and leakage are well correlated. moted, whereas arachidonic acid inhibited fusion More examples can be given. A 21 amino acid pore formation. The intermediates of hemifusion and peptide (AcE4K) coupled to a distearoylglycerol fusion pores in phospholipid membranes therefore lipid anchor and incorporated into POPC or egg involve different membrane monolayers and may PCâcholesterol (ePCâChol, 55:45) liposomes at pH have opposite net curvatures [121]. 7.5 affects membrane fusion. Vesicle stability is In one cryo-EM study, bilayer structures in vit- decreased and fusion probability is increased with rified thin films were often observed in coexistence higher lipopeptide concentration, decreasing pH, and with inverted hexagonal structures. The bilayer areas inclusion of cholesterol [117]. frequently exhibited complex, multiple contacts be- G. Cevc, H. Richardsen / Advanced Drug Delivery Reviews 38 (1999) 207 â232 227 tween stacked membranes [31]. Conversely, no 9. Final remarks lipidic particles or the hexagonal H phase wereII observed during the first few seconds of fusion Phospholipid vesicle models of fusion do not between the large (0.2 mm diameter) unilamellar perfectly mimic the process of cellular fusion since cardiolipinâphosphatidylcholine and phosphatidyl- the latter most often relies on fusogenic proteins. serineâphosphatidylethanolamine vesicles induced However, such models provide good insight into the 21by Ca [122,123]. Such entities were proposed to mechanism of bilayer fusion. They have shown that be the intermediate structures in membrane fusion. it is sufficient to disorder the bilayer locally to Lipidic particles were visualized, however, by quick- induce membrane fusion. Ideally the perturbation freezing freeze-fracture electron microscopy at method should catalyze and support membrane ag- equilibrium long after the initiation of fusion. They gregation as well. were particularly prevalent in the presence of gly- Non-lamellar lipid structures, in our opinion, are cerol [122,123]. not a prerequisite for vesicle fusion; however, mem- It is tempting to speculate that the non-leaky brane fusion is an inevitable consequence of the fusion might involve the stalk mechanism while the bilayer-to-nonbilayer phase transformations. merging of membranes accompanied with a volume Comparisons of experimental data with (i) the loss could rely more on the âmessy reorganizationâ calculations that describe fusion pores and (ii) the involving annealing and redistribution of defects. breakdown pores, support the view that fusion pores are initially passages through a single bilayer, as would be expected if membrane fusion proceeds via the hemifusion mechanism [128]. This would in- 8. Fusion kinetics volve highly bent intermediates-stalks, that is, local connections between the contacting monolayers of Bilayer fusion is a fast process completed within fusing membranes [129]. No experimental evidence less than a millisecond. This prevents reliable vis- is available, however, to prove the necessity for stalk ualization of the fusion intermediates, which require formation during the membrane fusion. certain time for external induction, and makes in- Membranes comprising lipids with a relatively low direct kinetic studies indispensable. While cryo-elec- affinity for water and non-charged lipids are more tron microscopy remains to be used to a greater likely to fuse than the bilayers that consist of very extent, the majority of existent information comes polar or ionized lipids. Phosphatidylethanolamine, all from the studies with kinetically trapped systems in common phospholipids at low pH and diacylglyc- which putative intermediates were captured in erols belong to the former class of lipids; phos- thermodynamically (meta)stable state with a suffi- phatidylcholine, charged lipids (e.g. phosphatidyl- cient lifetime for detailed inspection. In the living glycerol, phosphatidylserine or phosphatidic acid) in world, fusion intermediates are sometimes stabilized the neutral pH range or long-headed lipids (such as by the proteins. gangliosides and neoglycolipids) are representatives From the studies of exocytosis from mast cells it of the fusion-suppressing molecules. was concluded, for example, that several early events Lysolipids and other highly soluble biological in the life of a membrane fusion pore are reversible amphiphiles can take either role, depending on their [124]. Later steps are irreversible, however, and relative concentration, owing to their curvature-sup- affected by various parameters, such as the in- porting (and thus fusion-enhancing) effect on the one tramembrane tension. The latter acts directly [125] or hand and due to their repulsion-increasing (and thus indirectly, by influencing the formation of lipidâ fusion-decreasing) action on the other hand. protein complexes [126]. The irreversible expansion Lipids affect membrane fusion in a manner corre- of the fusion pore in the secretory granule mem- lated with their dynamic molecular shape, as this branes provides an example for this [125]. characteristic is believed to govern the individual The first milliseconds of pore formation by a monolayerâs propensity to bend in different direc- fusogenic viral envelope protein during membrane tions. fusion are reviewed in Ref. [127]. Membrane fusion is preceded or accompanied by 228 G. Cevc, H. Richardsen / Advanced Drug Delivery Reviews 38 (1999) 207 â232 ⢠close bilayer approach changes, lipid mixing, protein insertion or domain ⢠generation of defects or non-bilayer structures formation must be scrutinized â and used â in the (isothermal lipid phase transition) light of this concept. ⢠increased bilayer permeability ⢠(transiently) decreased interfacial dielectric con- stant References ⢠(transiently) increased surface tension ⢠exchange of lipid material between the originally [1] D. Papahadjopoulos, S. Nir, N. Duzgunes, Molecular mecha- nisms of calcium-induced membrane fusion, J. Bioenerg.separate bilayers Biomembr. 22 (1990) 157â179. ⢠vesicle morphology/size change [2] J.R. Monck, J.M. Fernandez, The fusion pore and mecha- nisms of biological membrane fusion, Curr. Opin. Cell. Biol. To study the phenomena listed above, differential 8 (1996) 524â533. scanning calorimetry, X-ray diffraction, and freeze- [3] N. Duzgunes, Molecular control of vesicle aggregation and membrane fusion: hypotheses on bile vesicles, Hepatologyfracture electron microscopy are most often used. 12 (1990) 67Sâ74S.Cryo-electron microscopy has much to offer in this [4] R.C. Macdonald, Surface chemistry of phospholipid vesicles respect. relevant to their aggregation and fusion, Hepatology 12 The initial fusion of small phosphatidylserine (1990) 56Sâ60S. vesicles is accompanied by 10% leakage of the [5] P.R. Cullis, M.J. Hope, B. de Kruijff, A.J. Verkleij, C.P.S. Tilcock, Structural properties and functional roles of phos-internal volume per fusion event; in contrast, the pholipids in biological membranes, in: J.F. Kuo (Ed.),fusion of large liposomes made from similar lipid is Phospholipids and Cellular Regulation, Vol. 1, CRC Press, essentially non-leaky. Massive release of vesicle Boca Raton, FL, 1985, pp. 1â59. contents therefore appears to be a secondary phe- [6] T. Landh, From entangled membranes to eclectic mor- nomenon related to the collapse of fused vesicles. phologies: cubic membranes as subcellular space organizers, FEBS Lett. 369 (1995) 13â17.The extent of membrane fusion correlates well [7] H. Ellens, D.P. Siegel, D. Alford, P.Y. Yeagle, L. Boni, L.J.with the lowering of interfacial dielectric constant Lis, P.J. Quinn, J. Bentz, Membrane fusion and inverted and with the increase in interfacial tension of the phases, Biochemistry 28 (1989) 3692â3702. membranes; both are indicative of a greater surface [8] P.R. Cullis, M.J. Hope, Effects of fusogenic agent on hydrophobicity [53]. membrane structure of erythrocyte ghosts and the mechanism of membrane fusion, Nature 271 (1978) 672â674.Protein organization into a macromolecular scaf- [9] G. Cevc, D. Marsh, Phospholipid Bilayers, Wiley-Inter-fold can bring adjacent membranes closer together science, Boca-Raton, FL, 1987. and induce hemifusion, that is, the fusion of the [10] W. Fenzl, L. Sigl, H. Richardsen, G. Cevc, The surface- apposing leaflets of the two membranes into a confined structures of dimyristoylphosphatidylcholine common bilayer. A small, dynamic fusion pore bilayers in contact with the vesicle suspension as studied by means of X-ray reflectivity, Colloid. Surf. A 102 (1995)forms in the common bilayer. 247â256.This usually expands irreversibly to complete [11] R.A. Parente, B.R. Lentz, Rate and extent of poly(ethylene membrane merging [2]. Other proteins, with less glycol)-induced large vesicle fusion monitored by bilayer pronounced tendency to aggregate, can still promote and internal contents mixing, Biochemistry 25 (1986) 6678. fusion by inducing defects in the membranes. Induc- [12] R. Lawaczeck, M. Kainosho, S.I. Chan, The formation and annealing of structural defects in lipid bilayer vesicles,tion of non-lamellar phases by the fusion-inducing Biochim. Biophys. Acta 443 (1976) 313â330.proteins or peptides has not been observed to date; if [13] Zellmer, S., Cevc, G., Risse, P. (1994) Temperature- and anything, such proteins are seen to interfere with the pH-controlled fusion of the complex lipid membranes. formation of non-bilayer phases. examples with diacylphosphatidylcholineâfatty acid mixed To control membrane fusion one must therefore liposomes, Biochim. Biophys. Acta, 101â113. [14] D. Hoekstra, Fluorescence assays to monitor membranefirst explore and then learn to master the multi- fusion: potential application in biliary lipid secretion andparameter space in which defect density and mem- vesicle interactions, Hepatology 12 (1990) 61Sâ66S.brane proximity are the final and decisive factors. [15] D.K. Struck, D. Hoekstra, R.E. Pagano, Use of resonance The effects of temperature, phase transitions, intra- energy transfer to monitor membrane fusion, Biochemistry bilayer tension, curvature, pH or salt concentration 20 (1981) 4093â4099. G. Cevc, H. Richardsen / Advanced Drug Delivery Reviews 38 (1999) 207 â232 229 [16] S.K. Huang, S.W. Hui, Fluorescence measurements of fusion [32] J.M. Crowley, Electrical breakdown of bimolecular lipid between human erythrocytes induced by poly(ethylene gly- membranes as an electromechanical instability, Biophys. J. col), Biophys. J 58 (1990) 1109â1117. 13 (1973) 711â724. [17] B.R. Lentz, G.F. Mcintyre, D.J. Parks, J.C. Yates, D. [33] G. Cevc, Polymorphism of the bilayer membranes in the Massenburg, Bilayer curvature and certain amphipaths pro- ordered phase and the molecular origin of the lipid pretransi- mote poly(ethylene glycol)-induced fusion of dipal- tion and rippled lamellae, Biochim. Biophys. Acta 1062 mitoylphosphatidylcholine unilamellar vesicles, Biochemis- (1991) 59â69. try 31 (1992) 2643â2653. [34] L.Y. Song, Q.F. Ahkong, D. Georgescauld, J.A. Lucy, [18] A.R. Viguera, M. Mencia, F.M. Goni, Time-resolved and Membrane fusion without cytoplasmic fusion (hemi-fusion) equilibrium measurements of the effects of poly(ethylene in erythrocytes that are subjected to electrical breakdown, glycol) on small unilamellar phospholipid vesicles, Biochem- Biochim. Biophys. Acta 1065 (1991) 54â62. istry 32 (1993) 3708â3713. [35] J. Wilschut, N. Duzgunes, D. Papahadjopoulos, Calcium/ [19] J. Wilschut, N. Duzgunes, R. Fraley, D. Papahadjopoulos, magnesium specificity in membrane fusion: kinetics of Studies on the mechanism of membrane fusion: kinetics of aggregation and fusion of phosphatidylserine vesicles and the calcium ion induced fusion of phosphatidylserine vesicles role of bilayer curvature, Biochemistry 20 (1981) 3126â followed by a new assay for mixing of aqueous vesicle 3133. contents, Biochemistry 19 (1980) 6011â6021. [36] S. Nir, J. Wilschut, J. Bentz, The rate of fusion of phos- [20] J. Sun, M. Petersheim, Lanthanide(III)-phosphatidic acid pholipid vesicles and the role of bilayer curvature, Biochim. complexes: binding site heterogeneity and phase separation, Biophys. Acta 688 (1982) 275â278. Biochim. Biophys. Acta 1024 (1990) 159â166. [37] W.A. Talbot, L.X. Zheng, B.R. Lentz, Acyl chain unsatura- [21] I.M. Robinson, J.M. Fernandez, The fusion pore interface: a tion and vesicle curvature alter outer leaflet packing and new biological frontier, Curr. Opin. Neurobiol. 4 (1994) promote poly(ethylene glycol) mediated membrane fusion, 330â336. Biochemistry 36 (1997) 5827â5836. [22] S.W. Hui, T.P. Stewart, L.T. Boni, P.L. Yeagle, Membrane [38] A. Relini, D. Cassinadri, Q. Fan, A. Gulik, Z. Mirghani, M. fusion through point defects in bilayers, Science 212 (1981) Derosa, A. gliozzi, Effect of physical constraints on the 921â923. mechanisms of membrane fusion: Bolaform lipid vesicles as model systems, Biophys. J. 71 (1996) 1789â1795.[23] A. Nicolussi, S. Massari, R. Colonna, Effect of lipid mixing on the permeability and fusion of saturated lecithin mem- [39] M.G.L. Elferink, J. van Breemen, W.N. Konings, A.J.M. branes, Biochemistry 21 (1982) 2134â2140. Driessen, J. Wilschut, Slow fusion of liposomes composed of membrane-spanning lipids, Chem. Phys. Lipids 88 (1997)[24] G. Basanez, G.D. Fidelio, F.M. Goni, B. Maggio, A. Alonso, 37â43.Dual inhibitory effect of gangliosides on phospholipase c- promoted fusion of lipidic vesicles, Biochemistry 35 (1996) [40] S. Zellmer, W. Pfeil, J. Lasch, Interaction of phosphatidyl- 7506â7513. choline liposomes with the human stratum corneum, Bio- chim. Biophys. Acta 1237 (1995) 176â182.[25] A. Mori, A.L. Klibanov, V.P. Torchilin, L. Huang, Influence of the steric barrier activity of amphipathic poly- [41] D.R. Kelsey, T.D. Flanagan, J.E. Young, P.L. Yeagle, Inhibi- ethyleneglycol and ganglioside GM1 on the circulation time tion of sendai virus fusion with phospholipid vesicles and of liposomes and on the target binding of immunoliposomes human erythrocyte membranes by hydrophobic peptides, in vivo, FEBS Lett. 284 (1991) 263â266. Virology 182 (1991) 690â702. [26] W.M. Gelbart, R. Bruinsma, Compositional mechanical [42] S. Ohki, A mechanism of divalent ion-induced phospha- instability of interacting mixed lipid membranes, Phys. Rev. tidylserine membrane fusion, Biochim. Biophys. Acta 689 E 55 (1997) 831â835. (1982) 1â11. [27] G. Cevc, J.M. Seddon, R. Hartung, W. Eggert, Phosphatidyl- [43] G. Cevc, Membrane electrostatics, Biochim. Biophys. Acta cholineâfatty acid membranes. I. Effects of protonation, salt 1031 (1990) 311â382. concentration, temperature and chain-length on the colloidal [44] G. Cevc, Colloidal and phase behaviour of biomacromole- and phase properties of mixed vesicles, bilayers and non- cules interdepend and are regulated by the supramolecular lamellar structures, Biochim. Biophys. Acta 940 (1988) 219â surface polarity. examples with lipid bilayer membranes, J. 240. Phys. (France) 50 (1989) 1117â1134. [28] A.L. Bailey, P.R. Cullis, Membrane fusion with cationic [45] G. Blume, G. Cevc, M.D. Crommelin, I.A. Bakker-Wouden- liposomes: Effects of target membrane lipid composition, berg, C. Kluft, G. Storm, Specific targeting with poly- Biochemistry 36 (1997) 1628â1634. (ethylene glycol)-modified liposomes: coupling of homing [29] G. Cevc, Isothermal lipid phase transitions, Chem. Phys. devices to the ends of the polymeric chains combines Lipids 57 (1991) 293â307. effective target binding with long circulation times, Biochim. Biophys. Acta 1149 (1993) 180â184.[30] Frederik et al. (1989), Biochim. Biophys. Acta, 979, 275â 278. [46] G. Cevc, M. Hauser, A.A. Kornyshev, Effects of the interfacial structure on the hydration force between laterally[31] P.M. Frederik, K.N. Burger, M.C. Stuart, A.J. Verkleij, Lipid uniform surfaces, Langmuir 11 (1995) 3103â3110.polymorphism as observed by cryo-electron microscopy, Biochim. Biophys. Acta 1062 (1991) 133â141. [47] G. Cevc, M. Hauser, A.A. Kornyshev, Effects of the 230 G. Cevc, H. Richardsen / Advanced Drug Delivery Reviews 38 (1999) 207 â232 interfacial structure on the hydration force between laterally Choppin, Control of virus-induced cell fusion by host cell lipid composition, Virology 175 (1990) 345â357.non-uniform surfaces, Langmuir 11 (1995) 3111â3118. 1[64] H. Ohno, Y. Maeda, E. Tsuchida, H-NMR study of the effect[48] A. Finkelstein, J. Zimmerberg, F.S. Cohen, Osmotic swelling of synthetic polymers on the fluidity, transition temperatureof vesicles: its role in the fusion of vesicles with planar and fusion of dipalmitoyl phosphatidylcholine small vesicles,phospholipid bilayer membranes and its possible role in Biochim. Biophys. Acta 642 (1981) 27â36.exocytosis, Annu. Rev. Physiol. 48 (1986) 163â174. [65] R. Rand, V. Parsegian, Hydration forces between phos-[49] B. Agostini, M. Stromer, W. Hasselbach, Effect of lipid pholipid bilayers, Biochim. Biophys. Acta 988 (1989) 351âmodification on fusion of sarcoplasmic reticulum vesicles, Z. 376.Naturforsch. C 33 (1978) 428â436. ¨[66] H. Trauble, Membrane electrostatics, in: S. Abrahamsson, I.[50] J.K. Lee, B.R. Lentz, Outer leaflet packing defects promote Pascher (Eds.), Structure of Biological Membranes, Plenumpoly(ethylene glycol) mediated fusion of large unilamellar Press, New York, 1977, pp. 509â550. vesicles, Biochemistry 36 (1997) 421â431. [67] S.A. Tatulian, Ionization and ion binding, in: G. Cevc (Ed.),[51] J.R. Monck, J.M. Fernandez, The exocytotic fusion pore and Phospholipids Handbook, Marcel-Dekker, New York, 1993, neurotransmitter release, Neuron 12 (1994) 707â716. pp. 511â552.[52] B.R. Lentz, J.R. Wu, L.X. Zheng, J. Prevratil, The interfacial [68] G. Cevc, How membrane chain melting properties are region of dipalmitoylphosphatidylcholine bilayers is per- regulated by the polar surface of the lipid bilayer, Biochem- turbed by fusogenic amphipaths, Biophys. J. 71 (1996) istry 26 (1987) 6305â6310. 3302â3310. [69] G. Cevc, J.M. Seddon, D. Marsh, Thermodynamic and[53] S. Ohki, K. Arnold, Surface dielectric constant, surface structural properties of phosphatidylserine bilayer mem- hydrophobicity and membrane fusion, J. Membr. Biol 114 branes in the presence of lithium ions and protons, Biochim.(1990) 195â203. Biophys. Acta 814 (1985) 141â146. [54] B.R. Lentz, Polymer-induced membrane fusion: potential [70] H. Hauser, G.G. Shipley, Crystallization of phosphatidylser- mechanism and relation to cell fusion events, Chem. Phys. ine bilayers induced by lithium, J. Biol. Chem. 256 (1981) Lipids 73 (1994) 91â106. 11377â11380. [55] T.J. Aldwinckle, Q.F. Ahkong, A.D. Bangham, D. Fisher, [71] D. Papahadjopoulos, W.J. Vail, C. Newton, S. Nir, K. J.A. Lucy, Effects of poly(ethylene glycol) on liposomes and Jacobson, G. Poste, R. Lazo, Studies on membrane fusion. erythrocytes. Permeability changes and membrane fusion, III. The role of calcium-induced phase changes, Biochim. Biochim. Biophys. Acta 689 (1982) 548â560. Biophys. Acta 465 (1977) 579â598. [56] T. Ito, M. Yamazaki, S. Ohnishi, Osmoelastic coupling in [72] M. Gratzl, G. Dahl, Fusion of secretory vesicles isolated biological structures: a comprehensive thermodynamic analy- from rat liver, J. Membr. Biol. 40 (1978) 343â364. sis of the osmotic response of phospholipid vesicles and a [73] G. Knoll, C. Braun, H. Plattner, Quenched flow analysis of reevaluation of the âdehydration forceâ theory, Biochemistry exocytosis in paramecium cells: time course, changes in 28 (1989) 5626â5630. membrane structure, and calcium requirements revealed after [57] M. Yamazaki, S. Ohnishi, T. Ito, Osmoelastic coupling in rapid mixing and rapid freezing of intact cells, J. Cell. Biol. biological structures: decrease in membrane fluidity and 113 (1991) 1295â1304. osmophobic association of phospholipid vesicles in response [74] S.A. Summers, B.A. Guebert, M.F. Shanahan, Polyphos- to osmotic stress, Biochemistry 28 (1989) 3710â3715. phoinositide inclusion in artificial lipid bilayer vesicles [58] M. Yamazaki, T. Ito, Deformation and instability in mem- promotes divalent cation dependent membrane fusion, Bio- brane structure of phospholipid vesicles caused by os- phys. J. 71 (1996) 3199â3206. mophobic association: mechanical stress model for the [75] W. Breisblatt, S. Ohki, Fusion in phospholipid spherical mechanism of poly(ethylene glycol)-induced membrane fu- membranes II. Effect of cholesterol, divalent ions and pH, J. sion, Biochemistry 29 (1990) 1309â1314. Membr. Biol. 29 (1976) 127â146. 21 21 21[59] J.M. Robinson, D.S. Roos, R.L. Davidson, M.J. Karnovsky, [76] F. Bellamare, R. Lesage, Mg , Ca and Mn bound to Membrane alterations and other morphological features anionic phospholipids resist desalting dialysis: evaluation of associated with polyethylene glycol-induced cell fusion, J. binding parameters using Stern approximation, J. Colloid Cell. Sci. 40 (1979) 63â75. Interface Sci. 147 (1991) 462â473. [60] S. Knutton, Studies of membrane fusion. III. Fusion of [77] A. Lau, A. McLaughlin, S. McLaughlin, The adsorption of erythrocytes with polyethylene glycol, J. Cell. Sci. 36 (1979) divalent cations to phosphatidylglycerol membranes, Bio- 61â72. chim. Biophys. Acta 557 (1978) 62â78. [61] S.W. Burgess, D. Massenburg, J. Yates, B.R. Lentz, Poly- [78] M. Toner, G. Vaio, A. McLaughlin, S. McLaughlin, Ad- (ethylene glycol)-induced lipid mixing but not fusion be- sorption of divalent cations to bilayer membranes containing tween synthetic phosphatidylcholine large unilamellar vesi- phosphatidylinositol 4,5-biphosphate, Biochemistry 27 cles, Biochemistry 30 (1991) 4193â4200. (1988) 7435â7443. [62] D. Hoekstra, Role of lipid phase separations and membrane [79] S. McLaughlin, N. Mulrine, T. Gresalfi, G.Vaio, A. McLaug- hydration in phospholipid vesicle fusion, Biochemistry 21 hlin, Adsorption of divalent cations to bilayer membranes (1982) 2833â2840. containing phosphatidylserine, L. Gen. Physiol. 77 (1981) 445â473.[63] D.S. Roos, C.S. Duchala, C.B. Stephensen, K.V. Holmes, P.W. G. Cevc, H. Richardsen / Advanced Drug Delivery Reviews 38 (1999) 207 â232 231 [80] H. Grasdalen, L.E.G. Eriksson, J. Westman, A. Ehrenberg, branes. Ethanolâdipalmitoylphosphatidylcholine mixtures, Surface potential effects on metal ion binding to PC mem- Biophys. J. 67 (1994) 1067â1072. 31branes. P-NMR study of lanthanide and calcium binding to [95] N.J. Zuidam, H.K. Gouw, Y. Barenholz, D.J. Crommelin, egg yolk lecithin vesicles, Biochim. Biophys. Acta 469 Physical (in) stability of liposomes upon chemical hydrol- (1977) 151â162. ysis: the role of lysophospholipids and fatty acids, Biochim. Biophys. Acta 1240 (1995) 101â110.[81] A. McLaughlin, C. Grathwohl, S. McLaughlin, The ad- sorption of divalent cations to phosphatidylcholine bilayer [96] S. Rufini, J.Z. Pedersen, A. Desideri, P. Luly, Beta-bun- membranes, Biochim. Biophys. Acta 557 (1978) 62â78. garotoxin- mediated liposome fusion: spectroscopic charac- terization by fluorescence and ESR, Biochemistry 29 (1990)[82] R.L. Misiorowski, M.A. Wells, Competition between cations 9644â9651.and water for binding to PC in organic solvents, Biochemis- try 12 (1973) 967â975. [97] L.V. Chernomordik, E. Leikina, V. Frolov, P. Bronk, J. Zimmerberg, An early stage of membrane fusion mediated[83] H. Oshima, Y. Inoko, T. Mitsui, Hamaker constant and 21 21 by the low pH conformation of influenza hemagglutininbinding constants of Ca and Mg in DPPCâwater depends upon membrane lipids, J. Cell Biol. 136 (1997)systems, J. Colloid Interface Sci. 86 (1982) 57â72. 81â93.[84] S.A. Tatulian, Binding of alkaline-earth metal cations and [98] P. Dunham, P. Babiarz, A. Israel, A. Zerial, G. Weissmann,some anions to PC liposomes, Eur. J. Biochem. 170 (1987) Membrane fusion: studies with a calcium-sensitive dye,413â420. arsenazo III, in liposomes, Proc. Natl. Acad. Sci. USA 74[85] S. Mabrey, J.M. Sturtevant, Incorporation of saturated fatty (1977) 1580â1584.acids into phosphatidylcholine bilayers, Biochim. Biophys. [99] P. Fromherz, Modulation of the edge tension of lipidActa 486 (1977) 444â450. membranes by cholate and cholesterol and the micelleâ[86] H.L. Kantor, J.H. Prestegard, Fusion of phosphatidylcholine vesicle transition in model bile systems, in: W. Swobodnik,bilayer vesicles: Role of free fatty acid, Biochemistry 17 H. Ditschuneit, R.D. Soloway (Eds.), Gallstone Disease.(1978) 3592â3597. Pathophysiology and Therapeutic Approaches, Springer Ver-[87] S. Massari, P. Arslan, A. Nicolussi, R. Colonna, Phos- lag, Berlin, 1990, pp. 27â33.pholipid exchange and size enlargement in sonicated vesicles [100] K.N. Burger, A.J. Verkleij, Membrane fusion, Experientiainduced by a âcriticalâ fatty acid concentration, Biochim. 46 (1990) 631â644.Biophys. Acta 599 (1980) 110â126. [101] L.V. Chernomordik, S.S. Vogel, A. Sokoloff, H.O. Onaran,[88] D. Marsh, J.M. Seddon, Gel-to-inverted hexagonal E.A. Leikina, J. Zimmerberg, Lysolipids reversibly inhibit(L fi H ) phase transitions in phosphatidylethanolaminesb II 21Ca -GTP- and pH-dependent fusion of biological mem-and fatty acidâphosphatidylcholine mixtures, demonstrated 31 branes, FEBS. Lett 318 (1993) 71â76.by P-NMR spectroscopy and X-ray diffraction, Biochim. [102] D.P. Siegel, Energetics of intermediates in membraneBiophys. Acta 690 (1982) 117â123. fusion: comparison of stalk and inverted micellar inter-[89] G. Cevc, J.M. Seddon, R. Hartung, W. Eggert, Properties of mediate mechanisms, Biophys. J. 65 (1993) 2124â2140.phosphatidylhcolineâfatty acid bilayer membranes I. Effects [103] M.M. Kozlov, V.S. Markin, Possible mechanism of mem-of protonation, salt, temperature, and chain length on the brane fusion, Biofizika 28 (1983) 255â261.colloidal and phase behaviour, Biochim. Biophys. Acta 940 [104] H. Parnas, I. Parnas, Neurotransmitter release at fast(1988) 219â240. synapses, J. Membr. Biol. 142 (1994) 267â279.[90] J.M. Seddon, R.H. Templer, N.A. Warrender, Z. Huang, G. [105] A. Morgan, Exocytosis, Essays. Biochem 30 (1995) 77â95.Cevc, D. Marsh, Phosphatidylcholineâfatty acid membranes: effects of headgroup hydration on the phase behaviour and [106] T.H. Sollner, J.E. Rothman, Molecular machinery mediat- structural parameters of the gel and inverse hexagonal (H ) ing vesicle budding, docking and fusion, Experientia 52II phases, Biochim. Biophys. Acta 1327 (1997) 131â147. (1996) 1021â1025. [107] J.E. Rothman, Intracellular membrane fusion, Adv. Second.[91] S. Zellmer, G. Cevc, Thermolabile liposomes with a high Messenger. Phosphoprotein. Res. 29 (1994) 81â96.fusion efficacy at 42 C can be made from dipalmitoylphos- phatidylcholine and fatty alcohols in the molar ratio of 1 /2, [108] H.B. Pollard, E. Rojas, R.W. Pastor, E.M. Rojas, H.R. Guy, J. Liposome Res. 4 (1994) 1091â1114. A.L. Burns, Synexin: molecular mechanism of calcium- dependent membrane fusion and voltage-dependent cal-[92] G. Cevc, Lipid properties as a basis for the modelling and cium-channel activity. Evidence in support of the âhydro-design of liposome membranes, in: G. Gregoriadis (Ed.), phobic bridge hypothesisâ for exocytotic membrane fusion,Liposome Technology, 2nd ed., CRC Press, Boca Raton, Fl, Ann. NY Acad. Sci. 635 (1991) 328â351.1992, pp. 1â36. [109] R.P. Rand, V.A. Parsegian, Mimicry and mechanism in[93] N.E. Nagel, G. Cevc, S. Kirchner, The mechanism of the phospholipid models of membrane fusion, Annu. Rev.solute-induced chain interdigitation in phosphatidylcholine Physiol. 48 (1986) 201â212.vesicles and characterization of the isothermal phase transi- tions by means of dynamic light scattering, Biochim. Bio- [110] P.P. Da Silva, K. Shimizu, C. Parkison, Fusion of human phys. Acta 1111 (1992) 263â269. erythrocytes induced by Sendai virus: Freeze-fracture as- pects, J. Cell. Sci. 43 (1980) 419â432.¨[94] U. Vierl, L. Lobbecke, N. Nagel, G. Cevc, Solute effects on the colloidal and phase behaviour of lipid bilayer mem- [111] S. Knutton, Studies of membrane fusion. VI. Mechanism of 232 G. Cevc, H. Richardsen / Advanced Drug Delivery Reviews 38 (1999) 207 â232 the membrane fusion and cell swelling stages of Sendai glutinin-induced hemifusion to full fusion, J. Cell. Biol. 136 virus-mediated cell fusion, J. Cell. Sci. 43 (1980) 103â118. (1997) 995â1005. [112] J.R. Monck, G. Alvarez de Toledo, J.M. Fernandez, Ten- [121] L. Chernomordik, A. Chanturiya, J. Green, J. Zimmerberg, sion in secretory granule membranes causes extensive The hemi-fusion intermediate and its conversion to com- membrane transfer through the exocytotic fusion pore, Proc. plete fusion: regulation by membrane composition, Bio- Natl. Acad. Sci. USA 87 (1990) 7804â7808. phys. J 69 (1995) 922â929. [113] R. Blumenthal, C.C. Pak, Y. Raviv, M. Krumbiegel, L.D. [122] E.L. Bearer, N. Duzgunes, D.S. Friend, D. Papahad- Bergelson, S.J. Morris, R.J. Lowy, Transient domains jopoulos, Fusion of phospholipid vesicles arrested by quick- induced by influenza haemagglutinin during membrane freezing. the question of lipidic particles as intermediates in fusion, Mol. Membr. Biol. 12 (1995) 135â142. membrane fusion, Biochim. Biophys. Acta 693 (1982) [114] R. Schubert, K. Beyer, H. Wolburg, K.-H. Schmidt, Struc- 93â98. tural changes in membranes of large unilamellar vesicles [123] D.C. Miller, G.P. Dahl, Early events in calcium-induced after binding of sodium cholate, Biochemistry 25 (1986) liposome fusion, Biochim. Biophys. Acta 689 (1982) 165â 5263â5269. 169. [115] K. Hong, N. Duzgunes, R. Ekerdt, D. Papahadjopoulos, [124] A.E. Spruce, L.J. Breckenridge, A.K. Lee, W. Almers, Synexin facilitates fusion of specific phospholipid mem- Properties of the fusion pore that forms during exocytosis branes at divalent cation concentrations found intracellular- of a mast cell secretory vesicle, Neuron 4 (1990) 643â654. ly, Proc. Natl. Acad. Sci. USA 79 (1982) 4642â4644. [125] J.R. Monck, A.F. Oberhauser, G. Alvarez de Toledo, J.M.[116] H. Tournois, C.H. Fabrie, K.N. Burger, J. Mandersloot, P. Fernandez, Is swelling of the secretory granule matrix theHilgers, H. van Dalen, J. de Gier, B. de Kruijff, Gramicidin force that dilates the exocytotic fusion pore?, Biophys. J 59A induced fusion of large unilamellar dioleoylphos- (1991) 39â47.phatidylcholine vesicles and its relation to the induction of [126] J. Zimmerberg, M. Curran, F.S. Cohen, A lipid /proteintype II nonbilayer structures, Biochemistry 29 (1990) complex hypothesis for exocytotic fusion pore formation,8297â8307. Ann. NY Acad. Sci 635 (1991) 307â317.[117] A.L. Bailey, M.A. Monck, P.R. Cullis, Ph induced destabili- [127] A.E. Spruce, A. Iwata, W. Almers, The first milliseconds ofzation of lipid bilayers by a lipopeptide derived from the pore formed by a fusogenic viral envelope proteininfluenza hemagglutinin, Biochim. Biophys. Acta 1324 during membrane fusion, Proc. Natl. Acad. Sci. USA 88(1997) 232â244. (1991) 3623â3627.[118] K.N. Burger, J.L. Nieva, A. Alonso, A.J. Verkleij, Phos- [128] J.R. Monck, A.F. Oberhauser, J.M. Fernandez, The ex-pholipase C activity-induced fusion of pure lipid model ocytotic fusion pore interface: a model of the site ofmembranes, A freeze fracture study., Biochim. Biophys. neurotransmitter release, Mol. Membr. Biol. 12 (1995)Acta 1068 (1991) 249â253. 151â156.[119] L.V. Chernomordik, J. Zimmerberg, Bending membranes to [129] L. Chernomordik, Non-bilayer lipids and biological fusionthe task: structural intermediates in bilayer fusion, Curr. intermediates, Chem. Phys. Lipids 81 (1996) 203â213.Opin. Struct. Biol. 5 (1995) 541â547. [120] G.B. Melikyan, S.A. Brener, D.C. Ok, F.S. Cohen, Inner [130] J.M. Seddon, Structure of the inverted hexagonal (H )II but not outer membrane leaflets control the transition from phase, and non-lamellar phase transitions of lipids, Bio- glycosylphosphatidylinositol-anchored influenza hemag- chim. Biophys. Acta 1031 (1990) 1â69.