Influenza hemagglutinin-mediated fusion pores connecting cells to planar membranes: flickering to final expansion

We have studied the fusion between voltage-clamped planar lipid bilayers and influenza virus infected MDCK cells, adhered to one side of the bilayer, using measurements of electrical admittance and fluorescence. The changes in currents in-phase and 90 degrees out-of- phase with respect to the applied sinusoidal voltage were used to monitor the addition of the cell membrane capacitance to that of the lipid bilayer through a fusion pore connecting the two membranes. When ethidium bromide was included in the solution of the cell-free side of the bilayer, increases in cell fluorescence accompanied tee admittance changes, independently confirming that these changes were due to formation of a fusion pore. Fusion required acidic pH on the cell- containing side and depended on temperature. For fusion to occur, the influenza hemagglutinin (HA) had to be cleaved into HA1 and HA2 subunits. The incorporation of gangliosides into the planar bilayers greatly augmented fusion. Fusion pores developed in four distinct stages after acidification: (a) a pre-pore, electrically quiescent stage; (b) a flickering stage, with 1-2 nS pores opening and closing repetitively; (c) an irreversibly opened stage, in which pore conductances varied between 2 and 100 nS and exhibited diverse kinetics; (d) a fully opened stage, initiated by an instantaneous, time- resolution limited, increase in conductance leveling at approximately 500 nS. The expansion of pores by stages has also been shown to occur during exocytosis in mast cells and fusion of HA-expressing cells and erythrocytes. We conclude that essential features of fusion pores are produced with proteins in just one of the two fusing membranes.     

Cholesterol stabilizes recombinant exocytic fusion pores by altering membrane bending rigidity

SNARE-mediated membrane fusion proceeds via the formation of a fusion pore. This intermediate structure is highly dynamic and can flicker between open and closed states. In cells, cholesterol has been reported to affect SNARE-mediated exocytosis and fusion pore dynamics. Here, we address the question of whether cholesterol directly affects the flickering rate of reconstituted fusion pores in vitro. These experiments were enabled by the recent development of a nanodisc⋅black lipid membrane recording system that monitors dynamic transitions between the open and closed states of nascent recombinant pores with submillisecond time resolution. The fusion pores formed between nanodiscs that bore the vesicular SNARE synaptobrevin 2 and black lipid membranes that harbored the target membrane SNAREs syntaxin 1A and SNAP-25B were markedly affected by cholesterol. These effects include strong reductions in flickering out of the open state, resulting in a significant increase in the open dwell-time. We attributed these effects to the known role of cholesterol in altering the elastic properties of lipid bilayers because manipulation of phospholipids to increase membrane stiffness mirrored the effects of cholesterol. In contrast to the observed effects on pore kinetics, cholesterol had no effect on the current that passed through individual pores and, hence, did not affect pore size. In conclusion, our results show that cholesterol dramatically stabilizes fusion pores in the open state by increasing membrane bending rigidity.     

Influenza virus hemagglutinin-induced cell-planar bilayer fusion: quantitative dissection of fusion pore kinetics into stages

Cells expressing the influenza virus hemagglutinin (HA) fuse to planar bilayer membranes under acidic conditions. After an electrically quiescent perfusion stage (Q), a fusion pore forms that enlarges in three subsequent stages. A repetitively flickering pore stage (R) develops into a securely open stage (S) that exhibits conductances ranging from a few to tens of nS. The pore then expands to a terminal stage (T) with a large conductance on the order of one microSiemens. We have studied how virus strain, HA receptors in the target bilayer membrane, and cytoskeleton affect the time a fusion pore remains in each stage. These intervals are referred to as waiting times. In the quiescent stage, waiting times were very sensitive to the virus strain and presence of gangliosides (HA receptors) in the bilayer. When bilayers contained gangliosides, the waiting times in the Q stage for cells infected with the PR/8/34 strain of virus were exponentially distributed, whereas waiting times for cells infected with the Japan/305/57 strain were not so distributed. Without gangliosides, the waiting time distribution for PR/8/34 infected cells was complex. The waiting times in the R and S stages of pore growth were exponentially distributed under all tested conditions. Within the R stage, we analyzed the kinetics of the flickering pore by fitting the open and closed time distributions with a sum of two exponentials. Neither the open and closed time distributions nor the flickering pore conductance distributions were appreciably affected by virus strain or gangliosides. Colchicine and cytochalasin B increased the flicker rates, without affecting the waiting time in the R stage. We conclude that variations in amino acid sequences of the HAs and the presence of gangliosides as receptors within the target membrane critically affect the kinetics of fusion pore formation, but do not affect subsequent stages.     

Comparison of transient and successful fusion pores connecting influenza hemagglutinin expressing cells to planar membranes

Time-resolved admittance measurements were used to investigate the evolution of fusion pores formed between cells expressing influenza virus hemagglutinin (HA) and planar bilayer membranes. The majority of fusion pores opened in a stepwise fashion to semistable conductance levels of several nS. About 20% of the pores had measurable rise times to nS conductances; some of these opened to conductances of approximately 500 pS where they briefly lingered before opening further to semistable conductances. The fall times of closing were statistically similar to the rise times of opening. All fusion pores exhibited semistable values of conductance, varying from approximately 2-20 nS; they would then either close or fully open to conductances on the order of 1 microS. The majority of pores closed; approximately 10% fully opened. Once within the semistable stage, all fusion pores, even those that eventually closed, tended to grow. Statistically, however, before closing, transient fusion pores ceased to grow and reversed their conductance pattern: conductances decreased with a measurable time course until a final drop to closure. In contrast, pore enlargement to the fully open state tended to occur from the largest conductance values attained during a pore's semistable stage. This final enlargement was characterized by a stepwise increase in conductance. The density of HA on the cell surface did not strongly affect pore dynamics. But increased proteolytic treatment of cell surfaces did lead to faster growth within the semistable range. Transient pores and pores that fully opened had indistinguishable initial conductances and statistically identical time courses of early growth, suggesting they were the same upon formation. We suggest that transient and fully open pores evolved from common structures with stochastic factors determining their fate.     

Multiple local contact sites are induced by GPI-linked influenza hemagglutinin during hemifusion and flickering pore formation

Membrane fusion intermediates induced by the glycosylphosphatidylinositol-linked ectodomain of influenza hemagglutinin (GPI-HA) were investigated by rapid freeze, freeze-substitution, thin section electron microscopy, and with simultaneous recordings of whole-cell admittance and fluorescence. Upon triggering, the previously separated membranes developed numerous hourglass shaped points of membrane contact (∼10–130 nm waist) when viewed by electron microscopy. Stereo pairs showed close membrane contact at peaks of complementary protrusions, arising from each membrane. With HA, there were fewer contacts, but wide fusion pores. Physiological measurements showed fast lipid dye mixing between cells after acidification, and either fusion pore formation or the lack thereof (true hemifusion). For the earliest pores, a similar conductance distribution and frequency of flickering pores were detected for both HA and GPI-HA. For GPI-HA, lipid mixing was detected prior to, during, or after pore opening, whereas for HA, lipid mixing is seen only after pore opening. Our findings are consistent with a pathway wherein conformational changes in the ectodomain of HA pull membranes towards each other to form a contact site, then hemifusion and pore formation initiate in a small percentage of these contact sites. Finally, the transmembrane domain of HA is needed to complete membrane fusion for macromolecular content mixing.     

Restricted movement of lipid and aqueous dyes through pores formed by influenza hemagglutinin during cell fusion

The fusion of cells by influenza hemagglutinin (HA) is the best characterized example of protein-mediated membrane fusion. In simultaneous measurements of pairs of assays for fusion, we determined the order of detectable events during fusion. Fusion pore formation in HA-triggered cell-cell fusion was first detected by changes in cell membrane capacitance, next by a flux of fluorescent lipid, and finally by flux of aqueous fluorescent dye. Fusion pore conductance increased by small steps. A retardation of lipid and aqueous dyes occurred during fusion pore fluctuations. The flux of aqueous dye depended on the size of the molecule. The lack of movement of aqueous dyes while total fusion pore conductance increased suggests that initial HA-triggered fusion events are characterized by the opening of multiple small pores: the formation of a "sieve".     

Initial size and dynamics of viral fusion pores are a function of the fusion protein mediating membrane fusion

Background information. Protein‐mediated merger of biological membranes, membrane fusion, is an important process. To investigate the role of fusogenic proteins in the initial size and dynamics of the fusion pore (a narrow aqueous pathway, which widens to finalize membrane fusion), two different fusion proteins expressed in the same cell line were investigated: the major glycoprotein of baculovirus Autographa californica (GP64) and the HA (haemagglutinin) of influenza X31. Results. The host Sf9 cells expressing these viral proteins, irrespective of protein species, fused to human RBCs (red blood cells) upon acidification of the medium. A high‐time‐resolution electrophysiological study of fusion pore conductance revealed fundamental differences in (i) the initial pore conductance; pores created by HA were smaller than those created by GP64; (ii) the ability of pores to flicker; only HA‐mediated pores flickered; and (iii) the time required for pore formation; HA‐mediated pores took much longer to form after acidification. Conclusion. HA and GP64 have divergent electrophysiological phenotypes even when they fuse identical membranes, and fusion proteins play a crucial role in determining initial fusion pore characteristics. The structure of the initial fusion pore detected by electrical conductance measurements is sensitive to the nature of the fusion protein.     

The lipid-anchored ectodomain of influenza virus hemagglutinin (GPI-HA) is capable of inducing nonenlarging fusion pores

GPI-linked hemagglutinin (GPI-HA) of influenza virus was thought to induce hemifusion without pore formation. Cells expressing either HA or GPI-HA were bound to red blood cells, and their fusion was compared by patch-clamp capacitance measurements and fluorescence microscopy. It is now shown that under more optimal fusion conditions than have been used previously, GPI-HA is also able to induce fusion pore formation before lipid dye spread, although with fewer pores formed than those induced by HA. The GPI-HA pores did not enlarge substantially, as determined by the inability of a small aqueous dye to pass through them. The presence of 1,1'-dioctadecyl-3, 3,3',3'-tetramethylindocarbocyanine perchlorate or octadecylrhodamine B in red blood cells significantly increased the probability of pore formation by GPI-HA; the dyes affected pore formation to a much lesser degree for HA. This greater sensitivity of pore formation to lipid composition suggests that lipids are a more abundant component of a GPI-HA fusion pore than of an HA pore. The finding that GPI-HA can induce pores indicates that the ectodomain of HA is responsible for all steps up to the initial membrane merger and that the transmembrane domain, although not absolutely required, ensures reliable pore formation and is essential for pore growth. GPI-HA is the minimal unit identified to date that supports fusion to the point of pore formation.     

The initial fusion pore induced by baculovirus GP64 is large and forms quickly

The formation of the fusion pore is the first detectable event in membrane fusion (Zimmerberg, J., R. Blumenthal, D.P. Sarkar, M. Curran, and S.J. Morris. 1994. J. Cell Biol. 127:1885-1894). To date, fusion pores measured in exocytosis and viral fusion have shared features that include reversible closure (flickering), highly fluctuating semistable stages, and a lag time of at least several seconds between the triggering and the pore opening. We investigated baculovirus GP64- induced Sf9 cell-cell fusion, triggered by external acid solution, using two different electrophysiological techniques: double whole-cell recording (for high time resolution, model-independent measurements), and the more conventional time-resolved admittance recordings. Both methods gave essentially the same results, thus validating the use of the admittance measurements for fusion pore conductance calculations. Fusion was first detected by abrupt pore formation with a wide distribution of initial conductance, centered around 1 nS. Often the initial fusion pore conductance was stable for many seconds. Fluctuations in semistable conductances were much less than those of other fusion pores. The waiting time distribution, measured between pH onset and initial pore appearance, fits best to a model with many (approximately 19) independent elements. Thus, unlike previously measured fusion pores, GP64-mediated pores do not flicker, can have large, stable initial pore conductances lasting up to a minute, and have typical lag times of < 1 s. These findings are consistent with a barrel-shaped model of an initial fusion pore consisting of five to eight GP64 trimers that is lined with lipid.     

Amino Acid Sequence Requirements of the Transmembrane and Cytoplasmic Domains of Influenza Virus Hemagglutinin for Viable Membrane Fusion

The amino acid sequence requirements of the transmembrane (TM) domain and cytoplasmic tail (CT) of the hemagglutinin (HA) of influenza virus in membrane fusion have been investigated. Fusion properties of wild-type HA were compared with those of chimeras consisting of the ectodomain of HA and the TM domain and/or CT of polyimmunoglobulin receptor, a nonviral integral membrane protein. The presence of a CT was not required for fusion. But when a TM domain and CT were present, fusion activity was greater when they were derived from the same protein than derived from different proteins. In fact, the chimera with a TM domain of HA and truncated CT of polyimmunoglobulin receptor did not support full fusion, indicating that the two regions are not functionally independent. Despite the fact that there is wide latitude in the sequence of the TM domain that supports fusion, a point mutation of a semiconserved residue within the TM domain of HA inhibited fusion. The ability of a foreign TM domain to support fusion contradicts the hypothesis that a pore is composed solely of fusion proteins and supports the theory that the TM domain creates fusion pores after a stage of hemifusion has been achieved.     

Hemifusion between cells expressing hemagglutinin of influenza virus and planar membranes can precede the formation of fusion pores that subsequently fully enlarge

The chronological relation between the establishment of lipid continuity and fusion pore formation has been investigated for fusion of cells expressing hemagglutinin (HA) of influenza virus to planar bilayer membranes. Self-quenching concentrations of lipid dye were placed in the planar membrane to monitor lipid mixing, and time-resolved admittance measurements were used to measure fusion pores. For rhodamine-PE, fusion pores always occurred before a detectable amount of dye moved into an HA-expressing cell. However, with DiI in the planar membrane, the relationship was reversed: the spread of dye preceded formation of small pores. In other words, by using DiI as probe, hemifusion was clearly observed to occur before pore formation. For hemifused cells, a small pore could form and subsequently fully enlarge. In contrast, for cells that express a glycosylphosphatidylinositol-anchored ectodomain of HA, hemifusion occurred, but no fully enlarged pores were observed. Therefore, the transmembrane domain of HA is required for the formation of fully enlarging pores. Thus, with the planar bilayer membranes as target, hemifusion can precede pore formation, and the occurrence of lipid dye spread does not preclude formation of pores that can enlarge fully.     

Reversible merger of membranes at the early stage of influenza hemagglutinin-mediated fusion

Fusion mediated by influenza hemagglutinin (HA), a prototype fusion protein, is commonly detected as lipid and content mixing between fusing cells. Decreasing the surface density of fusion-competent HA inhibited these advanced fusion phenotypes and allowed us to identify an early stage of fusion at physiological temperature. Although lipid flow between membranes was restricted, the contacting membrane monolayers were apparently transiently connected, as detected by the transformation of this fusion intermediate into complete fusion after treatments known to destabilize hemifusion diaphragms. These reversible connections disappeared within 10-20 min after application of low pH, indicating that after the energy released by HA refolding dissipated, the final low pH conformation of HA did not support membrane merger. Although the dynamic character and the lack of lipid mixing at 37 degrees C distinguish the newly identified fusion intermediate from the intermediate arrested at 4 degrees C described previously, both intermediates apparently belong to the same family of restricted hemifusion (RH) structures. Because the formation of transient RH structures at physiological temperatures was as fast as fusion pore opening and required less HA, we hypothesize that fusion starts with the formation of multiple RH sites, only a few of which then evolve to become expanding fusion pores.     

Effects of membrane potential and sphingolipid structures on fusion of Semliki Forest virus

Cells expressing the E1 and E2 envelope proteins of Semliki Forest virus (SFV) were fused to voltage-clamped planar lipid bilayer membranes at low pH. Formation and evolution of fusion pores were electrically monitored by capacitance measurements, and membrane continuity was tracked by video fluorescence microscopy by including rhodamine-phosphatidylethanolamine in the bilayer. Fusion occurred without leakage for a negative potential applied to the trans side of the planar membrane. When a positive potential was applied, leakage was severe, obscuring the observation of any fusion. E1-mediated cell-cell fusion occurred without leakage for negative intracellular potentials but with substantial leakage for zero membrane potential. Thus, negative membrane potentials are generally required for nonleaky fusion. With planar bilayers as the target, the first fusion pore that formed almost always enlarged; pore flickering was a rare event. Similar to other target membranes, fusion required cholesterol and sphingolipids in the planar membrane. Sphingosine did not support fusion, but both ceramide, with even a minimal acyl chain (C(2)-ceramide), and lysosphingomyelin (lyso-SM) promoted fusion with the same kinetics. Thus, unrelated modifications to different parts of sphingosine yielded sphingolipids that supported fusion to the same degree. Fusion studies of pyrene-labeled SFV with cholesterol-containing liposomes showed that C(2)-ceramide supported fusion while lyso-SM did not, apparently due to its positive curvature effects. A model is proposed in which the hydroxyls of C-1 and C-3 as well as N of C-2 of the sphingosine backbone must orient so as to form multiple hydrogen bonds to amino acids of SFV E1 for fusion to proceed.     

Reconstituted fusion pore

Fusion pores or porosomes are basket-like structures at the cell plasma membrane, at the base of which, membrane-bound secretory vesicles dock and fuse to release vesicular contents. Earlier studies using atomic force microscopy (AFM) demonstrated the presence of fusion pores at the cell plasma membrane in a number of live secretory cells, revealing their morphology and dynamics at nm resolution and in real time. ImmunoAFM studies demonstrated the release of vesicular contents through the pores. Transmission electron microscopy (TEM) further confirmed the presence of fusion pores, and immunoAFM, and immunochemical studies demonstrated t-SNAREs to localize at the base of the fusion pore. In the present study, the morphology, function, and composition of the immunoisolated fusion pore was investigated. TEM studies reveal in further detail the structure of the fusion pore. Immunoblot analysis of the immunoisolated fusion pore reveals the presence of several isoforms of the proteins, identified earlier in addition to the association of chloride channels. TEM and AFM micrographs of the immunoisolated fusion pore complex were superimposable, revealing its detail structure. Fusion pore reconstituted into liposomes and examined by TEM, revealed a cup-shaped basket-like morphology, and were functional, as demonstrated by their ability to fuse with isolated secretory vesicles.     

Effects of Spontaneous Bilayer Curvature on Influenza Virus–mediated Fusion Pores

Cells expressing the hemagglutinin protein of influenza virus were fused to planar bilayer membranes containing the fluorescent lipid probes octadecylrhodamine (R18) or indocarbocyanine (DiI) to investigate whether spontaneous curvature of each monolayer of a target membrane affects the growth of fusion pores. R18 and DiI lowered the transition temperatures for formation of an inverted hexagonal phase, indicating that these probes facilitate the formation of negative curvature structures. The probes are known to translocate from one monolayer of a bilayer membrane to the other in a voltage-dependent manner. The spontaneous curvature of the cis monolayer (facing the cells) or the trans monolayer could therefore be made more negative through control of the polarity of voltage across the planar membrane. Electrical admittance measurements showed that the open times of flickering fusion pores were shorter when probes were in trans monolayers and longer when in cis monolayers compared with times when probe was symmetrically distributed. Open times were the same for probe symmetrically distributed as when probes were not present. Thus, open times were a function of the asymmetry of the spontaneous curvature between the trans and cis monolayers. Enriching the cis monolayer with a negative curvature probe reduced the probability that a small pore would fully enlarge, whereas enriching the trans monolayer promoted enlargement. Lysophosphatidylcholine has positive spontaneous curvature and does not translocate. When lysophosphatidylcholine was placed in trans leaflets of planar membranes, closing of fusion pores was rare. The effects of the negative and positive spontaneous curvature probes do not support the hypothesis that a flickering pore closes from an open state within a hemifusion diaphragm (essentially a “flat” structure). Rather, such effects support the hypothesis that the membrane surrounding the open pore forms a three-dimensional hourglass shape from which the pore flickers shut.     

Membrane permeability changes at early stages of influenza hemagglutinin-mediated fusion

While biological membrane fusion is classically defined as the leak-free merger of membranes and contents, leakage is a finding in both experimental and theoretical studies. The fusion stages, if any, that allow membrane permeation are uncharted. In this study we monitored membrane ionic permeability at early stages of fusion mediated by the fusogenic protein influenza hemagglutinin (HA). HAb2 cells, expressing HA on their plasma membrane, fused with human red blood cells, cultured liver cells PLC/PRF/5, or planar phospholipid bilayer membranes. With a probability that depended upon the target membrane, an increase of the electrical conductance of the fusing membranes (leakage) by up to several nS was generally detected. This leakage was recorded at the initial stages of fusion, when fusion pores formed. This leakage usually accompanied the "flickering" stage of the early fusion pore development. As the pore widened, the leakage reduced; concomitantly, the lipid exchange between the fusing membranes accelerated. We conclude that during fusion pore formation, HA locally and temporarily increases the permeability of fusing membranes. Subsequent rearrangement in the fusion complex leads to the resealing of the leaky membranes and enlargement of the pore.     

Release of small transmitters through kiss-and-run fusion pores in rat pancreatic beta cells

Exocytosis of secretory vesicles begins with a fusion pore connecting the vesicle lumen to the extracellular space. This pore may then expand or it may close to recapture the vesicle intact. The contribution of the latter, termed kiss-and-run, to exocytosis of pancreatic beta cell large dense-core vesicles (LDCVs) is controversial. Examination of single vesicle fusion pores demonstrated that rat beta cell LDCVs can undergo exocytosis by rapid pore expansion, by the formation of stable pores, or via small transient kiss-and-run fusion pores. Elevation of cAMP shifted LDCV fusion pore openings to the transient mode. Under this condition, the small fusion pores were sufficient for release of ATP, stored within LDCVs together with insulin. Individual ATP release events occurred coincident with amperometric "stand alone feet" representing kiss-and-run. Therefore, the LDCV kiss-and-run fusion pores allow small transmitter release but likely retain the larger insulin peptide. This may represent a mechanism for selective intraislet signaling.     

The Pathway of Membrane Fusion Catalyzed by Influenza Hemagglutinin: Restriction of Lipids, Hemifusion, and Lipidic Fusion Pore Formation

The mechanism of bilayer unification in biological fusion is unclear. We reversibly arrested hemagglutinin (HA)-mediated cell–cell fusion right before fusion pore opening. A low-pH conformation of HA was required to form this intermediate and to ensure fusion beyond it. We present evidence indicating that outer monolayers of the fusing membranes were merged and continuous in this intermediate, but HA restricted lipid mixing. Depending on the surface density of HA and the membrane lipid composition, this restricted hemifusion intermediate either transformed into a fusion pore or expanded into an unrestricted hemifusion, without pores but with unrestricted lipid mixing. Our results suggest that restriction of lipid flux by a ring of activated HA is necessary for successful fusion, during which a lipidic fusion pore develops in a local and transient hemifusion diaphragm.     

Sterols and sphingolipids strongly affect the growth of fusion pores induced by the hemagglutinin of influenza virus

Cells expressing the hemagglutinin (HA) of influenza virus were fused to planar phospholipid bilayer membranes to evaluate the effects of sterols and sphingolipids in the target bilayer membranes on properties of fusion pores. Typically, in the absence of sterol, flickering pores are observed, followed by a successful pore (i.e., a pore that fully opens). The incorporation of cholesterol into the lipid bilayer had a marked effect: it greatly decreased the number of flickers, and the first pore formed was usually successful. Similar effects were produced by the sterols epicholesterol and 5beta-cholestanol. In contrast, the sterols cholesteryl acetate, coprostanol, and stanolone did not affect pore flickering, and a successful pore was observed to follow the typical number of flickers. 5alpha-cholestanol gave intermediate results. From these results, it follows that the 3-OH of cholesterol is essential to reduce flickering, but it does not matter if the 3-OH is in an alpha or beta configuration. The double bond is also not critical for the actions of cholesterol nor is the fact that it is a flat molecule. The sphingolipids sphingomyelin, lactosyl cerebroside, and glucosyl cerebroside tended to inhibit full pore enlargement, prolonging the stage of pore flickering. If a sphingolipid and a sterol that strongly interact were both included in the planar membrane, the pattern of flickering was the same as if neither had been included in the bilayer. However, if a sphingolipid and sterol that do not interact with each other were included in the bilayer, the reduced flickering characteristic of the sterol was observed.     

Membrane perturbation and fusion pore formation in influenza hemagglutinin-mediated membrane fusion. A new model for fusion

Low pH-induced fusion mediated by the hemagglutinin (HA) of influenza virus involves conformational changes in the protein that lead to the insertion of a "fusion peptide" domain of this protein into the target membrane and is thought to perturb the membrane, triggering fusion. By using whole virus, purified HA, or HA ectodomains, we found that shortly after insertion, pores of less than 26 A in diameter were formed in liposomal membranes. As measured by a novel assay, these pores stay open, or continue to close and open, for minutes to hours and persist after pH neutralization. With virus and purified HA, larger pores, allowing the leakage of dextrans, were seen at times well after insertion. For virus, dextran leakage was simultaneous with lipid mixing and the formation of "fusion pores," allowing the transfer of dextrans from the liposomal to the viral interior or vice versa. Pores did not form in the viral membrane in the absence of a target membrane. Based on these data, we propose a new model for fusion, in which HA initially forms a proteinaceous pore in the target, but not in the viral membrane, before a lipidic hemifusion intermediate is formed.