Fusion Peptides Derived from the HIV Type 1 Glycoprotein 41 Associate within Phospholipid Membranes and Inhibit Cell-Cell Fusion STRUCTURE-FUNCTION

The fusion domain of human immunodeficiency virus (HIV-1) envelope glycoprotein (gp120-gp41) is a conserved hydrophobic region located at the N terminus of the transmembrane glycoprotein (gp41). A V2E mutant has been shown to dominantly interfere with wild-type envelope-mediated syncytium formation and virus infectivity. To understand this phenomenon, a 33-residue peptide (wild type, WT) identical to the N-terminal seg-ment of gp41 and its V2E mutant were synthesized, fluo- rescently labeled, and characterized. Both peptides inhibited HIV-1 envelope-mediated cell-cell fusion and had similar (cid:97) -helical content in membrane mimetic en-vironments. Studies with fluorescently labeled peptide analogues revealed that both peptides have high affinity for phospholipid membranes, are susceptible to diges- tion by proteinase-K in their membrane-bound state, and tend to self- and coassemble in the membranes. In SDS-polyacrylamide gel electrophoresis the WT peptide formed dimers as well as higher order oligomers, whereas the V2E mutant only formed dimers. The WT, but not the V2E mutant, induced liposome aggregation, destabilization, and fusion. Moreover, the V2E mutant inhibited vesicle fusion induced by the WT peptide, probably by forming inactive heteroaggregates. These data form the basis for an explanation of the mechanism by which the gp41 V2E

To infect mammalian cells, enveloped viruses have to deposit their nucleocapsids into the cytoplasm of a host cell. Membrane fusion represents a key element in this entry mechanism. The basic unit of most viral fusion proteins is one or two type 1 integral membrane glycoproteins. These often combine into oligomers that project from the viral envelope (1). A key feature of most viral fusion proteins is a fusion peptide, a stretch of highly hydrophobic amino acids that is believed to trigger the fusion process. It was proposed that fusion is initiated by the insertion of the fusion peptide into either the target membrane (2)(3)(4)(5)(6), the viral membrane (7,8), or both (9 -11).
With HIV, 1 fusion between the virus and CD4 ϩ cells is a critical step in the viral infection. The fusion peptide located at the N terminus of the transmembrane viral protein gp41 is assumed to have a major role in this fusion process (12). A polar amino acid substitution at position 2 of the fusion peptide of gp41 (V2E) results in an envelope glycoprotein that dominantly interferes with both syncytium formation and infection mediated by the wild-type HIV-1 envelope glycoprotein (13,14). Although the mechanism of the interference by the V2E mutant is not clear, it does not result from aberrant envelope glycoprotein synthesis, processing, or transport. This mutant elicits a dominant interfering effect even in the presence of excess wild-type glycoprotein, which suggests that a higher order envelope glycoprotein complex is involved in the membrane fusion (13).
The fusion process is a complex phenomenon that involves an entire range of biochemical and biophysical interactions (15). In an attempt to understand the steps involved in the fusion process, synthetic peptides that resemble or mimic the putative fusion peptide's regions of envelope proteins of viruses, or model peptides, have been synthesized, and their interactions with liposomes or cells have been examined. Among them there are peptides corresponding to the fusion sequences of influenza virus (8, 16 -18), Sendai virus (19), simian immunodeficiency virus (20), and HIV (21)(22)(23).
Despite extensive studies, the mode of action of fusion peptides that promote fusion is still not clear. Interaction of viral fusion peptides with host-cell membranes was simulated by computer analysis (24), which led to the conclusion that the fusogenic helices were obliquely oriented with respect to the lipid-water interface. This conclusion was experimentally supported by the finding that a mutation that modified the oblique orientation of the fusion peptide of simian immunodeficiency virus gp32 reduced the peptide's fusogenic activity (25). Further support for this oblique orientation comes from a recent study (19) showing that the fusion peptide of the Sendai virus is obliquely oriented in its membrane-bound state. Furthermore, the peptide could self-assemble in its membrane-bound state, thus suggesting its role in assisting in the assembly of the envelope protein of the virus. More recently, Martin et al. (26) showed, using attenuated total reflection fourier transform infrared spectroscopy, that a short portion of the fusion peptide of HIV is obliquely oriented in its membrane-bound state. This oblique orientation was postulated to locally disorganize the structure of the lipid bilayer and to generate new lipid phases that are thought to be associated with the initial steps of membrane fusion.
To understand better the role of fusion peptides in the mediation of cell fusion, a 33-residue peptide that represents the N terminus of HIV-1 gp41, and its V2E mutant were synthesized and fluorescently labeled. The peptides were then used in a variety of biophysical and functional studies to characterize their structure, their abilities to interact with and permeate phospholipid membranes, to self-and coassemble within membranes, to induce fusion of phospholipid membranes, and to inhibit cell-cell fusion. The results revealed that the fusogenic properties of the intact gp41 and its V2E mutant can be modeled using synthetic peptides. Thus, this synthetic peptide model was used to explore the mechanism in which the V2E mutant interferes with the fusion peptide activity.
Peptide Synthesis and Fluorescent Labeling-The peptides were synthesized by a solid phase method on phenylacetamidomethyl-amino acid resin (0.15 milliequivalents) (27), as described previously (28,29). Labeling of the N terminus of the peptides was achieved as described previously (30,31); briefly, resin-bound peptides, with their amino acid side chains fully protected, were treated with trifluoroacetic acid, to remove the BOC protecting group from their N-terminal amino groups, while keeping all the other reactive amine groups of the attached peptides still protected. When needed the resin-bound peptides were reacted with the desired fluorescent probe and finally cleaved from the resins by hydrofluoric acid, extracted with trifluoroacetic acid, and precipitated with ether. This procedure yielded peptides selectively labeled with fluorescent probes at their N-terminal amino acid. The synthetic peptides were purified (Ͼ95% homogenicity) by reverse-phase high performance liquid chromatography on a C18 column using a linear gradient of 25-80% acetonitrile in 0.05% trifluoroacetic acid, for 40 min. The peptides were subjected to amino acid analysis and mass spectrometry to confirm their composition. Unless stated elsewhere, stock solutions of concentrated peptides in Me 2 SO were used to avoid aggregation of the peptides prior to their use. The final concentration of the Me 2 SO in each experiment did not exceed 2%, a concentration that did not have any effect on the system tested.
Fluorescence Video Imaging Microscopy-The actions of the peptides upon human cell-cell fusion were investigated using fluorescence video imaging microscopy. TF228 cells, expressing the HIV type 1 envelope protein, were labeled with DiI, a lipophilic dye. SupT1 cells, expressing CD4 receptors, were labeled with CMFDA, an aqueous dye. Transfer of DiI and CMFDA to another cell indicates membrane mixing and cytosolic mixing, respectively. Both cell lines are derived from lymphocytes. SupT1 cells were loaded with CMFDA as follows. Cells were pelleted and incubated in fresh RPMI containing 10 M CMFDA for 45 min at 37°C, 5% CO 2 . Inside the cytosol, acetate groups of CMFDA are cleaved by esterases, thereby converting a nonfluorescent compound into a green fluorescent one. Also, the chloromethyl moiety is conjugated to intracellular thiols, producing cell-impermeant dye-thioether adducts. The reaction is believed to be mediated by glutathione S-transferase (Molecular Probes data). The cells were washed twice in RPMI and then incubated at 37°C, 5% CO 2 for another 45 min to ensure complete conversion of the dye. The cells were washed twice again and allowed to settle onto polylysine-coated glass coverslips in serum-free medium for 15 min before addition of DiI-loaded TF228 cells and serum. TF228 cells were labeled by incubation in 3 M DiI(C18:3) for 10 min at room temperature in a 1:1 mixture of diluent C and RPMI medium, followed by three washes in RPMI.
TF228 and SupT1 cells were incubated for 2 h at 37°C, 5% CO 2 and video fluorescence images were taken. After 30 min of this incubation period, different doses of WT, V2E, or N-succinyl-Sendai fusion peptide were applied. N-Succinyl-Sendai fusion peptide consists of N-succinylated 33 N-terminal amino acids of the Sendai fusion peptide, and it is not expected to interact specifically with HIV-1. The proportion of TF228 cells, in contact with SupT1 cells, that had fused to SupT1 cells was counted. Fused cells are those labeled with both dyes. A control experiment was performed in which TF228 cells were preincubated at 37°C for 2 h in the presence or absence of 20 g/ml soluble CD4 immediately prior to co-culture with the SupT1 cells.
Preparation of Lipid Vesicles-Small unilamellar vesicles (SUV) were prepared by sonication, from PC/PS/Chol. (4:4:1, w/w). The cholesterol was included to reduce the curvature of the small unilamellar vesicles (32). Large unilamellar vesicles (LUV) were also prepared from PC/PS/Chol. (4:4:1, w/w) and, when necessary, with different amounts of Rho-PE and NBD-PE. The procedure was as follows. Dry lipids were hydrated in buffer and dispersed by vortexing to produce large multilamellar vesicles. The lipid suspension was freeze-thawed five times and then extruded eight times through a polycarbonate membranes with 0.1-or 0.4-m diameter pores (Nuclepore Corp., Pleasanton, CA).
Peptide-induced Lipid Mixing-Lipid mixing of large unilamellar vesicles was measured using a fluorescence probe dilution assay, based on resonance energy transfer measurements (33). Lipid vesicles containing 0.6 mol % each of NBD-PE (energy donor) and Rho-PE (energy acceptor) were prepared in PBS. A 1:4 mixture of labeled and unlabeled vesicles was suspended in 400 l of the buffer at room temperature, and a small volume of peptide in Me 2 SO was added. The increase in NBD fluorescence at 530 nm was monitored with the excitation wavelength set at 467 nm. The inner filter effect was minimized by using a 0.5-cm pathlength cuvette. The fluorescence intensity before the addition of the peptide was referred to as 0% lipid mixing, and the fluorescence intensity upon the addition of Triton X-100 (0.25% v/v) was referred to as 100% lipid mixing. All the fluorescence measurements in the present study were done on a Perkin-Elmer LS-50B Spectrofluorometer.
Electron Microscopy-The effects of the peptides on liposomal suspensions were examined by negative staining electron microscopy. A drop containing LUV alone or a mixture of LUV and peptide was deposited onto a carbon-coated grid and negatively stained with 2% uranyl acetate. The grids were examined using a JEOL JEM 100B electron microscope (Japan Electron Optics Laboratory Co. Tokyo, Japan).
Visible Absorbance Measurements-The changes in the vesicles' size were measured by visible absorbance measurements. Aliquots of peptide stock solutions were added to 200-l suspensions of 89 M PC/PS/ Chol (4:4:1) LUV in PBS. Absorbance was measured using Bio-Tek Instruments microplate before and after the addition of a peptide.
Membrane Permeability Studies-Membrane destabilization, in the form of diffusion potential collapse, was detected fluorimetrically as described previously (28,34). Briefly, a liposome suspension, prepared in "K ϩ buffer," was added to an isotonic buffer (K ϩ -free buffer), to which the dye diS-C 2 -5 was then added. Subsequent addition of a valinomycin created a negative diffusion potential inside the vesicles by a selective efflux of K ϩ ions, resulting in a quenching of the dye's fluorescence. Peptide-induced membrane permeability toward all the ions in the solution caused dissipation of the diffusion potential, as monitored by an increase of fluorescence. Fluorescence was monitored using excitation at 620 nm and emission at 670 nm. The percentage of fluorescence recovery (F t ) was defined as shown in Equation 1.
Where I t is the fluorescence observed after addition of peptide, at time t; I 0 is the fluorescence after addition of valinomycin; and I f is the total fluorescence prior to addition of valinomycin.
Determining the Oligomeric States of the Peptides by SDS-PAGE-The experiments were done as described (35), except for a change in the sample preparation; high performance liquid chromatography-purified peptide and SDS (1:1, w/w) were dissolved in CHCl 3 /MeOH (2:1 v/v). The solvents were evaporated under a stream of nitrogen and then lyophilized. The peptide and SDS mixtures were resuspended in buffer composed of 0.065 M Tris-HCl, pH 6.8, and 10% glycerol by sonication. Fixing, staining, and destaining times were shorted to 1 min, 1 h, and overnight, respectively, to decrease diffusion effects.
Circular Dichroism (CD) Spectroscopy-CD spectra were obtained using a Jasco J-500A spectropolarimeter. The spectra were scanned with a quartz optical cell with pathlength of 2 mm, at room temperature. Each spectrum was the average of eight scans at wavelengths of 250 to 195 nm. Fractional helicities (36) were calculated as shown in Equation 2.
NBD Fluorescence Measurements-Changes in the fluorescence of NBD-labeled peptides were measured upon their binding to vesicles. NBD-labeled peptide (0.1 M) was added to 2 ml of PBS, containing SUV (403 M). Emission spectra were recorded, with excitation set at 467 nm (10 nm slit), and compared with the emission spectra of the NBD-labeled peptide in a liposome-free buffer.
Binding Experiments-The degree of peptide association with lipid vesicles was measured by adding lipid vesicles to 0.1 M of NBD-labeled peptides at 28°C, as has been previously described with tryptophan containing peptides (37,38). The fluorescence intensity was measured as a function of the lipid/peptide molar ratio, with an excitation set at 467 nm (10 nm slit), and with emission set at 530 nm (5 nm slit). The fluorescence values were corrected by taking into account the dilution factor corresponding to the addition of microliter amounts of liposomes and by subtracting the corresponding blank (buffer with the same amount of vesicles).
Enzymatic Digestion of Membrane-bound Peptides-SUV (500 M) composed of PC/PS/Chol. (4:4:1, w/w) were added to 0.1 M NBD-labeled peptide, followed by the addition of proteinase K (1 g/400 l). Fluorescence intensities as a function of time were obtained before and after the addition of the enzyme. In these experiments, the peptide/lipid molar ratio was kept at a level such that more than 90% of the peptides are assumed to be bound to the vesicles. To estimate the percent of cleavage, a control experiment was done, in which the enzyme was added before the addition of the liposomes. The emission at the end of the control experiment was referred to as 100% cleavage.
Resonance Energy Transfer Measurements-Fluorescence resonance energy transfer was measured using NBD-labeled peptides serving as donors and Rho-labeled peptides serving as energy acceptors (39). Fluorescence spectra were obtained at room temperature, with excitation set at 467 nm using a 10-nm slit width. In a typical experiment, donor peptide (final concentration 0.04 M) was added to a dispersion of PC/PS/Chol. (4:4:1, w/w) SUV (224 M) in PBS, followed by the addition of acceptor peptide in several sequential doses. Fluorescence spectra were obtained before and after addition of the acceptor. The efficiency of energy transfer (E) was determined by measuring the decrease in the quantum yield of the donor as a result of the presence of acceptor. E was determined experimentally from the ratio of the fluorescence intensities of the donor in the presence (I da ) and in the absence (I d ) of the acceptor, at the donor's maximum emission wavelength (524 nm in the case of the WT peptide, 522 nm in the case of the V2E mutant, and 531 nm in the case of the control peptide, A 13 Paradaxin). The percentage of transfer efficiency (E) is shown by Equation 3, Correction for the contribution of acceptor emission as a result of direct excitation was made by subtracting the signal produced by the acceptor-labeled analogue alone. The contribution of buffer and vesicles was subtracted from all measurements.

RESULTS
Peptides representing the 33-amino acid residues N-terminal of gp41 of HIV-1 (LAV1a), as well as its V2E analogue, were synthesized and fluorescently labeled at their N-terminal amino acid with either NBD (to serve in the binding experiments and as an energy donor) or rhodamine (to serve as an energy acceptor). The sequences of the peptides and their designations are given in Table I. The table gives also the sequences of other two membranous ␣-helical peptides, the Nterminal succinylated analogue of the fusion peptide of Sendai virus, and a pore forming toxin, A 13 -pardaxin.

Functional Properties of the Peptides
Inhibition of Cell-Cell Fusion Induced by WT and V2E Peptides-Both WT and V2E peptides inhibited fusion of HIV-1 envelope glycoprotein-expressing cells with a CD4 ϩ cell line (Fig. 1). Statistical analysis using Student's t test showed that the data sets of control compared with WT peptides are different to the 97.5% confidence level and that the control compared with V2E data sets are different to Ͼ99% confidence ( p ϭ 0.0038). The data indicate 50% inhibition of fusion at 0.01 M peptide. CEM cells, expressing CD4 receptors but not the envelope protein, were used as negative controls. These cells were labeled with DiI in the same manner as the TF228 cells and incubated for 2 h with SupT1 cells. Only 2% of CEM cells showed dye overlap with SupT1 cells showing that background overlap is low. Fig. 2 shows representative images of fluorescent cells in the presence of WT, V2E, and control peptides at 4 g/ml. Images D, H, and L show regions of overlap of CMFDA and DiI fluorescence (AND function of the "Metamorph" software, Universal Imaging). Little fluorescence overlap is seen in the case of the WT and V2E peptides, whereas considerable overlap is observed using the control peptide.
Preincubation for 2 h at 37°C in the presence or absence of soluble CD4 (20 g/ml) prior to co-culture of the TF228 and SupT1 cells showed that soluble CD4 inhibits cell-cell fusion as measured by the dye mixing assay. Fusion in the presence of soluble CD4 was 20.6 Ϯ 2.2 and 42.3 Ϯ 5.5% in the absence of soluble CD4 (mean Ϯ S.E., six experiments). This is the same sCD4 concentration as that required for half-maximal inhibition of syncytium formation using these cells (40).
Lipid Mixing Induced by Peptides-The induction of intervesicular lipid mixing by the peptides, as a measure of their

Assembly of Fusion Peptides of HIV
fusogenic activity, was tested with PC/PS/Chol. (4:4:1) LUV (100-nm diameter) utilizing the probe dilution assay (33). The dependence of the extent of the lipid mixing process, on the peptide's concentration, was examined. In separate experiments, increasing amounts of each peptide were added to a fixed amount of vesicles. To compare the activity of the two peptides, the levels of lipid mixing, reached 10 min after the addition of the peptides, were plotted as a function of the [peptide]/[lipid] ratio (Fig. 3A). It is evident that only the WT peptide was able to cause lipid mixing. The ability of the V2E mutant to inhibit lipid mixing induced by the WT peptide was tested using the same experimental system. Addition of premixed WT and V2E using a molar ratio of 1:2.4, respectively, almost totally abolished the ability of the WT to induce membrane fusion (Fig. 3B). Furthermore, in a similar experiment done with 220-nm diameter LUV, 5.3 M of the WT peptide induce 28.6% lipid mixing, whereas a mixture of WT and V2E at a molar ratio of 1:0.75, respectively, did not cause any fusion (graph not shown). Thus, we conclude that V2E is able to interfere with the fusion activity of the WT peptide.
Electron Microscopy-To confirm that the intervesicular lipid mixing was the result of membrane fusion, electron microscopy was used. 100-nm diameter PC/PS/Chol. (4:4:1) LUV (90 M), were visualized with an electron microscope before and after addition of peptides (5 M). Fig. 4 shows representative micrographs of the LUV taken at pH 7.3 without peptide (A), with WT peptide (B), with the V2E peptide (C), and with a mixture of the WT and V2E (1:3 w/w) (D). The micrographs demonstrate that the lipid mixing observed with the WT peptide appears concurrently with a size increase of a portion of the vesicles. Such size increase was not observed with either the V2E mutant alone or with the mixture of the WT and V2E.

Mechanism for the Loss of the Activity of the V2E Mutant
Induction of Membrane Aggregation by the WT but Not by the V2E Mutant-Changes in vesicle size distribution resulting from aggregation and/or fusion can be monitored by following the absorbance of the liposome suspension. The changes in the absorbance at 405 nm as a function of the peptide to lipid molar ratio are shown in Fig. 5. The data reveal that in all concentrations tested, the WT peptide caused aggregation and/or fusion of vesicles, whereas the V2E mutant did not. Using subfusion quantities, it was revealed that the WT, but not the V2E mutant, can cause vesicles' aggregation.
Induction of Membrane Destabilization by the WT but Not by the V2E Mutant-To study the potential of the peptides to destabilize lipid bilayers, their ability to dissipate the diffusion potential in SUV prepared from PC/PS/Chol. (4:4:1 w/w) was tested. Peptides, at various concentrations, were mixed with a solution containing the K ϩ -entrapped vesicles, the fluorescent dye diS-C 2 -5, and valinomycin. Addition of valinomycin created a negative diffusion potential inside the vesicles by a selective efflux of K ϩ ions, resulting in a self-quenching of the dye's There is considerably more overlap of red and green fluorescence in the presence of control peptide than in WT or V2E peptides. Bright field images were used to confirm that overlap was due to the presence of both dyes in the same cell rather than just due to one cell lying over another.
fluorescence. Peptide-induced membrane permeability toward all the ions in the solution caused dissipation of the diffusion potential, as monitored by an increase of fluorescence. Recovery of fluorescence was monitored with time using excitation at 620 nm and emission at 670 nm (see inset to Fig. 6). The maximum level reached as a function of peptide concentration is shown in Fig. 6. The results demonstrate that the WT peptide is very active, and the V2E mutant is not active, in membrane destabilization.
Oligomerization of the Peptides in SDS-PAGE-The state of aggregation of several membrane proteins was determined in SDS, a membrane mimetics environment (41,42). Here, SDS-PAGE revealed that the dominant form of both peptides is a dimer, but the WT peptide, unlike the V2E mutant, also forms higher order oligomers, which seems to be tetramers (Fig. 7). The inability of the mutant to form higher order oligomers, may be correlated with its lost of activity in induction vesicle fusion, and the loss of activity of the mutated envelope protein.

Structural Properties of the Peptides Adoption of Partial ␣-Helix Conformation in Membrane-mimetic Environments, by Both WT and V2E-
The secondary structure of the peptides was evaluated from their CD spectra in 40% 2,2,2-trifluoroethanol and in 1% SDS. Both peptides displayed spectra with minima at 208 and 222 nm in both solvents (Fig. 8) which is typical of ␣-helix structure. However, the ␣-helical content of both peptides is higher in SDS which is considered as a better membrane mimetic environment than 40% trifluoroethanol. The ␣-helical content of V2E in SDS (40%) is slightly higher than that of the WT peptide (30%). It should be noted that only ϳ20 amino acids out of the 33 composed the actual fusion peptide and that 6 of them are glycines.
Localization of the N Terminals of WT and V2E within the Hydrophobic Core of the Membrane-The fluorescence emission spectra of the NBD-labeled peptides were monitored in aqueous solutions and in the presence of vesicles. In aqueous solution both peptides exhibited emission spectra similar to the NBD moiety dissolved in water (31,43), with a maximum at 549 Ϯ 1 nm (graph not shown). Upon addition of SUV (403 M) composed of PC/PS/Chol. (4:4:1 w/w) to the solution (pH 7.4), the fluorescence emission intensity increased significantly (3.1 and 5.3 times for the WT and V2E, respectively) concomitant with blue shifts of the emission maxima of the peptides. The shift was slightly larger for the V2E peptide (maximum of 522 Ϯ 1 nm) than for WT (maximum of 524 Ϯ 1 nm) which agrees with the higher increase of the fluorescence of V2E. Blue shifts of these magnitudes have been observed when surfaceactive NBD-labeled peptides interacted with lipid membranes (31,44,45) and are consistent with the NBD probe located within the hydrophobic core of the membrane (43). Note that in these experiments, the lipid/peptide molar ratio was consistently high (Ͼ4000:1) so that spectral contributions of free peptides could be considered negligible.
Binding of WT and V2E to Phospholipid Membranes-The increases in the fluorescence intensities of the NBD-labeled peptides, due to membrane partition, were recorded as a function of the lipid/peptide molar ratios. The fractions of membrane-bound peptides are plotted versus the lipid/peptide molar ratios in Fig. 9. The shapes of the binding curve of both peptides were very similar, indicating that they have similar surface partition coefficients. Since the peptides aggregate in the aqueous solution, their binding isotherms were not analyzed further. The shape of the binding curve which seems to become saturated at lipid/peptide molar ratios similar to those observed with the NBD-labeled antibacterial peptides DS-b (46) suggests a partition coefficient on the order of 10 4 M Ϫ1 for the WT and for the V2E mutant. These values are 1 order of magnitude smaller than the value obtained for the Sendai virus fusion peptide (19).
Accessibility of WT and V2E to Proteolytic Digestion in Their Membrane-bound State-The susceptibility of membranebound NBD-WT and NBD-V2E to proteolytic digestion by proteinase K was investigated by using PC/PS/Chol. (4:4:1 w/w) SUV (500 M) as described under "Experimental Procedures." The addition of the enzyme to a mixture containing an NBDlabeled peptide and vesicles caused a fast decrease in the NBD fluorescence, demonstrating its release from the hydrophobic environment of the vesicles. The level of the final fluorescence intensity was the same as that obtained for an unbound peptide. The data reveal that both peptides are accessible to proteolytic digestion, with faster kinetics for the V2E mutant (graph not shown). Full digestion of V2E was accomplished within 6 min, whereas it took more than 10 min to accomplish full digestion of WT.

Mechanism of Inhibition
Self-and Coassembly of WT and V2E in Their Membranebound State-The aggregation state of the peptides in their membrane-bound state was monitored by resonance energy transfer measurements. Peptides that were labeled at their N-terminal with either NBD, serving as an energy donor, or with rhodamine, serving as a fluorescence acceptor, were used. An example of a typical profile of the energy transfer from NBD-WT to Rho-WT, in the presence of lipid vesicles, is de-picted in Fig. 10A. When Rho-WT (final concentration of 0.04 -0.12 M) was added to a mixture of NBD-WT (0.04 M) and PC/PS/Chol. (4:4:1) lipid vesicles (224 M), a dose-dependent quenching of the donor's emission, which is consistent with energy transfer, was observed (Fig. 10A). Dose-dependent quenching was also observed for the donor/acceptor V2E and for the donor-V2E with acceptor-WT combination (Fig. 10B). Note, that the acceptor-peptide was added only after the donorpeptide was already bound to the membrane, thus preventing any association in solution. The lipid/peptide ratio in these experiments was kept high to create low surface density of donors and acceptors to reduce energy transfer between unassociated peptide monomers. To confirm that the observed energy transfer is due to peptide aggregation, the transfer effi-  , Fig. 10B). The levels of energy transfer between all the three combinations are significantly higher then those expected for randomly distributed donors and acceptors. The energy transfer between NBD-A 13pardaxin and Rho-WT was measured as a control, and the values obtained were similar to those of random distribution. The data for the random distribution was calculated assuming R 0 value for the NBD/Rho donor/acceptor pair to be 51 Å (39). The acceptor concentrations presented in Fig. 10B are the membrane-bound values, which were estimated using their binding curve (Fig. 9). The data reveal that the fusion peptides are not randomly distributed throughout the membrane but rather are associated. The finding that the WT and V2E can coassemble in the membrane may account for the ability of the V2E analogue to inhibit the fusion activity of the WT peptide.
Self-Association of WT and V2E in Solution-WT and V2E self-association in solution was tested using rhodamine-labeled peptides. Since the fluorescence of rhodamine is quenched when several molecules are in close proximity, an increase in fluorescence should occur when an aggregated rhodamine-labeled peptide is dissociated, a process that can take place when the peptide is cleaved by a proteolytic enzyme. When equal concentrations of Rho-WT or Rho-V2E (0.05 M each, as determined by UV absorbance at 567 nm in Me 2 SO) were added to PBS, the fluorescence of Rho-V2E was 3.7-fold higher than that of Rho-WT, which suggests that the WT peptide is more aggregated in solution than the V2E mutant. Furthermore, upon the addition of proteinase K to solutions of the peptides (0.05 M each), the fluorescence of Rho-WT increased 6.4-fold and that of Rho-V2E increased only 1.3-fold (data not shown). To ensure complete digestion, the labeled peptides were treated for ϳ30 min with the enzyme, a time that is longer than required for total cleavage as described above in the section of the accessibility of the peptides to proteolytic digestion in their membrane-bound state.

DISCUSSION
The interesting observations in this study are the 50% inhibition of HIV-1 envelope glycoprotein-mediated cell fusion at 0.01 M peptide concentration for both wild-type and the V2E mutant ( Figs. 1 and 2), and the inhibition of WT induced vesicles fusion by the V2E mutant (Figs. 3B and 4). These findings are consistent with the trans-dominant effect of the gp41 mutant protein in inhibiting HIV-1 infectivity in vivo when co-expressed with WT gp41 (13). We have shown recently that the trans-dominant mutation also inhibits HIV-1 envelope glycoprotein-mediated cell fusion when expressed in target cells (14). Previously, short peptides with sequences corresponding to the N terminus of gp41 were shown to inhibit HIV-1 envelope glycoprotein-induced syncytium formation. However, these short peptides were only effective at concentration of 1 mM (for the 6-aa long peptide (47)) and 0.01 mM (for the 11-aa long peptide (48)). In another study, Slepushkin et al. (49) showed that a 22-aa long fusion peptide and its conjugate with bovine serum albumin inhibited HIV-1 infection at concentrations of 1 M. The effect of the peptide is certainly enhanced as its length increases. Here, we found that the 33-aa fusion peptides inhibited fusion at a concentration of 2 orders of magnitude lower than that reported for the 22-aa peptide (49). The fusion assay we used is based on the redistribution of fluorescent dyes about 2 h after incubation of the gp120-gp41expressing cells with target cells and therefore measures initial events.
Other peptides from the external domain of HIV-1 gp41 also exhibit antiviral activity. The most potent peptide was derived from the aa 643-678 region and was active at 0.001 M con-centration (50). Peptides derived from other regions (a.a 637-666 (51) and aa 558 -595 (52)) (90% inhibition at 0.1 M) are less potent than those reported herein.
The present study also provides a possible mechanism for membrane fusion induced by the WT N-terminal gp41 peptide and its inhibition by the V2E mutant (Figs. 3B and 4). The observation that, as in the case of viral fusion induced by the full proteins (13), the parent WT peptide is fusogenic and a single amino acid substitution results in inactivation of the peptide suggests that the properties of membrane interaction discussed below might also play a crucial role in virus-cell fusion. Fusion of phospholipid membranes is thought to involve three steps, vesicle aggregation, membrane destabilization, and merging of membranes (53)(54)(55). The WT, but not the V2E peptide, can induce these three events as discussed below.
First, the WT peptide does not need Ca 2ϩ or Mg 2ϩ to mediate fusion, which suggests that the peptide itself is sufficient to induce vesicle aggregation (Figs. 3A and 4). Indeed, subfusion quantities of the WT, but not of the V2E mutant, induced vesicle aggregation (Fig. 5). Contrary to the finding of the present study, it has previously been reported that a 23-aa peptide resembling the sequence of HIV could induce phosphatidyl oleoyl palmitoyl glycerol LUV fusion only after their pre-aggregation by Ca 2ϩ or Mg 2ϩ ions (21). However, unlike the parent glycoprotein, a 23-residue synthetic peptide representing the N terminus of the V2E mutant was unable to inhibit the fusion activity of the wild-type peptide (56). The 33-mer peptides studied herein represent not only the hydrophobic stretch of 21 aa but also the polar border of this hydrophobic region, which consists of a highly conserved series of polar amino acids, which makes the peptide soluble in aqueous solution albeit at low concentrations.
Second, the high potency of the WT to dissipate the diffusion potential demonstrates that this peptide can significantly destabilize lipid membranes (Fig. 6). The potency of the peptide was higher than that of the fusion peptide of Sendai virus (19), and the peptide was as potent as pore-forming peptides such as alamethicin (37) and pardaxin (29). Nevertheless, the kinetics of membrane permeation by the WT peptide suggest that it forms irreversible aggregates in the membrane (57). The different membrane-permeating abilities of the WT and the V2E peptides are not due to a difference in the amount of membrane-bound peptide, since both peptides have similar partition coefficients (Fig. 9). However, it may result from the ability of the WT to form higher order aggregates, as revealed using SDS-PAGE (Fig. 7).
Third, the lipid mixing assay revealed that the WT, but not the V2E peptide, induced membrane fusion (Fig. 3A). Furthermore, V2E inhibited the ability of the WT peptide to induce membrane fusion (Fig. 3B). The findings that subfusion quantities of the WT, but not of the V2E mutant, induced vesicle  aggregation (Fig. 5), suggest that the inactivity of the mutant originates at, or before, the stage of the fusion process, where the bilayers approach each other. Circular dichroism spectra (Fig. 8) suggest that there is no significant difference in the secondary structure of the WT and the V2E peptides. The partial ␣-helical content in 1% SDS (30 and 40% for the WT and the V2E peptides, respectively) can be explained by the presence of the polar border of the fusion peptide. However, the fusion peptide can also contain some ␤-sheet conformation, as suggested by others using Fourier transform infrared spectroscopy (21,26). However, several reports showed that while helix formation is probably a requirement, it is not sufficient to trigger membrane fusion (8,16). Some peptides failed to induce efficient membrane fusion, even though they could bind to lipid bilayers and exhibited the appropriate secondary structures. Therefore, it was proposed that in addition to the other requirements for fusogenic activity, such as membrane binding, destabilization of the bilayer, and helical conformation, self-aggregation of the peptide monomers within the membrane and their appropriate orientation are also essential.
Attempts to define the oligomeric state of the HIV-1 envelope glycoprotein have yielded conflicting results. Several reports indicated that the envelope glycoprotein of HIV-1 is a tetramer in its membrane-bound state (58 -60). On the other hand, other reports revealed that it forms trimers (61)(62)(63). In the gp41 non-fusogenic form the fusion peptide is buried; thus it is not responsible for the oligomerization of the protein. However, once the envelope glycoproteins are recruited to form a fusion complex the fusion peptides will self-assemble. Here, fluorescently labeled WT and V2E peptides were used to show that both peptides tend to self-aggregate efficiently within PC/PS/ Chol. vesicles. Furthermore, the two peptides can coassemble in their membrane-bound state (Fig. 10B). The size of the aggregates was determined by using SDS as a membrane mimetic environment, as has been done with several other mem-brane proteins (41,42). SDS-PAGE revealed that the dominant form of both peptides is a dimer but that the WT peptide also forms higher order oligomers, which seem to be tetramers (Fig.  7). The inability of the V2E mutant to form higher order oligomers (Fig. 7) may be correlated with its loss of activity. Furthermore, the coassembly of the two peptides in their membrane-bound state (Fig. 10B) suggests that the inhibition of membrane fusion exhibited by the V2E mutant occurs via its association with the WT. This inhibition may occur either because the V2E mutant decreases the portion of higher order oligomers of WT or because the V2E mutant forms nonfunctional higher order oligomers with the WT peptide. Since the V2E mutant gp41 elicits a dominant interfering effect even in the presence of excess wild-type glycoprotein (13), the second possibility is a more likely mechanism. Furthermore, when the V2E peptide was pre-mixed with the WT peptide, prior to SDS-PAGE, there was no significant difference in the proportion of higher order aggregates (data not shown).
Whether the peptides also self-associate in solution as well was assessed using rhodamine-labeled peptides and exposure to a proteolytic enzyme. Quenching experiments with rhodamine-labeled peptides in an aqueous solution revealed that both WT and V2E peptides self-associate in aqueous solutions. The self-quenching of the rhodamine-labeled WT peptide was 6.4-fold and that of the rhodamine-labeled V2E mutant was only 1.3-fold, which suggests that the WT peptide forms higher order aggregates than the V2E mutant also in aqueous solutions.
Another characteristic of a peptide's interaction with membranes is the extent of a peptide's penetration into the lipid bilayers and its orientation. The studies with the NBD-labeled WT and V2E peptides revealed that the N terminus of both peptides was inserted into the membrane, with that of the non-fusogenic V2E mutant being inserted a little deeper than that of the WT. This might be due to the repulsion between the negative charge of the carboxylate at the N terminus of the V2E peptide and the acidic head groups of the lipids. These findings are similar to those obtained with the Sendai virus fusion peptides, in which the N terminus of the more fusogenic G12A mutant was exposed to the surface of the membrane more than the N terminus of the WT (19). Even though the N termini of both peptides are inserted into the hydrophobic core of the membrane, both peptides were cleaved efficiently by proteinase K when bound to membranes (data not shown), which is in contrast to other membrane-inserted helices, which were totally protected from enzymatic cleavage in their membranebound state (64). These findings suggest an oblique orientation for the fusion peptides, rather than a transmembrane orientation, as has been suggested by others (26). Such an oblique orientation is consistent with recent data (65,66) that indicate no merger of the inner with the outer leaflets of viral or liposomal membranes during fusion. Therefore, the peptides do not need to penetrate deeply into the inner leaflet to induce fusion. The more rapid enzymatic digestion of membrane-bound V2E, as compared with WT, is consistent with the higher order aggregates formed by WT, which may better protect the peptide from enzymatic digestion.
Synthetic peptides can only partly mimic the complex fusogenic properties of a viral protein (67). However, many of the properties of the HIV gp41 fusion peptide demonstrated in the present study, such as strong binding to membranes and assembly therein, orientation in the membrane-bound state, ability to destabilize membrane-packing, and ␣-helix secondary structure, were proposed to be requirements of viral proteininduced fusion (68). The similarities and distinctions between the physicochemical properties of the WT and V2E shown in  II may shed some light on the mode of action of the fusion peptide and the mechanisms of inhibition of this fusion by V2E. The WT and the V2E mutant are similar in their partial ␣-helical structure, ability to bind to membranes, and to self-assemble therein, and susceptibility to proteinase K digestion in the membrane-bound state. However, the WT and mutant differ in their ability to form higher order peptide aggregates in membranes, to induce vesicle aggregation, to destabilize lipid packing, and to cause membrane fusion. A model for the peptideinduced membrane fusion is shown in Fig. 11. Homotetramers of the fusion peptide cause the membranes to approach each other (top), which is followed by membrane mixing (bottom).