INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Fusion of influenza virus with its target membrane is mediated by the viral envelope protein HA at low pH. Acidification converts the HA into a fusogenic conformation, thereby exposing the hydrophobic N terminus of the HA2 subunit (36). This fusion peptide, which is highly conserved among different influenza virus strains, is believed to interact with the target membrane, where it causes a transient destabilization of the lipid bilayer (33). While the relevance of this peptide for fusion has been well documented, the role of other HA sequences in fusion is less established. Recent investigations emphasize the relevance of the membrane-spanning (15) and cytoplasmic (12, 32) domains of HA for the formation of a fusion pore and virus infectivity. Replacement of the TMR by a GPI anchor suppressed the formation of an aqueous pore between HA-expressing cells and RBC ghosts, while membrane (hemi)-fusion was not inhibited (15, 17, 19, 25). Furthermore, deletion of the CT of HA has been suggested to affect fusion kinetics (12) and shown to modulate virus infectivity (14, 32). Thus, while the interaction of the fusion sequence with the adjacent membrane is sufficient to trigger membrane fusion, the TMR and possibly the CT are essential principally for formation and widening of the aqueous fusion pore.

It has been shown for several other enveloped viruses that modification of the membrane-spanning domain and the CT of the glycoprotein may affect fusion activity. Alterations of the TMR of the gp41 envelope glycoprotein of human immunodeficiency virus by various point mutations (10, 26) or its substitution by the respective domain of vesicular stomatitis virus G protein (26) decreased or abolished cell fusion. Truncation of the cytoplasmic sequence of gp41 blocked the fusion activity (26) and infectivity (7) of virus. Specific truncation of the cytoplasmic domain of the virus envelope protein of simian immunodeficiency virus enhanced (3, 29) or diminished (34) syncytium formation. Truncation of the COOH-terminal region (i.e., the CT) of the fusion protein F of paramyxovirus SV5 led to a decrease of cytoplasmic content mixing activity, which correlated with the extent of the deletion (2). However, modification of those domains does not necessarily affect fusion activity. For example, while deletion of the entire cytoplasmic domain of the fusion protein of human parainfluenza virus type 3 did eliminate cell fusion activity (37), it was not impaired in a mutant of human parainfluenza virus type 2 with a truncated CT (37).

Whether for influenza virus HA the presence of any transmembrane and cytoplasmic domains per se is sufficient to trigger the formation of an aqueous fusion pore or whether the specific HA sequences are required for complete fusion has not been addressed systematically. Earlier studies suggested that the parallel replacement of both domains of the wt HA by related domains of another enveloped virus fusion protein does not inhibit fusion activity (6, 30). However, those studies left open whether both domains form a functional entity and whether the replacement of only one domain would affect fusion. Moreover, it remains to be elucidated whether the kinetics of fusion mediated by chimeric constructs with changed transmembrane and/or cytoplasmic domains were altered. Here, we addressed these questions by investigating the fusogenic ability of various chimeric HA. In one set of chimeras, the CT and/or TMR of HA (subtype H7) was substituted by the corresponding domains of the glycoprotein F of Sendai virus. Glycoprotein F has been shown to be responsible for triggering fusion between Sendai virus and its target (reviewed in reference 26). HA and F differ significantly in their amino acid sequence of both domains as well as in the length of their CT, with about 42 aa for F and 10 aa for HA. In two other constructs, we replaced the cytoplasmic domain of HA by large peptides of the human neurofibromin type 1 (NF1; ras regulator) or c-Raf-1 (ras effector), NF78 (residues 1441 to 1518; 78 aa), and Raf81 (residues 51 to 131; 81 aa), respectively. Both fragments (8, 23) are derived from cytoplasm-soluble proteins and differ from the HA or F CT in amino acid sequence, peptide length, and function. Thus, if specific sequences of the C terminus of HA were required for fusion, we would expect significant effects on the fusogenic properties, in particular of the last two chimeric constructs.

We expressed wt HA and the various chimeric protein constructs in monkey kidney cells (CV-1) by using the vaccinia virus-T7 polymerase transient-expression system (1, 9, 28). This transient-expression system allowed us to efficiently screen chimeras for differences in their low-pH-induced fusion activity with RBCs as targets. By fluorescence microscopy and spectroscopy, both membrane mixing and the formation of an aqueous fusion pore were assessed by double labeling of RBCs with appropriate fluorophores, the lipid analog R18, and the water-soluble fluorophore calcein. Our results show that neither sequence specificity nor a certain length of the CT is required for HA to induce membrane fusion or pore formation.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Abbreviations used in this paper. aa, amino acids; BSA, bovine serum albumin; CT, cytoplasmic tail; DMEM, Dulbecco's modified Eagle's medium; DMEM+, DMEM supplemented with 5% FCS; DMEM, DMEM without FCS; DMEM/met, methionine-deficient DMEM; FACS, fluorescence-activated cell sorter; FCS, fetal calf serum; FDQ, fluorescence dequenching; GPI, glycosylphosphatidylinositol; HA, influenza virus hemagglutinin; PBS, phosphate-buffered saline; PBS++, PBS containing 1 mM CaCl₂ and 1 mM MgCl₂; RBC, human erythrocytes; R18, octadecyl rhodamine B chloride; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; TMR, transmembrane region; WGA, wheat germ agglutinin; wt, wild type.

Materials. The fluorophores R18 and calcein-AM were purchased from Molecular Probes; and Tran[³⁵S]-label (>1,000 Ci/mmol) was purchased from ICN Radiochemicals. The enzymes used in molecular cloning were obtained from New England Biolabs. DMEM, trypsin versene, and FCS were purchased from BioWhittaker; DMEM without L-methionine and L-glutamine, L-glutamine, 2YT medium, and Lipofectin reagent were purchased from Life Technologies Inc.; protein A-Sepharose beads were purchased from Pharmacia Biotech. AmpliTaq DNA polymerase was obtained from Perkin-Elmer. The ExSite PCR-based site-directed mutagenesis kit was purchased from Stratagene; the Sculptor in vitro mutagenesis system was purchased from Amersham. Oligonucleotide primers used for sequencing and PCR were synthesized by Pharmacia Biotech and Gibco BRL. Neuraminidase (VCNA from Vibrio cholerae) was obtained from Behringwerke AG, trypsin-EDTA solution (0.05% trypsin and 0.02% EDTA in PBS) was purchased from Biochrome KG, and secondary antibody was purchased from Dakopatts. Cell culture dishes were purchased from Nunc (Biochrome KG).

Recombinant DNA procedures. All basic DNA procedures were done as described by Sambrook et al. (31). For transient expression in CV-1 cells, the F wt cDNA and the HA wt (subtype H7) cDNA were subcloned into the EcoRI site in the polylinker of the pTM1 vector (22), resulting in pTM/F and pTM/HA, respectively.

Cell culture and vaccinia virus-based expression of foreign genes. CV-1 cells were maintained as monolayer cultures in DMEM+. Transient expression of HA as well as the chimeric proteins in CV-1 cells was performed as described previously by Fuerst et al. (9). Subconfluent monolayers of cells grown on 35-mm-diameter plastic dishes were infected with recombinant vaccinia virus vTF7-3 (9), which expresses T7 RNA polymerase, at a multiplicity of infection of 10 PFU per cell and incubated at 37°C for 60 min. The virus inoculum was removed, and the cells were washed once with DMEM and transfected with 3 µg of the respective plasmid DNA, using Lipofectin as the transfecting agent as described by the manufacturer. At 4 h posttransfection, the cells were washed again with DMEM and further treated for metabolic labeling, FACS analysis, or fusion assays.

Isotopic labeling of polypeptides, immunoprecipitation, and SDS-PAGE. Cells expressing the recombinant glycoproteins were incubated in DMEM/met for 30 min at 4 h posttransfection and labeled with 50 µCi of Tran[³⁵S]-label in 600 µl of DMEM/met. After 3 h, the cells were washed once with cold PBS and lysed in ice-cold RIPA buffer (0.15 M NaCl and 20 mM Tris-HCl [pH 7.4] containing 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS, 1 mM phenylmethylsulfonyl fluoride, and 10 mM iodoacetamide). Lysates were centrifuged at 14,000 × g for 5 min, and the resulting supernatants were immunoprecipitated overnight at 4°C with polyclonal antibodies directed against influenza A virus (subtype H7). Immunoprecipitates were analyzed by SDS-PAGE on 12% acrylamide gels and visualized by fluorography (1 day) with Kodak X-Omat AR films.

FACS analysis. After being washed with DMEM 5 h postinfection, CV-1 cells expressing viral protein were incubated overnight at 32°C in DMEM+. They were then washed three times with ice-cold PBS, removed from the dish by gentle scraping, and fixed with paraformaldehyde (3% in PBS) for 1 h on ice. After three washes with PBS containing 0.5% BSA, the cells were incubated with polyclonal HA antibody (anti-influenza A virus rabbit serum, diluted 1:200 in PBS-0.5% BSA) and then treated with a secondary antibody (fluorescein-conjugated porcine immunoglobulin directed against rabbit immunoglobulins, diluted 1:30). Unbound antibodies were washed out at every step with PBS supplemented with 0.5% BSA. Finally the antibody-cell complexes were suspended in PBS and analyzed with FACScan equipment and Lysys II software (Becton Dickinson, Heidelberg, Germany).

Labeling of RBCs with R18 and calcein-AM. Human RBCs were labeled with the lipid probe octadecyl rhodamine B chloride (R18) as described previously (20, 21). A 10-µl sample of R18 (2 mM in ethanol) was added to RBCs (1% hematocrit in 5 ml of PBS) under vortexing. After the addition of 5 ml of PBS and incubation for 30 min at room temperature in the dark, 20 ml of DMEM+ was added to the suspension to absorb unbound probe. After further incubation at room temperature in the dark for 20 min, the RBC suspension was washed five times in 40 ml of PBS, and each wash was followed by centrifugation. The last washing step was in PBS containing 1 mM CaCl₂ and 1 mM MgCl₂ (PBS++). The cells were kept in PBS++ at 4°C for up to 48 h. The fluorescence intensity of intact labeled RBCs due to R18 self-quenching was 25 to 30% of that of labeled RBCs treated with 0.5% Triton X-100 (100% fluorescence because of infinite dilution of R18). Double labeling was performed by the method of Morris et al. (21) with slight modifications. Briefly, calcein-AM (1 mM in dimethyl sulfoxide) was added to R18-labeled RBCs (10% in PBS) to a final concentration of 20 µM. After incubation for 30 min at 37°C in the dark, the RBC suspension was washed and further incubated in PBS for 20 min at 37°C in the dark. The wash in PBS was repeated twice, and doubly labeled RBCs were kept in PBS++ at room temperature until use within 2 h.

Binding of fluorescently labeled RBCs to cells expressing viral proteins. After being washed with DMEM 5 h postinfection, the cells were incubated overnight at 32°C in DMEM+, washed once with DMEM (without serum), and incubated for 1 h at 37°C with 100 mU of neuraminidase per ml in DMEM. A 1-ml volume of labeled RBCs (0.2% hematocrit) in PBS++ was added to the monolayer and incubated for 15 min at room temperature in the dark with occasional gentle agitation. Unbound RBCs were removed by five washes with PBS.

R18 fusion assay. R18 RBC-acceptor cell complexes were suspended by brief treatment (2 min at room temperature) with 100 µl trypsin-EDTA solution (pretempered to 37°C), subjected to repeated washings with PBS++ at 4°C, and placed on ice until further use.

Fluorescence microscopy. After binding of doubly labeled RBCs to acceptor cells, complexes on the dish were studied at room temperature. By using the inverted microscope Axiovert 100 (Zeiss) with a 40× objective (LD Achroplan; Zeiss), fusion was observed after PBS++ (pH 7.4) was replaced by NaAc++ (20 mM NaC₂H₃O₂ plus 150 mM NaCl containing 1 mM CaCl₂ and 1 mM MgCl₂ [pH 5.0]). Calcein fluorescence was monitored with a "blue" filter set (450- to 490-nm excitation filter, long-path 520-nm emission filter). R18 fluorescence was visualized with a "green" filter set (510- to 560-nm excitation filter, long-path 590-nm emission filter). Photomicrographs were taken with an MC 80 microscope camera (Zeiss) on an Ektachrome EPH P1600x color reversal film for push processing (Kodak) at 3,200 ASA.

Binding of RBCs to CV-1 cells by lectin and fusion assay. Control experiments were performed to exclude fusion induction by vaccinia virus proteins as follows. Cells expressing only vaccinia virus proteins or, additionally wt HA were incubated for 5 min at room temperature with 500 µl of WGA (8 µg/ml; Sigma) in PBS++, and 500 µl of labeled RBCs (0.2% hematocrit) was added. After further incubation for 15 min at room temperature in the dark under occasional gentle agitation, unbound RBCs were removed by five washes with PBS. Fusion was monitored by fluorescence microscopy as described above.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Expression and characterization of HA and HA chimeras in CV-1 cells. wt HA and chimeric proteins were expressed in CV-1 cells with the vaccinia virus-T7 polymerase transient-expression system.

View larger version (46K): [in this window] [in a new window]   FIG. 1.   Expression of HA and HA-derived chimeras in CV-1 cells. CV-1 transfected with wt HA (H/H/H), H/H/F, H/F/H, H/F/F, and H/HF/F (left gel) or with vaccinia virus (VAC), H/H/H, H/H/N, and H/H/R (right gel) were labeled with [35S]methionine/cysteine. Cell extracts were immunoprecipitated with antibody against influenza virus and subjected to SDS-PAGE (12% polyacrylamide gel) followed by fluorography (1 day). The positions of uncleaved HA0 and its two subunits, HA1 and HA2, in the gel are marked on the left.

View larger version (32K): [in this window] [in a new window]   FIG. 2.   Histograms of the FACS analysis of CV-1 cells. The cells cultured in 35-mm dishes were infected with vTF7-3 vaccinia virus and transfected with plasmids encoding the respective protein as designated. After being labeled with fluorescent antibodies, the cells were treated as described in Materials and Methods and analyzed by FACS. VAC represents CV-1 cells infected with wt vaccinia virus (control); H/H/H, H/H/F, H/HF/F, H/F/F, H/F/H, H/H/N, and H/H/R represent for cells transfected with the respective constructs. Cell numbers (Counts) are plotted against fluorescence intensity (FL1-Height). About 50,000 were analyzed in each experiment, except for the H/F/F and H/F/H experiments in which about 20,000 cells were measured. The calculated transfection efficiency and mean fluorescence index are shown in Table 2.

Fusion of HA-expressing CV-1 cells with labeled RBCs. To probe for functional differences between HA wt and chimeric constructs, we measured low-pH-triggered fusion of CV-1 cells expressing the different HA-derived chimeras with human RBCs labeled with the lipidic fluorophore R18 as well as the small water-soluble dye calcein. The doubly labeled RBCs were efficiently bound to H/H/H as well as to the chimeric constructs at neutral pH (Fig. 3). The differences in the extent of RBC binding are probably due to the different surface expression levels (Fig. 2). Since RBC binding corresponds to the amounts of protein present on the surface, the structural differences in the TMRs and CTs of the constructs had no specific influence on the capacity of the RBC binding. At room temperature (25°C) within 3 to 4 min after lowering the pH by replacing neutral buffer (pH 7.4) with acidic buffer (pH 5.0), we could observe the emergence of membrane-bound fluorescent dye (R18) in the plasma membrane of the CV-1 cells. This indicates the occurrence of membrane mixing, i.e., membrane fusion (see also Fig. 5 through 7). A parallel flow of calcein from the lumen of the RBCs into the cytoplasm of the CV-1 cells expressing the above proteins could be observed within 3 to 5 min after pH lowering, thus indicating the formation of an aqueous continuity, i.e., fusion pore, between the RBC and CV-1 cell (Fig. 3 and 4).

View larger version (91K): [in this window] [in a new window]   FIG. 3.   Fluorographs of doubly labeled RBCs bound to CV-1 cells expressing chimeric or wt HA proteins. Chimeric proteins mediate membrane mixing as well as cytoplasmic content mixing, as demonstrated by fluorescence microscopy. Doubly labeled (R18 and calcein) human RBCs were bound to monkey kidney cells expressing the respective protein as described in Materials and Methods. At 4 min after the pH was reduced to 5.0, R18 label was present in the membranes of the CV-1 cells, indicating the mixing of membrane lipids. The flow of calcein from the lumen of the bound RBCs into the cytoplasm of the CV-1 cells was observed as well. This reveals that pores which permit the transfer of aqueous compounds with a low molecular weight have been formed. All observations were taken at room temperature. (Top row) R18 fluorescence at pH 7.4; (second row) calcein fluorescence at pH 7.4; (third row) R18 fluorescence 4 min after the pH was reduced to 5.0; (bottom row) calcein fluorescence 4 min after the pH was reduced to 5.0; (left column) H/H/H; (second column) H/H/F; (third column) H/F/H; (right column) H/F/F.

View larger version (92K): [in this window] [in a new window]   FIG. 4.   Fluorographs of doubly labeled RBCs bound to CV-1 cells expressing H/H/N or H/H/R proteins. Doubly labeled (R18 and calcein) human RBCs were bound to monkey kidney cells expressing H/H/N or H/H/R constructs at pH 7.4 (not shown). At 6 min after the pH was reduced to 5.0, R18 label was present in the membranes of the CV-1 cells, indicating the mixing of the membrane lipids. The flow of calcein from the lumen of the bound RBC into the cytoplasm of the CV-1 cell was observed as well. This reveals that pores which permit the transfer of aqueous compounds with low molecular weights have been formed. The experiments were performed at room temperature. (Top row) R18 fluorescence 6 min after the pH was reduced to 5.0; (bottom row) calcein fluorescence 6 min after the pH was reduced to 5.0; (left column) H/H/N; (right column) H/H/R.

Kinetics of membrane fusion. Since no difference between the fusogenicity of the various proteins was apparent from these microscopic studies, we also compared the kinetics of membrane fusion induced by chimeras with that of wt HA at various pHs. In Fig. 5, the typical kinetics are shown for wt HA (H/H/H) and the chimeric constructs between HA and F at pH 4.8, 5.8, and 6.7 and 37°C. For all the proteins, no significant increase of R18 fluorescence intensity was observed after the addition of RBC-CV-1 cell complexes to the buffer set to pH 6.7 (Fig. 5). On the other hand, immediately after addition of the RBC-CV-1 cell complexes to the buffer set to pH 4.8, a steep increase of R18 fluorescence intensity, reaching a plateau after about 100 s, was observed (Fig. 5). Measurements at an intermediate pH value (pH 5.8) resulted in a slower increase of the R18 fluorescence for all the studied proteins (Fig. 5). We observed similar time courses of FDQ for H/H/N and H/H/R, as shown for pH 5.0 in Fig. 6. For both constructs, we observed a smaller fusion extent, and in some preparations we found a tendency for slower lipid mixing. The fluorescence increase is caused by the relief of R18 self-quenching due to redistribution of the lipid-like fluorophore between the RBC and CV-1 cell membrane upon fusion of plasma membranes. The normalized data show that in all cases fusion activity was established with about the same kinetics but with some differences in the extents, most probably due to the different expression levels, in particular for H/H/N and H/H/R (Fig. 2; Table 2).

View larger version (66K): [in this window] [in a new window]   FIG. 5.   Kinetics of FDQ of R18-labeled RBCs bound to transfected CV-1 cells at different pHs. Typical time courses of the fusion of CV-1 cells expressing H/H/H or the H/F chimeric constructs with R18-labeled human RBCs at 37°C are shown. The cell-RBC suspension was added to the buffer in the cuvette at the respective pHs as indicated, and fusion monitoring started immediately (t = 0). At t = 590 s, Triton X-100 was added for infinite dilution of the fluorophore. The kinetics were normalized as described by Morris et al. (20). The fluorescence intensity at pH 7.4 was set to 0%, and fluorescence intensity after the addition of Triton X-100 was set to 100% FDQ (for details, see Materials and Methods). The results of two independent experiments with wt HA (H/H/H) are presented to show the experimental variability.

View larger version (49K): [in this window] [in a new window]   FIG. 6.   Kinetics of the FDQ of R18-labeled RBCs bound to H/H/N- and H/H/R-transfected CV-1 cells at different pHs. Typical time courses of fusion of CV-1-cells expressing H/H/N and H/H/R chimeric constructs with R18-labeled human RBCs at 37°C are shown. The experimental procedures were the same as those described for Fig. 5 (note the different scaling). Curve A, H/H/N at pH 5.0; curve B, H/H/R at pH 5.0; curve C, H/H/N and H/H/R at pH 6.7.

View larger version (22K): [in this window] [in a new window]   FIG. 7.   pH dependence of the maximal FDQ of R18-labeled erythrocytes bound to CV-1 cells expressing foreign protein. The kinetics of the FDQ of R18 at different pH values were recorded and normalized as described for Fig. 5. For each studied protein, the maximal FDQ values before the addition of Triton X-100 are plotted against the respective pH values. The inserted graph vac-control in panel B represents the control with CV-1 cells infected by vaccinia virus without any foreign DNA. For this measurement, labeled RBCs were bound to CV-1 cells via WGA. For all probes except the vac-control, the FDQ values were normalized to the respective FDQ maximum of the pH dependence. The maximal FDQ ranged between 30 and 60%.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The use of chimeric constructs offers the opportunity to search for dependencies of protein-induced membrane fusion on specific amino acid sequences and properties of the transmembrane and/or cytoplasmic domain of the fusion protein. We have constructed chimeras of the well-characterized fusion protein of influenza virus, HA, with distinct alterations in its membrane-spanning and cytoplasmic domain, and studied the fusion properties after expression. The domains in question were replaced in a combinatorial fashion by the respective parts of the fusion protein of Sendai virus, F. Although the two proteins are related in their functions, the sequences and physical properties of their transmembrane and cytoplasmic domains differ considerably (Table 1). In addition, we studied two HA chimeras which contained peptides derived from totally unrelated soluble cytoplasmic proteins as their cytoplasmic domains of HA. wt HA and all constructs were expressed on the plasma membrane of the transfected CV-1 cells as trimers and showed the appropriate posttranslational modifications. Serious differences were observed only in the degree of palmitoylation. However, reports from other groups and our previous studies had shown that this hydrophobic modification has no major effect on either membrane fusion or formation of an aqueous fusion pore mediated by influenza virus HA (13, 27). Thus, the constructed chimeras were suitable for studying the potential contribution of the particular domains in question on fusion induction competence in comparison to that of wt HA.

The fusion activity of the presented constructs and wt HA (designated H/H/H) were studied by using fluorescently labeled human RBCs. After the pH of the surrounding medium was reduced from 7.4 to 5.0, all our expression products induced the merging of the unlabeled HA-bearing plasma membrane of CV-1 cells with R18-labeled membranes of human RBCs with a similar time course (Fig. 3, 5, and 6). Furthermore, the fluorescence label calcein redistributed into the lumen of the transfected CV-1 cells after the pH of the surrounding medium was reduced, no matter which of the proteins was expressed. As observed for membrane mixing, the time course of the cytoplasmic mixing triggered by the chimeras was comparable to that induced by wt HA (Fig. 3 and 4), showing that all the chimeric constructs used in this study were able to induce both membrane mixing and formation of an aqueous fusion pore, allowing free movement of cytoplasmic water-soluble molecules as calcein. Thus, for both of these events, the influenza virus HA requires neither its own specific transmembrane and cytoplasmic domain sequences nor any particular length of its CT. It has been reported that a GPI anchor is not sufficient for promoting the formation of fusion pores (15), which indicates that at least the membrane-spanning domain is required. On the other hand, our data suggest that this is due not to a lack of specific transmembrane and/or cytoplasmic amino acid sequences but probably to the complete deletion of the TMR and CT in GPI-HA. The role of the CT in our system for fusion remains open. Recently, it has been shown for HA of another influenza virus subtype (H3) that this domain is not required for membrane mixing and formation of a small aqueous pore (13).

Recent models of HA-mediated fusion emphasize the role of both domains, in particular of the TMR, for the formation and widening of an aqueous fusion pore (35, 38). One of these models describes a lipidic "stalk" as an intermediate which is formed by merging of the contacting outer monolayers of the fusing membranes, the so-called hemifusion diaphragm (4, 38). Destabilization of this diaphragm is necessary for the formation and widening of an aqueous fusion pore as the successful completion of the fusion event. As suggested by Melikyan et al. (17), it is probably the mechanical coupling of the HA ectodomain to the TMR which causes such a destabilization. The diaphragm can be also disturbed by enhancement of the positive spontaneous curvature of the nonfused inner leaflets (reviewed in reference 4). Whatever the detailed molecular mechanism is, our results imply that this mechanical coupling does not require the specific sequence of TMR and CT of wt HA.