The 19-27 amino acid segment of gp51 adopts an amphiphilic structure and plays a key role in the fusion events induced by bovine leukemia virus.

Previous results indicate that the external glycoprotein gp51 of bovine leukemia virus plays an important role in the process of cell fusion induced by bovine leukemia virus (Bruck, C., Mathot, S., Portetelle, D., Berte, C., Franssen, J. D., Herion, P., and Burny, A. (1982) Virology 122, 342-352; Vonèche, V., Portetelle., D., Kettmann, R., Willems, L., Limbach, K., Paoletti, E., Ruysschaert, J. M., Burny, A., and Brasseur, R. (1992) Proc. Natl. Acad. Sci. U. S. A. 89, 3810-3814) and suggest that a region encompassing residues 23 and 25 of gp51 is involved in this process (Portetelle, D., Couez, D., Bruck, C., Kettmann, R., Mammerickx, M., Van der Maaten, M., Brasseur, R., and Burny, A. (1989) Virology 169, 27-33; Mamoun, R., Morisson, M., Rebeyrotte, N., Busetta, B., Couez, D., Kettmann, R., Hospital, M., and Guillemain, B. (1990) J. Virol. 64, 4180-4188). X-ray diffraction studies performed on envelope glycoproteins of influenza virus indicate that the NH2-terminal part of the external glycoprotein lies very close to the fusion peptide. The same overall structure seems to exist in human immunodeficiency virus as suggested by site-directed mutagenesis followed by syncytia induction assays. Our theoretical studies indicate that a segment expanding between residues 19 and 27 of gp51 probably adopts an amphipathic beta-strand structure. We hypothesize that the amphipathic 19-27 structure of gp51 plays an important role in the process of membrane fusion by interacting with the fusion peptide or with another region of gp30. Mutational analysis disrupting the amphipathy of the 19-27 region strongly altered the fusogenic capacity of the gp51-gp30 complex.

BLV is the etiological agent of enzootic bovine leukosis, the most frequent bovine neoplastic disease. The envelope glycoproteins of BLV are derived from a precursor, gp72, through endoproteolytic cleavage. gp51, the external glycosylated component, is responsible for the binding to a cell receptor determining the tropism of BLV. gp30, the transmembrane subunit, anchors the envelope glycoprotein complex into the plasma membrane of virions and infected cells (Burny et al., 1987).
The BLV glycoproteins play a central role in membrane fusion during infection of host cells and syncytia formation; polyclonal sera and some monoclonal antibodies directed against gp51 indeed inhibit syncytia formation (Graves et al., 1981;Bruck et al., 1982aBruck et al., , 1982b. On the other hand, mutational analyses indicate that 1) both gp51 and gp30 are necessary and sufficient for cell fusion, 2) the cleavage of the gp72 precursor into gp51 and gp30 is required for syncytia formation, 3) the hydrophobic NHz-terminal segment of gp30 plays a critical role in the fusion process, and 4) the oblique orientation of this segment in a lipid bilayer parallels the fusogenic capacity of the BLV envelope glycoprotein complex (Vonkche et al., 1992).
These results suggest that the process of membrane fusion depends on at least one region of BLV gp51. In order to identify that putative region of gp51, we compared the antigenicity and amino acid sequence of the envelope glycoproteins of seven BLV natural variants Mamoun et al., 1990). These variants differed by their reactivity to monoclonal antibodies directed against gp51, notably mono H, able to block syncytia formation by the prototype BLV. Two variants exhibited an Hphenotype. Their fusogenic capacity was comparable with that of the wild-type virus, but they were unreactive to the monoclonal antibody mono H. Their gp51 amino acid sequence differed from that of the wild-type protein by 1 amino acid residue located either at position 23 (Phe instead of Ser) or 25 (Ala instead of Ser). It is also of interest to note that, besides its capacity to inhibit syncytia formation, the monoclonal antibody mono H blocks BLV/VSV pseudotype infectivity (Bruck et al., 198213). These data thus suggest that the region of gp51 encompassing resi-The abbreviations used are: BLV, bovine leukemia virus; HIV, human immunodeficiency virus; ELISA, enzyme-linked immunosorbent assay.
15193 dues 23 and 25 plays an important role in the fusion events induced by BLV; residues 23 and 25 are not important for the fusogenic capacity of BLV, but other specific residues of this region are probably critical for the fusion events.
We noticed also that the anti-gp51 monoclonal antibody, mono H, was less efficient than anti-BLV polyclonal sera in syncytia inhibition assays. This suggests that mono H probably does not affect early steps of membrane fusion involving essentially receptor binding, whereas the sera containing polyclonal antibodies inhibit the different steps of the membrane fusion process. Moreover, mono H is not able to inhibit the binding of gp51 to a putative receptor of BLV, whereas antibodies directed against the central part of gp51 prevent it.' On the other hand, the binding site to the host receptor has been identified for the hemagglutinin HA1 of the influenza virus (Wilson et al., 1981;Weis et al., 1988) and for the external glycoprotein gp120 of human immunodeficiency virus (HIV) (Lasky et al., 1987;Kowalski et al., 1987) and localized in the middle part of HA1 and in the COOH-terminal part of gp120. However mutations introduced into the NH2terminal region of HIV gp120 completely abrogated syncytia formation without affecting gp120 binding to the CD4 receptor (Kowalski et al., 1987). More recently, Helseth et al. (1991) showed that residues 36-45 of gp120 of HIV-1 contribute to noncovalent association with the transmembrane glycoprotein gp41. By analogy to other enveloped viruses, such as influenza and HIV, the NH2-terminal segment of gp51 probably does not interact with a host cell receptor and seems to be involved in post-binding events.
In this paper, we modified the predicted amphipathic structure of the amino acid stretch 19-27 of gp51 via mutations at selected amino acid positions. We hypothesized that the role of the amphipathic structure could possibly be to interact through its charged or its hydrophobic face with a given region of gp30. The data collected, indeed, indicate that amphiphilicity of the 19-27 peptide segment must be preserved in order to allow membrane fusion to occur. As x-ray diffraction studies of HA1 locate the NHz end of the external membrane glycoprotein in the vicinity of the fusion peptide of HA2, we tentatively conclude that the same overall configuration exists in BLV glycoproteins and that the hydrophobic side of peptide 19-27 interacts with the hydrophobic fusion peptide.

EXPERIMENTAL PROCEDURES
Cells and Viruses-The transformed cat cell line (CC81) was described previously by Fischinger et al. (1974); CV1 and VERO cell lines were listed by ATCC: CCL 70 and CCL 81, respectively.
Vaccinia ENV and MU6 recombinants expressing wild-type and mutant BLV envelope glycoproteins were derived from the Elstree strain and were obtained as described previously (Vonkhe et al., 1992). The wild-type vaccinia virus (Elstree strain) was used as control.
Enzyme-linked Immunosorbent Assays (ELISAS)-A "sandwich" ELISA was performed for quantitation of gp51. Ninety-six multiwell plates were coated with anti-gp51 monoclonal antibody (mono E: 300 ng/well; Immunoplate I-Nunc), incubated at 4 "C for 3 h, washed with phosphate-buffered saline, 0.2% Tween 80, and saturated with bovine serum albumin 2% and 4% Tween 80 in phosphate-buffered saline. Three-fold serial dilutions of cells and supernatants (100 pl) were added and incubated overnight at 4 "C. After washing, gp51 was detected by addition of a mixture of anti-gp51 monoclonal antibodies  (Meulemans et al., 1978). The test was also performed on nonfixed cells. Western Blot-Lysates of recombinant vaccinia virus-infected cells were subjected to sodium dodecyl sulfate, 12.5% polyacrylamide gel electrophoresis, blotted, and detected using a panel of anti-gp51 monoclonal antibodies (mono A, B, B', D, D', E) (Portetelle et al., 1989). Syncytia Formation Assay-Two methods were used (a) VERO cells, grown in 96-well plates (10,000 cells/well), were infected with recombinant vaccinia virus at a multiplicity of infection of 1 plaqueforming unit/cell and co-cultivated with CC81 cells (40,000 cells/ well) 6 h after infection. Twenty-four hours after infection, the cells were washed and fixed. Syncytia were counted microscopically in six different fields at a magnification of X 400. Fifty cells were counted per field, and multinucleated cells containing more than four nuclei were considered as a syncytium; (b) CV1 cells (50,000 cells/well) were infected with recombinant vaccinia virus at a multiplicity of infection of 1 plaque-forming unit/cell. Twenty-four hours later, the cells were treated as described in a.

Theoretical Analysis of the Amino-terminal Part of gp51-
Theoretical studies were performed on the NHz-terminal part of gp51 of a BLV variant characterized by a H' phenotype (a variant recognized by the anti-gp51 monoclonal antibody mono H) in order to predict possible secondary structures adopted by the peptide region encompassing amino acid positions 23 and 25. A peptide stretch expanding between amino acids 19 and 27 could adopt a @-strand structure as indicated by the prediction methods of Gamier (1978) and Chou-Fasman (1978). The hydrophobic cluster analysis (HCA) method (Lemesle-Varloot et al., 1990) also predicts an amphipathic @-strand in this region but does not exclude other two-dimensional structures, considering that similar hydrophobic cluster analysis profiles correspond to helical regions of HA2 influenza hemagglutinin (Wiley and Skehel, 1987): Possible Role of the 19-27 Segment of gp51 in Cell Fuswn-X-ray diffraction studies of the influenza hemagglutinin established that the NHz-terminal region of HA1 (the external envelope component) lies very close to the fusogenic peptide, the NHz-terminal segment of HA2 (the transmembrane envelope subunit) (Wilson et al., 1981;Weis et aL, 1988;Wiley and Skehel, 1987). Moreover, the ectodomain of HA generated by bromelaine digestion can be cleaved by trypsin into a soluble fragment and an aggregating complex consisting of residues 1-27 of HA1 and 1-175 of HA2 (Skehel et al., 1982;Weis et al., 1990). On the other hand, molecular modeling of the BLV envelope glycoproteins using HA1 and HA2 as reference structures shows that here also the NHz-terminal region of gp51 can fold in the vicinity of the fusion peptide (Busetta, 1989;Mamoun et al., 1990).
The fusion peptide of BLV probably adopts a @-strand structure in solution as it was demonstrated by infrared spectroscopy for the fusion peptide of simian immunodeficiency virus (Martin et al., 1991) and is characterized by relative amphipathy as defined by the position of the most hydrophobic residues on one face and the hydrophilic and less hydrophobic residues on the other face. Fig. 1 illustrates two possible modes of interaction between two amphipathic @strands such as the fusogenic peptide and the 19-27 segment of gp51; the peptides interact via their hydrophobic side only or via their hydrophobic faces and their hydrophilic faces. In both situations, the amphipathic structure of the @-strands is I. Callebaut, manuscript in preparation. involved in the stability of the folding. We cannot, however, exclude a second hypothesis, suggesting interaction to occur between the 19-27 segment and other amphipathic structures of the external part of gp30.
Mutagenesis of the 19-27 Segment of gp51-In order to test the biological relevance of these hypotheses, single amino acid changes were introduced into the 19-27 region of gp51. Amino acids 22 (Phe), 24 (Ile), and 26 (Ile) were replaced by His, Gln, and Gln, respectively, in order to disrupt the amphipathy of this region. Disruption of the amphipathic character of the original structure could affect its possible interaction with the fusion peptide or with another region of gp30 involved in the interaction with gp51 and thereby alter the fusogenic activity of the gp51-gp30 complex.
Mutant and wild-type BLV envelope glycoproteins were expressed using vaccinia recombinants: recombinant ENV expressed wild-type gp51 and gp30 and recombinant MU6 expressed mutated gp51 and wild-type gp30.
Analysis of Mutated gp51-Level of expression and correct processing of the viral antigens were monitored by Western blot (Fig. 2). It appeared that cells infected with ENV or MU6 recombinant expressed gp51 and gp51-related products (precursor and degradation products) in comparable amounts, indicating that mutations introduced in the NH2-terminal region of gp51 did not affect expression and processing of BLV glycoproteins.
To determine whether gp51 was expressed at the cell surface, indirect immunofluorescence assays were performed using a mixture of anti-gp51 monoclonal antibodies (Fig. 3A). This analysis showed that cells infected by vaccinia recombinant ENV or MU6 expressed comparable amounts of viral antigens at the surface, suggesting that the intracellular transport of gp51 to the plasma membrane was not affected by the mutations introduced into the 19-27 region of gp51.
The antigenic reactivity of gp51 produced in cells infected with the ENV or MU6 recombinants was examined in an ELISA test using monoclonal antibodies directed against the biologically active and conformational epitopes F, G , and H $ $ h 3 FIG. 2. Western blot analysis of BLV envelope glycoproteins. CV1 cells were infected with vaccinia recombinants; twentyfour h after infection cell lysates were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis, blotted, and incubated with a mixture of monoclonal antibodies directed against gp51 (Bruck et aL, 1982a) as described previously (Portetelle et al., 1989). WT, wild-type vaccinia virus; CVI, uninfected CV1 cells; ENV, vaccinia recombinant producing wild-type gp51 and gp30; MU6, vaccinia recombinant producing mutated gp51 and wild-type gp30. Position of gp51 glycoprotein is indicated in the left margin. (Fig. 4). The reactivity of the monoclonals, including mono H, was not altered.
The amounts of gp51 expressed by recombinants ENV and MU6 were quantitatively compared by ELISA assay using the same mono F, mono G, and mono H antibodies (Fig. 4) and found to be similar.
Fusogenic Capacity of the Mutated Envelope Glycoproteins of BLV-The MU6 recombinant was then tested for its fusogenic capacity in two cell culture systems (Fig. 3, B and C).
Briefly, VERO cells were infected with recombinants ENV or MU6 expressing equal amounts of gp51 as estimated by ELISA and co-cultivated with CC81 cells, indicators of syncytia formation. In the second system, CV1 cells were infected with the recombinants. Twenty-four hours after infection, syncytia were counted. These experiments revealed that the MU6 recombinant induced syncytia with a significantly lower efficiency (35%) as compared with the ENV recombinant (loo%), indicating that the mutations introduced in the 19-27 region of gp51 indeed altered the fusogenic activity of the BLV envelope glycoproteins.
Reports from several groups have emphasized the role of various regions of the external glycoprotein of enveloped viruses in the fusion events. In addition to binding domains necessary for the interaction with the CD4 receptor (Lasky et al., 1987), the glycoprotein gp120 of HIV-1 contains regions crucial for membrane fusion but not involved in CD4 binding (Kowalski et al., 1987;Skinner et al., 1988;Ho et al., 1988;Willey et al., 1988).
Here we investigated the possible role of the amino-terminal region of gp51 in the cell fusion induced by BLV. This region was supposed to be part of the neutralizable conformational epitope H as variants characterized by the Hphenotype contain point mutations at amino acid positions 23 and 25. Using various algorithms we concluded that the  (Bruck et d., 1982a) and fluoresceinated goat anti-mouse serum (magnification X 400) as described previously (Meulemans et QL, 1978). B, syncytia induced on monolayers of CV1 cells infected by vaccinia recombinants (magnification X 400) as described previously (VonBche et d., 1992). C, the fusogenic capacity of the vaccinia virus recombinants was expressed as a percentage of the number of syncytia obtained for ENV, which was arbitrarily recorded as 100%. Data represent the mean of four independent assays; the standard deviation is less than 5%. WT, wild-type vaccinia virus; CVI, uninfected CV1 cells; ENV, vaccinia recombinant expressing wild-type gp51 and gp30; MU6, vaccinia recombinant expressing mutated gp51 and wild-type gp30.

FIG. 4. ELISA of recombinant vaccinia virus-infected CV1
cell lysates using monoclonal antibodies directed against conformational epitopes of gp51 (F, G, H) as described by Bruck et al. (1984). Results are expressed in terms of the maximal optical density observed. W, mono F; 0, mono G; 0, mono H. peptide segment between amino acids 19 and 27 most probably adopted an amphipathic P-sheet structure that could interact with the fusogenic peptide of BLV gp30 or other amphipathic structures.
In order to test this hypothesis, gp51 was mutated within the 19-27 segment and expressed using the vaccinia recombinant MU6. The mutated complex gp51-gp30 produced by MU6 displayed a strongly reduced fusogenic capacity relative to the wild-type complex produced by recombinant ENV, demonstrating the crucial role of the 19-27 region of gp51 in the fusogenic process.
The antigenic structure of gp51 produced by the recombinants ENV and MU6 was examined the three monoclonal antibodies, mono F, mono G, and mono H, displayed similar reactivity with the corresponding epitopes present on mutant and wild-type gp51. This suggested that the amino acid substitutions introduced in the 19-27 peptide segment of gp51 did not affect the mono H reactivity, although this region was supposed to contain or be part of the H epitope.
Amino acid changes characterizing the natural H-variant affect the hydrophilic face of the amphipathic 19-27 structure of gp51 ( SerZ3 and Ser 25 replaced by Phe and Ala, respectively (Mamoun et al., 1990)). These mutations prevent the recognition of gp51 by the monoclonal antibody, mono H (Portetelle et al., 1989), but do not alter the fusogenic capacity of the H-variant (data not shown). Here we introduced amino acid substitutions on the hydrophobic face of the amphipathic 19-27 segment (Phe", IleZ4, and IleZ6 replaced by His, Gln, and Gln, respectively). These modifications diminished the fusogenic activity of the gp51-gp30 complex but did not affect the recognition of gp51 by mono H. These observations suggest that one face of the amphipathic 19-27 segment influences the recognition of gp51 by mono H, whereas the other face is involved in the process of cell fusion.
Although we may not completely exclude a possible role of this region in receptor binding, the reduced fusogenic capacity obtained for the mutated gp51-gp30 complex agrees with our hypothesis concerning the interaction of the NHz-terminal region of gp51 with the fusogenic peptide or with another region of gp30.
In a multimeric model, three or four fusogenic peptides could be surrounded by three or four amino-terminal domains of gp51, as a sword in a sheath, and thus be isolated from the aqueous environment. This is in agreement with the model proposed by White et al. (1983,1987) for the post-binding events in the case of the hemagglutinin HA of influenza. After internalization of the virus in endocytotic vesicles, acidic pH induces conformational change in HA leading to fusion between viral and endosomal membrane; after modification of interaction between HA1 and HA2, the trimer of HA1 opens as a trilobated flower, drawing the fusogenic peptide (NHzterminal segment of HA2) until it inserts into the endosomal membrane and induces fusion. Except for the drop in pH, the mechanistic model proposed for HA1-HA2 probably holds for gp51-gp30 (Burny et al., 1988).