Individual N-Glycans Added at Intervals along the Stalk of the Nipah Virus G Protein Prevent Fusion but Do Not Block the Interaction with the Homologous F Protein

Cells.

Vero and PK13 (ephrinB2- and -B3-deficient) cells were obtained from the American Type Culture Collection (Manassas, VA). Vero and BHK-21F cells (gift of Rebecca Dutch) were maintained in Dulbecco's modified Eagle medium (DMEM) with high glucose, supplemented with 5% fetal calf serum, 20 mM l-glutamine, 4 U/ml of penicillin, and 4 μg/ml of streptomycin. PK13 cells were maintained in the same medium except for the use of 10% fetal calf serum and 1 mM sodium pyruvate. 293T cells (gift of Abraham Brass) were maintained in high-glucose DMEM supplemented with 10% fetal calf serum, 0.1 mM nonessential amino acids, 20 mM l-glutamine, 4 U/ml of penicillin, and 4 μg/ml of streptomycin.

Recombinant plasmids and site-directed mutagenesis.

The preparation of pCAGGS expression vectors for the NiV G and F proteins has been described previously (32). Mutations were introduced into the NiV G gene in pBluescript SK(+) (Stratagene Cloning Systems, La Jolla, CA) using the QuikChange site-directed mutagenesis kit (Stratagene), and subsequently, the mutated genes were transferred into pCAGGS by blunt-end ligation. The presence of the desired mutation was confirmed by sequencing.

Construction of the 188G-HN124 chimeric attachment protein gene.

A chimeric attachment protein having an N-terminal segment composed of 188 NiV G-derived residues and a C-terminal segment beginning at NDV HN residue 124 (chimera 188G-HN124) was constructed in pBluescript SK(+), facilitated by the introduction of HindIII sites at the desired positions in both NDV HN and NiV G. The sequence was subsequently corrected such that the transition from G residue 188 to HN residue 124 was seamless in the chimera.

Transfections and quantitation of cell surface expression.

For most experiments, cells were seeded in six-well plates at 2 × 10⁵/well 1 day prior to transfection. Wild-type (wt) and mutant proteins were expressed using the Lipofectamine 2000 transfection reagent (Invitrogen Corp., Carlsbad, CA) and 1 μg of each DNA per well according to protocols provided by the company. All assays were performed at 48 h posttransfection except staining for fusion by the chimera, which was also performed at 24 h. For fusion staining at 24 h posttransfection, Vero and BHK-21F cells were seeded at 3 × 10⁵/well. For fusion staining at 24 h and 48 h posttransfection, 293T cells were seeded at 4 × 10⁵/well and at 3 × 10⁵/well, respectively, using plates that were pretreated with 0.1 mM polylysine, rinsed with water, and allowed to dry before plating.

Cell surface expression of wt and mutated G proteins in Vero cells was quantified by flow cytometry (performed by the University of Massachusetts Medical School Flow Cytometry Core Laboratory), using a mixture of conformation-dependent G-specific monoclonal antibodies (MAbs) (33). A mixture of conformational HN-specific MAbs (34–37) was used for the G-HN chimera. Secondary antibodies (Alexa Fluor) were obtained from Invitrogen (Eugene, OR) or KPL (Gaithersburg, MD).

EphrinB2 binding assay.

The ability of HN-G chimeras to bind ephrinB2 was determined by a modification of the procedure described by Negrete et al. (15). PK13 cells were transfected as described above. The medium was removed, and the monolayers were incubated for 1 h at room temperature with 2 μg of soluble ephrinB2-human Fc protein (ephrinB2-Fc) (R&D Systems, Minneapolis, MN). Binding of ephrinB2 was quantified by flow cytometry.

Receptor binding enhancement (RBE) assay.

The effect of receptor binding on the recognition of the mutated G proteins by Mab45 was determined by preincubating a monolayer of PK13 cells expressing the chimeras with and without 10 nM soluble ephrinB2 and then quantifying antibody binding by flow cytometry as described in reference 33.

Hemadsorption (HAd) and NA assays.

The receptor binding activity of the chimera was assayed by its ability to adsorb guinea pig erythrocytes (Bio-Link Laboratories, Liverpool, NY) (23). The NA activity of the chimera was determined with 2-(4-methylumbelliferyl)-α-d-N-acetylneuraminic acid (MUN) as the substrate, as described by Tappert et al. (38).

Staining for fusion.

For pictures of fusion, transfected monolayers were fixed with methanol and stained with Giemsa stain either 24 h or 48 h posttransfection (Sigma Chemical Co., St. Louis, MO).

Content-mixing assay for fusion.

The ability of the mutated G proteins or the 188G-HN124 chimera to complement NiV F in the promotion of cell-cell fusion was quantified using a modification of a content-mixing assay (32), which measures β-galactosidase activity in target cells following fusion induced by the glycoprotein-expressing effector cells. Effector Vero cells were transfected with 1 μg each of wt or mutant DNA and the NiV F DNA, as well as 1 μg of pCAGT7 DNA (39). The following day, another set of Vero cell monolayers (target) was infected with wt vaccinia virus (multiplicity of infection [MOI] of 1) and transfected with 1 μg of pG1NT7β-gal (40). Five hours later, the cells were trypsinized, and equal numbers of the two cell populations were combined and incubated overnight. The next day, the extent of fusion was quantified colorimetrically.

Immunoprecipitation.

At 44 h posttransfection, transfected Vero cell monolayers were starved for 1 h at 37°C in medium lacking cysteine and methionine, radiolabeled for 3 h with 1 ml of medium containing 100 μCi of Express protein labeling mix ([³⁵S]Cys-Met) (PerkinElmer, Boston, MA) and chased for 90 min with medium (41). Cells were lysed and proteins were immunoprecipitated as described previously (32), using a G-specific polyclonal serum. Peptide-N-glycosidase F (PNGase F) digestion was performed as described previously (31). Proteins were separated by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) in the presence or absence of β-mercaptoethanol (BME). Rainbow markers were obtained from GE Healthcare Biosciences Corp. (Piscataway, NJ).

Coimmunoprecipitation of NiV G and F.

The NiV F and wt or mutant G proteins were pseudotyped onto a reporter vesicular stomatitis virus (VSV) expressing the Renilla Luc gene as described previously (42, 43). NiV/VSV-rLuc virions containing wt NiV F and wt or mutant G were lysed and subjected to immunoprecipitation, as previously described (43, 44), using a mixture of conformational G-specific antibodies. Coimmunoprecipitated proteins were analyzed by Western blotting and quantified using a Li-cor Odyssey fluorimager (Li-cor Biosciences, Lincoln, NE).

Introduction of potential N-linked glycosyation sites in the NiV G stalk.

To probe the role of the stalk of the NiV G protein in mediating the interaction with the homotypic F protein, we introduced individual potential N-linked glycosylation sites at intervals along the stalk (Fig. 1). The stalk of NiV G spans a total of 118 residues, encompassing residues 71 to 188. However, the F-interactive competence of the NiV G deletion mutant lacking stalk residues 146 to 182 (28) enabled us to focus on the remaining, N-terminal region.

N-linked carbohydrates are covalently attached to asparagines on nascent polypeptides at the motif N-X-T/S, where X is any amino acid except aspartic acid or proline. We have introduced potential N-linked glycosylation sites at convenient intervals in the segment spanning residues 75 to 133, using the mutagenesis strategy shown in Table 1 and naming the mutant according to the position that would potentially be glycosylated. This results in the potential for the addition of an N-glycan at residue 75, 79, 85, 108, 115, 123, or 133, with only the N85 mutant requiring more than a single point mutation. As it was subsequently demonstrated that the site at residue 75 is not utilized, we also attempted to add a site at the nearby residue 79 via an A81S mutation (mutant N79).

Cell surface expression of the mutated proteins and usage of the added glycosylation sites.

Prior to functional analysis of the mutants, expression at the surface of Vero cells was quantified by flow cytometry (Fig. 2). While two of the more N-terminal glycosylation mutants, N75 and N85, exhibited reduced cell surface expression levels of 57 and 63% of the wt, respectively, mutants N79, N108, N115, N123, and N133 were all expressed at the surface at levels comparable to that of wt G, ranging from 83 to 115% of the wt (Fig. 2). This verifies that the potential N-linked glycosylation mutants are expressed and likely properly folded.

To determine whether the potential glycosylation sites are actually utilized, the mutants were expressed at the surface of Vero cells, radiolabeled, and immunoprecipitated with a G-specific polyclonal serum followed by SDS-PAGE under reducing conditions (Fig. 3A). All of the mutated G proteins, except the two most N-terminal ones, N75 and N79, migrated at a lower rate in the gel than did wt G, suggesting that the potential N-linked glycosylation site in each of these mutants was, indeed, utilized. However, the added sites in mutants N75 and N79 were apparently not glycosylated, as their migration rates were indistinguishable from that of wt G (Fig. 3A).

To confirm that the lower migration rate of the putative glycosylation mutants was, in fact, due to the addition of an N-glycan, the immunoprecipitated proteins were treated with PNGase F, which cleaves the N-glycan linkage between the asparagine side group and the carbohydrate (45). After digestion with the enzyme, all the slower-migrating mutants comigrated with wt G treated in the same way, confirming that the lower migration rate of these mutants was due to a difference in N-linked glycosylation (Fig. 3B). Thus, we have successfully added N-glycans at positions 85, 108, 115, 123, and 133 along the stalk of NiV G.

Some of the added N-glycans compromise fusion promotion without affecting receptor-binding activity.

The effects of the added N-glycans on the ability of NiV G to complement F in the promotion of fusion were examined by a content-mixing assay following coexpression with NiV F in Vero cells (Fig. 4). Four of the mutants, N108, N115, N123, and N133, were unable to promote a detectable level of fusion. Actually, of the five overglycosylated mutants, only N85 promoted detectable fusion, at approximately 20% of the wt level (Fig. 4). This is analogous to previous findings with NDV and PIV5 HN, which showed that addition of an N-glycan at any of several positions along the stalks of those proteins eliminated fusion-promoting activity (30, 31). Not surprisingly, mutants N75 and N79, in which the added glycosylation site was not utilized, promoted fusion quite efficiently, at 75 and 78% of the wt level, respectively (Fig. 4). This suggests that the introduced point mutations themselves, R75N and A81S, affect fusion only minimally.

To be certain that the overglycosylated mutants were truly nonfusogenic, we compared their fusion-promoting activities to that of the wt G protein at a later time posttransfection. As shown in Fig. 5, at 48 h posttransfection, the wt G-promoted fusion is so extensive that the monolayer is almost destroyed. Similar results are obtained with the unglycosylated N75 and N79 mutants. Small syncytia are seen with the N85 mutant, consistent with the low level of fusion detected in the content-mixing assay (Fig. 4). However, even at this late time point, no syncytia are visible for either the N108, N115, N123, or N133 mutant.

Since the initial event in the fusion-triggering cascade in NiV G is thought to be binding to receptors, the possibility exists that the fusion deficiency of the glycosylation mutants could be the result of an effect on this function of G. To examine this possibility, we expressed the mutated proteins in PK13 cells, which are devoid of NiV receptors, and quantified the ability of the proteins expressed at the cell surface to bind soluble ephrinB2. As shown in Fig. 4, all of the mutants retain at least 90% of the ephrinB2-binding activity of the wt G protein, with the exception of the unglycosylated N79 mutant, which still exhibits 71% of wt activity. Indeed, some mutants, e.g., N108 and N123, exhibit more than a 20% increase in binding activity relative to that of the wt protein. Thus, the fusion deficiency of the glycosylation mutants is not the result of a defect in receptor binding.

The NiV G stalk N-glycan mutants oligomerize at an efficiency comparable to that of the wt protein.

The oligomeric structure of the stalk glycosylation mutants was examined by immunoprecipitation of the radiolabeled G protein that had been chased to the cell surface, using a G-specific polyclonal antiserum followed by SDS-PAGE under nonreducing conditions. The NiV G protein migrates in the gel predominantly as a mixture of disulfide-linked dimers and tetramers, with a smaller amount of monomer (Fig. 6). The intermolecular disulfide bonds responsible for these oligomers are mediated by cysteines at positions 146, 158, and 162 in the stalk region of NiV G (28).

As shown in Fig. 6, all of the mutants, including those with added N-glycans, are capable of forming both dimers and tetramers at the cell surface to an extent comparable to that of the wt G protein, although the N85 and N133 mutants appear to exhibit slightly reduced amounts of tetramers. Nonetheless, these data indicate that the fusion deficiency exhibited by the stalk N-glycan mutants is not the result of altered oligomerization. It is also noteworthy that unlike wt G and the other mutants, the N133 mutant fails to exhibit a doublet monomer band (Fig. 6). The significance of this is unclear.

The fusion-deficient G stalk glycosylation mutants retain the ability to interact with the F protein.

It has previously been demonstrated that the fusion deficiencies of both NDV HN (31) and MV H (26) stalk N-glycan mutants correlate with an interference with the ability of the protein to interact with its cognate F protein. This is even true for NDV HN N-glycans added at sites in the stalk that are distant from the putative F-interactive site (23, 31).

To determine whether the fusion-deficient phenotypes of the NiV G stalk glycosylation mutants similarly correlate with a block in the interaction with the homologous F protein, we have determined the ability of the NiV G mutants to interact with F in VSV virions pseudotyped with wt NiV F and either wt or mutant NiV G (43). We coimmunoprecipitated NiV F_o and F₁ from viral lysates using anti-G antibodies (Fig. 7). As an additional control, we included the N75 mutant, in which the added glycosylation site is not utilized. As expected from the flow cytometric data (Fig. 2) and the SDS-PAGE analyses (Fig. 3 and 6), each of the G mutants was immunoprecipitated at an efficiency comparable to that of the wt protein (G IP [Fig. 7]). Also, the efficiencies of MAb-mediated immunoprecipitation of the wt F protein from the respective virions were comparable whether they were coexpressed with wt G, the unglycosylated mutant, or the glycosylated mutants (F IP [Fig. 7]). Interestingly, all the G proteins retained the ability to interact with F (F co-IP [Fig. 7]). This includes all mutants, including N85, N108, N115, N123, and N133, that exhibited severely compromised or undetectable fusion. Indeed, only the N85 mutant seemed to exhibit a discernible slight reduction in the level of coimmunoprecipitation. The lack of F in the F-only coimmunoprecipitation (Fig. 7, last lane) indicates that the coimmunoprecipitation of the F proteins through their interaction with G is specific. Thus, these data indicate that there is no correlation between the fusion deficiency of an overglycosylated G stalk mutant and an effect on the ability of the protein to interact with the F protein.

The fusion deficiency of the G stalk N-glycan mutants correlates with a loss of the RBE phenotype of an MAb that appears to detect a step in the triggering cascade in NiV G.

Mab45 is a conformational, anti-G MAb whose binding inhibits virus entry and is enhanced upon receptor engagement (33). The receptor binding enhancement (RBE) for this MAb is predicted to reflect a required step in the fusion-triggering cascade in NiV G. Based on a deletion mutant analysis, the epitope recognized is thought to reside near the base of the globular domain quite distant from the receptor binding site (33).

To determine whether the Mab45 RBE is affected by the overglycosylation of the stalk of G, we used flow cytometry to compare the extents of antibody binding to wt or mutated G expressed at the surface of PK13 cells in the presence of 10 nM soluble ephrinB2 to that in the absence of receptor (Fig. 8). As expected, the wt G protein (190% RBE) as well as the unglycosylated N75 (177% RBE) and N79 (173% RBE) mutants exhibited a nearly 2-fold enhancement of Mab45 binding in the presence of receptor relative to that in its absence. Indeed, even the N85 mutant, which is only relatively weakly fusogenic (19% of the value for wt G), retained the Mab45 RBE (177% of the value for the wt). These values approach the 2.5-fold RBE originally reported for NiV G (33).

However, each of the overglycosylated, fusion-deficient stalk mutants failed to exhibit the Mab45 RBE. Indeed, the N-glycans at positions 108 and 123 actually result in a significant receptor-induced decrease in the binding of Mab45, with antibody binding reduced to 49 and 64%, respectively, in the presence of ephrinB2 compared to that in its absence (Fig. 8). Even the N115 and N133 glycosylation mutants exhibit no significant Mab45 RBE (99 and 87% of binding, respectively, with receptor relative to that in its absence) (Fig. 8). Thus, the fusion deficiency of all of the glycosylation mutants correlates completely with a loss of the Mab45 RBE phenotype.

A chimeric attachment protein with a complete NiV G stalk and intact NDV HN head triggers NiV F for fusion only minimally.

The properties of the NiV G N-glycan stalk mutants are consistent with our previous characterization of chimeras having NiV G-derived stalks and NDV HN-derived heads (G-HN chimeras) (32). Whereas several chimeras having NDV HN-derived stalks and NiV G-derived heads (HN-G chimeras) are capable of efficiently complementing the NDV F protein for fusion promotion (32), the reciprocal was not true. Attachment proteins having NiV G-derived N-terminal segments of 144 and 166 amino acids and a complete, functional HN-derived head were expressed and bound receptor, but they did not trigger NiV F to a detectable extent in a content-mixing assay (32). This lack of fusion was observed in BHK-21F, Vero, and 293T cells. We concluded from this that the G triggering cascade may be more complex than that of HN (32).

Subsequently, it was reported by another group that a chimera with a 188-amino-acid G-derived N-terminal segment and a complete HN head beginning at HN residue 124 (chimera 188G-HN124, using our nomenclature) could trigger NiV F-mediated fusion of erythrocytes, but only if the NA activity of the chimera was inhibited with zanamivir (46). It was concluded that the inhibitor was required to enable the protein to stay attached to receptors for a longer period. This finding also raised the possibility that NiV G residues 183 to 188 might somehow be critical for fusion triggering (46).

To address these findings, we prepared the 188G-HN124 chimera and evaluated its functional properties. The chimera exhibits HAd activity (38% ± 9% of wt HN) commensurate with its cell surface expression (42% ± 5% of wt HN), indicating that receptor binding activity is unaffected. However, in the reporter gene content-mixing assay, it did not trigger a detectable level of NiV F-mediated fusion of BHK-21F cells either with or without treatment with 5 mM zanamivir (Fig. 9). In Vero cells, only extremely weak triggering of NiV F was detected with this assay (4% ± 2% of wt G), and it was actually decreased even further with zanamivir treatment (1% ± 1% of wt G) (Fig. 9). Thus, in our hands, the 188G-HN124 chimera was not capable of triggering NiV F-induced fusion of either BHK-21F or Vero cells to a significant extent in a quantitative content-mixing assay, even with zanamivir treatment.

The lack of an effect of zanamivir in two cell types is consistent with the demonstration that the chimera has only 2% ± 1% of the NA activity of wt NDV HN. Thus, even after correcting for somewhat reduced cell surface expression, the chimera has <10% of wt NA activity. This is somewhat expected because the NA activity of NDV HN is known to be hypersensitive to even point mutations in the stalk (23, 47, 48), which may be related to the cooperative substrate saturation kinetics exhibited by the NA of this strain of NDV (49). Given this hypersensitivity, it is not surprising that replacement of the entire stalk would modulate NA activity.

Since the same group subsequently reported that a 186G-HN124 chimera is capable of triggering NiV F for fusion about 1/3 as effectively as wt G, with no mention of the need for zanamivir treatment (50), we wondered whether the promotion of fusion by the 188G-HN124 chimera could be detected at later times posttransfection. Thus, the extents of syncytium formation induced by the chimera and by wt G were compared at 24 h and 48 h posttransfection by visualization of stained monolayers. Consistent with the content-mixing assay data, syncytia were not visible at 24 h posttransfection in Vero cell monolayers coexpressing the 188G-HN124 chimera and NiV F (Fig. 10A), while robust fusion was obtained with wt G and F, as well as with NDV HN and F, but not with NDV HN and NiV F. At 48 h posttransfection, some chimera-triggered, NiV F-induced syncytium formation was detected. However, at this time, fusion promotion by NiV G and F (as well as NDV HN and F) was far more robust. Indeed, the fusion in NiV G-F-expressing Vero cells after 48 h was so extensive that the monolayer was almost completely destroyed (Fig. 10A). Similar results were obtained in BHK-21F cells (data not shown). We also obtained weak fusion at 48 h posttransfection in both Vero and BHK-21F cells with a chimera (182G-HN124) (31) having a shorter N-terminal, G-derived segment (data not shown). Thus, the weak syncytium formation in Vero and BHK-21F monolayers coexpressing the G-HN chimera and NiV F is visible only at times when fusion promotion induced by wt NiV G and F is so extensive that the monolayer is essentially destroyed.

Finally, since 293T cells are inherently more susceptible to syncytium formation and these cells were used by the other group (50), we also tested the 188G-HN124 chimera for its ability to trigger NiV F-mediated fusion in these cells. In 293T monolayers, NiV G and F gave extremely robust fusion after only 24 h (Fig. 10B). In these cells, NDV HN and F gave weaker, but clearly visible, syncytium formation at 24 h and much more robust fusion after 48 h (Fig. 10B). However, no syncytium formation was visible in 293T cell monolayers coexpressing the 188G-HN124 chimera and NiV F, even at 48 h posttransfection. They are indistinguishable from cells expressing only NiV F. Thus, we were unable to demonstrate the triggering of NiV F by the 188G-HN124 chimera in three different mammalian cell lines.