Residues of the Human Metapneumovirus Fusion (F) Protein Critical for Its Strain-Related Fusion Phenotype: Implications for the Virus Replication Cycle

Abstract

The paramyxovirus F protein promotes fusion of the viral and cell membranes for virus entry, as well as cell-cell fusion for syncytium formation. Most paramyxovirus F proteins are triggered at neutral pH to initiate membrane fusion. Previous studies, however, demonstrated that human metapneumovirus (hMPV) F proteins are triggered at neutral or acidic pH in transfected cells, depending on the strain origin of the F sequences (S. Herfst et al., J. Virol. 82:8891–8895, 2008). We now report an extensive mutational analysis which identifies four variable residues (294, 296, 396, and 404) as the main determinants of the different syncytial phenotypes found among hMPV F proteins. These residues lie near two conserved histidines (H368 and H435) in a three-dimensional (3D) model of the pretriggered hMPV F trimer. Mutagenesis of H368 and H435 indicates that protonation of these histidines (particularly His435) is a key event to destabilize the hMPV F proteins that require low pH for cell-cell fusion. The syncytial phenotypes were reproduced in cells infected with the corresponding hMPV strains. However, the low-pH dependency for syncytium formation could not be related with a virus entry pathway dependent on an acidic environment. It is postulated that low pH may be acting for some hMPV strains as certain destabilizing mutations found in unusual strains of other paramyxoviruses. In any case, the results presented here and those reported by Schowalter et al. (J. Virol. 83:1511–1522, 2009) highlight the relevance of certain residues in the linker region and domain II of the pretriggered hMPV F protein for the process of membrane fusion.

INTRODUCTION

Human metapneumovirus (hMPV) was first isolated in 2001 from respiratory specimens obtained from children in The Netherlands (40) and was soon recognized as a frequent respiratory pathogen of children, immunocompromised individuals, and the elderly worldwide (1, 8, 9, 26, 39). Analysis of sequence homology, gene constellation, and virological data were used to classify hMPV as a member of the family Paramyxoviridae, subfamily Pneumovirinae, genus Metapneumovirus. Clinical signs and symptoms associated with hMPV infection are similar to those caused by the human respiratory syncytial virus (hRSV), the prototype of the genus Pneumovirus within the Pneumovirinae subfamily. Both viruses cause illnesses ranging from mild respiratory distress to bronchiolitis and pneumonia, although incidence and severity are normally higher with hRSV (10, 18, 42). hMPV isolates have been classified into two main genetic lineages (A and B), each divided into two sublineages (A1, A2, B1, and B2) (41).

As other enveloped viruses, hMPV enters host cells by fusion of the viral and cellular membranes, a step promoted by the viral fusion (F) glycoprotein. In addition to virus-cell membrane fusion, most paramyxovirus F proteins also promote cell-cell fusion during the late stages of infection, leading to a characteristic cytopathic effect in which syncytia (multinucleated giant cells) are observed. The cytopathic effect caused by hMPV strains in tissue culture is variable; while some strains induce hRSV-like syncytia, others produce cell rounding and death, almost without syncytium formation (13).

The hMPV F protein is relatively well conserved among viral strains (41, 46) and shares structural features with other paramyxovirus F proteins (7). The paramyxovirus F proteins are homotrimeric type I integral membrane glycoproteins, which are synthesized as biologically inactive precursors (F0) that must be cleaved to become fusion competent. Cleavage takes place at a mono- or multibasic cleavage site immediately upstream of a hydrophobic fusion peptide, generating two polypeptide chains (F2 N-terminally located to F1) that remain covalently linked by at least one disulfide bond. It has been reported that the hMPV F protein has a monobasic cleavage site that can be recognized by trypsin (30, 32) or by TMPRSS2 (35), a transmembrane serine protease present in the human airway epithelium. The ectodomain of paramyxovirus F proteins also contains two conserved heptad repeat (HR) regions, designated HRA and HRB, which are located downstream of the fusion peptide and upstream of the transmembrane (TM) domain, respectively.

hMPV studies and data obtained for related viral fusion proteins suggest that the F protein undergoes discrete/stepwise conformational changes, from a metastable (prefusion) conformation to a lower free-energy (postfusion) structure, during membrane fusion (3, 7, 47, 48). After an initial activation event, the prefusion protein starts refolding into transient unstable intermediates, including the formation of a prehairpin structure in which the fusion peptide is inserted into the target cell membrane. Subsequent conformational changes of this prehairpin involve assembly of the HRA and HRB regions into a six-helix bundle (6HB) that brings the viral and cell membranes into proximity, ending up in fusion of both membranes. Formation of the 6HB and the associated free-energy change are tightly linked to the merger of the viral and cellular membranes and the formation of the so-called fusion pore (5, 22, 29).

The paramyxovirus F proteins are activated by specific triggers, allowing virus-cell fusion to occur “at the right time and at the right place” in the viral replication cycle (20), i.e., when the virus is bound at the cell surface of the target cell. For paramyxoviruses, in general, fusion mediated by the F protein requires cooperation by the homotypic viral attachment protein (G, H, or HN, depending on the virus) that binds to cellular receptors (16, 45). It has been proposed that following this interaction, conformational changes in the attachment protein are transduced to the F protein to trigger virus-cell membrane fusion, which for paramyxoviruses occurs at the cell surface and at neutral pH.

In the case of Pneumovirinae, it is clear that F protein activation does not necessarily rely on cooperation with the attachment protein, since spontaneous mutants of hRSV (17) or genetically engineered recombinants of hMPV (2) or hRSV (37, 38), expressing F as the only surface glycoprotein, can infect cells in culture although they are attenuated in vivo.

The hRSV F protein is unique among paramyxoviruses in having two multibasic cleavage sites (site I, RARR_109, and site II, KKRKRR_136), instead of one, separated by a region of 27 amino acids (pep27) (11). Studies with chimeric proteins and recombinant Sendai viruses, in which the monobasic cleavage site of Sendai virus F protein was replaced by the two cleavage sites of hRSV F, indicated that the presence of two multibasic cleavage sites may be an alternative mechanism to regulate activation of a paramyxovirus fusion protein without cooperation by the attachment protein (27, 28). While this mechanism may be at work in hRSV, the hMPV F protein has a single monobasic cleavage site and, therefore, its activation may not depend on a double-cleavage mechanism.

Schowalter et al. (32) found that cell-cell fusion promoted by the F protein of an hMPV strain, CAN97-83 (sublineage A2), required exposure of the transfected cultures to a pulse of low pH but did not require coexpression of the attachment protein, suggesting an alternative activation mechanism for the hMPV F protein, based on contact with an acidic environment. However, further studies demonstrated that the ability of the hMPV F protein to induce syncytia in transfected cells after a low pH pulse was not a general phenomenon; instead, it was related to the strain from which the F protein was derived (15). Thus, F proteins from lineage B viruses were fusogenic in transfected cells independently of pH, whereas lineage A proteins were either poorly fusogenic or required a short pulse at pH 5.0 to induce syncytia.

In the current study, a search for key variable residues of the hMPV F protein associated with a pH-dependent or -independent phenotype has been carried out. Residues at positions 294, 296, 396, and 404 were found to be the main determinants of the different phenotypes. These amino acids lay in close proximity to two conserved histidines, H368 and H435, in a three-dimensional (3D) model of the hMPV F prefusion conformation, suggesting that histidine protonation at low pH can trigger fusion promoted by certain F proteins. This low-pH dependency correlated with higher thermostability of the virus particle and pH-dependent syncytium formation in infected cultures, but it was unrelated to an acid-dependent mechanism of virus entry. These results, however, support the notion that destabilization of the F protein is required to initiate the process of membrane fusion and that low pH may be a surrogate of the actual trigger for certain hMPV strains, as heat has been found to be an alternative trigger of F proteins of other paramyxoviruses, e.g., Sendai virus (44).

MATERIALS AND METHODS

Cell lines, viruses, and plasmids.

Vero cells, subclone 118 (Vero-118) (19), were grown in Iscove's modified Dulbecco's medium (IMDM; Gibco Invitrogen) buffered with 25 mM HEPES and supplemented with 10% fetal calf serum (FCS) (HyClone; Perbio), penicillin (100 IU/ml), streptomycin (100 μg/ml), and glutamine (4 mM).

hMPV viruses were first isolated on tertiary monkey kidney (tMK) cells as described previously (40). The construction of hMPV recombinant green fluorescent protein (GFP)-expressing viruses rNL/1/00_A1-GFP (prototype of lineage A1) and rNL/1/99_B1-GFP (prototype of lineage B1) has been described previously (6). High-titer stocks of hMPV were produced in Vero-118 cells in infection medium: IMDM supplemented with penicillin (100 U/ml), streptomycin (100 U/ml), glutamine (4 mM), and trypsin (4 μg/ml). After 5 days, cultures were harvested and stored in 25% sucrose at −80°C. Recombinant respiratory syncytial virus of the A2 strain expressing GFP (rhRSV-GFP) (12) was provided by M. Peeples. Vesicular stomatitis virus (VSV) was obtained from L. Carrasco. For rhRSV-GFP and VSV production, trypsin was not included in the infection medium, cells were harvested 2 days after infection, and virus stocks were stored without sucrose.

The F proteins of hMPV strains NL/1/00_A1 (NCBI accession no. AAK62968), NL/17/00_A2 (NCBI AAQ90144), NL/1/99_B1 (NCBI AAQ90145), and NL/1/94_B2 (NCBI AAQ90146) (41) were cloned in the plasmid pCAGGS (24). To clone each F protein gene, the same forward primer, 5′-CGCGAATTCACCATGTCTTGGAAAG-3′, containing a EcoRI site (underlined) was used. Different reverse primers containing a XhoI site (underlined) were used for the four strains: 5′-CGCCTCGAGCTAATTATGTG-3′ for NL/1/00_A1, 5′-CGCCTCGAGCTAACTGTGTG-3′ for NL/17/00_A2, 5′-CGCCTCGAGCTAACTATGTG-3′ for NL/1/99_B1, and 5′-CGCCTCGAGCTAACTATGCG-3′ for NL/1/94_B2. The F genes of NL/1/00_A1 and NL/1/99_B1 were amplified from cloned full-length cDNA copies previously described (14). Reverse transcription (RT)-PCR products of a virus stock were used as PCR templates for the cloning of NL/17/00_A2 and NL/1/94_B2 F genes. For this purpose, viral RNA was isolated from virus stock using the High Pure RNA isolation kit, by following the manufacturer's instructions (Roche). All constructs contained the F sequences reported earlier (41), with the exceptions of one amino acid substitution, T270M, in the NL/17/00_A2 F protein_, and two amino acid substitutions, V231I and Q307R, in the NL/1/94_B2 F protein. hMPV F protein mutants were generated using the QuikChange Multi site-directed mutagenesis kit (Stratagene), as recommended by the manufacturer. Primers and templates used for each mutant are available from the authors upon request. All constructs were sequenced to confirm the presence of the desired mutations and the absence of additional mutations.

Syncytium formation assay.

For transfections, Vero-118 cells growing in 8-well microchambers at 90% confluence were incubated with 1 μg/well of plasmids and FuGENE, in accordance with the manufacturer's instructions (Roche). The transfection mixtures were removed 6 h later, and the cells were incubated in IMDM containing 2.5% FCS. For infections, Vero-118 cells growing in 24-well plates were inoculated at a multiplicity of infection (MOI) of 0.5 PFU/cell with hMPV. Virus was adsorbed for 4 h at 37°C in IMDM without trypsin. After adsorption, the media was replaced, and the cells were incubated for 16 h at 37°C. At 20 h posttransfection or postinoculation, cells were rinsed three times with Dulbecco's phosphate-buffered saline with Ca²⁺ and Mg²⁺ (PBS⁺; Lonza) and treated for 4 h at 37°C with 1 μg/ml trypsin (Sigma-Aldrich) in serum-free medium. Then the cells were rinsed once and incubated for 5 min at 37°C with prewarmed PBS⁺, 10 mM HEPES (Sigma-Aldrich), and 5 mM MES (morpholineethanesulfonic acid) (Sigma-Aldrich) buffered at the indicated pHs. Subsequently, the medium was replaced with IMDM containing 2.5% FCS, and cells were incubated for 2 to 4 h at 37°C to allow cellular rearrangements to occur.

Transfected cells were then washed with Dulbecco's phosphate-buffered saline without Ca²⁺ and Mg²⁺ (PBS⁻; Lonza) and fixed with cold methanol for 10 min, followed by cold acetone for 30 s. Fixed cells were immunostained using a rabbit polyclonal anti-F specific antiserum (15), followed by an anti-rabbit fluorescein-labeled antibody (Sigma-Aldrich). Nuclei were stained for 10 min with 1 ng/ml DAPI (4′,6-diamidino-2-phenylindole) dihydrochloride (Calbiochem) in PBS⁻. Images were taken with a Leica TCS SP5 AOBS confocal microscope with 10× and 40× lenses. To quantify the extent of fusion, a relative syncytial index (RSI) was calculated by counting the number of nuclei in syncytia in several random fields (10×) and then dividing that number by the total number of nuclei (>10³ nuclei), normalized to the same index obtained with transfected cells expressing the F protein of the NL/1/00_A1 strain and exposed to pH 5.0.

The formation of syncytia in infected cells was observed by phase-contrast microscopy or by UV fluorescence in the case of GFP-expressing hMPV recombinants.

Quantitation of hMPV F protein expression at the cell surface by flow cytometry.

Subconfluent monolayers of Vero-118 cells growing in 24-well plates were either transfected with 2 μg/well of plasmids using FuGENE or inoculated with hMPV viruses at an MOI 0.5 PFU/cell. After 20 h of incubation at 37°C, the cells were detached using 1 mM EDTA in PBS⁻. Then, cells were washed with PBS⁻ and incubated with polyclonal anti-F specific antibodies in PBS⁻ containing 2% FCS (15). After washing, cells were stained by subsequent incubations with anti-rabbit immunoglobulin (Amersham Biosciences) and streptavidin-R–phycoerythrin (Southern Biotechnology Associates). Finally, cells were fixed in 1% paraformaldehyde, and fluorescence of 1 × 10⁴ cells was measured using a Becton Dickinson FACSCalibur analyzer with CellQuest software. Data were analyzed using FlowJo software (Tree Star, Inc.). By comparison to a negative control, only positive cells were gated to estimate the geometric mean fluorescence intensity.

Virus temperature stability.

rhMPV-GFP stocks were diluted in IMDM without trypsin and incubated in separate aliquots at the temperatures and pHs indicated in the figure legends. Samples were shifted down to 4°C at different time points and left at this temperature until the end of the longest incubation period. Then the viruses were used to inoculate Vero-118 cells at an MOI of 0.5 PFU/cell as described above. After 20 h, the cells were detached and fixed in 1% paraformaldehyde, and the extent of infection was estimated by measuring GFP expression by flow cytometry. Values obtained were normalized to the value for control (100%) rhMPV-GFP, which was maintained at 4°C all the time.

Drug treatments.

Stock solutions of concanamycin A and bafilomycin A (both manufactured by Sigma-Aldrich) were made in dimethyl sulfoxide (DMSO). Ammonium chloride (Sigma-Aldrich) was dissolved in PBS⁺. Subconfluent monolayers of Vero-118 cells grown in 24-well plates were pretreated for 30 min with the concentrations of each drug indicated in the figure legends. Infections (MOI 0.5 PFU/cell) were carried out in the presence of each drug dissolved in IMDM serum-free medium without trypsin. After 4 h at 37°C, the cells were washed and incubated in fresh medium for an additional 2 h (VSV) or 16 h (rhMPV-GFP, rhRSV-GFP, and hMPV isolates). Then, cells were detached and processed for flow cytometry as described above. For recombinant viruses (rhMPV-GFP and rhRSV-GFP), GFP expression was quantified by flow cytometry, whereas for hMPV strains, the level of hMPV F expression was measured by flow cytometry after incubation with specific antibodies, as mentioned above. In the case of VSV, the level of G protein expression was quantified by flow cytometry after fixation and permeabilization with an IntraStain kit (Dako), followed by incubation with a monoclonal (clone P5D4) anti-VSV G protein antibody conjugated with Cy3 (Sigma-Aldrich).

Sequence alignment and analysis.

Complete or partial F protein amino acid sequences of 398 hMPV strains were retrieved from GenBank and aligned using the ClustalW algorithm found in the BioEdit software package version 7.0.5.2.

hMPV F modeling.

SWISS-MODEL server facilities (http://swissmodel.expasy.org/) were used to generate a model of the hMPV F protein prefusion conformation, built with the atomic coordinates of the preactive structure of the parainfluenza virus 5 F protein (48). The alignment used to create the model is shown in Fig. 1.

Fig. 1.

Alignment of hMPV F sequences from prototype strains and 3D model of hMPV F protein. (A) Alignment of hMPV F protein sequences of the four prototype viruses shown on the left. Only amino acid changes with respect to NL/1/00 (A1 lineage) are shown. Dots indicate identical amino acids. Hyphens indicate gaps. The PIV5 F protein sequence is included for comparison. Structural domains of panel C are highlighted in different colors. Key residues for the syncytial phenotype identified in this study are in bold. (B) Surface representation of the NL/1/00 (A1) hMPV F trimer, modeled in the prefusion conformation using the atomic coordinates of the parainfluenza virus 5 (PIV5) F protein (48). (C) Ribbon diagram of one monomer, with structural domains colored as in panel A. An arrow indicates the cleavage site. (D) Close-up view of the DII domain and HRB linker. Variable residues 294, 296, 396, and 404 (underlined), corresponding to the F_NL/1/00(A1) protein, and conserved histidines 368 and 435 are shown as balls.

RESULTS

Identification of critical residues for low-pH-dependent membrane fusion mediated by hMPV F protein.

It has been reported previously that transfected Vero cells expressing the hMPV F protein of lineage A viruses with Gly at position 294 required a pulse at pH 5.0 to induce syncytia, whereas those having Glu at the same position failed to induce syncytia at either pH 5.0 or pH 7.4 (15, 32). In contrast, F proteins of lineage B viruses with Glu at position 294 induced syncytia at both pHs. Therefore, although residue 294 influenced the low-pH requirement of the F protein for fusion, other sequence differences between lineages A and B should have an impact on the F protein phenotype.

To identify other residues that could influence the syncytial phenotype, a comparison was made of the hMPV F amino acid sequences available in GenBank. Figure 1A shows F protein sequences representative of the four sublineages (A1, A2, B1, and B2) in which hMPV isolates have been classified. The amino acid G294 was found only in the sequence of the A1 sublineage, while the other three proteins had E294. Residue 294 is located in the DII region of a prefusion hMPV F protein model, built with the atomic coordinates of the structure determined for the parainfluenza virus type 5 (PIV5) F protein (Fig. 1B) (48). Two conserved histidines, H368 (DII) and H435 (HRB linker), sit near residue 294 in the model as well as three other residues, 296, 396, and 404, which are variable among hMPV isolates (Fig. 1C and D). We hypothesized that protonation of H368 and/or H435 might be required by certain hMPV F proteins for activity, depending on the local environment. Therefore, we decided to test the influence of the variable amino acids at positions 294, 296, 396, and 404 on hMPV F protein fusion activity.

Initially, the F protein of NL/1/99 of the B1 lineage [F_{NL/1/99 (B1)}], which has a low-pH-independent membrane fusion activity, was selected to test the effect of mutations at the four variable sites mentioned above. Residues of the tetrad ENWP at positions 294, 296, 396, and 404 were replaced stepwise by the amino acids of the tetrad GKRN found in F_NL/1/00(A1), which requires low pH for activity, and each mutant was tested for syncytium formation in transfected cells (Fig. 2A). As reported previously (15), the exchange E294G (tetrad GNWP) impaired syncytium formation by F_NL/1/99(B1) at both pH 5.0 and pH 7.4. Essentially, the same phenotype was maintained when the substitution N296K (tetrad GKWP) was added to the previous change. However, addition of the W396R change (tetrad GKRP) to the two previous substitutions substantially increased the fusogenic activity of the F_NL/1/99(B1) protein, particularly at low pH. Finally, the replacement P404N added to the previous three changes (tetrad GKRN) inhibited syncytium formation slightly, closely reproducing the low-pH requirement of F_NL/1/00(A1) in the F_NL/1/99(B1) protein. Quantification of the number of nuclei in syncytia, expressed as the relative syncytial index (RSI), substantiated the observations made under the microscope, although the effect of P404N (tetrad GKRN) was not statistically significant in the RSI (Fig. 2B). Finally, certain mutants tested in a previously described cell content mixing assay (15) mimicked the results obtained in the syncytium formation assay (not shown).

Fig. 2.

Effects of amino acid substitutions at positions 294, 296, 396, and 404 in the F_NL/1/99(B1) phenotype. (A) Vero-118 cells were transfected with plasmids carrying F_NL/1/99(B1) or mutants derived from it as indicated. For comparison, cells were transfected with a plasmid carrying F_NL/1/00(A1) (two lower right panels). The amino acid tetrads at positions 294, 296, 396, and 404 of the different F proteins are shown in each respective panel. Twenty-four hours after transfection, cells were treated with trypsin and exposed to low pH (pH 5.0, bottom panel) or neutral pH (pH 7.4, top panel) as indicated in Materials and Methods. Subsequently, the medium was replaced and cells were incubated for an additional 2 to 4 h. Finally, cells were fixed and stained with a polyclonal anti-F antibody and DAPI. Representative 40× images collected with a confocal microscope are shown. (B) To measure the extent of fusion, nuclei were counted in several random fields (>10³ nuclei) and a syncytial index relative to F_NL/1/00(A1) exposed to pH 5.0 was calculated as indicated in Materials and Methods. Data are means ± standard deviations (SDs) from three independent experiments. White bars, pH 5.0; shaded bars, pH 7.4.

The same strategy was used to introduce stepwise all the amino acid changes found in the F protein of hMPV strains at positions 294, 296, 396, and 404 in the sequence backbone of the four virus prototypes of Fig. 1A. Each mutant was tested in transfected cells for both cell surface expression by flow cytometry and syncytium formation by confocal microscopy. The results are summarized in Fig. 3A, with some examples illustrated in Fig. 3B. None of the mutations substantially altered the level of F protein fluorescence intensity (Fig. 3A), excluding the possibility that differences in cell-cell fusion may be attributed to the amount of F protein expressed at the cell surface. However, replacement in a given F protein of residues 294, 296, 396, and 404 by those found in another protein changed the “recipient” protein phenotype to that of the “donating” molecule. For instance, replacement of the tetrad GKRN found in the F_NL/1/00(A1) by the tetrad EDRP found in the F_NL/1/94(B2) converted a low-pH-dependent F protein to a pH-independent protein for syncytium formation (Fig. 3A and B). Reciprocally, replacement of the tetrad EDRP of the F_NL/1/94(B2) protein by the tetrad GKRN of the F_NL/1/00(A1) protein converted a pH-independent protein to a low-pH-dependent protein for cell-cell fusion. Since not all possible amino acid combinations of the four different tetrads have been tested (the number of mutants would have been extremely high), it is not possible to conclude which of the four individual amino acids is contributing more significantly to the fusogenic phenotype of each protein. However, partial mutant studies with single-amino-acid changes suggested that it was the tetrad combination rather than individual residues that influenced the fusogenic properties of the different proteins (not shown).

Fig. 3.

Effects on cell surface expression and syncytium formation of exchanges at positions 294, 296, 396, and 404 between hMPV F proteins. (A) Vero-118 cells were transfected with plasmids carrying the genes encoding the indicated hMPV F proteins. The tetrads GKRN, EKRN, ENWP, and EDRP indicate the actual amino acids for each protein at positions 294, 296, 396, and 404. The tetrad found in each original protein is in boldface. Cell surface expression of each F protein was quantified by flow cytometry 24 h after transfection with specific anti-F antibodies. Values of mean fluorescence (FL) intensity were normalized to the fluorescence of F_NL/1/00(A1) GKRN. For the syncytium assay, cells were processed as described in the legend to Fig. 2. The relative syncytial index (RSI) was calculated as described in the legend to Fig. 2, normalized to the value of F_NL/1/00(A1) GKRN at pH 5.0. Values above 0.5 are highlighted in gray. Data are grouped by the F protein used as backbone to generate the different mutants. Data are means ± SDs from three independent experiments. N.D., not determined. (B) Representative confocal microscope images (40×) of transfected cultures expressing the hMPV F proteins labeled with an asterisk in panel A.

Even a poorly fusogenic protein, such as F_NL/17/00(A2), with a low RSI (0.27 at pH 7.4 and 0.23 at pH 5.0) could be converted to either a highly fusogenic pH-independent protein when the tetrad EKRN was replaced by EDRP (RSI, 1.01 and 0.92 at pH 7.4 and pH 5.0, respectively) or to a low-pH-dependent protein when GKRN replaced the original tetrad (RSI, 0.17 and 0.59 at pH 7.4 and pH 5.0, respectively). Independently of the sequence backbone, F proteins with the tetrad EKRN always had low RSI values, although syncytium formation could be increased in these cases with very low pH treatments (pH 3.5 to 4.0) (not shown). It is worth stressing that replacement of the four-amino-acid tetrad in a “recipient” F protein by those from a “donating” strain affected syncytium formation independently of the other 35 variable residues found in the sequences shown in Fig. 1A.

Histidines 368 and 435 also influence the syncytial phenotype.

As mentioned above, the amino acids at positions 294, 296, 396, and 404 sit near two conserved histidines at residues 368 and 435 in a 3D model of the hMPV F prefusion protein. Depending on the actual residues in those four positions, protonation of the two histidines may be of critical importance to induce electrostatic repulsions which would allow initiation of the conformational changes that hMPV F experiences during the process of membrane fusion. To explore this hypothesis, H368 and H435 were mutagenized in F_NL/1/00(A1) and F_NL/1/94(B2), since these two proteins represent the best prototypes of low-pH-dependent and pH-independent phenotypes, respectively (see Fig. 3A). None of the mutations introduced in those two residues substantially altered the level of F protein expression at the surface of transfected cells, except in the case of double mutants generated in F_NL/1/94(B2) that seemed to double that level, as assessed by flow cytometry (Fig. 4A).

Fig. 4.

Influence of histidines 368 and 435 in the syncytial phenotype of hMPV F. (A) Vero-118 cells were transfected with plasmids carrying the genes encoding the indicated hMPV F proteins. The motif at positions 294, 296, 396, and 404 in each F protein is indicated below the name of the hMPV strain. Single and double changes in H368 and H345 are shown for each F protein. Cell surface expression and syncytium formation were tested as described in the legend to Fig. 3. Values are means ± SDs from two independent experiments. Values above 0.5 are highlighted in gray. (B) Representative confocal microscope images (40×) of transfected cultures expressing the hMPV F proteins labeled with an asterisk in panel A.

The results of syncytium formation assays done with the His mutants are summarized in Fig. 4A, with examples of confocal microscopy shown in Fig. 4B. In agreement with previous results reported by Schowalter et al., single mutations of H368 in F_NL/1/00(A1) had generally less effect on the syncytial phenotype than those in H435 (31). For instance, replacement of H368 by negatively charged amino acids (H368D or H368E) had no significant effect on the F protein phenotype, which maintained its low-pH dependence, whereas H435D and H435E mutations abrogated the fusogenicity of F_NL/1/00(A1) at low and neutral pHs (Fig. 4A).

The influence of changes in H368 and H435 was even more evident in double mutants of F_NL/1/00(A1). Thus, although H368A and H435A partially reduced the syncytia at pH 5.0, the H368A/H435A double mutant lost the capacity to induce syncytia at both pH 5.0 and pH 7.4 entirely (Fig. 4A and B, upper panel). Interestingly, the same double mutation had only a modest effect on the fusogenic properties of F_NL/1/94(B2) (a protein which does not require low pH for activity) (Fig. 4A and B, middle panel), but it again had a major effect when residues 294, 296, 396, and 404 of F_NL/1/00(A1) (tetrad GKRN) were introduced in F_NL/1/94(B2) (Fig. 4A and B, bottom panel). These results strongly support the idea that His residues at positions 368 and 435 are required only by hMPV F proteins that show fusion activity at low pH. This idea is also reinforced by the results obtained with the H368Q/H435K double mutants (Fig. 4B, right panels). In this case, the presence of residues that mimic a protonated histidine (368Q and 435K) switched F_NL/1/00(A1) and F_{NL/1/94(B2),GKRN} from a pH-dependent to a pH-independent phenotype.

Cytopathology of cultures infected with hMPV encoding either low-pH-dependent or pH-independent F proteins.

To test if the differences in cell-cell fusion observed in transfected cells were reflected in infected cultures, Vero-118 cells were inoculated with the four hMPV virus prototypes shown in Fig. 1. Cultures were pulsed 20 h later with medium at pH 5.0 or pH 7.4 and observed 2 to 4 h later by phase-contrast microscopy. Syncytium formation replicated the results described in previous sections for transfected cells (Fig. 5A); i.e., the two lineage B strains (NL/1/99_B1 and NL/1/94_B2) induced syncytia irrespective of the pH, whereas NL/1/00_A1 required a low-pH pulse for syncytium formation in infected cultures. Cells infected with NL/17/00_A2 did not show syncytia at either pH 5.0 or pH 7.4.

Fig. 5.

Syncytium formation in hMPV-infected cells. hMPV isolates (A) or GFP-expressing hMPV recombinant viruses (B) were used to infect Vero-118 cultures at an MOI of 0.5 PFU/cell without trypsin. Twenty hours after infection, cells were treated with trypsin and exposed to low pH (pH 5.0, bottom panels) or neutral pH (pH 7.4, top panels). Subsequently, the medium was replaced and cells were incubated for an additional 2 to 4 h. Then, the cultures were observed (at 20× magnification) by either phase-contrast (A) or fluorescence (B) microscopy. (C) Summary of the results obtained with the different viruses, including cell surface expression of the F protein measured by flow cytometry as described in the legend to Fig. 2 and 3 and pH effect on syncytium formation (+, increased fusion; −, no effect).

To obtain a better view of syncytia, two recombinant viruses, rNL/1/00_A1-GFP and rNL/1/99_B1-GFP, encoding F proteins representative of the low-pH-dependent and pH-independent phenotypes, respectively, were used to infect Vero cells. Both viruses carried the green fluorescent protein (GFP) gene inserted between the P and M genes (6). Twenty hours after infection, a marked difference in the cytopathology of cultures infected with either virus was apparent (Fig. 5B). Whereas the cells infected with rNL/1/00_A1-GFP were highly fluorescent but remained mostly individual, the cells infected with rNL/1/99_B1-GFP were seen predominantly as part of large syncytia. A brief pulse of both cultures at pH 5.0 had a drastic effect on the cytopathology of rNL/1/00_A1-GFP-infected cells, which were now forming large syncytia; however, the low-pH pulse had no significant effect on the morphology of rNL/1/99_B1-GFP-infected cultures. In conclusion, as summarized in Fig. 5C, the low-pH dependence for syncytium formation observed with the F protein of certain hMPV strains in transfected cells was duplicated in cultures infected with the same viruses, suggesting that no other viral protein substantially influenced the extent of cell-cell fusion in infected cells. Similar expression levels of the different hMPV F proteins were found in cultures infected with the different viruses (Fig. 5C), excluding the possibility that differences in syncytium formation were due to differences in F expression levels.

Stability of hMPV with either low-pH-dependent or pH-independent fusion proteins.

The results described in earlier sections suggested that low pH might be required by certain hMPV F proteins to protonate H368 and H435 and thus destabilize their prefusion conformation to initiate the process of membrane fusion. It could therefore be expected that these proteins were more thermostable than those that do not require a pulse of low pH for fusion and that viruses bearing low-pH-dependent F proteins might be more thermostable at neutral pH than those with pH-independent F proteins. To test this hypothesis, rNL/1/00_A1-GFP (low-pH-dependent) and rNL/1/99_B1-GFP (pH-independent) viruses were incubated at different temperatures before their infectivity was tested in Vero cells (Fig. 6). Both viruses were stable at 33°C for 4 h (not shown). In contrast, at 37°C, the infectivity of rNL/1/99_B1-GFP dropped more rapidly than that of rNL/1/00_A1-GFP when both viruses were incubated at pH 7.4. However, the thermostability of rNL/1/00_A1-GFP was reduced considerably if the incubation was done at pH 5.0, whereas the thermostability of rNL/1/99_B1-GFP was unaffected (or slightly increased) at pH 5.0 compared with that at pH 7.4. At 41°C, both viruses were quite unstable and the pH had only a marginal effect. In summary, these results (despite the limitations of testing whole virus instead of purified proteins in the prefusion conformation) support the notion that hMPV F proteins requiring low pH for fusion are more thermostable than those that are pH independent and that an acidic environment or high temperatures reduce these differences.

Fig. 6.

Stability of hMPV at different temperatures. Aliquots of the indicated GFP-expressing recombinant viruses were incubated at 37°C or 41°C for the indicated time periods (x axis) and the indicated pHs before being used to inoculate Vero-118 cells at an MOI of 0.5 PFU/cell. The extent of infection was evaluated 24 h later by flow cytometry analysis of GFP expression. Values were normalized to those of the control treatment (4°C), which was set to 100% for each virus. Results are representative of three independent experiments.

The pH dependence of the hMPV F protein for fusion is not reflected in an acidic requirement to infect cells.

Some enveloped viruses, such as influenza virus, enter the cells via an endocytic pathway that requires acidification of the endosome to promote fusion of the viral membrane with that of the endocytic vesicle. Exposure of the influenza hemagglutinin (HA) to low pH induces conformational changes in this molecule that mimic those occurring during the membrane fusion process (36, 43). Accordingly, endocytic acidification inhibitors inhibit influenza infectivity and that of other viruses which follow a similar endocytic pathway to enter into the host cell (for a review, see references 21 and 36). In contrast, it is thought that paramyxoviruses fuse their membranes with those of the target cell at the cell surface and in a neutral environment. However, since certain hMPV F proteins required low pH for cell-cell fusion, we decided to test the susceptibility of hMPV infection to inhibitors of endocytic acidification. Again, rNL/1/00_A1-GFP and rNL/1/99_B1-GFP viruses were selected as representatives of strains with low-pH-dependent and pH-independent F proteins for syncytium formation, respectively. Vesicular stomatitis virus (VSV) was chosen as a control virus that is sensitive to endocytic acidification inhibitors and rhRSV-GFP (a recombinant virus carrying GFP as an extra promoter-proximal gene) (12) as a control virus that is insensitive to those inhibitors.

Figure 7 shows the results obtained with two endosomal acidification inhibitors, ammonium chloride, a weak base that buffers the endosomal pH gradient, and concanamycin A, a specific inhibitor of the vacuolar ATPase that is required for endosome acidification (23). As expected, both inhibitors reduced VSV infectivity to almost background levels, even at the lowest dose tested, while they had no significant effect on rhRSV-GFP. Ammonium chloride did not impair hMPV infectivity irrespective of the F protein expressed by the recombinant viruses, while concanamycin A reduced the infectivity of rNL/1/99_B1-GFP (with a pH-independent F protein) to about 50% of that of the untreated control at the lowest dose (5 nM), but this effect was not augmented at higher doses (10 and 20 nM). Furthermore, the infectivity of rNL/1/00_A1-GFP (with a low-pH-dependent F protein) was essentially unaffected by the concanamycin A treatment, even at the highest dose tested.

Fig. 7.

Effect of endosomal acidification inhibitors on virus infectivity. Vero-118 cells were treated with either concanamycin A (upper panels) or ammonium chloride (lower panels) at the indicated concentrations for 30 min at 37°C. Then, cultures were inoculated at an MOI of 0.5 PFU/cell with either GFP-expressing recombinant viruses (left panels, except VSV) or the hMPV prototypes shown in Fig. 1 (right panels). Infections were carried out at 37°C in the presence of the drugs. Four hours after inoculation, cultures were washed and incubated at 37°C with fresh medium for an additional 2 h (VSV) or 16 h (rhMPV-GFP, rhRSV-GFP, and hMPV isolates). The cells were then detached and processed for flow cytometry analysis of GFP (rhMPV-GFP and rhRSV-GFP), hMPV F protein (HMPV isolates), or VSV G protein, as described in Materials and Methods. The number of fluorescent cells is represented in each case as a percentage of the untreated control, which was set to 100% for each virus. Data are means ± SDs from three independent experiments.

To rule out the possibility that GFP might have had any effect on the results obtained, the susceptibility of the four hMPV prototypes shown in Fig. 1 to acidification inhibitors was tested in a similar manner. Again, ammonium chloride had no impact on the infectivity of any of those viruses and concanamycin A reduced about 2-fold the infectivity of lineage B viruses. These partial effects on lineage B viruses were also observed with bafilomycin A (not shown), another specific inhibitor of the vacuolar ATPase. In conclusion, the inhibitors of endosomal acidification had only moderate effects on hMPV infectivity and no correlation was observed between the partial and dose-independent inhibition of infectivity afforded by vacuolar ATPase inhibitors and the sensitivity of the F protein to low pH in a syncytium formation assay.

DISCUSSION

Virus-cell membrane fusion is a key step in the replication cycle of enveloped viruses to deliver the viral genome into the interior of the cell (4). This process is mediated by viral glycoproteins that are present in a metastable conformation in the virus particle. Upon binding of the virus to the target cell, membrane fusion is triggered. In some viruses, the attachment and fusion activities reside in the same glycoprotein which is activated upon binding to cell receptors and/or coreceptors (e.g., human immunodeficiency virus) or by acidification of the endosome for viruses that use this entry pathway (e.g., influenza virus).

In the case of paramyxoviruses, the attachment and fusion activities reside in separate glycoproteins. It has been hypothesized that conformational changes that occur in the attachment protein of viruses of the Paramyxovirinae subfamily following binding to cells activate the F protein by poorly understood mechanisms. However, viruses of the Pneumovirinae subfamily, such as hRSV and hMPV, can infect certain cell types in vitro even in the absence of the attachment protein. In these cases, the F protein should perform virus binding to cells and fusion of the viral and cell membranes without cooperation with a second viral glycoprotein.

RSV is unique among the paramyxoviruses, since the RSV F protein contains two proteolytic cleavage sites instead of one (11). It has been shown that the insertion of two cleavage sites in the Sendai virus F protein led to decreased dependence on the attachment protein for syncytium formation in transfected cells (28). Furthermore, insertion of two cleavage sites in the F protein of recombinant Sendai viruses made these viruses less thermostable and less dependent on the interaction of the attachment protein (HN) with its receptor (sialic acid) for infection. Thus, the presence of two cleavage sites, as in RSV F, is apparently an alternative mechanism of regulating activation of a paramyxovirus F protein, independently of the attachment protein (27).

hMPV F differs from hRSV F in having one cleavage site instead of two, and therefore, its activation could not be based on the mechanism described in the previous paragraph. Recently, Schowalter et al. (32) reported that syncytium formation promoted by the hMPV F protein in transfected cells required exposure of the cultures to a short pulse of low pH. These results suggested that hMPV might use another mechanism, unique among the paramyxoviruses, to regulate F protein activation, based on exposure to low pH. However, it was found later that the low-pH requirement of hMPV F for cell-cell fusion was a strain-dependent phenomenon (15). Therefore, low pH could not be a general mechanism of regulating hMPV F activation.

The studies mentioned above (15, 32) have been extended in the present study to assess the effect of amino acid differences among hMPV F proteins of natural isolates upon their syncytial phenotypes. Sequence comparison of the F proteins of representatives of the four hMPV genetic lineages led to the identification of four variable residues positioned near two histidines (H368 and H435) in a 3D model of hMPV F. Substitution of these amino acids in the backbone of the F protein from a different lineage was sufficient to change the syncytial phenotype of the “recipient” protein to that of the “donating” molecule, particularly, its low-pH dependence for cell-cell fusion. These differences were reproduced in cell cultures infected with hMPV isolates, indicating that no other viral protein influences the syncytial phenotype of hMPV F. Therefore, the amino acids at positions 294, 296, 396, and 404 of hMPV F are largely responsible for the differences found among hMPV isolates, regarding their syncytial phenotype in infected cultures.

The results presented here suggest that protonation of H368 and H435 may be required by certain hMPV F proteins but not by others, depending on the actual amino acids present at positions 294, 296, 396, and 404. Protonation of H368 and H435 may be needed to break local interactions that clamp the prefusion conformation of the low-pH-dependent F proteins but not of those that fuse at neutral pH. This hypothesis is supported by the results shown in Fig. 4, in which changes in H368 and, particularly, in H435 had a significant effect on the phenotype of a protein which requires low pH for activity, such as F_NL/1/00(A1), but had no effect on the phenotype of a pH-independent protein, such as F_NL/1/94(B2).

Since the low-pH requirement of the F protein correlated with a higher thermostability of the virus (Fig. 6) but not with a requirement of a low-pH step for virus entry in cells (Fig. 7), we postulate that interactions mediated by the four variable residues and the two conserved histidines modulate F protein stability, either preventing (clade A proteins) or promoting (clade B proteins) syncytium formation. It is worth mentioning that syncytium formation by hMPV F proteins with a low cell-cell fusion index at pH 5.0, such as those that contain the tetrad EKRN, could be augmented by exposure to extreme unphysiological pHs (e.g., pH 4.0) (not shown).

The tetrads that allow fusion at neutral pH (ENWP, EDRP) could be acting in certain hMPV F proteins like certain destabilizing mutations found in some strains of PIV5 and Newcastle disease virus (NDV). The F proteins of these viruses do not require cooperation of the attachment protein for syncytium formation (25, 33), because, presumably, they require lower activation energy to initiate membrane fusion. Furthermore, paramyxovirus variants carrying F proteins with destabilizing substitutions replicate faster in cell culture (34). In agreement with this observation, we have reported that NL/1/99_B1 bearing a pH-independent F protein displays faster multistep growth kinetics in Vero-118 cells than NL/1/00_A1, with a low-pH-dependent F protein (6, 14), although the final titers achieved by the two viruses were almost the same.

A recent study suggested that endocytosis might play a role in the entry of the hMPV strain CAN 97-83_A2 into cells (31). These authors found that certain inhibitors of endosome acidification (concanamycin A and bafilomycin A1) but not others (ammonium chloride) partially inhibited the infectivity of the CAN 97-83_A2 strain. Partial inhibition of hMPV infectivity by concanamycin A (and bafilomycin A1; not shown) was reproduced in this study (Fig. 7), but this effect was observed with F proteins derived from lineage B viruses that do not require low pH for syncytium formation in either transfected or infected cells. This result is therefore the opposite that would be expected if the pH dependence of hMPV F for cell-cell fusion reflected a requirement for an acidic environment during virus entry. Although the entry pathway of hMPV requires further investigation, our results and those of Schowalter et al. (31) underline the relevance of residues in the linker region and in domain II of the hMPV F protein for the activation and initiation of membrane fusion.