The Heptad Repeat C Domain of the Respiratory Syncytial Virus Fusion Protein Plays a Key Role in Membrane Fusion

ABSTRACT

Respiratory syncytial virus (RSV) mediates host cell entry through the fusion (F) protein, which undergoes a conformational change to facilitate the merger of viral and host lipid membrane envelopes. The RSV F protein comprises a trimer of disulfide-bonded F₁ and F₂ subunits that is present on the virion surface in a metastable prefusion state. This prefusion form is readily triggered to undergo refolding to bring two heptad repeats (heptad repeat A [HRA] and HRB) into close proximity to form a six-helix bundle that stabilizes the postfusion form and provides the free energy required for membrane fusion. This process can be triggered independently of other proteins. Here, we have performed a comprehensive analysis of a third heptad repeat region, HRC (amino acids 75 to 97), an amphipathic α-helix that lies at the interface of the prefusion F trimer and is a major structural feature of the F₂ subunit. We performed alanine scanning mutagenesis from Lys-75 to Met-97 and assessed all mutations in transient cell culture for expression, proteolytic processing, cell surface localization, protein conformation, and membrane fusion. Functional characterization revealed a striking distribution of activity in which fusion-increasing mutations localized to one side of the helical face, while fusion-decreasing mutations clustered on the opposing face. Here, we propose a model in which HRC plays a stabilizing role within the globular head for the prefusion F trimer and is potentially involved in the early events of triggering, prompting fusion peptide release and transition into the postfusion state.

IMPORTANCE RSV is recognized as the most important viral pathogen among pediatric populations worldwide, yet no vaccine or widely available therapeutic treatment is available. The F protein is critical for the viral replication process and is the major target for neutralizing antibodies. Recent years have seen the development of prefusion stabilized F protein-based approaches to vaccine design. A detailed understanding of the specific domains and residues that contribute to protein stability and fusion function is fundamental to such efforts. Here, we present a comprehensive mutagenesis-based study of a region of the RSV F₂ subunit (amino acids 75 to 97), referred to as HRC, and propose a role for this helical region in maintaining the delicate stability of the prefusion form.

INTRODUCTION

Respiratory syncytial virus (RSV) is widely considered the most significant viral pediatric pathogen worldwide. Belonging to the Pneumovirinae family, RSV causes substantial morbidity and mortality worldwide, with an estimated 33.8 million acute lower respiratory tract infections per year in children under 5 years of age, and has been linked to almost 200,000 deaths annually, 99% of which are in developing countries (1, 2). RSV is also recognized as an important pathogen of high-risk adult and elderly populations (3). Despite this, no targeted drug or vaccine is available, and immunoprophylaxis is reserved for only a select group of high-risk infants and does not effectively reduce disease burden (4). The fusion glycoprotein, F, is highly conserved between the two subgroups of RSV and is the major target for neutralizing antibodies (5, 6), features that have made the F protein the major focus of vaccine and antiviral development.

The F protein is initially expressed as a precursor (F₀) that is cleaved at two sites by a furin-like protease in the trans-Golgi network, which results in two disulfide-linked subunits, F₁ and F₂, and releases a 27-amino-acid (aa) fragment, p27, of unknown function (7–9). Trimerized and fully cleaved F adopts a fusion-competent, metastable prefusion state which is found at the surface of infected cells where it is incorporated into budding infectious virions. Anchored by a hydrophobic transmembrane domain at its C terminus, F₁ comprises the hallmark features of a type 1 fusion protein, having a hydrophobic fusion peptide (FP) at the N terminus and two major heptad repeats, A and B (HRA and HRB, respectively), that ultimately refold to form the hyperstable six-helix bundle (6HB) that constitutes the fusion core of the postfusion structure. Triggering of the protein is thought to be provoked by receptor engagement, which prompts the release of the hydrophobic FP into the target membrane and the accompanied elongation of the HRA helices beyond the head of the protein (10, 11). Refolding of HRA and HRB brings the viral and host membranes into close proximity, resulting in membrane fusion driven by the energy difference between the pre- and postfusion conformations of the protein (10, 11). In addition to facilitating the fusion of viral and cellular membranes, cell surface F protein also drives fusion between infected and neighboring cells, resulting in the formation of large multinucleated cells, or syncytia (12, 13).

While pre- and postfusion forms of F have been crystallized, many of the specific events that underlie the process are still unknown (14, 15). Since HRA and HRB have been extensively studied due to their role in the formation of the critical 6HB, we shifted our focus to a heptad repeat in the F₂ subunit (aa 75 to 97), referred to herein as HRC, which has remained largely uncharacterized in RSV. HRC is an amphipathic α-helix that lies at the interface between two protomers of the prefusion protein, with the C terminus of the helix in close proximity to the FP (14) (Fig. 1). HRC lies adjacent to HRA within the globular head of F in its prefusion form and also packs against a fourth heptad repeat (aa 214 to 241), referred to herein as HRD.

FIG 1.

RSV F structure. (A) Schematic diagram of the RSV F monomer showing the two cleavage sites (black arrows) and the resultant p27 region that is liberated from the disulfide-linked (s-s; solid line) F₁ and F₂ subunits that comprise the heterodimer. (B) F forms a trimeric prefusion structure that undergoes conformational rearrangements to the postfusion form. Both forms have been solved by X-ray crystallography (PDB accession numbers 4JHW and 3RK1) (14, 15). Two protomers are shown in two shades of gray with surface representation, while the third is shown as a ribbon with coloring as described for panel A. TM, transmembrane region.

To investigate this region, an alanine scan was performed throughout the helix, and 22 alanine mutations were generated in a recombinant codon-optimized F plasmid. Transfection of this plasmid alone has previously been shown to be sufficient to drive extensive syncytium formation, providing an ideal system to assay the effect of mutation on fusion promotion without the need for cotransfection of other viral proteins (13). Transfected cells were assayed for F protein expression, processing, cell surface localization, conformational state, and fusion phenotype. The results of this analysis indicate that HRC may provide important stabilizing interactions in the globular head of the protein and suggests a critical role for HRC in the fusion process.

RESULTS

Generation of mutations in the HRC domain.

In order to investigate the role of HRC in fusion, an alanine scan was performed at each amino acid within the region of Lys-75 to Met-97. Twenty-two mutated plasmids were generated by site-directed mutagenesis, each containing a single alanine substitution in the codon-optimized wild-type (WT) F sequence (based on the RSV A2 strain), with the exception of position 89, which is an alanine within the WT sequence. In the pIRES2-EGFP plasmid (Clontech), the F sequence is immediately followed by enhanced green fluorescent protein (EGFP) under the control of an internal ribosome entry site (IRES), allowing visualization of successfully transfected cells. The entire open reading frame for each mutation was sequence verified and confirmed to be free from polymerase-induced errors prior to functional analysis.

Expression and proteolytic processing of F constructs.

Following translation, the F protein must undergo proteolytic processing to become fusogenic; the protein is initially made as an inactive F₀ precursor that is cleaved at two sites by a furin-like protease in the trans-Golgi network (site 1, RARR¹⁰⁹; site 2, KKRKRR¹³⁶). The intervening 27-amino-acid peptide, p27, is then released, allowing the remaining disulfide-linked F₁-F₂ subunits to form the fusion-active trimer of heterodimers (8, 9). To assess the expression and processing of F protein mutations relative to those of the WT protein, monolayers of COS-7 cells were transiently transfected, lysed with a mild detergent after 24 h, and assessed by SDS-PAGE and Western blot analysis (Fig. 2) using the monoclonal antibody motavizumab that recognizes the site II epitope present in both the 70-kDa precursor F₀ and the fully cleaved 50-kDa F₁ subunit (16). This allowed the simultaneous confirmation of both protein expression and proteolytic processing.

FIG 2.

Expression and processing of HRC mutations. Transiently transfected monolayers of COS-7 cells were lysed at 24 h posttransfection (hpt) and separated by SDS-PAGE in a reducing buffer. On each gel, positive (WT F) and negative controls (empty vector [ev]) were examined. Proteins were transferred to nitrocellulose and probed with motavizumab (human IgG) and with GAPDH (rabbit IgG) monoclonal antibody as a host protein control and revealed by anti-human 800 and anti-rabbit 680. Membranes were scanned on the LiCor Odyssey Imaging System, and band intensity was measured by Image Studio Software (LiCor) and normalized to GAPDH intensity. Cleavage efficiency is expressed as the intensity of F₁ divided by that of F₁ plus F₀. n/a, not applicable; F_DEG, degradation product of WT F.

Western blot analysis revealed that the majority of mutant proteins were expressed at levels similar to the level of WT F, with some reduction observed for the mutations E82A, Y86A, L93A, and L96A (Fig. 2). For one mutation, Q81A, no expression was observed either by Western blotting or immunofluorescence (IF) on fixed and permeabilized cells with anti-F polyclonal rabbit serum (data not shown). For this expression construct, EGFP fluorescence was visualized, confirming efficient transfection and transcription. It is therefore likely that this mutation is not viable, with the protein likely misfolded, retained in the endoplasmic reticulum (ER), and degraded.

Western blot analysis also allows the efficiency of proteolytic cleavage relative to that of the WT F to be quantified through comparison of the ratio of cleaved F₁ to its uncleaved precursor F₀. Of the 22 mutations generated, approximately half showed some cleavage deficiency. This is most noticeable in E82A, L83A, V90A, L93A, and, to a lesser extent, in Y86A and L96A, where a larger portion of the protein present is in the 70-kDa F₀ form that is not cleaved at either of the two cleavage sites.

SDS-PAGE analysis of the cell lysates also revealed a degradation product of WT F of about 20 kDa, which matches the expected size of an unpublished crystal structure of the F head domain (PDB accession number 4CCF). This deposited fragment contains the motavizumab epitope and, according to the description in the Protein Data Bank (PDB), was derived as a by-product of cellular protease cleavage at some time during expression or purification. An alternate 30-kDa by-product corresponding to the remaining C terminus of F₁ (as determined by Western blotting to a C-terminally bound purification tag) has also been observed by other investigators during recombinant F protein expression (17). This product is absent in the cleavage-deficient mutations (E82A, L83A, V90A, and L93A) and reduced in others (Y86A and L96A). Interestingly, this degradation product is also reduced in E92A, despite cleavage efficiency similar to that of the WT.

Cell surface expression of F protein mutations.

Translocation of nascent RSV F protein to the plasma membrane is an essential step in the viral life cycle, required for both F-mediated cell fusion and incorporation into budding virions. Flow cytometry of transfected cells was used to assess the total levels of F protein on the cell surface (using monoclonal antibody motavizumab; the site II epitope is available on both the pre- and postfusion forms) and protein conformation (using the prefusion-specific monoclonal antibody D25) (Fig. 3A).

FIG 3.

Cell surface expression and conformation of transiently expressed F proteins. (A) At 24 hpt, the antibody motavizumab was used in flow cytometry to assess the level of protein expression at the cell surface. GFP fluorescence from the pIRES2-EGFP vector was used to gate successfully transfected cells and the monoclonal antibodies motavizumab and prefusion-specific D25 used to quantify F protein expression. The average median fluorescence intensity (MFI) of three independent replicates is shown relative to that of the WT F ± standard error of the mean. The log-transformed surface expression levels of total F bound by motavizumab (B) and prefusion F bound by D25 (C) are plotted against cleavage efficiency. Both surface expression and cleavage efficiency values are presented relative to the levels of the WT F, with linear regression used to model the relationship.

Mutations that resulted in decreased processing (Fig. 2) were found to result in lower cell surface expression. Mutations E82A, L83A, V90A, and L93A, all of which showed less than 50% of WT F cleavage levels, were present on the cell surface at levels below 50% of the level detected for WT F. Mutation Q81A was not detected on the cell surface, consistent with this protein not being detected by Western blotting (Fig. 2). Mutation L93A also led to the corresponding protein not being detected on the cell surface, indicating that it was likely retained within the secretory pathway. Generally, cleavage efficiency was positively correlated with cell surface expression and correct protein conformation (Fig. 3B and C).

As motavizumab has been previously shown to bind both cleaved and uncleaved RSV F with similar dissociation constants, quantification of cell surface expression should not be directly affected by processing (18). This experiment was also duplicated and confirmed with 18B2 (Argene), which targets an unknown epitope on F₁ that is exposed in both the pre- and postfusion forms (data not shown).

RSV F protein stability is a critical determinant of function. F protein must remain in a metastable conformation to allow appropriate triggering of the conformational change into the postfusion form. Increased stability could prevent this transition, while decreased stability could result in premature triggering to the postfusion form. This transition is irreversible under physiological conditions and renders the fusion protein inactive. Heat, low-molarity buffer, and freeze-thaw cycles have all been shown to cause this unidirectional conformation change (17, 19, 20). Recently, a number of conformation-specific antibodies have been characterized. We utilized D25, which is specific for the prefusion structure of RSV F (14), to probe the amount of prefusion F on the cell surface (Fig. 3A).

Of all the HRC mutations, none was found to have a significantly higher cell surface representation of prefusion F than the WT, indicating that none of these mutations had the effect of stabilizing the protein in the prefusion conformation. A number of mutations that had similar or slightly reduced levels of total F surface expression relative to that of the WT (inferred by motavizumab binding levels) showed significantly reduced amounts of D25-reactive prefusion F. These include K75A, L78A, and I79A. Most dramatically, I79A, which had 68% of the WT level of total F surface expression (as detected by motavizumab), had only 4% of the level of WT prefusion F surface expression (as detected by D25). A similar trend was observed for mutations K77A, L78A, and Y86A although to a lesser extent. Cell surface representation of less than 10% of WT F prefusion was also true of the processing-deficient mutations that had low levels of total F, specifically, E82A, L83A, Y86A, V90A, L93A, and L96A.

Fusogenicity and helical localization.

After expression, processing, and cell surface conformation were confirmed, all mutations were functionally assessed on their ability to drive cell fusion through both direct visualization and a quantitative electrical impedance-based fusion assay, a previously described method that detects syncytium formation through a decrease in electrical impedance across a cell monolayer (21). For direct visualization, monolayers of COS-7 cells were transiently transfected with mutated and control plasmids that coexpressed GFP under an internal ribosomal entry site. Fusion was visually graded on the extent of GFP diffusion caused by F-mediated cell-cell fusion and the subsequent cell content mixing (22). Almost all mutations were observed to have some effect on syncytium formation, with a range of phenotypes being observed, ranging from complete abrogation of fusion to WT levels. For some mutations, an increase over and above WT F levels was seen. Fusion phenotype was scored based on both the number of syncytia observed as well as the size of any syncytia present (Fig. 4B) and further quantified with an electrical impedance fusion assay (Fig. 4A).

FIG 4.

Two fusion assays that measure syncytium spread in transiently transfected COS-7 cells show distinct fusion-increasing and fusion-decreasing phenotypes. (A) Cell-cell fusion was measured by an electrical impedance fusion assay (21). At 20 hpt, impedance was recorded and normalized to the levels of the WT F and empty vector controls. Data are the averages ± standard errors of the means of two repeats (n = 3). Using the pIRES2-EGFP vector, fusion was measured by observing EGFP fluorescence after images (n = 5) were captured at 24 hpt using an IN Cell Analyzer. A visual rating system for fusion phenotype was developed to score mutations against the empty vector (ev; −) and WT F (+++) for number and extent of syncytia, with an example for each rating shown in panel B.

The results of both assays were largely consistent; however, each assay had particular advantages and limitations. The impedance assay was particularly useful in quantifying the increases in fusion above the level of WT F that were difficult to visually rate. Conversely, direct visualization was advantageous for observing smaller fusion events that were below the sensitivity of the cell impedance system. Both assays were considered for assessment of the fusion phenotype.

To visualize the effects of our site-directed mutations on fusion phenotype, we projected the fusion phenotype onto the prefusion structure of RSV F using a color spectrum to represent the extent of fusion (Fig. 5A and B). This revealed a striking correspondence between amino acid position and fusion phenotype, where mutations that were observed to decrease or completely ablate fusion were found localized to one face of the HRC α-helix while those that increased fusion were found on the opposing face. This is particularly evident along the longitudinal axis of the helix (Fig. 5B, inset).

FIG 5.

Localization of HRC mutations in the prefusion F structure. (A) A heptad repeat schematic showing the HRC residues in positions a to g, with the hydrophobic face highlighted (gray). (B and C) HRC is shown as a ribbon against the cognate (light gray) and opposing (dark gray) protomers in the prefusion structure (PDB accession numbers 4JHW in panel B and 5K6C in panel C). Mutations that ablated fusion (red, rating −), dramatically reduced fusion (orange, rating +), or increased fusion (light green, rating +++ +, or dark green, rating +++ ++) have been color coded to reflect phenotype and are shown in stick representation (B). In the inset, the helix is shown in isolation viewed from the N (foreground) to C (back) terminus, and it is evident that mutations cluster to opposing faces relative to each phenotype. At the N terminus of HRC, shown in panel C (in coral) against a surface representation of the protomers, two lysine residues (Lys-75 and Lys-77) lie against a bed of negative charge from HRD residues. Basic amino acids are shown in red while acidic residues are shown in blue. In panels B and C, a small region of the flexible loop N-terminal to HRC has been hidden to avoid obstructing the view of the helix.

Ablation of fusion activity for mutations Q81A and L93A was anticipated as these mutations were either not expressed (Q81A) or not translocated to the plasma membrane (L93A) where the protein mediates cell-cell fusion. Likewise, the mutations E82A, L83A, Y86A, V90A, and L96A all lacked any syncytium formation, likely a reflection of their low cleavage efficiencies and reduced translocation to the cell surface. These results are in line with previous findings, whereby proteolytic processing and cell surface localization have been shown to be critical in F-mediated membrane fusion (8).

Interestingly, two mutations (I79A and E92A) resulted in the absence of fusion for expressed F, even though the levels of processing and cell surface localization were similar to those of WT F. When the relative levels of total F to those of the prefusion conformation were compared, the relative representation of prefusion F for the I79A protein is greatly reduced while for E92A the ratio is similar to that of the WT. These results likely indicate two distinct mechanisms responsible for fusion ablation: decreased stability of the prefusion conformation leading to premature transition to the postfusion conformation (I79A) and an inability to trigger the conformational change into the postfusion form despite the presence of the prefusion conformation (E92A).

Five mutations showed decreased levels of fusion. Of note, four of the five were lysine residues (K75A, K77A, K85A, and K87A), and the fifth was leucine (L78A). Finally, V76A, D84A, N88A, T91A, L95A, and M97A were all seen to increase fusion above the level of the WT F. These residues all cluster to one face of the helix and are solvent exposed. Of these six mutations, only D84A had reduced representation of F on the cell surface and an even lower representation of the prefusion form. However, D84A was the most fusogenic construct of all mutations analyzed, despite a large decrease in cell surface expression.

DISCUSSION

RSV F plays a central role in viral replication and pathogenesis as a mediator of both receptor binding and membrane fusion and as the primary target of the host's protective antibody response. To date, the majority of research into the fusion function of RSV F has been focused on regions within the F₁ subunit, in particular, the HRA and HRB coiled coil (23–25). While these studies have provided deeper insight into formation and antiviral targeting of the terminal 6HB fusion core (26), the roles of residues and domains within the F₂ subunit remain relatively unexplored. Here, we performed a comprehensive mutational analysis of the HRC α-helix (Lys-75 to Met-97) and demonstrated that the region has an important role in RSV F-mediated membrane fusion. Each mutation was assessed for expression level, proteolytic processing, surface expression, conformational state, and fusion activity. Of particular interest, we noted a functional asymmetry across the region, where fusion-increasing mutations localized to the solvent-exposed face of the HRC helix and fusion-decreasing and ablating mutations located to residues on the side of the helix that is buried in the F protein structure.

Of the fusion-decreasing and -ablating mutations we have identified in RSV F, a number showed a reduction in levels of proteolytic processing and cell surface expression. Such deficiencies prevent a full assessment of the direct role these residues may play in the fusion process as both proteolytic cleavage and cell surface expression are critical for F protein-mediated membrane fusion. Changes to the HRC (aa 75 to 97) region could be expected to affect processing as it is located N-terminal to the first cleavage site between amino acids 109 and 110. While there are currently no crystal structures of RSV F in its uncleaved state, it has previously been shown that the F protein is likely monomeric prior to cleavage and p27 removal (18, 27). Changes in side chain packing may alter helical positioning within monomeric F, thereby affecting presentation of cleavage sites to the required protease(s).

Our findings also indicate that the buried residues of HRC (Gln-81, Leu-83, Tyr-86, Val-90, Leu-93, and Leu-96) appear to play an important role in the structural integrity of the protein, where alteration leads to a dramatic reduction in proteolytic processing and export to the cellular membrane. Of note, a clear trend was established linking proteolytic processing and transport to the cellular membrane. It is likely that mutations resulting in protein misfolding are retained in the ER and therefore not processed within the Golgi apparatus.

Of the paramyxo- and pneumovirus fusion proteins, RSV and human metapneumovirus F proteins are unique in that they are able to mediate membrane fusion in the absence of other viral proteins (13, 28). As such, RSV F is able to maintain a functionally active metastable state without the aid of additional viral glycoproteins. In this state, the protein must maintain a state of delicate equilibrium that is both sufficiently unstable to undergo the rapid conformational change from the pre- to the postfusion form while also maintaining enough structural integrity to stay in the prefusion state without premature triggering. Transition to the postfusion form in the absence of a target membrane is irreversible and therefore renders the F protein inactive. If all F proteins (or an amount beyond a certain threshold) on a virion are prematurely triggered, that virus is essentially noninfectious.

Viewed within this context, our findings suggest that many of the residues within HRC appear to play a role in conformational stability. Mutations of residues at the N terminus of HRC (K75A, K77A, L78A, and I79A) likely have the effect of destabilizing F, which is reflected in the relatively low abundance of prefusion F at the surface, despite no significant decrease in the total levels of F surface expression compared to the WT protein level. N-terminal residues form a network of hydrophobic interactions near the apex of the RSV prefusion trimer alongside residues in HRA and the flexible loop between HRA and HRD, another heptad repeat in F₁ from Ile-214 to Ala-241. Substitution with the smaller side chain of alanine may disrupt the packing of these helices in the head region and possibly lead to premature rearrangement of the HRA helices and release of the hydrophobic FP. Flexibility within this region is also evident when the atomic structures of the stabilized prefusion F constructs (e.g., PDB accession numbers 5K6C and 4MMT) are compared (data not shown) and is also apparent from the reported crystal structure B-factors that are consistent with a high intrinsic atomic mobility around the apex of the prefusion head, in particular the N-terminal region of HRC (29, 30).

Also at the N terminus of HRC are two lysine residues, Lys-75 and Lys-77, alanine substitution of which resulted in a decrease in fusion. In addition to the hydrophobic belt discussed above, these charged residues may facilitate additional stabilizing contacts between interprotomer HRC and HRD through electrostatic interactions with Glu-218 and Glu-222, respectively, and may provide additional conformational stability (Fig. 5C). Again, the localized conformation of this region is altered in the various reported prefusion crystal structures. Both K75A and K77A mutations show a decrease in the prefusion protein level on the cell surface, relative to the WT level, suggesting that disruption of these potential interactions promotes destabilization in the region and premature conformational triggering.

Another mutation of interest that ablated fusogenicity was E92A, even though it showed WT levels of processing and cell surface representation, including the presence of the prefusion conformation. The side chains of the glutamic acid are close enough to hydrogen bond with Ser-238 in the HRD helix of the prefusion structure. However, the alanine substitution does not seem to induce any destabilization since E92A shows WT levels of the prefusion conformation F on the cell surface. Interestingly, an E92D change was one of six mutations previously used in combination to generate a stabilized prefusion F, though the authors of that study hypothesized that the native glutamic acid was unfavorable due to clashes with Asn-254 in the prefusion conformation (30). It is possible that side chain packing at this site is important during the early stages of triggering and refolding that cannot be mediated by the small, nonpolar alanine.

Our results also suggest that the levels of expression, processing, and cell surface expression and the relative presence of the prefusion protein conformation are not always predictive of fusogenicity. For example, despite showing similar phenotypic levels for each of these parameters tested, E92A completely ablated fusion while L95A led to a slight increase. For the fusion-increasing mutations (V76A, D84A, N88A T91A, L95A, and M97A), there does not seem to be a trend in interactions made in the static structures of the pre- and postfusion forms of RSV F although this face of the helix is somewhat solvent exposed. Thus, we hypothesize that these residues may play a role during the transitional rearrangements in F and may make fleeting interactions that contribute to the thermodynamics of refolding. Removal of these side chains by alanine mutagenesis may reduce the mechanical energy of conformational change, leading to a more favorable transition into the postfusion form. This is seen as an increase in fusogenicity as measured by syncytium spread. In contrast, excess instability could lead to complete misfolding and degradation, as is likely with Q81A, which is not expressed at all.

Our results point to a stabilizing role for many residues at the N terminus of the helix. Leu-78 and Ile-79 contribute to a hydrophobic belt that lines the inner cavity of the prefusion F trimer, while Lys-75 and Lys-77 facilitate trimer stabilization through electrostatic interactions with the HRD helix in the adjacent protomer. Disruption of any of these residues had the effect of destabilizing the protein, as seen by a reduced detection of the prefusion form by flow cytometry and a corresponding reduction in fusion. Increased flexibility in this region could potentiate premature triggering, with HRC playing a critical role in the initiation of the pre-to-postfusion transition. Of interest, a recent study showed that the N-terminal region of HRC in parainfluenza virus 5 (PIV5) is among the first sections of the protein head to move in the prefusion structure along with the fusion peptide, indicating that HRC movement may be one of the earliest events in the fusion process (31). These findings, in combination with results presented here, may point to a conserved role for this helix across the Paramyxoviridae family. Supporting this hypothesis, efforts to engineer a stabilized prefusion F have been achieved in one study through limiting mobility of the flexible loop immediately N-terminal to HRC by an N67I mutation. This mutation was proposed to increase hydrophobic interactions between the loop, HRC, and part of HRA and effectively “stick” this loop to these domains. Another mutation at position 215 acted to reduce repulsion forces in this apex region where the HRA, HRD, and HRC helices meet (27).

Conversely, an F₂ mutant identified in a live attenuated vaccine strain was shown to increase syncytium formation in transfected cells and in in vitro infection with whole virus (32). This raises the interesting question of the tolerance of fusion protein mutations in the context of infectious virus, especially in relation to our fusion-increasing mutations. This study and others have seen the fusogenicity of transfected F mutations mimic the same phenotype when they are part of an infectious virus (32, 33). Even though many of our mutations show an increase in the fusion phenotype, the amino acid sequence of HRC is genetically stable, suggesting that the area does not tolerate change. In the context of transfected F protein in cell monolayers, some instability in the protein can be beneficial for syncytium spread since the target membrane is present once the protein reaches the cell surface, and an easier transition into the postfusion form will simply lead to larger syncytia. However, circulating virus is faced with the challenge of withholding transition until an appropriate host cell membrane is available. Correlation has been found between an increase in fusion activity and higher viral load and pathogenesis in mice for both RSV (33) and other paramyxoviruses (34). Thus, circulating virus faces competing selective pressures of balancing the infectious half-life of transmitted virions and fusion capacity.

In conclusion, we have shown that the HRC region of RSV F plays a critical role in both protein metastability and fusion activity. Minor changes in the sequence of this helix had a range of destabilizing effects, from abrogation of fusion activity to slight decreases or, indeed, increases in fusogenicity. This is the most comprehensive study of this region in the F₂ subunit in RSV and, together with findings from closely related paramyxoviruses, suggests a likely stabilizing role for this helix in the globular head of the prefusion F trimer and potential involvement in the early events of triggering. Stabilization of the prefusion conformation of the F protein has been a focus of recent research as antibodies to the prefusion protein account for the majority of the neutralizing antibody response (35). In this study, we have provided a functional and biophysical analysis of putative interactions within the globular head, particularly between HRC and HRD residues, which could be used to inform future vaccine design. Additionally, as has been demonstrated for a closely related paramyxovirus fusion protein (36), compounds able to prematurely trigger the fusion process can render viruses noninfective. Understanding the domains involved in triggering of the fusion process may also facilitate the rational design of antiviral therapeutics.

MATERIALS AND METHODS

Cells and transfection.

COS-7 cells were maintained in Opti-MEM I reduced serum medium (Gibco) supplemented with 3% fetal bovine serum (FBS) and maintained in a humidified incubator at 37°C with 5% CO₂. For transfection, monolayers were seeded 1 day before transfection in Opti-MEM supplemented with 2% FBS. DNA for transfection was mixed in a 1:4 ratio of 1 mg/ml polyethylenimine in MOPS [3-(N-morpholino)propanesulfonic acid] buffer at pH 7 in serum-free Opti-MEM, incubated at room temperature (RT) for 15 min, and then added to wells with sufficient medium to cover the monolayer. At 5 h posttransfection (hpt), cells were washed with phosphate-buffered saline (PBS) and then supplemented with Opti-MEM (2% FBS).

Antibody cloning, expression, and purification.

The variable regions of motavizumab and D25 antibodies were cloned and expressed in recombinant form. Briefly, the variable domains of the antibodies were cloned into separate plasmids encoding the heavy- and light-chain human IgG backbones described by Jones et al. (37). Vectors were transfected into CHO cells in suspension culture, and supernatant was harvested 7 days later for secreted antibody, which was purified by protein A chromatography and buffer exchanged into PBS as previously described (38).

Plasmids and PCR mutagenesis.

Codon-optimized RSV F (Fopt.FL) developed by Morton et al. was subcloned into the pIRES2-EGFP vector (Clontech) using the BamHI and PstI restriction sites and used as the template DNA in single-step site-directed mutagenesis PCR (13). Primers were designed as described by Zheng et al., and successful mutation was confirmed by Sanger sequencing at the Australian Genome Research Facility (39). The entire coding sequence of F for each mutation was screened to ensure that the gene was free of polymerase-induced errors. All primers are described in Table 1.

TABLE 1.

List of primers used for mutagenesis^a

Mutation	Forward primer	Reverse primer
K75A	GACGCCGCAGTGAAGCTGATCAAG	TCACTGCGGCGTCGGTGCCGTT
V76A	GCCAAGGCAAAGCTGATCAAGCAAG	GCTTTGCCTTGGCGTCGGT
K77A	AAGGTGGCACTGATCAAGCAAGAGCT	TCAGTGCCACCTTGGCGTC
L78A	GTGAAGGCAATCAAGCAAGAGCTG	TGATTGCCTTCACCTTGGCGTC
I79A	AAGCTGGCAAAGCAAGAGCTGG	GCTTTGCCAGCTTCACCTTGGC
K80A	CTGATCGCACAAGAGCTGGAC	CTTGTGCGATCAGCTTCACCTTGG
Q81A	ATCAAGGCAGAGCTGGACAAGTAC	GCTCTGCCTTGATCAGCTTCAC
E82A	CAAGCAAGCACTGGACAAGTAC	CCAGTGCTTGCTTGATCAGC
L83A	CAAGAGGCAGACAAGTACAAGAAC	TGTCTGCCTCTTGCTTGATCAG
D84A	GAGCTGGCAAAGTACAAGAACG	ACTTTGCCAGCTCTTGCTTG
K85A	CTGGACGCATACAAGAACGCCGTG	TGTATGCGTCCAGCTCTTGCTTG
Y86A	GACAAGGCAAAGAACGCCGTGACCG	TCTTTGCCTTGTCCAGCTCTTGC
K87A	TACGCAAACGCCGTGACCG	GCGTTTGCGTACTTGTCCAGCTC
N88A	GTACAAGGCAGCCGTGACC	CGGCTGCCTTGTACTTGTCCAG
V90A	AACGCCGCAACCGAGCTGCAAC	CGGTTGCGGCGTTCTTGTACTTGTC
T91A	CGTGGCAGAGCTGCAACT	GCTCTGCCACGGCGTTCTTGTA
E92A	GTGACCGCACTGCAACTGCTGATG	GTTGCAGTGCGGTCACGGCGTTC
L93A	ACCGAGGCACAACTGCTGATGC	AGTTGTGCCTCGGTCACG
Q94A	GAGCTGGCACTGCTGATGC	CAGCAGTGCCAGCTCGGTCA
L95A	GCAAGCACTGATGCAGTCG	TCAGTGCTTGCAGCTCGGTC
L96A	CAACTGGCAATGCAGTCGACTCAA	GCATTGCCAGTTGCAGCTCGGT
M97A	CTGCTGGCACAGTCGACTCA	ACTGTGCCAGCAGTTGCAGC

^aAlanine substitutions are underlined.

Western blot analysis.

Subconfluent monolayers of COS-7 cells in six-well plates (Nunc) were transiently transfected with 1 μg of mutated or control pIRES2-EGFP.Fopt plasmids. At 24 hpt cells were scraped and then centrifuged at 1,000 rpm for 2 min at 4°C. Pellets were resuspended in 50 μl of membrane lysis buffer (10 mM Tris-HCl, pH 7.2, 0.5 M NaCl, 0.5% NP-40) with cOmplete Mini EDTA-free protease inhibitor cocktail (Sigma-Aldrich) for 20 min on ice. Lysate was centrifuged at 10,000 × g for 10 min at 4°C, and then the supernatant was boiled for 4 min with 4× SDS-PAGE loading buffer and a 1:100 dilution of 2-mercaptoethanol. Samples were run on a 4% stacking, 10% resolving SDS-PAGE gel for 10 min at 100 V and then for 1 h at 170 V. Proteins were transferred to nitrocellulose and then blocked overnight in PBS-TMP (0.05% Tween 20, 3% skim milk powder). Membranes were incubated with human motavizumab (1:1,000) and rabbit anti-glyceraldehyde-3-phosphate dehydrogenase (GAPDH; Cell Signaling Technology) (1:1,000) for 1 h at RT, washed three times, incubated with goat anti-rabbit 680 and goat anti-human 800 (1:5,000) (LiCor), washed three times again, and read on a LiCor Odyssey Imaging System.

Flow cytometry.

The night before transfection, COS-7 cells were seeded into six-well plates (Nunc) at a density of 10⁴ cells/well in Opti-MEM (2% FBS) (Gibco). This concentration was chosen to prevent fusion occurring between neighboring cells, which was observed at higher densities. At 24 hpt, cells were harvested and probed with a 1:500 dilution of 18B2 (Argene), D25, or motavizumab (1 mg/ml stock) and revealed by a 1:1,000 dilution of anti-mouse IgG1 or anti-human IgG Alexa-647 (Thermo Fisher Scientific). Samples were read on an Accuri C6 flow cytometer (Becton Dickinson Biosciences) and analyzed in FlowJo, version 10.1 (Tree Star). Three biological replicates were performed, and results are presented as the means ± standard deviations.

Cell-cell fusion assay.

COS-7 cells were seeded into a 24-well plate (Nunc) at a concentration of 10⁵ cells/well and transfected the next day with 500 ng of DNA using the pIRES2-EGFP plasmid constructs. After transfection, cells were washed with PBS and supplemented with 2% FBS and Opti-MEM. Live cells were imaged at 24 hpt with an IN Cell 1000 instrument (GE Life Sciences) at a magnification of ×10 using bright-field microscopy and fluorescence microscopy with 485-nm excitation and 535-nm emission filters to detect EGFP. At 24 hpt syncytium formation was visualized through the expression of EGFP and visually assessed based on syncytium number and size relative to numbers and sizes for the positive (pIRES2-EGFP.Fopt) and negative (pIRES2-EGFP empty vector) controls.

Cell impedance assay.

We have recently described a label-free system to measure viral fusion through changes in electrical impedance across a cell monolayer (21). One night prior to transfection, 2 × 10⁵ COS-7 cells were seeded into 12-well plates (Nunc) and transfected with 1 μg of DNA the following day. At 4 hpt, transfection medium was removed, and monolayers were washed with PBS, removed with trypsin-EDTA solution (Thermo Fisher Scientific), and transferred to a 96-well E-plate (ACEA Biosciences) at 4 × 10⁴ cells/well. The E-plate was placed in an xCELLigence RTCA SP (real-time cell analysis, single plate) station (ACEA Biosciences), and electrical impedance was measured and recorded every 15 min for 72 h. Prior to cell replating, the system was blanked with Opti-MEM (2% FBS).