A single amino acid in the F2 subunit of respiratory syncytial virus fusion protein alters growth and fusogenicity

Respiratory syncytial virus (RSV) causes severe lower respiratory tract infection in children, especially in infants less than 1 year of age. There are currently no licensed vaccines against RSV. rA2ΔM2-2 is a promising live-attenuated vaccine candidate that is currently being evaluated in the clinic. Attenuation of rA2ΔM2-2 is achieved by a single deletion of the M2-2 gene, which disrupts the balance between viral transcription and replication. Whilst performing a manufacturing feasibility study in a serum-free adapted Vero cell line, differences in growth kinetics and cytopathic effect (CPE) were identified between two rA2ΔM2-2 vaccine candidates. Comparative sequence analysis identified four amino acid differences between the two vaccine viruses. Recombinant rA2ΔM2-2 viruses carrying each of the four amino acid differences identified a K66E mutation in the F2 fragment of the fusion (F) protein as the cause of the growth and CPE differences. Syncytium-formation experiments with RSV F protein carrying mutations at aa 66 suggested that a change in charge at this residue within the F2 fragment can have a significant impact on fusion.


INTRODUCTION
Respiratory syncytial virus (RSV) is an enveloped, singlestranded, negative-sense RNA virus of the family Paramyxoviridae, subfamily Pneumovirinae. RSV causes lower respiratory tract infection in children, with a high incidence of serious disease in infants less than 1 year old (Nair et al., 2010). Although prophylaxis with palivizumab antibody is available for high-risk infants, there are currently no licensed vaccines to prevent severe RSV disease (Johnson et al., 1997). A variety of approaches to RSV vaccination have been evaluated over the years including subunits, virus-like particles and live-attenuated vaccines (Collins & Melero, 2011;Schickli et al., 2009). The use of a non-live RSV vaccine for naïve infants is problematic because formalin-inactivated RSV, the first and only non-live RSV vaccine to be tested in naïve infants, not only was ineffective but the vaccinees also experienced a more severe disease upon subsequent reinfection with RSV than unvaccinated children. This phenomenon has been termed RSV enhanced disease (Kapikian et al., 1969;Kim et al., 1969). In contrast, live-attenuated vaccines are promising, and have been evaluated extensively in RSVnaïve children and infants in the clinic. None of the liveattenuated RSV vaccine candidates tested to date has been shown to cause enhanced disease in RSV-naïve infants or children (Karron et al., 2005;Wright et al., 2007). From an immunological perspective, live-attenuated virus is expected to mimic most closely the natural route of infection and, in turn, stimulate protective mucosal, humoral and cellular immune responses without the need for adjuvants.
One of the major challenges in developing a safe and effective live-attenuated RSV vaccine is maintaining the delicate balance between limiting virus replication in the host and delivering a high enough antigen load to induce a protective immune response. Most live-attenuated RSV vaccine candidates studied to date rely on point mutations to attenuate growth. Reversion of these point mutations can result in partial reversion of the attenuation phenotype as was observed in the rA2cp248/404/1030DSH clinical trial (Karron et al., 2005). Partial reversion raises concerns about transmission of less-attenuated virus to vulnerable contacts of vaccinees.
Potential reversion of the attenuation phenotype can be mitigated by employing gene deletion strategies such as that in rA2DM2-2, a promising live-attenuated vaccine candidate. Attenuation is achieved by a deletion in the RSV genome that eliminates expression of the M2-2 gene (Cheng et al., 2001;Jin et al., 2000Jin et al., , 2003Teng et al., 2000), the product of which is thought to regulate the switch from transcription of viral mRNA to replication of the viral genome (Bermingham & Collins, 1999). Although rA2DM2-2 virus can grow to high titres in Vero cells, it is attenuated in mice, cotton rats and non-human primates (Cheng et al., 2001;Jin et al., 2000Jin et al., , 2003Teng et al., 2000). The advantages that distinguish rA2DM2-2 vaccine candidates from other live-attenuated RSV viruses are: (i) an attenuation phenotype that results from a gene deletion, which is less likely to revert than point mutations, and (ii) attenuated growth that does not highly compromise the expression level of viral antigens, thereby helping to maintain a high level of antigen load.
As part of a manufacturing feasibility assessment, two versions of rA2DM2-2, designated here as rA2DM2-2(MEDI) and rA2DM2-2(NIH), were evaluated in a serum-free (SF) adapted Vero cell line previously used to manufacture clinical trial vaccines (Yuk et al., 2006;Kaur et al., 2008). In this manufacturing SF Vero cell line, the two versions of rA2DM2-2 showed different growth kinetics and cytopathic effect (CPE). The growth differences were unexpected, because both versions are derived from the A2 strain of RSV and share .98 % sequence identity. Alignment of their genome sequences identified four predicted amino acid differences in three viral proteins: the non-structural protein NS2, the nucleoprotein (N) and the fusion protein (F). Each of the different amino acids found in rA2DM2-2(NIH) was introduced into rA2DM2-2(MEDI) in order to assess their effect on growth. In this way, we identified the amino acid at position 66 in the F 2 fragment of the RSV F protein as the genetic determinant of the observed growth differences between the two rA2DM2-2 viruses. Substitution of different amino acids at position 66 of RSV F further demonstrated that basic amino acids with positively charged side chains promote fusion activity and growth of rA2DM2-2 viruses.

RESULTS
Differences in growth between the two rA2DM2-2 vaccine candidates Two laboratories have previously reported the generation of rA2DM2-2 virus (Bermingham & Collins, 1999;Jin et al., 2000). Both reported that rA2DM2-2 virus had reduced growth in certain cell lines, as well as attenuated growth in rodents and non-human primates (Teng et al., 2000;Jin et al., 2000). The multicycle growth-curve analysis of rA2DM2-2(MEDI) and rA2DM2-2(NIH) in three different cell lines is shown in Fig. 1. In HEp-2 cells, both rA2DM2-2 viruses grew poorly, with peak titres .100-fold lower than titres of wt RSV A2 (Fig. 1a). These results were in alignment with previous reports describing reduced growth of rA2DM2-2 in human cell lines such as Hep-2 (Jin et al., 2000;Bermingham & Collins, 1999). Next, we compared growth in a Vero cell line obtained from ATCC as well as a SF-adapted Vero cell line used in the manufacturing of a clinical trial vaccine (Yuk et al., 2006). A previous report demonstrated that growth of rA2DM2-2 in Vero cells was equivalent to wt RSV A2 (Jin et al., 2000). In our experiments in Vero cells, rA2DM2-2(MEDI) had faster growth kinetics compared with rA2DM2-2(NIH) (Fig. 1b). On day 2 post-infection (p.i.), rA2DM2-2(MEDI) had a titre of 6.3 log 10 p.f.u ml 21 and rA2DM2-2(NIH) had 4.7 log 10 p.f.u ml 21 , although both rA2DM2-2(MEDI) and rA2DM2-2(NIH) reached a titre of approximately 6.5 log 10 p.f.u ml 21 on day 5. The difference in growth kinetics was even more evident in SF Vero cells, where rA2DM2-2(MEDI) had 100-fold higher titre than rA2DM2-2(NIH) by day 2 p.i. (Fig. 1c). In SF Vero cells, rA2DM2-2(MEDI) reached a peak titre of 6.6 log 10 p.f.u ml 21 , whilst rA2DM2-2(NIH) reached a peak titre of 4.6 log 10 p.f.u ml 21 .
In addition to the differences in growth kinetics, these viruses showed marked differences in CPE. Vero cells infected with rA2DM2-2(MEDI) showed large syncytia, which spread over the entire cell monolayer by 48 h p.i. (Fig. 1d). In contrast, the rA2DM2-2(NIH) virus produced small syncytia that were associated with phase-bright cell clusters (Fig. 1e). Although these phase-bright cell clusters increased in number over time, the rA2DM2-2(NIH) virus never generated the larger syncytia characterized by rA2DM2-2(MEDI). Similar differences in CPE were observed in the SF Vero cell line. These results showed that the rA2DM2-2(MEDI) virus had faster growth kinetics and generated larger syncytia in Vero cells compared with rA2DM2-2(NIH).

Identification of K66E as the major genetic determinant for altered growth in Vero cells
In order to identify the genetic determinants responsible for the growth differences between these two viruses, we performed an alignment of their cDNA sequences. Although both rA2DM2-2(MEDI) and rA2DM2-2(NIH) are derived from RSV strain A2, there are differences in their genomic sequences and in the manner of the M2-2 deletion. The rA2DM2-2(MEDI) virus was generated by deletion of 234 nt encoding the C-terminal 78 aa of the M2-2 protein (Jin et al., 2000). The rA2DM2-2(NIH) virus was generated by deletion of 241 nt from the same region. The results of the cDNA alignment identified 34 nt differences: four differences encoding amino acid changes in the NS2, N and F genes, 15 differences in coding regions that did not alter amino acid sequence and eight differences in the non-coding regions and the remainder in the M2-2 deletion ( Table 1).
The four nucleotide differences encoding amino acid changes were introduced individually into the rA2DM2-2(MEDI) cDNA. A fifth cDNA was generated in which the M2-2 gene deletion in rA2DM2-2(MEDI) was changed to mimic the analogous deletion in rA2DM2-2(NIH). Four rA2DM2-2(MEDI) viruses each carrying a single amino acid change (R51K in NS2,A24T in N,K66E in F and Q101P in F) and one carrying the M2-2 deletion of rA2DM2-2(NIH) were generated from these cDNAs by reverse genetics for comparison of growth kinetics and CPE.
In contrast, the variants harbouring R51K in NS2, A24T in N, Q101P in F and the NIH M2-2 deletion sequence all generated CPE and peak titres that were similar to rA2DM2-2(MEDI) (Fig. 2). These results suggest that the K66E change in the F protein is the major genetic determinant for the reduced growth and altered CPE of rA2DM2-2(NIH) as compared with rA2DM2-2(MEDI).

A change at aa 66 in RSV F alters fusion activity
In order to analyse the fusion activity of the RSV F protein outside the context of virus replication, a sequenceoptimized version of the RSV F gene was cloned into plasmid pCMV-Script. Transfection of Vero cells with this plasmid carrying a lysine (K) at aa 66 in the RSV F gene (pF/66K) (Fig. 3a) generated large syncytia by 72 h (Fig.  3b). In contrast, Vero cells transfected with the same plasmid carrying a glutamic acid residue (E) in the RSV F gene at aa 66 (pF/66E) (Fig. 3a) formed only small syncytia (Fig. 3b). The differences in syncytium formation observed in Vero cells transfected with pF/66K and pF/66E matched the CPE differences seen in Vero cells infected with the rA2DM2-2(MEDI) and rA2DM2-2(NIH) viruses, respectively. These results suggested that a single amino acid at position 66 in RSV F may play an important role in promoting fusion.

A positive charge at aa 66 promotes fusogenicity
To determine whether the charge at position 66 was responsible for the differences in fusion, we generated additional pCMV/RSVF plasmids carrying either a positively charged arginine (pF/66R) or a negatively charged aspartic acid (pF/66D) at position 66. Transfection experiments showed that the RSV F mutant containing 66R produced large syncytia by 48 h, whereas the RSV F mutant containing 66D produced only small syncytia (Fig. 4a). Thus, the 66K and 66R substitutions with positively charged side chains promoted fusion, whilst 66E and 66D substitutions with negatively charged side chains hindered fusion.
To further test the influence of amino acid side chain on fusion, RSV F mutants containing substitutions with uncharged side chains were similarly generated. Alanine (A) and serine (S) are commonly used in functional mutagenesis studies and carry very short side chains that are hydrophobic and neutral, respectively. Neither of these substitutions restored fusogenicity to the levels of the 66K or 66R mutants (Fig. 4a). At the opposite end of the spectrum, proline (P) and tyrosine (Y) have side chains that contain large, bulky rings that could affect secondary structure or exert steric hindrance. As expected, neither 66P nor 66Y substitution enhanced the fusogenicity of RSV F (Fig. 4a). Interestingly, a glutamine (Q) substitution at residue 66 also failed to promote fusogenicity (Fig. 4a). Glutamine is structurally similar to glutamic acid except for the charge of the side chain, with glutamine carrying a neutral side chain and glutamic acid (E) carrying a negatively charged side chain. This indicated that simple removal of the negative charge at residue 66 was not sufficient to restore the fusogenicity of the 66E mutant to the level of the 66K mutant. Taken together, these results strongly suggested that electrostatic charge interactions at position 66 in the F 2 fragment of RSV F play a role in fusion.
The RSV F protein is initially produced as a full-length precursor (F 0 ), which is cleaved by a furin-like protease to form two disulfide-linked fragments (F 1 and F 2 ) of~50 and~25 kDa, respectively. To confirm that the level of expression and proteolytic cleavage was equivalent among the different RSV F mutants, SDS-PAGE and Western blotting was performed on lysates of transfected Vero cells. Blots probed with motavizumab, an anti-RSV F mAb, to visualize F 0 and F 1 indicated that the mutants had similar levels of expression and processing at the furin cleavage site (Fig. 4b). Blots were reprobed with anti-b-actin to show equivalent amounts of protein loaded in each lane (Fig. 4b).
As different levels of RSV F on the cell surface could also have an effect on syncytium formation, we compared cellsurface expression levels of the various F mutants using flow cytometry. 293T cells were transfected with each plasmid, stained with motavizumab to detect cell-surface RSV F and subjected to FACS analysis. Surprisingly, the two constructs that caused the most cell-to-cell fusion, pF/ 66K and pF/66R, actually showed slightly less RSV F on the cell surface compared with cells that had been transfected with the other F plasmids (Fig. 4c). These results suggested that the larger syncytia produced by pF/66K and pF/66R were not due to a larger amount of F protein on the cell surface but were due to the ability of a positively charged residue at position 66 in F to promote fusion.

DISCUSSION
rA2DM2-2(NIH) and rA2DM2-2(MEDI) are two RSV vaccine candidates that are attenuated by deletion of the M2-2 ORF. Although both viruses are derived from RSV strain A2, they carry four amino acid differences and have different deletions of the M2-2 gene. In a SF-adapted Vero cell line employed for manufacturing clinical trial vaccines, rA2DM2-2(MEDI) grew to 100-fold higher titres than rA2DM2-2(NIH), with markedly larger syncytia. Genetic mapping demonstrated that an amino acid difference at position 66 of the F protein was the major determinant responsible for the observed growth and fusion differences seen between these two viruses. Moreover, syncytiumformation experiments showed that positively charged amino acids at position 66 promoted fusion, whilst negatively charged amino acids hindered fusion activity, suggesting that electrostatic charge interactions at aa 66 of the F 2 fragment play an important role in fusion. Table 1. Genetic differences between rA2DM2-2(MEDI) and rA2DM2-2(NIH) Numbering is based on rA2DM2-2(MEDI) sequence. The K66E mutation has been described previously as part of a group of 'HEK' mutations identified following seven passages of a 1961 RSV A2 clinical isolate in human embryonic kidney (HEK) cells (Connors et al., 1995;Whitehead et al., 1998). The growth and CPE differences that we observed in both Vero and SF Vero cells were surprising, because the HEK-passaged virus from which rA2DM2-2(NIH) is derived was shown previously to grow well in Vero cells (Teng et al., 2001;Teng & Collins, 1999). In addition, a low level of natural polymorphism at aa 66 in the F sequence has been observed in circulating wt RSV strains with no reported deleterious effect on viral fitness (Zhu et al., 2012   Single change in RSV F alters growth and fusogenicity affect virus replication and fusion. Deletion of the M2-2 gene, which regulates the switch from transcription to replication, results in an imbalance that favours transcription and, in turn, viral protein expression (Bermingham & Collins, 1999). It is possible that this imbalance facilitated the identification of the K66E functional mutation described here. In addition, rA2DM2-2 virus has been shown to grow well in non-human cell lines such as Vero, which made it easier to identify a mutation that hindered growth as opposed to a cell line such as Hep-2 where growth of rA2DM2-2 virus is already significantly impaired (Jin et al., 2000).

Nucleotide position Genotype (MEDI A NIH) Mutation
Aa 66 is located in the F 2 fragment of the fully processed RSV F protein. Much of the published work on the function of RSV F has focused on the role of specific domains in the larger, membrane-attached F 1 fragment, such as the fusion peptide and heptad repeats A and B (Fig.  3a). Studies on the function of specific residues within the F 2 fragment have been limited to the role of moieties such as the cysteine (C) residue at position 69, which is involved in intermolecular disulfide bond formation, and the asparagine (N) residue at position 70, which is a potential site for N-glycosylation (Day et al., 2006;Zimmer et al., 2001). Although aa 66 is in close sequence proximity to both C69 and N70, no specific functional role has been assigned to this residue. Our syncytium-formation experiments showed that a change at aa 66 of RSV F can have a significant impact on fusion, with basic positively charged amino acids (K or R) promoting fusion and acidic negatively charged amino acids (E or D) impairing fusion. Because the differences in fusion activity seen between the F mutants cannot be explained by differences in proteolytic processing or cell-surface expression, we believe that amino acid changes at position 66 are probably having a direct effect on the fusogenicity of F.
The F 2 fragment was identified previously by Schlender et al. (2003) as being responsible for the host-cell specificity of RSV, suggesting that it is exposed and available for direct contact with host cells during virus infection. In more recent work by McLellan et al. (2013), a pre-fusion structure model of RSV F was generated by co-crystallization with an antibody specific for the pre-fusion form. In this model, aa 66 is localized to the outer surface of the homotrimer near the top of the head region (Fig. 5a). The structure model of the post-fusion form also places aa 66 on the outer surface of the homotrimer (Fig. 5b). This postulated location of aa 66 allows us to propose two mechanisms by which disruption of charge could alter fusion activity.
The first hypothesis suggests that a change in charge at aa 66 alters the ability of F to bind cell-surface receptors, thereby influencing syncytium formation and spread of the virus. Nucleolin and glycosaminoglycans (GAGs) have been identified as potential cell-surface receptors for RSV virus, and there is evidence that RSV F alone can also bind GAGs (Hallak et al., 2000a, b;Feldman et al., 2000;Martínez & Melero, 2000;Tayyari et al., 2011). Because GAGs such as heparan sulfate are negatively charged, one could hypothesize that increasing the positive charge on the outer surface of the F protein enhances virus binding and, in turn, fusion. A study by Feldman et al. (2000) identified a putative heparinbinding domain within the F 2 fragment that included aa 66, whilst work by Crim et al. (2007) using overlapping, linear peptides showed that RSV F peptides encompassing aa 66 could bind to GAGs and to Vero cells. However, binding of these peptides to Vero cells failed to inhibit subsequent binding of RSV (Crim et al., 2007). In addition, recently published experiments by McLellan et al. (2013) demonstrated that binding of a mAb to a pre-fusion epitope that included aa 66 had no effect on viral attachment, suggesting that this residue does not play an important role in binding of RSV to host cells.
The second hypothesis proposes that the charge of the amino acid at position 66 in RSV F affects local intra-and/ or intermolecular electrostatic interactions and, in turn, the ability to transition from pre-to post-fusion conformation. Gardner & Dutch (2007) identified a region spanning the C terminus of the F 2 fragment that is relatively well conserved in a variety of paramyxoviruses and found that mutations in this conserved region affected fusogenicity. Chang et al. (2012) also demonstrated the importance of charged residues in the F 2 fragment for electrostatic interactions and the overall stability of the human metapneumovirus F protein.
In both the pre-fusion (Fig. 5a) and post-fusion (Fig. 5b) models of RSV F, aa 66 is located on an exposed loop that is not in close proximity to known functional domains (McLellan et al., 2011(McLellan et al., , 2013. Our analysis of these structure models failed to identify any potential sidechain interactions between aa 66 and neighbouring residues, making speculation on the effect of the K66E mutation difficult. The overall charge distribution in the region surrounding aa 66 is highly positive; therefore, insertion of a negatively charged residue could stabilize the pre-fusion structure and, in turn, increase the threshold for triggering. Alternatively, the slight inward shift of the loop containing residue 66 between the pre-and post-fusion structure models raises the possibility that the side chain of aa 66 is interacting with other unknown residues during the massive structural rearrangement that constitutes fusion. Our work has demonstrated that a change in charge at aa 66 can have a significant impact on the fusogenicity of RSV F; however, elucidation of structural intermediates of fusion may be required in order to understand fully the precise role of this residue.

METHODS
Cell lines and virus. Vero cells (ATCC) were maintained in minimal essential medium (MEM; Gibco) supplemented with 5 % heatinactivated FBS (Hyclone), 2 mM L-glutamine (Gibco), and 100 U penicillin ml 21 and 100 mg streptomycin ml 21 (BioWhittaker). SFadapted Vero cells have been described previously (Yuk et al., 2006) and were maintained in OptiPro SFM (Gibco) supplemented with 2 mM Lglutamine, 100 U penicillin ml 21 and 100 mg streptomycin ml 21 . 293T cells (ATCC) were maintained in Dulbecco's MEM (Gibco) supplemented with 10 % heat-inactivated FBS, 2 mM L-glutamine, 100 penicillin U ml 21 and 100 mg streptomycin ml 21 . BSR/T7 cells (kindly provided by K. K. Conzelmann, Ludwig-Maximilians-Universitat Munich) were maintained in Glasgow MEM (Gibco) supplemented with 10 % heatinactivated FBS, 2 % tryptone-phosphate broth (Sigma) and 100 mg geniticin (Gibco) ml 21 . All cell lines were cultured at 37 uC, 5 % CO 2 . wt RSV strain A2 (ATCC) was passaged in Vero cells. Plasmids. Two subclones spanning the areas of interest were used to make nucleotide changes in the full-length rA2DM2-2(MEDI) cDNA. The subclones were derived from plasmid pA2DM2-2 as described previously (Jin et al., 2000). The first subclone was generated by digesting pA2DM2-2 with KpnI and XhoI (NEB) and ligating the 4482 bp fragment into plasmid pCITE-2a (Agilent). The resulting clone was designated pCITERSV/K-X, and comprised nt 1-4482 of the rA2DM2-2(MEDI) cDNA. The second subclone was generated by digesting plasmid pA2DM2-2 with XhoI and BamHI (NEB) and ligating the 3785 bp fragment into plasmid pCR-2.1 (Life Technologies). The resulting subclone was designated pCR2.1RSVDM2-2/X-B and comprised nt 4482-8267 of the rA2DM2-2(MEDI) genome. Nucleotide changes in each subclone were made using a QuikChange Site-Directed Mutagenesis kit, following the manufacturer's instructions (Agilent). Nucleotide changes were confirmed by sequencing, and the fragments were inserted back into the full-length pA2DM2-2 cDNA using the same paired restriction enzymes described above for each subclone. For transfection experiments requiring expression of the full-length RSVA2 F protein, the 1725 bp sequence of the RSV F ORF was optimized at MedImmune and the DNA synthesized by DNA2.0. The ORF was amplified by PCR and cloned into plasmid pCMV-Script (Agilent). This plasmid was designated pCMV/RSVF (pF). Nucleotide changes in the RSV F sequence were made using a QuikChange Site-Directed Mutagenesis kit. The following day, the transfection mix was removed and replaced with 2 ml Opti-MEM I. After 5 days' incubation at 35 uC in a 5 % CO 2 incubator, the cells and supernatant were harvested together as described above. Rescued virus was amplified by two to three passages in Vero cells. Viral titres were determined by plaque assay.
The sequence of each recombinant virus was confirmed by reverse transcription (RT)-polymerase chain reaction (PCR). Briefly, the viral RNA was isolated using a Qiamp Viral RNA Mini kit (Qiagen). RT-PCR was performed using a OneStep RT-PCR kit (Qiagen) and oligonucleotide primers that generated overlapping PCR products covering the entire genome. Gel-extracted PCR products (Qiagen) were sent to Sequetech Inc. for sequencing.
Plaque assay. Virus stocks were serially diluted and 0.5 ml of each dilution was used to infect one well of a six-well plate containing subconfluent Vero cells. After 1 h of rocking at room temperature, the virus was aspirated and wells were overlaid with a 1 : 1 mixture of 2 % methylcellulose and 26 L-15/EMEM (SAFC) supplemented with 2 % heat-inactivated FBS, 4 mM L-glutamine, 200 U penicillin ml 21 and 200 mg streptomycin ml 21 . Plates were incubated at 35 uC in a 5 % CO 2 incubator. After 5-6 days' incubation, the overlay was removed by aspiration, the plates were fixed in methanol and the fixed cells were immunostained using polyclonal anti-RSV antibody (Millipore) diluted 1 : 1000 in 5 % powdered milk (w/v) in PBS, followed by HRP-conjugated rabbit anti-goat antibody (Dako). Plaques were visualized with 3-amino-9-ethylcarbazole (Dako). Virus titre is reported as p.f.u. ml 21 .
Multicycle growth analysis of recombinant rRSVA2DM2-2 virus. Six-well plates of subconfluent Vero cells were infected at m.o.i. of 0.1 in 0.5 ml Opti-MEM I per well. Plates were rocked at room temperature for 1 h to facilitate virus absorption and washed once with Opti-MEM I, followed by the addition of 2 ml per well of fresh Opti-MEM I. Plates were incubated at 35 uC with 5 % CO 2 , and virus was harvested at 24 h intervals as described above. Samples were stored at 270 uC and titrated by plaque assay as described above.
Syncytium-formation assay. Subconfluent Vero cells in six-well plates were transfected overnight with 1 mg per well of plasmid pF or its derivatives. Briefly, transfection mix was generated by mixing 4 ml Lipofectamine 2000 per 1 mg plasmid DNA in a final volume of 0.2 ml Opti-MEM I. Cells were washed once and 0.5 ml Opti-MEM I was added per well, followed by 0.2 ml transfection mix per well.
After overnight incubation at 37 uC with 5 % CO 2 , the plates were washed and 2 ml per well of Opti-MEM I was added before continued incubation at 37 uC. Syncytium formation was examined at various time points post-transfection, and images were captured using a Nikon Eclipse TS100 microscope.
Western blotting. Six-well plates of Vero cells were transfected as described above. At 48 h post-transfection, cell lysates were harvested by aspirating the medium, washing the well with PBS and adding 0.3 ml Laemmli buffer plus b-mercaptoethanol directly to each well. Before loading onto 12 % Tris/glycine SDS-PAGE gels, lysates were incubated at 95 uC for 10 min. Gels were blotted onto PVDF membranes (Life Technologies) and probed with motavizumab (0.1 mg ml 21 ), a highly potent mAb against RSV F (Wu et al., 2007). This step was followed by incubation with HRP-conjugated anti-human secondary antibody (Dako) and electrochemiluminescent (ECL) detection using Supersignal Dura West ECL substrate (Pierce) and an ImageQuant LAS4000 imager (GE Healthcare). Blots were then rinsed with PBS and reprobed with an anti-chicken b-actin mAb (Millipore) to normalize for protein loading. This was followed by HRP-conjugated anti-mouse secondary antibody (Dako) and ECL detection as described above.
Immunofluorescence. Vero cells were seeded to 90 % confluency in 12-well plates containing sterile glass coverslips. Transfections were performed as described above but scaled down for 12-well plates. At 48 h post-transfection, cells were fixed with 4 % paraformaldehyde in PBS for 20 min at room temperature. Plates were blocked with PBS containing 1 % BSA for 1 h at 37 uC followed by incubation with motavizumab (0.5 mg ml 21 in PBS containing 1 % BSA and 0.1 % saponin) for 1 h at 37 uC. Plates were washed with PBS/Tween 20 and incubated with Alexa Fluor 488-conjugated goat anti-human IgG (Life Technologies; 1 mg ml 21 in PBS containing 1 % BSA and 0.1 % saponin). After 1 h at 37 uC, the plates were washed with PBS/Tween 20. Coverslips were inverted and mounted on glass slides using Vectashield mounting medium with DAPI (Vector Laboratories). Images were captured at 64 magnification using a Nikon Eclipse 80i microscope with a CoolSnapES2 camera and Simple PCI6 software.
Flow cytometry. To assess cell-surface expression of RSV F, 293T cells were transfected as described above. At 20 h post-transfection, cells were stained for FACS analysis using motavizumab (1 mg ml 21 ) followed by Alexa Fluor 488-conjugated goat anti-human IgG antibody (1 mg ml 21 ), each diluted in FACS Stain Buffer (BD Pharmingen). Cells were fixed with Cytofix Fixation Buffer (BD Pharmingen) and analysed on an LSR-II flow cytometer (BD Biosciences). Mean fluorescence intensity was determined using FACSDiva software.
Statistical analysis. Growth-curve data were analysed by t-tests using a mixed-effect model with correlated error structure. FACS data were also analysed by t-tests using a mixed-effect model with heterogeneous group variability.