Effects of Alterations to the CX3C Motif and Secreted Form of Human Respiratory Syncytial Virus (RSV) G Protein on Immune Responses to a Parainfluenza Virus Vector Expressing the RSV G Protein

Human RSV is the leading viral cause of severe pediatric respiratory illness. An RSV vaccine is not yet available. The RSV attachment protein G is an important protective and neutralization antigen. G contains a conserved fractalkine-like CX3C motif and is expressed in mG and sG forms. sG and the CX3C motif are thought to interfere with host immune responses, but this remains poorly characterized. Here, we used an attenuated chimeric bovine/human parainfluenza virus type 3 (rB/HPIV3) vector to express various modified forms of RSV G. We demonstrated that strong antibody and protective responses could be induced by G alone, and that this was highly dependent on the integrity of the CX3C motif. There was no evidence that sG or the CX3C motif impaired immune responses against RSV G or the rB/HPIV3 vector. rB/HPIV3 expressing wt RSV G provides a bivalent vaccine against RSV and HPIV3.

ABSTRACT

Human respiratory syncytial virus (RSV) is a major pediatric respiratory pathogen. The attachment (G) and fusion (F) glycoproteins are major neutralization and protective antigens. RSV G is expressed as membrane-anchored (mG) and -secreted (sG) forms, both containing a central fractalkine-like CX3C motif. The CX3C motif and sG are thought to interfere with host immune responses and have been suggested to be omitted from a vaccine. We used a chimeric bovine/human parainfluenza virus type 3 (rB/HPIV3) vector to express RSV wild-type (wt) G and modified forms, including sG alone, mG alone, mutants with ablated CX3C, and G with enhanced packaging into vector virions. In hamsters, these viruses replicated to similar titers. When assayed with a complement-enhanced neutralization assay in Vero cells, sG did not reduce the serum RSV- or PIV3-neutralizing antibody (NAb) responses, whereas ablating CX3C drastically reduced the RSV NAb response. Protective efficacy against RSV challenge was not reduced by sG but was strongly dependent on the CX3C motif. In ciliated human airway epithelial (HAE) cells, NAbs induced by wt G, but not by wt F, completely blocked RSV infection in the absence of added complement. This activity was dependent on the integrity of the CX3C motif. In hamsters, the rB/HPIV3 expressing wt G conferred better protection against RSV challenge than that expressing wt F. Codon optimization of the wt G further increased its immunogenicity and protective efficacy. This study showed that ablation of the CX3C motif or sG in an RSV vaccine, as has been suggested previously, would be ill advised.

IMPORTANCE Human RSV is the leading viral cause of severe pediatric respiratory illness. An RSV vaccine is not yet available. The RSV attachment protein G is an important protective and neutralization antigen. G contains a conserved fractalkine-like CX3C motif and is expressed in mG and sG forms. sG and the CX3C motif are thought to interfere with host immune responses, but this remains poorly characterized. Here, we used an attenuated chimeric bovine/human parainfluenza virus type 3 (rB/HPIV3) vector to express various modified forms of RSV G. We demonstrated that strong antibody and protective responses could be induced by G alone, and that this was highly dependent on the integrity of the CX3C motif. There was no evidence that sG or the CX3C motif impaired immune responses against RSV G or the rB/HPIV3 vector. rB/HPIV3 expressing wt RSV G provides a bivalent vaccine against RSV and HPIV3.

INTRODUCTION

Respiratory syncytial virus (RSV) is an enveloped nonsegmented negative-strand RNA virus in the family Pneumoviridae, genus Orthopneumovirus. It is the most common viral cause of bronchiolitis and pneumonia among children in their first year of life (1, 2). RSV also can cause symptomatic reinfections, usually with reduced disease, throughout life, especially among young children, the elderly, and those with compromised cardiac, pulmonary, or immune systems. Passive immunoprophylaxis currently is available for use in children at high risk for severe RSV disease, such as infants with prematurity, bronchopulmonary dysplasia, or congenital heart disease (3). Despite the burden of RSV infection, especially in young children, no licensed RSV vaccine or antiviral drug suitable for routine use is available.

RSV has a single serotype and two antigenic subgroups, A and B (4, 5). Its tropism is highly restricted to the respiratory tract, and infection occurs primarily in the superficial mucociliary airway epithelial cells and pneumocytes (6). The viral attachment protein (G) and fusion protein (F) are transmembrane surface glycoproteins that are the only known targets for neutralizing antibodies (NAbs) and are the most important protective antigens. RSV F is highly conserved and is essential for infectivity, while RSV G is more variable and is not required for infection and replication in immortalized cell lines (7, 8) but is required for efficient infection in cultures of primary ciliated human airway epithelial (HAE) cells and in vivo (9–11). Antibodies against RSV G reduce RSV viral load and disease severity upon challenge in animal models (12–17). Clinically, a higher concentration of RSV G antibodies in serum is associated with reduced severity of RSV disease in infants and young children (18). Thus, RSV G induces immune protection that is clinically important.

Full-length RSV G protein (mG) is a type II transmembrane glycoprotein that has an N-terminal cytoplasmic tail (CT; predicted to comprise amino acids [aa] 1 to 37 in strain A2 [Fig. 1A and B]; note that all numbering is relative to that of strain A2), a hydrophobic transmembrane domain (TM; comprising approximately amino acids 38 to 65 [Fig. 1A and B]), and a C-terminal ectodomain (comprising approximately amino acids 66 to 298). RSV G also is expressed as a secreted form (sG) that is produced by alternative translation initiation at the second AUG codon (M48) in the open reading frame (ORF), whose corresponding position in the protein lies within the TM domain (Fig. 1A and B) (19–21). The N terminus is then subjected to intracellular proteolytic trimming that creates a new N terminus at N66 (Fig. 1B) (19, 20). The G ectodomain consists of two large divergent domains that flank a short central conserved region (CCR) at amino acids 164 to 186 (22). The two divergent domains are called “mucin-like” because, like mucin, they have a high content of proline, alanine, threonine, and serine amino acids and a high content of carbohydrate; e.g., strain A2 has an estimated four N-linked and 24 to 25 O-linked carbohydrate side chains (7). The CCR contains a cystine noose (i.e., a tight turn, stabilized in this case by two disulfide bonds) that bears a conserved CX3C motif (CWAIC, aa 182 to 186). Apart from the truncated N terminus of sG, the mG and sG forms are believed to be essentially the same with regard to glycosylation and protein structure, except that mG forms a multimer that probably is a trimer or tetramer, whereas sG remains a monomer (23, 24).

FIG 1.

Diagrams of RSV G protein mutants and rB/HPIV3 vector. (A) Diagram of wt G of RSV strain A2 showing its functional domains and motifs. CT and TM are the presumed cytoplasmic tail and transmembrane domains at the N terminus. Two mucin-like domains (I and II) are shown. The following features are indicated: the central highly conserved region between amino acids 164 and 176 (green); the two disulfide bonds (S-S) of the cystine noose; the heparin binding domain (HBD) between amino acids 184 and 198 (orange) (62); the second start codon (Met 48) initiating the expression of the secreted G (sG); and the CX3C motif (amino acids 182 to 186) and its sequence. The map is not to scale. (B) Amino acid sequences of the TM and CT domains of RSV wt G protein (top). Modified versions of RSV G, namely, mG, sG, G_B3CT, and G_B3TMCT, and the wt BPIV3 HN (strain Kansas), were used. The sequences were aligned according to the beginning of the G ectodomain at amino acid 66. The presumed CT, TM, and ectodomains are demarcated with dashed lines. Sequences from RSV G and BPIV3 HN are distinguished in red and blue, respectively. Note that the N terminus indicated for sG is that of the primary translation product, which subsequently gets trimmed proteolytically to yield a predominant N terminus at N66 and a secondary N terminus at I75 that is not shown. The location of the second start codon (Met 48) is underlined. The sG silencing mutations (M48I/I49V) in the mG construct are shown above the protein sequence. (C) Diagram of the rB/HPIV3 vector with inserts of the various G mutants. The rB/HPIV3 gene map is shown at the top, and the second line shows the details of the sequence flanking the RSV G gene inserted at the second gene position, between the vector N and P genes, using the indicated AscI sites (underlined), with the following sequence features indicated: N GE, gene end signal from the BPIV3 N gene; P GS, gene start signal from the BPIV3 P gene. Also shown are the ATG initiation codon (boldface) and TGA stop codon (underlined) of the G ORF. Diagrams of the following RSV G constructs are shown: (i) wt G, unmodified wt RSV G, annotated to indicate the M48 that can serve as an alternative translational start site to generate sG, and CWAIC, the CX3C motif at amino acids 182 to 186; (ii) mG, expressing only the transmembrane form of RSV G (mG) due to the M48I mutation that ablates expression of secreted G (sG); (iii) sG, expressing only sG, due to deletion of the N-terminal 47 codons, so that the ORF begins with the M48 codon; (iv) G_B3CT, with the CT of RSV G replaced by the CT of BPIV3 HN, the region in blue; (v) G_B3TMCT, with the TM and CT of RSV G replaced by the TM and CT of BPIV3 HN, the region in blue; (vi) G_dCX3C, with the C186R mutation in the CX3C motif (yielding CX3R); (vii) G_wCX4C, with the addition of A186 to change the CX3C motif (yielding CX3AC); (viii) G_dCX3C_B3CT, with the C186R mutation in the CX3C motif and bearing the CT of BPIV3 HN; (ix) G_dCX3C_B3TMCT, with the C186R mutation in the CX3C motif and bearing the TM and CT of BPIV3 HN. As indicated, some G ORFs contain an additional TAG stop codon, indicated with three asterisks. This trinucleotide was added when needed to adjust the sequence length to conform to the “rule of six” (63). The map is not to scale.

A CX3C motif is characteristic of the chemokine called fractalkine, and the sequences flanking the CX3C domains in RSV G and fractalkine also share sequence relatedness (25). The G protein has been shown to mimic fractalkine in the ability to induce leukocyte chemotaxis in vitro. The G protein can bind to the fractalkine receptor CX3CR1, which is expressed on the apical surface of airway epithelial cells and certain immune cells, to initiate RSV infection (25, 26). Like the G protein, fractalkine is expressed as a full-length transmembrane form and a truncated secreted form (27). For fractalkine, the full-length transmembrane form acts as an adhesion molecule that interacts with the fractalkine receptor CX3CR1 expressed on T cells, NK cells, and monocytes, and the secreted form acts as a chemoattractant for the same cell types (27–29). The role of the CX3C domain in RSV attachment seems straightforward (i.e., allowing the virus to bind to cells expressing CX3CR1), but the effects of the sG and the CX3C domain in the G protein on host immune responses remain unclear. In principle, ablation of the CX3C domain and/or the expression of sG would be expected to reduce the chemotactic influx of immune cells and might reduce disease, and this is supported by several studies with mutant RSVs (30, 31), although there also are contradictory data (32). Other effects of G also have been described. For example, sG was shown to act as an antigen decoy to reduce neutralization by antibodies and also appeared to interfere with clearance of RSV by macrophages and complement (33, 34). As another example, in vitro, the G protein interfered with human dendritic cell activation (35). In addition, the binding of RSV to CX3CR1 expressed on human neonatal regulatory B cells leads to a Th2-polarized response (36). The effects of mutations to the CX3C motif in the G protein have been difficult to evaluate using recombinant RSV, because these mutations can alter RSV replication in vivo, which in itself affects immune responses.

rB/HPIV3 is a fully replication-competent virus that consists of a bovine PIV3 (BPIV3) backbone with the BPIV3 F protein and hemagglutinin-neuraminidase (HN) replaced by their human PIV3 (HPIV3) counterparts (Fig. 1C). rB/HPIV3 is attenuated in nonhuman primates and humans due to the BPIV3 backbone, and a version of rB/HPIV3 expressing wild-type (wt) RSV F was shown to be well tolerated when administered intranasally in young children as an experimental bivalent vaccine against RSV and HPIV3 (37). This favorable safety profile will expedite clinical evaluation of improved vaccines based on this vector. In the present study, we used rB/HPIV3 as a vector to express various modified forms of RSV G with silenced sG expression, mutated CX3C motif, enhanced packaging in the vector virions, and increased expression by codon optimization. Using a hamster model, we evaluated the effects of these mutations on the induction of RSV-specific and PIV3-specific serum neutralizing antibodies (NAbs) and protection against RSV challenge. Because PIV3 is a respiratory virus that is related to RSV and shares features of tropism and replication with RSV, analysis of the replication of PIV3 expressing various forms of RSV G in hamsters provided further means to evaluate effects of RSV G on the host pulmonary immune milieu during viral infection. We also measured the efficiency of RSV G-induced serum antibodies in blocking RSV infection in HAE cultures in the absence of complement compared to that with RSV F-induced serum antibodies, and we determined the quality of NAbs induced by RSV G and its dependence on the CX3C motif.

RESULTS

Construction of rB/HPIV3 vectors expressing modified forms of RSV G.

rB/HPIV3 was engineered to express the wt RSV G protein and eight G protein mutants. These constructs are shown in Fig. 1C. Construct i is wt RSV G (wt G). Constructs ii and iii encode mG and sG, respectively. The mG construct was made by M48I/I49V mutations previously shown to ablate the expression of sG without significantly altering the hydrophobicity of the TM domain (Fig. 1B) (9, 19), and the sG construct was made by deletion of the first 47 codons of the G ORF so that codon M48 initiates the ORF (Fig. 1B) (9, 19). Constructs iv and v (G_B3CT and G_B3TMCT) are chimeric proteins that have the CT and TMCT (i.e., TM and CT) of wt G replaced by those of the BPIV3 HN protein (a type II transmembrane protein with the same membrane orientation as that of G; Fig. 1B) in an attempt to promote efficient packaging of RSV G into the vector particles (38). Sequence boundaries of the TM and CT of BPIV3 HN were determined by inspection and alignment with HPIV3 HN (data not shown). Constructs vi and vii (G_dCX3C and G_wCX4C) have mutations that change the CX3C motif in the G protein. Specifically, G_dCX3C has a C186R mutation that changes the assignment of the second cysteine residue in the CX3C motif (Fig. 1C) (39), and G_wCX4C has the insertion of an alanine residue between positions 185 and 186 that disrupts the spacing of the motif (Fig. 1C) (30). Constructs viii and ix (G_dCX3C_B3CT and G_dCX3C_B3TMCT) have the C186R mutation in combination with the B3CT and B3TMCT substitutions.

Each RSV G construct was inserted between the N and P genes of the rB/HPIV3 vector flanked by BPIV3 gene start and gene end signals, so that the G insert would be expressed as a separate mRNA (Fig. 1C). All vectors with RSV G were successfully rescued and grew to high titers (7.6 to 8.6 log₁₀ 50% tissue culture infectious doses [TCID₅₀] per ml) in LLC-MK2 cells.

Expression of RSV G by rB/HPIV3 vectors.

The intracellular expression of RSV G by the various constructs was evaluated in Vero (Fig. 2A and B) and LLC-MK2 (data not shown) cells by Western blot analysis. Cells were infected with each vector at a multiplicity of infection (MOI) of 10 TCID₅₀ per cell or by wt RSV at an MOI of 3 PFU per cell.

FIG 2.

Western blot analysis of mutant RSV G proteins expressed in Vero cells from rB/HPIV3-RSV G vectors. Vero cells were infected with the indicated rB/HPIV3-RSV G vector or wt RSV at an MOI of 10 TCID₅₀ or 3 PFU per cell, respectively. At 24 h postinfection (h.p.i.), the cells were harvested for analysis, and at 48 h.p.i., the cell culture medium supernatants from duplicate cultures were harvested for analysis. The cells and medium supernatants were analyzed separately by gel electrophoresis under denaturing and reducing conditions, followed by Western blotting with antisera raised separately against RSV and HPIV3 and then by secondary antibodies conjugated with infrared fluorescent dyes. Densitometric analysis then was performed on the large predominant band of fully glycosylated, 90- to 120-kDa RSV G protein using an Odyssey imaging system (LI-COR). (A) Western blots of proteins from infected cells. (Top) Bars to the left indicate fully and partially glycosylated forms of RSV G (upper and lower bars, respectively). Lane 11 shows the RSV N, P, and M proteins expressed by wt RSV. (Middle) BPIV3 N protein. (Bottom) GAPDH protein used as a loading control. (B) Relative levels of fully glycosylated (90 to 120 kDa) RSV G quantified from the experiment shown in panel A by densitometry using LI-COR Image Studio software calibrated using the GAPDH signal and normalized to wt G (lane 2), set as 1.0. (C) Western blot of G protein in cell culture medium supernatants. (D) Relative levels of fully glycosylated G protein from panel C, normalized to wt G.

At 24 h postinfection (p.i.), cell-associated RSV G expressed by the rB/HPIV3 vector expressing wt RSV G (Fig. 2A, lane 2) and by wt RSV (lane 11) was detected by Western blotting as a predominant diffuse band of 90 to 120 kDa, which corresponds to the full-length, fully glycosylated form, and as less abundant bands of 35 to 50 kDa that represent processing intermediates of the G protein with incomplete O-glycosylation. Expression of the various G protein species described above by the mG construct was essentially the same as that for wt G (Fig. 2A and B, lane 4 versus 2). In contrast, consistent with efficient secretion, the sG construct had only trace intracellular amounts of a large diffuse band and the incompletely glycosylated bands, all of which were reduced in size due to the N-terminal truncation. The CT and TMCT substitutions increased expression of the large G band by 50% to 80% (Fig. 2A and B, lanes 5 and 6 versus 2), an effect that is unexplained, and an increase in expression was previously observed with the chimeric form of RSV F with its CT and TMCT replaced by their counterparts of the vector F (38). In contrast, mutation of the CX3C motif reduced the accumulation of the large G band by 20% to 50%, which appeared to be due at least in part to an increase in the accumulation of the 35- to 50-kDa incompletely glycosylated forms (Fig. 2A and B, lanes 7 and 8 versus 2). The reduction in accumulation of the large G band associated with the CX3C mutations was compensated for when combined with the B3CT and B3TMCT substitutions (Fig. 2A and B, lanes 9 and 10). These effects on the efficiency of glycosylation may reflect destabilization of the structure of G by the CX3C mutations and stabilization by the B3CT and B3TMCT substitutions. There were no significant differences in the expression of the BPIV3 N protein between the constructs, including the empty vector (Fig. 2A), suggesting that the presence of the various mutated G protein genes had little effect on vector gene expression in vitro. Intracellular expression of the RSV G and BPIV3 N protein in LLC-MK2 cells (data not shown) was similar to that in Vero cells.

Secretion of RSV G was evaluated in Vero cell cultures that were infected in parallel as described above and incubated for 48 h (Fig. 2C and D). The medium supernatant was collected and clarified by centrifugation to remove debris as well as virions. The clarified supernatants were then subjected to Western blot analysis with polyclonal antibodies against RSV and HPIV3. In the case of cells infected with rB/HPIV3 expressing wt RSV G (Fig. 2C, lane 2) or infected with RSV (lane 11), the only form of G protein that was detected in the clarified, virus-depleted medium was a large, diffuse band representing fully glycosylated sG, indicating that only the completely glycosylated form of sG was secreted into medium (Fig. 2C, lanes 2 and 11). As expected, very little G protein was detected in the medium from cells infected with the mG construct (Fig. 2C and D, lane 4), while sG protein produced by the sG construct was 70% more abundant than that of the wt G construct (Fig. 2C and D, lane 3 versus 2). The two versions of RSV G with mutated CX3C motif also expressed sG protein (Fig. 2C and D, lanes 7 and 8). The G_dCX3C mutant produced 50% less sG than wt G, likely due to its overall reduced expression (Fig. 2A and B, lane 7). The B3CT and B3TMCT substitutions completely abrogated the expression of sG protein (Fig. 2C and D, lanes 5, 6, 9, and 10). This perhaps was not surprising in the case of B3TMCT, since the native M48 codon that normally is used to initiate synthesis of sG protein is located in the TM domain of G that was replaced with the TM domain from HPIV3 HN protein (although two other AUG codons are present in the HPIV3 HN TM domain that apparently were not utilized; Fig. 1B). In the case of the B3CT substitution, the native TM with its M48 codon is present, and it is not known why it apparently was not utilized. One possibility is that the nucleotide sequence upstream of this region influences initiation at the M48 codon.

The stability of RSV G expression by the rB/HPIV3 constructs following passage in vitro was determined by a double-immunostaining plaque assay (Fig. 3A), similar to assays described previously for rB/HPIV3 constructs expressing the RSV F protein (40). Viral plaques that had formed on monolayers of Vero cells under a methylcellulose overlay were fixed and stained with primary rabbit anti-HPIV3 antibodies and goat anti-RSV antibodies, followed by the two respective species-specific secondary antibodies, each conjugated to a different infrared dye (HPIV3, green; RSV, red; Fig. 3A). Coexpression of HPIV3 antigens and RSV G by a single PFU was visualized as yellow. All of the vector stocks expressing RSV G were found to have almost 100% of PFU expressing RSV G, with the exception that this could not be determined in the case of the sG construct because the rapid secretion of G into the medium precluded staining of the plaques for this protein (Fig. 3A, lane 3).

FIG 3.

Double-immunostaining plaque assays characterizing the expression of modified forms of RSV G protein from rB/HPIV3-RSV G. rB/HPIV3 constructs expressing the indicated modified G proteins were inoculated in 10-fold dilution series on Vero cell monolayers in 24-well plates and incubated under methylcellulose overlay for 6 days. The monolayers were fixed with 80% methanol and analyzed with primary antibodies, followed by secondary antibodies conjugated to infrared dye. (A) Images of plaques that were probed with rabbit antisera raised against HPIV3 and a goat hyperimmune serum to RSV. HPIV3 antigens alone were visualized as green, RSV G protein alone as red, and coexpression as yellow. (B) Images of plaques that were probed with the same HPIV3-specific rabbit hyperimmune serum and RSV G MAb 131-2 G targeting the central conserved region. HPIV3 antigens alone are visualized in green, RSV G containing the 131-2 G epitope alone is red, and coexpression is yellow.

Additionally, a double-immunostaining plaque assay was performed using an RSV G monoclonal antibody (MAb; 131-2G) that binds to the CCR (Fig. 3B) (41, 42). This antibody bound to plaques for all of the versions of G in which the CX3C motif was intact (Fig. 3B, lanes 2, 4, 5, and 6), except for sG, which was secreted (lane 3) but did not bind to the versions with mutations in the CX3C motif (Fig. 3B, lanes 7 to 10). This indicated that mutation of the CX3C motif disrupted the 131-2 G binding epitope on RSV G.

Packaging of RSV G into rB/HPIV3 vector virions.

To evaluate the packaging efficiency of RSV G into rB/HPIV3 and RSV virions, vectors that were grown in LLC-MK2 cells and wt RSV that was grown in Vero cells were purified by centrifugation on 30% to 60% discontinuous sucrose gradients. Western blot analysis was performed to quantify the relative packaging efficiency of RSV G for each construct. For each construct, 4 μg of purified virus was analyzed (Fig. 4A), and the immunostained bands of G protein were quantified. This showed that the unmodified wt G was detectable in vector virions as a very faint band (Fig. 4A, lane 2), and its packaging efficiency was less than 5% per μg of virion protein that of G protein in wt RSV virions (Fig. 4A and 5B, lanes 2 and 6). Note that most of the G protein that was packaged in RSV virions grown in Vero cells had a lower molecular mass (50 to 60 kDa) than the normally predominant 90- to 120-kDa form. This was previously shown to arise due to cleavage by the cellular protease cathepsin L during long incubations in Vero cells (43) and was not seen with the vectors because they were grown in LLC-MK2 cells. The TMCT replacement with that of the BPIV3 HN protein enhanced the packaging efficiency of RSV G by 8.3-fold (Fig. 4A and B, lanes 2 and 5, G_B3TMCT construct), and the same magnitude of effect also was observed for the G_dCX3C_B3TMCT construct (not shown). Per μg of virion protein, the amount of RSV G protein packaged for the G_B3TMCT construct was 38% that of wt RSV. The BPIV3 HN CT substitution only marginally increased the packaging of RSV G into vector virions (Fig. 4A and B, lanes 2 and 4). As expected, sG protein was not packaged in the vector (Fig. 4A, lane 3). The enhanced incorporation of RSV G by the TMCT substitution did not significantly affect the packaging efficiency of PIV3 HN (Fig. 4A, bottom), suggesting the TMCT-modified form of RSV G was not strongly competitive against the packaging of the vector glycoprotein bearing the same TMCT domains.

FIG 4.

Quantification of RSV G packaging by Western blot analysis of purified rB/HPIV3-RSV G particles. Approximately 4 μg of each purified virus grown in LLC-MK2 cells and wt RSV grown in Vero cells were subjected to gel electrophoresis under denaturing and reducing conditions. Western blots were prepared and analyzed with antisera raised separately against RSV and HPIV3 or rabbit polyclonal antibodies against HPIV3 HN. (A) Western blot of empty rB/HPIV3 vector (lane 1), vector expressing wt G (lane 2), vectors expressing the indicated modified forms of RSV G (lanes 3 to 5), and wt RSV (lane 6). The upper panel was probed with polyclonal RSV antibodies, showing the fully glycosylated 90- to 120-kDa form of G, particularly evident in lane 5, and RSV structural proteins N, P, M, and G, including a smaller, broad band of G (∼48 to 62 kDa) often observed with Vero cells (lane 6), as previously described (43). The middle panel shows the BPIV3 N protein. The bottom panel shows the HPIV3 HN protein. (B) RSV G packaging efficiency based on quantification of panel A and normalized to wt G (lane 2), set as 1.0.

FIG 5.

Replication of rB/HPIV3-RSV G vectors in the upper and lower respiratory tracts of hamsters. Hamsters in groups of six were infected i.n. with 10⁵ TCID₅₀ of the indicated vectors. Nasal turbinates and lungs were collected on day 5 postimmunization and homogenized. Virus titers in nasal turbinates (A) and lungs (B) were determined by TCID₅₀ hemadsorption assays on LLC-MK2 cells, with each dot representing the titer of an individual animal. Mean titers and standard errors of the means (SEM) are shown as horizontal lines with error bars. The statistical significance of the differences in mean titers among all groups in the present study was analyzed by one-way analysis of variance (ANOVA), followed by the Tukey-Kramer test. All groups were designated with the same letter, A, because there were no statistically significant differences. The limit of detection of TCID₅₀ is indicated as a dashed line.

Vector replication in hamsters.

The effect of RSV G on vector replication in vivo was evaluated in a hamster model (Fig. 5). Hamsters in groups of six were immunized intranasally (i.n.) with 10⁵ TCID₅₀ of the indicated vector. Note that the vector inocula used in this study (10⁵ TCID₅₀ in Fig. 5 and 7 and 10⁴ TCID₅₀ in Fig. 10) are 10- to 100-fold lower than the usual inoculum of 10⁶ TCID₅₀ used for PIVs in hamsters (44, 45). We used a lower dose because we have found that the resulting lower immunogenicity provides greater sensitivity in detecting differences in immunogenicity between vectors. Lungs and nasal turbinates were collected on day 5 after immunization, and the titers of the rB/HPIV3 vectors were determined by TCID₅₀ hemadsorption assay on LLC-MK2 cell monolayers, whereas RSV titers were determined by plaque assay on Vero monolayers. In the nasal turbinates, all of the vectors bearing an RSV G insert replicated to similar titers as the empty rB/HPIV3 vector (Fig. 5A), whereas in the lungs they were slightly more attenuated than the empty vector, although the differences were not significant (Fig. 5B). There were no significant differences among the mean titers of the vectors expressing various forms of RSV G, allowing direct comparison of the immune responses in hamsters.

FIG 7.

Protection of immunized hamsters against wt RSV A2 challenge. The groups of hamsters were challenged 31 days postimmunization by i.n. inoculation with wt RSV at 10⁶ PFU per animal. On day 3 postchallenge, animals were sacrificed and viral titers were determined in homogenates of the nasal turbinates (NT) (A) and lungs (B) by plaque assay in Vero cells. The limit of detection is indicated with a dashed line. Each dot represents the titer of an individual animal. Mean titers are indicated as short horizontal bars, and error bars represent the SEM. The statistical significance of differences in mean titers among all groups was analyzed using one-way ANOVA followed by Tukey-Kramer test. Mean titers of groups designated with the same letter (A, B, C, or D) are not statistically different. The correlation of the serum RSV neutralization titers (log₂ PRNT₆₀) with challenge virus load (log₁₀ PFU/g) in the NT (C) and lungs (D) was evaluated by Pearson’s linear regression model using Prism 7.0. The solid lines indicate the fitted model. Calculated R² values, P values, and sample sizes (n) are indicated.

FIG 10.

Improving the immunogenicity of RSV G by codon optimization. (A to D) LLC-MK2 (A) and Vero (B) cells were infected with rB/HPIV3 expressing wt G/GS-opt or wt G at an MOI of 10 TCID₅₀ per cell or with wt RSV at an MOI of 3 PFU per cell. Cells were harvested at 48 h postinfection, and cell lysates were subjected to gel electrophoresis under denaturing and reducing conditions and analyzed by Western blotting using polyclonal antibodies against RSV. Expression of G (the large 90- to 120-kDa fully glycosylated form) in LLC-MK2 (C) and Vero (D) cells was quantified by densitometry using Image Studio software (Li-Cor). (E to G) The immunogenicity of RSV G was evaluated in a hamster model. Nine hamsters per group were intranasally immunized with 10⁴ TCID₅₀ of rB/HPIV3 vector expressing wt G, wt G/GS-opt (i.e., wt G with optimized codon usage by GenScript), RSV wt F, and 10⁶ PFU of wt RSV in a volume of 100 μl. Sera were collected on day 28 p.i. Hamsters were challenged intranasally with 10⁶ PFU of wt RSV (A2) on day 31 and sacrificed on day 3 after challenge, and NT and lungs were harvested and homogenized for RSV titration. (E) The RSV neutralizing serum antibodies were quantified by PRNT₆₀ assay in the presence of added guinea pig complement in Vero cells. RSV challenge virus titers in the NT (F) and lungs (G) were determined by plaque assays in Vero cells. The significance of difference between selected groups was determined by unpaired t test: *, P < 0.05; ns (not significant), P > 0.05.

Neutralizing antibody response induced by vectors expressing RSV G in hamsters.

To evaluate immunogenicity, hamsters were immunized i.n. with 10⁵ TCID₅₀ of vectors as described above or with 10⁶ PFU of wt RSV, and sera were collected on day 28 postimmunization. Plaque reduction neutralization assays (PRNT) were carried out in Vero cells to quantify the RSV-specific and HPIV3-specific serum NAb titers (Fig. 6). The assays were performed using recombinant RSV or HPIV3 expressing green fluorescent protein (GFP) and were performed in the presence of added complement (Fig. 6A to C and E) or in its absence (Fig. 6D and F). The data from Fig. 6A, B, and C are tabulated in Table 1.

FIG 6.

Serum RSV- and HPIV3-neutralizing antibody titers induced by rB/HPIV3-RSV G vectors and wt RSV in immunized hamsters. Hamsters in groups of six were infected i.n. with 10⁵ TCID₅₀ of the indicated rB/HPIV3-RSV G vector or 10⁶ PFU of wt RSV. Sera were collected on day 28 postimmunization. The serum neutralization titers were determined by PRNT₆₀ on Vero cell monolayers. For comparison, sera from a group of hamsters immunized with the same dose of rB/HPIV3 expressing unmodified RSV wt F (from a separate experiment performed by following essentially the same protocol) (40) were included in the RSV neutralization assays. Neutralization titers were assayed against wt RSV A2 with complement (A); RSV B1, of subgroup B, with complement (B); RSV CX3C mutant (CWAIS) with complement (C); wt RSV A2 without complement (D); HPIV3 with complement (E); and HPIV3 without complement (F). The limit of detection is indicated with a dashed line. Each dot represents the titer of an individual animal. The solid lines indicate the mean titers of the groups, and error bars represent the SEM. The statistical significance of differences in mean titers among all groups in each assay was analyzed using one-way ANOVA followed by Tukey-Kramer test. Mean titers designated with the same letter (A, B, C, or D) are not statistically different.

TABLE 1.

Mean RSV serum PRNT₆₀ titers induced in hamsters by the indicated vectors and wt RSV A2^a

Virus for immunization	PRNT60 titers for strains of RSV used in neutralization assays
	RSV A2b	RSV B1c	RSV A2 CWAISd
Empty vector	<4e	<4	<4
sG	<4	<4	<4
wt G	3,040	226	52
mG	2,077	152	16
G_B3CT	1,746	125	25
G_B3TMCT	1,911	70	25
G_dCX3C	40	<4	21
G_wCX4C	58	4	35
G_dCX3C_B3CT	83	6	42
G_dCX3C_B3TMCT	54	<4	36
wt RSV A2	6,339	2,521	4,640
wt Ff	6,889	3,040	6,654

^aHamsters in groups of 6 were infected intranasally with 10⁵ TCID₅₀ of rB/HPIV3 vectors or 10⁶ PFU of wt RSV as described in the legend to Fig. 6, and their mean PRNT₆₀ titers on day 28 p.i. are shown. Titers were assayed in Vero cells in the presence of added complement.

^bwt RSV strain A2 strain expressing eGFP was used in the neutralization assays in this column (data from Fig. 6A).

^cwt RSV strain B1 was used in the neutralization assays in this column (data from Fig. 6B).

^dRSV strain A2, in which the CX3C-like motif (CWAIC) in G was mutated to CWAIS, was used in the neutralization assay in this column (data from Fig. 6C).

^eThe limit of detection is 4 PRNT₆₀.

^fSera of six hamsters infected with 10⁵ TCID₅₀ of rB/HPIV3 expressing unmodified wild-type RSV F from a previous study (40) were included in these assays for comparison.

When serum RSV-NAb titers were evaluated in the presence of added complement, vectors expressing wt G, mG, G_B3CT, and G_B3TMCT were shown to induce similarly high titers (>1:1,024) that were not statistically different from that induced by a 10-fold higher (10⁶ PFU) inoculum of wt (i.e., not attenuated) RSV (Fig. 6A, lanes 3 to 6 versus 11, and Table 1). In contrast, the sG construct did not induce detectable serum RSV NAbs (Fig. 6A, lane 2) despite its robust secretory expression (Fig. 2C, lane 3). Interestingly, mutation of the CX3C motif in full-length G drastically reduced the serum RSV-NAb titers by 50- to 75-fold (Fig. 6A, lanes 7 to 10, and Table 1), indicating the CX3C motif is important for the induction of RSV-NAbs. To compare G- versus F-induced neutralization activity, sera of hamsters immunized with the same dose of rB/HPIV3 expressing unmodified wt RSV F from a previous study (40) were assayed in parallel (Fig. 6A, lane 12). The wt G induced 2-fold lower titers of RSV serum NAbs than the wt F, but the difference was not significant (Fig. 6A, lanes 3 and 12, and Table 1).

Hamster sera were also assayed for serum NAbs against a subgroup B strain of RSV (B1) in the presence of added complement. While the neutralization activity against the B1 strain remained high for wt RSV and wt F sera, with a modest reduction of ∼2-fold (Fig. 6B, lanes 11 to 12), the B1 cross-neutralization activity induced by RSV G constructs was reduced by 14-fold (Fig. 6B, lanes 3 to 6, and Table 1), and disruption of the CX3C motif almost completely abolished the neutralization activity (Fig. 6B, lanes 7 to 10, and Table 1). This indicates that the G protein of RSV A2 induced moderate titers of cross-subgroup NAbs; furthermore, this was highly dependent upon the presence of the CX3C motif.

To further characterize the role of the CX3C motif in the induction of RSV-NAbs, an A2 RSV strain bearing a mutated CX3C (CWAIS) of G was used to analyze the serum RSV-NAb titers in the presence of added complement (Fig. 6C and Table 1). While the serum neutralizing activity against the CX3C-mutated RSV induced by the four constructs bearing mutations of the CX3C motif (Fig. 6C, lanes 7 to 10, and Table 1) was low compared to that of wt RSV and vector expressing wt F (Fig. 6C, lanes 11 and 12, and Table 1), the neutralization activities in sera from animals immunized with wt G, mG, G_B3CT, and G_B3TMCT against this CWAIS mutant were equally low (Fig. 6C, lanes 3 to 6, and Table 1). This showed that the integrity of the CX3C motif is critical for efficient neutralization by G-specific NAbs, as well as being critical for the induction of G-specific NAbs.

When the sera were evaluated for neutralization of wt RSV strain A2 in the absence of added complement (Fig. 6D), to provide a more stringent assessment of neutralization activity in Vero cells, sera from animals immunized with wt RSV and vector-expressed wt F contained moderate titers of complement-independent RSV-NAbs (Fig. 6D, lanes 11 and 12), but sera from animals immunized with any of the G constructs were not able to neutralize RSV in the absence of complement (Fig. 6D, lanes 2 to 10).

When serum HPIV3-NAb titers (i.e., against the vector) were evaluated by an assay with added complement, all of the vectors expressing various forms of RSV G were similarly immunogenic (Fig. 6E, lanes 2 to 10), with the exception of the G_B3TMCT construct, which induced a slightly but significantly lower level of serum HPIV3-NAbs than the empty vector (Fig. 6E, lanes 1 and 6), which might reflect a reduction in vector glycoprotein immunogenicity due to the packaging of RSV G into the particle. In a neutralization assay in the absence of added complement (Fig. 6F), titers of serum HPIV3-NAbs were reduced 4- to 8-fold overall compared to those with added complement (Fig. 6E), but all vectors were similarly immunogenic (Fig. 6F). There was no significant difference between the G_B3TMCT construct and empty vector in this case (Fig. 6F, lanes 1 and 6), likely due to the reduced sensitivity of the assay in the absence of added complement. The presence or absence of the sG protein or the CX3C motif in the expressed G protein did not have a significant effect on the antibody response against the vector.

Vector-induced protection against RSV challenge in hamsters.

To measure protective efficacy, the hamsters that were immunized in the experiment shown in Fig. 6 were challenged i.n. with 10⁶ PFU of wt RSV on day 31 postimmunization. Hamsters were sacrificed 3 days later. Nasal turbinates and lungs were collected and homogenized for RSV titration by plaque assay (Fig. 7).

In the nasal turbinates, vector expressing wt G (Fig. 7A, lane 3) was significantly more protective than any of the other G constructs (Fig. 7A, lanes 2 and 4 to 10). In contrast, the sG construct was not protective (Fig. 7A, lane 2), correlating with its inability to induce serum RSV-NAbs (Fig. 6A, lane 2). Constructs mG, G_B3CT, and G_B3TMCT induced substantial protection (Fig. 7A, lanes 4 to 6), while mutation of the CX3C motif (Fig. 7A, lanes 7 and 8) drastically reduced the protection compared to that of the unmodified wt G (Fig. 7A, lane 3), to a level that was not statistically different from those of sG and the empty vector (Fig. 7A, lanes 1 and 2). Enhanced packaging of the dCX3C mutant by B3CT and B3TMCT (Fig. 7A, lanes 9 and 10) appeared to slightly improve the protective efficacy to a level that was significantly better than that of sG and the empty vector (Fig. 7A, lanes 1 and 2). Only the wt RSV inoculation achieved complete protection in the nasal turbinates (Fig. 7A, lane 11). Superior protection with wt RSV was not unexpected, since it was administered at a 10-fold higher dose, is not attenuated, bears both the F and G neutralization antigens, and expresses all of the viral antigens and, thus, would induce a broader humoral and cellular immune response. In studies with PIVs in hamsters, internal proteins in chimeric viruses have been shown to induce protection in a short-term (1- to 2-month) challenge, but this wanes by 4 months (46).

Vector-induced protection against RSV was much better in the lungs than in the nasal turbinates. The vectors expressing wt G, mG, G_B3CT, and G_B3TMCT conferred almost complete protection (Fig. 7B, lanes 3 to 6). However, the vectors expressing any of the four CX3C mutants were less protective, with 1 to 3 hamsters per group not completely protected (Fig. 7B, lanes 7 to 10). As was observed in the nasal turbinates, the TMCT substitution exhibited some improvement of protection with the dCX3C construct (Fig. 7B, lane 10). The level of protection in both the nasal turbinates and lungs strongly correlated with the titer of serum complement-dependent RSV-NAbs (Fig. 7C and D). The greater level of protection that was observed in the lungs than nasal turbinates also has been observed in our previous studies with PIV vectors (40, 47) and likely reflects greater access to the lungs by serum antibodies and cellular immunity.

RSV-GFP infection of ciliated human airway epithelial cells expressing CX3CR1.

Vero cells are deficient in the CX3CR1 receptor thought to play a significant role in attachment in vivo (26). Therefore, we prepared an in vitro-differentiated mucociliary HAE that expresses a CX3CR1 receptor functional in RSV attachment (26, 48) as a model to investigate RSV infection in airway epithelium. The HAE cells were fully differentiated at the air-liquid interface. Differentiation was verified by the formation of ciliated cells and tight junctions on the apical surface of the culture imaged by confocal microscopy (data not shown). Differentiated HAE cultures were infected apically with 2,000 PFU of RSV-GFP in culture medium per 12-mm well. At 3 days p.i., the HAE cultures were fixed and processed for confocal microscopy analysis as described in Materials and Methods. Cilia were stained with a β-tubulin antibody and an Alexa 647-conjugated secondary antibody (red), CX3CR1 was stained with a phycoerythrin (PE)-conjugated CX3CR1 antibody (yellow), RSV-GFP-infected cells were green, and nuclei were visualized by 4′,6-diamidino-2-phenylindole (DAPI) staining (blue). Images of a single representative HAE culture are shown in Fig. 8. Figure 8B and C are magnified images of the boxed region in Fig. 8A (Fig. 8B is a four-color merged image, and Fig. 8C is a two-color image, as indicated). Figure 8D and E are vertical cross sections of the images in Fig. 8B and C. This showed that the vast majority of infected HAE cells (green) were ciliated (red), with CX3CR1 (yellow) expressed exclusively on ciliated cells (Fig. 8A). These observations are in agreement with previous reports and support the notion that RSV uses CX3CR1 as a receptor to initiate infection in HAE cells (26, 48, 49). Interestingly, we found that, under higher magnification, CX3CR1 was located at the base of cilia (Fig. 8D and E), which differs from an earlier report indicating that CX3CR1 is expressed on the distal end of the cilia (49).

FIG 8.

In vitro model of differentiated mucociliary HAE tissue for CX3CR1 expression and infection with RSV-GFP. RSV-GFP (2,000 PFU per 12-mm insert) was diluted in culture medium and added to the apical surfaces of HAE cultures. After 2 h of incubation at 37°C, the inoculum was removed and the infected HAE cultures were incubated at 37°C for 3 days. Infected HAE culture were fixed and stained with DAPI to visualize nuclei (in blue) and stained with antibodies against β-tublin for cilia (in red) and CX3CR1 (in yellow). Cells infected with RSV-GFP are green. Samples were imaged by confocal microscopy. (A) An image of a single field of a representative HAE culture, with different colors merged as indicated in the upper right corner: RSV-GFP infected cells (green), β-tubulin (red), CX3CR1 (yellow), and nuclei (blue). (B and C) Magnification of the area covered by the dashed frame shown in panel A. Panel B is a four-colored merged image, and panel C is a two-colored merged image, as indicated in the upper right corner. (D and E) A vertical cross-section view of the same area shown in panels B and C.

Blocking RSV infection of ciliated human airway epithelial cells by RSV G-induced serum antibodies.

To further evaluate the neutralizing activity of NAbs specific to the CX3C motif, sera of immunized hamsters were tested for their efficiency in blocking RSV infection in CX3CR1-expressing HAE cells. Each airway epithelial culture was infected with 2,000 PFU RSV-GFP (A2 strain) that had been preincubated with hamster sera from this study and control sera from two previous studies for 30 min in the absence of added complement. The inoculum was added to the cells for 2 h and then removed, and the incubation was continued. At day 2 p.i., the GFP foci of infected cells were imaged and quantified (Fig. 9). In the uninfected culture, there was a background of approximately 6 foci per 12-mm culture insert (Fig. 9, lane 1), likely due to autofluorescence. In the RSV-GFP-infected culture, an average of about 200 (range, 53 to 518) fluorescent foci were detected (Fig. 9, lane 2). Preincubation of RSV-GFP with sera from hamsters infected with the empty vector had no effect on the number of fluorescent foci (Fig. 9, lane 3). Incubation of the RSV-GFP inoculum with sera from hamsters immunized with wt G or wt RSV almost completely prevented the formation of fluorescent foci, reducing the number down to the level of autofluorescence observed in uninfected wells (Fig. 9, lanes 4 and 6). Remarkably, the control sera from hamsters immunized with the vector expressing wt RSV F in a previous study (40) as described in Fig. 7 displayed little inhibitory activity in this model, but the sera from hamsters immunized with the vector expressing a stabilized prefusion F from a previous study (38) almost completely blocked the formation of foci (Fig. 9, lane 8). The sera of hamsters infected with the G_dCX3C construct were only partially protective compared with wt G sera (Fig. 9, lanes 4 and 5). These results indicated that wt RSV G induced high-quality NAbs that can effectively prevent CX3CR1-mediated RSV infection in the HAE cells in the absence of added complement, and that the induction of these high-quality NAbs depended on the intact CX3C motif.

FIG 9.

Evaluation of the ability of serum RSV-neutralizing antibodies to block RSV-GFP infection in an in vitro model of differentiated mucociliary HAE tissue. RSV-GFP was incubated with hamster sera in the absence of added complement for 30 min at 37°C and then inoculated onto HAE cultures (one well per serum sample). After 2 h of incubation at 37°C, the inoculum was removed and the infected HAE culture was incubated at 37°C. The GFP foci of infected cells were imaged and quantified 48 h after infection. Lane 1, mock treatment (incubated with L15 medium without RSV-GFP; n = 6 wells); lane 2, 2,000 PFU of RSV-GFP in L15 medium without added serum, n = 6 wells; lanes 3 to 8, 2,000 PFU of RSV-GFP in L15 medium with added sera from the indicated groups (n = 6). Sera used in lanes 3 to 6 were from the experiment described in the legend to Fig. 7. Sera used in lanes 7 and 8 were from hamsters immunized with the same dose and by following the same protocol as that used in references 40 and 38, respectively. The significance of difference between selected groups was determined by unpaired t test. **, P < 0.01; ns (not significant), P > 0.05.

Enhancing the immunogenicity of RSV G by codon optimization.

The codon usage of RSV wt G was optimized by GenScript for mammalian expression with no changes in amino acid coding, resulting in the construct wt G/GS-opt. In Vero and LLC-MK2 cells, replication of rB/HPIV3 expressing wt G/GS-opt was similar to that of the other vector constructs (data not shown), and Western blot analysis showed that wt G/GS-opt was expressed 2.2- to 2.3-fold more efficiently than the nonoptimized wt G ORF (Fig. 10A to D). In hamsters, the serum RSV-NAb titer induced by wt G/GS-opt, collected 28 days postimmunization, was slightly higher than that induced by wt G, which in turn was slightly higher than that induced by wt F, but these pairwise differences were not significant (Fig. 10E, lanes 2 to 4). However, wt G/GS-opt induced a 2-fold higher titer of serum RSV-NAbs than wt F, which was statistically significant (Fig. 10E, lanes 2 and 4). This indicated that codon optimization of RSV G and the resulting increase in protein expression resulted in increased induction of serum RSV-NAbs. The wt G and wt G/GS-opt constructs also conferred increased protection against RSV challenge replication in the nasal turbinates and lungs compared to wt F (Fig. 10F and G). These results indicated that rB/HPIV3 expressing wt G was more immunogenic and protective than rB/HPIV3 expressing wt F in hamsters, and the codon optimization of wt G increased its immunogenicity and protective efficacy.

DISCUSSION

The RSV G and F glycoproteins are the two RSV neutralization antigens and the most important protective antigens; thus, they are important targets for vaccine development. Previously, we modified RSV F expressed by the rB/HPIV3 vector to have increased stabilization in the prefusion conformation, to be efficiently packaged into the vector particle, and to be codon optimized. These features resulted in increased expression and increases in immunogenicity and in the quality of the induced NAbs, as assayed by neutralization of RSV in vitro in the absence of added complement. Since RSV G is an independent neutralization and protective antigen, its inclusion in a vector-based RSV vaccine may further increase immunogenicity and protective efficacy. However, it also has been suggested that the G protein, and in particular sG and the CX3C motif, interfere with or alter the immune response to RSV and should be omitted from an RSV vaccine (30, 50). In the present study, we evaluated the effects of several modifications to the RSV G protein expressed from the rB/HPIV3 vector, specifically (i) mutations in the CX3C motif, (ii) ablation of sG expression, (iii) modification of G so it is efficiently packaged into the vector particle, and (iv) codon optimization of the G ORF.

An important feature of the vector used in this study is that PIV3 has a tropism and replication similar to those of RSV (48, 51). Thus, any significant modulation of the pulmonary immune milieu by RSV G or its mutants taking place during vector infection should be evident in effects on vector replication and immune responses to the vector, in addition to immune responses to the RSV G insert. However, no changes in vector replication or immunogenicity were observed in association with any of the G mutations. Because the RSV G protein is a supernumerary protein not needed for replication of the rB/HPIV3 vector, mutations to G also should not directly affect vector replication, and indeed no effects were observed. Therefore, the level of expression of the RSV G gene was not a variable in these experiments, and immune responses to the various G inserts could be compared directly.

Previously, the expression of sG by RSV was shown to increase the RSV viral load and the severity of disease in a murine model and to suppress the innate immune response and the expression of proinflammatory cytokines in human monocytes (31, 52). In the present study, the expression of sG alone by the rB/HPIV3 vector failed to induce detectable NAbs or confer any protection in hamsters. This differs from the substantial immunogenicity and protective efficacy of secreted soluble RSV F expressed by the rB/HPIV3 vector (53). The poor immunogenicity of sG did not appear to be due to interference by sG with the induction of G-specific NAbs. This was evident by the similarity of NAb titers induced by wt G (which expresses both sG and mG) compared to those of the three constructs expressing only mG, namely, mG, G_B3CT, and G_B3TMCT. Furthermore, the presence or absence of sG expression had no effect on the induction of serum NAbs against the rB/HPIV3 vector and had no effect on replication of the vector (i.e., immune restriction of vector replication). It may be that the poor immunogenicity of sG compared to that of mG is due to sG being expressed as a monomer rather than a multimer (23, 24). In any event, the expression or lack of expression of sG did not appear to have a substantial effect on the induction of NAbs against the PIV3 vector or the RSV G insert, as would have been expected if sG interfered significantly with host immunity.

One minor effect that might be attributable to sG is that the protective efficacy of wt G was significantly greater in the upper respiratory tract than that of the three constructs expressing only mG (namely, mG, G_B3CT, and G_B3TMCT). Since all of these constructs induced the same level of G-specific NAbs, the greater protection with wt G might be due to some other mechanism. For example, sG was reported to promote the recruitment of CD8⁺ cytotoxic T lymphocytes to the respiratory tract (54), and cell-mediated immunity has been shown to confer short-term protection against PIVs in hamsters (46). We also observed a similar trend of greater protection associated with the expression of sG in other experiments in hamsters and mice, although in those experiments the difference was not significant (data not shown). Thus, sG might confer a modest increase in protective immunity. However, the expression of sG, or lack of expression of sG, did not appear to have an impact on the immunobiological milieu of the hamster lung sufficient to affect the replication of the rB/HPIV3 vector. Thus, apart from the difference in protective immunity in the nasal turbinates in the single experiment noted above, there was no evidence of significant effects of sG on host immunity in these experiments. In addition, since sG also acts as an antigen decoy to reduce neutralization and clearance of RSV by preexisting antibodies (33, 34), ablating sG in an RSV vaccine virus presumably would have the undesired effect of rendering it more susceptible to passive antibodies, such as maternal antibodies or monoclonal antibody immunoprophylaxis. Therefore, deletion of sG in a live RSV vaccine might reduce antigen load and immunogenicity.

The CX3C motif in the highly conserved cystine noose is a distinct feature of RSV G and has been suggested to be involved in the evasion of antibody-mediated immunity, suppression of innate and adaptive immune responses, and exacerbation of pulmonary inflammation and pathology. In the present study, we altered the CX3C motif in two different ways, by mutating one of the conserved cysteines or by increasing the spacing between the two conserved cysteine residues. One outcome of either mutation was the reduced glycosylation efficiency of expressed RSV G by 20% to 50%. It may be that mutations in the CX3C motif affect protein structure and perhaps the efficiency of intracellular transport, resulting in the observed reduced glycosylation. Mutation to the CX3C motif also abolished the binding of a neutralizing monoclonal antibody (131-2G) to the CCR, confirming that the CX3C motif is important for the integrity of this neutralization epitope. In addition, each of the two mutations to the CX3C motif drastically reduced the induction of serum RSV NAbs by 50- to 75-fold, indicating the integrity of CX3C is very important for the induction of G-specific NAbs. Furthermore, immune sera from animals that had been immunized with wt G and CX3C mutants were equally inefficient in neutralizing RSV in which the CX3C motif had been mutated. This confirmed that NAbs induced by wt G indeed target epitopes whose integrity involves the CX3C motif. Finally, it is noteworthy that mutation of the CX3C motif had no discernible effect on the replication of the rB/HPIV3 vector, suggesting that any immunomodulatory activities associated with this motif have no detectable effect in restricting vector replication.

It is reasonable to suggest that the CX3C motif is essential for the formation of the cystine noose, which presumably is a relatively rigid structural feature. In contrast, recent structural analyses of the binding of RSV G NAbs to the CCR indicated that the structure around the rigid cystine noose is relatively flexible and can display distinct conformations when targeted by different NAbs (55, 56). Despite the flexibility of the conserved region surrounding the CX3C motif, our results show that the integrity of this motif is essential for both the induction of an efficient G-specific cross-subgroup neutralizing antibody response and the neutralization of RSV by these antibodies. Given the importance of the integrity of the CX3C motif to the immunogenicity of RSV G, alteration of this motif in a live-attenuated or vectored RSV vaccine would create a major gap in the repertoire of induced NAbs. Additionally, given that the CX3C motif is highly conserved between the two antigenic subgroups A and B of RSV, and given that its receptor binding function would likely prevent any major antigenic changes at this site, this epitope represents a major antigenic site that should be included in an effective RSV vaccine.

The standard RSV plaque reduction neutralization assay, as used in this study, is performed in immortalized cell lines (e.g., Vero cells) that lack the CX3CR1 receptor. RSV infection in these cell lines depends on the use of glycosaminoglycans as alternative receptors for attachment. In the present study, neutralization of RSV by G-specific antibodies in Vero cells was highly dependent on the presence of added complement. Neutralization under these conditions is presumed to be mediated by virus lysis or steric hindrance by the added complement acting in concert with bound antibodies, which thereby confer neutralizing activity to antibodies that otherwise would be poorly neutralizing or nonneutralizing (57). Strongly neutralizing antibodies can be identified by a lack of dependence on complement for neutralization in vitro. In contrast to the results with Vero cells, evaluation of RSV neutralization by G-specific sera in HAE cells, which express CX3CR1, demonstrated efficient neutralization in the absence of added complement. We suggest that these complement-independent G-specific NAbs were detected in HAE cells but not Vero cells because they neutralize by directly blocking the interaction between the CX3C motif in the virus and CX3CR1 acting as a viral receptor, which is found in HAE cells but not Vero cells. This is consistent with the observation that mutation of the CX3C motif greatly reduced the induction of G-specific NAbs, detected in Vero cells in the presence of complement, and in HAE cells in the absence of complement. Taken together, these findings suggest that most of the NAbs induced by vector-expressed wt G depended on the integrity of the CX3C motif and are capable of blocking attachment to CX3CR1. It was interesting to find that serum antibodies induced by vector-expressed wt F were less efficient in blocking infection in HAE in the absence of complement. This shows that wt G protein induces a potent set of NAbs that are desirable for an RSV vaccine.

To validate the particular HAE cultures used in this study, we investigated the distribution of CX3CR1 in HAE cells and its correlation with RSV infection. We confirmed that RSV preferentially infected ciliated epithelial cells and that CX3CR1 was expressed only on the ciliated cells, as previously described (48, 49). The correlation of RSV infection with the CX3CR1-expressing ciliated cells supported the idea that CX3CR1 is the primary receptor for RSV in HAE cells (26, 48, 49). In addition, we observed that CX3CR1 was expressed at the base of the cilia, which differs from an earlier description of CX3CR1 being at the distal end of the cilia (49). The location of the receptor at the base of cilia might be more efficient for viral entry.

We investigated the strategy of increasing the immunogenicity of RSV G by increasing its packaging into the vector particle, which was done by replacing the CT plus TM domains of G with those from the vector HN protein. We previously applied this strategy to the RSV F protein expressed from this vector and showed that enhanced packaging resulted in a substantial increase in antibodies that neutralized RSV in vitro in the absence of added complement and were prefusion specific (38, 53). It was unclear whether this effect with RSV F was primarily due to increased stability of the prefusion conformation that somehow occurred with packaging or perhaps because packaging placed the RSV F protein in a multimeric array that provided better antigen presentation. In the present study, replacement of the TM and CT domains of G increased its packaging efficiency by 8.3-fold, to a level that, for an equivalent mass of particles, was 38% of that of wt RSV. However, this did not confer a significant increase in the induction of G-specific complement-dependent or complement-independent NAbs. Thus, in this case, packaging into a (presumably) more uniform conformation did not enhance antigen presentation. However, the packaging efficiency of G-TMCT was less (∼38% of that observed with wt RSV) than that of F-TMCT (∼100% of that of wt RSV), and this difference also was reflected in the packaging of the corresponding vector glycoprotein: in the previous study, incorporation of F-TMCT into the vector reduced the packaging of vector-specific F by 50% to 60% (38), whereas in the present study incorporation of G-TMCT did not reduce packaging of vector-specific HN. It will be of interest to increase the expression of G-TMCT by moving the gene to the promoter-proximal position or codon optimization, which may enhance the packaging, antigen presentation, and immunogenicity of RSV G-TMCT expressed by the vector.

Codon optimization of RSV wt G increased its expression in vitro by 2.2- to 2.3-fold. In hamsters, the induction of serum RSV-neutralizing antibodies by the vector-expressed codon-optimized G was indeed significantly increased. In addition, the codon-optimized G conferred significantly increased protection against RSV challenge compared to that of wt F. The codon optimization might have enhanced the packaging efficiency and the antigenic load of RSV G. Thus, the codon-optimized form of RSV G was more immunogenic and would be a preferred choice for future vaccine design.

In summary, this study demonstrated that the RSV wt G expressed by rB/HPIV3 is a strong protective and neutralization antigen that compared favorably to RSV wt F. In particular, potent NAbs whose induction depended on the CX3C motif, and which probably blocked viral attachment to CX3CR1, appeared to represent a large proportion of the G-specific response. There was no evidence that proposed immunomodulatory activities of the CX3C motif or sG had any deleterious effect on the induction of host immune responses; thus, there was no indication that removing these features from an RSV vaccine would improve its performance. The addition of RSV G to RSV F as protective antigens might further improve the immunogenicity and protective efficacy of vector-based RSV vaccines. The rB/HPIV3 vector expressing RSV wt F previously was shown to have an excellent safety profile in seronegative infants and children in a phase I trial (58); thus, this vector expressing RSV G can be expeditiously evaluated in this age group.

MATERIALS AND METHODS

Cells and viruses.

Rhesus monkey LLC-MK2 epithelial cells (ATCC CCL-7) and African green monkey Vero epithelial cells (ATCC CCL-8) were maintained as previously described (40). A recombinant wt RSV strain A2 (GenBank accession number KT992094) and B1 strain RSV (GenBank accession number AF013254.1; except for the G ORF, which was codon optimized for mammalian expression) were grown in Vero cells cultured at 37°C. A recombinant RSV A2 strain with an enhanced GFP (eGFP) gene inserted between the P and M genes, described previously (59), was used in RSV neutralization assays. The rB/HPIV3 vector and methods of propagation were described previously (44). The parental sequence of wt RSV G was from the A2 strain described above. All constructs of RSV G ORF contained the same set of noncoding flanking sequences and two AscI sites, shown in Fig. 1C, were synthesized by GenScript, and were cloned into the AscI site of the rB/HPIV3 vector.

The rB/HPIV3 vectors expressing RSV G were rescued in BHK BSR-T7/5 cells that constitutively express T7 polymerase, as described previously (60). Rescued viruses were passaged twice on LLC-MK2 cells at 32°C. Viral sequences were confirmed in their entirety (except for the 30 and 120 nucleotides at the 3′- and 5′-terminal ends, respectively, due to the positioning of sequencing primers) with automated Sanger sequencing of uncloned PCR products using reverse-transcribed cDNA of the viral RNA genome as the template.

Western blot analysis of RSV G expression.

Vero cells in 12-well plates were infected in duplicate with rB/HPIV3 vectors at a multiplicity of infection (MOI) of 10 50% tissue culture infectious doses (TCID₅₀) per cell or with wt RSV at an MOI of 3 PFU per cell. Infected cells were incubated at 32°C. At 24 h postinfection, cell monolayers of one of the duplicate wells were washed with 1× phosphate-buffered saline (PBS), lysed in-well with 200 μl of 1× LDS buffer (Thermo Fisher Scientific), and heated at 95°C for 5 min in the presence of reducing reagent (Thermo Fisher Scientific). At 48 h postinfection, the medium supernatants of the other duplicate well were collected and clarified by centrifugation at 10,000 × g for 30 min. The clarified supernatants were mixed with 1× LDS buffer and heated at 95°C for 5 min with added reducing reagent. The heat-treated lysates and supernatant were subjected to electrophoresis in 4 to 12% Bis-Tris NuPAGE gels (Thermo Fisher Scientific) and transferred to polyvinylidene difluoride (PVDF) membranes. Membranes were blocked and incubated with primary antibodies diluted 1:1,000 in blocking buffer (LI-COR), followed by incubation with secondary antibodies diluted 1:10,000 in blocking buffer. Images of blots were captured using the Odyssey imaging system (LI-COR). Primary antibodies included goat anti-RSV polyclonal antibodies (ab20745; Abcam), rabbit anti-HPIV3 antiserum (MS456) (61), and murine anti-glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (G8795; Sigma-Aldrich). Secondary antibodies included donkey anti-mouse IRDye 800CW, donkey anti-rabbit IRDye 680, and donkey anti-goat IRDye 680RD (LI-COR).

Double-immunostaining plaque assay.

The double-immunostaining plaque assays were carried out as described previously (40). Briefly, infected monolayers of Vero cells in 24-well plates were fixed with 80% methanol and then incubated with blocking buffer (LI-COR) containing primary antibodies diluted 1:1,000. After washing with PBS, the monolayers were incubated with blocking buffer containing secondary antibodies diluted 1:800. Goat anti-RSV polyclonal (ab20531; Abcam) and rabbit anti-HPIV3 antiserum (MS456) were used in conjunction with secondary donkey anti-goat 680RD (LI-COR) and donkey anti-rabbit 800CW (LI-COR). RSV G mouse monoclonal antibody (131-2G; Millipore) and rabbit anti-HPIV3 antiserum were used in conjunction with secondary goat anti-mouse 680LT (LI-COR) and goat anti-rabbit 800CW. The plaques were scanned and visualized with the Odyssey infrared imaging system (LI-COR) using 800-nm (shown as green, PIV3) and 680-nm (shown as red, RSV G) channels. When merged (800 and 680 nm), plaques coexpressing RSV G and vector proteins were shown in yellow.

Hamster studies.

Six-week-old Golden Syrian hamsters (Envigo Laboratories) were confirmed to be seronegative for HPIV3 and RSV using HPIV3-specific hemagglutination inhibition (HAI) and RSV neutralization assays, respectively, and were immunized i.n. with 10⁵ or 10⁴ TCID₅₀ of rB/HPIV3 vectors (as indicated in each study) and 10⁶ PFU of wt RSV (A2 strain) in a volume of 0.1 ml. The vector doses were 10- to 100-fold lower than the usual dose of 10⁶ TCID₅₀ for PIVs in hamsters in order to provide more sensitive detection of differences in immunogenicity among vectors. To assess viral replication, infected hamsters were euthanized with CO₂ on day 5. Nasal turbinates (NT) and lungs were collected and homogenized in L15 medium at a ratio of 1:10 (wt/vol). Homogenates were clarified by centrifugation at 2,500 rpm for 10 min at 4°C, and viral titers in the clarified homogenates were determined by TCID₅₀ assays measuring the hemadsorption in LLC-MK2 cells; wt RSV was not included here for comparison. In a separate study for immunogenicity and protective efficacy, additional hamsters in groups of 6 or 9 per virus were inoculated in the same way, and a wt RSV group was included for comparison. Sera were collected on day 28 p.i., and titers of RSV- and HPIV3-NAbs in sera were quantified by 60% plaque reduction neutralization test (PRNT₆₀) in the presence or absence of guinea pig complement (Cederlane, NC) as described previously (40). On day 31 p.i., the hamsters were challenged i.n. with 10⁶ PFU of wt RSV per hamster in a volume of 0.1 ml. Hamsters were sacrificed 3 days after challenge. The titers of wt RSV in the NT and lungs were quantified by plaque assay in Vero cells as described above.

RSV neutralization assay in HAE culture model.

Differentiated HAE cultures were prepared using normal human bronchial epithelial cells (NHBE) from a healthy 6-year-old Caucasian female donor (MatTek, Ashland, MA). Cells were expanded in basal medium containing growth supplements (MatTek). After expansion, 3 × 10⁵ cells were seeded onto the collagen-coated 12-mm polycarbonate membrane inserts and cultured overnight in basal medium added to the basal and apical sides of the inserts. Medium was replaced the next day, with Air-Lift medium (MatTek) added to the basal compartment. Medium was changed every other day. From week three through week six, the cultures were maintained in maintenance medium (MatTek). Differentiation of cells was confirmed by confocal microscopy analysis of HAE culture stained with antibodies against β-tubulin (Sigma, Saint Louis, MO) for cilia or ZO-1 (ThermoFisher Scientific) for tight junctions by following the procedure described in the next section.

The hamster immune sera were diluted 1:1 with L15 medium and heated at 56°C for 30 min for complement inactivation. Per sample, 100 μl of heat-treated serum was incubated with 2,000 PFU of RSV-GFP in the absence of added complement for 30 min at 37°C. The sample was then added to the apical surface of an HAE culture and incubated at 37°C in a cell culture incubator for 2 h. The inoculum was removed and the infected HAE culture was maintained in maintenance medium for 2 days, at which time GFP fluorescent foci on the HAE culture were imaged with a Leica AF6000LX inverted fluorescence microscope and DFC360FX camera (Leica Microsystems, Exton, PA), using a 10× dry objective (numeric aperture [NA], 0.3). The number of total GFP foci was quantified by Imaris software (version 9.1.2; Bitplane AG, Zurich, Switzerland) using the modified spot counting feature.

Confocal microscopic analysis of CX3CR1 expression in HAE culture infected with RSV-GFP.

The HAE cultures were generated and infected with RSV-GFP as described above. At day 3 p.i., membranes of the HAE cultures were fixed with 4% paraformaldehyde for 20 min. Membranes were incubated with 0.2% Triton X-100 for 2 h at room temperature and blocked with 0.5% bovine serum albumin (BSA) in PBS for 1 h at room temperature. After blocking, membranes were incubated with 0.5% BSA containing a mouse monoclonal antibody against β-tubulin (Sigma) and a PE-conjugated rat IgG2b antibody against CX3CR1 (2A9-1; ThermoFisher Scientific) overnight at 4°C. Parallel staining with isotype control antibodies, including a mouse IgG isotype control (Abcam) and a PE-rat IgG2b isotype control (ThermoFisher Scientific), was also performed. Membranes were washed twice with PBS and incubated with an Alexa 647-conjugated anti-mouse antibody (ThermoFisher Scientific) for 1 h at 4°C. Membranes were washed three times with PBS and fixed on glass slides with mounting media containing DAPI (ThermoFisher Scientific) and covered with a coverslip. Images were obtained using a Leica SP8 inverted 5-channel laser scanning confocal imaging system (Leica Microsystems, Exton, PA) equipped with the whole range of visible lasers and two HyD ultrasensitive detectors. Both 40× (NA, 1.3) and 63× (NA, 1.4) oil immersion objectives were used. DAPI was excited using a 405 UV diode laser, eGFP was excited using an argon 488 laser, Alexa 647 was excited using an HeNe 633 laser, and PE was excited using an HeNe 594 laser. Data were deconvolved using Huygens Professional software (version 18.04.0-p5; Scientific Volume Imaging BV, Hilversum, Netherlands), and images were exported using Imaris software (version 9.2.0; Bitplane AG, Zurich, Switzerland). Sequential Z sections of stained cells were acquired for three-dimensional reconstruction and surface modeling with the Imaris software.

Ethics statement.

All rodent studies were performed under animal protocol LID34E, reviewed, and approved by the Animal Care and Use Committee of the NIAID, NIH, according to the U.S. Animal Welfare Act (7 U.S.C. 2131, et seq, and contained in 9 CFR, Parts 1, 2, and 3).

The NHBE cells from a healthy 6-year-old Caucasian female donor were purchased from MatTek (Ashland, MA) as part of a commercial kit. Cells had been obtained by the vendor from anonymized donors with Institutional Review Board (IRB) approval. Experiments using these cells were determined to be exempt from the requirement of IRB review and approval under U.S. Department of Health and Human Services exemption 4 of Part 46.101(b) in 45 CFR.