Influence of Respiratory Syncytial Virus F Glycoprotein Conformation on Induction of Protective Immune Responses

ABSTRACT

Human respiratory syncytial virus (hRSV) vaccine development has received new impetus from structure-based studies of its main protective antigen, the fusion (F) glycoprotein. Three soluble forms of F have been described: monomeric, trimeric prefusion, and trimeric postfusion. Most human neutralizing antibodies recognize epitopes found exclusively in prefusion F. Although prefusion F induces higher levels of neutralizing antibodies than does postfusion F, postfusion F can also induce protection against virus challenge in animals. However, the immunogenicity and protective efficacy of the three forms of F have not hitherto been directly compared. Hence, BALB/c mice were immunized with a single dose of the three proteins adjuvanted with CpG and challenged 4 weeks later with virus. Serum antibodies, lung virus titers, weight loss, and pulmonary pathology were evaluated after challenge. Whereas small amounts of postfusion F were sufficient to protect mice, larger amounts of monomeric and prefusion F proteins were required for protection. However, postfusion and monomeric F proteins were associated with more pathology after challenge than was prefusion F. Antibodies induced by all doses of prefusion F, in contrast to other F protein forms, reacted predominantly with the prefusion F conformation. At high doses, prefusion F also induced the highest titers of neutralizing antibodies, and all mice were protected, yet at low doses of the immunogen, these antibodies neutralized virus poorly, and mice were not protected. These findings should be considered when developing new hRSV vaccine candidates.

IMPORTANCE Protection against hRSV infection is afforded mainly by neutralizing antibodies, which recognize mostly epitopes found exclusively in the viral fusion (F) glycoprotein trimer, folded in its prefusion conformation, i.e., before activation for membrane fusion. Although prefusion F is able to induce high levels of neutralizing antibodies, highly stable postfusion F (found after membrane fusion) is also able to induce neutralizing antibodies and protect against infection. In addition, a monomeric form of hRSV F that shares epitopes with prefusion F was recently reported. Since each of the indicated forms of hRSV F may have advantages and disadvantages for the development of safe and efficacious subunit vaccines, a direct comparison of the immunogenic properties and protective efficacies of the different forms of hRSV F was made in a mouse model. The results obtained show important differences between the noted immunogens that should be borne in mind when considering the development of hRSV vaccines.

INTRODUCTION

Human respiratory syncytial virus (RSV) (hRSV) is the most frequent cause of severe lower respiratory tract infections (bronchiolitis and pneumonia) in infants and young children throughout the world. It is estimated that each year, the virus causes severe disease in ∼34 million children <5 years of age, with >3.5 million requiring hospitalization, and is responsible for 66,000 to 199,000 deaths, mainly in developing countries (1). Licensed vaccines or effective drugs are not available but are urgently needed. Development of a hRSV vaccine has been hampered by the history of enhanced disease associated with a formalin-inactivated (FI) virus vaccine in the 1960s (2). Children who were <6 months of age at the time of vaccination were not protected against natural infection, and most of them were primed for enhanced respiratory disease after hRSV infection. Retrospectively, the lack of protection by the inactivated vaccine was associated with a failure to induce protective levels of neutralizing antibodies despite induction of high levels of binding and complement-fixing antibodies (3). Such poorly neutralizing antibodies might have contributed to immune complex deposition in small airways and, hence, to enhanced pathology (4, 5). In addition, a Th2-biased CD4 T-cell response, characterized by the production of allergic inflammation, including interleukin-4 (IL-4) production, may also have contributed to the enhanced disease observed in the FI hRSV vaccine trial (6). However, disease enhancement is not observed with live attenuated hRSV strains (7) or with subunit vaccines in individuals who have experienced previous RSV infections (8).

A wealth of knowledge supports the notion that protection against hRSV is conferred mainly by neutralizing antibodies: (i) passive transfer of this type of antibody protects mice (9) and cotton rats (10) against a hRSV challenge; (ii) infants at high risk of severe hRSV disease can be protected, at least partially, by prophylactic administration of neutralizing polyclonal antibodies (11) or monoclonal antibodies (MAbs) (12); and (iii) a positive correlation between high titers of serum neutralizing antibodies and protection of human volunteers against hRSV challenge (13), as well as protection of children (14) and the elderly (15) against natural hRSV infections, was found.

Like other paramyxoviruses, hRSV has two main glycoproteins (G and F) inserted into the viral membrane (16). The G glycoprotein was originally described as the receptor-binding protein (17) that binds to cell surface proteoglycans (18–20). The fusion (F) glycoprotein mediates fusion of the viral and cell membranes to allow entry of the virus ribonucleoprotein into the cell cytoplasm and initiation of a new infectious cycle (21). The F and G glycoproteins, expressed from vaccinia virus recombinants, are the only antigens able to induce neutralizing antibodies and confer long-lived protection against hRSV challenge in mice and cotton rats (22, 23), with F being more efficient than G (24).

The hRSV F glycoprotein is synthesized as an inactive precursor (F0) that is posttranslationally cleaved by furin at two polybasic sites, separated by 27 amino acids (aa), to become fusion competent (25). Cleavage generates two chains (F2 N terminal to F1) that remain covalently linked by two disulfide bridges. The mature F protein is a homotrimer of F1 and F2 subunits that is assembled in the virus particle in a metastable conformation, called the prefusion conformation. Once the virus is bound to the target cell surface, the F protein is activated to initiate fusion of the viral and cell membranes (26). During this process, F experiences a series of conformational changes, which results in a highly stable conformation called the postfusion conformation. Important knowledge about these changes was recently gained by solving the atomic structures of soluble forms of hRSV F folded in either the prefusion or the postfusion conformation (27–32).

We reported a few years ago that most neutralizing antibodies present in human immunoglobulin (Ig) preparations were directed against epitopes found exclusively in the prefusion form of hRSV F (33, 34). New evidence for the dominance of prefusion-specific neutralizing antibodies in individual human sera was also recently provided (35). Furthermore, highly neutralizing MAbs isolated from immortalized human B lymphocytes (36) were found to bind epitopes of a new antigenic site (site Ø) found only in the prefusion form of hRSV F (29). McLellan et al. (30) and Krarup et al. (32) reported that stabilized soluble forms of prefusion F were able to induce higher levels of neutralizing antibodies than was the soluble postfusion F protein in laboratory animals (30). Other highly neutralizing antibodies that bind to unique epitopes in prefusion F have been reported. Corti et al. (37) isolated a human MAb (MPE8) that recognized one of these epitopes and cross-neutralized four related members of the Pneumovirinae, i.e., human and bovine RSVs, human metapneumovirus, and pneumonia virus of mice. Recently, Gilman et al. (38) described a neutralizing MAb (AM14) that was dependent on cleavage for binding specifically to prefusion hRSV F. Prefusion F therefore stands out as the most promising immunogen for a subunit hRSV vaccine. However, postfusion F also induces neutralizing and protective responses and has the advantage of being highly stable (27).

Recently, a soluble monomeric form of hRSV F was described (39), which retained some of the epitopes recognized by neutralizing MAbs specific for prefusion F. Therefore, to have an inclusive comparison of the immunogenic potentials of the three available soluble forms of hRSV F (prefusion, postfusion, and monomer), mice were immunized with the three antigens, and their antibody and protective responses following challenge with hRSV were evaluated. The results obtained unveiled substantial differences in the antibody specificities induced by the three proteins as well as in protection against virus replication and the associated pathology upon challenge. These differences should be borne in mind when considering vaccine development.

MATERIALS AND METHODS

Vaccinia viruses and recombinant hRSV F proteins.

Vaccinia viruses encoding the proteins depicted in Fig. 1A were produced as described previously (40). Cloning and expression of the soluble hRSV F protein ectodomain (Long strain, aa 1 to 524) were initially achieved by introducing a stop codon (Ile525Stop) before the transmembrane region (41). The soluble protein used in this study (Fig. 1A, line 1) contains an Xa cleavage site followed by a 6-His tag (GGEIEGRHHHHHH) at the C terminus to facilitate purification in Ni²⁺ columns. The additional proteins shown in Fig. 1A contain the foldon oligomerization domain (GSGYIPEAPRDGQAYVRKDGEWVLLSTFL) used previously by McLellan et al. (30), inserted before the Xa cleavage site. To generate the uncleaved postfusion form of hRSV F (Fig. 1A, line 2), all basic residues in furin cleavage site I (aa 106 and 108 to 109) and site II (aa 131 to 136) were replaced by Asn (42). Cleaved postfusion F (Fig. 1A, line 3) has the first 10 residues of the fusion peptide deleted to avoid aggregation (28, 43). Stabilized prefusion hRSV F was made by incorporating the amino acid changes described previously by McLellan et al. (30) in their DSCav-1 mutant (S155C, S190F, V207L, and S290C) into our hRSV Long strain-derived F protein ectodomain. To obtain the uncleaved prefusion F protein (Fig. 1A, line 4), the basic residues of cleavage sites I and II were changed to Asn as indicated above. All mutagenesis procedures (insertions, deletions, and amino acid changes) were done with the Phusion site-directed mutagenesis kit (Thermo Fisher Scientific) according to the manufacturer's instructions.

FIG 1.

Proteins used in this study. (A) Diagrams of the hRSV F protein ectodomain (aa 1 to 524) denoting the 6-His tag and foldon sequences added at the C terminus, as indicated. Furin cleavage sites I and II are depicted as arrows, and their absence (by mutation of the basic residues to Asn) is depicted as crosses. Proteins produced in each case are indicated on the left: F monomer (F mon) (line 1), uncleaved postfusion F (Unc F post) (line 2), cleaved postfusion F (F post) (line 3), uncleaved prefusion F (Unc F pre) (line 4), and prefusion F (Fpre) (line 5). This nomenclature is used throughout the paper. The first 10 amino acids of the fusion peptide, following cleavage site II, were deleted in postfusion F (line 3), as indicated. Mutations described previously by McLellan et al. (30) to stabilize their prefusion DSCav-1 protein are shown in prefusion F proteins (lines 4 and 5). (B) Gel filtration chromatograms of the indicated proteins. OD, optical density. (C) Coomassie blue-stained gel of the major peaks of each chromatogram. Samples were heated in SDS-PAGE buffer before (left) or after (right) being treated at room temperature for 5 min with 0.1% glutaraldehyde. (D) Negative-stained EM images of the indicated proteins (the F monomer is not shown since no identifiable protein molecules were discerned in the grids). Bar, 50 nm.

The proteins shown in Fig. 1A were purified from supernatants of CV1 cells infected for 48 h with the corresponding vaccinia virus recombinants (0.1 PFU/cell). Culture supernatants were concentrated; buffer was exchanged by using Vivaflow membranes (Sartorius) and then loaded onto Ni²⁺ columns in buffer containing 50 mM Na₂HPO₄ (pH 8.0), 300 mM NaCl, and 10 mM imidazole; and, after washing, proteins were eluted with the same buffer containing 250 mM imidazole. Finally, the proteins were concentrated by using Amicon Ultracel centrifugal filter units (50 kDa; Millipore) and exchanged to buffer without imidazole before being loaded onto a HiLoad 16/600 Superdex 200-pg gel filtration column (GE Healthcare) equilibrated and eluted with the same buffer. Elution was monitored by reading the absorbance at 280 nm. Protein integrity and purity were checked by SDS-PAGE. Reactivity of the purified proteins with conformation-specific antibodies was checked by an enzyme-linked immunosorbent assay (ELISA) (see below).

Purified proteins were applied onto glow-discharged carbon-coated grids and negatively stained with 1% aqueous uranyl formate. Micrographs were recorded with either a JEOL JEM-1100 electron microscope operated at 100 kV or a Tecnai 12 FEI microscope operated at 120 kV.

Mouse immunization and virus challenge.

Specific-pathogen-free female BALB/c mice weighing ∼20 g were obtained from Charles River UK Ltd. Groups of five mice were inoculated bilaterally in the quadriceps muscles (50 μl/site) (intramuscularly [i.m.]) with the indicated amounts of proteins (see the figure legends) in phosphate-buffered saline (PBS) mixed with an equal volume of CpG (Magic Mouse adjuvant; Creative Diagnostics). Four weeks after vaccination, mice were bled from the tail vein and then challenged intranasally (i.n.), under isoflurane general anesthesia, with 4 × 10⁶ PFU of the A2 strain of hRSV in a volume of 50 μl. Mice were weighed daily for 5 or 6 days following hRSV challenge. Groups of 5 mice were euthanized at 5 days postchallenge, and hRSV titers in lung homogenates were determined by a plaque assay on Vero cells. Further groups of 5 mice were euthanized 6 days after challenge, and their lungs were subjected to bronchoalveolar lavage (BAL) with 1 ml of PBS and then fixed in 10% buffered formalin. Blood was also collected when the mice were euthanized on day 5 or 6 after challenge. Cytocentrifuged preparations of BAL fluid cells were fixed and stained with Diff Quick (Thermo Fisher Scientific), and differential cell counts of ∼300 cells/slide were made under oil immersion microscopy. Lungs were paraffin wax embedded, and sections were stained with hematoxylin and eosin. Histopathological lesions were scored as the sum of the scores for each of 3 different lung lobes/mouse. The extent of peribronchiolar and perivascular inflammation was scored on a scale from 0 to 3 depending upon the thickness of cells surrounding the bronchiole or blood vessel multiplied by the proportion of the lung section showing peribronchiolar inflammation on a scale from 0 to 4, giving a maximum bronchiolitis score of 36 for each mouse lung. The extent of alveolitis was scored on a scale from 0 to 3 depending upon the number of inflammatory cells in the air spaces multiplied by the proportion of the lung section with alveolitis on a scale from 0 to 4, giving a maximum alveolitis score of 36 for each mouse lung.

Enzyme-linked immunosorbent assay.

For protein characterization (Fig. 2), 96-well microtiter plates were coated overnight at 4°C with the indicated purified MAbs at a concentration of 6 μg/ml in PBS. Unreacted sites were blocked with 2% pig serum in PBS plus 0.05% Tween 20 for 1 h, followed by washing with H₂O. Serial dilutions of purified proteins were then added for 1 h, and after washing, bound protein was detected with a biotinylated anti-His antibody (0.3 μg/ml) (Bio-Rad), streptavidin-horseradish peroxidase (HRP) (1:2,000; GE Healthcare), and o-phenilenediamine dihydrochloride (OPD) (Sigma). Reactions were stopped with 2 N sulfuric acid, and the absorbance was read at 490 nm.

FIG 2.

ELISA of hRSV F proteins with conformation-specific monoclonal antibodies. MAbs shown in each panel were used to capture serial dilutions of the indicated F proteins. The amount of each protein bound to the different MAbs was determined with an excess of anti-His antibody as described in Materials and Methods.

For analysis of mouse sera by an ELISA, palivizumab (Pz) (12) was used to coat 96-well microtiter plates overnight at 4°C. Nonspecific binding was blocked with 0.5% bovine serum albumin (BSA) in PBS with 0.05% Tween 20 for 30 min at room temperature. Twenty micrograms of purified proteins per well was then added, and incubation was continued for 1 h at room temperature, followed by the addition of serial dilutions of mouse sera and incubation for another hour. Extensive washing with water was done after each step. Finally, bound antibodies were detected with biotin-labeled goat anti-mouse IgG, streptavidin-HRP, and OPD as the substrate (GE Healthcare). The reaction was stopped, and the color was read as indicated above (34).

Neutralization assay.

A neutralization assay was performed based on a microneutralization assay described by Anderson et al. (44), as previously described (45). Briefly, dilutions of mouse sera were incubated with 2 × 10² PFU of the RSV Long strain for 30 min at room temperature in a total volume of 50 μl. These mixtures were used to infect 5 × 10⁴ HEp-2 cells growing in 96-well plates with Dulbecco's modified Eagle's medium (DMEM) supplemented with 2.5% fetal calf serum (FCS) inactivated for 30 min at 60°C. After 1 h of adsorption, DMEM with 2.5% FCS was added, and the cultures were incubated for 72 h at 37°C with 5% CO₂. The plates were then washed three times with 0.05% Tween 20 in PBS and fixed with 80% cold acetone in PBS. After air drying, viral antigen production in the fixed monolayers was measured by an ELISA with a pool of anti-G (021/1G and 021/21G) (46) and anti-F (47F, 101F, 56F, and 2F) (47, 48) murine MAbs, essentially as described above.

Palivizumab (12) was run in parallel to mouse sera as a neutralization control. The neutralization index of each serum sample was calculated as the ratio of nanograms of palivizumab to microliters of serum that reduced antigen production by 50% and shown as log₁₀ values.

Statistical analysis.

Results are expressed as the means ± standard deviations (SD) of the means or as individual values and their means. Statistical significance was calculated by analysis of variance followed by Tukey's test using GraphPad Prism 6.

Ethics statement.

Animal studies were performed under the regulations of the Home Office Scientific Procedures Act (1986) with Office Project License PPL 30/3024. The studies were approved by The Pirbright Institute Animal Welfare and Ethical Review Body.

RESULTS

Soluble hRSV F proteins used in this study.

The different forms of the hRSV F proteins used in this study are shown in Fig. 1A. They were all purified from the supernatants of CV1 cells infected with the corresponding vaccinia virus recombinants by the use of affinity Ni²⁺ columns followed by gel filtration chromatography.

The soluble hRSV F monomer (Fig. 1A) spanned amino acids 1 to 524 of the protein ectodomain, with all basic residues of the two cleavage sites being changed to Asn, as described previously (49). This protein was secreted from cells as a mixture of monomeric and trimeric molecules. Stabilization of trimeric postfusion hRSV F required the insertion of the foldon trimerization domain (50) at the C terminus of the protein ectodomain. Two versions were made, one with the basic residues of the cleavage sites changed to Asn and the other with unaltered cleavage sites but with the first 10 amino acids of the fusion peptide deleted to avoid aggregation (28, 43) (Fig. 1A). Finally, uncleaved and cleaved soluble forms of prefusion hRSV F were obtained by introducing the stabilizing mutations described previously by McLellan et al. (30) into the F protein ectodomain with altered and unaltered cleavage sites, respectively (Fig. 1A).

The gel filtration profile of each protein is shown in Fig. 1B. In the case of monomeric F, about half of the protein actually eluted from the gel filtration column as a trimer. The monomeric peak from this chromatograph was collected and loaded again onto a second gel filtration column to obtain the monomer peak shown in Fig. 1B. The addition of the foldon oligomerization domain at the C terminus increased the proportion of trimeric uncleaved postfusion F to nearly 100%.

The difference in the elution volumes between cleaved and uncleaved postfusion F proteins probably reflects the reduction in the size of the cleaved form, since it lacks the 27-aa peptide between the two cleavage sites and has a 10-aa deletion within its fusion peptide. However, a more drastic difference between the elution volumes of uncleaved and cleaved prefusion F proteins was noticed, which could not be accounted for solely by size differences, and it is hence indicative of significant conformational changes in this protein form upon cleavage.

The integrity and oligomeric state of the different proteins were assessed by SDS-PAGE under reducing and denaturing conditions (Fig. 1C). The uncleaved postfusion, prefusion, and monomeric forms of hRSV F migrated as a single band of the expected size for an uncleaved F0 polypeptide, whereas the cleaved forms of postfusion and prefusion F migrated as separated F1 and F2 bands. A brief cross-linking treatment with glutaraldehyde before SDS-PAGE induced a drastic change in the postfusion and prefusion F proteins, which migrated as high-molecular-weight bands of the expected size for either uncleaved F0 trimers (3×) or cleaved trimers of F1 and F2 subunits. In contrast, migration of the monomeric F form remained essentially unchanged after glutaraldehyde treatment, confirming its monomeric state.

Examination of the F proteins by electron microscopy (EM) after negative staining provided information about their overall shapes (Fig. 1D). As reported previously (39), monomeric F did not generate any discernible form distinguishable from the background, likely due to its relatively small size (not shown). Postfusion F molecules, either uncleaved or fully cleaved at the two furin sites, were seen as the characteristics cones originally described by Calder et al. (41). Deletion of the first residues of the fusion peptide in postfusion F prevented aggregation in rosettes of the cone-shaped molecules upon cleavage, as reported previously (43). Uncleaved prefusion F generated molecules of a heterogeneous shape (Fig. 1D), despite having eluted as a homogeneous peak from the gel filtration column (Fig. 1B). We believe that uncleaved prefusion F may be affected by the conditions used to prepare the EM grids, which is indicative of a certain intrinsic instability. After cleavage, prefusion F was seen as relatively homogenous round molecules, similar to those reported previously by McLellan et al. (30). The change in morphology associated with the cleavage of prefusion F parallels the difference in the elution volumes of the two molecules from the gel filtration column (Fig. 1B).

To verify the antigenic properties of the purified F proteins, they were tested by an ELISA for reactivity with conformation-specific MAbs (Fig. 2). MAb 101F, which recognizes an epitope of antigenic site IV shared by prefusion and postfusion F proteins, reacted similarly with the five proteins shown in Fig. 1A, and the same was true for Pz, a MAb specific for site II which recognizes both prefusion and postfusion F proteins, although the reactivity of Pz with monomeric F was slightly decreased. Site I-specific MAb 2F reacted similarly with postfusion and monomeric F proteins, but binding was significantly weaker with prefusion F. MAb 114F, specific for the six-helix bundle motif characteristic of postfusion F, reacted with this protein (either cleaved or uncleaved) but failed to react with prefusion F and reacted very poorly with the F monomer.

The remaining MAbs shown in Fig. 2 have been reported to be specific for prefusion F. Thus, MAb AM14 has been shown to bind to a quaternary epitope of prefusion F that requires cleavage (38). Accordingly, AM14 reacted with only the cleaved form of prefusion F (Fig. 2). MAbs D25 and 5C4 were described previously as recognizing epitopes of prefusion-specific antigenic site Ø (29). As shown in Fig. 2, both MAbs reacted efficiently with cleaved and uncleaved prefusion F but also with the F monomer and reacted poorly with uncleaved postfusion F, as reported previously (38). Finally, the MPE8 epitope is highly conserved among Pneumovirinae prefusion F proteins, although it is also preserved to some extent in postfusion F (38), as shown in Fig. 2.

In summary, the proteins illustrated in Fig. 1A reacted with the MAbs shown in Fig. 2 as expected from their structural characteristics. Thus, both cleaved and uncleaved prefusion F forms reacted similarly with all antibodies, except AM14, which was exclusive for the cleaved protein. Also, cleaved and uncleaved postfusion F forms showed similar patterns of reactivity with all MAbs. Finally, the reactivities of the F monomer were similar to those of prefusion F, except that monomeric F reacted better with 2F than did prefusion F and did not react with AM14.

Immunization of BALB/c mice with 5 μg or 15 μg of soluble antigens.

The first approach to compare the immunogenicities of different forms of soluble hRSV F was done by using the monomeric and the uncleaved postfusion or prefusion trimeric soluble forms of hRSV F. A single dose rather than multiple doses of each protein was administered i.m., since single doses might facilitate discrimination of the immune responses induced by each immunogen. Groups of 10 mice were immunized with either 5 μg or 15 μg of each protein, mixed with CpG as an adjuvant, and challenged i.n. 4 weeks later with hRSV (A2 strain). Half of the mice were euthanized 5 days after infection to quantitate lung virus titers, and the other half were weighed daily and euthanized 6 days after challenge to evaluate lung pathology.

Each mouse serum, irrespective of the protein used as the immunogen, was titrated by an ELISA for antibodies binding to monomeric, uncleaved postfusion, or uncleaved prefusion hRSV F. ELISA titers of sera collected at the time of challenge and 5 to 6 days later, when mice were euthanized, were indistinguishable. Therefore, results are shown only for sera obtained at the later date.

Figure 3 shows that sera from mice inoculated with either monomeric F (Fig. 3A) or postfusion F (Fig. 3B) contained antibodies that reacted similarly with the three soluble forms of the F protein, although antibody titers in sera from mice immunized with monomeric F were highest against prefusion F, and antibody titers in sera from mice immunized with postfusion F were highest against postfusion F, but with low significance values. However, it is worth stressing that no statistically significant differences in the ELISA titers of mice inoculated with either 5 μg or 15 μg of monomeric or postfusion hRSV F were noted.

FIG 3.

ELISA and neutralization titers of sera from mice inoculated with either 5 μg or 15 μg of the immunogen. (A to C) Groups of 10 female BALB/c mice were inoculated with the indicated amounts (abscissa) of monomeric F (F mon) (A), uncleaved postfusion F (Unc F post) (B), or uncleaved prefusion F (Unc F pre) (C). Four weeks later, mice were challenged with 4 × 10⁶ PFU of the A2 strain of hRSV, as indicated in Materials and Methods. All sera were tested by an ELISA against monomeric F, uncleaved postfusion F, and uncleaved prefusion F, irrespective of the antigen used for immunization. ELISA titers were calculated as the inverse of the serum dilution that gave 50% of the maximum value for each antigen. (D) Sera were tested by a microneutralization assay, and neutralization indexes in reference to palivizumab were calculated as described in Materials and Methods. Controls received CpG only. Mean ELISA titers and mean neutralization indexes for each group are shown by horizontal bars. Horizontal red lines in each panel indicate detection limits. P values were calculated as indicated in the text. Only relevant P values are shown.

In contrast, antibodies present in sera from mice inoculated with uncleaved prefusion hRSV F (Fig. 3C) showed a highly significant preference for binding to this form of the protein in comparison with uncleaved postfusion F, whereas reactivity with monomeric F was intermediate. Furthermore, ELISA titers increased ∼10-fold when mice were inoculated with 15 μg of prefusion F compared with 5 μg of the same antigen.

Mouse sera were also analyzed in a microneutralization test and run in parallel with the MAb palivizumab (12). Neutralization indexes for each serum (Fig. 3D) were calculated by reference to palivizumab. The sera of mice inoculated with either 5 μg or 15 μg of monomeric or postfusion F had similar mean neutralization indexes. In contrast, the mean neutralization index of sera from mice inoculated with prefusion F increased substantially when the immunizing dose was increased from 5 μg up to 15 μg. Furthermore, the mean neutralization index in the latter group of mice was the highest among all groups.

Interestingly, some of the sera from mice inoculated with 5 μg of prefusion F had neutralization indexes below the detection level, while others had indexes comparable to those of sera from mice inoculated with 15 μg of prefusion F (Fig. 3D). This wide range of values was observed in an otherwise homogenous group of ELISA titers for the 5-μg group (Fig. 3C), resulting in a lack of correlation between neutralizing and ELISA binding antibody titers for certain mouse sera. Figure 4 illustrates three such examples of mice inoculated with 5 μg of uncleaved prefusion F. Mice 2 and 3 showed similar ELISA titers against prefusion F, although as noted above, they had very low titers for uncleaved postfusion F and intermediate titers for monomeric F (Fig. 4A). However, mouse 2 serum neutralized virus infectivity as efficiently as palivizumab at the concentrations tested, whereas serum from mouse 3 lacked any significant virus-neutralizing activity (Fig. 4B). The serum from mouse 4 had slightly lower ELISA titers than those of sera from mice 2 and 3, but again, its neutralization index was disproportionally lower than that of serum from mouse 2 in the neutralization test. These differences in neutralization were reflected in the level of protection. Thus, whereas virus was not isolated from the lungs of mouse 2, virus was isolated from the lungs of mouse 3 (2.65 log₁₀ PFU/ml) and mouse 4 (1.74 log₁₀ PFU/ml).

FIG 4.

Examples of ELISA and neutralization results with sera from mice inoculated with 5 μg of prefusion F. (A) Serial dilutions of sera from mice 2, 3, and 4 inoculated with 5 μg of prefusion F were tested by an ELISA against prefusion F, postfusion F, and monomeric F, as indicated. (B) The same three sera were tested by a microneutralization assay, in parallel to palivizumab and control serum (CpG only). Results are shown as the amount of viral antigen produced in the presence of sera/amount of viral antigen produced in the absence of sera × 100. Lung virus titers for the three mice are shown at the bottom. The limit of detection of the plaque assay was 0.6 log₁₀ PFU/ml.

Overall, most mice included in this set of experiments were fully protected, in the sense that virus could not be isolated from the lungs, 5 days after infection (Fig. 5A). However, three out of five mice inoculated with 5 μg of either monomeric F or uncleaved prefusion F exhibited low virus titers in their lungs compared with the high virus titers in all mice in the control group inoculated with only CpG.

FIG 5.

Virus titers and pathology parameters for mice inoculated with 5 μg or 15 μg of antigen following hRSV challenge. (A) Mean lung virus titers in mice inoculated with the proteins indicated in abscissas expressed as log₁₀ PFU per milliliter (±SD), as described in Materials and Methods. Numbers above bars indicate the number of mice with detectable virus versus the total number of mice in each group (*, P < 0.0001 between vaccinated mice and CpG controls). (B) Mean number of cells in BAL fluid of the indicated groups of mice. (C and D) Mean weight loss of mice inoculated with 5 μg (C) and 15 μg (D) of the indicated proteins following challenge (days) (*, P < 0.02; **, P < 0.001 [between vaccinated mice and CpG controls]). (E and F) Bronchiolitis (E) and alveolitis (F) scores determined as indicated in Materials and Methods for each group of mice.

In mice that were euthanized 6 days after hRSV challenge, the total number of cells recovered in the BAL fluid ranged from 1.3 × 10⁵ cells/ml to 1 × 10⁶ cells/ml, and the numbers of cells in BAL fluid from vaccinated mice were not significantly different from those in BAL fluid from the control CpG-only group (Fig. 5B). However, the mean number of cells in BAL fluid from mice immunized with 15 μg of the F monomer or 15 μg of uncleaved postfusion F was significantly higher than that in BAL fluid from mice inoculated with 15 μg of uncleaved prefusion F (P < 0.03 and P < 0.003, respectively). The proportions of neutrophils in BAL fluid ranged from 7 to 29%, and no statistically significant differences in the proportions of cell types between mouse groups were noted (not shown). All animals inoculated with 5 μg of the F protein lost weight after virus challenge, with weight loss being maximal on day 5 after infection (Fig. 5C). However, weight loss in these mice was not as great as that in CpG controls 6 days after hRSV challenge, although the differences were not statistically significant. In contrast, mice inoculated with 15 μg of uncleaved prefusion F did not lose weight (Fig. 5D); mice immunized with the same dose of monomeric F had minimal, transient weight loss at day 5 postchallenge; and mice inoculated with 15 μg of postfusion F developed a more rapid onset of weight loss than did control mice given the CpG adjuvant alone (Fig. 5D). Nevertheless, all groups of mice immunized with 15 μg of F protein were significantly protected against weight loss compared to CpG controls 6 days after hRSV challenge (P < 0.02, P < 0.01, and P < 0.001 for mice vaccinated with the F monomer, uncleaved postfusion F, and uncleaved prefusion F, respectively). Mice inoculated with CpG lost weight from about day 4 postchallenge. The greater weight loss in these mice is consistent with a higher titer of replicating virus after challenge.

Analysis of the severity of lung histopathology showed that most vaccinated mice had bronchiolitis and alveolitis scores slightly higher than those of the control group (Fig. 5E and F). Bronchiolitis scores were significantly higher than those of CpG controls for mice immunized with 5 μg or 15 μg of the F monomer or with 5 μg of uncleaved postfusion F (P < 0.0001, P < 0.0001, and P < 0.02, respectively). However, it is noteworthy that mice immunized with 15 μg of uncleaved prefusion F showed minimal alveolitis (Fig. 5F), which was significantly less than that seen in mice immunized with either 5 μg or 15 μg of the F monomer (P < 0.005 and P < 0.02) or mice immunized with 5 μg of uncleaved prefusion F (P < 0.01). The histopathology is illustrated in the examples of stained sections from lungs of mice inoculated with 15 μg of monomeric F (Fig. 6A and A′), 15 μg of prefusion F (Fig. 6B and B′), 15 μg of postfusion F (Fig. 6C and C′), or the CpG-only controls (Fig. 6D and D′).

FIG 6.

Lung histology of immunized mice. Hematoxylin-and-eosin-stained lung sections (10× objective [left] and 40× objective [right]) from mice inoculated with 15 μg of monomeric F (A and A′), prefusion F (B and B′), postfusion F (C and C′), or CpG only (D and D′) 6 days after challenge with hRSV. Bars, 400 μm (left) and 100 μm (right).

In summary, although the three immunogens used in this set of experiments (monomeric, uncleaved postfusion, and uncleaved prefusion hRSV F proteins) were able to induce neutralizing antibodies and conferred protection against virus challenge, substantial differences between the antibody and pathological responses were observed. Whereas 15 μg of uncleaved prefusion F induced the highest titers of neutralizing antibodies and all mice were protected against virus challenge, mice inoculated with 5 μg of the same protein had variable titers of neutralizing antibodies, and despite having relatively high titers of ELISA antibodies, some mice were not fully protected against virus replication in the lungs. Of note, antibodies induced by the uncleaved prefusion F protein were particularly focused against epitopes found exclusively in prefusion F. Both monomeric and postfusion F proteins were capable of inducing neutralizing antibodies and protection against virus challenge (particularly postfusion F), but mice showed more pulmonary pathology (alveolitis) and a more rapid onset of weight loss with high doses of postfusion F than did those immunized with prefusion F.

Immunization of BALB/c mice with 1 μg, 3 μg, or 9 μg of soluble antigens.

Since the experiments described above showed almost no dose effect of monomeric and postfusion F proteins on antibody titers, a lower and wider dose range was chosen for the next experiment. Additionally, since cleavage of postfusion and prefusion F proteins had significant effects on their profiles of elution from the gel filtration column and since reactivity with certain MAbs has been shown to be cleavage dependent (38), 1 μg, 3 μg, and 9 μg of monomeric as well as cleaved and uncleaved forms of postfusion and prefusion F proteins were included in the next experiment. Cleaved monomeric F could not be used due to its intrinsic tendency to trimerize upon cleavage (39). The experimental design was the same as the one described above; i.e., groups of five mice were immunized and challenged 4 weeks later. Mice were weighed daily and euthanized 5 days after challenge to assess antibody responses and lung virus titers.

The ELISA titers of sera from mice inoculated with monomeric F (Fig. 7A) showed a sizeable increase from 1 μg up to 3 μg, but a dose of 9 μg did not increase ELISA titers further. The antibodies bound essentially the same to postfusion and prefusion F proteins irrespective of the cleavage status of the F protein and, if anything, bound slightly less well to monomeric F.

FIG 7.

ELISA and neutralization titers of sera from mice inoculated with 1 μg, 3 μg, or 9 μg of the immunogen. The results are shown as described in the legend of Fig. 2, except that only 5 mice were used per group. (A and C to F) ELISA titers of mice inoculated with the indicated doses and immunogens. (B) Neutralization titers of the same sera.

The ELISA titers of sera from mice inoculated with each of the three doses (1 μg, 3 μg, or 9 μg) of uncleaved postfusion F were essentially the same with all the proteins tested, although binding to cleaved F proteins was slightly higher (Fig. 7C). Similarly, the ELISA titers in sera from mice inoculated with cleaved postfusion F were dose independent and similar to those in sera from mice inoculated with uncleaved postfusion F (Fig. 7D), except for those in sera from mice inoculated with the lowest dose (1 μg) of cleaved postfusion F, which were slightly higher than those in sera from mice inoculated with 1 μg of uncleaved postfusion F.

As seen previously for prefusion F, there was a clear increase in antibody titers with increasing doses of uncleaved prefusion F (Fig. 7E) and, in contrast to sera from mice immunized with postfusion F, a preference for binding to prefusion F, regardless of cleavage. Mice inoculated with cleaved prefusion F showed ELISA results similar to those for mice immunized with uncleaved prefusion F (Fig. 7F) except that titers were slightly lower than those with uncleaved prefusion F.

The neutralization results (Fig. 7B) were also generally concordant with those of the first experiment (see above). Thus, sera from mice inoculated with 1 μg of monomeric F showed low neutralization indexes that increased moderately in mice immunized with 3-μg and 9-μg doses. Sera from mice inoculated with even the lowest dose of uncleaved postfusion F had moderate neutralization indexes that did not increase following immunization with 3 μg or 9 μg of the same protein. Similar results were obtained with the cleaved postfusion F protein, except that the neutralization indexes were slightly lower than those induced with the uncleaved protein.

The sera of mice inoculated with prefusion F showed the same neutralization trend as that in the first experiment; i.e., the sera of mice inoculated with the two lowest doses (1 μg and 3 μg) had very low neutralization indexes, and some were even below the limit of detection (Fig. 7B). However, with the 9-μg dose, a very significant increase in the mean neutralization index was observed, reaching the highest index among all groups of mice. This result was essentially the same irrespective of whether the inoculated prefusion F protein was cleaved or uncleaved.

The virus titers in the lungs of mice analyzed 5 days after challenge reflected the neutralization results obtained with their sera (Fig. 8A). Thus, most mice inoculated with 1 μg or 3 μg of either monomeric or prefusion F (cleaved or uncleaved) had sizeable amounts of virus in their lungs, whereas mice inoculated with 9 μg of the same proteins were essentially free of virus. In contrast, most mice inoculated with any of the three doses of postfusion F (cleaved or uncleaved) had no detectable virus in their lungs.

FIG 8.

Virus titers and weight loss of mice inoculated with 1 μg, 3 μg, or 9 μg of antigen following hRSV challenge. The results are shown as described in the legend of Fig. 4. In this experiment, the control group was inoculated with 50 μl of CpG. (A) Mean lung virus titers (±SD). (*, P < 0.0001 between vaccinated mice and CpG controls). (B to D) mean weight losses of mice inoculated with 1 μg (B), 3 μg (C), and 9 μg (D) of the indicated proteins following challenge (days) (*, P < 0.05; **, P < 0.01 [between immunized mice and CpG controls]) (#, P < 0.01; ##, P < 0.001 [between immunized and uninfected mice]). CpG-only and nonimmunized controls are included in all weight loss panels for comparison.

Mice inoculated with 1 μg (Fig. 8B), 3 μg (Fig. 8C), or 9 μg (Fig. 8D) of cleaved or uncleaved postfusion F or 9 μg of monomeric F lost weight more rapidly than did control mice inoculated with CpG, as shown by statistically significant differences in weight loss 3 days after hRSV challenge (P < 0.05 and P < 0.01 between immunized mice and CpG controls). In contrast, weight loss in mice immunized with 1 μg or 3 μg of prefusion F proteins or the F monomer was not significantly different from that in CpG controls. Furthermore, mice inoculated with 9 μg of prefusion F (Fig. 8D), particularly with the uncleaved protein, lost less weight than did the CpG controls, and weight change in these mice was not significantly different from that of the unchallenged controls.

In summary, the results of the second experiment confirmed those of the first experiment and demonstrated the unanticipated capacity of postfusion F to induce antibody responses and confer protection against virus challenge even at very low doses in comparison with monomeric and prefusion F proteins. However, mice inoculated with postfusion F as well as monomeric F were predisposed to weight loss after virus challenge, whereas prefusion F was highly protective at the highest dose (9 μg) and prevented weight loss upon virus challenge. At lower doses, however, prefusion F induced variable levels of neutralizing antibodies, and most mice were not protected.

DISCUSSION

Development of an efficacious hRSV vaccine faces two main challenges. One is the induction of a protective immune response, and the other is the avoidance of priming for enhanced disease (51). Recent studies have provided persuasive evidence that protection is afforded mainly by neutralizing antibodies and that most neutralizing antibodies in humans recognize epitopes found exclusively in the prefusion conformation of the hRSV F glycoprotein (34, 35). Hence, it is desirable that any vaccine candidate should be able to induce high levels of neutralizing antibodies directed against prefusion F without predisposing to enhanced disease.

Recently, soluble forms of hRSV F stabilized in the prefusion conformation were shown to induce higher levels of neutralizing antibodies in mice (30, 32), cotton rats (32), and nonhuman primates (30) (rhesus macaques) than those induced by soluble postfusion F; however, postfusion soluble F, which has the advantage of being highly stable, can also induce sizeable titers of neutralizing antibodies in cotton rats and afford protection against hRSV (27). This is likely because certain neutralizing epitopes of prefusion F are also present in postfusion F (34). Since a monomeric form of hRSV F that retained at least some prefusion-specific epitopes was recently described (39), the time was ripe to compare the three soluble forms of hRSV F (prefusion, postfusion, and monomer) as possible candidates for a subunit vaccine.

The proteins used in this study were designed to be stable forms of either prefusion or postfusion trimers (cleaved or uncleaved) or the uncleaved monomer. Their behavior upon gel filtration chromatography, SDS-PAGE before or after glutaraldehyde cross-linking, and negative-staining EM supported the expected state of oligomerization and overall structure. Of note, the significant differences between uncleaved and cleaved prefusion F proteins observed by gel filtration chromatography and by EM may reflect the recently proposed differences in the head compactness of prefusion F before and after cleavage (32). In any case, the reactivity of the uncleaved and cleaved prefusion F proteins with the MAbs shown in Fig. 2 was the same, except for the above-mentioned MAb AM14, which recognizes a quaternary epitope spanning two adjacent protomers of the cleaved molecule (38). The reactivities of postfusion F and monomeric F with the MAbs shown in Fig. 2 were as expected.

Although a single i.m. dose of prefusion F induced the highest levels of neutralizing antibodies and afforded efficient protection at doses of 9 μg and 15 μg, induction of neutralizing antibodies at lower doses of the immunogen (1 μg, 3 μg, and 5 μg) was rather erratic; i.e., some mice developed high titers of neutralizing antibodies, but others developed titers that were below the limit of detection, despite all mice having similar ELISA titers. Generally, mice inoculated with prefusion F that had high neutralization titers did not have virus in their lungs, indicative of protection, while others with low neutralization titers were not protected even if they developed relatively high ELISA titers, resembling the situation found in children immunized with the FI-hRSV vaccine (3). This lack of correlation between ELISA binding and neutralizing antibodies in mice inoculated with low doses of prefusion F was unanticipated and was not found with postfusion or monomeric F. Since it is known that even stabilized prefusion F is less thermostable (30) in vitro than postfusion F (42), it is possible that the former is degraded more rapidly, once inoculated into mice, than other forms of the protein and favors the generation of nonneutralizing antibodies that can bind to prefusion F but are inefficient at inhibiting F-mediated membrane fusion. Only at high doses of prefusion F does enough unaltered protein remain after inoculation to induce antibodies directed against epitopes of prefusion F that are able to block membrane fusion and neutralize virus infectivity.

Another unexpected finding was the biased antibody response toward prefusion-specific epitopes in mice inoculated with soluble prefusion F. This preference was dose independent and contrasted with the results for mice inoculated with soluble postfusion F or monomeric F, where no evidence of conformation dominance was seen. The reasons for this disparity are unknown, but it is worth stressing that proteins with essentially identical primary structures may induce significantly different antibody responses depending on their conformations. A relatively similar preference for prefusion-specific antibodies was recently reported for mice inoculated with Newcastle disease virus-like particles (VLPs) that incorporated engineered prefusion hRSV F (52).

The unanticipated finding that postfusion F could induce essentially the same level of antibodies and protect mice after inoculation of 1 μg to 15 μg of protein was also remarkable and may be related to the high stability of this protein (42), which possibly remains unaltered for prolonged periods of time after inoculation, facilitating the development of an antibody response. However, mice inoculated with postfusion F were among those with the most severe pathological signs, such as weight loss and pulmonary inflammation, and this is a negative aspect of this protein for a subunit vaccine. The weight loss and enhanced pulmonary pathology seen in all vaccinated mice except those immunized with the highest doses of prefusion F may be due to the activation of T cells as a result of early virus replication in the lungs, even in those mice that were free from infectious virus 5 days after challenge. In contrast, the absence of weight loss and alveolitis in mice immunized with the highest doses of prefusion F suggests that high levels of these neutralizing antibodies completely prevented virus replication. Previous vaccines based on poorly characterized F protein probably used postfusion F as the immunogen. Although in most cases, induction of neutralizing antibodies and protection were achieved in mice or cotton rats (53, 54), there were cases in which pathology was also noted, depending on the adjuvant and route of vaccination. Hence, postfusion F may have the drawback of priming for pathology, although this is probably not the case in individuals who have been previously infected with hRSV (8). Our results indicate that protective doses of prefusion F may be less pathogenic, but a more extensive safety profile needs to be investigated.

There are other variables that need to be explored, such as the number of doses, different adjuvants, and vaccination routes, etc. Although others have shown that two doses of prefusion F, either as a soluble protein (32) or incorporated into VLPs (52), can induce protection against RSV challenge in rodent models of RSV infection, our results indicate that at low antigen concentrations, individuals may remain susceptible to infection following the first vaccination and before the second dose has been administered. However, in aggregate, prefusion F is capable of inducing the highest levels of neutralizing antibodies after a single dose and conferring protection with minimal pathology at high doses of the immunogen. Thus, two main messages stand out from this study: (i) the induction of high levels of ELISA antibodies, even if they are preferentially directed against prefusion F, could not be regarded as a correlate of neutralization and protection, and (ii) as a consequence of the above-described results, the dose of prefusion F to be incorporated into a subunit vaccine should be carefully worked out to avoid undesirable effects.