Nonglycosylated G-Protein Vaccine Protects against Homologous and Heterologous Respiratory Syncytial Virus (RSV) Challenge, while Glycosylated G Enhances RSV Lung Pathology and Cytokine Levels

ABSTRACT

New efforts are under way to develop a vaccine against respiratory syncytial virus (RSV) that will provide protective immunity without the potential for vaccine-associated disease enhancement such as that observed in infants following vaccination with formalin-inactivated RSV vaccine. In addition to the F fusion protein, the G attachment surface protein is a target for neutralizing antibodies and thus represents an important vaccine candidate. However, glycosylated G protein expressed in mammalian cells has been shown to induce pulmonary eosinophilia upon RSV infection in a mouse model. In the current study, we evaluated in parallel the safety and protective efficacy of the RSV A2 recombinant unglycosylated G protein ectodomain (amino acids 67 to 298) expressed in Escherichia coli (REG) and those of glycosylated G produced in mammalian cells (RMG) in a mouse RSV challenge model. Vaccination with REG generated neutralizing antibodies against RSV A2 in 7/11 BALB/c mice, while RMG did not elicit neutralizing antibodies. Total serum binding antibodies against the recombinant proteins (both REG and RMG) were measured by surface plasmon resonance (SPR) and were found to be >10-fold higher for REG- than for RMG-vaccinated animals. Reduction of lung viral loads to undetectable levels after homologous (RSV-A2) and heterologous (RSV-B1) viral challenge was observed in 7/8 animals vaccinated with REG but not in RMG-vaccinated animals. Furthermore, enhanced lung pathology and elevated Th2 cytokines/chemokines were observed exclusively in animals vaccinated with RMG (but not in those vaccinated with REG or phosphate-buffered saline [PBS]) after homologous or heterologous RSV challenge. This study suggests that bacterially produced unglycosylated G protein could be developed alone or as a component of a protective vaccine against RSV disease.

IMPORTANCE New efforts are under way to develop vaccines against RSV that will provide protective immunity without the potential for disease enhancement. The G attachment protein represents an important candidate for inclusion in an effective RSV vaccine. In the current study, we evaluated the safety and protective efficacy of the RSV A2 recombinant unglycosylated G protein ectodomain produced in E. coli (REG) and those of glycosylated G produced in mammalian cells (RMG) in a mouse RSV challenge model (strains A2 and B1). The unglycosylated G generated high protective immunity and no lung pathology, even in animals that lacked anti-RSV neutralizing antibodies prior to RSV challenge. Control of viral loads correlated with antibody binding to the G protein. In contrast, the glycosylated G protein provided poor protection and enhanced lung pathology after RSV challenge. Therefore, bacterially produced unglycosylated G protein holds promise as an economical approach to a protective vaccine against RSV.

INTRODUCTION

Respiratory syncytial virus (RSV) is the leading cause of virus-mediated lower respiratory tract illness (LRI) in infants and children worldwide. In the United States, RSV is a major cause of morbidity, second only to influenza virus (1). For infants, more than 2% of hospitalizations are attributable to RSV infection annually (1). Although traditionally regarded as a pediatric pathogen, RSV can cause life-threatening pulmonary disease in bone marrow transplant recipients and immunocompromised patients (2, 3). In developing countries, most RSV-mediated severe disease occurs in infants younger than 2 years and results in significant infant mortality (4). Among the elderly, RSV is also a common cause of severe respiratory infections that require hospitalization (4).

Although the importance of RSV as a respiratory pathogen has been recognized for more than 50 years, no vaccine is available yet because of several problems inherent in RSV vaccine development. These barriers to development include the very young age of the target population, recurrent infections in spite of prior exposure, and a history of enhanced disease in young children who were immunized with a formaldehyde-inactivated RSV (FI-RSV) vaccine in the 1960s (3, 5). Subsequent studies with samples from these children showed poor functional antibody responses with low neutralization or fusion-inhibition titers (6, 7). There was also evidence for deposition of immune complexes in the small airways (8); however, the mechanism of the FI-RSV vaccine-induced enhanced disease is poorly understood. Animal models of the FI-RSV vaccine-associated enhanced respiratory disease (VAERD) suggested a possible combination of poor functional antibody responses and Th2-biased hypercytokine release, leading to eosinophilic infiltration in the lungs (9, 10).

RSV live-attenuated vaccines (LAV) are an attractive vaccine modality for young children. These vaccines present to the immune system many viral genes with potential protective targets, including the F and G membrane proteins. By using reverse genetics, attenuating mutations were incorporated into RSV A2 in different combinations, and this strategy has been explored extensively, with an emphasis on reaching a good balance between safety and immunogenicity (11). However, the stability of the engineered mutations is an important technical challenge (12). A recent RSV LAV candidate (rA2cp248/404/1030deltaSH) was found to be safe in infants but poorly immunogenic (13). However, new RSV LAV candidates are being evaluated.

Vaccines based on recombinant proteins in different cell substrates have been pursued as well (3, 14). Earlier, RSV F glycoprotein (PFP-2) formulated in alum was well tolerated in clinical trials but only modestly immunogenic in adults, pregnant women, and the elderly (15). A mixture of F, G, and M recombinant proteins was tested in individuals >65 years old and was found to induce >4-fold increases in serum neutralizing activity in 58% of subjects with low prevaccine titers (16). Recently, the structures of the F protective targets recognized by the monoclonal antibodies (MAbs) palivizumab and 101F were resolved, as well as the prefusion form of the F protein trimer, leading to the structure-based design of stabilized F protein vaccine candidates (17–19). The G protein was also evaluated as a vaccine candidate in preclinical studies. A nanoparticle vaccine encompassing the RSV G protein CX3C chemokine motif protected BALB/c mice from RSV challenge (20). A subunit vaccine based on the central conserved region of the G attachment surface glycoprotein was fused to the albumin-binding domain from streptococcal protein G, produced in prokaryotic cells, and formulated with an alum-based adjuvant. After promising results in murine models, challenge studies with rhesus macaques showed no reduction in viral loads, and studies with human adults showed a relatively low capacity for inducing neutralizing antibodies (21, 22). Ideally, an optimally effective RSV vaccine must protect against antigenically divergent group A and B RSV strains.

In the current study, we evaluate side by side the immunogenicity, safety, and protective capacity of a recombinant RSV G protein ectodomain (amino acids 67 to 298) vaccine produced either as an unglycosylated protein in an Escherichia coli prokaryotic system (REG) or as a fully glycosylated protein in mammalian cells (RMG) in a mouse challenge model using homologous RSV A2 and heterologous RSV B1 strains. We demonstrate that the unglycosylated REG generated higher protective immunity against both homologous and heterologous RSV strains, while the glycosylated RMG immunogen resulted in enhanced lung pathology and increased cytokine levels in lungs following virus challenge.

MATERIALS AND METHODS

Cell, viruses, and plasmids.

Vero cells (CCL-81) and A549 cells (CCL-185) were obtained from the ATCC. 293-Flp-in cells (R750-07) were obtained from Invitrogen. Vero cells, A549 cells, and 293-Flp-In cells were grown in Eagle's minimal essential medium (EMEM), F-12K medium, and Dulbecco's modified Eagle medium (DMEM) (high glucose), respectively. All cell lines were supplemented with 10% heat-inactivated fetal bovine serum (FBS), 1× penicillin-streptomycin (P-S), and l-glutamine and were maintained in an incubator at 37°C under 5% CO₂.

RSV A2 (NR-12149) and B1 (NR-4052) strains were obtained from BEI Resources, NIAID, NIH. All virus stocks were prepared by infecting subconfluent A549 cell monolayers with virus in F-12K medium with l-glutamine supplemented with 2% FBS and 1× P-S (infection medium). Virus was collected 3 to 5 days postinfection (dpi) by freeze-thawing cells twice and combining them with the supernatant. Harvested viruses were cleared of cell debris by centrifugation at 3,000 rpm for 15 min. Virus stocks to be used in challenge studies were pelleted by centrifugation at 7,000 rpm overnight. The pelleted virus was resuspended in F-12K medium supplemented with 50 mM HEPES and 100 mM MgSO₄, aliquoted, and frozen at −80°C until needed. Virus titers were determined by plaque assays in Vero cells.

Codon-optimized RSV G coding DNA for E. coli and mammalian cells was chemically synthesized. NotI and PacI sites were used for cloning the RSV A2 G ectodomain coding sequence (amino acids 67 to 298) into the T7-based pSK expression vector (23) for bacterial expression and the pSecR vector for mammalian expression (24) to express G protein in E. coli and 293Flp-In cells, respectively.

Purified RSV A2 F protein (amino acids 22 to 529) fused to a polyhistidine tag produced in insect cells, with endotoxin levels of <1 endotoxin unit (EU)/μg of protein, was obtained from Sino Biologicals.

Production of REG.

The recombinant RSV G extracellular domain (amino acids 67 to 298) was expressed in E. coli BL21(DE3) cells (Novagen) and was purified as described previously (23, 25). Briefly, G protein expressed and localized in E. coli inclusion bodies (IB) was isolated by cell lysis and multiple washing steps with 1% Triton X-100. The pelleted IB containing G protein were resuspended in denaturation buffer and were centrifuged to remove debris. The protein supernatant was renatured by slowly diluting in the redox folding buffer. The renatured protein solution was dialyzed against 20 mM Tris-HCl (pH 8.0) to remove the denaturing agents. The dialysate was filtered through a 0.45-μm filter and was purified through a HisTrap FF chromatography column (GE Healthcare). The protein concentration was analyzed by a bicinchoninic acid (BCA) assay (Pierce), and the purity of the recombinant G protein from E. coli (REG) was determined by SDS-PAGE. The endotoxin levels of the purified protein were <1 EU/μg of protein.

Production of recombinant glycosylated G protein using 293 Flp-In cells (RMG).

293-Flp-In cells and pOG44 (plasmid expressing Flp-In recombinase) were obtained from Invitrogen (Carlsbad, CA). The 293-Flp-In cell line stably expressing the RSV A2 G protein with a secretory signal peptide from the IgG κ chain was developed as described previously (24). Briefly, 293-Flp-in cells were cotransfected with the plasmids expressing Flp-in recombinase and the RSV A2 G ectodomain in DMEM (Invitrogen). Twenty-four hours after transfection, the culture medium was replaced with fresh DMEM containing 150 μg/ml of hygromycin for the selection of stably transfected cells. For protein expression, cells were maintained in 293 Expression Medium (Invitrogen), and the culture supernatant was collected every 3 to 4 days. The supernatant was cleared by centrifugation and was filtered through a 0.45-μm filter before purification through a HisTrap FF column (GE Healthcare).

Gel filtration chromatography.

Proteins at a concentration of 5 mg/ml were analyzed on a Superdex S200 XK 16/60 column (GE Healthcare) preequilibrated with phosphate-buffered saline (PBS), and protein elution was monitored at 280 nm. Protein molecular weight (MW) marker standards (GE Healthcare) were used for column calibration and for the generation of standard curves to identify the molecular weights of the test protein samples.

PRNT.

For the plaque reduction neutralization test (PRNT), heat-inactivated serum was diluted 4-fold and was incubated with 20 to 60 PFU of RSV from the A2 or B1 strain for 1 h at 37°C under 5% CO₂. The assay was performed in the presence of 5% guinea pig complement as described previously (26). Briefly, Vero cells were infected with a serum-virus mixture and were incubated for 1 h before the removal of the inoculum and the addition of an overlay of 0.8% methylcellulose in the infection medium. Plates were incubated for 5 to 7 days, and plaques were detected by immunostaining. Neutralization titers were calculated by adding a trend line to the neutralization curves and using the following formula to calculate 50% endpoint titers: antilog of [(50 + y-intercept)/slope].

Mouse immunization, RSV challenge, and sample collection.

All animal experiments were approved under animal protocol study number 2009-20 by the U.S. FDA institutional animal care and use committee. Four- to 6-week-old female BALB/c mice were obtained from the NCI (Frederick, MD). Mice were immunized intramuscularly (i.m.) at day 0 and day 20 with 5 μg of purified RSV protein combined with Emulsigen adjuvant. Blood was collected from the tail vein on days 0, 14, and 30. On day 34, mice were anesthetized with a ketamine-xylazine cocktail and were infected intranasally (i.n.) with 10⁶ PFU of RSV from the commonly used lab strain A2 or B1. Mice were sacrificed by CO₂ asphyxiation either 2 or 4 days post-RSV challenge, when blood and lungs were collected. For histopathological analysis of the lungs, the right lobe of the lung was collected on days 2 and 4 post-RSV challenge by inflation with 10% neutral buffered formalin. For determination of the viral load and cytokine analysis, the left lobe of the lung was collected.

Determination of viral loads in lungs.

Lungs were weighed and homogenized in F-12K–2% FBS–1× P-S (5 ml medium/g of lung) using an Omni (Kennesaw, GA) tissue homogenizer. The supernatant was cleared by centrifugation at 3,000 rpm for 10 min and was used immediately for viral titration by a plaque assay in Vero cells as described above.

Measurement of cytokine levels in lungs.

All lungs were weighed and were homogenized in 5 ml of medium/g of lung, as described above, to normalize the amount of lung tissue used per sample. Homogenized lungs were further diluted in infection culture medium containing a 2× concentration of Complete EDTA Free protease inhibitor cocktail (Roche, Basel, Switzerland) and were used in a Bio-Plex Pro mouse cytokine 23-plex assay according to the manufacturer's recommendations. Plates were read using a Bio-Plex 200 system (Bio-Rad, Hercules, CA).

Lung histopathology.

Lungs were fixed in situ with 10% neutral buffered formalin and were removed from the chest cavity. Fixed lungs were embedded in a paraffin block, sectioned, and stained with hematoxylin and eosin (H&E) by Histoserv in a blinded fashion. Three mice per immunization and two slides per mouse were analyzed for histopathology. The slides were reviewed with an Olympus BX41 microscope. Photomicrographs were taken with the microscope and an Olympus DP71 digital camera, and images were captured using Olympus cellSens software. Lung lesions in each mouse were graded (scored) according to severity in each of the following sites: whole-lung section, bronchiolar lesions, vascular lesions, and alveolar lesions. The severity scores were as follows: 0, no lesions; 1, minimal lesions; 2, mild lesions; 3, moderate lesions; 4, severe lesions. Cell populations in five different perivascular lung inflammatory foci in three mice per vaccine group were counted following RSV challenge.

SPR.

Steady-state equilibrium binding of postvaccination mouse sera was monitored at 25°C using a ProteOn surface plasmon resonance (SPR) biosensor (Bio-Rad). The recombinant G proteins from E. coli (REG) or 293T cells (RMG) were coupled to a GLC sensor chip via amine coupling with 500 resonance units (RU) in the test flow channels. Samples of 100 μl freshly prepared sera at a 10-fold dilution or MAbs (starting at 1 μg/ml) were injected at a flow rate of 50 μl/min (contact duration, 120 s) for association, and disassociation was performed over a 600-s interval. Responses from the protein surface were corrected for the response from a mock surface and for responses from a buffer-only injection. Prevaccination mouse sera were used as a negative control. Total antibody binding and data analysis results were calculated with Bio-Rad ProteOn Manager software (version 2.0.1).

Statistical analyses.

The statistical significances of group differences were determined using one-way analysis of variance (ANOVA) and a Bonferroni multiple-comparison test. Correlations were calculated with a Spearman two-tailed test. P values less than 0.05 were considered significant with a 95% confidence interval.

RESULTS

Expression and purification of glycosylated and nonglycosylated RSV G protein.

The RSV G protein from the A2 strain is a heavily glycosylated protein of 298 amino acids with N-linked and O-linked glycans in its native form (Fig. 1A) (27). To determine the impact of glycosylation on G protein immunogenicity, a nonglycosylated extracellular domain of G protein from RSV strain A2 was recombinantly produced in E. coli and was termed REG (Fig. 1A). Insoluble G protein was purified from E. coli inclusion bodies, denatured, renatured under controlled redox refolding conditions, and purified by Ni-NTA chromatography as described in Materials and Methods. The purified REG protein displayed a single band of ∼36 kDa under reducing conditions in SDS-PAGE that was recognized by Western blotting with the protective anti-G MAb 131-2G (28, 29) (Fig. 1B). To produce a native glycosylated extracellular G protein (Fig. 1A), 293-Flp-In cells were stably transfected with a plasmid expressing the ectodomain of the RSV A2 G coding sequence. The glycosylated G protein secreted in the mammalian cell culture supernatant was purified by Ni-NTA chromatography and was termed RMG. Purified glycosylated G protein (RMG) exists as a band of ∼120 kDa under reducing conditions, like the native virus-derived G protein that was recognized by anti-G MAb 131-2G by Western blotting (Fig. 1C). In addition, the two purified G ectodomains (REG and RMG) were recognized equally by MAb 131-2G in solution phase SPR (28, 29) (Fig. 1D).

FIG 1.

Purification of recombinant G protein from E. coli and 293 cells. (A) Schematic representation of the RSV G protein. RSV G protein purified from E. coli (REG) inclusion bodies lacks the cytoplasmic and transmembrane domains (CT-TM) and is not glycosylated, while the RSV G protein secreted from 293-Flp-In cells (RMG) is glycosylated. N-linked glycosylation sites, as predicted by NetNGlyc software, version 1.0, are indicated above the diagrams, and predicted O-linked glycosylation sites, as predicted by NetOGlyc software, version 4.0, are indicated below the diagrams, for the full-length RSV G and RMG proteins. REG was purified from E. coli inclusion bodies as described in Materials and Methods. RMG was purified from the clarified supernatant of 293-Flip-In cells stably expressing the RSV G protein. (B and C) Both REG (B) and RMG (C) were purified through a Ni-NTA column, and the final products were each detected as a single band by Western blotting under reducing conditions using MAb 131-2G. (D) SPR interaction profiles of binding of REG and RMG to the protective MAb 131-2G, which targets the central conserved domain of RSV G. (E and F) Superdex S200 gel filtration chromatography of RSV G protein produced in a bacterial system (REG) (E) or a mammalian system (RMG) (F). Elution profiles of purified RSV G proteins (red lines) are overlaid with calibration standards (gray lines).

To determine if these purified recombinant proteins contain higher-order quaternary forms, the purified RSV G proteins were subjected to size exclusion gel filtration chromatography (Fig. 1E and F). The in vitro-refolded, bacterially produced recombinant RSV G (REG) extracellular domain (amino acids 67 to 298) contained three MW forms representing approximately equal amounts of monomers, homotetramers, and a higher-order oligomeric form (Fig. 1E). In comparison, the glycosylated RSV G (RMG) extracellular domain (amino acids 67 to 298), purified from proteins secreted from a 293-Flp-In mammalian cell culture, contained only the homotetrameric form (Fig. 1F).

Immunization of mice with nonglycosylated REG protein generates higher binding and neutralizing antibody titers than immunization with glycosylated RMG protein.

The RSV G protein is one of the two surface proteins containing neutralizing targets of RSV (20, 30). To test the antigenicity of the glycosylated (RMG) and nonglycosylated (REG) G proteins, BALB/c mice either were immunized intramuscularly twice, 20 days apart, with 5 μg of REG, RMG, or F protein from the RSV A2 strain, with Emulsigen as an adjuvant, or were mock vaccinated with PBS (Fig. 2A). Since the glycosylated and unglycosylated G ectodomains (RMG and REG, respectively) run at different MWs (Fig. 1), the doses used for vaccination were normalized by protein content to 5 μg of protein per dose (therefore, equal molarity) as determined by a BCA assay and SPR-based quantification using MAb 131-2G. The sera collected after the second immunization were tested for RSV-neutralizing activity by a plaque reduction neutralization test (PRNT). Prevaccination (day zero) serum samples from all the animals tested negative by PRNT (data not shown). As expected, mice immunized with the F protein had high levels of neutralizing antibodies against both RSV strains, while mice immunized with PBS did not have detectable levels of anti-RSV neutralizing antibodies (Fig. 2B and C). Seven of 11 mice immunized with REG (67%) had neutralizing antibodies against homologous RSV A2 (Fig. 2B), but all had very weak or no measurable neutralizing antibodies against heterologous RSV strain B1, as measured by the PRNT (Fig. 2C). Surprisingly, mice vaccinated with the glycosylated RMG did not develop RSV-neutralizing antibodies against either strain (as measured by the PRNT) after two immunizations (Fig. 2B and C).

FIG 2.

Neutralizing antibody response following RSV G (REG or RMG) or F protein immunization. (A) Schematic representation of mouse immunization and challenge schedule. BALB/c mice were immunized i.m. with 5 μg of RSV strain A2 REG, RMG, or F protein with Emulsigen adjuvant, or with PBS as a control, on days 0 and 20. Ten days after the second immunization, blood was collected from the tail veins. Fourteen days after the second immunization, mice were challenged intranasally with 10⁶ PFU of either RSV A2 or RSV B1 (6 to 11 mice per group). Mice were sacrificed on day 2 or 4 postchallenge, when lungs and blood were collected. (B and C) Serum samples collected from individual mice on day 10 after the second immunization were tested for neutralization by a PRNT against the homologous RSV A2 strain (B) or the heterologous RSV B1 strain (C). Neutralizing antibody titers represent 50% inhibition of plaque numbers. The average for each group is indicated by a horizontal line. Prevaccination (day zero) serum samples from all the animals tested negative by the PRNT (data not shown). The dotted lines indicate cutoff values based on the 1:5 dilution of sera used in PRNT. Statistical significance was tested by one-way ANOVA and Bonferroni multiple-comparison tests. ***, P < 0.0001; **, P < 0.001; *, P < 0.05.

To obtain more-complete information about the antibody responses to the two recombinant G proteins in mice, an SPR-based real-time kinetics assay was used. Total anti-G binding antibody titers were measured against both the REG and RMG proteins captured on the SPR chip surface (Fig. 3A and B, respectively). As can be seen, sera from REG-immunized mice demonstrated high levels of antibodies binding to REG (Fig. 3A, red dots). Importantly, good binding to the glycosylated RMG was also observed for sera from REG-immunized mice (Fig. 3B, red dots). In comparison, sera from mice immunized with the glycosylated RMG protein gave >10-fold-lower binding to both the nonglycosylated REG and glycosylated RMG proteins by SPR (Fig. 3A and B, green dots). Therefore, the nonglycosylated REG was a better immunogen than the glycosylated RMG, generating higher levels of antibodies that recognized both the G protein sequence (unglycosylated) and the glycan-covered RSV G protein.

FIG 3.

SPR analysis of postvaccination serum antibodies to REG, RMG, and different antigenic regions within RSV G. (A and B) The same individual postvaccination mouse sera for which results are shown in Fig. 2B and C were tested for total antibody binding to the REG protein (A) or the RMG protein (B) by SPR. (C) Antigenic peptides representing amino acids 66 to 90, 90 to 110, 148 to 178, 169 to 207, or 236 to 263 of the RSV G protein were chemically synthesized and were tested for binding to serum antibodies from REG- or RMG-immunized mice in a real-time SPR kinetics experiment. Total antibody binding is represented as resonance units detected by SPR. Statistical significance was tested by one-way ANOVA and Bonferroni multiple-comparison tests. ***, P < 0.0001; **, P < 0.001; *, P < 0.05.

REG vaccination generates a higher diversity of the antibody immune response than RMG vaccination.

To test which antigenic regions are recognized by antibodies generated following REG or RMG vaccination, a series of antigenic peptides derived from the N terminus, central conserved domain (CCD) (amino acid residues 164 to 186), and C terminus of the G protein were designed and were tested by SPR. Serum samples from REG-vaccinated mice showed strong binding to all G peptides, suggesting that REG induces a diverse antibody immune response that encompasses most of the RSV G protein (Fig. 3C). In contrast, sera from mice vaccinated with RMG generated antibodies that recognized peptides from the CCD and C terminus, but not the N terminus, of RSV G protein. In addition, the RMG-induced antibody titers were lower than those generated following REG vaccination for all peptides tested. These data suggested that the nonglycosylated RSV G protein can induce stronger binding antibodies against more diverse epitopes of the G protein than the glycosylated RMG immunogen.

REG immunization provides better protection than RMG immunization against both homologous and heterologous RSV challenge: G binding antibodies correlate with a reduction in lung viral loads following RSV challenge.

To examine whether REG or RMG immunization could protect from homologous (RSV A2) and heterologous (RSV B1) viral challenge, mice were intranasally (i.n.) infected with 1 × 10⁶ PFU of either RSV A2 or RSV B1, 14 days after the second immunization with REG, RMG, F (positive control), or PBS (negative control). Since neither virus is lethal for mice, we measured viral loads in the lungs on day 4 postchallenge. A 2-log10 reduction in lung viral loads from those in mock (PBS)-vaccinated animals is considered good control of virus replication. The majority of F-vaccinated animals reduced viral loads to undetectable levels after challenge with either the RSV A2 or B1 strain (Fig. 4A and B, purple symbols). Among the REG-vaccinated animals, seven of eight mice were completely protected from viral replication in the lungs, and the eighth animal in each group showed a >100-fold reduction in the viral load following challenge with either the RSV A2 or B1 strain (Fig. 4A and B, red symbols). In contrast, RMG immunization conferred more-variable protection and reduction of viral loads after homologous RSV A2 (2/8 animals protected) or heterologous RSV B1 (4/8 animals protected) virus challenge (Fig. 4A and B, green symbols).

FIG 4.

Immunization with REG protects against homologous and heterologous RSV challenge. (A and B) Mice immunized twice with PBS, REG, RMG, or F protein were challenged i.n. with 1 × 10⁶ PFU of RSV A2 (A) or RSV B1 (B) 14 days after the second immunization. Four days post-virus challenge, mouse lungs were collected and homogenized as described in Materials and Methods, and lung viral loads were determined by a plaque assay. Statistical significance was tested by one-way ANOVA and Bonferroni multiple-comparison tests, *, P < 0.05. (C) The relationship between the total anti-G antibody binding in each individual postvaccination serum sample from PBS-, REG-, and RMG-immunized mice (color coded as in panel A) measured in SPR and the lung viral load at 4 days post-RSV A2 challenge (shown in panel A) was analyzed, and Spearman's correlation was calculated. Total antibody binding to REG protein is expressed in resonance units, and lung viral loads are expressed in PFU per gram of lung. P values less than 0.05 were considered significant.

The lack of detectable neutralizing antibodies in mice that were completely or partially protected from challenge suggests that protection is mediated by immunological functions not captured in the traditional RSV PRNT. This is in contrast to RSV F-immunized animals, for which RSV-PRNT is a predictive assay. A relationship plot between the anti-G binding serum antibody titers of individual animals after the second immunization and lung viral loads after RSV challenge shows that anti-G binding antibody titers have a statistically significant inverse correlation with viral loads of the homologous RSV A2 strain in the lungs (Fig. 4C) (r = −0.6248; P = 0.0014). Results for the REG-vaccinated and RMG-vaccinated animals are represented by red dots and green dots, respectively (the same animals as those for which results are shown in Fig. 4A). For RSV B1, a weaker inverse correlation was found between anti-G binding antibodies and lung viral loads (r = −0.3961; P = 0.0613 [data not shown]).

Therefore, it is apparent that the REG immunogen confers better protective immunity than RMG against both homologous and heterologous RSV challenge. The correlates of protection may include immune mechanisms not measured using a classical RSV PRNT. However, total anti-G binding antibodies as measured in the SPR assay provided a strong correlate with homologous protection as measured by control of virus replication in the lungs.

Glycosylated RMG induces enhanced pathology in the lungs following virus infection, while nonglycosylated REG protein provides protection from lung pathology.

Previous studies suggested that vaccines based on RSV G protein can prime mice for enhanced lung pathology following RSV challenge, which is associated with increased cytokine production and cellular infiltration, including infiltration of eosinophils, neutrophils, and NK cells (31–35). Therefore, we investigated whether intramuscular immunization with REG or RMG leads to enhanced lung pathology following intranasal RSV challenge. The two RSV strains used in the current study, RSV A2 and B1, are known to cause mild lung pathology in mice. Lungs were collected from RSV-infected animals and were analyzed by histopathology as described in Materials and Methods. At 2 days following RSV challenge, lungs from PBS (control)-vaccinated mice showed mild pathology (Fig. 5). In contrast, lungs from RMG-immunized mice demonstrated more cellular infiltrates and higher perivasculitis and interstitial pneumonia scores than lungs from PBS-immunized mice following either homologous or heterologous RSV challenge. In contrast, REG-immunized mice showed lung pathology scores lower than those of PBS-immunized mice after infection with RSV A2. Surprisingly, scores were similarly low in REG-immunized mice following challenge with RSV B1, even though these mice lacked serum neutralizing antibodies against RSV as measured by PRNT (Fig. 5A and B). In addition, cellular infiltrates in the lungs of PBS- and REG-immunized mice were composed mostly of lymphocytes and macrophages with few eosinophils, while the lungs of RMG-immunized mice contained significantly higher numbers of eosinophils (100-fold) and neutrophils (10-fold) than the lungs of REG- or mock-immunized mice (Fig. 5C). Similar lung pathology was observed in animals on day 4 following RSV challenge (data not shown). These results suggest that the nonglycosylated REG immunogen provides good protection from RSV-mediated disease by control of lung viral loads and reduction of lung disease, while the glycosylated RMG counterpart induces enhanced pathology in the lungs following RSV infection.

FIG 5.

Histopathology analysis of lungs from REG- and RMG-immunized mice after virus challenge. REG protects, while RMG induces perivasculitis and cellular infiltrates in the lungs. Lungs collected 2 days after challenge with RSV A2 or RSV B1 were stained with hematoxylin and eosin and were scored for inflammation in bronchioles, near veins (vascular), and alveoli. (A) Lung histology scores represent the averages for 2 slides per mouse and 3 mice per group. Pathology was scored as follows: 0, no lesions; 1, minimal; 2, mild; 3, modest; 4, severe. The maximum pathology score (sum of peribronchiolitis, perivasculitis, and interstitial pneumonia scores) was 12. Neutralizing antibody titers represent 50% inhibition of plaque numbers as measured by PRNT against the respective RSV strains for 3 mice per group. (B) Images (magnification, ×400) of hematoxylin-and-eosin-stained lung sections. (C) Populations of eosinophils, neutrophils, and lymphocytes/macrophages in five different perivascular lung inflammatory foci in three mice per vaccine group, counted 2 days following RSV A2 challenge.

Cytokine profiles in the lungs of REG- and RMG-immunized mice following virus challenge.

The hallmark of RSV vaccine-induced enhanced pathology in mice is an increase in cytokine expression and the development of a skewed Th2 response. Therefore, levels of Th1 or Th2 cytokines and chemokines were measured in lung homogenates of immunized mice following RSV infection (Table 1). Levels of interleukin 4 (IL-4), known to be important for the development of a Th2 response, were >100-fold higher in the lungs of RMG-immunized mice than in those of placebo (PBS)-immunized mice, while IL-4 levels in REG-immunized mice were comparable to those for the PBS control (Fig. 6A). Levels of gamma interferon (IFN-γ), a Th1 cytokine, were only ∼2-fold higher in RMG-immunized mice than in placebo- or REG-immunized mice (Fig. 6E). As a consequence, the ratio of Th2 to Th1 cytokines (IL-4/IFN-γ ratio) was significantly higher in the lungs of mice immunized with the RMG immunogen than in the lungs of placebo- or REG-immunized mice following RSV infection (Fig. 6I). The Th2 cytokines IL-5 and IL-13, known to be involved in eosinophil recruitment, airway hyperresponsiveness, and mucus production (32, 36), were found at significantly higher levels in mice immunized with RMG than in placebo-immunized mice. In contrast, in the lungs of REG-immunized mice, levels of IL-5 and IL-13 were present in ranges similar to those of the placebo-immunized control lungs (Fig. 6B and D). We also identified significant production of IL-6, a proinflammatory cytokine often associated with increased C-reactive protein (CRP) levels and an elevation in body temperature (Fig. 6C). The chemokines eotaxin, monocyte chemoattractant protein 1 (MCP-1), and macrophage inflammatory protein 1α (MIP-1α), previously found to be involved in RSV-induced enhanced lung pathology (37), were also observed at significantly higher levels in the lungs of RSV-challenged RMG-immunized mice than in those of REG-immunized or PBS control mice (Fig. 6F to H). The elevated levels of cytokines and chemokines in RMG-immunized mice following viral challenge seemed to be immune mediated and were independent of lung viral loads following RSV challenge, since similar or higher viral loads were observed in PBS control mice (Fig. 4).

TABLE 1.

Cytokine production in the lungs of immunized and placebo-treated mice on day 2 following RSV A2 challenge

Cytokinea	Concn (pg/ml)b in lungs of animals immunized with:
	PBS	REG	RMG
Th1 cytokines
IL-1α	39.46 ± 1.50	63.77 ± 15.70*	99.71 ± 10.76*#
IL-1β	460.95 ± 61.36	499.99 ± 108.97	1,341.63 ± 396.63*#
IL-2	33.07 ± 1.86	36.06 ± 6.22	57.91 ± 6.27*#
IL-12p40	105.52 ± 3.40	212.82 ± 109.13	93.97 ± 69.25
IL-12p70	96.64 ± 14.02	137.21 ± 37.08	308.91 ± 88.60*#
GM-CSF	300.06 ± 20.84	292.35 ± 24.43	336.40 ± 25.04#
IFN-γ	18.96 ± 1.83	25.48 ± 6.37	46.22 ± 4.33*#
TNF-α	328.87 ± 31.03	374.53 ± 42.91	644.31 ± 60.47*#
Th2 cytokines
IL-4	1 ± 0	3.11 ± 3.68	451 ± 223.96*#
IL-5	5.82 ± 1.82	9.83 ± 3.74	176.67 ± 72.08*#
IL-6	8.1 ± 1.80	11.92 ± 14.14	64.44 ± 20.06*#
IL-9	199.42 ± 34.23	95.41 ± 43.94	356.9 ± 268.40#
IL-13	260.81 ± 106.76	267.51 ± 56.78	581.69 ± 55.35*#
Th17 cytokines, IL-17	9.94 ± 0.84	11.36 ± 2.93	13.45 ± 3.17
Chemokines and growth factors
Eotaxin	1,018.32 ± 143.14	1,065.05 ± 132.29	1,413.19 ± 162.62*#
G-CSF	40.35 ± 8.87	139.19 ± 158.40	136.7 ± 65.89
KC	196.76 ± 48.80	307.05 ± 174.47	564.91 ± 130.99*#
MCP-1	381.98 ± 32.60	626.59 ± 431.41	2,474.1 ± 1,153.93*#
MIP-1α	50.98 ± 4.27	63.21 ± 23.06	2,025.14 ± 1,767.41*#
MIP-1β	59.54 ± 7.74	64.41 ± 11.47	289.51 ± 155.53*#
RANTES	272.79 ± 7.69	436.25 ± 194.68	711.81 ± 227.38*
Th1 + Th2-expressed cytokines
IL-3	11.72 ± 0.84	14.19 ± 2.72	13.65 ± 2.22
IL-10	48.7 ± 4.12	61.12 ± 16.27	148.83 ± 16.42*#

^aGM-CSF, granulocyte-macrophage colony-stimulating factor; TNF-α, tumor necrosis factor α; G-CSF, granulocyte colony-stimulating factor; KC, keratinocyte chemoattractant.

^bValues are averages for three mice ± standard deviations. Symbols indicate significant differences (P < 0.05) from levels in PBS-immunized (*) or REG-immunized (^#) animals.

FIG 6.

RMG (but not REG) immunization induces high levels of Th2 cytokines and chemokines following RSV i.n. challenge. (A to H) Cytokines in lung homogenates from day 2 postchallenge were measured in a Bio-Plex Pro mouse cytokine assay. Values are concentrations, expressed in picograms per milliliter, for 3 mice per group. The box extends from the 25th to the 75th percentile, and the error bars represent the lowest and highest values. Values below the limit of detection of the assay were assigned a number according to the minimum detection limit of the cytokine. The mean for each group is shown. (I) Ratio of the observed concentration of Th2 cytokines to that of Th1 cytokines as measured by the IL-4/IFN-γ ratio. Statistical significance was analyzed by one-way ANOVA and Bonferroni multiple-comparison tests. ***, P < 0.0001; **, P < 0.001; *, P < 0.05.

Taken together, the findings of this study demonstrated that RMG vaccination induces enhanced lung cellular infiltration, as measured by both histopathology and cytokine levels after RSV challenge. At the same time, the RMG vaccine does not elicit strong protective immunity. On the other hand, vaccination with the unglycosylated REG did not induce enhanced lung pathology and did confer protection against both homologous and heterologous RSV strains, even in the absence of a neutralizing antibody response as determined by PRNT.

DISCUSSION

New efforts are under way to develop subunit vaccines against RSV that will provide protective immunity without the potential for disease enhancement that was observed in infants following vaccination with the formalin-inactivated RSV (FI-RSV) vaccine in the 1960s (5, 38–40). Most recent RSV vaccine development efforts are focused on the F protein, which is relatively conserved among strains (3) and can be expressed in mammalian and insect cells (14, 18, 19, 41). The G attachment protein, expressed on the virus surface, also represents an important target of protective immunity. However, in earlier studies, glycosylated G protein expressed in mammalian cells or a recombinant vaccinia virus expressing G protein was shown in mouse models to induce pulmonary eosinophilia upon RSV infection (36, 42, 43). Therefore, we decided to evaluate side by side the immunogenicity and safety of the recombinant unglycosylated G protein ectodomain (amino acids 67 to 298) expressed in E. coli (REG) and those of a fully glycosylated G ectodomain produced in 293-Flp-In mammalian cells (RMG) in a mouse RSV challenge model. The FI-RSV model was not included in the current study, because it has been documented in previous studies, demonstrating immune response imbalance and levels of lung cytokines and cellular infiltrates that were associated with enhanced lung pathology. The targets of the nonneutralizing antibodies were not completely deciphered. Furthermore, F and G subunit vaccines studied early on have a history of disease enhancement, albeit to a lesser extent than FI-RSV (3).

Previous studies in which glycosylated G protein was shown to be protective in mice typically used three vaccinations with ≥10 μg protein per dose (44–47). In this study, we immunized mice with only two vaccinations of a lower dose of the G proteins (5 μg/dose) to determine the impacts of glycosylation on virus neutralization and the disease enhancement phenomenon after RSV infection. Two vaccinations with REG generated neutralizing antibodies against RSV A2 in 7/11 animals, while glycosylated RMG did not elicit neutralizing antibodies as measured by the PRNT. Furthermore, total binding antibodies against the recombinant proteins (both REG and RMG) were found to be >10-fold higher in REG-immunized mice sera than in RMG-vaccinated mice. REG immune sera also demonstrated broader epitope recognition, spanning the entire G ectodomain, than RMG immune sera. Homologous and heterologous protection was evaluated by measuring lung viral loads after challenge with either the RSV A2 (homologous) or the RSV B1 (heterologous) strain. Complete control of viral loads was observed in the majority of animals vaccinated with REG and challenged with either RSV strain. In contrast, the majority of RMG-vaccinated animals did not control virus replication after homologous or heterologous virus challenge. There were animals with low or no serum neutralization titers that controlled lung viral loads very well. This was most evident in animals challenged with the heterologous RSV B1 strain. It has been shown before that anti-G antibodies may neutralize virus in vivo but not in the in vitro PRNT (29). Antibody-dependent cellular cytotoxicity (ADCC) and/or cell-mediated immunity may also play a role in virus clearance after RSV challenge. However, previous studies showed that BALB/c mice immunized with either FI-RSV or a recombinant vaccinia virus that expresses the glycosylated RSV G attachment protein generated RSV-specific CD4⁺ T cells but did not prime CD8⁺ T cells (48, 49). In our study, a strong inverse correlation was observed between the total anti-G binding serum antibody titers (measured by SPR) and lung viral loads after RSV challenge. Therefore, it is likely that anti-G binding antibodies that are protective in vivo do not neutralize RSV in the PRNT, as was observed for anti-G MAb 131-2G (29). Therefore, this study demonstrates that for an RSV G protein-based vaccine, the correlate of protection mediated by total G-binding antibodies as measured by SPR may provide a reasonably good predictor of control of viral replication after intranasal RSV challenge. Additional studies with patient-derived RSV isolates (as they become available) are planned to confirm and expand our findings. These analytical tools could also be evaluated further in preclinical and clinical studies of other RSV vaccines containing both F and G proteins, including live-attenuated vaccines under investigation in children (11, 12).

An important safety consideration for the use of subunit and nonreplicating candidate vaccines against RSV is enhanced Th2 cytokine response and its potential to increase disease severity upon virus challenge. The RSV G glycoprotein in particular has been implicated as an RSV antigen that promotes Th2 CD4⁺ T lymphocytes and induces eosinophilic infiltrates in the lungs after RSV challenge (31, 32, 36, 50–53). In the current study, animals vaccinated with glycosylated RMG, but not PBS- or REG-vaccinated animals, demonstrated Th2-biased cytokine and elevated chemokine responses in the lungs that correlated with significant lung histopathology after either RSV A2 or RSV B1 challenge. The higher numbers of cellular infiltrates containing eosinophils (100-fold) and neutrophils (10-fold) in RMG-vaccinated mice following RSV challenge recapitulated the lung pathology observed previously in models of enhanced RSV disease (33, 34). The enhanced lung pathology in RMG-immunized mice following viral challenge seems to be immune mediated, since similar or higher viral loads were observed in PBS control mice, which showed lower cytokine levels and less lung pathology.

The two purified recombinant G proteins were identical in terms of their primary amino acid sequence and were administered in combination with the same adjuvant, Emulsigen. The only difference between the REG and RMG proteins was the presence of glycosylation as N- and O-linked glycans in the RMG protein. Therefore, the observed shift toward a higher Th2/Th1 ratio after RMG vaccination could not be attributed simply to binding of the CX3C motif within the CCD to CX3CR1 (the fractalkine receptor) on several cell types, as suggested previously (54–56). Instead, the enhanced Th2 cytokine and chemokine levels could possibly be induced by the high level of O-linked sugars characteristic of mammalian-cell-expressed glycosylated RMG and also of the native-virus-associated G attachment protein. A role for processing by different glycosylation-specific antigen-presenting cell subsets and routes of immunization that could influence the subsequent balance between Th2 and Th1 cytokines has been reported and should be further investigated (57–59). Carbonylation of protein is one of several factors that have been proposed to be responsible for the induction of Th2 disease in this model. However, we did not observe any difference in the carbonyl contents of the REG and RMG proteins used in our studies.

Our findings are in agreement with those of previous studies using the E. coli-produced fusion protein BBG2Na, which contained the central conserved region of the A2 G gene (amino acids 130 to 230) (G2Na) fused to the albumin-binding domain of streptococcal protein G (BB) formulated with an aluminum adjuvant (44–47). In comparison with BBG2Na, our REG encompasses the entire G ectodomain with no “foreign” sequences or fusion protein that may influence the immune response to the irrelevant target (BB rather than G2Na) during vaccinations. Most of the studies with BBG2Na used 20 μg of the vaccine in a series of three vaccinations. In contrast, in our study, only two vaccinations with 5 μg REG protein/dose elicited strong protection. Also, REG immunization induced antibodies with a more-diverse epitope repertoire than BBG2Na. In summary, our study for the first time compared side by side fully glycosylated and unglycosylated G proteins for immunogenicity and safety in a murine RSV challenge model. The combination of low virus neutralizing activity, lower G-binding antibody titers, and enhanced lung pathology observed in the RMG-vaccinated animals is in agreement with the findings of previous studies using various forms of glycosylated G proteins (31, 32, 34, 52, 60). In contrast, the lack of enhanced lung pathology after REG vaccination was an unexpected, encouraging finding that provides support for further development of this vaccine approach. It also provided data supporting the feasibility of developing a simple recombinant RSV G-based vaccine produced in an E. coli expression system that elicits protective antibodies against both homologous and heterologous RSV challenge. The bacterial production system for a G protein-based vaccine provides an economical, simple, and rapid alternative to cell-based subunit vaccines, as was previously demonstrated for influenza virus HA1 proteins from multiple virus strains (25, 61–63), and could be applied for the expression of multiple G proteins from circulating RSV strains. Such a multicomponent G immunogen could be tested alone or in combination with F-subunit vaccines for effective protection against RSV disease.