Subcellular Site of Expression and Route of Vaccination Influence Pulmonary Eosinophilia Following Respiratory Syncytial Virus Challenge in BALB/c Mice Sensitized to the Attachment G Protein¹

Tcells play a central role in the regulation of immune responses to pathogens. Two functional phenotypes of CD4⁺ T cells have been defined based on different profiles of cytokine secretion; the development of one response to the exclusion of the other can have profound implications for resistance to infection and in the pathogenesis of disease. Th1 cells secrete cytokines such as IFN-γ, IL-2, and TNF-β, which mediate macrophage activation and delayed-type hypersensitivity reactions. Th2 cells secrete cytokines such as IL-4, IL-5, IL-6, and IL-10, which are important growth and differentiation factors for B cells. There is evidence that the selective differentiation of Th subsets can be influenced by a variety of factors, such as the nature of the APC and the selective expression of costimulatory molecules on the APC, the cytokine environment in which priming occurs, the nature and concentration of Ag, and the binding affinity of the antigenic peptide for MHC class II or TCR (1).

The selective differentiation of T cell subsets appears to play a role in determining the outcome of infection with respiratory syncytial virus (RSV),³ which is the single most common cause of viral bronchiolitis in young children. Although the development of a vaccine against RSV is a high priority, no effective vaccine against human RSV (HRSV) is available. Indeed, vaccine development has proceeded with caution following the specter of vaccine-augmented disease observed following vaccination with formalin-inactivated RSV (2, 3). The BALB/c mouse model of RSV infection has revealed that T cells and the cytokines they produce play an important role in determining disease outcome. Mice sensitized to individual RSV proteins show characteristic changes in pulmonary pathology and Th priming following RSV challenge. Thus, although mice vaccinated with recombinant vaccinia virus (rVV) expressing either the fusion (F) or the attachment protein (G) of RSV are protected against subsequent RSV infection, vaccinated mice develop increased pulmonary pathology compared with control animals undergoing a primary RSV infection (4). Lung lesions in mice primed with the F protein are characterized by peribronchiolar and perivascular infiltrations of lymphocytes and polymorphonuclear leukocytes (4), whereas lesions in mice primed with the G protein are characterized by lymphocytes and an extensive eosinophil infiltration (5, 6). The G protein primes Th2 CD4⁺ T cells but fails to induce class I-restricted CD8⁺ T cells. In contrast, the F protein primes Th1 CD4⁺ T cells and CD8⁺ CTLs (7). The reasons why these two glycoproteins delivered in the same form and by the same route should prime different T cell subsets and induce distinct patterns of pathology after RSV challenge are not clear.

The F protein of RSV is a type I, membrane-bound glycoprotein that mediates fusion of the viral membrane with that of the host cell to initiate a new infective cycle (8). The protein is synthesized as a single polypeptide, Fo precursor, that is posttranslationally cleaved into two subunits, F1 (49 kDa) and F2 (20 kDa), which remain associated by disulfide bridges. The F1 subunit has a single potential site for N-glycosylation, whereas the F2 subunit has four to five potential sites (9). The G protein of RSV is naturally synthesized as a type II, membrane-anchored glycoprotein and a smaller, soluble form, which lacks the cytoplasmic domain and part of the membrane anchor sequence. Synthesis of these two distinct forms is achieved by initiating translation at either of two different AUG codons near the 5′ end of the G open reading frame (10). Although the predicted mass of the unglycosylated G protein is ∼33,000 Da, the full length mature protein has a molecular mass of 86,000 to 90,000 Da as estimated by SDS-PAGE (11, 12). This apparent difference in M_r is due to extensive posttranslational modification, the G gene encoding 4 potential N-linked and 77 potential O-linked glycosylation sites (13, 14, 15).

In this study, the contribution of the different forms of the G protein to the pattern of pulmonary pathology and the character of the cytokine response in the spleens of BALB/c mice following intranasal (i.n.) challenge with RSV was investigated. Furthermore, since previous studies had failed to identify eosinophils in the lungs of mice vaccinated with rVVG by the i.p. route and challenged with RSV (4), whereas eosinophils were apparent in mice vaccinated by scarification (5, 6), the effect of the route of vaccination on induction of pulmonary pathology and T cell priming was also investigated.

HRSV Long strain was grown in HEp-2 cells as previously described (16). The A2 strain of HRSV was grown in fetal calf kidney cells. Virus pools that contained 10⁶ pfu/ml were stored in liquid nitrogen and used in all experiments. A list of rVV used in this study is given in Table I. HRSV genes were inserted in either of two sites of the vaccinia genome: the tk locus or the VP37 locus. Insertion in the tk locus was achieved by the procedure of Chakrabarti et al. (17). The foreign genes were cloned into pSC11 vector under the early/late 7.5 promoter. CV1 cells were infected with the WR strain of VV and transfected with plasmid DNA. Recombinants were selected from progeny virus by infecting HuTK⁻143B cells in the presence of 25 μg/ml⁻¹ of BrdU. rVV (VA-F) expressing the F protein of the Long strain has been reported previously (9). It has been renamed rVVF in this paper to normalize nomenclature. Insertion in the VP37 locus was achieved by the procedure of Blasco and Moss (18). In this case, the foreign genes were cloned into the pRB21 plasmid, which carries a complete copy of the VP37 gene, under a synthetic early/late promoter. CV-1 cells were infected with VRB12 VV, which has the VP37 gene disrupted and is unable to form extracellular virus, and transfected with pRB21-derived plasmids. Recombinant viruses were selected by their ability to form plaques. The G protein gene of the Long strain, previously cloned into pGEM-4 (plasmid LG3A (19)), was released by digestion with EcoRI and HindIII and subcloned into plasmid pRB21 digested with the same enzymes to obtain plasmid pRBG. From this plasmid, the recombinant (VRBG) rVVG was obtained. Two other G genes were derived from LG3A, following the work of Roberts et al. (10), to encode either the membrane-bound (pGM48I) or the soluble form (pGM48) of the Long strain G protein. Plasmid pGM48I was obtained by PCR-based site-directed mutagenesis (20) as previously described (21), changing codons 48 ATG–ATC (Met-Ile) and 49 ATA–GTA (Ile-Val). Plasmid pGM48 was obtained by PCR using a primer that started in the second ATG of the G protein gene and a second primer that included the last 18 nucleotides of the G gene. The two G inserts were transferred from pGEM-4 to pRB21 and recombinants (VRBGM48I) rVVGmem and (VRBGM48) rVVGsol were obtained as described before. rVV were propagated in CV-1 cells, purified by sucrose gradient centrifugation, and titrated on HTK cells as described previously (4).

Six-week old, specific pathogen-free, female BALB/c mice were bred at the Institute for Animal Health. Mice were inoculated i.p. or by scarification, with 2 × 10⁶ pfu of rVV. Serum samples were taken 3 wk post immunization by bleeding mice from the tail vein and 5 days after challenge at postmortem. Groups of four to five mice were challenged i.n. with ∼10⁵ pfu of the A2 strain of RSV. Five days after challenge, mice were killed with an overdose of pentobarbitone administered i.p., and lungs were removed. RSV titers in lung homogenates were determined by plaque assay (22). Further groups of mice were killed 5 days post-RSV challenge and their lungs subjected to repeated bronchoalveolar lavage (BAL) (23). Each round of BAL consisted of inflating the lungs with 1 ml of 12 mM lidocaine in PBS, three times over a 1- to 2-min period. The first round of BAL from each mouse was used for total cell counts, and cytocentrifuge preparations of cells were stained with May-Grunwald Giemsa. Differential counts of 350 to 450 cells/slide were made using oil immersion. More cells were obtained from mice by inflating lungs with 1 ml of 12 mM lidocaine a further three times on two occasions. Cells obtained from further rounds of BAL were pooled together in ice cold RPMI with 10% heated FCS together with the cells obtained from the first round of BAL (after samples had been removed for cell counts and differentials) for flow cytometric analysis. All experiments were repeated on at least two occasions.

Cells obtained by repeated BAL were washed in RPMI-10% heat-inactivated FCS and resuspended to 1 to 5 × 10⁶ cells/ml. Two-color flow cytometric analysis of isolated cells used rat anti-mouse CD4 coupled to FITC (Sigma, Poole, U.K.) and biotinylated rat anti-mouse CD8 (PharMingen, San Diego, CA) followed by streptavidin-phycoerythrin (Southern Biotechnology Associates, Birmingham, AL). Staining was analyzed on a FACScan (Becton Dickinson, Mountain View, CA).

The presence of serum Abs to RSV were determined by ELISA using the A2 strain of HRSV, as described previously (24). Different Ab isotypes were determined using horseradish peroxidase-conjugated rabbit anti-mouse IgG1, IgG2a (ICN Biomedicals, Thame, U.K.) and horseradish peroxidase-rat anti-mouse IgE (Serotec, Oxford, U.K.).

HEp-2 cells growing in 60-mm petri dishes were infected with either HRSV Long strain (m.o.i., 1–2 pfu/cell) or rVV (m.o.i., 5 pfu/cell) in DMEM supplemented with 2.5% heated FCS. Twenty-four (VV) or forty-eight (Long) hours later, culture supernatants were saved, and cell extracts were made in 0.2 ml of buffer A (10 mM Tris-HCl, pH 7.6, 140 mM NaCl, 5 mM EDTA, 1% Triton X-100, 1% sodium deoxycholate). Cell extracts (1/20) and culture supernatants (1/100) were diluted with sample buffer and separated by SDS-PAGE. Proteins were electrotransferred to Immobilon-P membranes (Millipore, Watford, U.K.) and developed with a pool of anti-G mAbs by the enhanced chemiluminescence (ECL) procedure as recommended by the manufacturer (Amersham, Little Chalfont, U.K.).

HEp-2 cells growing in tissue culture chamber slides (Nunc, Naperville, IL) were infected with VV as indicated before. Twenty-four hours later, the cells were fixed with either cold methanol for 5 min and acetone for 30 s or with 3.5% formaldehyde in PBS for 30 min. Then, the cells were processed for indirect immunofluorescence as described previously (25).

Lymphocytes were isolated from spleens obtained from immunized mice 5 days after RSV challenge and restimulated with RSV-infected autologous splenocytes as described previously (26). Supernatants from restimulated lymphocytes were harvested on a daily basis and assessed for cytokine production.

The concentrations of IL-2, IL-4, IL-5, IL-10, and IFN-γ in supernatants from lymphocyte cultures was measured using commercial ELISA reagents under conditions recommended by the manufacturer (PharMingen).

Statistical analysis was performed by a two-sample t test using the Minitab for Windows statistical software package (Bioscience IT Services, Harpenden, U.K.).

The recombinant viruses encoding different forms of the G protein are listed in Table I. Recombinant VVG contains the wild-type gene and thus encodes both the membrane-bound and the soluble form of the G protein by alternative initiation of translation at the first or second AUG of the open reading frame, respectively (10). The rVVGmem, in which the second AUG was eliminated by introducing the changes M48I and I49V, is predicted to encode only the membrane-bound form of the G protein. rVVGsol, which was obtained by PCR with primers designed to eliminate the segment preceding the second AUG, is predicted to encode only the soluble form of the G protein.

The G protein produced in cells infected by the various VV recombinants was analyzed by Western blot (Fig. 1). A wide band of 80 to 90 kDa, typical of the mature G protein (10, 15), was detected in extracts of HEp-2 cells infected with HRSV. A wide band with slightly higher mobility, corresponding to the soluble form of the G protein that lacks the N-terminal cytoplasmic and part of the transmembrane domain, was observed in the supernatant of HRSV-infected cells. Similarly, G protein was detected in the extracts and supernatants of cells infected with rVVG that have the HRSV Long G protein gene inserted in the VP37 locus. It should be noted that the G protein encoded by the vaccinia recombinant has a slightly retarded migration compared with HRSV Long virus, due to differences in glycosylation (our unpublished observations). As expected, the G protein produced by rVVGmem was detected in cell extracts, but not in culture supernatants. In contrast, the G protein produced by rVVGsol was found at low levels in the cell extracts and at normal levels in the supernatants, indicating that most of the protein was being secreted and only traces of the shorter form, in its traffic to the cell exterior, remained cell associated.

Immunofluorescence was used to examine the sites of expression of the G protein encoded by the different rVV in HEp-2 cells. Anti-G mAbs stained both rVVGmem- and rVVGsol-infected cells when fixed with methanol/acetone, although as expected, fluorescence intensity was lower in rVVGsol than in rVVGmem cells (Fig. 2). In contrast, in cells fixed with formaldehyde to preserve cell membrane impermeability to Abs, only rVVGmem expressed detectable G protein on the surface of infected cells (Fig. 2). The pattern of staining of cells infected with rVVG was the same as that of rVVGmem-infected cells (not shown).

To determine whether the site of expression of the G glycoprotein or the route of immunization influenced humoral immune responses, RSV-specific Ab was analyzed in sera by ELISA 3 wk postvaccination and 5 days post-RSV challenge, (Table II). The titers of RSV-specific serum Ab in mice immunized with rVVF, rVVG, or rVVGmem were similar irrespective of the route of vaccination (Table II). However, marked differences in Ab titers were observed in mice immunized with rVV expressing the soluble form of the G protein by scarification compared with those immunized via the i.p. route (Table II). Thus, i.p. immunization resulted in ∼550-fold more RSV-specific Ab than that induced by scarification. This large difference reflects the very low levels of Ab induced by rVVGsol after scarification. To determine the effect of site of expression of the G protein and route of vaccination on the isotype of RSV-specific Abs, the same sera were examined by ELISA, utilizing IgG1-, IgG2a-, or IgE-specific secondary Abs. Three weeks after vaccination with rVVF or rVVGmem, by either i.p. injection or scarification, the titers of IgG1 and IgG2a were similar (Fig. 3), whereas titers of IgG1 were ∼5- or 9-fold greater than IgG2a titers in mice vaccinated i.p. with rVVG or vaccinia expressing the soluble form of the G protein, respectively (Fig. 3). RSV-specific IgE was not detected in sera from any of the vaccinated mice.

Protection against RSV infection was examined 3 to 4 wk after vaccination. Vaccination with rVVF, rVVG, rVVGmem, and rVVGsol conferred protection against a subsequent RSV infection, as virus was not recovered from the lungs of any mice 5 days after RSV challenge (Table II). In contrast, high titers of RSV were recovered from the lungs of mice vaccinated with rVV-β-galactosidase (rVVβgal) (Table II).

To determine the influence of the site of expression of the G protein on the cellular response in the lungs of scarified mice, the cellular content of BAL fluid was examined 5 days after RSV challenge. Preliminary studies showed that there were no differences in the pulmonary inflammatory response after i.n. challenge with the A2 strain of HRSV in mice vaccinated with rVV expressing the G protein of the A2 strain of HRSV, inserted in the tk locus, compared with animals vaccinated with rVV expressing the G protein of the Long strain of HRSV, inserted in the VP37 locus. After RSV challenge, there was a 3- to 10-fold increase in the number of cells in BAL from immunized mice compared with rVVβgal controls (Table II). Cytocentrifuge preparations of BAL cells from individual mice were examined for changes in polymorphonuclear granulocyte content. As seen previously (5), scarification with rVVF induced a marked neutrophil efflux into the lungs after RSV challenge, whereas scarification with rVVG induced a pulmonary eosinophilic response. The contribution of the different forms of the G protein to the eosinophilic response following RSV challenge was assessed in mice scarified with rVV expressing either soluble, membrane-anchored, or both (wild-type) forms of the G protein. In three independent experiments, greater numbers of eosinophils were present, following RSV challenge, in the BAL from mice scarified with rVVGsol or rVVG compared with mice scarified with rVVGmem (p < 0.005) (Table III). These observations indicate that the eosinophilic response in mice scarified with rVVG is largely due to the soluble form of the G protein. However, the membrane-anchored form of the G protein retains some capacity to prime for an eosinophilic response following RSV challenge. Although the differences were not statistically significant, the numbers of eosinophils in BAL from mice scarified with rVVGsol exceeded those from mice scarified with rVVG in each of three independent experiments. However, if the eosinophil counts from all three experiments were amalgamated, there was a significant increase in the number of eosinophils in BAL from mice scarified with rVVGsol (mean cell count, 160.3 × 10³ ± 63 cells/ml) compared with that from mice scarified with rVVG (mean cell count, 109.5 × 10³ ± 48) (p < 0.04).

Mice vaccinated by the i.p. route were also subjected to BAL, following RSV challenge, to determine the influence of the route of vaccination on the cellular response in the lung and the induction of pulmonary eosinophilia. The cellular response was influenced by the route of vaccination. Thus, few eosinophils were observed in the BAL from mice vaccinated i.p. with any of the rVV expressing different forms of the G protein after RSV challenge (Table III). Furthermore, following RSV challenge, the numbers of neutrophils in BAL from mice scarified with either rVVF or rVVGmem were significantly greater than in BAL from mice immunized by the i.p. route (p < 0.005 and p < 0.001, respectively) (Table III).

The effect of vaccination with rVV that expressed different forms of the G protein on the recruitment of CD4⁺ and CD8⁺ T cell subsets into the lung was assessed 5 days after RSV challenge by flow cytometry of pooled BAL from groups of five mice. Approximately equal numbers of CD8⁺ and CD4⁺ cells were present in the BAL from mice scarified with rVV expressing the F protein (Table IV), whereas the numbers of CD4⁺ cells present in the BAL from mice scarified with rVVG, rVVGmem, or rVVGsol exceeded the numbers of CD8⁺ T cells by ∼1.5 to 2:1 (Table IV). However, the route of vaccination did influence the ratio of CD4⁺:CD8⁺ T cells in BAL of mice following RSV challenge. Thus, CD8⁺ T cells outnumbered CD4⁺ cells by 2.3:1 in the BAL from mice vaccinated i.p. with rVVF and by 1.5 to 2:1 in the BAL from mice vaccinated i.p. with rVVG, rVVGmem, or rVVGsol (Table IV). Thus, priming of mice by scarification favored an influx of CD4⁺ T cells into the lungs after RSV challenge, whereas i.p. vaccination resulted in a bias toward CD8⁺ T cells in BAL, irrespective of the RSV Ag used to prime the mice.

To determine whether differences in the pulmonary eosinophilic response to RSV challenge were due to differences in Th priming, the cytokines produced by memory/effector T cells were analyzed after stimulation in vitro with RSV. Previous studies have shown that BALB/c mice scarified with rVVF are primed for a Th1-like response to RSV, whereas mice scarified with rVVG are primed for a Th2-like immune response (7). These initial investigations made use of biologic assays to assess the cytokines present in culture supernatants of in vitro restimulated splenocytes from immunized mice (7). However, using Ag-capture ELISA, Srikiatkhachorn and Braciale (6) found that the culture supernatants from restimulated splenocytes of rVVF- or rVVG-primed mice contain predominantly IL-2 and IFN-γ following one round of in vitro restimulation with RSV. Th1- or Th2-like cytokine production from rVVF- or rVVG-primed mice was observed only after in vitro stimulation with RSV of spleen cells from vaccinated mice 5 days after RSV challenge (6). Our preliminary results were essentially similar to those of Srikiatkhachorn and Braciale (6), and we therefore focused our attention on analyses of cytokine production by spleen cells following challenge of vaccinated mice. Spleen cells obtained 5 days after RSV challenge were stimulated in vitro with RSV-infected autologous splenocytes. Lymphocytes from mice scarified with rVVF produced predominantly IL-2 and IFN-γ, with little or no IL-4 or IL-5, whereas lymphocytes from mice scarified with rVVG produced IL-4 and IL-5 and lower levels of IL-2 and IFN-γ (Fig. 4,A). A similar pattern of cytokine production was observed in lymphocytes from mice scarified with rVV expressing membrane-anchored or soluble forms of the G protein, although the latter primed for significantly greater levels of IL-5 production than the other recombinants. The greater IL-5 production and lower levels of IL-2 and IFN-γ production were associated with a greater pulmonary eosinophilia in mice scarified with rVVGsol compared with those scarified with rVVGmem. Whereas the route of vaccination influenced pulmonary pathology after RSV challenge, it did not significantly influence the pattern of cytokine production by spleen lymphocytes. Thus, lymphocytes from mice vaccinated i.p. with rVVF produced high levels of IL-2 and IFN-γ with little or no IL-5, and mice vaccinated i.p. with rVVGmem or rVVGsol produced high levels of IL-5 and only low levels of IFN-γ (Fig. 4,B). However, there were differences in IL-4 production between mice vaccinated i.p. compared with those vaccinated by scarification. Spleen cells from mice vaccinated i.p. with rVVF or rVVGmem consistently produced higher levels of IL-4 than lymphocytes from mice scarified with these recombinants. In fact, i.p. vaccination with rVVGmem primed for (∼10-fold) higher levels of IL-4 production compared with splenocytes from rVVGsol- or rVVF-vaccinated mice (Fig. 4,B). Splenocytes from mice vaccinated by either route with rVVβgal failed to produce detectable levels of IL-5 or IL-4 and only low levels of IL-2 and IFN-γ after restimulation in vitro with RSV (Fig. 4, A and B).

The induction of pulmonary eosinophilia in rVVG-primed mice was influenced both by the nature of the G protein and by the route of vaccination. Thus, greater numbers of eosinophils were detected in BAL from mice scarified with rVV expressing the soluble form of the G protein than from those primed with rVV expressing the membrane-anchored version. However, priming for the induction of pulmonary eosinophilia was not solely dependent on immunization with soluble Ag, since scarification with rVV expressing the membrane-anchored form of the G protein does contribute to the induction of pulmonary eosinophilia. Thus, the numbers of eosinophils detected in BAL from rVVGmem-primed mice, although less than those from mice primed with the soluble form of the G protein, were still greater than those in BAL from rVVF-primed mice. The relative amounts of the two forms of the G protein produced in vivo by rVV expressing wild-type G protein are not known. However, since the pulmonary eosinophil response induced by rVV expressing the wild-type G protein was only ∼30% less than that induced by rVVGsol, it may be that the soluble form of G protein is the predominant form produced in vivo. Although the highest levels of IL-5 and the lowest levels of IL-2 and IFN-γ were produced by spleen T cells from mice primed with soluble G, the reduced eosinophilic response in the lungs of mice primed with the membrane-anchored form of the G protein, compared with those primed with wild-type G protein, was not associated with reduced levels of Th2 cytokines in the spleen.

Priming with high vs low doses of Ag can skew the CD4⁺ T cell response in favor of Th1- or Th2-like responses. Low doses of bacterial flagellin (27), Leishmania major (28), or Trichuris muris (29) induce Th1-like responses. In contrast, infection with increasing numbers of parasites leads to the generation of a Th2-like response (1, 29). Other studies have demonstrated the opposite dichotomy in responses with respect to high vs low doses of Ag (1). The levels of expression of the G protein from the different rVV constructs were similar in vitro, but the relative amounts of G protein produced by rVV following vaccination by the different routes will be difficult to ascertain. The higher Ab titers in mice vaccinated i.p. with rVVG compared with those following scarification may be related to higher doses of Ag.

Differences in the induction of Th2 cells and the eosinophilic response may be related to the physical characteristics rather than the amount of G protein produced by the rVV. Indeed, other studies have shown that soluble Ags preferentially induce Th2 responses (30). Furthermore, scarification of BALB/c mice with a rVV expressing a soluble form of the F protein primed spleen T cells for the production of Th2 cytokines after i.n. challenge with RSV (Bembridge et al., manuscript in preparation). However, these mice did not develop a pulmonary eosinophilia after subsequent RSV challenge. These observations, together with the finding that >40% of BAL cells are eosinophils in mice immunized with purified G protein and QS-21 adjuvant, whereas only 4% of BAL cells are eosinophils in mice immunized with purified F protein and QS-21 (31), suggest that the G protein may have intrinsic properties that prime for an eosinophilic response. It may be that the extensive glycosylation of the G protein (9) influences the priming of T cells and the induction of eosinophils.

The RSV-specific Ab response induced by rVV expressing the soluble form of the G protein was affected by the route of vaccination. Thus, the serum Ab titer in mice scarified with rVV expressing the soluble form of the G protein was greatly reduced in comparison with that induced following i.p. vaccination with this recombinant virus. Furthermore, Ab titers were not increased following RSV challenge. Nevertheless, the mice were completely protected against RSV infection. Since protection in mice vaccinated by rVVG appears to be mediated by Ab (32), it is not clear whether the low level of Ab induced in mice scarified with rVVGsol is sufficient to mediate protection. Since adoptively transferred G protein-specific CD4⁺ T cell lines with Th2 cytokine production profiles can reduce RSV replication in the lungs (33), it may be that resistance to RSV challenge in mice scarified with rVVGsol is mediated by T cells. The poor Ab response induced by scarification with rVV expressing the soluble form of the G protein contrasts with the high levels of serum Ab induced following scarification with rVV expressing a soluble form of the F protein (Bembridge et al., manuscript in preparation). The reasons for the poor Ab response, despite efficient priming of T cells, in mice scarified with rVVGsol are not clear but may be related to the interaction of the soluble G protein with APCs or indeed B cells.

Priming for the induction of pulmonary eosinophilia by rVV expressing the G protein was also influenced by the route of vaccination. Thus, following RSV challenge, pulmonary eosinophilia was not induced in mice vaccinated i.p. with rVV expressing any of the different forms of the G protein. Nevertheless, i.p. vaccination with rVV expressing either the membrane-anchored or the secreted form of the G protein primed spleen T cells for the production of high levels of IL-5 and low levels of IFN-γ. In fact, the levels of IL-4 and IL-5 were similar to or higher, and the levels of IFN-γ were similar to or lower than those from mice primed by scarification. It may be that the pattern of cytokine production by spleen cells in vaccinated mice after RSV challenge does not accurately reflect the pattern of cytokine production in the lung and that further studies are needed to characterize the pulmonary cytokine response following RSV challenge in these mice.

Recent studies indicate that virus-specific CD8⁺ T cells appear to play a critical role in the regulation of Th2 cytokine secretion and recruitment of eosinophils into the lungs following RSV infection (33, 34, 35). Following RSV challenge of mice vaccinated by the i.p. route, the proportion of CD8⁺ T cells in BAL consistently outnumbered CD4⁺ T cells, whereas the reverse was true for mice vaccinated by scarification. It is possible, therefore, that the presence of excess CD8⁺ T cells in the lungs of mice vaccinated by the i.p. route and challenged 5 days previously with RSV suppressed eosinophil recruitment into the lungs. Since rVVG does not appear to prime RSV-specific CD8⁺ cytotoxic T cells in BALB/c mice vaccinated either by the i.p. route or after scarification (5, 36), further work is needed to analyze the specificity and effector functions of the pulmonary CD8⁺ T cells present in the lungs after RSV challenge of mice vaccinated by the i.p. route. Although the presence of greater numbers of CD8⁺ T cells in the lungs of mice immunized by the i.p. route may be responsible for down-regulating the recruitment of eosinophils into the lungs, differences in lymphocyte subsets in the lung do not account for differences in eosinophil responses between mice scarified with rVVGsol and those scarified with rVVGmem. Furthermore, since there were no significant differences in cytokine responses in spleens from mice scarified with either rVVG or rVVGmem, and RSV-specific IgE Abs were not detected in any of the vaccinated mice, the reasons for the reduced pulmonary eosinophil response in mice primed with rVVGmem are not clear.

There are a number of studies that suggest that IFN-γ produced by CD8⁺ T cells can suppress CD4⁺ Th2 responses in the airways (37, 38). However, other studies have shown that the cytokine profile of virus-specific CD8⁺ T cells can be switched from IFN-γ to IL-5 production by bystander CD4⁺ Th2 responses, leading to the accumulation of eosinophils in the lung (39). Therefore, further studies comparing the recruitment and activation of T cells by rVVG after vaccination by different routes will not only increase our understanding of the selective differentiation of Th subsets but will also aid in the design of a vaccine against RSV that avoids potentially damaging immune responses.

Our findings indicate that a vaccine in which the G protein is expressed only as a membrane-anchored form would reduce the potential of the vaccine to prime for pulmonary eosinophilia. Furthermore, since the pulmonary pathology induced in vaccinated mice after RSV challenge is influenced by the route of vaccination, it may be possible to manipulate the APCs and the lymphoid environment to avoid vaccine-enhanced pathology.