|Home | About | Journals | Submit | Contact Us | Français|
Enhanced disease is the term used to describe the aberrant Th2 skewed responses to naturally-acquired human respiratory syncytial virus (hRSV) infection observed in individuals vaccinated with formalin-inactivated viral antigens. Here we explore this paradigm with pneumonia virus of mice (PVM), a pathogen that faithfully reproduces features of severe hRSV infection in a rodent host. We demonstrate that PVM infection in mice vaccinated with formalin-inactivated antigens from PVM-infected cells (PVM Ags) yields Th2-skewed hypersensitivity, analogous to that observed in response to hRSV. Specifically, we detect elevated levels of IL-4, IL-5, IL-13, and eosinophils in bronchoalveolar lavage (BAL) fluid of PVM-infected mice that were vaccinated with PVM Ags, but not among mice vaccinated with formalin-inactivated antigens from uninfected cells (Ctrl Ags). Interestingly, infection in PVM Ag-vaccinated mice was associated with a ~10-fold reduction in lung virus titer and protection against weight loss when compared to infected mice vaccinated with Ctrl Ags, despite the absence of serum neutralizing antibodies. Given recent findings documenting a role for eosinophils in promoting clearance of hRSV in vivo, we explored the role of eosinophils in altering the pathogenesis of disease with eosinophil-deficient mice. We found that eosinophil deficiency had no impact on virus titer in PVM Ags-vaccinated mice, nor on weight loss or levels of CCL11 (eotaxin-1), interferon-γ, interleukin (IL)-5, or IL-13 in BAL fluid. However, levels of both IL-4 and CCL3 (macrophage inflammatory protein-1α) in BAL fluid were markedly diminished in PVM Ag-vaccinated, PVM-infected eosinophil-deficient mice when compared to wild type controls (246 words).
The development of safe and effective vaccines remains a standard and a benchmark for the control of respiratory virus infection. For infections with the unbiquitous human pathogen, respiratory syncytial virus (hRSV), this goal has not yet been reached. Along with efforts directed toward human clinical trials [1, 2], there is also significant interest in exploring alternative models for preclinical experimental research. The cotton rat and mouse models of hRSV challenge have been used extensively to study seroconversion and other immune responses to this virus, but these models face limitations with respect to issues of infectivity and virus replication in vivo [3, 4]. The bovine pathogen, bRSV, shares remarkable sequence similarity with hRSV, although RSV disease in cattle and humans are distinguished by markedly different clinical findings and tissue pathology [5, 6]. Live-attenuated vaccines against bovine respiratory syncytial virus (bRSV) have been available for more than 20 years, although these formulations do not provide full protection against naturally-acquired disease .
We and others have characterized pneumonia virus of mice (PVM) infection, as disease pathogenesis in inbred strains of mice reproduces features of the more severe forms of hRSV (reviewed in [8, 9]). Results of several recent studies suggest that PVM may also serve as a useful model for developing vaccine strategies, particularly those focused on acquisition of mucosal immunity to replicating virus pathogens. Specifically, Ellis and colleagues  demonstrated that mucosal inoculation with tissue-culture attenuated PVM resulted in protection against lethal challenge with the fully virulent strain via an interferon-γ independent mechanism. Two groups have recently developed recombinant PVM using reverse genetics methodology [11, 12] which may prove to be useful for experimental vaccine-related studies.
However, among the factors that are credited with having delayed progress toward an effective hRSV vaccine is the feature known as “enhanced disease”, a hypersensitivity phenomenon initially observed in children inoculated with a formalin-inactivated vaccine who then responded aberrantly to a subsequent natural infection (reviewed in [13, 14]). Rather than developing protective immunity, children inoculated with the formalin-inactivated vaccine developed pulmonary infiltrates containing mononuclear cells and eosinophils, both features characteristic of a skewed Th2 immune response. This response has been studied extensively, and has been modeled in BALB/c mice inoculated with formalin-inactivated RSV components; the characteristic lung findings have been associated with overexpression of Th2 cytokines as well as with immune complex formation [13 - 18]. Most recently, Moghaddam and colleagues  have provided biochemical evidence correlating this response with the presence of formalin-induced aldehydes on virus proteins, and Delgado and colleagues  have associated this response with lack of antibody affinity maturation and poor toll-like receptor (TLR) signaling.
A Th2-skewed response may be a specific characteristic of formalin-inactivated paramyxovirus antigen vaccines, as hypersensitivity responses have been observed in similar preparations of bRSV [21, 22], measles virus , and human metapneumovirus . Given the interest in using PVM as a model for vaccination strategy, it will be crucial to establish its similarities to other infection and vaccine models. As such, here we evaluate the responses of mice to formalin-fixed PVM antigens, prepared in manner analogous to the “lot 100” vaccine described in the initial clinical trials . In addition, we explored the responses to formalin-fixed PVM antigens in eosinophil-deficient ΔdblGATA and TgPHIL mouse models [26, 27]. Interestingly, despite substantial focus on eosinophil recruitment, it is not at all clear whether eosinophils actually promote systemic disease in response to formalin-inactivated vaccine challenge [28, 29]. Given recent findings of Phipps and colleagues  who documented the role of eosinophils in augmenting virion clearance in the hRSV challenge model, we have focused on determining whether eosinophils serve to reduce virus titer and thereby promote host defense against PVM infection in response to this challenge.
Wild type BALB/c and C57BL/6 mice were obtained from Taconic Laboratories (Rockville, MD). Eosinophil-deficient ΔdblGATA mice  backcrossed on the BALB/c background were originally obtained from Dr. Craig Gerard and Dr. Alison Humbles. Eosinophil-deficient TgPHIL mice  were originally obtained from Dr. James Lee and Dr. Nancy Lee. All animal studies are carried out under the review of National Institutes of Allergy and Infectious Diseases Animal Study Protocol LAD8E.
The fully pathogenic pneumonia virus of mice (PVM) strain J3666 was originally obtained from Dr. Andrew Easton (University of Warwick, Coventry, UK) from virus stocks originating at the Rockefeller University. PVM strain J3666 has been maintained via continuous passage in vivo. Tissue-culture attenuated PVM strain 15 was obtained from American Type Culture Collection, Manassas, VA. Sequence similarities and differences between these two strains and the second variant strain 15 are as described . PVM strain 15 was used to prepare formalin-fixed vaccine antigen in a manner similar to that described for the RSV “lot 100” vaccine [25, 32, 33]. In previous studies, mice inoculated with PVM strain 15 that recovered from acute infection, seroconverted to PVM (SMART-M12, Biotech Trading Partners, El Cerrito, CA), and were completely protected from subsequent challenge with pathogenic PVM strain J3666, indicating conservation of crucial protective epitopes. To prepare vaccine antigens (PVM Ags), cells of the mouse monocyte-macrophage RAW 264.7 cell line (ATCC), which permits moderate rates of replication of PVM at the permissive temperature of 32°C [34, Supplemental Figure 1] were grown to 50% confluence in Iscove's Modified Dulbecco's medium (IMDM) with 10% heat-inactivated fetal calf serum, 2 mM glutamine and penicillin/streptomycin and inoculated with PVM strain 15 at an ~MOI of 0.1. At 96 hours after inoculation, the RAW 264.7 cells were scraped into the growth medium and sonicated to release cell-associated virions. Supernatants were clarified by centrifugation, and formalin was added (1:4000) to the clarified supernatants which were then incubated at 37°C for 48 hrs. Precipitates were collected by high speed centrifugation, washed twice with cold phosphate buffered saline and resuspended in phosphate buffered saline at 100 μg/mL (as determined by Bradford assay) prior to addition of alum (Pierce Biotechnology, Rockford, IL). Control vaccine antigens (Ctrl Ags) were prepared precisely as above from parallel cultures of uninfected RAW 264.7 cells.
Sera from age and gender-matched mice were evaluated for seronegativity to PVM (SMART M-12) prior to initiating the experimental protocols [Figure 1A]. Ctrl Ags or PVM Ags were inoculated by sterile subcutaneous injection (100 μl / mouse) on days 0 and 14. On day 21, vaccinated mice were weighed, bled for sera to be used for detection of neutralizing antibodies to PVM, and were then challenged with 30 pfu  pathogenic PVM strain J3666 in an 80 μL volume, with an identical inoculum of heat-inactivated virus, or diluent control. Mice were weighed on days 25 and 26 (days 4 and 5 after challenge) and euthanized by cervical translocation under mild anaesthesia on day 26, which was day 5 after infectious challenge. On day 5, differential physiologic responses (weight loss) are most pronounced [Tables I and II]. PVM-infected, but otherwise unmanipulated mice [35, 36] and PVM-infected mice vaccinated with Ctrl Ags undergo rapid onset mortality shortly thereafter.
Blood was taken from mice prior to vaccination on day 0 (prior to vaccination) and at day 21 (after the vaccinations on days 0 and 14, prior to infectious challenge) and serum separated by standard methods. To determine whether or not neutralizing antibodies developed in response to PVM Ags vaccination, paired pre- and post-immune sera from individual mice were diluted 1:10 in phosphate buffered saline (40 μL serum and 360 μL PBS). Virus (50 μL undiluted PVM strain J3666) was added and samples were rotated end over end for 2 hrs at 4°C. 120 μL of the virus plus serum was added to RAW 264.7 cells growing in culture at 50% confluence in a 6 well plate (triplicate samples). Cultures were harvested 4 days after inoculation, and triton-lysates and RNA were prepared. Virus replication was detected by probing Western blots containing cell lysates with a 1:300 dilution of rabbit polyclonal anti-PVM N protein antibody, followed by a 1:1000 dilution of alkaline-phosphatase conjugated goat-anti-rabbit IgG and standard developing reagents (BioRad). Parallel blots were probed with a 1:300 dilution of chicken anti-actin following by a 1:1000 dilution of alkaline phosphatase conjugated rabbit anti-chicken IgY (US Biologicals) and developing reagents. Virus replication was also detected in a more quantitative fashion by quantitative RT-PCR using the dual standard curve method as described below. Negative control was sera from a normal, unmanipulated mouse; positive control was convalescent sera from C57BL/6 mice challenged with a sublethal inoculum of PVM strain 15 .
For evaluation of histopathology, lung tissue was inflated slightly by transtracheal instillation of 0.2 to 0.3 ml cold phosphate buffered 10% formalin. Lungs were removed and fixed in cold phosphate buffered formalin prior to paraffin embedding. Hematoxylin and eosin-stained fixed tissue sections were prepared by Histoserv, Inc., Germantown, MD. Eosinophils were identified in formalin-fixed lung tissue sections via serial dilution of polyclonal rabbit anti-mouse eosinophil major basic protein antisera (generous gift of Drs. James and Nancy Lee, Mayo Scottsdale) followed by goat anti-rabbit peroxidase staining (performed by Histoserv, Inc., Germantown, MD).
RNA was isolated from lung tissue that was isolated from mice and immersed in RNAlater (Ambion, Inc., Austin, TX) and stored as per manufacturer's instructions. RNA was isolated using the RNAzol Bee reagent (Friendswood, TX)  and virus titer was determined by quantitative RT-PCR with dual standard curve. This assay has been used to measure absolute copies of virus genome (PVMSH/106 GAPDH) generated per unit time with appropriate replication kinetics both in vivo, in mouse lung tissue, and in vitro, in the mouse macrophage RAW 264.7 cell line [Supplemental Figure 1]. In specific detail, RNA isolated as described was treated with DNase I to remove genomic contaminants. Reverse transcription to cDNA was performed using the First strand cDNA synthesis kit (Roche, catalog number 1 483 188), using random primers, and including a no reverse transcriptase control. The quantitative PCR reactions were run in triplicate, with the ABI 2X TaqMan reagent, primer-probe mixes, and cDNA or plasmid standard in a 25 uL final volume. Thermal cycling parameters for the ABI 7500, absolute quantitation program include 50°C for 2 minutes (UNG incubation), 95°C for 10 minutes (AmpliTaq Gold activation) and 40 amplifcation cycles alternating 95°C for 15 seconds and 60°C for 1 minute. Primer-probe mixes include GAPDH-Vic (ABI catalog number 4308313) and PVMSH-Fam (custom design, primer 1, 5' GCC GTC ATC AAC ACA GTG TGT 3'; primer 2, 5' GCC TGA TGT GGC AGT GCT T 3'; probe 6FAM-CGC TGA TAA TGG CCT GCA GCA-TAMRA). GAPDH standard curve includes serial dilutions of a mouse GAPDH nucleotide sequence (Ambion DECA probe template, #7330) to 109, 108, 107, 106, 105 molecules per reaction. PVMSH standard curve includes serial dilutions of the full length PVM SH open reading frame (GenBank (http://ncbi.nlm.nih.gov/sites/entrez) accession number AY573815) in pBacPAK8 to 109, 108, 107, 106, 105 molecules per reaction. Experimental triplicate datapoints are interpolated to linear standard curves over the concentration ranges indicated.
Bronchoalveolar lavage (BAL) fluid was collected from vaccinated and challenged mice by instillation and withdrawal of 0.8 ml cold phosphate buffered saline with 0.5 mM bovine serum albumin. Cytospin preparations were stained with modified Giemsa (Dade Behring, Marburg, Germany) and the leukocyte differential determined by visual inspection of 10 independent fields from 6 – 8 mice per group (> 200 cells / mouse). Concentrations of various proinflammatory mediators (eotaxin, interferon-γ, macrophage inflammatory protein 1-α, IL-4, IL-5 and IL-13) were determined by ELISA (R&D Systems, Minneapolis, MN) as per manufacturer's instructions.
Each datapoint was determined from duplicate or triplicate trials of samples obtained from 4 – 10 mice as indicated in the text. Each independent experiment was repeated two or three times. Statistically significant differences were determined via Student's t-test or Wilcoxon rank sum test, with p values obtained as indicated.
The experimental time line is shown in Figure 1A. Lung tissue sections from mice vaccinated on days 0 and 14 with formalin-inactivated antigens prepared from uninfected cells (Ctrl Ags) and from PVM-infected cells (PVM Ags) prior to challenge on day 21 with actively-replicating PVM are shown in Figure 1B – 1E. The histopathology observed in response to vaccination with Ctrl Ags followed by PVM infection is similar to that observed in PVM infection in naïve, otherwise unmanipulated mice [Supplemental Figure 2]; at day 26 we observed profound inflammation throughout the lung parenchyma [Figure 1B], with neutrophils predominating at this time point in both lung tissue and in BAL fluid [35; Figure 1C]. In contrast, the inflammatory infiltrates in lung tissue from mice vaccinated with PVM Ags exhibit a more diffuse, patchy pattern [Figure 1D] with denser inflammatory infiltrates in the peribronchiolar and perivascular regions [Figure 1E]. Profound eosinophilia was observed in the BAL fluid [Figure 1F], and within the lung parenchyma [Figure 2].
The detailed pattern of leukocyte recruitment is shown in Figure 3A. Leukocyte recruitment in PVM-infected mice vaccinated with Ctrl Ags is similar to what we have observed previously in PVM infected, but otherwise naïve mice . Specifically, by day 5 of infection (day 26 of the protocol, see Figure 1A), eosinophils are present (2.7 ± 1.5%), but neutrophils predominate (93 ± 1.8 %). In contrast, 66 ± 4.0 % of the leukocytes in BAL fluid from PVM-infected mice that were vaccinated with PVM Ag were eosinophils. The BAL eosinophilia observed in the PVM Ag-vaccinated, PVM-infected mice was accompanied by elevated levels of the Th2 cytokine, interleukin-5 (IL-5) [Figure 3B] similar to what has been reported in mouse models of aberrant hypersensitivity to hRSV [16, 37]. Minimal IL-5 (below detectable limits of the assay) was detected in response to PVM infection in mice inoculated with Ctrl Ags. Other Th2 cytokines (IL-4, IL-13) detected in response infection in PVM Ags and Ctrl Ags-vaccinated mice are described in sections to follow.
We measured weights of all mice on days 21, 25, and 26 of the experimental protocol (see Figure 1A), which are days 0, 4, and 5 after inoculation of 30 pfu replication-competent PVM, 30 pfu-equivalents of heat-inactivated PVM, or diluent alone. As shown in Table I, we observed no statistically significant weight loss over the 5-day examination period (days 21 – 26) in response to diluent or in response to a single-dose of heat-inactivated PVM regardless of the vaccination antigen (Ctrl Ags or PVM Ags). We did observe significant weight loss in response to actively-replicating PVM in both Ctrl Ags- and PVM Ags-vaccinated mice. Interestingly, PVM-infected mice vaccinated with PVM Ags had lost only 6 ± 4% of their initial body weight at day 5, compared to infected mice vaccinated with Ctrl Ags, which had lost 12 ± 3% (p < 0.002) of their initial body weight at this time point.
In association with protection against weight loss, PVM-infected mice that were vaccinated with PVM Ags exhibit lower lung virus titers than those that were vaccinated with Ctrl Ags (*p < 0.05, Figure 4A). Likewise, we detected CCL3 (MIP-1α), a chemokine produced locally by respiratory epithelial cells in response to virus infection and a biomarker of disease severity [38 - 40], in all Ag-inoculated and PVM infected mice, but in substantially diminished quantities in BAL fluid of mice that were inoculated with PVM Ags [Figure 4B].
Taken together, thus far, our results indicate that vaccination with PVM Ags elicits the characteristic hypersensitivity response, which includes eosinophil recruitment to the lung tissue in response to active virus infection. Interestingly, vaccination with PVM Ags likewise elicits an incomplete but clearly significant level of protection against subsequently acquired acute respiratory virus infection.
All mice used here were documented seronegative for PVM prior to the start of the experiments (data not shown). Sera obtained prior to vaccination (before day 0) and paired sera obtained on day 21 (after antigen vaccination, but prior to infectious challenge) were evaluated for their ability to neutralize PVM in the tissue culture assay described in the Methods. No neutralizing activity was detected among paired individual antisera at low dilution (1:10), which were evaluated qualitatively by Western blotting [Figure 5A]. Similarly, no neutralizing activity was observed when sera were tested, again at low (1:10) dilution, and virus genome equivalents were evaluated by quantitative RT-PCR [Figure 5B].
Eosinophils have been shown to reduce the infectivity of the hRSV for target cells in vitro [41, 42]. Most recently, Phipps and colleagues  have shown that hRSV clearance in a mouse model is augmented in the presence of pulmonary eosinophilia. As the antiviral response elicited by PVM Ags is not dependent on the presence of serum neutralizing antibodies, we proceeded to examine the role of the eosinophils in promoting these protective responses.
The ΔdblGATA (BALB/c background) and TgPHIL mice (C57BL/6 background) are devoid of eosinophils at baseline and remain so in response to profound Th2 stimulation [26, 43, 44]. The ΔdblGATA mice have undergone a 21 nucleotide deletion of a palindromic GATA-binding enhancer site in the hematopoietic promoter of the gene encoding GATA-1. This deletion results in a unique deficiency of the eosinophil lineage with sparing of all other hematopoietic lineages . TgPHIL mice express a diphtheria toxin A transgene under the control of the eosinophil-specific eosinophil-peroxidase promoter, thereby resulting in lineage-specific cytosuicide . Eosinophils can be identified in the BAL fluid of PVM Ags vaccinated, PVM infected C57BL/6 wild type mice, but, as anticipated no eosinophils are detected in identically-treated eosinophil-deficient TgPHIL or ΔdblGATA mice [Figure 6A].
As originally shown in Figure 4A, virus titers among PVM-infected mice that had been inoculated with PVM Ags are diminished when compared to those inoculated with Ctrl Ags, although the difference is not dramatic. Interestingly, we find that eosinophil-deficiency has no impact on virus titer in this experimental setting. Virus titer in lung tissue of PVM Ags-vaccinated, PVM infected eosinophil-deficient mice (both ΔdblGATA and TgPHIL) are indistinguishable from those determined for their respective wild type; diminished virus titers are detected among all mice vaccinated with PVM Ags, compared to those vaccinated with Ctrl Ags, regardless of the presence or absence of eosinophils [Figure 6B and and6C6C].
As originally shown in Table I, we observe weight loss in response to PVM infection in both Ctrl Ags and PVM Ags-vaccinated mice, with a small but statistically significant degree of protection observed when comparing those inoculated with PVM Ags to those vaccinated with Ctrl Ags (p < 0.03, Table II). Similar findings were observed among the Ctrl Ags and PVM Ags-vaccinated and PVM infected ΔdblGATA mice; eosinophil ablation had no impact on the degree of protection observed.
A prominent Th2 response, including cytokines IL-4, IL-5 and IL-13, was detected in BAL fluid of all mice that were inoculated with PVM Ags prior to PVM infection. Eosinophil ablation had no impact on detection of IL-5 or IL-13 [Figure 7A and and7B].7B]. In contrast, IL-4 detection in the eosinophil-deficient ΔdblGATA mice was markedly diminished; IL-4 levels in PVM Ags-vaccinated, PVM-infected ΔdblGATA mice were not significantly higher than those detected in response to Ctrl Ags vaccination [Figure 7C]. This finding is similar (although interestingly, not quite identical) to that reported by Jacobsen and colleagues  in their study of eosinophil-mediated recruitment of Th2 cells in response to allergen sensitization and challenge in vivo.
Finally, we compared the biochemical inflammatory responses among the Ag-vaccinated, PVM-infected wild type and eosinophil-deficient ΔdblGATA mice. Both wild type and ΔdblGATA mice respond to PVM infection by producing a variety of proinflammatory mediators. Among the mediators of interest, CCL11 (eotaxin-1) is produced by respiratory epithelial cells in response to virus infection and is a unique mediator of eosinophil chemoattraction , CCL3 (MIP-1α) recruits both neutrophils and eosinophils to the airways, and, as noted earlier, is a prominent biomarker of hRSV disease severity [39, 40]; and interferon-gamma, a cytokine with pleiotropic immunomodulatory properties , recently shown to exert hierarchical control over the activities of CCL3 in the setting of PVM infection in vivo . Here, we find that BAL levels of CCL11, interferon-gamma, and CCL3 were markedly diminished in PVM-infected mice that had been vaccinated with PVM Ags compared to those vaccinated with Ctrl Ags [Figure 8]. Eosinophil deficiency had no impact on levels of CCL11 or interferon-γ detected in BAL fluid. In contrast, we observed a marked reduction in the level of CCL3 detected in BAL fluid of PVM-infected eosinophil-deficient ΔdblGATA mice, both the Ctrl Ags-vaccinated as well as those responding to PVM Ags vaccination. CCL3 is produced primarily by respiratory epithelial cells in response to PVM infection; the role of eosinophils in promoting virus-induced expression of CCL3 is completely unexplored.
In this work we characterize the pathophysiologic responses of mice vaccinated with formalin-inactivated PVM antigens (PVM Ags), and demonstrate that that are similar to those reported for formalin-inactivated hRSV in the human vaccine trial and in exploratory rodent models. Specifically, we detect eosinophil-enriched inflammatory infiltrates in lung tissue and BAL fluid associated with augmented levels of BAL Th2 cytokines. Interestingly, hypersensitivity responses have been observed in response to formalin-inactivated antigens derived from a variety of paramyxovirus pathogens, including measles virus, bRSV, and most recently, human metapneumovirus [21 - 24]. The precise molecular mechanisms underlying these responses remain uncertain. Several groups originally reproduced the hRSV response in mice by overexpression of G (attachment) protein alone in the absence of formalin [49 - 51]. However, recent analyses suggested that, although there were some similarities, specifically pulmonary eosinophilia in association with Th2 cytokines, the responses to G protein overexpression occurred via distinct cellular mechanisms (reviewed in [13, 52]). Among these, G-protein dependent eosinophilia required Vβ14+ T cells, while responses to formalin-inactivated hRSV antigens did not , and likewise, formalin-inactivated hRSV antigens devoid of G protein were fully capable of eliciting hypersensitivity responses in BALB/c mice . From another perspective, the recent evidence suggesting that formalin-induced hRSV protein oxidation plays a role in promoting aberrant responses  is quite intriguing, and it will be interesting to determine, given that formalin-inactivation results in more or less indiscriminate protein oxidation, why only a specific formalin-oxidized paramyxovirus protein (or proteins) induces Th2 skewing in vivo. In fact, it will be interesting to determine whether or not Th2 skewing is in fact limited to formalin-oxidized paramyxovirus proteins only. The PVM model may be useful in experiments designed to address these issues.
Regarding specificity, Piedra and colleagues  and Boelen and colleagues  both reported independently that formalin-inactivated non-virus antigens elicited immunopathology in vaccinated cotton rats and BALB/c mice, respectively, an issue that is of concern regarding validity of this model. We did not observe hypersensitivity responses to formalin-inactivated antigens from uninfected cells in our experiments, but this may be attributed to the experimental design more than specificity of the response. Among the differences between the models, the PVM strain J3666 used for infection is passaged in vivo, and is thus devoid of tissue culture antigens that could readily cross-react with those used for preparing PVM for inoculation; in vivo passage of the infectious pathogen is physiologically relevant to natural infection, and not necessarily the case in all reported experimental trials. Furthermore, in contrast to the hRSV challenge model in mice and cotton rats, in which high titer, highly-concentrated viral inocula grown in tissue culture are cleared rapidly, and generate limited physiologic responses, the PVM infectious inoculating dose is minimal and dilute, and replicates to very high virus titer, sufficient for isolation and re-infection of a naïve host, in mouse lung tissue in vivo [3, 4 Supplemental Figure 1]. Thus, the only antigens ultimately present in high concentration come from replicated virus, similar to what one would expect to occur during a natural infection in a vaccinated host. As such, we do not truly understand the underlying specificity of this response. There actually may not be anything extraordinary or unique about formalin-inactivated paramyxovirus proteins, other than the fact that, in a natural setting, only replication of the respiratory virus in situ ultimately results in an antigen challenge of sufficient magnitude to initiate a hypersensitivity response to the formalin-inactivated vaccine components.
In addition to establishing that “enhanced disease” or hypersensitivity response is a characteristic of PVM Ags, we proceeded to use this model to explore the potential of pulmonary eosinophils elicited in this fashion to promote pathology or, as suggested in several recent publications [30, 41, 57, 58], to promote antiviral host defense. Of note, despite a remarkable amount of effort devoted to understanding the mechanisms via which eosinophils are recruited to the lung in response to formalin-inactivated virus antigens, it is not at all clear that eosinophils are in fact contributing to the physiologic dysfunction, specifically, the respiratory dysfunction and systemic disease [28, 29], and recent results relating eosinophil recruitment, airway dysfunction, and systemic disease in response to hRSV G protein vaccination, a phenomenon recently shown to occur via mechanisms that are distinct from those contributing to formalin-inactivation, suggest that eosinophils may be more or less “innocent bystanders” in the former process . In our earlier studies, we showed that isolated human eosinophils could reduce the infectivity of hRSV for target cells via the actions of their secretory mediators [41, 42]; more recently, Phipps and colleagues  showed augmented hRSV clearance in IL-5 transgenic hypereosinophilic mice and induction of antiviral proteins in mouse eosinophils upon interaction with virions in vivo.
As a first intriguing finding, we found that mice vaccinated with PVM Ags were protected against virus challenge; despite the absence of neutralizing antibodies, we observed reduced total genome equivalents and diminished weight loss. These results suggest that the presence of eosinophils in the lung tissue and airways are associated with protection against virus infection. We repeated the vaccination and challenge protocol in eosinophil-deficient mice, and although this yielded several important biochemical alterations (to be discussed below), we found that eosinophil deficiency resulted in no change in virus titer and no change in the overall clinical picture.
Although we are still without explanation as to what is promoting protection against virus challenge in response to formalin-inactivated PVM antigens, among the larger questions vis à vis this work is why the responses of eosinophils differ here from what was observed in the hRSV challenge model , and why prominent virus clearance observed in the latter setting was not evident here. Among several possibilities, the most apparent relates to the nature of the pathogens and their relationship to the mouse model as a whole and with mouse eosinophils specifically. Overall, PVM is more virulent in mice than is hRSV, as PVM not only replicates in mouse lung epithelial tissue , but we have recently shown that PVM infects and replicates directly within mouse eosinophils [60, 61]. We have not yet examined the how PVM infection might alter the antiviral responses of eosinophils, but it is possible that infection might disable these cells, reduce their capacity to respond productively and to clear virus infection, all elements which may serve to enhance the virulence of PVM in vivo.
As noted above, eosinophil ablation is associated with specific biochemical alterations in this model. While Th2 cytokine responses are detected in both wild type and ΔdblGATA mice inoculated with PVM Ags, production of IL-4 is diminished specifically among the ΔdblGATA mice. This finding is consistent with those of Jacobsen and colleagues  who reported diminished recruitment of CD4+ T cells and IL-13 in BAL fluid in eosinophil-ablated TgPHIL mice subjected to ovalbumin sensitization and challenge, in a study which led to the conclusion that pulmonary eosinophils are crucial for the recruitment of Th2 lymphocytes in this traditional model of allergic lung disease. Here, IL-5 and IL-13 production persists among the ΔdblGATA mice, possibly derived from activated mast cells and basophils, NK and NKT cells, respectively [62 - 65], although the role of these cells in response to formalin-inactivated antigens has not been formally explored. Interestingly, dimished Th2 cytokine production in the absence of eosinophils is not a universal finding; we detected elevated levels of IL-5 in both the ΔdblGATA and TgPHIL eosinophil-deficient mouse strains during the Th2 phase of acute infection with the helminthic parasite, Schistosoma mansoni .
In summary, we have clearly documented that a Th2 skewed hypersensitivity response, otherwise known as “enhanced disease” is elicited by PVM infection following inoculation with formalin-inactivated PVM antigens, analogous to what is observed in response to formalin-inactivated hRSV, and in response to other formalin-inactivated paramyxovirus vaccines. While formalin-inactivated hRSV vaccine formulations are certainly not currently under consideration for human use, the developers of novel anti-RSV vaccine strategies continue to make a significant effort to demonstrate the absence of hypersensitivity reactions due to the serious nature of this problem [66 - 68]. Given the recent interest in PVM as a model for exploring mechanisms of disease and vaccine strategies, we have established an important point documenting similar immunopathologic responses to both human and mouse pathogens. Furthermore, we have also explored the role of eosinophils in this setting and found that they did not promote PVM clearance, despite clear evidence for eosinophil-mediated clearance of hRSV. The nature of the interactions between PVM and eosinophils remain a subject for future investigation.
The authors thank Drs. James and Nancy Lee for the generous gift of the TgPHIL males for our breeding colony and anti-mouse MBP antibody, and Drs. Craig Gerard and Alison Humbles for providing the ΔdblGATA males for breeding. We also thank members of the Eosinophil Biology Section and Molecular Signal Transduction Section, LAD, NIAID, for helpful discussions that contributed to this work.
This work was supported by NIAID Division of Intramural Research funding to HFR.