INTRODUCTION
Emerging in 2002 in Guandong Province, China, severe acute respiratory syndrome (SARS) presented as an atypical pneumonia with an overall mortality rate of 10 to 12%, but exceeding 50% in aged (>60-year-old) populations (
3,
12,
36). The etiological agent was the novel SARS coronavirus (SARS-CoV), a zoonotic virus that likely emerged from bats and spread into civets and raccoon dogs either concurrent with or prior to the human epidemic (
8,
22,
62). While the epidemic strain was controlled by aggressive public health intervention strategies, the possibility of a reemergence is fueled by the presence of SARS-like CoV strains circulating in animal reservoirs (
22,
23,
35). Indeed, phylogenetic analysis of outbreak strains isolated during the late 2003/early 2004 epidemic suggest multiple independent emergences into the human population (
49,
62).
SARS-CoV is a cytoplasmically replicating, positive-polarity, single-stranded RNA (ssRNA) virus with three major membrane-bound structural proteins, spike (S), envelope (E), and membrane (M); several unique glycoproteins; and one structural protein within the virus core, the nucleocapsid (N) protein. Multiple candidate antiviral and immunomodulatory therapeutics have been developed in response to the epidemic, and vaccines would likely be a major tool in controlling any new SARS-CoV outbreak (
51). Key to the development of effective SARS vaccines appears to be the generation of neutralizing antibodies targeting the S glycoprotein, which provide complete protection upon passive transfer and are consistently associated with protection in multiple vaccine formulations (
15,
44,
52,
67). SARS vaccine strategies consist of varied formulations of inactivated (
24,
40), live attenuated (
33), recombinant subunit (
41), DNA (
28,
60), or subunit-vectored vaccines (
2,
11,
13,
48). Live attenuated vaccines with deletions in nonessential proteins show some efficacy in young mice, but low antibody titers preclude sterilizing immunity, and they remain untested in more vulnerable aged animals (
33). Vectored vaccines incorporating the spike glycoprotein alone show significant protection but are limited by strain specificity and immunosenescence (
48). Inactivated whole-virus vaccines have the advantages of relative ease of production in large quantities, stable expression of conformation-dependent antigenic epitopes, and the contribution of multiple viral immunogens. However, the disadvantages of inactivated formulations include the risk of vaccine preparations containing infectious virus, as well as the inclusion of antigenic determinants not associated with protection that may unpredictably skew the immune response (
27). With few exceptions, SARS vaccine formulations have not been tested against heterologous challenges in immunosenescent models of severe end stage lung disease (
48).
Effective SARS vaccines must meet several criteria, including (i) the ability to protect against heterologous viral variants that arise during independent emergence events, since many S-targeted antibodies have significantly reduced neutralization titers against heterologous spike glycoproteins (
11,
19,
44); (ii) the ability to elicit robust immune responses in elderly populations that are difficult to immunize and at increased risk for SARS-CoV-induced morbidity and mortality (
14,
29); and (iii) avoidance of adverse vaccine outcomes, such as the vaccine-induced immune pathology that has been demonstrated following vaccination with the SARS N protein (
11,
61). Whole inactivated SARS-CoV vaccines have demonstrated efficacy in young-animal models, generating high titers of neutralizing antibodies, yet most challenge studies have used a virus replication model devoid of clinical disease (
17,
46,
50). In humans, inactivated SARS-CoV vaccines have been shown to induce neutralizing antibodies in healthy young subjects in phase 1 clinical trials (
24,
28,
41). However, in neither humans nor animal models have inactivated vaccines been assessed for their ability to provide protection in aged populations or to protect against heterologous challenges. Given the severity of disease in aged populations and the possibility that emergent SARS viruses will be antigenically distinct from the 2002 epidemic strain, animal models that capture severe age-related disease and allow assessment of heterologous SARS challenges are essential for the preclinical validation of any vaccine or therapeutic candidate. The aged BALB/c mouse model reproduces age-related susceptibility to SARS-CoV disease similar to that noted in human infections, including increased levels of SARS-CoV replication, more severe clinical disease, and enhanced pulmonary histopathology (
42,
45,
59). When challenged with zoonotic and human chimeric SARS-CoV incorporating variant spike glycoproteins, the aged BALB/c mouse model reproduces severe lung damage associated with human disease, including diffuse alveolar damage, hyaline membrane formation, and death, thereby also providing a model for assessing vaccine-mediated protection against heterologous viruses (
45).
To test these hypotheses, we characterize the efficacy of an inactivated whole SARS-CoV vaccine in a highly lethal homologous and heterologous challenge model that recapitulates the age-related susceptibility and pathological findings seen in lethal human cases. The vaccine used was the CDC strain (AY714217) of SARS-CoV, doubly inactivated by formalin and UV irradiation, herein referred to as double-inactivated virus (DIV) (
50). The vaccine had initially been characterized in tissue culture and young mice, where it was shown to induce neutralizing antibodies and to provide protection from viral replication. Adjuvanting with alum had minimal effect on the serum neutralizing titers or protection in these young-mouse protection studies (
50). In this study, we chose to advance the protection and safety studies of DIV by assessing homologous and heterologous challenges in mice. We initially assess the vaccine's efficacy and potential for enhancement in a nonlethal animal model using icGD03-S. This synthetically derived virus incorporates the spike protein of a human strain isolated in 2004, providing a human virus challenge that is nonetheless divergent from the vaccine strain (
45). Extending this protection study to the more stringent test of a lethal challenge, we utilize a mouse-adapted virus, icMA15, which is lethal in both young and old BALB/c mice and is minimally different from the vaccine strain (
39). A chimeric virus incorporating the spike protein of a civet strain (HC/SZ/61/03) into the Urbani backbone provides a lethal heterologous and zoonotic challenge model (
45). These three viral challenge regimens, varied adjuvants, and an aged-mouse model help to accurately model potential challenges of vaccinating a human population against future emergences involving a SARS-CoV-like zoonotic virus. Our results demonstrate vaccine-induced enhancement of eosinophilia and inflammatory response following challenge, as well as failure to protect against heterologous challenge in an aged-animal model. This work highlights the challenge of vaccine design for zoonotic viruses, the need for developing broadly neutralizing therapeutics, and the particular difficulty of immunizing aged populations and offers new routes for understanding SARS-CoV pathogenesis.
MATERIALS AND METHODS
The generation and characterization of each of the recombinant infectious clones (icUrbani, icGD03-S, and icHC/SZ/61/03-S) have been described previously (
45,
63). Briefly, all recombinant icSARS-CoV strains were propagated on Vero E6 cells in Eagle's minimal essential medium (MEM) (Invitrogen, Carlsbad, CA) supplemented with 10% fetal calf serum (HyClone, Logan, UT), kanamycin (0.25 μg/ml), and gentamicin (0.05 μg/ml) at 37°C in a humidified CO
2 incubator. All work was performed in a biological safety cabinet in a biosafety level 3 (BSL3) laboratory containing redundant exhaust fans. Personnel were equipped with powered air-purifying respirators with high-efficiency particulate air and organic vapor filters (3M, St. Paul, MN), wore Tyvek suits (DuPont, Research Triangle Park, NC), and were double gloved.
Viruses and cells.
The icGD03-S (AY525636) (
47), icMA15 (FJ882957), and icHC/SZ/61/03-S (
45) strains of SARS-CoV were propagated on Vero E6 cells in Eagle's MEM supplemented with 10% fetal calf serum, kanamycin (0.25 μg/ml), and gentamicin (0.05 μg/ml) at 37°C in a humidified CO
2 incubator. For virus growth, cultures of Vero E6 cells were infected at an approximate multiplicity of infection (MOI) of 1 for 1 h, and the monolayer was washed twice with 2 ml of phosphate-buffered saline (PBS) and then overlaid with complete medium. At 30 h postinfection, the supernatant was clarified by centrifugation at 1,600 rpm for 10 min, aliquoted, and frozen at −70°C. Virus stocks were titrated by plaque assay.
Mouse vaccination and challenge.
Due to the poor availability of aged mice, two slightly divergent mouse strains were utilized during the course of this research. BALB/c mice (Harlan Laboratoriess, Indianapolis, IN) were challenged with lethal viruses (icMA15 and icHC/SZ/61/03-S), while the National Institute of Aging (NIA) provided BALB/cBy mice for nonlethal/epidemic strain challenges. Prior studies in our laboratory have shown conserved susceptibility phenotypes in these mouse strains following SARS-CoV challenge, with slightly increased morbidity and mortality in the NIA (BALB/cBy) strain. Therefore, in the nonlethal icGD03-S challenge, we expected slightly more morbidity than normally would have been predicted in Harlan mice.
Female BALB/cAnNHsd mice (young [6 to 8 weeks old] and aged [12 to 14 months old]; Harlan Laboratories, Indianapolis, IN) were separated into 4 groups of 12 young and 12 aged mice. The mice within each group were vaccinated by footpad injection with 20-μl volumes consisting of 0.2 μg of double-inactivated SARS-CoV vaccine, 0.2 μg of double-inactivated SARS-CoV vaccine with alum, 0.2 μg of inactivated influenza virus (iFlu), or 0.2 μg of inactivated influenza virus with alum. The mice were boosted with the same regimen 22 to 28 days later. Aged female BALB/cBy mice (12 to 14 months old; NIA) were vaccinated with PBS-, alum-, or VAP (VEE [Venezuelan equine encephalitis virus] replicon [VRP] adjuvanting particle)-adjuvanted iFlu (
n = 8, 10, and 9, respectively) or DIV (
n = 10, 9, and 10, respectively) immunogen. The vaccine formulations consisted of 0.2 μg of DIV or iFlu plus either PBS, 0.69 mg/ml alum, or 10
5 infectious units (IU) of VAP, a dose which had previously been demonstrated to provide protective immunity in young mice (
25,
54). These mice were then boosted with the same regimen 22 to 28 days later.
We collected blood from tail veins prior to challenge with icMA15, icHC/SZ/61/03, or icGD03-S on day 36 postvaccination. Mice were anesthetized with a ketamine (1.3 mg/mouse)-xylazine (0.38 mg/mouse) mixture administered intraperitoneally in a 50-μl volume. The mice were intranasally inoculated with 10
5 PFU of icMA15, icHC/SZ/61/03, or icGD03-S in 50-μl volumes and weighed daily. At 2 or 4 days postinfection, the mice were euthanized with isoflurane, and lung and serum samples were collected for analysis. For studies involving VRP vaccinations, 5-week-old female BALB/c mice were immunized with 10
5 IU of VRPs expressing SARS-CoV N, a bat coronavirus (BtCoV.279) N, or hemagglutinin (HA) in a 10-μl volume by footpad injections. Three weeks later, blood was collected by the tail nick method for enzyme-linked immunosorbent assay (ELISA), and the mice were boosted with 10
5 IU of the respective VRPs. Three weeks postboost, blood was collected by tail nick for assessing antibody responses. The mice were moved to a satellite facility under BSL3 conditions, acclimatized, and challenged with 10
5 PFU of recombinant MA15 (rMA15) icGDO3 virus (
48) by intranasal inoculation as described above.
All mice were housed under sterile conditions in individually ventilated Sealsafe cages using the SlimLine system (Tecniplast, Exton, PA). Experimental protocols were reviewed and approved by the Institutional Animal Care and Use Committee at the University of North Carolina, Chapel Hill, NC.
Plaque assay titration of virus from lungs.
One-quarter of each lung was taken for determination of the viral titer. Samples were weighed and homogenized for 60 s at 6,000 rpm in four equivalent volumes of PBS to generate a 20% solution. The solution was centrifuged at 13,000 rpm under aerosol containment in a table top centrifuge for 5 min, the clarified supernatant was serially diluted in PBS, and 200-μl volumes of the dilutions were placed onto monolayers of Vero E6 cells in six-well plates. Following 1 h of incubation at 37°C, the cells were overlaid with 0.8% agarose-containing medium. Two days later, the plates were stained with neutral red, and the plaques were counted.
Plaque reduction neutralization titer assays.
We heat inactivated mouse serum at 55°C for 30 min and then serially diluted it to 1:50, 1:100, 1:200, 1:400, and 1:800 in PBS to a volume of 125 μl. Next, we added 125 μl of PBS containing 125 PFU of icSARS-CoV to each serum dilution, incubated the virus-serum mixtures at 37°C for 30 min, added 200 μl of each mixture to confluent cultures of Vero E6 monolayers, and allowed them to incubate at 37°C for 1 hour. Following the 1-h infection, we covered each monolayer with 4 ml of 0.8% agarose melted in standard Vero E6 cell medium and resolved the plaques with neutral red staining 2 days later. Finally, we calculated the 50% plaque reduction neutralization titer (PRNT50) values, the serum dilutions at which plaque formation was reduced by 50% relative to that of virus not treated with serum.
Lung histopathology.
One-half of each lung was fixed in 4% paraformaldehyde (PFA) in PBS (pH 7.4) for at least 7 days, embedded in paraffin, sectioned to 5 μm, and stained with hemotoxylin and eosin (H&E). Sections were blindly evaluated by W. Funkhouser for the extent of tissue damage and characterization of inflammation.
Visual enumeration of eosinophils.
Lung tissues were prepared as described above and stained with H&E or Congo red (plus hematoxylin) (
30). For each slide, an initial assessment of gross lung pathology was followed by selection of a lung section and enumeration of eosinophils within the viewing field at ×400 magnification. Representative images were minimally and identically processed to enhance contrast in Adobe Photoshop CS4. For both H&E- and Congo red-stained slides, multiple 160-μm
2 sections proximate to airways were assessed, and the eosinophils counted were averaged per lung.
Quantitative real-time reverse-transcription (RT)-PCR.
One-quarter of a lung from each mouse was placed into RNAlater (Ambion) for 4 days at 4°C and then frozen at −70°C. Lung samples were transferred from RNAlater to TRIzol and homogenized for 60 s at 6,000 rpm, and RNA was extracted by chloroform/isopropanol precipitation. cDNA was prepared by standard protocols using random hexamers and SuperScript II Reverse Transcriptase (Invitrogen). Quantitative PCR was conducted on a Lightcycler 480II (Roche) using ABI TaqMan Gene Expression Assays specific for mouse GAPDH (glyceraldehyde-3-phosphate dehydrogenase) or mouse interleukin 4 (IL-4), IL-5, gamma interferon (IFN-γ), IL-13, CCL11 (eotaxin), or Cxcl1 Keratinocyte-derived chemokine (KC). Relative quantification was calculated as the log10 fold change (2ΔΔCT) relative to mock-vaccinated, mock-challenged controls.
Flow cytometry.
Mice vaccinated with PBS, DIV, or DIV plus alum were challenged with 10
5 PFU of icHC/SZ/61/03, weighed and monitored daily for morbidity, and euthanized 4 days postinfection by isoflurane inhalation. The lungs were perfused with 10 ml PBS by cardiac puncture, dissected, manually minced, and vigorously agitated for 2 h in digestion medium (RPMI, 10% fetal bovine serum [FBS], 15 mM HEPES, 1.7 mg/ml DNase 1 [Sigma], 2.5 mg/ml collagenase A [Roche], 1× streptomycin, 1× gentamicin). The lungs were then passed through a 75-μm cell strainer, resuspended in RPMI medium (RPMI, 10% FBS, 15 mM HEPES), and overlaid on a density gradient of iodixanol diluted to a density of 1.079 gm/cm
3 with RPMI 1640 containing 10% FBS (Optiprep, Sigma-Aldrich Co., St. Louis, MO). Following centrifugation, cells were collected from the interface and washed, and viable cells were counted with a Countess automated cell counter (Invitrogen). The cells were then incubated with the following panel of antibodies: allophycocyanin (APC) anti-leukocyte common antigen (LCA), PE-Cy7 anti-CD11b, and phycoerythrin (PE)-Cy5 anti-major histocompatibility complex (MHC) class II antigens, all from eBioscience (San Diego, CA); fluorescein isothiocyanate (FITC) anti-Gr-1 and PE anti-SiglecF, from BD-Pharmingen (San Diego, CA); and PE-Texas Red anti-CD11c (Molecular Probes Invitrogen, Carlsbad, CA). Following staining, the cells were washed with fluorescence-activated cell sorter (FACS) wash buffer (1× Hanks balanced salt solution [HBSS], 1% FBS) and fixed with 2% formalin, and flow cytometry was conducted on a CyAn ADP (Beckman-Coulter) with 300,000 live-cell events gathered per lung sample. Analysis was performed with Summit software (version 5.2; Beckman-Coulter). First, we gated on LCA
+ and CD11c
+ cells by plotting those parameters against forward scatter and gating on positive cells. Then, Gr-1 was plotted against SigLecF (see
Fig. 8A). SigLecF-high, Gr-1-intermediate cells were selected, and CD11b signal versus CD11c signal was plotted for cells in that region. CD11b
+, CD11c
− cells are classified as eosinophils, while alveolar macrophages are CD11c
+ (see
Fig. 8b). After gating on LCA
+ and CD11c
+ cells, Gr-1-positive cells are classified as neutrophils (see
Fig. 8a). Of the cells that remain after gating out SiglecF
+ cells and neutrophils, we classify MHC class II-negative, CD11b
+, and B220-negative cells as monocyte-derived dendritic cells (mDCs). The cell counts per lung were calculated as the product of the total viable lung cell population by the percentage of gated cells in live-cell events. We used a two-factor analysis of variance (ANOVA) to assess the statistical significance of age and vaccine on the overall number of cells. If the ANOVA determined a factor was significant,
post hoc analyses using Tukey's honestly significant differences (HSD) were used to further determine the effects of treatment on cell counts.
Enzyme-linked immunosorbent assay.
Antigen-specific IgG and IgG subisotype titers were determined by ELISA. Briefly, purified recombinant SARS N protein or S protein was coupled to high-binding 96-well ELISA plates (Greiner) in basic carbonate buffer (pH = 9.6). After washing with ELISA wash buffer (EWB) (PBS with 0.016% Tween 20), diluted serum was added to the wells in EWB with 10% blocking buffer (Sigma-Aldrich). After 2 h at 4°C, the plates were washed again, and horseradish peroxidase-conjugated goat anti-mouse IgG, IgG1, or IgG2a was added to the appropriate wells diluted in EWB plus blocking buffer. After another 2 h, chromogenic substrate (o-phenylenediamine in citrate buffer with added hydrogen peroxide) was added to each well. After 30 min, the reaction was stopped with the addition of 0.1 M sodium fluoride and read at 450 nm. A sigmoidal curve was fit to each set of optical density (OD) versus log10 serum dilution values using the curve-fitting software of the SigmaPlot graphics package (Systat Software, Inc.), and the inflection point (where the OD is one-half of the maximum value recorded for that isotype-antigen combination) is calculated and reported as the “half-max titer.”
DISCUSSION
The human coronaviruses HCoV 229E, HCoV OC43, and SARS-CoV, each likely originating from animal reservoirs, have demonstrated a high proclivity for coronaviruses to cross the species barrier, adapt, and colonize the human host. Though not currently circulating in humans, SARS-CoV, like other zoonotic viruses, remains a significant reemerging disease threat given its maintenance in animal reservoirs. The development of vaccines or therapeutics for SARS-CoV is complicated by several challenges: the presence of a large heterogeneous zoonotic reservoir of related strains, the resistance of highly susceptible aged populations to vaccination, and potential disease-enhancing complications of the vaccine formulations (
11,
48,
61). Though many experimental SARS vaccine formulations have been developed, whole inactivated virus vaccines have the advantage of large-scale production, presentation of multiple epitopes, and conformation stability (
2,
11,
13,
28,
33,
48,
60).
Aged populations are classically difficult to vaccinate and suffer increased disease pathologies following infection with a variety of respiratory viruses, including influenza virus, RSV, HCoV OC43, and SARS-CoV (
14). Until recently, most published assessments of SARS-CoV vaccine efficacy utilized models capable of assessing only viral replication or antibody induction, and they are routinely conducted in animal models that neglect important human disease presentations (
11,
17,
33,
40,
46,
50). While necessary, these assessments are incomplete, especially because more robust lethal-challenge models have been developed that recapitulate severe end stage lung disease pathologies and allow assessment of the potential complications of senescence (
42,
43,
45,
48,
59). As shown here, a vaccine that appears protective in young animals is much less protective, and potentially pathogenic, in an aged-animal model.
SARS-CoV emerged from a heterologous pool of closely related animal strains, suggesting that future outbreak emergences will likely involve strains with unique changes in the S glycoprotein. The viral strains used in this study represent a homologous lethal challenge virus (icMA15), as well as a nonlethal human heterologous virus (icGD03-S) and a lethal zoonotic virus (icHC/SZ/61/03-S), which allowed us to directly test whether DIV was capable of providing effective protection against heterologous viruses in both young and highly susceptible aged populations. Importantly, the DIV vaccine provided partial protection against both homologous and heterologous challenge in young animals, and this protective effect was enhanced by alum adjuvant. In contrast, even the adjuvanted DIV vaccine failed to protect against virus-induced disease and viral replication following homologous or heterologous challenge in aged animals. Perhaps most importantly, though DIV plus alum protected against viral replication, disease, and respiratory pathology following homologous viral challenge (icMA15) in young animals, which is consistent with prior reports (
17,
46,
50), both DIV alone and DIV plus alum failed to protect against respiratory pathology following homologous challenge in aged animals, and both vaccine formulations failed to protect against respiratory pathology following heterologous challenge in mice of any age (
Fig. 4 and
5). These results further demonstrate the difficulty in eliciting protective immune responses in highly susceptible elderly population and indicate that in the face of a reemergent SARS virus, likely antigenically heterologous from the 2002 outbreak strain, existing vaccine formulations are unlikely to provide protective immunity.
In addition to the general failure of the DIV or DIV-plus-alum vaccines to elicit protective immunity against heterologous SARS viruses or to provide protection even against homologous viral challenge in aged animals, both the DIV and DIV-plus-alum vaccine formulations result in significantly enhanced immune pathology within the lungs compared to control animals. Although adjuvanted DIV protected young animals from morbidity and mortality following lethal challenges, the heterologous virus, icHC/SZ/61/03-S, induced a lung pathology that was more severe in vaccinated than in unvaccinated mice (
Fig. 4). This increased pathology was not correlated with weight loss or mortality through day 4 postinfection (
Fig. 3), but the increased immune infiltrate indicates the vaccine is not fully protective against heterologous challenges. Further, in aged animals recalcitrant to immunization, insufficient protective immunity correlated with significantly increased immunopathology. Thus, evidence of enhanced disease subsequent to vaccination was evident in both heterologous challenge models and models of immune senescence.
In each of the experiments conducted here, immunization with the whole inactivated SARS vaccine induced increased inflammatory infiltrates and pulmonary eosinophilia upon subsequent challenge, demonstrating the potential for dangerous clinical complications. This is consistent with two prior studies of vaccine formulations incorporating SARS N, where N-specific immune responses resulted in enhanced eosinophilic immune pathology (
11,
61). This pathological signature is reminiscent of the two known human examples of vaccine-induced immunopathology, atypical measles and enhanced RSV. For both of these vaccine-induced immunopathologies, infection subsequent to vaccination resulted in failure to control viral replication, enhanced clinical disease, and a pathology characterized by increased complement deposition and inflammation, skewing toward Th2 responses, and eosinophilic influx (
38). The cytokine profiles of DIV-plus-alum-vaccinated and icHC/SZ/61/03-challenged mice showed increased levels of Th2 effector cytokines and eosinophil chemokines (IL-5, IL-13, and CCL11/eotaxin) compared to mock-vaccinated groups (
Fig. 8). In contrast, IFN-γ and IL-4 (Th1- and Th2-inducing cytokines) were unchanged at 2 and 4 days postinfection, likely because the peak mRNA levels for these inducing cytokines were earlier in the time course of infection.
As previous studies had indicated peak eosinophilia at 4 days postinfection, we assessed lung eosinophilia by both histopathology and flow cytometry at this time point, quantifying significant increases in the lungs of vaccinated mice following CoV challenge (
11). Eosinophilia was present independent of age and independent of the alum adjuvant, although adjuvant did increase the magnitude of the eosinophilic influx. DIV-induced eosinophilic influx was present even in the animals that were protected from morbidity and mortality. When this protection was absent, eosinophilic immunopathology was a dominant response more severe than the viral pathology seen in unvaccinated controls. The eosinophilia in clinically protected animals may thus serve as a marker for potentially pathogenic immune responses. While recent studies have argued that eosinophils are not the primary mediators of RSV vaccine-induced immune pathology, they may contribute to increased airway hyperresponsive conditions, including asthma, and the pathophysiology of viral infections (
6,
20).
Only eosinophils were consistently and significantly increased in response to vaccination with DIV. In contrast, neutrophils and monocytic DC populations were significantly affected by age, and monocytic DCs were decreased as a function of vaccination. The greater population of neutrophils in aged animals following challenge, independent of vaccination, suggests that neutrophils may contribute to the increased severity of SARS-CoV pathogenesis in the aged. A pathogenic role for neutrophils in infection has been demonstrated for other respiratory viruses, including influenza virus, suggesting conserved mechanisms of enhanced respiratory pathology in the aged (
26,
32).
SARS-CoV-targeted neutralizing antibodies are sufficient to provide complete immunity against lethal SARS challenges in multiple animal models and show evidence of controlling disease severity in human infections (
4,
5,
11,
15,
44,
64,
67). Spike-specific antibodies are both neutralizing and protective up to 1 year postvaccination in mice, while anti-nucleocapsid antibodies are neither neutralizing nor protective and, further, appear to be detrimental to the longevity of protective antibodies (
11,
61). This deleterious effect does not appear to be an antibody-dependent enhancement (ADE), since passive transfer was unable to replicate the immunopathology, though low posttransfer antibody titers preclude definitive exclusion of this potential mechanism (
11,
58).
While multiple major and minor SARS-CoV antigens are incorporated in DIV, the N protein is the most likely agent of eosinophilic immunopathology (
34). N is a strongly immunogenic protein (
7,
9,
37,
65), is the most abundant protein in infection (
34), and has been shown in prior studies to induce immunopathology when delivered in isolation (
66). This nucleocapsid-induced enhancement is not apparent in animals with appreciable levels of neutralizing antibodies, indicating that the induction of sufficiently neutralizing antibody responses can protect against SARS vaccine-induced immune pathology. However, the results presented here clearly demonstrate that in situations where individuals fail to mount protective anti-SARS responses, as is the case with heterologous viral challenge or in immune senescence, the DIV-vaccinated individuals are at significant risk for vaccine-induced immune pathology. With this in mind, it will also be important to determine the mechanisms by which N vaccination promotes immune pathology, with a key question being whether pathology is simply due to a nonprotective response against N or if N vaccination actively skews the host immune response to promote immune pathology. Given that N has been shown to modulate innate immunity and to act as an interferon antagonist, it is possible that N sufficiently alters the host immune response to induce a Th2 skew and subsequent inflammatory pathology (
21). Indeed, immunization with N appears to induce a Th2 skewing of the immune response regardless of the adjuvant or formulation, suggesting the nucleocapsid protein alone may well be the defining factor in CoV vaccine-induced enhancement (
11,
61; K. Long, D. Deming, R. Baric, and M. Heise, unpublished data). Therefore, additional studies are required to assess whether N′s innate immunomodulatory activity is linked to the N vaccine-induced immune pathology, and if so, whether this reflects a more general trait of viral interferon antagonists in modulating the downstream host adaptive immune response.
The major conclusion that can be drawn from these studies is that although DIV SARS vaccines do elicit protection under optimal conditions (homologous challenge in immunocompetent individuals), more stringent challenges reveal likely failures. If DIV vaccine approaches are to be used for SARS in the future, efforts must be made to improve the quality and magnitude of the vaccine-induced immune response while limiting the vaccine's capacity to induce immune pathology. The whole-virus vaccine used in this study was doubly inactivated by UV irradiation and formalin (
50). Formalin-inactivated vaccines are suggested to skew the immune response toward a Th2 response, producing higher levels of IL-4 and increasing the relative contribution of IgG2a isotypes (
31,
57). Previously, formalin inactivation leading to a disruption of fusion glycoproteins or addition of carbonyl groups had been blamed for the skewing of formalin-inactivated RSV (FI-RSV) immune responses (
31,
38). However, recent studies suggest that inactivation by alternate methods still results in a Th2 skew and immunopathology and that it is a failure of affinity maturation that results in nonprotective responses and subsequent antibody-mediated enhancement (
10,
57). Furthermore, the fact that the DIV vaccine did elicit neutralizing antibody responses and protection against homologous challenge in young animals suggests that the DIV-induced pathology did not simply represent a loss of antigenic epitopes. Therefore, we do not think the method of inactivation is necessarily responsible for immunopathology associated with DIV, but rather, that any SARS vaccines that include the nucleocapsid protein should be investigated for challenge-induced enhancement.
The results presented here reinforce the need to find methods to enhance the protective S-specific immune response while minimizing potentially pathological anti-N response. Our observation that immunization with BtCoV.279 N, which has high sequence similarity to SARS N, also induces eosinophila indicates that sequences intrinsic to N protein that are conserved across group 2b coronaviruses may drive this immune-mediated pathology. The amino acids responsible for this response need to be mapped. Assessment of N-induced immune pathology by sequentially divergent CoV N proteins may allow the design of chimeric SARS CoVs that could serve as vaccines devoid of immune pathology. In both young and aged mice, adjuvanting with alum increases the immunogenicity of the DIV, concordant with many earlier studies using this adjuvant (
50). In young mice, approximately one-half of the animals mounted neutralizing antibodies following DIV vaccination, and all achieved neutralizing titers when DIV was adjuvanted with alum (
Fig. 2). However, only half the aged mice were capable of mounting neutralizing antibody titers to DIV plus alum, and none mounted such responses without the adjuvant. Alum is one of the few adjuvants approved for use in human vaccine formulations and functions to stimulate Th2 immunity, a potentially confounding factor in the induction of immunopathology. We briefly assessed an alternative adjuvant, VAP, which is reported to stimulate Th1 immunity (
18,
53,
56,
57). In contrast to reports of success in young mice, the VAP-adjuvanted formulation in aged mice ablated rather than enhanced the protective response to DIV; subsequently challenged mice showed morbidity and mortality rates comparable to those in unvaccinated controls (data not shown) (
55,
56). While the mechanism of this ablation has not been defined, the VAP adjuvant likely functions through cross-presentation of exogenous antigens within antigen-presenting cells, a process impaired in age-associated immunosenescence (
16). VEE formulations incorporating the wild-type 3000 glycoprotein show better immunogenicity in aged animals, likely due to improved cross-presentation over the attenuated 3014 glycoprotein, suggesting that alternative VAP formulations may be more effective (
48). Therefore, if DIV approaches are to be considered for SARS or other respiratory coronaviruses, we feel that it will be important to rigorously test the vaccine and potential adjuvant in the aged-mouse model to assess both efficacy and potential immune pathology in the face of immune senescence and/or heterologous viral challenges.
Emergent zoonotic viruses present novel and shifting targets for vaccine design. The clear deficiency of the double-inactivated SARS vaccine against challenge models with divergent spike glycoproteins highlights the need for vaccines that induce broadly neutralizing immune responses. Further, impartial conservation of viral antigens cannot be considered a benefit to vaccine formulations for coronaviruses when select immunogens induce detrimental immune responses upon challenge. The N-induced pathogenic responses appear to be masked by S-targeted neutralizing antibodies but become dominant once the protective immunity wanes. In the case of SARS-like CoV, we cannot expect zoonotic variation to reduce specificity for N more readily than for S, as the S glycoproteins of multiple isolates show greater sequence variation and readily evolve over the course of an epidemic (
19,
22,
44). Further, the conservation of N across group 2b coronaviruses and the demonstrated conservation of N-induced immunopathology raise the possibility that challenge with nonepidemic coronavirus strains may induce eosinophilic immunopathology in vaccinated populations. Despite the difficulty of vaccine design for zoonotic viruses such as coronaviruses, paramyxoviruses, and filoviruses, a growing pool of sequence data for zoonotic isolates; the rapidity of sequencing and isolation in the case of outbreaks; synthetic-gene design; and the multiple vectored, inactivation, or antibody-generating platform technologies available ensure that vaccines can be readily formulated in case of novel outbreaks (
1). The challenge for researchers and clinicians is to validate these vaccines in strong animal models and to confirm and enhance vaccine efficacy in aged individuals. Identifying the vaccine components that induce protective immunity in aging individuals will be essential to protecting this vulnerable population. The data presented here indicate that SARS-CoV, coupled with a panel of heterologous zoonotic precursor viruses, represents a tractable model system to evaluate the molecular mechanisms governing immunosenescence and its impact on emerging virus pathogenesis and vaccine efficacy.