Prior Immunization with Severe Acute Respiratory Syndrome (SARS)-Associated Coronavirus (SARS-CoV) Nucleocapsid Protein Causes Severe Pneumonia in Mice Infected with SARS-CoV¹

From November 2002 to July 2003, an outbreak of severe acute respiratory syndrome (SARS),⁴ which originated in China, spread worldwide, resulting in 8098 cases with 774 deaths (http://www.who.int/csr/sars/country/en/index.html). Patients with SARS usually develop high fever followed by severe clinical symptoms, which include acute respiratory distress syndrome with diffuse alveolar damage, and ultimately death. A novel type of coronavirus (CoV), termed SARS-associated CoV (SARS-CoV), was identified as the etiologic agent of SARS (1, 2, 3). The genome of SARS-CoV is a single strand of positive-sense RNA of ∼30 kb in length with 14 putative open reading frames, which encode nonstructural replicase polyproteins and several structural proteins, including spike (S), envelope (E), membrane (M), and nucleocapsid (N) proteins (4). The S protein of SARS-CoV, like the S proteins of other CoVs, plays an important role in the first step of viral infection by binding to a host cell receptor. Angiotensin-converting enzyme 2 was identified as the host receptor for SARS-CoV (5). Angiotensin-converting enzyme 2 is abundantly expressed in the epithelia of the lung and small intestine and may mediate SARS-CoV entry in humans (6). Although intensive investigations rapidly unraveled the sequence of the SARS-CoV genome and its receptor in humans, the precise molecular mechanism underlying the development of SARS is not fully understood.

The possible roles of host anti-SARS-CoV immune responses have been suggested in severe clinical cases. The uncontrolled release of immune mediators, called a “cytokine storm,” has been implicated in the pathogenesis of SARS. However, the cytokine profiles of SARS patient sera do not correlate with the severity of pneumonia because of their diversity. For example, Jones et al. (7) have reported a decreased number of IL-2-, IL-4-, IL-10-, and IL-12-producing cells in SARS-CoV-infected patients. In contrast, Wong et al. (8) have demonstrated increased production of IFN-γ, IL-1, IL-6, and IL-12 p70, but not of IL-2, IL-4, IL-10 or TNF-α, which is consistent with a Th1 response. The data from these adult patients with SARS show no clear trend toward either a Th1 or Th2 bias. These results might be related to patient anamnesis. Therefore, the development of animal models for SARS is needed to understand the pathogenesis of SARS. Non-human primates, mice, ferrets, and hamsters have been found to support the replication of SARS-CoV (9, 10, 11, 12, 13, 14). However, an animal model that mimics the clinical symptoms and pathology observed in SARS patients has not been reported to date. Recently, Roberts et al. (15) reported that aged BALB/c mice (older than 12 mo) exhibited high and prolonged levels of viral replication, signs of clinical symptoms, and histopathologic changes in the lung. Aged BALB/c mice represent a conventional animal model that mimics the findings in elderly SARS patients, many of whom exhibit severe disease requiring intensive care and ventilation support, as well as increased mortality.

In the present study, we investigated the pulmonary immune responses and pathologies of intranasally SARS-CoV-infected BALB/c mice older than 6 mo of age that were previously immunized with SARS-CoV structural proteins using vaccinia virus (VV) vectors, by measuring various cytokine mRNAs and histopathologies of the lungs.

RK13 cells (CCL-37) from the American Type Culture Collection (ATCC) and Vero E6 cells (CRL-1586) from ATCC were cultured in MEM (Nissui Pharmaceutical) that contained 5% FCS. To generate recombinant VV LC16m8, which expresses the structural proteins of SARS-CoV, primary rabbit kidney cell cultures were prepared by overnight digestion with 100 PU/ml dispase (Sanko Jun-yaku) of kidneys extirpated from 7-day-old inbred JW rabbits (Kitayama Labs). The cells were grown in T175 flasks in lactalbumin medium with Hank’s salts (LH) that contained 5% FCS, 100 U/ml penicillin, and 100 μg/ml streptomycin. When the cell confluency was ∼50%, the culture medium was replaced with lactalbumin medium with Eagle’s salts (LE) that contained 5% FCS, 100 U/ml penicillin, and 100 μg/ml streptomycin. SARS-CoV Vietnam/NB-04/2003 strain, which was isolated from the throat wash fluid of one patient (16), was provided by Dr. M. Quynh Le. VVs LC16m8 (m8) and LC16mO (mO) were provided by the Chemo-Sero-Therapeutic Research Institute (Kumamoto, Japan). All work using SARS-CoV was performed in BioSafety Level 3 facilities by personnel wearing powered air-purifying respirators (Shigematsu Works).

To generate a pBR322-based plasmid vector (pBMSF) for homologous recombination into the hemagglutinin (HA) locus of m8, we cloned the HA gene, which contained the ATI/p7.5 synthetic hybrid promoter, from the pSFJ1–10 plasmid and inserted it into the pBM vector, which was reconstructed in our laboratory. Full-length cDNAs for the SARS-CoV nucleocapsid (N), membrane (M), and envelope (E) proteins were cloned from the Vietnam/NB-04/2003 strain of SARS-CoV by RT-PCR (16). Full-length SARS-CoV spike (S) protein gene was prepared from pSFJ1-10-SARS-S, which is described in our previous report (17). Next, the genes that encode the SARS-CoV structural proteins were ligated by inserting internal ribosomal entry site sequence of hepatitis C virus (genotypes 2a and 1b/2b) fused with the 2A sequence of foot and mouth disease virus and Thosea asigna virus or encephalomyocarditis virus by PCR (see Fig. 1 A). The generated DNA fragment was digested with EcoRI and inserted downstream of the ATI/p7.5 hybrid promoter of pBR322-based plasmid vector pBMSF, thereby generating pBMSF-SARS-NMES. The pBMSF-SARS-NMES plasmid was linearized with PvuI, and transfected into primary rabbit kidney cells that had been infected with m8 at a multiplicity of infection (MOI) of 10. After 36 h, the virus-cell mixture were harvested by scraping, and frozen at −80°C until use. The resulting HA-negative recombinant viruses were purified as previously described (17), and named m8rVV-NMES. Furthermore, recombinant mO that expressed the SARS-CoV N, M, or E protein with a six histidine tag at the C terminus was generated (mOrVV-NHis, mOrVV-MHis, and mOrVV-EHis), as was mO that expressed six histidine-tagged S protein (mOrVV-SHis), as previously described (17).

Vero E6 cells were infected with m8rVV-NMES at an MOI of 5. After 18 h, the cells were lysed with lysis buffer (10 mM Tris (pH 7.4), 150 mM NaCl, 1% SDS, 0.5% Nonidet P-40, protease inhibitor cocktail). The cell lysates (30 μg) were resolved by SDS-PAGE and transferred to a polyvinylidene difluoride membrane (Immobilon-P; Millipore). After blocking the membranes with 5% skim milk solution at room temperature for 1 h, the membrane was incubated with polyclonal Abs against the N, M, E, or S protein. Vero E6 cell lysates infected with mOrVV-NHis, mOrVV-MHis, mOrVV-EHis, or mOrVV-SHis was used as positive controls. We used the anti-S polyclonal Abs described in our previous study (17). Polyclonal Abs against N and E proteins were prepared from rabbit sera immunized with KLH-conjugated N peptide (residues aa 250–263) and E peptide (residues aa 61–73). Polyclonal Abs against the M protein were provided by Dr. Mizutani (National Institute of Infectious Diseases, Musashimurayama, Tokyo). We purified the IgG fractions of these antisera using the protein A Ampure PA kit (Amersham Biosciences). After washing with TBS that contained 0.1% Tween 20 (TBST), the membranes were reacted with HRP-conjugated F(ab′)₂ of anti-rabbit IgG (GE Healthcare). Each specific protein band was visualized using the ECL system (GE Healthcare).

Vero E6 cells were infected with m8rVV-NMES at an MOI of 5 at 30°C for 4 h. The cells were washed with PBS and fixed with cold acetone/methanol (1/1) mixture for 10 min. After blocking with TNB blocking buffer (NEN Life Science Products) at room temperature for 1 h, the fixed cells were incubated with polyclonal Abs against the N, M, or E protein or mAb against the S protein (designated as anti-S-His protein, clone no. 13B8), which was originally prepared in our laboratory, at 4°C overnight. After washing, the cells were incubated with Alexa Fluor 488-conjugated anti-rabbit IgG or mouse IgG Ab at room temperature for 1 h. Nuclei were stained with DAPI (4′,6-diamidino-2-phenylindole). Fluorescence images were acquired using a confocal microscope (LSM510 META; Carl Zeiss).

RK13 cells were cultured in 150-mm dishes, and then infected with m8rVV-NMES at an MOI of 5. After 48 h of incubation, the culture supernatants were collected and centrifuged to remove cell debris at 3000 rpm for 30 min at 4°C. The supernatants were concentrated ∼100-fold using the Pellicon XL (cut off molecular weight 3 × 10⁵; Millipore). The isolation of virus-like particles (VLP) was performed as previously described, with a slight modification (18). Briefly, the concentrated supernatant was placed on 60% (w/w) sucrose cushion and centrifuged at 4.0 × 10⁴ rpm for 5 h. The opalescent band was collected and centrifuged in a 20–60% (w/w) sucrose gradient at 2.7 × 10⁴ rpm for 4 h, and then divided into 20 fractions. The protein content of each fraction was determined with the DC protein assay kit (Bio-Rad). The 20 μl of each fraction were separated by SDS-PAGE (7.5%, 10%, or 15% polyacrylamide gel), and transferred onto a polyvinylidene difluoride membrane. The membrane was incubated with mAb against S protein (13B8), mAb against N protein (IMG-654; Imgenex) or polyclonal Abs against the M or E protein. After washing, the membranes were reacted and visualized as described. The VLPs in the concentrated culture supernatant were visualized using transmission electron microscopy. For immunogold staining, VLPs were loaded onto a collodion-coated electron microscopy grid for 5 min. After the removal of excess sample solution, polyclonal Ab against S protein was added onto the grid and incubated at room temperature for 1 h. The grids were washed six times with Sorensen’s phosphate buffer at room temperature and incubated with 5-nm gold-conjugated anti-rabbit IgG for 1 h. After washing with Sorensen’s phosphate buffer for 10 s, the samples were stained with 2% phosphotungstic acid for 1 min. After draining off the excess phosphotungstic acid, the samples were observed under the electron microscopy.

Groups of three New Zealand White rabbits (SLC) were immunized intradermally with 1 × 10⁸ PFU/body of m8rVV-NMES or with 1 × 10⁸ PFU/body of m8, at 0 and 6 wk. Sera were collected at the indicated time points (see Fig. 2 A), and used in ELISA analysis and in the in vitro neutralization assay described below. All animal experiments using rabbits were approved by The Tokyo Metropolitan Institute of Medical Science Animal Experiment Committee and were performed in accordance with the animal experimentation guidelines of The Tokyo Metropolitan Institute of Medical Science.

Recombinant SARS-CoV N, M, E, and S proteins tagged with six histidines at the C terminus were expressed in RK13 cells by infecting with mOrVV-N-His, mOrVV-E-His, mOrVV-M-His, or mOrVV-S-His at an MOI of 5. These proteins were purified using nickel Sepharose (6 Fast Flow; GE Healthcare). His-tagged E and M proteins were further purified by SDS-PAGE. These full-length structural proteins (0.2 μg/ml, 50 μl/well) were coated onto 96-well plates at 4°C overnight. The plates were blocked with 1% BSA in PBS(−) that contained 0.5% Tween 20 and 2.5 mM EDTA, and then incubated with serial 2-fold dilutions of sera from the rabbits immunized with m8rVV-NMES or m8. After extensive washing, the plates were assayed as previously described, except that o-phenylenediamine was used as the substrate (17). The individual SARS-CoV structural protein-specific IgG titers are presented as the end point dilution Ab titers. The end point titer was defined as the reciprocal of the highest dilution of serum at which the absorbance at 490 nm (A₄₉₀) ratio (A₄₉₀ of m8rVV-NMES-immunized serum/A₄₉₀ of m8-immunized serum (negative control)) was greater than 2.0, as previously described (19).

The neutralizing Ab titers of the sera of rabbits immunized with m8rVV-NMES or m8 were determined as previously described (17). Briefly, serial 2-fold dilutions of heat-inactivated sera were mixed with equal volumes of 200 tissue culture ID₅₀ (TCID₅₀) of SARS-CoV and incubated at 37°C for 1 h. Vero E6 cells were then infected with 100 μl of the virus-serum mixtures in 96-well plates. After 5 days (or 6 days in the SARS-CoV challenge experiment) of infection, the neutralization titer was determined as the end point dilution of the serum at which there was 50% inhibition of the SARS-CoV-induced cytopathic effect. The method used for end point calculation was that described by Reed and Muench (20).

Female BALB/c mice older than the 6 mo of age (SLC) were used in this study. Four groups of eight BALB/c mice (seven mice in the vehicle-immunized group) were inoculated intradermally with either 1 × 10⁷ PFU/body of m8, m8rVV-S, or m8rVV-NMES or 70 μl of vehicle (MEM without FCS). At 7–8 wk postimmunization, the mice were infected intranasally with 1 × 10⁵ TCID₅₀/body of SARS-CoV (20 μl/mouse), as previously described (11). Four mice from each group were sacrificed 2 and 9 days later, except for the three mice of the vehicle-immunized group, which were sacrificed 2 days later. The mice were sacrificed under anesthesia and the lung, liver, small intestine, and spleen were extirpated. Aliquots of these tissues were frozen immediately at −80°C or fixed with 10% formalin. The collected blood was used for the in vitro neutralization assay. In addition, BALB/c mice were injected intradermally with 1 × 10⁷ PFU/body of recombinant VV that expressed each structural protein of SARS-CoV (mOrVV-NHis, mOrVV-MHis, mOrVV-EHis, mOrVV-SHis) with or without LC16mOrVV-SHis (i.e., LC16mOrVV-N, -M, -E, -S alone or LC16mOrVV-N + LC16mOrVV-S, -M + LC16mOrVV-S, or -E + LC16mOrVV-S), and infected with 1 × 10⁵ TCID₅₀/body of SARS-CoV more than 4 wk later. After 2 and 9 days, mice (n = 3–5 per group) were sacrificed following blood collection under anesthesia, and their lungs were extirpated. All animal experiments using mice were approved by the Animal Experiment Committee at The Institute of Medical Science, University of Tokyo, and were performed in accordance with the animal experimentation guidelines of The Institute of Medical Science, University of Tokyo.

The SARS-CoV titers in the mouse organs were determined as previously described (11). Briefly, tissue samples (i.e., lung, liver, small intestine, and spleen) were homogenized in a 10-fold volume of Leibovitz 15 medium (Invitrogen). The homogenates were centrifuged at 2000 rpm for 10 min at 4°C. Serial 10-fold dilutions of the supernatants of these homogenates were added to Vero E6 cells seeded on 96-well plates. After 6 days of incubation, the cells were fixed with 10% formalin. Viral titer was determined as the 50% end point dilution of the homogenate that induced the cytopathic effect. The method used for end point calculation was that described by Reed and Muench (20).

In accordance with a previous report (11), 10% formalin-fixed lung tissues of the SARS-CoV-infected mice were embedded in paraffin. Paraffin block sections (4-μm thickness) were stained with H&E staining. The peribronchial and perivascular scores were recorded in a blinded fashion by a pathologist. We evaluated pulmonary pathology using the histopathologic scoring systems developed by Cimolai et al. (21), in which the scoring system is weighted heavily for bronchial lesions. This scoring system allowed us to differentiate the severity of pulmonary pathology in small groups of animals. The pathology grading system consisted of a numerical score ranging from 0 to 26. In brief, each section was scored based upon a cumulative total from five categories that incorporated evaluations of the following: A) number of bronchiolar and bronchial sites affected by the periluminal infiltrate (range, 0 to 3); B) severity of the periluminal infiltrate (range, 0 to 3); C) luminal exudate severity (range, 0 to 2); D) frequency of perivascular infiltrate (range, 0 to 3); and E) severity of parenchymal pneumonia (range, 0 to 5). The accumulated numeric score was derived from the sum of the subscores: A+3(B+C)+D+E. Eosinophils were detected in tissue sections by method of Luna (22).

To measure the levels of cytokine or chemokine mRNA, total RNA samples were extracted from the lungs using the RNeasy Mini kit (Qiagen). Quantitative RT-PCR was conducted with TaqMan Gene Expression assays (Applied Biosystems) using the ABI Prism 7700 and Sequence Detection System software v.1.7. The fold change in copy number of each cytokine/chemokine mRNA was revealed using the 2^−ΔΔCt method using 18 S rRNA as an endogenous calibrator.

Data are presented as mean ± SD. Statistical analysis was performed by one-way ANOVA, followed by the Dunnett or Bonferroni test. A value of p < 0.05 was considered to be statistically significant.

A multicistronic transgene that expresses simultaneously four structural proteins (N, M, E, and S proteins) of SARS-CoV was constructed and inserted into the HA locus of LC16m8 (m8) by homologous recombination (Fig. 1,A). Expression of the transgene was placed under the control of the powerful ATI/p7.5 hybrid promoter. We screened for m8rVV-NMES using the erythrocyte agglutination assay (17), and confirmed the insertion of the transgene by PCR. Expression of the N, M, E, and S proteins in Vero E6 cells infected with m8rVV-NMES was detected by Western blot analysis. Recombinant LC16mO (mO) expressing the C-terminal histidine-tagged N, M, E or S protein (mOrVV-NHis, -MHis, -EHis, and -SHis) was generated as previously described, and used as a positive control for each protein. We also used m8rVV-S (17). As shown in Fig. 1,B, the expression levels of the N and S proteins in the m8rVV-NMES-infected cells were high and moderate, respectively. In contrast, the expression levels of the M and E proteins in m8rVV-NMES-infected cells were weaker than those in mOrVV-MHis- and mOrVV-EHis-infected cells. The M protein in the m8rVV-NMES-infected cells was 20 kDa, whereas that in the mOrVV-MHis-infected cells was observed as forms of ∼20 kDa (nonglycosylated form) and 25 kDa (glycosylated form) (23). Furthermore, we investigated the cellular localizations of these structural proteins by indirect immunofluorescence (Fig. 1,C). In m8rVV-NMES-infected cells, all of the SARS-CoV proteins were localized in the perinuclear regions. In particular, the localization of the N protein in m8rVV-NMES-infected cells was different from that in mOrVV-NHis-infected cells, in which the N-His protein was found diffusely in the cytoplasm. VLPs are formed by the assembly of structural proteins in the cytoplasm, followed by release into the culture medium. By infecting m8rVV-NMES into RK13 cells, we confirmed the formation of VLPs in the culture medium. After sucrose gradient centrifugation, 20 fractions (500 μl each) were collected (Fig. 1,D). The four SARS-CoV structural proteins were monitored by Western blot analysis. As shown in Fig. 1,E, fraction number 10 contained all the SARS-CoV proteins, and the buoyant density of this fraction was ∼1.15 g/ml, a value that is consistent with previous reports (18, 24, 25). Moreover, we confirmed the formation of VLPs in the concentrated culture supernatant using scanning electron microscopy and immunogold-labeling with the anti-S protein polyclonal Ab. The particles were 70–100 nm in diameter, which is consistent with the sizes as reported previously (18, 24, 25). The particles were positively stained with immunogold (Fig. 1 F).

To investigate the immunogenicity of m8rVV-NMES, 1 × 10⁸ PFU/body of either m8rVV-NMES or m8, its parental strain, was inoculated intradermally on the backs of New Zealand White rabbits at 0 and 6 wk (Fig. 2,A). Rabbit antisera specific for the full-length structural proteins of SARS-CoV were detected by ELISA (Fig. 2,B). In agreement with previous reports (26, 27, 28), the N and S proteins both exhibited strong immunogenicity in rabbits. IgG-specific for the N or S protein was induced as early as 1 wk after m8rVV-NMES immunization, and the titer exceeded 1:10000 2 wk later. The titers of Abs against the N and S proteins were dramatically increased by booster immunization with m8rVV-NMES. It was also observed that the Ab titer of the N protein, but not that of the S protein, decreased after reaching the peak titer. Immunization with m8rVV-NMES did not induce Abs specific for the E and M proteins, even after booster immunization (Fig. 2 B). The antigenicity of the purified E and M proteins coated onto the ELISA plates was confirmed using each rabbit anti-E or anti-M peptide Ab (data not shown). Therefore, we believe that the lack of induction of Abs specific for the E and M proteins in the rabbit sera results from the poor immunogenicity and lower expression levels of these proteins.

We determined the neutralization titers against SARS-CoV using the same rabbit antisera. The neutralization titer was ∼1:30 (range, 1:25 to 1:36) after 2 wk, and was sustained for 6 wk (Fig. 2,C). Booster immunization with m8rVV-NMES further increased the neutralization titer more than 10-fold 2 wk later. These values are somewhat lower than those induced by m8rVV-S in our previous report (17). In contrast, the antisera from rabbits immunized with m8 did not exhibit any neutralizing activity against SARS-CoV (Fig. 2 C).

As m8rVV-NMES and m8rVV-S could induce high levels of neutralizing Abs against SARS-CoV (Fig. 2,C), we investigated the influences of m8rVV-NMES and m8rVV-S on SARS-CoV challenge of BALB/c mice (Fig. 3,A). The m8rVV-NMES and m8rVV-S constructs were inoculated intradermally on the backs of BALB/c mice at 1 × 10⁷ PFU/body. At 7–8 wk after this single immunization, the mice were infected intranasally with SARS-CoV at 1 × 10⁵ TCID₅₀/body. After 2 and 9 days, the lung, liver, small intestine, and spleen were extirpated from the mice under anesthesia, and the SARS-CoV titers were measured. As shown in Fig. 3 B, 200- and 100-fold reductions in pulmonary virus titers were observed in the m8rVV-NMES-immunized and m8rVV-S-immunized groups 2 days after infection. The virus titers in the lungs of the m8rVV-NMES-immunized and m8rVV-S-immunized groups were 5.40 × 10⁵ and 1.52 × 10⁶ TCID₅₀/g of lung, respectively. In contrast, the vehicle-immunized and LC16m8-immunized groups exhibited virus titers of 1.07 × 10⁸ and 1.18 × 10⁸ TCID₅₀/g of lung, respectively. The virus was not detected in the lungs of any group 9 days later, as reported previously (11, 15). In contrast, virus titers in other organs, including liver, small intestine, and spleen, were lower than that of the detection limit 2 and 9 days after infection (data not shown).

We also measured the neutralization titers in these mice sera 2 and 9 days after SARS-CoV infection (Fig. 3 C). Two days postinfection, the neutralization titers of the m8rVV-NMES-immunized and m8rVV-S-immunized groups were 1:11.1 ± 1.01 and 1:14 ± 3.94, respectively, whereas those of the negative control groups were below the limit of detection. At 9 days postinfection, the serum neutralization titers of m8rVV-NMES-immunized and m8rVV-S-immunized groups had increased to 1:838.0 ± 681.0 and 1:367.9 ± 132.1, respectively. In contrast, the serum neutralizing titers of the vehicle-immunized and m8-immunized groups were 1:59.7 ± 35.4 and 1:67.8 ± 18.6, respectively. These results suggest that both the m8rVV-NMES- and m8rVV-S-immunized groups could elicit neutralizing Abs against SARS-CoV and alleviate SARS-CoV infection.

We performed histopathologic analyses of lung tissues. Two days after SARS-CoV infection, the vehicle-, m8-, and m8rVV-S-immunized groups showed only slight pulmonary inflammation (Fig. 4,A, a, b, and d), whereas the m8rVV-NMES-immunized group showed infiltration of lymphocytes into the areas surrounding the bronchi and slight thickening of the alveolar epithelium (Fig. 4,A, c). We scored pulmonary inflammation in all the groups 2 days after SARS-CoV infection as follows (Fig. 4,B): in the m8rVV-NMES-immunized group, 5.00 ± 2.71; in the vehicle-immunized group, 2.00 ± 2.00; in the m8-immunized group, 1.33 ± 0.82; and in the m8rVV-S-immunized group, 2.50 ± 1.00. At 9 days postinfection, the vehicle-, m8-, and m8rVV-NMES-immunized groups exhibited severe pulmonary inflammation, i.e., infiltration of inflammatory cells and thickening of alveolar epithelia (Fig. 4,A, e, f, and g). In contrast, the m8rVV-S-immunized group showed only slight pulmonary inflammation (Fig. 4,A, h). As shown in Fig. 4,B, the pulmonary inflammation score for the m8rVV-NMES-immunized group (12.75 ± 2.87) 9 days after SARS-CoV infection was significantly higher than that for the m8rVV-S-immunized group (3.50 ± 3.00). In contrast, this score was comparable to those obtained for the vehicle-immunized and m8-immunized groups (9.75 ± 2.87 and 8.33 ± 2.31, respectively). The m8rVV-NMES-immunized group exhibited as severe inflammation as the control groups, although m8rVV-NMES contains the S protein and protects as well as m8rVV-S against SARS-CoV infection. In addition, marked infiltration of neutrophils, eosinophils, plasma-like cells, and lymphocytes was observed in the m8rVV-NMES-immunized group, as compared with the control groups, after SARS-CoV infection (Fig. 4 C, b and D).

These results suggest that the severe pulmonary inflammation seen in m8rVV-NMES-immunized mice after SARS-CoV infection results from host immune responses rather than a direct cytopathic effect of SARS-CoV, because the virus titers for all the group were negligible 9 days after SARS-CoV infection and the virus titer of the m8rVV-NMES-immunized group was significantly decreased 2 days postinfection.

We hypothesized that the severe pulmonary inflammation seen in the m8rVV-NMES-immunized mice resulted from the host immune responses to SARS-CoV components expressed by m8rVV-NMES. This notion was supported by the observation of negligible virus titers 9 days after SARS-CoV infection. Therefore, we investigated the influence of recombinant VV expressing each structural protein of SARS-CoV (mOrVV-NHis, mOrVV-MHis, mOrVV-EHis, and mOrVV-SHis) on subsequent intranasal infection with SARS-CoV. BALB/c mice were immunized with mOrVV-NHis, -MHis, -EHis, and -SHis at 1 × 10⁷ PFU/body, and 4 wk later infected intradermally with 1 × 10⁵ TCID₅₀ of SARS-CoV (Fig. 5,A). After 2 and 9 days, three mice from each group were sacrificed following blood collection under anesthesia, and their lungs were extirpated. Consistent with earlier results, a significant reduction of pulmonary virus titer was observed after 2 days in only the mOrVV-SHis-immunized group (Fig. 5,B). In contrast, immunization with the other SARS-CoV structural proteins, including the N, M, and E proteins, did not confer protection against the subsequent SARS-CoV infection. As shown in Fig. 5,C, the alleviation of pulmonary inflammation was also observed in the mOrVV-SHis-immunized group. Severe infiltration of lymphocytes and thickening of the alveolar epithelia were observed in the lung tissues of the mOrVV-NHis-immunized mice 9 days after SARS-CoV infection (Fig. 5,C). The pulmonary damage in the mOrVV-NHis-immunized mice (15.00 ± 5.56) was significantly more severe than that in the mOrVV-SHis-immunized mice (5.67 ± 2.52) (Fig. 5,D). However, there were no significant differences among the other groups. Furthermore, infiltration of neutrophils, eosinophils, and lymphocytes was observed in the mOrVV-NHis-immunized mice after SARS-CoV infection (Fig. 5,E, b), although the extent of infiltration of these cells into the lungs of these mice was somewhat lower than that observed in the m8rVV-NMES-immunized mice after SARS-CoV infection (Fig. 4 D). This may explain the observed differences in the histopathologic findings for the mOrVV-NHis-immunized mice and m8rVV-NMES-immunized mice.

To elucidate the reason for the severe pulmonary inflammation observed in the mOrVV-NHis-immunized mice after SARS-CoV infection, we measured by quantitative RT-PCR the mRNA levels for various cytokines and chemokines in the lungs of BALB/c mice preimmunized with mOrVV-NHis, -MHis, -EHis, -SHis, or mO. Several proinflammatory cytokine and chemokine mRNAs, including those for IL-6, CXCL10, CCL2, and CCL3, were increased in all the groups, with the exception of the mOrVV-SHis group, 2 days after SARS-CoV infection (Fig. 6,A). In contrast, the mOrVV-SHis-immunized group showed low levels of mRNA expression for these proinflammatory cytokines or chemokines, especially IL-6, resulting in reduced lung pathology after immunization. The mRNA levels for IFN-γ, IL-2, IL-4, and IL-5 were highest in the mOrVV-NHis-immunized group (Fig. 6, A and B). None of the other groups showed up-regulation of these cytokines, with the exception of the IL-5 mRNA level in the mOrVV-SHis-immunized group. Furthermore, the mRNA expression levels of anti-inflammatory cytokines (IL-10 and TGF-β) in the mOrVV-NHis-immunized group were markedly lower than expression levels in any of the other groups, which exhibited high virus titers, and were comparable to those of the mOrVV-SHis group, in which pulmonary inflammation was alleviated (Fig. 6 C).

To verify the exacerbating effect of N protein immunization, we investigated the pulmonary virus titers and histopathology in BALB/c mice that were previously immunized with the combination of mOrVV-N and mOrVV-S (mOrVV-N+S-immunized group) 2 and 9 days after SARS-CoV infection, and compared them to those of all other groups, including the mO-, mOrVV-M+S-, mOrVV-E+S-, and mOrVV-S-immunized groups. The mOrVV-N+S-immunized group showed significantly decreased pulmonary virus titers compared with the mO-immunized group (Fig. 7,A). However, the mOrVV-N+S-immunized group exhibited as severe pneumonia as the mO-immunized group (Fig. 7, B and C). In contrast, both the mOrVV-M+S-immunized group and the mOrVV-E+S-immunized group were protected against SARS-CoV infection to the same extent as the mOrVV-S-immunized group (Fig. 7, A–C).

SARS-CoV is newly identified as an agent of SARS. However, the detailed mechanism by which SARS-CoV causes severe pneumonia remains unclear. The uncontrolled release of immune mediators has been implicated in the pathogenesis of SARS, whereas the cytokine profiles of SARS patients have not elucidated the cause the pneumonia owing to their diversity. It seems likely that the diverse cytokine profiles noted among adult SARS patients are related to patient anamnesis.

In the present study, we observed severe pulmonary inflammation in m8rVV-NMES-immunized BALB/c mice 9 days after SARS-CoV infection (Fig. 4,A, g), even though the initial virus titer was significantly lower than those of the control groups, which included vehicle- and m8-immunized mice (Fig. 3,B). The severity of pulmonary inflammation did not correlate with the virus titer in the m8rVV-NMES-immunized mice, in contrast to the correlations observed for the vehicle-, m8-, and m8rVV-S-immunized groups. We identified the N protein of SARS-CoV as the cause of the severe pneumonia observed during SARS-CoV infection (Fig. 5, C and D, and 7, B and C). To date, no studies have been reported to our knowledge regarding SARS patients with severe pneumonia who were previously immunized with either SARS-CoV or a highly related species. In contrast, there are several reports of antisera against human CoV (229E and OC43) and host factor IL-11 cross-reacting with the SARS-CoV Ag (29, 30). Furthermore, the N protein of SARS-CoV has been shown to induce both cellular and humoral immune responses (31, 32, 33). Taken together, these results raise the possibility that a percentage of SARS patients already possess the adaptive immune response elements that can interact with SARS-CoV components, including the N protein, and that their adaptive immune response may be involved in the exacerbation of pneumonia. The temporal changes in immune response and the pathogenesis after SARS-CoV infection of an animal model that had previously been immunized with SARS-CoV components are not well understood, as almost all the previous studies reported only protection within a few days of SARS-CoV infection (34, 35, 36, 37, 38, 39). In the present study, we demonstrate that mOrVV-NHis-immunized mice after SARS-CoV infection exhibit an imbalance between T cell activation (high expression levels of IFN-γ, IL-2, IL-4, and IL-5) and subsequent suppression (low expression levels of IL-10 and TGF-β), as well as high-level production of proinflammatory cytokines (IL-6 and TNF-α) and chemokines (CCL2, CCL3, and CXCL10). Jiang et al. (40) reported elevation of CXCL10 or IP-10 production in the pneumocytes, CD3⁺ T cells, and monocytes and macrophages of the lungs of patients with SARS. CXCL10 may be responsible for the infiltration of activated T cells and monocytes or macrophages, which is a pathologic finding in SARS patients (41, 42, 43). It has been reported that elevated expression of monocyte or macrophage activation factors (CCL2 and CCL3) was observed in SARS patients (8, 44). Furthermore, the highest expression of IL-6 in mOrVV-NHis-immunized mice is reasonable (Fig. 6 A), as the elevation of IL-6 levels is considered one of the causes in the severe pneumonia of SARS patients. Zhang et al. (45) reported recently the molecular mechanism of IL-6 expression induction by the N protein of SARS-CoV. In contrast, both IL-10 and TGF-β play important roles in suppressing inflammatory responses (46). Thus, the reduced production of both anti-inflammatory cytokines in the mOrVV-NHis-immunized mice after SARS-CoV infection may be related to the severity of the pulmonary inflammation in these mice. Weingartl et al. (47) and Czub et al. (48) reported that immunization with S protein expressing-recombinant modified VV Ankara (rMVA-S) induced stronger inflammatory responses and focal necrosis in liver tissues after SARS-CoV challenge than in control animals. However, the precise mechanism underlying this liver inflammation has not been clarified. Feline infectious peritonitis virus, which is another member of the coronavirus family, exhibits enhanced infection into monocytes or macrophages through virus-specific Ab binding to the Fc receptors of these cells and causes enhanced inflammation (49). It has also been reported for dengue virus that secondary infection with a different genotype results in more severe symptoms, including dengue hemorrhagic fever and dengue shock syndrome. The exacerbation of this symptom is also positively associated with pre-existing Abs with specificity for dengue virus (50). In the case of SARS-CoV, Ab-dependent enhancement of infection has not been reported previously. We hypothesized that the severe pneumonia observed in mOrVV-NHis-immunized mice after SARS-CoV infection does not result from Ab-dependent enhancement because the virus titers in the mouse lungs 9 days later were below the detection limit. Deming et al. (51) reported recently the intensive infiltration of eosinophils as well as lymphocytes after SARS-CoV infection of aged BALB/c mice previously immunized with the N protein of SARS-CoV. It has also been reported that immunization with formalin-inactivated respiratory syncytial virus vaccine and VV that expresses the G glycoprotein of respiratory syncytial virus correlates with the augmentation of Th2-type immune responses and enhanced pulmonary disease (52, 53). Therefore, the authors speculated that the Th2-biased responses of vaccinated hosts after SARS-CoV infection might aggravate pulmonary inflammation, although the main host response remains unknown. In contrast, our current data suggest that N protein-immunized mice exhibit activation of both Th1 and Th2 responses after SARS-CoV infection. In agreement with our data, Jin et al. (54) have demonstrated that prior immunization with N protein generates stronger Ag-specific Th1 and Th2 responses than immunization with M or E protein. In addition, we demonstrate the suppression of anti-inflammatory cytokine responses in N protein-immunized mice. Interestingly, Shi et al. (55) demonstrated that coinjection of M protein with N protein not only enhanced the production of Th1 cytokines (IFN-γ and IL-2), but also reduced the rates of mortality and pathologic change in SARS-CoV-infected voles. These results suggest that further studies, including epitope analysis, are required to reveal the precise mechanism underlying the severe pulmonary inflammation that results from SARS-CoV infection of BALB/c mice immunized with the N protein of SARS-CoV.

In contrast, intradermal immunization of aged BALB/c mice with m8rVV-S at 1 × 10⁷ PFU/body significantly reduced the pulmonary virus titer 2 days after SARS-CoV infection (Fig. 3 B). Furthermore, the m8rVV-S-immunized group exhibited alleviation of the pulmonary histopathology, as compared with both control groups after 9 days. To date, various types of SARS vaccine, including recombinant vaccines, inactivated vaccines, and DNA vaccine, have been reported (34, 35, 36, 37, 38, 39). There are only a few reports on the effect of a single immunization with recombinant SARS vaccines, namely SARS-CoV S protein-expressing vaccines based on rabies virus (56), vesicular stomatitis virus (39), and adeno-associated virus (57). It is noteworthy that a single i.m. immunization with recombinant adeno-associated virus that expresses the receptor-binding domain of S protein conferred long-term protection against SARS-CoV infection (57). In the present study, we also show that a single immunization with m8rVV-S reduces viral load and improves the histopathologic findings in the lungs of BALB/c mice infected with high-titer (1 × 10⁵ TCID₅₀/body) SARS-CoV, although a relatively low titer of SARS-CoV was used in the previous study conducted by Du et al. (57). These results suggest that the systemic immune responses induced by a single immunization with SARS vaccine successfully protect the animal model against intranasal SARS-CoV infection.

In summary, we demonstrate that the immunization of BALB/c mice with the N protein of SARS-CoV causes severe pulmonary inflammation upon subsequent SARS-CoV infection, probably via the imbalance created between T cell activation and suppression, as well as by massive proinflammatory cytokine production. These results provide new insights into the mechanisms involved in the pathogenesis of SARS and help in the development of safe vaccines.

We are grateful to Dr. Ryuichi Miura (University of Tokyo) for arranging the SARS-CoV challenge experiment. We are also grateful to Iyo Kataoka (Institute of Medical Science, University of Tokyo). We thank Dr. Masahiro Shuda of the University of Pittsburgh for helpful discussions. We also thank Dr. Tetsuya Mizutani and Dr. Shigeru Morikawa (Department of Virology I, National Institute of Infectious Diseases) for providing antisera from rabbits immunized with the M protein peptide and inactivated SARS-CoV particles.

The authors have no financial conflict of interest.