A Mouse-Adapted SARS-Coronavirus Causes Disease and Mortality in BALB/c Mice

No single animal model for severe acute respiratory syndrome (SARS) reproduces all aspects of the human disease. Young inbred mice support SARS-coronavirus (SARS-CoV) replication in the respiratory tract and are available in sufficient numbers for statistical evaluation. They are relatively inexpensive and easily accessible, but their use in SARS research is limited because they do not develop illness following infection. Older (12- to 14-mo-old) BALB/c mice develop clinical illness and pneumonitis, but they can be hard to procure, and immune senescence complicates pathogenesis studies. We adapted the SARS-CoV (Urbani strain) by serial passage in the respiratory tract of young BALB/c mice. Fifteen passages resulted in a virus (MA15) that is lethal for mice following intranasal inoculation. Lethality is preceded by rapid and high titer viral replication in lungs, viremia, and dissemination of virus to extrapulmonary sites accompanied by lymphopenia, neutrophilia, and pathological changes in the lungs. Abundant viral antigen is extensively distributed in bronchial epithelial cells and alveolar pneumocytes, and necrotic cellular debris is present in airways and alveoli, with only mild and focal pneumonitis. These observations suggest that mice infected with MA15 die from an overwhelming viral infection with extensive, virally mediated destruction of pneumocytes and ciliated epithelial cells. The MA15 virus has six coding mutations associated with adaptation and increased virulence; when introduced into a recombinant SARS-CoV, these mutations result in a highly virulent and lethal virus (rMA15), duplicating the phenotype of the biologically derived MA15 virus. Intranasal inoculation with MA15 reproduces many aspects of disease seen in severe human cases of SARS. The availability of the MA15 virus will enhance the use of the mouse model for SARS because infection with MA15 causes morbidity, mortality, and pulmonary pathology. This virus will be of value as a stringent challenge in evaluation of the efficacy of vaccines and antivirals.


Introduction
The occurrence in late 2002 and early 2003 of cases of severe acute respiratory syndrome (SARS) in southeast China quickly drew international attention as the disease sickened more than 8,000 people and spread to more than 30 countries within six months.Since the identification of the etiological agent, the SARS-coronavirus (SARS-CoV), in 2003, development and characterization of animal models for evaluation of prophylaxis and treatment strategies have been of great interest.Although SARS-CoV has not been associated with a subsequent widespread outbreak since 2002-2003, the potential for such an outbreak remains.Identification of a SARS-like coronavirus in Chinese horseshoe bats (Rhinolophus species) that are indigenous across Southeast Asia suggests that they may represent a natural reservoir from which viruses may be introduced into the human population [1].The course of infection in animal models is abbreviated compared with the course of SARS in humans; however, many aspects of SARS-CoV-associated disease are reproducible in animal models, including age-dependent susceptibility, re-covery of SARS-CoV from respiratory tissues and secretions, infection of type I and type II pneumocytes and bronchial epithelial cells, detection of viral genome in blood and extrapulmonary tissues, and pulmonary pathology (including pneumonitis, edema, necrotic debris, and hyaline membrane formation) [2,3].Clinical symptoms have been reported in some species, but the findings are not entirely reproducible in outbred species.Variability in clinical symptoms seen in outbred species may result from additional factors, including infection with co-pathogens, stress, existence of sub-species of test animals, and use of different virus strains.This variation can be problematic in studies of pathogenesis and vaccine efficacy unless a large enough number of animals are included in each treatment group [2].The ideal animal model would demonstrate viral replication in respiratory tissues, histopathologic evidence of respiratory disease, and consistent clinical signs of disease, including mortality.A small animal model in which all of these aspects of virus-associated disease are seen would be desirable because reproducible data can be generated in inbred animals, and larger numbers of animals can be included for statistical analysis of biological outcomes.
To generate a model that satisfies these criteria, we have serially passaged SARS-CoV in the respiratory tract of young BALB/c mice, resulting in a lethal virus that causes dosedependent weight loss and mortality associated with higher viral titers in the respiratory tract than are seen with the wildtype virus and with histopathologic findings of severe pulmonary disease.The characteristics of this lethal mouseadapted SARS-CoV, (MA15), are reported here.

Adaptation of SARS-CoV for Increased Virulence in a Young BALB/c Mouse Model
Adaptation of SARS-CoV (Urbani) was achieved by serial passage through lungs of BALB/c mice as previously described for influenza A and influenza B viruses [4,5].Lightly anesthetized mice were inoculated intranasally with 10 5 50% tissue culture infectious dose (TCID 50 ) per mouse of SARS-CoV (Urbani).Two to three days post-infection (d.p.i.), when peak viral titers are observed, lungs were harvested from infected mice and clarified homogenates were used as inocula for continued serial passage via intranasal inoculation in mice.To screen for virulence in mice, groups of young naı ¨ve mice (N ¼ 5-8) were inoculated with lung homogenates collected at passage 2, 10, and 15 (P2, P10, and P15, respectively), weighed daily, and observed twice daily for signs of morbidity or mortality.Deaths were not observed following inoculation with P2 or P10; however, increased morbidity, as indicated by weight loss, was noted in P10inoculated mice at day 3 post-infection (p.i.) (unpublished data).Mortality was observed in P15-inoculated mice; 60% of P15-inoculated mice died or displayed extreme morbidity and were euthanized by 5 d.p.i.Mortality was not observed in mice infected with intermediate passages (P11-P14), and accompanying morbidity was not measured (unpublished data).Supernatant from lung homogenates at P15 contained a heterogeneous virus pool as evident upon sequence analysis of cDNA fragments generated by reverse transcriptase PCR (RT-PCR) amplification from purified viral RNA (vRNA).Dual peaks were observed on sequence chromatograms at several nucleotide residues, indicating mixed populations in the virus pool.To obtain a clonal population, P15 virus was biologically cloned by three rounds of terminal dilution in Vero cells.Five clones were screened for lethality in 6-to 8wk-old BALB/c mice; these clones caused mortality from 50% to 100%.One clone, designated MA15, resulted in 100% mortality within 6 d.BALB/c mice aged 6 to 8 wk, 4 mo, and 13 mo were all susceptible to MA15 infection, with severe morbidity or death occurring within 3 to 5 d following intranasal inoculation.

Mutations Identified in the Mouse-Adapted Virus
In order to identify the mutations in MA15 associated with the lethal phenotype, the sequence of MA15 was compared with that of SARS-CoV (Urbani).The sequence of the initial SARS-CoV (P0) virus was identical to the published SARS-CoV (Urbani) sequence, and six nucleotide substitutions were identified in the MA15 genome compared with that of SARS-CoV (Urbani).All six substitutions were predicted to cause amino acid substitutions.These six mutations were localized to open reading frame (ORF) 1ab (four mutations) and ORFs S and M (one each) of SARS-CoV (Figure 1).Independent analysis conducted by The Institute for Genomic Research (Rockville, Maryland, United States), under a National Institute of Allergy and Infectious Diseases (NIAID) contract, confirmed the same six mutations in the MA15 sequence compared with those in the SARS-CoV (Urbani) sequence.

Generation of Recombinant Clones
To confirm that the six mutations identified in the MA15 virus were responsible for lethality in mice, the mutations were introduced into cDNA clones from which recombinant SARS-CoVs were recovered.Sequence analysis confirmed that recombinant viruses contained the appropriate mutation sets that were derived from the MA15 virus.In addition to generating a recombinant virus that included the six mutations (rMA15), two additional recombinants were generated containing either the two mutations in the structural protein genes or the four mutations in the ORF 1ab (rMA15 SM and rMA15 ORF1ab , respectively).

MA15 and Recombinant Viruses Replicate In Vitro with Similar Kinetics
The three recombinant viruses demonstrated similar kinetics and levels of viral replication compared with that

Author Summary
Severe acute respiratory syndrome (SARS) is a severe, sometimes fatal respiratory disease caused by a coronavirus (SARS-CoV).In order to study the disease and evaluate vaccines and antiviral drugs, animal models that mimic the disease are necessary.However, no single animal model for SARS reproduces all aspects of the disease as it affects humans.SARS-CoV replicates in the lungs of young mice, but they do not show signs of illness.Adaptation of SARS-CoV by serial passage in the lungs of mice resulted in a virus (MA15) that is lethal for young mice following intranasal inoculation.Lethality is preceded by rapid and high titer viral replication in lungs, viremia, and dissemination of virus to extrapulmonary sites accompanied by hematological changes and pathological changes in the lungs.Mice infected with MA15 virus die from an overwhelming viral infection with extensive, virally mediated destruction of pneumocytes, and ciliated epithelial cells.The MA15 virus has six coding mutations in its genome, which, when introduced into a recombinant SARS-CoV, confer lethality.The MA15 virus will enhance the use of the mouse model for SARS because infection with this virus in mice reproduces many aspects of severe human disease, including morbidity, mortality, and pulmonary pathology.
of SARS-CoV infectious clone (icSARS-CoV), a wild-type recombinant SARS-CoV (Urbani) generated from cDNAs, the biologically derived MA15 virus, and SARS-CoV (Urbani) in in vitro single-cycle growth analyses in Vero E6 cells.At a multiplicity of infection (MOI) of 0.1, viruses reached peak replication (10 7.0-7.5 pfu/mL) at ;24 h.p.i., with a slight delay in peak titers for rMA15 ORF1ab (Figure S1).Northern blot analysis of RNA from infected Vero E6 cells indicated that genomic vRNA and viral mRNA and all eight sub-genomic mRNAs were present in similar ratios for the recombinant viruses and MA15 virus as for SARS-CoV (Urbani) (Figure 2A).The level of expression and mobility of the structural proteins (S and N) and an accessory protein (ORF 3a, also called X1) of the recombinant viruses and MA15 virus were comparable to those of SARS-CoV (Urbani), as determined by Western blot analysis (Figure 2B).Although S, N, and X1 are somewhat reduced in the lane (Figure 2B) from icSARS-CoVinfected cultures, this is associated with reduced amounts of total protein added to the well, rather than with any specific reduction in X1 expression.Thus, the recombinant viruses (rMA15 SM , rMA15 ORF1ab , rMA15, and icSARS-CoV) and the MA15 virus demonstrate no defects in replication or in RNA or protein expression compared with SARS-CoV (Urbani) in Vero E6 cells.

Replication of MA15 and rMA15 in Mice
Groups of naı ¨ve mice were inoculated with serial 10-fold dilutions of MA15 virus, and the dose-dependent weight loss and lethality observed following infection is summarized in Table 1.Mice receiving a dose !10 3.9 TCID 50 of MA15 virus died or lost more than 20% of their initial body weight between days 3 and 5 p.i. Mice experiencing weight loss in excess of 20% initial body weight are euthanized in accordance with our animal study protocol.The 50% lethal dose (LD 50 ) was 10 4.6 TCID 50 .At an intermediate, non-lethal dose (10 2.9 /mouse), mice lost 8.4% 6 2.5% of initial body weight by day 4 p.i.At a lower dose, weight loss was not significant (4.3% 6 1.0% by day 4 p.i.) Similarly, groups of naı ¨ve mice were inoculated with serial 10-fold dilutions of rMA15, rMA15 ORF1ab , or rMA15 SM , and followed daily for signs of morbidity (indicated by weight loss) and mortality.Mortality was observed only in rMA15inoculated mice at doses equal to or in excess of 10 4.4 TCID 50 /mouse (Table 1).The LD 50 for rMA15 of 10 3.9 TCID 50 is similar to the LD 50 observed for the MA15 virus.Consistent with observations following infection with MA15 virus, mice inoculated with low concentrations of rMA15 (10 0.4-2.4TCID 50 /mouse) had little to no significant weight loss, but an intermediate dose of rMA15 (10 3.4 TCID 50 /mouse) resulted in significant weight loss but no mortality.Thus rMA15 recapitulated the phenotype of the MA15 virus in mice.Inoculation with the highest doses of rMA15 SM and rMA15 ORF1ab (10 6.2 TCID 50 /mouse) did not result in mortality (Table 2), and only rMA15 ORF1ab -infected mice demonstrated significant morbidity as indicated by weight loss.
The rapid lethality observed following administration of MA15 virus (or its recombinant clone rMA15) to BALB/c mice could result from changes in tissue tropism with or without viremia, increased viral load and subsequent necrosis in pulmonary or extrapulmonary tissues, failure of innate or early adaptive immune responses, immunopathology in pulmonary or extrapulmonary tissues, or a combination of these or other factors.In order to evaluate whether changes in tissue tropism or levels of viral replication could contribute to the lethal phenotype of the MA15 virus, viral titers in lungs, spleen, liver, and brain of BALB/c mice were determined at various time points following intranasal inoculation with SARS-CoV (Urbani) or MA15.

Efficient Replication of MA15 Virus in Pulmonary Tissues
Mice were inoculated intranasally with SARS-CoV (Urbani) (10 5 TCID 50 /mouse, a dose that results in peak viral titers in lungs of mice 2 d.p.i.) or with lethal (10 5.6 TCID 50 /mouse) or sub-lethal (10 3.6 TCID 50 /mouse) doses of MA15 virus.At various time points p.i., mice were sacrificed and tissues harvested for subsequent processing.Data from replicate experiments were combined.High titers of virus were detected in the lungs of mice through day 5 p.i. (Figure 3).One day after inoculation, the amount of SARS-CoV (Urbani) in lungs was ;10 6 TCID 50 /g tissue.Following administration of a lethal dose, MA15 virus was found to replicate to significantly higher titers of ;10 9 TCID 50 /g tissue within 24 h (p ¼ 0.0001; Figure 3).In this study and in previous studies (K.Subbarao, A. Roberts, L. Vogel, E. Lamirande, et al., unpublished data), peak virus titers (;10 7.0 TCID 50 /g tissue) occur at day 2 following inoculation with SARS-CoV (Urbani).In comparison, by day 2 p.i., viral titers in MA15-inoculated mice reached, or persisted at, titers of ;10 9 TCID 50 /g tissue, regardless of inoculating dose (Figure 3).Furthermore, in mice inoculated with a sublethal dose of MA15, virus was detected at 10 7.7 TCID 50 /g lung on day 5 p.i.By comparison, in SARS-CoV (Urbani)-infected mice, virus was detected at titers of 10 5.1 TCID 50 /g lung on day 5 p.i.By 6 d.p.i., virus was no longer consistently detected in mice inoculated with SARS-CoV (Urbani) or with sub-lethal doses of MA15 virus (Figure 3).(B) Western analysis.Cell lysates were separated on two 7.5% SDS-PAGE gels, transferred to polyvinylidene fluoride and probed with either mouse anti-S antisera (top panel) or probed first with a mouse anti-X1 antisera (sera raised to accessory protein X1 [37]; middle panel), and then stripped and probed again with a mouse anti-N antisera (bottom panel).Each primary antibody was followed by goat anti-mouse HRP-conjugated secondary antibody and visualized by enhanced chemiluminescence.doi:10.1371/journal.ppat.0030005.g002

MA15 Virus Detected in Extrapulmonary Tissues
Following inoculation with a lethal dose of MA15 virus, virus was also recovered from spleen, brain, and liver from day 1 through day 4 p.i. at titers ranging from 10 1.8 to 10 2.7 TCID 50 /g (brain) and from 10 1.8 to 10 3.3 TCID 50 /g (spleen) (Table 3).Virus was detected more frequently in the liver and at higher titers (10 2.1 -10 4.3 TCID 50 /g) than in brain or spleen.Virus was not recovered from any of these organs following inoculation with SARS-CoV (Urbani), except in one mouse, sacrificed 5 d.p.i., in which virus was detected in the spleen at a titer just above the limit of detection (Table 3).Although the levels of viral replication detected in these tissues in MA15-infected mice are modest, the number of mice in which virus was detected is remarkable, and both the frequency and the titers are significantly higher than those observed in SARS-CoV (Urbani)-infected mice.
In a separate experiment, infectious virus was also detected in sera of mice infected with MA15 virus or recombinant SARS-CoVs on days 2 and 4 p.i. Data from this experiment suggest that the presence of virus in serum may be facilitated by the mutations in the S and M genes, since virus was detected more frequently and at higher titers in sera from mice infected with MA15, rMA15, or rMA15 SM than in sera from mice infected with icSARS-CoV or MA15 ORF1ab (Figure S2).
vRNA Detected in Pulmonary and Extrapulmonary Tissues In addition to recovery of infectious virus from these tissues, total RNA was also isolated from whole blood, brain, kidney, liver, intestine, spleen, thymus, heart, and lungs of BALB/c mice after infection.Although quantitative virology holds biological relevance since it indicates viral burden by measuring infectious virus, RT-PCR of vRNA or sub-genomic mRNAs may be more sensitive assays for the presence of virus.Due to the replication strategy of SARS-CoV, primers that specifically amplify sub-genomic mRNAs allow detection of products of viral transcription.Furthermore, amplification of virus-specific mRNAs allows distinction from vRNA that may represent non-viable virus [6].RT-PCR amplification of genomic, antigenomic, and sub-genomic mRNA-specific sequences was employed to confirm the presence of SARS-CoV nucleic acid or viral transcription products in different tissues.
Groups of four mice were inoculated with a lethal dose of MA15 virus or with SARS-CoV (Urbani) (10 5 TCID 50 /mouse).Mice inoculated with MA15 virus did not survive past day 4 p.i. SARS-CoV (Urbani)-infected mice were followed through day 14 p.i. vRNA, and viral mRNA were amplified from lungs of MA15-infected mice on days 1-4 p.i.In contrast, vRNA was amplified from SARS-CoV (Urbani)-infected mice through day 14 p.i., but mRNA, indicating viral transcription, was consistently amplified only through day 5 p.i. mRNA was detected as late as day 7 p.i. by RT-PCR in one of four of the SARS-CoV (Urbani)-infected mice (Tables 4 and 5).
We were also able to amplify vRNA and mRNA from blood, brain, thymus, heart, and spleen of MA15-infected mice (Tables 4 and 5).Viral mRNA was consistently detected by RT-PCR in the tissues of MA15-inoculated mice from day 2 p.i. through day 4 p.i. and in the thymus and heart of MA15infected mice as early as day 1 p.i.In contrast, vRNA and viral mRNA were detected transiently in very few SARS-CoV (Urbani)-infected mice and only in the blood, spleen, and  thymus.vRNA and viral mRNA were not detected by RT-PCR analysis in liver, intestines, or kidneys from MA15 or SARS-CoV (Urbani)-infected mice.These tissues may contain factors that inhibit RT-PCR amplification of SARS-CoV RNAs, since virus was isolated from liver homogenates of MA15-infected mice but was not detected by RT-PCR, and viral mRNA was detected in intestines by in situ hybridization (unpublished data).In summary, virus or viral-specific RNA was detected in blood, lung, thymus, brain, spleen, liver, and heart from almost all MA15-infected mice.In contrast, virus or viral-specific RNA was detected in the lungs, and only sporadically in blood, spleen, and thymus from SARS-CoV (Urbani)-infected mice (Tables 3-5).

Histopathologic and Immunohistochemical Findings
The lungs of mice infected with SARS-CoV (Urbani) showed a rapid progression from focal, mild, perivascular, and peribronchiolar mononuclear inflammatory cell infiltrates to a diffuse, but transient, interstitial pneumonitis on day 3 p.i. (Figure 4A, 4C, 4E, and 4G).Occasional ciliated columnar epithelial cells of the bronchioles and alveolar pneumocytes stained for viral antigen by immunohistochemistry (IHC) (Figure 5A, 5C, 5E, and 5G).No viral antigens were detected at day 14 p.i., and most mice at this time point showed no significant pulmonary histopathology (unpublished data).
In comparison, the pulmonary pathology of mice infected with MA15 virus showed a rapid progression of inflammatory changes (Figure 4B, 4D, 4F, and 4H), but with more extensive damage to bronchiolar and alveolar epithelial cells (Figure 6A  and 6B).These changes were especially evident in pneumocytes, and many detached, pyknotic, and necrotic cells were identified in alveolar spaces (Figure 6A).Inflammatory infiltrates, although consistently identified, were generally similar in intensity to those observed in the lungs of SARS-CoV (Urbani)-infected mice (Figure 4).However, the distribution and amount of viral antigens identified by IHC was far greater in the lungs of MA15-infected mice than in the lungs of mice infected with SARS-CoV (Urbani) (Figure 5).Intracellular SARS-CoV antigens were extensively and abundantly distributed in bronchiolar epithelium, alveolar pneumocytes, and in necrotic debris within the alveoli and bronchiole lumens of MA15-infected mice (Figure 6C and 6D).
No significant histopathologic changes were identified in extrapulmonary organs, including the liver, spleen, thymus, or brain, of mice infected with SARS-CoV (Urbani) or MA15 virus.Furthermore, SARS-CoV antigens were not detected in any of these tissues by IHC staining.

Changes in Blood Counts and Chemistry in MA15-Infected Mice
Infection of young BALB/c mice with 10 6.4 TCID 50 /mouse of MA15 virus resulted in elevated levels of the liver enzyme alkaline phosphatase (ALP) in sera collected on days 1-6 p.i. ALP levels were significantly higher in sera of MA15inoculated mice than they were in pre-infection sera from the same mice.Serum ALP levels in MA15-inoculated mice were also significantly higher than those of mice inoculated with 10 6.4 TCID 50 /mouse of SARS-CoV (p ¼ 0.0045; Table S5).Levels of other liver enzymes, including aspartate aminotransferase, alanine aminotransferase, gamma glutamyl transferase, creatine kinase, urea nitrogen, total bilirubin, and albumin, were not significantly altered following inoculation with MA15 virus (unpublished data).Although moderately elevated levels of creatinine were observed in MA15inoculated mice, no significant differences were seen between MA15-and SARS-CoV-inoculated mice, suggesting that this was not associated with the lethal phenotype of the MA15 virus.Furthermore, significant lymphopenia and neutrophilia were observed following infection with SARS-CoV and MA15 virus.Although some samples were lost due to coagulation of blood, the alterations in the lymphocyte and neutrophil counts were more severe following infection with MA15 virus than they were following SARS-CoV infection (Table S5).

Primary Infection with SARS-CoV (Urbani) Protects Mice from Lethal Challenge with MA15 Virus
In order to determine whether MA15 virus can be used as a more stringent challenge in evaluating vaccines and antiviral therapy designed for SARS-CoV than could the non-lethal SARS-CoV (Urbani), we inoculated eight mice with 50 lL of SARS-CoV (Urbani) at 10 5 TCID 50 /mouse and mock-immunized an additional eight mice.Four weeks after inoculation, the mice were bled for determination of SARS-CoV-specific serum neutralizing antibodies and challenged with 10 6.9 TCID 50 of MA15 virus.Mice were followed daily for signs of  7).

Discussion
Serial passage of SARS-CoV (Urbani) in the lungs of BALB/ c mice resulted in a mouse-adapted SARS-CoV (MA15 virus) that is lethal for young (6-to 8-wk-old) BALB/c mice.The virulence and lethality of the MA15 virus result from six mutations in the SARS-CoV genome that occurred within 15 passages through BALB/c mice.Introduction of these six mutations into a recombinant SARS-CoV infectious clone, rMA15, conferred a lethal phenotype on the virus for young BALB/c mice.Five of these six mutations (not including the T67A mutation occurring in the non-structural protein [nsp] 9 of ORF 1ab) occur within nsp 5 (two mutations), nsp 13, S, and M.These genes have been reported as ones where sequence evolution occurred during adaptation of SARS-CoV in humans [7].The mutations in nsp 9, nsp 5 (H133Y and E269A within the main protease, 3CL pro ), and nsp 13 (A4V within the helicase protein) do not occur within any known functional domains and do not alter known cleavage sites utilized in the processing of the ORF 1ab polyprotein.Furthermore, the mutations in the MA15 virus do not occur at any specific amino acid positions identified by the Chinese SARS Molecular Epidemiology Consortium [7].Only the mutation in S (Y436H) occurs within a known functional domain, the receptor-binding motif (RBM).Other reports have indicated that mutations within the RBM may account for increased affinity of the virus for its cellular receptor angiotensin converting enzyme 2 [8].However, the Y436H mutation in MA15 virus does not occur at previously identified sites of the RBM and angiotensin converting enzyme 2 interaction, and preliminary findings suggest that the Y436H mutation does not increase binding of the SARS-CoV (Tor2) RBD to murine angiotensin-converting enzyme 2 [9].
Amplification of vRNA used primer pairs For28411 and Rev28725.doi:10.1371/journal.ppat.0030005.t004 Amplification of mRNA used primer pairs PCR-L and Rev28509.doi:10.1371/journal.ppat.0030005.t005 Reports of adaptation of an influenza A virus for increased virulence in mice indicate that multiple gene products may interact or contribute independently to virulence.In one adaptation of influenza A/FM/1/47 (FM-MA), findings indicated that four viral gene segments contributed to increased virulence [4], but additional analyses indicated that mutations occurring in at least two and likely three gene products acted synergistically to account for the increased virulence [10][11][12].Recombinant icSARS-CoV produced mild pneumonia identified by X-ray changes in macaques, similar to the clinical disease noted with wild-type SARS-CoV (Urbani) [13].Consistent with these findings, recombinant SARS-CoV bearing the six novel mouse-adapted mutations, rMA15, recapitulated a fatal respiratory disease phenotype in mice, demonstrating the ability of the SARS molecular clones to capture complex disease phenotypes.We generated two other recombinant SARS-CoVs, rMA15 SM (encoding the two mutations in the S and M genes) and rMA15 ORF1ab (encoding the four mutations in ORF 1ab).Neither of these was lethal (Table 2) in BALB/c mice, indicating that SARS-CoV adaptation for BALB/c mice involves mutations in at least two and possibly three genes ).Although recombinant viruses bearing two or four of the mutations were not lethal, each had a different phenotype than that of SARS-CoV (Urbani).It is possible that the mutations in S and M contribute to increased viremia (Figure S2) and that the mutations in ORF 1ab contribute to increased pathogenicity indicated by weight loss (Table 2).
Both quantitative virology and IHC analysis of the lungs of MA15-infected mice indicate extraordinarily high levels of viral replication in pulmonary tissues as early as 24 h.p.i., and the level of replication remains high through day 4 p.i. Unlike SARS-CoV-infected mice, in which viral antigen is detected by IHC staining at modest levels in bronchial epithelium and in alveolar pneumocytes on days 1 and 2 p.i., the bronchial epithelium of MA15-infected mice is replete with viral antigen at day 1 p.i., as are alveolar pneumocytes on days 1 and 2 p.i.By day 3 p.i., viral antigen is rarely detected in the In this model, day 3 to day 4 p.i. seems to be a critical time for the outcome of SARS-CoV infection.SARS-CoV (Urbani)-infected mice demonstrate a pronounced but transient interstitial inflammation at day 3 p.i. that is absent in MA15infected mice.This transient inflammation is not associated with significant weight loss but coincides with the beginning of viral clearance from the lungs.In contrast, by day 3 p.i., MA15 virus-infected mice lose weight, and necrotic cellular debris begins to fill the bronchioles and alveoli.By day 4 p.i., MA15-infected mice have lost in excess of 20% of their initial body weight, and several die without other overt clinical signs such as ataxia, paralysis, hunched posture, etc. Viral titers in lungs of MA15-infected mice are up to 1,000-fold higher than those in SARS-CoV-infected mice by 24 h.p.i. and remain higher than peak levels in SARS-CoV-infected mice through day 4 p.i.In situ hybridization further confirms the intense and persistent signal of MA15 (and rMA15) vRNA in lung tissues at day 4 p.i. when vRNA from SARS-CoV (Urbani) and the non-lethal rMA15 SM and rMA15 ORF1ab viruses are less abundant (Figure S3).Objective data related to respiratory distress such as plethysmography and measurement of blood gases could not be collected because of practical and logistical constraints on experiments carried out in an Animal Biosafety Level 3 laboratory.However, our findings indicate that MA15 virus-infected mice die as a result of overwhelming pulmonary viral infection and destruction of bronchiolar epithelial cells and alveolar pneumocytes.Viral load in SARS cases was an important determinant of severe disease and death [14], but the mechanism of disease leading to fatal outcome in human cases of SARS may be different than that observed in MA15-infected BALB/c mice.The mechanism of death following SARS-CoV infection in humans, and particularly the relative contribution of virusinduced damage and immunopathology, are not fully under-  stood.The histopathology documented in the mice infected with the MA15 virus includes a rapid progression and extensive damage to bronchiolar and alveolar epithelial cells (Figure 6A and 6B), but it does not show some features such as alveolar edema and hyaline membranes that were reported in many SARS-CoV-infected patients, or in aged mice infected with the Urbani strain of SARS-CoV [15].This can be explained by the relative virulence of the MA15 strain, and the time p.i. when the lungs were evaluated.Mice infected with the MA15 virus developed severe morbidity and died within 3 to 5 d following infection and did not survive long enough to show progression of the diffuse alveolar damage seen in aged mice or in human patients, who survive for a relatively prolonged length of time before succumbing to complications of the acute infection.In aged mice following SARS-CoV infection, histopathologic changes indicative of progressive diffuse alveolar damage, including proteinaceous deposits around alveolar walls and intraalveolar edema, are seen beginning on day 5 [15].
The rapid progression of infection and the extensive and persistent pulmonary replication of the virus are accompanied by viremia and detection of virus in extrapulmonary sites, suggesting that other factors may contribute to the increased pathogenicity of the MA15 virus.A prolonged viremic state or a secondary viremia is seen in mice infected with MA15 and rMA15 that is not observed following infection with the recombinant SARS-CoVs (Urbani), rMA15 SM , or rMA15 ORF1ab .The MA15 virus model captures viremia and multi-organ involvement noted in human SARS patients [16].However, as in all SARS cases, the primary site of infection is the lung.
Perfusion of tissues harvested from mice in these experiments was not performed, and therefore, cannot be certain that the detection of vRNA in various extrapulmonary sites in MA15-inoculated mice is not due to virus carried to and remaining in the organs and blood.However, on days 2 through 4 p.i., MA15-specific mRNAs in various tissues are detected at a higher frequency and signal intensity than in whole blood, and the presence of MA15 vRNA in extrapulmonary tissues is detected by in situ hybridization.Furthermore, infectious virus was detected on day 1 p.i. in homogenates of extrapulmonary tissues when RT-PCR of RNA from blood did not amplify any viral specific products (Tables 3-5).Taken together, these observations support a hypothesis that virus is present and replicating at low levels in extrapulmonary tissues.
In BALB/c mice infected with MA15 virus or SARS-CoV, significant lymphopenia and neutrophilia were observed compared with pre-infection levels.MA15 virus-infected mice had significantly greater lymphopenia and neutrophilia than SARS-CoV-infected mice, reflecting hematological evidence of increased disease severity following MA15 virus infection.MA15 virus-infected mice also had significantly higher levels of ALP than uninfected or SARS-CoV-infected mice.Elevated levels of ALP, an enzyme found in all tissues, but especially concentrated in the liver, may indicate cell destruction in the liver, intestines, or other tissues.Reports of elevated ALP levels in SARS patients are rare [17,18], but neutrophilia and lymphopenia have been reported frequently in human cases of SARS [19][20][21][22][23][24][25].Although steroid treatment and secondary bacterial infections may be speculated to be the cause of neutrophilia, some patients with SARS had neutrophilia prior to initiation of steroid treatment and in the absence of bacterial infections [26].Additionally, elevated absolute neutrophil counts at presentation were independent indicators of severe outcome of SARS in human cases [21,27,28].The definitive causes of lymphopenia, neutrophilia, and elevated ALP levels have not been identified in human cases of SARS or in this mouse model.
The mouse-adapted SARS-CoV MA15 virus will be a valuable tool in evaluating SARS-CoV vaccines and antiviral therapy.Quantitative virology was the only outcome measure available in young BALB/c mice challenged with SARS-CoV (Urbani).Because the MA15 virus replicates rapidly to high titer and is lethal for young BALB/c mice, this virus provides a more stringent challenge than the SARS-CoV (Urbani) virus in evaluating the efficacy of therapeutic interventions or prevention strategies.Prophylaxis that is able to rescue MA15-infected mice from a lethal outcome must lower viral burden in lungs very rapidly (e.g., within the first 24 h).Prophylaxis that accomplishes such reduction in the burden of MA15 virus may also reduce the severity of immunopathology that follows.A reduction of immunopathology was demonstrated in a SARS-CoV hamster model when a monoclonal antibody specific to SARS-CoV spike protein administered the day after infection was able to arrest a further rise in viral titer in the lungs [29].If similar protection can be demonstrated against challenge with MA15 virus in BALB/c mice, the efficacy of intervention will be well proven.
Infection of young BALB/c mice with the MA15 virus provides a model that is small and accessible and that can be evaluated extensively at an immunological level.Finally, and most importantly, MA15 virus infection of young BALB/c mice provides many elements that replicate observations in acute (and chronic) cases of SARS infection in humans, including sequence changes during adaptation in several genes (nsp 5, nsp 13, S, and M); viral replication and histopathological changes in lungs of infected animals; viremia; detection of vRNA in extrapulmonary sites, including the intestines; clinical indicators of illness, including mortality; and changes in blood counts, including lymphopenia and neutrophilia.The availability of a molecular clone of the MA15 virus, and the ability to generate additional SARS-CoV recombinant viruses that recapitulate the in vivo phenotype, will allow for detailed probing of the mechanisms mediating SARS-CoV pathogenesis and acute lung damage.

Materials and Methods
All animal and viral experiments were conducted in Biosafety Level 3 laboratories or animal facilities, and all personnel wore personal protective equipment, including Tyvek suits and hoods and positive pressure HEPA-filtered air respirators.All animal protocols employed in these studies have been approved by NIAID's Animal Care and Use Committee.

Figure 1 .
Figure 1.Schematic Diagram of SARS-CoV Genome Indicating Mutations Found in MA15 Virus (A) The 29,727 nucleotide positive-sense RNA genome of SARS-CoV is depicted in this to-scale drawing with ORFs indicated by shaded boxes (dark gray, structural and non-structural proteins; light gray, accessory genes X1-X5 [37]; and straight lines, non-coding regions).Asterisks indicate the sites of the six nucleotide changes (compared with the published SARS-CoV (Urbani) sequence) resulting in six coding mutations found in the mouse-adapted SARS-CoV (MA15).(B) The six mutations found in MA15. a ORF, open reading frame.b CDS, coding sequence, sequence of nucleotides that corresponds with the sequence of amino acids in a protein (location includes start and stop codon).c nsp, non-structural protein, cleavage product of ORF 1ab; Main pro , main 3C-like protease; Hel, helicase.d RBM, receptor binding motif (amino acids 424-494).doi: 10.1371/journal.ppat.0030005.g001

Figure 2 .
Figure 2. Recombinant SARS-CoVs Demonstrate Normal Processing of vRNAs and Proteins (A) Northern analysis.Intracellular RNA was isolated 10.5 h.p.i.from Vero E6 cells infected with indicated viruses or from mock-infected cells.RNA (0.1 lg) was treated with glyoxal, separated on 1% agarose gel, transferred to a BrightStar-Plus membrane, and probed with an N gene-specific biotinylated oligomer as described in Materials and Methods.(B)Western analysis.Cell lysates were separated on two 7.5% SDS-PAGE gels, transferred to polyvinylidene fluoride and probed with either mouse anti-S antisera (top panel) or probed first with a mouse anti-X1 antisera (sera raised to accessory protein X1[37]; middle panel), and then stripped and probed again with a mouse anti-N antisera (bottom panel).Each primary antibody was followed by goat anti-mouse HRP-conjugated secondary antibody and visualized by enhanced chemiluminescence.doi:10.1371/journal.ppat.0030005.g002

Figure 3 .
Figure 3. Virus Titers in Lungs of BALB/c Mice Inoculated with SARS-CoV or MA15 Virus Data represents a compilation of two experiments.In each experiment, groups of four mice were inoculated intranasally with 50 lL of SARS-CoV (Urbani) (10 5.0 TCID 50 /mouse, black bars) or MA15 virus at lethal (10 5.6 TCID 50 /mouse, white bars) or sub-lethal (10 3.6 TCID 50 /mouse, light gray bars) doses.Mice were sacrificed on indicated d.p.i.Mice receiving lethal doses of MA15 virus did not survive beyond day 4. Bars represent mean viral titers; error bars indicate standard error.Asterisks indicate significant differences (p , 0.05) compared with titers in mice receiving lethal doses of MA15 virus.Dotted line indicates lower limit of detection (10 1.5 TCID 50 /g).doi:10.1371/journal.ppat.0030005.g003

Figure 6 .
Figure 6.Histopathology and Immunohistochemical Localization of SARS-CoV Antigens in the Lungs of Mice Infected with MA15 Virus Abundant necrotic cellular debris (arrows) in alveoli (A) and a bronchiole lumen (B) of mice at days 2 and 3 p.i., respectively.Abundant SARS-CoV antigens (arrowheads) within alveolar pneumocytes (C) and in necrotic alveolar and bronchiolar cellular debris in mice at day 2 p.i. (D).(A and B) Hematoxylin and eosin stain; (C and D) primary antibody, rabbit anti-SARS-CoV antibody; secondary antibody conjugated with alkaline phosphatase with naphthol fast-red and hematoxylin counterstain; original magnifications3100.Mice were inoculated with 10 5.6 TCID 50 MA15 virus/mouse.doi:10.1371/journal.ppat.0030005.g006

Table 2 .
Morbidity and Mortality in Mice following Intranasal Administration of MA15 and Recombinant Viruses

Table 3 .
Detection of SARS-CoV and MA15 Infectious Virus PLoS Pathogens | www.plospathogens.orgmorbidity and mortality.Serum neutralizing antibodies were detected at low titers (1:10) in SARS-CoV-immunized mice but not in mock-immunized mice (,1:8).Nevertheless, SARS-CoV-immunized mice were protected from lethal challenge and demonstrated reduced morbidity (;11% weight loss compared with .20%weight loss in mock-immunized mice) following a high titer challenge with MA15 virus.Mockimmunized mice did not survive past day 4 following challenge with MA15 virus (Figure

Table 5 .
Detection of SARS-CoV and MA15 mRNA