![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Previous Article | Next Article 
Journal of Virology, July 2007, p. 7410-7423, Vol. 81, No. 14
0022-538X/07/$08.00+0 doi:10.1128/JVI.00505-07
Copyright © 2007, American Society for Microbiology. All Rights Reserved.
Department of Epidemiology, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina,1 Department of Microbiology and Immunology, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina,2 Carolina Vaccine Institute, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina,3 Department of Pathobiology and Diagnostic Investigation, College of Veterinary Medicine, Michigan State University, East Lansing, Michigan,4 Cystic Fibrosis/Pulmonary Research and Treatment Center, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina,5 University Health Network, University of Toronto, Toronto, Ontario, Canada6
Received 9 March 2007/ Accepted 3 May 2007
 |
ABSTRACT |
|---|
 Top ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The epidemic has been divided into zoonotic, early, middle, and late phases based on molecular epidemiological studies (7). The early phase is characterized by a series of independent cases, possibly of zoonotic origin. The middle phase is characterized by extensive local transmission of the virus, whereas global transmission to over 30 countries was seen in the late phase of the epidemic. Unfortunately, few if any zoonotic strains of SARS-CoV have been successfully isolated and maintained in culture, preventing their use in vaccine and pathogenesis studies.
Phylogenetic analysis, coupled with serological and epidemiological studies, has implicated Himalayan palm civets as a probable source for zoonotic transmission of SARS-CoV to humans (13, 23). However, recent studies showing a limited distribution of SARS-CoV in wild animals as well as the observation that palm civets show relevant clinical signs after experimental infection with SARS-CoV (61) make these animals an unlikely reservoir. Currently Chinese horseshoe bats are believed to be the most likely reservoir of SARS-CoV, with palm civets and raccoon dogs acting as intermediate hosts (31).
Comparative analyses of the SARS-CoV genomes of different isolates from both humans and animals throughout the different phases of the epidemic showed a high rate of evolution in the viral attachment protein, the spike (S) glycoprotein that was critical for the transition from animal-to-human to human-to-human transmission (7, 23, 42, 53). The S glycoprotein binds to the receptor angiotensin-converting enzyme 2 (ACE-2), mediating viral entry and establishing host range (30, 32). A receptor binding domain (RBD) has been defined. Not surprisingly, the S protein has also been identified as a major component of protective immunity (17, 56). Several vaccine candidates based on the S glycoprotein have been successfully tested in different animal models, showing complete protection against homologous SARS-CoV challenge (3, 54). In addition, passive transfer of antibodies directed against the S glycoprotein was successful in protecting mice and hamsters against subsequent challenge with homologous SARS-CoV (46, 55). Unfortunately, most vaccine candidates have been developed using nearly identical isolates from the late phase in the epidemic, and it is not clear whether these isolates will provide robust cross protection against zoonotic forms likely to emerge in the future, especially in vulnerable elderly populations (1, 10). In fact, studies using pseudotyped viruses bearing the S glycoprotein from different human and animal isolates demonstrated mixed results ranging from no cross-neutralization with human monoclonal antibodies (MAbs) and even enhanced infection to robust cross-neutralization (10, 16, 63). Given the variable responses of pseudotyped viruses, the goal of this study was to develop an isogenic panel of recombinant SARS-CoV isolates bearing variant epidemic and zoonotic S glycoproteins and evaluate the role of S glycoprotein heterogeneity in host tropism and pathogenesis. Developing robust animal models that recapitulate the complex age-related pathogenic phenotype, including ARDS, noted during the SARS-CoV epidemic in humans, is essential for deciphering the SARS-CoV pathogenic mechanisms. Recombinant viruses were derived using synthetic biology and reverse genetics to reconstruct a comprehensive panel of representative SARS-CoV viruses bearing variant S glycoproteins identified during the different phases of the epidemic.
|
MATERIALS AND METHODS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Growth curve analyses were performed using Vero E6 cultures inoculated with the different recombinant viruses at a multiplicity of infection (MOI) of 2 for 1 h and overlaid with medium. Virus samples were collected at various time points postinfection and stored at –70°C until viral titers were determined by plaque assay.
Virus titers were determined as PFU by plating six-well plates with 5 x 105 Vero E6 cells per well and inoculating cultures with 200 µl from 10-fold serial dilutions. Cells were incubated with virus for 1 h at 37°C and overlaid with 3 ml of 0.8% agarose in media. Plates were incubated for 2 days at 37°C, and plaques were visualized by staining with neutral red for 3 h and counted. Virus concentrations were calculated as PFU/ml.
All work was performed in a biological safety cabinet in a biosafety level 3 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.
Bioinformatic analysis of the SARS-CoV spike gene. The S gene of viral sequences representing early, middle, and late phases of the SARS-CoV epidemic in humans and zoonotic strains of SARS-CoV isolated from palm civets and raccoon dogs found in Chinese live-animal markets or housed on farms in China were aligned using ClustalX 1.83 with default settings (6). Molecular phylogenetic reconstruction was conducted using Bayesian inference as implemented in the program MrBayes v3.0b4 (21). Briefly, the alignment was exported in the nexus format, the amino acid substitution model was set to JTT (22) with the lset command, and Markov chain Monte Carlo simulation (15, 21) was used to approximate the posterior probabilities of trees, with sampling conducted on four chains over 500,000 generations (48). Trees were sampled every 100 generations, and the 5,001 trees collected were summarized with the sumt command set to a burnin of 1,000, which generated a consensus tree using the 50% majority rule (48). The burnin value was determined using the sump command with an arbitrary burnin of 250, which demonstrated that stationarity occurred prior to the 100,000th generation, indicating that a burnin of 1,000 was appropriate for the sumt command (48).
Five representative sequences were selected as representatives of important clusters, including the Urbani strain (GenBank accession no. AY27841), representing the late phase of the human epidemic, CUHK-W1 (GenBank accession no. AY278554), representing the middle phase, GZ02 (GenBank accession no. AY390556), representing the early phase, HC/SZ/61/03 (GenBank accession no. AY515512), representing zoonotic SARS-CoV isolated from a Himalayan palm civet, and A031G (GenBank accession no. AY687358), representing zoonotic SARS-CoV isolated from a raccoon dog.
Construction of recombinant viruses. Spike glycoprotein sequences for strains CUHK-W1 (GenBank accession no. AY278554), GZ02 (GenBank accession no. AY390556), HC/SZ/61/03 (GenBank accession no. AY515512), and A031G (GenBank accession no. AY687358) were used to replace the S glycoprotein of the Urbani strain (GenBank accession no. AY27841) in the E and F fragments of the icUrbani. For the CUHK-W1 and GZ02 strains, S glycoprotein synthetic DNA (Blue Heron Biotechnology, Bothell, WA), a BglI, starting at an NcoI restriction site at nucleotide position 21920 and ending at nucleotide position 24067, was purchased. The NcoI/BglI fragment was excised, purified, and ligated into the icUrbani-CoV E fragment, replacing the Urbani S glycoprotein sequence, and subclones were sequenced. For CUHK-W1 no additional nucleotide changes were needed in the C terminus of the S glycoprotein, which is coded in the F fragment of icUrbani. For GZ02, the F fragment of GD03T0013 (GenBank accession no. AY304486), a cDNA previously constructed in our laboratory (10), was used, as the sequences are identical in this domain.
For strain HC/SZ/61/03, a synthetic DNA was purchased that encodes nucleotide positions 21542 to 24067 containing the appropriate mutations and is flanked by AgeI and BglI restriction sites. The synthetic DNA was subcloned into the E fragment of icUrbani, and appropriate subclones were identified by sequence. An additional mutation in the F fragment was generated by overlap PCR mutagenesis. Amplicon A was generated using primer set BR29 (5'-GTGTTTAGTGGCACTTCTTG-3') and HC-673rv (5'-GAATAACCACCAATTTGGTA-3'), amplicon B was generated using HC-673fw (5'-CACAAGCAGCCCCGC-3') and BR30 (5'-CAAGAAGTGCCACTAAACAC-3'), and the overlap PCR was performed with primer sets BR29 and BR30. The product was subcloned into the F fragment of GD03T0013 by use of flanking BbsI restriction sites, and the sequence was verified.
To generate an A031G S gene, two additional mutations in the E fragment of HC/SZ/61/03 were generated by overlap PCR mutagenesis. Amplicon C was generated using primer set BR24 (5'-GCTGCAGCCTATTTTGTTG-3') and BR25 (5'-GCCATCAGAAGAGAAAGG-3'), amplicon D was generated using BR26 (5'-CCTTTCTCTTCTGATGGC-3') and BR27 (5'-CTAGGCATTGGCCATATTG-3'), and amplicon E was generated using BR28 (5'-CAATATGGCCAATGCCTAG-3') and Erv (5'-GAGCAGCCGTGTAGGCAGCAAT-3'). An overlapping product of amplicons C, D, and E was generated by use of primer set BR24 and Erv, the product was subsequently subcloned into the E fragment of HC/SZ/61/03 by use of flanking PstI and BglI restriction sites, and the sequence was verified. For A031G, no additional changes were needed in the F fragment of HC/SZ/61/03, as the sequences are identical in this domain.
Full-length cDNAs were constructed and transfection of full-length transcripts was performed as previously described (65). The transfected Vero E6 cells were seeded in a 75-cm2 flask and monitored for cytopathic effect (CPE). If the resulting icSARS-CoV isolate did not induce CPE at passage 0, cells were passed in Vero E6 cells until a strong CPE phenotype emerged. Viruses were plaque purified and intracellular RNA was isolated from infected cultures by use of TRIzol reagents as directed by the manufacturer (Invitrogen, Carlsbad, CA). The S gene was amplified by reverse transcription-PCR into two products by use of primer sets Leader (5'-AAAGCCAACCAACCTCGATC-3') and S2rv (5'-GGTCTCAAGCAATTGAACTATCAGC-3') to produce amplicon Spike-1, while SR2fw (5'-GGTCTCTTGCTTACTCTAATAACACC-3') and SRv (5'-GGTCTCCGTTTATGTGTAATGTAATTT-3') were used to amplify the Spike-2 products. Products were cloned into a pCR-XL-TOPO vector (Invitrogen, Carlsbad, CA) and sequenced. Stocks of viruses were grown in 150-cm2 flasks and stored at –70°C until further use.
ACE2 blocking assay. Vero E6 cultures were seeded at 5 x 105 cells/well in six-well plates. The cells were incubated with twofold dilutions of an anti-human ACE-1 or ACE-2 ectodomain antibody (AF929 or AF933; R&D Systems, Minneapolis, MN) starting at 10 µg/well for 1 h at 37°C. A control well without antibody was included on each plate. Cells were subsequently inoculated with 50 PFU of icUrbani, icCUHK-W1, icGZ02, icHC/SZ/61/03, or icA031G in the presence of antibodies. After a 1 h incubation period at 37°C, unbound virus and antibody was removed by washing the cells three times with phosphate-buffered saline (PBS). After the last wash, cells were overlaid with 3 ml of 0.8% agarose in media. Plates were incubated for 2 days at 37°C and stained with neutral red for 3 h, and plaques were counted. The percentage of blocking was calculated as follows: 1 – (number of plaques with antibody/number of plaques without antibody) x 100%.
Cathepsin inhibition assay. Vero E6 cells were cultured at 5 x 105 cells/well in six-well plates. The cells were incubated with twofold dilutions of cathepsin K inhibitor II, cathepsin L inhibitor III, or cathepsin S inhibitor (Calbiochem, San Diego, CA) starting at 10–4 M/well for 1 h at 37°C. A control well without a cathepsin inhibitor was included on each plate. Cells were subsequently inoculated with 50 PFU of icUrbani, icCUHK-W1, icGZ02, icHC/SZ/61/03, or icA031G in the presence of the respective cathepsin inhibitor. After a 1 h incubation period at 37°C, cells were overlaid with 3 ml of 0.8% agarose in media containing the respective cathepsin inhibitors. Plates were incubated for 2 days at 37°C and stained with neutral red for 3 h, and plaques were counted. The percentage of blocking was calculated as follows: 1 – (number of plaques with cathepsin inhibitor/number of plaques without cathepsin inhibitor) x 100%.
HAE cell cultures. Human ciliated airway epithelial (HAE) cell cultures were prepared and maintained as described previously (52). Briefly, freshly isolated tracheobronchial primary cells were expanded on plastic to generate passage 1 cells and then plated onto semipermeable Transwell-Col (12-mm-diameter) supports. HAE cell cultures were generated by provision of an air-liquid interface for 4 to 6 weeks to form well-differentiated polarized cultures that recapitulate the human pseudostratified ciliated airway epithelium.
HAE cell cultures from two different patients were inoculated via the apical surface with 106 PFU in 200 µl of each recombinant virus at a high (>2) MOI in duplicate. Prior to inoculation, the apical surfaces of HAE cell cultures were rinsed three times over 30 min with PBS at 37°C to remove accumulated secretions. Following a 2-h inoculation at 37°C, all inocula were removed and HAE cell cultures were maintained at an air-liquid interface for the remainder of the experiment. Samples for growth curve analysis were taken by incubating 500 µl of medium for 10 min at 37°C, collecting the medium, and storing at –70°C before plaque titration.
Histological sections of paraffin-embedded HAE cell cultures were prepared as previously described (52). Briefly, sections were incubated with mouse monoclonal antibodies raised against SARS nucleoprotein (immunoglobulin gG2a [IgG2a]) and acetylated alpha-tubulin (IgG2b). Immunoreactivity was visualized by use of secondary antibodies against mouse IgG2a conjugated to fluorescein and mouse IgG2b conjugated to rhodamine. Images were taken on a Leica fluorescence microscope using a tricolor filter as previously described (52).
Virus neutralization assay.
Three MAbs against Urbani (341CD, 283CD, and 540CD; kindly provided by L. J. Anderson, CDC, Atlanta, GA) and seven human convalescent-phase serum samples from patients infected with a late-phase isolate of SARS-CoV (kindly provided by D. Kelvin, Toronto General Research Institute, Toronto, Ontario, Canada) were used in a neutralization assay. Samples were collected at discharge from hospital from convalescent Toronto-area SARS patients enrolled without bias to age, sex, or previous medical history. SARS patients had been confirmed for SARS-CoV infection by positive PCR and/or seroconversion results. The cohort was comprised of three males and four females, with a median age of 39 years. Informed consent was obtained from all subjects with the approval of the Research Ethics Boards of the University Health Network and participating Toronto-area hospitals. Monoclonal antibodies or convalescent-phase serum samples were serially diluted twofold, starting at 500 µg/ml for the mouse MAbs, and incubated with 100 PFU of strain icUrbani, icCUHK-W1, icGZ02, icHC/SZ/61/03, or icA031G for 1 h at 37°C. Virus and antibodies were then added to a 96-well plate with 5 x 103 Vero E6/well at 10 wells per antibody dilution. Wells were checked for CPE at 4 to 5 days postinfection. A 50% neutralization titer was determined as the antibody dilution at which 
5 wells showed no CPE.
Mouse infection. Female BALB/c mice (6-week-old BALB/cAnNCrl mice from Charles River Laboratories and 12-month-old BALB/cBy mice from the National Institute on Aging) were anesthetized with a ketamine (1.3 mg/mouse)-xylazine (0.38 mg/mouse) mixture administered intraperitoneally with a 50-µl volume. Each mouse was intranasally inoculated with 105 PFU (strains icUrbani, icCUHK-W1, icGZ02, and icA031G) or 2 x 104 PFU (strain icHC/SZ/61/03) of virus in a 50-µl volume (highest dose possible for icHC/SZ/61/03). At 2 and 4 days postinfection, lung, spleen, liver, kidney, and brain serum and tissue samples were removed and frozen at –70°C for later plaque assay determination of viral titers. Lung, spleen, liver, kidney, and intestinal tissues were removed for histological examination.
All mice were housed under sterile conditions using individually ventilated Sealsafe cages and the SlimLine system (Tecniplast, Exton, PA). Experimental protocols were reviewed and approved by the Institutional Animal Care and Use Committee at University of North Carolina, Chapel Hill.
Virus titers in tissue samples. Tissue samples were weighed and homogenized in five equivalent volumes of PBS to generate a 20% solution. The solution was centrifuged at 13,000 rpm under conditions of aerosol containment in a table-top centrifuge for 5 min, the clarified supernatant serially diluted in PBS, and 200-µl volumes of the dilutions placed onto monolayers of Vero E6 cells in six-well plates. Following a 1-h incubation at 37°C, cells were overlaid with 0.8% agarose-containing medium. Two days later, plates were stained with neutral red and plaques counted.
Histology. All tissues were fixed in 4% paraformaldehyde in PBS (pH 7.4) prior to being submitted to the Histopathology Core Facility (University of North Carolina, Chapel Hill) for paraffin embedding, sectioning at 5-µm thickness, and hematoxylin and eosin staining. Approximately one-quarter of the total lungs were sectioned, with three sections mounted from cuts taken at three different depths within the paraffin-embedded tissue. The same sets of tissues were also evaluated qualitatively by a veterinary respiratory pathologist.
|
RESULTS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
Cell culture adaptation of variant SARS-CoV.
To determine whether tissue culture passage had selected for second-site mutations that enhanced the cytopathology of the zoonotic strains in vitro, the S genes from all rescued, plaque-purified recombinant viruses were cloned and sequenced. Sequence analysis revealed no codon changes within the S glycoprotein of isolates icUrbani, icCUHK-W1, and icGZ02. In contrast, a single amino acid change was identified at position 578 and at position 577 in icHC/SZ/61/03 and icA031G, respectively. To identify possible cell culture-adapting mutations in the other structural and accessory open reading frames (ORFs) the remaining 3' ends of the icSARS-CoV isolates were sequenced, revealing no additional amino acid changes throughout the 3' end of each recombinant virus genome (
8.5 kb).
Effect of Spike mutations on in vitro replication. The replication efficiency of recombinant viruses bearing heterologous S glycoproteins was evaluated in Vero E6 cells. Cultures were infected at an MOI of 2 and growth kinetics determined over the next 32 h. All viruses replicated efficiently in Vero E6 cells, achieving peak titers after 24 to 32 h postinfection (Fig. 2A). Growth kinetics of recombinant viruses bearing zoonotic S glycoproteins were delayed in comparison to those of the parental icUrbani control. The virus bearing the palm civet S glycoprotein (HC/SZ/61/03) achieved peak titers at 32 h postinoculation (p.i.) of only around 105 PFU/ml, a reduction of 1 log (t test; P < 0.05). Although they were delayed, the raccoon dog icA031G isolate displayed growth kinetics similar to those of epidemic strains whereas the icHC/SZ/61/03 isolate replicated less efficiently.
|
Effect of Spike mutations on receptor usage and entry. Previous studies have suggested that some zoonotic strains may use alternative receptors for docking and entry into the host cell (63). In order to determine whether the variant S strains of SARS-CoV used ACE-2 as a receptor, the receptor was blocked by human ACE-2-specific polyclonal antibodies prior to infection with icSARS-CoV bearing the zoonotic and epidemic S variants. Infection was completely inhibited by blocking ACE-2 on Vero E6 for all variants (Fig. 2E); these data suggest that zoonotic and epidemic strains all use ACE-2 as a receptor for entry into Vero E6 cells. No blockade was observed when cells were pretreated with polyclonal antibodies directed against ACE-1 as a negative control.
Recently, two studies demonstrated the dependence of SARS-CoV on the presence of cathepsin L for entry and infection of cells (19, 51). In order to determine whether cathepsin L was essential for epidemic and zoonotic virus entry into cells, cultures of Vero E6 cells were pretreated with cathepsin L inhibitor III prior to infection with the variant S panel. Infection of Vero E6 cells with SARS-CoV bearing variant spikes was highly sensitive to pretreatment with cathepsin L inhibitor (Fig. 2F). No obvious difference in infection results was observed, although both icGZ02 and icHC/SZ/61/03 were about a half-log more sensitive to cathepsin L inhibition than the other viruses. No block in virus entry was observed when using cathepsin K or S inhibitors as controls, demonstrating the specificity of the block to entry of the panel of S variants (data not shown).
Effect of Spike mutations on binding of neutralizing antibodies. Several of the S mutations fall within the RBD as well as other regions shown to elicit neutralizing antibodies (24, 28, 57). To study the effect of these mutations on the ability of monoclonal antibodies to neutralize both homologous and heterologous viruses, SARS-CoV isolates bearing the spike variants were incubated with murine monoclonal antibodies previously shown to neutralize the Urbani isolate (58). For all three MAbs, the highest neutralizing titers were directed against the homologous icUrbani isolate (Fig. 3). Similar titers were observed with the other two human isolates, icCUHK-W1 and icGZ02, and MAb 341CD. However, recombinant viruses bearing GZ02 S glycoproteins were not neutralized efficiently by 540CD or 283CD (analysis of variance [ANOVA]; P < 0.01). Not only were neutralization titers against the zoonotic spike variants significantly reduced but measurable values were observed only when using >10-fold concentrations of MAb. When convalescent-phase serum samples from patients infected with a late-phase SARS-CoV isolate were used, the homologous icUrbani variant was most potently neutralized (ANOVA; P < 0.01) (Fig. 3). Moreover, both the other human and zoonotic S variants were effectively neutralized using human convalescent-phase serum samples but were 5-to 10-fold more resistant that icUrbani.
|
|
|
Lung histopathology. Six-week-old mice that were infected with the strain icUrbani, icCUHK-W1, or icA031G S variants showed minimal or no bronchiolar or alveolar pathology, whereas mice infected with icGZ02 or icHC/SZ/61/03 had only minimal to mild bronchiolitis. Minimal or no alveolitis was present in the pulmonary parenchyma of these young mice at either 2 or 4 days p.i. (Fig. 6). The bronchiolar epithelium of infected young mice had only widely scattered individual cell degeneration and necrosis at 2 days p.i., with no obvious epithelial loss or attenuation. Some associated peribronchiolar and perivascular inflammatory cell infiltration (mainly of lymphocytes with lesser numbers of neutrophils) was also observed. By 4 days p.i. there was some microscopic evidence of epithelial regeneration, i.e., mild hypertrophy-hyperplasia and scattered intraepithelial mitotic figures.
|
|
|
|
DISCUSSION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
For use in a model system to develop synthetic genomics as a means of reconstructing viable viruses from sequence databases, the SARS-CoV outbreak is uniquely defined by a chronological set of sequence changes that span the epidemic strains, providing an unparalleled opportunity to identify the genetic basis for zoonotic virus cross-species transmission and pathogenesis during an expanding epidemic (7). Our phylogenetic studies, in agreement with earlier studies using fewer strains (7), identified a core set of conserved mutations that evolved and were maintained throughout each phase of the expanding epidemic. The early phase was initially characterized by a series of independent cases likely due to zoonotic transmission events (reviewed in reference 66). The middle phase refers to the major outbreak of SARS-CoV in a Guang Dong hospital, where a superspreader event was associated with more than 130 primary and secondary infections. The late phase started at hotel M in Hong Kong, from which virus was transmitted to Vietnam, Singapore, Canada, and the United States. Briefly, by comparing the early-, middle-, and late-phase human isolates (GZ02, CUHK-W1, and Urbani, respectively) to the palm civet isolate HZ/SZ/61/03 and the raccoon dog isolate A031G, 23 amino acid variations were noted in the S glycoprotein which varied in a coordinated time-sensitive fashion during the outbreak. Structural analysis has identified changes at positions L479N and T487S showing that these are key residues for efficient human ACE2 receptor usage (29, 32); however, the functional significance of the remaining 21 codon changes is unknown. In addition to the S glycoprotein mutations, it has been hypothesized that many replicase- and group-specific ORF mutations were also key to an expanding epidemic, and yet no defined roles have been described (7).
Recombinant SARS-CoV bearing variant spikes from zoonotic and early-, middle-, and late-phase human isolates were viable, formed plaques, and replicated efficiently in Vero E6 cells, although replication was delayed for the zoonotic strains. Zoonotic strains were incapable of producing extensive CPE but evolved a CPE phenotype with serial passage. Since RNA viruses are known for their high mutational rate, serial passage of SARS-CoV has been shown to select for tissue culture-associated mutations in the spike glycoprotein as well as other structural and nonstructural ORFs (43). In addition, the occurrence of a CPE phenotype for the SARS-CoV bearing zoonotic spikes after several passages indicates that one or more mutations likely evolved with passage in Vero E6 cells. Surprisingly, independent passage of the zoonotic isolates selected amino acid changes located at flanking amino acids 577 and 578. The S glycoprotein mutations mapped in a region of unknown function flanked by the fusion peptide and receptor binding domains roughly +200 and –77 codons away, respectively. Similar adaptations have been shown previously for another icSARS-CoV bearing the GD03 S glycoprotein at positions 7 and 613 (10). Adaptive mutations in the S glycoprotein have been implicated in coronavirus adaptation and extended host studies (9), although involvement of unknown mutations in ORF1a/ORF1b cannot be excluded. The lack of mutations in the S genes of strains icUrbani, icCUHK-W1, and icGZ02 as well as in the rest of the 3' end of the SARS-CoV genome indicates that the reverse genetic system can be readily used to introduce mutations in the S gene without subsequent reversion or extensive adaptation, making this an attractive tool for studying the effects of heterogeneity in the S gene in a stable genetic background of the Urbani isolate.
The similarity between the growth curve kinetics of the raccoon dog spike variant A031G and those of the human isolates suggests that the delay occurred at entry. This is in agreement with previous observations that binding efficiency of SARS-CoV to its receptor ACE-2 decreases dramatically in the presence of one or two amino acid changes within the RBD (32). The two amino acids (479N and 487T) are located in the binding interface of the S-ACE-2 complex (29), and while most of the early- to late-phase human isolates contain an asparagine and threonine, respectively, animal isolates contain an arginine (isolate HC/SZ/61/03; A013G) or lysine at position 479 and a serine at position 487, potentially interfering with interactions between human ACE-2 and the RBD of SARS-CoV S glycoprotein. Although mutations within the RBD may affect the efficiency of binding to ACE-2, all spike variant SARS-CoV isolates depended on the presence of ACE-2 for binding and entry in Vero E6 cells. In a more relevant model, HAE cell cultures mimicking the architecture of the human ciliated airway cells have been shown to express abundant levels of human ACE-2 on their apical surfaces, which perhaps even serve as an initial site of virus replication (52). Surprisingly, zoonotic S glycoprotein variants were inefficient at infecting HAE cell cultures despite the expression of the SARS-CoV receptor ACE-2 on target cells in the cultures (52). This is in agreement with the observation that zoonotic SARS-CoV isolates produce attenuated pathogenesis in humans (23) and suggests either a block postentry, as these viruses were shown to depend on ACE-2 and cathepsin L for binding and entry in a fashion similar to that seen with the human S variants, or insufficient availability of ACE-2 to initiate productive infections. Our data provide empirical support for the hypothesis that most zoonotic strains harbored in palm civets and raccoon dogs would likely require additional rounds of evolution and change to promote efficient replication and human-to-human transmission (23).
Studies focusing on virus-receptor interactions and blocking thereof have mainly used a limited subset of pseudotyped viruses expressing the different SARS-CoV S glycoproteins (14, 38, 63). Although pseudotyped viruses allow for the easy introduction of mutations into the SARS-CoV S gene and present with limited biosafety issues, this system remains artificial and structural differences in the S glycoprotein cannot be excluded. Potential problems with these systems are demonstrated by contradicting observations of antibody-mediated enhancement (16, 63, 64). By using the reverse genetics systems developed for Urbani SARS-CoV to introduce the variant S genes from the zoonotic and human isolates, we ensured the proper structural expression of the S glycoprotein on the virions and demonstrated that all mutants tested used ACE-2 and cathepsin L efficiently to mediate entry into susceptible Vero E6 cells.
Currently a variety of animal models exist for studying SARS-CoV, including mice, ferrets, hamsters, cats, and nonhuman primates (27, 34, 35, 47, 55). In young BALB/c mice, peak virus replication occurred on day 1 or 2, with virus clearance in a week in the absence of any clinical disease or pathology (55). For senescent mice, virus titers are increased and mild pathological lesions are noted in the lungs of infected animals at 4 days postinfection (45). Some animals may experience limited weight loss, but symptoms are mild overall. As none of these models recapitulate the complex pathogenic phenotype noted during the SARS-CoV epidemic in humans, animal models that are more robust are needed to evaluate the efficacy of SARS-CoV vaccines and therapeutics, especially in the more vulnerable elderly populations. The recent adaptation of the Urbani isolate to mice by serial passage of the virus in the respiratory tract of young BALB/c mice resulted in a virus that is lethal in young BALB/c mice after intranasal inoculation and that showed extensive pathological changes (44). Histopathologic changes after inoculation with this mouse-adapted virus included damage to bronchiolar and alveolar epithelial cells but not the alveolar edema and damage to hyaline membranes that were reported for many SARS-CoV-infected patients (12, 18, 44). During passage the virus did, however, acquire six coding mutations, including one in the S glycoprotein, all of which were necessary for the lethal phenotype (44). Two other groups described lethal SARS-CoV Urbani models of mouse infection involving development of transgenic mice expressing human ACE-2 (36, 60). Although high titers of SARS-CoV could be found in lungs of the transgenic mice, infection rapidly spread to the central nervous system, resulting in encephalitis and death. These models do not recapitulate the novel age-related phenotype of SARS pathogenesis or the principle histological changes of diffuse alveolar damage, hyaline membrane formation, respiratory failure, and death.
In this study, we developed two new lethal SARS-CoV challenge models employing mice with SARS-CoV isolates bearing S glycoproteins from an early human isolate (GZ02) and a palm civet isolate (HC/SZ/61/03), making use of mutations in the S glycoprotein that naturally occurred during the epidemic. Three amino acid changes in the S glycoprotein of SARS-CoV were associated with a lethal phenotype in aged mice, including one each in the receptor binding domain, the putative FP region, and the HR-2 region. Mutations in the HR and FP domains of mouse hepatitis virus have previously been shown to affect host ranges in vitro and pathogenesis in vivo (9, 59). Mice infected with viruses bearing these three mutations showed clinical signs, lost significant amounts of weight, and died by day 4. In addition, virus replication was primarily restricted to the lungs, with low levels of replication found in peripheral organs like the liver, spleen, and kidney but not the brain. Surprisingly, peak viral titers in the lung were not higher in mice infected with the lethal isolates; in fact, mice infected with the lethal icHC/Z/61/03 isolate had titers in the lungs that were 1 to 2 logs lower. This discrepancy between viral titers and increased pathogenesis may be explained by observations that SARS is characterized by immunopathological damage caused by a "cytokine storm" (20) and not by direct pathological damage due to viral replication.
The pathology observed in mice infected with these two lethal S glycoprotein variants was very similar to the pathology observed in acute human cases of SARS-CoV. Mice died of ARDS characterized by DAD and hyaline membrane formation within 3 to 4 days after intranasal infection (8, 12). Both the development of ARDS and the increased pathogenic phenotype in older individuals are in complete agreement with data from human SARS-CoV cases (5, 33). Two phases have been identified during SARS-CoV infection in humans (41). ARDS develops within the first 10 days with DAD, edema, and hyaline membrane formation (39, 40). After the acute phase, organizing DAD with increased fibrosis is observed (40). ARDS has been shown to be the major cause of death in cases of human infection by SARS-CoV and avian influenza A (H5N1) (39) and likely the 1918 influenza virus (25). Robust model systems designed to study the complex pathogenesis of virus-induced ARDS are rare (25, 62) and are limited to select influenza viruses and the two SARS-CoV models described in the current report; these models provide a rare opportunity to study the complex virus-host interactions that contribute to this severe acute pathogenic phenotype, especially in association with aging.
To our knowledge these are the only lethal SARS-CoV infection models using mice and the SARS-CoV spike glycoprotein isolates that occurred naturally during the outbreak and that recapitulate the age-related pathogenic phenotype. Further studies into the pathogenesis of these lethal S glycoprotein variants of SARS-CoV are under way.
This panel of lethal viruses will be invaluable for testing vaccine candidates and monoclonal antibodies for their ability to protect and/or neutralize heterologous SARS-CoV isolates. To date, several vaccine candidates have been developed and tested in different animal models. However, all of these studies, except one, have focused on vaccination and challenge with the homologous virus strain most often isolated from the late phase of the SARS-CoV epidemic (3, 10, 49, 54). By the use of a recombinant virus bearing a GD03 S glycoprotein, vaccine efficacy was reduced in aged animals, suggesting that more-rigorous testing of vaccines in models reflecting the vulnerability of elderly populations is central to improved public health (10). In addition, SARS-CoV isolates bearing zoonotic S glycoproteins were only neutralized at high concentrations of monoclonal antibodies which readily neutralize the human isolates. This is in agreement with other studies showing limited cross-neutralization when using pseudotyped viruses (16, 63). The absence of human cases of SARS-CoV in the last two years suggests that a new outbreak of SARS-CoV would likely result from zoonotic transmission (2). Therefore, vaccines against SARS-CoV should protect against a wide variety of strains, particularly from the zoonotic and early epidemic phases, to prevent further transmission and adaptation.
Our data show that the S glycoprotein of SARS-CoV can be replaced by variant S glycoproteins, thereby allowing construction of a panel of SARS-CoV isolates bearing S glycoproteins from different phases of the epidemic of 2002 and 2003. We developed two new mouse models for heterologous SARS-CoV infection that reflect human cases of SARS by both age dependence and pathological changes. As immune senescence is a critical mediator of disproportionate disease burdens in the elderly (37), we believe that this panel will be a powerful tool for studying the complex age-related relationship to ARDS and the effect of spike glycoprotein heterogeneity on virus-receptor interactions, host-interactions, tropism, and pathogenesis as well as an essential model for successful vaccine development for aged populations.
|
ACKNOWLEDGMENTS |
|---|
This work was supported by NIH grants P01-AI059443 and -AI059136.
|
FOOTNOTES |
|---|
Published ahead of print on 16 May 2007.
|
REFERENCES |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
with dashed black line, icCUHK-W1; x with dotted black line, icGZ02; 
with solid gray line, icHC/SZ/61/03; 
with dotted gray line, icA031G. Error bars represent standard deviations.
