An outbreak of severe acute respiratory syndrome (SARS) occurred in Guangdong Province, People's Republic of China, in November 2002 (
39). From China, SARS spread to 30 other countries and as of September 23, 2003, this outbreak had resulted in 8,098 reported cases, of which 774 were fatal (
http://www.who.int/csr/sars/country/table2003_09_23/en/ ). Through the coordinated efforts of laboratories around the world, a novel coronavirus (CoV), designated SARS-coronavirus (SARS-CoV), was identified as the causative agent of SARS (
6,
9,
14,
24,
25). This discovery was quickly followed by the publication of the complete genomic sequences of two SARS-CoV isolates and identification of specific subgenomic RNAs (sgRNAs) and proteins involved in replication (
19,
28,
33). The origin of SARS-CoV has not been determined, but evidence strongly suggests that its emergence may be the result of zoonotic transmission(s) (
10).
CoVs (order Nidovirales, family
Coronaviridae) are diverse, enveloped, positive-stranded RNA viruses that produce a nested set of sgRNA molecules during replication (
11,
16). CoVs are placed into group 1, 2, or 3 based on antigenic and genetic criteria (
11,
16). The genome of SARS-CoV has considerable nucleotide divergence from that of other known human CoVs (HCoVs) (
14,
19,
28). However, phylogenetic analysis of the SARS-CoV replicase gene demonstrated that, despite a number of unique features, SARS-CoV is most closely related to group 2 CoVs (
32). Group 2 includes mouse hepatitis virus (MHV), bovine coronavirus, and HCoV-OC43. The CoV genome, approximately 27 to 32 kb in length, is the largest found in any of the RNA viruses. Large spike (S) glycoproteins protrude from the virus particle, giving CoVs a distinctive corona-like appearance when visualized by electron microscopy. The S protein is the major viral-attachment protein, critical to virus binding and fusion of the viral envelope with cellular membranes. CoVs infect a wide variety of species, including dogs, cats, pigs, mice, cows, birds, and humans (
11,
16). However, the natural host range of each CoV strain is narrow, typically consisting of a single species (
11,
16).
CoVs enter cells by receptor-mediated endocytosis or by fusion with the plasma membrane (
11,
16). The S protein-receptor interaction is a major determinant of species specificity and tissue tropism for both group 1 and group 2 CoVs. This finding is best illustrated by the fact that CoV genomic RNA (gRNA) is infectious when transfected into nonpermissive cells and that transfection of nonpermissive cells with constructs expressing CoV receptors renders them susceptible to infection (
4,
8,
11,
13,
16,
36). The receptor for group 1 CoVs, including HCoV-229E, feline CoVs, and porcine CoVs, is aminopeptidase N (APN) (
4,
34,
38). Although APN is highly conserved, it is generally used in a species-specific manner (
13,
35). CEACAM1a, the best-characterized CoV receptor and a principal determinant of the tissue tropism and restricted host range of MHV (
3,
7,
8,
30), is utilized by different strains of MHV, which is a group 2 CoV (
11).
SARS-CoV was first isolated in African green monkey kidney (VeroE6) cells and fetal rhesus monkey kidney (FRhMK) cells inoculated with clinical specimens (
6,
14,
24,
26). Based on cytopathic effect (CPE), other cells routinely used for identification of respiratory pathogens (MDCK, A549, NCI-H292, HeLa, LLC-MK2, Hut-292, B95-8, MRC-5, RDE, and Hep-2) were determined to be nonpermissive to SARS-CoV infection (
6,
14,
24). Additionally, human peripheral blood mononuclear cells (PBMCs) were shown by reverse transcriptase PCR (RT-PCR) to support SARS-CoV replication (
17). To determine the in vitro host range, analyze potential receptors, and identify additional human cell lines permissive to SARS-CoV, a multiplex RT-PCR assay for the detection of SARS-CoV replication was developed. Cells routinely used by clinical diagnostic laboratories for pathogen screening were specifically analyzed to determine their susceptibility to SARS-CoV. Additionally, primary cells and continuous cell lines derived from a number of species and tissues were analyzed for their susceptibilities to SARS-CoV. This study identified novel human and animal cells that support SARS-CoV replication.
DISCUSSION
This study was designed to analyze the species specificity and tissue tropism of SARS-CoV in vitro and to identify novel human and animal cells that are susceptible to SARS-CoV infection. The multiplex RT-PCR assay described rapidly and specifically detects early replication of SARS-CoV RNA and is an indirect measure of viral entry (Fig.
1B). Importantly, this assay can identify cells that are susceptible to SARS-CoV entry and RNA transcription yet result in minimal CPE or that have blocks between RNA transcription and the efficient production of progeny viruses (Table
1). Previously described RT-PCR procedures detect SARS-CoV gRNA; however, the detection of sgRNA production described here provides an effective approach to differentiate input virus from replicating virus (
6,
25,
26,
37). This assay may be useful in a diagnostic setting and for the analysis of animal models. In addition, it could be used to screen for potential inhibitors of viral replication.
Previously published reports predicted that all monkey kidney cells would be susceptible to SARS-CoV (
6,
14,
24,
26). In agreement with previous findings, the data presented here show that kidney cells derived from three species of monkey (African green monkey, rhesus macaque, and cynomolgus macaque) are productively infected with SARS-CoV. Fouchier et al. and Kuiken et al. demonstrated that cynomolgus macaques inoculated with SARS-CoV develop clinical symptoms consistent with infection (
9,
15). Although SARS-CoV was not detected in the kidneys of these animals by immunohistochemical techniques, our data coupled with work by Ksiazek et al. suggests that the kidney supports SARS-CoV replication (
14). Evaluation of these tissues with more-sensitive methods may identify SARS-CoV in the kidneys of experimentally infected animals. Interestingly, infection of pCMK and pRhMK cells resulted in lower viral titers than infection of VeroE6 cells. The reason for these results remains to be determined, but a potential explanation is that VeroE6 is a cell line deficient in interferon production, whereas pCMK and pRhMK are both primary cell populations. Additionally, pCMK and pRhMK are both mixed cell populations; thus, the cells susceptible to SARS-CoV may make up only a percentage of the total cell population.
At the beginning of the SARS-CoV outbreak, panels of cells were inoculated with clinical specimens to identify the causative agent of SARS (
6,
14,
24). Based on CPE, VeroE6 and FRhMK cells were identified as susceptible to SARS-CoV infection (
6,
14,
24). However, many CoVs can establish persistent infection in cells without inducing CPE, and we have detected replication of SARS-CoV in the absence of CPE (
11). The analysis of SARS-CoV infection of cells that are routinely used by clinical virology laboratories is critical from both biological safety and diagnostic standpoints. The clinical presentation of SARS-CoV infection is very similar to that of influenza virus and other respiratory pathogens. Given the possibility for SARS-coronavirus re-emergence to occur during influenza or other respiratory disease seasons, it is important to identify cells used by clinical laboratories that support SARS-CoV replication so that SARS-CoV is not inadvertently produced without appropriate biosafety precautions. A panel of cells used to isolate respiratory viruses was analyzed for susceptibility to SARS-CoV. MRC-5, HEL, and A549 cells, lung-derived cells used to screen specimens for rhinovirus, RSV, adenovirus, and influenza virus, were nonpermissive to SARS-CoV replication based on the multiplex RT-PCR assay. However, R-Mix, a mixed cell population used for the detection of influenza A and B viruses, RSV, adenovirus, and parainfluenza viruses 1, 2, and 3, was found to be productively infected by SARS-CoV. We identified Mv1Lu as the cell population within R-Mix that supports productive SARS-CoV infection. Despite the low yield of SARS-CoV in Mv1Lu and R-Mix cells, their widespread use in clinical diagnostic laboratories warrants further work to devise methods of enhancing their safety. Ongoing approaches include the incorporation of SARS-CoVspecific monoclonal antibodies or antiviral agents into the maintenance medium and determining the effect medium components have on SARS-CoV replication. Diagnostic laboratories utilizing R-Mix and Mv1Lu for the detection of respiratory specimens perform inoculations in medium that contains different components from those of the medium used in this study (e.g., it lacks FBS and contains trypsin). Additional evaluation of the susceptibility of these cell lines is being carried out under conditions utilized by clinical laboratories, as they may provide enhanced safeguards against amplification of SARS-CoV. Preliminary evidence strongly suggests that the passage history and/or medium components during passage or infection alter the production of SARS-CoV by Mv1Lu cells (data not shown). Our results demonstrate the importance of conducting further work to analyze cells used to diagnose illnesses that have clinical symptoms similar to those of influenza virus and SARS-CoV infection. For example, rhesus monkey kidney cells (LLC-MK2), used for the detection of rhinovirus, did not show CPE when inoculated with a clinical SARS-CoV specimen (
14). However, both pRhMK and FRhMK cells are susceptible to SARS-CoV, suggesting that LLC-MK2 cells may also be susceptible (
24). The discovery that Mv1Lu cells are permissive to SARS-CoV infection is also important in the development of safer diagnostic assays, identification of receptors, and understanding of the virus life cycle and also suggests that mink and other related species are potential animal models or natural reservoirs (
10,
20).
To investigate the possibility that cellular receptors utilized by other CoVs could function as receptors for SARS-CoV, cell lines known to be permissive to CoVs from groups 1, 2, and 3 were analyzed by multiplex RT-PCR for SARS-CoV RNA replication. APN (also called CD13), is a 150-kDa metalloprotease that serves as a receptor for related CoVs of humans (HCoV-229E), pigs (transmissible gastroenteritis virus and porcine respiratory coronavirus), and cats (feline infectious peritonitis virus and feline coronavirus) (
4,
34,
38). APN is expressed in tissues such as the liver and kidney, as well as on granulocytes, monocytes, dendritic cells, and endothelial cells and by epithelial cells of the intestine and respiratory tract, where it allows initial entry of CoVs (
22,
27). To examine the potential involvement of hAPN in SARS-CoV entry, we inoculated human lung cells, canine kidney cells, and feline lung cells that are productively infected by HCov-229E, canine CoV, and feline CoVs, respectively. In agreement with Ksiazek et al., no evidence of SARS-CoV entry was observed (
14). Furthermore, cells derived from two other species (mouse and hamster) expressing high levels of hAPN and previously demonstrated to be susceptible to HCoV-229E were also demonstrated to be nonpermissive to SARS-CoV (
35). These results show that hAPN is not sufficient for entry of SARS-CoV. Similar results were found by K. V. Holmes and collaborators (personal communication). SARS-CoV is most closely related to group 2 CoVs, suggesting that it may use a receptor utilized by another group 2 CoV (
32). However, murine cells expressing CEACAM1a (L2 and CMT-93), the receptor for MHV, and HRT-18 cells, which are susceptible to HCoV-OC43 and bovine CoV, were both found to be nonpermissive to SARS-CoV infection. Finally, avian cells susceptible to a group 3 avian CoV (CEF) were also found to be nonpermissive for SARS-CoV infection. These data are in agreement with a recent study that shows that angiotensin-2-converting enzyme is a functional receptor for SARS-CoV (
18).
The SARS-CoV outbreak affected the human population, yet human cell lines productively infected by SARS-CoV were not demonstrated prior to this study. We examined human cell lines derived from lung (HEL A549 and MRC-5), intestine (HRT-18), kidney (HEK-293T), and liver (Huh-7). HEL, MRC-5, and HRT-18 cells were all nonpermissive to SARS-CoV infection; however, we identified two susceptible human-derived cell lines, HEK-293T and Huh-7. Huh-7 is a hepatocellular carcinoma cell line that is widely utilized to study hepatitis B and C viruses. Our results indicate that Huh-7 cells are productively infected by SARS-CoV and that they produce higher levels of virus than do HEK-293T cells. Interestingly, Huh-7 cells were previously shown to be susceptible to the DL and DS variants of MHV-JHM, a neurotropic group 2 CoV (
12). Furthermore, infection of Huh-7 cells by JHM was mediated by the S protein (
12). A Blastp GenBank search with the predicted amino acid sequence of the SARS-CoV S protein reveals that it shares the most amino acid similarity with the S protein of MHV-JHM (
1,
31). MHV-JHM utilizes murine CEACAM1a as a primary receptor, but it is one of very few isolates of MHV that has shown a broader host range (
2,
12,
21,
30), which leads to the possibility that conserved amino acids or motifs in the S protein of JHM and SARS-CoV interact with the receptor on host cells.
The susceptibility of HEK-293T cells correlates with the susceptibility of primate kidney cells to SARS-CoV and the isolation of SARS-CoV from the kidney of an infected patient (
14). While HEK-293T and Huh-7 cells are susceptible to SARS-CoV infection, HEK-293T cells do not produce high titers of virus when inoculated with a low multiplicity of virus. The reason for this low level of virus production remains to be determined but may be due to low-level surface expression of angiotensin-2-converting enzyme (
18). Like Mv1Lu cells, HEK-293T cells may provide a safer diagnostic alternative to VeroE6 cells, which produce >10,000-fold-greater titers of SARS-CoV.
The data presented here are important to vaccine production, diagnostic assay development, and elucidation of animal models and reservoirs susceptible to SARS-CoV. Furthermore, the data suggest that other human and animal cells are likely to be susceptible to SARS-CoV upon examination and that laboratories may need to adopt different practices to prevent accidental exposure to SARS-CoV during routine screening of clinical specimens.
Coronaviruses tend to be species specific, yet SARS-CoV appears to have crossed species barriers and infected humans, resulting in high morbidity and mortality (
10,
23,
39). Our in vitro data, in conjunction with data from multiple animal studies, show that SARS-CoV has a broad host range (
9,
10,
15,
20), which suggests that SARS-CoV or closely related variants may be circulating in multiple animal reservoirs, increasing the likelihood of its re-emergence in humans.