VIRAL IMMUNOLOGY Volume 16, Number 4, 2003 © Mary Ann Liebert, Inc. Pp. 461–474 Review SARS: Lessons Learned from Other Coronaviruses SONIA NAVAS-MARTIN and SUSAN R. WEISS ABSTRACT The identification of a new coronavirus as the etiological agent of severe acute respiratory syndrome (SARS) has evoked much new interest in the molecular biology and pathogenesis of coronaviruses. This review summarizes present knowledge on coronavirus molecular bi- ology and pathogenesis with particular emphasis on mouse hepatitis virus (MHV). MHV, a member of coronavirus group 2, is a natural pathogen of the mouse; MHV infection of the mouse is considered one of the best models for the study of demyelinating disease, such as multiple sclerosis, in humans. As a result of the SARS epidemic, coronaviruses can now be considered as emerging pathogens. Future research on SARS needs to be based on all the knowledge that coronavirologists have generated over more than 30 years of research. INTRODUCTION S EVERE ACUTE RESPIRATORY SYNDROME (SARS) appears to have originated in Guangdong Province, China, in late 2002. It spread to locations in Asia including Hong Kong, Vietnam, Singapore, and Taiwan and in North America, most notably to Toronto. By the end of the epidemic, the CDC and WHO were report- ing more than 8000 cases, with more than 800 deaths and some evidence of recurring disease (50). At pres- ent, there is no effective treatment for SARS. The epidemic has been controlled essentially by isolation. Given the fast spread of the disease, the high mortality rates (.10%, up to 40% for certain age groups), the lack of any treatment and the uncertainties of whether the disease will return and at what frequency, development of a SARS vaccine as well as anti-viral strategies are of a high priority. A newly identified coronavirus was isolated from SARS patients (50); proof that this virus is the etio- logic agent for SARS was provided by results of infections carried out in non-human primates, in which Koch’s postulates were fulfilled (33). Several groups of scientists, including those at CDC, have quickly sequenced the genome of this new coronavirus referred to as SARS-CoV (81,87); the sequence informa- tion demonstrated that this is a previously unrecognized coronavirus. The identification of a new coron- avirus as the etiological agent of SARS has evoked much new interest in the molecular biology and patho- genesis of coronaviruses. Coronaviruses form a group of pathogenic, enveloped RNA viruses that have the largest genome (around 30 kb) among RNA viruses. In 1931, Schalk and Hawn published a description of “a new respiratory tract 461 Department of Microbiology, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania. disease affecting baby chicks” (82). That discovery lead to a series of reports on animal diseases, such as the “hepatitis virus of mice” reported by Gledhill et al. in 1951 (37). The name coronavirus was introduced in 1968 by Tyrrel et al. (96); it was based on the “corona”-like morphology of these viruses observed by electron microscopy. However, it was not until 1975 that the Coronaviridae family was established by the International Committee on the Taxonomy of Viruses. Coronaviruses are divided into three antigenic groups (termed groups 1, 2, and 3), that infect many species of animals including humans (63) (Table 1). Corona- viruses infect mainly mucosal surfaces, causing respiratory and enteric diseases that may account for im- portant economic loss, particularly in the case of porcine and bovine coronavirus infections. Of note, some coronaviruses induce hepatic and central nervous system diseases (mouse hepatitis virus, MHV) and even systemic infections (feline infectious peritonitis virus, FIPV). Under experimental conditions, susceptibil- ity to coronavirus infections depends on factors such as the genetic background, age of the host, virus strain, dose, and route of inoculation (55,86). Until recently, there were two prototype human coronaviruses, HCoV- 229E and HCoV-OC43, which, together cause about the 30% of the common colds (6,62). The newly iso- lated SARS-CoV is related in sequence to other known coronaviruses. SARS-CoV has been suggested to define a new fourth antigenic group of viruses (81); however, one report suggests it may be a related mem- ber (“early split-off”) of group 2 coronaviruses (87) along with the prototype coronavirus, murine hepati- tis virus (MHV) and the human cold virus OC43. SARS-CoV is most closely related in sequence to group 2 coronaviruses, including murine coronavirus, mouse hepatitis virus (MHV), human coronavirus OC43 and bovine coronavirus (BCV) (47,81,87,98,99). This would suggest that the information gathered with MHV and other coronaviruses may be quite important and relevant to the understanding of the SARS-CoV. CORONAVIRUSES Taxonomy. Coronavirus taxonomy has recently been updated (27,29). Coronaviruses together with toroviruses (that cause enteric diseases in cattle and possibly in humans) form the Coronaviridae family, NAVAS-MARTIN AND WEISS 462 TABLE 1. CORONAVIRUSES HOST TROPISM AND DISEASES Disease Host Group Virus Respiratory Enteric Other Human 2 HCoV-OC43, 1 (1) 1 HCoV-229E, 1 — ? SARS-CoV 1 (1) Kidney Chicken 3 IBV 1 1 Kidney Turkey 2 TCV 1 1 Cat 1 FIPV 1 1 Systemic Cat 1 FECV — 1 Dog 1 CCV 1 1 Pig 1 TGEV 1 1 Pig 2 HEV 1 1 CNS Cow 2 BCV 1 — Rabbit 2 RbEVC — 1 Rat 2 SDAV — — CNS Mouse 2 MHV 1 1 CNS HCoV-OC43 and HCoV-OC229 are the previously known human coronaviruses that cause the 30% of common colds in humans; SARS-CoV, severe acute respiratory syndrome; IBV, infectious bronchitis virus; TCV, turkey coronavirus; FIPV, feline infectious peritonitis virus; FECV, feline enteric coronavirus; CCV, canine coronavirus; TGEV, porcine transmissible gastroenteritis virus; HEV, porcine hemagglutinating encephalomyelitis virus; BCV, bovine coronavirus; RbEVC, rabbit coronavirus; SADV sialodacryadenitis virus; MHV, mouse hepatitis virus. SARS-CoV has been detected in kidneys from infected patients; however, the clinical implications remain unknown. included in the Nidovirales order that also includes the Arteviridae and Roniviridae families. Whereas Roniviridae comprises a new genera of invertebrate viruses (16), some members of the Arteviridae family, cause important economic loss in the swine and equine industries (21,80). Virion structure. Coronaviruses are 60–200 nm round, somewhat pleiomorphic, particles containing an internal helical RNA-protein nucleocapsid surrounded by an envelope containing glycoprotein peplomers (15) (Fig. 1). The coronavirus genome is a very long single stranded, positive sense, polyadenylated RNA of 27–32 kb (39,40,88) (Fig. 2). The virions of all coronaviruses contains the following structural proteins: nucleocapsid (N) protein, a phosphoprotein (377–454 amino acids) that is complexed with genome RNA; M, a transmembrane glycoprotein (225–262 amino acids); S (spike), the highly glycosylated peplomer gly- coprotein (1173–1452 amino acids) mediates attachment to cells and induction of fusion with the host cell membrane during entry for all coronaviruses and the induction of cell to cell fusion during infection for some group 2 some coronaviruses (74,90); E, a membrane associated protein (with a range of 76–108 amino acids) that plays a role in assembly of virions (107). Some group 2 viruses including OC43 and some strains of MHV, express a fifth structural protein, the hemagglutinin-esterase protein (HE), which forms smaller spikes on virions (59); its function in the virus life cycle is as yet unknown. An HE protein is not encoded in the SARS-CoV genome (81). Replication. During lytic infection of cultured mouse cells, the virus enters the cell via attachment of S protein to a receptor which has been identified as a carcinoembryonic antigen (CEACAM) for MHV (23,24), and aminopeptidase N (APN or CD13) for the group 1 viruses such as 229E and transmissible gastroen- teritis virus (TGEV), feline infectious peritonitis virus (FIPV) and canine coronavirus (CCV) (93,94). Re- ceptor interaction triggers fusion of the viral and plasma membranes allowing entry of the nucleocapsid into the cytoplasm (61,108). Virus-specific RNA and proteins are synthesized probably entirely in the cy- toplasm (57,101) (Fig. 3). The 59 two thirds of the genome encode the replicase gene (approximately 21 kilobases) which is expressed via two very large ORFs, 1a and 1b. Expression of coronavirus proteins starts with translation of two polyproteins, pp1a and pp1ab, with predicted lengths of approx 4000 and 7000 amino acids, respectively. The latter is the result of a translational frameshifting event at the end of ORF1a (8). These polyproteins undergo co-translational proteolytic processing into at least four key enzymes: an RNA- dependent RNA polymerase (RdRp), a picornavirus 3C-like proteinase (3CLpro), a papain-like proteinase SARS AND OTHER CORONAVIRUSES 463 FIG. 1. Model of coronavirus virion structure. Viral particles contain an internal helical RNA-protein nucleocap- sid surrounded by an envelope containing glycoprotein peplomers. Nucleocapsid (N) protein is a phosphoprotein that is complexed with genome RNA. S, spike glycoprotein, forms the large glycosylated peplomers that are characteristic of coronaviruses. M, transmembrane protein, is highly hydrophobic and spans the membrane three times. E, a mem- brane associated protein, is a minor component of the membrane. Some group 2 viruses including OC43 and some strains of MHV, express a fifth structural protein, the hemagglutinin-esterase protein (HE) which forms smaller spikes on virions. (PLP), and a helicase (38). The replicase complex is used to transcribe a 39-coterminal set of nested subge- nomic mRNAs, as well as genomic RNA, that have a common 59 “leader” sequence derived from the 59 end of the genome. While the actual mechanism of synthesis of mRNAs is not well understood, it is cur- rently believed that subgenomic negative strand mRNAs serve as templates for mRNA (7). The replicase carries out “discontinuous transcription” in the fusion of body and leader sequences in subgenomic RNAs NAVAS-MARTIN AND WEISS 464 FIG. 2. Coronavirus genomes representing three (or 4?) antigenic groups. Infectious bronchitis virus (IBV, group 3), human coronavirus 229E (group 1) and murine coronavirus (MHV, group 2) are shown along with the SARS-CoV genome. The replicase gene (ORF 1a, 1b) is shown by open bars; structural genes (S, E, M, N and HE) are depicted with striped bars; non-structural genes (black bars) are variable in number and location in the coronavirus genome among the different viral groups. Small open reading frames (ORFs) are depicted in solid bars. FIG. 3. Model of coronavirus replication. The virus life cycle is described in the text. Polymerase (pol); rough en- doplasmic reticulum (RER); nucleus (N). Polyprotein 1a,1b (pp 1a.1b). and also during recombination events which occur at high frequency during coronavirus replication. In the case of some coronaviruses, such as MHV, syncytia are formed by fusion from within, mediated by S pro- tein on the plasma membrane. New virions are assembled by budding into intracellular membranes and are released from the cells probably through vesicles by the cell secretory mechanisms. Cellular syncytia even- tually lift off the tissue culture plastic and the cells die. In addition to lytic infection, some coronaviruses may cause persistent infection in cells in vitro as well as in animals (56,89). In the case of non-fusogenic viruses such as SARS-CoV there is lytic destruction of the cells (50). Pathogenesis. Coronaviruses infect many species of animals including humans. While human corona- viruses 229E and OC43 cause 30% of colds (46); other coronaviruses infect other organ systems, such as the gastrointestinal track and central nervous system, causing more severe and sometimes economically im- portant diseases in pigs, cows, chickens as well as in laboratory mice. Coronavirus infections can some- times cause fatal diseases; these include FIPV of the cat, hemagglutinating encephalomyelitis (HEV) of the pig, infectious bronchitis virus (IBV) in the chicken and MHV in the mouse. MHV infection of the mouse, using the commonly studied strains of A59 and JHM, provides a convenient small animal model for the study of acute hepatitis and encephalitis, as well as for chronic demyelinating disease such as multiple scle- rosis (42). In immune competent animals, infection with coronaviruses induces a protective immune re- sponse including cell mediated immunity and antibodies. However the immune response may also con- tribute to disease, at least in the case of MHV (79) and FIVP (13). The possibility of immunopathology must be considered in designing vaccine strategies for SARS-CoV. As a general rule (with some exceptions), coronaviruses are limited to replication in cells of one species. In the case of MHV, only murine cells lines may be productively infected. However when mammalian cells are transfected to express the receptor CEACAMI, they do become permissive for replication of MHV (24). Likewise, the human 229E virus can infect cells of other species when transfected with the human APN receptor (104). Thus, for coronaviruses, the ability to replicate in a cell type is usually dependent on viral entry and not intracellular events. There have been several studies adapting MHV to replicating in other cells types. Variant MHVs, able to replicate in human and hamster cells and monkey cells for example were isolated either by persistent infection of murine cells or infection of mixed cultures of murine and non per- missive hamster cells (4,83). The adaptations required mutations in the spike protein and the selection of viruses able to use the mammalian CEACAMI receptor; furthermore, these mutants retain broader host range, infecting cells of multiple species. These studies concluded that, in persistently infected murine cell lines, there is a strong selective advantage for virus with altered interactions with receptor. This selection is likely driven by the very low levels of expression of viral receptor MHVR and/or expression of alterna- tive virus receptors of lesser efficiency. Furthermore, MHV persistence may promote virus cross-species transmissibility by selecting for virus variants that recognize phylogenetic homologues of the normal re- ceptor. ROLE OF CORONAVIRUS GENES IN PATHOGENESIS Spike protein. Coronavirus spike proteins are believed to be a major determinant of pathogenic pheno- type. Spike proteins both interact with receptor and contain many of the determinants of immune response and thus are often considered as the first target for coronavirus vaccine development (28,31,77). Much in- direct evidence indicated that spike was important in determining pathogenic outcome of infection. Early studies demonstrated a relationship between variations in the spike and changes in viral pathogenesis, such as attenuation of neurovirulence of MHV (9,10,18,36,44). Studies on viral determinants of pathogenesis have been limited to the comparison of naturally occurring strains with mutant viruses. This limitation was mainly due to the lack of reverse genetics systems and infectious molecular clones for coronaviruses, and the inability to carry out reverse genetics, which was in turn largely due to the tremendous size of corona- virus genomes. However, in the last 10 years, a targeted RNA recombination system was developed for MHV by Dr. Paul Masters. This reverse genetics system exploits the high frequency of homologous RNA recombination of coronaviruses and allows the introduction of site directed mutations within most of the genes within the coronavirus genome (60). Targeted RNA recombination has been extended to other coro- SARS AND OTHER CORONAVIRUSES 465 navirus systems and has been used to directly demonstrate that the spike is a major determinant of patho- genic phenotype for FIPV (42) as well as MHV (64,65,78) and IBV (11). Of note, infectious clones have only been developed very recently for TGEV (2,91,105), MHV (106), and HCoV-229 (91) coronaviruses, using various strategies. There are no crystal structures available for any coronavirus spikes as yet; the model depicted in Figure 4 is based on other viral attachment/fusion proteins, and shows a diagram of the murine coronavirus spike protein. Figure 4A shows a model of the structure of the trimerized spike on the viral and/or cell mem- brane, while Figure 4B shows a linear map of the protein. Spike is synthesized as a 120-kDa precursor, which is co-translationally glycosylated to a 180-kDa polypeptide, which is cleaved intracellularly into two approximately 90-kDa non-covalently associated subunits, S1 and S2. S2 is anchored in the viral or cell membrane, while S1 is modeled as the globular head. S1 contains the receptor binding domain (RBD) as well as a hypervariable domain (HVR) while S2 is highly conserved, containing features common to many viral fusion proteins, including two heptad repeat domains (HR1 and HR2) as well as a transmembrane do- main (TM) (5,34,51,71). These domains are believed to be important in viral entry and in the cell-to-cell fusion process (35). In the case of the murine coronavirus, MHV strain JHM, for example, the spike pro- tein contains most of the (H-2b) CD4 (up arrows) and CD81 T (down arrows) cell epitopes (75). However, the immunodominant CD41 epitope is in the membrane protein (103). Spike is also a major target of neu- tralizing antibodies. There are B cell epitopes in both S1 and S2; the major neutralizing epitope, however, are within S1 (12,76). Thus, S1 is the more variable portion of the spike and contains most of the targets for B and T cells. By sequence comparison of the SARS-CoV spike with the already described and se- quenced coronavirus spike proteins, it appears that S2 is more conserved than S1, and that the SARS-CoV spike has the two heptad repeats as well as TM domain (see Fig. 1C). The sequence of the SARS-CoV spike suggests that it is not cleaved in that the SARS-CoV spike lacks the BBXBB (B 5 basic residue) se- quence believed to be the recognition site for cleavage by furin like enzymes. This lack of cleavage is prob- ably the reason for the inability to induce cell to cell fusion as for the known coronaviruses, cleavage ap- pears to be prerequisite for the induction of efficient cell to cell fusion. The S1 domains, among the known coronaviruses, share much less homology than S2 domains and there is presently no way to predict loca- tion of the SARS-spike receptor interacting domain and to determine whether there is hypervariable do- main, which in the murine coronavirus contains determinants of virulence. The understanding of the SARS- CoV spike gene is important in terms of vaccine design. Small sequence changes, as little as one amino acid substitution within the spike protein can vastly ef- fect pathogenesis; for example, one amino acid change in the amino terminus of spike of MHV strain A59 NAVAS-MARTIN AND WEISS 466 FIG. 4. Coronavirus spike proteins. (A) A model of a murine coronavirus (MHV) spike in the membrane. (B) A linear diagram of the MHV (strain JHM) and its functional domains. S1 and S2 are the two non-covalently linked sub- units. The cleavage site is shown as are the CD41 and CD81 (H-2b) T cell epitopes. RBD (receptor binding domain); HVR (hypervariable domain); heptad repeat domains (HR); TM (transmembrane domain) are shown. (C) A diagram of the SARS protein. The HR and TM domains are predicted to be conserved among coronavirus spikes, including SARS-CoV. (Q159L) can abrogate hepatotropism while retaining virulence in the central nervous system (45,58). A one amino acid substitution within the spike protein of MHV, JHM strain, can highly attenuate and restrict vi- ral antigen to the olfactory bulbs of infected mice (72,95) . Consequently, changes in SARS-CoV spike are expected to have a key role in pathogenesis. Although the targeted RNA recombination is a powerful technique to manipulate coronavirus genome, and its application has unequivocally demonstrate the role of the spike gene in coronavirus host tropism and pathogenesis, there is still a challenge, that is to understand in mechanistic terms the role of corona- virus spikes in viral entry and pathogenesis. To achieve this goal, other approaches should be considered such as the generation and use of viral pseudotypes. Viral pseudotyping has been exploited for the molec- ular analysis of envelope glycoproteins of HIV, Ebola, and hepatitis C viruses, among others (26,48,53,102). To our knowledge, there have not yet been any published reports of the application of this technology to coronavirus spike proteins. We have recently been able to incorporate MHV spike proteins into HIV par- ticles. As illustrated in Figure 5, such pseudotypes enter murine L2 and 17.CL-1 cells, but not human (293T cells) or green monkey kidney cells (Vero E6) (Navas and Weiss, manuscript in preparation) (Fig. 5). This methodology will allow the precise investigation of coronavirus spikes in terms of receptors, inhibitors of entry, host-range, and neutralization by antibodies. Genes other than the spike, so-called “background genes,” also influence pathogenesis. Interestingly, a recombinant virus expressing the spike gene of the hepatropic A59 strain of MHV within the background of the neurotropic JHM strain, is very poorly hepatotropic. Thus, the JHM genetic background, plays a dom- inant role over the spike in the determination of hepatotropism (66). M protein. The coronavirus M protein is the most abundant envelope component. While group 1 and group 3 coronaviruses have M proteins with N-linked sugars, the M proteins of the group 2 coronaviruses [e.g., mouse hepatitis virus (MHV)] are O-glycosylated. It is not clear whether there is biological signifi- cance associated with the difference in glycosylation among the M proteins of the groups of coronaviruses. Although glycosylation is not required for viral assembly, its role in pathogenesis remains unknown. For the porcine coronavirus, TGEV, mutations in the M protein ectodomain that impair N-glycosylation, de- crease the interferogenic activity (54), and antibodies directed to the TGEV M ectodomain block IFNa in- duction (14). For MHV, recombinant viruses expressing N-, O-, or unglycoyslated M proteins were selected and compared (19). The M protein glycosylation state did not influence the tissue culture growth charac- teristics of the recombinant viruses. However, in mice, the recombinant virus differed in their ability to replicate in the liver, but not in the brain. In tissue culture, the recombinant virus expressing the N-glyco- sylated M protein replicated to higher titer than that expressing the O-glycosylated M , which in turn repli- cated to higher titer than the virus expressing an unglycosylated M protein. Curiously while, the ability to induce interferon in vivo did not appear to be affected by their glycosylation status, the ability to replicate in the liver correlated with the ability to induce interferon in vitro. SARS AND OTHER CORONAVIRUSES 467 FIG. 5. Infectivity of MHV-A59 spike-pseudotyped HIV particles in murine fibroblast cells (L2 and 17.CL-1), hu- man cells (HeLa, and 293T cells), and African green monkey kidney cells (Vero E6). Pseudotype infectivity correlates with MHV tropism. N protein. The N gene has been implicated in MHV induced fulminant hepatitis (67,68) In particular, Ding et al. (22) demonstrated that the nucleocapsid protein of MHV-3 strain induces transcriptional acti- vation of a novel prothrombinase gene (fgl2), which is implicated in the coagulation cascade; however, the mechanism remains poorly understood. The expression of fgl2 prothrombinase is restricted to macrophages (Kuppfer cells) and endothelial cells in the liver, but not in hepatocytes. The possible relevance of these findings to SARS disease needs to be further evaluated, as the presence of vascular thrombosis has been reported in some postmortem lung tissues from SARS patients (73). The genomes of several group 2 coronaviruses, including mouse hepatitis virus (MHV), contain an in- ternal open reading frame (ORF) within the 59 half of the nucleocapsid gene. This ORF, translated in the 11 reading frame with respect to the N protein, encodes a mostly hydrophobic 23-kDa polypeptide. In the case of MHV, the I gene product is expressed in MHV-infected cells and found within the virions as well. Selection and characterization of recombinant viruses in which the I gene is disrupted demonstrated that I protein is not essential for the replication of MHV either in tissue culture or for pathogenesis in the mouse (32). However, it may have an as yet unknown, subtle role in pathogenesis. Envelope (E) protein. The E protein is an integral membrane protein and a minor component of the coronavirus envelope (107). E plays an important role in coronavirus assembly (97). Expression of the E protein alone results in its incorporation into vesicles that are released from cells, and the coexpression of the E protein with the membrane protein M leads to the assembly of coronavirus-like particles. In vitro studies have demonstrated that E induces apoptosis in MHV-A59–infected 17Cl-1 cells via a caspase de- pendent mechanism (3). Recently reverse genetics techniques were used to select recombinant MHVs in which the E gene was deleted (along with ORFS 4 and 5a) (52). Surprisingly such viruses were viable demonstrating that E protein is not essential for replication in vitro. However, these viruses had small plaque size, low growth rate, and low infectious titer providing further support for the importance of E protein in MHV replication; however, the role of the E gene is pathogenesis beyond affecting replication rate has not been explored in vivo. Small open reading frames (ORFs). All coronaviruses, including SARS-CoV, encode, in addition to structural proteins and replicase proteins, small unique likely nonstructural proteins of unknown functions. In the case of MHV, these are ORFS 2a, 4, and 5a; the proteins encoded in these three ORFs appear or be non essential for replication. In some strains of MHV, ORF 4 is interrupted and becomes ORFs 4a and 4b (100). There is a report of an MHV isolate with a deletion of ORF2a (84). A recombinant MHV lacking gene 4 has been shown to be as virulent as wild type in mice (69). There is some evidence that the protein encoded in these small ORFs may have functions in pathogenesis. While, a recombinant MHV lacking ORFs 2a, 4, and 5 replicates almost as efficiently as wild type virus in tissue culture, it is highly attenu- ated in vivo (20). Furthermore gene 7 of the porcine coronavirus TGEV is non-essential for replication but does influence in vivo replication and virulence (70). The SARS coronavirus genome sequence predicts 8 such ORFs (ORFs 3a, 3b, 6, 7a,7b, 8a, 8b, and 9b) (87). While there are not yet any specific examples of coronavirus proteins involved in anti-host defense there are many examples of non-essential proteins that have such functions in other viral systems of such proteins. The HIV Vif protein, which is dispensable in cell types yet essential, is some cell such as primary T cells, is believed to overcome an antiretroviral ac- tivity of the innate immune response that is mediated by deamination of viral DNA (43). The Herpes sim- plex proteins ICP47 blocks the loading of peptides onto MHC class I and subsequent cell surface expres- sion of the peptide/MHC complex (25). The nonstructural (NS) gene of the highly lethal H5N1influenza viruses, unlike other human, avian and swine influenza viruses, confers resistance to the antiviral effects of interferons and tumor necrosis factor alpha (85). Replicase. Lastly the replicase proteins, and perhaps other non structural proteins, could affect tropism and pathogenesis by determining rate of replication of virus perhaps by interactions with cell type specific factors or with elements of the immune response. This hypothesis needs to be further evaluated in the future. PERSPECTIVES: CORONAVIRUS AS EMERGING PATHOGENS The origins of SARS-CoV are not known. The genome does not appear to be recombinant of any known coronaviruses. One prominent theory is that SARS-CoV has a reservoir in another species and “jumped” NAVAS-MARTIN AND WEISS 468 to humans (30). This has been given credibility by the isolation of SARS like coronavirus from several species of wild animals (six civet cats among them) (30) in markets of Southern China and the report that pet cats were found to be infected and sick from the SARS-CoV in the Amboy Garden apartment complex in Hong Kong, location of more than 100 human SARS in infections (1,17). It has recently been reported that the genomes of viruses isolated from civet cats are close in sequence to the human isolates; there are differences in the spike gene and in one of the small, likely nonstructural, open reading frames (41). A re- cent comparison of the genomes of human and animal isolates of SARS-CoV indicates ten consistent amino acid differences between the spike proteins of human and animal isolates (41); interestingly, these substi- tutions suggest the loss of multiple predicted glycosylation sites in the spikes of the animal isolates. This is a potentially exciting finding and may indicate a difference in biological properties. The loss of glyco- sylation sites in the glycoprotein of mutant simian immunodeficiency virus has been shown to increase neu- tralization of these viruses (49). Among the differences between the animal and human isolates, there are also two amino acid substitutions within the heptad repeat domains regions of the spike; these domains are involved in membrane fusion and aa substitutions are known in fusion domains to effect pathogenic prop- erties in the murine coronavirus (72,95). The other consistent difference observed in the comparison of the genomes of SARS-CoV isolates from humans and civet cats, was a 29-nt insertion in the genomes of ani- mal isolates. These results in the merging of ORFs 8a and 8b so there is only ORF8 (87). In another re- port, an in frame deletion of 45 nt occurs in ORF 7b after three passages of SARS CoV in tissue culture (92). These small ORFs may play a role in biological properties of SARS-CoV isolates and should be ex- plored further. Thus, at this point, it not clear what the natural host of the SARS-CoV is and how it “jumped” into the human population; it will be important to understand more about the sequence differences in the spike pro- teins between animal and human isolates and how the spike protein controls species tropism. The knowl- edge accumulated by the study of the coronaviruses has already accelerated the understanding of the SARS coronavirus and will continue to do so in the future. ACKNOWLEDGMENTS We are indebted to many colleagues for helpful discussions and thoughts. We apologize to any investi- gators whose work we have inadvertently omitted. We acknowledge NIH grants AI17418, NS21954, NS30606, and AI 47800, as well as RG2585B5 from the National Multiple Sclerosis Society. REFERENCES 1. Abbott, A., and D. Cyranoski. 2003. Biologists seek to head off future sources of infection. Nature 423:3. 2. Almazan, F., J.M. Gonzalez, Z. Penzes, et al. 2000. Engineering the largest RNA virus genome as an infectious bacterial artificial chromosome. Proc. Natl. Acad. Sci. USA 97:5516–5521. 3. An, S., C.J. Chen, X. Yu, et al. 1999. Induction of apoptosis in murine coronavirus-infected cultured cells and demonstration of E protein as an apoptosis inducer. J. Virol. 73:7853–7859. 4. Baric, R.S., E. Sullivan, L. Hensley, et al. 1999. Persistent infection promotes cross-species transmissibility of mouse hepatitis virus. J. Virol. 73:638–649. 5. Bosch, B.J., R. van der Zee, C.A. de Haan, et al. 2003. The coronavirus spike protein is a class I virus fusion pro- tein: structural and functional characterization of the fusion core complex. J. Virol. 77:8801–8811. 6. Bradburne, A.F., M.L. Bynoe, and D.A. Tyrrel. 1967. Effects of a “new” human respiratory virus in volunteers. Br. Med. J. 3:767–769. 7. Brian, D.A., R.Y. Chang, M.A. Hofmann, et al. 1994. Role of subgenomic minus-strand RNA in coronavirus replication. Arch. Virol. Suppl. 9:173–180. 8. Brierley, I., M.E. Boursnell, M.M. Binns, et al. 1987. An efficient ribosomal frame-shifting signal in the poly- merase-encoding region of the coronavirus IBV. EMBO J. 6:3779–3785. SARS AND OTHER CORONAVIRUSES 469 9. Buchmeier, M.J., R.G. Dalziel, and M.J. Koolen. 1988. Coronavirus-induced CNS disease: a model for virus-in- duced demyelination. J. Neuroimmunol. 20:111–116. 10. Buchmeier, M.J., R.G. Dalziel, M.J. Koolen, et al. 1987. Molecular determinants of CNS virulence of MHV-4. Adv. Exp. Med. Biol. 218:287–295. 11. Casais, R., B. Dove, D. Cavanagh, et al. 2003. Recombinant avian infectious bronchitis virus expressing a heterologous spike gene demonstrates that the spike protein is a determinant of cell tropism. J. Virol. 77:9084–9089. 12. Castro, R.F., and S. Perlman. 1995. CD81 T-cell epitopes within the surface glycoprotein of a neurotropic coro- navirus and correlation with pathogenicity. J. Virol. 69:8127–8131. 13. Chalmers, W.S., B.C. Horsburgh, W. Baxendale, et al. 1993. Enhancement of FIP in cats immunised with vac- cinia virus recombinants expressing CCV and TGEV spike glycoproteins. Adv. Exp. Med. Biol. 342:359–364. 14. Charley, B., and H. Laude. 1988. Induction of alpha interferon by transmissible gastroenteritis coronavirus: role of transmembrane glycoprotein E1. J. Virol. 62:8–11. 15. Cheley, S., V.L. Morris, M.J. Cupples, et al. 1981. RNA and polypeptide homology among murine coronaviruses. Virology 115:310–321. 16. Cowley, J.A., C.M. Dimmock, K.M. Spann, et al. 2000. Detection of Australian gill-associated virus (GAV) and lymphoid organ virus (LOV) of Penaeus monodon by RT-nested PCR. Dis. Aquat. Organ. 39:159–167. 17. Cyranoski, D., and A. Abbott. 2003. Apartment complex holds clues to pandemic potential of SARS. Nature 423:3–4. 18. Dalziel, R.G., P.W. Lampert, P.J. Talbot, et al. 1986. Site-specific alteration of murine hepatitis virus type 4 peplomer glycoprotein E2 results in reduced neurovirulence. J. Virol. 59:463–471. 19. de Haan, C.A., M. de Wit, L. Kuo, et al. 2002. O-Glycosylation of the mouse hepatitis coronavirus membrane protein. Virus Res. 82:77–81. 20. de Haan, C.A., P.S. Masters, X. Shen, et al. 2002. The group-specific murine coronavirus genes are not essential, but their deletion, by reverse genetics, is attenuating in the natural host. Virology 296:177–189. 21. Del Piero, F., and P.A. Wilkins. 2001. Pulmonary vasculotropic EHV-1 infection in equids. Vet. Pathol. 38:474. 22. Ding, J.W., Q. Ning, M.F. Liu, et al. 1997. Fulminant hepatic failure in murine hepatitis virus strain 3 infection: tissue-specific expression of a novel fgl2 prothrombinase. J. Virol. 71:9223–9230. 23. Dveksler, G.S., C.W. Dieffenbach, C.B. Cardellichio, et al. 1993. Several members of the mouse carcinoembry- onic antigen-related glycoprotein family are functional receptors for the coronavirus mouse hepatitis virus-A59. J. Virol. 67:1–8. 24. Dveksler, G.S., M.N. Pensiero, C.B. Cardellichio, et al. 1991. Cloning of the mouse hepatitis virus (MHV) re- ceptor: expression in human and hamster cell lines confers susceptibility to MHV. J. Virol. 65:6881–6891. 25. Easterfield, A.J., B.M. Austen, and O.M. Westwood. 2001. Inhibition of antigen transport by expression of in- fected cell peptide 47 (ICP47) prevents cell surface expression of HLA in choriocarcinoma cell lines. J. Reprod. Immunol. 50:19–40. 26. Endres, M.J., S. Jaffer, B. Haggarty, et al. 1997. Targeting of HIV- and SIV-infected cells by CD4–chemokine receptor pseudotypes. Science 278:1462–1464. 27. Enjuanes, L., D. Brian, D. Cavanagh, et al. 2000. Coronaviridae., p. 835–849. In: C.M.F.M.H.V. van Regenmor- tel, D.H.L. Bishop, E.B. Carsten, et al. (eds.), Virus Taxonomy. Classification and Nomenclature of Viruses. Aca- demic Press, New York. 28. Enjuanes, L., C. Smerdou, J. Castilla, et al. 1995. Development of protection against coronavirus induced dis- eases. A review. Adv. Exp. Med. Biol. 380:197–211. 29. Enjuanes, L., W. Spaan, E. Snijder, et al. 2000. Nidovirales, pp. 827–834. In: C.M.F.M.H.V. van Regenmortel, D.H.L. Bishop, E.B. Carsten, et al. (eds.), Virus Taxonomy. Classification and Nomenclature of Viruses. Acade- mic Press, New York. 30. Enserink, M. 2003. Infectious diseases. Clues to the animal origins of SARS. Science 300:1351. NAVAS-MARTIN AND WEISS 470