Severe acute respiratory syndrome (SARS) represents the 21st century's first pandemic of a transmissible disease with a previously unknown cause. The pandemic started in November 2002 and was brought under control in July 2003, after it had spread to 33 countries on 5 continents, resulting in >8,000 infections and >700 deaths ( 1 ). The outbreaks were caused by a newly emerged coronavirus, now known as the SARS coronavirus (SARS-CoV). In late 2003 and early 2004, sporadic outbreaks were reported in the region of the People's Republic of China where the 2002–2003 outbreaks originated ( 2 ). However, molecular epidemiologic studies showed that the viruses responsible for the 2003–2004 outbreaks were not the same as those isolated during the 2002–2003 outbreaks ( 3 ). These findings indicate independent species-crossing events. They also indicate that a SARS epidemic may recur in the future and that SARS-like coronaviruses (SARS-like–CoVs) that originate from different reservoir host populations may lead to epidemics at different times or in different regions, depending on the distribution of the reservoirs and transmitting hosts. The recent discovery of a group of diverse SARS-like–CoVs in bats supports the possibility of these events and further highlights the need to understand reservoir distribution and transmission to prevent future outbreaks. Animal Origin of SARS Coronaviruses Because of the sudden and unpredictable nature of the SARS outbreaks that started in November 2002 in southern People's Republic of China, structured and reliable epidemiologic studies to conclusively trace the origin of SARS-CoV were not conducted. However, accumulated studies from different groups, which used a variety of approaches, indicated an animal origin on the basis of the following findings. 1) Genome sequencing indicated that SARS-CoV is a new virus with no genetic relatedness to any known human coronaviruses ( 4 , 5 ). 2) Retrospective serologic studies found no evidence of seroprevalence to SARS-CoV or related viruses in the human population ( 6 ). 3) Serologic surveys among market traders during the 2002–2003 outbreaks showed that antibodies against SARS-CoV or related viruses were present at a higher ratio in animal traders than control populations ( 7 – 9 ). 4) Epidemiologic studies indicated that early case-patients were more likely than later case-patients to report living near a produce market but not near a farm, and almost half of them were food handlers with probable animal contact ( 7 ). 5) SARS-CoVs isolated from animals in markets were almost identical to human isolates ( 9 ). 6) Molecular epidemiologic analyses indicated that human SARS-CoV isolates could be divided into 3 groups from the early, middle, and late phases of the outbreaks and that early-phase isolates were more closely related to the animal isolates ( 10 ). 7) Human SARS-CoVs isolates from the 2003–2004 outbreaks had higher sequence identity to animal isolates of the same period than to human isolates from the 2002–2003 outbreaks ( 3 ). Susceptible Animals in Markets and Laboratories The first evidence of SARS-CoV infection in animals came from a study conducted in a live animal market in early 2003 ( 9 ). From the 25 animals sampled, viruses closely related to SARS-CoV were detected in 3 masked palm civets (Paguma larvata) and 1 raccoon dog (Nyctereutes procyonoides). In addition, neutralizing antibodies against SARS-CoV were detected in 2 Chinese ferret badgers (Melogale moschata). This initial study indicated that at least 3 different animal species in the Shenzhen market were infected by coronaviruses that are closely related to SARS-CoV. Given the vast number of live animals being traded in animal markets in southern People's Republic of China, knowing which other animals are also susceptible to these viruses is crucial. Unfortunately, for a variety of reasons no systematic studies were conducted on traded animals during the outbreak period. Experimental infection of different animals therefore became a component of the SARS-CoV investigation. Currently, >10 mammalian species have been proven to be susceptible to infection by SARS-CoV or related viruses (Table 1). Rats were also implicated as potentially susceptible animals that may have played a role in the transmission and spread of SARS-CoV in the well-publicized SARS outbreaks in the Amoy Gardens apartment block in Hong Kong Special Administrative Region, People's Republic of China ( 23 ). In Guangdong in 2004, the first human with a confirmed case of SARS was reported to have had no contact with any animals except rats ( 2 ). Experimentally, we have obtained serologic evidence that SARS-CoV replicates asymptomatically in rats (B.T. Eaton et al., unpub. data). Further studies are needed to clarify the potential role of rats in the transmission of SARS-CoV. Studies by 2 independent groups suggested that avian species were not susceptible to SARS-CoV infection and that, hence, domestic poultry were unlikely to be the reservoir or associated with the dissemination of SARS-CoV in the animal markets of southern People's Republic of China ( 22 , 24 ). Table 1 Animal species susceptible to infection by SARS coronavirus* Animal Mode of infection Clinical signs References Common name Taxonomic name Masked palm civet Paguma larvata Natural None observed ( 9 ) Experimental Fever, lethargy, reduced appetite ( 11 ) Racoon dog Nyctereutes procyonoides Natural None observed ( 9 ) Chinese ferret badger Melogale moschata Natural None observed ( 9 ) Cynomolgus macaque Macaca facicularis Experimental Lethargy, skin rash, respiratory distress ( 12 ) Rhesus macaque Macaca mulatta Experimental Fever, low appetite ( 13 , 14 ) African green monkey Cercopithecus aethiops Experimental None observed ( 15 ) Ferret Mustela furo Experimental Lethargy, mild pulmonary lesions ( 16 ) Golden hamster Mesocricetus auratus Experimental None observed ( 17 ) Guinea pig Cavia porcellus Experimental None observed ( 18 ) Mouse Mus musculus Experimental Aged animal (12–14 mo): weight loss, hunched posture, ruffled fur, slight dehydration ( 19 ) Young animal (4–6 weeks): none observed ( 20 ) Rat Rattus rattus Experimental None observed B.T. Eaton et al., unpub. data Domestic cat Felis domesticus Natural Not reported ( 16 ) Experimental None observed ( 16 ) Pig Sus scrofa Natural Not reported ( 21 ) Experimental None observed ( 22 ) *SARS, severe acute respiratory syndrome. Role of Masked Palm Civets Although in 1 live animal market, 3 species were found to be infected by viruses related to SARS-CoV ( 9 ), all subsequent studies have focused mainly on palm civets, possibly because the rate of detection was higher in civets or because the number of civets traded in southern People's Republic of China exceeds that of other wildlife groups. The isolation of closely related SARS-CoV in civets during the 2002–2003 and 2003–2004 outbreaks and the close match of virus sequences between the human and civet isolates from each outbreak ( 3 , 9 , 25 ) strongly suggest that civets are a direct source of human infection. However, these studies did not clarify whether animals other than civets were involved in transmission of SARS-CoV to humans or whether civets were an intermediate host or the natural reservoir host of SARS-CoVs. During the 2002–2003 outbreaks, none of the animal traders surveyed in the markets, who supposedly had very close contact with live civets, displayed SARS symptoms ( 7 – 9 ). During the 2003–2004 outbreaks, at least 1 human SARS patient had had no contact with civets ( 2 ). These observations seem to indicate that >1 other animal species may play a role in transmission of SARS-CoV to humans. Most, if not all, civets traded in the markets are not truly wildlife animals; rather, they are farmed animals. Civet farming is relatively new in People's Republic of China and has rapidly expanded during the past 15 years or so. Tu et al. conducted the first comparative study of market and farmed civets ( 26 ). Serologic testing was performed on 103 serum samples taken from civets in an animal market in Guangdong and several civet farms in different regions of People's Republic of China in June 2003 and January 2004. No significant level of SARS-CoV antibody was detected in any of the 75 samples taken from 6 farms in 3 provinces. In contrast, of the 18 samples taken from an animal market in Guangdong Province in January 2004, 14 (79%) had neutralizing antibodies to SARS-CoV. In a parallel study conducted between January and September 2004 ( 27 ), molecular analysis was used to investigate the distribution of SARS-CoV in palm civets in markets and on farms. PCR analysis of samples from 91 palm civets and 15 raccoon dogs in 1 animal market and 1,107 civets from 25 farms in 12 provinces showed positive results for all animals from the market and negative results for all animals from the farms. Similar results were obtained in wild-trapped civets in Hong Kong; none of the 21 wild civets sampled had positive antibody or PCR results for SARS-CoV ( 28 ). Although not universally true, natural reservoir hosts tend to have coevolved with their viruses and usually do not display clinical signs of infection ( 29 ). However, when palm civets were experimentally infected with 2 strains of human SARS-CoV, all developed clinical signs of fever, lethargy, and loss of aggressiveness ( 11 ). Civits' high susceptibility to SARS-CoV infection and wide presence in markets and restaurants strongly indicates an important role for civets in the 2002–2003 and 2003–2004 SARS outbreaks. However, the lack of widespread infection in wild or farmed palm civets makes them unlikely to have been the natural reservoir host. SARS-like Coronaviruses in Bats The presence of SARS-like–CoVs in different species of horseshoe bats in the genus Rhinolophus has recently been reported. We found, in a study of horseshoe bat species in different regions of mainland People's Republic of China in 2004 ( 30 ), that each of the 4 species surveyed had evidence of infection by a SARS-like–CoV: 2 species (R. pearsoni and R. macrotis) had positive results by both serologic and PCR tests, and 2 (R. pussilus and R. ferrumequinum) had positive results by either serologic or PCR tests, respectively. Bats with positive results were detected in the provinces of Hubei and Guangxi, which are >1,000 km apart. A group in Hong Kong ( 31 ) found that, when analyzed by PCR, 23 (39%) of 59 anal swabs of wild Chinese horseshoe bats (R. sinicus) contained genetic material closely related to SARS-CoV. They also found that as many as 84% of the horseshoe bats examined contained antibodies to a recombinant N protein of SARS-CoV. A previous study indicated a certain level of antigenic cross-reactivity between SARS-CoV and some group 1 coronaviruses ( 6 ) and that several group 1 coronaviruses had recently been found in bats. Therefore, the actual seropositive proportion of R. sinicus might be 1 intermediate host before it could efficiently infect humans. The existence of at least 3 discontinuous highly variable genomic regions between SARS-CoV and SARS-like–CoV indicates that the second mechanism is more likely. In conclusion, the discovery of bat SARS-like–CoVs and the great genetic diversity of coronaviruses in bats have shed new light on the origin and transmission of SARS-CoV. Although the exact natural reservoir host for the progenitor virus of SARS-CoV is still unknown, we believe that a continued search in different bat populations in People's Republic of China and neighboring countries, combined with experimental infection of different bat species with SARS-CoV, will eventually identify the native reservoir species. A positive outcome of these investigations will greatly enhance our understanding of spillover mechanisms, which will in turn facilitate development and implementation of effective prevention strategies. The discovery of SARS-like–CoVs in bats highlights the increasingly recognized importance of bats as reservoirs of emerging viruses ( 36 ). Moreover, the recent emergence of SARS-CoVs and other bat-associated viruses such as henipaviruses ( 37 , 38 ), Menangle, and Tioman viruses ( 36 ), and variants of rabies viruses and bat lyssaviruses ( 38 , 39 ) also supports the contention that viruses, especially RNA viruses, possess more risk than other pathogens for disease emergence in human and domestic mammals because of their higher mutation rates ( 40 ).
