Pathfinder
Editor, Senior Moderator
Broad host range of SARS-CoV-2 predicted by comparative and structural analysis of ACE2 in vertebrates
View ORCID ProfileJoana Damas, View ORCID ProfileGraham M. Hughes, View ORCID ProfileKathleen C. Keough, View ORCID ProfileCorrie A. Painter, View ORCID ProfileNicole S. Persky, View ORCID ProfileMarco Corbo, View ORCID ProfileMichael Hiller, View ORCID ProfileKlaus-Peter Koepfli, View ORCID ProfileAndreas R. Pfenning, View ORCID ProfileHuabin Zhao, View ORCID ProfileDiane P. Genereux, View ORCID ProfileRoss Swofford, View ORCID ProfileKatherine S. Pollard, View ORCID ProfileOliver A. Ryder, View ORCID ProfileMartin T. Nweeia, View ORCID ProfileKerstin Lindblad-Toh, View ORCID ProfileEmma C. Teeling, View ORCID ProfileElinor K. Karlsson, and View ORCID ProfileHarris A. Lewin
PNAS first published August 21, 2020 https://doi.org/10.1073/pnas.2010146117
Edited by Scott V. Edwards, Harvard University, Cambridge, MA, and approved July 31, 2020 (received for review June 2, 2020)
Article Figures & SI Info & Metrics PDF
Significance
The novel severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is the cause of COVID-19, a major pandemic that threatens millions of human lives and the global economy. We identified a large number of mammals that can potentially be infected by SARS-CoV-2 via their ACE2 proteins. This can assist the identification of intermediate hosts for SARS-CoV-2 and hence reduce the opportunity for a future outbreak of COVID-19. Among the species we found with the highest risk for SARS-CoV-2 infection are wildlife and endangered species. These species represent an opportunity for spillover of SARS-CoV-2 from humans to other susceptible animals. Given the limited infectivity data for the species studied, we urge caution not to overinterpret the predictions of the present study.
Abstract
The novel coronavirus severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is the cause of COVID-19. The main receptor of SARS-CoV-2, angiotensin I converting enzyme 2 (ACE2), is now undergoing extensive scrutiny to understand the routes of transmission and sensitivity in different species. Here, we utilized a unique dataset of ACE2 sequences from 410 vertebrate species, including 252 mammals, to study the conservation of ACE2 and its potential to be used as a receptor by SARS-CoV-2. We designed a five-category binding score based on the conservation properties of 25 amino acids important for the binding between ACE2 and the SARS-CoV-2 spike protein. Only mammals fell into the medium to very high categories and only catarrhine primates into the very high category, suggesting that they are at high risk for SARS-CoV-2 infection. We employed a protein structural analysis to qualitatively assess whether amino acid changes at variable residues would be likely to disrupt ACE2/SARS-CoV-2 spike protein binding and found the number of predicted unfavorable changes significantly correlated with the binding score. Extending this analysis to human population data, we found only rare (frequency <0.001) variants in 10/25 binding sites. In addition, we found significant signals of selection and accelerated evolution in the ACE2 coding sequence across all mammals, and specific to the bat lineage. Our results, if confirmed by additional experimental data, may lead to the identification of intermediate host species for SARS-CoV-2, guide the selection of animal models of COVID-19, and assist the conservation of animals both in native habitats and in human care.
...
Discussion
Phylogenetic analysis of coronaviruses has demonstrated that the immediate ancestor of SARS-CoV-2 most likely originated in a bat species (1). However, whether SARS-CoV-2 or the progenitor of this virus was transmitted directly to humans or through an intermediate host is not yet resolved. To identify candidate intermediate host species and species at risk for SARS-CoV-2 infection, we undertook a deep comparative genomic, evolutionary, and structural analysis of ACE2, which serves as the SARS-CoV-2 receptor in humans. We drew on the rapidly growing database of annotated vertebrate genomes, including new genomes produced by the Genomes 10K-affiliated Bat1K Consortium, Zoonomia, and Vertebrate Genomes Project, and other sources (30, 31). We conducted a phylogenetic analysis of ACE2 orthologs from 410 vertebrate species and predicted their propensity to bind the SARS-CoV-2 S using a score based on amino acid substitutions at 25 consensus human ACE2 binding residues (12, 20). Similarity-based methods are frequently used for predicting cross-species transmission of viruses (32, 33), including SARS-CoV (2). We supported these predictions with comprehensive structural analysis of the ACE2 binding site complexed with SARS-CoV-2 S. We also tested the hypothesis that the ACE2 receptor is under selective constraints in mammalian lineages with different susceptibilities to coronaviruses.
We predict that species scoring as very high and high for propensity of SARS-CoV-2 S binding to ACE2 will have a high probability of becoming infected by the virus and thus may be potential intermediate hosts for virus transmission. We also predict that many species having a medium score have some risk of infection, and species scored as very low and low are less likely to be infected by SARS-CoV-2 via the ACE2 receptor. Importantly, our predictions are based solely on in silico analyses and must be confirmed by direct experimental data. The prediction accuracy of the model may be improved in the future as more extensive data are generated showing the impact of ACE2 mutations on its binding affinity for SARS-CoV-2 S, which may enable knowledge-based weighting of residues in the scoring algorithm. Until the present model’s accuracy can be confirmed with additional experimental data, we urge caution not to overinterpret the predictions of the present study. This is especially important with regards to species, endangered or otherwise, in human care. While species ranked high or medium may be susceptible to infection based on the features of their ACE2 residues, pathological outcomes may be very different among species depending on other mechanisms, such as immune response, that could affect virus replication and spread to target cells, tissues, and organs. Furthermore, we cannot exclude the possibility that infection in any species occurs via another cellular receptor (for a review see ref. 34), as shown for other betacoronaviruses (35), or lower-affinity interactions with ACE2 as proposed for SARS-CoV (2). Nonetheless, our predictions provide a useful starting point for the selection of appropriate animal models for COVID-19 research and identification of species that may be at risk for human-to-animal or animal-to-animal transmissions of SARS-CoV-2.
Several recent studies examined the role of ACE2 in SARS-CoV-2 binding and cellular infection and its relationship to experimental and natural infections in different species (26, 35⇓⇓⇓⇓–40). Our study design differs substantially from those in several aspects: 1) we analyzed a larger number of primates, carnivores, rodents, cetartiodactyls, and other mammalian orders and an extensive phylogenetic sampling of fishes, birds, amphibians, and reptiles; 2) we analyzed the full set of S-binding residues across the ACE2 binding site, which was based on a consensus set from two independent studies (12, 20); 3) we used different methodologies to assess ACE2 binding capacity for SARS-CoV-2 S; and 4) our study tested for selection and accelerated evolution across the entire ACE2 protein. While our results are consistent with the results and conclusions of Melin et al. (38) on the predicted susceptibility of primates to SARS-CoV-2, particularly Old-World primates, we made predictions for a larger number of primates (n = 39 vs. n = 27), bats (n = 37 vs. n = 7), other mammals (n = 176 vs. n = 5), and other vertebrates (n = 158 vs. n = 0). When ACE2 from species in our study were compared with results of other studies there were many consistencies, such as the low risk for rodents, but some predictions differ, such as the relatively high risk predicted by others for SARS-CoV-2 S binding in pangolin and horse (39), civet (40), Chinese rufous horseshoe bat (40), and turtles (22). Our results are generally consistent with a study that tested binding affinity of soluble ACE2 for the SARS-CoV-2 S RBD using saturation mutagenesis (27), particularly in the binding hot-spot region of ACE2 residues 353 to 357 (SI Appendix, Fig. S1). Importantly, as compared with other studies, our results greatly expanded the number of candidate intermediate hosts and identified many additional threatened species that could be at risk for SARS-CoV-2 infection via their ACE2 receptors.
Evolution of ACE2.
Variation in ACE2 in the human population is rare (16). Overall, ACE2 is intolerant of loss-of-function mutations [pLI = 0.998; LOEUF = 0.25 in gnomAD v2.1.1 (16)]. We examined a large set of ACE2 variants for their potential differences in binding to SARS-CoV-2 S and their relationship to selected and accelerated sites. We found rare coding variants that would result in missense mutations causing substitutions in 7/25 binding residues (Dataset S2). Some of those [e.g., E35K, f = 0.00001636 (16)] could reduce the virus binding affinity as per our structural analysis (Dataset S2) but would potentially lower the susceptibility to the virus only in a very small fraction of the population. Our analysis suggests that some variants (e.g., D38E) might not affect binding propensity while the potential impact of others (e.g., S19P) could not be determined. Further investigations on the effects of these rare variants on ACE2/SARS-CoV-2 binding affinity are needed.
When exploring patterns of codon evolution in ACE2, we found that multiple ACE2 residues important for the binding of SARS-CoV-2 S are evolving rapidly across mammals, with two (Q24 and H34) under positive selection (Fig. 4 A and B and SI Appendix, Fig. S5). Relative to other lineages analyzed, Chiroptera has a greater proportion of accelerated versus conserved codons (SI Appendix, Fig. S6), particularly in the SARS-CoV-2 S-binding region, suggesting the possibility of selective forces on these codons in Chiroptera driven by their interactions with SARS-CoV-2-like viruses (Fig. 4 C and D and Dataset S4.13). Indeed, distinct signatures of positive selection found in bat ACE2 (41) and in the SARS-CoV-2 S protein (42) support the hypothesis that bats are evolving to tolerate SARS-CoV-2-like viruses (discussed further below).
Relationship of the ACE2 Binding Score to Known Infectivity of SARS-CoV-2.
Data on susceptibility of nonhuman species to SARS-CoV-2 is still very limited (SI Appendix, Fig. S10) but mostly agree with our predictions of ACE2 binding propensity for SARS-CoV-2 S (Figs. 1 and 2 and Dataset S1). Five out of six species with demonstrated susceptibility to SARS-CoV-2 infection score very high [rhesus macaque (43) and cynomolgus macaque (44)] or medium [domestic cat (45, 46), tiger (8) and golden hamster (47)]. Both species susceptible to infection but asymptomatic scored low [dog (45, 48) and Egyptian rousette bat (49)], and the three species resistant to infection scored either low [pig (45, 49)] or very low [mallard and red junglefowl (45, 49)].
A discrepancy was observed for ferret, which had a low ACE2 binding score but is susceptible to infection (45, 49⇓–51). Ferrets may be a special case because of their unique respiratory biology (52). Ferrets are highly susceptible to upper respiratory tract infections and serve as models of respiratory diseases. They are susceptible to many viral diseases, including influenza type A and type B, canine distemper, and SARS-CoV (53). It has been proposed that ACE2 receptor distribution does not match the tropism of SARS-CoV in ferrets, because in ferrets viruses may use LSECTin receptor(s) to enable or enhance infectivity (52, 54). This may also be true for SARS-CoV-2 because the virus can potentially be glycosylated at 22 N-linked sites (55). Several studies have demonstrated SARS-CoV-2 infection in ferrets through intranasal inoculation of high doses (>105 plaque-forming units) of tissue-cultured virus, followed by direct or indirect transmission to na?ve ferrets (45, 49⇓–51). However, experimental infection via direct inoculation of high concentrations of tissue-cultured virus does not necessarily indicate infectability under natural conditions, and clinical signs of infection differed among studies. These data indicate that experimentally inoculated ferrets may become infected by another mechanism, possibly via high expression levels of low-affinity ACE2 and/or their very efficient LSECTin system.
Mammals with Predicted High Risk of SARS-CoV-2 Infection.
Of the 19 catarrhine primates analyzed, 18/19 scored very high for binding of their ACE2 to SARS-CoV-2 S and one scored high (the Angola colobus); the 18 species scoring very high had 25/25 binding residues identical to human ACE2, including rhesus macaques, which are known to be infected by SARS-CoV-2 and develop COVID-19-like clinical symptoms (3, 43). Our analysis predicts that all Old-World primates are susceptible to infection by SARS-CoV-2 via ACE2. Thus, many of the 21 primate species native to China could be a potential reservoir for SARS-CoV-2. The remaining primate species were scored as high or medium, with only the gray mouse lemur and the Philippine tarsier scoring as low.
Although inconsistent with the species phylogeny, and overall similarity to human ACE2, we found that all three species of cervid deer and 12/14 cetacean species have high scores for binding of their ACE2s to SARS-CoV-2 S. There are 18 species of cervids found in China. While coronavirus sequences have been found in white-tailed deer (56) and gammacoronaviruses have been found in beluga whales (57, 58) and bottlenose dolphins (59), in which they are associated with respiratory diseases, the cellular receptor used by these viruses is not known. Studies of cellular infectivity in these species would provide important data for validating the prediction model.
Other Artiodactyls.
A relatively large fraction (21/30) of artiodactyl mammals were classified with medium score for ACE2 binding to SARS-CoV-2 S. These include many species that are found in Hubei Province and around the world, such as domesticated cattle, sheep, and goats, as well as many species commonly found in zoos and wildlife parks (e.g., Masai giraffe, okapi, hippopotamus, water buffalo, scimitar-horned oryx, and dama gazelle). Although the cattle-derived MDBK cell line was shown in one study to be resistant to SARS-CoV-2 in vitro (60), our predictions suggest that ruminant artiodactyls can serve as a reservoir for SARS-CoV-2, which would have significant epidemiological implications as well as implications for food production and wildlife management (discussed below). It is noteworthy that camels and pigs, known for their ability to be infected by other coronaviruses (28), both score low in our analysis. These data are consistent with results (discussed above) indicating that pigs cannot be infected with SARS-CoV-2 either in vivo (45) or in vitro (60) but inconsistent with transfection studies using pig ACE2 receptors expressed in HeLa cells (1).
Rodents.
Among the rodents, 7/46 species score high for ACE2 binding to SARS-CoV-2 S, and the remaining 11, 10, and 18 score medium, low, or very low, respectively. House mouse scored very low, consistent with infectivity studies (1, 60). Given that wild rodent species likely come in contact with bats as well as with other predicted high-risk species, rodents with high and medium scores cannot be excluded as possible intermediate hosts for SARS-CoV-2.
Bats and Other Species of Interest.
Chiroptera represents a clade of mammals that are of high interest in COVID-19 research because several bat species are known to harbor coronaviruses, including those most closely related to SARS-CoV-2 (1). We analyzed ACE2 from 37 bat species, of which 8 and 29 scored low and very low, respectively. These results were intriguing because the three Rhinolophus spp. tested, including the Chinese rufous horseshoe bat, are major suspects in the transmission of SARS-CoV-2, or a closely related virus, to humans (1). Bats have been shown to harbor the highest diversity of betacoronaviruses among mammals (28) and show little pathology in individuals carrying these viruses (61).
Do bat ACE2 receptors bind SARS-CoV-2 S? Zhou et al. (1) transfected human ACE2-negative HeLa cells with ACE2 from a Chinese rufous horseshoe bat and obtained a low-efficiency infection with SARS-CoV-2. A recent report indicates that SARS-CoV-2 S protein can bind vesicular stomatitis virus (VSV) pseudotypes expressing halcyon horseshoe bat (Rhinolophus alcyone) ACE2 in BHK-21 cells (60). However, cell lines derived from big brown bat (Eptesicus fuscus) (62), Lander’s horseshoe bat (Rhinolophus landeri), and Daubenton’s bat (Myotis daubentonii) could not be infected with SARS-CoV-2 (60). Relatedly, cell lines from six different species of bats could not be infected with SARS-CoV, which also uses human ACE2 as a receptor (63). These data suggest that some bat species have evolved ACE2 receptors that do not bind SARS-CoV-like viruses or bind them with very low affinity, which is supported by our results showing positive selection and accelerated evolution of ACE2 in chiropterans. Alternatively, ACE2 expression could be very low in the bat cell lines, or SARS-CoV-2-like viruses can use other receptors, such as the MERS-CoV, a betacoronavirus that uses CD26/DPP4 (15), and porcine transmissible enteritis virus, an alphacoronavirus that uses aminopeptidase N (64). Also, other molecules required for SARS-CoV infection, such as TMPRSS2, might not be sufficiently expressed or function differently in bats.
Whether an ancestor of SARS-CoV-2, such as RaTG13, utilizes bat ACE2 is an important question related to whether bat ACE2 receptors bind SARS-CoV-2 S (discussed above). RaTG13 was found in feces of the intermediate horseshoe bat (Rhinolophus affinis) (1), but to our knowledge this virus has not been shown to bind to ACE2 of R. affinis or any other bat species. In addition, RaTG13 was reported not to infect human cells expressing Rhinolophus sinicus ACE2 in a recent study (65). Relatedly, Hoffman et al. (63) were unable to infect bat kidney- and lung-derived cell lines derived from six different species with VSV pseudotypes bearing SARS-CoV S protein or pseudotypes of two bat SARS-related CoV (Bg08 and Rp3) (63). Lack of concordance between the presence of bat SARS-CoV-like coronaviruses and binding to bat ACE2 may arise because of variations in susceptibility among bat species to SARS-CoV-like coronaviruses or due to one of the mechanisms discussed above.
Carnivores.
Recent reports of a Malayan tiger and a domestic cat infected by SARS-CoV-2 suggest that the virus can be transmitted to other felids (8, 45). Our results are consistent with these studies; 9/9 felids we analyzed scored medium for ACE2 binding of SARS-CoV-2 S. However, the masked palm civet, a member of the Viverridae family that is related to but distinct from Felidae and proposed as the intermediate host for SARS-CoV, scored as very low. While our results are inconsistent with transfection studies using civet ACE2 receptors expressed in HeLa cells (1), these experiments have limitations as discussed above, and no data are available on infectivity in civet cells or animals. While carnivores closely related to dogs (dingoes, maned wolves, and foxes) all scored low, experimental data consistently show that dogs are not readily infected or symptomatic (45, 60, 66).
Pangolins.
Considerable controversy surrounds reports that pangolins can serve as an intermediate host for SARS-CoV-2, with some reports proposing that SARS-CoV-2 arose as a recombinant between bat and pangolin betacoronaviruses (21, 67), while another study rejected that claim (68). In our study, ACE2 of Chinese pangolin, Sunda pangolin, and white-bellied pangolin had low or very low binding score for SARS-CoV-2 S. Binding of pangolin ACE2 to SARS-CoV-2 S was predicted using molecular binding simulations (67); however, neither experimental infection nor in vitro infection with SARS-CoV-2 has been reported for pangolins. Further studies are necessary to resolve whether SARS-CoV2 S binds to pangolin ACE2.
Other Vertebrates.
Our analysis of species in 29 orders of fishes, 29 orders of birds, 3 orders of reptiles, and 2 orders of amphibians predicts that the ACE2 proteins of species within these vertebrate classes are not likely to bind SARS-CoV-2 S. Thus, vertebrate classes other than mammals are not likely to be an intermediate host or reservoir for the virus, despite predictions reported in a recent study (39), unless SARS-CoV-2 uses another receptor for infection. With diverse nonmammal vertebrates sold in the seafood and wildlife markets of Asia and elsewhere, it is important to determine if SARS-CoV-2 can be found in nonmammalian vertebrates.
Animal Models for COVID-19.
Presently, there is a tremendous need for animal models to study SARS-CoV-2 infection and pathogenesis, as the only species currently known to be infected and show similar symptoms of COVID-19 is rhesus macaque. Nonhuman primate models have proven to be highly valuable for other infectious diseases but are expensive to maintain and numbers of experimental animals are limited. Our results provide an extended list of potential animal models for SARS-CoV-2 infection and pathogenesis, including large animals maintained for biomedical and agricultural research (e.g., domesticated sheep and cattle), and Chinese hamster and Syrian/golden hamster (47), which may be preferred due to their easier handling and already established value as models for other human diseases caused by viruses (69).
Relevance to Threatened Species.
Among the 103 species that scored very high, high, and medium for ACE2/SARS-CoV-2 S binding, 41 (40%) are classified in one of three “threatened” categories (vulnerable, endangered, and critically endangered) on the International Union of Conservation of Nature (IUCN) Red List of Threatened Species, five are classified as near threatened, and two species are classified as extinct in the wild (70) (Dataset S1). This represents only a small fraction of the threatened species potentially susceptible to SARS-CoV-2. For example, all 20 catarrhine primate species in our analysis, representing three families (Cercopithecidae, Hylobatidae, and Hominidae) scored very high, suggesting that all 185 species of catarrhine primates, including 62 classified as threatened, are potentially susceptible to SARS-CoV-2. Similarly, all three species of deer, representatives of a family of ∼92 species (Cervidae), including 25 classified as threatened, scored as high. In contrast, some threatened species scored low or very low, such as the giant panda (low), potentially positive news for these at-risk populations.
In Cetacea, 12 of 14 species score as high, and of those two are threatened. Toothed whales have potential for viral outbreaks and have lost function of a gene that is key to the antiviral response in other mammalian lineages (71). If they are susceptible to SARS-CoV-2, human-to-animal transmission could pose a risk through sewage outfall (72) and contaminated refuse from cities, commercial vessels, and cruise liners (73). Our results have practical implications for populations of threatened species in the wild and those under human care (including those in zoos). Established guidelines for minimizing potential human-to-animal transmission should be implemented and strictly followed. Guidelines for field researchers working on great apes established by the IUCN have been in place since 2015 in response to previous human disease outbreaks (74) and have received renewed attention because of SARS-CoV-2 (74⇓–76). For zoos, guidelines in response to SARS-CoV-2 have been distributed by several taxon advisory groups of the North American Association of Zoos and Aquariums, the American Association of Zoo Veterinarians, and the European Association of Zoo and Wildlife Veterinarians, and these organizations are actively monitoring and updating knowledge of species in human care considered to be potentially sensitive to infection (77, 78). Although in silico studies suggest potential susceptibility of diverse species, verification of infection potential is warranted, using cell cultures, stem cells, organoids, and other methods that do not require direct animal infection studies. Zoos and other facilities that maintain living animal collections are in a position to provide such samples for generating crucial research resources by banking tissues and cryobanking viable cell cultures in support of these efforts.
...
https://www.pnas.org/content/early/2020/08/20/2010146117
View ORCID ProfileJoana Damas, View ORCID ProfileGraham M. Hughes, View ORCID ProfileKathleen C. Keough, View ORCID ProfileCorrie A. Painter, View ORCID ProfileNicole S. Persky, View ORCID ProfileMarco Corbo, View ORCID ProfileMichael Hiller, View ORCID ProfileKlaus-Peter Koepfli, View ORCID ProfileAndreas R. Pfenning, View ORCID ProfileHuabin Zhao, View ORCID ProfileDiane P. Genereux, View ORCID ProfileRoss Swofford, View ORCID ProfileKatherine S. Pollard, View ORCID ProfileOliver A. Ryder, View ORCID ProfileMartin T. Nweeia, View ORCID ProfileKerstin Lindblad-Toh, View ORCID ProfileEmma C. Teeling, View ORCID ProfileElinor K. Karlsson, and View ORCID ProfileHarris A. Lewin
PNAS first published August 21, 2020 https://doi.org/10.1073/pnas.2010146117
Edited by Scott V. Edwards, Harvard University, Cambridge, MA, and approved July 31, 2020 (received for review June 2, 2020)
Article Figures & SI Info & Metrics PDF
Significance
The novel severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is the cause of COVID-19, a major pandemic that threatens millions of human lives and the global economy. We identified a large number of mammals that can potentially be infected by SARS-CoV-2 via their ACE2 proteins. This can assist the identification of intermediate hosts for SARS-CoV-2 and hence reduce the opportunity for a future outbreak of COVID-19. Among the species we found with the highest risk for SARS-CoV-2 infection are wildlife and endangered species. These species represent an opportunity for spillover of SARS-CoV-2 from humans to other susceptible animals. Given the limited infectivity data for the species studied, we urge caution not to overinterpret the predictions of the present study.
Abstract
The novel coronavirus severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is the cause of COVID-19. The main receptor of SARS-CoV-2, angiotensin I converting enzyme 2 (ACE2), is now undergoing extensive scrutiny to understand the routes of transmission and sensitivity in different species. Here, we utilized a unique dataset of ACE2 sequences from 410 vertebrate species, including 252 mammals, to study the conservation of ACE2 and its potential to be used as a receptor by SARS-CoV-2. We designed a five-category binding score based on the conservation properties of 25 amino acids important for the binding between ACE2 and the SARS-CoV-2 spike protein. Only mammals fell into the medium to very high categories and only catarrhine primates into the very high category, suggesting that they are at high risk for SARS-CoV-2 infection. We employed a protein structural analysis to qualitatively assess whether amino acid changes at variable residues would be likely to disrupt ACE2/SARS-CoV-2 spike protein binding and found the number of predicted unfavorable changes significantly correlated with the binding score. Extending this analysis to human population data, we found only rare (frequency <0.001) variants in 10/25 binding sites. In addition, we found significant signals of selection and accelerated evolution in the ACE2 coding sequence across all mammals, and specific to the bat lineage. Our results, if confirmed by additional experimental data, may lead to the identification of intermediate host species for SARS-CoV-2, guide the selection of animal models of COVID-19, and assist the conservation of animals both in native habitats and in human care.
...
Discussion
Phylogenetic analysis of coronaviruses has demonstrated that the immediate ancestor of SARS-CoV-2 most likely originated in a bat species (1). However, whether SARS-CoV-2 or the progenitor of this virus was transmitted directly to humans or through an intermediate host is not yet resolved. To identify candidate intermediate host species and species at risk for SARS-CoV-2 infection, we undertook a deep comparative genomic, evolutionary, and structural analysis of ACE2, which serves as the SARS-CoV-2 receptor in humans. We drew on the rapidly growing database of annotated vertebrate genomes, including new genomes produced by the Genomes 10K-affiliated Bat1K Consortium, Zoonomia, and Vertebrate Genomes Project, and other sources (30, 31). We conducted a phylogenetic analysis of ACE2 orthologs from 410 vertebrate species and predicted their propensity to bind the SARS-CoV-2 S using a score based on amino acid substitutions at 25 consensus human ACE2 binding residues (12, 20). Similarity-based methods are frequently used for predicting cross-species transmission of viruses (32, 33), including SARS-CoV (2). We supported these predictions with comprehensive structural analysis of the ACE2 binding site complexed with SARS-CoV-2 S. We also tested the hypothesis that the ACE2 receptor is under selective constraints in mammalian lineages with different susceptibilities to coronaviruses.
We predict that species scoring as very high and high for propensity of SARS-CoV-2 S binding to ACE2 will have a high probability of becoming infected by the virus and thus may be potential intermediate hosts for virus transmission. We also predict that many species having a medium score have some risk of infection, and species scored as very low and low are less likely to be infected by SARS-CoV-2 via the ACE2 receptor. Importantly, our predictions are based solely on in silico analyses and must be confirmed by direct experimental data. The prediction accuracy of the model may be improved in the future as more extensive data are generated showing the impact of ACE2 mutations on its binding affinity for SARS-CoV-2 S, which may enable knowledge-based weighting of residues in the scoring algorithm. Until the present model’s accuracy can be confirmed with additional experimental data, we urge caution not to overinterpret the predictions of the present study. This is especially important with regards to species, endangered or otherwise, in human care. While species ranked high or medium may be susceptible to infection based on the features of their ACE2 residues, pathological outcomes may be very different among species depending on other mechanisms, such as immune response, that could affect virus replication and spread to target cells, tissues, and organs. Furthermore, we cannot exclude the possibility that infection in any species occurs via another cellular receptor (for a review see ref. 34), as shown for other betacoronaviruses (35), or lower-affinity interactions with ACE2 as proposed for SARS-CoV (2). Nonetheless, our predictions provide a useful starting point for the selection of appropriate animal models for COVID-19 research and identification of species that may be at risk for human-to-animal or animal-to-animal transmissions of SARS-CoV-2.
Several recent studies examined the role of ACE2 in SARS-CoV-2 binding and cellular infection and its relationship to experimental and natural infections in different species (26, 35⇓⇓⇓⇓–40). Our study design differs substantially from those in several aspects: 1) we analyzed a larger number of primates, carnivores, rodents, cetartiodactyls, and other mammalian orders and an extensive phylogenetic sampling of fishes, birds, amphibians, and reptiles; 2) we analyzed the full set of S-binding residues across the ACE2 binding site, which was based on a consensus set from two independent studies (12, 20); 3) we used different methodologies to assess ACE2 binding capacity for SARS-CoV-2 S; and 4) our study tested for selection and accelerated evolution across the entire ACE2 protein. While our results are consistent with the results and conclusions of Melin et al. (38) on the predicted susceptibility of primates to SARS-CoV-2, particularly Old-World primates, we made predictions for a larger number of primates (n = 39 vs. n = 27), bats (n = 37 vs. n = 7), other mammals (n = 176 vs. n = 5), and other vertebrates (n = 158 vs. n = 0). When ACE2 from species in our study were compared with results of other studies there were many consistencies, such as the low risk for rodents, but some predictions differ, such as the relatively high risk predicted by others for SARS-CoV-2 S binding in pangolin and horse (39), civet (40), Chinese rufous horseshoe bat (40), and turtles (22). Our results are generally consistent with a study that tested binding affinity of soluble ACE2 for the SARS-CoV-2 S RBD using saturation mutagenesis (27), particularly in the binding hot-spot region of ACE2 residues 353 to 357 (SI Appendix, Fig. S1). Importantly, as compared with other studies, our results greatly expanded the number of candidate intermediate hosts and identified many additional threatened species that could be at risk for SARS-CoV-2 infection via their ACE2 receptors.
Evolution of ACE2.
Variation in ACE2 in the human population is rare (16). Overall, ACE2 is intolerant of loss-of-function mutations [pLI = 0.998; LOEUF = 0.25 in gnomAD v2.1.1 (16)]. We examined a large set of ACE2 variants for their potential differences in binding to SARS-CoV-2 S and their relationship to selected and accelerated sites. We found rare coding variants that would result in missense mutations causing substitutions in 7/25 binding residues (Dataset S2). Some of those [e.g., E35K, f = 0.00001636 (16)] could reduce the virus binding affinity as per our structural analysis (Dataset S2) but would potentially lower the susceptibility to the virus only in a very small fraction of the population. Our analysis suggests that some variants (e.g., D38E) might not affect binding propensity while the potential impact of others (e.g., S19P) could not be determined. Further investigations on the effects of these rare variants on ACE2/SARS-CoV-2 binding affinity are needed.
When exploring patterns of codon evolution in ACE2, we found that multiple ACE2 residues important for the binding of SARS-CoV-2 S are evolving rapidly across mammals, with two (Q24 and H34) under positive selection (Fig. 4 A and B and SI Appendix, Fig. S5). Relative to other lineages analyzed, Chiroptera has a greater proportion of accelerated versus conserved codons (SI Appendix, Fig. S6), particularly in the SARS-CoV-2 S-binding region, suggesting the possibility of selective forces on these codons in Chiroptera driven by their interactions with SARS-CoV-2-like viruses (Fig. 4 C and D and Dataset S4.13). Indeed, distinct signatures of positive selection found in bat ACE2 (41) and in the SARS-CoV-2 S protein (42) support the hypothesis that bats are evolving to tolerate SARS-CoV-2-like viruses (discussed further below).
Relationship of the ACE2 Binding Score to Known Infectivity of SARS-CoV-2.
Data on susceptibility of nonhuman species to SARS-CoV-2 is still very limited (SI Appendix, Fig. S10) but mostly agree with our predictions of ACE2 binding propensity for SARS-CoV-2 S (Figs. 1 and 2 and Dataset S1). Five out of six species with demonstrated susceptibility to SARS-CoV-2 infection score very high [rhesus macaque (43) and cynomolgus macaque (44)] or medium [domestic cat (45, 46), tiger (8) and golden hamster (47)]. Both species susceptible to infection but asymptomatic scored low [dog (45, 48) and Egyptian rousette bat (49)], and the three species resistant to infection scored either low [pig (45, 49)] or very low [mallard and red junglefowl (45, 49)].
A discrepancy was observed for ferret, which had a low ACE2 binding score but is susceptible to infection (45, 49⇓–51). Ferrets may be a special case because of their unique respiratory biology (52). Ferrets are highly susceptible to upper respiratory tract infections and serve as models of respiratory diseases. They are susceptible to many viral diseases, including influenza type A and type B, canine distemper, and SARS-CoV (53). It has been proposed that ACE2 receptor distribution does not match the tropism of SARS-CoV in ferrets, because in ferrets viruses may use LSECTin receptor(s) to enable or enhance infectivity (52, 54). This may also be true for SARS-CoV-2 because the virus can potentially be glycosylated at 22 N-linked sites (55). Several studies have demonstrated SARS-CoV-2 infection in ferrets through intranasal inoculation of high doses (>105 plaque-forming units) of tissue-cultured virus, followed by direct or indirect transmission to na?ve ferrets (45, 49⇓–51). However, experimental infection via direct inoculation of high concentrations of tissue-cultured virus does not necessarily indicate infectability under natural conditions, and clinical signs of infection differed among studies. These data indicate that experimentally inoculated ferrets may become infected by another mechanism, possibly via high expression levels of low-affinity ACE2 and/or their very efficient LSECTin system.
Mammals with Predicted High Risk of SARS-CoV-2 Infection.
Of the 19 catarrhine primates analyzed, 18/19 scored very high for binding of their ACE2 to SARS-CoV-2 S and one scored high (the Angola colobus); the 18 species scoring very high had 25/25 binding residues identical to human ACE2, including rhesus macaques, which are known to be infected by SARS-CoV-2 and develop COVID-19-like clinical symptoms (3, 43). Our analysis predicts that all Old-World primates are susceptible to infection by SARS-CoV-2 via ACE2. Thus, many of the 21 primate species native to China could be a potential reservoir for SARS-CoV-2. The remaining primate species were scored as high or medium, with only the gray mouse lemur and the Philippine tarsier scoring as low.
Although inconsistent with the species phylogeny, and overall similarity to human ACE2, we found that all three species of cervid deer and 12/14 cetacean species have high scores for binding of their ACE2s to SARS-CoV-2 S. There are 18 species of cervids found in China. While coronavirus sequences have been found in white-tailed deer (56) and gammacoronaviruses have been found in beluga whales (57, 58) and bottlenose dolphins (59), in which they are associated with respiratory diseases, the cellular receptor used by these viruses is not known. Studies of cellular infectivity in these species would provide important data for validating the prediction model.
Other Artiodactyls.
A relatively large fraction (21/30) of artiodactyl mammals were classified with medium score for ACE2 binding to SARS-CoV-2 S. These include many species that are found in Hubei Province and around the world, such as domesticated cattle, sheep, and goats, as well as many species commonly found in zoos and wildlife parks (e.g., Masai giraffe, okapi, hippopotamus, water buffalo, scimitar-horned oryx, and dama gazelle). Although the cattle-derived MDBK cell line was shown in one study to be resistant to SARS-CoV-2 in vitro (60), our predictions suggest that ruminant artiodactyls can serve as a reservoir for SARS-CoV-2, which would have significant epidemiological implications as well as implications for food production and wildlife management (discussed below). It is noteworthy that camels and pigs, known for their ability to be infected by other coronaviruses (28), both score low in our analysis. These data are consistent with results (discussed above) indicating that pigs cannot be infected with SARS-CoV-2 either in vivo (45) or in vitro (60) but inconsistent with transfection studies using pig ACE2 receptors expressed in HeLa cells (1).
Rodents.
Among the rodents, 7/46 species score high for ACE2 binding to SARS-CoV-2 S, and the remaining 11, 10, and 18 score medium, low, or very low, respectively. House mouse scored very low, consistent with infectivity studies (1, 60). Given that wild rodent species likely come in contact with bats as well as with other predicted high-risk species, rodents with high and medium scores cannot be excluded as possible intermediate hosts for SARS-CoV-2.
Bats and Other Species of Interest.
Chiroptera represents a clade of mammals that are of high interest in COVID-19 research because several bat species are known to harbor coronaviruses, including those most closely related to SARS-CoV-2 (1). We analyzed ACE2 from 37 bat species, of which 8 and 29 scored low and very low, respectively. These results were intriguing because the three Rhinolophus spp. tested, including the Chinese rufous horseshoe bat, are major suspects in the transmission of SARS-CoV-2, or a closely related virus, to humans (1). Bats have been shown to harbor the highest diversity of betacoronaviruses among mammals (28) and show little pathology in individuals carrying these viruses (61).
Do bat ACE2 receptors bind SARS-CoV-2 S? Zhou et al. (1) transfected human ACE2-negative HeLa cells with ACE2 from a Chinese rufous horseshoe bat and obtained a low-efficiency infection with SARS-CoV-2. A recent report indicates that SARS-CoV-2 S protein can bind vesicular stomatitis virus (VSV) pseudotypes expressing halcyon horseshoe bat (Rhinolophus alcyone) ACE2 in BHK-21 cells (60). However, cell lines derived from big brown bat (Eptesicus fuscus) (62), Lander’s horseshoe bat (Rhinolophus landeri), and Daubenton’s bat (Myotis daubentonii) could not be infected with SARS-CoV-2 (60). Relatedly, cell lines from six different species of bats could not be infected with SARS-CoV, which also uses human ACE2 as a receptor (63). These data suggest that some bat species have evolved ACE2 receptors that do not bind SARS-CoV-like viruses or bind them with very low affinity, which is supported by our results showing positive selection and accelerated evolution of ACE2 in chiropterans. Alternatively, ACE2 expression could be very low in the bat cell lines, or SARS-CoV-2-like viruses can use other receptors, such as the MERS-CoV, a betacoronavirus that uses CD26/DPP4 (15), and porcine transmissible enteritis virus, an alphacoronavirus that uses aminopeptidase N (64). Also, other molecules required for SARS-CoV infection, such as TMPRSS2, might not be sufficiently expressed or function differently in bats.
Whether an ancestor of SARS-CoV-2, such as RaTG13, utilizes bat ACE2 is an important question related to whether bat ACE2 receptors bind SARS-CoV-2 S (discussed above). RaTG13 was found in feces of the intermediate horseshoe bat (Rhinolophus affinis) (1), but to our knowledge this virus has not been shown to bind to ACE2 of R. affinis or any other bat species. In addition, RaTG13 was reported not to infect human cells expressing Rhinolophus sinicus ACE2 in a recent study (65). Relatedly, Hoffman et al. (63) were unable to infect bat kidney- and lung-derived cell lines derived from six different species with VSV pseudotypes bearing SARS-CoV S protein or pseudotypes of two bat SARS-related CoV (Bg08 and Rp3) (63). Lack of concordance between the presence of bat SARS-CoV-like coronaviruses and binding to bat ACE2 may arise because of variations in susceptibility among bat species to SARS-CoV-like coronaviruses or due to one of the mechanisms discussed above.
Carnivores.
Recent reports of a Malayan tiger and a domestic cat infected by SARS-CoV-2 suggest that the virus can be transmitted to other felids (8, 45). Our results are consistent with these studies; 9/9 felids we analyzed scored medium for ACE2 binding of SARS-CoV-2 S. However, the masked palm civet, a member of the Viverridae family that is related to but distinct from Felidae and proposed as the intermediate host for SARS-CoV, scored as very low. While our results are inconsistent with transfection studies using civet ACE2 receptors expressed in HeLa cells (1), these experiments have limitations as discussed above, and no data are available on infectivity in civet cells or animals. While carnivores closely related to dogs (dingoes, maned wolves, and foxes) all scored low, experimental data consistently show that dogs are not readily infected or symptomatic (45, 60, 66).
Pangolins.
Considerable controversy surrounds reports that pangolins can serve as an intermediate host for SARS-CoV-2, with some reports proposing that SARS-CoV-2 arose as a recombinant between bat and pangolin betacoronaviruses (21, 67), while another study rejected that claim (68). In our study, ACE2 of Chinese pangolin, Sunda pangolin, and white-bellied pangolin had low or very low binding score for SARS-CoV-2 S. Binding of pangolin ACE2 to SARS-CoV-2 S was predicted using molecular binding simulations (67); however, neither experimental infection nor in vitro infection with SARS-CoV-2 has been reported for pangolins. Further studies are necessary to resolve whether SARS-CoV2 S binds to pangolin ACE2.
Other Vertebrates.
Our analysis of species in 29 orders of fishes, 29 orders of birds, 3 orders of reptiles, and 2 orders of amphibians predicts that the ACE2 proteins of species within these vertebrate classes are not likely to bind SARS-CoV-2 S. Thus, vertebrate classes other than mammals are not likely to be an intermediate host or reservoir for the virus, despite predictions reported in a recent study (39), unless SARS-CoV-2 uses another receptor for infection. With diverse nonmammal vertebrates sold in the seafood and wildlife markets of Asia and elsewhere, it is important to determine if SARS-CoV-2 can be found in nonmammalian vertebrates.
Animal Models for COVID-19.
Presently, there is a tremendous need for animal models to study SARS-CoV-2 infection and pathogenesis, as the only species currently known to be infected and show similar symptoms of COVID-19 is rhesus macaque. Nonhuman primate models have proven to be highly valuable for other infectious diseases but are expensive to maintain and numbers of experimental animals are limited. Our results provide an extended list of potential animal models for SARS-CoV-2 infection and pathogenesis, including large animals maintained for biomedical and agricultural research (e.g., domesticated sheep and cattle), and Chinese hamster and Syrian/golden hamster (47), which may be preferred due to their easier handling and already established value as models for other human diseases caused by viruses (69).
Relevance to Threatened Species.
Among the 103 species that scored very high, high, and medium for ACE2/SARS-CoV-2 S binding, 41 (40%) are classified in one of three “threatened” categories (vulnerable, endangered, and critically endangered) on the International Union of Conservation of Nature (IUCN) Red List of Threatened Species, five are classified as near threatened, and two species are classified as extinct in the wild (70) (Dataset S1). This represents only a small fraction of the threatened species potentially susceptible to SARS-CoV-2. For example, all 20 catarrhine primate species in our analysis, representing three families (Cercopithecidae, Hylobatidae, and Hominidae) scored very high, suggesting that all 185 species of catarrhine primates, including 62 classified as threatened, are potentially susceptible to SARS-CoV-2. Similarly, all three species of deer, representatives of a family of ∼92 species (Cervidae), including 25 classified as threatened, scored as high. In contrast, some threatened species scored low or very low, such as the giant panda (low), potentially positive news for these at-risk populations.
In Cetacea, 12 of 14 species score as high, and of those two are threatened. Toothed whales have potential for viral outbreaks and have lost function of a gene that is key to the antiviral response in other mammalian lineages (71). If they are susceptible to SARS-CoV-2, human-to-animal transmission could pose a risk through sewage outfall (72) and contaminated refuse from cities, commercial vessels, and cruise liners (73). Our results have practical implications for populations of threatened species in the wild and those under human care (including those in zoos). Established guidelines for minimizing potential human-to-animal transmission should be implemented and strictly followed. Guidelines for field researchers working on great apes established by the IUCN have been in place since 2015 in response to previous human disease outbreaks (74) and have received renewed attention because of SARS-CoV-2 (74⇓–76). For zoos, guidelines in response to SARS-CoV-2 have been distributed by several taxon advisory groups of the North American Association of Zoos and Aquariums, the American Association of Zoo Veterinarians, and the European Association of Zoo and Wildlife Veterinarians, and these organizations are actively monitoring and updating knowledge of species in human care considered to be potentially sensitive to infection (77, 78). Although in silico studies suggest potential susceptibility of diverse species, verification of infection potential is warranted, using cell cultures, stem cells, organoids, and other methods that do not require direct animal infection studies. Zoos and other facilities that maintain living animal collections are in a position to provide such samples for generating crucial research resources by banking tissues and cryobanking viable cell cultures in support of these efforts.
...
https://www.pnas.org/content/early/2020/08/20/2010146117