Soon after the discovery of SARS-CoV-2, the cell surface exopeptidase ACE2 was found to serve as viral receptor in human and the first investigation of species susceptibility to this new virus demonstrated that SARS-CoV-2 is able to use Chinese horseshoe bat, swine, but not mouse ACE2 to bind host cells 41. Since this pioneering work, several laboratories have intended to predict the utilizing capability by SARS-CoV-2 of ACE2 from different species using amino acids sequence comparisons aimed at identifying the possible intermediate hosts of SARS-CoV-2. This was made possible after the crystallographic analyzes determining which amino acids of ACE2 are essential for the attachment of the viral spike protein 11.
Our investigation suggests that SARS-CoV-like ancestral coronaviruses have adapted the ACE2 receptor to replicate in bats. However, our analysis also suggests that probably not all bat species support SARS-CoV-like coronavirus replication. According to multisequence alignment, Rhinolophus bats appear to be appropriated candidates for replication of SARS- CoV-2 related viruses, yet a species polymorphism is observed among the Rhinolophus. Rhinolophus sinicus with K31, Y41H, N82, N90, K353 is a good candidate for SARS-CoV-2- like virus replication whereas Rhinolophus ferrumequinum with K31D, Y41H, N82, N90, K353 can be predicted less susceptible to the virus. ACE2 sequences from other bat species show increased amino acids substitutions at positions considered required for viral spike binding (e.g., Desmodus rotundus with K31N, Y41, N82T, N90D, K353N). In species expressing variant ACE2 not suitable for virus binding another surface receptor could serves for viral entry into cells but such viruses will be less likely to cross species barriers using an
ACE2 protein as receptor in an intermediate host species. This can support the hypothesis of a long bat and virus co-evolution with bat species which replicate ACE2-tropic viruses like SARS-CoV and other species which replicate CD26-tropic viruses like MERS-CoV.
In order that a SARS-CoV-like virus leave bats to infect a intermediate host in between bat and human, infected bat must come into contact with an animal expressing an ACE2 receptor adapted to SARS-CoV-like virus binding. In agreement with other studies 38, 39,41, our in silico search for intermediate host species able to pass the SARS-CoV-2 to humans supports the hypothesis that species bearing K31 and K353 amino acids are more likely to be susceptible to SARS-CoV-2. For example, ACE2 from Manis javanica, Mustela putorius furo and Felis catus, considered SARS-CoV-2 susceptible species, show K31 and K353 amino acids whereas Mus musculus considered SARS-CoV-2 resistant species shows a K31N and K353H variant. A Y41 also seems to be important, yet Rhinolophus sinicus ACE2 expresses a Y41H variant. It may account for the requirement of an intermediate host before being able to infect humans. A position not particularly stressed in previous SARS-CoV-2 studies that appear important, is N90. Indeed the N90 that is found in Homo sapiens final host and Rhinolophus sinicus early host, is also found in Macaca mulatta, Manis javanica, Felis catus, and Pelodiscus sinensis, previously described susceptible to SARS-CoV-2 and possible intermediate host whereas N90D or N90T variants are found in the other species studied. This is also consistent with the earlier observation indicating that N90 was important for SARS- CoV binding to ACE2 32. However, what is surprising is the sequence of the Paguma larvata with K31T, Y41, N82T, N90D, and K353, since palm civet has been considered as the intermediate host for SARS-CoV and suggested to also serve as an intermediate host for SARS-CoV- 2 40, whereas with the absence of K31, the absence of N82 and N90 (which are expected to be glycosylated thereby favoring interaction with the viral spike), palm civet appears to be an animal unlikely to be infected through ACE2. This discrepancy should be further explored. Obviously, not all the species tested are theoretically susceptible to infection by SARS-CoV-2. The ACE2 protein should express amino acids essential for the viral spike binding and variants ACE2 that lack such amino acids are not likely to allow virus binding and entry. Qiu and colleagues 39 compared the ACE2 sequences from 250 species with a specific focus on T20, K31, Y41, K68, Y83, S218, A246, K353, D355, R357, M383, P426,
T593, N636, A714, R716, and A774 and concluded that SARS-CoV-2 might bind Manis javanica (pangolin), Felis catus (cat), Bos taurus (cow), Bubalus bubalus (buffalo), Capra hircus (goat), Ovis aries (sheep) and Columbia livia (pigeon) ACE2 but not (Mus musculus) murine ACE2. They also suggested to paid attention to Protobothrops mucrosquamatus
(pallas pit viper) a common snake living in the Hubei Province of China. In their study Luan and colleagues 40, investigated 42 mammalian ACE2 proteins from the wild animal protection list of Hubei Province. In their study the authors focused attention on key amino acids K31, E35, D38, M82, and K353. According to their predictions, they considered that beside humans, the mammals whose ACE2 could bind to the S protein of SARSCoV-2 are bats (Rhinolophus macrotis, Rhinolophus sinicus, Rhinolophus pearsonii, Pteropus vampyrus, Rousettus leschenaultii,), pangolin (Manis javanica), palm civet (Paguma larvata), monkeys (Macaca mulatta, Pan troglodytes, Pongo abelii, Papio Anubis, Callithrix jacchus), cat (Felis catus), dog (Canis lupus familiaris), ferret (Mustela putorius furo), pig (Sus scrofa domesticus) among others (Rhinopithecus roxellana, Mustela erminea, Sus scrofa, Equus caballus, Bos taurus, Ovis aries, Oryctolagus cuniculus, Vulpes vulpes, Phodopus campbelli, Mesocricetus auratus, Heterocephalus glaber, Ictidomys tridecemlineatus, and Cricetulus griseus). The mammals whose ACE2 appeared unable to bind to S protein of SARS-CoV-2 included the Rhinolophus ferrumequinum bats, rat (Rattus norvegicus), mouse (Mus musculus), camel (Camelus dromedarius), and others (Procyon lotor, Ornithorhynchus anatinus, Loxodonta africana, Erinaceus europaeus, Nyctereutes procyonoides, Suricata suricatta, Dipodomys ordii, and Cavia porcellus). They draw particular attention to N82 amino acid in the ACE2 protein. Another study by Liu and colleagues 38, based on prediction of interactions between the S protein of SARS-CoV-2 and ACE2, that investigated monkey (Gorilla, Macaca), bat (Rhinolophus sinicus; Rhinolophus pearsonii), pangolin (Manis javanica), snake (Ophiophagus hannah), turtles (Chrysemys picta bellii, Chelonia mydas and Pelodiscus sinensis), and others (dog, cat, mouse), stressed a possible role as intermediate host animal reservoir for turtles. This study which focused on positions T27, F28, D30, K31, H34, D38, Y41, Q42, M82, E329, K353, G354, D355, and R357, indicated that mouse and dog ACE2 showed multiple substitutions (>5) among the 14 amino acids that retained their attention, an observation in agreement with the relative resistance of these species to infection by SARS-CoV-2. They suggested K31, Y41 and K353 to be key amino acids for viral spike binding.
Although the in silico studies have the advantage of being able to quickly investigate the probability of infection of a large number of species, nothing can replace the in vivo experimentation. Interestingly golden hamster (Mesocricetus auratus) and Chinese hamster (Cricetulus griseus) are known as animal models for SARS-CoV 42,43. Monkeys (Macaca mulatta; Macaca fascicularis ; Chlorocebus aethiops) were also found to be animal models
for SARS-CoV with reports of pneumonitis in infected monkeys 44,45. Ferret (Mustela putorius furo), were also used as animal model for SARS-CoV and showed productive infection 46,47. Although mouse (Mus musculus) ACE2 was considered unable to bind SARS- CoV-2 spike, it was previously reported that young inbred mice supported SARS-CoV viral replication but failed to show clinical sign of disease 48,49. A recent (not peer-reviewed) study 50, describes the investigation of the in vivo susceptibility of animals to replicate SARS-CoV-2. The authors claim that the virus replicated poorly in dogs, pigs, chickens and ducks but efficiently infected ferrets and cats. In addition, these authors found that the virus can be transmitted from cat to cat by respiratory droplets.
If the SARS-CoV-2 like ancestral virus from bats can meet an intermediate host bearing an ACE2 molecule to which the virus spike can bind and if this lead to productive infection of the intermediate host, then another conjunction of events must occur for the virus to pass the species barrier and infect humans. Time span between these events can be very long. Upon contact with the infected intermediate animal host, the SARS-CoV-2 can meet the ACE2 protein at the surface of human lung epithelial cells allowing infection to occur. ACE2 is expressed on both type I and II alveolar epithelial lung cells as well as epithelial cells of oral mucosa, enterocytes of the small intestine, and arterial and venous endothelial cells contributing to the COVID-19 disease 51-53. In human, ACE2 is a 100kDa type I cell-surface glycoprotein of 805 amino acids. It is characterized by a NH2-terminal signal peptide of 17 amino acid residues, a peptidase domain (PD) (residues 19-615) with its zinc binding metalloprotease HEXXH motif, a C-terminal Collectrin-like domain (CLD) (residues 616- 768) that includes a ferredoxin-like domain (615-726), a transmembrane region of 22 amino acid residues followed by an intracellular segment of 43 amino acid residues 54,55. Polymorphism of ACE2 in human populations was recently well documented 56.
The question must be asked why not all coronaviruses capable of infecting humans use the ACE2 receptor to infect target cells. There is probably an initial response with MERS-CoV. Instead of binding ACE2, MERS-CoV bind the dipeptidyl peptidase 4 (DPP4)/CD26), a serine peptidase expressed by T cells 57-59. After attachment of MRES-CoV spike (S) glycoprotein to DPP4 positive human cells the S protein undergoes proteolytic activation through the cellular serine protease TMPRSS2 and cysteine protease cathepsin L once inside endosomes 60. It was recently reported that among fourteen characterized mutants forms of DPP4, four polymorphisms (K267E, K267N, A291P and D346-348) strongly reduce the binding and penetration of MERS-CoV into target cells and the viral replication 61. Indeed, MERS-CoV did not emerge in China from Rhinolophus bats, it emerged in Saudi Arabia from Pipistrellus bats. it was found that MERS-CoV was closely related to Ty-BatCoV HKU4 borne by Tylonycteris pachypus and Pi-BatCoV HKU5 borne by Pipistrellus abramus 62,63. It should be noted that a Betacoronavirus (BatCoV-P.davyi49/Mexico/2012) isolated from bats captured in Mexico presented 72% nucleotide identity with MERS-like CoV circulating in Rousettus and Pipistrellus bats 64. Although we had available only a single sequence of ACE2 from Pipistrellus bat and three sequences of Rousettus to compare to other ACE2 from other bats, it should be noted that both Pipistrellus and Rousettus express a N90D variant. If this position is really critical as we suggest, one might hypothesize that this is perhaps the reason why CD26-tropic coronaviruses would have spread in Pipistrellus abramus and Tylonycteris pachypus bats, rather than ACE2-tropic coronaviruses. Before transmission to human, the ancestral MESR-like CoV from bats crossed the species barrier to infect camels. Humans were infected through camel contacts 65-67.
In conclusion, our results suggest that species carrying a sequence with K31, Y41, N90, and K353 are likely to be susceptible to infection by SARS-CoV-2 (including Homo sapiens, Macaca mulatta, Felis catus, Rhinolophus sinicus, Manis javanica, and Pelodiscus sinensis) while others should be less susceptible to infection except if the virus adapts a second receptor for cellular binding and entry. The combination of 3-D structure analysis and electrostatic potential surface was quite informative. We found that the substitution of human ACE2 regions 30-41, 82-93 and 353-358 by the corresponding regions from Rhinolophus sinicus, Mus musculus, and Xenopus tropicalis species did not significantly changed the 3-D structure of ACE2 but modify the electrostatic potential surface of the molecule. These modifications are likely to be sufficient to alter the interaction of SARS-CoV-2 spike with the variants ACE2. The crystal structure analysis of ACE2 suggested the presence of several hinge regions and N-glycosylations 68, including the glycosylation of N90 considered essential for SARS- CoV-2 binding. This may explain why N90 may be very important for infection of host cells. Among the various in vitro antiviral activities of chloroquine described to date, it has been suggested that this molecule could prevent the glycosylation of ACE2 69,70. We could hypothesize that chloroquine blocks the N-glycosylation at position 90 of the ACE2 sequence, thereby preventing the attachment of SARS-CoV and SARS-CoV-2 spike to the receptor.