3. Sialyl Glycan Receptor-Dependent Recognition of Influenza A (H1–H16) Viruses, Egyptian Fruit Bat-Isolated Influenza A Virus, Influenza B, C and D Viruses and Lineage A βCoVs (β1CoVs)

3.1. Influenza A (H1–H16) Viruses Use Siaα2,3/2,6Gal Receptors
HAs of H1–H16 viruses recognize specific sialyl glycans on the host epithelial cell surface, as a crucial step mediating virus infection (Figure 3a). IAVs from avians, either wild birds or domestic birds, typically prefer the α2,3Sia terminal. Surprisingly, a recent study showed that some gull/tern H16 viruses prefer α2,6Neu5Ac over or equal to the α2,3Neu5Ac terminal of synthetic sialylglycopolymers [23]. It was suggested that the particularly distinctive receptor-binding specificity of H16 viruses may be related to their HAs containing A138S (found in human 1977-derived H1N1 viruses, reducing binding to α2,3Neu5Ac receptors) and E190T (the amino acid (aa) at position 190 determining binding specificity of H1N1 viruses to the sialyl linkage type) [23]. Usually, viruses adapt to bind to sialyl glycans dominant in the host target tissues. Information on virus collection from oral, nasal, nasopharyngeal, cloacal or feces swabs and sialyl glycan analysis of tissues of specific wild birds shedding the virus in the sample collection may lead to a better understanding of why some H16 viruses display binding distinct from that of other avian viruses. Binding preference for the internal part of sialyl glycans appears to differ among different viruses based on birds of isolation as indicated in Table 1.
Several zoonotic influenza virus subtypes (Table 1) including avian subtypes H5N1, H7N9 and H9N2 and swine subtypes H1N1, H1N2 and H3N2 have been occasionally reported to cross the species barrier to infect humans [57]. However, human-to-human transmission of nonhuman viruses has been limited and non-sustained [154]. Viruses in four historical pandemics acquired strong binding to human-type α2,6Neu5Ac receptors for efficient human-to-human transmission [19,155,156,157]. (i) The H1N1 Spanish pandemic in 1918–1919 was found to include at least two strains with distinct receptor-binding properties during the pandemic period [155]. First, viral HAs have a single aa substitution, E190D, in the receptor-binding site (RBS) and bind to both avian-type and human-type receptors. Second, there are two aa substitutions in the HA RBS, E190D and G225D, that enable HA adaptation to bind only to α2,6Neu5Ac receptors. (ii) The H2N2 Asian pandemic in 1957–1958 had virus isolates from two stages of the pandemic. In the early pandemic stage, virus isolates can be divided into three subpopulations based on receptor binding specificities: avian-like viruses with 226Q and 228G in the HA RBS, atypical viruses with Q226L and 228G, and classic human viruses with Q226L and G228S that have preferential binding to avian-type receptors, both avian-type and human-type receptors, and human-type receptors, respectively. In the subsequent stage, all virus isolates have Q226L and G228S substitutions with preferential binding to human-type receptors [156]. (iii) The virus in the H3N2 Hong Kong pandemic in 1968–1969 had the same acquisition of Q226L and G228S substitutions in the HA RBS as that in the H2 pandemic for switching from avian-type to human-type receptor binding preference [157]. (iv) The virus in the H1N1 swine pandemic in 2009–2010 had 190D and 225D in the HA RBS as in the swine H1 HA RBS recognizing human-type α2,6Neu5Ac receptors that are abundant in the porcine lung, which is the main site of swine IAV replication [19,39]. Protein engineering by chimeragenesis and site-directed mutagenesis of H1 proteins suggested that A200T and A227E substitutions in the H1 swine pandemic were responsible for efficient and sustained human-to-human transmission. Molecular modeling revealed hydrogen bond formation between T200 and Q191 in the 190-helix that is important for receptor binding preference of H1 HAs and between E227 and Gal next to Sia [158].
Based on historical data, after a pandemic virus continued to circulate as a seasonal strain, the preexisting seasonal virus, which donated at least three gene segments to the pandemic virus, disappeared from human circulation. The disappearance of 1918-derived H1N1, 1957-derived H2N2 and 1977-derived H1N1 (being the 1918-derived H1N1, recurrent from a research laboratory in 1977, same as the classical swine H1N1 viruses) [159] has resulted in only 1968-derived H3N2 and 2009-derived H1N1 viruses remaining in human circulation. Despite binding to human-type receptors being essential for influenza virus transmission among humans, the binding of 1968-derived H3N2 viruses to the human-type receptor analog α2,6 sialyl N-acetyllactosamine (6′SLN)-polyacrylamide started to decrease significantly in 2001 and seemed to be completely lost in 2010 [160]. However, after the discovery [161] and the widespread use of long α2,6 sialylated N-glycans with multiple LN repeats for studies on influenza virus binding specificity, it appeared that 1968-derived H3N2 viruses have evolved binding preference for human-type receptors with LacNAc (LN) repeats [60,162]. Based on the binding preferences to short 3′SLN and 6′SLN and long 3′SLNLNLN and 6′SLNLNLN linked to a polyglutamic acid, IAVs can be divided into two groups [19]. Group 1 are avian viruses, including H5N1 and H5N3 viruses, that preferentially bind to terminal α2,3Neu5Ac with either short or long LN chains. Group 2 consists of viruses that preferentially bind to terminal α2,6Neu5Ac, and the viruses can be further divided into two subgroups. Subgroup 2-1 includes swine H1N2/2008 and pdm H1N1/2009 viruses, which can bind to both short and long α2,6 sialylated glycans. These results support the hypothesis that pigs are vessels to generate viral HAs with pandemic potential [41]. However, the pdm H1N1/2009 viruses acquired at least two amino acids that are different from the swine H1 HA, A200T and A227E, and they are responsible for the binding differences in fetuin, chicken erythrocytes and human erythrocytes and are believed to be determinants of the shift in binding specificity from swine-type to human-type [158]. Further investigation to find the α2,6 sialyl glycan structure that is able to clearly distinguish binding specificity between swine and pandemic viruses is needed since such sialyl glycans could be useful for surveillance and prevention of a pandemic. Subgroup 2-2 consists of long-term circulating human viruses including human H3N2/2008 viruses and human H1N1/2004 and H1N1/2006 viruses, which have binding preference to the long α2,6 sialyl glycan.
Structural comparison of avian-type and human-type receptors interacting with the receptor binding sites of avian H3/1963, pandemic (pdm) H3/1968 and human H3/2007 (Figure 3b) revealed that trisaccharide Neu5Acα2,3Galβ1,3GlcNAc of lactoseries tetrasaccharide a (LSTa) interacts with the avian H3/1963 binding site in a cone-like topology (1mqm [149]), whereas 6′SLNLN interacts with pandemic H3/1968 (6tzb [150]) and human H3/2007 binding sites in an umbrella-like topology (6aov [151]). Residues 226 and 228 are important in determining sialyl linkage binding specificity. As shown in Figure 3b, Q226 in the avian H3/1963 binding site directly forms hydrogen bonds with Sia-1 and Gal-2 of LSTa, whereas S228 in the pdm H3/1968 binding site directly forms a hydrogen bond with Sia-1 of 6′SLNLN and S228 in the human H3/2007 binding site directly forms two hydrogen bonds with Sia-1 of 6′SLNLN. Although no direct interaction of residue 226 in pdm and human H3 HAs was found, a previous study suggested that L226Q mutation in the HA decreases α2,6Neu5Ac binding preference. Both G228 and S228 can be found in H3 avian HAs, whereas only S228 is found in human H3 HAs [149]. L226 is not conserved during circulation in humans; L226V and V226I substitutions were observed before 2001 and in 2004, respectively [160].
Similar to the H1 HA receptor binding site [10], two sets of human receptor binding residues provide networks to make contact with the long human-type receptor that results in an umbrella-like topology of the receptor. (i) A base region Neu5Acα2,6Galβ1- motif is governed by residues 131–138 in a 130-loop, residues 140–145 in a 140-loop and residues 219–228 in a 220-loop [163]. Figure 3b shows direct H-bond formation between Y98, G135, S136, N137, H183, E190 and S228 in the pdm H3/1968 binding site and Sia-1 of 6′SLNLN. In the human H3/2007 binding site, Y98, T135, S136, S137, S228, R222, and N225 make direct H-bonds with Sia-1 and Gal-2, respectively, of 6′SLNLN. (ii) The extension region -4GlcNAcβ1,4Galβ1,4GlcNAc motif is governed by residues 190–196 in a 190-helix and residues 156–160 in a 150-loop [163], and S193 and K156 in the pdm H3/1968 binding site were observed to generate direct H-bonds with GlcNAc-5 of 6′SLNLN. Amino acid change in HA during co-evolution with humans occurs to evade human immunity. Not only is there a change in antigenicity but the number of glycosylation sites masking antigenicity also increases over time as shown in Figure 3c; the numbers of glycosylation sites/monomer are two for avian H3/1968 HA, two for pdm H3/1968 HA, seven for human H3/2007 HA and six for human H3/2014. The limitation of increase in the number of glycosylation sites might be because the change of the virus must have a balance between mutation and selection for optimal immune evasion and infection. Taken together, the change in receptor binding specificity of long-term circulating human IAVs from short and long to long α2,6 sialylated glycans may have resulted from aa change in the RBS (Figure 3b,d) and an increase in glycosylation sites surrounding the RBS, possibly making the shallow RBS deeper (Figure 3c). The differences in receptor binding preferences of avian, pandemic and long-term circulating human IAVs are associated with viral pathology along the human respiratory tract containing different sialylated glycan structures. The preferential binding of avian and pandemic viruses to both short and extended receptors can typically cause diffuse alveolar damage, resulting in greater severity than that caused by long-term circulating human viruses with preference for long receptors that rarely infect human alveoli [164,165]. This correlates well with our finding that human alveolar N-glycans consist of mainly short receptors, 22.32: 0.17: 16.10: 0.15 mol% (Neu5Acα2,3LN: Neu5Acα2,3(α1,3fucosylated LN): Neu5Acα2,6LN: Neu5Ac-LN-LN), of total human alveolar N-glycans [19]. Sialyl N-glycans with various numbers of LN units (up to 10 units) have been reported in human lungs (principally terminated in α2,3Neu5Ac [166]) and the human bronchus, whereas fewer extended LN profiles can be detected in the human nasopharynx [167]. Although structures of glycans in the human trachea have not be determined, the pdm H1N1/2009 virus was found at higher levels in tracheal aspirate specimens than in throat or nasopharyngeal swabs [168]. Uncomplicated long-term circulating human viruses are related to tracheobronchitis [169].

3.2. Egyptian Fruit Bat-Isolated Influenza A Virus Uses Siaα2,3Gal Receptors
In 2019, Kandeil et al. [64] reported a new IAV isolated in 2017 from Egyptian fruit bats (Rousettus aegyptiacus, family Pteropodidae) in an abandoned mudbrick house in a densely inhabited agricultural village in the Nile Delta, Egypt. The new IAV was found more frequently in oral swabs than in rectal swabs. Each of eight genomic segments of this newly characterized bat influenza A/bat/Egypt/381OP/2017 virus was shown to have nucleotide (nt) and aa sequences similar to those in genes of other avian IAVs isolated from wild birds, except for those in the PA gene, which are similar to those in the PA gene of an equine IAV (H7N7). The HA protein of the Egyptian bat IAVs is closely related to the group 1 cluster of HA subtypes with highest similarity (73% identity) to the H9 HA of influenza A/mallard/Ohio/13OS3856/2013 virus (H9N2). A receptor binding assay indicated that the Egyptian bat virus possessing Q226 (H3 numbering) in the RBS showed a clear binding preference for α2,3sialyllactose receptors over α2,6sialyllactose receptors, suggesting that Siaα2,3Gal receptors might be abundant in the infection sites in Egyptian fruit bats. Further investigation of that possibility is required. The virus was speculated to originate from an avian host, and that speculation was supported by the finding that the virus can grow well in allantoic fluid cavities of embryonated chicken eggs. The virus can also propagate in MDCK cells and in the lungs of C57BL/6 mice and BALB/c mice, indicating the possibility of the virus causing infection in other mammalian species. Thus, surveillance of IAVs among bats and distribution in other animals should be performed.

3.3. Influenza B Viruses Use Siaα2,3/2,6Gal Receptors
In 1940, a new serotype of influenza viruses was isolated and designated type B. The first strain was named B/Lee/1940. Influenza B viruses have continued to cause respiratory disease in humans with antigenic change. Although HA and NA antigenic differences within influenza B viruses (IBVs) are not sufficient to separate antigenic subtypes, there were sufficient antigenic differences to classify IBVs into two lineages: (i) Victoria lineage B/Victoria/2/1987-like and (ii) Yamagata lineage B/Yamagata/16/1988-like viruses [170]. Consequently, morbidity and mortality-associated seasonal influenza is currently caused by the two lineages of IBVs and two subtypes of IAVs, 1968-derived H3N2 and 2009-derived H1N1 viruses. In contrast to IAVs, IBVs infect mainly humans, although there are sporadic reports of IBV infection in seals, pigs, horses, pheasants and dogs [65]. Similar to IAVs, IBVs have eight (-)ssRNA genome segments and possess receptor-binding and -destroying activities on different molecules, homo-trimeric HA and homo-tetrameric NA glycoproteins. Clinically approved NA inhibitors (NAIs), including zanamivir, oseltamivir, laninamivir and peramivir, are now used for treatment of infection with not only IAVs but also IBVs [171]. Several chemical compounds that have been developed as anti-influenza A NAs, including Neu5Ac2en mimetics for minimizing side effects on human Neu1-Neu4 enzymes [172,173] and NA covalent inhibitors for irreversible NA inhibition [174] and Psidium guajava Linn. (guava) tea [175] and povidone-iodine that possess anti-influenza A sialidase activities [176] might be able to inhibit influenza B viruses.
Receptor binding specificity determines the site of virus infection. It appears that wild-type influenza B/Victoria HAs possessing G141, R162 and D196 [67] and B/Yamagata HAs with F95 and N194 [68] clearly exhibit binding preference to human-type α2,6Neu5Ac receptors. Investigation of receptor binding preference of IBV clinical isolates in Taiwan during the period from 2001 to 2007 (Table 1) revealed that (i) 83% of Yamagata-like strains prefer α2,6Sia receptors, whereas 17% of them prefer both α2,3Sia and α2,6Sia and (ii) 54% of Victoria-like strains prefer both α2,3Sia and α2,6Sia, whereas 25% of them prefer sulfated glycan, either β-Gal-3-sulfate or 6-HSO3-Galβ1,4GlcNAc, and 21% of them prefer α2,6Sia. The viruses with dual α2,3Sia and α2,6Sia-binding preferences were shown to be associated with bronchopneumonia and gastrointestinal symptoms [66]. These findings indicate that the evolution of receptor binding specificity in IBVs in circulation is different from that in IAVs and indicate tissue tropism and pathogenicity of IBVs, possibly affecting virus transmission.

3.4. Influenza C Viruses Use Neu5,9Ac2
In 1947, a new influenza virus without cross-reactive antisera against IAV (PR8) and IBV (Lee) was first isolated by R.M. Taylor from throat washings of a New York man during an influenza outbreak [177]. It was later designated type C and the first strain was named C/Taylor/1233/1947. ICV usually causes mild upper respiratory infection but can cause lower respiratory infection in children less than 2 years of age [178]. Most humans acquire antibodies to ICV at a young age [178,179] and antigenicity of ICV is stable, with no antigenic change being detected for at least 30 years [71]. These facts may be related to the limited outbreaks of ICV in humans, mainly in children. Although ICV antigenicity is stable, comparison of HE gene sequences in viruses isolated from 1947 to 2014 demonstrated that there are six lineages comprised of C/Taylor/1233/1947, C/Kanagawa/1/1976, C/Mississippi/1980, C/Aichi/1/1981, C/Yamagata/26/1981 and C/Sao Paulo/378/1982 [71]. ICVs have also been isolated from pigs [69] and cattle [70] (Table 1).
Different from IAVs and IBVs, ICV possesses hemagglutinin and receptor-destroying enzyme (RDE) on the same homotrimeric glycoprotein having multifunctional hemagglutinin (receptor-binding and membrane fusion activities) and esterase (receptor-destroying activity) and so-called hemagglutinin-esterase-fusion (HEF) protein [180]. The glycoprotein HEF spikes are encoded by the fourth gene segment, and only the ICV (-)ssRNA genome is comprised of only seven gene segments [181].
Thin-layer chromatography (TLC), gas-liquid chromatography (GLC) and high-performance liquid chromatography (HPLC) analyses of rat alpha 1-macroglobulin (RMG) and bovine submaxillary mucin (BSM) incubated with ICV in comparison with those incubated with neuraminidase from A. ureafaciens revealed that RMG and BSM incubated with ICV have a reduced amount of Neu5,9Ac2 but an increased amount of Neu5Ac. After confirmation by using purified Neu5,9Ac2 instead of RMG and BSM, it was concluded that RDE of ICV is neuraminate O-acetylesterase (9-O-acetyl N-acetylneuraminate O-acetylhydrolase (EC 3.1.1.53) catalyzing removal of the 9-O-acetyl group from Neu5,9Ac2, not cleaving the terminal Neu5Ac from glycoconjugate [182]. RMG and BSM can potentially inhibit hemagglutination by ICV at 4oC, and their inhibitory effects were abolished by pre-incubation of RMG and BSM with ICV at 37oC [182]. This evidence suggested that Neu5,9Ac2 is a receptor of ICV on the cell surface.
Receptor binding analysis of C/Johannesburg/1/66 classified in C/Aichi lineage [71] on a sialoglycan microarray showed that the virus predominantly binds to Neu5,9Ac2α2,6Galβ1,4GlcNAc β1,2Manα3(Neu5,9Ac2α2,6Galβ1,4GlcNAcβ1,2Manα6)Manβ1,4GlcNAcβ1,4GlcNAcitol-AEAB [72]. Further studies by using ICVs from other lineages may help to clarify whether receptor binding specificity of all ICVs to Neu5,9Ac2 depends on the α2,6 linkage or not.

3.5. Influenza D Viruses Use Neu5,9Ac2 and Neu5Gc9Ac Receptors
In 2011, a novel virus isolated from a nasal swab of a 15-week-old pig with influenza-like symptoms in Oklahoma in the USA was found to possess seven (-)ssRNA genomic segments and HEF spike glycoproteins and to share approximately 50% overall aa sequence identity with human ICVs, and it was named C/swine/Oklahoma/1334/2011 (C/OK) [183]. At first, it was suggested to be a new subtype of ICVs due to (i) no cross-reaction of C/OK with human ICVs determined by hemagglutination inhibition assays and (ii) a wider cellular tropism of C/OK than that of a human ICV determined by cell culture studies [183]. In 2016, however, it was determined by the International Committee on Taxonomy of Viruses that this novel influenza virus is distinct from other types, and it was officially classified in a new genus, Deltainfluenzavirus, and so-called influenza D virus (IDV, type (species) D). As shown in Table 1, in addition to pigs, IDVs have been isolated from cattle and have so far been classified into three lineages: D/OK (D/swine/Oklahoma/1334/2011-like viruses), D/660 (D/bovine/Oklahoma/660/2013-like viruses) and D/Japanese, with D/Japanese lineage being further classified into 2 sublineages, D/Yama2016 (D/bovine/Yamagata/10710/2016-like viruses) and D/Yama2019 (D/bovine/Yamagata/1/2019-like viruses), based on phylogenetic and antigenic analyses [73]. Although there has been only serological evidence suggesting that IDV can infect humans [15], the virus may acquire mutations to potentially infect humans and to cause influenza illness in humans.
The host range of IVs is primarily determined by receptor binding specificity of the viruses. Recently, Liu et al. compared receptor binding specificities of IDVs and their related ICVs by a sialoglycan microarray approach [72]. Strain D/swine/Oklahoma/1334/2011 (D/OK) showed preferential binding to Neu5,9Ac2 and Neu5Gc9Ac either linked to α2,6Gal or α2,3Gal and strain D/bovine/Oklahoma/660/2013 /660) preferred to bind to Neu5,9Ac2α2,6Gal, Neu5Gc9Acα2,6Gal and Neu5Gc9Acα2,3Gal, whereas strain C/Johannesburg/1/1966 dominantly recognized Neu5,9Ac2α2,6Gal. The broader receptor recognition by IDVs than by human ICV could explain why cellular tropism of IDVs is wider than that of human ICVs. Binding of IDVs to both Neu5,9Ac2 and Neu5Gc9Ac, different from human ICV binding to Neu5,9Ac2, could be determined by their different HEF-binding pockets. It was shown that different from human ICV HEF of C/Johannesburg/1/1966, swine IDV HEF of D/OK has an open cavity between the 230-helix and 270-loop in the receptor-binding site, which is thought to allow for accommodation of diverse glycan receptors, including Neu5Gc9Ac harboring an extra hydroxyl group on the N-acetyl group of C5 Neu5Gc and different sialyl linkages [184]. Further investigation of the structure of the bovine IDV HEF-binding pocket might lead to an understanding of different receptor binding preferences of swine and bovine IDVs. Receptor binding specificity of viruses is believed to be associated with receptors present on the target tissue. Glycoconjugate structures terminated with Neu5,9Ac2 and Neu5Gc9Ac along the bovine, porcine and human respiratory tracts have not been determined and further investigation is therefore needed. Previous findings that there is no Neu5Gc production in healthy humans due to mutation of a gene encoding CMP-Neu5Ac hydroxylase, which converts CMP-Neu5Ac to CMP-Neu5Gc [42,185], could explain why human ICVs prefer binding to Neu5,9Ac2, whereas swine and bovine IDVs can bind preferentially to both Neu5,9Ac2 and 9-O-acetylated Neu5Gc.

3.6. β1 HCoV-OC43 and β1 HCoV-HKU1 Use Neu5,9Ac2 Receptors
HCoV-OC43 strain was first detected in 1967 by an organ culture technique from throat washings of patients with common colds [186], but its complete genomic sequence was not reported until 2004 [187]. HCoV-HKU1 was first characterized in 2005 by Woo et al. at the University of Hong Kong (HKU) from a nasopharyngeal aspirate of a patient with pneumonia [188]. Based on genomic sequences reported so far, there is no bat CoV classified as a βCoV lineage A. Based on phylogenetic analysis, both HCoV-OC43 and HCoV-HKU1 βCoV lineage A probably originated in rodents [85]. While an intermediate host of HCoV-HKU1 remains unknown, HCoV-OC43 is believed to have cattle serving as intermediate hosts from rodents to humans [189].
HCoV-OC43 does not bind to and agglutinate erythrocytes pretreated with 9-O-acetyl esterase from either influenza C virus or bovine CoV [190]. HCoV-HKU1 does not infect primary human ciliated airway epithelial cells pretreated with an expressed HKU1 hemagglutinin-esterase (HE) protein possessing 9-O-acetylesterase activity [191]. These findings suggest that both HCoV-OC43 and HCoV-HKU1 bind to 9-O-acetylated sialyl glycans (Figure 4a) on the host cell surface for mediating virus infection. As shown in Table 2, the 9-O-Ac-Sia receptor-binding function of homodimeric HE proteins, comprised of a receptor-binding (lectin) domain and receptor-destroying domain, of HCoV-OC43 and HCoV-HKU1 was reported to be lost, and its loss was reported to be associated with an accumulation of mutations in the OC43-HE lectin domain or massive deletions found in the HKU1-HE lectin domain during evolution in humans [94]. Binding of the S1 subunit of another type of spike, a homotrimeric spike (S) protein (Figure 2), of HCoV-OC43 and HCoV-HKU1 on human rhabdomyosarcoma cells was shown and was reported to be reduced by pretreating the cells with HKU1-HE, OC43-HE or BCoV-HE, but not by pretreating the cells with MHV-S-HE, possessing 4-O-acetylesterase activity [191]. These findings suggested that 9-O-Ac-Sia is an essential receptor for infection of HCoV-OC43 and HCoV-HKU1 mediated by the S1 subunit of their S proteins.
The S1 subunit of the S protein is composed of four domains, A through D (S1A through S1D) domains from the N-terminus [112]. By using OC43 or HKU1 S1A–Fc proteins in a direct binding assay, HCoV-OC43 and HCoV-HKU1 were shown to bind to the receptors via domain A (S1-NTD (Figure 5a), residues 15–302 based on the S protein of OC43 strain ATCC VR-759) [92]. However, binding of HKU1 S1A to its receptors on rat erythrocytes can be detected when HKU1 S1A–Fc proteins have been conjugated to nanoparticles but cannot be detected by using free HKU1 S1A–Fc proteins (the standard method), indicating the requirement of multivalency of HKU1 S1A–Fc proteins for binding to rat erythrocytes. Based on structural analysis, residues 28–34 (element 1) and/or residues 243–252 (element 2) in HKU1 S1A (Figure 5b) were thought to hamper the binding of HKU1 S1A. The mutant HKU1 S1A was generated by replacement of one or both of their elements with the corresponding element(s) from bovine coronavirus (BCoV), which is believed to be the ancestor of OC43. The free mutant HKU1 S1A–Fc proteins with only one replacement at element 2 were found to bind to rat erythrocytes. The free mutant HKU1 S1A–Fc proteins with replacement of both elements showed greater binding to rat erythrocytes. In comparison with binding of the wild-type HKU1 S1A conjugated with nanoparticles to rat erythrocytes, the mutant HKU1 S1A with removal of a glycosylation site at element 2 (N251Q) showed increased binding, and the mutant HKU1 S1A with removal of the glycosylation sites in both elements (N29Q in element 1 + N251Q in element 2) showed greater binding. These findings indicated that binding of HKU1 S1A to its receptors on rat erythrocytes is impeded by both the RBS architecture and N-glycans on the RBS [92]. Binding of free HKU1 S1 or free HKU1 S1A not only to rat erythrocytes but also to mouse erythrocytes and to BSM cannot be detected by the standard method unlike other 9-O-Ac-Sia-binding β1CoVs, including HCoV-OC43 for which their free S1 and S1A detectably bind to those erythrocytes and BSM [92,191]. The difference of HKU1 from other 9-O-Ac-Sia-binding β1CoVs was suggested to be due to receptor fine-specificity determined by elements 1 and 2. The effects of the internal part of the glycan structure, such as Siaα2,3/2,6Gal and LN repeats, on binding of HKU1 in comparison with other 9-O-Ac-Sia-binding β1CoVs should be further determined.
The cryo-electron microscopy structure of an HCoV-OC43 S trimer in complex with a 9-O-Ac-Me-Sia revealed that a sialoside-binding site was located at the surface-exposed groove of each S1A monomer (Figure 5a) [193]. The sialoside-binding groove (Figure 5b,c) is formed by two loops, L1 consisting of 27-NDKDTG-32 and L2 consisting of 80-LKGSVLL-86 at the RBS edges, two hydrophobic pockets separated by the indole side chain of W90, P1 consisting of L85, L86 and W90 and P2 consisting of L80, W90 and F95 [193], and a residue, S87, interacting with L1 [92]. Substitutions of N27 having an H-bond with OA9 of the 9-O-acetyl carbonyl group, K81 forming H-bonds with O1 and N5 of Sia, or S83 containing an H-bond with O3 of Sia C1, with alanine and mutations of L80, L86 or W90 in hydrophobic pockets accommodating the 5-N-acyl moiety and the 9-O-acetyl-methyl moiety provided a mutant HCoV-OC43 S1A that had lost the ability to bind to 9-O-Ac-6SLN. Substitutions at N27, T31, L80, K81, S83, L86 and W90 completely blocked the entry of pseudotyped VSVΔG particles harboring HCoV-OC43 S proteins into HEK293T cells. These results confirmed that residues in the surface-exposed groove are critical for interaction with the 9-O-Ac-Sia receptor and that their interactions are essential for mediating viral entry [193]. Interestingly, HCoV-OC43 S1A recognized 9-O-Ac-Sia bound to Gal via α2,6 linkage. More research on binding specificity of both animal and human 9-O-Ac-Sia-binding β1CoVs to the internal part of the receptor in combination with analysis of 9-O-Ac-Sia-containing glycan structures expressed on host tissues and analysis of changes in the viral S1A proteins could reveal which part of 9-O-Ac-Sia-containing glycans determines host/tissue tropism of β1CoVs and changes in the viral S1A proteins associated with host/tissue tropism.