Human coronaviruses OC43 and HKU1 bind to 9-O-acetylated sialic acids via a conserved receptor-binding site in spike protein domain A

Abstract

Human betacoronaviruses OC43 and HKU1 are endemic respiratory pathogens and, while related, originated from independent zoonotic introductions. OC43 is in fact a host-range variant of the species Betacoronavirus-1, and more closely related to bovine coronavirus (BCoV)—its presumptive ancestor—and porcine hemagglutinating encephalomyelitis virus (PHEV). The β1-coronaviruses (β1CoVs) and HKU1 employ glycan-based receptors carrying 9-O-acetylated sialic acid (9-O-Ac-Sia). Receptor binding is mediated by spike protein S, the main determinant of coronavirus host specificity. For BCoV, a crystal structure for the receptor-binding domain S1A is available and for HKU1 a cryoelectron microscopy structure of the complete S ectodomain. However, the location of the receptor-binding site (RBS), arguably the single-most important piece of information, is unknown. Here we solved the 3.0-Å crystal structure of PHEV S1A. We then took a comparative structural analysis approach to map the β1CoV S RBS, using the general design of 9-O-Ac-Sia-binding sites as blueprint, backed-up by automated ligand docking, structure-guided mutagenesis of OC43, BCoV, and PHEV S1A, and infectivity assays with BCoV-S–pseudotyped vesicular stomatitis viruses. The RBS is not exclusive to OC43 and related animal viruses, but is apparently conserved and functional also in HKU1 S1A. The binding affinity of the HKU1 S RBS toward short sialoglycans is significantly lower than that of OC43, which we attribute to differences in local architecture and accessibility, and which may be indicative for differences between the two viruses in receptor fine-specificity. Our findings challenge reports that would map the OC43 RBS elsewhere in S1A and that of HKU1 in domain S1B.