The sequence of the spike (also called peplomer or E2) protein gene of the Mebus strain of bovine coronavirus (BCV) was obtained from cDNA clones of genomic RNA. The gene sequence predicts a 150,825 mol wt apoprotein of 1363 amino acids having an N-terminal hydrophobic signal sequence of 17 amino acids, 19 potential N-linked glycosylation sites, a hydrophobic anchor sequence of approximately 17 amino acids near the C terminus, and a hydrophilic cysteine-rich C terminus of 35 amino acids. An internal Lys-Arg-Arg-Ser-Arg-Arg sequence predicts a protease cleavage site between amino acids 768 and 769 that would separate the S apoprotein into S1 and S2 segments of 85690 and 65153 mol wt, respectively. Amino terminal amino acid sequencing of the virion-derived gp 100 spike subunit confirmed the location of the predicted cleavage site, and established that gp 120 and gp 100 are the glycosylated virion forms of the S1 and S2 subunits, respectively. Sequence comparisons between BCV and the antigenically related mouse hepatitis coronavirus revealed more sequence divergence in the putative knob region of the spike protein (S1) than in the stem region (S2).

The open reading frame potentially encoding a 78 amino acid, 9101 Da hydrophobic protein (HP) and, mapping at the 3' end of the porcine transmissible gastroenteritis coronavirus (TGEV) genome, was shown to be expressed during virus replication. The cloned HP gene was placed in a plasmid under control of the T7 RNA polymerase promoter and in vitro translation of transcripts generated in vitro yielded a 9.1-kDa protein that was immunoprecipitable with porcine hyperimmune anti-TGEV serum. Antiserum raised in rabbits against a 31 amino acid synthetic polypeptide that represented the central hydrophilic region of HP specifically immunoprecipitated HP from TGEV-infected cells. HP was further shown to become associated with microsomal membranes during synthesis in vitro and was found to be closely associated with the endoplasmic reticulum and cell surface membranes in infected cells. The intracellular location of HP suggests that it may play a role in the membrane association of replication complexes or in virion assembly.

The sequence of the hemagglutinin-esterase (HE) gene for the Mebus strain of bovine coronavirus was obtained from cDNA clones, and its deduced product is a 47,700-kilodalton apoprotein of 424 amino acids. Expression of the HE protein in vitro in the presence of microsomes revealed N-terminal signal peptide cleavage and C-terminal anchorage but not disulfide-linked dimerization. Dimerization was observed only after expression in vivo, during which HE was also transported to the cell surface.

The nucleotide sequence between the spike and membrane protein genes in the bovine coronavirus (BCV) genome was determined by sequencing cDNA clones of the genome, and open reading frames potentially encoding proteins of 4.9, 4.8, 12.7, and 9.5 kDa, in that order, were identified. The 4.9- and 4.8-kDa proteins appear to be vestiges of an 11-kDa protein for which a single nucleotide deletion event in the central part of the gene gave rise to a stop codon. The consensus CYAAAC sequence precedes the 4.9-, 12.7-, and 9.5-kDa ORFs and predicts that transcription will start from each of these sites. Northern analyses using sequence-specific probes and oligo(dT)-selected RNA demonstrated that the predicted transcripts are made, and that these correspond to mRNAs 4, 5, and 5-1. BCV mRNA 4 appears to be a counterpart to mouse hepatitis virus (MHV) mRNA 4 which, in the MHV JHM strain, encodes the putative 15.2-kDa nonstructural protein. BCV mRNAs 5 and 5-1 appear to be used for the synthesis of the 12.7- and 9.5-kDa proteins, respectively, which demonstrates a pattern of expression strikingly different from that utilized by MHV. MHV makes its homologs of the 12.7- and 9.5-kDa proteins from the single mRNA 5. In vitro translation analyses demonstrated that the BCV 9.5-kDa protein, unlike its MHV counterpart, is poorly made from downstream initiation of translation. Thus, from a comparison between BCV and MHV we find evolutionary evidence for the importance of the CYAAAC sequence in regulating coronavirus transcription.

cDNA clones prepared from genomic RNA of the Mebus strain of bovine coronavirus (BCV) were sequenced to reveal the hemagglutinin (H) gene of 1,272 bases that predicts a 47,700 mol. wt. apoprotein of 424 amino acids. The H gene mapped on the immediate 5' side of the peplomer gene. The H protein sequence revealed a putative N-terminal signal peptide of 18 amino acids, 9 potential glycosylation sites, 14 cysteine residues, and a potential C-terminal anchor region of 26 amino acids. When transcripts of the gene were translated in vitro in the presence of microsomes, signal cleavage, glycosylation, and membrane anchorage were observed, but not disulfide-linked dimerization. Translation of a truncated mRNA having no sequence for the C-terminal anchor resulted in a nonanchored, intraluminal (intramicrosomal) protein. When the H protein was expressed in cells in the absence of other coronaviral proteins, it became glycosylated, dimerized, and transported to the cell surface. The BCV hemagglutinin protein, therefore, is a type 1 glycoprotein that contains all the information it needs for signal cleavage, glycosylation, disulfide-linked dimerization, and transport to the cell surface. MATERIALS AND METHODS cDNA Cloning of BCV Genomic RNA The Mebus strain of BCV was grown on human rectal tumor (HRT) cells and purified as previously described (Lapps et. al., 1987). cDNA cloning was done essentially as described (Lapps et. al., 1987), except that random 5-mer oligodeoxynucleotides (Pharmacia) were used as primers for first strand synthesis to generate several clones, one of which was II (Fig. lA). A synthetically made primer (5' ATTATGACCGCACACC 3') was used to extend genomic sequence 5'-ward from clone II, and this was used to generate several clones one of which was LA6 (Fig. lA). Clones were selected by colony hybridization to randomly primed cDNA prepared from