Human coronavirus HCoV-229E is an enveloped positive-stranded RNA virus in group 1 that causes common colds (
18). Human aminopeptidase N (hAPN), a cell surface metalloprotease found on apical membranes of intestinal cells, lung and kidney epithelial cells, on macrophages, and at synaptic junctions, serves as receptor for HCoV-229E (
57). The deduced amino acid sequence of the gene that encodes the 200-kDa spike glycoprotein (S) of HCoV-229E suggests that the S protein is composed of a 15-amino acid (aa) signal sequence, an N-terminal S1 domain (aa 16 to 560), an S2 domain (aa 561 to 1173) containing several heptad repeat regions, a transmembrane domain (aa 1117 to 1138), and a short cytoplasmic tail (aa 1139 to 1173) with a carboxy-terminal cysteine cluster (
31). The S protein has 30 potential sites for N-glycosylation. It is likely that the spikes on the HCoV-229E viral envelope are trimers of the S glycoprotein, as shown for the spikes of a closely related coronavirus of pigs, transmissible gastroenteritis virus (TGEV) (
9). This paper reports studies to identify the receptor-binding domain of the viral spike glycoprotein.
To initiate infection, the spikes of HCoV-229E virions bind to hAPN on the plasma membrane of human cell lines or on the apical membranes of respiratory epithelial cells (
47,
57). HCoV-229E virions are neutralized by incubation at 37°C with soluble hAPN (4a). Fusion of the TGEV envelope with host cell membranes occurs in endosomes (
16). The molecular events that follow binding of a virion to its APN receptor have not been defined for group 1 coronaviruses. However, for mouse hepatitis virus in serogroup 2, binding at 37°C of the S protein to its receptor, murine CEACAM1a (CD66a), leads to conformational changes in the carboxyl-terminal domain (S2) that may facilitate fusion of the viral envelope with the host cell plasma membrane (
44,
58).
Three coronaviruses in serogroup 1, HCoV-229E, TGEV, and feline coronavirus, and probably also canine coronavirus (CCoV), utilize the APN glycoprotein of their normal host species as a virus receptor (
3,
7,
17,
22,
24,
43,
57). Although HCoV-229E can utilize either hAPN or feline APN (fAPN) as a receptor (
43,
57), it cannot utilize porcine APN (pAPN) (
7,
23). Kolb and collaborators identified a region of hAPN from aa 288 to 295 as essential for HCoV-229E infection. A different region, aa 717 to 813, on pAPN and the corresponding sites on fAPN and canine APN are essential for infection by TGEV, feline coronavirus, and CCoV, respectively (
6,
17). Recent studies showed that the HCoV-229E receptor activity of hAPN can be abrogated by the addition of a single N-linked glycosylation site at amino acid 291 of hAPN, corresponding to a naturally occurring N-glycosylation site on pAPN (
48). Thus, it is possible that the four group 1 coronaviruses studied may bind to a common site found on APN proteins of all of these host species and that the species specificity of APN recognition by the viruses depends on other domains of the APN proteins that differ from one host species to another and impede the approach of S proteins of group 1 coronaviruses from other hosts to the putative common conserved binding site (
48). It is not yet known what domains of the S glycoproteins of human and porcine coronaviruses are responsible for their selective utilization of hAPN or pAPN, respectively, or what domain of S binds to the putative common receptor site on APN proteins of human and pigs.
This report describes the expression of purified soluble, truncated S glycoproteins of HCoV-229E by using a recombinant baculovirus expression system, and it analyzes the interactions of these truncated S proteins with polyclonal anti-HCoV-229E antiserum, anti-S monoclonal antibodies (MAbs), and hAPN on the surface of living cells. The region between aa 417 and 547 of the HCoV-229E S protein was found to be essential for binding to the hAPN receptor for HCoV-229E.
MATERIALS AND METHODS
Cells.
Spodoptera frugiperda (Sf9) cells (Invitrogen, Carlsbad, Calif.) were maintained at 27°C in TC-100 medium (JRH BioSciences, Lenexa, Kans.) with 10% fetal bovine serum (Gemini Bioproducts, Calabasas, Calif.) and 2% antibiotics (penicillin, streptomycin, and amphotericin B; Gibco/BRL, Gaithersburg, Md.). Murine 3T3 cells transfected with hAPN and 3T3 cells transfected with vector alone were obtained from T. Look (St. Judes, Memphis, Tenn.) and were cultured as previously described (
57). The myeloma cell line P3X63-AG8.653 and porcine ST cells were obtained from the American Type Culture Collection (Rockville, Md.) and were propagated as recommended.
Virus.
HCoV-229E was obtained from the American Type Culture Collection (VR-740). The RW stock of HCoV-229E was propagated at 34°C in MRC5 cells as previously described (A. Bonavia and K. V. Holmes, submitted for publication). Plaque assays were done with MRC5 cells at 34°C.
Antibodies.
A polyclonal goat antiserum (G1) directed against sucrose density gradient-purified, NP-40-disrupted HCoV-229E virions, and preimmune goat serum were prepared in collaboration with L. Sturman, New York State Department of Health, Albany, N.Y.
Anti-HCoV-229E MAbs were prepared by immunizing BALB/c mice with 100 μg of semipurified HCoV-229E emulsified in complete Freund's adjuvant (Difco Laboratories, Detroit, Mich.). At 3-week intervals two boosters of 100 μg of semipurified HCoV-229E in incomplete Freund's adjuvant (Difco Laboratories) were administered, and mice were sacrificed 4 days after the last booster inoculation. Spleen cells from immunized mice were fused with P3X63-AG8.653 myeloma cells. Positive hybridoma clones were selected by immunofluorescence on L-132 cells infected with HCoV-229E.
WM15, a MAb directed against human aminopeptidase N (hAPN CD13), was obtained from Biodesign International (Kennebunk, Maine). Anti-hAPN MAb-BB1 was prepared by injecting Swiss Webster mice with membrane preparations of 3T3 cells expressing hAPN. Treatment of mouse cell lines or rodent cell lines expressing hAPN with anti-hAPN MAb-BB1 blocked HCoV-229E infection in these cells. A control mouse MAb of the same immunoglobulin G1 (IgG1) isotype against cholera toxin B subunit was kindly provided by R. Holmes (University of Colorado Health Sciences Center, Denver, Colo.). IgG from polyclonal goat anti-HCoV-229E antiserum G1 was purified on a protein A column (Pierce, Rockford, Ill.) and then was adsorbed with either S547 (G1/S547), S417 (G1/S417), or S268 (G1/S268) on Ni-nitrilotriacetic acid superflow columns (Qiagen, Valencia, Calif.).
Virus neutralization.
Five thousand PFU of HCoV-229E was incubated at 4°C with serial 10-fold dilutions of polyclonal goat anti-HCoV-229E antiserum in MRC5 medium (pH 7) or control goat preimmune serum in a total volume of 200 μl. After 2 h of incubation, 300 μl of medium was added and plaque assays were performed with MRC5 cells at 34°C by using an overlay medium of modified Eagle's medium containing 0.45% Seakem LEagarose (FMC Bioproducts, Rockland, Maine), 8% heat-inactivated fetal bovine serum, and antibiotics.
Expression of histidine-tagged, soluble, truncated HCoV-229E spike glycoproteins.
The S gene from HCoV-229E was reverse transcribed by using primer AB25 (Table
1) with Superscript 2 (Gibco, Grand Island, N.Y.) in a 20-μl reaction mixture, as described in the manufacturer's instructions. The gene was amplified by PCR by using 1 μl of cDNA and primers S-23 and S-3600 (Table
1) and then was cloned into pBCSk+ (Stratagene, La Jolla, Calif.). The cDNA encoding the S gene was reamplified by using
Pfu-cloned polymerase (Stratagene) with primers BZ21 and BZ22 (Table
1) and then was cloned into pCITE at the
NcoI and
XbaI sites (Novagen, Madison, Wis.). The latter clone was used to amplify the cDNA sequences encoding the leader and the various regions of S with
Pfu-cloned polymerase (Stratagene). We used the baculovirus expression vector pAcMP2(TH) (PharMingen, San Diego, Calif.) that encodes a thrombin cleavage site and six-histidine tag added to the C-terminal part of the expressed proteins, which were previously described (
59). The cDNA encoding the N-terminal 547 aa (S
547) including the signal sequence was amplified with oligonucleotides BZ21 and BZ22; 417 aa (S
417) amplified with BZ21 and AB84 and 268 aa (S
268) amplified with BZ21 and AB91.
NcoI and
XbaI restriction sites were added to the 5′ and 3′ ends, respectively, to allow cloning into pAcMP2(TH). Each construct was sequenced at the University of Colorado Health Sciences Center Cancer Center Core Sequencing Facility with primers BZ10, BZ14, AB29, AB31, AB41, and AB42 (see Table
1). The baculovirus expression constructs were cotransfected with BaculoGold DNA (PharMingen) into Sf9 cells, and progeny viruses were plaque purified according to the manufacturer's instructions. The resulting constructs, called S
547, S
417, and S
268, produced soluble six-histidine-tagged glycoproteins (Fig.
1).
Even though there is no protease cleavage site in the HCoV-229E S protein as there is in the spike protein of mouse virus (MHV) (
28), we used the S1/S2 nomenclature to refer to S1 as the amino-terminal half of the protein and S2 as the carboxy-terminal half.
Purification of soluble, histidine-tagged, truncated HCoV-229E S glycoproteins.
Large-scale preparations of supernatant medium containing the soluble truncated proteins were purified with HiTrap nickel affinity columns (Pharmacia, Piscataway, N. J.) as previously described (
59). After analysis by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and staining with silver (data not shown), the fractions containing the truncated S proteins were pooled and stored at −80°C.
Immunoblots.
The soluble receptor glycoproteins and prestained molecular size standards (Bio-Rad, Hercules, Calif.) were separated by SDS-12.5% PAGE, transferred to Immobilon-P membranes (Millipore, Bedford, Mass.), and blocked overnight at 4°C in TBST (25 mM Tris [pH 7.6], 150 mM NaCl, 0.1% Tween 20) supplemented with 5% powdered milk. All subsequent steps were done in TBST buffer with 0.5% powdered milk. Blots were incubated for 1 h with a 1:3,000 dilution of polyclonal goat anti-HCoV-229E antiserum preabsorbed to mouse 3T3 cells. After being washed, the blots were incubated for 1 h with a 1:3,000 dilution of rabbit anti-goat IgG conjugated to horseradish peroxidase (HRP) (Cappel, Aurora, Ohio). Bound HRP-conjugated antibody complexes were detected with Renaissance Chemiluminescence Reagent (DuPont/NEN, Boston, Mass.), and images were exposed to autoradiography film (Kodak, Rochester, N.Y.).
PNGase treatments of truncated HCoV-229E S proteins.
Fifteen micrograms of the truncated HCoV-229E spike glycoprotein was incubated in Denaturing Buffer (New England Biolabs, Beverly, Mass.) at 100°C for 10 min. Upon cooling, G7 Buffer (New England Biolabs) and NP-40 (New England Biolabs) buffer were added to a 1× final concentration by following the manufacturer's directions. Fifteen hundred units of PNGase F (New England Biolabs) was added to each sample, and the samples were incubated for 1 h at 37°C. The deglycosylated proteins were separated by SDS-12.5% PAGE, transferred to Immobilon-P membrane (Millipore) paper, and immunoblotted by using polyclonal goat anti-HCoV-229E antiserum.
Enzyme-linked immunosorbent assay (ELISA).
Ninety-six-well Immulon plates (Dynex Technologies, Inc., Franklin, Mass.) were coated with (350 ng per well) the purified truncated HCoV-229E spike proteins in 0.05 M carbonate buffer (pH 9.6) (Sigma, St. Louis, Mo.). The plates were blocked overnight at 4°C in 5% bovine serum albumin (BSA)-B3 buffer (0.15 M NaCl, 0.05 M Tris-HCl, 0.8 mM EDTA, 0.05% Tween 20, 0.1% BSA, pH 7.4). Plates were incubated with anti-S MAbs (1:10) or polyclonal anti-HCoV-229E antiserum (1:30) for 1 h at room temperature. Bound antibodies were detected by using HRP-conjugated goat anti-mouse IgG or HRP-conjugated rabbit anti-goat IgG (Cappel) developed with o-phenylenediaminedihydrochloride peroxidase substrate (Sigma) by following the manufacturer's directions. Washes after every step and antibody dilution were performed in B3 buffer.
Receptor-binding assays.
Flow cytometry analysis was used to assay binding of truncated HCoV-229E S proteins to hAPN on live-cell membranes. Mouse 3T3 cells stably transfected with empty vector (3T3) or 3T3 cells expressing recombinant hAPN (hAPN/3T3) were incubated for 1 h at 4°C with different amounts of nickel affinity-purified truncated S proteins. After three washes, bound truncated HCoV-229E S proteins were detected by flow cytometry analysis by using various anti-S protein MAbs or polyclonal anti-HCoV-229E antiserum and their respective isotypic or normal serum controls. All incubations and washes were performed at 4°C in F buffer (0.1% BSA in phosphate-buffered saline, pH 7.4).
DISCUSSION
The binding of viral attachment proteins (VAP) to their cognate cellular receptors and postbinding conformational changes in VAPs that lead to virus entry and uncoating have been studied intensively during the last 10 years. The receptor-binding domains of the VAPs of many different types of viruses have been mapped. In general, the amino acids that bind to the receptor are well conserved among related viruses that utilize the same receptor. Often the VAP binds to an exposed region on the surface of a receptor protein, frequently a site containing one or more hydrophobic residues (
42). For example, a conserved RGD domain in the surface-exposed GH loop of the VP1 VAP of foot-and-mouth disease virus, a picornavirus, binds to integrin αvβ3 (
21). The amino acid immediately following the RGD motif influences binding to another integrin, αvβ1 (
20). Mutations in this amino acid of the VAPs of different foot-and-mouth disease virus strains may determine the ability of the virus to use different integrins as receptors in vivo. The hemagglutinin (HA) glycoprotein of influenza A virus is the VAP that binds to sialic acid receptors (
52). HA forms a trimeric spike on the viral envelope. The crystal structures of HA proteins of many influenza A virus strains have been determined previously (
54). Protease cleavage of HA by furin during virus maturation or by extracellular trypsin-like enzymes occurs at a site between the HA1 and HA2 proteins. The N-terminal domain, HA1, consists of a ball of predominantly beta sheets and contains the major antigenic determinants of the protein. The receptor-binding site is a shallow pocket at the tip of HA1 that contains six residues that are conserved among all influenza A and B viruses (
53). The C-terminal HA2 domain contains a hydrophobic sequence called the fusion peptide at its amino terminus, followed by a coiled-coil domain that undergoes a profound conformational change upon shifting the pH to 5.5 (
36,
50,
51), a transmembrane domain, and a short cytoplasmic domain. The structure and functions of HA provide a model for class I virus fusion glycoproteins of many other enveloped viruses, including the paramyxovirus simian virus 5, measles, Ebola, vesicular stomatitis virus, and human immunodeficiency virus type 1 (
1,
11,
29,
56). In human immunodeficiency virus, the gp160 spike glycoprotein is cleaved to yield gp120 and gp41 (
19). The N-terminal gp120 binds to CD4 on the cell surface, triggering a conformational change in the gp120 protein that exposes the binding site on gp120 for the chemokine coreceptor (
33,
56). Interaction of the spike with the coreceptor triggers a conformational change that activates gp41 to position the fusion peptide in close proximity to the cell membrane and allow insertion of the fusion peptide into the membrane of the target cell (
51). The N-terminal Ig-like domain of human CD4 binds to a recessed pocket in gp120, and the critical contact amino acids have been mapped (
55).
Analyses of the deduced amino acid sequences of the 180- to 200-kDa coronavirus spike glycoproteins suggest that these proteins have the same general structure as the prototypic class 1 viral fusion protein, the HA protein of influenza A, although they are much larger than HA. The Chow and Fassman algorithms predict that the N-terminal domain of coronavirus S proteins consists primarily of beta sheets, and the C-terminal domain includes a coiled-coil domain, transmembrane domain, and short cytoplasmic domain. For some coronaviruses in group 2, such as MHV and bovine coronavirus, there is a protease cleavage site near the center of the S protein that defines the N-terminal S1 and the C-terminal S2 domains. Protease cleavage at that site can facilitate virus-induced cell fusion and viral infectivity (
10,
39), although cleavage between S1 and S2 is not essential for viral infectivity (
4). Cleavage has not been observed in S proteins of group 1 coronaviruses.
The spike glycoprotein (S) of murine coronavirus MHV is the most extensively studied coronavirus VAP. The principal receptors for MHV are isoforms of murine CEACAM1a (CD66a; formerly called Bgp, mmCGM, and MHVR) (
2). Mutational analysis and X-ray crystallographic studies show that the protruding CC′ loop on the surface of the N-terminal Ig domain of murine CEACAM1a serves as the viral binding site (
32,
42,
49). A hydrophobic amino acid, Ile41, that projects from this loop has been postulated to insert into a possible hydrophobic pocket in the viral S protein. The N-terminal 330 aa of the S1 protein of MHV is sufficient for binding to the murine CEACAM1a receptor protein (
41). On MHV virions, binding of soluble murine CEACAM1a to the receptor-binding domain of S1 at 37°C induces a conformational change in the S2 protein (
44,
58) and facilitates detachment of the S1 protein of MHV-JHM (
12). There is no hydrophobic fusion peptide at the N terminus of the S2 protein, but several possible fusion peptides have been identified within the S2 protein (
26,
27).
To date there are no structural data and only limited functional information about the spike glycoproteins of coronaviruses in serogroup 1, which includes HCoV-229E, TGEV, FCoV, and CCoV. Since each of these viruses uses APN as a receptor to enter cells (
7,
43,
57), similar or conserved regions in the spike glycoproteins could be candidates for the binding site for viral S proteins. The amino acid identity between the S glycoproteins of serogroup 1 coronaviruses ranges from 41 to 45% (by using MacVector; International Biotechnologies, Inc., New Haven, Conn.). The best-studied S protein of the group 1 viruses is that of TGEV. In addition to binding to pAPN, TGEV virions treated with neuraminidase can bind via the S protein to N-acetyl neuraminic acid or N-glycolyl neuraminic acid moieties on the cell surface (
25,
37). Two amino acids in the S1 domain play a role in sialic acid binding. A domain of more than 220 aa in the S1 domain of the TGEV S protein is not required for binding to pAPN, because deletions of this segment of S1 occur spontaneously, generating the porcine respiratory coronavirus that cause epizootics of mild respiratory disease in pigs (
34,
35,
38). Like the TGEV viruses from which they are derived, porcine respiratory coronavirus isolates can also utilize pAPN as receptors (
8). Epitopes for MAbs have been mapped to the S protein of TGEV (
5). Because all strains of TGEV are neutralized by a MAb to epitope A, this region of S is a candidate for the receptor-binding domain (
13,
40). There is a single report about the interaction of truncated TGEV S protein with pAPN (
14). When a peptide from aa 506 to 655 of TGEV S protein was coexpressed in insect cells with pAPN, the two proteins could be coimmunoprecipitated. Therefore, this region of S was implicated as a possible receptor-binding domain. This aa 506 to 655 region of TGEV S protein corresponds by sequence alignment to aa 279 to 417 of the S protein of HCoV-229E, but there is only 28% identity of these peptides at the amino acid level. These coexpression experiments do not represent physiological binding of a viral spike to its receptor on a host cell. For example, coimmunoprecipitation could occur if the insect cells had not been thoroughly disrupted prior to antibody addition.
To identify the receptor-binding domain of the HCoV-229E spike protein, we expressed three C-terminal-truncated S proteins from the N-terminal domain of S1. S547 included the entire S1 domain. S417 was truncated at the C terminus corresponding to the putative receptor-binding domain of TGEV S protein, and S268 was the smallest N-terminal domain tested. The proteins were purified, and binding of MAbs demonstrated that they retained the antigenic determinants of the intact S protein. Epitopes recognized by eight nonneutralizing anti-S MAbs were mapped to these three peptides. We showed by flow cytometry that the soluble N-terminal domain of HCoV-229E (S547) could bind directly to hAPN on the membranes of hAPN/3T3 cells, while no binding was detected on 3T3 cells transfected with vector alone. Binding of the peptide to hAPN was specific and could be blocked either by pretreatment of the cells with anti-hAPN MAb-BB1 or by incubation of the S547 peptide with goat polyclonal anti-HCoV-229E antiserum. Neither S417 nor S268 was found to bind to hAPN/3T3 cells by this assay. This observation implicates the region between aa 417 and 547 of HCoV-229E S protein as an important determinant for binding to hAPN. Further evidence in support of this conclusion is the observation that region-specific polyclonal antibody directed against the region from aa 417 to the C terminus of S could block binding of S547 to hAPN on hAPN/3T3 cells. The putative receptor-binding domain of aa 417 to 547 of HCoV-229E protein is not homologous to the putative aa 506 to 655 receptor-binding domain of TGEV S protein (which corresponds to aa 279 to 417 of HCoV-229E S protein). The reason for this difference between TGEV and HCoV-229E is not clear. Possibly, although the N-terminal 417 aa of HCoV-229E S protein were not sufficient for binding to hAPN, they may contribute to the species specificity and/or the affinity of receptor binding.
Further mutational analysis and crystallographic studies of the S protein of HCoV-229E are needed to identify the specific amino acid residues on S that are required for binding to hAPN as well as residues important for possible induction of receptor-induced conformational changes in S. These studies will also elucidate the molecular determinants of species-specific recognition of APN proteins by S proteins of group 1 coronaviruses. Finally, understanding the interactions of HCoV-229E spike glycoprotein with its hAPN receptor may lead to the development of novel antiviral drugs to prevent or treat coronavirus-induced colds.