key: cord-0814887-y92e5gvw
authors: Künkel, F.; Herrler, G.
title: Structural and functional analysis of the S proteins of two human coronavirus OC43 strains adapted to growth in different cells
date: 1996
journal: Arch Virol
DOI: 10.1007/bf01718615
sha: 106b769328f09f2f27eb26ecfe9d11f60c8325e2
doc_id: 814887
cord_uid: y92e5gvw

The receptor-binding activity of strain CU (grown in MDCK I cells) and of strain VA (adapted to Vero cells) of human coronavirus OC43 was analyzed and compared with the binding activity of bovine coronavirus (BCV) and of the OC43 strain provided by the American Type Culture Collection (AT). Results obtained with resialylated erythrocytes indicated that the ability of the viruses to recognize 9-O-acetylated sialic acid in an α2,6-linkage decreased in the following order: AT>CU>BCV>VA. Only minor differences were observed with respect to the α2,3-linkage. The amino acid sequence of the S protein of strain CU and VA was derived from the nucleotide sequence of the cloned gene. Strain VA differed from strain CU in 34 positions, 18 in the S1 and 16 in the S2 subunit.

that differed in the amount ofNeu5,9Ac 2 present on the cell surface. The different strains of OC43 were analyzed for their ability to agglutinate the resialylated cells. Strain OC43-AT has been reported recently to recognize ~2,6-1inked Neu5,gAc 2 more efficiently than does BCV. As shown in Table 1 , strain OC43-CU was somewhat superior to BCV agglutinating cells that had been resialylated in the presence of 1-2 nmol of CMP-activated 9-O-acetylated sialic acid compared to 4 nmol required for BCV. With strain OC43-AT, however, hemagglutination was observed when the erythrocytes had been resialylated with amounts of CMP-Neu5,9Ac 2 as low as 0.25 nmol. The Vero cell adapted strain OC43-VA, on the other hand, appeared to be unable to recognize 9-O-acetylated sialic acid in an ~2,6-tinkage, because it failed to agglutinate cells that had been resialylated at the highest amount of CMP-Neu5,9Ac 2 tested, 16 nmol. The differences between individual strains were less pronounced when the recognition of ~2,3-1inked sialic acid was analyzed. Strain OC43-VA and BCV were able to agglutinate erythrocytes after resialylation in the presence of 1 nmol of CMP-Neu5,9Ac2, whereas OC43-AT and OC43-CU required 2 nmol.

In order to explain the functional differences between the strains of HCV-OC43 on a molecular level, the nucleotide sequence of the S gene of OC43-CU and OC43-VA was determined and compared with the corresponding sequence of strain OC43-AT reported recently [41. Virus was purified by discontinuous sucrose gradient centrifugation as described recently [4] . The final virus pellet was resuspended in a small volume of PBS. RNA was prepared from purified virus after proteinase-K digestion and phenol-chloroform extraction. After precipitation with 3M NaAc, pH 5.3, and 99% EtOH, vRNA was dissolved in DEPC-treated water as reported previously [4] . The S genes of the strains OC43-CU and OC43-VA were cloned by transcribing vRNA into two overlapping cDNA fragments using MoMuLV reverse transcriptase (Boehringer Mannheim). The product was amplified by PCR and then cloned into pBluescript (Stratagene).

Full length S genes were generated as described recently [4] and two clones were independently sequenced in both directions. In addition to sequencing the cloned S genes, the viral RNA from OC43-CU was directly sequenced twice with a modification of the dideoxy chain termination method [7] using AMV reverse transcriptase (Boehringer Mannheim). Hydridization of vRNA and primers was performed in a total volume of 8.5 ~tl using a minimum of 2 lag RNA and 20 ng primer (18-26 nt long) in the presence of a final concentration of 205 mM KC1. After 90 sec of heating to 95°C, the mixture was cooled over a period of 30 min to room temperature. The following steps were done at 4 °C. Two lal 35S-dATP, 2.5gl 10 x reaction buffer (600mM Tris/HC1, pH 8.3, 100mM MgC12 and 100 mM DTT) and 1 unit AMV reverse transcriptase were added to the annealing mixture. Aliquots of 3 gl were added to four nucleotide mixtures containing 50 gM dATP, 300 laM dCTP, 200 gM dGTP, 200 gM dTTP; and either of the dideoxy-NTPs (2 gM in the case of ddATP, and 10 gM in the case of ddCTP, ddGTP or ddTTP, respectively). After incubation for 20 min at 42 °C, 1 gl, 

100 pM dATP was added and incubation was continued at 42 °C for 10 minutes. The reaction was stopped with 10gl stop-solution (95% formamide, 20mM EDTA, 0.05% bromphenot blue and 0.05% xylene cyanol if). Sequencing of DNA was performed using Sequenase 2.0 (USB) as described recently [4] . [14] is marked by a wavy line. The proteolytic cleavage site is printed in bold letters and underlined twice. The amino acids in parentheses above the sequence indicate the differences found in the cDNA compared to the vRNA sequence of OC43-CU 4 amino acid exchanges (Fig. 1) . The structural properties of the two S proteins were similar to those reported for OC43-AT [4] . Both open reading frames have a coding capacity of 1363 amino acids corresponding to a Mr of nearly 150 kDa in the unprocessed form. They are two amino acids longer than the S protein of OC43-AT and identical in length to the corresponding protein of BCV-Mebus [1] . OC43-VA and OC43-CU have 18 potential sites for N-glycosylation, 9 in each subunit, in contrast to the S protein of OC43-AT, which has 20 of them. The two additional glycosylation sites are located in the S1 subunit (Fig. 1) . The three OC43 isolates differ also in the sequence of the potential cleavage site. The AT isolate has the sequence K N R R S R G in contrast to T K R R S R R in the two other isolates. The latter sequence is identical to that found in BCV-Mebus [1] . In contrast to the S proteins of OC43-CU and OC43-VA, which were cleaved to an extent of nearly 40% in infected cells, cleavage products of the S protein of the AT-isolate were never detected irrespective of the cell line used (data not shown). The sequence of the S protein of OC43-CU was found to be very similar to the corresponding sequence of BCV-Mebus differing in only 12 amino acids, 7 in the S1 and 5 in the $2 subunit (Table 2 ). By contrast, the S proteins of strains CU and AT of HCV-OC43 differed from each other in 115 amino acid positions, the majority of them being located in the SI subunit. The S protein of strain OC43-VA differed from OC43-CU in 34 amino acids, but it was still more closely related to BCV than to OC43-AT ( Table 2) . As OC43-CU and OC43-VA are derived from the same source, we assume that most of the differences between both strains are due to mutations that occurred during the adaptation to Vero cells.

The V e r o -a d a p a t e d strain recognized 9-O-acetylated sialic acid in an a2,3linkage as efficiently as did the other strains tested; however, it was unable to agglutinate erythrocytes resialylated to contain a2,6-tinked sialic acid -at least at the range of substrate concentrations tested. Assuming that the receptorbinding site is located in the S1 subunit, one or m o r e of the 18 amino acids differing between O C 4 3 -C U and O C 4 3 -V A are expected to be part of the sialic acid binding site. They should be helpful in future attempts to localize the binding site on the S protein. As the sialic acid binding activity of OC43-VA is diminished rather than improved, the change in this activity does not explain the ability of this virus to grow in Vero cells. Therefore, the mutations selected during the adaptation process are expected to affect an additional property in the S protein of the virus. Apart from the primary attachment to the cell surface, the S protein is also responsible for the fusion of the viral envelope with the cell membrane. The events involved in the fusion reaction of coronaviruses are not welt characterized. A concept that has been proposed suggests that, after primary attachment to sialic acid-containing receptors, the S protein may interact with a post-adsorption receptor and this interaction may induce the fusion activity [10] . If this concept is true, the sequence changes during the adaptation process may have enabled strain OC43-VA to recognize such a post-adsoption receptor on Vero cells. Future work has to determine the property of HCV-OC43 that is critical for growing in Vero cells.

The S proteins of strains OC43-CU and OC43-VA are more related to BCV than to OC43-AT by several characteristics: (i) the sequence similarity, (ii) the length of the open reading frame, (iii) the number of sites for N-linked glycosylation. This might suggest a contamination by BCV. This is, however, unlikely for several reasons. The seventh brain passage of HCV-OC43 in suckling mice has been reported by McIntosh et at. [5] at a time when BCV has not yet been described. At Marburg, we have obtained brain material derived from the eighth passage from the former British Common Cold Unit. The virus grew readily in MDCK I cells and did not require adaptation. The stock virus used for the present analysis was derived from the third passage in cell culture. Independently, HCV-OC43 obtained from the Common Cold Unit was adapted at Munich to growth in Vero cells [2] . The fact that this strain (OC43-VA) as well as strain OC43-CU are both more related to BCV than to OC43-AT argues against a laboratory contamination. Strain OC43-AT obtained fi'om the American Type Culture Collection has also been described as originating from the suckling mouse brain material reported by McIntosh et al [5] . An explanation of our findings may be that the mouse brain material contained a mixture of viruses and that different viruses were selected for by growth in different cell cultures (MDCK I cells for the CU, Vero cells for the VA, and HRT cells for the AT isolate). Alternatively, the mutations might have been introduced during the different passage history. In this context it should be noted that OC43 obtained from the American Type Culture Collection has been sequenced independently in two laboratories [4, 6] . The amino acid sequence derived from these nucleotide sequences differed in 26 amino acids, 23 in S1 and 3 in $2, including 11 amino acids present in our virus and missing in the sequence reported by Mounir and Talbot [61. These differences may be due to different passage history in the two laboratories. Analysis of more HCV isolates is required to determine the extent of sequence variation within human coronaviruses and the similarity to BCV. The Vero cell adapted strain OC43-VA should be helpful in characterizing functional domains of the S protein.

Deduced sequence of the coronavirus spike protein and identification of the internal proteolytic cleavage site

Structural proteins of human coronavirus OC43

Structural and functional analysis of the surface protein of human coronavirus OC43

Growth in suckling-mouse brain of "IBV-like" viruses from patients with upper respiratory tract disease

Molecular characterization of the S protein gene of human coronavirus OC43

DNA sequencing with chain-terminating inhibitors

The S protein of bovine coronavirus is a hemagglutinin recognizing 9-O-acetylated sialic acid as a receptor determinant

Hemagglutinating encephalomyelitis virus attaches to N-acetyl-9-O-acetylneuraminic acid containing receptors on erythrocytes: comparison with bovine coronavirus and influenza C virus

Recognition of cellular receptors by bovine coronavirus

The biology of coronaviruses

Coronaviruses: structure and genome expression

Human and bovine coronaviruses recognize sialic acid-containing receptors similar to those of influenza C viruses

Structural analysis of the conformational domains involved in neutralization of bovine coronavirus using deletion mutants of the spike glycoprotein S1 subunit expressed by recombinant baculoviruses

We thank Dr. C.-P. Czerny for kindly providing HCV-OC43-VA and Prof. Dr. S. Siddell for HCV-OC43-CU. Financial support was provided by a grant from Deutsche Forschungsgemeinschaft (He 1168/2-2 and 2-3). Sequences have been submitted to EMBL data library under accession number Z32768 (OC43-CU) and Z32769 (OC43-VA).