A 12-Amino Acid Stretch in the Hypervariable Region of the Spike Protein S1 Subunit Is Critical for Cell Fusion Activity of Mouse Hepatitis Virus*

The spike (S) glycoprotein of mouse hepatitis virus (MHV) plays a major role in the viral pathogenesis. It is often processed into the N-terminal S1 and the C-terminal S2 subunits that were evidently important for binding to cell receptor and inducing cell-cell fusion, respectively. As a consequence of cell-cell fusion, most of the naturally occurring infections of MHV are associated with syncytia formation. So far, only MHV-2 was identified to be fusion-negative. In this study, the S gene of MHV-2 was molecularly cloned, and the nucleotide sequence was determined. The MHV-2 S protein lacks a 12-amino acid stretch in the S1 hypervariable region from amino acid residue 446 to 457 when compared with the fusion-positive strain MHV-JHM. In addition, there are three amino acid substitutions in the S2 subunit, Tyr-1144 to Asp, Glu-1165 to Asp, and Arg-1209 to Lys. The cloned MHV-2 S protein exhibited the fusion-negative property in DBT cells as the intrinsic viral protein. Furthermore, similar to the fusion-positive MHV-JHM strain, proteolytic cleavage activity was detected both in DBT cells infected with the fusion-negative MHV-2 and in the transfected cells that expressed the cloned MHV-2 S protein. Domain swapping experiments demonstrated that the 12-amino acid stretch missing in the MHV-2 S1 subunit, but not the proteolytic cleavage site, was critical for the cell-fusion activity of MHV.

Mouse hepatitis virus (MHV) 1 is a member of the Coronaviridae family that causes inapparent enteric and respiratory infection, hepatitis, and acute and chronic demyelinating diseases of the central nervous system (1). It is an enveloped virus that contains a positive sense, single-stranded genomic RNA of approximately 31 kilobases in length (2,3). The viral particle contains four major structural proteins: the nucleocapsid (N) protein that interacts with the viral genomic RNA and membrane (M) protein to form a spherical core (4,5) and the spike (S) and small membrane (E) proteins that together with the M protein constitute the envelope (6,7). In some strains of MHV, there exists an additional enveloped protein, the hemagglutinin esterase (8,9). The E and M proteins were demonstrated to be required for virion assembly (5,10). The S protein of MHV forms large characteristic projections of coronaviruses and plays a major role in viral pathogenesis.
The S protein is cotranslationally glycosylated in the endoplasmic reticulum, giving a molecular mass of approximately 180 kDa, and trimerized (11). Following a transport through the Golgi apparatus, the high mannose oligosaccharides are trimmed, and the protein is further acylated (12). After the modifications, the S protein is often cleaved by a cellular protease into two noncovalently bound 90-kDa subunits, S1 and S2, derived from the N-terminal and C-terminal halves, respectively (13,14). S protein that is not assembled into virions is transported by the secretory system to the cell surface, where it may participate in inducing fusion between adjacent cells (15). In addition to the induction of pH independent cell-to-cell fusion, the S protein also mediates the viral attachment to the receptor of susceptible cells and is involved in the elicitation of neutralizing antibodies, the tissue tropism, and the variation of viral pathogenicity (7, 16 -21).
The process of S protein-mediated membrane fusion has been intensively studied, but the mechanism is still poorly understood. Proteolytic cleavage of the S protein precursor into the S1 and S2 subunits was thought to be a prerequisite for the virus to induce cell fusion (14); cell fusion activity correlated with the cleavage level of S protein. The cleavage site was located at the C terminus of the amino acid sequence RRARR in the middle of the S protein (22). Nevertheless, mutations at the cleavage site did not completely abolish the cell fusion activity (23)(24)(25)(26). Although the S2 subunit appeared to be responsible for the cell fusion activity of MHV spike protein (27,28), both S1 and S2 subunits were suggested to be correlated with the cell fusion activity and viral infectivity (29,30).
There is considerable polymorphism in both the length and nucleotide sequence of the S1 subunit that possibly resulted from mutation and recombination (31)(32)(33). Several neuroattenuated variants of MHV-JHM have been isolated. Deletions and mutations in the hypervariable region of the S1 subunit of the variants were demonstrated to be associated with altered antigenicity and virulence (18,33,34). Studies performed by infection of cultured cells demonstrated that most of the MHV strains identified so far are capable of inducing membrane fusion between infected and uninfected cells, resulting in multinucleated polykaryons (syncytia) (2,35). Among the identified naturally occurring strain of MHV, MHV-2 is the only one that cannot induce syncytia (36). Although MHV-2 lacks the cell fusion activity, it causes fulminant hepatitis in mice (37).
To examine the cell fusion activity of MHV in cultured DBT * This work was supported by National Science Council of the Republic of China Grants NSC82-0412-B-002-257 and NSC86-2314-B-002-139. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF107212.
ʈ To whom correspondence should be addressed: Inst. of Biochemistry, National Taiwan University, College of Medicine, 1 Jen-Ai Rd., First Section, Taipei, 100 Taiwan. Tel.: 2-2397-0800 (ext. 8217); Fax: 2-2391-5295; E-mail: mfchang@ha.mc.ntu.edu.tw. 1 The abbreviations used are: MHV, mouse hepatitis virus; nt, nucleotide(s). cells, a recombinant vaccinia virus that encodes the T7 RNA polymerase (38) and a recombinant plasmid containing the cDNA of MHV S protein driven by the promoter of T7 RNA polymerase were used in this study. We found that the 12amino acid stretch from residue 446 to 457 of the MHV-JHM was important for the cell fusion activity. In addition, the S protein of MHV-2 was capable of undergoing proteolytic cleavage, but it lost the ability to induce cell-cell fusion.

Cell Line and Viruses
The DBT cell line was derived from a Schmidt-Ruppin Rous Sarcoma virus-induced murine astrocytoma (39) and was maintained as monolayer culture in Eagle's minimal essential medium (Life Technologies, Inc.) supplemented with 10% heat-inactivated fetal bovine serum, 10% tryptose phosphate broth (Difco), and 0.06 mg/ml kanamycin. MHV-JHM and MHV-2 were kindly provided by Dr. Fujiwara (University of Tokyo) and were propagated on DBT cells.

Virus Titration
Virus titers were determined by plaquing on DBT cells using the following procedures. Viruses were serially diluted and layered over DBT cells grown on 60-mm Petri dishes to allow adsorption for 90 min at 37°C. Five milliliters of overlay medium containing Eagle's minimum essential medium, 5% calf serum, 10% tryptose phosphate broth, and 1% Bacto-agar were then added, and the cultures were incubated at 37°C for 36 h. Following an addition of 3 ml of staining solution (1% Bacto-agar and 0.4 mg/ml neutral red) and incubation for 6 -8 h, plaques were counted. Virus titer was expressed in plaque-forming units.

Virus Purification
Virus was propagated on DBT cells and purified as described previously (40) with modifications. Briefly, cultured DBT cells were infected with MHV-JHM or MHV-2 at a multiplicity of infection of 25 plaqueforming units/cell and maintained at 37°C for 16 h. Virus particles released into the cultured medium were collected, and cell debris was clarified by centrifugation for 30 min at 8000 rpm in a JA-14 rotor (Beckman). Viruses in the supernatant were precipitated with 50% saturated ammonium sulfate at 4°C for 1.5 h followed by centrifugation at 8000 rpm for 30 min. The virus pellets were resuspended in NTE buffer (100 mM NaCl, 10 mM Tris-HCl, pH 7.2, and 1 mM EDTA), and the suspensions were further centrifuged at 2000 rpm for 10 min in the same rotor. The supernatants that contain the virus particles were collected and subjected to two rounds of discontinuous sucrose gradient consisting of 20, 30, 50, and 60% sucrose in NTE buffer at 26,000 rpm for 3.5 h in an SW28 rotor (Beckman) at 4°C. Viral particles that formed a band at the interface of 30 and 50% sucrose layers were harvested and pelleted by centrifugation at 40,000 rpm for 1.5 h in an SW41 rotor (Beckman). The virus pellets were resuspended in NTE buffer.

Viral RNA Isolation and cDNA Cloning
To obtain double-stranded cDNA of the S gene of MHV-2, genomic RNA was isolated from sucrose gradient-purified viral particles by the acid guanidinum thiocyanate/phenol/chloroform extraction method (41). Complementary DNA synthesis was performed with avian myeloblastosis virus reverse transcriptase and DNA polymerase I as described by the manufacturer (Roche Molecular Biochemicals). In addition, the MHV-specific oligonucleotide primer ␣3732 TCTTTCCAGGAGAGGCTGTGATAG ␣3709 (nucleotides (nt) are numbered as described by Schmidt et al. (42) but starting from the translational initiation site, and the letters ␣ indicate sequences of the complementary strand) was used in the reverse transcriptase reaction. Following the second-stranded cDNA synthesis, HindIII linkers were ligated to the ends of the double-stranded cDNAs. The resultant cDNAs were then treated with HindIII restriction endonuclease and inserted into a modified pGEM-4Z vector (Promega) that had been treated with HindIII restriction endonuclease and calf intestine alkaline phosphatase. These generated a cDNA library of the MHV-2 S gene.

DNA Sequence Analysis
The complete sequence of the MHV-2 S gene was determined by analyzing its cDNA clones with Sequenase (U.S. Biochemical Corp.) and primers SP6 (5Ј-GATTTAGGTGACACTATA-3Ј) and T7 (5Ј-TAAT-ACGACTCACTATAG-3Ј) following the dideoxy chain termination method (43). In addition, all of the plasmids described below were sequenced across the ligating junctions of DNA fragments to confirm the accuracy of cloning.

Plasmids
Plasmid pCS-dl-Plasmid pCS-dl contains the full-length coding region of the JHM-DL S gene inserted into the MscI site of pCITE-4a(ϩ) vector (Novagen). It was reconstructed from three overlapping cDNA clones, pTS-n, pTS-m, and pTS-c, generously provided by Dr. Michael M. C. Lai at the University of Southern California.
Plasmids pCS-j and pGEM-j-(1125-2042)-For construction of plasmid pCS-j, which contains the full-length coding region of the MHV-JHM S gene, plasmid pGEM-j-(1125-2042) was first generated. Plasmid pGEM-j-(1125-2042) bears the MHV-JHM-specific region of the S gene from nt 1125 to 2042 inserted into the SmaI site of pGEM-4Z vector (Promega). The partial cDNA fragment was obtained from MHV-JHM genomic RNA by reverse transcriptase-polymerase chain reaction with the primer ␣2100 CTGCTGTCTGCATGCAGC ␣2083 and the amplification primer set 1125 TGATGCGTCCAAAGTG 1140 and ␣2042 TTGGGAGCCCT-TATCTG ␣2026 . Plasmid pCS-j was then generated from plasmid pCS-dl by replacing its 680-base pair BspEI-ClaI fragment with the cognate cDNA fragment of pGEM-j-(1125-2042).
Plasmid pCS-m2-Plasmid pCS-m2 contains the full-length coding region of the MHV-2 S gene inserted into the MscI site of pCITE-4a(ϩ) vector. For construction of plasmid pCS-m2, seven overlapping cDNA clones were selected from the cDNA library of MHV-2 S gene generated in this study.
Plasmids pCS-j/m2 and pCS-m2/j-Plasmid PCS-j/m2 was generated from pCS-j by replacing its 680-base pair BspEI-ClaI fragment with the cognate fragment of pCS-m2. Plasmid pCS-m2/j was generated from pCS-m2 by replacing its 644-base pair BspEI-ClaI fragment with the cognate fragment of pCS-j. Both plasmids were constructed for performing domain swapping experiments.
Plasmid pES-e/p-For construction of plasmid pES-e/p, a 539-base pair EcoRI-PstI fragment was obtained from pTS-m that contains the middle region of the JHM-DL S gene. The DNA fragment was blunted with the Klenow fragment of DNA polymerase I and inserted into a modified pET-15b expression vector (Novagen) that had been treated with NdeI restriction endonuclease and the Klenow fragment of DNA polymerase I. Plasmid pES-e/p encodes a His-tagged MHV S2 subunit from amino acid residue 45 to 223 with a predicted molecular size of 22 kDa, designated S2-(45-223). The His tag facilitated purification of the recombinant S2-(45-223) protein by Ni 2ϩ -chelating affinity chromatography (44).
Plasmid pGEM-A59 -Plasmid pGEM-A59 contains the full-length coding region of the MHV-A59 S gene inserted into the SmaI site of pGEM-3Z vector and was kindly provided by Dr. Michael M. C. Lai.

Production of His-tagged MHV S Protein and Generation of the S Protein-specific Antisera
For generation of antisera against MHV S protein, the His-tagged recombinant S2-(45-223) protein was produced by transforming Escherichia coli BL21(DE3) (45) with plasmid pES-e/p. Expression and purification of the recombinant protein were performed essentially according to the procedures described by the manufacturer (Novagen). The isopropyl-1-thio-␤-D-galactopyranoside-induced recombinant S2-(45-223) protein was found to be present in the insoluble fraction (data not shown). To obtain the recombinant S protein, cell pellet was resuspended in a 50 mM sodium phosphate buffer (pH 8.0) containing 30 mM NaCl and 6 M urea and further purified on Ni 2ϩ -nitrilotriacetic acid resin following the procedures as described by the manufacturer (Amersham Pharmacia Biotech). The recombinant S2-(45-223) protein was used as the antigen to generate rabbit antisera against MHV S protein.
Preparation of rabbit polyclonal antisera was carried out as described previously (46). The final bleeding was carried out 1 week after the last immunization, and IgG fractions of the antisera were purified through immobilized protein G (Pierce). The specificity of the rabbit antiserum to the S2 subunit was confirmed by immunoprecipitation of in vitro translated products of plasmid pGEM-A59 (data not shown). In vitro transcription-translation of the MHV S protein was performed as described previously (46,47) with T7 RNA polymerase (Promega) and the mRNA-dependent rabbit reticulocyte lysate system (Promega), and immunoprecipitation was performed as described previously (48).

Cell Fusion Assay and Indirect Double Immunofluorescence Staining
For analysis of the cell fusion activity of the S protein of MHV-JHM, MHV-2, and their hybrids, plasmids pCS-j, pCS-m2, and pCS-j/m2 and pCS-m2/j, respectively, were transfected into the vTF7-3-infected DBT cells grown on 60-mm Petri dishes or 21 ϫ 20-mm four-chamber glass slides (Nunc). The cells were monitored at 6-h intervals under microscope and photographed. Indirect immunofluorescence staining was performed as described previously (24,49) with modifications. In brief, cells cultured on chamber slides were washed with phosphate-buffered saline at 48 h posttransfection, fixed with precooled acetone-methanol (1:1) for 5 min at Ϫ20°C, and air-dried. The fixed cells were blocking at room temperature for 15 min with 1% bovine serum albumin in phosphate-buffered saline and then overlaid with a mixture of a rabbit antiserum (1:60 dilution) specific for MHV S protein and a mouse antiserum (1:120 dilution) specific for glial fibrillary acidic protein and incubated in a 37°C moist chamber for 1 h. The antiserum against glial fibrillary acidic protein was used as a glial cell-specific cytoplasmic marker (50). After three rinses in phosphate-buffered saline, fluorescein isothiocyanate-conjugated goat anti-rabbit IgG (1:200 dilution) and rhodamine-conjugated goat anti-mouse IgG (1:200 dilution) were added and incubated for 40 min. The immunostained cells were washed thoroughly with phosphate-buffered saline and mounted by SlowFade (Molecular Probes, Inc., Eugene, OR). Photographs were taken under a fluorescence microscope. the monolayer cells were lysed in a buffer containing 50 mM Tris-HCl (pH 7.5), 100 mM NaCl, 0.1% SDS, 1% Nonidet P-40, 1% sodium deoxycholate, and 100 g of phenylmethylsulfonyl fluoride per ml. The cell lysates were clarified, and supernatants were collected to perform immunoprecipitation as described previously (48) with rabbit antiserum against the recombinant MHV S2-(45-223) protein.

RESULTS
Cloning and Sequencing of MHV-2 S Gene-To learn the possible sequence variations of the S gene of MHV that conferred the cell fusion activity of MHV-JHM in contrast to the fusion-negative strain MHV-2, cDNA of MHV-2 S gene was cloned from polyadenylated viral genomic RNA. Partial cDNA clones that were obtained as outlined in Fig. 1A were used to generate the full-length cDNA. The complete nucleotide sequence of the MHV-2 S gene was determined, and an open reading frame was identified that consists of 3669 nucleotides encoding an S protein of 1223 amino acid residues. Deduced amino acid sequences indicated that the S protein contains a signal peptide at the amino terminus, a transmembrane domain at the carboxyl terminus, and a putative proteolytic cleavage sequence RRARR in the middle portion. Sequence alignment to the MHV-JHM strain (42) indicated that there is a 36-nt deletion in the MHV-2 S gene from nt 1336 to 1371 ( and MHV-2, respectively, under the control of the T7 promoter were transiently transfected into DBT cells following infection of the recombinant vaccinia virus. Functional analysis of the S protein was determined by examining syncytium formation. In pCS-j-transfected DBT cells, the expression of MHV-JHM S protein was detected at 20 h posttransfection when polykaryons were also observed, and syncytia were detected by 48 h (Fig. 2, A and B). The syncytia induced by transfection of the S-encoding plasmid were morphologically indistinguishable from those induced by the infection of MHV-JHM (data not shown). In addition, the size of polykaryons continued to increase until the cells detaching from the culture dishes that occurred 72 h postinfection of MHV-JHM (data not shown). Although the expression of MHV-2 S protein was obvious in cells transfected with pCS-m2, no syncytium formation but only rounding up of the cells was observed (Fig. 2, C and D). Syncytium formation was not observed in DBT cells transfected with vector alone either (Fig. 2, E and F).
Cell Fusion Activity Could Be Rescued by Replacing the Deletion Domain of MHV-2 with MHV-JHM S Protein-S protein of MHV acts as a key factor to induce cell fusion in cultured cells. To explore the possible involvement of the 12-amino acid deletion and 3-amino acid substitutions of the MHV-2 S protein in its fusion-negative property, domain swapping experiments were employed between the S genes of MHV-JHM and MHV-2. By replacing the BspEI-ClaI fragment of the MHV-2 S gene that lacks the 12-amino acid stretch with the cognate fragment of MHV-JHM, the hybrid S-m2/j protein was found to cause syncytia of cultured DBT cells 48 h posttransfection (Fig. 3, C  and D), whereas by replacing the BspEI-ClaI fragment of MHV-JHM S gene with the cognate fragment of MHV-2, no polykaryon formation was evident, although the hybrid S-j/m2 protein was expressed equally well to the S-m2/j in the trans-fected cells (Fig. 3, A and B). These results indicated that the 12 amino acid residues in the hypervariable region of the S1 subunit between 446 and 457 are important, but the three variable amino acid residues in the C-terminal domain may have little contribution for the cell fusion activity of MHV S protein.
Both Fusion-positive MHV-JHM and Fusion-negative MHV-2 S Proteins Are Capable of Undergoing Proteolytic Cleavage in Infected and Transfected Cells-Previous studies have demonstrated that in MHV-infected cells, the primary 150-kDa S gene product (p150) was glycosylated to form gp180, which was then proteolytically cleaved into two 90-kDa subunits (gp90) (14). To examine whether there is a correlation between proteolytic cleavage and fusion property of the MHV S protein, the cleavability of MHV S protein was analyzed. DBT cells that were infected with MHV-JHM or MHV-2 or transfected with an S protein-encoding plasmid were metabolically labeled with [ 35 S]methionine. Cell lysates were precipitated with the antiserum against MHV S protein and resolved by SDS-polyacrylamide gel electrophoresis. Both gp180 and gp90 were detected in DBT cells that were either infected with MHV-JHM or MHV-2 or transiently transfected with the plasmid encoding the S protein of MHV-JHM or MHV-2 (Fig. 4). The indistinguishable cleavabilities of the MHV-JHM and MHV-2 S proteins indicate that the fusion-negative property of MHV-2 is not due to the loss of proteolytic cleavage activity of the S protein.

DISCUSSION
In this study, we have cloned and sequenced the S gene of the fusion-negative MHV-2. The MHV-2 S protein possesses the predicted proteolytic cleavage sequence RRARR and was capable of undergoing cleavage both in MHV-2-infected cells and cells transfected with the S-encoding plasmid. The MHV-2 S1 subunit lacks a 12-amino acid stretch that resulted in its fusion-negative characteristics.
Previous studies have identified amino acid substitutions in a predicted heptad repeat region of S2 subunit that abolished the neutral pH-dependent cell fusion activity of MHV-4 (52). Bos et al. (26) postulated that mutations in the conserved cysteines of the hydrophobic stretch of S2 transmembrane region altered the cell fusion activity of MHV-A59. However, none of the amino acid mutations was found in the fusionnegative MHV-2 strain used in this study. In addition to the 12-amino acid deletion in the S1 subunit of the MHV-2, only three amino acid substitutions in the S2 subunit were identified when compared with the fusion-positive MHV-JHM. Domain swapping experiments indicated that the three amino acid substitutions had little effect on the cell fusion activity; DBT cells transfected with plasmid pCS-m2/j that encodes an MHV S protein with the three point mutations possessed cell fusion activity (Fig. 3). By analyzing a fusion-positive variant of MHV-2, MHV-2f, Yamada et al. (53) concluded that MHV-2 does not possess cell fusion activity due to the lack of the cleavage sequence in the S protein. Nevertheless, the possible effect of a mutation in the signal peptide cannot be ruled out. In addition, although both MHV-2 are nonfusogenic, our MHV-2 strain is different from that of Yamada (53) in that no mutation was found in the signal peptide and proteolytic cleavage sequence. The genetic variation suggested that the fusion-negative MHV-2 strain has evolved variants.
Proteolytic activation of cell fusion has been observed for the negative-stranded orthomyxovirus and paramyxovirus (54,55). In these cases, the newly generated amino terminus after cleavage seems to be critical for cell fusion; synthetic oligopeptides of the amino-terminal sequence inhibited the biological activity of the naturally occurring cleavage products (56). Although cell fusion has been observed with uncleavable S protein of MHV-JHM variants (23)(24)(25)(26), previous studies also demonstrated that proteolytic cleavage of S glycoprotein into S1 and S2 subunits was required to promote cell fusion in some MHV strains (14). In addition, the S2 subunit of MHV appeared to play an important role in the cell fusion activity (19, 26 -28, 52, 57). Nevertheless, the amino terminus of MHV S2 subunit does not have the hydrophobic characteristics of those adjoining the cleavage sites of the orthomyxovirus HA and paramyxovirus F glycoprotein (22,58,59). A putative fusogenic peptide (PEP1) in the longer heptad repeat of the S2 subunit of MHV-A59 was recently identified that possibly functions as an internal fusion peptide (60). The fusion peptide is a relatively hydrophobic alanine/glycine-rich stretch of 19 amino acids. However, MHV-2 S protein was unable to induce fusion in cultured DBT cells, although it retained the proteolytic cleavage activity and the putative fusogenic peptide of MHV-A59. Taken together, both the proteolytic cleavage sequence and the conserved S2 fusogenic peptide may be involved, but may not be sufficient to determine the cell fusion activity of MHV. Additional factors are likely to be required.
Sequence analysis of the S1 subunit of our MHV-2 did not find any amino acid stretch that resembles the putative fusogenic peptides of influenza virus hemagglutinin HA2, Newcastle disease virus glycoprotein F1, human immunodeficiency virus env glycoprotein gp41, semliki forest virus spike protein E1, and vesicular stomatitis virus glycoprotein G (59,61). In addition, the S1 subunit is relatively variable among MHV; deletions and mutations often occur in the hypervariable region (22,31,62). When compared with the S protein of MHV-4, there are three deletions of one, three, and eight amino acid residues in Yamada's MHV-2 strain (31,42,53) and a large deletion of 141 and 153 amino acid residues in the fusion-positive MHV-JHM strain and our fusion-negative MHV-2 strain, respectively, in the hypervariable region of the S1 subunit. Domain swapping experiments indicated that the 12-amino acid stretch, 446 WNRRYGFKVNDR 457 , missing in the S1 subunit of our fusion-negative MHV-2 is critical for the cell fusion activity of MHV-JHM (Fig. 3). Earlier studies have demonstrated that the S1 subunit was involved in binding to cell surface receptor (63) and that the S2 subunit was responsible for cell-cell fusion (19,27,28). Based on our results, the S1 subunit also contributed to the cell fusion activity of MHV.
Conformational change of fusogenic proteins is a current model for the mechanism of protein-mediated membrane fusion (10,64). Deletions and mutations that caused conformational changes of putative fusogenic peptides significantly affected the cell fusion activity of MHV-4, MHV-A59, Newcastle disease virus, human cytomegalovirus, and human immunodeficiency virus (52,60,(65)(66)(67). In these cases, amino acid modification often occurred in the hydrophobic regions of the viral fusogenic proteins. Fusion peptides are often rich in alanine and glycine and have hydrophobicity values ranging from 0.5 to 0.8 (59). Therefore, the 12-amino acid stretch 446 WNRRYGFKVNDR 457 that is lacking in the S1 subunit of MHV-2 does not seem by itself to be a good candidate of fusion domain. Nevertheless, domain swapping experiments did demonstrate a role of the 12-amino acid critical for cell-cell fusion. We thus proposed that the 12-amino acid stretch may not function as a fusion peptide but is involved in maintaining a correct folding of the S protein that would allow the exposure of internal fusion peptides and, in turn, facilitate the cell fusion process upon viral infection. In the fusion-negative MHV-2, although the S protein is capable of undergoing cleavage into S1 and S2 subunits, deletion of the 12-amino acid stretch in the S1 subunit may result in a conformation to which the putative fusion peptides are embedded. Further structural analysis of the S protein of naturally occurring MHV-2 would shed light on the mechanisms of its fusionnegative property.