INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Enveloped viruses acquire their lipid envelopes by budding through cellular membranes. The viral envelope contains integral membrane proteins that play important roles in envelope formation and virus entry. Three models have been proposed for the budding of enveloped viruses. Two of these are based on nucleocapsid-dependent assembly. In retroviruses the nucleocapsid is all that is required, since the Gag core particle can direct its envelopment without the need for viral envelope proteins (9, 23, 31). In contrast, both core and spike proteins are essential for the formation of togaviruses (26, 71), envelopment being driven by direct interactions between the viral envelope proteins and the nucleocapsid (42, 78). In the case of rhabdoviruses, optimal budding efficiency and virus production are achieved by a concerted action of spike and internal virus proteins (48). In the third model, budding is nucleocapsid independent. Coexpression of flavivirus prM and E results in the formation of subviral particles resembling capsidless viral envelopes (1, 44). The hepatitis B virus (HBV) surface proteins can be independently secreted from cells as subviral particles, which, however, are morphologically quite distinct from HBV virions (58, 67). Recently we demonstrated that coronavirus envelope proteins have the capacity to assemble uniform envelopes, which have the same appearance and dimensions as virus particles, independently of nucleocapsid proteins (73).

Coronaviruses are positive-stranded RNA viruses with 30-kb genomes packaged in helical nucleocapsids. The nucleocapsid is incorporated into a viral particle by budding into the intermediate compartment between the endoplasmic reticulum (ER) and the Golgi complex (33, 35, 72). The coronavirus membrane contains three or four viral proteins. The membrane (M) glycoprotein is the most abundant structural protein; it spans the membrane bilayer three times, leaving a short NH₂-terminal domain outside the virus (or exposed luminally in intracellular membranes) and a long COOH terminus (cytoplasmic domain) inside the virion (reviewed by Rottier [62]). The spike protein (S) is a type I membrane glycoprotein that constitutes the peplomers. The small envelope protein (E) has been detected as a minor structural component in avian infectious bronchitis virus (IBV), transmissible gastroenteritis virus (TGEV), and mouse hepatitis virus (MHV) particles (reviewed by Siddell [66]), but it has not been extensively characterized. Some coronaviruses also contain a hemagglutinin-esterase protein (HE) (reviewed by Brian et al. [7]).

Molecular interactions between the envelope proteins are thought to determine the formation and composition of the coronaviral membrane. M plays a predominant role in the intracellular formation of virus particles, for which S appears not to be required. Growth of coronaviruses in the presence of tunicamycin gave rise to the production of spikeless, noninfectious virions (27, 51, 63, 69). These particles were devoid of S but contained M. Independently synthesized MHV M protein accumulates in the Golgi complex (33, 34, 65) in homomultimeric complexes (36). Heterotypic interactions between the M and S proteins have been identified by coimmunoprecipitation and sedimentation analyses (55). The S protein, on its own, is transported to the cell surface, but when it is associated with the M protein, it is retained in the Golgi complex. Upon coexpression of M, S, and E by using the vaccinia virus T7 system (20), virus-like particles (VLPs) containing these three viral membrane proteins were assembled in and released from cells. However, only M and E were required for particle formation. The S protein was dispensable but was incorporated when present (6, 73). The envelope particles produced by this system were shown to form a homogeneous population of spherical particles indistinguishable from authentic virions in size and shape (73).

One of our main interests is to understand the process of coronavirus assembly. We are particularly interested in the interactions and the mechanisms that drive the formation of the viral particles. With the VLP assembly system, we have developed an ideal tool for the study of coronavirus envelope formation and for the analysis of the interactions between the viral membrane proteins in molecular detail. In the present study, we have used the VLP assembly system to investigate the primary structure requirements of the M protein in envelope formation. Site-directed mutations in the carboxy terminus of M were also incorporated into the MHV genome by targeted recombination. The results demonstrate that particle assembly is critically sensitive to changes in all domains of the M molecule.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Cells and viruses. Recombinant vaccinia virus encoding the T7 RNA polymerase (vTF7-3) (20) was obtained from B. Moss. OST7-1 cells (15) (obtained from B. Moss) were maintained as monolayer cultures in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal calf serum (FCS), 100 IU of penicillin/ml, 100 µg of streptomycin/ml, and 400 µg of G418/ml (all from Life Technologies, Ltd., Paisley, United Kingdom). Baby hamster kidney cells (BHK-21) obtained from the American Type Culture Collection (ATCC; Manassas, Va.) were maintained in the same medium lacking G418.

Expression vectors and transcription vectors. Expression construct pTM5ab contains the MHV open reading frames (ORFs) 5a and 5b, the latter encoding the E protein (73) in pTUG3 (74). Expression construct pTUMM contains the MHV strain A59 M gene (obtained from H. Niemann) cloned in the same vector as an XhoI fragment (73). The carboxy-terminal amino acid of the M protein coded by this clone is Thr (46, 54) rather than Ile, the terminal residue originally reported (2). A number of mutations were introduced into the M gene within this construct. Mutations in the amino-terminal domain (designated S₂N, A₂A₃, A₄A₅, A₈A₁₀, Ains2, and His) were made by PCR mutagenesis using 5' terminal primers (Table 1, primers 2 through 7) directing the desired mutations and a 3' internal primer (primer 1) corresponding to the region of the M gene that contains the unique KpnI site. PCR fragments were first cloned into the pNOTA/T7 shuttle vector according to the Prime PCR Cloner procedure (5 Prime3 Prime, Inc.) and were subsequently retrieved by cutting with KpnI, after which the purified fragments were cloned into the expression vector pTUMM, from which the corresponding M fragment had been removed by using KpnI. Mutants N, C and G₁₁N₁₃ were made by using single-stranded phagemid DNA according to the method of Zoller and Smith (79) as described previously (64). Mutant S₃N was made similarly by using primer 8. These mutants were expressed from the transcription vector pTZ19R (47). For the construction of carboxy-terminal M mutants for VLP expression, an intermediate cloning vector was made as follows. By using a 5' flanking primer (primer 9) and a 3' terminal primer (primer 20), the M gene was prepared by PCR from a vector (pSFV1; Life Technologies, Inc.) containing the MHV M gene as a BamHI fragment. The PCR fragment was cloned into the pNOTA/T7 shuttle vector. This vector was recleaved with BamHI, and the resulting fragment was cloned into pBR322. The resulting cloning plasmid (pBM5) contains the mutant M gene 5 as a BamHI fragment and has a unique XbaI site flanking the 3' terminus. The 3, 2, 1, T₂₂₈I, T₂₂₈L, T₂₂₈V, T₂₂₈N, OC, +5, and R₂₂₇A mutant M genes were all made by PCR mutagenesis using a 5' internal primer (primer 10) containing the unique KpnI site and a 3' terminal primer containing the desired mutation and an XbaI site (primers 11 through 20). The PCR fragments were digested with KpnI and XbaI and cloned into pBM5 that had been treated with the same enzymes. The mutant M genes were finally cleaved out with BamHI and cloned into expression vector pTUG3. Mutants 18, 11, and 5 were also made by PCR mutagenesis using a 5' internal primer (primer 10) and 3' terminal primers (primers 21 through 23), and the PCR fragments were cloned into pGEM-T (Promega). The plasmids were digested with KpnI and SpeI, and the resulting fragments were cloned into expression vector pTUMM treated with KpnI and XbaI. The Y₂₁₁G mutant was made with the Altered Sites site-directed mutagenesis kit purchased from Promega. The MHV M gene was cloned as a BamHI fragment into pALTER-1. Primer 24 was used to introduce the mutations, and pALTER-1 was used as the expression vector.

Infection and transfection. Subconfluent monolayers of OST-7 and BHK-21 cells in 10-cm² tissue culture dishes were inoculated at 37°C with vTF7-3 in DMEM at a multiplicity of infection of 10. After 1 h (t = 1 h), cells were washed with DMEM and medium was replaced with transfection mixture, consisting of 0.2 ml of DMEM without FCS but containing 10 µl of Lipofectin (Life Technologies) and 5 µg of each selected construct. After 10 min at room temperature (RT), 0.8 ml of DMEM was added and incubation was continued at 37°C. At t = 2 h, cells were transferred to 32°C and incubation was continued.

Metabolic labeling and immunoprecipitation. At t = 4.5 h, cells were washed with phosphate-buffered saline (PBS) containing Ca²⁺ and Mg²⁺ (PBS⁺⁺) and were starved for 30 min in cysteine- and methionine-free MEM, containing 10 mM HEPES, pH 7.2, without FCS. The medium was then replaced by 600 µl of the same medium containing 100 µCi of ³⁵S in vitro cell labeling mix (Amersham). After 3 h, cells were placed on ice, and the media were collected and cleared by centrifugation for 15 min at 4,000 × g and 4°C. Cells were washed with ice-cold PBS⁺⁺ and lysed with lysis buffer, consisting of 20 mM Tris-HCl (pH 7.6), 150 mM NaCl, 1% Nonidet P-40 (NP-40), 0.5% sodium deoxycholic acid (NaDOC), 0.1% sodium dodecyl sulfate (SDS), 2 µg of aprotinin/ml, 2 µg of leupeptin/ml, and 1 µg of pepstatin A/ml. Lysates were cleared by centrifugation for 10 min at 10,000 × g and 4°C. Radioimmunoprecipitation was performed on lysates diluted 5 times with immunoprecipitation buffer, consisting of 20 mM Tris-HCl (pH 7.6), 150 mM NaCl, 5 mM EDTA, 0.5% NP-40, 0.1% NaDOC, 0.1% SDS, and protease inhibitors. Culture media were prepared for immunoprecipitation by addition of 1/4 volume of 5-times-concentrated lysis buffer. Rabbit anti-MHV serum k134 (61) was used at a 500-fold dilution for immunoprecipitation of MHV proteins at 4°C. The immune complexes were adsorbed to Pansorbin cells (Calbiochem) for 30 min at 4°C and were subsequently collected by centrifugation. Pellets were washed three times by resuspension and centrifugation using 20 mM Tris-HCl (pH 7.6)-150 mM NaCl-5 mM EDTA-0.1% NP-40 followed by a single wash using 20 mM Tris-HCl (pH 7.6)-0.1% NP-40. The final pellets were suspended in electrophoresis sample buffer and heated at 95°C for 2 min before analysis by SDS-polyacrylamide gel electrophoresis (PAGE) using a 15% polyacrylamide gel according to the method of Laemmli (38). In some cases immunoprecipitates were digested with endoglycosidase F/N-glycosidase F (glyco F; Boehringer Mannheim) as described earlier (10) before analysis by SDS-PAGE.

Indirect immunofluorescence. Indirect immunofluorescence was performed on BHK-21 cells grown on 12-mm coverslips. The morphology of these cells makes them more convenient than OST7-1 cells for this assay. Cells were fixed at t = 5 h, permeabilized, and stained for immunofluorescence as described previously (33). The rabbit anti-MHV serum k134 was used at a 1:400 dilution.

Construction of MHV mutants. Carboxy-terminal M mutations were incorporated into the genome of MHV by targeted recombination between synthetic donor RNA from HindIII-truncated transcription vectors and the thermolabile N-gene deletion mutant Alb4 as described previously (16, 45, 59). Candidate recombinant viruses were plaque purified and analyzed by reverse transcription-PCR (RT-PCR) and sequencing of RNA from infected cells. Final confirmation of the construction of individual mutants came from direct sequencing of RNA isolated from purified virions (16, 59).

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Effects of cytoplasmic-domain mutations on VLP assembly. In order to elucidate the primary structure requirements of the MHV M cytoplasmic domain for virus assembly, we constructed a number of mutants. Mutant C has a large internal deletion, removing residues E₁₂₁ through D₁₉₅, which comprise most of the amphiphilic domain. Mutant 18 lacks the carboxy-terminal hydrophilic domain. Since this deletion appeared to have a drastic effect on VLP assembly, a series of mutants was made with progressively smaller deletions at the carboxy terminus, ranging from 11 amino acids to a single amino acid (Fig. 1). The abilities of these mutant M molecules to function in assembly were tested by coexpressing each of the mutant genes with the E protein gene. Genes were expressed by using the recombinant vaccinia virus bacteriophage T7 RNA polymerase system in OST7-1 cells. Proteins were labeled with ³⁵S-labeled amino acids from 5 to 8 h postinfection. Cells and media were collected separately and processed for immunoprecipitation with a polyclonal rabbit anti-MHV serum, followed by SDS-PAGE using a 15% polyacrylamide gel. Analysis of the cell lysates (Fig. 2) of the single expressions demonstrated that all mutant constructs were expressed and yielded products of expected sizes. In all cases M appeared as a set of proteins differing in apparent molecular size, due to different extents of O glycosylation (34, 72). The patterns of the glycosylated species of M mutants were not much different from that of the wild-type (WT) M protein, indicating that the mutations in the carboxy terminus did not affect the ability of the amino terminus to become glycosylated, nor did they affect the ability of the proteins to be transported to the Golgi complex, as the slowest-migrating forms of M are known to result from modifications occurring in this organelle (34). Transport to the Golgi complex was confirmed by immunofluorescence analyses as illustrated in Fig. 3. The mutant M proteins were also analyzed for their stability by pulse-chase experiments and appeared to be as stable as WT M. In the double expressions, the presence of the E protein did not seem to affect the synthesis of M qualitatively under the experimental conditions used; in some cases, the expression level of M was somewhat decreased. The E protein itself was not resolved due to poor recognition by the antiserum, but its synthesis was confirmed by using an E-specific serum (data not shown). Particle assembly and secretion were assayed by measuring the release of the M protein into the culture medium. The E protein was extremely difficult to detect in VLPs, due to its small size and very low abundance (73). It is clear from observation of WT M protein that no M was released into the medium unless the E protein was coexpressed, consistent with our earlier findings (73). However, all mutant M proteins failed to be secreted into the culture medium when they were expressed in combination with the E protein. Not only an internal deletion in the cytoplasmic domain but also deletions at the carboxy terminus were fatal for VLP assembly. Even the deletion of one single amino acid at the extreme carboxy terminus abolished VLP formation almost completely (Fig. 2), indicating that the carboxy-terminal residue is very important.

View larger version (26K): [in this window] [in a new window]   FIG. 1.   Overview of mutant M proteins. In the middle is a schematic representation of the structure of the M protein, with the three transmembrane (TM) domains indicated. Amino acid sequences of the amino-terminal and carboxy-terminal domains, and of the mutants with mutations in these domains, are shown above and below the diagram, respectively. Dashes represent unchanged residues, gaps represent deletions. The domain deleted in mutant C (residues E121 through D195) is indicated by a horizontal line. The abilities of the different M proteins to support VLP assembly is indicated at the right. The ratio of the amount of M present in the culture medium to that in cells was taken as a measure for VLP assembly by using WT M as a reference. The semiquantitative scores ++, +, +/, and  indicate efficient, moderately efficient, inefficient, and nonexistent VLP synthesis, respectively.

View larger version (35K): [in this window] [in a new window]   FIG. 2.   Effects of deletions in the M cytoplasmic domain on VLP assembly. OST7-1 cells infected with recombinant vaccinia virus vTF7-3 were transfected with a plasmid containing the WT or mutant M gene either alone or in combination with a plasmid containing the E protein gene, each gene behind a T7 promoter. Cells were labeled for 3 h with 35S-labeled amino acids. Both cells (upper panel) and the culture medium (lower panel) were prepared and used for immunoprecipitation, and the precipitates were analyzed by SDS-PAGE. The different M genes expressed are indicated above each set.

View larger version (66K): [in this window] [in a new window]   FIG. 3.   Indirect immunofluorescence analysis of mutant M proteins. vTF7-3 infected BHK-21 cells were transfected with plasmids encoding WT M (A) or the C (B), 18 (C), N (D), A4A5 (E), A8A10 (F), or A8A10C14 (G) mutant or were mock transfected (H and I). Cells were fixed at 5 h postinfection and processed for immunofluorescence with the anti-MHV serum k134 (A through H) or a polyclonal rabbit serum against the resident Golgi protein -mannosidase II (I) (a kind gift from K. Moremen [50]).

View larger version (34K): [in this window] [in a new window]   FIG. 4.   Effects of mutations in the M carboxy-terminal domain on VLP assembly. Expression of M and E genes was performed as described in the legend to Fig. 2. The different M genes tested are indicated above the gels.

Effects of transmembrane deletions on VLP assembly. To assess the role of the transmembrane domains in VLP assembly, mutant proteins lacking either the first (a), the second (b), or the third (c) transmembrane domain or combinations thereof [(a+b), (b+c), and ac] were tested for their VLP-forming abilities. The construction of these mutants and their expression in vitro and in vivo have been described earlier (37). When these mutant M proteins were coexpressed with E, no VLPs were detected in any case (data not shown). These results indicate the importance of the transmembrane domains in preserving the functional structure, orientation, and localization of the M molecule.

Effects of amino-terminal domain mutations on VLP assembly. Next, we wanted to investigate whether VLP formation is also sensitive to changes in the amino-terminal domain, i.e., the luminal domain of the M protein. For this purpose another set of mutants was constructed with various mutations in this domain. In mutant A₂A₃, the serines at positions 2 and 3 were replaced by alanines, while in mutant A₄A₅, the threonines at positions 4 and 5 were replaced by alanines (Fig. 1). Mutant N lacks residues A₇ through F₂₂, resulting in an internal deletion of 16 amino acids. Three mutants have substitutions in this domain. In mutant A₈A₁₀, the prolines at positions 8 and 10 were replaced by alanines. A₈A₁₀C₁₄ also has a cysteine substitution for the tryptophan at position 14, which was fortuitously obtained during the construction of mutant A₈A₁₀. Mutant G₁₁N₁₃ has replacements of valine and glutamine at positions 11 and 13 with glycine and asparagine, respectively. Furthermore, two mutants with insertions between the initiating methionine and the serine at position 2 were constructed. Mutant Ains2 has an insertion of 1 alanine, while mutant His has a stretch of 6 histidines inserted for purification purposes. The mutant constructs were expressed alone and in combination with the E protein gene. Cells and media were processed and analyzed as described above. Analysis of the cell lysates (Fig. 5) demonstrated that in all cases mutant M proteins of expected sizes were expressed but that their glycosylation patterns were variously affected. It should be noted that O glycosylation of MHV M occurs at a threonine in the amino-terminal domain (unpublished data). Thus, mutant A₄A₅ did not become glycosylated, and this did not reflect an inability to be transported to the Golgi complex, as was verified by immunofluorescence (Fig. 3E). The same holds true for mutant N; the deletion blocked glycosylation at the threonines without affecting intracellular transport (Fig. 3D). Transport was also unaffected for mutants A₈A₁₀ and A₈A₁₀C₁₄ (Fig. 3F and G). The extents of glycosylation of these mutants were decreased, apparently due to the replacements of the prolines. The normal pattern of differently glycosylated species was observed with mutants A₂A₃, G₁₁N₁₃ and Ains2. The His mutant also showed the usual pattern of glycosylation, indicating that the insertion of the histidine stretch did not interfere with the membrane translocation of the amino-terminal domain, which occurs through the action of an internal signal sequence (61). The normal glycosylation also indicates that the protein's transport to the Golgi complex was not affected. One reason to prepare mutant G₁₁N₁₃ was to obtain an N-glycosylated form of MHV M. The N glycosylation consensus sequence generated by the introduction of the asparagine at position 13 appeared, however, not to be used by the cell. The protein's modifications are indistinguishable from those of WT M and are insensitive to endoglycosidases that remove N-linked sugars.

View larger version (31K): [in this window] [in a new window]   FIG. 5.   Effects of mutations in the amino-terminal domain of M on VLP assembly. M and E genes were expressed as described in the legend to Fig. 2.

Assembly of N-glycosylated M protein in VLPs. Coronavirus M proteins are either N glycosylated or O glycosylated in the amino-terminal domain (62). Murine coronaviruses belong to the latter category. Above we showed that O glycosylation is not required for MHV membrane assembly. In order to investigate whether MHV M would still be able to function in assembly as an N-glycosylated protein, we constructed a mutant (S₂N) in which the serine residue at position 2 was replaced by an asparagine, thus creating an N glycosylation consensus sequence (Asn-X-Ser/Thr [22]). The mutant construct was expressed alone or in combination with the E protein gene. Initial expression studies with the mutant showed that the protein can become both O and N glycosylated in a complex pattern that will be described elsewhere (unpublished data). One major complication was the maturation of the N-linked side chain by extensive heterogeneous modifications, resulting in a diffuse smear in the gel, which hampered the detection of N-glycosylated M protein in cells and VLPs. In order to avoid this problem, we used an inhibitor of oligosaccharide maturation, 1-deoxy-mannojirimycin (DMJ), which interferes with the action of -mannosidase I, keeping the sugars in a simple, endoglycosidase H (endo H)-sensitive form (21). In other cases, cells were treated with tunicamycin, a general inhibitor of N glycosylation (reviewed by Elbein [14]). Cells and media were processed and analyzed as described above. Prior to gel electrophoresis, some immunoprecipitates were treated with glyco F to remove the N-linked sugars. Analysis of the cell lysates (Fig. 6) showed the mutant S₂N protein to appear both in an unglycosylated form (lower band; about 23 kDa) and as some N-glycosylated species (lanes 5 and 6). The distinct band of about 28 kDa consisted of M protein carrying N-linked sugars that were endo H sensitive (data not shown). The endo-H-resistant S₂N was differentially glycosylated and could not be distinguished from the background due to its heterogeneity. After treatment with glyco F, the N-linked sugars were removed, resulting in the typical pattern of differently O-glycosylated M species (Fig. 6, lanes 7 and 8). Hence, S₂N was both N and O glycosylated. After treatment with DMJ, the N-glycosylated M proteins appeared as a 28- and a 30-kDa species (Fig. 6, lanes 9 and 10). When immunoprecipitates were treated with glyco F, the typical pattern of differently O-glycosylated M species was again observed (Fig. 6, lanes 11 and 12). DMJ did not influence expression of WT M quantitatively or qualitatively (Fig. 6, lanes 3 and 4). Treatment of S₂N-expressing cells with tunicamycin resulted in S₂N that was O glycosylated but not N glycosylated (Fig. 6, lanes 13 and 14).

View larger version (45K): [in this window] [in a new window]   FIG. 6.   Assembly of an N-glycosylated form of the MHV M protein into VLPs. OST7-1 cells infected with recombinant vaccinia virus vTF7-3 were transfected with a plasmid containing the WT or the S2N mutant M gene, either alone or in combination with a plasmid containing the E protein gene, each gene behind a T7 promoter. Cells were labeled for 3 h with 35S-labeled amino acids. In some cases, cells were treated with 1 mM DMJ (lanes 3 and 4, 9 to 12, and 16); in other cases, cells were treated with 5 µg of tunicamycin (TM)/ml (lanes 13 and 14). Both cells (upper panel) and the culture medium (lower panel) were prepared and used for immunoprecipitation, and the precipitates were analyzed by SDS-PAGE. Prior to gel electrophoresis, some immunoprecipitates were treated with glyco F to remove N-linked sugars (lanes 7, 8, 11, and 12).

Inhibition of normal VLP formation by M proteins with carboxy-terminal tail deletions. Interactions between M molecules are considered essential in coronavirus envelope assembly. It was therefore of interest to analyze whether and how mutant M proteins that are themselves deficient in VLP formation would interfere with the assembly of VLPs driven by WT M and E. Therefore, a triple-expression experiment in which these proteins were coexpressed with different carboxy-terminal tail mutants was performed. Fixed amounts of plasmid DNA encoding WT M and E were used in transfection, while an equal or a 5-times-lower amount of the plasmid DNAs specifying the mutant M proteins was used. Cells and media were processed as described above. Analysis of the cell lysates (Fig. 7) in all cases showed the differently glycosylated M species. Due to the small differences in size, mutants 5, 2, and 1 could not be discriminated from WT M. The unglycosylated form of mutant 18 could be distinguished, but its glycosylated forms were also obscured by WT M. In this case the glycosylated species of WT M could be discriminated from those of the mutant because the former run slightly slower. The results indicate that the expression of WT M was hardly affected by the coexpression of mutant 18, even when equal amounts of their plasmids had been cotransfected. Judging from the amounts of the unglycosylated forms, it seems that the efficiency of expression was higher for WT M than for the mutant.

View larger version (45K): [in this window] [in a new window]   FIG. 7.   Inhibition of VLP formation by mutant M proteins with carboxy-terminal deletions. After infection of OST7-1 cells with recombinant vaccinia virus vTF7-3, cells were transfected with plasmid DNA encoding WT M and E (5 µg of each) together with 1 or 5 µg of plasmid DNA encoding mutant M. The different M mutants expressed are indicated above the gels. Cells were labeled for 3 h with 35S-labeled amino acids. Both cells (upper panel) and the culture medium (lower panel) were prepared and used for immunoprecipitation, and the precipitates were analyzed by SDS-PAGE.

View larger version (43K): [in this window] [in a new window]   FIG. 8.   Quantitation of the concentration-dependent inhibition of VLP formation by mutant M protein 18. The relative amounts of glycosylated WT or unglycosylated 18 mutant M protein present in cells and media for each transfection condition shown in Fig. 6 were determined by using a phosphorimager. The relative amount of WT M secreted (i.e., VLPs in media) is also shown.

Rescue of deletion mutants into VLPs. Our inability to discriminate between the WT M protein and the mutant M proteins in these competition experiments did not allow an accurate analysis of the possible rescue of the mutant proteins into VLPs. To circumvent this problem and to increase the sensitivity of detection, we made use of mutant A₂A₃, which we showed above to function efficiently in VLP assembly (Fig. 5). Moreover, we found that the mutations in this M protein destroyed the epitope recognized by the monoclonal antibody J1.3 (17). Replacing WT M by this mutant therefore allowed the desired discrimination. Rescue experiments were performed as described above. To allow sufficient VLP production, a 5-times-lower amount of plasmid DNA specifying the M proteins to be rescued was used compared to the amount of A₂A₃ plasmid. Analysis of the cell lysates (Fig. 9) showed that A₂A₃ protein is precipitated with the polyclonal anti-MHV serum k134 but not with the monoclonal antibody J1.3. When mutants 18, 5, 2, or 1 or WT M was coexpressed with mutant A₂A₃, M proteins were immunoprecipitated with the polyclonal anti-MHV serum. Only after prolonged exposure could M protein be detected in the samples immunoprecipitated with J1.3, indicating that under the conditions used, the bulk of the expressed M protein is A₂A₃.

View larger version (35K): [in this window] [in a new window]   FIG. 9.   Rescue of carboxy-terminal deletion mutant M proteins into VLPs. After infection of OST7-1 cells with recombinant vaccinia virus vTF7-3, cells were transfected with plasmid DNA encoding mutant M protein A2A3 and E protein (5 µg of each). In some cases, these were cotransfected with 1 µg of plasmid DNA encoding the indicated mutant or WT M protein. Cells were labeled for 3 h with 35S-labeled amino acids. Both cells (upper panel) and the culture medium (lower panel) were prepared and used for immunoprecipitation, either with the polyclonal anti-MHV serum k134 or with the monoclonal antibody J1.3, and the precipitates were analyzed by SDS-PAGE.

Incorporation of cytoplasmic domain mutations into virions. To further our understanding of the consequences of M protein carboxy-terminal tail mutations in the presence of the full complement of virion structural components, we sought to directly introduce many of these mutations into the genome of MHV A59. This was accomplished by targeted RNA recombination between a synthetic defective interfering (DI) RNA analog containing one of the intended mutations and the N-gene deletion mutant Alb4 as the recipient virus (16, 45, 59) (Fig. 10A). In this manner we were able to isolate several recombinant viruses containing each of the carboxy-terminal residue substitutions that had been studied in the VLP system: T₂₂₈I, T₂₂₈L, T₂₂₈V, and T₂₂₈N (Fig. 10B). Another recombinant was constructed in which the carboxy-terminal half of the MHV M protein (amino acid residues 134 to 228) was replaced with its counterpart from BCV (residues 133 to 230), a region containing 17 amino acid differences from MHV, including the residues NNI_228-230 in place of T₂₂₈ (Fig. 10B). In addition, the substitution mutants T₂₂₈F, T₂₂₈Y, and T₂₂₈M were also created in the viral M protein (Fig. 10B). (The T₂₂₈F mutant also contained a secondary mutation, T₂₂₈I.) With one exception, all of these M protein mutants had plaque sizes indistinguishable from that of the WT virus, and at passage 3 all had high yields in tissue culture (>10⁸ PFU/ml), comparable to or better than that of the WT. For a subset of the mutants, T₂₂₈I, T₂₂₈V, and T₂₂₈N, virus stocks were serially propagated for six passages. Following every passage the genetic stability of each mutant was monitored by direct sequencing of an RT-PCR product of the region containing the constructed mutation. None of the sequences showed any detectable reversion or other secondary genetic alteration, indicating that the introduced genomic changes were stable for at least six passages. The only phenotypic difference noted was with T₂₂₈V mutants, which, for both of the independent recombinants analyzed, exhibited a slightly smaller plaque size than the WT at 39°C. These results showed that assembly of the complete virion of MHV tolerates changes in the carboxy-terminal residue of the M protein that are considerably more deleterious for the formation of particles from just the M and E proteins (Fig. 4). In the most extreme case, the T₂₂₈N mutation, which severely diminished VLP production, had no obvious phenotypic effect when incorporated into MHV. As can be seen in Fig. 10B, direct sequencing of RNA isolated from highly purified virions of the T₂₂₈N mutant Alb206 revealed a substantial amount of overlapping sequence from the leader region of subgenomic RNA 7, which was read by the same primer. This was observed in multiple independent preparations of purified virions of this mutant and of an independent T₂₂₈N mutant, Alb205. For Alb205, we confirmed the presence of the mutant asparagine codon by sequencing an RT-PCR product from the M-N junction region of the viral genome. We have previously noted that highly purified MHV contains a small amount of packaged subgenomic RNAs (60). It remains for further work to determine if the T₂₂₈N mutants aberrantly package greater amounts of these RNA species.

View larger version (56K): [in this window] [in a new window]   FIG. 10.   Incorporation of M protein cytoplasmic-domain mutations into the MHV genome. (A) Schematic for construction of M mutants by targeted recombination between donor synthetic RNA and the N-gene deletion mutant Alb4. Each donor RNA containing a codon 228 mutation in the M gene (denoted by a star) was transcribed from a vector derived from pCFS8 (16), which includes the entire portion of the MHV genome 3' to the start of the S gene and is tagged with a 19-base marker in gene 4 (denoted by a triangle). In the case shown, the MHV mutant, generated by a single upstream crossover, has inherited the gene 4 tag, the constructed M mutation, and the region repairing the N-gene deletion. (B) Sequence of the relevant region of genomic RNA isolated from passage-3 purified virions of one mutant of each type. For each codon 228 mutation, recombinants were obtained that contained and that lacked the upstream gene 4 tag. The particular mutants shown are all positive for the gene 4 tag, except for Alb206 and Alb200. For Alb200, a BCV-MHV chimeric M protein mutant, the donor RNA was derived from pFV1 (16) and did not contain the gene 4 tag.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Despite a high degree of sequence variation among the coronavirus M proteins, there is a surprising conservation of their overall chemical features (reviewed by Rottier [62]). The dominant common feature is the occurrence of three hydrophobic domains alternating with short hydrophilic regions. The carboxy-terminal half of all M proteins is largely amphiphilic, with a hydrophilic end. This conservation indicates that there are rigid structural constraints on M as a result of functional requirements. Indeed, M has been shown to play a key role in coronavirus assembly. M and E are the only requirements for particle formation (73). Using an envelope assembly assay, we have now shown that all domains of the M protein are important. Changes in the primary structure of the luminal domain, the transmembrane cluster, the amphiphilic domain, and the carboxy-terminal domain had effects on the assembly of M into enveloped particles. Clearly, in order for M to function in assembly, stringent requirements must be satisfied.

The carboxy-terminal domain is located in the cytoplasm. In infected cells this domain is probably important for virus assembly by interacting with the nucleocapsid. An affinity of M for nucleocapsids has been observed in vitro with several coronaviruses. Subviral particles prepared by NP-40 disruption of purified MHV (70, 75), avian IBV (39), or BCV (32) still contained M protein associated with the nucleocapsids. Our results indicate that this cytoplasmic domain is also of crucial importance for the assembly of the viral envelope. Both the exposed carboxy end and the amphiphilic, protease-resistant domain between this end and the transmembrane domains appeared to be essential structural requirements. Strikingly, the extreme carboxy-terminal residues were found to be crucial. The VLP system provided the most sensitive indicator of this: deletion of as little as the single terminal residue of M protein (1) was fatal, while substitutions or extensions at this position generally had a strong negative effect on assembly. Relative to this, in the whole virus the threshold for loss of function was slightly displaced. The 1 mutation had no detectable effect when incorporated into the MHV genome, and only one of eight different substitutions of the carboxy-terminal residue had a weakly measurable impact on viral phenotype. These results suggest that in the complete virion an additional component, most likely the nucleocapsid, provides a further measure of stabilization not available in VLPs composed purely of M and E proteins. Moreover, there may be other factors, such as the kinetics of virion assembly or the relative or absolute levels of viral protein expression, that render the whole virus more tolerant to mutational changes that are devastating in the VLP system. Nevertheless, further truncation of the M protein carboxy terminus in the virion, as in the VLP system, appeared to abrogate its function. In multiple attempts, we were unable to incorporate the 2, 5, or 18 mutation into the MHV genome, suggesting that these mutations, if not absolutely lethal, were at least as harmful to the virus as the lesion in Alb4, the N deletion mutant from which recombinants were selected. These observations all point to a sensitive role of the extreme terminal residues in some aspect of envelope assembly. The exact nature of this role remains unclear, but it is not unlikely that the very carboxy terminus is involved in interactions either intramolecularly, to establish a particular secondary structure in the M molecule, or intermolecularly, with membrane lipids, with E, or with other M proteins. A role of the carboxy terminus in homotypic interactions is, however, unlikely in view of earlier data. Using sucrose gradient analysis, we have shown previously that the MHV M protein alone can associate into large homomeric complexes (36), but this oligomerization still occurred when the carboxy-terminal 22 residues of the protein were deleted.

The cytoplasmic domain of envelope proteins has been assigned an important role in the assembly of many enveloped viruses. It was shown to be essential for incorporation of the glycoprotein of vesicular stomatitis virus (57, 76). For alphaviruses the cytoplasmic domain of the E2 envelope glycoprotein appeared to be critical, with particular roles for a tyrosine, a leucine, and a set of cysteines important for palmitoylation (29, 56, 78). Interestingly, the 2-residue cytoplasmic tail of glycoprotein E1 is dispensable for virus growth (4). In the case of retroviruses, a direct interaction in vitro was reported between the matrix protein and the cytoplasmic domain of the Env protein of human immunodeficiency virus type 1 (HIV-1) (8). Deletions in the cytoplasmic domain of the Env protein of HIV-1 (11, 18, 77) and murine leukemia virus (24) affect Env incorporation into virions. Remarkably, HIV-1 Env mutants with large cytoplasmic deletions can be incorporated into virions in a matrix-independent manner (19, 43). The influenza virus hemagglutinin (HA) cytoplasmic tail is not essential for virus assembly (30, 52, 68). However, deletion of the cytoplasmic tail of influenza virus neuraminidase (NA) severely compromised the incorporation of mutant NA molecules into virions (5, 49).

The amino-terminal domain of the coronaviral M protein is exposed luminally in cellular organelles. This domain varies considerably in length, from some 16 residues in the mature protein of the human coronavirus (HCV) 229E to around 36 in that of feline coronaviruses (see reference 62). No function has been assigned to it yet. Somewhat to our surprise, some changes in this domain did affect MHV VLP formation. At the very amino terminus, insertion of a single residue (alanine) following the initiating methionine was allowed, but a six-His stretch at this position was nearly fatal. Changes in the clusters of hydroxyl amino acids flanking the methionine, or substitutions of the two nearby prolines, were without effect. In contrast, deletion of the middle part of the ectodomain or some substitutions in this domain had severe effects. One such substitution involved a fortuitously obtained tryptophan-to-cysteine mutation at a position close to that where a cysteine naturally occurs in the HCV 229E M protein. While the latter cysteine gives rise to the formation of homodimers (3), no disulfide bonds were observed with the MHV M mutant protein. For alphaviruses, mutations in the ectodomain of the envelope proteins have been shown to affect virus assembly (12, 25, 41). It was suggested that these mutations resulted in impaired lateral interactions causing the budding defect. This might also be the explanation for the behavior of the amino-terminal M mutants.

Coronavirus M proteins are invariably glycosylated, carrying either N-linked or O-linked oligosaccharides (see reference 62). The oligosaccharides are attached to the amino-terminal region of the M molecule and are thus exposed at the virion surface. Since a function for this modification has not been identified, we used our VLP system to study the possible involvement of glycosylation in virus assembly. Our results showed that for MHV, O glycosylation of the M protein is not required. We then created an N-glycosylated form of the protein which appeared still to produce particles, though with decreased efficiency. Only when the maturation of the N-linked oligosaccharide was inhibited did VLP production become normal, indicating that N glycosylation per se was not interfering with assembly. Altogether, these observations demonstrate that glycosylation plays no role in assembly, consistent with the findings of earlier studies that used tunicamycin (69) and monensin (53) to inhibit glycosylation in infected cells. In addition, Laude et al. (40) isolated a mutant of the porcine TGEV in which the sole N-glycosylation site of the M protein had been disrupted without affecting the viability of the virus.

As mentioned above, homotypic interactions between M molecules are supposed to be essential in coronavirus envelope formation. It was therefore of interest to investigate whether M molecules that are themselves deficient in VLP formation would interfere with the formation of particles by assembly-competent M and E. From the set of carboxy-terminal tail deletion mutants of M that we studied, it was clear that interference indeed occurred, in a concentration-dependent manner. Furthermore, assembly-competent M protein was able to rescue assembly-negative M molecules into particles. These observations support the importance of lateral interactions between M molecules in the assembly process. Moreover, they also support our earlier conclusion that such interactions do not require the cytoplasmic tail of the M protein. This conclusion was based on the finding that a mutant M protein lacking the carboxy-terminal 22 amino acids was able to associate into large heterogeneous complexes as does WT M, as shown by sucrose density gradient analysis (36). Lateral interactions between envelope proteins have similarly been demonstrated for Semliki Forest virus. Here it was shown that nucleocapsid binding-deficient p62-E1 heterodimers did inhibit normal virus budding in a concentration-dependent way and that under suitable conditions these heterodimers could be rescued into virus particles (13). It is quite likely that lateral interactions between viral glycoproteins generally are instrumental in the assembly of enveloped viruses.