Role of Matrix and Fusion Proteins in Budding of Sendai Virus

Abstract

Paramyxoviruses are assembled at the surface of infected cells, where virions are formed by the process of budding. We investigated the roles of three Sendai virus (SV) membrane proteins in the production of virus-like particles. Expression of matrix (M) proteins from cDNA induced the budding and release of virus-like particles that contained M, as was previously observed with human parainfluenza virus type 1 (hPIV1). Expression of SV fusion (F) glycoprotein from cDNA caused the release of virus-like particles bearing surface F, although their release was less efficient than that of particles bearing M protein. Cells that expressed only hemagglutinin-neuraminidase (HN) released no HN-containing vesicles. Coexpression of M and F proteins enhanced the release of F protein by a factor greater than 4. The virus-like particles containing F and M were found in different density gradient fractions of the media of cells that coexpressed M and F, a finding that suggests that the two proteins formed separate vesicles and did not interact directly. Vesicles released by M or F proteins also contained cellular actin; therefore, actin may be involved in the budding process induced by viral M or F proteins. Deletion of C-terminal residues of M protein, which has a sequence similar to that of an actin-binding domain, significantly reduced release of the particles into medium. Site-directed mutagenesis of the cytoplasmic tail of F revealed two regions that affect the efficiency of budding: one domain comprising five consecutive amino acids conserved in SV and hPIV1 and one domain that is similar to the actin-binding domain required for budding induced by M protein. Our results indicate that both M and F proteins are able to drive the budding of SV and propose the possible role of actin in the budding process.

Enveloped viruses possess a membrane that encases the viral nucleocapsid or core, which contains the viral genome. The membrane and its viral glycoproteins are acquired during the final stage of viral assembly and budding. Sendai virus (SV), a prototype of the Paramyxoviridae, is a pleomorphic enveloped virus whose membrane has the hemagglutinin-neuraminidase (HN) and fusion (F) glycoproteins embedded on its outer surface and matrix (M) protein embedded on its inner surface. The M proteins interact with each other at the inner surface of the lipid bilayer to form a sheet that interacts with viral nucleocapsid and glycoproteins (26). The M protein can promote vesiculation of the membrane and release of M-containing particles into the extracellular medium without the aid of other viral proteins (16, 19). When M is coexpressed with homologous viral nucleoprotein (NP), vesicles containing M and nucleocapsid-like structures are produced (6). It is therefore believed that the M protein orchestrates the budding of paramyxovirus by acting as a bridge between the nucleocapsid and the plasma membrane.

There is also evidence that the cytoplasmic domains of spike glycoproteins play a crucial role in virus budding and assembly at the plasma membrane. For example, the particle formation of a rabies virus mutant that is deficient for glycoprotein G was enhanced 6- and 30-fold in the presence of tailless G or G, respectively (19). Also, the budding of influenza virus encoding tailless hemagglutinin and neuraminidase proteins was found to be significantly impaired (15). In paramyxoviruses, truncation of the cytoplasmic tail of HN of simian virus 5 or SV HN or F caused inefficient release of progeny virus particles from infected cells (11, 32).

The cellular cytoskeleton has also been reported to play an important role in the assembly of paramyxoviruses. Cytoskeletal components appear to be directly involved in the transport of viral glycoproteins to the assembly site (10). Cellular actin has been found in purified preparations of paramyxoviruses (17, 23, 38), rabies virus (21), and human immunodeficiency virus (24). The M1 protein of influenza virus colocalizes with the actin cytoskeleton in vivo (4), and paramyxovirus M protein and human immunodeficiency virus Gag protein bind directly to filamentous actin (F-actin) (12, 29). Despite these findings, however, the role of actin in virus assembly and budding is not well understood.

In this study, we investigated the ability of individually expressed SV membrane proteins (M, F, and HN) to induce the formation of vesicles that contain the respective proteins. Expression of SV M induced the production of M-bearing virus-like particles, and expression of SV F induced the production of virus-like particles bearing F spikes at their surface. By analyzing mutant M and F proteins, we identified the domains required for their induction of budding.

MATERIALS AND METHODS

Cells and viruses.

We cultured 293T cells (9) in Dulbecco's modified Eagle's medium with 10% fetal calf serum. SV was propagated in 10-day-old chicken embryonated eggs.

cDNA clones.

The M and F genes of SV were cloned from viral RNAs by using the Titan RT-PCR system (Roche Molecular Biochemicals, Indianapolis, Ind.). The primers were specific for noncoding regions of the genes. The cDNAs were cloned into the transient expression vector pCAGGS (22), and the plasmids containing the SV M and F genes were designated pCAGGS-SVM and pCAGGS-SVF, respectively. The creation of pCAGGS vectors for the SV HN gene and human parainfluenza virus type 1 (hPIV1) M gene has been described elsewhere (6, 36). Mutant cDNA clones which express deletion mutants of the SV and hPIV1 M proteins were constructed by inserting a stop codon into the respective M genes with the Transformer site-directed mutagenesis kit (Clontech, Palo Alto, Calif.).

Mutant SV F genes were generated by using the same kit.

Detection of release of vesicles into culture medium and immunoprecipitation of the expressed proteins.

We used 16 μg of Lipofectamine (Life Technologies, Grand Island, N.Y.) and 2 μg of expression vectors encoding M, F, or HN genes to transfect 293T cells growing in six-well plates. To obtain cells that coexpressed two proteins, we used 1 μg of each expression vector for transfection. Cells were cultured for 24 h and then labeled overnight with 100 μCi of [³⁵S]Trans-Label (ICN, Costa Mesa, Calif.) at 33°C. The culture medium was briefly clarified, and vesicles released into the medium were then purified by ultracentrifugation at 190,000 × g for 2 h at 4°C through 4 ml of 30% glycerol in phosphate-buffered saline (PBS). Vesicles were resuspended in Laemmli reducing sample buffer and analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE).

To quantify the expression of proteins in the transfected cells, we first washed the cells with PBS and lysed them with 1 ml of TNE buffer (10 mM Tris [pH 7.4], 150 mM NaCl, 0.5% NP-40, 1 mM EDTA). The lysates were then clarified by centrifugation at 15,000 × g for 10 min, and the supernatants were analyzed by immunoprecipitation. Two microliters of a cocktail of monoclonal antibodies (MAbs) to SV M, F, or HN were incubated with 20 μl of Dynabeads (Dynal, Lake Success, N.Y.) in TNE buffer at 4°C for 30 min. The MAb-Dynabead complexes were washed with TNE buffer and incubated with 100 μl of cell lysate in TNE buffer at 4°C for 30 min. The immunocomplexes were washed with TNE buffer and analyzed by SDS-PAGE. Proteins were quantified by using a STORM 860 imaging system (Amersham Pharmacia Biotech, Piscataway, N.J.).

Cell surface expression of F protein.

Expression of F protein at the cell surface was quantified by cell surface enzyme-linked immunosorbent assay (3). Cells were incubated with the cocktail of anti-SV F MAbs and then with horseradish peroxidase-conjugated sheep anti-mouse immunoglobulin G (Bio-Rad, Hercules, Calif.) in PBS containing 0.1% bovine serum albumin. The cells were then incubated with the substrate 2,2′-azinobis(3-ethylbenzothiazoline-6-sulfonic acid). The absorbance of the solution at 405 nm was determined by spectrophotometry.

Electron microscopy.

Vesicles in the culture supernatants of transfected cells were prepared for analysis as described previously (6). Briefly, 293T cells were transfected with expression vectors containing M or F cDNAs as described above and cultured for 2 days in Opti-MEM (Life Technologies, Rockville, Md.). The culture medium was then clarified by centrifugation at 15,000 × g for 10 min, and the supernatants were concentrated by centrifugation through a Centricon 100 filter (Millipore, Bedford, Mass.). The vesicles in the concentrated medium were adsorbed to carbon-coated grids, negatively stained with 1% aqueous uranyl acetate, and examined with a Phillips 301 electron microscope operated at 60 kV.

Analysis of proteins in density gradient fractions.

293T cells were infected with SV at an multiplicity of infection of 5 or were transfected with expression vectors containing M, F, or HN cDNAs and then were metabolically labeled, as described above. Culture supernatants (1 ml each) were centrifuged through 10 ml of a 5 to 40% sucrose gradient in PBS. Eight equal fractions were then taken from the top of each sucrose layer. An aliquot of each fraction was removed for density measurement by refractometry (Bausch & Lomb, Rochester, N.Y.). The remainder of each fraction was diluted with 3 ml of PBS and centrifuged at 190,000 × g for 2 h at 4°C. The pellets were resuspended in Laemmli reducing sample buffer and analyzed by SDS-PAGE.

Western blot detection of actin in vesicles.

After transfection with expression vectors encoding SV M or F, the cells were metabolically labeled with [³⁵S]Cys and [³⁵S]Met, and the proteins released into culture medium were fractionated by ultracentrifugation through a 5 to 40% sucrose gradient. Proteins in the eight fractions collected from each sample were subjected to electrophoresis and transferred to Immobilon-P membranes (Millipore). The membranes were then incubated with anti-actin MAb1501 (Chemicon, Temecula, Calif.). After reaction with goat anti-mouse immunoglobulin G conjugated to horseradish peroxidase (Bio-Rad), membranes were treated with SuperSignal West Pico chemiluminescent substrate (Pierce, Rockford, Ill.). After chemiluminescent signals dissipated, the membrane was exposed to Kodak BioMax X-ray film overnight to detect ³⁵S-labeled proteins.

RESULTS

Protein-induced release of virus-like particles into culture medium.

We previously reported that expression of hPIV1 M protein from cDNA induces the release of virus-like particles into culture medium, a finding that indicates that M protein plays a role in the budding of PIV (6). To further characterize the process of virus particle formation, we investigated the roles of the HN, F, and M proteins in the budding process of SV. We transfected cells with expression vectors encoding the M, F, or HN proteins and assessed their induction of virus-like particle formation. After transfection, the cells were metabolically labeled with [³⁵S]Cys and [³⁵S]Met, and the proteins released into culture medium were purified by ultracentrifugation through 30% glycerol and analyzed by SDS-PAGE. As shown in Fig. 1, SV M expressed from cDNA was detected in the culture medium, as hPIV1 M had been previously (6). SV F protein expressed from cDNA was also detected in the culture medium. Cells expressing SV HN, however, did not release the HN protein into the medium, although HN was expressed abundantly in the cells.

FIG. 1.

Release of viral proteins into culture medium. Cells were transfected with cDNAs encoding SV M, F, or HN and were labeled with [³⁵S]Met and [³⁵S]Cys for 16 h. (A) Protein expression in transfected cells. Cells were lysed in 1 ml of TNE buffer, and 100 μl of the clarified lysate was used for immunoprecipitation with specific MAbs. (B) Proteins released into culture medium were collected, purified through 30% glycerol in PBS, and analyzed by SDS-PAGE. The upper bands observed in the material of the M vesicles are charge-induced M aggregates (28). The arrows indicate cellular actin.

We next determined whether the F and M proteins released into the medium were included in the plasma membrane-derived vesicles or were protein aggregates. The culture media of cells expressing M or F were concentrated by membrane filtration, and the samples were examined by electron microscopy (Fig. 2). The medium of cells expressing M protein contained a number of virus-like particles that were 50 to 150 nm in diameter (Fig. 2A), as were those found in the medium of cells expressing hPIV1 M (6). The medium of cells expressing F protein contained round, virus-like particles bearing spikes at the surface (Fig. 2B and C). These particles were 30 to 100 nm in diameter, which is about one-third the standard size of SV. These results indicate that SV F protein and SV M protein can induce budding that releases virus-like particles into culture medium.

FIG. 2.

Electron micrographs of virus-like particles released into medium. Supernatants from cultures of cells transfected with SV M (A) or SV F (B and C) cDNAs were concentrated by membrane filtration. The samples were adsorbed to carbon-coated grids, negatively stained, and examined by electron microscopy. Bars, 86 nm.

Density gradient analysis of virus-like particles released into medium.

We further characterized the virus-like particles released from cells expressing M or F by centrifugation of the culture media on a 5 to 40% sucrose density gradient. Eight fractions were collected from each sample, and the proteins in each fraction were analyzed by SDS-PAGE. The proteins released from SV-infected cells were assayed for comparison. The structural proteins derived from whole SV virions were recovered mainly from the fractions whose densities were between 1.14 and 1.17 g/ml (Fig. 3, fractions 6 to 8). Fractions 4 and 5 contained the M and F proteins but very little of the other structural proteins (P, HN, NP, or L). This result suggests that the SV-infected cells released vesicles that contained only M, only F, or both M and F, in addition to whole virus particles.

FIG. 3.

Density gradient analysis of the vesicles released from cells expressing viral proteins. Cells infected with SV or transfected with cDNAs encoding M, F, or HN alone or in combination were labeled with [³⁵S]Met and [³⁵S]Cys. The clarified cell culture medium was centrifuged through a 5 to 40% sucrose gradient. Eight fractions were collected, and the proteins in each fraction were analyzed by SDS-PAGE.

In the culture supernatant of cells transfected with the SV M expression vector, most of the M protein was detected in the fractions whose densities were between 1.09 and 1.11 g/ml (Fig. 3, fractions 4 and 5). F-containing vesicles released from cells transfected with SV F cDNA were detected mainly in the fractions with densities of 1.11 to 1.14 g/ml (Fig. 3, fractions 5 and 6). These results and the electron microscopic findings indicate that the M or F proteins released into medium were incorporated into the lipid membrane.

Coexpression of M and F proteins significantly enhances the release of vesicles containing M and F.

We next compared the efficiency with which budding was induced by coexpression of M and F proteins. Released vesicles containing radiolabeled proteins were purified by centrifugation through 30% glycerol (Fig. 4A). Radiolabeled proteins in cell lysates were purified by immunoprecipitation (Fig. 4B). After SDS-PAGE analysis, the M and F proteins in each sample were quantified by using the PhosphorImager system, and the proportions of the synthesized proteins released into the medium were calculated. M protein was released from cells expressing M protein much more efficiently than F was released from F-expressing cells (52 versus 5%) (Fig. 4C). Interestingly, the release of F protein into culture medium was significantly enhanced in cells that coexpressed M and F proteins (Fig. 4A, lanes 3 and 5). Quantification of the proteins revealed that cells that coexpressed F and M released F protein with an efficiency about 4.2 times that of cells expressing F alone and released M with an efficiency about 1.4 times that of cells expressing M alone (Fig. 4C). Although SV HN was not released into medium when expressed alone, HN was detected in the media of cells that coexpressed HN and M (Fig. 4A, lane 6). HN was not released from cells that coexpressed F and HN (Fig. 4A, lane 7). However, because coexpression of F resulted in the downregulation of HN expression (Fig. 4B, lane 10), we cannot evaluate the role of coexpression of F in the budding of HN-containing vesicles.

FIG. 4.

Efficiency of protein release from transfected cells. Cells transfected with expression vectors containing M, F, or HN cDNAs alone or in combination were labeled with [³⁵S]Met and [³⁵S]Cys. (A) Vesicles released into the medium were purified and analyzed by SDS-PAGE. (B) Protein expression in cells was analyzed by immunoprecipitation and SDS-PAGE. (C) Proteins in cell lysates and in culture medium were quantified from SDS-PAGE gels (A and B) by using a STORM 860 imaging system, and the efficiency of protein release was calculated. Each value represents the mean of three independent experiments.

To determine whether the enhanced release of coexpressed viral membrane proteins might have been caused by their direct interaction, we analyzed the distribution of the coexpressed proteins in density fractions to investigate whether they had been included in the same vesicles or in separate vesicles (Fig. 3, bottom two panels). Most of the F-containing vesicles released from cells coexpressing F and M were recovered from fractions 5 and 6, the fractions that contained F vesicles released from cells that expressed only F. Similarly, the M-containing vesicles were recovered mainly from fractions 4 and 5, which were the same fractions that contained M released from cells expressing only M protein. Although some M- or F-containing vesicles overlap in fraction 5, the distribution patterns of M- or F-containing vesicles obtained from coexpressing cells were not different from the samples obtained from cells transfected with M or F cDNA separately. These results suggest that vesicles containing M and F were released separately from cells that coexpressed the two proteins, while we cannot exclude the possibility of the presence of the vesicles containing both M and F in fraction 5. The same result was obtained with cells expressing both HN and M proteins. The distribution of HN was similar to that of F protein and not to that of M protein (Fig. 3, bottom panel).

Cellular actin is involved in the formation of particles.

Purified virus-like particles induced by expression of M or F proteins included several proteins derived from transfected cells. One of the proteins (Fig. 1) migrated at the position of 43 kDa, which is similar to that of the cellular actin molecule. The SV virion is known to contain cellular actin (17), which is hypothesized to play a role in paramyxovirus replication and assembly (7, 8, 14). Therefore, we next investigated whether the virus-like particles containing M or F protein contained actin as well. Cells transfected with SV M or SV F expression vectors were labeled with [³⁵S]Cys and [³⁵S]Met, and the culture supernatant was fractionated by ultracentrifugation on a 5 to 40% sucrose gradient. Each fraction was assayed by Western blotting with an anti-actin MAb, and the membrane was exposed to film overnight to detect the radiolabeled M or F proteins. Cellular actin was detected in fractions 4, 5, and 6, a distribution that was correlated well with that of M protein (Fig. 5). The same result was observed in the supernatant of cells expressing F: most of the F protein and actin was found in fractions 5 and 6. These results indicate that actin was included in the vesicles that contained M or F and are consistent with the requirement of actin for the formation of the virus-like particles.

FIG. 5.

Cellular actin was present in vesicles that contained M or F. Cells expressing SV M or F protein were labeled with [³⁵S]Met and [³⁵S]Cys, and the clarified cell culture supernatants were centrifuged on 5 to 40% sucrose gradients. Eight density fractions were collected, and Western blots of proteins in each fraction were analyzed with an anti-actin MAb. After chemiluminescent signals dissipated, the membrane was exposed to X-ray film overnight to detect ³⁵S-labeled proteins.

The C-terminal region of SV M is required for particle formation.

To gain insight into the budding process induced by viral proteins, we next identified the domain of the M protein required for budding. SV M contains a C-terminal sequence similar to that of an actin-binding domain (KLKK motif), and the sequence is conserved in SV and hPIV1 (Fig. 6A). Because the KLKK motif of thymosin β4 was identified by mutational studies as an important residue in actin binding (39), we investigated whether the C-terminal sequence that contains the KIRK sequence is required for the budding induced by M protein. A mutant SV M(−5) cDNA that encodes M protein lacking the C-terminal five amino acids was constructed and expressed in cells as described above. The deletion of the C-terminal five amino acids did not affect the level of protein expression in the cells, but the release of the mutant M(−5) into culture medium was significantly less than that of SV M (Fig. 6B). A membrane flotation assay showed no difference between the two proteins in their membrane association (data not shown). It should be noted, however, that a membrane flotation assay provides association not only with the plasma membrane but with all the membranes in transfected cells.

FIG. 6.

C-terminal regions of SV and hPIV1 M proteins are required for the release of M-containing vesicles. (A) Sequences of the C-terminal region of SV and hPIV1 M proteins. (B and C) Release of the deletion mutant of SV M (B) or of a series of deletion mutants of hPIV1 M (C) into the culture medium was determined as described in the legend to Fig. 1. Expression of M proteins in the transfected cells was quantified by immunoprecipitation with an anti-M MAb.

The M protein of hPIV1 can induce the budding of virus-like particles (6), and it contains a C-terminal sequence like that of SV. To confirm the role of C-terminal residues in budding, we made a series of cDNAs encoding deletion mutants of hPIV1 M and assessed the release of the encoded M proteins into culture medium. Deletion of one or two C-terminal amino acids did not diminish the protein's release into medium. However, deletion of three to five C-terminal amino acids significantly reduced the release of protein (Fig. 6C). These results indicate that the C-terminal regions of SV and hPIV1 M proteins include residues essential to the budding of M-containing vesicles.

Amino acids in the cytoplasmic domain of SV F are required for particle formation.

To identify residues of the cytoplasmic domain of SV F protein that are necessary for particle formation, we made cDNAs that encoded SV F with deletions or site-directed mutations. Figure 7A shows the sequence identity of the cytoplasmic domains of SV and hPIV1 F proteins. Although the overall homology of these two virus F proteins is very high (68%) (20), amino acids in the cytoplasmic domain of F are not highly conserved in the two viruses. However, there are five consecutive amino acids (TYTLE) that are conserved between the two F proteins, suggesting that these residues play an essential role in the life cycles of the closely related viruses. hPIV3 and bovine PIV3 also share a five-residue sequence (PYVLT) in the cytoplasmic tail of F. Two amino acids (YxL) are conserved between the F proteins of all the respiroviruses (Fig. 7A). To determine the role of these residues in the production of F-containing vesicles, we made two deletion mutants, SVF546 and SVF541, that encode SV F proteins that lack 19 or 24 C-terminal residues, respectively. We also created a mutant SV F (SVFYLAA) in which alanine was substituted for Y543 and L545. SVF546, which contained the conserved sequence (TYTLE), was easily detected in the culture medium of cells transfected with the SVF546 cDNA, whereas SVF541 was barely detectable. The release of SVFYLAA into culture medium was also significantly reduced. These mutants were expressed at the cell surface at wild-type levels (Fig. 7C). These findings suggest that the five consecutive amino acids conserved in SV F and hPIV1 F are required for the production of F-containing vesicles.

FIG. 7.

Identification of SV F residues that affect budding of F-containing vesicles. (A) Sequence of the F protein cytoplasmic domain. SV and hPIV1 F share five consecutive amino acids (TYTLE). Sequences of the F mutants used are also shown. (B) Release of the SV F mutants into the medium was assessed as described in the legend to Fig. 1. (C) Surface expression of the mutant F proteins by transfected cells was quantified by cell-surface enzyme-linked immunosorbent assay with a cocktail of anti-F MAbs.

Results of deletion mutant analysis of M protein indicated that the C-terminal domain composed of basic-hydrophobic-basic-basic amino acids plays a role in the budding induced by M protein. The cytoplasmic domain of SV F contains a similar motif just beneath the membrane anchor region (Fig. 7A). To evaluate the role of this motif, we created a mutant in which alanine is substituted for the four residues, 524 to 527. We also created a mutant that has alanines at three residues (525 to 527) and left residue 524 unchanged, because a positively charged residue flanking the hydrophobic domain is important in maintaining the correct orientation of the glycoprotein in the membrane (25). The expression of these mutants at the cell surface was comparable to that of wild-type F (Fig. 7C). However, the efficiency with which the mutant proteins were released into medium was reduced to 13% (F525-7A) and 19% (F524-7A) of that of wild-type F (Fig. 7B). This finding, together with the results of studies of M deletion mutants, suggests that the domain composed of basic and hydrophobic residues is required for the efficient budding of virus-like particles induced by M and F proteins.

DISCUSSION

Viral M protein localizes at the inner surface of the plasma membrane of infected cell and plays a central role in viral assembly and budding. When expressed alone, M proteins of hPIV1, vesicular stomatitis virus, and influenza virus can induce the formation and release of M-containing particles into culture medium (6, 13, 16). When M is coexpressed with homologous NP, the vesicles contain both M and a nucleocapsid-like structure composed of NP (6). M protein is therefore considered to be the central organizer in virus assembly. Induction of the formation of virus-like particles by SV M protein, which is highly homologous to hPIV1 M (87% identity) (28), indicates that M drives the budding process in SV.

We found that SV F protein alone can also induce the budding and release of virus-like particles that carry F spikes. F protein appears to participate actively in the assembly of measles virus and SV particles. Most measles viruses isolated from the brains of patients with subacute sclerosing panencephalitis are defective in producing infectious virus and have defects in the M gene, the F gene, or both. In many viruses isolated from these patients, the cytoplasmic domain of F is shortened, elongated, or markedly altered (5, 31); these findings suggest that the cytoplasmic domain of F protein is required for the assembly of measles virus. Furthermore, in a study of recombinant SV carrying deletions in the cytoplasmic tail of F, a specific sequence in the cytoplasmic tail between residues 538 and 550 was shown to affect virus particle production (11). The TYTLE sequence (residues 542 to 546) in the F cytoplasmic domain, which we found to be required for the budding of F-containing vesicles, lies between residues 538 and 550. The virus particle production of their mutant SV from infected cells was about half as efficient as that of the wild type, which was not significant, considering our result with mutant F541 that showed no release of F vesicles (Fig. 7B). However, their mutant SV contained all the other structural proteins intact, including M. Therefore, it is not surprising that particle formation by the mutant SV was not significantly impaired because M can induce budding much more efficiently than F (Fig. 4). In our present study, we showed the importance of the TYTLE domain in the budding induced by F protein. This is also supported by the fact that the TYTLE sequence is conserved between SV and hPIV1 F proteins although the overall homology of the F cytoplasmic tails is very low (Fig. 7A).

No HN protein was released into the culture medium of cells expressing HN alone, despite the nearly equivalent expression of HN and F proteins. This result indicates that HN alone is unable to induce the budding of HN-containing vesicles. Coexpression of M with HN, however, resulted in the release of HN into medium; thus, coexpressed M protein may supply the machinery required to produce HN-containing vesicles. Because HN and M were found in fractions of different densities, it is clear that HN was not released because of direct physical interaction with M. Instead, it is likely that a function of M triggers the budding of vesicles containing HN.

Virions that contain no HN have been efficiently produced at a nonpermissive temperature from cells infected with a temperature-sensitive SV mutant (ts271) (27, 35), and recombinant SV that contains no HN gene has been recovered from cDNA (18), indicating that HN is not essential for the production of virions. However, characterization of recombinant SV or simian virus 5 with truncated HN cytoplasmic tails showed that the cytoplasmic tail sequence affects the budding efficiency of the mutant viruses (11, 32). In cells infected with recombinant simian virus 5 that expressed a truncated HN, viral proteins failed to accumulate at presumptive budding sites; therefore, the HN cytoplasmic tail may be involved in the formation of budding complexes required for the efficient budding of simian virus 5. In virus-infected cells, the cytoplasmic tail of HN may enhance the efficiency of budding by participating in the budding complex that includes M and F.

Direct interaction between SV M and F proteins has been reported. One group, who treated cells expressing SV M and F with detergent, reported that M protein interacts specifically with both the transmembrane domain and the cytoplasmic tail of F (1). They also observed specific colocalization of M with F at the plasma membrane by using confocal microscopy. Another group of investigators reported that coexpression of F and M reinforced the association of M with the membrane, a finding that suggests direct interaction between F and M (30); however, these results have not been confirmed (34). We have shown that when M was coexpressed with F, the release of vesicles containing M and F was strongly enhanced. However, sucrose gradient analysis suggested that the M- and F-containing vesicles were produced separately. These apparently contradictory findings may reflect a weak physical interaction between M and F in the absence of other viral proteins or an interaction between M and F that interferes with the machinery that drives the budding of vesicles that contain both proteins.

Interestingly, vesicles released from cells expressing SV M or F contained cellular actin in addition to these viral proteins. This finding supports the hypothesis that actin is involved in virus budding. Interaction between virus and actin is suggested by the presence of actin in highly purified virions. Paramyxoviruses were among the first viruses shown to contain actin (17, 23, 38). The electron microscopic observation of actin filaments in association with budding measles virus suggests the involvement of actin in paramyxovirus assembly (2). Cytochalasin B, an agent that disrupts actin microfilaments, completely inhibited the release of measles virions, a finding that suggests that actin is essential for virus budding (33). Furthermore, there is evidence of direct binding between actin and the M proteins of Newcastle disease virus and SV (12). Our findings with deletion mutants of SV and hPIV1 M indicate that C-terminal residues are required for the budding of M-containing vesicles. Interestingly, the residues we found to be required for budding are similar to the actin-binding domain (KLKK motif) identified in several F-actin-binding proteins, such as thymosin β4, villin, and vasodilator-stimulated phosphoprotein (37, 39). M protein contains the domain only at the C-terminus of the molecule. Remarkably, SV F contains a similar sequence in its cytoplasmic tail, and mutation of these residues affected the efficiency of budding. We do not have biochemical evidence to indicate whether M and F proteins directly bind to actin through these domains. However, our mutational analysis together with previous reports at least suggests the possible role of cellular actin in the process of virus budding. Further characterization of the interaction between actin and M and/or F proteins will be required to unveil the role of actin in the budding process and the mechanism of the final steps of virus assembly.