Association of Respiratory Syncytial Virus M Protein with Viral Nucleocapsids Is Mediated by the M2-1 Protein

Abstract

Cytoplasmic inclusions in respiratory syncytial virus-infected cells comprising viral nucleocapsid proteins (L, N, P, and M2-1) and the viral genome are sites of viral transcription. Although not believed to be necessary for transcription, the matrix (M) protein is also present in these inclusions, and we have previously shown that M inhibits viral transcription. In this study, we have investigated the mechanisms for the association of the M protein with cytoplasmic inclusions. Our data demonstrate for the first time that the M protein associates with cytoplasmic inclusions via an interaction with the M2-1 protein. The M protein colocalizes with M2-1 in the cytoplasm of cells expressing only the M and M2-1 proteins and directly interacts with M2-1 in a cell-free binding assay. Using a cotransfection system, we confirmed that the N and P proteins are sufficient to form cytoplasmic inclusions and that M2-1 localizes to these inclusions; additionally, we show that M associates with cytoplasmic inclusions only in the presence of the M2-1 protein. Using truncated mutants, we show that the N-terminal 110 amino acids of M mediate the interaction with M2-1 and the subsequent association with nucleocapsids. The interaction of M2-1 with M and, in particular, the N-terminal region of M may represent a target for novel antivirals that block the association of M with nucleocapsids, thereby inhibiting virus assembly.

Respiratory syncytial virus (RSV), a member of the family Paramyxoviridae, is one of the most important viral agents causing lower respiratory tract disease in infants, the elderly, and immunocompromised patients of all ages (3, 6, 21). The genome of RSV is a nonsegmented negative-strand RNA encoding 11 proteins (3). In RSV-infected cells, the viral RNA with the L (polymerase), N (nucleocapsid), P (phosphoprotein), and M2-1 proteins form the polymerase complex in which the transcription of messenger and genomic RNA takes place. As the cytoplasmic inclusions in RSV-infected cells have been shown to contain all elements of the polymerase complex and are capable of transcription in isolation (1, 7), it is presumed that they are major sites of viral transcription.

Several studies have shown that the N protein is the major driver for the formation of these cytoplasmic inclusions. N associates with viral RNA, and N-RNA complexes are resistant to RNase treatment (18). Inclusion-like structures are formed when the N and P proteins are coexpressed in cells (7), and this association results from a specific protein-protein interaction between N and P, which can be disrupted by mutagenesis (8, 24). Garcia et al. (7) also showed that the M2-1 protein is present in cytoplasmic inclusions; subsequent investigations confirmed that the association of the M2-1 protein with inclusions resulted from its association with P (16).

We previously reported that the RSV M protein is also found in cytoplasmic inclusions late during infection, in association with the N, P, and M2-1 proteins (11). Since we have also shown that the M protein inhibits virus transcription (11), the role of the M protein in cytoplasmic inclusions may be to inhibit viral transcription as a prelude to viral assembly and budding, driven by the M protein bringing cytoplasmic nucleocapsids into association with RSV envelope proteins (10). The concept is supported by data indicating specific interactions between M and the cytoplasmic domains of envelope glycoproteins (10).

To date, it is not known how M becomes associated with the nucleocapsid complex. In the current study, we demonstrate that the N terminus of M can bind directly to M2-1 in a cell-free assay and that M colocalizes with M2-1 in the cytoplasm of cells either infected with RSV or expressing only M and M2-1 proteins. Using a cotransfection system, it was demonstrated that M associates with inclusion-like structures formed by N and P only in the presence of M2-1.

MATERIALS AND METHODS

Cells and virus.

Human epithelial (HEp2) cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum (FCS) at 37°C and 5% CO₂. RSV subgroup A strain A2 (a gift from Paul Young, University of Queensland, Brisbane, Australia) was grown in HEp2 cells as previously described (9). To prepare the virus stock, an 80% confluent cell monolayer was infected with RSV at a multiplicity of infection of 1 for 1 h before additional maintenance medium (Dulbecco's modified Eagle's medium supplemented with 2% FCS) was added. The cells were incubated for 3 to 4 days until an extensive cytopathic effect was observed; at which time virus was harvested by one cycle of freezing and thawing, followed by centrifugation at 3,500 rpm for 15 min to remove cellular debris. Virus titers were determined using HEp2 cells by end-point dilution as previously described (9).

Antibodies.

Rabbit anti-M and guinea pig anti-M2-1 antisera were generated in our laboratory (11). The antisera were cleared of nonspecific binding elements by absorbing with HEp2 cell lysate before use. The mouse monoclonal antibodies directed against the M, N, P, and M2-1 proteins were gifts from Erling Norrby and Mariethe Ehnlund, Karolinska Institute, Sweden (19). Rabbit and mouse antibodies to the influenza hemagglutinin (HA) epitope tag (rabbit or mouse anti-HA) were purchased from Sigma. The mouse, rabbit, and guinea pig immunoglobulin G (IgG)-specific secondary antibodies conjugated with horseradish peroxidase and the mouse IgG-specific antibody conjugated with rhodamine were purchased from Chemicon. The guinea pig- and rabbit-specific antibodies conjugated with fluorescein isothiocyanate were purchased from Dako.

Plasmid constructs.

The construct used to express the RSV M protein in mammalian cells (pSD-M-HA) was described previously (10). Briefly, the M gene fused with an HA tag at the C terminus was inserted into vector pSD4.2. To generate the N-terminally (M-ΔN-HA)-deleted M mutant, a specific forward primer (AGAGCTCGGGGCAAATATGGCATGTAGTCTAACATGCC [the underlined sequence indicates a SacI site, and boldface type indicates the gene start signal of M]) and a reverse primer complementary to the HA epitope tag with a HindIII site (TAAGCTTAGGCGTAGTCGGGCAC) were used to amplify the desired region of M (114 to 256 amino acids [aa]) with the 9-aa HA epitope tag from pSD-M-HA used as a template. To generate the C-terminally-truncated M mutants [M-ΔC-HA and M-Δ(C+ZFD)-HA], a forward primer complementary to the vector sequence upstream of the M start site and specific reverse primers (TGGATCCTTAGGCGTAGTCGGGCACGTCGTAGGGGTATGTTACTATGTTTTCAAATTG and TGGATCCTTAGGCGTAGTCGGGCACGTCGTAGGGGTAACAGGGTGTG GTTACATCATATGC [the underlined sequence indicates the BamHI recognition sequence, and boldface type indicates the HA tag]) were used to amplify the desired sequence with a C-terminal HA epitope tag from pSD-M-HA used as a template.

Vector pSD4.2 was also used to make pSD-N-HA, pSD-P-HA, and pSD-M2-1-HA, which contain N, P, and M2-1 genes with a C-terminal HA tag. The relevant genes were amplified from full-length RSV cDNA using PCR (13); the reverse primer included the coding sequence for the HA epitope tag. The primers for N are CAAGAGCTCGGGGCAAATACAAAGATGGCTCTTAGC and AAGCTTAGGCGTAGTCGGGCACGTCGTAGGGGTAAAGCTCTACATCATTATCTTTTGG, primers for P are CAAGAGCTCGGGGCAAATAAATCATCATGGAAAAGTTTGC and AAGCTTAGGCGTAGTCGGGCACGTCGTAGGGGTAGAAATCTTCAAGTGATAGATCATTGTC, and the primers for M2-1 are CAAGAGCTCGGGGCAAATATGTCACGAAGGAATCC and AAGCTTAGGCGTAGTCGGGCACGTCGTAGGGGTAGGTAGTATCATTATTTTTGGCATGGTC (restriction sites are in italic type, and the HA sequence is underlined). The PCR products were blunt cloned into pSD4.2 using SacI restriction.

Cloning of the full-length M and M2 genes into vector pET30(a) and subsequent expression and purification from Escherichia coli were described previously (10).

The Semliki Forest virus (SFV) replicon system (14) was utilized for expressing the N (SFV-N), P (SFV-P), and M2-1 (SFV-M2-1) proteins in mammalian cells. The M2 gene was subcloned from the pET30(a) clone by excision with BamHI and cloning into vector JMPpSFV1 (17). The N and P genes were amplified by reverse transcription-PCR from total RNA extracted from RSV-infected cells. The N- and P-gene reverse primers (5′-CCGGGCCCGGGCCATGGAATTCAGGAGC-3′ and 5′-CCGGGCCCGGGGTTAGTTTGTTGG-3′ [restriction sites are underlined]) contained 5′ SmaI and ApaI restriction sites, and forward primers (5′-CCCATCGATGGGATCCCGCATAACTATACTCC-3′ and 5′-CCCATCGATCCGCAGAAGAACTAGAGGC-3′) contained 5′ ClaI restriction sites. These sites were used to clone the N and P genes into JMPpSFV-1. All clones were sequenced for authenticity.

RNA transcription and transfection.

All constructs prepared using pSD4.2 were linearized by SalI and constructs prepared using the SFV replicon system were linearized by SpeI. RNAs were transcribed using the MEGAscript transcription system (Ambion) according to the manufacturer's protocol. The quality and quantity of the RNA was analyzed by agarose gel electrophoresis, and the concentration was determined by optical density measurements. For transfection with a single RNA species, 0.8 μg RNA was used per well of a 24-well plate, and 5 μg RNA was used per well of a 6-well plate. In the case of transfection with multiple RNAs, 1.5 μg of total RNA was used per well of a 24-well plate. Lipofectamine 2000 transfection reagent (Invitrogen) was used for all transfections according to the manufacturer's protocol. Cells transfected with RNA transcribed from the empty SFV replicon were used as controls (mock). Except where indicated, the transfection reagent-RNA complexes were incubated with cells for 15 to 16 h at 37°C in 5% CO₂ prior to analysis.

Cell lysates.

Transfected cells were harvested using nondenaturing buffer (1% Triton X-100, 50 mM Tris [pH 7.4], 300 mM NaCl, 5 mM EDTA), and the relative concentrations of N-HA, P-HA, and M2-1-HA in cell lysates were determined by Western blotting; equal volumes of cell lysate were treated with denaturing sample buffer (50 mM Tris [pH 6.8], 2% sodium dodecyl sulfate [SDS], 5% 2-mercaptoethanol, 8% glycerol, 0.01% bromophenol blue) and resolved by 13% SDS-polyacrylamide gel electrophoresis (PAGE). The proteins were transferred onto a nitrocellulose membrane (Hybond-C Extra; Amersham) using a semidry transfer cell (Bio-Rad). Nonspecific binding sites were blocked by incubation with 5% nonfat dry milk-2% FCS in phosphate-buffered saline (PBS) at room temperature for 1 h, followed by overnight incubation at 4°C with rabbit anti-HA antibody (diluted 1:1,000 in blocking buffer), two washes with PBS plus 0.1% Tween 20, one wash with PBS, and incubation for 1 h at room temperature in species-specific antibodies conjugated to horseradish peroxidase. The membrane was washed, and bound antibodies were detected by chemiluminescence according to the manufacturer's (Amersham) instructions.

Immunofluorescence.

Subconfluent (80%) HEp2 cell monolayers grown on glass coverslips were infected with RSV or transfected with various mRNAs and cultured for the indicated times after infection or transfection. Cells were washed with ice-cold PBS and fixed with 4% paraformaldehyde for 10 min at room temperature, followed by the permeabilization of membranes with 0.2% Triton X-100 for 5 min. Fixed cells were washed thoroughly in PBS and incubated for 30 min in specific antibody (or a mix of antibodies) diluted 1:100 in bovine serum albumin (BSA)-PBS (1% BSA in PBS). Bound antibodies were detected with species-specific fluorochrome-conjugated secondary antibodies. Coverslips were mounted in fluorescent mounting medium (Dako) and analyzed by confocal laser scanning microscopy (CLSM) as described previously (15). Images of more than 20 cells were analyzed for each sample. We quantified the fraction of inclusions containing M using the formula (number of inclusions containing M)/(number of inclusions containing M2-1 or P); data from ≥20 cells are presented (means ± standard errors).

Cell-free translation of M mutants.

In vitro translation reactions were performed in a 50-μl rabbit reticulocyte lysate (RRL) (Promega) reaction mixture loaded with 1 μg of the respective M-HA, M-ΔN-HA, M-ΔC-HA, and M-(ΔC+ZFD)-HA in vitro-transcribed mRNA and incubated at 30°C for 90 min, and the quality and concentration of the product were determined by immunoblotting; 5 μl of each translated product was analyzed as described above.

Immunoprecipitation.

Equivalent amounts of in vitro-translated wild-type or mutant M-HA were incubated for 1 h at room temperature with lysates from HEp2 cells transfected to express M2-1. The mixture was then incubated with M2-1 antibody bound to protein A-Sepharose (Bio-Rad) for 2 h at 4°C with rotation. The beads were washed twice with wash buffer (0.1% Triton X-100, 50 mM Tris Cl [pH 7.4], 300 mM NaCl, and 5 mM EDTA), followed by two washes with ice-cold PBS and elution by boiling in elution buffer (1% SDS, 100 mM Tris Cl [pH 7.4], 10 mM dithiothreitol). Eluted proteins were analyzed for the presence of M-HA variants by Western blotting as described above.

Cell-free binding assay.

Binding of recombinant proteins expressed in bacteria to RSV proteins in transfected cell lysates was assayed as described previously (10). Briefly, recombinant M protein and the unrelated hepatitis C virus NS3 protein (used as a negative control) were expressed in bacteria and purified by affinity chromatography. One hundred nanograms per well of M or NS3 was coated onto microtiter plates in carbonate buffer. Nonspecific sites were blocked for 2 h with 1% BSA-PBS, followed by incubation for 1 h in the presence of 60 μg/ml of RNase A with lysates from mammalian cells expressing recombinant N-HA, P-HA, and M2-1-HA proteins. Bound M2-1, N, and P proteins were detected with rabbit anti-HA (diluted 1:800 in 0.1 mg/ml BSA-PBS containing 0.05% Tween 20), followed by horseradish peroxidase-conjugated anti-rabbit IgG antibody and detection using tetramethyl benzidine substrate.

RESULTS

RSV M protein associates with cytoplasmic inclusions in RSV-infected cells.

To investigate the association of the M protein with cytoplasmic inclusions in RSV-infected cells, we probed cells with various dual-antibody combinations and examined them by CLSM (Fig. 1). At 18 h postinfection, the N and P proteins were detected predominantly in cytoplasmic inclusions, with only minimal amounts present diffusely in the cytoplasm (Fig. 1A, B, and C, left). The M2-1 protein was found in virtually every cytoplasmic inclusion, but a considerable amount was also present diffusely throughout the cytoplasm (Fig. 1A, B, and D, middle). M2-1 colocalized with the N and P proteins in cytoplasmic inclusions but not in the cytoplasm outside of the inclusions (Fig. 1A and B, right). Interestingly, the M protein was present diffusely throughout the cytoplasm and in the nucleus as well as in some of the cytoplasmic inclusions (Fig. 1C, middle, and D, left) with M2-1, N, and P proteins as shown in Fig. 1C and D (right) (data for the N protein are not shown). These results indicate that N and P are present almost exclusively in cytoplasmic inclusions, while M2-1 and M are found both associated with inclusions and diffusely in the cytoplasm.

FIG. 1.

The M protein is present in cytoplasmic inclusions in RSV-infected cells. RSV-infected HEp2 cells were fixed 18 h after infection and were double stained with various antibody combinations, followed by CLSM analysis. The antibodies used are indicated above each panel. All primary antibodies were used at a dilution of 1/100 in PBS and incubated for 30 min; cells were washed with PBS and incubated for 30 min with species-specific secondary antibodies conjugated to fluorescein isothiocyanate or Texas Red. Cells were washed again, and localization was observed using CLSM. The left and middle columns of images are the red and green channel outputs of the same cell at the same optical plane; the right columns of images are computer-generated, merged images of the two channels, with yellow coloration indicating colocalization. A rabbit anti-M antiserum was used in C, and a mouse anti-M antibody was used in D.

Changes in the association of M and M2-1 with cytoplasmic inclusions over time were assayed by double immunofluorescence and CLSM analysis at 14 h, 16 h, and 20 h postinfection. An increase in the proportion of cytoplasmic inclusions containing M over time was noted (Fig. 2). Similarly, we found an increase in the M protein content of purified nucleocapsids relative to that of the N protein over time (data not shown). These findings confirm our previous data showing that the M protein associates with nucleocapsid complexes also containing the N, P, and M2-1 proteins in RSV-infected cells (11) and that the association of M with nucleocapsids increases over time.

FIG. 2.

Number of M-containing cytoplasmic inclusions increases over time. RSV-infected HEp2 cells were fixed at 14 h, 16 h, and 20 h postinfection and double stained with a mouse anti-M antibody and guinea pig anti-M2-1 antiserum, followed by CLSM, as described in legend of Fig. 1. Numbers on the right indicate proportions of inclusions containing M2-1 that also contain M [(number of inclusions containing M)/(number of inclusions containing M2-1)]; data from ≥20 cells are presented (means ± standard errors).

M protein can colocalize with M2-1 but not with N and P proteins in transfected cells.

To determine if M directly associates with the N, P, or M2-1 protein, M was coexpressed with the M2-1, N, or P protein in HEp2 cells. The expression and localization of each protein were examined 15 h after transfection by double immunofluorescence, followed by CLSM analysis (Fig. 3A). In the absence of other viral components, M colocalized with the M2-1 protein in punctate regions within the cytoplasm (Fig. 3A, top) but not with the N or P protein (Fig. 3A, middle and bottom). Colocalization of the intracellular M with the M2-1 protein suggested a direct interaction between the two proteins. Interestingly, the association of M with M2-1 appeared to correlate with decreased nuclear localization of M (Fig. 3A, compare left image in top row with middle images in middle and bottom rows).

FIG. 3.

M interacts directly with the M2-1 protein. (A) HA epitope-tagged M mRNA was transfected into HEp2 cells along with SFV-M2-1 (top), SFV-N (middle), or SFV-P mRNAs (bottom). The cells were fixed 15 h after transfection, followed by double staining with the indicated antibodies and CLSM analysis. (B) Five microliters of cell lysate from nontransfected cells and cells transfected to express M2-1-HA, P-HA, and N-HA was resolved by 12% SDS-PAGE, followed by Western blotting using rabbit anti-HA antibody. (C and D) Equivalent amounts of M2-1-HA, P-HA, and P-HA (as estimated from the Western blot) and a similar volume of untransfected cell lysate were diluted and incubated with bacterially expressed M (C) or hepatitis C virus NS3 (D) immobilized on microtiter plate wells. Bound protein was detected with rabbit anti-HA antibody followed by horseradish peroxidase-conjugated secondary antibody and development with tetramethyl benzidine substrate.

M protein interacts with M2-1 but not with N or P proteins in a cell-free system.

To confirm a direct interaction between the M and M2-1 proteins, a cell-free binding assay was used. Cell lysates from cells expressing equivalent levels of N-HA, P-HA, or M2-1-HA (as determined by Western blotting) (Fig. 3B) were added to purified recombinant M or hepatitis C virus NS3, and bound RSV proteins were detected with antibody to the HA tag. The M protein bound M2-1-HA in a dose-dependent manner but did not bind to either N-HA or P-HA (Fig. 3C). None of the RSV proteins bound to hepatitis C virus NS3 (Fig. 3D).

Taken together, the data presented in Fig. 1 to 3 strongly suggest that M colocalizes with cytoplasmic inclusions in RSV-infected cells via a direct association with M2-1.

Association of the M protein with the cytoplasmic inclusions requires the M2-1 protein.

The direct interaction of M with M2-1, but not with N or P, suggests that the association of the M protein with cytoplasmic inclusions may be mediated by an M-M2-1 interaction. To test this, we performed a series of transfections and cotransfections to express the relevant RSV proteins intracellularly either individually or in various combinations. As reported previously (7), when N, P, or M2-1 was expressed alone, each one was distributed diffusely in the cytoplasm; the coexpression of the N and P proteins formed cytoplasmic inclusion-like structures, although they tended to be smaller than those seen in RSV-infected cells (data not shown).

We next examined the association of the M2-1 and M proteins with the inclusion-like structures formed by the coexpression of N and P (Fig. 4). As previously reported (7), M2-1 colocalized with N and P in the inclusion-like structures (Fig. 4, top). In contrast, the M protein failed to associate with the inclusion-like structures formed by N and P (Fig. 4, second row) but did colocalize with inclusion-like structures containing N, P, and M2-1 (Fig. 4, third row). The results provide further evidence supporting an interaction between M and M2-1 that mediates the recruitment of the M protein to nucleocapsid complexes.

FIG. 4.

M associates with inclusions only in the presence of M2-1. N and P mRNAs were cotransfected into HEp2 cells with M2-1 (top row), M-HA (second row), or M2-1 and M-HA mRNAs (third row). Control (mock) cells were transfected with RNA from an empty SFV replicon (bottom row). Cells were fixed at 15 h after transfection and double stained with the indicated antibodies, and fluorescence was visualized using CLSM.

The N terminus of the M protein is essential for its interaction with M2-1.

To investigate the M protein domain interacting with M2-1, we generated one construct with an N-terminal deletion (M-ΔN-HA) and two constructs with C-terminal deletions [M-ΔC-HA and M-Δ(C+ZFD)-HA] (Fig. 5A), all with a C-terminal HA epitope. Like M, all three mutants were distributed in the nucleus as well as the cytoplasm at 15 h posttransfection (Fig. 5A), but the M-ΔC-HA protein was generally expressed at lower levels and had a cytoplasmic distribution in some cells. The association of M mutants with inclusion-like structures was examined by double-label-immunofluorescence CLSM of HEp2 cells in which N, P, and M2-1 were coexpressed. M-ΔC-HA colocalized with the P protein in the inclusion-like structures in a manner similar to that of full-length M (the fractions of inclusions containing P that also contained M [#M/#P] were 0.88 and 0.86, respectively), while M-ΔN-HA had a considerably reduced association (#M/#P = 0.37). Further deletion of the zinc finger domain (ZFD) from the C terminus [M-Δ(C+ZFD)-HA] led to a reduced association (mean #M/#P = 0.54) but still higher than that observed with M-ΔN-HA. In cells transfected to coexpress M2-1 with the M deletion mutants, M-ΔC-HA and M-Δ(C+ZFD)-HA, but not M-ΔN-HA, colocalized with M2-1 (data not shown).

FIG. 5.

The N terminus of M appears to mediate associations with inclusions and binding to M2-1. (A) Schematic diagram showing constructs of M used. The sequence motifs are based on predictive software analysis of the amino acid sequence and our previous work. NES, nuclear export signal; NLS, nuclear localization signal; LRR, leucine-rich region. Numbers indicate amino acids within the M sequence; each construct has a C-terminal HA tag. Broken lines in the mutants indicate the actual boundaries of the ZFD in the full-length protein. Full-length M and all three mutants were localized throughout the cell, as shown by the respective CLSM images on the right. (B) SFV-N, SFV-P, and SFV-M2-1 mRNAs were cotransfected into HEp2 cells with M-HA, M-ΔN-HA, M-ΔC-HA, or M-Δ(C+ZFD)-HA mRNA, and cells were fixed 15 h after transfection, followed by immunofluorescence assays using anti-HA and anti-P antibodies, as indicated, and CLSM analysis. Arrowheads point to cytoplasmic inclusions that have M (or its mutant). Numbers on the right indicate the proportion of inclusions containing P that also contain M [(number of inclusions containing M)/(number of inclusions containing P)] (means ± standard errors). (C) M-HA, M-ΔN-HA, M-ΔC-HA, M-Δ(C+ZFD)-HA, and luciferase were translated in RRL from in vitro-transcribed mRNAs. Five microliters of each translation reaction mixture was resolved by 12% SDS-PAGE, followed by immunoblotting using rabbit anti-HA antibody. (D) Equivalent amounts of translated M-HA, M-ΔN-HA, M-ΔC-HA, M-Δ(C+ZFD)-HA, and luciferase were incubated with same amounts of lysates of cells expressing M2-1. Another control consisted of M-HA incubated with untransfected (mock) cell lysate. The mixtures were immunoprecipitated using anti-M2-1 antibody, and bound M-HA (or mutants) in the immunoprecipitate was eluted and detected by Western blotting using rabbit anti-HA antibody. (E) Densities of specific M-HA (or mutants) bands from Western blots like those shown in C and D were measured using a Molecular Imager system (Bio-Rad). The percentage of total protein recovered is shown. The data represent means ± standard deviations of data from three independent experiments.

The reduced association of M-ΔN-HA and M-Δ(C+ZFD)-HA with the inclusion-like structures was investigated by a cell-free binding assay developed using M-HA, M-ΔN-HA, M-ΔC-HA, and M-Δ(C+ZFD)-HA translated in a cell-free system with RRL. The quality and concentration of M and the mutants in each reaction were estimated by immunoblotting using an anti-HA antibody (Fig. 5C). An equivalent amount of M-HA, M-ΔN-HA, M-ΔC-HA, or M-Δ(C+ZFD)-HA was incubated with a constant amount of cell lysate from HEp2 cells transfected to express M2-1. M2-1 with any bound protein was collected by immunoprecipitation with an M2-1 antibody. Bound M or its mutants in immunoprecipitation products were detected by immunoblotting using the HA antibody (Fig. 5D). Densitometry of the immunoblot revealed that 20 to 25% of total M-HA, M-ΔC-HA, and M-Δ(C+ZFD)-HA was recovered in the immunoprecipitation product, while only 5% of M-ΔN-HA and less than 3% of luciferase was recovered (Fig. 5E). No M-HA was recovered when M2-1-containing lysate was substituted with cell lysate from nontransfected cells (Fig. 5E). These results show that the N terminus of M, and the first 110 aa in particular, mediates M's binding to M2-1.

DISCUSSION

In this study, we examined the basis of the RSV M association with viral nucleocapsids, demonstrating for the first time that the M protein associates with nucleocapsids through the M2-1 protein, with no requirement for RNA or other cellular proteins; we show further that the interaction with M2-1 is strongly dependent on the N-terminal 110 aa of M.

The paramyxovirus M proteins play a major role in virus assembly, acting by concentrating viral envelope glycoproteins and nucleocapsids at the site of virus assembly (20). In recombinant paramyxoviruses where M is either mutated or deleted, colocalization of envelope glycoproteins and nucleocapsids is lost, with a consequent reduction in virus release (2). Interactions of M with other viral proteins are not well defined and appear to vary between different paramyxovirus species. A direct, virus-specific interaction of the M protein with nucleocapsids has been described for parainfluenza virus type 1 and Sendai virus; M interacts with the NP protein of homologous but not heterologous virus (5). In parainfluenza virus type 1, M alone is sufficient to form virus-like particles, while in simian virus 5, the nucleocapsid and at least one envelope glycoprotein are required, along with M, to form virus-like particles (23).

Previous work with RSV has shown that M is an essential requirement for packaging and passaging of minigenomes, along with F, N, and P (26). Our previous work has further defined the roles of RSV M in virus assembly; M interacts with the G protein, associates with nucleocapsids, and inhibits viral transcriptase activity (10, 11). Here, we extend our previous findings to show that the N terminus of M is involved in the interaction with M2-1; the latter would appear to function as an adaptor to facilitate the association of M with the nucleocapsid complex and the subsequent inhibition of viral transcriptase activity.

In cells transfected to express various combinations of the N, P, M, and M2-1 proteins, we found that N and P were sufficient to form inclusion-like structures and that the M2-1 protein localized to the inclusions thus formed, as previously shown (7). M did not localize to inclusions formed by N and P except in the presence of the M2-1 protein. Our cell-free binding data show that M and M2-1 can bind directly to one another without the requirement for other viral components. These data are consistent with the notion that the association of M with nucleocapsids is linked to the presence of M2-1. The added observation that the amount of M bound to nucleocapsids relative to M2-1 increases with time is also consistent with the possibility that M2-1 is acting as an “adaptor”.

The finding that M and M2-1, even though present throughout the cytoplasm in infected cells, associate in the cytoplasm is supported by our transfection data showing colocalization when expressed together in the absence of other viral proteins. This is in contrast to N and M, which do not colocalize when expressed together, even though both are present diffusely in the cytoplasm. Importantly, data from our cell-free binding assay confirm that M can associate directly with M2-1 but not with N or P.

The finding that M-M2-1 binding was not detectable using coimmunoprecipitation approaches in transfected cells may be attributable to one or more factors: the transfection efficiency may be low in cells transfected with multiple RNA species (16); the interaction of M with M2-1 may be transient, serving only to bring M into nucleocapsids, followed by other more lasting interactions; or the conditions used during the immunoprecipitation may have led to the disruption of the interaction.

To determine the domain within M protein that interacts with M2-1, we initially made two truncations of equivalent sizes at the N (M-ΔN-HA) and C (M-ΔC-HA) termini. M-ΔC-HA and M-ΔN-HA share an overlapping sequence, which is predicted to be a ZFD (10) and contains a putative RNA-binding motif (22). As demonstrated by cotransfections and cell-free binding assays, the ability to interact with M2-1 and associate with cytoplasmic inclusions was retained by M-ΔC-HA but markedly reduced in M-ΔN-HA, suggesting that the N terminus of the M protein, but not the ZFD or RNA binding domain, plays an important role in the association with the M2-1 protein and inclusions. This finding was confirmed by another C-terminal deletion, M-Δ(C+ZFD), which lacks the ZFD and RNA binding domain but bound to the M2-1 protein in a manner similar to that of the full-length M protein in a cell-free system; M-Δ(C+ZFD) had a reduced association with nucleocapsids in a cotransfection system. Taken together, the data suggest that the ZFD is not necessary for M and M2-1 binding but may contribute to nucleocapsid association in infected cells. As described previously (20), M (full length or mutant) formed aggregates in the cytoplasm. M has a propensity to form homo-oligomers that may serve to regulate M's diverse assembly functions (20). The expression of M protein variants lacking the C-terminal nuclear localization signal [M-ΔC, M-Δ(C+ZFD)] still resulted in abundant intranuclear M protein, probably due to free diffusion across the nuclear membrane resulting from the small size of the mutants [12.5 kDa for M-Δ(C+ZFD)]. Interestingly, the overall expression level of the M mutants appeared to be higher in singly transfected cells than in cells coexpressing N, P, and M2-1; this was especially true of the M-ΔC and M-Δ(C+ZFD) mutants. This seeming difference in expression may simply be the result of the change in the distribution of the M proteins from being diffused all over cytoplasm and nucleus to being localized in inclusions or a direct consequence of its interactions with nucleocapsids.

M2-1 functions as a transcriptional processivity factor, preventing premature termination during transcription, thus enhancing transcriptional readthrough at gene junctions and permitting access of the RSV polymerase to downstream transcriptional units (4, 12, 25). In the RSV replication cycle, nucleocapsid complexes released after infection first initiate transcription to produce individual mRNAs for viral protein synthesis. The newly synthesized M2-1 protein associates with nucleocapsids through its interaction with P (16) to prevent the termination of transcription and promoting readthrough at gene junctions. Our previous and current data taken together suggest that M associates with nucleocapsids through its interaction with M2-1 to shut down virus transcriptase activity, presumably to initiate assembly and budding by interacting with envelope glycoproteins (10, 11; R. Ghildyal et al., unpublished data).

Although RSV infections are prime candidates for early childhood immunization and antiviral drug therapy, to date, the considerable efforts to develop these prevention and treatment modalities have been unsuccessful. The results here indicate that M2-1 plays a key role in localizing M to the nucleocapsids; conceivably, the interaction between M2-1 and M (or rather, its N-terminal region) may represent a target for the development of antivirals to inhibit M association with nucleocapsids and thereby virus assembly, with a consequent reduction in disease severity. This interaction could also be used to develop attenuating mutations suitable for candidate vaccines. The current focus of this laboratory is to define the key sequences mediating the M2-1-M interaction as a prelude to the design of inhibitors of this interaction that may prove to be useful antivirals in the future.