The Influenza Virus M2 Protein Cytoplasmic Tail Interacts with the M1 Protein and Influences Virus Assembly at the Site of Virus Budding

Abstract

The cytoplasmic tail of the influenza A virus M2 proton-selective ion channel has been shown to be important for virus replication. Previous analysis of M2 cytoplasmic tail truncation mutants demonstrated a defect in incorporation of viral RNA (vRNA) into virions, suggesting a role for M2 in the recruitment of M1-vRNA complexes. To further characterize the effect of the M2 cytoplasmic tail mutations on virus assembly and budding, we constructed a series of alanine substitution mutants of M2 with mutations in the cytoplasmic tail, from residues 71 to 97. Mutant proteins M2-Mut1 and M2-Mut2, with mutations of residues 71 to 73 and 74 to 76, respectively, appeared to have the greatest effect on virus-like particle and virus budding, showing a defect in M1 incorporation. Mutant viruses containing M2-Mut1 and M2-Mut2 failed to replicate in multistep growth analyses on wild-type (wt) MDCK cells and were able to form plaques only on MDCK cells stably expressing wt M2 protein. Compared to wt M2 protein, M2-Mut1 and M2-Mut2 were unable to efficiently coimmunoprecipitate with M1. Furthermore, statistical analysis of planar sheets of membrane from cells infected by virus containing M2-Mut1 revealed a reduction in M1-hemagglutinin (HA) and M2-HA clustering as well as a severe loss of clustering between M1 and M2. These results suggest an essential, direct interaction between the cytoplasmic tail of M2 and M1 that promotes the recruitment of the internal viral proteins and vRNA to the plasma membrane for efficient virus assembly to occur.

The assembly and budding of influenza A virus is a complex, multistep process involving the proper organization of viral proteins at the plasma membrane, incorporation of a segmented genome, and formation of viral particles by pinching off from the cell surface (reviewed in reference 51). The segmented genome of influenza virus is made up of eight single-stranded, negative-sense RNA segments that are wrapped around the outside of the nucleoprotein (NP) subunits to form viral ribonucleoprotein (vRNP) complexes. The matrix protein (M1) is the major internal structural protein and underlies the viral envelope, and it is thought to form contacts between the vRNPs and the cytoplasmic tails of the viral integral membrane proteins. The spike glycoproteins hemagglutinin (HA) and neuraminidase (NA) are the predominant viral surface proteins and are the major antigenic determinants of influenza virus (reviewed in reference 27). The M2 protein is translated from a spliced mRNA derived from genome segment 7 (28); the protein is abundantly expressed on the cell surface, and small amounts of M2 are incorporated into virions (∼15 tetramers per virion) (19, 29, 54, 62, 63). The M2 protein has a proton-selective ion channel activity (13, 36, 43). Although only a low level of expression of M2 is necessary for virus replication (56), the ion channel activity is required for acidification of the viral particle within the late endosome during virus entry. The influx of H⁺ ions disrupts interactions between the vRNPs and M1, allowing the RNPs to be released into the cytoplasm devoid of M1 protein after fusion of the virus and cellular membrane has occurred (reviewed in references 17, 26, and 53).

During virus assembly, there is evidence that each of the viral surface integral membrane proteins participates in forming a complete viral particle. Data obtained using mutant HA and NA protein-containing viruses generated by reverse genetics indicated a role for the HA and NA cytoplasmic tails in controlling virus morphology (25), virus assembly (66), and genome packaging (65). The accumulation of HA and NA into lipid microdomains on the plasma membrane is an intrinsic property of each protein that is required to facilitate efficient virus budding and replication (4, 18, 30, 49, 55). Furthermore, biochemical evidence supports a model in which HA and NA recruit M1 to lipid microdomains (1, 15, 66), presumably through specific features of the cytoplasmic domains of HA (e.g., palmitoylation sites [12]) and NA (e.g., a critical proline residue [5, 35]). The importance of HA and NA in virus assembly was also observed in a biologically relevant influenza virus-like particle (influenza VLP) system (11).

Whereas the M2 ion channel activity is essential for influenza virus replication primarily during virus entry, structural elements of the M2 protein appear to be important at other stages of the virus replicative cycle. Several studies have now shown that, although posttranslational modifications such as phosphorylation (20, 57) and palmitoylation (8, 20) of the M2 cytoplasmic tail are not required for virus replication, specific regions of the M2 cytoplasmic tail are important for and presumably have a role in virus assembly. Initial electrophysiology studies of Xenopus oocytes suggested the M2 cytoplasmic tail was necessary to stabilize the ion channel pore against premature closure (58). Examination of M2 cytoplasmic tail truncation-containing mutant viruses that retained ion channel activity showed a defect in virus assembly, particularly a defect in genome packaging (34). Similarly, the cytoplasmic domain of the influenza B virus BM2 ion channel protein (37) is required for vRNP incorporation into influenza B virions during virus assembly (22, 23). Further studies of the influenza A virus M2 cytoplasmic tail have reinforced the notion that the cytoplasmic tail of M2 plays a role in virus assembly (24, 33), presumably through contacts with M1 and vRNP complexes.

A direct interaction between M1 and the M2 cytoplasmic tail has been suggested by studies examining antibodies against M2 and their ability to restrict virus growth (62). Under selective pressure from the monoclonal antibody (MAb) 14C2, which recognizes an epitope in the ectodomain of M2 (62), mutant escape variants genetically linked to segment 7 were recovered. Although the M2 proteins produced by the mutant viruses retained reactivity to MAb 14C2, nucleotide sequencing identified amino acid mutations in both the M2 cytoplasmic tail and M1 (61), suggesting that changes in the cytoplasmic compartment can overcome the extracellular effect of MAb 14C2 on virus budding. Mutations in the M2 cytoplasmic tail can also affect virus morphology (24, 47), similar to the case for observed filamentous and spherical morphologies that are dependent on residue changes in M1 (6, 7, 14). Furthermore, viruses generated in the presence of M2-M1 fusion proteins, as well as glutathione S-transferase (GST) pull-down assays, have suggested that an interaction between the M1 and M2 proteins occurs during a virus infection (33).

In this study, our aim was to further characterize the effects of mutations in the M2 cytoplasmic tail on virus assembly. We generated a series of M2 mutants by introducing alanine substitutions along the M2 cytoplasmic tail from residues 71 to 97 and analyzed VLPs and viruses containing these mutant M2 proteins. We also performed a statistical analysis of the two-dimensional spatial distribution of viral proteins in the membrane of virus-infected cells. Our data from these experiments using the human influenza A/Udorn/72 virus (H3N2; Ud) genetic background provide evidence for an interaction between M2 and M1 that is crucial for complete virus assembly at sites of virus budding.

MATERIALS AND METHODS

Cells and viruses.

Madin-Darby canine kidney (MDCK) cells and 293T cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS). M2-MDCK cells (i.e., stably expressing the wild-type [wt] influenza virus Ud M2 protein) were maintained in DMEM supplemented with 10% FBS, 200 μg/ml Geneticin (G418; InvivoGen, San Diego, CA), and 2 μM amantadine (Sigma-Aldrich [Sigma], St. Louis, MO). All cells were maintained in a humidified incubator at 37°C with 5% CO₂.

wt and mutant Ud influenza A viruses were generated by reverse genetics from cDNAs essentially as described previously (39). Briefly, 293T cells were transfected with eight genome-sense plasmids along with four protein-expressing plasmids, encoding PB1, PB2, PA, and NP. Approximately 16 to 18 h posttransfection (p.t.), the 293T cells were dispersed and cocultured with M2-MDCK cells in DMEM supplemented with 10% FBS. After the cells had formed a monolayer, the medium was replaced with DMEM supplemented with 3.0 μg/ml N-acetyl trypsin (NAT; Sigma). Virus stocks were propagated in M2-MDCK cells, and virus titers were determined by plaque assay on M2-MDCK cells. Mutant viruses are designated by the origin and number of wt gene segments, followed by the origin and gene mutant. For example, Ud7+UdM1/UdM2-Mut1 virus contains the wt Ud PB1, PB2, PA, NP, NS, HA, and NA genes and the Ud M gene with a mutation in the M2 coding sequence. Viral RNA was extracted from virus stocks by using a QIAamp viral RNA kit (Qiagen, Valencia, CA), transcribed into DNA with AMV Super reverse transcriptase (Molecular Genetic Resources, Tampa, FL), and amplified with AmpliTaq DNA polymerase (Applied Biosystems, Foster City, CA). The complete nucleotide sequence of the M gene from each mutant virus was determined using a 3100-Avant genetic analyzer (Applied Biosystems).

Plasmids.

The eight genome-sense (pHH21) plasmids and four protein-expressing (pcDNA3.1) plasmids used to generate Ud influenza virus by reverse genetics have been described previously (16, 39, 55, 56). The eukaryotic expression plasmid pCAGGS (40) was used to transiently express viral proteins in 293T cells. Plasmids used for the VLP system have been described previously (11). Overlapping PCR or QuikChange (Stratagene, La Jolla, CA) was used to introduce site-specific mutations into the M gene (pHH21) or M1 and M2 protein expression constructs (pCAGGS). 293T cells were transfected by using TransIT-LT1 (Mirus, Madison, WI), for reverse genetics experiments, or by using Lipofectamine and Plus reagents (Invitrogen, Carlsbad, CA), for protein expression experiments, according to manufacturer protocols.

Antibodies.

Goat serum raised to purified, detergent-disrupted Ud virus (goat anti-Ud) was used to detect HA, NP, and M1 by immunoblotting. MAb 14C2 specific for M2 (62) was used to detect M2 by immunoprecipitation, immunoblotting, and electron microscopy. HA was detected by using goat anti-influenza virus HA H3 (A/Equine/Miami/1/63; American Type Culture Center [ATCC]), and M1 was stained by using goat anti-matrix protein [A/NWS/34(HO)-Equine/Prague/1/56; ATCC] or MAb JZc8. For immunoelectron microscopy, antibody binding was detected by using donkey immunoglobulins (Igs) specific for either goat or mouse IgG conjugated to 6-nm or 12-nm gold particles (Jackson Immunoresearch Laboratories, West Grove, PA).

Pulse-chase and surface expression analyses of mutant M2 proteins.

293T cells (2 × 10⁵ cells) were grown in 35-mm dishes and transiently transfected in duplicate with wt or mutant M2 expression plasmids. For pulse-chase analysis, at 20 h p.t., the cells were incubated in DMEM deficient in methionine and cysteine for 30 min and then metabolically labeled with 50 μCi l-[³⁵S]methionine-cysteine-containing Redivue Promix in vitro cell-labeling mix (Amersham Biosciences, Piscataway, NJ) in DMEM deficient in methionine and cysteine for 30 min. The cells from one set of samples was then lysed in radioimmunoprecipitation assay buffer (1% deoxycholic acid, 1% Triton X-100 [TX-100], 0.1% sodium dodecyl sulfate [SDS], 10 mM Tris, pH 7.4, 150 mM NaCl) with 1 mM phenylmethanylsulfonyl fluoride (Sigma) and 1× protease inhibitor cocktail (Sigma). The other set of cells was incubated in complete DMEM with 10% FBS for 2 h and then lysed in radioimmunoprecipitation assay buffer. Samples were sonicated, centrifuged at 14,000 × g for 10 min, and immunoprecipitated with M2-specific MAb 14C2. Immune complexes were pulled down with a protein A/G-Sepharose mixture (3:1; Amersham Biosciences) and processed as described previously (12). Radiolabeled proteins were separated by SDS-polyacrylamide gel electrophoresis (PAGE) on a 17.5% polyacrylamide gel with 4 M urea and detected on a Fujifilm FLA-5100 system and analyzed by using MultiGauge version 3.0 software (Fuji Medical Systems, Stanford, CT).

For surface expression analysis, at 20 h p.t., cells were washed with phosphate-buffered saline (PBS), stained with MAb 14C2 followed by fluorescein isothiocyanate-conjugated goat anti-mouse IgG secondary antibody (Jackson Immunoresearch Laboratories), fixed in 0.5% formaldehyde (Sigma), and analyzed by flow cytometry by using a FACSCalibur flow cytometer (Becton Dickinson, Franklin Lakes, NJ).

Analysis of VLP budding and infectivity.

VLPs were generated and analyzed for infectivity as described previously (11). Briefly, 293T cells (1 × 10⁶ cells) grown in 6-cm dishes were transfected with protein expression plasmids expressing PB2, PB1, PA, M1, HA, NA, NP, M2, and NS2 (also known as NEP). An artificial pseudogene reporter construct encoding green fluorescent protein (GFP) in the negative sense, flanked by RNA segment 7 noncoding regions (vGFP), was also included. At 48 h p.t., the culture medium was harvested and cellular debris was pelleted by centrifugation at 2,000 × g for 10 min. The culture medium was then layered onto a 30% sucrose-NTE (100 mM NaCl, 10 mM Tris, pH 7.4, 1 mM EDTA) (wt/vol) cushion and centrifuged at 200,000 × g for 2 h at 4°C in a 70.1 Ti rotor (Beckman Coulter, Fullerton, CA). The pellet was resuspended in SDS loading buffer (50 mM Tris, pH 6.8, 100 mM dithiothreitol, 2% SDS, 0.1% bromophenol blue, 10% glycerol), boiled for 5 min, and analyzed by SDS-PAGE on a 15% polyacrylamide gel. To prepare the cell lysate, transfected cells were lysed in SDS loading buffer, sonicated for ∼20 s, boiled for 5 min, and analyzed by SDS-PAGE on a 15% polyacrylamide gel. Polypeptides were detected by standard immunoblotting techniques and quantified with an Odyssey infrared imaging system (Li-Cor Biosciences, Lincoln, NE).

To test VLP infectivity, VLPs were prepared by transfecting 293T cells as described above. After the culture medium was harvested, VLPs were treated with 5 μg/ml NAT for 15 min at 37°C to cleave and activate HA. Soybean trypsin inhibitor (0.1 mg/ml; Sigma) was then added to inhibit trypsin activity. 293T cells (5 × 10⁵ cells) grown in 6-cm-diameter dishes were transfected with pcDNA-PB1, pcDNA-PB2, pcDNA-PA, and pCAGGS-NP. At 6 h p.t., the transfection mixture was removed from the target cells and replaced with trypsin-treated VLPs. Twenty hours after VLPs were added, the cells were washed with PBS, scraped into PBS with 50 mM EDTA, and fixed in 0.5% formaldehyde. GFP fluorescence in fixed cells was quantified by using a FACSCalibur flow cytometer.

Virus growth, plaque assay, and budding analysis.

Confluent MDCK cells were inoculated with virus at a multiplicity of infection (MOI) of 0.001 in DMEM containing 1% bovine serum albumin (BSA) for 1 h at 37°C. Unbound virus was removed by washing the cells with PBS, and replacement medium was serum-free DMEM supplemented with 2 μg/ml NAT. At 12-h time points, an aliquot of medium was removed from each sample and replaced with an equal volume of DMEM-NAT. The virus titer at each time point was determined by plaque assay on M2-MDCK cells.

For plaque assays, confluent MDCK or M2-MDCK cells in six-well plates were inoculated with appropriate dilutions of virus in DMEM containing 1% BSA for 1 h at 37°C. Unbound virus was removed by washing the cells with PBS. Cells were then overlaid with DMEM-1% agarose supplemented with 1.5 μg/ml NAT and incubated at 37°C for 3 days. Plaques were visualized by immunostaining. Briefly, cells were fixed in 1% glutaraldehyde (Sigma) in PBS for 2 h and then incubated in blocking solution (3% egg albumin [Sigma] in PBS) for 30 min. Cells were incubated with goat anti-Ud serum diluted in blocking solution for 1 h, washed, and then incubated with horseradish peroxidase-conjugated donkey anti-goat IgG (Jackson Immunoresearch Laboratories) secondary antibody diluted in blocking solution for 1 h. Plaques were visualized using the ImmunoPure metal-enhanced DAB substrate kit (Pierce Biotechnology, Rockford, IL) and photographed.

To analyze virus budding, confluent MDCK cells grown in 6-cm dishes were inoculated with virus at an MOI of 1 in DMEM-1% BSA for 1 h at 37°C. The inoculum was removed and replaced with DMEM supplemented with 10% FBS and 0.1 mg/ml soybean trypsin inhibitor to prevent multiple rounds of infection. Twenty hours postinfection, the culture medium and cells were harvested and processed as described for the assaying of VLP budding.

Coimmunoprecipitation.

293T cells (3 × 10⁶ cells) grown in 10-cm dishes were transfected with pCAGGS-M1 (3 μg) and pCAGGS-M2 (1 μg). At 20 h p.t., the cells were washed with PBS and then lysed in lysis buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 1% polyethylene glycol 400 dodecyl ether [Thesit]) with 1 mM phenylmethanylsulfonyl fluoride and 1× protease inhibitor cocktail for 15 min with gentle rocking. All coimmunoprecipitation steps were performed at 4°C. Cell lysates were transferred to centrifuge tubes, and insoluble material was pelleted at 14,000 × g for 10 min. M2-specific MAb 14C2 was then added to the clarified cell lysate, and samples were incubated for 1 h. Samples were then incubated with protein A/G-Sepharose for 30 min to precipitate immune complexes. Samples were washed three times with lysis buffer and twice with wash buffer (50 mM Tris, pH 7.4, 150 mM NaCl). Samples were then boiled in SDS loading buffer, separated by SDS-PAGE on a 17.5% polyacrylamide gel with 4 M urea, and analyzed by immunoblotting.

Analysis of planar sheets of plasma membrane.

MDCK cells were infected with virus at an MOI of 4 for 12 h. Cells were blocked in 0.1 M phosphate buffer, pH 7.4, containing 0.2% ovalbumin (Sigma) and 0.1% cold water fish gelatin (Electron Microscopy Sciences, Hatfield, PA) and stained for HA and/or M2 where indicated. Antibody binding was detected using 6-nm or 12-nm-gold-particle-conjugated secondary antibodies diluted in blocking buffer. All immunogold reagents were examined with an electron microscope (EM) and found to be monodisperse and free of aggregates. Sheets of plasma membrane were prepared as described previously (48, 59, 64). Membrane sheets were fixed with 2% formaldehyde. If M1 on the cytoplasmic face of the plasma membrane was to be labeled, M1 was stained with primary and gold-conjugated secondary antibodies as described above. Grids were then fixed with 2% glutaraldehyde, washed, and treated with 1% tannic acid. The plasma membrane of Ud-infected MDCK cells exhibits areas of considerable electron density; therefore, to aid in the identification of gold particles, the membrane sheets were not postfixed with osmium tetroxide or stained with uranyl acetate. Membrane preparations were examined with a JEOL 1230 EM (JEOL Ltd., Tokyo, Japan) operating at 100 kV. Digital images were acquired at a magnification of ×40,000. At least 10 images were analyzed for each experimental condition, with each image derived from a different membrane sheet. Each image encompassed 11.1 μm² of membrane. The level of background labeling was determined by staining membrane sheets prepared from mock-infected MDCK cells and ranged from 0.61 to 1.49 gold particles per μm² of membrane.

Thresholds were determined for micrographs, and an x/y coordinate list of all gold particles was generated using CorelDraw (Corel, Fremont, CA) and a customized plug-in (Center for the Spatiotemporal Modeling of the Cell, University of New Mexico) for Image J (http://rsb.info.nih.gov/ij/). Coclustering of gold-particle populations was analyzed with the Ripley bivariate analysis (45, 46) by using the Cellspan Matlab tools (Center for the Spatiotemporal Modeling of the Cell, University of New Mexico) and the SpatStat library for analyzing spatial point patterns by using R (2).

RESULTS

M2 cytoplasmic tail mutant constructs.

Previous studies have shown that the cytoplasmic tail domain of M2 is important for influenza virus replication (9, 24, 33, 34). Truncations between residues 45 to 70 impact ion channel activity (58), whereas truncations beyond residue 70 permit the M2 protein to retain ion channel activity and to remain able to form tetramers (24, 34, 58). Therefore, M2 cytoplasmic tail mutants were generated by starting from Ud M2 serine-71 and substituting triplets of residues with alanine at the C terminus of the protein (M2-Mut1 [with mutation of residues 71 to 73] through M2-Mut9 [with mutation of residues 95 to 97]) (Fig. 1A). An M2 mutant protein designated M2-Stop71 was constructed by changing nucleotides encoding residues 71 and 72 to stop codons, and this produced an M2 protein with a C-terminal 27-amino-acid truncation.

FIG. 1.

Influenza virus Ud M2 cytoplasmic tail mutants. (A) The amino acid sequence of the Ud M2 cytoplasmic tail (CYT) is shown. Nomenclature and amino acid changes in the M2 mutants created by alanine-scanning mutagenesis that were used in this study are listed below the wt M2 cytoplasmic tail sequence. Amino acid residues that are the same as those of the wt are indicated by periods, and asterisks indicates the M2 stop codons. ECTO, ectodomain; TM, transmembrane domain. (B) The stabilities of the mutant M2 proteins were tested by pulse-chase analysis. 293T cells were transfected with protein expression plasmids encoding wt M2 or the mutant M2 proteins. At 20 h p.t., the cells were metabolically labeled with [³⁵S]methionine and [³⁵S]cysteine. Cells were then lysed (pulse samples) or incubated in nonradioactive medium for 2 h and then lysed (chase samples). M2 protein in the cell lysate was immunoprecipitated with 14C2 MAb and analyzed by SDS-PAGE. Radiolabeled M2 proteins were detected by using a Fujifilm FLA-5100 system, and the ratio of the amount of protein in the chase sample to the amount of protein in the pulse sample was calculated by using MultiGauge software. (C) The surface expression levels of the mutant M2 proteins were determined by flow cytometry. Transfected 293T cells were labeled with MAb 14C2 followed by a fluorescein isothiocyanate-conjugated secondary antibody. Labeled cells were detected by flow cytometry, and the mean fluorescence intensity of cells expressing mutant M2 proteins was normalized to the mean fluorescence intensity of cells expressing wt M2.

The stability of the M2 mutant proteins was tested by pulse-chase analysis in 293T cells. Cells were transfected with plasmids expressing the M2 cytoplasmic tail mutants and metabolically labeled with [³⁵S]methionine and [³⁵S]cysteine for 30 min and chased for 2 h in nonradioactive medium. Comparable amounts of wt M2 and all mutant M2 proteins remained after the 2-h chase period, indicating that the mutant M2 proteins were stable in transfected cells (Fig. 1B). Less M2-Mut1 was detected using this assay because the only methionine residue in M2, methionine-72, was mutated to alanine, resulting in only incorporation of radioactive cysteine. M2-Mut1 expression was comparable to that of wt M2 as determined by immunoblotting (see Fig. 2A). Cell surface expression levels of M2 mutants transiently expressed in cells were analyzed by flow cytometry and were found to be comparable to wt M2 surface expression levels (Fig. 1C).

FIG. 2.

Release and infectivity of VLPs containing mutant M2 proteins. (A) VLPs were prepared with transfected 293T cells as described in Materials and Methods. As indicated, an M2 protein expression plasmid was omitted from the VLP preparation (−M2) or wt M2 expression plasmid was replaced with the indicated M2 mutant expression plasmid. The culture medium was harvested at 48 h p.t. and pelleted through a 30% sucrose cushion, and cell lysates were prepared as described in Materials and Methods. Samples were analyzed by SDS-PAGE on a 15% polyacrylamide gel followed by immunoblotting to detect viral proteins. (B) M1 protein in the 30% sucrose pellet was quantified by using the Odyssey infrared imaging system, and the value was normalized to the amount of M1 protein found in the cell lysate. The amount of M1 released from cells expressing all VLP proteins (WT) was set to 1.0. Error bars represent standard deviations from the averages for three experiments. (C) VLPs prepared as described above were harvested at 48 h p.t. and treated with trypsin. VLPs that were not treated with trypsin were used as a negative control. VLPs were transferred to target 293T cells expressing PB1, PB2, PA, and NP. Twenty-four hours later, the target cells were fixed, and GFP-positive cells were quantified by flow cytometry. The number of GFP-positive target cells overlaid with WT VLPs was set to 100%. Error bars represent standard deviations from the averages for three experiments.

M2 cytoplasmic tail mutants affect protein composition and infectivity of influenza VLPs.

By using a previously described VLP system that produces particles that resemble infectious virions (11), the protein compositions of VLPs containing mutant M2 proteins were assessed (Fig. 2A and B). As demonstrated previously (11), omission of M2 from VLPs resulted in about a 50% decrease in the amount of M1 incorporated into the VLP (Fig. 2A, lane 2) compared to VLPs containing wt M2 (Fig. 2A, lane 1). M2-Mut1, M2-Mut2, and M2-Mut7 resulted in a 40 to 60% decrease in M1 incorporation into VLPs, suggesting that these mutant proteins produced a defect in VLP assembly (Fig. 2A, lanes 3, 4, and 9). M2 appears as two bands in the cell lysate due to proteolytic cleavage by cellular caspases (67).

The ability of VLPs containing mutant M2 to transfer a pseudogene construct encoding GFP in the negative sense (vGFP) was tested in a VLP infectivity assay (11). VLPs were treated with trypsin and overlaid onto target 293T cells expressing influenza virus NP and polymerase proteins PB1, PB2, and PA. GFP-positive cells were then counted by flow cytometry (Fig. 2C). Omission of M2 from VLPs resulted in a reduction in GFP-positive cells to a level similar to the background number of GFP-positive cells observed when VLPs were not treated with trypsin. Incorporation of M2-Mut1 and M2-Mut2 into VLPs also resulted in a dramatic reduction in VLP infectivity, suggesting a correlation between reduced M1 packaging and genome packaging for these mutants. However, VLPs with M2-Mut7 had a reduced amount of M1, but VLP infectivity did not appear to be compromised. In the case of M2-Stop71, as suggested previously (34), VLP infectivity was reduced significantly, suggesting a defect in genome packaging. However, M1 incorporation into VLPs did not appear to be affected by M2-Stop71. Thus, although there is a trend linking M1 incorporation and genome incorporation into VLPs, discrepancies between M1 packaging and genome packaging into VLPs could point to overlapping or redundant mechanisms or a limitation of the VLP system.

Specific residues in the M2 cytoplasmic tail are required for efficient virus replication and virus budding.

To understand better the effects of M2 mutations on virus replication, mutant viruses in the Ud genetic background were generated by reverse genetics. To facilitate virus recovery, all mutants were propagated on MDCK cells stably expressing wt M2 protein (M2-MDCK cells). Ud7+UdM1/M2-Mut1 to Ud7+UdM1/M2-Mut8 and Ud7+UdM1/M2-Stop71 were generated successfully in this manner. Despite multiple attempts, Ud7+UdM1/M2-Mut9 could not be recovered for reasons unknown, although a possibility is the disruption of crucial RNA packaging signals in genome RNA segment 7 (38, 41).

All of the recovered M2 mutant viruses could form plaques on M2-MDCK cells, although Ud7+UdM1/M2-Mut1 and Ud7+UdM1/M2-Mut2 formed smaller plaques on M2-MDCK cells than wt Ud virus and the other M2 mutant viruses did (Fig. 3A). On MDCK cells, Ud7+UdM1/M2-Mut1, Ud7+UdM1/M2-Mut2, and Ud7+UdM1/M2-Stop71 failed to form detectable plaques. This observation correlated with the multistep replication kinetics of the mutant viruses in MDCK cells after infection at a low MOI (0.001) (Fig. 3B). Ud7+UdM1/M2-Mut1, Ud7+UdM1/M2-Mut2, and Ud7+UdM1/M2-Stop71 replicated poorly, barely reaching a titer of 10 PFU/ml after 48 h, whereas wt Ud virus reached a titer of 1 × 10⁸ PFU/ml after 48 h. Replication of Ud7+UdM1/M2-Mut3 was delayed compared to that of wt Ud virus, reaching a titer of 1 × 10⁵ PFU/ml after 48 h and forming small plaques on MDCK cells. Ud7+UdM1/M2-Mut4 to Ud7+UdM1/M2-Mut8 all formed plaques on MDCK cells, similar to wt Ud virus, and their levels of replication were within 2 logs of that of wt Ud virus after 48 h.

FIG. 3.

Plaque phenotype and replication of Ud M2 mutant viruses. (A) Ud influenza virus and mutant viruses containing M2 cytoplasmic tail mutations were generated by reverse genetics and assayed for plaque formation on MDCK and M2-MDCK cells. After 3 days of incubation, plaques were immunostained using goat anti-Ud serum followed by horseradish peroxidase-conjugated secondary antibody and metal-enhanced DAB substrate. All mutant viruses are Ud7+UdM1/M2 mutants. (B) MDCK cells were infected at an MOI of 0.001. At 12-h time points, an aliquot of the culture medium was removed, and virus titers were determined by plaque assay on M2-MDCK cells. Error bars represent standard deviations from the averages for three experiments.

The polypeptide compositions of virus particles harvested from a virus budding assay showed ∼60% decreases in the amount of M1 incorporated into Ud7+UdM1/M2-Mut1, Ud7+UdM1/M2-Mut2, and Ud7+UdM1/M2-Stop71 (Fig. 4A, lanes 2, 3, and 10) compared to the amount of M1 incorporated into wt Ud virus (Fig. 4A, lane 1), consistent with the poor growth observed. A more modest reduction (20 to 40%) in M1 incorporation was also observed for Ud7+UdM1/M2-Mut3, Ud7+UdM1/M2-Mut4, and Ud7+UdM1/M2-Mut6 (Fig. 4A, lanes 4, 5, and 7), reflecting the 3-log decrease in growth of Ud7+UdM1/M2-Mut3 and 1-log decrease in growth of Ud7+UdM1/M2-Mut4 and Ud7+UdM1/M2-Mut6 (Fig. 3B). Interestingly, although Ud7+UdM1/M2-Mut7 replicated similarly to wt Ud virus, reaching a titer of about 5 × 10⁶ after 48 h, it exhibits an apparent defect in M1 packaging, both for virus (Fig. 4A, lane 8) and VLPs (Fig. 2A, lane 9). McCown and Pekosz (33, 34) found a requirement for the M2 cytoplasmic tail in packaging the RNPs into virions. Thus, the finding that the infectivity of VLPs with M2-Mut7 is comparable to that of wt VLPs (Fig. 2C) suggests that although M1 packaging may be disrupted, the genome packaging function of M2 remains intact, resulting in a relatively minor effect on virus replication due to this mutation. In contrast, mutations in the more proximal region spanned by M2-Mut1 and M2-Mut2 appear to impact both M1 recruitment and genome packaging.

FIG. 4.

Assembly and release of Ud M2 mutant viruses. (A) MDCK cells were infected with wt Ud or Ud7+UdM1/M2 mutant virus at an MOI of 1 for 20 h. The culture medium was harvested, virions were pelleted through a 30% sucrose cushion, and cell lysates were prepared as described in Materials and Methods. Samples were analyzed by SDS-PAGE on a 15% polyacrylamide gel followed by immunoblotting to detect viral proteins. (B) M1 protein in the 30% sucrose pellet was quantified by using the Odyssey infrared imaging system and normalized to the amount of M1 protein found in the cell lysate. The amount of M1 released from cells infected by wt Ud was set to 1.0. Error bars represent standard deviations from the averages for three experiments.

M2-Mut1 and M2-Mut2 exhibit a decreased interaction with M1.

To probe whether there is an interaction between residues 71 to 76 of the M2 cytoplasmic tail and M1, coimmunoprecipitation of cDNA-expressed M1 and M2 was performed by using the MAb 14C2 to pull down M2. Although there was a relatively high background level of M1 pulled down in the absence of MAb 14C2, probably due to nonspecific association with the protein A- and protein G-conjugated Sepharose, there was a reproducible increase in the level of M1 detected when MAb 14C2 was used to pull down wt M2 (Fig. 5). When M1 was coexpressed with either M2-Mut1 or M2-Mut2, the amount of coimmunoprecipitated M1 fell reproducibly to background levels, suggesting a decreased association between M1 and M2-Mut1 or M2-Mut2 (Fig. 5). When M1 was coexpressed with M2-Mut3 through M2-Mut9, the amount of coimmunoprecipitated M1 was not significantly affected (data not shown).

FIG. 5.

Coimmunoprecipitation (coIP) of M1 and mutant M2 proteins. 293T cells were transfected with plasmids expressing Ud M1 and wt M2, M2-Mut1, or M2-Mut2. Cells were lysed and M1-M2 complexes were coimmunoprecipitated by using MAb 14C2. Samples were analyzed by SDS-PAGE on a 17.5% polyacrylamide gel with 4 M urea followed by immunoblotting to detect M1 and M2. The amount of M1 detected was quantified by using the Odyssey infrared imaging system and normalized to the amount of M1 pulled down by wt M2 in the presence of MAb 14C2. Error bars represent standard deviations from the averages for three experiments.

Experiments involving coimmunoprecipitation of M1 and M2 from virus-infected cells were not performed, because expression of the M2 mutant proteins was not consistent (Fig. 4A), and this would confound quantification, although all M2 mutant proteins were expressed at the surface of virus-infected cells (data not shown). The reason for the variability in M2 mutant stability only in virus-infected cells is currently unclear, but this phenomenon has been observed previously (34).

Analysis of coclustering of M2, M1, and HA in membrane sheets prepared from M2-Mut1-infected cells.

Whereas there are biochemical data showing the intrinsic lipid microdomain association of HA and NA (3, 4, 18, 50, 55, 66), it has been observed that M2 is susceptible to TX-100 extraction at 4°C (52, 66), suggesting that M2 is not found in detergent-resistant membrane microdomains. EM analysis of serial thin sections from virus-infected cells has also revealed large patches of HA and NA on the plasma membrane that are approximately 400 nm in diameter, with M2 associated at the periphery of these clusters (30). Thus, although M2 is not associated with lipid microdomains as defined by TX-100 solubility and as observed in thin-section electron microscopy, M2 is nonetheless incorporated in small amounts into virions.

As M2-Mut1 had a decreased association with M1 (Fig. 5), we examined whether this M2 cytoplasmic tail mutation would alter the distribution of M2 or other viral proteins on the virus-infected cell membrane. To test biochemically whether M2-Mut1 affected the lipid microdomain association of HA and M1, the association of HA and M1 with detergent-resistant microdomains was analyzed first by TX-100 extraction of virus-infected cells at 4°C. HA and M1 from both wt Ud and Ud7+UdM1/M2-Mut1 virus-infected cells were found to be predominantly associated with the insoluble membrane fraction, indicating that M2-Mut1 did not grossly affect the lipid microdomain association of HA and M1 (data not shown).

To investigate further associations of M1, M2, and HA, the spatial distribution of viral proteins in planar sheets of plasma membrane prepared from virus-infected cells was examined. This procedure has been used to visualize and to analyze statistically the distribution of a variety of membrane and membrane-associated proteins, including that of influenza virus HA (18), Ras (44), Thy-1, and LAT (60), and to analyze changes in membrane protein distribution upon activation of T cells (31). MDCK cells were grown on poly-l-lysine-coated coverslips, infected for 12 h with wt Ud virus or Ud7+UdM1/M2-Mut1 virus, and reacted with antibodies specific for surface-expressed HA and M2. Cell surfaces were then reacted with secondary antibodies conjugated to 12-nm or 6-nm gold particles. Membrane sheets from these labeled cells were then prepared by methods described previously (48, 59) which give access to antigens associated with the cytoplasmic face of the plasma membrane. M1 is then exposed and can be labeled with immune reagents. Portions of typical membrane sheets from MDCK cells infected with wt Ud virus (Fig. 6A) or Ud7+UdM1/M2-Mut 1 virus (Fig. 6C), labeled for M2 (6-nm gold) and M1 (12-nm gold), are shown. Clathrin-coated pits can be observed (Fig. 6A and C), and these indicate preservation of the structural integrity of the membrane following the entire procedure. Low-magnification digital electron micrographs of labeled plasma membrane were processed to generate an x/y coordinate list of all gold particles. In Fig. 6B and D, the gold-particle coordinates for the portion of corresponding membrane sheets in Fig. 6A and C, respectively, are shown.

FIG. 6.

Distributions of M2 and M1 in membranes of virus-infected cells. Planar sheets of plasma membrane stained for viral proteins were prepared from infected MDCK cells as described in Materials and Methods. A portion of membrane sheet stained for M1 and M2, visualized with 12-nm and 6-nm gold particles, respectively, is shown for MDCK cells infected with wt Ud (A) or Ud7+UdM1/M2-Mut1 (C). Clathrin-coated pits are indicated by red arrows. Areas of clustering identified by visual inspection are outlined by boxes. Bar, 0.5 μm. Images of membrane sheets for which the thresholds had been determined were obtained, and an x/y coordinate list of all gold particles was generated. Graphic representations of the coordinate maps corresponding to panels A and C are shown in panels B and D, in which the locations of M1 are in blue and those of M2 are in red. Boxes indicate areas of clustering.

To quantify the potential colocalizations of two different viral proteins, the positioning of each gold particle was statistically analyzed using Ripley's bivariate K function, which examines for each gold particle the density of neighboring gold particles of a second population as a function of distance (Fig. 7) (45, 46). These data were then compared to the distribution expected if both particle populations were distributed randomly with respect to one another. The results of the bivariate Ripley analysis were compared to the averages of 100 simulations of two random-particle populations of the same average density as the populations from which the experimental data were obtained. If two proteins are distributed randomly with respect to one another, the linear transformation of the bivariate Ripley K function [L(r) − r] has a theoretical value of 0. The 99% confidence interval for the random simulations is also shown. If the experimental data fall above the confidence envelope, the proteins can be considered to be colocalized, with the height of the line proportional to the magnitude of coclustering. If the experimental data fall below the envelope, the proteins are considered to be explicitly separated or excluded. Experimental data that fall within the envelope represent proteins that exhibit a random distribution with respect to one another. For each analysis whose results are shown in Fig. 7, the data from wt Ud virus-infected cells are shown in blue and the data from Ud7+UdM1/M2-Mut1 virus-infected cells are shown in red. At least 10 images, each from an individual cell, were analyzed.

FIG. 7.

Statistical analysis of M2, M1, and HA clustering. The relative distributions of M1 and HA (A), M2 and HA (B), or M2 and M1 (C) were analyzed with the bivariate Ripley function, and these distributions for wt Ud (blue) and Ud7+UdM1/M2-Mut1 (red) were compared. The slight initial drop in the data for distances 10 nm and below is likely due to an artifact of antibody binding. The solid, black line represents a bivariate Ripley K function with a theoretical value of 0. The dashed lines represent the 99% confidence interval defining complete spatial randomness. The data that fall above the envelope are considered to represent coclustering, the magnitude of colocalization being proportional to the height of the line. When data fall within the envelope, the two protein populations are said to be distributed randomly with respect to one another.

A comparison of the clusterings of M1 and HA is shown in Fig. 7A. In cells infected with mutant virus Ud7+UdM1/M2-Mut1, M1 and HA coclustered significantly, although to a lesser extent (∼40% at 100 nm) than these two proteins clustered in wt virus-infected cells. The difference in clustering is much greater in the comparison of the colocalization between M2 and HA (Fig. 7B). M2 protein expressed in Ud7+UdM1/M2-Mut1 virus-infected cells exhibits an approximately 80% reduction at 100 nm in coclustering with HA compared to wt M2 and HA. However, it should be noted that the distribution of M2-Mut1 and HA still falls above the confidence envelope, indicative of significant clustering.

The most dramatic difference in clustering was observed when the colocalizations of M2 and M1 in wt Ud virus-infected cells and Ud7+UdM1/M2-Mut1 virus-infected cells were compared (Fig. 7C). Whereas M1 and M2 expressed by wt Ud virus exhibited a significant association of a magnitude similar to that observed between M1 and HA, the M1 and M2 expressed by the Ud7+UdM1/M2-Mut1 virus did not cocluster; the data fell within the envelope, indicating complete spatial randomness. This observation was unchanged when the experiment was repeated with the sizes of the gold particles reversed (data not shown). Thus, expression of M2-Mut1 in Ud7+UdM1/M2-Mut1 virus-infected cells appeared to influence the distribution and disrupt coclustering of M1 and HA at the cell surface.

DISCUSSION

The mechanism by which influenza virus proteins at the cell surface assemble a budding virion appears to involve the coordination of all three viral integral membrane proteins (HA, NA, and M2) with the soluble viral components (M1, vRNPs, and NS2 [or NEP]). Although the functions of the cytoplasmic domains of HA, NA, and M2 appear to overlap somewhat, perhaps providing redundancy to the system, disrupting specific regions of HA, NA, and M2 can lead to defects in virus assembly and replication. In line with results from previous studies, results from our experiments highlight specific regions in the M2 cytoplasmic tail that are required for virus replication, and the results also provide evidence for a role for M2 in virus assembly that involves the spatial distribution of viral proteins on the cell surface, presumably through contacts between M2 and M1.

By use of our panel of mutants, we found that mutation of residues 71 to 73 (M2-Mut1) and 74 to 76 (M2-Mut2) drastically impacted virus replication (Fig. 3). In both VLP and virus budding assays (Fig. 2A and 4), inclusion of M2-Mut1 or M2-Mut2 resulted in decreased amounts of M1 released into the medium, reflecting a defect in M1 incorporation into budded particles. Previous VLP studies have shown that M1 is required for genome packaging (11), and thus, a decrease in M1 packaging due to mutant M2 proteins was also reflected in a decrease in VLP infectivity and pseudogene packaging (Fig. 2B). These results complement previous findings that demonstrated that the M2 cytoplasmic tail is required for efficient genome packaging (34). The finding that M2-Mut7, in both VLPs and virus, resulted in a decrease in the amount of M1 incorporation but produced a viable virus and could package the vGFP pseudogene presents a discrepancy that requires additional examination. Interestingly, there was no change in the amount of M1 coimmunoprecipitated with M2-Mut7 (data not shown). It was found previously, however, that mutations in this region was compatible with virus growth (24, 33), and this is reflected in the moderate growth of Ud7+UdM1/M2-Mut7 (Fig. 3). Therefore, it is possible that M1 and vRNP binding to the M2 cytoplasmic tail may vary along the length of the protein and that mutation in M2 residues 86 through 88 results in a more subtle defect, a result which is in line with conclusions from prior studies (33).

The viral genome contents of Ud7+UdM1/M2-Mut1 and Ud7+UdM1/M2-Mut2 were not tested, because a sufficient amount of particles could not be obtained, due to the drastic growth defect of the mutant viruses, precluding attempts to properly normalize RNA levels to infectious units. Similarly, mutant virus morphology was difficult to assess, although examination of purified Ud7+UdM1/M2-Mut1 and Ud7+UdM1/M2-Mut2 viruses did not reveal a gross change in morphology (data not shown), as suggested in another study (24). However, a statistical analysis of particle size was not performed, due to the small numbers of particles obtained from infection of MDCK cells.

It has proven difficult to demonstrate a physical interaction between M1 and the viral integral membrane proteins by using biochemical methods, probably due to relatively weak interactions that are easily disrupted by common methods. However, we did detect a difference between the abilities of M2-Mut1 and M2-Mut2 and of wt M2 to coimmunoprecipitate with M1 (Fig. 5). These results differed somewhat from those from experiments that used only the cytoplasmic tail portion of Ud M2 fused to GST to pull down M1 from A/WSN/33 (WSN)-infected cells (33). In those experiments, it was concluded that mutation of residues 70 to 77 did not disrupt WSN M1 binding. In addition, because expression of only residues 70 to 97 fused to GST failed to pull down WSN M1, it was concluded that the WSN M1 binding domain resided upstream of M2 residue 70, although this was not demonstrated experimentally. In the context of a full-length Ud M2 protein, it is possible that the cytoplasmic tail is able to adopt a conformation in which Ud M1 can interact with residues 71 to 76.

The finding that M2 residues 71 to 73 are important for an interaction with M1 is interesting because the mutation of serine-71 was one of the changes observed for Ud viruses that could escape MAb 14C2 selection (61). MAb 14C2 selection resulted in mutations in the M2 cytoplasmic domain and M1 but did not change the reactivity of M2 to MAb 14C2. A mechanism for the escape from MAb 14C2-mediated growth restriction could be a change in the strength of interaction between M1 and M2. MAb 14C2 has been shown to reduce the level of M2 expressed on the virus-infected cell surface (21), but an increased interaction between M2 and M1 may be able to overcome this decrease in surface expression and enable viruses to assemble with a sufficient amount of M2 incorporated.

Interpreting the role of the M2 cytoplasmic tail in virus assembly has not been straightforward. Prior studies examining either WSN or Ud strains of viruses have yielded differing results for the two strains with respect to the requirement of M2 for virus replication, suggesting that important differences exist between WSN and Ud virus strains. In a comparative study, cell lines expressing Ud M2 cytoplasmic tail mutants differed in their abilities to trans-complement the growth of Ud or WSN viruses from which functional M2 was deleted (33). These differences were mapped to the genetic origin of the M1 protein because the WSN M1 protein could rescue growth of M2-deleted viruses grown in cells with M2 residues 70 to 73 mutated to alanine, whereas M2-deleted viruses containing the Ud M1 protein failed to grow (33). WSN viruses with M2 mutations at residues 74 to 76 or residues 77 to 79 that were generated by reverse genetics were found to be altered in growth, although it was concluded that the major effect of the M2 mutations was the production of smaller viruses (24), a result that is probably linked to impaired virus assembly. Our results with mutant Ud viruses generated by reverse genetics are consistent with results from these previous studies in that assembly defects were observed when amino acids near residue 71 were mutated to alanine. The mechanism by which the six amino acid differences between Ud and WSN M1 can so profoundly contribute to differences in virus morphology (6, 7, 14), vRNP binding (32), HA palmitoylation requirements (12), and M2 cytoplasmic tail interactions (33) remains to be elucidated. In addition, the M2 ectodomain (42) and a potential cholesterol binding domain (52) found in the M2 protein of WSN, A/Germany/27 Weybridge, and A/Singapore 1/57 have been implicated in virus assembly. However, the potential cholesterol binding motif is not present in the M2 protein of Ud and other influenza virus strains, suggesting that different virus strains may utilize a diversity of features in M2 and other viral proteins during virus assembly.

The en face analysis of planar membrane sheets to analyze statistically the two dimensional distribution of membrane and membrane-associated proteins shows that M2-Mut1 influences the distribution of both M1 and HA on the surface of cells infected with Ud7+UdM1/M2-Mut1 virus (Fig. 6 and 7). Whereas the presence of M2-Mut1 did not significantly disrupt M1-HA and M2-HA clustering, the extent of clustering in each case was reduced. More interestingly, and consistent with coimmunoprecipitation data demonstrating a decreased association between M2-Mut1 and M1 (Fig. 5), M2 and M1 were found to be randomly distributed on the surface of Ud7+UdM1/M2-Mut1 virus-infected cells. Together, these data argue strongly for an interaction between the M2 cytoplasmic tail and M1 that is important for virus assembly and, consequently, essential for virus replication.

In the future, analysis of the distributions of the vRNPs with HA, NA, M2, and M1 and putative host proteins is required for a complete understanding of the spatial assembly of proteins at the viral budozone (10, 51). Nonetheless, the present experimental data suggest a model for influenza virus assembly that includes all three viral envelope proteins and their interactions with internal viral proteins. M2 appears to have the potential to serve as a scaffold for recruiting M1, perhaps a subset of M1 bound to vRNPs. How this assembly mechanism might lead to organization of HA and NA into complete, budding virions and lead to the selective incorporation of viral genome segments will require further investigation.