|Home | About | Journals | Submit | Contact Us | Français|
Influenza A virus buds from cells as spherical (~100-nm diameter) and filamentous (~100 nm × 2 to 20 μm) virions. Previous work has determined that the matrix protein (M1) confers the ability of the virus to form filaments; however, additional work has suggested that the influenza virus M2 integral membrane protein also plays a role in viral filament formation. In examining the role of the M2 protein in filament formation, we observed that the cytoplasmic tail of M2 contains several sites that are essential for filament formation. Additionally, whereas M2 is a nonraft protein, expression of other viral proteins in the context of influenza virus infection leads to the colocalization of M2 with sites of virus budding and lipid raft domains. We found that an amphipathic helix located within the M2 cytoplasmic tail is able to bind cholesterol, and we speculate that M2 cholesterol binding is essential for both filament formation and the stability of existing viral filaments.
Influenza virus is an enveloped virus with a segmented negative-sense RNA genome. The eight RNA segments encode 11 proteins (45). There are two major spike glycoproteins, the receptor binding/membrane fusion protein hemagglutinin (HA) and the enzyme neuraminidase (NA). A third integral membrane protein, M2, has a proton-selective ion channel activity and is involved in virus assembly. Three proteins, PB1, PB2, and PA, form the RNA polymerase complex and, together with the nucleocapsid protein (NP), form the ribonucleoprotein (RNP) core. A matrix protein (M1) forms peripheral interactions with the lipid envelope and the RNP. In addition, the virus encodes the multifunctional protein NS1, which is an interferon antagonist, and the protein NEP or NS2, involved in export of RNA from the nucleus. Most, but not all, strains of influenza virus also encode PB1-F2, a proapoptotic protein.
Influenza virus assembles at and buds from lipid raft domains in the plasma membranes of infected cells (10, 34, 59). Influenza virus buds either as filamentous virus particles (100 nm × 2 to 20 μm) or as spherical particles (100-nm diameter). Infection of cells with strain A/WSN/33 (H1N1) (strain WSN) produces almost entirely spherical virions, whereas infection of cells with strain A/Udorn/72 (H3N2) (strain Udorn) produces a mixture of spherical and filamentous virions. Filamentous virions appear to be the predominant form isolated from the human upper respiratory tract (14, 31) and were seen in the recent 2009 H1N1 pandemic virus (41). Indeed, filamentous influenza A viruses have been observed since the earliest days of examining viruses in the electron microscope (1, 7, 11-14, 31, 40). Although fresh or limited-passage isolates of human influenza viruses are predominantly filamentous in appearance, conversion of filamentous to spherical viruses occurs after serial passage in eggs (13). It has been reported that the spherical and filamentous particles are equally infectious and that both contain one copy of the viral genome (44, 51). However, the transmission efficiency of 2- to 20-μm filamentous particles in comparison to that of 100-nm spherical particles is currently unknown.
Influenza virus filament formation is a genetic trait that has been mapped to the M1 protein (5, 6, 19, 51). Transfer of the Udorn M1 protein coding sequence into the WSN genetic background allows filament formation, whereas substitution of specific WSN M1 residues in the Udorn genetic background results in the production of spherical virions (5, 19). Additional studies have suggested that while it is not sufficient to confer filament formation, the M2 protein may also be involved in filament formation (36).
The M2 protein proton-selective ion channel activity permits protons to enter virus particles during uncoating in endosomes. Acidification of the virus interior causes dissociation of the M1 protein from the RNP core, a necessary step for the import of the RNPs into the nucleus (reviewed in references 32 and 49). The M2 protein is a homotetramer consisting of four polypeptide chains of 97 residues with a 24-residue N-terminal extracellular domain, a single internal hydrophobic domain that acts as a transmembrane (TM) domain and forms the pore of the ion channel, and a 54-residue cytoplasmic tail (23, 33, 47, 64). The M2 ion channel is the target of the antiviral drug amantadine (reviewed in reference 48). Recently, the atomic structure of the M2 TM domain was determined by X-ray crystallography (57), and the atomic structure of a larger peptide (residues 18 to 60) was determined by nuclear magnetic resonance (NMR) methods (53). Both models of the structure of the M2 TM domain indicate that it forms a 4-helix bundle, as predicted previously (46). Confirming earlier work (43, 60), the NMR-derived structural model indicates that residues 46 to 62 of the cytoplasmic tail form an amphipathic helix that lies parallel to the cytoplasmic side of the membrane.
In addition to ion channel activity, the M2 protein has a role(s) in virus assembly. Recent evidence suggests that the M2 cytoplasmic tail is involved in binding to M1 (9). Also, truncation of the M2 cytoplasmic tail alters the ability of strain Udorn to form filaments, suggesting that M2 may modulate the ability of M1 to form filamentous virions (28, 36). Further evidence for the interaction between M1 and M2 comes from studies with an M2 ectodomain-specific monoclonal antibody (MAb), 14C2 (63). This MAb mediates growth restriction of certain strains of influenza A virus (62, 63), and the minimal binding motif has been mapped to residues 4 to 15 in the M2 ectodomain (38). Viral mutants that escape the MAb 14C2-mediated growth restriction have been identified, and unusually, the mutations do not occur in the M2 ectodomain epitope but map to the M1 protein and to the cytoplasmic tail of M2 (62). Analysis of these escape mutants showed that many had lost the ability to form filamentous virions (51). Thus, while M1 may be sufficient to confer filament formation, interactions with M2 during budding may modulate this ability. In this study, we have investigated the contributions of M2 to viral filament formation and have found that M2 is a cholesterol-binding protein that is essential for the formation of filamentous influenza virions.
Madin-Darby canine kidney (MDCK) cells, M2-MDCK cells, 293T cells, viral stock propagation, and viral infections were as previously described (9). Recombinant A/Udorn/72 viruses were recovered by reverse genetics, using QuikChange mutagenesis (Stratagene, La Jolla, CA) of the M segment pHH21 plasmid as previously described (9). Recombinant viruses with mutations in M2 were recovered and titrated in M2-MDCK cells. M2-Mut1 and ΔM2 recombinant viruses have been described previously (9, 29). Antibodies used were anti-Udorn (9), anti-HA (NR3118; BEI, Manassas, VA), anti-M2 ectodomain-specific MAbs 14C2 (63) and 5C4 (23), and Alexa-488, Alexa-594, and Alexa-680 fluorophore-conjugated secondary antibodies (Invitrogen, Carlsbad, CA).
MDCK cells were infected with the indicated virus at a multiplicity of infection (MOI) of 0.001 for 64 h. Viral supernatant was then harvested, clarified by centrifugation at 3,000 × g for 30 min, and purified through 15% sucrose by centrifugation at 100,000 × g for 1 h at 4°C. The pelleted virions were resuspended in phosphate-buffered saline (PBS), and the total virion protein concentration was determined by bicinchoninic acid (BCA) assay (Pierce, Rockford, IL). Five micrograms of total protein was diluted into protein lysis buffer and subjected to SDS-PAGE and immunoblotting as previously described (9).
Cells were grown on glass coverslips, treated as indicated, fixed in 10% formalin-PBS (EMS, Hatsfield, PA), and stained without permeabilization. Cholera toxin subunit B (CTX-B) cross-linking was performed using a Vybrant lipid raft labeling kit (Invitrogen, Carlsbad, CA) per the manufacturer's instructions. Post-cross-linking, cells were fixed and stained for M2 and CTX-B. For MAb 5C4 and 14C2 treatments, cells were washed with warm Dulbecco's modified Eagle's medium (DMEM) and incubated in DMEM with 15 μg/ml of MAb 5C4 or MAb 14C2 for 1 h at 37°C. Cells were then fixed and stained for HA. All images were collected on an LSM5 Pa (Zeiss, Thornwood, NY) confocal microscope, using a 63× Plan-Apochromat objective (Zeiss) in multitrack mode. Alexa-488-stained samples were imaged using the 488-nm line of an argon laser (Lasos, Jena, Germany), a 488/543/633-nm excitation filter, a 545-nm dichroic filter, and a 505- to 530-nm emission filter. Alexa-594-stained samples were imaged using a 594-nm HeNe laser line (Lasos, Jena, Germany), a 488/543/633-nm excitation filter, a 545-nm dichroic filter, and a 560-nm emission filter. Settings were optimized to eliminate cross talk between detection channels. Three-dimensional (3D) optical sectioning was performed during imaging, and 2D maximal intensity projections and 3D rotations were generated using AIM software (Zeiss). Postimaging manipulation was performed in Photoshop (Adobe, San Jose, CA) and was limited to image cropping and equal adjustments of image levels. Colocalization was quantified using Velocity software (Improvision, Waltham, MA) on unedited z-series image files. Images were compared to a threshold, and Pearson's R coefficient was calculated over an average of 50 cells, with over- and underexposed cells removed from analysis.
Viral supernatant was treated as indicated, absorbed to glow discharge-treated (model 208C High Vacuum Turbo carbon coater; Ted Pella, Redding, CA), carbon-Formvar-coated, 300-mesh copper grids (Electron Microscopy Sciences, Hatfield, PA), stained with 2% phosphotungstic acid (pH 6.6) for 30 s, and imaged on a JEOL 1230 electron microscope (JEOL, Tokyo, Japan) at 100 keV. Images were captured using a Gatan (Pleasanton, CA) digital camera. Postimaging manipulation was performed in Photoshop (Adobe, San Jose, CA) and was limited to image cropping and equal adjustments of image levels.
Six-centimeter dishes of M2-MDCK cells or virus-infected MDCK cells were extracted with 1% Triton X-100 in NTE (100 mM NaCl, 10 mM Tris, pH 7.4, 1 mM EDTA) on ice for 30 min. Soluble and insoluble fractions were separated by centrifugation at 14,000 × g for 30 min at 4°C and subjected to SDS-PAGE and immunoblotting as previously described (9), with band intensities quantified in Photoshop (Adobe, San Jose, CA). The 6-h time point could not be examined due to the low levels of M2 present at this time, and the 24-h time point could not be examined due to infection-induced cytopathic effects in the majority of the cells.
M2 was immunoprecipitated from 10-cm dishes of MDCK, M2-MDCK, and virus-infected MDCK cells by use of Thesit lysis buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 1% Thesit), MAb 14C2, and protein G Sepharose beads as previously described (9). Immunoprecipitated M2 was quantified by SDS-PAGE and immunoblotting, with band intensities measured in Photoshop (Adobe, San Jose, CA). M2-bound cholesterol was quantified with an Amplex Red cholesterol assay kit (Invitrogen, Carlsbad, CA) per the manufacturer's instructions. At the 1-h time point, the reaction was read on a SpectraMax M5 fluorescence plate reader (Molecular Devices, Sunnyvale, CA).
M2 current recordings in Xenopus oocytes were performed as previously described (2-4). In brief, capped cRNA was in vitro transcribed, using T7 RNA polymerase (mMessage mMachine; Ambion, Austin, TX), from a linearized plasmid encoding the influenza virus A/Udorn/72 M2 protein. Stage V to VI Xenopus laevis oocytes were prepared as described previously (56). Oocytes were injected with 5 to 10 ng of cRNA in 50 nl/oocyte and were assayed 48 to 72 h later. Two-electrode voltage clamp (TEVC) recordings were performed, using a TEV-200 instrument (Dagan, Minneapolis, MN) connected to Digidata 1440A and pCLAMP10 devices (Molecular Devices, Sunnyvale, CA). Oocytes were superfused with Ca2+-free normal frog Ringer (NFR) solution (115 mM NaCl, 2.5 mM KCl, and 1.8 mM MgCl2) with 15 mM HEPES for pH 8.5 or 15 mM morpholineethanesulfonic acid (MES) for pH 5.5. Where indicated, oocytes were superfused with 15 μg/ml of MAb 14C2 in NFR solution, pH 8.5. Currents were recorded at −20 mV and analyzed using ORIGIN 8.0 software (OriginLab, Northampton, MA).
The amino acid sequence of the influenza virus M2 protein cytoplasmic tail is shown in Fig. Fig.1A.1A. Residues 45 to 62 have been shown to form an amphipathic helix (43, 53, 60) containing a potential cholesterol recognition/interaction amino acid consensus (CRAC) motif (54, 60) and a caveolin-1 binding domain (CBD) (67). Furthermore, cysteine residue 50 is modified by the covalent addition of palmitate (24, 58, 61). A role for palmitoylation in virus replication in vitro has not been found (8, 22), although there is a subtle effect of palmitoylation on virulence in mice (22). The M2 protein has also been shown to interact with the M1 protein, and the interaction site has been mapped to M2 residues 71 to 73 (9) (Fig. (Fig.1A).1A). The amphipathic helix is illustrated in Fig. Fig.1B1B to show the charged and hydrophobic faces of the helix. It has been shown that purified M2 protein binds cholesterol in vitro (54). To investigate the role of the amphipathic helix in filament formation, five highly conserved residues (F47, F48, I51, Y52, and F55) on the hydrophobic face of the M2 amphipathic helix were changed to alanine. These five mutations were introduced into the viral genome by reverse genetics (42), individually and in combination (M2-Helix), and mutant viruses were recovered. The single-alanine-substituted viruses grew to near-wild-type (near-wt) titers (data not shown), whereas the M2-Helix virus showed a growth impairment of >4 log (Fig. (Fig.1C).1C). Since the M2 amphipathic helix is inserted into the cytoplasmic face of lipid bilayers (43, 60), single point mutations to alanine are unlikely to disrupt the helix for lipid interaction, but the mutation of five highly conserved, bulky, hydrophobic residues to alanine is likely to have a more significant effect, as we have observed. We do not think it likely that the mutations in M2-Helix altered the entire M2 protein structure, as our analysis of M2-Helix ion channel activity showed that it was indistinguishable from wt M2 ion channel activity (35).
Because of the poor growth of the M2-Helix virus, we analyzed the protein composition of M2-Helix virions in comparison to that of wt virions. Approximately equal amounts of virus protein (μg) were analyzed by SDS-PAGE, with polypeptides identified by immunoblotting (Fig. (Fig.1D).1D). It was observed that the M2-Helix virions appeared to contain greater amounts of M2 protein than wt virus and smaller amounts of NP and M1 proteins than wt virus. Examination of M2 expression in wt and M2-Helix virus-infected cells showed comparable expression levels (data not shown). The increased amounts of M2 in M2-Helix virions may have stemmed from altered localization during the course of budding. Lowered levels of M1 (9) and NP incorporation have been reported previously for several different mutations in the M2 protein (28, 36, 37) and may reflect an important role of the M2 protein in assembly and budding.
It has been shown that immunostaining and light microscopy can be used to visualize filamentous virions (15, 50). An example of filamentous virion budding from A/Udorn/72 virus-infected MDCK cells is shown in Fig. Fig.2A.2A. To analyze the role of specific residues in the M2 cytoplasmic tail in filamentous virus formation, a series of mutations were generated, with mutant viruses recovered by reverse genetics. A mutant virus that does not express the M2 protein (ΔM2) lacked filamentous virion production (Fig. (Fig.2B),2B), confirming previous data (36). Recently, using scanning alanine mutagenesis, we identified a region of M2 involved in binding to M1 (9). Replacement of residues 71 to 73 in the M2 cytoplasmic tail with alanine (M2-Mut1) yielded a mutant virus that exhibited a reduced association between M1 and M2, an impaired release of virus-like particles (VLPs), and significantly attenuated replication (9). We observed that M2-Mut1 was also unable to form filamentous virions, suggesting an additional link between M1 and M2 in filament formation (Fig. (Fig.2C2C).
Single alanine substitutions in the M2 amphipathic helix, including mutation of a crucial reside in the CRAC domain (M2-Y52A) (30), had little effect on filament formation (Fig. 2D to H). However, in cells infected with the M2-Helix virus, filament formation was significantly impaired, suggesting that the combination of the five single alanine substitutions in this virus may be sufficient to disrupt either the properties of this helical region or the local structure, possibly modifying interactions with the membrane and preventing filament formation (Fig. (Fig.2I).2I). Extraction of membrane cholesterol by methyl-β-cyclodextrin (MβCD) treatment also eliminated filament formation, suggesting a possible role for cholesterol in filament formation (Fig. (Fig.2J).2J). Replacement of the palmitoylated cysteine residue (C50) with alanine was shown to have no effect on ion channel activity or on virus growth (8, 22), and infection of cells with the mutant virus (M2-C50A) had no effect on filament formation (Fig. (Fig.2K),2K), as found previously (22). Importantly, filament formation by wt virus was not affected when the M2 ion channel activity was inhibited by addition of the antiviral drug amantadine (Fig. (Fig.2L),2L), suggesting that the role of M2 in facilitating viral filament formation is independent of its ion channel activity.
M2 is expressed at high levels on the surfaces of virus-infected cells (33, 34). However, released virions contain, on average, only 5 to 15 tetramers of M2 protein, suggesting that M2 is actively excluded from the budding virion (63). It has been hypothesized that the exclusion of M2 from the sites of budding is accomplished by targeting of HA, NA, and M2 to different membrane microdomains (34, 54, 66). Given the involvement of M2 in filament formation, we investigated if M2 is able to localize to sites of virus budding. Virus-infected MDCK cells at 18 h postinfection (p.i.) were stained with antibodies specific for HA and M2. It was observed that M2 forms clusters that localize to the base of budding viral filaments (Fig. (Fig.3).3). M2 staining in the budding virion was difficult to detect, likely due to the active exclusion of M2 from the virion (Fig. (Fig.3).3). Examination of single z-section images allowed for individual filaments to be traced to their origin on the cell surface, where M2 foci were observed (Fig. (Fig.3).3). Additional M2 foci, independent of established filaments, were also observed, though these foci occasionally appeared near or under existing viral filaments, increasing the difficulty of demonstrating the localization of M2 to the base of budding filaments (Fig. (Fig.3).3). To confirm M2 colocalization with sites of viral budding, single z-section images at the cell surface were obtained and the colocalization between M2 and HA was quantified (Fig. (Fig.4).4). Extensive colocalization was observed between HA and M2 foci in wt virus-infected cells (Fig. (Fig.4A).4A). In M2-Helix virus-infected cells, M2 localized to HA foci that resembled sites of viral budding (Fig. (Fig.4b),4b), although minimal filament formation was observed (Fig. (Fig.2I).2I). Quantification of the colocalization between M2 and HA within the cell showed a significant association for both wt and M2-Helix viruses, suggesting that the M2-Helix mutation does not impair M2 recruitment to sites of virus budding (Fig. (Fig.4C4C).
The observed association between M2 and HA is similar to that seen previously when we analyzed the distributions of various viral proteins on planar sheets of virus-infected cell membrane by use of immuno-electron microscopy (9). Significant associations were observed between M1 and HA, between M1 and M2, and between M2 and HA (9). Coclustering between M2 and HA suggests the incorporation of M2 at sites of virus budding. The association between M2 and HA was eliminated when the binding between M2 and M1 was disrupted in the M2-Mut1 virus, though M1 was still able to associate with HA (9). Since M1 is thought to bind the cytoplasmic tails of both HA and M2, it is possible that the interaction with M1 is responsible for drawing M2 into the sites of budding and that the absence of M2 at these budding foci may lead to the impaired growth seen with the M2-Mut1 virus (9).
It has been shown that influenza virus budding occurs from lipid raft domains of the plasma membrane, whereas M2 is thought to largely be excluded from these lipid raft microdomains (34, 54, 66). Since we have shown that M2 is localized to sites of virus budding (Fig. (Fig.33 and and4),4), we investigated the possibility that M2 dynamically associates with lipid rafts at different times postinfection. By utilizing CTX-B as a marker for lipid rafts, we observed that in MDCK cells that constitutively express the M2 protein (M2-MDCK cells), M2 does not colocalize with lipid rafts (Fig. 5A and B), consistent with previous results (34, 65). However, in influenza virus-infected MDCK cells, with increasing time p.i., M2 was progressively incorporated into raft domains until 18 h p.i., after which there was a reduction in M2-raft association (Fig. 5A and B). The levels of M2 expression also increased in this period, but quantification of colocalized pixels showed that the changes in lipid raft association were not due to variations in M2 expression levels (Fig. (Fig.5B).5B). Infection of MDCK cells with M2-Helix virus showed only slight impairment of the association between M2 and lipid rafts (Fig. 5A and B), suggesting that this amphipathic helix region is not essential for lipid raft association.
Lipid raft association can also be demonstrated by the insolubility of a protein to 1% Triton X-100 extraction at 4°C. It was found that M2 expressed in M2-MDCK cells is largely soluble in Triton X-100 (Fig. (Fig.5C).5C). However, in influenza virus-infected cells, there was a consistent, time-dependent increase in the insolubility of M2 to Triton X-100 extraction for up to 18 h p.i., and this insolubility was slightly reduced in cells infected with the M2-Helix mutant virus (Fig. (Fig.5C5C).
Since previous results have shown that M2 is capable of binding cholesterol in vitro (54), we sought to determine if the localization of M2 to sites of virus budding, which are enriched in cholesterol-containing lipid raft domains, enables cholesterol binding via the M2 amphipathic helix. To investigate this further, M2 was immunoprecipitated from influenza virus-infected cells and from M2-MDCK cells, and the amount of bound cholesterol was determined. As shown in Fig. Fig.6,6, M2 expressed in virus-infected cells bound a significantly greater amount of cholesterol than did M2 expressed in M2-MDCK cells. This is consistent with our observation that M2 associates with lipid rafts only upon expression of other viral proteins during the course of viral infection (Fig. (Fig.5B).5B). Furthermore, cholesterol binding by M2 was significantly reduced in cells infected with the M2-Helix mutant virus (Fig. (Fig.6).6). This suggests that while disruption of the M2 amphipathic helix does not significantly alter its association with HA or lipid rafts, the M2 amphipathic helix is required for cholesterol binding. The observation that both the M2-Helix virus and wt influenza virus treated with MβCD are deficient in filament formation (Fig. 2I and J) suggests that there may be a link between M2-cholesterol binding and the ability of influenza virus to form filaments.
To further investigate the importance of cholesterol in viral filament formation, we treated virions with MβCD and analyzed the effects on morphology by electron microscopy. The wt influenza virus formed many filamentous virions, as well as some spherical particles (Fig. (Fig.7A).7A). MβCD treatment of virus appeared to cause a fragmentation of the filamentous virions, resulting in the generation of many spherical and amorphous particles (Fig. (Fig.7B),7B), raising the possibility that cholesterol binding by M2 is important for filamentous virion stability as well as formation. However, cholesterol extraction by MβCD may disrupt the structure of filamentous virions through an M2-independent mechanism.
We investigated if there is a connection between the MβCD-mediated disruption of filamentous virions and the effect of the M2 ectodomain-specific MAb 14C2 in reducing filament formation (27, 51). Similar to earlier observations (27), we found that treatment with MAb 14C2 for 1 h induced a loss of filaments from the surfaces of virus-infected cells (Fig. (Fig.8a)8a) compared to the case for untreated samples and a control treatment with the distinct M2 ectodomain-specific MAb 5C4, of the same IgG1 isotype as MAb 14C2 (Fig. (Fig.8a8a).
The above data do not allow us to discern whether MAb 14C2 treatment blocks the formation of filamentous particles or has a direct effect on preexisting filamentous particles. To test if MAb 14C2 affects the structure of influenza virus filaments, we treated filamentous virions with MAb 14C2 for 1 h. It was found that antibody treatment caused fragmentation of the filamentous virions (Fig. 8b and c), similar to that seen with MβCD treatment (Fig. (Fig.7B).7B). Importantly, treatment with the isotype control antibody, 5C4, had no effect on the virions (Fig. (Fig.8b),8b), showing that general cross-linking of M2 proteins is not sufficient for the fragmentation event to occur; rather, fragmentation is a specific consequence of MAb 14C2 binding to the M2 protein. This suggests that MAb 14C2 alters the M2 protein conformation in such a way as to mimic, though not necessarily cause, M2 cholesterol withdrawal, further illustrating the importance of M2 and cholesterol in filament formation and stability.
Nonetheless, MAb 14C2 treatment could also have other effects on the M2 protein, separate from its effects on filament formation. To investigate this further, we determined whether MAb 14C2 would affect M2 ion channel activity. It has been shown that MAb 14C2 reacts with M2 protein expressed at the surfaces of oocytes (47). M2 protein was expressed in oocytes of X. laevis, and two-electrode voltage clamp recordings were made as described previously (2-4, 47). Superfusing M2-expressing oocytes with MAb 14C2 did not induce proton flux through the closed M2 channel (pH 8.5), nor did MAb 14C2 treatment alter proton flux when the channel was subsequently activated at pH 5.5 (Fig. (Fig.9).9). These data are in agreement with the observation that treatment with the ion channel inhibitor amantadine had no effect on filament formation (Fig. (Fig.2L)2L) and suggest that the disruption of viral filaments by MAb 14C2 is not accomplished through modulation of ion channel activity.
To determine if MAb 14C2 disruption of filamentous virions had an effect on virus infectivity, filamentous virions were treated with MAb 14C2 for 1 h before titration in a plaque assay. It was found that fragmentation of filamentous virions by MAb 14C2 treatment had little to no effect on virus infectivity, whereas low-pH treatment led to a 2-log reduction in titer, consistent with the irreversible, premature triggering of HA to its low-pH-induced postfusion form, which is biologically inactive for virus entry into cells (10). Interestingly, these data indirectly support the hypothesis that filamentous virions contain only one copy of the viral genome (44). If the filamentous virions were multigenomic, then MAb 14C2-mediated fragmentation of one filamentous virion into multiple sphere-like particles should have increased the virus titer, but no change in virus titer was observed (Fig. (Fig.10).10). Overall, these data suggest that MAb 14C2 induces a specific conformational change in the M2 protein, resulting in the fragmentation of filamentous virions and the inhibition of their formation.
Influenza virus filament formation is a genetic trait, and it has been established clearly that it is linked to the M1 protein (5, 6, 19, 51). It can readily be envisaged that filament elongation would be dependent on M1 protein polymerization. A growing body of evidence also links the M2 protein to filament formation. Cells infected with the A/Udorn/72 strain of influenza virus produce copious amounts of filamentous virus in addition to spherical particles. Treatment of these infected cells with the M2 ectodomain-specific MAb 14C2 causes a loss of filament formation (27, 51) (Fig. (Fig.8).8). Furthermore, MAb 14C2 restricts the growth of some strains of influenza virus, e.g., A/Udorn/72, and viruses that escape the growth restriction contain mutations in the M2 cytoplasmic tail or the M1 protein (26, 63). Additional evidence for a role of M2 in filament formation comes from the finding that truncation of the A/Udorn/72 M2 protein at residue 70 leads to a loss of filaments on the surfaces of influenza virus-infected cells (36).
We show here that a three-alanine substitution at residues 71 to 73 in the M2 cytoplasmic tail is also effective at disrupting filament formation (Fig. (Fig.2C).2C). It has been speculated that this region of the M2 protein is involved in binding to the M1 protein and that this interaction is essential for filamentous virion formation (9, 36). Surprisingly, truncation of the M2 protein at amino acid 70 in the spherical strain A/WSN/33 resulted in the formation of filamentous virions (28, 36), though it has been speculated that the particles were not actual filamentous virions; rather, they may reflect a defect in scission during the budding process (36).
Although an interaction between M1 and M2 for virus formation was deduced from genetic studies (62), the nature of the M2 protein involvement is unknown. It is possible that the interaction between M1 and M2 is required for recruitment of M2 to sites of virus budding. Since the M2 protein is normally excluded from the lipid raft domains that serve as sites of budding, expression of other viral proteins in the context of virus infection may be necessary for the recruitment of M2 to sites of virus budding (Fig. (Fig.33 to to5).5). It is possible that M2 remains a peripheral raft protein, even at the sites of virus budding, but the close association with lipid rafts may be sufficient to allow for cholesterol binding by the M2 amphipathic helix (Fig. (Fig.6).6). We speculate that M1 mediates recruitment of M2 to lipid raft-enriched sites of virus budding, allowing for cholesterol binding by the M2 amphipathic helix, which may mediate a crucial step in viral filament formation and may explain the dependence of viral filament formation on both membrane cholesterol and the M2 amphipathic helix (Fig. 2I and J).
Many proteins, including several viral proteins, utilize an amphipathic helix to modify membrane curvature (17, 20, 21, 25, 39, 52, 55). It is possible that cholesterol binding by the M2 amphipathic helix may modify the conformation of the M2 protein in a manner that is essential for filament formation and stability, possibly through alterations in membrane curvature. Previous work has shown that cholesterol binding can alter the ability of an amphipathic helix to modify membrane curvature (16, 18). Cholesterol withdrawal by MβCD treatment (Fig. (Fig.7)7) may cause a change in the conformation of the M2 protein similar to the effects of MAb 14C2 binding (Fig. 8b and c), with subsequent changes in membrane curvature resulting in filament fragmentation and the prevention of filament formation during virus budding.
It is likely that the same MAb 14C2-induced conformational change in M2 that leads to the fragmentation of filamentous virions also leads to the loss of budding filaments from the surfaces of infected cells (Fig. (Fig.8);8); however, it is also possible that MAb 14C2 only fragments filamentous virions. Filamentous virions may remain capable of forming in the presence of MAb 14C2, but they would quickly be fragmented, leading to the loss of filaments on the surfaces of infected cells and the increase in released spherical particles that have been reported previously (27, 51). Further study will be necessary to determine whether conformational changes in the M2 protein have a role in virus budding as well as filament integrity.
The notion of conformational changes within the M2 amphipathic helix is supported by recent structural data on the M2 protein. It is known that low-pH activation of the M2 ion channel causes movement in the TM helices that is translated into the M2 cytoplasmic tail (43). The M2 ion channel is activated by low pH (47), and electron paramagnetic resonance (EPR) spectroscopy data have shown that this low-pH activation results in the clockwise rotation of the amphipathic helix and in deeper embedding of the helix in the plasma membrane (43). Similarly, MAb 14C2 binding to the ectodomain of M2 may induce a conformational change that is translated into the M2 amphipathic helix without affecting ion channel activity. A deeper embedding and rotation of the amphipathic helix in the plasma membrane may mimic changes that occur during M2 cholesterol withdrawal.
It has been shown that the presence of cholesterol can reduce the insertion of an amphipathic peptide into the plasma membrane (18). Thus, withdrawal of cholesterol from M2, just like MAb 14C2 triggering, may increase the insertion of the M2 amphipathic helix into the membrane, leading to an alteration of membrane curvature that may fragment the filamentous virions and prevent their formation. Conversely, M2-cholesterol binding during the course of virus budding may alter the local membrane curvature, perhaps stabilizing the site of budding to allow for M1 polymerization and filament elongation. Disruption of the M2 amphipathic helix in the M2-Helix virus inhibited viral filament formation (Fig. (Fig.2I).2I). It is possible that the deficiency in filament formation for this virus can be attributed to its inability to bind cholesterol (Fig. (Fig.6)6) and to the loss of subsequent alterations in membrane curvature mediated by the M2 amphipathic helix in the presence of cholesterol. Taken together, these data show that M2 is an essential player in the formation of filamentous virions, and we suggest that cholesterol-induced conformational changes within the M2 amphipathic helix may be able to modify membrane curvature, enabling viral filament formation.
We thank members of the Lamb laboratory for helpful discussions and critical readings of the manuscript. Electron microscopy was performed in the Northwestern University Biological Imaging Facility (Evanston Campus).
This research was supported in part by grant R01 AI-20201 from the National Institute of Allergy and Infectious Diseases. J.S.R. is an associate and R.A.L. is an investigator of the Howard Hughes Medical Institute.
Published ahead of print on 10 March 2010.