Ubiquitination of the Cytoplasmic Domain of Influenza A Virus M2 Protein Is Crucial for Production of Infectious Virus Particles

ABSTRACT

Virus replication is mediated by interactions between the virus and host. Here, we demonstrate that influenza A virus membrane protein 2 (M2) can be ubiquitinated. The lysine residue at position 78, which is located in the cytoplasmic domain of M2, is essential for M2 ubiquitination. An M2-K78R (Lys78→Arg78) mutant, which produces ubiquitination-deficient M2, showed a severe defect in the production of infectious virus particles. M2-K78R mutant progeny contained more hemagglutinin (HA) proteins, less viral RNAs, and less internal viral proteins, including M1 and NP, than the wild-type virus. Furthermore, most of the M2-K78R mutant viral particles lacked viral ribonucleoproteins upon examination by electron microscopy and exhibited slightly lower densities. We also found that mutant M2 colocalized with the M1 protein to a lesser extent than for the wild-type virus. These findings may account for the reduced incorporation of viral ribonucleoprotein into virions. By blocking the second round of virus infection, we showed that the M2 ubiquitination-defective mutant exhibited normal levels of virus replication during the first round of infection, thereby proving that M2 ubiquitination is involved in the virus production step. Finally, we found that the M2-K78R mutant virus induced autophagy and apoptosis earlier than did the wild-type virus. Collectively, these results suggest that M2 ubiquitination plays an important role in infectious virus production by coordinating the efficient packaging of the viral genome into virus particles and the timing of virus-induced cell death.

IMPORTANCE Annual epidemics and recurring pandemics of influenza viruses represent very high global health and economic burdens. The influenza virus M2 protein has been extensively studied for its important roles in virus replication, particularly in virus entry and release. Rimantadine, one of the most commonly used antiviral drugs, binds to the channel lumen near the N terminus of M2 proteins. However, viruses that are resistant to rimantadine have emerged. M2 undergoes several posttranslational modifications, such as phosphorylation and palmitoylation. Here, we reveal that ubiquitination mediates the functional role of M2. A ubiquitination-deficient M2 mutant predominately produced virus particles either lacking viral ribonucleoproteins or containing smaller amounts of internal viral components, resulting in lower infectivity. Our findings offer insights into the mechanism of influenza virus morphogenesis, particularly the functional role of M1-M2 interactions in viral particle assembly, and can be applied to the development of new influenza therapies.

INTRODUCTION

Influenza A virus (IAV) is an important pathogen that threatens human public health and the global economy. It is an enveloped RNA virus with eight segments of negative-sense RNA primarily encoding 11 to 12 viral proteins (1). At the beginning of infection, the viral hemagglutinin (HA) protein binds to sialic acid-containing receptors on the host cell surface, causing endocytosis of viral particles. The viral envelope then fuses with the membrane of the late endosome, leading to the release of viral ribonucleoprotein (vRNP) into the cytoplasm. Subsequently, vRNP is imported into the nucleus (2), where replication and transcription of the influenza virus genome occur. In the late stage of the viral life cycle, the viral envelope proteins HA and neuraminidase (NA) are targeted to lipid rafts of the cell membrane, causing the deformation of the membrane and the initiation of virus-budding events. In association with vRNP, matrix protein 1 (M1) then binds to the cytoplasmic tails of HA and NA. Finally, the membrane protein M2 causes membrane scission and the release of progeny virions (3).

M2 has been shown to play several important roles in the viral life cycle. During early stages of infection, M2 functions as an ion channel that allows virion acidification for efficient uncoating in the acidifying endosome, thereby assisting virus entry (4). In late infection stages, M2 not only manipulates programmed cell death (5) but also mediates virus morphogenesis, including virus assembly, budding, and release (6). In addition, M2 plays a key role in transporting proton ions out of the lumen of the Golgi apparatus, leading to pH neutralization of the trans-Golgi network (TGN) and preventing HA (which contains polybasic cleavage sites) from becoming fusogenic in the acidic TGN (7). Moreover, the localization of M2 to the Golgi apparatus can stimulate the NOD-like receptor protein 3 (NLRP3) inflammasome pathway (8). M2 is a 97-residue single-pass membrane protein and consists of an externally oriented N-terminal domain (ectodomain) (amino acids [aa] 1 to 24), a transmembrane domain (aa 25 to 43), and an internally oriented cytoplasmic tail (aa 44 to 97) (9). The transmembrane domain mediates the channel activity of M2 and is important for virus uncoating (4), the activation of inflammasomes (8), and the prevention of HA, which contains polybasic cleavage sites, from becoming fusogenic (7). The cytoplasmic tail of M2 plays important roles in the late stage of infection, particularly in virus assembly and release (10, 11). However, the mechanisms by which M2 regulates virus production remain unclear.

Ubiquitination is a highly versatile posttranslational modification that governs countless cellular processes and has been demonstrated to play important roles in the replication of various viruses (12, 13), including influenza A virus. Several studies have highlighted the significance of the cellular ubiquitin (Ub) system in influenza virus replication, particularly for virus entry (14–16). In the presence of MG132, a proteasome inhibitor, influenza virus was sequestered in endosomes, and as a consequence, its entry into the nucleus was blocked (14). Furthermore, a genome-wide RNA interference (RNAi) screen identified ubiquitination as a crucial biological process for the early stage of influenza virus replication (15). More direct roles of ubiquitin in influenza virus entry have been evidenced by the requirement for a ubiquitin-interacting protein, Epsin 1 (16), and an E3 ubiquitin ligase, ITCH (17). In addition, histone deacetylase 6 (HDAC6), a key component in ubiquitin-dependent aggresome formation and disassembly, plays an important role in virus uncoating via its ubiquitin-binding domain (18). Furthermore, the ubiquitination of viral nucleoprotein (NP) facilitates influenza virus RNA replication (19). Moreover, the existence of free ubiquitin in influenza virus particles has been demonstrated by proteomics analysis (20) and immunofluorescence staining (18). However, whether the cellular ubiquitin system plays any roles in the late stage of the influenza virus life cycle remains unknown.

In a preliminary study examining all IAV proteins, we found that M2 was ubiquitinated. In view of the possible contribution of the cellular ubiquitin system to influenza virus production, we focused on the ubiquitination of M2. In this study, we show that the M2 protein can be ubiquitinated at the lysine 78 residue located in the cytoplasmic domain. Based on the investigation of wild-type (WT) and M2-K78R (Lys78→Arg78) mutant viruses, we conclude that M2 ubiquitination plays important roles in the intricate control of the production of infectious influenza virus.

RESULTS

Influenza A virus M2 protein undergoes ubiquitination.

In our previous screening to detect potentially ubiquitinated IAV proteins, we found that, besides NP, M2 may also be ubiquitinated (data not shown). To further characterize this modification, we first used an in vivo ubiquitination assay to examine whether the influenza virus M2 protein could indeed be ubiquitinated. HEK293T cells were cotransfected with myc-tagged Ub (molecular mass, 9.5 kDa) and HA-tagged M2 (molecular mass, 17 kDa), and the lysates were immunoprecipitated with anti-HA agarose and then subjected to Western blot analysis. As shown in Fig. 1A (left), anti-myc antibody detected several proteins, with estimated molecular masses of around 26 kDa (corresponding to ≅ M2 + Ub), 35 kDa (≅ M2 + 2Ub, i.e., two ubiquitin molecules are conjugated to M2), and 45-kDa (≅ M2 + 3Ub), respectively, all migrating slower than HA-tagged M2. These results suggest that viral M2 might be modified by multiple monoubiquitin or polyubiquitin moieties. We confirmed this interpretation using anti-HA antibody, which detected similar migrating proteins, although the relative amounts of these bands were different (Fig. 1A, right). To demonstrate that the viral M2 protein was indeed ubiquitinated in virus-infected cells, human A549 lung carcinoma cells were infected with influenza A/WSN/33 virus and harvested at 10 h postinfection (hpi). The ubiquitinated proteins were isolated by using a UbiQapture-Q kit and further analyzed by Western blotting. As shown in Fig. 1B, ubiquitinated proteins could be detected from cells with or without influenza A virus infection (Fig. 1B, left). In contrast, modified M2 proteins that migrated at various distances in gels equivalent to those of 26-kDa and 35-kDa proteins could be detected only in influenza virus-infected cells and not in uninfected cells (Fig. 1B, right), suggesting that viral M2 is ubiquitinated during IAV replication.

FIG 1.

Influenza virus M2 undergoes ubiquitination. (A) HEK293T cells were cotransfected with plasmids expressing myc-tagged ubiquitin and HA-tagged M2 protein. The HA-tagged proteins were immunoprecipitated by HA agarose and subjected to Western blot analysis with anti-myc (left) or anti-HA (right) antibody. (B) A549 cells were infected with influenza A/WSN/33 virus (MOI = 10) and harvested at 10 h postinfection. The ubiquitinated proteins were pulled down and subjected to Western blot analysis with anti-Ub antibody (left). The same blot was then stripped and reprobed with anti-M2 antibody (right). The ubiquitinated M2 proteins are indicated by arrows. The upper bands (labeled *) may be caused by incomplete stripping or nonspecific signals. (C) HEK293T cells were cotransfected with the plasmids expressing HA-tagged M2 protein and various myc-tagged ubiquitins, including the WT and the K48R and K63R mutants. The HA-tagged proteins were immunoprecipitated by HA agarose and subjected to Western blot analysis with anti-myc (left) or anti-HA (right) antibody.

Ubiquitin contains seven lysine residues, each of which can potentially be used for isopeptide bond formation, resulting in different polyubiquitin chain linkages and different roles in various biological processes (21). Given that K48- and K63-linked ubiquitinations are the most commonly observed ubiquitin chains, we examined whether the M2 protein possessed either of these ubiquitin-linked chains. We performed an in vivo ubiquitination assay using myc-tagged Ub-K48R (Lys48→Arg48) and Ub-K63R (Lys63→Arg63) mutants as the source of ubiquitins. We found that the Ub-K48R and Ub-K63R mutants were still conjugated to the M2 protein, similar to WT ubiquitin (Fig. 1C), implying that M2 undergoes neither K48- nor K63-linked ubiquitination. Although we have not yet established the precise ubiquitin linkage type of M2, our results indicate that both ectopically expressed and influenza virus-produced M2 protein can be ubiquitinated.

The lysine 78 residue is essential for M2 ubiquitination.

To search for the potential ubiquitination site on M2, we analyzed the amino acid sequence of the influenza virus M2 protein using influenza virus alignment software (http://www.ncbi.nlm.nih.gov/genomes/FLU/Database/nph-select.cgi?go=alignment). Here, we first focused on the lysine residues of this protein, because lysine is regarded as the primary ubiquitination site of most ubiquitinated proteins (22). We found that four M2 lysine residues, Lys49, -56, -60, and -78, are highly conserved among the majority of H1N1 and H3N2 virus isolates (Fig. 2A). However, these lysine residues are mutated in some recent virus strains, and we discuss this further in Discussion. We replaced these four lysine residues individually or as a group with arginine residues. We then subjected these K-to-R mutants to in vivo ubiquitination assays to detect ubiquitinated M2. As expected, multiple ubiquitin moieties were detected in the WT M2 immunoprecipitates, whereas none of them were detected in the 4K-to-4R mutant (Fig. 2B). Strikingly, among the four single K-to-R mutants, only the M2-K78R mutant failed to be ubiquitinated, suggesting that lysine 78 is the essential site for M2 ubiquitination.

FIG 2.

Lysine 78 is essential for M2 ubiquitination. (A, top) Alignment of M2 protein sequences from influenza virus H1N1 and H3N2 strains. The lysine residues are indicated by stars. (Bottom) Schematic representation of the four conserved lysine residues in influenza virus M2 protein. a.a., amino acid. (B) HEK293T cells were cotransfected with plasmids expressing myc-tagged ubiquitin and various HA-tagged M2 proteins, including the WT and K-to-R mutants of M2, as indicated. The HA-tagged proteins were immunoprecipitated and subjected to Western blot analysis with anti-myc (left) or anti-HA (right) antibody. MW, molecular weight (in thousands).

The M2-K78R mutant produces defective virion particles.

To investigate the role of M2 ubiquitination in influenza A virus replication, we generated A/WSN/33 virus containing the M2-K78R mutation by a reverse-genetics approach using eight plasmids and determined the titer of virus released into the medium by a plaque assay (23). The titer of the M2-K78R mutant virus was about two times lower than that of the WT virus (data not shown). We then examined the physical properties and chemical compositions of the viruses. The same PFU of WT and M2-K78R mutant viruses produced from the recombinant plasmids were subjected to Western blot analysis to detect the M1 protein since M1 is the most abundant protein in virus particles. Surprisingly, the M2-K78R mutant virus contained 3-fold more M1 protein (Fig. 3A) than did the WT virus, implying that the mutant virus was less infectious than the WT virus. Therefore, the infectivities and physical attributes of the WT and mutant viruses are different. In most of the following experiments, we compared the properties of WT and mutant viruses at the same multiplicity of infection (MOI), so that more mutant viral particles were applied in the experiment to reach the same infectious viral titers as those of the WT virus.

FIG 3.

Characteristics of the M2-K78R mutant virus. WT and M2-K78R mutant viruses were generated by a reverse-genetics approach. (A) Different volumes of WT and M2-K78R viruses containing the same infectious units (determined by a plaque-forming assay) were used for Western blot analysis with anti-M1 antibody. (B to D) MDCK cells were infected with influenza A viruses (MOI = 0.1) for 24 h. (B) The supernatants were collected, and their titers were determined by a plaque assay. The same PFU of WT and M2-K78R mutant viruses were subjected to a hemagglutination assay. Diffuse red staining indicates a hemagglutination-positive test. (C) Virus particles were purified from equivalent volumes of the supernatant and resuspended in 200 μl of saline buffer. Resuspended virus particles (20 μl) were analyzed for protein composition by Western blotting. Amounts of detected proteins were quantified by using ImageJ software and normalized to those of the WT virus (as shown below the blots). Two independent experiments were performed, and a representative datum is shown. (D) Resuspended virus particles (100 μl) were used for isolation of vRNA. The vRNA levels of viral NP segments in virus particles were determined by quantitative PCR. The values represent the means ± SD of data from two independent experiments. ***, P value of <0.001 (D).

To gain further insight into the effect of the M2-K78R mutation on viral infectivity, we examined whether the amounts of viral particles produced from virus-infected cells were impacted by the M2-K78R mutation. We first attempted to increase the titer of the inoculating virus by infecting Madin-Darby canine kidney (MDCK) cells with WT and M2-K78R mutant viruses. The culture supernatants were collected at 24 hpi, and the virus titer was determined by a plaque assay. We then used the same PFU of WT and M2-K78R mutant viruses for a hemagglutination assay. Interestingly, the hemagglutination titer of the M2-K78R mutant virus was 4- to 8-fold higher than that of the WT virus (Fig. 3B). We next purified WT and mutant viral particles from equivalent volumes of the supernatants and studied their protein compositions. Compared to WT virions, M2-K78R mutant virions contained about twice the amount of HA protein (Fig. 3C). Interestingly, there were lower levels of NP, M1, and M2 proteins in M2-K78R mutant virions than in WT virions (Fig. 3C). Furthermore, the amount of viral RNA (vRNA) in M2-K78R mutant virions was about 20% of that in WT virions (Fig. 3D).

To further investigate the physical properties of WT and M2-K78R mutant virions, virus particles from the supernatant of infected cells were subjected to density gradient centrifugation analysis. To trace the distribution of the virus, the influenza virus M1 protein was used for Western blot analysis. As shown in Fig. 4A, the majority of the mutant virus was detected in fractions 7 to 10, whereas the WT virus was detected in fractions 8 to 11, suggesting that the WT virus has a higher density than that of the M2-K78R mutant virus. Influenza A viruses are enveloped, spherical or filamentous structures ranging from 80 to 120 nm in diameter. We next monitored the morphology of the virus particles by transmission electron microscopy. In cell cultures of both WT and M2-K78R mutant viruses, most of the A/WSN/33 virions released into the culture medium were spherical, although some filamentous virions were also observed (Fig. 4B). There was no difference in the ratios of different types of virus particles of the WT and mutant viruses. Interestingly, the electron-dense dots that represent the rod-like structures of vRNP within virions differed (Fig. 4C). WT virions contain eight dots, arranged as a central dot surrounded in a circle by seven others (24). In contrast, we found that many M2-K78R mutant virions lacked this vRNP structure. These “empty” cross sections might correspond to viral particles that did not incorporate vRNPs or might indicate that cross-sectioning had missed the vRNPs. To minimize possible artifacts due to the sectioning process, we assessed 500 cells from different cross sections. Around 80% of the WT virus particles contained vRNP dots, whereas <50% of M2-K78R mutant virions exhibited vRNP dots, implying that >50% of M2-K78R mutant virus particles lacked vRNP (Fig. 4D). Together with our analyses of the protein and RNA contents of the virions, these results suggest that the ubiquitination of M2 residue Lys78 facilitates the production of complete infectious viral particles.

FIG 4.

The M2-K78R mutant causes defective virus production. (A) WT and M2-K78R mutant viruses were used to infect MDCK cells (MOI = 0.1) for 24 h. Virus particles were purified from infected cell supernatants by ultracentrifugation and further fractionated by Opti-Prep density gradient ultracentrifugation. The distribution of M1 proteins was monitored by Western blotting (left) and further quantified by using ImageJ software (right). The respective amounts of M1 protein were analyzed by dividing the signal for each fraction by the total signal. Two independent experiments were performed, and a representative datum is shown. (B to D) WT and M2-K78R mutant viruses were used to infect A549 cells (MOI = 1). (B and C) At 24 hpi, virus-infected cells were processed for transmission electron microscopy under low magnification (B) and high magnification (C). (D) Virus containing electron-dense dots was considered normal virus, whereas virus lacking clear electron-dense dots was counted as “empty.” We examined 500 viral particles for this analysis. Values represent the means ± SD of data from two independent experiments.

Production of infectious virus particles is affected in the M2-K78R mutant.

The above-described experiments show that the M2-K78R mutation results in a defect in the production of virus particles. To confirm these findings, we next investigated the role of M2 ubiquitination in influenza A virus replication. WT and mutant viruses were used to infect A549 cells at an MOI of 1. Since M2 is an ion channel protein involved in virus uncoating (4), we initially examined whether the M2-K78R mutation affects this step. We assessed the distribution of the incoming vRNA in the cytoplasm and nucleus since the nuclear import of RNA immediately follows virus uncoating, so the nucleus/cytoplasm ratio of RNA can be used as an indicator of the efficiency of the uncoating process, assuming that subsequent cytoplasmic-nuclear import is not affected. Our results show that the ratios of cytosolic and nuclear vRNAs were similar between WT and M2-K78R viruses (Fig. 5A). As a control, we knocked down the ubiquitin ligase ITCH, which has been shown to be a cellular factor for virus uncoating (17), and found that it decreased the ratio of nuclear/cytoplasmic vRNA, confirming the feasibility of this assay. These results suggest that the M2-K78R mutation does not influence IAV entry steps up to uncoating and vRNA import into the nucleus (Fig. 5A).

FIG 5.

The M2-K78R mutation affects influenza A virus replication. Viruses were used to infect A549 cells at an MOI of 1 for the following experiments. (A) A549 cells were incubated with influenza A viruses for 30 min at an MOI of 5. The cytoplasmic and nuclear viral RNAs were isolated and used for qRT-PCR. The percentage of vRNA was calculated as the ratio of the vRNA from each compartment to the vRNA from whole cells. Values represent the means ± SD of data from three independent experiments. (B) Infected A549 cells were harvested at the indicated times, and their vRNA levels of NP segments were determined by quantitative PCR. Values represent the means ± SD of data from three independent experiments. ***, P value of <0.001. (C) Same as panel B except that 50 mM NH₄Cl was added at 3 hpi. Values represent the means ± SD of data from three independent experiments. (D) A multistep virus growth curve was established for A549 cells infected with influenza viruses. The progeny viruses were collected at the indicated times, and their titers were determined by a plaque assay. Values represent the means ± SD of data from three independent experiments.

We next examined the effect of the M2-K78R mutation on the transcription and replication of the viral genome by examining amounts of vRNA during virus infection. Up to 6 hpi, amounts of vRNA in M2-K78R mutant virus-infected cells were equal to or even slightly higher than those in WT virus-infected cells, implying that the early stage of replication is not affected by the M2-K78R mutation. The presence of slightly larger amounts of mutant vRNA may have been due to having to inoculate more mutant viral particles to compensate for the lower infectivity of the mutant virus (Fig. 3). Strikingly, from 12 hpi, mutant-infected cells exhibited lower levels of vRNAs than did WT-infected cells at every time point examined (Fig. 5B). Since the influenza virus produced from the infected cells could reinfect cells, the difference in the amounts of vRNA between the WT and mutant viruses may be due to this second-round infection. To assess this possibility, we restricted infection to a single cycle by adding an inhibitor of endosome acidification, ammonium chloride (25), at 3 hpi. In the presence of ammonium chloride, the amounts of vRNA were similar between WT- and M2-K78R mutant-infected cells throughout the assay period (Fig. 5C), confirming that the M2-K78R mutation does not affect the early stages of viral replication. Furthermore, these data strongly suggest that the increased amount of WT vRNA in the late stage of infection (Fig. 5B) was due to virus reinfection. Therefore, we conclude that the M2-K78R mutation impairs the production of infectious virus particles in the late stage of the viral life cycle, reducing reinfection by the virus particles produced. We confirmed this finding by examining the kinetics of infectious virus production. Titers of infectious viruses released into the medium were determined by plaque assays at various time points after infection. The M2-K78R mutant produced significantly lower titers of infectious virus at 12 hpi and thereafter (Fig. 5D). However, at 6 hpi or 3 hpi, mutant virus titers were similar to or even slightly higher than those of the WT virus (Fig. 5D and data not shown). Thus, most of the virus detected at the early time points most likely represented the residual inoculated virus. Notably, the kinetics curve of the virus titer is nearly flat for the M2-K78R mutant virus (Fig. 5D), indicating that the virus titer did not significantly increase throughout the infection course. Collectively, these findings suggest that the M2-K78R mutation does not affect viral entry or replication at the early stages of viral replication but limits the assembly or production of infectious virus particles at the late stages of its life cycle.

The M2-K78R mutation interferes with the interaction between M1 and M2.

The finding that M2-K78R mutant virions have less NP and vRNA (Fig. 3C and D) prompted us to examine whether the M2-K78R mutant may be defective in protein-protein interactions with M1 since it has been reported that M2 interacts with M1 (11) and that M1, in turn, interacts with NP in the form of the vRNP (26). Defects in the M1-M2 interaction would inhibit virus assembly and/or release. Therefore, we examined the subcellular distributions of M1 and M2 at various stages of the viral life cycle. M2 proteins of both the WT and M2-K78R mutant viruses were located primarily in the perinuclear region at 4 hpi, some of which were colocalized with M1 (Fig. 6A). Their colocalization coefficient revealed that less mutant M2 than WT M2 was colocalized with M1 at both 4 and 6 hpi. However, by 8 hpi, the colocalization coefficients of M1 and M2 became similar between the WT and M2-K78R mutant viruses (Fig. 6A and B). This result suggests that the direct or indirect M1-M2 interaction may be weaker in the mutant virus. Next, we investigated the interaction between M1 and M2 by a coimmunoprecipitation assay. A549 cells infected with the WT or M2-K78R mutant virus at the same MOI were immunoprecipitated with anti-M1 antibody. Expression levels of the intracellular M2 protein were similar between WT- and M2-K78R mutant-infected cells, but the amount of coimmunoprecipitated M2 was significantly smaller in the M2-K78R mutant-infected cells (Fig. 6C and D), suggesting that the direct or indirect interaction between the M1 and M2 proteins is reduced when M2 residue Lys78 is altered. This finding may partially explain why many M2-K78R virus particles failed to incorporate internal viral proteins and vRNAs.

FIG 6.

The M2-K78R mutation reduces the interaction of M2 with M1. (A and B) A549 cells were infected with WT and M2-K78R mutant viruses (MOI = 5) and harvested for immunofluorescence staining at the indicated times. M2 and M1 stainings are shown. Bars, 20 μm. (B) Colocalization coefficients of M1 and M2 signals were measured by using Zen 2011 microscope software. The colocalization coefficients from 30 cells were further analyzed by using Student's t test. **, P value of <0.01. (C and D) A549 cells were infected with WT and M2-K78R mutant viruses (MOI = 3) for 20 h and harvested for immunoprecipitation with anti-M1 antibody. (C) The input and immunoprecipitated (IP) proteins were subjected to Western blot analysis with anti-M1 or anti-M2 antibody. Three independent experiments were performed, and a representative datum is shown. The amounts of detected proteins were quantified by using ImageJ software and normalized to those of the WT virus. Values represent the means ± SD of data from three independent experiments. *, P value of <0.05. (D) Relative ratios of M2 in the immunoprecipitates determined from data in panel C.

The M2-K78R mutation promotes autophagy.

The above-described differences in the biological, physical, and chemical properties of M2 proteins between the WT and M2-K78R mutant viruses may impact virus-host interactions. Previous studies showed that influenza A virus infection can trigger the formation and accumulation of autophagosomes and that M2 is a critical viral protein regulating autophagy in host cells (5, 27). Therefore, we used the mammalian Atg8 homolog, microtubule-associated protein light chain 3 (LC3), to test if the ubiquitination of M2 plays a role in the induction of autophagy by assessing the distribution of green fluorescent protein-tagged LC3 (GFP-LC3) upon virus infection. A549 cell lines stably expressing GFP-LC3 were infected with the WT or M2-K78R mutant virus. At 6 hpi, cells infected with M2-K78R mutant viruses showed punctate or dot-like green fluorescent structures in the perinuclear region, representing autophagic vesicles, whereas WT virus-infected cells exhibited primarily a diffuse green fluorescent signal (Fig. 7A). M2 protein levels were similar between WT virus- and M2-K78R mutant virus-infected cells, highlighting that the M2 protein itself has the ability to induce autophagy and that this ability is affected by the M2-K78R mutation. These results suggest that the M2-K78R mutant virus triggers autophagy more easily and earlier than the WT virus. We further confirmed this result by assessing the levels of LC3-II, a marker protein for autophagic vesicle formation. It has been recognized that the amount of LC3-II relative to the amount of total LC3 reflects the degree of autophagosome formation. As expected, we found that LC3-II levels were increased for both WT and M2-K78R mutant viruses as the infection time was prolonged (Fig. 7B). However, notably, levels of LC3-II were 3-fold higher in M2-K78R mutant virus-infected cells than in WT-infected cells at 6 hpi (Fig. 7B). Moreover, it was previously demonstrated that M2 is necessary for blocking autophagosome fusion with lysosomes (5). Using LysoTracker Red, a red fluorescent dye for labeling lysosomes, we observed only a very limited overlap between GFP-LC3 and LysoTracker staining in WT virus- and M2-K78R mutant virus-infected cells (Fig. 7C), implying that both viruses prevented the fusion of autophagosomes with lysosomes. Taken together, these data suggest that the ubiquitination of the Lys78 residue of M2 may accelerate the induction of autophagy, but it does not affect the ability of M2 to block the completion of autophagy.

FIG 7.

The M2-K78R mutation promotes autophagy upon virus infection. (A) GFP-LC3/A549 cells that stably expressed GFP-LC3 were infected with WT or M2-K78R mutant viruses (MOI = 1) and harvested for indirect-immunofluorescence analysis at 6 hpi. The GFP-LC3 signal, M2, and DAPI staining of nuclei are shown. Bars, 20 μm. (B) A549 cells were infected with WT or M2-K78R mutant viruses (MOI = 1) and harvested at the indicated times. The cell lysates were used for Western blot analysis using anti-LC3 and antiactin antibodies. Band intensities were quantified, and the relative ratios of LC3-II/LC3-I plus LC3-II are shown below the blots. (C) Conditions similar to those described above for panel A except that the cells were treated with LysoTracker Red at 10 hpi and harvested at 12 hpi. The GFP-LC3 signal, LysoTracker Red, and DAPI staining of nuclei are shown. Bars, 10 μm.

M2-K78R mutant virus is more susceptible to induction of apoptosis.

It has been demonstrated that the M2 protein induces incomplete autophagy, which is related to cell death (5). Given that the M2-K78R mutant virus induced autophagy earlier than the WT virus, we next investigated whether the ubiquitination of M2 affects the susceptibility of influenza virus-infected cells to undergoing apoptosis. In general, caspase-3 activity serves as a measure of early apoptosis, whereas DNA fragmentation serves as an indicator of late apoptosis events. As expected, cleaved caspase-3 was detected at 18 hpi, i.e., during the late stage of virus infection (Fig. 8A). Interestingly, levels of cleaved caspase-3 were higher in M2-K78R mutant virus-infected cells than in WT virus-infected cells. To further verify that the M2-K78R mutant virus has a greater potential to induce apoptosis, we applied the fluorometric terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling (TUNEL) assay. As shown in Fig. 8B, influenza virus-induced apoptosis was more pronounced in M2-K78R mutant virus-infected cells than in WT virus-infected cells. In addition, the viability of M2-K78R mutant virus-infected cells significantly declined by 24 hpi (Fig. 8C). Collectively, these data demonstrate that the M2-K78R mutation enhances early and late apoptosis, consequently leading to more cell death.

FIG 8.

The M2-K78R mutation facilitates apoptosis upon virus infection. (A) A549 cells were infected with WT or M2-K78R mutant viruses (MOI = 1) and harvested at the indicated times. The cell lysates were used for Western blot analysis using anti-caspase-3 and anti-GAPDH antibodies. Procaspase-3 and cleaved caspase-3 are indicated by arrows. (B and C) A549 cells were infected with WT or M2-K78R mutant viruses (MOI = 0.01) for 24 h and then subjected to a TUNEL assay (B) and an MTS assay (C). (B) TUNEL-positive cells (green) and DAPI staining of nuclei (blue). Bars, 25 μm. (C) Cell viability, as measured by an MTS assay. The viability of uninfected cells was defined as 100%. The ratio between WT- and M2-K78R mutant-infected cells was determined. Values represent the means ± SD of data from three independent experiments. **, P value of <0.01.

DISCUSSION

Posttranslational modifications of M2 have been extensively investigated. M2 can be posttranslationally modified by the formation of intermolecular disulfide bonds at cysteine residues 17 and 19 (28), phosphorylation predominantly at serine residue 64, and acylation (particularly palmitoylation) at cysteine residue 50 (29). However, the functional roles of these posttranslational modifications remain unclear (29). In this study, we reveal an additional type of posttranslational modification of M2 and demonstrate its functional role in influenza A virus replication through the regulation of viral particle formation and host cell apoptosis.

Viruses can subvert or exploit the host cellular ubiquitination machinery in different phases of their replication cycle, among which viral budding and release are most relevant. Highly conserved L domains, containing either PT/SAP, PPXY, or YPXL/LXXLF, have been found in the structural proteins of most RNA-enveloped viruses such as retroviruses, rhabdoviruses, filoviruses, arenaviruses, and paramyxoviruses (30). The L domains of Gag proteins of variant retroviruses are responsible for Gag ubiquitination and are exploited by viruses to efficiently execute budding and egress (31, 32). However, these short proline-rich motifs do not exist in the influenza virus M2 protein. Nevertheless, we show that M2 can be ubiquitinated and that this modification is important for the formation of infectious particles.

It has been demonstrated that a C-terminal 28-amino-acid truncation mutant of M2 produced a less infectious influenza virus (10). Although the mutant virus particles were similar to the WT virus in morphology, the mutant virions contained reduced amounts of vRNP (10). The region of the M2 cytoplasmic tail important for the interaction with M1 and virus assembly was mapped to residues 71 to 76 of M2 (11), which is very close to the ubiquitination-related residue (M2-K78) described here. Thus far, M2 structures have been solved by X-ray crystallography and nuclear magnetic resonance (NMR). However, conformations of the C-terminal amphipathic helices in these structures are very different, probably due to environmental sensitivities (33). The structural flexibility of the C-terminal domain allows the possibility that the ubiquitination of M2 residue Lys78 may cause a conformational change in the domain, thereby facilitating the interaction between M1 and M2 and allowing the correct incorporation of vRNP into progeny virions. Notably, residue K78 is highly conserved among the M2 proteins of historic H1N1 and H3N2 isolates, but in some recent pandemic H1N1 strains (e.g., A/New York/18/2009), residue 78 is changed from Lys to Gln (Fig. 2A), which is likely not ubiquitinated. There are other mutations in M2 apart from residue 78 in recent H1N1 isolates. A recent study indicated that a lethal mutation in the M2 cytoplasmic tail can be compensated for by mutations in the M1 protein to restore the production of infectious virus (34). Together, the K78 mutation plus other mutations may affect virus replication.

A recent study highlighted that influenza virus vRNAs are not exported individually from the nucleus but rather are assembled en route to the plasma membrane via dynamic colocalization events (35). To maintain the integrity of progeny virions, influenza virus has been shown to undergo selective packaging, which might take place during cytoplasmic trafficking and involve recycling endosomes and an intact microtubule network (36) or during budding of the viral particles (37). During virus infection, M2 has been demonstrated to be distributed in the Golgi apparatus during early infection and to be incorporated into apical vesicles and finally to localize at the plasma membrane (38). It would be interesting to explore the cellular location of M2-mediated virus packing.

Influenza virus NS1, HA, and M2 proteins are involved in the induction of autophagy in host cells (5, 39). Among these proteins, M2 is emerging as a critical protein regulating autophagy. M2 is able to block the fusion of autophagosomes with lysosomes (5), and it promotes LC3 relocalization to the plasma membrane (40). Our study shows that the M2-K78R mutant virus, like the WT virus, can prevent autophagosomes from fusing with lysosomes (Fig. 7C), suggesting that the ubiquitination of M2 is not required for blocking late-stage autophagy. In fact, a recent finding revealed that the proton channel activity of M2 is responsible for blocking the fusion of autophagosomes with lysosomes (41).

As for most viruses, influenza virus-mediated apoptosis is a complicated process. During early infection, the viral NS1 protein acts as an antiapoptotic agent to counteract apoptosis (38). In a later stage of infection, the translocation of fragment 2 of the polymerase basic 1 (PB1-F2) protein to the mitochondrial membrane activates the intrinsic apoptotic pathway (39, 40). Additionally, M1 binds to heat shock protein 70 (Hsp70) to reduce the interaction between Hsp70 and apoptosis protease-activating factor 1 (Apaf-1), thereby facilitating apoptosome formation (41). M2 also induces incomplete autophagy, leading to cell death (5). Intriguingly, we found that the M2-K78R mutant virus induced autophagy and apoptosis earlier than did the WT virus (Fig. 7 and 8). Since membrane alterations can trigger a variety of cell signaling pathways, including apoptosis, we propose that the M2-K78R mutant virus may promote apoptosis signaling by inadequate localization on the plasma membrane. It will be important to investigate further how the ubiquitination of M2 regulates viral pathogenesis.

MATERIALS AND METHODS

Cell culture.

Human lung adenocarcinoma epithelial cells (A549) were maintained in F-12K medium (Gibco) supplemented with 10% fetal bovine serum (FBS) (HyClone) and antibiotics (100 U/ml penicillin G and 100 μg/ml streptomycin). MDCK cells and human embryonic kidney cells (HEK293T) were maintained in Dulbecco's modified Eagle's medium (DMEM) (Gibco) supplemented with 10% FBS and antibiotics (100 U/ml penicillin G and 100 μg/ml streptomycin). All cells were maintained in a humidified incubator at 37°C with 5% CO₂. GFP-LC3/A549 cells that stably expressed GFP-LC3 were produced by the transduction of cells with lentiviruses expressing GFP-LC3 (42). Positive clones were selected in the presence of 10 μg/ml blasticidin and maintained in the presence of 4 μg/ml blasticidin.

Plasmids and viruses.

The eight plasmids used to generate influenza A/WSN/33 virus were described previously (43). pCAG2-M2-HA, expressing HA-tagged M2, and pUI-myc-Ub, expressing myc-tagged ubiquitin (19), were used for in vivo ubiquitination assays. To introduce site-specific mutations, an overlapping-PCR method (44) was used for the construction of M2 mutants, including K49R, K56R, K60R, and K78R M2 mutants. QuikChange site-directed mutagenesis (Stratagene, La Jolla, CA) was used for myc-tagged ubiquitin mutants, including Ub-K48R and Ub-K63R. Recombinant viruses were generated by reverse genetics according to a previously reported protocol (23). Briefly, HEK293T cells and MDCK cells were cocultured and transfected with eight plasmids expressing 10 viral proteins and 8 vRNAs. The supernatants containing the viruses were harvested at around 90 h posttransfection and frozen at −80°C until use.

Plaque assay.

MDCK cells were infected with 10-fold serial dilutions of IAV for 1 h, washed twice with phosphate-buffered saline (PBS), and then overlaid with 0.5% agarose-containing minimum essential medium alpha (MEM-alpha). After 2 days, the cells were fixed with 10% formaldehyde and stained with a 0.1% crystal violet solution to calculate the number of plaques.

Antibodies and reagents.

Anti-HA and antiactin antibodies were purchased from Millipore (catalog numbers 04-902 and MAB1501). Anti-IAV HA, NP, and M2 antibodies were purchased from GeneTex (catalog numbers GTX127357, GTX629633, and GTX125951). Anti-IAV M1 antibody was obtained from AbD Serotec (catalog number MCA401). Anti-caspase-3 and anti-glyceraldehyde-3-phosphate dehydrogenase (GAPDH) antibodies were purchased from GeneTex (catalog numbers GTX110543 and GTX627408). Anti-LC3 antibodies were purchased from Abgent (catalog number AP1802a). Anti-myc antibody was obtained from Santa Cruz Biotechnology (catalog number sc-789). All Alexa Fluor-conjugated secondary antibodies used for immunofluorescence were procured from Molecular Probes (Invitrogen). 4′,6′-Diamidino-2-phenylindole dihydrochloride (DAPI) was purchased from Sigma-Aldrich. Anti-HA agarose, M-PER mammalian protein extraction reagent, and the Cyto-to-Nuc isolation kit were obtained from Thermo Scientific.

In vivo ubiquitination.

HEK293T cells were transfected with pCAG2-M2-HA and pUI-myc-Ub for the expression of HA-tagged M2 and myc-tagged Ub. At 48 h posttransfection, cells were harvested by using M-PER mammalian protein extraction reagent containing protease inhibitors and 5 μM N-ethylmaleimide (NEM) (a deubiquitinase inhibitor) and centrifuged to remove the insoluble fraction. The clarified supernatant was subjected to immunoprecipitation with an anti-HA affinity matrix at 4°C for 4 h. The matrix was washed three times with wash buffer (50 mM Tris [pH 7.5], 150 mM NaCl, 0.5% Triton X-100, and 0.05% SDS) and then subjected to Western blot analysis. To detect the intrinsic ubiquitination of M2, 300 μg of IAV-infected A549 cell lysates was incubated with UbiQuapture-Q matrix (Enzo Life Sciences) at 4°C for 4 h. The matrix was washed three times with PBS and then subjected to Western blot analysis.

Hemagglutination assay.

Red blood cells (RBCs) were prepared from fresh human blood, washed with PBS, and diluted in PBS to make a working solution of 0.75% RBCs in PBS. We added 25 μl of serially diluted viruses to a round-bottomed 96-well plate and then added 25 μl of 0.75% RBCs to each well. After mixing gently, the plate was maintained at room temperature for 30 to 60 min. In the absence of influenza virus, a dot appears in the center of the well. In the presence of the virus, a uniform reddish color forms across the well, indicating hemagglutination of red blood cells.

Virus purification and isolation.

Growth medium from one T75 flask of infected cells at 24 hpi was collected and spun at 2,200 × g for 10 min to remove cell debris. For virus purification, viruses were pelleted by ultracentrifugation at 54,000 × g for 2 h at 4°C and resuspended in NTC (100 mM NaCl, 20 mM Tris-HCl [pH 7.4], 5 mM CaCl₂). For virus isolation, viruses were pelleted by ultracentrifugation at 25,000 rpm for 3 h at 4°C and then resuspended in 350 μl homogenization buffer (HB) (0.25 M sucrose, 1 mM EDTA, 10 mM Tris-HCl [pH 7.4]) with a protease inhibitor and adjusted to a concentration of 40% OptiPrep by using 60% OptiPrep (Sigma-Aldrich). Diluted virus (1 ml) was transferred to an ultracentrifuge tube (Beckman) (total volume of 12 ml) and overlaid with a sequence of 1 ml 30%, 2 ml 20%, 2 ml 15%, 2 ml 10%, and 2 ml 5% OptiPrep and finally 1 ml HB at the top, which was centrifuged at 30,000 rpm for 16 h at 4°C. Following centrifugation, 1-ml samples were collected from top to bottom and numbered as fractions 1 to 11.

Apoptosis assay.

To perform the TUNEL assay, we used an In Situ Cell Death Detection kit (Roche). Briefly, the cells were fixed with 4% paraformaldehyde for 20 min and then permeabilized with 0.5% Triton X-100 for 5 min. The cells were then incubated with 100 μl of a TUNEL reaction mixture for 60 min at 37°C and washed with PBS. Images were taken by using a confocal microscope. To detect cell viability, cells were seeded onto 96-well plates. After virus infection, the cells were incubated with 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium (MTS) (Promega) for 10 to 30 min, and the absorbance at 490 nm was recorded by using a plate reader.

Quantitative reverse transcription-PCR.

Total cellular RNA was extracted by using a High Pure RNA isolation kit (Roche Diagnostics) according to the manufacturer's protocol. cDNA was synthesized by using the SuperScript III first-strand synthesis system (Invitrogen). The primers for reverse transcription of mRNA and vRNA were oligo(dT)₂₀ and the IAV-specific reverse transcription (RT) primer uni-12 (5′-AGCAAAAGCAGG-3′), respectively. We followed the standard TaqMan method with the Universal Probe Library system (Roche) for quantitative RT-PCR (qRT-PCR) analysis. GAPDH was used as a normalization control for cellular mRNA and intracellular viral RNA. Primers and probes were as follows: sense primer 5′-GATGGAGACTGATGGAGAACG-3′ and antisense primer 5′-TCATTTTTCCGACAGATGCTC-3′ with Universal Probe 59 for the IAV_NP segment and sense primer 5′-AGCCACATCGCTCAGACAC-3′ and antisense primer 5′-GCCCAATACGACCAAATCC-3′ with Universal Probe 60 for GAPDH.

Transmission electron microscopy.

All procedures were done at 4°C or on ice. Cells were cultured on Aclar embedding film (Electron Microscopy Sciences) and then rinsed with 0.1 M cacodylate buffer (0.1 M sodium cacodylate, 3.4% sucrose [pH 7.4]), fixed with 2.5% glutaraldehyde in 0.1 M cacodylate, washed with 0.1 M cacodylate buffer, postfixed in 1% osmium tetroxide, prestained with 1% uranyl acetate, dehydrated in a graded series of ethanol concentrations, and embedded in Spurr's resin. We then cut 60-nm sections, which we stained with 5% uranyl acetate–50% methanol, before viewing them on a Tecnai G2 Spirit Twin transmission electron microscope (FEI Company) with a Gatan 794 MultiScan charge-coupled-device (CCD) camera.

Statistical analysis.

Statistical significances were determined by using conventional Student's t test. Most of the assays are representative of data from three independent experiments. Data are shown as means ± standard deviations (SD). A P value of <0.05 was considered statistically significant.