Influenza A virus is one of the most important causes of respiratory infection. During viral infection, multiple cell signaling cascades are activated, resulting in the production of antiviral cytokines and initiation of programmed cell death of virus-infected cells. In the present study, we have used subcellular proteomics to reveal the host response to influenza A infection at the protein level in human macrophages. Macrophages were infected with influenza A virus, after which the cytosolic and mitochondrial cell fractions were prepared and analyzed by using two-dimensional electrophoresis for protein separation and mass spectrometry for protein identification. In cytosolic proteomes, the level of several heat shock proteins and fragments of cytoskeletal proteins was clearly up-regulated during influenza A virus infection. In mitochondrial proteomes, simultaneously with the expression of viral proteins, the level of intact actin and tubulin was highly up-regulated. This was followed by translocation of the components of antiviral RNA recognition machinery, including RIG-I (retinoic acid-inducible protein I), TRADD (TNFR1-associated death domain protein), TRIM25 (tripartite motif protein 25), and IKKε (inducible IκB kinase), onto the mitochondria. Cytochalasin D, a potent inhibitor of actin polymerization, clearly inhibited influenza A virus-induced expression of IFN-β, IL-29, and TNF-α, suggesting that intact actin cytoskeleton structure is crucial for proper activation of antiviral response. At late phases of infection mitochondrial fragmentation of actin was seen, indicating that actin fragments, fractins, are involved in disruption of mitochondrial membranes during apoptosis of virus-infected cells. In conclusion, our results suggest that actin network interacts with mitochondria to regulate both antiviral and cell death signals during influenza A virus infection.
The innate immune system has a critical role in recognizing virus infection and mounting antiviral responses. One of the most important innate immune effector cell types is macrophages. They detect viral infection by recognizing single- or double-stranded viral RNA through so-called pattern recognition receptors (PRRs).3 Receptor ligation results in the activation of signaling pathways leading to the production of type I IFN and inflammatory cytokines (1). Secreted IFN-α/β induces expression of IFN-stimulated genes that are important for the elimination of viruses (2). Additionally, a programmed cell death, apoptosis, is induced in infected cells by yet undefined signaling mechanism to restrict viral infection.
Influenza A virus is one of the most important causes of respiratory infection, and the virus primarily targets epithelial cells of the respiratory tract. However, it also infects macrophages and dendritic cells (3). Host cells detect the incoming influenza virus by recognizing viruses’ single-stranded genomic RNA by cytoplasmic RNA helicase retinoic acid-inducible protein I (RIG-I) (4, 5). The activity of ubiquitin ligase TRIM25 (tripartite motif protein 25) is necessary for RIG-I interaction with its downstream adaptor molecule, mitochondrial antiviral signaling protein (MAVS), also known as VISA, Cardif, and IPS-1 (6). This interaction stimulates signaling through inducible IκB kinase (IKKε) and TANK-binding kinase 1 to IFN regulatory factor 3 (IRF-3) and NF-κB transcription factors, respectively, culminating in the synthesis of type I IFNs and IFN-stimulated genes (7). The adaptor protein TRADD (TNFR1-associated death domain protein) is a multifunctional protein, which is an essential component of proinflammatory TNFR1 signaling (8). A very recent report showed that TRADD is also important for RIG-I-mediated signal transduction by forming a complex with MAVS and other components present in the RIG-I/MAVS signaling pathway (9).
MAVS localizes to mitochondria via its C-terminal hydrophobic region. Functional studies have shown that MAVS localization on mitochondrial membrane is required for signal transduction (10) providing a link between mitochondria and the innate immune defense system. Mitochondria are the major organelles involved in the apoptosis (11), and the role of mitochondria in controlling downstream apoptotic events is relatively well characterized. However, the mechanism of viral infection-induced apoptosis remains largely unknown. In the present study, we have characterized the mitochondrial and cytosolic proteomes of human primary macrophages upon influenza A virus infection to gain novel insight into cell signaling mechanisms involved in viral infection. We show that the level and/or subcellular location of several key proteins involved in cell death as well as in antiviral and stress response are regulated during influenza A virus infection of human primary macrophages. Our results also highlight the importance of mitochondria in the regulation of both antiviral and cell death signals during influenza A virus infection.
Materials and Methods
Abs and reagents
The mAb against TRIM25 was obtained from BD Transduction Laboratories. General caspase inhibitor z-VAD-FMK as well as IKKε, voltage-dependent anion channel-1, and actin Abs were purchased from Santa Cruz Biotechnology. The Ab against caspase-3 was obtained from Cell Signaling Technology. IL-18, RIG-I, MAVS, and influenza A virus H3N2 Abs have been previously described (12, 13, 14, 15, 16) and they were kindly provided by Dr. Ilkka Julkunen. MDA-5 Ab (17) was a generous gift form Dr. Paul B. Fisher. Cytochalasin D was purchased from Sigma-Aldrich.
Macrophages and cell stimulation
Human macrophages were obtained from leukocyte-rich buffy coats from healthy blood donors (Finnish Red Cross Blood Transfusion Service, Helsinki, Finland). Monocytes were isolated as described previously (12
Influenza A strain Udorn/72 (H3N2) was grown in 11-day-old embryonated eggs as previously described (12). The virus stock had a hemagglutination titer of 256 when a standard method was used (18). Macrophages were infected with the virus (1/50 dilution) in 6-well plates in 1 ml of complete Macrophage-SFM medium for the times indicated. The virus dose used has been evaluated in previous studies and has resulted in maximal induction of viral proteins at 18 h after infection (12, 16) (data not shown).
Isolation of mitochondrial and cytoplasmic cell fractions
The mitochondrial and cytoplasmic cell fractions were isolated by a Qproteome mitochondria isolation kit (Qiagen) according to the manufacturer’s instructions. Cytoplasmic fractions were further purified with a 2-D Clean-Up kit (GE Healthcare). For isolation, ∼10 × 106 macrophages were used. The purity of the subcellular fractions was confirmed with known mitochondrial marker proteins voltage-dependent anion channel-1 and MAVS.
Two-dimensional electrophoresis (2-DE) and protein identification
The first dimension of 2-DE was performed with 11-cm (pH 4–7) immobilized pH gradient strips (Bio-Rad), and the second dimension was performed with Criterion Tris-HCl 8–16% gradient precast gels (Bio-Rad). Isolated mitochondrial and cytoplasmic fractions from influenza A virus-infected and untreated macrophages were dissolved in 200 μl of rehydration buffer (7 M urea, 2 M thiourea, 3.5% CHAPS, 0.6% DTT, 0.5% immobilized pH gradient buffer (pH 4–7)), and the proteins were absorbed into immobilized pH gradient strips for 24 h at room temperature. Isoelectric focusing to a total of 55 kVh was done at 20°C, and focused strips were equilibrated twice at room temperature for 15 min with SDS-equilibration buffer containing 6 M urea, 50 mM Tris-HCl (pH 6.8), 2% SDS, 20% glycerol, and 10 mg/ml DTT for the first equilibration, and 25 mg/ml iodoacetamide for the second equilibration. SDS-PAGE was run at 200 V using a Criterion electrophoresis unit (Bio-Rad) at 4°C. Proteins were detected with mass spectrometry-compatible silver staining (19). Only those protein spots that were reproducibly clearly up- or down-regulated in at least three independent experiments were chosen for identification.
For identification, the proteins were in-gel digested with trypsin, and the resulting peptides were analyzed by peptide mass fingerprinting (PMF) as previously described (20, 21). The mass spectra were acquired using an Ultraflex TOF/TOF instrument (Bruker Daltonics) in positive ion reflector mode. Database searches were done with publicly available Mascot search engine against NCBInr database (Matrix Science). The search criteria were: human-specific taxonomy; trypsin digestion with one missed cleavage allowed; carbamidomethyl modification of cysteine as a fixed modification and oxidation of methionine as a variable modification; peptide tolerance maximum ±50 ppm. All of the protein identification scores were significant (p < 0.05).
Isolated mitochondria and cytoplasmic fractions and whole cell lysates from influenza A virus-infected and untreated macrophages were dissolved in 100 μl of Laemmli sample buffer. Samples (5 μl) were run on SDS-PAGE, and proteins were visualized with silver staining. For Western blotting an equal protein loading was determined from the silver-stained gel. Proteins from each fraction were separated with SDS-PAGE and transferred onto polyvinylidene difluoride membrane. Membranes were blocked with 5% nonfat milk and stained with different Abs overnight and detected by ECL.
Immunofluorescence staining and confocal microscopy
For indirect immunofluorescence staining, human macrophages were grown on coverslips. After 12 h of infection, cells were incubated with 1 μM MitoTracker Red580 (Molecular Probes) for 30 min at 37°C (cell incubator) at 5% CO2. The cells were fixed with 4% paraformaldehyde in PBS and permeabilized with ice-cold acetone for 5 min. The aldehyde groups were quenched with 50 mM NH4Cl for 20 min. Cells were treated with specific Abs for 1 h at room temperature followed by treatment with secondary, species-specific Abs conjugated with Alexa 488 (Molecular Probes) for 1 h. The samples were mounted in Mowiol and viewed under a Leica TCS SP5 confocal microscopy. A HCX APO ×63/1.30 (glycerol) objective was used and images were processed using the LAS AF program (Leica Application Suite Advanced Fluorescence) and Adobe Photoshop.
RNA isolation and cDNA synthesis
Quantitative real-time PCR and data analysis
CT) of a sample was selected according to the manufacturer’s guidelines. Relative units were calculated by a comparative CT method. First, the difference between 18S rRNA CT and the corresponding target cytokine CT of a sample was calculated to get ΔCT. To get the calibrator ΔCT, the average of 18S rRNA CTs of samples was subtracted from the calibrator CT value of 40, which was obtained from NTC. Next, the calibrator ΔCT was subtracted from the ΔCT of each experimental sample to obtain ΔΔCT. Finally, the amount of target normalized to an endogenous control, which was relative to the NTC calibrator, was calculated by the equation 2−ΔΔCT.
Proteome analysis of influenza A virus-infected human macrophages reveals up-regulation of apoptosis and cell stress-related proteins
During viral infection, the host cell activates multiple signaling cascades that lead to the production of IFNs and other cytokines and often to initiation of apoptosis. Here, we have used subcellular proteomics to identify influenza A virus target proteins in human primary macrophages. Macrophages were infected with influenza A virus for 18 h or left untreated, and cells were fractionated into cytosolic and mitochondrial fractions. Proteins in these fractions were separated by 2-DE and detected by silver staining. A representative 2-DE gel image from the cytoplasmic cell fraction of influenza A virus-infected macrophages, as well as the mitochondrial proteomes of control and infected macrophages, are shown in Fig. 1⇓, A and B, respectively. The differentially regulated protein spots are marked to the images.
Comparison analysis of the 2-DE gels revealed 17 clearly up-regulated and 8 down-regulated protein spots in the cytoplasmic proteome of influenza A virus-infected cells (Fig. 1⇑A). The differentially regulated proteins were cut out from the gels, subjected to in-gel tryptic digestion, and identified using PMF. The identification results are shown in Table I⇓. The up-regulated proteins identified from infected cells include cytoskeletal proteins (vimentin, cytoplasmic actin, gelsolin), proteins involved in signal transduction and antiviral response (annexin V, annexin IV, MxA (myxovirus resistance A)), as well as several proteins involved in stress response (calreticulin, BiP (heat shock 70-kDa protein 5, GRP78), heat shock protein (HSP)60, HSP90, glucose-regulated protein (GRP)94). The estimated molecular masses for cytoskeletal proteins vimentin, actin, and gelsolin in our 2-DE gels are ∼18–26, 30, and 45 kDa, respectively, whereas the theoretical molecular masses for these proteins are larger (vimentin, 54 kDa; actin, 42 kDa; gelsolin, 80 kDa). With PMF analysis we found peptides matching only to C-terminal parts of vimentin and gelsolin, and to an internal part of actin. Vimentin was also identified from two spots with peptides matching to aa 146–390 and 101–282, respectively (internal fragments). This shows that these proteins are fragmented upon influenza A virus infection. Interestingly, actin, gelsolin, and vimentin are known caspase-3 cleavage targets (22).
In cytoplasmic proteomes, cathepsins B and D were down-regulated upon influenza A virus infection. The mature cathepsin D consists of L and H chain, with a theoretical molecular mass of 45 kDa. With PMF we found only peptides matching to the H chain of cathepsin D, and also the estimated molecular mass for this spot in the 2-DE gel corresponds to the H chain of cathepsin D. Additionally, the level of Rho GDP dissociation inhibitor 2 (Ly-GDI) was clearly down-regulated upon virus infection. Ly-GDI is a known target protein of caspase-3 (23, 24), and this further indicates that apoptotic caspase cascade is activated in influenza A virus-infected macrophages.
In mitochondrial proteomes, actin and its fragments were highly up-regulated upon influenza A virus infection (Fig. 1⇑B and Table I⇑). Additionally, we found fragmention of microtubules (tubulins), gelsolin, and the HSP90 chaperone GRP94, and mitochondrial translocation of these fragments during infection (Fig. 1⇑B and Table I⇑). To further characterize actin translocation during influenza A virus infection, we performed 2-DE Western blot analysis with actin Ab against the C-terminal part of human actin. There was a strong up-regulation of intact 42-kDa actin expression in mitochondria after influenza A virus infection (Fig. 1⇑C). Additionally, actin fragmentation was clearly seen in 2-DE Western blot analysis. These findings suggest that influenza A virus infection results in the rearrangement of cytoskeletal proteins in human macrophages.
Actin and RIG-I/MAVS signaling components translocate from the cytoplasm to the mitochondria during influenza A virus infection
Subcellular proteome analysis revealed that cytoplasmic actin and its fragments accumulate onto mitochondria during influenza A virus infection. Also, MAVS resides in the mitochondrial outer membrane, and its localization is essential for its function (10). However, many signaling molecules involved in antiviral response have been reported to localize in cytoplasm. Therefore, we studied the effect of influenza A virus infection and a synthetic cytosolic dsRNA analog poly(I:C), a known activator of MAVS signaling pathway, to the subcellular localization of the signaling molecules in this pathway. As reported earlier, MAVS was localized mainly to mitochondrial fraction (Fig. 2⇓A). RIG-I and its downstream signaling molecule IKKε were detected only in the cytoplasmic fractions in control cells (Fig. 2⇓A). During viral infection, RIG-I is expected to interact with MAVS, which, in turn, associates with IKKε, resulting in the expression of IFN-β (6). Influenza A virus infection clearly induced translocation of RIG-I and IKKε from the cytoplasm to mitochondria (Fig. 2⇓A). TRIM25 is a cytoplasmic ubiquitin ligase crucial for RIG-I-mediated signaling events by facilitating its interaction with MAVS (25). In control macrophages, TRIM25 was observed as two bands in cytoplasmic fractions. The upper band of TRIM25 is most likely a posttranscriptionally modified form of the molecule since TRIM25 has been suggested to act as an E3 ligase for self-ISGylation and self-ubiquitination (26). A clear mitochondrial translocation of TRIM25 was seen upon viral infection (Fig. 2⇓A). In contrast, X-linked inhibitor of apoptosis (XIAP), a cytoplasmic protein that is the most potent member of the inhibitor of apoptosis protein family, did not translocate to the mitochondria upon infection. The expression level of XIAP decreased during infection, suggesting that apoptosis is initiated in infected macrophages. Mitochondrial translocation of RIG-I, IKKε, TRIM25, and actin after influenza A virus infection was also verified by confocal microscopy (Fig. 2⇓B).
Transfected RIG-I has previously been shown to interact with TRIM25 in HEK cells (25), but it has not been established whether the binding occurs in endogenous level and whether it needs an activation signal. Therefore, we next conducted coimmunoprecipitation experiments to determine whether endogenous RIG-I binds to TRIM25 in human primary macrophages upon influenza A virus infection. We could not detect any RIG-I-TRIM25 binding in control cells, but, importantly, endogenous RIG-I and TRIM25 coprecipitated from influenza A virus-infected macrophages (Fig. 2⇑C). Our results imply that a multiprotein signaling complex is formed onto mitochondria in influenza A virus-infected human macrophages.
We also studied whether cytosolic poly(I:C) had similar effects than influenza A virus on protein dynamics. Poly(I:C) transfection increased RIG-I expression, and only a very small amount of RIG-I was translocated to the mitochondria (Fig. 2⇑A). The same was observed with TRIM25 and IKKε after transfection of poly(I:C). In contrast, the other cytoplasmic RNA helicase, mda-5 (melanoma differentiation-associated gene 5), which also signals through MAVS, did not translocate to mitochondrial fraction upon viral infection. However, small but clearly detectable amount of the protein translocated to the mitochondria upon poly(I:C) stimulation. Our results are in accordance with previous reports showing that the influenza A virus-induced signaling pathway utilizes RIG-I and cytosolic dsRNA signals mainly through mda-5 (4, 27, 28).
TRADD is a cytosolic adapter protein that was very recently shown to interact with MAVS and thereby participate in RIG-I/MAVS-dependent antiviral immune responses (9). Thus, we were interested in studying the subcellular localization of TRADD in influenza A virus-infected macrophages. Like RIG-I, IKKε, and TRIM25, TRADD was detected only on cytoplasmic fraction in control cells (Fig. 2⇑D). Viral infection clearly induced its translocation onto mitochondria. Mitochondrial accumulation enables TRADD to interact with MAVS that is supposedly followed by signal complex formation and IRF-3 activation.
RIG-I protein is rapidly up-regulated during influenza A virus infection, but it translocates onto mitochondria with slower kinetics than actin, TRIM25, TRADD, and influenza A virus proteins
We next examined the effect of influenza A virus infection to the total cellular protein amount of actin, RIG-I, TRIM25, and TRADD, as well as to their translocation kinetics onto mitochondria. Macrophages were infected with influenza A virus for different time periods, and total cellular lysates and cytoplasmic and mitochondrial fractions were prepared. Infection had a very modest decreasing effect on total expression level of actin (Fig. 3⇓A). In control cells, actin was almost exclusively found in cytoplasmic fractions, whereas mitochondrial translocation of actin was detected as early as 6 h postinfection, and the amount of 42-kDa actin in mitochondria further increased when the infection proceeded (Fig. 3⇓B). The actin cytoskeleton is a caspase cleavage target during apoptosis (29). Accordingly, actin cleavage was detected 16 h after infection, and actin fragments were solely localized in mitochondrial fractions (Fig. 3⇓B). The protein amount of RIG-I was clearly induced already at 6 h postinfection, and the amount continued to increase up to 24 h postinfection (Fig. 3⇓A). After 16 h of infection, a part of the RIG-I protein was detected in mitochondrial fraction, and mitochondrial protein levels of RIG-I were significantly increased until 24 h postinfection (Fig. 3⇓B). TRIM25 translocation to mitochondria preceded RIG-I translocation: as early as 10 h after infection a weak signal of TRIM25 was seen in mitochondrial fraction, and after 16 h, TRIM25 was translocated almost completely to mitochondria (Fig. 3⇓B). At the same time, the cytoplasmic level of TRIM25 decreased, and the analysis of total cellular lysates showed that viral infection had a modest decreasing effect on TRIM25 protein expression (Fig. 3⇓A). Also, TRADD translocation to mitochondria during influenza A virus infection preceded that of RIG-I (Fig. 3⇓B). Influenza A virus infection had little effect on total TRADD protein expression (Fig. 3⇓A).
To characterize the progression of influenza A virus infection in macrophages, we analyzed the expression kinetics of viral proteins in total lysates as well as mitochondrial and cytoplasmic fractions with influenza A virus H3N2 Ab. The viral Ab recognizes three major bands (∼75, 55, and 25 kDa), but the proteins in these bands have not been previously characterized. We used in-gel digestion and mass spectrometry to identify these virus proteins from a corresponding silver-stained gel. The 75-kDa band contains polymerase basic protein 2, polymerase acidic protein, and hemagglutinin; the 55-kDa band contains nucleoprotein and hemagglutinin; and the 25-kDa band contains nonstructural protein 1 and matrix protein 1. In total cell lysates, expression of viral proteins was clearly seen at 6 after infection (Fig. 3⇑A). Interestingly, expression of viral proteins in mitochondrial fractions was seen already at 3 h postinfection, whereas in cytoplasmic fractions viral proteins became detectable at 6 h after infection (Fig. 3⇑B).
RIG-I mRNA is more rapidly up-regulated than IFNs and TNF-α upon influenza A virus infection
To further characterize the effect of influenza A virus infection on macrophages, we studied the mRNA expression profiles of IFN-β, IL-29, TNF-α, and RIG-I in influenza A virus-infected macrophages. The mRNA expression levels of type I and III IFNs, IFN-β, and IL-29, as well as proinflammatory cytokine TNF-α, started to increase in response to influenza A virus infection in human macrophages at 6 h postinfection, and reached maximum at 10 or 16 h postinfection (Fig. 4⇓). However, RIG-I transcription preceded these events: it was elevated already at 3 h after infection, and was maximal at 6 h postinfection (Fig. 4⇓).
Mitochondrial translocation of actin and RIG-I/MAVS signaling components is not dependent on caspase activation
Our subcellular proteome analysis of influenza A virus-infected macrophages resulted in the identification of several cleaved protein fragments that are known caspase-cleavage targets. As RIG-I/MAVS signaling has been linked to the activity of caspase-3 (30), we studied the activation kinetics of caspase-3 upon influenza A virus infection. The proteolytic processing of pro-caspase-3 into active caspase-3 was observed after 10 h of infection (Fig. 5⇓A), suggesting ongoing and simultaneous antiviral and apoptotic events in influenza A virus-infected human macrophages. Moreover, influenza A virus infection induced caspase-3-dependent proteolysis of proinflammatory cytokine IL-18 to IL-18 p15 and p16 fragments with the same kinetics as caspase-3 activation (Fig. 5⇓A).
To investigate whether mitochondrial translocation of cytoplasmic actin, RIG-I, TRIM25, and TRADD requires the activation of caspases, macrophages were pretreated with the general caspase inhibitor z-VAD, which inhibits caspases 1, 3, 4, and 7, before influenza A virus infection. Pretreatment of macrophages with z-VAD totally inhibited the proteolytic processing of pro-caspase-3 into active caspase-3 in response to influenza A virus infection (Fig. 5⇑B). Additionally, caspase-dependent cleavage of actin was totally hindered since no actin fragments were detected (Fig. 5⇑C). A specific inhibitor of caspases 1 and 4, z-YVAD, did not block actin fragmentation (data not shown), suggesting that intact actin is cleaved by caspases 3 and/or 7. The general caspase inhibitor z-VAD did not decrease the accumulation of intact actin onto mitochondria. Moreover, z-VAD did not prevent mitochondrial translocation of RIG-I, TRIM25, and TRADD upon viral infection. This shows that mitochondrial translocation of intact actin and RIG-I/MAVS signaling components are not dependent on caspase activation.
An intact actin cytoskeleton is essential for full expression of IFNs and TNF-α mRNA induced by influenza A virus
To examine the role of actin in antiviral response in more detail, we treated the cells with cytochalasin D, which disassembles the actin network. Previous studies have shown that cytochalasin D has an effect on viral internalization, assembly, and budding of influenza A virus in a virus strain- and cell polarity-dependent manner (31, 32, 33). Therefore, we first examined the effect of cytochalasin D on the expression of influenza proteins. Macrophages were treated with cytochalasin D for 30 min before infection, and cell lysates were analyzed by Western blotting with specific viral H3N2 protein Ab. The results show that cytochalasin D does not have an influence on viral protein expression (Fig. 6⇓A).
We then determined the role of actin network in caspase-3 activation in response to influenza A virus infection. Macrophages were treated with cytochalasin D for 30 min before infection, after which cytoplasmic fractions were analyzed by Western blotting. During viral infection, pro-caspase-3 was cleaved into active caspase-3, and cytochalasin D had no effect on this or formation of caspase-3 cleavage of pro-IL-18 into a biologically inactive form of IL-18 (Fig. 6⇑B). To study the effect of cytochalasin D on cytokine response, we treated macrophages with cytochalasin D for 30 min before infection, after which IFN-β, IL-29, and TNF-α mRNA expression was determined by real-time PCR. Cytochalasin D significantly inhibited the mRNA expression of IFN-β, IL-29, and TNF-α in influenza A virus-infected macrophages (Fig. 6⇑C). These results indicate that intact actin cytoskeleton structure is crucial for cytokine response during viral infection.
Influenza A viruses are important pathogens that cause acute respiratory diseases in humans and different animal species. Host defense against influenza A virus infection is initiated by the innate immune system, operating on the basis of general pathogen features. Pathogens are identified through PRRs that detect pathogen-associated molecular patterns (PAMPs) and signal the presence of infection to the host, thereby activating host defense, including antiviral and proinflammatory cytokine production. Additionally, apoptosis, programmed cell death, of infected cells is initiated by an as yet unidentified mechanism. In the present report we have used subcellular proteomics to characterize the cellular responses of human primary macrophages to influenza A virus infection. Proteomics experiments are still technically challenging compared with, for example, DNA microarray studies, and a general drawback with current proteomic techniques is that the sensitivity of the methods is usually not enough to identify and quantify low-abundant proteins in the cells, including many signaling molecules such as RIG-I and IKKε. However, the information provided with proteomics has a key role in understanding the details of cellular signaling mechanisms. Proteins are the active molecules involved in signal transduction, and it is well known that protein and mRNA levels do not correlate well. Additionally, many of the cellular events in signal transduction are such that they cannot be detected at the RNA level, for example, cleavage products of the intact target proteins, and protein modifications and translocations inside the cell. These were also major findings in the present study underscoring the importance on proteome-level studies.
Viruses utilize cellular machinery to complete their life cycle, and a large amount of viral proteins are synthesized in infected cells. These viral polypeptides stimulate endoplasmic reticulum stress response, which activates the unfolded protein response. The unfolded stress response is antagonized by at least three hierarchical protein-folding machineries. These include the HSP70 molecular chaperones BiP (GRP78), the HSP90 chaperone GRP94, and the lectin chaperones, such as calreticulin (34). BiP prefers completely unfolded or unstructured proteins as substrates, whereas GRP94 and calreticulin function downstream of BiP and they preferentially work on partially folded proteins (34). In our experiments, expression of BiP, GRP94, and calreticulin was highly up-regulated in cytosolic proteomes of influenza A virus-infected macrophages, suggesting an enhanced activity for all three major protein-folding machineries. In general, endoplasmic reticulum stress is associated with transcriptional up-regulation of genes encoding proteins that facilitate folding or degradation of proteins, and therefore it is conceivable that expression of BiP, GRP94, and calreticulin is up-regulated during influenza A virus infection.
In addition to BiP, GRP94, and calreticulin, the expression of three other proteins involved in cell rescue and defense response was up-regulated in cytosolic proteomes of influenza A virus-infected macrophages: HSP60, HSP90α, and HSP90β. In contrast to other general chaperones, HSP90 seems to be more selective, as it preferentially interacts with a specific subset of the proteome (35). In addition to chaperone function, HSP90 induces conformational changes in folded, native-like proteins that results in their activation or stabilization (36). A recent study showed that HSP90 interacts with and stabilizes NALP3 inflammasome structure (37), which is involved in caspase-1 activation and secretion of inflammatory cytokines IL-1β and IL-18 during microbial infection (38). We have previously shown that influenza A virus infection of human primary macrophages leads to the activation of inflammasome-associated caspase-1 and secretion of IL-1β and IL-18 (12, 13, 30). It is likely that cytosolic up-regulation of HSP90 expression results in further stabilization of inflammasome structure and leads to more efficient activation of caspase-1, resulting in enhanced inflammatory response in virus-infected macrophages.
Cytoplasmic actin constitutes microfilament that is the major component of cellular cytoskeleton. One of the major findings of our present report is a significant subcellular rearrangement and cleavage of actin microfilaments during influenza A virus infection. In control cells, actin was almost exclusively found in cytoplasmic fractions, whereas mitochondrial translocation of actin was detected at early phases of influenza A virus infection. Interestingly, influenza A virus proteins were localized to mitochondria already at 3–6 h after infection, suggesting that viral proteins could be involved in triggering movement of actin to mitochondria. However, further studies are required to define the mechanism through which viral infection triggers actin translocation to the mitochondria. At later phases of infection, actin cleavage was detected, and its fragments were solely localized in mitochondrial fractions. Actin has been shown to be cleaved by caspases into N-terminal and 15-kDa C-terminal fragments. C-terminal fragment can undergo N-myristoylation, which targets it to mitochondria (39). Importantly, the 15-kDa actin fragment was detected in mitochondrial fractions of influenza A virus-infected cells when 2-DE was performed using a isoelectric point range of 3–10 in the first dimension separation (data not shown). In our experiments actin cleavage was completely dependent on caspase activity (Fig. 5⇑C), which is in line with results showing that the actin cytoskeleton is a caspase cleavage target during apoptosis. It has been suggested that actin fragments, fractins, induce morphological changes resembling apoptotic cells, and they facilitate the mitochondrial recruitment of proapoptotic proteins from the cytosol (29). Fragmentation of actin network may also inhibit assembly and exocytosis of virus particles. In conclusion, our results clearly show that actin fragments specifically translocate to mitochondria during influenza A virus infection where they are likely to disrupt mitochondrial membrane and enhance apoptosis of infected cells.
Caspases cleave also a number of other structural proteins, such as intermediate filaments, resulting in morphological changes of apoptotic cells (22). In our experiments, a C-terminal fragment of gelsolin, a cytoplasmic actin-regulatory protein, was found to be up-regulated in both cytoplasmic and mitochondrial proteomes of virus-infected macrophages. This C-terminal fragment of gelsolin has been reported to have antiapoptotic activity (40). The structural protein vimentin and its cleaved fragments showed clear up-regulation in the cytoplasmic fraction upon influenza A virus infection. Vimentin is an intermediate filament protein that is degraded in response of apoptotic inducers (41, 42). Caspase proteolysis of vimentin promotes apoptosis by dismantling intermediate filaments, and by generating a proapoptotic amino-terminal cleavage product that serves to amplify the cell death signal (43). Microtubules have also been reported to undergo rearrangement during apoptosis, but conflicting reports reflect differences in the properties of microtubules in diverse cell lineages (44). Here, we observed tubulin fragments in mitochondrial fraction of virus-infected human macrophages, indicating that microtubules are cleaved during infection, and fragments are targeted to mitochondria.
Our present results show that in addition to RIG-I, other signaling molecules required for IFN-α/β production, including IKKε and TRIM25, translocate onto mitochondria in influenza A virus-infected macrophages. Translocation of IKKε to mitochondrial membrane has been previously described in IKKε-transfected and vesicular stomatitis virus-infected lung epithelial cells (15). In contrast, this is the first report showing mitochondrial translocation of TRIM25 in response to virus infection. TRIM25 delivers a lysine 63-linked ubiquitin moiety to the N-terminal domain of RIG-I, which strengthens interactions of RIG-I with MAVS, resulting in enhanced IFN-α/β transcription (25). Viral infection did not increase TRIM25 protein amount, whereas RIG-I protein level was very rapidly up-regulated in infected cells (Fig. 5⇑A). Also, translocation of TRIM25 to mitochondrial outer membrane was seen earlier than that of RIG-I, and coimmunoprecipitation experiments showed that RIG-I and TRIM25 form a complex in the surface of mitochondria. These findings suggest that RIG-I ubiquitinylation by TRIM25 occurs in mitochondrial outer membrane.
Activation of transcription factor NF-κB is a fundamental early step of TNF-α and RIG-I/MAVS signaling pathways. Upon activation of TNFR1 at the plasma membrane, TNFRI serves as a docking site for the adaptor protein TRADD. TRADD in turn recruits TNFR-associated factor (TRAF)2 and the serine/threonine kinase RIP1 (receptor-interacting protein 1), which rapidly signals for NF-κB activation. At later time points, TRADD, RIP1, and TRAF2 dissociate from TNFR1 and recruit FADD (Fas-associated death domain protein) to the complex to activate caspase-8, resulting in the initiation of programmed cell death (8). A very recent report showed that TRADD is also a central mediator of RIG-I-induced antiviral response that leads to activation of transcription factors IRF-3 and NF-κB and production of IFN-α/β (9). After influenza A virus infection of macrophages, TRADD was highly concentrated on mitochondrial outer membrane where it can interact with other components of the RIG-I/MAVS signaling pathway. TRADD participates in TNF-α-induced apoptosis (45), and it may be that TRADD is involved in triggering apoptosis in virus-infected cells.
A key event of the antiviral innate immune response is the production of IFN-α/β. Transcription of IFN-α/β in influenza A virus infection is triggered by genomic viral RNA and/or by RNA products formed during viral replication. Viral replication products can be sensed by endolysomal TLR3 or by cytoplasmic RNA sensors RIG-I and MDA-5 (1). We have previously shown that cytoplasmic RNA recognition pathway, but not TLR3 pathway, activates strong expression of antiviral cytokines in human macrophages (30). In accordance with these results, several studies have shown that cytosolic RIG-I and its interaction with mitochondrial adapter protein MAVS are essential for influenza A virus-induced IFN-α/β production (4, 28, 46). In our present study, RIG-I gene expression was activated with faster kinetics compared with IFN-β and IL-29 genes, suggesting that enhanced RIG-I expression is required for efficient production of antiviral cytokines. Our data also demonstrate that RIG-I is translocated to mitochondrial outer membrane during influenza A virus infection. Interestingly, actin was translocated onto mitochondria with faster kinetics compared with RIG-I. Our current results and previous studies have demonstrated that actin participates in transducing cell death signals. Additionally, it has been shown that disruption of actin cytoskeleton results in NF-κB activation and production of chemokines in intestinal epithelial cells (47). It is possible that actin is also involved in the activation of antiviral response. It is tempting to speculate that actin filaments serve as tracks directing RIG-I to mitochondria where RIG-I can interact with MAVS and trigger the signaling pathway needed for IFN-α/β production. Interestingly, cytochalasin D, a cell-permeable and potent inhibitor of actin polymerization that disrupts actin microfilaments, clearly inhibited influenza A virus-induced expression of IFN-β, IL-29, and TNF-α. In contrast, caspase-3 activation, which is a hallmark of apoptosis, was not affected by cytochalasin D treatment. Our results suggest that intact actin cytoskeleton structure is crucial for antiviral response during viral infection.
The adapter molecule involved in RIG-I-induced IFN-α/β gene expression, MAVS, resides in the mitochondrial outer membrane (10). It was recently reported that mitochondrial Nod-like receptor (NLR) gene family member, NLRX1, acts as a negative regulator of virus infection-induced IFN-α/β production (48). NLRX1 was found to interact with MAVS and inhibit RIG-I-mediated signaling after Sendai virus infection. These studies imply that mitochondria have a unique function in viral infection and important role in innate immunity. Our present data show that several components of the RIG-I/MAVS signaling pathway, including RIG-I, TRADD, TRIM25, and IKKε, translocate onto mitochondria in response to influenza A virus infection in macrophages. These results further emphasize the role of mitochondria in controlling antiviral innate immune response. In addition to activation of antiviral response, initiation of programmed cell death of infected macrophages was simultaneously seen with mitochondrial translocation of actin and its fragments. In conclusion, our results suggest that actin network interacts with mitochondria to regulate both antiviral and cell death signals during influenza A virus infection.
The authors have no financial conflicts of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵1 This work was supported by Academy of Finland Grant 114437 and the Sigrid Jusélius Foundation.
↵2 Address correspondence and reprint requests to Dr. Tuula Nyman, Protein Chemistry Research Group, Institute of Biotechnology, P.O. Box 65, FI-00014, University of Helsinki, Helskinki, Finland. E-mail address:
↵3 Abbreviations used in this paper: PRR, pattern recognition receptor; IKKε, inducible IκB kinase; BiP, heat shock 70-kDa protein 5, GRP78; GRP, glucose-regulated protein; HSP, heat shock protein; IRF-3, IFN regulator factor 3; MAVS, mitochondrial antiviral signaling protein; NTC, no template control; PMF, peptide mass fingerprint; RIG-I, retinoic acid-inducible protein I; TRADD, TNFR1-associated death domain protein; TRIM25, tripartite motif protein 25; 2-DE, two-dimensional electrophoresis; XIAP, X-linked inhibitor of apoptosis.
- Received September 17, 2008.
- Accepted February 23, 2009.
- Copyright © 2009 by The American Association of Immunologists, Inc.