The Leader Protein of Theiler's Virus Prevents the Activation of PKR

The leader (L) protein encoded by cardioviruses is a very short multifunctional protein that contributes to evasion of the host innate immune response. This protein notably prevents the formation of stress granules in infected cells. Using Theiler’s virus as a model, we show that L proteins can act at two levels in the stress response pathway leading to stress granule formation, the most striking one being the inhibition of eucaryotic translation initiation factor 2 alpha kinase 2 (PKR) activation. Interestingly, the leader protein appears to inhibit PKR via a novel mechanism by rendering this kinase unable to detect double-stranded RNA, its typical activator. Unlike other viral proteins, such as influenza virus NS1, the leader protein appears to interact with neither PKR nor double-stranded RNA, suggesting that it acts indirectly to trigger the inhibition of the kinase.

ABSTRACT

Leader (L) proteins encoded by cardioviruses are multifunctional proteins that contribute to innate immunity evasion. L proteins of Theiler’s murine encephalomyelitis virus (TMEV), Saffold virus (SAFV), and encephalomyocarditis virus (EMCV) were reported to inhibit stress granule assembly in infected cells. Here, we show that TMEV L can act at two levels in the stress granule formation pathway: on the one hand, it can inhibit sodium arsenite-induced stress granule assembly without preventing eIF2α phosphorylation and, thus, acts downstream of eIF2α; on the other hand, it can inhibit eucaryotic translation initiation factor 2 alpha kinase 2 (PKR) activation and the consequent PKR-mediated eIF2α phosphorylation. Interestingly, coimmunostaining experiments revealed that PKR colocalizes with viral double-stranded RNA (dsRNA) in cells infected with L-mutant viruses but not in cells infected with the wild-type virus. Furthermore, PKR coprecipitated with dsRNA from cells infected with L-mutant viruses significantly more than from cells infected with the wild-type virus. These data strongly suggest that L blocks PKR activation by preventing the interaction between PKR and viral dsRNA. In infected cells, L also rendered PKR refractory to subsequent activation by poly(I·C). However, no interaction was observed between L and either dsRNA or PKR. Taken together, our results suggest that, unlike other viral proteins, L indirectly acts on PKR to negatively regulate its responsiveness to dsRNA.

IMPORTANCE The leader (L) protein encoded by cardioviruses is a very short multifunctional protein that contributes to evasion of the host innate immune response. This protein notably prevents the formation of stress granules in infected cells. Using Theiler’s virus as a model, we show that L proteins can act at two levels in the stress response pathway leading to stress granule formation, the most striking one being the inhibition of eucaryotic translation initiation factor 2 alpha kinase 2 (PKR) activation. Interestingly, the leader protein appears to inhibit PKR via a novel mechanism by rendering this kinase unable to detect double-stranded RNA, its typical activator. Unlike other viral proteins, such as influenza virus NS1, the leader protein appears to interact with neither PKR nor double-stranded RNA, suggesting that it acts indirectly to trigger the inhibition of the kinase.

INTRODUCTION

Theiler’s murine encephalomyelitis virus (TMEV or Theiler’s virus) belongs to the species Theilovirus, within the genus Cardiovirus from the family Picornaviridae. The DA strain of TMEV is known to produce a persistent infection of the central nervous system of the mouse. This infection leads to chronic demyelinating lesions reminiscent of those found in multiple sclerosis patients (1, 2). Other cardioviruses are Saffold virus (SAFV), a TMEV-related virus isolated from humans, and encephalomyocarditis virus (EMCV), including the mengovirus strain.

The genome of cardioviruses consists of one nonsegmented positive-stranded RNA molecule of approximately 8 kb. It encodes a polyprotein that is processed to generate the structural and nonstructural proteins. A small protein (of about 70 amino acids) is cleaved off from the amino-terminal end of the polyprotein and is therefore called the leader (L) protein. Cardiovirus L proteins are closely related multifunctional proteins shown to interfere with critical cellular processes (3), such as type I interferon and chemokine production (4, 5), nucleocytoplasmic trafficking (6–12), apoptosis (13, 14), and mitogen-activated protein (MAP) kinase activity (15). Several domains have been defined in L: (i) an amino-terminal zinc finger (CHCC), (ii) a central glutamate/aspartate-rich domain that confers a very low isoelectric point (3.8) to the protein and that was shown in the case of EMCV to interact with Ran GTPase (16), and (iii) a carboxy-terminal domain found only in the Theilovirus species (TMEV and SAFV) and therefore called the Theilo domain (17). Mutations introduced in either the zinc finger or the Theilo domain affect all the known activities of Theilovirus L proteins (17). We and others previously reported that stress granules (SGs) appear in cells infected with L mutant but not wild-type TMEV or mengovirus (18, 19). Therefore, Cardiovirus infection is a stimulus that triggers SG assembly, which is simultaneously blocked by the L protein. SGs are cytoplasmic aggregates of stalled mRNA translation complexes that form in cells exposed to different cellular stresses (20). In addition to their obvious role in the stress-induced translational blockade, SGs are considered to serve as a platform for innate immunity signaling (21). Accordingly, cytoplasmic pathogen sensors such as RIG-I, MDA5, and eucaryotic translation initiation factor 2 alpha kinase 2 (PKR) have been detected in virus-induced SGs (19, 21), and inhibition of SG assembly in response to ΔNS1-IAV and EMCV infection correlated with decreased type I interferon transcription (21, 22). Thus, it is not surprising that many viruses have developed strategies to inhibit SG assembly (reviewed in reference 23).

The main event triggering SG assembly is the phosphorylation of the alpha subunit of the eukaryotic initiation factor 2 (eIF2α) by stress-activated kinases (24). Four mammalian eIF2α kinases have been identified: PKR, PERK, GCN2, and HRI. PKR is a well-known antiviral kinase whose expression is upregulated by type I interferons and whose catalytic activity is activated by double-stranded RNA, a common by-product of viral replication (reviewed in reference 25). Once activated, PKR phosphorylates eIF2α, and this leads to inhibition of mRNA translation initiation. PKR contains two amino-terminal dsRNA binding domains (dsRBD) and a carboxy-terminal catalytic protein kinase domain. Binding of dsRNA induces PKR homodimerization and transphosphorylation on various Ser and Thr residues, leading to full activation of the kinase. Phosphorylation of Thr446 and Thr451, located in the activation loop, is crucial for PKR activation (26) and therefore can be used as a marker to monitor the activation status of PKR. As PKR activation leads to translation arrest, many viruses evolved to antagonize PKR activity. Examples of viral countermeasures include the production of dsRNA binding proteins to prevent its detection by PKR (27, 28), triggering lysosome-mediated (29) or proteasome-mediated (30, 31) PKR degradation, limiting the PKR-activating dsRNA or defective interfering particle production (32, 33), production of RNA secondary structures that bind to but do not activate PKR (34, 35), direct binding of a viral protein to PKR to prevent dimerization or inhibit catalytic activity (36), and recruitment of a cellular phosphatase to dephosphorylate the PKR substrate, eIF2α (37).

Using PKR^−/− mouse embryonic fibroblasts, Langereis et al. showed that PKR was required for the formation of SGs after infection with a mengovirus carrying a mutation in the zinc finger domain of L (L^Zn) (19). On the other hand, Ng et al. (22) reported that wild-type EMCV infection triggers transient SG formation and then inhibited the process by G3BP1 cleavage at late stages of infection in a way similar to that used by poliovirus. In G3BP1/2 double knockout (KO) cells, L^Zn mengovirus infection failed to trigger SG formation, whereas it still triggered PKR activation (38).

Using TMEV as a model, we further explored the mechanism by which the leader proteins of cardioviruses inhibit infection-induced SGs. We show that L can act at two independent steps in the SG formation pathway. Importantly, our data suggest that L can act on PKR activation by rendering PKR insensitive to the presence of double-stranded RNA.

RESULTS

TMEV-induced stress granule assembly correlates with eIF2α phosphorylation and PKR activation.

We first confirmed previous observations showing that SGs are formed in cells infected with TMEV derivatives carrying mutations in either the zinc finger (L^Zn) or the Theilo domain (L^M60V) of the L protein but not in cells infected with wild-type TMEV (L^WT) (Fig. 1A). We also examined the phosphorylation status of eIF2α (Ser51) and of PKR (Thr446 and Thr451) in infected cells. Western blots presented in Fig. 1B show that, in cells infected with L-mutant viruses, SG assembly correlates with eIF2α phosphorylation and PKR activation. Our data therefore suggest that L inhibits SG assembly by blocking PKR activation and the consequent eIF2α phosphorylation.

FIG 1.

TMEV-induced stress granule assembly correlates with eIF2α phosphorylation and PKR activation. (A) Confocal microscopy images showing the coimmunostaining of eIF3 (green) and of the viral capsid protein VP1 (red) in HeLa cells infected for 10 h with the wild-type virus (L^WT) or L-mutant viruses (L^Zn and L^M60V). Note that some VP1-negative cells also display SG in the wells infected with the L-mutant viruses. This is probably due to the fact that VP1 had not reached a detectable level in some infected cells at the time the cells were fixed. (B) Western blot analysis of eIF2α and PKR phosphorylation in HeLa cells infected as described for panel A. 3D is the viral polymerase; β-actin detection was used as an additional loading control.

PKR is responsible for TMEV-induced eIF2α phosphorylation and stress granule assembly.

To test whether PKR was indeed the kinase responsible for SG formation in cells infected with L-mutant TMEV, we used both pharmacological inhibition of PKR and knockdown of PKR expression. Figure 2A shows that the PKR inhibitory compound 16 (C16), an ATP competitor, effectively decreases TMEV-induced PKR activation (i.e., phosphorylation of both T446 and T451) and eIF2α phosphorylation. This inhibition of PKR activity was sufficient to block SG assembly induced by both TMEV-L^Zn and TMEV-L^M60V (Fig. 2B). At the concentration used, C16 did not prevent eIF2α phosphorylation and SG assembly induced by sodium arsenite, which mostly acts through the kinase HRI (39).

FIG 2.

Pharmacological inhibition of PKR prevents TMEV-induced eIF2α phosphorylation and stress granule assembly. HeLa cells were treated with either DMSO or C16 and infected for 10 h with the wild-type virus (L^WT) or L-mutant viruses (L^Zn and L^M60V). As a control, DMSO- and C16-treated cells were treated with sodium arsenite (Ars) to induce PKR-independent phosphorylation of eIF2α. (A) Western blot analysis of eIF2α and PKR phosphorylation. (B) Confocal microscopy images showing the coimmunostaining of eIF3 (green) and the viral capsid (red).

To confirm that PKR was the cause of SG formation in TMEV-infected cells, we transduced HeLa cells with lentiviral vectors coding for three different short hairpin RNAs (shRNAs) directed against PKR mRNA. All three shRNAs triggered a strong inhibition of PKR expression (Fig. 3A). We chose to continue our experiments with shRNA2 and shRNA3 because some residual expression of PKR could be detected with shRNA1 (not shown). As shown in Fig. 3B and C, shRNA3 strongly inhibited PKR expression, eIF2α phosphorylation, and SG assembly in response to TMEV-L^Zn or TMEV-L^M60V infection. Importantly, PKR inhibition and infection did not prevent eIF2α phosphorylation and SG assembly in response to arsenite treatment (Fig. 3C, bottom). Similar results were obtained with shRNA2, while some stress granules were still detectable with shRNA1, in agreement with the residual expression of PKR observed with this shRNA (not shown).

FIG 3.

PKR is responsible for TMEV-induced eIF2α phosphorylation and stress granule assembly. (A) Western blot analysis of PKR expression in HeLa cells transduced with an empty pLKO-1 lentiviral vector (control) or with lentiviral vectors (FB52, FB53, and FB54) expressing shRNA1, shRNA2, or shRNA3, directed against PKR. Transduced cells were selected for 1 week with puromycin before being assessed for PKR expression. (B) Control HeLa cells and HeLa cells transduced with FB54 (shRNA3) were infected for 10 h with wild-type (L^WT) or L-mutant (L^Zn and L^M60V) viruses. Mock-infected and L^M60V-infected cell samples were treated for 30 min with sodium arsenite prior to cell lysis. Lysates were harvested at 10 hpi, and PKR expression and eIF2α phosphorylation were analyzed by Western blotting. (C) Confocal microscopy images showing the coimmunostaining of eIF3 (green) and the viral capsid (red) in control and PKR knockdown HeLa cells infected or treated as described for panel B.

Altogether, these results show that the L protein of TMEV inhibits stress granule assembly by blocking PKR activation.

L acts at two levels to inhibit SG formation.

The data described above (Fig. 2 and 3) suggest that L can prevent PKR activation. We previously observed, however, that in both infected and transfected cells, L^WT also prevented the formation of sodium arsenite-induced SG, suggesting that L was acting downstream of PKR-mediated eIF2α phosphorylation (18). In the case of mengovirus, it was reported that, in cells infected with the L^Zn mengovirus mutant, eIF2α was phosphorylated in a PKR-dependent fashion and that this eIF2α phosphorylation was inhibited by L (19). It was therefore unclear whether L acted upstream or downstream of PKR activation.

As shown in Fig. 4A and D, and in agreement with our previous report (18), infection with L^WT TMEV consistently suppressed SG formation induced by sodium arsenite treatment. In this case, however, L did not prevent eIF2α phosphorylation (Fig. 4B and C). These data confirm that L can act downstream of PKR to inhibit SG formation.

FIG 4.

L acts at two levels in the stress granule formation pathway. (A to D) L^WT prevents NaAsO₂-mediated SG assembly. HeLa cells were left noninfected (NI) of were infected for 10 h with KJ6 (L^WT). Where indicated, cells were treated for 30 min with NaAsO₂ before being fixed and processed for eIF3 and VP1 immunostaining (A and D) or before protein harvest and Western blot analysis of eIF2α phosphorylation (B and C). (A) Representative confocal microscopy images showing the absence of stress granules in L^WT-infected NaAsO₂-treated cells (upper right). (B) Representative Western blot monitoring of eIF2α (Ser51) phosphorylation and total eIF2α in uninfected and infected cells treated with increasing concentrations of NaAsO₂. (C) Quantification (mean and standard deviation [SD] values) of the data shown in panel B (n = 3). (D) Percentage (mean and SD values) of SG-positive cells, among infected cells, as counted from microscopy images taken from uninfected or infected NaAsO₂-treated cells (n = 3). (E and F) TMEV and mengovirus L^WT proteins can inhibit PKR activation. HeLa cells were infected with L^WT or the indicated L mutant viruses (12 h, 2 PFU per cell for TMEV; 6 h, 5 PFU per cell for mengovirus). Phospho-PKR (Thr446) and total PKR were quantified by Western blotting (mean and SD values; n = 5 for TMEV, n = 4 for mengovirus). *, P < 0.05; **, P < 0.01; ***, P < 0.001; all by analysis of variance comparisons.

On the other hand, we confirmed that L was also acting on PKR activation, thereby preventing PKR Thr446 phosphorylation. As shown in Fig. 4E and F, both mengovirus and TMEV L^WT inhibited PKR phosphorylation in infected cells, contrary to L mutants.

We conclude that the L protein can act both upstream and downstream of the PKR-mediated eIF2α phosphorylation step. Given the strong antagonism of L on PKR activation, we further examined the mechanism of this inhibition.

PKR is an antagonist of TMEV replication.

We compared the replication levels of wild-type and L-mutant viruses in control and in PKR-knockdown HeLa cells. In agreement with the data reported for mengovirus (38), Fig. 5 shows that the absence of PKR has a modest positive influence on replication of wild-type TMEV but a strong impact on L^Zn- and L^M60V-mutant virus replication. Thus, PKR is an antagonist of TMEV replication, and its inhibition by L is critical for optimal TMEV infection.

FIG 5.

PKR is a strong antagonist of TMEV replication. Control- or PKR^Knockdown-HeLa cells were infected with wild-type virus (L^WT) or L-mutant viruses (L^Zn and L^M60V), and supernatants were harvested at 16 h postinfection. Viral titers were measured by plaque assay. Histograms show the mean and SD values from three independent infection experiments.

The L protein prevents the interaction between viral dsRNA and PKR.

We next investigated the mechanisms of PKR inhibition by L. First, we sought a potential interaction between L and PKR by coimmunoprecipitation. Immunoprecipitation of L^WT from cells infected with a recombinant virus expressing a 3×FLAG-tagged L^WT protein failed to reveal any interaction between TMEV-L and PKR, although this virus inhibited PKR activation fairly well (Fig. 6A). Reineke et al. (40, 41) reported that SG-located G3BP1 could be involved in the activation of PKR through formation of heterotrimeric complex also involving Caprin1. However, as for PKR, coimmunoprecipitation experiments failed to reveal any interaction between L and G3BP1 (Fig. 6A).

FIG 6.

L prevents the interaction between viral dsRNA and PKR. (A) HeLa cells were infected with 5 PFU per cell of TM994, TM1016, and TM1017, recombinant TMEV expressing 3×FLAG-L^WT, 3×FLAG-L^Zn, and 3×FLAG-L^M60V, respectively. At 10 hpi, cells were lysed and 3×FLAG-L proteins were immunoprecipitated using an anti-FLAG antibody. The presence of L (FLAG detection), PKR, phospho-PKR (Thr446), and G3BP1 was assessed by Western blotting in cell lysates (ly), in postimmunoprecipitation lysate fractions (ly_p), and in immunoprecipitates (IP). (B) Confocal microscopy images showing the coimmunostaining of dsRNA and PKR in HeLa cells infected for 10 h with wild-type (L^WT) or L-mutant (L^Zn and L^M60V) viruses. Note the marked redistribution of PKR to punctate structures in cells infected with the L-mutant viruses. (C) Confocal microscopy images showing the coimmunostaining of PKR, eIF3, and G3BP (left) and of PKR, eIF3, and dsRNA (right) in HeLa cells infected, as described for panel B, with the L^M60V mutant virus. (D) Cartoon summarizing the observations. (E) HeLa cells were infected for 10 h with the wild-type or L-mutant viruses or transfected with poly(I·C) for 4 h before lysis and dsRNA immunoprecipitation. The presence of PKR, phospho-PKR, and L was monitored by Western blotting in equivalent amounts of whole-cell lysates and IP products. Graphs show the mean and SD values of PKR amounts detected in lysates and dsRNA-immunoprecipitated fractions from 3 experiments. *, P < 0.05 by analysis of variance pairwise comparisons between lysates and IP fractions.

We next analyzed the localization of PKR in infected cells by immunostaining. Interestingly, in cells infected with L-mutant viruses (L^Zn and L^M60V), PKR was markedly redistributed to cytoplasmic granules, where it partly colocalized with dsRNA. In sharp contrast, in cells infected with the wild-type virus, PKR displayed diffuse staining and did not clearly colocalize with dsRNA (Fig. 6B). In cells infected with the L^M60V mutant virus, PKR was detected in partially distinct groups of granules: (i) conspicuous granules likely corresponding to bona fide SG, often perinuclear, containing G3BP1, eIF3, and PKR, and (ii) generally more diffuse granules containing dsRNA, PKR, and some amount of G3BP1 (Fig. 6C and D). The lack of PKR relocalization to dsRNA-containing granules in L^WT virus-infected cells suggests that L blocks PKR activation by preventing its association with viral dsRNA.

To test the association of PKR with dsRNA in another setting, we immunoprecipitated dsRNA from infected cell lysates and examined PKR coimmunoprecipitation. Figure 6E shows that PKR coprecipitates with dsRNA in samples infected with L-mutant viruses significantly more than in samples infected with the wild-type virus. Thus, wild-type TMEV L protein inhibits PKR interaction with viral dsRNA.

In the same experiment, we analyzed whether L^WT would bind viral dsRNA and thereby prevent its detection by PKR. However, neither L^WT, L^Zn, nor L^M60V coimmunoprecipitated with dsRNA (Fig. 6E).

The L protein dampens the responsiveness of PKR to dsRNA.

As L appeared to interact neither with dsRNA nor with PKR, we were puzzled by the way L may impede dsRNA recognition by PKR. One the one hand, L may act by preventing PKR access to dsRNA (for example, by reshaping or hiding viral replication complexes). On the other hand, L may indirectly modify PKR to render this kinase insensitive to dsRNA. In the first hypothesis, PKR would keep the potential to become activated by another stimulus [e.g., poly(I·C)]. In the second hypothesis (true PKR inhibition), PKR should be refractory to poly(I·C) activation in cells infected with the wild-type virus. Therefore, we compared PKR activation in response to poly(I·C) transfection in mock-infected cells and in cells infected with L^WT or L^Zn virus. Figure 7A and B show that PKR activation in response to poly(I·C) was significantly decreased in cells infected with the wild-type virus. In contrast, PKR was hyperphosphorylated in L^Zn-infected cells, showing that infection per se did not prevent PKR activation by poly(I·C).

FIG 7.

L affects the responsiveness of PKR to dsRNA. (A and B) HeLa cells were infected with the wild-type (L^WT) and L^Zn mutant viruses. At 6 hpi, cells were transfected with 500 ng of poly(I·C) or control treated with the transfection reagent only. At 10 hpi, whole-cell lysates were harvested and analyzed by Western blotting to monitor PKR phosphorylation on threonine 446 and 451. (B) Quantification of the phospho-Thr446/total PKR ratio from 3 experiments (mean and SD values). *, P < 0.05; **, P < 0.01; ***, P < 0.001; all by analysis of variance comparisons. (C and D) HeLa cells were infected for 10 h with the wild-type virus (L^WT) or with a mutant virus that expresses GFP instead of L (ΔL-GFP) or were coinfected with both viruses. PKR phosphorylation was assessed by Western blotting (C), and its distribution was monitored by classical immunofluorescence microscopy (D). White arrows point to some infected cells presenting a typical punctate redistribution of PKR.

The inactive status of PKR in cells infected with the wild-type virus was confirmed in a coinfection experiment. HeLa cells were infected with a wild-type virus (L^WT), with KJ7, a TMEV derivative expressing green fluorescent protein (GFP) instead of L (ΔL-GFP), or coinfected with both viruses. As expected, the ΔL-GFP mutant virus triggered PKR redistribution and phosphorylation (Fig. 7C and D). After coinfection with the L^WT virus, PKR phosphorylation was inhibited (Fig. 7C) and PKR did not relocalize to cytoplasmic punctae in ΔL-GFP-infected cells (Fig. 7D).

Therefore, we concluded that the L protein of TMEV exerts a specific action that dampens the responsiveness of PKR to dsRNA.

DISCUSSION

We showed that TMEV L can inhibit SG formation by acting on the stress response both downstream of eIF2α phosphorylation and at the level of PKR activation. We then focused our analysis on the mechanism of PKR inhibition. PKR is a well-known interferon-stimulated gene product that restricts viral translation/replication through eIF2α phosphorylation and consequent inhibition of the cellular translation machinery. Accordingly, as previously observed in the case of mengovirus (19), PKR exerts a strong inhibitory effect on TMEV replication, which is counteracted by L.

Several pieces of data concur to suggest that L acts by preventing PKR association with dsRNA: (i) PKR coimmunoprecipitated with dsRNA from cells infected with L^Zn or L^M60V significantly more than from cells infected with L^WT viruses (Fig. 6E); (ii) PKR relocalized to punctate cytoplasmic dsRNA clusters after L-mutant but not L^WT virus infection (Fig. 6B); (iii) L^WT decreased PKR activation by poly(I·C) (Fig. 7A and B); and (iv) L^WT suppressed PKR activation in cells that were coinfected with L^WT and L-mutant viruses (Fig. 7C).

Other viral proteins, such as NS1 (influenza virus) or E3L (vaccinia virus), were shown to block dsRNA-mediated PKR activation (28, 42). These proteins possess a dsRNA binding domain (DRBD) that competes with PKR for dsRNA binding. In the case of cardioviruses, the leader sequence does not contain any putative dsRNA binding domain, and dsRNA immunoprecipitation from TMEV-infected cells failed to reveal any interaction between L and dsRNA. Moreover, sequestration of dsRNA by L is unlikely, because TMEV mutants that express truncated L* proteins (L* is an RNase L antagonist) but still express L^WT do not inhibit the RNase L pathway (43), which depends on dsRNA detection by oligoadenylate synthetases.

Alternatively, PKR inhibition may result from L binding to PKR DRBDs, thereby obstructing dsRNA recognition. However, experimental evidence does not support this mechanism, since we did not detect any interaction between the L protein and PKR.

Thus, the L-induced PKR inability to interact with dsRNA may result from an indirect mechanism. Posttranslational modifications of PKR that impact its activity have been described; ISGylation of lysine residues in the DRBD was shown to activate PKR independently of dsRNA (44), and SUMOylation of other lysine residues was shown to be important for efficient dsRNA binding and activation (45, 46). However, such posttranslational modification by covalent linkage of small proteins is unlikely to explain L activity, because the migration of PKR was not shifted in Western blots in samples from L^WT-infected cells.

Many phosphorylatable residues (Ser, Thr, and Tyr) are present in the DRBD of PKR. It is unclear if, and how, these residues regulate the activity of PKR. However, an appealing possibility is that phosphorylation of some of these residues dampens the capacity of PKR to bind dsRNA, and that L triggers their phosphorylation by acting on cellular kinases or phosphatases. The involvement of a cellular kinase in the activities of L is supported by the work of Porter et al. (15), who showed the involvement of MAP kinases in L-mediated activities.

The C protein of measles virus (MV) was shown to prevent PKR activation without targeting the classical PKR activation pathway (47). The mechanism exploits the cellular adenosine deaminase ADAR1, a dsRNA editing enzyme that structurally destabilizes endogenous dsRNA (e.g., transcripts containing inverted repeats in their 3′ untranslated regions) in order to avoid aberrant activation of cellular dsRNA sensors such as Mda5, RIG-I, and PKR (48). In an elegant paper, Pfaller et al. demonstrated that MV C protein regulates the amount of viral dsRNA and of defective interfering (DI) genomes that are produced during MV replication (49). They proposed that C protein acts on viral polymerase processivity and thereby contributes to maintaining aberrant viral dsRNA levels low enough to allow their destabilization by ADAR1 and prevent PKR activation (32). The C protein of Sendai virus, another paramyxovirus, was found to act in a similar fashion by limiting dsRNA generation in the course of viral replication (33). Unlike what we observed for the L protein, the C protein was unable to prevent PKR activation in response to poly(I·C) transfection (33), suggesting that L uses a distinct mechanism to target the PKR activation pathway.

In conclusion, our work suggests that the leader protein of TMEV acts on PKR by a novel indirect mechanism to render this kinase nonresponsive to dsRNA.

MATERIALS AND METHODS

Cells and viruses.

HeLa cells used in this study (kindly provided by R. H. Silverman) were of the HeLa M subclone, which reportedly has low endogenous RNase L activity (50). They were maintained in Dulbecco's modified Eagle medium (Lonza) supplemented with 10% fetal calf serum (MP Biomedicals), 100 IU penicillin/ml, and 100 μg streptomycin/ml. BHK-21 cells (ATCC) were cultured in Glasgow's modified Eagle's medium (GMEM) (Sigma) supplemented with 10% newborn bovine serum (Gibco), 100 IU penicillin/ml, 100 μg streptomycin/ml (Lonza), and 2.6 g of tryptose phosphate broth per liter (Difco).

The virus referred to as the wild type in this study is named KJ6 and is a derivative of the TMEV DA1 persistent strain, which expresses the wild-type L protein and harbors capsid mutations that adapt the virus to infect L929 cells efficiently (51). L-mutant viruses derived from KJ6 are TM659, carrying mutations disrupting the L zinc finger (L^Zn) (4), and FB09, carrying an M60V substitution in the Theilo domain (L^M60V) (17). KJ7 (ΔL-GFP) is a KJ6 derivative carrying the GFP-coding sequence between codons 5 and 67 of the L-coding region.

TMEV variants expressing a 3×FLAG-tagged L protein were obtained by cloning a 3×FLAG-coding sequence 5′ to the L protein initiation codon in the plasmids carrying the full-length cDNA of viruses KJ6 (pKJ6), TM659 (pTM659), and FB09 (pFB09) (18). Resulting plasmids bearing the cDNA of 3×FLAG-L^wt, 3×FLAG-L^Zn, and 3×FLAG L^M60V viruses were called pTM994, pTM1016, and pTM1017, respectively. For mengovirus, the FLAG sequence and Zn finger mutations (C17A to C19A) were introduced into the attenuated strain derived from pMC24 (52), kindly provided by Ann Palmenberg. Viruses were produced by reverse genetics from these plasmid constructs. Viruses were titrated by plaque assay on BHK-21 cells, and unless otherwise indicated, infections were performed at 10 PFU per cell for the indicated time. Viruses used in this study are listed in Table 1.

TABLE 1.

Viruses used in this study

Virusa	Parental strain	Characteristics
TMEV derivatives
KJ6	DA1	LWT; capsid adapted to L929 cells
TM659	KJ6	LZn; capsid adapted to L929 cells
FB09	KJ6	LM60V; capsid adapted to L929 cells
KJ7	KJ6	ΔL-GFP; capsid adapted to L929 cells
TM994	KJ6	3×FLAG-LWT; capsid adapted to L929 cells
TM1016	KJ6	3×FLAG-LZn; capsid adapted to L929 cells
TM1017	KJ6	3×FLAG-LM60V; capsid adapted to L929 cells
Mengovirus derivatives
FS269	MC24	FLAG-LWT; 5′ poly(C) tract reduced (24 C)
TM1097	FS269	FLAG-LZn (C19A-C22A) double mutation in L
Lentiviral vectors
FB52	pLKO.1	shRNA1 PKR, probe TRCN0000001379; puromycin resistance
FB53	pLKO.1	shRNA2 PKR, probe TRCN0000001381; puromycin resistance
FB54	pLKO.1	shRNA3 PKR, probe TRCN0000196400; puromycin resistance

^aPlasmids carrying the corresponding full-length cDNA are designated with a prefix of “p” (i.e., plasmid pKJ6 for virus KJ6).

Immunostaining.

Immunostainings were performed on cells that were cultivated on glass coverslips treated with poly-l-lysine and placed in 24-well plates. Prior to immunostaining, cells were fixed for 5 to 10 min in 300 μl of phosphate-buffered saline (PBS) containing 4% paraformaldehyde (PFA). Cells were then washed in 500 μl of PBS and permeabilized for 5 min at room temperature in 500 μl of PBS–0.1% Triton X-100. Blocking occurred for 1 h at room temperature in 300 μl of TNB blocking reagent (Perkin Elmer). Cells were next incubated with the primary antibody diluted in TNB at the following dilutions: TMEV VP1 (mouse; F12B3 clone; a kind gift of M. Brahic), 1/10; anti-TMEV capsid (rabbit polyclonal antibody; a kind gift of M. Brahic), 1/500; eIF3 (goat; sc-16377; Santa Cruz), 1/200 to 1/800; J2 (mouse, anti-double-stranded RNA [anti-dsRNA]; English & Scientific Consulting Bt.), 1/200; and PKR (rabbit; GTX 61151; Genetex), 1/250. After 1 h of incubation at room temperature, cells were washed 3 times for 5 min in 500 μl PBS–0.1% Tween 20. Secondary antibodies (Alexa Fluor 488- or 594-conjugated antibodies; Invitrogen) were incubated for 1 h at a 1/800 dilution in TNB. Finally, cells were washed 3 times in 500 μl PBS–0.1% Tween 20 and mounted with Mowiol for fluorescence microscopy analyses.

Fluorescence microscopy was performed with a DM IRB inverted microscope (Leica) equipped with a high-resolution AxioCam MRm digital camera (Zeiss) or with a spinning disk confocal microscope (Zeiss). Intensity, contrast, and color balance of images were equally equilibrated across conditions using ImageJ or Adobe Photoshop.

Western blotting.

Proteins extracted in 1× Laemmli buffer were heated for 5 min at 95°C, run on Tris-glycine-SDS (8 or 10%) polyacrylamide gels, and transferred onto nitrocellulose or poly(vinylidene fluoride) membranes. Blockage was performed in PBS–5% nonfat dry milk or in PBS–5% bovine serum albumin (BSA) for detection of phosphorylated peptides. The SuperSignal enhancer kit (Pierce) was used with nitrocellulose membranes to improve detection of the L protein. Primary antibodies were anti-VP1 (described above), anti-phospho-eIF2α (rabbit; number 3597; Cell Signaling), anti-eIF2α (rabbit; number 9722; Cell Signaling), anti-PKR (rabbit; GTX61151 [Genetex] or 18244-1-AP [Proteintech]), anti-phospho T446 PKR (rabbit; GTX61049 [Genetex] or Ab32036 [Abcam]), anti-phospho T451 (rabbit; GTX61867 [Genetex] or Ab81303 [Abcam]), anti-β actin (mouse; A5441; Sigma-Aldrich), anti-L (rabbit; home-made polyclonal antipeptide antibody), and anti-TMEV 3D polymerase (rabbit; kindly provided by M. Brahic). All primary antibodies were used at 1/1,000 dilution in PBS-nonfat dry milk or PBS-BSA. Primary antibodies were detected using goat anti-rabbit or goat anti-mouse IgG (H+L) horseradish peroxidase-conjugated antibodies (1/1,000; Dako). Secondary antibodies were detected with SuperSignal chemiluminescent substrates (Thermo Scientific).

Pharmacological inhibition of PKR.

Mock-infected or infected cells were treated sequentially with increasing concentrations of C16 (imidazolo-oxindole; I9785; Sigma) a first time at 1 h postinfection with 1 μM C16 diluted in dimethyl sulfoxide (DMSO) and a second time at 7.5 h postinfection with 10 μM to give an inhibitory boost. Cells were fixed for immunostaining or harvested for Western blot analyses at 10 h postinfection.

Knockdown of PKR expression.

Three different shRNAs targeting PKR were selected from the Sigma mission collection (reference code TRCN0000001379, shRNA1; TRCN0000001381, shRNA2; TRCN000019400, shRNA3) and cloned in the pLKO.1 lentiviral vector by following Addgene instructions (Addgene plasmid 10878, protocol version 1.0). HeLa cells were then transduced with the corresponding lentiviruses produced from these vectors, and selection of transduced cells was performed with increasing concentrations of puromycin (2 to 4 μg/ml). Efficacy of the knockdown was tested by Western blotting.

Immunoprecipitation.

Cells cultured in 10-cm dishes (Sarstedt) were washed twice with cold PBS and lysed with 800 μl of lysis buffer per dish (150 mM NaCl, 50 mM Tris, 1 mM EDTA, 1% NP-40, 1 mM phenylmethylsulfonyl fluoride, 1/1600 Ribolock RNase inhibitor, 1 tablet of Pierce phosphatase/protease inhibitor cocktail [number 88669] per 10 ml of lysis buffer, pH 7.5). Cell lysates were harvested in 1.5-ml tubes, homogenized with 18-gauge or 25-gauge syringes, and cleared by centrifugation at 12,000 × g for 10 min at 4°C. Supernatants were then transferred to new 1.5-ml tubes, and a sample of 80 μl per condition was mixed with 40 μl of 3× Laemmli buffer as a control for the total cell lysate. Remaining samples were cleared 3 times with 15 μl of protein A/G UltraLink resin (50% slurry) (Pierce) and incubated overnight on a rotating wheel at 4°C with a 1/100 dilution of the adequate primary antibody (same antibodies as those described in the Western blotting and immunostaining sections). The following day, 50 μl of protein A/G UltraLink resin (50% slurry) was added to the samples. These were kept on a rotating wheel at 4°C for 2 h. Beads were washed 3 times with lysis buffer and resuspended in 50 μl of Laemmli buffer (1.5×) for Western blot analyses. For 3×FLAG-L immunoprecipitations, cells grown in 6-cm dishes were collected in 300 μl of lysis buffer. Lysates were cleared once for 30 min with 20 μl of protein A/G UltraLink resin (50% slurry) (Pierce), and immunoprecipitation was performed by incubating lysates with 25 μl anti-FLAG magnetic beads (Sigma) for 2 h at 4°C. Other steps were as described above.

Poly(I·C) transfection.

Cells seeded in a 24-well plate were transfected by dropwise deposition of a mix containing 45 μl of transfection reagent (Lipofectamine 2000) and 7.5 μl of poly(I·C) (stock at 2 mg/ml) diluted in 1.5 ml of Dulbecco’s modified Eagle’s medium (DMEM) (Lonza). Transfected cells were harvested 4.5 h after transfection.