mRNA Decapping Enzyme 1a (Dcp1a)-induced Translational Arrest through Protein Kinase R (PKR) Activation Requires the N-terminal Enabled Vasodilator-stimulated Protein Homology 1 (EVH1) Domain*

Background: Dcp1a is a protein involved in mRNA decapping, processing body formation, and signal transduction. Results: GFP-Dcp1a, through its amino-terminal domain, can activate protein kinase R (PKR). Conclusion: Dcp1a may play novel roles in translation regulation and innate immunity. Significance: We present a novel instance of RNA degradation proteins affecting translation regulation. We have shown previously that poliovirus infection disrupts cytoplasmic P-bodies in infected mammalian cells. During the infectious cycle, poliovirus causes the directed cleavage of Dcp1a and Pan3, coincident with the dispersion of P-bodies. We now show that expression of Dcp1a prior to infection, surprisingly, restricts poliovirus infection. This inhibition of infection was independent of P-body formation because expression of GFP-Dcp1a mutants that cannot enter P-bodies restricted poliovirus infection similar to wild-type GFP-Dcp1a. Expression of wild-type or mutant GFP-Dcp1a induced phosphorylation of eIF2α through the eIF2α kinase protein kinase R (PKR). Activation of PKR required the amino-terminal EVH1 domain of Dcp1a. This PKR-induced translational inhibition appears to be specific to Dcp1a because the expression of other P-body components, Pan2, Pan3, Ccr4, or Caf1, did not result in the inhibition of poliovirus gene expression or induce eIF2α phosphorylation. The translation blockade induced by Dcp1a expression suggests novel signaling linking RNA degradation/decapping and regulation of translation.

non-translating mRNA within a cell. Conditions that increase the level of non-translating or deadenylated mRNA increase the size and number of P-bodies (4 -6). P-bodies are readily identified by their protein constituents, which include the RNA helicase Rck/p54, the cellular 5Ј-3Ј exoribonuclease 1 (Xrn1), components of the decapping complex, mRNA-decapping enzymes 1a and 2 (Dcp1a and Dcp2), and components of the deadenylation complexes, i.e. PABP-dependent poly(A) nucleases 2 and 3, Pan2/Pan3 (1,(7)(8)(9). The degradation of cellular mRNA can occur within P-bodies (5,10) and typically begins with a two-stage shortening of the poly(A) tail mediated by the poly(A) nucleases Ccr4, Caf1, Pan2, and Pan3 (11,12). P-body formation is also modulated by several stimuli ranging from stage of the cell cycle to virus infection. Numerous examples of viral modulation of P-body size and number as well as modification or cleavage of P-body constituent proteins have been described (13). Poliovirus infection leads to the loss of P-bodies by 3-4 h post-infection (hpi) and the cleavage or degradation of several key proteins critical for cellular RNA decay (14). Infection with West Nile virus or hepatitis C virus leads to a progressive reduction in the number of P-body foci 24 -36 hpi and relocalization of P-body components to viral replication centers (15)(16)(17). The genomic RNA of dengue virus interacts with the P-body component Rck/p54 (DDX6) and is presumed to be important for viral replication (18).
Previous studies have indicated roles for Dcp1a in P-body formation, maintenance, and regulation (6,32,33). Dcp1a in P-bodies constantly and quickly exchanges with Dcp1a in the cytoplasmic pool during fluorescence recovery after photobleaching analysis (34). Phosphorylation at Ser-315 mediates Dcp1a release from P-bodies as well as the regulation of P-body formation during translational stress (32), which suggests that Dcp1a may be involved in modulation of, or signal transduction from, P-bodies. In contrast, Dcp2 has a slower recovery rate during fluorescence recovery after photobleaching experiments, suggesting that it is a core P-body protein (34). The C-terminal domain of Dcp1a is highly conserved in plants and metazoa, folds in an ␣-helical conformation, and mediates homotrimerization that is required for Dcp1a incorporation into activated decapping complexes and localization to P-bodies (35).
Dcp1a has additional roles in cellular signaling that are not as well studied as its role in decapping. Previous studies have demonstrated the involvement of Dcp1a in cellular signaling pathways through protein-protein interactions (36,37). All eukaryotic orthologues of Dcp1a contain an N-terminal EVH1 (enabled vasodilator-stimulated protein homology 1) domain (30,31,37,38). EVH1 domains are known to interact with a variety of proline-rich protein ligands (39 -41). Dcp1a interacts with Dcp2 and Xrn1 to facilitate decapping reactions through the EVH1 domain (36,42). A recent report suggests that ERK kinase-mediated phosphorylation of Dcp1a enhances its interaction with Dcp2 (36). The EVH1 domain of Dcp1a also has a role in transforming growth factor ␤ (TGF-␤) signaling through a SMAD4 interaction (37, 43). Interestingly, the EVH1 domain of Dcp1a is highly conserved, but the substrate specificity varies across species (42). Recent studies have looked at the status of Dcp1a and P-bodies during stages of the cell cycle and the importance of Dcp1a phosphorylation for its function (32,33,36). These studies have highlighted the importance of Dcp1a phosphorylation in several signaling pathways that affect P-body formation and maintenance.
Our previous studies demonstrated that Dcp1a is cleaved during poliovirus infection by poliovirus protease 3C pro and that P-bodies were disrupted during infection (14). In an effort to understand the importance of Dcp1a cleavage to P-body disruption and further elucidate Dcp1a biology, we utilized the well documented GFP-Dcp1a expression system (32,33,36). Previous studies utilizing the GFP-Dcp1a overexpression system have demonstrated that GFP-Dcp1a retains the same functions as endogenous Dcp1a (32,33,35,36) and can enter and function in active decapping complexes and P-bodies (35). Additionally, GFP-Dcp1a expression has been used to determine alterations in the phosphorylation status of Dcp1a throughout the cell cycle and in response to cellular stress (32,33). These studies demonstrated the phosphorylation of GFP-Dcp1a by JNK or ERK kinases is important for regulating Dcp1a function and localization to P-bodies (32,36). Expression of GFP-Dcp1a does not significantly alter mRNA degradation or P-body formation, even in the presence of truncations that inhibit GFP-Dcp1a function (32,33,35,36). Taken together, these studies suggest that GFP-Dcp1a expression is a valid tool for studying the role of Dcp1a in cellular biology. In this study, we report the inhibition of cellular translation upon GFP-Dcp1a expression through the activation of double-stranded RNA-dependent protein kinase (PKR) and subsequent phosphorylation of eIF2␣. We also demonstrate the importance of the amino-terminal EVH1 domain for translational inhibition. The results presented here demonstrate a new link between the mRNA degradation machinery, the regulation of translation initiation, and cellular stress sensors.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfections-A549, PKR Ϫ/Ϫ , GCN2 Ϫ/Ϫ , PERK Ϫ/Ϫ , and PACT Ϫ/Ϫ mouse embryonic fibroblast (MEF) cells were cultured in DMEM supplemented with 10% FBS and maintained at 37°C with 5% CO 2 . PKR Ϫ/Ϫ MEFs were a gift from Mauro Costa-Mattioli (Baylor College of Medicine, Houston, TX). GCN2 Ϫ/Ϫ , PERK Ϫ/Ϫ , and corresponding wild-type control MEFs were obtained from the ATCC and were originally developed in the laboratory of David Ron (University of Cambridge, Cambridge, United Kingdom). PACT Ϫ/Ϫ and corresponding wild-type control MEFs were a gift from Ganes Sen (Cleveland Clinic, Cleveland, Ohio). Cells were seeded at a density of 1.5 ϫ 10 5 cells/well in a 12-well plate. Eighteen hours later, cells were transfected with the indicated plasmids using Xtremegene-HP transfection reagent (Roche) according to the protocol of the manufacturer. Additionally, A549 cells or MEFs were electroporated with the indicated plasmids utilizing the Neon transfection system (Invitrogen) in accordance with the protocol of the manufacturer for each cell type. Unless indicated otherwise in the figure legends, cells were transfected with 1 g of DNA/1.5 ϫ 10 5 cells.
Virus and Infections-Twenty-four hours post-transfection, A549 cells were infected with poliovirus type 1 (Mahoney strain) with a multiplicity of infection of 10 in DMEM containing 2% FBS for the indicated times. Poliovirus was grown in HeLa S3 cells and purified on CsCl gradients as described previously (44). Poliovirus was titrated by plaque assay on HeLa monolayers overlaid with 2ϫ DMEM containing 1% methylcellulose.
Plasmids and Cloning-Plasmids encoding GFP-Dcp1a (peGFP-Dcp1a and peGFP-Dcp1a-⌬CTD) were provided by M. Kracht (32). The plasmid expressing adenoviral VA1/2 RNA, pAdVAntage, was purchased from Promega. N-terminal truncations and EVH1 expression fragments were created from the plasmid encoding GFP-Dcp1a using the In-Fusion HD cloning system (Clontech). Dcp1a coding sequences were amplified by PCR using the described primers. The PCR products were ligated into the pcDNA4-GFP vector that had been linearized by EcoRV digestion for 2 h at 37°C. Ligation products were transformed into DH5␣ Escherichia coli (New Eng-land Biolabs), and colonies were selected by growth on ampicillin plates.
Immunofluorescence-A549 or MEF cells were seeded on glass coverslips and transfected with the appropriate plasmids as described previously. Cells were allowed to express the transgene for 18 -24 h, after which the cells were fixed with 4% paraformaldehyde in PEM (80 mM PIPES, 5 mM EGTA, and 2 mM MgCl 2 ). Cells were then permeabilized in 0.5% Triton X-100 and blocked with 5% BSA in Tris-buffered saline-0.1% Tween 20 (TBS-T). The cells were then incubated with the appropriate primary antibodies for two hours at room temperature or overnight at 4°C. Cover slips were mounted on microscope slides with Vectashield mounting medium with DAPI (Vector Laboratories) to visualize nuclei. The coverslips were analyzed by microscopy on a Nikon TE2000 inverted microscope. Deconvolution microscopy was performed with an Applied Precision DeltaVision image restoration microscope with conservative deconvolution algorithms.
For single cell quantification of eIF2␣ phosphorylation, images of 50 -100 transfected cells/plasmid were assayed for anti-eIF2␣ P-S51 antibody reactivity using ImageJ to measure the mean cytoplasmic pixel intensity. The results are represented as a percentage of Dcp1a-transfected cells positive for P-S51 staining above a 6% threshold of maximal eIF2␣ phosphorylation response obtained with arsenite treatment. Experiments were repeated a minimum of three times. Error bars indicate mean Ϯ S.D. p values were determined by Welch's two-sample t test.
Ribopuromycylation Assay (RPA)-Cells were grown, seeded, and transfected as described previously. Ribopuromycylation assays were conducted in accordance with previous studies (47) using the 12D10 antibody raised against puromycin (48) Briefly, on the day following transfection, the medium was replaced with fresh 10% FBS DMEM. 4 -6 h later, cells were treated with 50 g/ml puromycin and 100 g/ml cycloheximide for 5 min at 37°C. Cells were then washed with cold PBS and extracted with digitonin containing permeabilization buffer (50 mM Tris-HCl (pH 7.5), 5 mM MgCl 2 , 25 mM KCl, 100 g/ml cycloheximide, 1ϫ protease inhibitor mixture (Thermo), 10 units/ml RNase inhibitor ( New England Biolabs, Ipswich, MA), and 0.015% digitonin) for 5 min on ice. Cells were then washed again with permeabilization buffer and processed for immunofluorescence as described previously.
Immunoblot Analysis-Following transfection and plasmid expression, the cells were collected and lysed in 1% Nonidet P-40 buffer (10 mM Tris-Cl (pH 7.4), 100 mM NaCl, and 1 mM EDTA) to release cytoplasmic contents. Nuclei were cleared, and the cytoplasmic fraction was added to 2ϫ Laemmli sample buffer prior to 10% SDS-PAGE. Protein was transferred to nitrocellulose membranes and immunoblotted in accordance with standard procedures. Secondary peroxidase-coupled antimouse (Pierce) or anti-rabbit (Bio-Rad) antibodies were utilized for the detection of proteins of interest.

GFP-Dcp1a Expression Inhibits Poliovirus Protein Expression-
Previous studies have demonstrated that a GFP fusion variant of Dcp1a (GFP-Dcp1a) colocalizes with endogenous P-body markers and is capable of entering active decapping complexes (32,33,35). We confirmed that expressed GFP-Dcp1a localized to constitutive P-bodies in A549 cells using the endogenous P-body marker Rck/p54 (Fig. 1A, first panel). Additionally, we detected only rare P-bodies devoid of a GFP-Dcp1a signal, suggesting that GFP-Dcp1a is ubiquitously localized to P-bodies in the transfected cell.
Our previous studies demonstrated that poliovirus infection disrupts endogenous P-bodies by 3-4 hpi and results in the cleavage of endogenous Dcp1a (14). To further explore Dcp1a cleavage during poliovirus infection, GFP-Dcp1a was expressed in A549 cells, and the cells were infected with poliovirus. At 5 hpi, the infected cells were processed for immunofluorescence with antibodies against poliovirus proteins. Interestingly, GFP-Dcp1a containing P-bodies were retained at late time points of infection ( Fig. 1A, third panel). Additionally, GFP-Dcp1a was not cleaved during poliovirus infection (data not shown). Furthermore, in cells expressing GFP-Dcp1a, poliovirus gene expression was undetectable by immunofluorescence. In contrast, adjacent non-transfected cells or cells expressing GFP readily expressed poliovirus protein and exhibited cytopathic effects (Fig. 1A, second and third panels). GFP-Dcp1a-⌬CTD contains a deletion of amino acids 515-582 and does not significantly localize to P-bodies. However, GFP-Dcp1a-⌬CTD also blocked poliovirus protein expression (Fig. 1A, fourth panel). In addition to observing a loss of poliovirus protein production in cells expressing GFP-Dcp1a or GFP-Dcp1a-⌬CTD, plaque assays of supernatants from infected cells at 4 and 8 hpi were conducted. We observed a 3-fold loss of poliovirus production in cell populations that were transfected with GFP-Dcp1a compared with cells expressing GFP alone (Fig. 1B). In these experiments, transfection/expression efficiencies averaged 70%, leaving roughly 30% of cells fully permissive for virus growth. Together, these data indicate that expression of GFP-Dcp1a can block virus protein expression and reduce progeny virus production.
GFP-Dcp1a Expression Induces eIF2␣ Phosphorylation-We next sought to identify a mechanism by which GFP-Dcp1a expression would inhibit poliovirus gene expression. The active translation status of GFP-Dcp1a-transfected cells was examined utilizing an RPA (47,48), which detects incorporation of puromycin into nascent peptides on ribosomes. Fig. 2A illustrates the effects of expression of control GFP, GFP-Dcp1a, and sodium arsenite treatment, a potent translation inhibitor (49), on nascent peptide generation. Cells expressing GFP alone displayed robust puromycin staining in the RPA, indicating active cellular translation. Sodium arsenite treatment blocked translation and the RPA signal as expected, indicating that the RPA was performing as described previously (47). Cells expressing GFP-Dcp1a did not display significant puromycin staining, similar to cells treated with arsenite, suggesting that translation is repressed in these cells when compared with surrounding non-transfected cells ( Fig. 2A). Quantitation of intensity of the RPA stain in single cells revealed that translation activity was reduced in cells expressing Dcp1 over a wide range of levels, including cells with low expression levels (Fig. 2F).
A restriction of translation induced by GFP-Dcp1a could explain the lack of viral gene products and reduced virus titer during poliovirus infection. Further, this suggested a potential cross-talk role for an RNA decay component with cellular translation regulation. Cellular translation is commonly regulated at the rate-limiting initiation step. Phosphorylation of serine 51 of the ␣ subunit of eukaryotic initiation factor 2 (eIF2␣) is a well described method of cellular control of translation initiation (50,51). eIF2␣ is phosphorylated by four eIF2␣ kinases (PERK, PKR, GCN2, and heme-regulated inhibitor (HRI)) (50,51). The effect of GFP-Dcp1a expression on the induction of eIF2␣ phosphorylation was examined using immunofluorescence and an antibody specific to eIF2␣ phosphorylated at Ser-51, as described previously (52). A549 cells expressing a GFP control plasmid did not contain phosphorylated eIF2␣ as expected (Fig. 2B, left panel). In contrast, cells treated with sodium arsenite, which activates HRI kinase, displayed significantly elevated eIF2␣ phosphorylation (Fig. 2B,  right panel). In the majority of cells expressing GFP-Dcp1a (92.6% of transfected cells), we detected elevated levels of eIF2␣ phosphorylation (Fig. 2, B (center panel) and C), which suggests that GFP-Dcp1a is inducing the activation of an eIF2␣ stress kinase. To ensure that the presence of a large protein tag was not substantially enhancing the phosphorylation of eIF2␣, we expressed Dcp1a with a small HA tag (HA-Dcp1a) that also induced similar high levels of phospho-eIF2␣ (Fig. 2C).
To determine which levels of Dcp1a expression were associated with production of phospho-eIF2␣, we performed an immunoblot analysis with a Dcp1a-specific antibody from lysates of cells where increasing amounts of peGFP-Dcp1a were transfected into A549 cells. It was found that, in cell populations with a 50% transfection efficiency, approximately three times more GFP-Dcp1a compared with endogenous Dcp1a was required to induce eIF2␣ phosphorylation to a level detectable by immunoblot analysis in lysates from the whole cell popula-  tion (Fig. 2D). This suggests an average population activation threshold of near 6-fold in transfected cells. However, quantitation of phospho-eIF2␣ immunofluorescence levels at a single cell level revealed that widely variable levels of Dcp1a triggered eIF2␣ phosphorylation, which was often quite significant in cells with low-level expression of the Dcp1a transgene (Fig. 2E).
Analysis of single cells was carried out to determine whether expression of GFP-Dcp1a altered the cytoplasmic distribution or partioning between P-bodies and the remainder of the cytoplasm. A low to moderate/average expression level of Dcp1a associated with eIF2␣ phosphorylation did not significantly increase the P-body number. Very high expression led to new potential P-body foci in a subset of cells, some of which did not contain p54 (data not shown). A detailed examination of Dcp1a-expressing cells was carried out with deconvolution microscopy to reduce scattered light from the very bright P-body foci. This procedure revealed that GFP-Dcp1a partitioned quite effectively to P-bodies and strongly colocalized with the canonical P-body marker Rck/p54 (Fig. 3C). In some cases the relative concentration of p54 and Dcp1a in individual P-bodies was varied. The degree of partioning of endogenous Dcp1a and GFP-Dcp1a into P-bodies was quite similar (Fig. 3, C  and D).
Expression of Other P-body Proteins Does Not Effectively Induce Phosphorylation of eIF2␣-To determine whether expression of other P-body proteins induced eIF2␣ phosphorylation, V5 epitope-tagged Caf1, Ccr4, Pan2, and Pan3 were expressed in A549 cells. Expression of V5-Caf1 or V5-Pan2 did not significantly induce eIF2␣ phosphorylation, similar to the results obtained by expressing GFP alone, where up to 10% of transfected cells displayed at least a weak signal (Figs. 2 and 3). Interestingly, expression of V5-Ccr4 or V5-Pan3 induced some low level of eIF2␣ phosphorylation in slightly more cells. However, expression did not trigger this response in the majority of cells (Fig. 3B). Cells expressing V5-Pan3 that did induce eIF2␣ phosphorylation also formed large foci containing the transgene, which colocalized with canonical stress granule markers (G3BP-1 and TIA-1, data not shown). Thus, the expression of V5-Pan3 appeared to induce stress granule formation in some cells, which has been shown to trigger eIF2␣ phosphorylation (52). Taken together, these results suggest that GFP-Dcp1ainduced eIF2␣ phosphorylation is not a generalized response to protein expression or to an increase in P-body protein concentration but, rather, a specific response to GFP-Dcp1a.
The EVH1 Domain of Dcp1a Is Required to Trigger eIF2␣ Phosphorylation-To determine which Dcp1a domains were required for induction of eIF2␣ phosphorylation by GFP-Dcp1a expression, a panel of GFP-tagged truncations of Dcp1a were expressed in A549 cells (Fig. 4A). The C-terminal deletion mutant (GFP-Dcp1a-⌬CTD) (32) lacked the ability to homotrimerize and enter P-bodies but induced a high level of eIF2␣ phosphorylation (Fig. 4B, first panel) and inhibited poliovirus gene expression, similar to full-length GFP-Dcp1a (Fig.  1A, fourth panel). This suggests that the ability of GFP-Dcp1a to induce eIF2␣ phosphorylation does not require the trimerization of the Dcp1a constructs, nor does it require localization of Dcp1a to P-bodies.
A truncation of the N-terminal portion of Dcp1a lacking the EVH1 protein interaction domain (GFP-Dcp1a-⌬NTD) was included in P-bodies, although less consistently than the fulllength protein. GFP-Dcp1a-⌬NTD poorly induced eIF2␣ phosphorylation when compared with the full-length protein (18 versus 92.6% of transfected cells) (Fig. 4, B (second panel) and C). This suggested that the N-terminal EVH1 domain is required for induction of eIF2␣ phosphorylation. To test whether the EVH1 domain was sufficient to induce eIF2␣ phosphorylation, we expressed the EVH1 domain of Dcp1a with either an N-or C-terminal tag (GFP-EVH1 or EVH1-cherry) in A549 cells and probed for eIF2␣ phosphorylation. Both fusion proteins containing the EVH1 domain failed to induce eIF2␣ phosphorylation at a frequency higher than cells expressing GFP (Fig. 4, B (third and fourth panels) and C). Immunoblotting of electroporated A549 lysates illustrates the level of GFPtagged Dcp1a truncations with the corresponding level of eIF2␣ phosphorylation, illustrating the reduction in phospho-eIF2␣ when the EVH1 domain is removed (Fig. 4D). Taken together, these findings suggest that the EVH1 domain of Dcp1a is required, but not sufficient, for the induction of eIF2␣ phosphorylation in response to GFP-Dcp1a expression.
PKR Is Activated by GFP-Dcp1a Expression-To determine which of the four eIF2␣ kinases was activated by GFP-Dcp1a, we expressed GFP-Dcp1a in knockout MEF cell lines, each lacking one functional eIF2␣ kinase. MEFs competent for PKR activity (PKR ϩ/ϩ ) phosphorylated eIF2␣ in response to GFP-Dcp1a expression but not when expressing GFP alone (Fig. 5, A  and E). Conversely, MEFs deficient for PKR activity (PKR Ϫ/Ϫ ) did not phosphorylate eIF2␣ in response to GFP-Dcp1a expression (Fig. 5, B and E). Levels of GFP-Dcp1a-induced eIF2␣ phosphorylation did not significantly differ between cells deficient for GCN2 or PERK activity and their wild-type counterparts ( Fig. 5E and data not shown). These data suggest that PKR is the principle eIF2␣ kinase activated by GFP-Dcp1a expression.
To further confirm the involvement of PKR in GFP-Dcp1ainduced eIF2␣ phosphorylation, an expression plasmid encoding the adenoviral VA1/2 RNA (pAdVAantage) was utilized. This viral RNA is a non-coding transcript expressed during adenoviral infection that interacts with and inhibits the function of PKR (53,54). As shown in Fig. 6A, coexpression of the Dcp1a Activates PKR and Induces eIF2␣ Phosphorylation FEBRUARY 14, 2014 • VOLUME 289 • NUMBER 7 JOURNAL OF BIOLOGICAL CHEMISTRY 3941 VA1/2 RNA with either GFP-Dcp1a or GFP-Dcp1a-⌬CTD significantly inhibited the phosphorylation of eIF2␣. Fig. 6B demonstrates that, in the presence of VA1/2 RNA, a significant reduction in eIF2␣ phosphorylation was observed for all of the tested constructs. These results, combined with the results from Fig. 5, demonstrate that catalytically active PKR was required for the phosphorylation of eIF2␣ induced by GFP-Dcp1a, whereas the activity of other eIF2␣ kinases was not essential.

Inhibition of PKR Rescued Poliovirus Gene Expression in Cells
Expressing GFP-Dcp1a-To verify that eIF2␣ phosphorylation and subsequent inhibition of translation initiation was responsible for the observed inhibition of poliovirus gene expression and reduction in viral titer, we attempted to rescue viral replication. A549 cells were electroporated with pcDNA4-GFP or peGFP-Dcp1a with or without coexpression of adenovirus VA1/2 RNA. The cells were infected with poliovirus for 5 h and processed for immunofluorescence. A549 cells that expressed GFP alone or GFP with VA1/2 RNA displayed viral protein production by 5 hpi (Fig. 7A, top and bottom left panels). In contrast, cells expressing GFP-Dcp1a alone displayed significantly reduced poliovirus gene expression (Fig. 7A, top right  panel). However, cells expressing both GFP-Dcp1a and adenoviral VA1/2 RNAs produced poliovirus proteins similar to untreated cells and caused partial disruption of P-bodies (Fig.  7A, bottom right panel). Additionally, we conducted plaque assays of supernatants from A549 cells expressing GFP or GFP-Dcp1a with or without adenoviral VA1/2 RNA. Inhibition of PKR activity by VA1/2 RNA expression alleviated the reduction in poliovirus replication (Fig. 7B). These data suggest that the inhibition of PKR rescued poliovirus protein expression and allowed for progression of the viral replication cycle.

Activation of PKR by GFP-Dcp1a
Requires PACT-PKR is known to exist in an inactive state in unstimulated cells. PKR activation requires the interaction with specific ligands, most notably viral double-stranded RNA. More recently, it has been discovered that a cellular protein, PACT, could stimulate PKR in response to environmental stresses (55,56). To address the mechanism by which GFP-Dcp1a expression activates PKR and induces eIF2␣ phosphorylation, we utilized PACT-deficient MEFs (PACT Ϫ/Ϫ ). Wild-type and PACT Ϫ/Ϫ MEFs were transfected with pcDNA4-GFP or peGFP-Dcp1a and analyzed for eIF2␣ phosphorylation by immunofluorescence (Fig. 8A). PACT wild-type and PACT Ϫ/Ϫ cells responded similarly to GFP expression. PACT wild-type MEFs had a significantly elevated number of cells with phospho-eIF2␣ when expressing GFP-Dcp1a. In contrast, PACT Ϫ/Ϫ MEFs had substantially reduced levels of phosphorylated eIF2␣ in response to GFP-Dcp1a expression (Fig. 8B). These data suggest that PACT is involved in GFP-Dcp1a-mediated activation of PKR and subsequent phosphorylation of eIF2␣.

DISCUSSION
In a previous study, we showed that poliovirus infection disrupts cellular P-bodies by 3-4 hpi and mediates the cleavage of Dcp1a through the activity of viral 3C pro (14). Here we demonstrate that poliovirus did not express viral proteins or disrupt P-bodies in cells expressing GFP-Dcp1a (Fig. 1A). The preser-vation of P-bodies was not due to a modest delay of viral replication kinetics because GFP-Dcp1a-expressing infected cells contained normal amounts of P-bodies at 24 hpi, which is four times longer than the normal poliovirus replication cycle (data not shown). Expression of a C-terminal truncation of Dcp1a (GFP-Dcp1a-⌬CTD), which does not localize to P-bodies, prevented the expression of poliovirus proteins (Fig. 1A). This suggests that GFP-Dcp1a and GFP-Dcp1a-⌬CTD inhibit poliovirus gene expression independent of Dcp1a localization to P-bodies. P-bodies are dynamic and rapidly exchange constituent proteins, such as Dcp1a, with the surrounding cytosol. Concomitant with the inhibition of viral protein expression, we
RPA analysis revealed reduced translation in cells expressing GFP-Dcp1a compared with non-transfected cells or GFP-expressing cells (Fig. 2A). Global restriction of cellular translation is most often regulated at the rate-limiting initiation step through reducing the availability of the Met-tRNA met i ⅐eIF2⅐GTP ternary complex via eIF2␣ phosphorylation by one of the four eIF2 kinases (50,51). Phosphorylation of eIF2␣ also inhibits poliovirus translation at early time points during infection. Poliovirus translation is driven by cap-independent internal ribosome entry site-mediated translation initiation. However, initiation begins on AUG and requires a ternary complex to deliver Met-tRNA to ribosomes (57). However, if viral 3C protease is expressed in sufficient quantity, viral RNA can be translated through an eIF2␣-independent pathway that involves production of the truncated form of eIF5B that functionally replaces the ternary complex (58). eIF2␣ phosphorylation was significantly elevated in GFP-Dcp1a-expressing cells, notably before virus infection was initiated, thus blocking the production of  FEBRUARY 14, 2014 • VOLUME 289 • NUMBER 7

JOURNAL OF BIOLOGICAL CHEMISTRY 3945
virus protein at the outset of infection and before 3C pro could be produced at all.
We determined that other P-body proteins, Caf1, Ccr4, Pan2, and Pan3, were not capable of inducing eIF2␣ phosphorylation comparably to GFP-Dcp1a (Fig. 3). This suggested that the observed increase in phospho-eIF2␣ was not a result of disruption of P-body protein equilibrium but, rather, a specific response to GFP-Dcp1a expression.
Using eIF2␣ kinase-deficient MEFs and PKR inhibitors, we determined that functional PKR, but not GCN2 or PERK, was required for GFP-Dcp1a induced eIF2␣ phosphorylation (Figs. 5 and 6). Finally, we demonstrated that PKR-mediated translational arrest was responsible for the inhibition of poliovirus replication by rescuing poliovirus translation using PKR inhibitors (Fig. 7).
Dcp1a contains several domains, including an N-terminal EVH1 domain, a 14-residue short motif (MI), as well as a C-terminal trimerization domain. The trimerization domain adopts an ␣-helical conformation and is important for Dcp1a incorporation into active decapping complexes and localization to P-bodies (35). GFP-Dcp1a-⌬CTD inhibited poliovirus gene

Dcp1a Activates PKR and Induces eIF2␣ Phosphorylation
expression and induced eIF2␣ phosphorylation similarly to the full-length protein (Figs. 1A and 4), which suggests that localization to P-bodies and inclusion into active decapping complexes is not required to induce eIF2␣ phosphorylation.
The N-terminal EVH1 domain of Dcp1a is multifunctional because it interacts with Dcp2 and Xrn1 to promote mRNA decapping and also interacts with SMAD4 to mediate TGF-␤ signaling (30, 31, 36 -38, 42, 43). Here we show that the EVH1 domain is involved in a mechanism that triggers PKR activation. Additionally, expression of the EVH1 domain alone did not significantly activate PKR (Fig. 4). This suggests that the EVH1 domain is necessary, but not sufficient, for eIF2␣ phosphorylation in response to GFP-Dcp1a expression. The EVH1 domain may operate in conjunction with other regions of Dcp1a to mediate protein interactions. One report that indicated EVH1 domain-mediated interactions are affected by phosphorylation of downstream residues in the Dcp1a prolinerich domain (36). As a result, GFP-EVH1 may not be able to interact with PKR or PKR complexes strongly enough to trigger activation without the Dcp1a C-terminal extension. Interestingly, the Dcp1a N terminus has been suggested to be a new class of EVH1 domain, largely because of its longer ␣-helix (30,31,42). As a result of additional conserved residues in the EVH1 domain, Dcp1a is able to bind a variety of proline-rich protein ligands over an extended surface (42). This potentially provides Dcp1a with the ability to interact with numerous protein targets within the mammalian cell, which remain to be elucidated. Modulation of this activity through posttranslational modifications of Dcp1a may provide further understanding of the role of Dcp1a as a coactivator and signal transducer. The data in this study suggest that GFP-Dcp1a may interact with other proteins through the N-terminal domain, either through signaling cascades or directly with PKR.
The cleavage of Dcp1a by poliovirus 3C pro during viral infection was initially a curious discovery. Poliovirus proteases tend to selectively target proteins involved in cellular gene expression or the antiviral response to make the cellular environment more amenable to viral replication (68). Although Dcp1a has a described role in RNA degradation, no form of poliovirus RNA contains a 5Ј-methylguanosine cap, and it was not expected to be targeted by Dcp1a for decapping reactions. So why would a poliovirus protease target Dcp1a? This study suggests that Dcp1a may play novel roles in translation regulation and innate immunity because PKR is a central factor in translation regulation and many innate immune pathways. In this regard, it is possible that poliovirus targets Dcp1a for cleavage to control a novel arm of the innate immune response as well as maintaining the cellular translation apparatus in an active configuration. The activation of cellular stress kinase PKR when the levels of Dcp1a are increased indicates that Dcp1a expression must be maintained in stoichiometric balance with other factors. One scenario is that the PKR activation function of Dcp1a is normally repressed by a Dcp1a inhibitor of some sort, which is overcome by increased concentration of Dcp1a. The activation of PKR in response to GFP-Dcp1a expression links Dcp1a to the cellular stress pathways in a way not described previously.
Although PKR autophosphorylates and activates in response to double-stranded RNA, it is also regulated by protein-protein interactions that are independent of double-stranded RNA. Notably, PACT associates with and activates PKR without double-stranded RNA binding (55,56). In contrast, the human immunodeficiency virus, type 1 transactivation response RNAbinding protein (TRBP) interacts with PKR and inhibits PKR activation (59). TRBP also interacts with PACT and prevents PACT-mediated activation of PKR (60). However, upon cellular stress, PACT is phosphorylated and dissociates from TRBP, allowing PACT-dependent PKR activation (55, 60 -64). In this study, we demonstrate a requirement of PACT for the phosphorylation of eIF2␣ in response to GFP-Dcp1a expression (Fig. 8). This suggests that GFP-Dcp1a expression may mimic a stress signal, leading to the phosphorylation of PACT, disassociation from TRBP, and activation of PKR. Alternatively, Dcp1a may complex with TRBP or PACT, leading to the relief of TRBP-mediated PKR repression. The latter is a possibility because examination of protein lysates immunoprecipitated with a Dcp1a-specific antibody indicated that PACT existed in complexes with the Dcp1a transgene and endogenous Dcp1a (data not shown). Other protein factors activate PKR, although the mechanisms are not as clearly elucidated (65)(66)(67). Whether and how GFP-Dcp1a complexes with PACT, TRBP, PKR, or with additional proteins to activate PKR is a subject for future experimentation. The mechanism of PKR activation, the involvement of accessory proteins, and the role of posttranslational modifications of Dcp1a will be interesting areas for future studies.