IRF3 and IRF7 Phosphorylation in Virus-infected Cells Does Not Require Double-stranded RNA-dependent Protein Kinase R or IκB Kinase but Is Blocked by Vaccinia Virus E3L Protein*

Induction of interferon-α (IFNα) gene expression in virus-infected cells requires phosphorylation-induced activation of the transcription factors IRF3 and IRF7. However, the kinase(s) that targets these proteins has not been identified. Using a combined pharmacological and genetic approach, we found that none of the kinases tested was responsible for IRF phosphorylation in cells infected with Newcastle disease virus (NDV). Although the broad-spectrum kinase inhibitor staurosporine potently blocked IRF3 and -7 phosphorylation, inhibitors for protein kinase C, protein kinase A, MEK, SAPK, IKK, and protein kinase R (PKR) were without effect. Both IκB kinase and PKR have been implicated in IFN induction, but cells genetically deficient in IκB kinase, PKR, or thePKR-related genes PERK, IRE1, orGCN2 retained the ability to phosphorylate IRF7 and induce IFNα. Interestingly, PKR mutant cells were defective for response to double-stranded (ds) RNA but not to virus infection, suggesting that dsRNA is not the only activating viral component. Consistent with this notion, protein synthesis was required for IRF7 phosphorylation in virus-infected cells, and the kinetics of phosphorylation and viral protein production were similar. Despite evidence for a lack of involvement of dsRNA and PKR, vaccinia virus E3L protein, a dsRNA-binding protein capable of inhibiting PKR, was an effective IRF3 and -7 phosphorylation inhibitor. These results suggest that a novel cellular protein that is activated by viral products in addition to dsRNA and is sensitive to E3L inhibition is responsible for IRF activation and reveal a novel mechanism for the anti-IFN effect of E3L distinct from its inhibition of PKR.

Type I interferon (IFN), 1 consisting of the single IFN␤ gene and a family of IFN␣ genes, is rapidly induced by infection of mammalian cells by a broad spectrum of viruses (1,2). Gene induction is rapid, and can be divided into two distinct phases initiated by induction of IFN␤ and IFN␣4 expression followed by induction of other members of the IFN␣ gene family (3,4). Differences in the gene expression profiles of the distinct type I IFN genes can be accounted for by unique properties of their promoter/enhancer structures. IFN␤ expression is controlled by an enhanceosome that binds three distinct transcription factor complexes in the context of chromatin-organizing proteins (5). Each of these complexes, c-jun/ATF2 (AP1), IRF, and NFB, become active following protein phosphorylation events induced in response to virus infection. IFN␣ promoters are also activated by IRF proteins, and the distinct DNA binding characteristics and patterns of induction and activation of IRF3 and IRF7 confer the differential expression of the IFN␣ gene family (3, 6 -9).
While several transcription factors required for IFN gene induction have been identified, delineation of the signaling pathways stimulated by virus infection that lead to phosphorylation-dependent activation remains incomplete. Activation of NFB requires phosphorylation-induced degradation of its inhibitor, IB, through the action of the IB kinase (IKK) complex composed of IKK␣, IKK␤, and IKK␥/NEMO (10). The catalytic activity of IKK is activated by a wide variety of inducers, including viral infection and double-stranded RNA (dsRNA), a common by product of viral infection (11). Activation of IKK by dsRNA and viral infection may depend on a second enzyme, protein kinase R (PKR); however, in this role PKR may function noncatalytically and serve as an adaptor protein (12), although even this requirement has been recently called into question (13). The AP1 transcription factor, also required for IFN␤ gene induction, is likewise activated by phosphorylation through the action of c-Jun kinase (JNK), which is also activated in virus-infected cells (12). The third transcription factor complex required for IFN gene induction, composed of IRF3 and/or IRF7, is also activated by phosphorylation specifically in virus-infected cells (3, 6, 14 -22). However, the kinase responsible for its activation remains to be identified.
Not only is the identity of the IRF3/IRF7 kinase unclear, the nature of the activating component produced during virus infection that leads to kinase activation remains unknown. One candidate has been dsRNA, a common intermediate or by product of many viral infections that directly activates PKR (23). Moreover, many viruses target inactivation of PKR as a strategy to overcome host antiviral responses, by binding dsRNA, interfering with the activation of PKR, blocking its recognition of substrates, or inducing its degradation (24). However, recent evidence suggests that dsRNA may not be the essential or at least not the only viral component leading to IFN gene expression. For instance, disruption of the gene for PKR abrogated the induction of IFN in response to dsRNA, but did not prevent responsiveness to viral infection (25)(26)(27). Moreover, cytomegalovirus, which is capable of activating IRF-dependent gene expression (20,28), can do so from the cell surface without entering cells or replicating (29).
We have investigated the mechanisms of phosphorylation of IRF7 and its close relative, IRF3, during activation of IFN gene expression following Newcastle disease virus (NDV) infection. By all parameters tested, IRF3 and IRF7 appeared to be activated by the same mechanism. Their phosphorylation occurred with similar kinetics which required ongoing protein synthesis during viral infection, was blocked by the broad-spectrum kinase inhibitor staurosporine but not by inhibitors specific for individual kinases, and could be inhibited by the vaccinia virus protein, E3L. However, IRF3 and -7 phosphorylation was unaffected by the inactivation of the PKR gene, and IFN␣ gene expression was still induced in virus-infected PKR(Ϫ/Ϫ) and IKK␥Ϫ/Ϫ) cells, as well as in cells deficient for several PKRrelated genes. These data suggest that IRF7 activation can be stimulated by a viral component other than or in addition to dsRNA, that a novel cellular kinase distinct from PKR is responsible for its phosphorylation, and that the E3L protein exerts its anti-IFN effect by inhibiting not only PKR but also the distinct IRF3/7 kinase(s).
For kinase inhibition studies, inhibitors were added to growth media 5 h post-infection, unless otherwise indicated, at the following final concentrations: staurosporine, 500 nM; genistein, 300 M; H7, 45 M; H8, 45 M; acetylsalicylic acid, 5 mM; sodium salicylate, 5 mM. Dimethyl sulfoxide treatment was performed by adding 0.1% dimethyl sulfoxide by volume to growth media. Cycloheximide (75 g/ml) was added to cells at the indicated time points post-infection. Additional inhibitors were used at the concentrations listed in Table I.
Plasmid Constructs-Mouse IRF3 and IRF7 (3), and E3L and K3L expression vectors (34), the kind gift of Robert Schneider (New York University, NY), were driven by the cytomegalovirus immediate-early promoter. The K167A point mutation in E3L was created by PCRmediated site-directed mutagenesis using the following primers: sense, 5Ј-CGATAAGGCAGATGGAAAATCTGCTCGAGATGCTAAAAATAAT-GC-3Ј; antisense, 5Ј-GCATTATTTTTAGCATCTCGAGCAGATTTTCC-ATCTGCCTTATCG-3Ј. DRBP76 (35) and Staufen expression vectors (36) were the kind gifts of Ganes Sen (Cleveland Clinic, OH) and Juan Ortín (Centro Nacional de Biotecnologia, Madrid, Spain), respectively. The adenovirus VA gene was a kind gift of Robert Schneider (New York University, NY) and the MKK7(D) construct was a kind gift of Jan Vilcek (New York University).
Transfections and Viral Infections-293T cells were transfected by the calcium phosphate method (37). 12 h after transfection, plates of similarly transfected cells were pooled and distributed onto 60-mm plates. Cells were infected 12 h later with NDV, Manhattan strain (3), or influenza virus (38), as described. Unless otherwise noted, cells were harvested 7 h post-infection, and nuclear and cytoplasmic fractions were prepared, as described (39,40).
RNA Analysis-Total RNA was prepared using 3 ml of Trizol reagent (Life Technologies, MD) per 100-mm plate. Semi-quantitative reverse transcriptase-PCR analysis was performed as previously described (3).
Protein Analysis-Electromobility shift assays (EMSA) were performed using nuclear extracts, as described (6,40). Western blots were performed by standard techniques (41), except that E3L proteins were transferred to polyvinylidene difluoride with a 0.2-m pore size. Rabbit antisera to IRF3 and IRF7 (Zymed Laboratories Inc.) were used at 0.5 g/ml. Chicken anti-NDV (SPAFAS, North Franklin, CT) was used at a dilution of 1:1000. Monoclonal anti-E3L antibody Tw2.3 (42) was an ammonium sulfate fraction from supernatants of hybridoma cultures, the kind gift of Jonathan Yewdell (National Institutes of Health, Bethesda, MD). DRBP76 and Staufen proteins were detected using antibodies to epitope tags. In vitro kinase assays for IKK were performed by immunoprecipitating IKK from mouse fibroblasts, and incubating the recovered protein with 1 g of GST fusion protein substrates in 15 l of kinase buffer (20 mM Hepes, pH 7.6, 20 mM magnesium chloride, 20 mM ␤Ϫglycerophosphate, 20 mM p-nitrophenyl phosphate, 1 mM Na 4 VO 3 , 1 mM EDTA, plus protease inhibitors) in the presence of 10 Ci of [␥-32 P]ATP at 30°C for 30 min, followed by analysis by 10% SDS-PAGE and autoradiography. Antibody against IKK was obtained from Santa Cruz Biotechnology (Santa Cruz, CA). GST-IRF7 was expressed in bacteria from the pGEX-2T vector (43). GST-IB and mutant IB were kind gifts of Jan Vilcek (New York University).

IRF3 and IRF7 Can Be Phosphorylated with Equivalent
Kinetics during NDV Infection-IRF3 is constitutively expressed in cells while IRF7 is induced in response to an early wave of IFN production following viral infection (44). This differential expression complicates the question of whether these two proteins can be activated by the same kinase. To alleviate this technical shortcoming, we ectopically expressed IRF3 or IRF7 by transient transfection so that expression would be normalized under the control of an artificial promoter and phosphorylation could be measured independent from protein induction. Infection of transfected cells with NDV led to phosphorylation, as judged by a mobility shift following SDS-PAGE that is dependent on phosphorylation (3,6,14,16), of both IRF7 (Fig. 1A) and IRF3 (Fig. 1B). Moreover, both proteins were activated with equivalent kinetics, becoming phosphorylated at 5 h post-infection and showing maximal phosphorylation 7-8 h post-infection that remained high for several hours (data not shown). Acquisition of DNA binding activity, as judged by EMSA, correlated precisely with phosphorylation for both proteins (data not shown). We validated this transfection system by examining the kinetics of phosphorylation of endogenous IRF3 (Fig. 1D) which was phosphorylated with the same kinetics as the ectopically expressed protein. Therefore, the sequential activation of IRF3-and IRF7-target genes normally observed in untransfected cells following NDV infection can be accounted for by the requirement for induction of the IRF7 protein rather than by utilization of distinct kinases.
Considerable viral protein synthesis occurred within the first 7 h of infection, as judged by immunoblotting for viral proteins using antisera directed against NDV virions (Fig. 1C). This observation prompted us to consider whether protein synthesis, including viral protein synthesis, might be involved in IRF activation. To test this notion, protein synthesis was blocked by addition of cycloheximide at various times after infection. Inhibition of protein synthesis at any point up until ϳ3 h postinfection prevented phosphorylation of IRF7 (Fig. 2, lanes 3-5). However, addition of cycloheximide after that point was without effect (Fig. 2, lanes 6 -9). Consistent with the requirement of phosphorylation of IRF3 (16) and IRF7 for acquisition of DNA binding (6), cycloheximide added early in infection also blocked IRF7 activity as measured by EMSA (data not shown). These data suggest that proteins accumulating during the first 3 h of NDV infection are required for activation of the IRF7 kinase. Since no serine kinases are encoded by NDV (45), it is likely that viral proteins are involved either directly or indirectly in activation of a cellular kinase that targets IRF7. Alternatively, cycloheximide might block synthesis of a shortlived cellular protein.
No Specific Kinase Inhibitor Blocks IRF7 Phosphorylation-Cellular kinases potentially involved in IRF phosphorylation were investigated initially by use of pharmacological inhibitors. To distinguish a direct effect of inhibition on an IRF kinase from a possible indirect effect caused by, for instance, inhibition of necessary viral protein synthesis, pharmacological inhibitors were added late in infection (5 h post-infection), just prior to the observed appearance of phosphorylated protein and at a time point at which ongoing protein synthesis was no longer required. IRF7 phosphorylation was blocked by staurosporine (Fig. 3A, lane 4), a broad-spectrum kinase inhibitor previously shown to block phosphorylation of IRF1 in virusinfected cells (46). IRF7 phosphorylation was also reduced by the related inhibitor, K252a (Table I). However, none of the more specific kinase inhibitors tested were capable of preventing IRF phosphorylation ( Fig. 3A and Table I). These included inhibitors of PKR (2-amino purine (47)) and of IKK (acetylsalacylic acid and salacylic acid (48)), kinases known to be activated in response to virus infection and involved in activation of NFB. Other agents ineffective in blocking IRF phosphorylation included inhibitors of PKC, PKA, p38, extracellular kinase, phosphatidylinositol 3-kinase, and tyrosine kinases. The phosphorylation status of IRF7 correlated with its ability to bind DNA, as detected by EMSA (Fig. 3A, lower panel). Specificity of DNA binding was confirmed by antibody and DNA competition reactions (Fig. 3B)  IRF7 (lane 4), or homologous DNA competitor (lane 5). IRF3 phosphorylation was also blocked by staurosporine treatment (Fig. 3C, lane 3) but not by any other inhibitors tested, and none of the pharmacologic agents tested led to IRF phosphorylation in the absence of infection (data not shown). Phosphorylation of ectopically expressed IRF3 paralleled that of the endogenous protein whose phosphorylation was blocked by staurosporine and cycloheximide (Fig. 3C, lanes 6 and 7) but not by genistein, H7, H8, or wortmanin (lanes 8 -11).
Another technique used to probe the nature of the IRF kinase involved a search for agents that might block a virally activated cellular signaling step, or, alternatively, that might activate IRF7 phosphorylation in the absence of viral infection. We treated cells with DNA damaging agents, calcium chelators, proteosome inhibitors, phosphatase inhibitors, and activators of the unfolded protein response. We also overexpressed a constitutively active form of MKK7, an activator of the JNK pathway (49,50). None of these treatments caused IRF7 phosphorylation in the absence of viral infection nor prevented IRF7 activation when added to virus-infected cells, allowing us to rule out involvement of the known stress-and virus-activated signaling pathways in stimulating IRF7 phosphorylation (Table I).

Genetic Evidence for Lack of Involvement of PKR and IKK-
Because both PKR and IKK are known to be activated in response to viral infection, we sought additional evidence for their apparent lack of involvement in IRF phosphorylation by examining mutant cell lines lacking these enzyme activities. Embryo fibroblasts were prepared from mice devoid of the PKR gene due to gene targeting (26). Induction of type I IFN genes following NDV infection was assessed as a measure of endogenous IRF phosphorylation. IFN␣4 and IFN␤ expression, dependent on IRF3 phosphorylation, and expression of additional members of the IFN␣ gene family (non-IFN␣4 genes), dependent on IRF7 phosphorylation, were detected in approximately equal abundance in both wild type and PKR-null cells (Fig. 4A). Similar results were obtained using PKR-null cells derived from a different mouse strain that sustained a distinct mutation (27), corroborating the results obtained with inhibitors (data not shown). However, IFN␣ induction in response to dsRNA as opposed to virus infection was impaired in PKR-null cells (data not shown), as previously reported (25).
Phosphorylation of IRF3 and IRF7 was directly tested in virus-infected, PKR-null cells. Flag epitope-tagged IRF3 or IRF7 were transfected into PKR-null cells which were subsequently infected with NDV and labeled with [ 32 P]inorganic phosphate. Protein extracts were immunoprecipitated and analyzed by SDS-PAGE and autoradiography. Phosphorylated IRF3 or IRF7 was recovered from PKR-mutant cells (Fig. 4B), Inhibition of IRF7 phosphorylation in response to NDV infection is indicated by (ϩ). K252a treatment (ϩ/Ϫ) resulted in partial inhibition at the highest concentration tested. None of these agents caused IRF7 phosphorylation in the absence of NDV infection. demonstrating the PKR independence of this process. It should be noted that both IRF3 and IRF7 are basally phosphorylated in the absence of viral infection and become additionally phosphorylated in response to infection, as indicated by their change in electrophoretic mobility. Neither basal nor induced phosphorylation was inhibited by the absence of the PKR gene.
Another enzyme known to be activated in response to virus infection and required for activation of NFB is IKK (12). To confirm the inhibitor studies that suggested the lack of its involvement in IRF phosphorylation, we took advantage of a mutant cell line lacking the IKK␥/NEMO subunit that is therefore completely defective for IKK activity (11). Consistent with a lack of requirement for IKK activity for IRF phosphorylation and no requirement of NFB function for IFN␣ gene transcription, IFN␣ gene expression was induced normally in NDVinfected mutant cells (Fig. 4C). As expected, IFN␤ induction was impaired (Fig. 4C, lane 4), due to its demonstrated requirement for NFB (51). The modest reduction in IFN␣ gene induction is likely due to the absence of IFN␤-dependent positive feedback (3,52). Finally, we asked whether IKK was capable of phosphorylating IRF7, even though it was not required. Activated IKK was immunoprecipitated from extracts of virusinfected cells and tested for its ability to phosphorylate IRF7 in vitro. Although phosphorylation of IB was readily detected with virus-activated IKK (Fig. 4D, lane 2) to levels similar to that achieved with IKK isolated from interleukin 1␤-treated cells (Fig. 4D, lane 1), no phosphorylation of IRF7 was observed using the virus-activated kinase (Fig. 4D, lane 4). A mutant form of IB in which serine residues 23 and 25 were changed to alanine (IBm) served as specificity control for activation of IKK (Fig. 4D, lane 3).
Several genes have been isolated recently that bear homology to PKR. One possibility raised by the ability of PKR-null cells to phosphorylate IRF proteins in response to virus infection while being deficient in response to dsRNA might be the existence of PKR-related enzymes that are activated by a viral component other than dsRNA. To test this notion, we examined IFN gene expression in response to virus infection of cells deficient for individual PKR-related genes due to gene targeting. Cells deficient for PERK, an eIF2␣ kinase (31), for IRE1␣ and -␤, enzymes involved in activation of JNK (32), or for mouse GCN2, another eIF2␣ kinase (33,53,54), were all capable of producing IFN␣, IFN␤, and the non-IFN␣4 subset of IFN␣ genes in response to viral infection (Fig. 5). Therefore, none of these PKR-related kinases is individually required for IFN gene induction and is unlikely to be a required IRF kinase.
IRF7 Phosphorylation Is Blocked by Vaccinia Virus E3L Protein-Viruses maintain numerous strategies to evade host innate immune defenses, including inhibition of PKR (55). Three such viral inhibitors that target inhibition of PKR are vaccinia virus E3L and K3L proteins and adenovirus VA RNA (56 -62). To test the potential of these genes as inhibitors of IRF phosphorylation, cells were co-transfected with E3L, K3L, or VA expression constructs along with IRF3 or IRF7, infected with NDV, and assayed for IRF phosphorylation by SDS-PAGE. Expression of K3L or VA were without effect (Fig. 6A, and data not shown); in contrast, expression of E3L blocked IRF7 phosphorylation in a dose-dependent manner (Fig. 6A,  top panel). Coexpression of E3L with IRF7 also blocked its ability to bind DNA following isolation from virus-infected cells (Fig. 6A, lower panel), likely a reflection of the lack of phosphorylation. Coexpression of K3L was without effect in either assay (Fig. 6A).
E3L binds dsRNA and has been thought of as a competitive inhibitor of PKR that sequesters dsRNA (56,60) and directly inhibits PKR through protein-protein interaction following binding to dsRNA (63,64). We tested the requirement of dsRNA binding for the inhibition of IRF7 phosphorylation by E3L using a point mutant version of E3L, in which lysine 167 was converted to alanine (K167A), and is incapable of binding dsRNA (65). Loss of RNA binding by E3L disrupted its ability to inhibit IRF7 phosphorylation (Fig. 6B, lane 6). To evaluate whether direct sequestering of dsRNA might be the sole mechanism of the inhibitory function of E3L, we tested two additional dsRNA-binding proteins. Expression of neither DRBP76 (35) nor mammalian Staufen (66) prevented IRF7 phosphorylation, nor did they block the inhibitory action of E3L ( Fig. 6C and data not shown), despite these proteins being implicated in regulation of PKR and viral infection (35,36,66). Additionally, E3L but not K3L effectively inhibited IRF3 phosphorylation following NDV infection (Fig. 6D). Therefore, it is likely that E3L directly inhibits a cellular kinase that targets both IRF3 and IRF7 rather than functioning merely to sequester dsRNA, although its ability to bind dsRNA is necessary for this inhibitory function, as it is for inhibition of PKR.

DISCUSSION
Activation of IFN gene expression is a cellular response resulting from innate immune recognition of virus infection, but how cells sense viral replication is only poorly characterized. While production of viral products such as dsRNA play a role in activation of IKK and possibly of JNK (12), the data reported here suggest that additional signals stimulate phosphorylation of IRF3 and IRF7, essential events in induction of the protective IFN response. This evidence is severalfold. First, the known target of dsRNA in mammalian cells, PKR, was not required for stimulation of IRF3 or -7 phosphorylation. Neither pharmacological inhibitors of PKR nor ablation of the PKR gene resulted in impaired IRF phosphorylation. Second, while PKR-mutant cells were defective in responses to dsRNA, they retained responsiveness to viral infection, demonstrating that infection provided additional signals beyond simply dsRNA. Finally, protein synthesis during viral infection was required for IRF phosphorylation, and the kinetics of phosphorylation and of the required protein synthesis correlated well with the major synthesis of viral proteins. Therefore, it is likely that newly synthesized viral proteins play a necessary role in activation of the cellular IRF kinase. A model for regulation of IRF phosphorylation is shown in Fig. 7. Viral infection leads to production of dsRNA and activation of PKR. Infection also leads to an independent activation event in a protein synthesisdependent manner, targeting a novel cellular kinase (X) and leading to IRF phosphorylation. dsRNA-activated PKR might directly phosphorylate IRF or might activate the same kinase X, leading to IRF phosphorylation. We cannot rule out the possibility that the required viral protein synthesis induces a secondary product (possibly an RNA) that is the ultimate activator of the IRF kinase. It is also possible that synthesis of a cellular protein is necessary during viral infection. In either case, these events are independent of PKR and its close homologues PERK, IRE1␣/␤, and GCN2.
Our data are consistent with a single kinase targeting both IRF3 and IRF7. By studying transfected cells, we avoided the requirement for IRF7 induction that normally limits its action to a secondary wave of IFN production (3). Under these experimental conditions, both IRF3 and IRF7 were phosphorylated with the same kinetics, and endogenous IRF3 was phosphorylated with similar kinetics, validating the transfection system. Moreover, IRF3 and IRF7 phosphorylation showed the same sensitivity and resistance to a panel of pharmacological agents. Indeed, only the broad-spectrum kinase inhibitors staurosporine and to a lesser extent K252a were effective IRF kinase inhibitors. Although staurosporine was first characterized as an inhibitor of PKC, other inhibitors of PKC or down-regulation of PKC protein by prolonged exposure to phorbol ester were ineffective in preventing IRF phosphorylation. Other cellular stress responses, such as induction of the unfolded protein response or stimulation of MAP kinase cascades, did not appear to play a role in the IRF response to virus, suggesting that a novel cellular signaling pathway allows the innate immune system to recognize viral infection. Hiscott and colleagues have recently reached a similar conclusion (67).
Despite the lack of evidence for dsRNA and PKR involvement in IRF phosphorylation by virus, we found that the vaccinia virus PKR inhibitor E3L, but not K3L, was an effective IRF phosphorylation inhibitor (Fig. 6). Moreover, the ability of E3L to bind RNA was necessary for its anti-IRF action. These data suggest that dsRNA is one and possibly an essential aspect of the recognition of viral infection. However, while the action of dsRNA alone is dependent on the activity of PKR, the action of dsRNA in the context of other components of viral replication acted in a PKR-independent manner, although it remained sensitive to E3L inhibition. These findings suggest that an enzyme related to PKR could be a likely candidate for the virus-activated IRF kinase; however, none of the known PKR-related genes tested was individually required, including PERK, IRE1␣, IRE1␤, and GCN2. It remains a possibility that another PKR-related enzyme, for instance, HRI (68) or an as yet to be identified related kinase, is the IRF kinase. Alternatively, a combination of these genes operating in a redundant fashion could have obscured the requirement for any one individual enzyme.
Whether IRF phosphorylation depends on a novel gene or a combination of PKR-related enzymes, it is likely that E3L protein directly inhibits the catalytic function of such an enzyme, as it does PKR. Inhibition by E3L likely required the presence of dsRNA, possibly serving to activate the inhibitory function of E3L. However, sequestering of dsRNA alone is probably insufficient to inhibit IRF phosphorylation since other RNA-binding proteins were unable to mimic this action of E3L. Interestingly, the non-RNA binding amino terminus of E3L is required for full inhibition of PKR (63,64) and must interact with PKR, not simply sequester its activator dsRNA to block catalytic activity. The amino terminus of E3L also induces multimerization (69) that may be necessary for its function. In contrast, the K3L protein inhibits PKR catalysis by acting as a psuedo-substrate, mimicking the PKR substrate eIF2␣ and blocking the catalytic site of the enzyme. The finding that K3L was not an effective inhibitor of the IRF3/7 kinase suggests a  panel (lanes, 2, 4, and 6). C, DRBP76 does not inhibit IRF7 phosphorylation. Cells were analyzed as described for panel B following transfection with Flag-tagged DRBP76, as indicated. D, IRF3 phosphorylation is inhibited by E3L. 293T cells were transfected with 10 g of an IRF3 expression vector and 10 g of E3L (lanes 3 and 4) or K3L (lanes 5 and 6), infected with NDV, as indicated, and nuclear extracts were analyzed for IRF3 by immunoblotting. FIG. 7. Proposed model of IRF3 and -7 phosphorylation. Virusinduced dsRNA activates PKR while a second, protein synthesis-dependent step activates an unknown kinase X which phosphorylates IRF3 and -7. Activated PKR either directly or indirectly through kinase X also leads to IRF3 and IRF7 phosphorylation. E3L protein blocks both PKR and kinase X in a dsRNA-dependent fashion. substrate specificity distinct from PKR. The ability of these two viral proteins to target independent aspects of the IFN pathway may explain why both genes have been maintained in the vaccinia virus genome.
The action of E3L to inhibit IRF3 and IRF7 phosphorylation in response to NDV infection is reminiscent of the recently reported action of the influenza viral protein NS1 to inhibit IRF3 nuclear translocation (70). Indeed, we also observed inhibition of IRF7 phosphorylation in the presence of NS1 protein (Table I). We would speculate that both E3L and NS1, and possibly additional, virally-encoded inhibitors of IFN induction, function by binding and inhibiting a cellular IRF protein kinase which is activated in response to dsRNA plus additional viral components.