Protein Kinase PKR Amplification of Interferon β Induction Occurs through Initiation Factor eIF-2α-mediated Translational Control*

Background: RNA-dependent protein kinase PKR regulates interferon induction. Results: Knockdown of PKR or mutation of initiation factor eIF-2α results in increased IκB-α protein levels and decreased IFN-β induction in RNA-transfected and virus-infected cells. Conclusion: PKR amplifies IFN-β induction through an eIF-2α translational control response. Significance: PKR functions together with additional RNA sensors to modulate signaling leading to interferon gene induction. The protein kinase PKR is activated by RNA with double-stranded (ds) structure and subsequently impairs translation through phosphorylation of protein synthesis initiation factor eIF-2α. PKR also mediates activation of signal transduction pathways leading to interferon beta (IFN-β) gene induction following virus-infection or RNA transfection. We previously demonstrated in measles virus-infected cells that PKR is required for the maximal induction of IFN-β gene expression by the interferon promoter stimulator gene 1 (IPS-1) adaptor-dependent cytosolic RNA sensor pathway. While both IPS-1 and PKR are important mediators of IFN-β induction, with PKR contributing to an enhanced NF-κB activation, the mechanism by which PKR enhances NF-κB activity and amplifies IFN-β induction is unresolved. Herein we tested the possibility that PKR could activate signal transduction pathways indirectly through translational control responses. Following transfection with synthetic or natural dsRNAs or infection with measles virus, we observed increased mRNA but decreased protein levels for the inhibitor of NF-κB signaling, IκB-α, that correlated with PKR activation and eIF-2α phosphorylation. Importantly, knockdown of PKR increased IκB-α protein levels and impaired IFN-β induction. Additionally, inhibition of translation by cycloheximide treatment rescued IFN-β induction following PKR knockdown but not IPS-1 knockdown. Mutation of eIF-2α to prevent phosphorylation also impaired IFN-β induction in PKR-sufficient virus-infected cells. These results suggest that an eIF-2α-dependent translation inhibition mechanism is sufficient to explain the PKR-mediated amplification of IPS-1-dependent IFN-β induction by foreign RNA.

been observed to activate NF-B signaling by activating IB kinase or NF-B-inducing kinase (20,21) and to interact with TRAF family of protein members (22), however conflicting results were found with two different mouse pkr knock-outderived cell lines regarding NF-B activation by PKR (23). In addition to a role for PKR in activation of NF-B signaling, PKR is implicated in the activation of proteins including the phosphatase PP2A, p53, stress-activated JNK and p38 kinases, and IRF-1 (19). The mechanism by which PKR activates these diverse pathways largely remains to be elucidated.
We have shown that PKR is required for maximal IFN-␤ induction in HeLa cells and that this induction of IFN-␤ gene expression is dependent upon the RLR interferon promoter signaling 1 (IPS-1) adaptor protein, but not the TLR3 adaptor protein TIR-domain-containing adaptor-inducing IFN-␤ (TRIF) protein (24). Transcriptional induction of IFN-␤ is known to require the activation of both IRF and NF-B transcription factors downstream of the TLRs and RLRs (25,26). We earlier established that while PKR is not required for IRF3 activation following virus infection or RNA transfection, PKR is required for maximal NF-B activity (24,27). The NF-B family members involved in type I IFN signaling include the p65 (RelA) and p50 subunits. These proteins heterodimerize and are kept in an inactive state in the cytoplasm of cells by a family of inhibitors of NF-B proteins, among which is IB-␣ (26). NF-B signaling is activated following the phosphorylation, ubiquitination and degradation of IB proteins by the inhibitor of NF-B kinase (IK) complex (28,29). Following degradation of the inhibitor, NF-B subunits localize to the nucleus and bind sites present within the promoter region of NF-B regulated genes. Among these is IB-␣, whose promoter possesses multiple NF-B binding sites, and following NF-B activation, IB-␣ thereby negatively regulates NF-B signaling (29,30).
To probe the mechanism by which PKR amplifies IFN-␤ induction, we tested the possibility that PKR-mediated inhibition of translation triggered by foreign RNA was sufficient to enhance NF-B activity and amplify IFN-␤ transcript induction. We found that the PKR-dependent inhibition of IB-␣ protein expression in RNA-transfected and measles virus-infected cells correlated with increased phosphorylation of PKR and eIF-2␣. This response was accompanied by enhanced expression of IB-␣ and IFN-␤ RNAs, but reduced levels of IB-␣ protein. Knockdown of PKR or expression of a serine-toalanine mutant initiation factor eIF-2␣ in PKR-sufficient cells resulted in increased IB-␣ protein expression and decreased IFN-␤ induction following RNA transfection or virus infection. Cycloheximide treatment, which inhibits translation independently of eIF-2␣ phosphorylation, rescued IFN-␤ RNA induction in PKR knockdown cells, but importantly did not rescue IFN-␤ induction following IPS-1 knockdown. These results suggest that PKR amplification of IFN-␤ RNA induction in response to foreign RNA occurs through a translational control mechanism.
Recombinant parental measles virus (MV) MVvac GFP(H) designated wild-type (WT) virus, and isogenic mutant viruses either V-deficient (V ko ) or C-deficient (C ko ), were generously provided by R. Cattaneo (Mayo Clinic, Rochester, MN) and were as previously described (33). The viruses were constructed based on the Moraten vaccine strain (34,35), except that a gene encoding green fluorescent protein (GFP) was inserted downstream of the H gene.
For MV infection, PKR ϩ or PKR kd cells were seeded into 12-well plates 24 h prior to infection. Cell monolayers were washed once with Opti-MEM I and then infected with WT, V ko , or C ko MV at a multiplicity of infection (MOI) of five 50% tissue culture infective doses (TCID 50 )/cell (35,36).
Transient Knockdown Experiments-Chemically synthesized siRNAs targeting PKR, IPS-1, and TRIF and a control siRNA targeting firefly luciferase (siLUC) were as previously described (24), and were purchased from Dharmacon. For transient knockdown experiments, a double transfection strategy with siRNA was utilized (27). HeLa cells in 60-mm dishes at ϳ60% confluency were transfected with 50 nM of siRNA with Lipofectamine 2000 on day 1 and again on day 3. On day 2, the cells were reseeded in 60-mm dishes for the second siRNA transfection, and on day 4 the cells were seeded in 12-well plates for inducer RNA transfection on day 5.
Double-stranded RNAs and Transfection-Synthetic poly(rI)-poly(rC) was from Sigma. Reovirus genome dsRNA was isolated from purified reovirions as previously described (37). T7 RNA polymerase synthesized 20-bp (Pds20) and 50-bp (Pds50) dsRNAs were made and purified using the Silencer siRNA construction kit (Ambion) according to the manufacturer's protocol. Further gel purification of the RNAs synthesized in vitro with T7 RNA polymerase was as described by Sambrook et al. (38). The sequence of the Pds20 was the same as the chemically synthesized siLUC; the sense sequence of the T7 transcribed 50-bp Pds50 included this sequence and firefly luciferase sequence 3Ј of this region, as previously described (24). For RNA transfections, cells were seeded in 24-or 12-well plates and transfected with the indicated amount of RNA complexed with Lipofectamine 2000 in Opti-MEM I.
DNA Plasmids and Transfection-The expression constructs for human eIF-2␣, mutated at serine 51 to either alanine (S51A) or aspartic acid (S51D) in the pMSCV-HA3iresGFP vector, were generously provided by M. Hatzoglou (Case Western University, Cleveland, Ohio). The empty vector (Vec) was generated by deletion of the eIF-2␣ cDNA insert from the S51D expression plasmid using EcoRI and XhoI. The eIF-2␣ double mutant construct S48,51A was generated by mutating serine 48 to alanine in the S51A plasmid background using the Quick-Change site-directed mutagenesis strategy (Stratagene) and the following primers: forward 5Ј-GAAGGCATGATTCTTCTT-GCTGAATTGGCCAGAAGGCG-3Ј and reverse 5Ј-CGC-CTTCTGGCCAATTCAGCAAGAAGAATCATGCCTTC-3Ј (mutations are designated by underlined font). Constructs were verified by direct sequencing and restriction enzyme analysis. HeLa cells in 6-well plates were transfected with the indicated eIF-2␣ expression plasmid or empty vector (1.6 g/well) using Fugene HD transfection reagent (Roche) according to the manufacturer's protocol. After 20 h the transfected cells were in-fected with C ko virus at a MOI of 1, maintained in DMEM containing 5% (V/V) fetal bovine serum for 24 h, and then RNA was isolated for qPCR analysis or extracts prepared for Western immunoblot analysis.
Cyclohexamide Treatment-For cycloheximide (CHX) (Sigma) treatment prior to protein or RNA isolation, cells were incubated with 10 g/ml of CHX as indicated for 4 -6 h in Opti-MEM I with or without transfection complexes. In experiments where the time to harvest was greater than 4 h, CHX containing maintenance media was placed on cells following the removal of transfection complexes at 4 h. Cells were then harvested for RNA isolation for qPCR analysis or extract preparation for Western immunoblot analysis.
Real Time PCR Analyses-Total cellular RNA was purified at the indicated time after transfection or infection using an RNeasy mini kit (Qiagen) for 24-well plates or Trizol (Invitrogen) for 12-well plates following the manufacturer's protocol. Random-primed reverse transcription was carried out using 500 ng of RNA and SuperScript II (Invitrogen) according to the manufacturer's protocol. For qPCR reactions, the GAPDH and IFN-␤ primer pairs were as previously described (27). qPCR reactions were performed in duplicate or triplicate with each RT template, using IQ SYBR Green Supermix (Bio-Rad) and a Bio-Rad MyIQ multicolor real-time qPCR instrument (3-min hot start followed by 30 s at 95°C, 45 s at 58°C, 45 s at 72°C, for 40 to 45 cycles). IFN-␤ values were normalized to GADPH values.

DsRNA Length-dependent Induction of IFN-␤ and PKR-dependent Reduction of IB-␣-We previously showed that T7
phage RNA polymerase synthesized RNAs (PRNAs) induce IFN-␤ in a size-dependent manner (24) as also has been shown by others (41,42). This size-dependent induction is illustrated in Fig. 1A, where the 50-bp dsRNA Pds50 induced IFN-␤ in HeLa cells as measured by qPCR, while the 20-bp dsRNA Pds20 did not induce IFN-␤. The long (Ͼ100 bp) synthetic dsRNA pIpC, tested as a positive control, induced IFN-␤ similar to Pds50.
Because NF-B is one of the transcription factors required for IFN-␤ induction and the activation of NF-B and induction of IFN-␤ are enhanced by PKR in measles virus-infected and RNA-transfected cells (24,27,43) we hypothesized that differential NF-B signaling may be responsible in part for the differences in IFN-␤ expression triggered by the different-sized dsRNAs. We previously observed that the inhibitor of NF-B signaling, IB-␣, was degraded following Pds50 and pIpC transfection (24). To compare this effect with Pds20, we measured the protein levels of IB-␣ following either Pds20 or Pds50 transfection in both parental PKR ϩ HeLa cells and the PKR stable knockdown cells. We observed that a decrease in IB-␣ was only seen following transfection with Pds50 and not Pds20, and that this IB-␣ decrease was PKR-dependent ( Fig. 1, B and C). Furthermore, the decrease in IB-␣ seen in transfected cells correlated with enhanced eIF-2␣ phosphorylation (Fig. 1B).

Measles Virus C ko Mutant Infection Reduces IB-␣ Protein but Increases IB-␣ mRNA in a PKR-dependent Manner-Us-
ing wild type (WT) measles virus or isogenic mutants lacking either the V (V ko ) or C (C ko ) viral accessory proteins that impair innate immune system activation (44), we previously established that the C ko virus induces the highest levels of IFN-␤ (27). We also observed significant PKR activation measured by Thr-446 phosphorylation, but no change in total PKR level following C ko virus infection, and we showed that maximal induc- A, relative IFN-␤ induction measured by quantitative PCR (qPCR) using RNA extracted 5 h following incubation with lipid alone (NT) or RNA transfection in parental PKR ϩ HeLa cells. IFN-␤ mRNA levels were normalized to GAPDH using the ⌬⌬CT method. Standard deviation determined from three independent experiments. B, Western blot for IB-␣, phosphorylated eIF-2␣ (P-eIF-2␣) and total eIF-2␣, and ␣-tubulin loading control, at 4 h and 8 h following transfection of either parental PKR ϩ or knockdown PKR kd HeLa cells with the indicated dsRNA. C, quantitation of IB-␣ protein level normalized to ␣-tubulin and compared with cells not transfected (NT). Mean and standard deviation determined from two independent experiments. *, p Ͻ 0.05 compared with NT and Pds20 at the corresponding time point in PKR ϩ cells, **, p Ͻ 0.01 compared with NT and Pds20.
tion of IFN-␤ required PKR and IPS-1. Using gel shift and NF-B luciferase reporter assays, we demonstrated that the measles virus C ko mutant activated NF-B signaling, and that maximal NF-B activation was dependent upon both PKR and IPS-1 (27).
To test whether the PKR-dependent decrease in IB-␣ protein that we observed following dsRNA transfection ( Fig. 1) also occurred in the context of viral infection, we examined the IB-␣ level in cells infected with WT, V ko or C ko measles virus at the 24 h time point where we previously observed maximal IFN-␤ induction, PKR phosphorylation, and NF-B activation (27,33). Following infection, C ko virus-infected cells displayed a lower IB-␣ protein level than uninfected cells or cells infected with either the WT or V ko virus in a PKR-dependent manner ( Fig. 2A).
We next examined IB-␣ RNA levels by qPCR in order to determine the relationship between the levels of IB-␣ protein and IB-␣ mRNA, and to assess whether the reduction in IB-␣ protein was because of a reduction in IB-␣ RNA. While untreated and WT virus-infected cells showed no significant change in IB-␣ mRNA transcript levels and the V ko virus only slightly increased IB-␣ mRNA in the parental PKR ϩ cells, the C ko virus significantly increased IB-␣ mRNA levels in a PKRdependent manner (Fig. 2B). IB-␣ protein levels were lowest in PKR ϩ cells following C ko virus infection even though the IB-␣ mRNA level was ϳ7-fold higher in the PKR ϩ cells than PKR kd cells. The enhanced induction of IB-␣ mRNA in PKR ϩ cells was not unexpected, because NF-B is known to induce IB-␣ (30) and, as described above, PKR ϩ cells display enhanced NF-B signaling. These results suggest that PKR either enhanced IB-␣ protein degradation or the inhibition of IB-␣ protein synthesis.
PKR-dependent Decrease of IB-␣ Protein and Increase of IB-␣ mRNA Occurs following Transfection with Naturally Occurring dsRNA-We next examined the PKR dependence of decreased IB-␣ protein and the relationship between the level of IB-␣ mRNA and IB-␣ protein in PKR ϩ and PKR kd cells transfected with purified reovirus genome RNA. Reovirus genome is composed of ten segments of dsRNA of lengths varying from 1.2 to 3.8 kbp (45), and is not known to include inosine as is present in the synthetic pIpC dsRNA. Reovirus dsRNA activates PKR (37) and both the RIG-I and MDA5 cytosolic RNA signaling pathways which utilize the IPS-1 adaptor protein (42). Following reovirus RNA transfection, we found a significant reduction in IB-␣ protein only in the PKR ϩ cells and not in the PKR kd cells (Fig. 3A). This reduction was observed between 2 and 12 h after transfection and was seen only in PKR ϩ cells where enhanced eIF-2␣ phosphorylation was observed (Fig. 3A).
Following transfection with either pIpC or reovirus genome dsRNA, the level of IB-␣ mRNA was highest in PKR ϩ cells and was reduced significantly in PKR kd cells (Fig. 3B), just the opposite of what we observed for IB-␣ protein levels (Fig. 3A). We next confirmed this result obtained with PKR kd cells stably deficient in PKR by using cells transiently knocked down for PKR and observed a similar reduction in IB-␣ mRNA was seen following reovirus genome RNA transfection (Fig. 3C). We also transiently knocked down IPS-1 to test whether IB-␣ mRNA induction was reduced, since IPS-1-dependent signaling has been shown to activate the IK complex through the NF-B modulator (NEMO, also known as IK-␥) protein (46). IPS-1 knockdown but not TRIF knockdown also reduced IB-␣ mRNA induction in response to transfected reovirus genome RNA (Fig. 3C), suggesting that NF-B signaling is activated, or at least enhanced, by PKR and IPS-1-dependent responses.
Inhibition of Translation by Cycloheximide in PKR kd Cells Reduces IB-␣ Protein Levels to That Seen in PKR ϩ Cells following dsRNA Transfection-The mechanism by which PKR enhances NF-B activation (27) and NF-B-dependent gene induction (Figs. 2 and 3) could involve the direct activation of NF-B signaling components. Alternatively, an indirect translational control mechanism also may be operative in which the activity of NF-B signaling would be affected through a PKRdependent eIF-2␣ phosphorylation and subsequent inhibition of synthesis of negative regulatory proteins such as IB-␣. For the latter explanation, IB-␣ protein degradation could be initially similar in PKR ϩ and PKR kd cells (47), but subsequent synthesis of IB-␣ protein would be reduced in PKR ϩ cells due to an inhibition of IB-␣ mRNA translation compared with that in PKR kd cells. The observed disparity between IB-␣ mRNA and protein levels (Figs. 2 and 3) is consistent with the notion that PKR may enhance activation of NF-B signaling through a translational control mechanism.  OCTOBER 19, 2012 • VOLUME 287 • NUMBER 43

JOURNAL OF BIOLOGICAL CHEMISTRY 36387
To test the possibility that PKR decreases IB-␣ protein through translational control thereby leading to the increase seen in IB-␣ mRNA (Fig. 3), we treated PKR kd cells with cycloheximide (CHX), a drug that prevents the elongation step of translation (48), and then examined IB-␣ protein levels. In PKR kd cells, CHX treatment significantly decreased IB-␣ protein levels (Fig. 4, lanes 11 and 12) to a level comparable to that of PKR ϩ cells, suggesting that PKR indeed might contribute to NF-B induction solely by preventing the translation of IB-␣ mRNA (47). No changes in eIF-2␣ phosphorylation were observed in PKR kd cells following CHX treatment.

Cycloheximide Treatment Rescues IFN-␤ RNA Induction Following PKR-knockdown, but Does Not Rescue Following IPS-1
Knockdown-Most importantly, CHX treatment rescued IFN-␤ RNA induction in PKR kd cells to a significant extent (p Ͻ 0.01) in response to transfected synthetic (pIpC) or natural (reovirus genome) dsRNA (Fig. 5A), indicating that inhibition of synthesis of inhibitors of NF-B signaling was a likely explanation for the PKR dependence of IFN-␤ induction. CHX treatment also restored IB-␣ RNA induction in PKR kd cells to levels found in PKR ϩ cells (Fig. 5B).
While our results are consistent with an amplification of IFN-␤ RNA induction by PKR-mediated translation inhibition of IB-␣ protein synthesis, they do not clearly differentiate between the IPS-1 and PKR contributions to IFN-␤ signaling. To test whether CHX treatment specifically rescues PKR-dependent but not IPS-1-dependent induction of IFN-␤ transcripts, we transiently knocked down either PKR or IPS-1 in parental HeLa cells. As we earlier reported (24), transient knockdown of either IPS-1 or PKR significantly reduced IFN-␤ induction (Fig. 5C). Following CHX treatment, however, IFN-␤ RNA induction increased to near control levels in the PKR transient knockdown cells, but remained significantly reduced in the IPS-1 knockdown cells (Fig. 5C). These results suggest that the translation inhibition-dependent contribution of PKR to IFN-␤ induction is independent of the activation of signaling by IPS-1.

Non-phosphorylatable Mutant eIF-2␣ Reduces IFN-␤ RNA Induction and Increases IB-␣ Protein Levels Following Infection of PKR-sufficient Cells-
To further test whether the PKRdependent enhancement of IFN-␤ transcript induction was translationally mediated through eIF-2␣, the effect of S48,51A, an unphosphorylatable mutant form of eIF-2␣ in which the serine phosphorylation sites were replaced with alanine, was examined. As shown in Fig. 6, the phosphorylation defective S48,51A mutant resulted in lower levels of both IFN-␤ transcript (Fig. 6A) and IB-␣ transcript (Fig. 6B) following C ko measles virus infection compared with either vector-transfected cells or cells expressing the phospho-mimetic S51D mutant. IFN-␤ transcript levels in the absence of the C ko infection were comparably low in the vector, S51D and S48,51Atransfected cells (Fig. 6).
Expression of the S48,51A phosphorylation defective mutant form of eIF-2␣ in PKR ϩ parental cells also led to enhanced IB-␣ protein production in C ko virus-infected cells (Fig. 7A,  lane 6), reaching a level of IB-␣ comparable to that seen in C ko -infected PKR kd cells (Fig. 7A, lane 8). By contrast, the level of IB-␣ in C ko -infected vector-or S51D-transfected cells was 2-3-fold lower (Fig. 7B). Interestingly, PKR ϩ cells expressing the phosphorylation defective S48,51A mutant also gave rise to increased viral H protein synthesis, as was observed in PKR kd cells, compared with the vector control or S51D-expressing cells (Fig. 7A).

FIGURE 3. PKR-dependent reduction in IB-␣ protein level and increase in IB-␣ mRNA level following transfection with reovirus genome dsRNA.
A, Western blot for IB-␣, phosphorylated eIF-2␣ (P-eIF-2␣) and total eIF-2␣, and ␣-tubulin loading control, at the indicated times after transfection of PKR ϩ or PKR kd cells with purified reovirus genome dsRNA. Percentage (%) indicates IB-␣ protein level normalized to ␣-tubulin and compared with cells not transfected. B, IB-␣ mRNA transcript level measured by qPCR with RNA isolated from PKR ϩ and PKR kd cells at 5 h following transfection with reovirus genome dsRNA or pIpC. Error bars represent the standard deviation determined from three independent experiments. C, qPCR measurement of IB-␣ mRNA transcript level following transfection with reovirus genome dsRNA of HeLa cells transiently knocked down for PKR, IPS-1 or TRIF or transfected with a control siRNA (siLUC). Standard deviations determined from two independent experiments analyzed in triplicate. *, p Ͻ 0.05; **, p Ͻ 0.01.

DISCUSSION
PKR has emerged as an important modulator of IFN-␤ gene induction following infection with RNA viruses, including negative-stranded (27,43), positive-stranded (49 -53) and doublestranded (54) RNA viruses. But, the molecular mechanism by which PKR affects IFN-␤ induction largely remains unresolved. The objective of our study was to determine the mechanistic basis by which the PKR kinase enhances IFN-␤ transcript RNA levels following virus infection and RNA transfection. The results reported herein suggest that a translational control mechanism through eIF-2␣ is likely sufficient to account for the PKR-dependent reduction in IB-␣ levels and enhancement of IFN-␤ induction. In fact, translational control may be broadly important and operative in multiple pathways of IFN signaling as the translational repressors 4E-BP1 and 4E-BP2 also have been shown to negatively regulate the type-I IFN response by inhibiting IRF7 mRNA translation during viral infection (55).
Impaired IFN-␤ transcript induction in HeLa cells stably knocked down for PKR is known following dsRNA transfection (24) or measles virus infection (27,43). Using a transient knockdown strategy we earlier had established that, in addition to PKR, the IPS-1 adaptor protein for the RLR cytosolic signaling pathway was required for maximal IFN-␤ induction by measles FIGURE 5. Translation inhibition by cycloheximide rescues PKR-dependent but not IPS-1-dependent induction of IFN-␤ transcripts. A, RNA was isolated from PKR ϩ or PKR kd cells at 5 h after transfection with pIpC or reovirus genome dsRNA in the presence or absence of CHX as indicated and the relative IFN-␤ mRNA transcript level was determined by qPCR as described in Fig. 1. The standard deviation was determined from two or three independent experiments. B, qPCR measurement of relative IB-␣ mRNA transcript levels following treatment conditions described in A. C, relative IFN-␤ mRNA transcript level in parental PKR ϩ HeLa cells transiently knocked down for either PKR or IPS-1, or transfected with a control siLUC siRNA prior to transfection of reovirus genome dsRNA in the absence or presence of CHX. Standard deviation is shown for two independent experiments. *, p Ͻ 0.05; **, p Ͻ 0.01.

increases IB-␣ protein levels in C ko measles virus infected PKR ؉ cells to levels seen in PKR kdinfected cells.
A, PKR ϩ or PKR kd cells were transfected with either an empty vector (Vec) or an expression construct encoding either S48,51A or S51D mutant eIF-2␣ as indicated. At 20 h after transfection cells were either left uninfected (Ϫ) or infected (ϩ) with MV C ko virus, and at 24 h after infection extracts were prepared and western immunoblot analyses performed using antibodies against IB-␣, MV H protein and eIF-2␣, and ␣-tubulin as a loading control. B, quantitation of IB-␣ protein levels normalized to ␣-tubulin; average and standard deviation determined from three independent analyses; **, p Ͻ 0.01.
virus deficient in C protein, but that the TRIF mediator of TLR3 signaling was not involved. Because both PKR and IPS-1 were necessary for maximal IFN-␤ transcript induction in response to either dsRNA transfection or MV infection (24,27), we attempted to further delineate and distinguish the requirements for these proteins. Short structured RNAs without a 5Ј-triphosphate are known to be weak inducers of IFN-␤ (42), but they still activate PKR (24,37), suggesting that PKR activation is not sufficient for IFN-␤ induction. Although the activation of IRF3 is PKR-independent in both RNA transfected and measles virus infected HeLa cells, the activation of NF-B is enhanced by PKR (27,43).
Negative regulators of NF-B signaling, exemplified by IB-␣, are known to be degraded in response to an NF-Bactivating stimuli, and then resynthesized to subsequently inhibit NF-B activity (29,47). Hence, we considered that translation inhibition mediated by PKR might lead to enhanced NF-B signaling and IFN-␤ induction. As shown herein, an inverse correlation was found between IB-␣ mRNA and protein levels following dsRNA transfection and virus infection. Furthermore, consistent with previous results (24), we observed an inverse correlation between eIF-2␣ phosphorylation and IB-␣ protein levels following these treatments. The steady-state protein expression level of IB-␣ was increased and largely rescued when the S49,51A mutant of eIF-2␣ was expressed, which suggests that the resynthesis rather than the stability of IB-␣ is the key parameter affected by PKR activation.
As a further test of the translational basis of the PKR-dependent effect, we used a pharmacologic inhibitor, CHX, to inhibit protein synthesis. Importantly, CHX treatment during RNA transfection rescued PKR-dependent, but not IPS-1-dependent, IFN-␤ transcriptional induction. This finding, together with our observation that the transient knockdown of either PKR or IPS-1 impaired IFN-␤ transcript induction, suggests that PKR and IPS-1 differentially function in the signaling response leading to activation of IFN-␤ transcription.
Our results furthermore demonstrate that the effect of PKR on signaling pathway responses following RNA transfection or virus infection is dependent on PKR enzymatic activity through its eIF-2␣ kinase activity (40,56) and hence cannot represent simply a protein adaptor or scaffold role for PKR (22,57). The observed PKR-dependent decrease in IB-␣ following infection or transfection is consistent with results obtained for the endoplasmic reticulum stress-responsive eIF-2␣ kinase, PERK, where maximal NF-B signaling following phosphorylation of eIF-2␣ by PERK likewise correlated with decreased protein levels of IB-␣ (56).
Our results taken together support the role of translation control of negative regulators, including IB-␣, in the modulation of NF-B signaling. Similar to the effects of PKR on IB-␣ expression established herein, it is conceivable that other negative regulators of the NF-B pathway in addition to IB-␣ also are modulated in a PKR-dependent manner. One intriguing possibility is the A20 negative regulatory protein, which prevents NF-B essential modulator (NEMO) activation through its ubiquitin-editing activity (58). Because A20 is induced by NF-B signaling in a manner similar to IB-␣ resulting in inhi-bition of NF-B signaling upstream of IB-␣ (59), it is reasonable to speculate that A20 would represent an additional potential target for PKR-dependent translational control of NF-B signaling. Since A20 regulates NEMO activity, and NEMO activity is required for the activation of IRF3 through TBK1 and IK (60), then PKR-dependent inhibition of A20 translation presumably could also affect IRF activation in some instances, perhaps explaining the PKR-dependent activation of IRF3 observed following mutant vaccinia virus infection (32). In fact, the absence of negative regulators of NF-B signaling may also explain the PKR-dependence of IB phosphorylation and activation of IK shown by others (20,21). It is also possible that cell type differences in NF-B signaling, especially in cancer cells (61), may account for the seemingly different effects of PKR on this pathway.
Our results are consistent with a translational inhibition response through eIF-2␣ phosphorylation for the mechanistic basis of the PKR-dependent amplification of IFN-␤ expression following transfection or infection (40,56). First, in PKR-sufficient PKR ϩ cells as shown herein, a mutant form of eIF-2␣ that cannot be phosphorylated impaired measles virus-induced activation of IFN-␤ gene expression and enhanced IB-␣ protein production following infection. Secondly, in PKR-deficient cells, catalytically active PKR is necessary in order to complement measles virus-induced PKR-dependent IFN-␤ transcript induction (40). Our findings, taken together, reveal that PKR functions at least in part through an eIF-2-mediated translational control mechanism to indirectly enhance IFN-␤ transcript levels, and that IPS-1 functions through direct activation of signaling pathways including initial NF-B and IRF3 activation as established by others (5).