Activation of Double-stranded RNA-activated Protein Kinase (PKR) by Interferon-stimulated Gene 15 (ISG15) Modification Down-regulates Protein Translation*

Background: Double-stranded RNA-activated protein kinase (PKR) is activated by virus-derived RNA and inhibits protein translation. Results: PKR is covalently modified by interferon-stimulated gene 15 (ISG15), and ISG15-PKR fusion protein is active without virus RNA. Conclusion: PKR is able to be activated by ISG15 modification. Significance: PKR might be an anti-tumor molecule by inhibiting protein translation in ISG15-positive cancer cells. The ubiquitin-like molecule ISG15 (UCRP) and protein modification by ISG15 (ISGylation) are strongly induced by interferon, genotoxic stress, and pathogen infection, suggesting that ISG15 plays an important role in innate immune responses. However, how ISGylation contributes to innate immune responses is not clear. The dsRNA-dependent protein kinase (PKR) inhibits translation by phosphorylating eIF2α to exert its anti-viral effect. ISG15 and PKR are induced by interferon, suggesting that a relationship exists between ISGylation and translational regulation. Here, we report that PKR is ISGylated at lysines 69 and 159. ISG15-modified PKR is active in the absence of virus infection and phosphorylates eIF2α to down-regulate protein translation. The present study describes a novel pathway for the activation of PKR and the regulation of protein translation.


The ubiquitin-like molecule ISG15 (UCRP) and protein modification by ISG15 (ISGylation) are strongly induced by interferon, genotoxic stress, and pathogen infection, suggesting that ISG15 plays an important role in innate immune responses. However, how ISGylation contributes to innate immune responses is not clear. The dsRNA-dependent protein kinase (PKR) inhibits translation by phosphorylating eIF2␣ to exert its
anti-viral effect. ISG15 and PKR are induced by interferon, suggesting that a relationship exists between ISGylation and translational regulation. Here, we report that PKR is ISGylated at lysines 69 and 159. ISG15-modified PKR is active in the absence of virus infection and phosphorylates eIF2␣ to down-regulate protein translation. The present study describes a novel pathway for the activation of PKR and the regulation of protein translation.
The expression of ISG15 and post-translational protein modification by ISG15 (ISGylation) are strongly induced by type I interferon (IFN) (1,3), and IFNs are critical cytokines involved in innate immune responses (17). IFNs, which play a role in the defense against bacterial or viral infections, are induced by several stimuli, such as lipopolysaccharide (LPS), lipoproteins, lipopeptides, double-stranded RNA, doublestranded DNA, single-stranded RNA, or unmethylated CpG motifs mainly through Toll-like receptors (18). Because ISG15 is highly induced by IFNs (19), ISGylation may regulate certain immune responses related to pathogen infections and various stresses. ISG15-deficient mice show no obvious phenotype against vesicular stomatitis virus and lymphocytic choriomeningitis virus (20); however, they are more susceptible to influenza A/WSN/33 and influenza B/Lee/40 virus infections (21). ISG15-deficient mice also exhibit increased susceptibility to herpes simplex virus type 1, murine ␥-herpesvirus 68, and Sindbis virus infection (21). Wild-type ISG15, but not a mutant form of ISG15 that cannot form conjugates with substrates, is able to rescue the increased susceptibility of ISG15-deficient mice to Sindbis virus infection, suggesting that ISGylation is important for the resistance against certain viral infections (21).
Double-stranded RNA (dsRNA)-activated protein kinase (PKR) phosphorylates the ␣-subunit of eukaryotic initiation factor 2 (eIF2␣) to inhibit cap-dependent translation (22)(23)(24)(25) and plays a role in the innate immune response to viral infection and several cellular signal transduction pathways (26). PKR is induced by interferon and activated by binding to dsRNA, which causes the homodimerization and autophosphorylation of the kinase (27)(28)(29). PKR is an anti-viral molecule that recognizes viral RNA and inhibits viral protein translation. Some RNA viruses, including influenza virus, hepatitis C virus, hepa-titis D virus, West Nile Virus, human immunodeficiency virus type 1 (HIV-1), Sindbis virus, encephalomyocarditis virus, footand-mouth disease virus, and even DNA viruses such as herpes simplex virus type 1, are regulated by PKR (30). Interestingly, PKR is required for the production of IFN-␣/␤ mRNA in response to a subset of RNA viruses including encephalomyocarditis virus, Theiler murine encephalomyelitis, and Semliki Forest virus (31). Although the importance of PKR in the protection against viral infection has been well established, how PKR recognizes different kinds of viruses and dsRNAs is not well understood.
The present study demonstrates that PKR is modified by ISG15 at lysine residues 69 and 159 located in the dsRNA-binding motifs. ISG15 modification activates PKR in the absence of virus infection and prevents protein translation. ISGylation and autophosphorylation are required for the activation of PKR in the absence of virus infection. These data suggest a novel pathway mediating the activation of PKR and the prevention of protein translation with no relation to virus infection.
Immunoprecipitation (IP), Ni-NTA Pulldown, and Western Blot Analysis-Ni-NTA pulldown and Western blot analysis were done as reported previously (32). To detect endogenous ISGylation of PKR, 0.5 g of anti-ISG15 antibodies (36) was incubated with 10 l of protein A-Sepharose (GE Healthcare) in lysis buffer containing 50 mM Tris-HCl (pH 7.6), 300 mM NaCl, 1% Triton X-100, 10 g/ml aprotinin, 10 g/ml leupeptin, 1 mM PMSF, 0.4 mM Na 3 VO 4 , 0.4 mM EDTA, 10 mM NaF, and 10 mM sodium pyrophosphate overnight, and then washed three times with same lysis buffer followed by incubation with RAW264.7 cell lysates overnight. The resulting immunoprecipitates were washed three times with the same lysis buffer and subjected to Western blotting with anti-ISG15 antibodies (37) or human PKR antibodies (Santa Cruz Biotechnology, sc-100378). Bands were visualized using Supersignal West Pico or Dura chemiluminescent substrate (Pierce).
Statistical Analysis-Student's t test was used to determine the statistical significance of the experimental data.

RESULTS
ISG15 Modification of PKR-More than 100 proteins have been identified as potential substrates for ISG15 modification or as ISG15-interacting proteins, and PKR is one of the candidates (39,40). Furthermore, ISG15 and PKR expressions are induced by interferon stimulation. We therefore investigated whether PKR is modified by ISG15. UBE1L, mUbcH8, His 6tagged ISG15, and HA-tagged PKR were expressed in HEK293T cells together with the nonspecific ISG15 ligase Herc5 to improve ISGylation (11)(12)(13). ISGylated proteins were purified by Ni-NTA-agarose chromatography and subjected to Western blotting against an anti-HA antibody to detect the ISGylation of HA-PKR (Fig. 1A). A band corresponding to HA-PKR and two additional bands around 70 kDa and 100 -115 kDa, respectively, were detected when all components were expressed, indicating that several ISG15 molecules were covalently attached to PKR (Fig. 1A). Reciprocal IP was performed by anti-HA antibody to confirm the ISGylation of HA-PKR (Fig.  1B). As a result, ISGylation of HA-PKR was detected when all components were expressed. Because E3 enzyme recognizes substrate, we checked the interaction between PKR and Herc5 and detected no interaction (Fig. 1B), suggesting that Herc5 may not be an E3 enzyme for PKR or the interaction is unstable to be detected. ISGylation of PKR was further confirmed under physiological conditions. The murine macrophage-like cell line RAW264.7 was stimulated by LPS to induce ISGylation (33). ISGylated proteins were purified using an anti-ISG15 antibody, and ISGylated PKR was detected by Western blotting against an anti-PKR antibody (Fig. 1C). Two ISGylated PKR species with different migration patterns were detected similar to those observed in the overexpressed condition. Nonmodified PKR was also nonspecifically pulled down during IP as shown in control IPs. Reciprocal IP was performed using wild-type mouse lung fibroblasts or ISG15-deficient fibroblasts stimulated by interferon ␤ for 1 day (Fig. 1D). As expected, ISGylated PKR was detected in wild-type fibroblasts stimulated by interferon but not in ISG15-deficient cells.
Activation of PKR by ISG15 Modification-Because PKR is a protein kinase that phosphorylates eIF2␣ and inhibits cap-dependent translation (22)(23)(24)(25), we hypothesized that the activity of PKR may be regulated by ISG15 modification. To analyze the activities of PKR and ISGylated PKR, we tried to purify His 6 -ISG15-modified FLAG-PKR by two-step purification using Ni-NTA-agarose (for His 6 -ISG15) and FLAG-agarose (for FLAG-PKR) as reported previously (10). However, we failed to purify enough amounts of ISGylated PKR to perform in vitro kinase assay. Because autophosphorylation of PKR at threonine 451 is required for the activation of PKR (28), we examined the activity of PKR by comparing autophosphorylation of nonmodified and ISGylated PKR ( Fig. 2A). Although nonmodified PKR was phosphorylated in 293T cells without virus infection, ISGylated PKR was strongly phosphorylated compared with that of nonmodified PKR, indicating that ISGylation activated PKR. To investigate the functional role of ISG15 modification, ISG15 was fused to either the N or C terminus of PKR (Fig. 2B). Such fusion proteins have been found to mimic the constitutively modified state of a protein and are particularly useful when the native modification site is near the terminus of a protein of interest (10,41). The C-terminal amino acid sequence of active ISG15 (LRLRGG) covalently binds to lysine residues on the substrate and is reversibly removed from the substrate by USP18 (16). The C-terminal glycine residue was deleted to prevent the cleavage of N-terminally fused ISG15 from PKR (Fig.  2B). Expression of FLAG-ISG15, FLAG-PKR, FLAG-ISG15-PKR, or FLAG-PKR-ISG15 in HEK293T cells resulted in the inhibition of cell growth in FLAG-ISG15-PKR-transfected cells compared with other transfectants 40 h after transfection (Fig.  2C). To confirm the expression of these proteins, cell lysates FIGURE 1. ISG15 modification of PKR. A, ISG15 modification of PKR by expression of the ISG15 conjugation system components. UBE1L, mUbcH8, His 6 -ISG15, FLAG-Herc5, and HA-PKR were expressed in HEK293T cells as indicated, followed by Ni-NTA pulldown and Western blotting with anti-HA, FLAG, or ISG15 antibodies. Note that two bands around 100 kDa were detected. B, His 6 -ISGylation system and HA-PKR were expressed as A, followed by IP by anti-HA antibody and Western blotting with anti-HA, FLAG, or ISG15 antibodies. C, endogenous ISG15 modification of PKR. Murine macrophage-like RAW264.7 cells were stimulated by LPS, followed by IP with anti-ISG15 or control antibodies. The resulting immunoprecipitates were subjected to Western blotting with anti-PKR or ISG15 antibodies. A short exposure is shown to provide a clearer image of the induction of PKR. D, ISG15-deficient fibroblasts were stimulated by interferon ␤, followed by IP with anti-PKR or control antibodies. The resulting immunoprecipitates were subjected to Western blotting with anti-PKR or ISG15 antibodies. Asterisk, nonspecific bands.
were prepared and analyzed by Western blotting (Fig. 2D). The inhibition of cell growth by FLAG-ISG15-PKR suggested that ISG15 modification may occur near the N terminus of PKR, resulting in its activation and the prevention of protein translation. The phosphorylation of eIF2␣ was up-regulated by the expression of FLAG-ISG15-PKR but not by other proteins (Fig.  2E). Because ISG15-PKR fusion protein is an artificial protein and may induce unfolded protein responses, which also results in phosphorylation of eIF2␣, we checked unfolded protein responses by monitoring immunoglobulin heavy chain binding protein (Bip) and C/EBP homologous protein (CHOP) by quantitative PCR, and no unfolded protein responses were detected (Fig. 2F). FLAG-PKR, FLAG-ISG15, and FLAG-ISG15-PKR were expressed in PKR-deficient immortalized fibroblasts (34), and eIF2␣ phosphorylation was analyzed (Fig. 2G). Reintroduction of wild-type PKR slightly up-regulated the phosphoryla- tion of eIF2␣, and this effect was enhanced by the expression of ISG15-PKR (Fig. 2, G and H). SUMO1, a ubiquitin-like protein similar to ISG15, was N-terminally fused to PKR to investigate the specificity of ISG15 modification for the activation of PKR. The phosphorylation of eIF2␣ was not up-regulated by SUMO1-PKR, indicating that ISG15 modification specifically activates PKR (Fig. 2I). The activities of PKR and ISG15-PKR were examined by comparing autophosphorylation, and only that of ISG15-PKR was detected, indicating that ISG15-PKR fusion protein is active (Fig. 2J). ISG15 Modification at Lysines 69 and 159 of PKR-Because poly-ISG15 modification has not been identified yet and ISG15 modification on different lysine residues could affect the migration pattern on SDS-PAGE (10), we hypothesized that at least two lysine residues on PKR may be the sites of modification by ISG15, which could result in different migration rates as shown in Fig. 1. Furthermore, N-terminally but not C-terminally fused ISG15 activated PKR (Fig.  2), suggesting that the native ISG15 modification sites are located near the N terminus. The N-terminal regulatory domain of PKR contains two dsRNA-binding motifs, dsRBM1 and dsRBM2 (29). The similarity between the amino acid sequences of dsRMB1 and dsRMB2 suggests that if dsRMB1 is ISGylated, then dsRMB2 may also be ISGylated. Several lysine residues are conserved between dsRBM1 and dsRBM2 as shown in Fig. 3A. Because neighbor lysine residues might be modified by ISG15 when the native ISGylation site is mutated, we mutated Lys-112 or Lys-61 in addition to Lys-111 or Lys-60, respectively, to prevent unexpected ISGylation (Fig. 3). Either wild-type or mutant PKR was expressed in HEK293T cells with the His 6 -ISG15 conjugation system (including UBE1L, mUbcH8, FLAG-Herc5, and His 6 -ISG15), and ISGy- PKR-deficient fibroblasts were transfected by plasmids encoding FLAG-PKR, FLAG-ISG15, or FLAG-ISG15-PKR as well as control plasmid. After 1 day of transfection, the cell lysates were subjected to Western blotting with anti-p-eIF2␣, total eIF2␣, ISG15, or PKR antibodies. H, quantification of phosphorylated eIF2␣ shown in F. The amount of phosphorylated eIF2␣, which was determined by scanning densitometry of the immunoblot, was divided by the signal intensity of the total eIF2␣ and normalized to that in the presence of FLAG-PKR. Data are expressed as means Ϯ S.D. from four independent experiments. I, SUMO1 modification did not activate PKR. HEK293T cells were transfected by plasmids encoding FLAG-PKR, FLAG-ISG15-PKR, or FLAG-SUMO1-PKR. After 1 day of transfection, the cell lysates were subjected to Western blotting with anti-PKR, p-eIF2␣, or total eIF2␣ antibodies. J, activation of PKR by ISG15 modification. FLAG-PKR or FLAG-ISG15-PKR was expressed in PKR-deficient fibroblasts, followed by Western blotting with anti-phosphorylated PKR (p-PKR) or total PKR antibodies. Conserved lysine residues and amino acids are boxed and indicated by asterisks, respectively. B, ISG15 modification of wild-type or mutant PKR. The His 6 -ISGylation system (including UBE1L, mUbcH8, His 6 -ISG15, FLAG-Herc5) was expressed in HEK293T cells with wild-type or mutant PKR as indicated, followed by Ni-NTA pulldown and Western blotting with anti-HA, or ISG15 antibodies. Note that the two bands around 110 kDa were not detected when lysine 69 and 159 were mutated to arginine. lated proteins were purified by Ni-NTA-agarose chromatography, followed by Western blotting with an anti-HA antibody to detect ISGylated HA-PKR (Fig. 3B). Although ISGylation of mutant PKR(K20R/K111R/K112R) was weaker than that of wild-type PKR, we concluded that lysines 69 and 159 are the natural ISGylation sites because mutant PKR(K69R/K159R) showed complete loss of ISGylation (Fig. 3B). Expression of mutant PKR(K60R/K61R/K150R) enhanced the total ISGylation, although the underlying mechanism was not clear and should be investigated in the future.
Down-regulation of Protein Translation by ISG15-modified PKR-pRL-TK encoding Renilla luciferase driven by herpes simplex virus thymidine kinase (HSV-TK) promoter was used to monitor protein translation. pRL-TK was co-transfected with plasmids encoding FLAG-PKR(wild-type or T451A), ISG15, ISG15-PKR(wild-type or T451A), or SUMO-PKR to PKR-deficient fibroblasts. mRNA and lysates were prepared to perform quantitative PCR for Renilla luciferase (Fig. 4A) and luciferase assay (Fig. 4B), respectively. The amounts of luciferase were normalized by that of mRNA, and protein translation was examined (Fig. 4C). Although PKR was inactive without virus-derived dsRNA, the reintroduction of wild-type PKR to PKR-deficient cells slightly down-regulated protein translation (Fig. 4C). Co-expression of ISG15 and PKR showed an effect similar to that of PKR single expression. Importantly, the ISG15-PKR fusion protein had a greater effect on the downregulation of protein translation than wild-type PKR (Fig. 4C). SUMO1-PKR fusion protein down-regulated protein translation but not as obviously as ISG15-PKR fusion protein did (Fig.  4C). Because autophosphorylation of PKR at threonine 451 is required for the activation of PKR (28), we investigated whether ISG15 modification is able to activate PKR without autophosphorylation. PKR(T451A) did not down-regulate protein translation, and ISG15 modification did not activate PKR(T451A) as wild-type PKR did, indicating that phosphorylation at threonine 451 of PKR was necessary for the ISG15 modification-induced activation of PKR (Fig. 4C). The physiological relationship between ISGylation and PKR was studied using ISG15-deficient fibroblasts (Fig. 4D) or PKR-deficient fibroblasts, which were reconstituted by wild-type PKR or PKR(K69R/K159R) (Fig. 4E). Wild-type or ISG15-deficient fibroblasts were stimulated by interferon ␤, and the phosphorylation of eIF2␣ and subsequent expression of activating transcription factor 4 (ATF4) were assessed (Fig. 4D). Unexpectedly, eIF2␣ was constantly phosphorylated in wild-type mouse lung fibroblasts by an unidentified mechanism. However, most importantly, the phosphorylation of eIF2␣ was under the detectable level by Western blotting in ISG15-deficient lung fibroblasts even after induction of PKR by interferon stimulation. In contrast, ATF4 was up-regulated by interferon stimulation in wild-type fibroblasts but not in ISG15-deficient fibroblasts (Fig. 4D). These data may suggest that an unidentified mechanism in addition to the phosphorylation of eIF2␣ regulates the expression of ATF4 in these cell lines. Next, we reintroduced wild-type PKR or PKR(K69R/K159R) to PKR-deficient fibroblasts (Fig. 4E). Reintroduction of PKR resulted in the phosphorylation of eIF2␣, and that was up-regulated by interferon stimulation. Reintroduction of PKR(K69R/K159R) also phosphorylated eIF2␣, but the effect was weaker than wild-type PKR. It may be explained by unstable expression of mutant PKR than wild-type PKR. Importantly, phosphorylation of eIF2␣ was not up-regulated by interferon stimulation in cells expressing PKR(K69R/K159R) (Fig. 4E). These data indicate that PKR is activated by ISG15 modification.

DISCUSSION
PKR is induced in an inactive form by interferon and activated by binding to viral dsRNA, which causes dimerization and autophosphorylation (27)(28)(29). However, in the present study, exogenously expressed wild-type PKR could prevent protein translation (Fig. 4C), indicating that high concentrations of PKR may cause homodimerization and the consequent activation of the kinase in the absence of dsRNA. Because PKR is involved in microRNA precursor-dependent apoptosis (42), another possibility is that endogenous microRNA-derived dsRNAs activate PKR to some extent.
ISG15 is a ubiquitin-like modifier (1, 2) that is covalently conjugated to cellular proteins in a process similar to that of ubiquitin conjugation (3). However, the physiological functions of ISGylation have not been elucidated to date. Our observations that ISGylation can activate PKR in the absence of viral infection suggest a novel mechanism for PKR activation. Interferon stimulation, virus infection, or even tumorigenesis induces ISGylation (43)(44)(45). The ISGylation of the dsRNAbinding motifs of PKR would open up the kinase domain and induces homodimerization through ISG15 because ISG15 can homodimerize (35). The subsequent activation and autophosphorylation of PKR at threonine 451 prevent translation through the phosphorylation of eIF2␣. The discovery that ISGylation activates PKR without virus infection demonstrates gain-of-function modulation, which is highly significant considering that ISG15 generally modifies a small fraction of any target protein upon IFN stimulation or pathogen infections. So far, ISGylation is extremely weak in any individual substrate. This is also true for modification by other ubiquitin like proteins, such as SUMO. It has been reported that these minor ISGylated fractions have important physiological roles, such as tumor-suppressive effect through down-regulation of ⌬Np63␣ (46), inhibition of influenza A virus replication (47), sustained IRF3 activation (48), and prevention of JNK cascade (49). These reports support the significant physiological role of ISGylated PKR on the protein translation although ISGylated PKR is minor fraction. Unfortunately endogenous ISGylation of PKR by IFN stimulation was not detected in a plain Western blotting using crude cell lysates (data not shown). In contrast, endogenous ISGylation of NS1A was clearly shown (47), may suggest that ISGylation of PKR is not the major mechanism to activate PKR, or ISGylated PKR is unstable in the cells examined. . Down-regulation of protein translation by ISG15-modified PKR. A, PKR-deficient fibroblasts were transfected with plasmids encoding FLAG-PKR, FLAG-ISG15-PKR, FLAG-ISG15, or FLAG-SUMO1-PKR as indicated, as well as the control plasmid and reporter plasmid. Forty hours after transfection, total RNA was purified, and quantitative PCR was performed to compare the amounts of Renilla luciferase mRNA. Representative data are shown. B, Renilla luciferase was quantified according to the manufacturer's protocol. Representative absolute units are shown. C, down-regulation of protein translation by PKR is shown. Protein translation was calculated by dividing the amount of Renilla luciferase by that of mRNA. Data are means Ϯ S.D. (error bars) of values from three independent experiments. D, no phosphorylation of eIF2␣ in ISG15-deficient fibroblasts was detected. ISG15-deficient fibroblasts were stimulated by interferon ␤ followed by Western blotting with anti-phosphorylated eIF2␣ (p-eIF2␣), eIF2␣, ATF4, ISG15, or PKR antibodies. E, eIF2␣ was phosphorylated by wild-type PKR but not PKR(K69R/K159R) and interferon ␤ stimulation. PKR-deficient fibroblasts reconstituted by wild-type PKR or PKR(K69R/K159R) were stimulated by interferon ␤ followed by Western blotting with anti-phosphorylated eIF2␣ (p-eIF2␣), eIF2␣, ISG15, or PKR antibodies. The relative intensities of p-eIF2␣ normalized by that of total eIF2␣ are shown below the blot.