The interferon-inducible protein kinase PKR modulates the transcriptional activation of immunoglobulin kappa gene.

PKR is an interferon (IFN)-induced serine/threonine protein kinase that regulates protein synthesis through phosphorylation of eukaryotic translation initiation factor-2 (eIF-2). In addition to its demonstrated role in translational control, recent findings suggest that PKR plays an important role in regulation of gene transcription, as PKR phosphorylates IκBα upon double-stranded RNA treatment resulting in activation of NF-κB DNA binding in vitro (Kumar, A., Haque, J., Lacoste, J., Hiscott, J., and Williams, B. R. G.(1994) Proc. Natl. Acad. Sci. U. S. A. 91, 6288-6292). To further investigate the role of PKR in transcriptional signaling, we expressed the wild type human PKR and a catalytically inactive dominant negative PKR mutant in the murine pre-B lymphoma 70Z/3 cells. Here, we report that expression of wild type PKR had no effect on κ-chain transcriptional activation induced by lipopolysaccharide or IFN-γ. However, expression of the dominant negative PKR mutant inhibited κ gene transcription independently of NF-κB activation. Phosphorylation of eIF-2α was not increased by lipopolysaccharide or IFN-γ, suggesting that PKR mediates κ gene transcriptional activation without affecting protein synthesis. Our findings further support a transcriptional role for PKR and demonstrate that there are at least two distinct PKR-mediated signal transduction pathways to the transcriptional machinery depending on cell type and stimuli, NF-κB-dependent and NF-κB-independent.

IFNs 1 induce a large number of genes whose products either singly or coordinately mediate antiviral, growth-inhibitory, or immunoregulatory activities (1,2). IFN-mediated gene induc-tion is accomplished by a cascade of events in which many positive and negative regulatory factors are involved. IFNinducible proteins initiate a cascade of activation of a second set of genes, whose expression requires continued protein synthesis (1,2).
One of the best characterized IFN-stimulated proteins is the double-stranded RNA-dependent protein kinase, PKR (also known as dsRNA-PK, dsI, and DAI) (3). PKR is a 68-kDa polypeptide in humans and 65-kDa in mice. There is also a yeast homologue, termed GCN2, that is involved in regulation of amino acid biosynthesis under starvation conditions (4). PKR is a serine/threonine-specific protein kinase (3) that displays two distinct kinase activities (i) activation by autophosphorylation upon treatment with dsRNA and (ii) phosphorylation of the ␣ subunit of the eukaryotic translation factor eIF-2 (5), a modification that causes inhibition of protein synthesis (6).
Cloning of the human and mouse PKR cDNAs (7-10) enabled a detailed analysis of the structure-function relationship of the proteins (8 -13). The dsRNA binding domain has been localized to the N-terminal half of the kinase (9,(11)(12)(13). The C-terminal half of the molecule contains all 11 conserved domains that are present in protein kinases (14). A single amino acid substitution in the invariant lysine 296 in catalytic domain II of human PKR (this invariant lysine is directly involved in ATP binding and the phosphotransfer reaction) (14) causes the inactivation of the human PKR, but the protein retains the ability to bind dsRNA (11).
Studies on the role of PKR in regulation of cell growth suggest that it may function as a tumor suppressor. Expression of wt PKR in yeast inhibits cell growth, which correlates with increased phosphorylation of eIF-2␣ (15). Expression of catalytically inactive mutants of human PKR in NIH 3T3 cells results in malignant transformation (16,17). The mutants studied consisted of either a deletion of 6 amino acids (Leu-Phe-Ile-Gln-Met-Glu; amino acids 361-366) in subdomain V (PKR⌬6) (16) or substitution of the invariant lysine 296 to arginine (PKR K296R) (11,17). These findings suggest that wt PKR is a tumor suppressor gene product whose activity can be inhibited by the presence of catalytically inactive PKR mutants. In this regard, a form of murine lymphoblastic leukemia is associated with an in-frame deletion in the PKR gene, which results in expression of an inactive protein. 2 The human PKR gene maps to chromosome region 2p21-22 (18 -20), and abnormalities involving this region are observed among patients with acute myelogenous leukemia (20), raising the possibility of a role for PKR in leukemogenesis.
The mechanism(s) of growth suppression by wt PKR remains to be established. In addition to its role in translational control, several reports have suggested a role for PKR in regulation of gene transcription (21)(22)(23)(24). For example, the PKR inhibitor 2-aminopurine inhibits gene transcription that is induced by virus infection or dsRNA treatment (25)(26)(27). Moreover, PKR activation by dsRNA results in phosphorylation of IB␣ leading to activation of NF-B (28). Furthermore, cells depleted of PKR activity were unresponsive to activation of NF-B by dsRNA (29). Other mechanisms which are NF-B independent cannot be excluded, however (27).
To investigate the role of PKR in signaling to the transcriptional machinery, we expressed wt human PKR, or the dominant negative catalytically inactive mutant PKR⌬6 (16), in 70Z/3 cells. 70Z/3 is a mouse pre-B lymphoma cell line which has been used successfully as a model system to study transcriptional regulation of the immunoglobulin gene. Transcription of the gene, which is thought to be the rate-limiting event for differentiation of pre-B to B cells (30), is induced by a variety of mitogens and lymphokines (31,32), leading to the expression of surface immunoglobulin M (sIgM). Here, we demonstrate that transcriptional activation of gene is mediated by PKR. Expression of the dominant negative PKR⌬6 resulted in inhibition of -chain transcription induced by either LPS or IFN-␥. In addition to cell growth regulation by PKR (15)(16)(17), these findings also provide evidence for a role of PKR in lymphoid cell differentiation.
Transfection and Selection of Stable Transfectants-Plasmids containing wt PKR, and PKR⌬6 cDNAs under the control of human cytomegalovirus promoter in the pcDNAI/neo vector (16) were used for expression in 70Z/3 cells. Plasmid DNA (10 g) was linearized with KpnI and electroporated into 1 ϫ 10 7 cells at 300 V-960 F (Bio-Rad) as described previously (33). After electroporation, 70Z/3 cells were cultured in nonselective medium and grown for 24 h to allow for expression of the transfected genes. Cells were recultured in 24-well plates at a concentration of 1 ϫ 10 5 /ml (1 ml/well) in medium containing G418 (Life Technologies, Inc.) at a final concentration 400 g/ml. Medium was replenished every 3 days. Cells (polyclonal populations) were expanded and characterized 15 days postselection. Independent clones were selected by a limiting dilution method as described previously (34).
Immunoprecipitation and Immunoblotting-Cells (1 ϫ 10 7 ) were washed three times with cold phosphate-buffered saline (PBS, 140 mM NaCl, 15 mM KH 2 PO4 (pH 7.2), and 2.7 mM KCl) and incubated on ice with an equal volume of 2 ϫ lysis RIPA (100 mM Tris⅐Cl (pH 7.5), 300 mM NaCl, 2% Nonident P-40, 1% sodium deoxycholate, and 0.2% SDS) supplemented with 2 mM dithiothreitol (DTT), 0.4 mM phenylmethylsulfonyl fluoride (PMSF), and 4 g/ml aprotinin. The lysate was centrifuged at 10,000 ϫ g for 10 min, and the supernatant was incubated with 2.5 l of anti-PKR polyclonal antibody for 2 h at 4°C. Then, 50 l of 50% suspension of protein A-Sepharose 4L (Pharmacia Biotech Inc.) in 1 ϫ RIPA were added, and incubation was continued for additional 4 h at 4°C under rotation. The Sepharose beads were washed with 1 ϫ RIPA plus 1 M NaCl twice and 1 ϫ RIPA twice. Immunoprecipitates were subjected to electrophoresis on SDS-8% polyacrylamide gel. The separated proteins were transferred to a nitrocellulose membrane (Schleicher & Schuell) in 25 mM Tris⅐Cl (pH 7.5), 190 mM glycine, and 20% (v/v) methanol for 2 h at 1 Å. The filter was first incubated with 5% (w/v) non-fat dried skimmed milk powder in PBS for 1 h at room temperature and then with 25% fetal bovine serum and 0.5% (v/v) Triton X-100 in PBS containing a mouse monoclonal antibody to human PKR (13B8-F9). The blot was incubated with peroxidase-conjugated rabbit antibody to mouse immunoglobulin G, and proteins were visual-ized using the enhanced chemiluminescence system (Amersham Corp.) according to the manufacturer's specifications.
Cell Induction and Immunofluorescent Analysis-Cells were incubated at concentration 5 ϫ 10 5 cells/ml with appropriate concentrations of inducing agents: 10 g/ml of Salmonella typhosa LPS (Sigma) or 100 U/ml murine recombinant IFN-␥ (Cedarlane, Canada). For prolonged inductions, cells were diluted daily to 5 ϫ 10 5 cells/ml with fresh medium supplemented with the inducing agent.
RNA Extraction and Northern Blotting-Total RNA was isolated by the guanidinium thiocyanate method (36). RNA (10 g) was denatured with glyoxal and dimethyl sulfoxide and subjected to electrophoresis on a 1% agarose gel in 10 mM sodium phosphate buffer (pH 7.0). For RNA stability experiments, total RNA was isolated from cells treated with actinomycin D (10 g/ml) for 30, 60, and 120 min. RNA was transferred onto a nylon membrane (BioTrans, ICN). Hybridization was performed at 65°C for 16 h with [␣-32 P]dATP-labeled random-primed cDNA probes (5 ϫ 10 6 cpm/ml) (37), consisting of either the 750-base pair SmaI-PstI fragment of the -chain cDNA together with a 3.0-kilobase HindIII fragment of the -chain cDNA (38) or the entire coding sequence of mouse ␤-actin. After hybridization, the filter was washed with 0.1 ϫ SSC (150 mM NaCl and 15 mM sodium citrate (pH 7.0)) plus 1% SDS for 1 h at 45°C. The filter was dried and exposed to an x-ray film for 10 h.
PKR Autophosphorylation and eIF-2␣ Phosphorylation Analysis-For in vitro autophosphorylation of PKR, 10 g of extracts from untreated HeLa S3 cells or HeLa S3 cells treated with human IFN-␤ for 18 h (1000 IU/ml; Lee Biomolecules) were suspended in kinase reaction buffer (10 mM Tris⅐Cl, pH 7.7, 50 mM KCl, 2 mM MgCl 2 , 5 mM ␤-mercaptoethanol, 2 g/ml aprotinin, 1 g/ml leupeptin, 1 g/ml pepstatin, and 0.1 mM PMSF) and 10 Ci of [␥-32 P]ATP. Reovirus dsRNA was added to the final concentration of 0.1 g/ml. After incubation at 30°C for 30 min, the reaction was diluted 5-fold with RIPA and split equally into two fractions. In one of the fractions, 2.5 l of anti-PKR monoclonal antibody (13B8-F9) were added, and immunoprecipitation of the autophosphorylated PKR was performed as described above. In the other fraction, 5 l of sheep anti-eIF-2␣ polyclonal antibody were added, and eIF-2␣ immunoprecipitation was performed as for PKR using protein G-Sepharose (Pharmacia) as a carrier. PKR immunoprecipitates were subjected on SDS-8% polyacrylamide gels, whereas eIF-2␣ immunoprecipitates were on SDS-10% polyacrylamide gels.
Electrophoretic Mobility Shift Assays-Nuclear protein extracts were prepared as described elsewhere (42). Five g of protein extracts were tested for NF-B activity by binding to 80 pg of a 32 P-5Ј-end-labeled dsDNA oligonucleotide (1 ϫ 10 6 cpm/ng; 5Ј-GATCCAAGGGGACTTTC-CATGGATCCAAGGGGACTTTCCATG-3Ј; Life Technologies, Inc.; the underlined sequences correspond to the NF-B-binding sites) as described previously (43). Antibody mobility supershift assays were performed by incubating 10 g of protein extracts together with 200 pg of the 32 P-5Ј-end-labeled dsDNA HIV-B oligonucleotide (44) (5 ϫ 10 6 cpm/ng; 5Ј-AGCTGGGACTTTCCGCTA-3Ј; the underlined sequence corresponds to the NF-B-binding site) and 1 l of the stock of affinitypurified rabbit polyclonal antibodies against rel (45), p65 (45), or p50 (46) proteins. For cold competition an 125-fold excess of unlabeled dsDNA oligonucleotides was added. The specificity of the supershifted bands was tested by antibody binding competition with 1 g of the epitope peptide (45,46) used for antisera preparation.

Expression of Wild Type PKR and Catalytically
Inactive PKR⌬6 in 70Z/3 Cells-70Z/3 cells were transfected with wt human PKR or PKR⌬6 (originally termed p68⌬6) (16) cDNA and selected in G418. Polyclonal populations of G418-resistant cells were expanded and characterized for protein expression by immunoblotting using a monoclonal antibody (13B8-F9) specific for the human PKR. 3 As expected, the anti-human PKR antibody failed to detect the endogenous mouse PKR in control 70Z/3 cells transfected with the neomycin-resistant gene only (Fig. 1, A and B, lane 2). Two bands were detected in wt PKR-transfected cells (1, A and B, lane 3) which correspond to the phosphorylated (upper band) and nonphosphorylated (lower band) forms of the kinase (5,47). Phosphatase treatment of wt PKR yielded the slower migrating nonphosphorylated form of the molecule (Fig. 1B, compare lanes 3 and 4). In contrast, expression of the mutant PKR⌬6 yielded one polypeptide species which is the nonphosphorylated form (Fig. 1A, lane 4). This is consistent with the dominant negative character of PKR⌬6 (16), as PKR⌬6 is neither autophosphorylated nor is it phosphorylated by endogenous mouse PKR. The native PKR is nonphosphorylated (Fig. 1, A and B, lane 1). It is noteworthy that wt PKR can be overexpressed in several transformed cell lines, 4 including 70Z/3 cells, without apparent inhibition of cell growth, in contrast to NIH 3T3 cells (16). This may be explained by modulation of PKR activity by a specific inhibitor(s) as reported for v-ras-transformed cells (48).
Inhibition of sIgM Expression in 70Z/3 Cells Expressing PKR⌬6 Mutant-The sIgM expression in 70Z/3 cells expressing wt PKR or PKR⌬6 was examined by cell surface staining with FITC anti-mouse antibody followed by cell analysis on a cell sorter (35). Upon LPS or IFN-␥ treatment, sIgM expression was diminished in 70Z/3 cells expressing PKR⌬6 (2-3-fold; Fig.  2, E and F), but not in cells expressing the neomycin-resistant gene only (herein referred as control 70Z/3 cells; Fig. 2, A and  B). Expression of sIgM in response to LPS or IFN-␥ was not affected in 70Z/3 cells overexpressing wt PKR relative to control 70Z/3 cells (Fig. 2, C and D). This is presumably because the endogenous mouse PKR elicits the maximal IgM expression upon LPS or IFN-␥ treatment. sIgM expression by the dominant-negative PKR⌬6, we isolated single clones from the polyclonal population of control 70Z/3 cells (herein referred as control clones) and 70Z/3 cells expressing PKR⌬6 (herein referred as PKR⌬6 clones) by the limiting dilution method (34). Several clones were examined for sIgM expression upon LPS or IFN-␥ treatment. All 12 of the control clones showed the same pattern of high levels of sIgM expression upon LPS or IFN-␥ treatment (data not shown). However, from the PKR⌬6 clones tested, 12 out of 16 showed a significant decrease (between 40 and 70% compared to control cells) in sIgM expression, whereas the rest of the PKR⌬6 clones (4 out of 16) showed a smaller but measurable effect (20 -40% decrease in sIgM expression) upon LPS or IFN-␥ treatment (data not shown).
PKR⌬6 Mediates Inhibition of -Chain Expression at Transcriptional Level-The effect of PKR⌬6 on -chain expression could be explained by either a transcriptional or a post-transcriptional mechanism(s). To distinguish between these possibilities, a nuclear run-on analysis ofand -chain transcription was performed with two of the PKR⌬6-expressing clones (PKR⌬6-1 and PKR⌬6-20). Run-on experiments with PKR⌬6-1 clone (Fig. 4A) showed that inhibition of -chain transcription relative to was 60% (compare lanes 2 and 5) and 65% (compare lanes 3 and 5) upon LPS and IFN-␥ treatment, respectively (the experiment was repeated twice and the results varied by no more than 10%). Inhibition of -chain transcription relative to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was 60% (lanes 2 and 5) and 70% (lanes 3 and 6) upon LPS and IFN-␥ treatment, respectively. In the case of PKR⌬6-20, 55 and 60% inhibition of transcription of -chain relative to the -chain was seen upon LPS and IFN-␥ treatment, respec- tively (Fig. 4B, compare lane 2 to 5 and lane 3 to 6; the experiment was repeated three times, and the results varied by less than 10%). These findings indicate that PKR⌬6 expression leads to inhibition of -chain expression at the level of transcription initiation.
Effects on stability of mRNA were tested by the following experiment. Following actinomycin D treatment, total RNA from control or PKR⌬6 cells was isolated, and the levels of -chain and -chain mRNA were compared by Northern blotting. Although -chain mRNA expression was decreased in PKR⌬6 cells upon LPS treatment (Fig. 4C, compare lanes 1 and  9) or IFN-␥ treatment (compare lanes 5 and 13), the ratio of  , lanes 1, 4,  and 7; B, lanes 1 and 4) and after treatment with LPS (10 g/ml; A, lanes 2, 5, and 8; B, lanes 2 and 5) or IFN-␥ (100 IU/ml; A, lanes 3, 6, and 9; B, lanes 3 and 6) for 24 h. RNA extraction and Northern analysis were performed as described under "Experimental Procedures." Quantitation of labeled bands was performed by scanning autoradiograms in the linear range of exposure with a Bio-Image system (Millipore).
-chain to -chain mRNA did not change either in control cells or in PKR⌬6 cells after actinomycin D treatment. This is consistent with previous studies showing that -chain mRNA is very stable (49) and indicates that PKR⌬6 does not affect -chain mRNA stability.
Phosphorylation of eIF-2␣ Is Not Required for the Transcriptional Activation of Gene-The regulation of -chain transcription by PKR is the second example of regulation of transcription by an eIF-2␣ kinase. The yeast eIF-2␣ kinase, GCN2, regulates expression of the transcriptional activator GCN4 at the translational level upon amino acid starvation conditions (4). It is possible that PKR mediates -chain transcription indirectly by regulating the protein synthesis of a transcriptional factor(s) through eIF-2␣ phosphorylation as does GCN2. To examine this possibility, we measured the extent of eIF-2␣ phosphorylation in vivo upon LPS or IFN-␥ treatment. We used two different assays for eIF-2␣ phosphorylation: (i) immunoprecipitation of eIF-2␣ from [ 32 P]orthophosphate-labeled cells (Fig. 5B) and (ii) isoelectric focusing followed by eIF-2␣ immunoblotting (Fig. 5C). In the first assay we used a sheep anti-eIF-2␣ polyclonal antibody, whose suitability was tested first (Fig. 5A). HeLa S3 cell extracts were incubated with reovirus dsRNA and [ 32 P-␥]ATP followed by immunoprecipitation with either anti-PKR antibody (Fig. 5A, lanes 1-4) or anti-eIF-2␣ antibody (Fig. 5A, lanes 5-8). Induction of PKR autophosphorylation by dsRNA before (lane 2) or after treatment with IFN-␤ (lane 4) resulted in increased phosphorylation (lanes 6 and 8, respectively) of a protein immunoprecipitated by anti-eIF-2␣ antibody, whose molecular size corresponds to eIF-2␣ (ϳ38 kDa). Based on these results, we used the anti-eIF-2␣ polyclonal antibody to immunoprecipitate the in vivo 32 P-labeled eIF-2␣. Phosphorylation of eIF-2␣ did not significantly differ between control (Fig. 5B, lanes 1-3) and PKR⌬6 cells (lanes 4 -6), which were stimulated either with LPS (lanes 2 and 5) or IFN-␥ (lanes 3 and 6). This experiment was performed three times with no significant variations in eIF-2␣ phosphorylation. Similarly, no significant differences in the levels of eIF-2␣ phosphorylation were observed when the isoelectric focusing and eIF-2␣ immunoblotting assay was used (Fig. 5C). These experiments suggest that eIF-2␣ is not a substrate for PKR activated by LPS or IFN-␥ in 70Z/3 cells.
Induction of NF-B Activity Is Not Inhibited by PKR⌬6 -The induction of -chain transcription in 70Z/3 cells upon LPS treatment requires the activation of NF-B, which binds to the -chain enhancer motif, GGGACTTTCC (50). It has been recently shown that the transcription inhibitor IB can be phosphorylated by PKR in vitro resulting in induction of NF-B DNA binding (28). Based on this finding, we examined the possibility that PKR⌬6 inhibits phosphorylation of IB by PKR resulting in inhibition of NF-B activation and consequently in a decrease of -chain transcription. To this end, we tested NF-B activity in nuclear extracts of 70Z/3 cells induced with LPS only since IFN-␥ induces -chain transcription in the absence of NF-B activation (51,52). NF-B binding to a DNA fragment containing two repeats of NF-B consensus sequence -GGGACTTTCC-was analyzed by the gel retardation assay (43). No detectable NF-B activity was observed in resting 70Z/3 control cells, as no protein-DNA complexes were formed (Fig. 6A, lane 1), but DNA binding was evident after LPS treatment (Fig. 6A, lane 2). The two inducible bands most likely correspond to the binding of one (lower band) or two NF-B complexes (upper band) to one or two B sites, respectively, in the DNA probe (see also below). Formation of these complexes was drastically reduced by competition with an unlabeled oligonucleotide containing the two NF-B-binding sites (Fig. 6A,  lane 3). Significantly, indistinguishable NF-B⅐DNA complexes were observed in control and two independent PKR⌬6-expressing clones (PKR⌬6 -1 and PKR⌬6 -20) upon LPS treatment (Fig. 6A, compare lane 2 to lanes 4 and 5).
Genes encoding B-binding proteins form a family of related genes that include NFKB1 (p50/p105), NFKB2 (p52/p100), vrel, c-rel, relA (p65), relA⌬ (p65⌬), and relB (for review, see Ref. 53). Recent findings suggest that treatment of pre-B cells with LPS changes the subunit composition of B-binding complexes from p50-p65 to p50-rel (54,55). Based on this observation we wished to investigate whether PKR⌬6 expression had an effect on NF-B subunit composition upon LPS induction. To examine which of the two B-binding complexes, p50-p65 or p50-rel, were involved in the binding to B site, we performed gel supershift assays by incubating nuclear extracts from a control clone (CON-8), wt PKR cells (polyclonal populations), and a PKR⌬6 clone (PKR⌬6-1) together with antibodies against p65, rel, or p50 protein. As shown in Fig. 6B, no differences in NF-B subunit composition between control, wt PKR, or PKR⌬6 cells were observed. The NF-B binding complexes consisted of p65 (lanes 5, 15, and 25), rel (lanes 7, 17, and 27), and p50 (lanes 9, 19, and 29) proteins. Similar results were obtained from three independent experiments after different periods of LPS stimulation (data not shown). These data are consistent with the existence of p50-p65 and p50-rel heterodimers in 70Z/3 cells (54,55). Thus, inhibition of -chain transcription in PKR⌬6-expressing cells is apparently not mediated through NF-B.

DISCUSSION
The interaction of mitogens and cytokines with their receptors triggers signaling cascades through the activation of kinases which result in the phosphorylation and activation of numerous proteins. The LPS-induced protein phosphorylation is mediated by mitogen-activated protein kinases (56,57), protein kinase C (58), protein kinase A (58), and tyrosine phosphorylation (59,60). Interaction of IFN-␥ with its receptor elicits a cascade of tyrosine phosphorylation of cytoplasmic and nuclear proteins resulting in transcriptional activation of genes (61). Recent findings suggest that serine phosphorylation is also important in IFN-␥ signaling (62). Analysis of mutant variants of 70Z/3 cells has shown that LPS and IFN-␥ share common signaling pathways (63)(64)(65). This is consistent with our data which demonstrate that PKR is a mediator of LPS and IFN-␥ signaling in 70Z/3 cells. However, the lack of complete inhibition of gene transcription by PKR⌬6 suggests that PKR activation is necessary but not sufficient for the induction of gene transcription and indicates the existence of other pathways which do not involve PKR.
PKR has also been implicated in several other signaling pathways. For example (i) activation of PKR is required for gene transcription induced by dsRNA (28, 29); (ii) stimulation of cell growth by interleukin-3 results in a decrease of PKR activity and eIF-2␣ phosphorylation concomitant with a stimulation of protein synthesis (66); (iii) induction of the tumoricidal activity of macrophages by LPS requires PKR (67); (iv) PKR mediates the induction of c-myc, c-fos, and JE genes upon platelet-derived growth factor treatment (68); and (v) induction of indoleamine 2,3-dioxygenase gene expression by IFN-␥ is mediated by PKR (69). These findings together with ours reveal a multifunctional and complex role for PKR in regulation of gene expression at two different levels, translation and transcription. It is not as yet clear how PKR activity is regulated by the different stimuli. One possibility is that PKR activity is induced by cellular dsRNA, whose nature and availability are dependent upon the cell type and/or stimuli. Our data show that the PKR-mediated effect of LPS or IFN-␥ is unlikely to proceed through eIF-2␣ phosphorylation, suggesting that phosphorylation of other protein(s) is required for this effect. This is consistent with earlier studies showing that new protein synthesis is not required for transcriptional activation of gene (50,70).
Like other eukaryotic genes, the gene is regulated by the interaction of sequence-specific DNA-binding proteins with cisacting DNA elements. NF-B transcription factor binds to the B site in the intron enhancer (J -C enhancer) of gene (50). Activation of NF-B requires IB phosphorylation and degradation (71,72). Interestingly, IB can be phosphorylated by the two eIF-2␣ kinases, heme control repressor (73) and PKR (28) in vitro. In 70Z/3 cells, LPS but not IFN-␥ induces NF-B activity, which is necessary but not sufficient for gene transcription (49,63,64,74). Our data show that PKR mediates transcription independently of NF-B. This is the second example of an NF-B-independent pathway of -gene transcription in 70Z/3 cells. Transforming growth factor-␤ inhibits LPSinduced gene transcription without affecting NF-B activation (51). Thus PKR activates at least two different pathways to the transcriptional machinery, NF-B dependent for dsRNA and NF-B independent for LPS or IFN-␥. However, it should be emphasized the difference in cell types used in these experiments. In this regard, IFN-␤ expression by dsRNA in mouse F9 embryonal carcinoma cells does not require NF-B activation (75), indicating that dsRNA signaling clearly differs between cell lines.
A second enhancer element, which lies 8.5 kilobases downstream of the gene, has been identified (3Ј enhancer) and contains an IFN consensus sequence (76). The 3Ј enhancer contains a binding site for B cell and macrophage-specific factor PU.I (77). PU.I recruits the binding of a second B cell-restricted nuclear factor, NF-EM5. DNA binding by NF-EM5 requires protein-protein interaction with PU.I and protein phosphorylation of PU.I (77). NF-EM5 is homologous to interferon regulatory proteins, 5 consistent with its function in IFN-␥ signaling. At the present time it is not known what kinase(s) regulates PU.I phosphorylation in vivo, and PKR is an intriguing possibility that remains to be examined. In conclusion, our data demonstrate that PKR is a mediator of LPS and IFN-␥ signaling to gene transcription and substantiate the transcriptional role of PKR in regulation of gene expression. Inasmuch as PKR plays a role in many pathophysiological events such as virus infections (78) including AIDS (79), and possibly cancer (16,17), the understanding of the mechanism of action of PKR is important for devising strategies to combat these diseases.