|1 November 1999, Volume 18, Number 45, Pages 6112-6120|
|Table of contents Previous Article Next [PDF]
|PKR; a sentinel kinase for cellular stress|
|Bryan RG Williams|
Department of Cancer Biology NB40, Lerner Research Institute, The Cleveland Clinic Foundation, 9500 Euclid Avenue, Cleveland, Ohio, OH 44195, USA
Correspondence to: Bryan RG Williams, Department of Cancer Biology NB40, Lerner Research Institute, The Cleveland Clinic Foundation, 9500 Euclid Avenue, Cleveland, Ohio, OH 44195, USA
The double stranded RNA (dsRNA)-activated protein kinase PKR is a ubiquitously expressed serine/threonine protein kinase that is induced by interferon and activated by dsRNA, cytokine, growth factor and stress signals. It is essential for cells to respond adequately to different stresses including growth factor deprivation, products of the inflammatory response (TNF) and bacterial (lipopolysaccharide) and viral (dsRNA) products. As a vital component of the cellular antiviral response pathway, PKR is autophosphorylated and activated on binding to dsRNA. This results in inhibition of protein synthesis via the phosphorylation of eIF2 and also induces transcription of inflammatory genes by PKR-dependent signaling of the activation of different transcription factors. Along with RNaseL, PKR constitutes the antiviral arm of a group of mammalian stress response proteins that have counterparts in yeast. What began as adaptation to amino acid deprivation and sensing unfolded proteins in the endoplasmic reticulum has evolved into a family of sophisticated mammalian stress response proteins able to mediate cellular responses to both physical and biological stress.
interferon; PKR; stress; apoptosis; signal transduction
PKR is an interferon (IFN)-induced protein, initially identified and characterized as a translational inhibitor in an antiviral pathway regulated by IFNs (Stark et al., 1998). More recently, it has become clear that PKR is a component of signal transduction pathways mediating cell growth control, and responses to stress (Tan and Katze, 1999). Although the most well characterized role of PKR is the inhibition of translation through phosphorylation of eukaryotic initiation factor 2 -subunit (eIF-2) (DeHaro et al., 1996), a process which is conserved from yeast to mammals, increasing evidence suggests that PKR may target other substrates and mediate multiple cellular processes. It is also possible that PKR may function through protein-protein interactions that are independent of catalytic activity. PKR has distinct auto- and substrate phosphorylation activities and may possess an unusually broad serine-threonine kinase substrate specificity. While PKR can be both activated and inhibited by dsRNA ligand, it is also activated via cytokine and stress signaling pathways that likely operate independent of dsRNA. Since the dsRNA binding and kinase activities of PKR are sequestered in distinct structural domains, the mechanisms of activation by different stimuli may be different. It is also possible that different stress stimuli lead to the production of dsRNA resulting in the activation of PKR through its dsRNA binding domains. In this review, the basic characteristics of PKR are discussed and evidence for its involvement in different stress activated signaling pathways presented.
Characteristics of PKR
PKR is a 551 (515 in mouse) amino acid protein consisting of two functionally distinct domains: an N-terminal dsRNA binding regulatory domain, and a C-terminal kinase catalytic domain (Figure 1a). The dsRNA binding domain (dsRBD; amino acids 1 - 170 of human PKR) contains two dsRNA binding motifs (dsRBMs) of approximately 70 amino acid residues each. The dsRBM is found in more than 40 different proteins of diverse function and origin, sometimes as single copy, but more often as multiple copies (St Johnson et al., 1992). This motif is rich in basic residues, clustered around its C-terminal region and in PKR a 20 amino acid linker connects the two dsRBMs. All dsRBM-containing proteins tested bind dsRNA independent of sequence but may recognize a specific higher ordered structure as in the case of Drosophila Staufen protein recognition of 3' UTR stem loops in bicoid mRNAs (St Johnston et al., 1991). Sequence homology between the two dsRBMs of PKR is 49%, with 29% of the residues identical. Homologous and identical residues in dsRBMs are preferentially located in the C-terminal basic region of the motif. The N-terminal structure of PKR including the two dsRBMs has been determined using NMR (Nanduri et al., 1998). The individual dsRBMs of PKR have identical ---- secondary structures, which are in turn identical to the structures of the individual dsRBM peptides derived from Staufen and RNase III (Nanduri et al., 1998; Kharrat et al., 1995; Bycroft et al., 1995). The 20 amino acid linker consists entirely of random coil conformation (Nanduri et al., 1998) and may allow the dsRBD to wrap around an A-form dsRNA helix and allow optimal interactions of each dsRBM with the RNA. The dsRBD structure coupled with mutagenesis data pinpoints the dsRNA contact sites and the specific requirement of 2'-OH groups within the RNA minor groove for PKR binding (Bevilacqua and Cech, 1996; McMillan et al., 1995). This allowed a model of the interaction between dsRBD and dsRNA (Nanduri et al., 1998) where conserved residues on the surface of each dsRBM make hydrogen bond contacts with the 2'-OH groups, as well as electrostatic interactions with the oxygens of the phosphate backbone (Nanduri et al., 1998). The flexible linker region allows the protein to wrap around the minor groove of the dsRNA helix, covering about 11 bp, consistent with biochemical studies (Bevilacqua and Cech, 1996; Manche et al., 1992). The accuracy of this model is supported by the crystal structure of a complex between dsRNA and the second dsRBM of Xenopus laevis RNA-binding protein A (Ryter and Schultz, 1998). The kinase domain of PKR contains the usual conserved protein kinase subdomains and the homology within subdomain VI identifies PKR as a serine/threonine kinase (Meurs et al., 1990). A distinctive feature of PKR, shared with the other eIF-2 kinases HRI and GCN2 is an insert region C-terminal of Domain IV. This insert region, and the dsRNA-binding regulatory domain, are key determinants of PKR function and regulation but as yet no structural information on the full length PKR molecule is available.
Activation by dsRNA
Activation and/or inhibition of PKR can occur in response to a number of agents, including dsRNA, polyanionic compounds such as heparin, and cellular and viral proteins. Cellular exposure to different stresses such as heat shock, viral infection, or translational inhibition increase the abundance of human Alu RNA which can form stable, discrete complexes with PKR in vitro and in vivo (Chu et al., 1998). Although initially characterized as PKR inhibitors, Alu RNAs are efficient activators of PKR at low concentrations (Figure 1b) and may act as physiological regulators of this kinase. The characteristic `bell curve' of PKR activation versus dsRNA concentration (Figure 1b) is a general feature of PKR activation with dsRNA. Viruses exploit this property by producing large quantities of dsRNA which can inhibit the elevated levels of PKR induced by IFNs in response to viral infection.
The kinetics of dsRNA binding to PKR are complicated by the presence of the two dsRBMs on PKR. Different binding scenarios include one or the other site occupied by a dsRNA molecule, both sites bound to different regions on the same dsRNA molecule, and two different dsRNA molecules bound to each protein site. There is also the possibility that several PKR molecules could attach themselves to a single long strand of dsRNA. The dsRBD of PKR can bind to any RNA containing sufficient A-form helical structure, regardless of non-Watson-Crick base pairs or mismatches (Bevilacqua et al., 1998) and most dsRNAs will bind to PKR or dsRBD with dissociation constants in the nano molar range. Point mutation and domain swapping experiments in the dsRBD have indicated that dsRBM1 is more important for dsRNA binding than dsRBM2 (Green and Mathews, 1992; McCormack et al., 1994; Green et al., 1995; Romano et al., 1995), although both motifs are required for optimal binding (Green and Mathews, 1992). The requirement of both dsRBMs for dsRNA binding, along with the greater importance of dsRBM1, suggests a cooperative mechanism for dsRNA binding and PKR activation.
Activation of PKR by dsRNA or other agents results in PKR autophosphorylation and activation of substrate phosphorylation (kinase) activity. DsRNA binding causes a major conformational change in PKR as evidenced by gel analyses of protein-RNA complexes (Manche et al., 1992), or by biophysical techniques using tryptophan fluorescence quenching and neutron scattering (Carpick et al., 1997). The conformational change likely serves to uncover a catalytic site(s) for autophosphorylation, or to shift domains from different parts of the same PKR molecule (or from different PKR molecules within a protein-protein complex) into an active conformation. The finding that dsRNA binding causes PKR to elongate rather than contract (Carpick et al., 1997) suggests a mechanism, whereby the dsRBD would swing away from the rest of the protein, exposing the catalytic site. The second-order kinetics of PKR autophosphorylation suggests this reaction is intermolecular (Kostura and Mathews, 1989). In fact dimerization is important for kinase activation (Langland and Jacobs, 1992; Thomis and Samuel, 1993; Romano et al., 1995; Patel et al., 1995; Ortega et al., 1996; Wu and Kaufman, 1996, 1997; Carpick et al., 1997; Tan et al., 1998) although exact molecular mechanisms and the protein structural and dsRNA requirements for dimerization, remain to be determined. PKR mutants which are capable of dimerization but lack dsRNA binding activity are biologically inactive, indicating that dimerization alone is not sufficient for PKR activity (Patel et al., 1995, 1996; Wu and Kaufman, 1996). Moreover, the inactive mutant PKR (K296R, Figure 1a) can dimerize in vitro (Carpick et al., 1997). PKR dimerization may be mediated in part through the dsRBD, either through direct protein-protein interactions in this region (Patel et al., 1995; 1996), or through dsRNA bridging the protein subunits (Cosentino et al., 1995; McMillan et al., 1995) but domains within the C-terminal half of PKR are also required and dimerization (Romano et al., 1995; Tan et al., 1998) can be blocked by a peptide corresponding to amino acids 244-296 (Tan et al., 1998). While multiple deletions within this region interfere with dimerization (Romano et al., 1995), the N-terminal 280 amino acids of PKR do not interact with the full length protein (Ortega et al., 1996). Highly purified PKR can form dimers in the absence of dsRNA, and is in equilibrium between the monomeric and dimeric forms (Carpick et al., 1997). Addition of dsRNA tends to shift this equilibrium towards the dimer (Carpick et al., 1997), either through a bridging effect or through a protein conformational change (or both). In addition, NMR experiments have shown that dsRBD is predominantly monomeric, even at very high (16 mg/mL) protein concentrations (Nanduri et al., 1998). Single dsRBMs from Staufen and RNase III were also shown to have monomeric solution structures by NMR (Kharrat et al., 1995; Bycroft et al., 1995). Clearly, the dsRBD alone is not sufficient to mediate PKR dimerization.
Non-dsRNA molecules can activate PKR including polyanions such as dextran sulfate, chondroitin sulfate, poly(L-glutamine) and heparin (Hovanessian and Galabru, 1987). PKR activation by heparin A does not require the dsRBD and a mutant lacking the N-terminal 145 amino acids can be activated by heparin (Patel et al., 1994). Recently protein activators of PKR have been described in both human (PACT) and mouse cells (RAX) (Patel and Sen, 1998; Ito et al., 1999). PACT contains two dsRBMs and heterodimerizes with PKR via the dsRBD but activates it in the absence of dsRNA (Patel and Sen, 1998). The activation of RAX and its association with PKR is stimulated by growth factor deprivation in IL3-dependent cells (Ito et al., 1999).
PKR autophosphorylation in response to dsRNA activation occurs following a structural rearrangement involving the C-terminal catalytic domain. Phosphopeptide analysis of PKR autophosphorylated in vitro has identified phosphorylation sites in the third basic region (Figure 1a) on Thr-258, Ser-242, and Thr-255 (Taylor et al., 1996). Mutation at Thr-258 reduced but did not eliminate PKR activity, while mutations at the other two residues exacerbated this effect (Taylor et al., 1996). MALDI/TOF MS phosphopeptide mapping has identified two other phosphorylation sites at Thr-446 and Thr-451 (Romano et al., 1998) which lie within the activation loop between kinase subdomains VII and VIII. Substitution of Thr-451 to alanine completely inactivated PKR, while a mutant with a Thr-446 to alanine substitution was partially active (Romano et al., 1998). Identical phosphorylation sites within the homologous loop region have been identified in the yeast eIF-2 kinase GCN2, and are found in homologous regions of a number of other protein kinases (Romano et al., 1998). Phosphorylation within the third basic region may serve to lock PKR in an active conformation with the catalytic domain irreversibly uncovered by the regulatory domain, while phosphorylation in the activation loop serves to facilitate substrate binding and catalysis (Romano et al., 1998). Mutation of Glu-490 to Gln partially restored the activity of the Ala-451 mutant, suggesting an interaction between the activation loop and this region, and indicating structural homology with MAP kinase (Romano et al. 1998). There is no sequence homology between the two phosphorylation sites in the basic region and activation loop of PKR, suggesting that PKR has multiple auto-substrate specificities.
As discussed above, PKR can be both activated and inhibited by dsRNA, and its inhibition by viral transcripts is a mechanism to aid viral evasion of the immune system. It is also possible that endogenous regulation of PKR may occur through expression of cellular RNAs. In recent years it has become increasingly clear that many proteins are capable of inhibiting PKR, either by direct interaction, or through binding to activators or substrates (reviewed in Clemens and Elia, 1997, Tan and Katze, 1999). Some of these proteins are virally encoded, while some are cellular. The endogenous cellular inhibitors may exist to regulate PKR in response to stress and can themselves facilitate PKR inhibitory pathways that are exploited by viruses (for example P58 discussed below). Some inhibitors, such as HIV Tat and Vaccinia K3L, are either substrates or pseudosubstrates for PKR phosphorylation. It is likely that those protein inhibitors that bind directly to PKR may serve to block the substrate interaction sites and the mapping of the PKR and inhibitor contact sites provides clues as to the mechanisms and structural requirements for kinase activation and dimerization, substrate recognition, and substrate phosphorylation. However, no consensus inhibition site has been identified and it is likely PKR is regulated through protein - protein interactions with different regulatory and/or effector molecules (for example P58, Stats, p53 and B56, see below) directing or targeting PKR to varied cellular processes.
To date, the only physiological substrate for PKR that has been characterized in detail is eIF-2 (DeHaro et al., 1996). PKR phosphorylates eIF-2 on Ser-51 which lies in a basic region and is flanked by four arginine residues on the C-terminal side (Colhurst et al., 1987). A peptide corresponding to residues 45 - 56 of eIF-2 (ILLSELSRRRIR) can also be phosphorylated on Ser-51 by PKR, another mammalian eIF-2 kinase HRI, and protein kinase C (PKC) (Mellor and Proud, 1991). A region of PKR (amino acids 362 - 370) immediately adjacent to the conserved eIF-2 kinase insert domain is required for PKR kinase and full eIF-2 binding activity (Cai and Williams, 1998). Hydrophobicity in this region is essential for kinase activity but not eIF-2 binding, while Glu-367 is required for full substrate binding but not PKR kinase activity (Cai and Williams, 1998). These results suggest that PKR structural features required for substrate binding are different from those required for auto- and/or substrate phosphorylation, although there is likely to be some overlap.
PKR phosphorylation sites on HIV-1 Tat and on PKR itself have no homology to the kinase site on eIF-2. Thus, there are no obvious clues in terms of sequence homology as to the identity of novel PKR substrates. PKR regulates NF-B through phosphorylation of its inhibitor, IB (Kumar et al., 1994). Although IB can be phosphorylated by PKR in vitro, it is probably not a direct PKR substrate in vivo. Rather, PKR regulates NF-B activation via the IB kinase complex but the direct substrate remains to be identified (DiDonato et al., 1997, Zamanian-Daryoush et al., 1999 submitted). PKR also physically interacts with Stat1 (Wong et al., 1997), Stat3 (Deb et al. submitted), and p53 but these do not appear to be direct substrates and consequently, there remains a dearth of information on the nature and variety of PKR phosphorylation sites.
Signaling via PKR
Different roles have been suggested for PKR in various cellular processes, including growth regulation, antiviral protection, signal transduction and differentiation. Most cells express constitutive levels of PKR, but exposure of cells to type I IFNs results in a several-fold increase in PKR levels as a result of transcriptional activation of the gene (Meurs et al., 1990). The discovery that dsRNA could activate NF-B led to the identification of PKR as the signal transducer (reviewed in Williams, 1995). It is now clear that this involves the IB kinase complex and results in the preferential degradation of IB (Zamanian-Daryoush et al., 1999 submitted, Figure 2). Although the PKR substrate in this pathway has not been identified, the kinase activity of PKR is required and it does not perform only a scaffolding role. Recombinant active PKR added to cell extracts can activate NF-B DNA binding activity and transdominant mutants of PKR disrupt NF-B dependent reporter activity in transient transfection assays (Kumar et al., 1994).
Several early reports suggested that PKR may respond to other extracellular signals (reviewed in Clemens and Elia, 1997; Williams, 1995). That tumor necrosis factor alpha (TNF-) could be added to this list was revealed by experiments in the promonocytic U937 cells where subclones of cells carrying a PKR antisense expression vector was resistant to TNF--induced apoptosis (Yeung et al., 1996). Unlike most cell lines, U937 cells, which express very low levels of PKR, can tolerate overexpression but such cells exhibit increased apoptosis. In accord with this, mouse embryo fibroblasts (MEFs) derived from PKR-null mice are resistant to TNF-induced apoptosis (Der et al., 1997). Remarkably, these fibroblasts also exhibit a 1000-fold increase in resistance to apoptosis-induced by lipopolysaccharide. They are also more resistant to serum deprivation and have altered cell cycle characteristics (Zamanian-Daryoush et al., 1999). TNF at relatively high concentrations can induce apoptosis in cultured cells by a mechanism that involves eIF-2 phosphorylation (Srivastava et al., 1998). The connection between protein synthesis inhibition resulting from eIF-2 phosphorylation and apoptosis is not clear but overexpression of Ser-51-Ala mutant eIF-2 that is resistant to phosphorylation protects cells from TNF-induced (and serum induced) apoptosis (Srivastava et al., 1998). The link between the TNF signaling pathway and PKR activation also remains to be determined. There is a deficiency in TNF signaling of NF-B in PKR null cells or in cells expressing transdominant mutants of PKR that is manifest in an inability to sustain NF-B activation (Zamanian-Daryoush et al., 1999; Chesire et al., 1999). Therefore, the defect in apoptosis induction by TNF in PKR null cells could be due in part to a failure to upregulate the expression of antiapoptotic genes dependent on NF-B. At present, a pathway can be described where several stress-related signals (IL1, TNF, dsRNA, LPS and IFNs) are channeled through PKR to activate NF-B (Figure 2). In the absence of PKR, NF-B signaling still occurs, but is attenuated.
PKR and environmental stress response
PKR belongs to an interesting family of stress regulated proteins (Silverman and Williams, 1999; Kaufman, 1999). In yeast, the eIF2 kinase GCN2 senses amino acid starvation via uncharged tRNA. Unfolded proteins in the endoplasmic reticulum (ER) are sensed by Ire1, which is both a kinase and an endoribonuclease. Thus, both GCN2 and Ire1 recognize RNA and phosphorylate other proteins. In mammals, Ire1 and and PKR-like ER kinase (PERK), are localized to the ER where they are sensors and effectors of the unfolded protein response. PKR, along with RNase L, constitute the major IFN-regulated antiviral pathway in mammals. RNase L which has an inactive kinase domain and a 2'-5' oligoadenylate (2-5A) binding domain requires 2-5A to bind to the inactive monomeric form of the enzyme promoting dimerization and activation of endoribonuclease activity. Ire1 proteins contain the RNase L consensus sequence at their C-termini responsible for the hydrolysis of RNA. A functional link exists among PKR, RNase L and Ire1 in that they mediate apoptosis induced by different stimuli (Silverman and Williams, 1999; Kaufman, 1999).
Accumulation of proteins in the ER membrane termed the ER overload response (EOR) can occur during virus infection leading to the release of Ca2+ from the ER, production of reactive oxygen intermediates and NF-B activation (Pahl and Baeuerle, 1997). While the kinase(s) responsible has not been identified, it remains possible that PKR can contribute to this effect. In PKR null cells, or in cells that overexpress the PKR-inhibitory VA1 RNAs, protein production is enhanced (Terenzi et al. submitted). Although the VA1 effect has been attributed to inhibition of PKR-dependent phosphorylation of eIF2, suppression of a PKR-mediated EOR may also be involved given its role in regulating NF-B activation.
PKR has been implicated as an ER stress regulated kinase, but since the discovery of PERK, the role of PKR is unclear. The lack of specific reagents to measure the activated form of these eIF2 kinases hinders their analysis. In neither case are phospho-specific antibodies available. PKR wild type and null cells do not exhibit differential sensitivity to protein synthesis inhibition by tunicamycin, an N-glycosylation inhibitor that stimulates an unfolded protein response (UPR) and an EOR and increases autophosphorylation of PERK (Harding et al., 1999). Treatment of cells with different ER stress stimuli including calcium ionophores, inhibitors of calcium-dependent ATPase, and ionomycin, stimulate PKR activation and eIF2 phosphorylation and under some circumstances, this can be blocked by a catalytically inactive mutant of PKR suggesting a role for PKR in perhaps ER stress induced by depletion of calcium stores. Okadaic acid, a protein phosphatase 2A (PP2A) inhibitor, can be used to discriminate between the UPR and the EOR as it stimulates the latter and eIF2 phosphorylation without effecting the UPR. Okadaic acid is also an activator of NF-B through the production of reactive oxygen intermediates (Pahl and Baeuerle, 1997; Schmidt et al., 1995). Interestingly, PKR interacts with and phosphorylates the PP2A regulatory subunit B56 (Xu and Williams submitted) suggesting the possibility that PKR may regulate the EOR by modulating the activity of PP2A.
PKR and p53
The tumor suppressor p53 is an important sensor of DNA damage induced by genotoxic and other stress. While PKR does not appear to be activated by some agents that damage DNA and activate p53 such as etoposide (Zamanian-Daryoush et al. unpublished observations), a link to p53-mediated activities has been observed under different conditions. For example, it has been shown in U937 cells where TNF- is an effective inducer of apoptosis, apoptosis induction by TNF coincides with an induction of p53 that is preceded by PKR induction (Yeung and Lau, 1998). Inhibition of p53 expression in U937 cells overexpressing PKR prevents TNF- induced apoptosis in these cells. Conversely, overexpressing wild type p53 in PKR-deficient U937 cells confers susceptibility to TNF--induced apoptosis. U937 cells overexpressing PKR have constitutively higher levels of p53, and spontaneously undergo apoptosis even without TNF- treatment. Therefore, in these cells it seems that PKR-dependent induction of p53 is necessary for TNF--induced apoptosis. PKR also physically associates with the C-terminus of p53 and can phosphorylate serine 392 in vitro (Cuddihy et al., 1999a). IFN enhances the interaction between p53 and PKR. Both the transcriptional activity and cell cycle arrest functions of p53 are impaired in PKR null MEFs. Defective phosphorylation of mouse p53 on Ser18 is seen in these cells which is consistent with an impaired transcriptional induction of the p53-inducible genes p21(WAF/Cip1) and Mdm2. Ser18 phosphorylation and transcriptional activation by mouse p53 were also decreased in PKR null cells after DNA damage induced by adriamycin or gamma but not UV radiation. A role for phosphatidylinositol-3 (PI-3) kinase in mediating PKR-dependent phosphorylation of Ser18 of p53 was suggested by the observation that the PI-3 kinase inhibitor LY294002 blocked adriamycin induction of phosphorylation (Cuddihy et al., 1999b). These findings suggest that PKR enhances p53 transcriptional function and implicate PKR in cell signaling elicited by a specific type of DNA damage that leads to p53 phosphorylation, possibly through a PI-3 kinase pathway. However, these data were generated from non-isogenic fibroblasts and needs to be confirmed using MEFs derived from matched wild type and null litter mates.
PKR and p38 MAPK
The mammalian p38 MAPK family are activated by cellular stress including UV irradiation, osmotic shock, heat shock, lipopolysaccharide, protein synthesis inhibitors, and cytokines such as IL-1, and TNF-. Recently, we have found that p38 MAPK is a necessary component for IFN signaling where it directs the phosphorylation and activation of cytosolic phospholipase A2, an essential step in the IFN activated assembly of the transcription factor complex, ISGF3 (Goh KC et al. submitted). IFN or activation of p38MAPK also results in the phosphorylation of the transcription factor Stat1 on Ser 727 since this can be blocked by the p38 MAPK inhibitor SB203580 (Goh KC et al. submitted). The transcriptional activation potential of IFNs and their antiviral activity can also be inhibited by SB203580 or by the expression of dominant negative p38. In accord with these results, the transcriptional activity mediated by ISGF3 or Stat1 can be enhanced by constitutively active MKK6, the upstream activator of p38. In PKR null cells, signal transduction and gene induction by IFN- is attenuated as a result of deficient activation of the transcription factor IRF-1 (Kumar et al., 1997). However, in PKR null cells, there is a defect in IFN--induced phosphorylation of Ser727 on Stat1 (Ramana et al. submitted) suggesting that PKR might act upstream of p38. In accord with this, Stat1 Ser727 phosphorylation in response to different p38 activators including LPS, TNF- and pIC is defective in PKR null MEFs (Goh et al. submitted). Since p38 can regulate NF-B-dependent transcription after its translocation into the nucleus (Norris and Baldwin, 1999), it remains possible that PKR can effect NF-B dependent gene transcription by both IKK-and p38 dependent pathways (Figures 2 and 3).
PKR and growth factor signaling
PKR has growth regulatory properties and acts as an oncogene when catalytically inactive mutants are overexpressed in NIH3T3 cells (reviewed in Clemens and Elia, 1997; Tan and Katze, 1999). Transformation could be driven in part by a deregulation of protein synthesis initiation since mutant forms of eIF2, a PKR substrate, are also transforming oncogenes in this assay (Donze et al., 1995). Since dominant negative mutants of PKR also inactivate its activities as a signal transducer, other mechanisms of transformation can be envisaged including a failure to maintain the homeostasis and inhibition of apoptosis. PKR has also been implicated in the activation of immediate early gene expression by platelet-derived growth factor (PDGF), and protects against apoptosis induced by serum withdrawal from fibroblasts or interleukin 3 (IL3) withdrawal from dependent cell lines (Mundschau and Faller, 1992, 1995; Ito et al., 1994). PDGF activates PKR autophosphorylation but this can be blocked by expression of activated p21ras which induces a cellular inhibitor of PKR activation (Mundschau and Faller, 1995). This has the same inhibitory effect on immediate early gene expression induced by PDGF as 2-amino purine, a chemical inhibitor of PKR. Down-regulation of PKR by antisense also blocks the induction of c-myc, c-fos, and JE by PDGF. While the molecular details are still to be clearly defined, PKR does not provide a growth promoting signal for PDGF as PKR null cells still undergo mitogenesis in response to PDGF (Deb et al. submitted). However, in PKR null cells PDGF induction of c-fos is deficient. In wild type cells, PKR physically interacts with Stat3 and appears necessary for Stat3 DNA binding activity since in the absence of PKR, this is defective (Deb et al. submitted). In addition, PKR appears to be necessary for the PDGF-mediated phosphorylation of Stat3 on serine 727 potentiating its transcriptional activation potential. PKR does not appear to be a direct Stat3 kinase but clearly promotes the PDGF-mediated phosphorylation of Stat3 on both tyrosine (allowing the formation of homo and hetero Stat dimers and DNA binding) and serine (increasing transactivation potential) (Deb et al. submitted and unpublished results). In contrast to the PKR-dependent, IFN-induced phosphorylation of serine 727 on Stat1, which is mediated by PKR and p38MAPK, the PKR-dependent, PDGF-induced phosphorylation of serine 727 on Stat3 is not mediated by p38, but is sensitive to inhibition by PD98059 a specific ERK1/ERK2 inhibitor (Deb et al. submitted). This is consistent with PKR signaling through the growth factor-driven ERKs and stress-driven p38MAPK (Figure 4). While this activity of PKR is not linked to growth promotion, it does appear to be necessary for promoting apoptosis under the stress signal of low serum and, furthermore, inhibition of PKR contributes to the transforming activity of activated Ras. Interestingly, reovirus which does not proliferate in NIH3T3 cells, is able to grow in ras (or V-erbB, or Sos) transformed cells as a result of inhibition of PKR activity, indicating that events upstream of ras contribute to PKR-dependent signaling (Strong et al., 1998). Since oncogenic ras can activate NF-B through a p38 or related stress activated kinase pathway and signals from an oncogenic ras pathway block PKR, then it follows that PKR should not be able to contribute to the activation of NF-B in transformed fibroblasts.
A connection between the tyrosine kinase activity of the PDGF receptor and serine phosphorylation required to activate PKR has not been made. There are similar gaps in our knowledge of how the Jak tyrosine kinase pathway is linked to PKR. However, in a murine interleukin 3 (IL-3)-dependent cell line, IL-3 deprivation results in PKR activation that may involve a protein activator (Ito et al., 1999). Addition of IL-3 to deprived cells induces a reciprocal response characterized by the rapid dephosphorylation and inactivation of PKR. The PKR-associated protein, RAX and its human homolog PACT (Patel and Sen, 1998) activate PKR in the absence of dsRNA in vitro. However, although overexpression of RAX does not induce PKR activation or inhibit IL-3-dependent cell growth, IL-3 deprivation induces the rapid phosphorylation of RAX followed by RAX-PKR association and activation of PKR (Ito et al., 1999, Figure 5). Interestingly, other cell stress inducers including arsenite, thapsigargin, and H2O2 also induce RAX phosphorylation and PKR activation. Therefore, cellular RAX (and by inference, PACT) is a candidate for a stress-activated, physiologic activator of PKR. The connection between IL-3 withdrawal, PKR activation and downstream events including apoptosis is not yet clear. However, it has been proposed that IL-3 suppresses apoptosis by phosphorylating Bcl2 at a conserved serine residue (Ser70) which is required for Bcl2 to suppress apoptosis after IL-3 withdrawal (Deng et al., 1998). Phosphorylation of Bcl2 occurs rapidly after the addition of IL-3 to deprived cells but can be reversed by the action of an okadaic acid (OA)-sensitive phosphatase. A role for protein phosphatase (PP) 2A as the Bcl2 regulatory phosphatase is possible since intracellular PP2A co-localizes and directly interacts with Bcl2 and the purified PP2Ac catalytic subunit directly dephosphorylates Bcl2 in vitro. Treatment of factor-deprived cells with the signaling agonist as bryostatin 1 increases the association between PP2A and Bcl2. This occurs rapidly and is most pronounced at the peak of Bcl2 phosphorylation before dephosphorylation occurs. The association of PP2A and Bcl2 requires an intact Ser70 site. PKR has not been implicated in these experiments but recently we have found that PKR physically interacts with and phosphorylates a PP2A regulatory subunit, B56 (Xu and Williams, submitted). The phosphorylation of B56 by PKR prevents it acting as an inhibitor of PP2A core dimer activity (using eIF-2 as a substrate). Thus, PKR may regulate different cellular activities involving PP2A including growth factor induced suppression of apoptosis.
PKR, viral mediated apoptosis and heat shock
Activation of PKR is necessary to efficiently signal the induction of expression of different genes involved in the cellular inflammatory response, including the synthesis of chemokines, class I MHC, and molecules involved in apoptosis. The loss of PKR function not only results in a decreased antiviral activity of IFNs but also attenuates the cellular response to certain cytokines and growth factors. To circumvent the antiviral effects of IFN and to reduce an inflammatory reaction mediated by PKR, different viruses have elaborated mechanisms to inhibit PKR (reviewed in Gale and Katze, 1998). These include the synthesis of inhibitory dsRNAs (adenoviruses, Epstein-Barr virus, human immunodeficiency virus) or the synthesis of proteins which can bind and sequester dsRNA activators of PKR (reovirus, vaccinia). Other viruses synthesize protein inhibitors of PKR (hepatitis C virus, herpes simplex virus, vaccinia) or proteases that cleave PKR (poliovirus, encephalomyocarditis virus). Inhibition of apoptosis is commonly employed by different viruses, but is likely a delaying tactic to allow sufficient viral replication for cells to eventually undergo apoptosis. In influenza virus infection, PKR has been implicated in apoptosis (Takizawa et al., 1996). This occurs even though influenza virus recruits a cellular inhibitor of PKR, P58IPK, to inhibit or delay apoptosis. P58IPK is a tetratricopeptide repeat-containing protein cochaperone that is present in normal cells as an inactive complex with the molecular chaperone Hsp40 and a Hsp90 related protein, P52rIPK. Influenza virus infection or heat shock disrupts the Hsp40-P58IPK complex thereby activating P58IPK that then associates with and stimulates the ATPase activity of Hsp70 possibly directing hsp/Hsc70 to refold, and thus inhibit kinase function (Lee et al., 1992; Gale et al., 1998; Melville et al., 1999). Although P58IPK can transform NIH3T3 cells, this property does not require binding to or inhibition of PKR since a mutant P58IPK which cannot bind PKR retains its transforming ability (Tang et al., 1999). However, in this case, apoptosis induced by TNF- is blocked suggesting that P58IPK can act as an inhibitor of apoptotic stimuli independent of PKR (Tang et al., 1999).
Because PKR is a cell growth inhibitor and transdominant mutants of the enzyme can transform NIH3T3 cells, some attempts have been made to link PKR dysregulation to cancer. PKR levels have been inversely correlated with proliferative activity in different human tumors and tumor cell lines (Haines et al., 1996; Zhou et al., 1998; Shimanda et al., 1998). Human invasive ductal breast carcinomas have high levels of PKR and in several breast carcinoma cell lines PKR levels are high compared to those found in lines derived from normal breast (Haines et al., 1996; Savinova et al., 1999). Despite high levels, the activity of PKR from the carcinoma cells is attenuated although it remains capable of binding dsRNA. Mixing experiments suggest these lines contain a transdominant inhibitor of PKR (Savinova et al., 1999). Unfortunately, other than eIF-2 phosphorylation, no other functional assays on PKR mediated activities have been performed and it remains to be determined whether PKR-dependent stress signaling pathways are perturbed. Nevertheless, disrupting the control of protein synthesis initiation by PKR may play a role. In support of this ribosomal protein L18, a 22 kDa protein that is overexpressed in colorectal cancer tissue, has been shown to bind to PKR through dsRBM1 (Kumar et al., 1999). L18 inhibits both PKR autophosphorylation and PKR-mediated phosphorylation of eIF-2 in vitro and in vivo. Although overexpression of L18 in tumors may promote protein synthesis and cell growth through inhibition of PKR activity, there are a large number of dsRNA-binding proteins that also bind PKR (Patel et al., 1999) and identifying any one of these as the crucial PKR regulator in tumorigenesis without more direct evidence is speculative.
Although the regulation of eIF-2 phosphorylation has remained a focus of several studies linking PKR to cell growth control, a connection to regulation of the proto-oncogene c-myc may be more relevant to tumor development. Interferon suppresses c-myc expression in M1 myeloid leukemia cells and inactive mutant forms of PKR can abrogate this (Raveh et al., 1996). Transfection of M1 cells with wild type human PKR inhibits proliferation in the absence of interferons, but this could be rescued by the ectopic expression of deregulated c-myc providing a link between PKR and c-Myc suppression. In accord with these results, transfection of breast carcinoma MCF-7 cells, with a dominant-negative mutant of PKR, relieved c-Myc down-regulation occurring as a consequence of the synergistic inhibition of growth with all-trans-retinoic acid and IFN- (Shang et al., 1998). Recent experiments show that both PKR and Stat1 are required for IFN-mediated down-regulation of c-myc (Ramana et al., submitted). In PKR or Stat1 null cells, c-Myc gene expression is stimulated, not repressed, by IFN. Since PKR-null cells exhibit deficient Stat1 serine 727 phosphorylation, PKR is clearly placed in pathways required for cell growth control and potentially, tumor suppression. Moreover, a link can be made to apoptosis since constitutive Stat1 expression and phosphorylation on serine 727 is necessary for the basal expression of caspase 3 and sensitivity to apoptosis induction by different stress stimuli (Kumar et al., 1997).
PKR is a key player in the cellular response to different situations of stress (Figure 6). The outcome of PKR activation is to inhibit protein synthesis initiation and to activate the transcription of genes involved in an inflammatory response. Whether these events are mutually exclusive likely depends on the nature and strength of the stress signal. With the exception of dsRNA, the nature of PKR activation by other stress stimuli remains to be determined. The link from PKR to stress-activated kinases such as p38 is being established, but providing an integrated picture of this complex network will continue to provide challenges.
I thank B Carpick for Figure 1b and R Silverman and KC Goh for comments on the manuscript and helpful discussions.
Bevilacqua PC and Cech TR. (1996). Biochemistry 35, 9983-9994. MEDLINE
Bevilacqua PC, George CX, Samuel CE and Cech TR. (1998). Biochemistry 37, 6303-6316. MEDLINE
Bycroft M, Grünert S, Murzin AG, Proctor M and St Johnston D. (1995). EMBO J. 14, 3563-3571. MEDLINE
Cai R and Williams BRG. (1998). J. Biol. Chem. 273, 11274-11280. MEDLINE
Carpick BW, Graziano V, Schneider D, Maitra RK, Lee X and Williams BRG. (1997). J. Biol. Chem. 272, 1291-1296. MEDLINE
Chesire JL, Williams BR and Baldwin Jr AS. (1999). J. Biol. Chem. 274, 4801-4806. MEDLINE
Chu W-M, Ballard R, Carpick BW, Williams BRG and Schmid CW. (1998). Mol. Cell. Biol. 18, 58-68. MEDLINE
Clemens MJ and Elia A. (1997). J. Interferon. Cyto. Res. 17, 503-524.
Colthurst DR, Campbell DG and Proud CG. (1987). Eur. J. Biochem. 166, 357-363. MEDLINE
Cosentino GP, Venkatesan S, Serluca FC, Green SR, Mathews MB and Sonenberg N. (1995). Proc. Natl. Acad. Sci. USA 92, 9445-9449. MEDLINE
Cuddihy AR, Wong AH, Tam NW, Li S and Koromilas AE. (1999a). Oncogene 18, 2690-2702. MEDLINE
Cuddihy AR, Li S, Wai Ning Tam N, Hoi-Tao Wong A, Taya Y, Abraham N, Bell JC and Korolmilas AE. (1999b). Mol. Cell. Biol. 19, 2475-2484. MEDLINE
De Haro C, Méndez R and Santoyo J. (1996). FASEB J. 10, 1378-1387. MEDLINE
Deng X, Ito T, Carr B, Mumby M and May Jr WS. (1998). J. Biol. Chem. 273, 34157-34163. Article MEDLINE
Der SD, Zhou A, Williams BRG and Silverman RH. (1997). Proc. Natl. Acad. Sci. USA 94, 3279-3283. Article MEDLINE
DiDonato JA, Hayakawa M, Rothwarf DM, Zandi E and Karin M. (1997). Nature 388, 548-553. Article MEDLINE
Donze O, Jagus R, Koromilas AE, Hershey JW and Sonenberg N. (1995). EMBO J. 14, 3828-3834. MEDLINE
Gale Jr M and Katze MG. (1998). Pharmacol. Ther. 78, 29-46. MEDLINE
Green SR, Manche L and Mathews MB. (1995). Mol. Cell. Biol. 15, 358-364. MEDLINE
Green SR and Mathews MB. (1992). Genes Dev. 6, 2478-2490. MEDLINE
Haines GK, Cajulis R, Hayden R, Duda R, Talamonti M and Radosevich JL. (1996). Tumour Biol. 17, 5-12. MEDLINE
Harding HP, Zhang Y and Ron D. (1999). Nature 397, 271-274. Article MEDLINE
Hovanessian AG and Galabru J. (1987). Eur. J. Biochem. 167, 467-473. MEDLINE
Ito T, Jagus R and May WS. (1994). Proc. Natl. Acad. Sci. USA 91, 7455-7459. MEDLINE
Ito T, Yang M and May WS. (1999). J. Biol. Chem. 274, 15427-15432. MEDLINE
Kaufman RJ. (1999). Genes Dev. 13, 1211-1233. MEDLINE
Kharrat A, Macais MJ, Gibson TJ, Nigles M and Pastore A. (1995). EMBO J. 14, 3572-3584. MEDLINE
Kostura M and Mathews M. (1989). Mol. Cell. Biol. 9, 1576-1586. MEDLINE
Kumar A, Commane M, Flickinger TW, Horvath CM and Stark GR. (1997). Science 278, 1630-1632. Article MEDLINE
Kumar A, Haque J, Lacoste J, Hiscott J and Williams BRG. (1994). Proc. Natl. Acad. Sci. USA 91, 6288-6292. MEDLINE
Kumar KU, Srivastava SP and Kaufman RJ. (1999). Mol. Cell. Biol. 19, 1116-1125. MEDLINE
Lee TG, Tomita J, Hovanessian AG and Katze MG. (1992). J. Biol. Chem. 267, 14238-14243. MEDLINE
Langland JO and Jacobs BL. (1992). J. Biol. Chem. 267, 10729-10736. MEDLINE
Manche L, Green S, Schmedt C and Mathews M. (1992). Mol. Cell. Biol. 12, 5238-5248. MEDLINE
McCormack SJ, Ortega LG, Doohan JP and Samuel CE. (1994). Virology 198, 92-99. MEDLINE
McMillan NAJ, Carpick BW, Hollis B, Toone WM, Zamanian-Daryoush M and Williams BRG. (1995). J. Biol. Chem. 270, 2601-2606. MEDLINE
Mellor H and Proud CG. (1991). Biochem. Biophys. Res. Comm. 178, 430-437. MEDLINE
Melville MW, Tan SL, Wambach M, Song J, Morimoto RI and Katze MG. (1999). J. Biol. Chem. 274, 37977-37803.
Meurs E, Chong K, Galabru J, Thomas NSB, Kerr IM, Williams BRG and Hovanessian AG. (1990). Cell 62, 379-390. MEDLINE
Mundschau LJ and Faller DV. (1992). J. Biol. Chem. 267, 23092-23098. MEDLINE
Mundschau LJ and Faller DV. (1995). J. Biol. Chem. 270, 3100-3106. MEDLINE
Nanduri S, Carpick BW, Yang Y, Williams BRG and Qin J. (1998). EMBO J. 17, 5458-5465.fc. MEDLINE
Norris JL and Baldwin Jr AS. (1999). J. Biol. Chem. 274, 13841-13846. MEDLINE
Ortega LG, McCotter MD, Henry GL, McCormack SJ, Thomis DC and Samuel CE. (1996). Virology 215, 31-39. MEDLINE
Pahl HL and Baeuerle PA. (1997). Trends Biochem. Sci. 22, 63-67. MEDLINE
Patel RC and Sen GC. (1998). EMBO J. 17, 4379-4390. Article MEDLINE
Patel RC, Stanton P, McMillan NAJ, Williams BRG and Sen G. (1995). J. Biol. Chem. 92, 8283-8287.
Patel RC, Stanton P and Sen GC. (1994). J. Biol. Chem. 269, 18593-18598. MEDLINE
Patel RC, Stanton P and Sen GC. (1996). J. Biol. Chem. 271, 25657-25663. MEDLINE
Patel RC, Vestal DJ, Xu Z, Bandyopadhyay S, Guo W, Erme S, Williams BRG and Sen GC. (1999). J. Biol. Chem., in press. .
Raveh T, Hovanessian AG, Meurs ER, Sonenberg N and Kimchi A. (1996). J. Biol. Chem. 271, 25479-25484. MEDLINE
Romano PR, Garcia-Barrio MT, Zhang X, Wang Q, Taylor DR, Zhang F, Herring C, Mathews MB, Qin J and Hinnebusch AG. (1998). Mol. Cell. Biol. 18, 2282-2297. MEDLINE
Romano PR, Green SR, Barber GN, Mathews MB and Hinnebusch AG. (1995). Mol. Cell. Biol. 15, 365-378. MEDLINE
Ryter JM and Schultz SC. (1998). EMBO J. 17, 7505-7513. Article MEDLINE
Savinova O, Joshi B and Jagus R. (1999). Int. J. Biochem. Cell. Biol. 31, 175-189. MEDLINE
Schmidt KN, Traencker Eb, Meier B and Baeuerle PA. (1995). J. Biol. Chem. 270, 27136-27142. MEDLINE
Shang Y, Baumrucker CR and Green MH. (1998). J. Biol. Chem. 273, 30608-30613. MEDLINE
Shimanda A, Shiota G, Miyata H, Kamahora T, Kawasaki H, Shiraki K, Hino S and Terada T. (1998). Cancer Res. 58, 4434-4438. MEDLINE
Silverman RH and Williams BRG. (1999). Nature 397, 208-211. Article MEDLINE
Srivastava SP, Kumar KU and Kaufman RJ. (1998). J. Biol. Chem. 273, 2416-2423. Article MEDLINE
St Johnson D, Brown NH, Gall JG and Jantsch M. (1992). Proc. Natl. Acad. Sci. USA 89, 10979-10983. MEDLINE
St Johnston D, Beuchle D and Nüsslein-Volhard C. (1991). Cell 66, 51-63. MEDLINE
Stark GR, Kerr IM, Williams BRG, Silverman RH and Schreiber RD. (1998). Annu. Rev. Biochem. 67, 227-264. MEDLINE
Strong JE, Coffey, MC, Tang D, Sabinin P and Lee PW. (1998). EMBO J. 17, 3351-3362. Article MEDLINE
Takizawa T, Ohashi K and Nakanishi Y. (1996). J. Virol. 70, 8128-8132. MEDLINE
Tan S-L, Gale Jr MJ and Katze MG. (1998). Mol. Cell. Biol. 18, 2431-2443. MEDLINE
Tan S-L and Katze MG,. (1999). JICR 19, 543-554.
Tang NM, Korth MJ, Gale Jr, M, Wambach M, Der SD, Bandyopadhyay SK, Williams BR and Katze MG. (1999). Mol. Cell. Biol. 19, 4757-4765. MEDLINE
Taylor DR, Lee SB, Romano PR, Marshak DR, Hinnebusch AG, Esteban M and Mathews MB. (1996). Mol. Cell. Biol. 16, 6295-6302. MEDLINE
Thomis DC and Samuel CE. (1993). J. Virol. 67, 7695-7700. MEDLINE
Williams BRG. (1995). Virology 6, 191-202.
Wong AHT, Tam NWN, Yang YL, Cuddihy AR, Li SY, Kirchhoff S, Hauser H, Decker T and Koromilas AE. (1997). EMBO J. 16, 1291-1304. Article MEDLINE
Wu S and Kaufman RJ. (1996). J. Biol. Chem. 271, 1756-1763. MEDLINE
Wu S and Kaufman RJ. (1997). J. Biol. Chem. 272, 1291-1296. MEDLINE
Yeung MC and Lau AS. (1998). J. Biol. Chem. 273, 25198-25202. MEDLINE
Yeung MC, Liu J and Lau AS. (1996). Proc. Natl. Acad. Sci. USA 93, 12451-12455. MEDLINE
Zamanian-Daryoush M, Der S and Williams BRG. (1999). Oncogene 18, 315-326. MEDLINE
Zhou Y, Gobl A, Wang S, Jacobsen MB, Janson ET, Haines GK III, Radosevich JA and Oberg K. (1998). Eur. J. Cancer 34, 2046-2052. MEDLINE
Figure 1 (a) Characteristic motifs of PKR. (b) Activation and inhibition of PKR by flAlu RNA. Highly purified PKR (2 pmol) was incubated with poly rI:rC (pIC, 60 ng) or with the indicated amounts of flAlu RNA and kinase activity assayed as described (Chu et al., 1998). An autoradiograph of samples analysed by SDS-polyacrylamide gel electrophoresis is shown
Figure 2 Activation of PKR by different stimuli leads to NF-B formation. With the exception of dsRNA, the mechanisms of PKR activation by the other stimuli remain to be uncovered. In PKR null cells or in cell expressing transdominant mutants of PKR, the activation of NF-B in response to these stimuli is deficient. In some cases (e.g. TNF) this is only apparent when the kinetics of activation are examined or when synergistic activities (e.g. with IFN-) are investigated
Figure 3 Different stress stimuli activate p38 MAPK via a PKR-dependent pathway. The activation of p38 by these stimuli is deficient in PKR-null cells. The activation of ATF2 phosphorylation and DNA binding activity by dsRNA is compromised in PKR null cells but PKR dependent phosphorylation of NF-B by a p38-dependent pathway has yet to be described
Figure 4 PDGF-signaling of c-fos is PKR dependent. In PKR null cells, PDGF-induced DNA binding activity and serine 727 phosphorylation of Stat3 is deficient and the induction of c-fos defective. In ras transformed cells an inhibitor is present which blocks the activity of PKR
Figure 5 Activation of PKR by IL3 deprivation. In IL3-dependent cells the withdrawal of IL3 results in the phosphorylation of RAX, a protein activator of PKR and subsequent eIF2 phosphorlyation, inhibition of protein synthesis and apoptosis
Figure 6 Activation of PKR by cellular stress results in transcriptional and translational signals leading to sensitization of cells to apoptotic and proinflammatory stimuli
|1 November 1999, Volume 18, Number 45, Pages 6112-6120|
|Table of contents Previous Article Next [PDF]