Double-stranded (ds) RNA binding and not dimerization correlates with the activation of the dsRNA-dependent protein kinase (PKR).

Upon binding to double-stranded (ds) RNA, the dsRNA-dependent protein kinase (PKR) sequentially undergoes autophosphorylation and activation. Activated PKR may exist as a dimer and phosphorylates the eukaryotic translation initiation factor 2 α subunit (eIF-2α) to inhibit polypeptide chain initiation. Transfection of COS-1 cells with a plasmid cDNA expression vector encoding a marker gene, activates endogenous PKR, and selectively inhibits translation of the marker mRNA, dihydrofolate reductase (DHFR). This system was used to study the dsRNA binding and dimerization requirements for overexpressed PKR mutants and subdomains to affect DHFR translation. DHFR translation was rescued by expression of either an ATP hydrolysis defective mutant PKR K296P, the amino-terminal 1-243 fragment containing two dsRNA binding motifs, or the isolated first RNA binding motif (amino acids 1-123). Mutation of K64E within the dsRNA binding motif 1 destroyed dsRNA binding and the ability to rescue DHFR translation. Immunoprecipitation of T7 epitope-tagged PKR derivatives from cell lysates detected interaction between intact PKR and the amino-terminal 1-243 fragment as well as a 1-243 fragment harboring the K64E mutation. Expression of adenovirus VAI RNA, a potent inhibitor of PKR activity, did not disrupt this interaction. In contrast, intact PKR did not interact with fragments containing the first dsRNA binding motif (1-123), the second dsRNA binding motif (98-243), or the isolated PKR kinase catalytic domain (228-551). These results demonstrate that the translational stimulation mediated by the dominant negative PKR mutant does not require dimerization, but requires the ability to bind dsRNA and indicate these mutants act by competition for binding to activators.

Phosphorylation of translation initiation and elongation factors is a fundamental mechanism that regulates the rate of protein synthesis as cells respond to their external environment (1). The best well characterized mechanism that regulates the rate of polypeptide chain initiation is phosphorylation of the ␣ subunit of the translation initiation factor 2 (eIF-2␣). 1 eIF-2 is a heterotrimer of ␣, ␤, and ␥ subunits that is essential to transfer initiator tRNA (Met-tRNAi) in a ternary complex with GTP to the 40 S ribosomal subunit in the first step of polypeptide chain initiation. Upon 60 S ribosomal subunit joining, GTP is hydrolyzed and the eIF-2⅐GDP complex is released. In order for eIF-2 to promote another round of initiation, the GDP must be exchanged for GTP. This reaction is catalyzed by the guanine nucleotide exchange factor (eIF-2B) (2,3). Control of eIF-2 utilization is mediated by phosphorylation of eIF-2␣ on serine residue 51. Phosphorylated eIF-2 cannot undergo GDP/ GTP exchange and forms a non-dissociable complex between eIF-2B and eIF-2⅐GDP (4,5). Since eIF-2B is present at a lower concentration than eIF-2, eIF-2B becomes sequestered by small increases in eIF-2␣ phosphorylation and prevents further initiation events (6).
Three protein kinases are known to control protein synthesis through eIF-2␣ phosphorylation (7)(8)(9). Most mammalian cells express the dsRNA-dependent protein kinase (PKR) (10). PKR expression is induced by interferon and its activation is dependent upon dsRNA. Many of the anti-viral activities of interferon are mediated by PKR. Interferon-resistant viruses have evolved specific mechanisms to inactivate PKR, such as the RNA polymerase III gene product VAI RNA from adenovirus (10,11). Overexpression of wild-type PKR inhibits protein synthesis and cell growth (12)(13)(14)(15), whereas overexpression of mutant PKR inhibits endogenous PKR activity and causes cell transformation (16,17).
The primary structure of PKR deduced from its cDNA sequence (18) identified the presence of 11 conserved Ser/Thr kinase subdomains (19). The NH 2 -terminal 171 amino acids of the protein comprise two copies of a 67-amino acid RNA-binding motif found in a number of different RNA-binding proteins such as the TAR RNA-binding protein 1, vaccinia virus E3L protein, Drosophila staufen protein, Xenopus proteins Xrlbpa and 4F, and Escherichia coli RNase III (20 -23). The consensus RNA binding sequence contains a positively charged ␣-helical region in the COOH-terminal third of the motif. Mutations of the positively charged residues within this motif destroy RNA binding (24,25).
At present, mutagenesis experiments have identified residues important for RNA binding. Both dsRNA binding domains apparently contribute to the stability of the RNA-protein complex. However, the first domain is more important for binding (21, 22, 24 -26). In addition, mutation of the conserved lysine 64 to glutamic acid (mutant K64E) significantly reduced RNA binding (21). The PKR amino acid requirements for dsRNA binding appear to be the same or overlap the requirements for dsRNA-dependent activation, however, very little is known about how dsRNA binding leads to PKR activation. RNA molecules that activate apparently bind the same sites as RNA molecules that inhibit activation (27,28). PKR activation requires critical concentrations of dsRNA. High concentrations of dsRNA inhibit PKR activation. dsRNA binding to PKR induces * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed: Tel.: 313-763-9037; Fax: 313-763-9323. 1 The abbreviations used are: eIF-2␣, eukaryotic translation initiation factor 2␣; PKR, double-stranded RNA dependent protein kinase; BD, PKR dsRNA binding fragment 1-243; D1, PKR dsRNA binding fragment 1-123; D2, PKR dsRNA binding fragment 98 -243; KD, PKR kinase domain; PAGE, polyacrylamide gel electrophoresis; DHFR, dihydrofolate reductase; eIF-2B, guanine nucleotide exchange factor. autophosphorylation (29,30). PKR is detected in the cytosol as a partially phosphorylated dimer (31). Although autophosphorylation can occur in trans (32,33), it is not known if dimerization is required for dsRNA-dependent activation. Two models proposed to explain the dsRNA dependent activation of PKR differ in their requirement for dimerization. One model proposes that each PKR molecule has one dsRNA binding site and activation results from bridging of two PKR molecules to elicit intermolecular autophosphorylation (32). Another model proposes that activation involves intramolecular autophosphorylation that is dependent upon proper occupancy of two dsRNA binding sites (34). Similarly, two models based on whether PKR dimerizes have been proposed to explain the capability of mutant PKR to dominantly down-regulate wild-type PKR. One model proposes that mutant PKR forms mixed heterodimers with the wild-type PKR and prevents intermolecular phosphorylation and activation (16,35). The second model proposes that overexpression of the mutant PKR sequesters PKR activators such as specific dsRNA sequences (12,36). We have evaluated the validity of these models through analysis of the functional activity and dimerization of PKR mutants and subdomains upon expression in COS-1 cells. Our results support the hypotheses that dsRNA activation does not require dimerization and that mutant PKR acts as a dominant negative through sequestering activators.
Direct studies on the activation of PKR in vivo have been limited due to the inhibition of protein synthesis observed in the presence of active PKR (12)(13)(14)(15)36). Transfection of selective plasmids into COS-1 cells activates PKR to selectively inhibit in a cis-acting manner the translation of reporter mRNAs derived from the plasmid DNA (37)(38)(39)(40). In this study we show that expression of a dominant negative mutant K296P PKR can rescue translation of the reporter mRNA in this system. This system was used to study the PKR structural requirements for PKR dimerization and dsRNA binding to down-regulate endogenous PKR activity. By immunoprecipitation of T7 epitope-tagged proteins transiently expressed in COS-1 cells, we demonstrated that dsRNA-independent dimerization of PKR is mediated through the intact dsRNA binding domain (residues 1-243). In addition, expression of a 1-243 fragment harboring a K64E mutation that is defective in dsRNA binding did not rescue protein synthesis although it dimerized with intact PKR. These results support the hypothesis that the PKR dominant negative phenotype observed by mutant PKR overexpression results from competition for binding to potential dsRNA activators and is not due to formation of inactive heterodimers.

EXPERIMENTAL PROCEDURES
Vector Construction-The expression vectors used in this study contain the same transcription unit as described in Fig. 1. The dihydrofolate reductase (DHFR) expression plasmid pD61 (38), the expression plasmid pETFVA Ϫ (39), and the eIF-2␣ expression vector pETFVA Ϫ -2␣wt (39) were previously described. The K64E mutant PKR was kindly provided by Dr. C. E. Samuel. All other PKR mutants were made by polymerase chain reaction using Vent DNA polymerase (New England Biolabs, Beverly, MA). The primers used for polymerase chain reaction were designed to contain a PstI site at the 5Ј end and a SalI site at the 3Ј end. For the T7-tagged PKR mutants, the 3Ј end primers were designed to contain an overhang encoding amino acids (MASMTG-GQQMG) for T7 antibody binding. The polymerase chain reaction fragments were subcloned into PstI-SalI sites of pETFVA Ϫ vector (Fig. 1). The sequences of the mutants were determined by the dideoxy nucleotide sequencing method (41).
DNA Transfection and Analysis-COS-1 monkey kidney cells were transfected by the DEAE-dextran procedure (42). After 48 h, cells were labeled with Expre 35 S 35 S protein labeling mixture (100 Ci/ml; 1,000 Ci/mmol; DuPont NEN) for 20 min in methionine/cysteine-free minimal essential medium (Life Technologies, Inc., Gaithersburg, MD). Cell extracts were prepared by lysis in Nonidet P-40 lysis buffer (38) and analyzed by SDS-PAGE (43). Proteins were immunoprecipitated using T7-tag monoclonal antibody (Novagen Corp., Madison, WI) or using polyclonal antibody against PKR kindly provided by Dr. C. E. Samuel. Gels were fixed, prepared for fluorography by treatment with En 3 Hance (DuPont NEN), dried, and autoradiographed using Kodak XAR-5 film. The coimmunoprecipitated PKR mutants and fragments were quantitated by PhosphorImage scanning (Molecular Dynamics, Sunnyvale, CA) and band intensity was measured using ImageQuant TM program (Molecular Dynamics, Sunnyvale, CA).
In Vivo Phosphorylation of eIF-2␣-COS-1 cells were co-transfected with pETFVA Ϫ -eIF-2␣ in the presence of different PKR mutants and labeled at 48 h post-transfection with 2 ml of 32 PO 4 (200 Ci/ml, DuPont NEN) for 4 h in phosphate-free medium (Life Technologies, Inc.). Cell extracts were prepared by lysis in Nonidet P-40 lysis buffer. The eIF-2 ␣-subunit was immunoprecipitated with anti-eIF-2␣ monoclonal antibody (kindly provided by Dr. Henshaw), resolved by SDS-PAGE, and electroblotted to nitrocellulose. The membrane was immunoblotted with anti-eIF-2␣ monoclonal antibody and the eIF-2␣ protein level was visualized by the alkaline phosphatase immunoblotting detection system (Sigma). 32 PO 4 incorporation was quantitated by autoradiography of the nitrocellulose membrane using Kodak XAR-5 film and a DuPont Cronex Lightning Plus screen.

FIG. 1. Expression vectors and PKR mutants and fragments used in this study.
The construction of the vectors (A) and PKR fragments (B) are described under "Experimental Procedures." The pETFVA Ϫ expression vector contains the SV40 origin of replication (SV40 ori), the adenovirus major late promoter (AdMLP), the adenovirus tripartite leader (TPL), a small intron (IVS), a polycloning site for insertion of foreign DNA, the encephamolomyocarditis internal ribosomal entry site (EMC), tissue factor cDNA (TF), and the SV40 early polyadenylation signal (SV40 poly(A)). pD61 contains the DHFR coding sequence and p9A contains the adenosine deaminase coding sequence. pETFVA Ϫ differs from pD61 and p9A in the plasmid backbone where pUC18 is in place of pBR322.

RESULTS
dsRNA Binding Is Required for the Dominant Negative Activity of PKR-A series of expression plasmids to direct synthesis of PKR mutants and/or fragments was constructed to evaluate structural requirements to mediate a dominant negative effect on protein synthesis (Fig. 1). The point mutants studied were K64E in the first dsRNA binding motif that is defective in PKR dsRNA binding activity (24) and K296P that is catalytically inactive (16). The dsRNA binding fragments D1 (residues 1-123), D2 (residues 98 -243), and the intact dsRNA binding domain BD (residues 1-243) were constructed in addition to the isolated kinase domain KD (residues 228 -551). The ability of these mutants to inhibit endogenous PKR activity was analyzed utilizing a transfection system that exploits the property of some expression plasmids (such as pD61 encoding DHFR) to produce mRNAs that are selectively inefficiently translated due to PKR activation and eIF-2␣ phosphorylation (38). The unique feature of this system is that inefficient translation occurs in a cis-acting manner with respect to the plasmidderived mRNA.
COS-1 cells were co-transfected with pD61 and the vector pETFVA Ϫ containing wild-type, mutant, or fragments of PKR. Since the mRNA derived from pETFVA Ϫ does not activate PKR and is efficiently translated (39), it is possible to measure the ability of the protein expressed from pETFVA Ϫ to rescue translation of DHFR mRNA derived from pD61. DHFR protein synthesis was quantitated by [ 35 S]methionine/cysteine pulse labeling of cells and analyses of harvested cell extracts by SDS-PAGE. Results were quantitated by PhosphorImage scanning (Table IA). In parallel cells were harvested for quantitation of mRNA by Northern blot hybridization analysis. Transfection of pD61 with pETFVA Ϫ vector alone into COS-1 cells detected a low level of DHFR synthesis ( Fig. 2A, lane 2) that was significantly above the background observed in cells that did not receive pD61 DNA (Fig. 2A, lane 1). Co-transfection of pD61 with wild-type PKR reduced DHFR synthesis to a non-detectable level above background (Fig. 2A, lane 3). Co-transfection of pD61 with PKR mutant K296P increased DHFR synthesis by 7.0-fold compared to co-transfection with the pETFVA Ϫ vector alone ( Fig. 2A, lanes 2 and 5). Whereas wild-type PKR synthesis was not detected, K296P PKR synthesis was observed as a polypeptide migrating at 69 kDa. Co-transfection with the mutant K64E PKR slightly reduced DHFR synthesis by 10% and K64E mutant PKR synthesis was significantly reduced compared to the K296P mutant PKR ( Fig. 2A, lanes 4 and 5). These results indicate that K64E mutant PKR displays intermediate activity between the K296P mutant and wild-type PKR. Cotransfection with KD yielded a low level of KD synthesis, likely resulting from its constitutive activation and eIF-2␣ phosphorylation (Ref. 46; see below). Expression of KD inhibited DHFR synthesis by 30% (Fig. 2B, lane 11). In contrast, expression of either PKR fragments BD or D1 rescued DHFR synthesis ( Fig.  2A, lanes 6 and 8), whereas expression of fragment D2 did not alter DHFR synthesis (Fig. 2A, lane 10). In addition, D1 harboring the K64E mutation did not rescue DHFR synthesis ( Fig.  2A, lanes 7 and 9). Co-transfection with BD harboring the K64E mutation increased DHFR synthesis 1.8-fold, possibly due to residual dsRNA binding of the BD K64E mutant. Northern blot analysis demonstrated that the DHFR mRNA level did not significantly change upon co-transfection with wild-type or mutant PKR (Fig. 2B, lanes 2-11), demonstrating the changes in DHFR synthesis were due to changes in mRNA translational efficiency. The different PKR fragments were detected as polypeptides migrating at the expected sizes (identified in Fig.  2A) and were all expressed at high levels. Northern blot analysis of the PKR mRNA demonstrated the translational efficiency of the mutant PKRs, for example, BD and D1, compared to K64E-BD or K64E-D1, were less affected than the translational efficiencies of DHFR. These results demonstrated that expression of either BD or D1 could rescue DHFR translation and that the K64E mutation destroyed this capability.
Intact Mutant K64E PKR and the KD Phosphorylate eIF-2␣-Since expression of intact mutant K64E PKR and KD inhibited DHFR translation, we evaluated the effect of their expression on eIF-2␣ phosphorylation. The phosphorylation status of eIF-2 ␣-subunit co-expressed with wild-type and mu-  Fig. 2 and the values are expressed relative to DHFR synthesis upon co-transfection with pETFVA Ϫ vector alone. The relative levels of eIF-2␣ phosphorylation were determined from Fig. 3. The amount of dimerization detected with intact K296P or BD were quantitated from the data in Fig. 4B. Asterisk indicates the relative RNA binding affinities previously reported in other studies (25,29,47).
B, the amount of co-immunoprecipitation was quantitated by PhosphorImage scanning. The ratio of the amount of anti-T7 antibody coimmunoprecipitated protein (from Fig. 4B) to the total amount of expressed PKR protein determined by anti-PKR antibody immunoprecipitation (from Fig. 4C) is shown. The amount of the PKR mutant or fragment was corrected for the methionine/cysteine content of the protein.

Co-transfected DNAs Dimerization
K296P/BD-T7 6.6% K296P/K64E-BD-T7 5.1% BD/BD-T7 11.0% BD/K64E-BD-T7 9.4% tant PKR was measured by 32 PO 4 labeling of co-transfected cells. The 32 PO 4 incorporation into eIF-2␣ was measured by immunoprecipitation and analysis by SDS-PAGE and autoradiography. The steady state level of eIF-2␣ was analyzed by Western blot analysis. Comparison of the steady state level of eIF-2␣ protein to the level of eIF-2␣ phosphorylation demonstrated that eIF-2␣ phosphorylation was increased in cells co-transfected with wild-type PKR, the K64E mutant PKR, or the kinase catalytic domain KD (Fig. 3, lanes 3, 4, and 8) compared to cells co-transfected with intact K296P PKR, the intact dsRNA binding domain BD, D1, or D2 (Fig. 3, lanes 5-7 and 9 -11). These data show that the kinase catalytic domain and the K64E mutant PKR phosphorylate eIF-2␣. However, we do not know if this phosphorylation occurs directly or through the endogenous wild-type PKR present in COS-1 cells. Dimerization Is Mediated by the Intact dsRNA Binding Domain-To elucidate whether rescue of DHFR translation by BD and D1 was dependent on the ability for the fragments to dimerize with PKR, a co-immunoprecipitation assay was established to measure dimerization. Expression vectors encoding intact PKR or the dsRNA binding fragment BD (1-243) were co-transfected with vectors encoding PKR fragments engineered to contain a bacteriophage T7-epitope tag in the carboxyl terminus. After labeling transfected cells with [ 35 S]methionine/cysteine, the potential for the different fragments to interact with intact PKR or with BD was measured by coimmunoprecipitation with monoclonal anti-T7 antibody. For these studies, the K296P mutant PKR was used since it was expressed at a higher level than wild-type PKR. Analysis of total protein synthesis by SDS-PAGE indicated that each of the fragments, except for the kinase domain KD, was expressed at a high level (Fig. 4A). In comparison to KD (Fig. 4A, lanes 3 and  10), the K296P mutant KD was expressed at a significantly higher level (Fig. 4A, lanes 4 and 11) indicating that kinase activity from the isolated catalytic domain could inhibit PKR protein synthesis. In addition, co-expression of BD or K296P did not rescue the synthesis of the isolated KD (Fig. 4A, lanes  3 and 10) indicating that either BD or K296P could not inhibit the activity of the KD in trans.
Immunoprecipitation of the cell extracts with anti-T7 antibody specifically adsorbed each T7-tagged fragment (Fig. 4B,  lanes 1-14) that was not detected in the absence of the T7epitope tag (Fig. 4B, lane 15). Immunoprecipitation of T7tagged BD yielded the expected polypeptide at 33 kDa as well  1 and 8), K64E-BD (lanes 2 and 9), KD (lanes 3 and 10), K296P-KD (lanes 4 and 11), D1 (lanes 5 and 12), K64E-D1 (lanes 6 and 13), or D2 (lanes 7 and 14) contained in the same vector. Co-transfection of untagged intact K296P PKR with untagged BD is shown in lane 15. Protein synthesis was analyzed as described under "Experimental Procedures." Panel B, T7-tagged proteins were immunoprecipitated using anti-T7 monoclonal antibody and the immunoprecipitates were analyzed by SDS-PAGE followed by autoradiography. The identity of the different PKR polypeptides is indicated on the right. Panel C, cell extracts were immunoprecipitated with anti-PKR polyclonal antibody and analyzed as described under "Experimental Procedures." as some 69-kDa polypeptide that represented the K296P intact PKR (Fig. 4B, lane 1). Immunoprecipitation of T7-tagged BD also co-immunoprecipitated untagged BD in co-transfected cells (Fig. 4B, lane 8). The K64E mutation contained in BD did not reduce the amount of co-immunoprecipitation detected with K296P PKR or BD (Fig. 4B, lanes 2 and 9). In addition, co-immunoprecipitation of K296P PKR or BD was not detected upon co-expression with D1, D2, or KD (Fig. 4B, lanes 3-7 and  10 -14).
The amount of dimerization was quantitated by Phosphor-Imager scanning of the amount of PKR or BD that was coimmunoprecipitated with the T7-tagged protein upon anti-T7 immunoprecipitation (Fig. 4B) and comparing it to the total amount of untagged BD or K64E BD measured by anti-PKR antibody immunoprecipitation (Fig. 4C). The amount of dimerization detected for BD and intact K296P PKR ranged from 5.1 to 11.0% (Table IB). Similar amounts of dimer were detected for the BD as for the K64E mutant BD. These values were also reproduced in several independent transfection experiments. If the efficiency of heterodimer formation between these species was equivalent to homodimers, then only one-half of the dimers would be detected in this analysis. If one corrects for the total amount of dimer, then 10.2 to 22.0% of the expressed protein would be present as either homodimer or heterodimer. The data suggest that heterodimerization between intact PKR and the RNA binding domain is less efficient homodimerization between the PKR RNA binding domains. At present we do not know the efficiency of homodimerization of intact PKR. The ability of the K64E mutant BD to dimerize is not sufficient to reverse protein synthesis inhibition. The results indicate that PKR dimerization results from protein-protein interactions of the RNA binding domain of PKR. In addition, both dsRNA binding domains were required for dimerization and the dsRNA binding ability of the protein did not affect the dimerization.
Dimerization Occurs Independent of dsRNA Binding-The K64E mutation in BD did not disrupt dimerization and suggested that dsRNA binding is not required for dimerization. To further evaluate the role of dsRNA binding in dimerization, we studied the effect of adenovirus VAI RNA on dimerization and the ability to rescue translation of a reporter mRNA, adenosine deaminase. Adenosine deaminase translation from the vector p9A is inefficient due to PKR activation, similar to DHFR translation from the pD61 vector (37). When p9A was co-transfected with pVASVOD, an expression vector encoding VAI RNA, adenosine deaminase translation was rescued compared to co-transfection with the control plasmid pSVOD (Fig. 5,  lanes 1 and 2). This demonstrates that under these conditions, VAI RNA from pVASVOD could inhibit activation of COS-1 cell endogenous PKR. The effect of VAI RNA on PKR dimerization was studied by co-transfection with intact K296P PKR and the T7 epitope-tagged BD. The synthesis of K296P PKR or BD from the pETFVA Ϫ vector were not affected by the presence of VAI RNA (Fig. 5, lanes 3 and 4). The presence of VAI RNA did not reduce the amount of dimerization detected upon immunoprecipitation with anti-T7 monoclonal antibody (Fig. 5, lanes 5 and  6). These results show that under conditions where VAI RNA inhibits endogenous PKR, there was no effect on dimerization detected between BD and PKR. DISCUSSION The regulation of PKR activation is a crucial control mechanism in protein synthesis initiation. Binding of dsRNA to PKR requires perfect duplexes of 30 base pairs with most efficient binding and activation occurring with duplexes of 85 base pairs (27,28). However, RNA molecules with extensive secondary structure such as VAI RNA that do not contain long stretches of perfect dsRNA duplex do bind efficiently. There are two dsRNA binding motifs in PKR and the first displays a higher affinity interaction than the second (24,25). Mutations within a basic ␣-helical region of the COOH-terminal portion of the motif (for example, K64E) disrupt dsRNA binding (24,25,46,47). Although dsRNA and protein structural requirements for interaction have been intensively studied, very little is known how this interaction results in activation of PKR. Two models for the dsRNA dependent activation of PKR were proposed. The first suggests that dsRNA acts as a bridge to promote dimerization and subsequent trans autophosphorylation and activation (32). This model is based on the second-order kinetics of activation and that one study revealed only a single dsRNA binding site (32). The second model proposes that autophosphorylation is intramolecular and that the difference between activators and inhibitors is their ability to productively interact with the two dsRNA binding motifs to activate autophosphorylation. This is based on the observation that PKR has two sites of different affinity for poly(rI):poly(rC) (34). Both models can account for the inhibition of activation at high concentrations of dsRNA. We have addressed the validity of these models by directly measuring PKR dimerization and functional activation in an in vivo COS-1 cell transfection system.
The requirements for dimerization were studied using mutants and fragments expressed in the presence and absence of a T7-epitope tag to specifically immunoprecipitate one species. The results show that the intact dsRNA binding domain of residues 1-243 could dimerize with itself and with intact K296P mutant PKR. In contrast, BD (residues 1-243) did not interact with fragment D1 (residues 1-123), fragment D2 (residues 98 -243), or the intact isolated PKR kinase catalytic domain (residues 228 -551). In addition, two methods were used to demonstrate dimerization was independent of dsRNA binding. First, a K64E mutant in BD which was previously shown to disrupt the ability to bind dsRNA (24,47) did not affect dimerization with the wild-type BD or K296 mutant intact PKR. Since McMillan et al. (47) observed that the K64E mutant has 5% dsRNA binding activity compared to the wildtype, we cannot rule out that dimerization was dependent on a low affinity dsRNA interaction. In addition, we do not know if the dsRNA-independent dimerization observed with the isolated BD extends to intact PKR. As a second method, the presence of VAI RNA, under conditions known to inhibit endogenous PKR activity, did not inhibit dimerization. These results demonstrate that dimerization can be mediated by interactions between intact dsRNA binding domains and that this does not require high affinity dsRNA interactions.
Overexpression of wild-type PKR inhibits cell growth and inhibits protein synthesis. In addition, translation of PKR itself is down-regulated and mutation of the kinase activity releases the translational repression (48). In contrast, overexpression of catalytically inactive PKR mutants inhibits endogenous PKR activity, stimulates protein translation, and causes cell transformation (16,17). Recent observations from expression of phosphorylation resistant eIF-2␣ suggest this growth promoting activity might result from reduced phosphorylation of eIF-2␣ (49). The ability of mutant PKR to inhibit endogenous PKR activity was proposed to occur by either the ability of PKR to form inactive heterodimers (16,35) or by the ability of mutant PKR to bind and sequester potential activators (12,36). We have provided evidence that the inhibition of endogenous PKR activity by overexpressed PKR mutant or fragments is dependent on a functional dsRNA binding site and not upon the ability to dimerize. This is most strongly supported by two observations. First, expression of the intact dsRNA binding domain 1 (residues 1-123) stimulated translation from a reporter mRNA but did not dimerize with PKR. Mutation of K64E within this domain destroyed the ability to stimulate translation. Second, expression of the intact dsRNA binding domain fragment 1-243 containing dsRNA binding motifs 1 and 2 also stimulated translation and was capable of dimerizing with intact PKR. However, mutation of K64E within motif 1 significantly reduced the ability to stimulate translation without any effect on the ability to dimerize. The results provide good evidence that the dominant negative interference by defective PKR occurs primarily through sequestering dsRNA activators rather than defective heterodimer formation in this COS-1 cell system and is consistent with recent observations by Patel et al. (46). This is in contrast to reports where aminoterminal deletion (50,51) and point mutations (47) that disrupt dsRNA binding can transform NIH 3T3 cells (50,51) and inhibit activation of NFB in response to dsRNA (47). It is possible that PKR in these different experimental systems responds to different activators, such as dsRNA structures or polyanions, or phosphorylates different substrates to mediate the final response measured. For example, a PKR mutant that does not bind dsRNA is activable by heparin (46). Further studies are required to determine if multiple mechanisms are responsible for the dominant interference by mutant PKR molecules.
Our results show that PKR can dimerize, however, dimerization was not sufficient or essential for activation. This was most strongly supported by the observation that VAI RNA inhibited PKR activation but did not affect dimerization. In addition, expression of the isolated PKR kinase domain induced eIF-2␣ phosphorylation and inhibited DHFR translation, although the expressed protein did not dimerize with intact kinase or with the dsRNA binding domain. It is unlikely that the endogenous wild-type COS-1 cell PKR dimerized and/or activated the isolated kinase domain because DHFR translation was also reduced by co-transfection of the intact kinase domain in the presence of either the intact dsRNA binding domain or the K296P mutant PKR. 2 Both of these molecules are known to inhibit endogenous PKR activity and stimulate DHFR translation. We propose that the intact kinase domain is a constitutively active kinase by deletion of the inhibitory dsRNA binding domain. Activation of intact PKR requires productive interactions with dsRNA to relieve the inhibition by the dsRNA binding domain. Although dimerization does not appear to be directly involved in activation of the kinase, it may play an important role in localizing and/or concentrating PKR at functionally important locations in the cell.