INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Cellular protein kinases play crucial roles in propagating, regulating, and coordinating signals necessary for many seminal biological processes, including metabolism, gene expression, cell growth, differentiation, and development. As a result, protein kinases are subjected to elaborate control mechanisms, including association with domains or subunits that inhibit kinase activity by an autoregulatory process (40, 44) or domains that target the kinase to different subcellular localizations and/or substrates (23, 36). In addition, association with activating or inhibitory proteins (21, 86), reversible protein phosphorylation (19, 32), and multimerization (31, 76) also may regulate kinase activity. While dimerization is a common regulatory mechanism for receptor protein kinases, it is less so for cytosolic nonreceptor protein kinases. The latter class of protein kinases, whose dimerization is implicated in their activation and/or function, includes the cGMP- and cAMP-dependent kinases (81), casein kinase 2 (9), Mst1 kinase (17), Raf-1 kinase (22), and the interferon (IFN)-induced, double-stranded (ds)-RNA-activated kinase (PKR) (60). PKR is novel in that it also regulates its own protein synthesis at the translational level (7, 82).

PKR is a pivotal component of the host antiviral defense system because of its translational inhibitory properties (58, 74). Viral replication produces dsRNA that can bind PKR via two dsRNA-binding motifs (DSRMs) located in the N-terminal portion of the kinase, resulting in autophosphorylation and consequently activation of the enzyme. Activated PKR, in turn, phosphorylates the  subunit of eukaryotic initiation factor-2 (eIF-2), leading to a complex series of biochemical events that culminate in a dramatic decrease in the initiation of protein synthesis (15, 59). This disables the use of the translational machinery for the production of viral proteins, and hence restricts viral replication within the cell. Due to its function in antiviral defense, PKR is a target of viral and cellular inhibitors (42, 51). The best-characterized cellular protein inhibitor of PKR is P58^IPK, which is activated upon influenza virus infection (53, 54). P58^IPK appears to be a member of a potential new class of molecular chaperones containing tetratricopeptide repeat motifs and the "J region" of the DnaJ family (52, 62). The non-enzymatic P58^IPK protein inhibits both the auto- and trans-phosphorylation activities of PKR (53, 54). However, the exact mode of P58^IPK action is not fully understood, although it likely involves direct physical interaction with PKR (25, 69).

In addition to its role in interferon-induced antiviral resistance, there is growing evidence that PKR is involved in the control of cell growth and proliferation. Overexpression of PKR in mouse (46), insect (4), and yeast (14) cells results in severe inhibition of growth due to increased eIF-2 phosphorylation. Furthermore, expression of catalytically inactive mutants of PKR elicits fibroblast transformation and tumor formation upon injection of the cells into nude mice, suggesting that PKR has tumor suppressor properties (6, 46, 61). In support of this view, the P58^IPK protein exhibits oncogenic potential; overexpression of the cellular PKR inhibitor causes a transformed phenotype and rapid tumor formation in nude mice (3).

The mechanism(s) by which the functionally defective PKR mutants or wild-type P58^IPK induce malignant transformation is not known. One hypothesis is that the PKR mutants inhibit kinase function by forming inactive heterodimers with endogenous wild-type PKR (6, 46). This also raises a fundamental question concerning the role of dimerization on PKR function. Indeed, evidence for PKR dimerization dependent on the DSRMs has been reported (16, 67), although its role in activation and/or function remains unclear (71, 85). Moreover, the role of P58^IPK in modulating PKR dimer formation has not been investigated.

Recent efforts to identify the region and mechanism responsible for PKR dimerization have led to conflicting results (16, 66, 67). This controversy can be explained, at least in part, by our demonstration herein that PKR can dimerize through a previously unrecognized region independent of the DSRMs. In addition, we show that P58^IPK, but not a nonfunctional form of the protein, prevents PKR dimer formation, suggesting that P58^IPK inhibits PKR by converting PKR dimers into stable monomers. To our knowledge, our findings present the first known example of a cytosolic kinase whose activity may be modulated at the level of dimerization through association with a nonenzymatic cellular protein inhibitor.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial strains, bacteriophage, and media. KH54 and the Escherichia coli strains AG1688 and JH372 (34) were kindly provided by J. C. Hu (Texas A&M University). AG1688, which carries the lacI^q gene for maintaining low expression level of the fusion protein, was used to assess immunity to the tester phage, KH54. KH54 has undergone a deletion of the cI repressor. JH372 strain, which is a derivative of AG1688, harbors the lacZ gene under the control of the P_R promoter and was used in -galactosidase (-Gal) activity assays. The E. coli strain XL-1 Blue (Stratagene) was used in the cloning of plasmids. E. coli strains used in this study were propagated in Luria broth (LB) or agar (73) and stored at 70°C in LB containing 20% (vol/vol) glycerol. All media contained 20 µg of chloramphenicol and/or 50 µg of ampicillin per ml for plasmid selection.

Plasmid constructions. repressor fusions containing various regions of PKR were constructed in the plasmids pC132 and pC168 (55), kindly provided by F. Gigliani (Universita La Sapienza). pC132 (12) carries a fusion between the N-terminal 132 residues of the  cI repressor (N) and the Rop protein. Expression of fusion protein is driven by the pLac promoter. pC168, a derivative of pC132, replicates under the control of the replication origin from p15A and is thus a low-copy-number replicon compatible with plasmids (such as pC132 and pGEX2T) that contain the ColE1 origin. Gene fusions were performed by cloning PCR-amplified PKR DNA fragments into the SalI and BamHI sites of pC132 or pC168. Digestion with SalI and BamHI completely removed the Rop part of the fusion in pC132 and pC168 and created the backbone vectors for the construction of the N fusions. DNA fragments were amplified by PCR with the appropriate primers to introduce the required restriction sites and the stop codon on the DNA fragments. PCR was performed with the Pfu DNA polymerase (Stratagene). Plasmid pcDNAI/NEO (Invitrogen) carrying PKR(K296R) (43) was used as a template for the amplification of the various fragments of PKR. DNA fragments obtained by PCR were purified on agarose gels, recovered with the GeneClean II kit (Bio 101), digested, and cloned into the backbone of pC132 or pC168. When possible, internal fragments were released from the resultant plasmids by using the appropriate restriction enzymes and replaced with the corresponding internal fragments from pcDNAI/NEO-PKR(K296R) to minimize PCR-generated mutations. pKH101 expressing only the N-terminal DNA-binding domain of the repressor (N) was obtained from J. C. Hu.

View larger version (18K): [in this window] [in a new window]   FIG. 1.   repressor-derived dimerization tests. (A) Structure and function of  cI repressor. Wild-type  cI repressor comprises a N-terminal DNA-binding domain (N) and a C-terminal dimerization domain that are tethered by a flexible linker. The cI protein binds to  early promoters as a dimer and represses the transcription of genes required for the phage lytic growth. N itself is unable to dimerize efficiently and thereby is inactive. (B) Rationale for the  repressor fusion system. Fusion of a heterologous protein (PKR in this case) containing a dimerization sequence restores dimer formation and DNA binding; cells become immune to  superinfection and form white colonies on X-Gal medium when carrying a lacZ reporter under the control of N-PKR. (C) Dimer disruption assay. The coexpression of a protein (a GST fusion in this case) capable of interfering with PKR dimer formation renders cells sensitive to  superinfection and turns them blue on X-Gal medium due to derepression of reporter expression.

repressor-derived dimerization tests. Bacterial cells expressing different N fusions or coexpressing N and GST fusions were assessed for immunity to KH54 by cross-streak tests. Phage (~10⁹ PFU) were striped down the center of an agar plate containing appropriate antibiotics and 100 nM isopropyl--D-thiogalactopyranoside (IPTG), and allowed to dry. Cells from liquid cultures grown overnight at 30°C in LB supplemented with appropriate drugs, 10 mM MgSO₄, and 0.2% maltose were streaked perpendicularly across the phage stripe with a toothpick. For dot plaque assay, lawns of E. coli strains containing various plasmids were infected by spotting with 5-µl aliquots of serial dilutions of phage lysates containing 10² to 10⁶ PFU at 10-fold intervals. Infected lawns were incubated overnight at 30°C. -Gal assays were done on agar plates layered with 3 ml of top agar containing the appropriate antibiotic(s), 20 µM 5-bromo-4-chloro-3-indolyl--D-galactopyranoside (X-Gal), 100 nM IPTG, and 4 mM phenylethyl--D-thiogalactoside. Colony color determination was made after streaking overnight cultures and overnight incubation at 30°C. Plasmids pBR322 and pACYC184 (New England Biolabs) were used as empty vector controls.

Yeast two-hybrid system. Yeast strain Hf7c (Clontech Laboratories), which carries the HIS3 reporter fused to a GAL1 promoter sequence, was used to assay for protein-protein interactions. The two-hybrid plasmids pGBT10 and pGAD425 were used for construction of GAL4 DNA-binding domain (GBD) and GAL4 transcriptional activation domain (GAD) fusions, respectively. Hf7c was cotransformed with pGBT10-PKR(244-551) and different pGAD425 plasmid constructs as indicated. Transformation was performed by the lithium acetate method (75). Transformants were plated on solid synthetic defined (SD) medium lacking tryptophan (Trp) and leucine (Leu) for selection of double transformants and determination of transformation efficiency. To select for histidine (His)-positive colonies, transformants grown on SD plates lacking Trp and Leu for 3 days at 30°C were streaked onto SD plates lacking Trp, Leu, and His and allowed to grow for 3 to 5 days to deplete endogenous His stores. Colonies were then streaked onto His-containing plates in the presence of 1 mM 3-amino-triazole (3-AT) and incubated for 3 to 5 days at 30°C, after which the plates were scored for growth.

PKR functional assay in yeast. To determine the effect of overexpression of P58^IPK and PKR fragments upon PKR function in vivo, we used a yeast growth suppression assay (26, 71). P58^IPK or PKR(244-296) was coexpressed from the pYES2 plasmid in Saccharomyces cerevisiae strain RY1-1, which carried two copies of human PKR integrated into the LEU2 locus, under the control of the galactose-inducible GAL1-CYC1 promoter (71). RY1-1 exhibits a growth arrest phenotype when grown on minimal medium containing galactose as the sole carbon source due to hyperphosphorylation of yeast eIF-2 by PKR. Thus, the ability of P58^IPK or PKR(244-296) to interfere with PKR function was assessed by reversion of the PKR-mediated growth suppression effect and analysis of the eIF-2 phosphorylation state in the appropriate RY1-1 strain harboring the corresponding expression plasmid. Reversal of yeast growth arrest and in vivo eIF-2 phosphorylation status was determined as previously described (26).

In vitro transcription and translation. Transcripts encoding full-length and mutant PKR were generated essentially as described previously (43) with the T7 promoter of pCDNA1/NEO. Linear transcripts encoding full-length PKR(K296R), PKR(1-242), and PKR(244-551) were prepared from pCDNA1/NEO-PKR(K296R) as described previously (25, 43). Transcripts encoding PKR(296-551) were generated by linearizing pCDNA1/NEO-PKR(296-551) with BamHI digestion. Transcript integrity was monitored by agarose gel electrophoresis. For in vitro translation, 2-µg portions of each in vitro transcription product were used to program a rabbit reticulocyte lysate system (Promega) in the presence of [³⁵S]methionine as described previously (43). To verify that the translation products were of the expected sizes, an aliquot of each was analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and fluorography. Translation product abundance was quantitated by scintillation counting of the trichloroacetic acid-precipitable material from each translation reaction.

Far-Western analysis. For the detection of in vitro self-interaction of PKR, increasing amounts of purified bacterially expressed GST and GST-PKR(244-296) fusion proteins were size-fractionated by SDS-PAGE on a 14% resolving gel. Proteins were transferred to nitrocellulose and renatured by overnight incubation in BLOTTO G (50 mM Tris-HCl [pH 7.5], 1% nonfat milk powder, 50 mM NaCl, 1 mM EDTA), 1 mM dithiothreitol, and 10% glycerol) at room temperature. Filters were then rinsed twice in Hyb75 buffer (20 mM HEPES, 1% nonfat milk powder, 75 mM KCl, 2.5 mM MgCl₂, 1 mM dithiothreitol, 0.05% Nonidet P-40) and subsequently overlaid with approximately 5.0 × 10⁵ cpm of [³⁵S]methionine-labeled in vitro-translated product from PKR amino acids (aa) 1 to 242, 244 to 551, or 296 to 551 (in fresh Hyb75 buffer) overnight at 4°C. Production, purification, and quantitation of GST fusions and in vitro translation products were carried out as described previously (25). Filters were washed three times with Hyb75 buffer and then subjected to autoradiography.

Immunoblot analysis. One milliliter of the same overnight culture (3 ml) used for the  repressor fusion tests was centrifuged, and the pellet was resuspended in 2× SDS sample buffer. Protein samples were subjected to SDS-PAGE on a 12% gel (equal amounts of protein estimated by Coomassie blue staining were loaded into each lane) and were then electroblotted onto nitrocellulose membranes. Immunoblot analyses were performed with the indicated antibodies and enhanced chemiluminescence (Amersham) as described previously (4).

DNA sequencing and sequence analysis. All DNA constructs were verified to be the correct reading frame by restriction mapping and nucleotide sequence analysis. Sequencing reactions were carried out by using the Applied Biosystems dye terminator system (University of Washington). DNA sequences were analyzed by DNA strider and the Wisconsin GCG package (Madison, Wis.). Homology searches were carried out using the BLAST (1) service at the National Center for Biotechnology Information (NCBI). Synthetic oligonucleotides were purchased from Life Technologies.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

The  repressor fusion system detects a novel region involved in PKR dimerization. We used the  repressor fusion assay (34) to determine the region(s) responsible for mediating PKR dimerization because it is a relatively simple system in which to study dimer interactions (33). A schematic of the  repressor fusion system is shown in Fig. 1. We first determined whether full-length PKR dimerized in this assay by scoring for bacterial immunity to  superinfection. Since wild-type PKR is partly toxic to bacteria (5), we used a catalytically inactive PKR protein, PKR(K296R) (43). This mutant protein is well characterized and retains the ability to dimerize (16, 66, 67) and interact with known PKR partners (10, 25, 26). As revealed by dot plaque assays (Fig. 2A), bacteria expressing a fusion between the  repressor N-terminal DNA-binding domain (N) and PKR(K296R), designated N-PKR(K296R), were clearly less susceptible to  superinfection than were bacteria expressing N alone. This indicates successful reconstitution of N DNA-binding activity promoted by PKR dimerization in E. coli. As an additional test, the plasmid was also introduced into a strain (JH372) that harbors the lacZ reporter gene under the control of the P_R promoter (34). In this scenario, dimerization is assayed by -Gal activity, whereby functionally reconstituted N fusions repress expression of lacZ. As shown in Fig. 2B, bacteria expressing a N fusion of PKR(K296R) or the Rop protein control (12) formed white colonies on indicator plates containing the chromogenic lactose analog X-Gal. On the contrary, bacterial cells that expressed only N or an empty vector control (pBR322) developed blue colonies.

View larger version (62K): [in this window] [in a new window]   View larger version (64K): [in this window] [in a new window]   View larger version (52K): [in this window] [in a new window]   View larger version (56K): [in this window] [in a new window]   FIG. 2.   Dimerization properties of N-PKR fusion proteins. (A) Dot plaque assay. Freshly poured lawns of E. coli (strain AG1688) expressing either N alone or N-PKR(K296R) fusion were infected by spotting with 5-µl aliquots of 10-fold dilutions of lysates of KH54. The titer and pattern of the aliquots are indicated. Sensitivity to  superinfection was scored by evaluating lysis plaques. (B) Repression of -Gal expression. Repressor activity of the indicated N fusions was assessed by expression from the PR lacZ reporter in strain JH372. A functional N fusion inhibits lacZ expression, resulting in white colony color. In contrast, cells lacking a functional repressor fusion form blue colonies due to -Gal production. (C) Cross-streak tests. Phages were striped down the center of an agar plate, and cells harboring the plasmids expressing the indicated N fusions were streaked perpendicularly across the phage stripe. Sensitive cells were scored by assessing lysis by the phage (i.e., the streak disappeared or was noticeably thinner after crossing the phage zone). (D) Expression of N chimeric proteins containing different PKR mutants. Crude cell extracts were subjected to SDS-PAGE on a 14% gel and visualized by immunoblot analysis by using a polyclonal antibody raised against PKR (4). Detection of N-PKR(244-296) was unsuccessful by this analysis. The positions of fusion proteins are indicated by arrows. The molecular size standards are shown in kilodaltons.

View larger version (13K): [in this window] [in a new window]   FIG. 3.   PKR dimerization activity maps two independent regions. Schematic representation of domain structures of wild-type (WT) PKR and PKR deletion constructs used in the  repressor fusion system. DSRM1 and DSRM2 denote the positions of the dsRNA-binding motifs 1 and 2, respectively. The protein kinase catalytic domain begins at residue 264 and contains the conserved kinase homology subdomains labeled I to XI (2). A summary of the dimerization activity of PKR variants is shown on the right. The solid bars at the top indicate the positions of the dimerization regions of PKR. The positions of terminal amino acids are also indicated.

Residues 244 to 296 mediate dsRNA-independent dimerization of PKR in the yeast two-hybrid system and in far-Western analysis. To validate that aa 244 to 296 were indeed involved in PKR-PKR interaction, we used the yeast two-hybrid system to detect protein-protein interactions as a result of their ability to reconstitute the trans-activating function of the GAL4 protein (24). Since PKR dimerization dependent on the DSRMs (aa 1 to 167) has been demonstrated by use of the two-hybrid assay (16, 67), we focused instead on the examination of PKR DSRM-independent interactions. A hybrid protein consisting of PKR aa 244 to 551 and the GAL4 DNA-binding domain was therefore constructed. The ability of this hybrid protein to associate with various regions of PKR fused to the GAL4 transcriptional activation domain was examined. All fusion constructs were expressed to detectable levels in the yeast cells as detected by Western blot analysis (25). As illustrated in Fig. 4, only hybrid proteins containing aa 244 to 296 interacted with PKR(244-551), as indicated by growth on His medium. Conversely, all cotransfectants grew on medium supplemented with His, indicating that the lack of growth on His medium is not due to toxicity caused by the overexpression of the relevant hybrid proteins in yeast.

View larger version (45K): [in this window] [in a new window]   FIG. 4.   PKR self-associates independently of the DSRMs in the yeast two-hybrid system. Yeast strain Hf7c containing pGBT10/PKR(244-251) was transformed with the indicated pGAD425 plasmids containing various PKR mutants or plasmid alone. Cotransformants were replica plated on His+ medium (left) and His medium (right) containing 1 mM 3-AT. Interaction was scored by growth on His medium. Western blot analysis indicated that all GAL4 fusions were efficiently expressed in yeast (not shown).

View larger version (46K): [in this window] [in a new window]   FIG. 5.   Direct PKR-PKR interaction via aa 244 to 296. Increasing amounts (lanes 1 and 6, 10 ng; lanes 2 and 7, 20 ng; lanes 3 and 8, 30 ng; lanes 4 and 9, 40 ng; lanes 5 and 10, 50 ng) of purified bacterially expressed GST and GST-PKR(244-296) were subjected to SDS-PAGE and transferred to nitrocellulose. The filter was overlaid with radiolabeled PKR aa 1-to-242 (top), 244-to-551 (middle), or 296-to-551 probe (bottom) and subjected to autoradiography. Arrows indicate positions of GST fusion protein and GST alone.

Coexpression of an amino acid fragment (aa 244 to 296) inhibits both PKR dimerization and kinase function. To further demonstrate that aa 244 to 296 were sufficient to mediate PKR dimerization, we asked whether coexpression of GST-PKR(244-296) could prevent dimerization of N-PKR(K296R) in the  repressor system via the formation of heterodimers (see Fig. 1C). If this were the case, the bacteria would turn blue on X-Gal-treated plates and become sensitive to phage superinfection due to the loss of functional N-PKR(K296R) dimers (and hence repressor activity). Bacteria producing both N-PKR(K296R) with GST-PKR(244-296), but not with GST alone, formed blue colonies on X-Gal-treated plates and were sensitive to phage superinfection (Table 1). No repressor activity was observed in bacteria coexpressing an empty vector control (pACYC184) with GST-PKR(244-296). Furthermore, GST-PKR(244-296) did not disrupt dimerization of the Rop protein control in the  repressor system.

View this table: [in this window] [in a new window]   TABLE 1.   Summary of  resistance and -Gal expression properties of JH372 or AG1688 clones coexpressing N-PKR(K296R) and GST fusionsa

View larger version (29K): [in this window] [in a new window]   View larger version (29K): [in this window] [in a new window]   View larger version (44K): [in this window] [in a new window]   FIG. 6.   PKR aa 244-to-296 fragment alone prevents full-length PKR dimerization and inhibits kinase activity. Expression of N-PKR(K296R) with the low-copy-number plasmid pC168 was sufficient to inhibit lacZ expression in strain JH372, resulting in white colony color (see Table 1). Coexpression from a high-copy-number plasmid of a second protein or protein domain [GST-PKR(244-296)] capable of disrupting the association of N-PKR(K296R) molecules resulted in derepression of lacZ and blue colonies.  immunity was determined by cross-streak assays as described for Fig. 2B. Bacteria coexpressing an inhibitor of PKR dimer formation are sensitive to  infection in this case. (A) Coexpression of aa 244 to 296 reduced the amount of PKR dimer formation in the yeast two-hybrid system. Yeast strain SFY526 cotransformed with constructs expressing PKR(K296R) fused to GBD and PKR(K296R) fused to GAD or empty vectors (pGBT10 and pGAD425) was mated with yeast strain Y187 transformed with the expression constructs pYX222-PKR(244-296) or the negative control vector pYX222. Diploids were selected on a minimal selective medium lacking Trp, Leu, and His. -Gal assays of yeast cell extracts cotransformed with the indicated plasmids were performed as described in Materials and Methods. The results shown represent the mean activity obtained from two experiments. (B) Expression of GAL4-PKR fusion constructs in yeast. An immunoblot analysis of extracts prepared from the indicated yeast transformants is shown. The panel represents the same blot probed sequentially with antibodies to PKR or actin as indicated. Not shown is PKR(244-296), which we were unable to detect by this analysis. The positions of GBD-PKR(K296R) and GAD-PKR(K296R) fusion proteins are indicated by arrows. (C) PKR aa 244-to-296 reverse wild-type kinase-mediated growth suppression in yeast. Yeast strain RY1-1 was transformed with the 2µm yeast expression constructs pYES2-PKR(244-296) or pEMBLYex-K3L (positive control) or the negative control vector pYES2. Transformants were spotted at various dilutions, as indicated in the middle of the panel, on uracil-deficient minimal synthetic medium containing 2% dextrose (SD) or 10% galactose-2% raffinose (SGAL) as the carbon source and scored for growth. Kinase function inhibition was scored by growth on the SGAL plate. Western blot analysis indicated that PKR was efficiently expressed in all yeast strains (not shown).

P58^IPK inhibits PKR activity through the disruption of PKR dimerization. Earlier studies had demonstrated that PKR aa 244 to 296 comprised the binding site for kinase inhibitors, including P58^IPK, the cellular inhibitor recruited by influenza virus to inactivate the protein kinase (25, 26). Given that dimerization may be essential for PKR activity and that the aa 244-to-296 region overlaps with the site of P58^IPK interaction, we next explored the possibility that P58^IPK adversely affected PKR dimerization. To test this possibility, we examined whether coexpression of a GST-P58^IPK fusion could prevent or reduce dimerization of N-PKR(K296R) in the  repressor fusion system. As summarized in Table 2, transformants coexpressing GST-P58^IPK but not those expressing GST alone turned blue on X-Gal plates and became sensitive to  superinfection, indicating that P58^IPK inhibits PKR dimerization. The specificity of GST-P58^IPK for disruption of PKR dimerization was demonstrated by the fact that GST-P58^IPK did not diminish dimerization of the Rop protein control in the  repressor system. Furthermore, coexpression of the GST fusion containing other known PKR inhibitors that bind to other regions of the kinase, including the vaccinia virus K3L protein (25) and the influenza virus NS1 protein (78), had little or no effect on the dimeric state of PKR (Table 2). P58^IPK did not abolish dimerization of a PKR fragment containing residues 1 to 167 (data not shown), suggesting that P58^IPK acts specifically at residues 244 to 296 and that the two regions can dimerize independently.

View this table: [in this window] [in a new window]   TABLE 2.   Summary of  resistance and -Gal expression properties of JH372 or AG1688 clones coexpressing N-PKR(K29R) and various GST fusionsa

View larger version (49K): [in this window] [in a new window]   FIG. 7.   P58IPK inhibits PKR dimer formation in the yeast two-hybrid system. (Top) -Gal activities of triply transformed yeast cells. The appropriate GAL4 two-hybrid tester strains were transformed with the indicated plasmids, and -Gal assay of the yeast cell extracts was performed as described for Fig. 6B. The results shown here represent the mean activity derived from two experiments. (Bottom) Expression of GAL4-PKR fusions in yeast cells. An immunoblot analysis of extracts prepared from the indicated yeast transformants is shown. The panel represents the same blot probed sequentially with antibodies to PKR or actin as indicated.

View larger version (35K): [in this window] [in a new window]   FIG. 8.   P58IPK prevents PKR dimerization in the  repressor fusion system. (A) Coexpression levels of GST fusion proteins. Crude cell extracts from bacteria coexpressing N-PKR(K296R) and various GST fusions were subjected to SDS-PAGE on a 14% gel and visualized by immunoblot analysis with a polyclonal antibody directed against GST. Positions of molecular size standards are indicated in kilodaltons. (B) Expression of N-PKR chimeric protein. The panel represents the same blot probed with a polyclonal antibody to PKR. The arrow indicates the position of N-PKR(K296R). (See also Table 2.)

View larger version (40K): [in this window] [in a new window]   FIG. 9.   A P58IPK mutant incapable of disrupting PKR dimerization does not inhibit PKR function. (A) Yeast growth analysis. Yeast strain RY1-1 was transformed with the 2µm expression plasmid pYES2 (vector control), pYES2-P58IPK, pYES2-P58IPK6, or pEMBLYex4-K3L (positive control) and streaked onto SD (not shown) or SGAL medium. Growth on the SGAL-treated plate indicated inhibition of PKR function in yeast cells. All transformants grew on SD medium (not shown). (B) eIF-2 phosphorylation analysis. Extracts from RY1-1 yeast cells harboring pYES2-P58IPK (lane 1), pYES2-P58IPK6 (lane 2), pYES2 (lane 3), or pEMBLYex4-K3L (control; lane 4) were separated by isoelectric focusing and subjected to immunoblot analysis with an antiserum to yeast eIF-2. Arrows indicate the positions of yeast eIF-2 phosphorylated on basal sites only (lower band) and yeast eIF-2 phosphorylated on Ser-51, the site of phosphorylation by PKR (upper band). (C) Protein expression. Strains were grown for 5 h in galactose-containing liquid medium, and extracts prepared as described previously (26). Proteins (25 µg) were separated by SDS-PAGE and subjected to immunoblot analysis. The panels represent the same blot probed sequentially with antibodies to PKR (top), P58IPK (middle), or actin (bottom).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Dimerization of PKR and its role in kinase regulation. PKR was originally inferred to function as a dimer because activation of PKR displayed second-order kinetics (47). This notion was later supported by numerous biochemical analyses demonstrating that PKR exists in solution predominantly as a dimer (11, 49). Furthermore, the dominant negative phenotype of nonfunctional PKR mutants suggests that they form inactive heterodimers with wild-type PKR (6, 46, 61). It is thought that dimerization may mediate activation of PKR through trans-phosphorylation between two PKR monomers brought into close proximity by binding to a single molecule of dsRNA (47). That the wild-type enzyme is able to phosphorylate a catalytically inactive PKR in trans is consistent with this view (83, 84). Moreover, Cosentino et al. (16) demonstrated that PKR dimerizes via the DSRMs, and that dimerization in vitro is dependent on the presence of dsRNA.

Model for PKR dimerization. Although dsRNA binding has been suggested to mediate the dimerization of PKR DSRMs, evidence for a dsRNA-independent mechanism also has been reported (66, 85). However, it will be difficult to decipher the exact mechanism of dimerization because both the dsRNA-binding and the dimerization properties of the DSRMs are closely embedded in the same regions (68). In any case, both the DSRMs and the aa 244-to-296 region may be required for the formation of a stable PKR dimer complex. Proteins that contain more than one dimerization region have been described, including the transcription factor AP-4 (35), c-Myc (18), the thyroid hormone receptor (63), the vitamin D receptor (64), and the early B-cell factor (29). However, it is beyond the scope of this study to compare the relative affinities of each of the two independent dimerization domains of PKR. Interestingly, we found that P58^IPK, which binds PKR at aa 244 to 296, effectively inhibited dimerization of full-length PKR but not a PKR fragment containing aa 1 to 167 (Fig. 7 and 8; data not shown).

View larger version (39K): [in this window] [in a new window]   FIG. 10.   PKR dimerization. (Top) Proposed model for the modulation of PKR dimerization. The catalytic domain (C) and the regulatory domain (R) containing DSRMs are shown. (I) dsRNA binding to the DSRMs bridges PKR molecules and induces a conformational change that unmasks the dimerization site with residues 244 to 296 (darkened region) to promote intermolecular association of PKR. (II) Binding of P58IPK to the aa 244-to-296 region leads to monomerization of PKR and presumably to inactivation of the kinase. (III) Deletion of the DSRM sequences results in a conformational alteration exposing this dimerization region, and thus the isolated catalytic domain is still capable of forming dimers via aa 244 to 296. (Bottom) Limited homology within the noncatalytic portion of the aa 244-to-296 dimerization region of PKR with human (h) and rabbit (r) glycogen-associated subunits of PP1 (PP1-G). Identical residues are shown by the shaded area; dotted lines indicate the less-conserved residues. The numbers refer to the positions of the amino acids listed. Homology was identified by using the NCBI BLAST program (1).

Does P58^IPK represent a new class of PKR inhibitors? Until the present study, regulation of kinase dimerization by a cellular protein inhibitor had not been reported. The PIN protein inhibitor of the neurotransmitter nitric oxide synthase apparently inhibits the enzyme by directly interfering with its dimeric state (37). The results reported here demonstrate that P58^IPK can specifically prevent PKR dimerization, providing insights into the mechanism by which P58^IPK inhibits PKR activity. In addition, since P58^IPK is an established inhibitor of PKR, the finding also lends support to the functional importance of dimerization for kinase activity. Consistent with the view that the aa 244-to-296 region mediates dimerization, PKR inhibitors that act elsewhere on the molecule, including the pseudosubstrate protein K3L (25) and the dsRNA-binding protein NS1 (78), did not perturb the dimeric state of PKR (Table 2). The biological significance of aa 244 to 296 of PKR is further exemplified by the fact that at least one other PKR inhibitor, the hepatitis C virus nonstructural protein 5A (NS5A), also associates with PKR in this region (26). Therefore, P58^IPK and NS5A may constitute a new group of PKR effectors, i.e., those that act via disruption of PKR dimerization.