Mechanism of PKR Activation: Dimerization and Kinase Activation in the Absence of Double-stranded RNA

doi:10.1016/j.jmb.2004.10.031

J. Mol. Biol. (2005) 345, 81–90

Mechanism of PKR Activation: Dimerization and Kinase Activation in the Absence of Double-stranded RNA Peter A. Lemaire1, Jeffrey Lary2 and James L. Cole1,2* 1

Department of Molecular and Cell Biology, University of Connecticut, Storrs, CT 06269 USA 2 National Analytical Ultracentrifugation Facility University of Connecticut Storrs, CT 06269, USA

The kinase PKR is a central component of the interferon antiviral pathway. PKR is activated upon binding double-stranded (ds) RNA to undergo autophosphorylation. Although PKR is known to dimerize, the relationship between dimerization and activation remains unclear. Here, we directly characterize dimerization of PKR in free solution using analytical ultracentrifugation and correlate self-association with autophosphorylation activity. Latent, unphosphorylated PKR exists predominantly as a monomer at protein concentrations below 2 mg/ml. A monomer sedimentation coefficient of s020;w Z 3:58 S and a frictional ratio of f/f0Z1.62 indicate an asymmetric shape. Sedimentation equilibrium measurements indicate that PKR undergoes a weak, reversible monomer–dimer equilibrium with KdZ450 mM. This dimerization reaction serves to initiate a previously unrecognized dsRNA-independent autophosphorylation reaction. The resulting activated enzyme is phosphorylated on the two critical threonine residues present in the activation loop and is competent to phosphorylate the physiological substrate, eIF2a. Dimer stability is enhanced by w500-fold upon autophosphorylation. We propose a chain reaction model for PKR dsRNA-independent activation where dimerization of latent enzyme followed by intermolecular phosphorylation serves as the initiation step. Subsequent propagation steps likely involve phosphorylation of latent PKR monomers by activated enzyme within high-affinity heterodimers. Our results support a model whereby dsRNA functions by bringing PKR monomers into close proximity in a manner that is analogous to the dimerization of free PKR. q 2004 Elsevier Ltd. All rights reserved.

*Corresponding author

Keywords: PKR; protein kinase; autophosphorylation; analytical ultracentrifugation; sedimentation equilibrium

Introduction Protein kinase R (PKR) is a double-stranded (ds) RNA-activated protein kinase that is induced by interferon.1 PKR plays a central role in the cellular antiviral response2 and functions as a signal transducer in a variety of cellular processes.3 PKR is synthesized in a latent state, but upon binding dsRNA, it undergoes autophosphorylation at multiple serine and threonine residues resulting in activation. The best characterized cellular substrate of PKR is the eukaryotic initiation factor eIF2a. Abbreviations used: dsRBD, double-stranded RNAbinding domain; PPase, protein phosphatase; dsRBM, double-stranded RNA-binding motif; PKR, protein kinase R; TCEP, Tris(2-carboxyethyl)phosphine. E-mail address of the corresponding author: [email protected]

Phosphorylation of eIF2a inhibits the initiation of translation, such that production of dsRNA during viral infection leads to PKR activation and inhibition of viral protein synthesis. PKR is comprised of an N-terminal doublestranded RNA-binding domain (dsRBD) and a C-terminal kinase domain. The dsRBD contains two tandem copies of the w70 amino acid residue dsRNA binding motif (dsRBM).4 In the NMR structure of the PKR dsRBD, each of the RNAbinding motifs adopts the typical abbba-fold.5 The catalytic domain contains conserved kinase subdomains and is most closely related to two kinases that also phosphorylate eIF2a: vertebrate hemeregulated eIF2a kinase and yeast GCN2.6 A large body of in vitro and in vivo data imply that dimerization of PKR is associated with activation, but the relationships between dimerization of free PKR, dimerization that may occur upon binding to

0022-2836/$ - see front matter q 2004 Elsevier Ltd. All rights reserved.

82 RNA and activation remain unclear. Initial evidence for dimerization came from studies indicating that the activation rate displays a second-order dependence on PKR concentration.7 The dimerization model also accounts for the observation that low concentrations of dsRNA activate PKR but higher concentrations are inhibitory; 8–10 presumably, dilution of PKR monomers onto separate molecules of dsRNA by a high concentration of activator prevents productive dimerization. Several studies have demonstrated RNA-independent dimerization of PKR11–16 or the isolated dsRBD12,17–19 using a variety of assays. Dimerization is believed to be mediated by interactions involving the dsRBD12,15,18,20 and a second dimerization motif lying between residues 244 and 296.19 In contrast, analytical ultracentrifugation 21 and NMR22 measurements indicate that the dsRBD exists as a monomer, even at high protein concentrations. RNA binding is not absolutely required for enzymatic activation. For example, polyanionic molecules such as heparin induce autophosphorylation.23 An endogenous protein activator of PKR, PACT, binds to PKR and activates the enzyme in the absence of RNA.24 Fusion of a heterologous dimerization domain with the PKR catalytic domain enhances autophosphorylation and eIF2a kinase function in vivo,25 suggesting that dimerization can mediate activation in the absence of RNA binding. In order to clarify the relationship between PKR dimerization and enzymatic activation, we have characterized self-association of full-length PKR in the unphosphorylated and phosphorylated states using analytical ultracentrifugation and correlate

Mechanism of PKR Activation

these data with functional assays. We observe that latent PKR undergoes a weak, reversible monomer– dimer equilibrium and dimerization affinity is enhanced by phosphorylation. Although autophosphorylation is dependent on dsRNA at low protein concentrations, a novel RNA-independent mode of activation is detected above 0.5 mM. Our data support a model where latent PKR exists predominantly as a monomer and dsRNA or other exogenous activators function by inducing PKR dimerization.

Results Activation and autophosphorylation of wild-type PKR As previously observed,26 PKR expresses very poorly in Escherichia coli BL21 (DE3) cells and cannot be detected in Coomassie-stained SDS/ polyacrylamide gels of whole lysates following induction (results not shown). This low expression level is due to codon bias, and both wild-type and a catalytically inactive K296R PKR mutant express to high levels using a host containing a plasmid encoding genes for tRNAs that are rare in E. coli (Figure 1A). As previously reported,26–29 wild-type PKR expressed in E. coli is phosphorylated in vivo as indicated by a reduction in mobility relative to K296R (Figure 1A) and staining for phosphoprotein (data not shown). However, co-expression with l PPase27,30 in pPET-PKR/PPase results in production of unphosphorylated wild-type PKR that co-migrates with the K296R mutant (Figure 1A, lane

Figure 1. SDS-PAGE of the induction (A) and in vitro phosphorylation (B) of PKR. A, wild-type (WT) and K296R mutant (K296R) PKR were expressed in BL21(DE3) Rosetta cells. Cultures were induced with 1 mM IPTG and analyzed on Coomassie blue-stained gels. l PPase: co-expression with phage l protein phosphatase. B, In vitro phosphorylation of PKR analyzed by Coomassie blue staining (left) and a stain specific for phosphoproteins (Pro-Q Diamond, Molecular Probes) (right). Unphosphorylated WT or K296R PKR were incubated at 3 mM for 20 minutes at 30 8C; where indicated, reactions contained 10 mg/ml of dsRNA (poly(rI:rC)) and 100 mM ATP.

Mechanism of PKR Activation

83

Figure 2. dsRNA-dependent autophosphorylation of PKR: 100 nM unphosphorylated wild-type PKR was incubated with various concentrations of dsRNA (poly(rI:rC)) for 20 minutes at 30 8C. Reactions were initiated with 0.1 mM ATP containing 1 mCi of [g-32P]ATP and were quenched with SDS loading buffer after 20 minutes.

4). Both unphosphorylated wild-type PKR and the K296R mutant are efficiently purified by a threecolumn protocol yielding 10–20 mg/liter of culture (Figure 1B). The molecular mass of 62,118 Da for K296R PKR obtained by matrix-assisted laser desorption/ionization (MALDI) mass spectroscopy agrees closely with the predicted value of 62,122 Da. For wild-type PKR, the observed mass of 61,980 Da is slightly less than predicted (62,094 Da), confirming that the enzyme is unphosphorylated. The lower mass is consistent with processing of the N-terminal methionine. As expected from the electrophoretic mobility and MALDI data, purified wild-type PKR does not stain for phosphoprotein (Figure 1B, lane 5). In vitro autophosphorylation of PKR was initially characterized by the decrease in electrophoretic mobility and by phosphoprotein staining. Wildtype but not K296R PKR undergoes in vitro autophosphorylation upon addition of ATP and dsRNA (Figure 1B). Surprisingly, addition of ATP results in autophosphorylation of wild-type PKR in the absence of dsRNA (Figure 1B, lane 6). Autophosphorylation of PKR has previously only been observed in the presence of dsRNA1 or other exogenous activators.23,24 However, the experiments depicted in Figure 1B were performed at a relatively high protein concentration (3 mM). At a 30-fold lower protein concentration (100 nM), autophosphorylation is not detected in the absence of dsRNA using a sensitive radioactive gel-based assay (Figure 2). In this case, autophosphorylation is strongly enhanced upon addition of poly(rI:rC) and higher concentrations of dsRNA above 10 mg/ ml are inhibitory, in agreement with previous observations.8–10 Thus, the recombinant PKR used in the present study exhibits the typical pattern of dsRNA-induced activation when assayed at low protein concentration. The dsRNA-independent autophosphorylation reaction was characterized in more detail using a quantitative radioactive filter-binding assay. Figure 3 demonstrates that the reaction progress curve is somewhat non-linear, with evidence of a lag phase at the early time-points. The reaction

Figure 3. Kinetics of dsRNA-independent autophosphorylation. PKR was incubated at 4 mM for 20 minutes at 25 8C and autophosphorylation was initiated with 3 mM ATP containing 4 mCi of [g-32P]ATP. Aliquots were withdrawn at the indicated time-points and were quenched with 100 ml of stop buffer. Phosphorylated products were quantified using the nitrocellulose filter-binding assay and scintillation counting.

approaches a plateau at long reaction times corresponding to incorporation of 15–20 phosphate groups/PKR. Previous studies have identified 14 PKR phosphorylation sites,31–34 indicating that essentially complete PKR autophosphorylation occurs in the absence of dsRNA. Figure 4A shows the dependence of the autophosphorylation rate on PKR concentration in the absence of dsRNA. The activity is extremely low up to a threshold protein concentration of w0.5 mM and then increases, approximately linearly, at higher concentrations. The same dependence of autophosphorylation activity on PKR concentration is observed in the presence of 0.1–1 mg/ml of acetylated BSA, indicating that the absence of activity at low enzyme concentrations is not an artifact associated with adsorption. The dsRNAindependent activation is highly specific: autophosphorylation of 100 nM wild-type PKR is not induced by addition of 0.1–1 mg/ml of BSA, PEG 20,000 or the dsRBD of PKR (data not shown). PKR autophosphorylation at higher protein concentrations is not abolished by pre-incubation of enzyme with RNase A or RNase III, indicating that this reaction is not associated with trace contaminants of single-stranded or doublestranded RNA. This activity was also not affected by passage of the enzyme through a Q-Sepharose column under conditions where phosphorylated PKR binds readily16,29 or by treatment with protein phosphatase 1, which is known to readily dephosphorylate PKR.16,28 Although the large number of PKR phosphorylation sites and the complexity of the linked dimerization and phosphorylation reactions preclude rigorous analysis of the autophosphorylation

84

Figure 4. Protein concentration dependence of dsRNAindependent autophosphorylation. A, dsRNA-independent autophosphorylation of wild-type PKR. B, Activation of wild-type PKR autophosphorylation by 100 nM pre-phosphorylated enzyme. C, Phosphorylation of K296R PKR by 100 nM pre-phosphorylated enzyme. Reactions were initiated with 3 mM ATP containing 4 mCi of [g-32P]ATP. Phosphorylated products were quantified using the nitrocellulose filter-binding assay. Inset: comparison of reaction rates at low protein concentrations. Wild-type (B); wild-typeCpre-phosphorylated enzyme (,); K296RCpre-phosphorylated enzyme ($).

kinetics, qualitative analysis provides useful insights. A lag phase is characteristic of autocatalytic processes and has been observed for the autophosphorylation of human src kinase.35 An intramolecular reaction mechanism is expected to give a simple linear dependence of reaction rates on enzyme concentration. The non-linear dependence of the rates (Figure 4A) is not consistent with

Mechanism of PKR Activation

an intramolecular process and supports an intermolecular activation mechanism. Higher rates and a more linear concentration dependence are obtained by “seeding” the unphosphorylated wild-type PKR with a low concentration (100 nM) of enzyme that has been pre-activated by dsRNA-independent autophosphorylation with unlabeled ATP (Figure 4B). The pre-activated enzyme serves to bypass a slow initiation step in the autoactivation process. Figure 4C shows an analogous phosphorylation experiment where 100 nM pre-activated wild-type enzyme is mixed with various concentrations of K296R PKR. In this case, the wild-type enzyme is fully phosphorylated and the K296R protein is inactive, so that all 32P incorporation must occur via an intermolecular mechanism and the K296R PKR functions purely as a substrate. Indeed, the reaction exhibits a typical hyperbolic concentration dependence and the data fit well to the Michaelis–Menten expression with KmZ3.1(G0.6) mM and VmaxZ25.0(G1.7) pmol 32P minK1. These data support the intermolecular autophosphorylation model. Previously, intermolecular autophosphorylation has been described in the context of dsRNA-dependent PKR activation.20,36,37 The inset to Figure 4C compares the autophosphorylation rates for all three reactions at low protein concentrations. Virtually no activity is detected for the wild-type enzyme alone. The reaction rates for wild-type and K296R PKR seeded with 100 nM pre-activated enzyme are essentially the same up to 0.2 mM, but at higher protein concentrations the reaction rates for the wild-type protein surpass that of K296R. These data indicate that at low concentrations of wild-type enzyme all of the phosphate incorporation is catalyzed by the 100 nM pre-activated PKR, but at higher concentrations newly activated wild-type enzyme contributes to the activity. In the context of an isolated kinase domain of PKR, it has recently been reported that PKR autophosphorylation does not lead to an activated enzyme that is competent to phosphorylate the physiological substrate, eIF2a.38 This defect in activation has been correlated with a lack of phosphorylation at Thr446 and Thr451 within the kinase activation loop. We have analyzed the phosphorylation state of these residues by probing PKR autophosphorylated in the absence of dsRNA with phosphopeptide-specific antibodies directed against Thr446 and Thr451. A Western blot in Figure 5A shows that these antibodies do not react with unphosphorylated enzyme but react strongly with PKR autophosphorylated by incubation at high enzyme concentration without dsRNA. A control antibody directed against the kinase domain of PKR reacts with both enzyme forms. Figure 5B shows that PKR autophosphorylated with unlabeled ATP in the absence of dsRNA is competent to phosphorylate eIF2a with [g-32P]ATP. However, the control, unphosphorylated PKR requires dsRNA to be activated for autophosphorylation as well as

Mechanism of PKR Activation

85

Figure 5. dsRNA-independent autophosphorylation of PKR generates fully activated enzyme. Wild-type PKR was autophosphorylated in vitro in the absence of dsRNA. A, PKR autophosphorylation in the kinase activation loop: 0.4 mg of unphosphorylated or phosphorylated enzyme was analyzed by Western blotting using phosphopeptide specific antibodies directed against Thr446 and Thr451 and a control secondary antibody directed against the kinase domain of PKR. B, Autophosphorylated PKR is a competent eIF2a kinase. Each enzyme form (100 nM) was incubated for 20 minutes at 30 8C G0.2 mg of eIF2a and 10 mg/ml of dsRNA (poly(rI:rC)). Phosphorylation reactions were initiated with 0.1 mM ATP containing 4 mCi of [g-32P]ATP and were quenched with SDS loading buffer after 20 minutes.

phosphorylation of eIF2a. Thus, in contrast to the autophosphorylation reactions catalyzed by the isolated kinase domain, autophosphorylation of intact PKR in the absence of dsRNA produces an enzyme that is phosphorylated at the critical threonine residues within the activation loop and is an active eIF2a kinase. Association state of PKR In preliminary sedimentation velocity studies of K296R PKR, a predominant feature was observed near sZ3.3 S in c(s) distributions, along with features at higher s corresponding to small amounts (10–15%) of higher oligomers (data not shown). Gelfiltration immediately prior to sedimentation removes the higher S species (Figure 6A), indicating that they represent irreversible aggregates. Using the Svedberg equation and fitted values of s and D, the mass of the sZ3.3 S species is 61.7 kDa, which closely corresponds to a mass of 62.1 kDa predicted for the monomer. Figure 6B shows that the sedimentation coefficient increases slightly with protein concentration from w3.3 to w3.4 over a range of 0.1–1.6 mg/ml. The slight increase in s with concentration indicates that K296R PKR undergoes rapidly reversible, low-affinity self-association. It was not convenient to characterize this weak selfassociation by sedimentation velocity because of the marked contribution of hydrodynamic non-ideality at the high protein concentrations required to substantially populate the oligomeric species. Thus, we employed sedimentation equilibrium measurements (see below). Sedimentation coefficients were extrapolated to zero protein concentration and buffer-corrected to calculate a sedimentation coefficient of s020;w Z 3:50 S and a frictional ratio of f/foZ1.58. Similar sedimentation velocity results were obtained with dephosphorylated wild-type PKR, with s020;w Z 3:41 S and f/foZ 1.62 (data not shown). Thus, the PKR monomer has a more extended shape than typical globular

Figure 6. Sedimentation velocity of K296R PKR. A, Continuous sedimentation coefficient distribution (c(s)) analysis of K296R PKR at 0.7 mg/ml (11.3 mM). B, Dependence of the sedimentation coefficient on protein concentration. Conditions: rotor speed, 50,000 rpm; temperature, 20 8C; data collection, interference optics at one minute intervals. K296R PKR was purified on Superdex 200 immediately before sedimentation.

86

Mechanism of PKR Activation

Table 1. Analysis of PKR self-association by sedimentation equilibrium Protein K296R K296R Wild-type, unphosphorylated Wild-type, phosphorylatedd

Model

ln Ka

Kb

RMSc

Monomer–dimer Monomer–trimer Monomer–dimer Monomer–dimer

7.713 [7.444, 7.972] 16.14 7.532 [6.885, 8.004] 13.87 [13.54, 14.22]

2.24 [1.71,2.90]!103 MK1 1.02!107 MK2 1.87 [0.98,2.99]!103 MK1 1.06 [0.76,1.50]!106 MK1

0.0128 0.0142 0.0112 0.0048d

Unless otherwise indicated, experiments were performed using interference optics as described in Materials and Methods. a Natural logarithm of the equilibrium binding constant. The values in square brackets represent the 95% joint confidence intervals. b Equilibrium binding constant. c Root-mean-square deviation of the fit in units of fringe displacement or absorbance. d Absorption optics at 230 nm.

proteins. Previously, an elongated shape for PKR was suggested based on neutron-scattering.14 Self-association of PKR was quantitatively defined by sedimentation equilibrium analysis. The stoichiometry and affinity of PKR K296R self-association were determined by global analysis of ten data channels. A good fit was obtained for a monomer–dimer model as indicated by a low RMS deviation (Table 1) and a lack of systematic deviations in the residuals (Figure 7). A monomer–trimer model gave a significantly worse fit to the data. When the stoichiometry was treated as an adjustable parameter, a best fit value of NZ2.38 was obtained, confirming that the monomer– dimer model is the best description of the data. The best-fit value of ln KZ 7:713 corresponds to KdZ446 mM, indicating that dimerization of K296R is quite weak. The narrow confidence intervals indicate that the equilibrium constant is well

defined in this analysis. Addition of 5 mM Mg2C did not significantly alter Kd. Addition of 2 mM of the non-hydrolysable ATP analogue AMP-PNP in the presence of 5 mM Mg2C decreased Kd slightly to 250 mM. Limited solubility of PKR precludes more extensive study of the effects of ionic strength, pH and temperature on the dimerization constant. A dimerization Kd of 535 mM was obtained for unphosphorylated wild-type PKR (Table 1), indicating that self-association of wild-type and K296R PKR are similar. In vitro autophosphorylation enhances dimerization of wild-type PKR. In sedimentation velocity experiments performed at 0.1–0.6 mg/ml, a new species is detected near sZ5.2 S corresponding to a PKR dimer (data not shown). A corrected sedimentation coefficient of s020;w of 5.17 S is obtained for the dimer, corresponding to f/fo of 1.71. Thus, the dimer also has an extended shape. Sedimentation

Figure 7. Sedimentation equilibrium of K296R PKR. Conditions: sample concentrations, 0.3, 0.6, 1.0, 1.5 and 2.0 mg/ml; rotor speeds, 18,000 rpm (B) and 24,000 rpm (,); temperature, 20 8C. The entire dataset was globally fit to monomer–dimer model. The lines show the best-fit model and the inset shows the residuals. The vertical offsets are arbitrary.

Mechanism of PKR Activation

equilibrium data fit well to a monomer–dimer model with Kd of 0.95 mM (Table 1). Thus, autophosphorylation results in a very large, w500-fold enhancement of PKR dimerization affinity.

Discussion Despite extensive investigation, the critical macromolecular interactions that modulate activation of PKR are not well understood. Here, we have characterized the energetics of PKR dimerization in free solution using analytical ultracentrifugation and correlate self-association with enzymatic activation. Our observation that latent PKR exists in a weak, reversible monomer–dimer equilibrium that is enhanced by autophosphorylation rationalizes some previous observations regarding the oligomerization and phosphorylation state of PKR. Gelfiltration measurements14,16 indicated that PKR exists as a monomer; however, a dimeric form was observed at a higher concentration of 1.7 mg/ml in neutron-scattering experiments.14 Presumably, the lower concentrations employed in gel-filtration results in dimer dissociation. Consistent with the higher dimerization affinity of phosphorylated PKR, it was previously reported that partially phosphorylated enzyme eluted as a dimer on gelfiltration.16 There is precedent for enhancement of dimerization of soluble kinases by phosphorylation. Phosphorylation of the MAP kinase EKR2 results in w3000-fold increase in dimerization affinity.39 Dimerization is believed to be mediated by interactions involving the dsRBD12,15,18,20 and possibly a second motif lying between residues 244 and 296 within the kinase domain.19 Under conditions comparable to those employed here for full-length PKR, we have not detected dimerization of the dsRBD21 nor heterodimerization of fulllength PKR with the dsRBD (J. W. Ucci & J.L.C., unpublished results). NMR22 measurements indicate that the dsRBD exists as a monomer at high protein concentrations. These results suggest that PKR dimerization must involve regions outside of the dsRBD. Although self-association of the unactivated PKR is weak, this dimerization reaction has mechanistic significance and serves to initiate a previously unrecognized dsRNA-independent PKR autophosphorylation reaction. Control experiments indicate that this reaction is not due to trace contaminants of dsRNA or pre-activated enzyme. The resulting enzyme is phosphorylated on the two critical threonine residues present in the activation loop and is functional as an eIF2a kinase. Thus, the dsRNA-independent pathway appears to generate an activated state of PKR that corresponds to the form generated upon activation by dsRNA. In contrast to dsRNA-dependent autophosphorylation, which readily occurs at 100 nM PKR, dsRNA-independent activation is essentially absent below 0.5 mM. The non-linear dependence of the

87 rates is not consistent with an intramolecular process and supports an intermolecular activation mechanism. In support of this intermolecular activation model, the K296R mutant functions as a Michaelis–Menten substrate for the activated enzyme. Both intramolecular and intermolecular autophosphorylation events have been observed in other kinases, and in some cases both mechanisms have been reported for the same kinase.35 Previously, intermolecular autophosphorylation has been described in the context of dsRNA-dependent PKR activation.20,36,37 Based on the analytical ultracentrifugation results presented here, and previous evidence of PKR dimerization, we propose that formation of dimers of latent PKR serves to initiate intermolecular autophosphorylation. Activation is specific for PKR and is not induced by addition of high concentrations of heterologous proteins or macromolecular crowding agents (PEG 20,000). It may appear that the threshold concentration for the dsRNA-independent PKR activation of w0.5 mM is inconsistent with a much higher dimerization Kd for the latent enzyme of w450 mM. However, it is important to recognize that the extent of intermolecular autophosphorylation does not reflect an equilibrium process but represents the endpoint of a series of autocatalytic reactions. Dimerization of latent enzyme followed by intermolecular phosphorylation serves as the initiation step. Subsequent chain propagation steps likely involve phosphorylation of latent PKR monomers by newly activated enzyme, possibly via formation of mixed heterodimers of phosphorylated and latent enzyme. Thus, initial formation of each dimer of latent PKR may eventually result in the generation of many molecules of phosphorylated enzyme via a branched chain reaction. This chain reaction would be further enhanced by an increase in heterodimer stability relative to the latent enzyme. The Kd for heterodimer formation is likely to fall between the Kd for the latent enzyme (450 mM) and the Kd for the activated homodimer (0.95 mM). An estimate for this dissociation constant is provided by the Km of 3.1 mM using the catalytically inactive K296R PKR as substrate. This chain reaction model may also be relevant to activation of PKR by dsRNA. It is known that phosphorylation of PKR greatly reduces dsRNA affinity.28 Thus, the dimerization and phosphorylation of PKR induced by dsRNA may be amplified following dissociation from dsRNA by autophosphorylation reactions analogous to those observed here in the absence of dsRNA. It is noteworthy that introduction of a single molecule of dsRNA per cell is capable of stimulating interferon synthesis,40 suggesting the requirement for a high-level amplification process. Although it is well known that interferon stimulation results in a fivefold to tenfold induction of PKR,41 absolute intracellular enzyme concentrations have not been reported. It appears likely that the typical PKR concentrations are far below

88 the Kd reported here, such that only a small fraction of latent enzyme associates to dimer under physiological conditions. However, it is not apparent whether typical PKR concentrations are also below a threshold concentration for dsRNA-independent activation. Although it is generally believed that PKR requires dsRNA or other exogenous activators to induce in vivo autophosphorylation, there is some evidence that PKR undergoes activation in vivo in the absence of dsRNA intermediates produced during viral infection. Phosphorylated PKR is recovered from unstimulated as well as interferon-treated cells.16,41 PKR over-expressed in E. coli26–29 or yeast34 is recovered in a phosphorylated state; presumably the high intracellular concentration of PKR following induction results in dimerization and activation in vivo without the requirement for dsRNA. Other studies have demonstrated activation of PKR in the absence of dsRNA binding. Binding of PKR to heparin induces autophosphorylation.23 Fusion of a heterologous dimerization domain with the PKR kinase domain enhances autophosphorylation and eIF2a kinase function in vivo.25 All of these PKR activation modes may share a common mechanism involving multimerization. In this model, binding to dsRNA or heparin serves to bring multiple PKR monomers into close proximity in a manner that is analogous to dimerization of free enzyme at higher protein concentrations. It has been suggested that PKR activation by dsRNA involves a conformational change that releases an inhibitory interactions with the dsRBD42 and NMR measurements indicate a specific interaction of the second dsRBM with the kinase domain.43 The ability of dimerization to mediate PKR activation in the absence of dsRNA suggests activation does not require dsRNAmediated release of the second dsRBM from an interaction with the kinase domain.

Materials and Methods The wild-type PKR gene was copied from the plasmid pET-PKR44 by PCR to introduce an NdeI restriction site immediately before the start codon and a BamHI restriction site after the stop codon. The PCR product was cloned into pET11a (Novagen) to obtain pPET-PKR. The K296/R-mutation was introduced to create pPETK296R, using the Quik-Change kit (Stratagene). In vivo dephosphorylated PKR was obtained by co-expression with l protein phosphatase (PPase).27,30 The l PPase gene was excised from the vector pGST-PKR/PPase using BamHI and BlpI, and was inserted downstream of the PKR gene in pPET-PKR to create the vector pPET-PKR/ PPase. Expression vectors were transformed into E. coli BL21(DE3) or BL21(DE3) Rosetta cells (Novagen). Cells were grown in LB medium at 37 8C until A600 nm w0.7 and protein expression was induced with 1 mM IPTG for three hours at w20 8C. For PKR purification, cells were resuspended in buffer A (20 mM Hepes (pH 7.5), 50 mM NaCl, 0.1 mM EDTA, 10 mM b-mercaptoethanol, 10% glycerol) supplemented with protease inhibitor cocktail (Sigma). Cells were lysed

Mechanism of PKR Activation

by incubation with 300 mg/ml of lysozyme for 30 minutes followed by sonication for three minutes. The lysate was centrifuged for 20 minutes at 20,000 g. The supernatant was applied to a heparin Sepharose column (Amersham) equilibrated in buffer A and PKR was eluted using a NaCl gradient. The peak fractions were diluted twofold with buffer A and applied to a poly(rI:rC) agarose column (Amersham) equilibrated in buffer A. PKR was eluted at 1.1 M NaCl, concentrated to w10 mg/ml and stored at K80 8C. Prior to each experiment, PKR aggregates were removed by gel-filtration on either Sephacryl S-200 or Superdex 200 (Amersham). Autophosphorylated wildtype PKR was prepared by incubating enzyme at 2 mg/ml with 3 mM ATP in phosphorylation buffer (20 mM Hepes (pH 7.5), 50 mM KCl, 5 mM MgCl2, 0.1 mM EDTA, 1 mM DTT) at 30 8C for two hours followed by purification of the dimer fraction on Superdex 200 (Amersham). Phosphorylation within the PKR activation loop was detected by Western blot analysis using antibodies specific for the phosphorylated forms of Thr446 and Thr451 (Cell Signaling Technology Inc.) Qualitative phosphorylation assays were performed using established methods45 and were analyzed by SDSPAGE and phosphorimaging. eIF2a was obtained from Biomol and E. coli RNase III was obtained from Ambion. Quantitative autophosphorylation measurements were performed using a filter-binding assay.46 For both assays, PKR was exchanged into phosphorylation buffer using Biogel P6 spin columns (Biorad). Reactions were performed at 25 8C for six minutes and quenched with 100 ml of stop buffer (0.2% SDS, 100 mM EDTA, pH 8.0). Then 100 ml of the mixture was spotted onto nitrocellulose filters (Schleicher and Schuell). Filters were washed for 15 minutes in 20 mM Tris, 200 mM NaCl and 20 mM sodium pyrophosphate (pH 8.0) and counted by liquid scintillation. Analytical ultracentrifugation was performed with a Beckman-Coulter XL-I instrument at 20 8C. Samples were equilibrated into analysis buffer (20 mM Hepes, 200 mM NaCl, 0.1 mM EDTA, 0.1 mM TCEP) using spin columns. The buffer density was measured to be 1.00864 g/ml using an Anton Paar DMA-60 density meter. Protein concentrations were assayed by measuring absorbance at 280 nm. The following properties of PKR were calculated (20 8C) using SEDNTERP:47 3280Z4.33!104 MK1 cmK1, MZ62,094 Da, nZ  0:732 g=ml. Standard two-channel and external-loading, six-channel centerpieces were used for sedimentation velocity and equilibrium experiments, respectively. Initially, sedimentation velocity data were fit to finite element solutions of the Lamm equation using a model of a large number of discrete non-interacting species with the program SEDFIT.48 Sedimentation coefficients and diffusion constants were obtained using a discrete species model.49 For sedimentation equilibrium, the approach to equilibrium was monitored using WinMatch (D. Yphantis & J.L., Storrs, CT). Sedimentation equilibrium data were fit using non-linear least-squares fitting algorithms with macros developed in IGOR Pro (Wavemetrics), as described.50

Acknowledgements We thank Peter Beal (University of Utah) for the pET-PKR vector and Tadashi Mastui (Kanazawa Medical University, Uchinada, Japan) for the

89

Mechanism of PKR Activation

pGST-PKR/PPase vector. This work was supported by grant number AI-53615 from the NIH to J.L.C. 17.

References 1. Clemens, M. J. (1997). PKR–a protein kinase regulated by double-stranded RNA. Int. J. Biochem. Cell. Biol. 29, 945–949. 2. Stark, G. R., Kerr, I. M., Williams, B. R., Silverman, R. H. & Schreiber, R. D. (1998). How cells respond to interferons. Annu. Rev. Biochem. 67, 227–264. 3. Williams, B. R. (2001). Signal integration via PKR. Sci. STKE, 89, RE2. 4. Fierro-Monti, I. & Mathews, M. B. (2000). Proteins binding to duplexed RNA: one motif, multiple functions. Trends Biochem. Sci. 25, 241–246. 5. Nanduri, S., Carpick, B. W., Yang, Y., Williams, B. R. & Qin, J. (1998). Structure of the double-stranded RNA binding domain of the protein kinase PKR reveals the molecular basis of its dsRNA-mediated activation. EMBO J. 17, 5458–5465. 6. Hanks, S. K. & Hunter, T. (1995). Protein kinases 6. The eukaryotic protein kinase superfamily: kinase (catalytic) domain structure and classification. FASEB J. 9, 576–596. 7. Kostura, M. & Mathews, M. B. (1989). Purification and activation of the double-stranded RNA-dependent eIF-2 kinase DAI. Mol. Cell. Biol. 9, 1576–1586. 8. Hovanessian, A. G. & Kerr, I. M. (1979). The (2 0 -5 0 ) oligoadenylate (pppA2 0 -5 0 A2 0 -5 0 A) synthetase and protein kinase(s) from interferon-treated cells. Eur. J. Biochem. 93, 515–526. 9. Hunter, T., Hunt, T., Jackson, R. J. & Robertson, H. D. (1975). The characteristics of inhibition of protein synthesis by double-stranded ribonucleic acid in reticulocyte lysates. J. Biol. Chem. 250, 409–417. 10. Kimchi, A., Zilberstein, A., Schmidt, A., Shulman, L. & Revel, M. (1979). The interferon-induced protein kinase PK-i from mouse L cells. J. Biol. Chem. 254, 9846–9853. 11. Ortega, L. G., McCotter, M. D., Henry, G. L., McCormack, S. J., Thomis, D. C. & Samuel, C. E. (1996). Mechanism of interferon action. Biochemical and genetic evidence for the intermolecular association of the RNA-dependent protein kinase PKR from human cells. Virology, 215, 31–39. 12. Patel, R. C., Stanton, P., McMillan, N. M., Williams, B. R. & Sen, G. C. (1995). The interferon-inducible double-stranded RNA-activated protein kinase selfassociates in vitro and in vivo. Proc. Natl Acad. Sci. USA, 92, 8283–8287. 13. Wu, S. & Kaufman, R. J. (1997). A model for the double-stranded RNA (dsRNA)-dependent dimerization and activation of the dsRNA-activated protein kinase PKR. J. Biol. Chem. 272, 1291–1296. 14. Carpick, B. W., Graziano, V., Schneider, D., Maitra, R. K., Lee, X. & Williams, B. R. G. (1997). Characterization of the solution complex between the interferon-induced double-stranded RNA-activated protein kinase and HIV-I trans-activating region RNA. J. Biol. Chem. 272, 9510–9516. 15. Cosentino, G. P., Venkatesan, S., Serluca, F. C., Green, S. R., Mathews, M. B. & Sonenberg, N. (1995). Doublestranded-RNA-dependent protein kinase and TAR RNA-binding protein form homo- and hetero-dimers in vivo. Proc. Natl Acad. Sci. USA, 92, 9445–9449. 16. Langland, J. O. & Jacobs, B. L. (1992). Cytosolic

18.

19.

20.

21. 22.

23. 24. 25.

26.

27.

28. 29.

30. 31.

32.

double-stranded RNA-dependent protein kinase is likely a dimer of partially phosphorylated MrZ66,000 subunits. J. Biol. Chem. 267, 10729–10736. Rende-Fournier, R., Ortega, L. G., George, C. X. & Samuel, C. E. (1997). Interaction of the human protein kinase PKR with the mouse PKR homolog occurs via the N-terminal region of PKR and does not inactivate autophosphorylation activity of mouse PKR. Virology, 238, 410–423. Tian, B. & Mathews, M. B. (2001). Functional characterization of and cooperation between doublestranded RNA binding motifs of the protein kinase PKR. J. Biol. Chem. 276, 9936–9944. Tan, S. L., Gale, M. J., Jr & Katze, M. G. (1998). Doublestranded RNA-independent dimerization of interferon-induced protein kinase PKR and inhibition of dimerization by the cellular P58IPK inhibitor. Mol. Cell. Biol. 18, 2431–2443. Wu, S. & Kaufman, R. J. (1996). Double-stranded (ds) RNA binding and not dimerization correlates with the activation of the dsRNA-dependent protein kinase (PKR). J. Biol. Chem. 271, 1756–1763. Ucci, J. W. & Cole, J. L. (2004). Global analysis of nonspecific protein–nucleic interactions by sedimentation equilibrium. Biophys. Chem. 108, 127–140. Nanduri, S., Carpick, B., Yang, Y., Williams, B. R. G. & Qin, J. (1998). 1H, 13C, 15N resonance assignments of the 20 kDa double stranded RNA binding domain of PKR. J. Biomol. NMR, 12, 349–351. Hovanessian, A. G. & Galabru, J. (1987). The doublestranded RNA-dependent protein kinase is also activated by heparin. Eur. J. Biochem. 167, 467–473. Patel, R. C. & Sen, G. C. (1998). PACT, a protein activator of the interferon-induced protein kinase PKR. EMBO J. 17, 4379–4390. Vattem, K. M., Staschke, K. A. & Wek, R. C. (2001). Mechanism of activation of the double-strandedRNA-dependent protein kinase, PKR: role of dimerization and cellular localization in the stimulation of PKR phosphorylation of eukaryotic initiation factor-2 (eIF2). Eur. J. Biochem. 268, 3674–3684. Barber, G. N., Tomita, J., Hovanessian, A. R., Meurs, E. & Katze, M. G. (1991). Functional expression and characterization of the interferon-induced doublestranded RNA activated P68 protein kinase from Escherichia coli. Biochemistry, 30, 10356–10361. Matsui, T., Tanihara, K. & Date, T. (2001). Expression of unphosphorylated form of human doublestranded RNA-activated protein kinase in Escherichia coli. Biochem. Biophys. Res. Commun. 284, 798–807. Jammi, N. V. & Beal, P. A. (2001). Phosphorylation of the RNA-dependent protein kinase regulates its RNA-binding activity. Nucl. Acids Res. 29, 3020–3029. Xu, Z., Wang, D., Lee, X. & Williams, B. R. (2004). Biochemical analyses of multiple fractions of PKR purified from Escherichia coli. J. Interferon Cytokine Res. 24, 522–535. Conn, G. L. (2003). Expression of active RNA-activated protein kinase (PKR) in bacteria. Biotechniques, 35, 682–684. 686. Zhang, X., Herring, C. J., Romano, P. R., Szczepanowska, J., Brzeska, H., Hinnebusch, A. G. & Qin, J. (1998). Identification of phosphorylation sites in proteins separated by polyacrylamide gel electrophoresis. Anal. Chem. 70, 2050–2059. Taylor, D. R., Lee, S. B., Romano, P. R., Marshak, D. R., Hinnebusch, A. G., Esteban, M. & Mathews, M. B.

90

33.

34.

35.

36.

37.

38.

39.

40.

Mechanism of PKR Activation

(1996). Autophosphorylation sites participate in the activation of the double-stranded-RNA-activated protein kinase PKR. Mol. Cell. Biol. 16, 6295–6302. Taylor, D. R., Tian, B., Romano, P. R., Hinnebusch, A. G., Lai, M. M. & Mathews, M. B. (2001). Hepatitis C virus envelope protein E2 does not inhibit PKR by simple competition with autophosphorylation sites in the RNA-binding domain. J. Virol. 75, 1265–1273. Romano, P. R., Garcia-Barrio, M. T., Zhang, X., Wang, Q., Taylor, D. R., Zhang, F. et al. (1998). Autophosphorylation in the activation loop is required for full kinase activity in vivo of human and yeast eukaryotic initiation factor 2a kinases PKR and GCN2. Mol. Cell. Biol. 18, 2282–2297. Barker, S. C., Kassel, D. B., Weigl, D., Huang, X., Luther, M. A. & Knight, W. B. (1995). Characterization of pp60c-src tyrosine kinase activities using a continuous assay: autoactivation of the enzyme is an intermolecular autophosphorylation process. Biochemistry, 34, 14843–14851. Thomis, D. C. & Samuel, C. E. (1993). Mechanism of interferon action: evidence for intermolecular autophosphorylation and autoactivation of the interferoninduced, RNA-dependent protein kinase PKR. J. Virol. 67, 7695–7700. Thomis, D. C. & Samuel, C. E. (1995). Mechanism of interferon action: characterization of the intermolecular autophosphorylation of PKR, the interferon-inducible, RNA-dependent protein kinase. J. Virol. 69, 5195–5198. Wu, S. & Kaufman, R. J. (2004). trans-Autophosphorylation by the isolated kinase domain is not sufficient for dimerization or activation of the dsRNA-activated protein kinase PKR. Biochemistry, 43, 11027–11034. Khokhlatchev, A. V., Canagarajah, B., Wilsbacher, J., Robinson, M., Atkinson, M., Goldsmith, E. & Cobb, M. H. (1998). Phosphorylation of the MAP kinase ERK2 promotes its homodimerization and nuclear translocation. Cell, 93, 605–615. Marcus, P. I. & Sekellick, M. J. (1977). Defective interfering particles with covalently linked [G]RNA induce interferon. Nature, 266, 815–819.

41. Lasky, S. R., Jacobs, B. L. & Samuel, C. E. (1982). Mechanism of interferon action. Characterization of sites of phosphorylation in the interferon-induced phosphoprotein p1 from mouse fibroblasts: evidence for two forms of p1. J. Biol. Chem. 257, 11087–11093. 42. Robertson, H. D. & Mathews, M. B. (1996). The regulation of the protein kinase PKR by RNA. Biochimie, 78, 909–914. 43. Nanduri, S., Rahman, F., Williams, B. R. G. & Qin, J. (2000). A dynamically tuned double-stranded RNA binding mechanism for the activation of antiviral kinase PKR. EMBO J. 19, 5567–5574. 44. Vuyisich, M. & Beal, P. (2000). Regulation of the RNAdependent protein kinase by triple helix formation. Nucl. Acids Res. 28, 2369–2374. 45. Gunnery, S. & Mathews, M. B. (1998). RNA binding and modulation of PKR activity. Methods, 15, 189–198. 46. Surette, M. G., Levit, M., Liu, Y., Lukat, G., Ninfa, E. G., Ninfa, A. & Stock, J. B. (1995). Dimerization is required for the activation of the protein histidine kinase cheA that mediates signal transduction in bacterial chemotaxis. J. Biol. Chem. 271, 939–945. 47. Laue, T. M., Shah, B. D., Ridgeway, T. M. & Pelletier, S. L. (1992). Computer-aided interpretation of analytical sedimentation data for proteins. In Analytical Ultracentrifugation in Biochemistry and Polymer Science (Harding, S., Rowe, A. & Horton, J., eds), pp. 90–125, Royal Society of Chemistry, Cambridge. 48. Schuck, P. (2000). Size-distribution analysis of macromolecules by sedimentation velocity ultracentrifugation and Lamm equation modeling. Biophys. J. 78, 1606–1619. 49. Schuck, P. (1998). Sedimentation analysis of noninteracting and self-associating solutes using numerical solutions to the Lamm equation. Biophys. J. 75, 1503–1512. 50. Snyder, D., Lary, J., Chen, Y., Gollnick, P. & Cole, J. L. (2004). Interaction of the trp RNA-binding attenuation protein (TRAP) with anti-TRAP. J. Mol. Biol. 338, 669–682.

Edited by D. E. Draper (Received 16 August 2004; received in revised form 7 October 2004; accepted 8 October 2004)