Recombinant Vaccinia Virus K3L Gene Product Prevents Activation of Double-stranded RNA-dependent, Initiation Factor 2a-specific Protein Kinase*

Deletion of the vaccinia virus K3L gene, a homologue of the alpha subunit of protein synthesis initiation factor 2, has been reported to reduce the ability of the virus to grow in interferon-treated cells (Beattie, E., Tattaglia, J., and Paoletti, E. (1991) Virology 183, 419-422). Purified recombinant K3L gene product, pK3r, has potent effects on activation of double-stranded (ds) RNA-dependent, initiation factor-2 alpha (eIF-2 alpha)-specific protein kinase (PKR) in in vitro reactions. Recombinant pK3 prevents the inhibition of protein synthesis by dsRNA in a cell-free translation system from rabbit reticulocytes at levels equal to, or lower than, the level of endogenous eIF-2 alpha. In the cell-free translation system, pK3r exerts its effects at all dsRNA concentrations tested, by preventing phosphorylation of eIF-2 alpha. In addition, pK3r reduces the autophosphorylation of immunopurified PKR, as well as its ability to phosphorylate the alpha subunit of purified eIF-2. At 400 mM NaCl, in vitro translated [35S]methionine-radiolabeled pK3 can be co-immunoprecipitated with human PKR, using a monoclonal antibody to PKR. This tight binding is consistent with a role for pK3 as a pseudosubstrate for the kinase, and identifies the amino-terminal 30% of eIF-2 alpha as the domain recognized by the eIF-2 alpha-specific protein kinases. In addition, the tight binding opens up the possibility of using binding assays to identify functional domains within the kinase and pK3. Recombinant pK3 also prevents activation of the heme-sensitive eIF-2 alpha-specific protein kinase, eIF-2 alpha-PKh, in both cell-free translation systems as well as in partially purified preparations. This suggests some similarity between the eIF-2 alpha binding domains of the two eIF-2 alpha specific protein kinases.

* The nomenclature used for PKR is that recommended by informal agreement by workers in the field (4,5). Nomenclature used for initiation factors is that recommended by the Nomenclature Committee of the International Union of Biochemists (52).

Virus erF-2~1 Homologue Prevents Activation
of PKR phosphorylation site on eIF-2a at residue 51 (I), although pK3 does not have a phosphorylatable serine at this position (37).3 A K3L deletion mutant of vaccinia virus has been reported to have increased sensitivity to interferon (I), and co-transfection of the K3L gene into COS cells enhances translation of a reporter gene by preventing eIF-Pa phosphorylation and activation of PKR (37).
The experiments described in this paper look at the in vitro activity of purified recombinant pK3 in a cell-free translation system from rabbit reticulocytes, as well as its effects on immunopurified PKR. The reticulocyte cell-free translation system contains two eIF-2a-specific protein kinases, PKR, as well as a heme-sensitive eIF-20-specific protein kinase (eIF-2a-PKh). We report that pK3, prevents eIF-2a phosphorylation and thus prevents and reverses the inhibition of translation resulting from the activation of PKR. In addition, we have found that pK3, prevents the autophosphorylation of immunopurified PKR, as well as its ability to phosphorylate purified eIF-Pa. Similarly, we have found that pK3, prevents the inhibition of translation by activation of eIF-2a-PKh, although higher concentrations of pK3, are necessary than for the prevention of activation of PKR. We also present evidence to show that pK3 can be co-immunoprecipitated with antibodies to PKR, in the presence of the kinase, suggesting a tight interaction between the two molecules.

MATERIALS AND METHODS
Cells and Viruses-HeLa and 293 cells were grown and maintained as monolayers in Dulbecco's modified Eagle's medium supplemented with 10% calf serum. Stocks of vaccinia virus (WR strain) were propagated and maintained in African green monkey kidney cells, BSC-1, as described (38).
Escherichia coli Strains and Plasmid Vectors-The coding region of K3L from vaccinia virus (Wisconsin strain) was isolated by polymerase chain reaction amplification using the primers 5'-GGGGCCATGGTTGCATTTTGTTATTCG-3' and 5"GGGGGG-ATCCTATTGATGTCTACACAT-3'. The resulting DNA product was restricted with NcoI and BamHI and cloned into pTM1 (39) between the NcoI and BamHI restriction sites. The sequence of the gene was confirmed by dideoxy DNA sequencing (40). The vector, pTM1, puts the 5"untranslated region of encephalomyocarditis virus mRNA upstream of the initiation codon of K3L, allowing efficient translation of mRNA without capping. The NcoI site engineered into K3L by use of specific oligonucleotide primers, provides the initiation codon, followed by the codon for Val instead of for Leu as the second amino acid. pTMl also contains the T7 polymerase promoter and transcription terminator, obviating the need to linearize K3L DNA prior to in vitro transcription (39). The same polymerase chain reaction-amplified fragment of K3L was ligated into the bacterial expression vector, pETlld (41), also between the NcoI and BamHI restriction sites. E. coli strain BL21(DE3) (Novagen) were transformed with the resulting pETlld/K3L construct. PET-lld contains the gene for the lac repressor and a multi-site cloning region downstream from a T7 promoter (41). Induction of the T7 polymerase, under control of the lac promoter, was initiated by the addition of 0.5 mM isopropyl-lthio-8-D-galactopyranoside (Sigma).
Expression of pK3, in E. coli-2-10-ml overnight cultures of BL21 (DE3) containing the plasmid pETlld/K3L were grown from a single plated colony at 37 "C overnight in the presence of 100 pg/ml ampicillin (Sigma). These cultures were used to inoculate 2 X 1-liter cultures in Luria-Bertani broth containing 150 pg/ml ampicillin, to give an initial density of Asg5 = 0.05-0.1. At a cell density of A695 = 0.5, isopropyl-1-thio-0-D-galactopyranoside was added to 0.5 mM and incubation continued for 30 min, at which time rifampicin was added to 200 rg/ml and the incubation continued for 2 h (41). Cells were harvested by centrifugation at 4 "C, for 10 min, at 7,000 rpm in a Beckman JA-14 rotor.
Purification of pK3,-The cell pellet was washed twice with 50 mM KCl, 50 mM HEPES, pH 7.5, 1 mM EDTA, pH 8.0. The cell pellet was resuspended in lysis buffer, 50 mM KCl, 50 mM HEPES, pH 7.5, R. Jagus and K. Carroll, unpublished observations. 1 mM EDTA, pH 8.0, 1 mM DTT, 0.5% Elugent (Calbiochem), using 50 rl/ml cell culture. The resuspended cells were maintained at 4 "C and sonicated using a cuphorn sonicator (Heat Systems Inc., model XL2020) at setting 7. The resuspended cells were frozen on dry ice and stored in a liquid nitrogen freezer overnight. The next day the frozen cells were thawed at 4 'C and sonication was repeated. The mixture was centrifuged for 15 min at 10,000 rpm in a Beckman JA-20 rotor, at 0 "C. The pellet was recovered and solubilized in 1 M KC], 50 mM HEPES, pH 7.5, 1 mM EDTA, pH 8.0, 1 mM DTT, using 50 pl of buffer/ml of original cell culture. This material was centrifuged for 15 min at 0 "C at 10,000 rpm in a JA-20 rotor in a Beckman 52-21 centrifuge. The supernatant was recovered and dialyzed overnight at 4 "C against 25% ammonium sulfate in 25 mM HEPES, pH 7.5, using Spectrapor 3 dialysis membrane with a molecular mass cut-off of 3,500 daltons. The precipitated material was recovered by centrifugation for 30 min at 0 "C at 12,000 rpm in a JA-20 rotor in a 52-21 Beckman centrifuge. The pellet was recovered and resuspended in 10% glycerol. This material was extensively dialyzed against the same buffer at 4 "C overnight. The purified protein was clarified with BPA-1000 (5 pl/ml) (Toso-Haas), followed by microcentrifugation. This protocol gave yields of pK3, of approximately 1 mg/100 ml of cell culture and was essentially homogeneous. Protein concentration of purified pK3, was determined by the method of Bradford (42), using a kit from Bio-Rad. Molarity was calculated from protein concentration using a molecular weight of 10,556 derived from the deduced amino acid sequence (1). The purified pK3, contained no measurable protease or ATPase activities (data not shown). Note that the use of M, 10,000-12,000 cut-off dialysis tubing results in significant loss of Rabbit Reticulocyte Translation System-Rabbit reticulocyte lysate was prepared and used as described (43). In vitro translation reactions were performed with the addition of 100 mM KC], 0.5 mM MgC12,150 p M amino acids minus valine, ['4C]valine (50 p M , 120 mCi/mmol, Du Pont-New England Nuclear), 10 mM phosphocreatine, 2 units/ml creatine phosphokinase, and 20 p M hemin. The translation mix contained approximately 20 pmol of ribosomes/ml, 55-65 pmol of eIF-P/ml, and exhibited a protein synthetic activity equivalent to 1 (pmol of globin/pmol of ribosomes)/min at 30 "C.
sentially as described, using a narrow pH range of 4.5-6.1, in the Proteins were electrophoresed at 2 mA/gel for 16 h, using reverse polarity, with 0.01 M glutamic acid at the anode and 0.05 M histidine at the cathode. Proteins were transferred to Immobilon-P at 60 V for 1 h in 250 mM glycine, 20 mM Tris, 20% (v/v) methanol, 0.01% SDS. After transfer, the blots were blocked with BLOTTO, 5% nonfat dried milk, 0.01% Antifoam A (Sigma), 0.01% sodium azide in 50 mM Tris-HC1, pH 7.4, 200 mM NaCl (45) and processed as described, using immunopurified polyclonal antibodies raised in sheep (44). An earlier evaluation of the use of VSIEF/immunoblotting with eIF-2a of known phosphorylation state has confirmed that the antibodies are eauallv reactive against phosphorylated and nonphosphorylated 100 mM KCI, 25 mM HEPES, pH 7.5, 0.1 mM EDTA, 1 mM DTT, pK3,.
Cell lysates were centrifuged at 10,000 X g, clarified with BPA-1000 (5 pl/ml), and stored in liquid nitrogen.
Immunoprecipitation of PKR,-50 pl of HeLa or 293 cell extracts were incubated, with rolling at 4 "C for 2 h with the monoclonal antibody to PKR previously coupled to CNBr-activated CL-4B Sepharose, as described (28). The immunoprecipitates were washed in 400 mM NaCl, 50 mM KCI, 25 mM HEPES, pH 7.2,O.l mM EDTA, 10% glycerol, 10 pg/ml aprotinin, 5 pg/ml leupeptin, 5 times, followed by 3 additional washings in 100 mM KCl, 25 mM HEPES, pH 7.2, 0.1 mM EDTA, 10% glycerol, 10 pg/ml aprotinin, 5 pg/ml leupeptin, essentially as described by K a t s e t al. Polyacrylamide Gel Electrophoresis (SDS-PAGE)-Samples were analyzed by 10% SDS-PAGE, essentially as described (47), with the exception of those containing pK3,, which were analyzed by SDS-PAGE using a Tris/Tricine buffer system for the resolution of smaller polypeptides as described (48).

RESULTS
Effect § of pK3? on the A~t~v~~ of the Cell-free Rabbit Retku k y t e T~u~~t~~n System- Fig. 1 ~~~~A ) shows the ability of increasing concentrations of pK3, to prevent translational inhibition in a reticulocyte translation system caused by activation of the endogenous dsRNA-dependent eIF-2a-specific protein kinase. The addition of poly(1) -poly(C) (Pharmacia LKB Biotechnology Inc.) at 125 ng/ml reduces the rate of protein synthesis to approximately 5% of control, after a lag of 15-20 min. This reduces the incorporation of [14C]valine into trichloracetic acid-precipitable material to approximately 50% of control value after a 30-min incubation. The inhibition reflects the inactivation of eIF-2 resulting from the activation of PKR. The addition of pK3, prevents the inhibition, at levels between 2-4 pmo1/100 pl. This is similar to the levels of endogenous eIF-2, estimated to be between 5.5 and 6.5 pmo1/100 pl in reticulocyte lysate by immunoblotting (49). To ensure that the effect on translation was due to the added pK3, and was not due to a trace contaminant from the bacterial cells, an equivalent protein fraction was generated A from untransformed cells, as well as from cells containing vector only. Such protein fractions had no effect on the rabbit reticulocyte trans~ation system (data not shown). In addition, pK3, that was boiled for 10 min lost its ability to prevent dsRNA translation inhibition (data not shown). Fig. 1 (panel B ) demonstrates that in addition to preventing PKR activation in the reticulocyte translation system, pK3, can also rescue translation in the reticulocyte translation system after inhibition by dsRNA has been established. In this experiment, dsRNA (poly(I).poiy(C)) was added t o a final concentration of 125 ng/ml at time zero. In the presence of dsRNA, protein synthetic rates proceed at control rates for 15-20 min and then decline to reach a final rate of about 5% of control values. At 19 min, pK3, was added, to 5 pmo1/100 p1, and reversed the inhibition of protein synthesis to give rates approaching those of control, after a lag of approximately 6 min. These data suggest that pK3, may act as a pseudosubstrate for PKR, and that the turnover rate of the activated state of PKR in the rabbit reticulocyte system is approximately 6 min.
The Effect of pK3, on T r a~~~i o n Is Not ~e p e n~e n t on ~s~~A C o~e n t r u~~n -T h e absolute concentrations of dsRNA required to activate PKR vary with the type of dsRNA, its molecular weight, and the system under study. Fig. 2 (panels A and B ) shows the effects of added pK3, ( 5 pmo1/100 pl) over a wide range of dsRNA (poly (1) .poIy(C)) on the translational activity of the rabbit reticulocyte translation system. The batch of poly(I).poly(C) used in these studies began to inhibit at 12.5 ng/ml, and gave maximum inhibition at 50 ng/ml; not until 5 pg/ml of dsRNA did this level of inhibition begin to decrease. The effect of two levels of pK3,, 4 and 18 pmo1/100 11, was examined over this extended concentration range of dsRNA. Both levels of pK3, prevented the inhibition of dsRNA on translation over this entire range. The function of pK3, is independent of dsRNA concentration, unlike the activity previously described as specific kinase inhibitory factor, or SKIF (9,15). SKIF has been shown to displace the concentration dependence of PKR for dsRNA, and probably corresponds to the vaccinia virus E3L gene product, a dsRNA-binding protein (33, 34). The   (1). poly(C)) was added as indicated. Protein synthetic activity was measured by ["Clvaline incorporation into trichloroacetic acid-precipitable material after a 30-min incubation at 30 "C. pK3, was added as indicated. Protein synthetic activity was measured by [14C]valine incorporation into trichloroacetic acid-precipitable material after a 30-min incubation at 30 "C. above data demonstrate that pK3 does not function as a dsRNA-binding protein and is not equivalent to the activity previously described as SKIF.
Effect of pK3, on eIF-2a-PKh in the Rabbit Reticulocyte System-The reticulocyte translation system contains another eIF-2a-specific kinase, eIF-2a-PKh, a heme-sensitive kinase, which is activated by incubation in the absence of added hemin. It was of interest to know whether pK3, was specific for PKR or whether it would also have an effect on the activity of eIF-2a-PKh. Fig. 3 shows the effect of pK3, on translational inhibition by hemin deprivation in the reticulocyte translation system. In the absence of hemin, protein synthesis continued at control rates for approximately 3-5 min, after which the rate decreased rapidly to give a final rate of approximately 5% of control. After a 30-min incubation, this resulted in the reduction of [14C]valine incorporation to approximately 30% of control. This reflects the inactivation of eIF-2 resulting from the activation of eIF-2a-PKh. Added pK3, also prevented translational inhibition by hemin deprivation, but much higher levels were needed than for preventing inhibition caused by dsRNA addition. Approximately 90-100 pmo1/100 p1 of pK3, were needed to prevent the inhibition of translation caused by hemin depletion, a concentration approximately 20-fold higher than that needed to prevent translational inhibition by dsRNA. These data suggest that although pK3, is acting as a competitive inhibitor for both kinases, it is a better competitor for PKR than it is for eIF-The ability of pK3 to prevent activation of eIF-2a-PKh as well as PKR is also distinct from the characteristics of SKIF, which is unable to prevent activation of eIF-2a-PKh during the incubation of a reticulocyte translation system in the absence of hemin (9).
pK3, Prevents el%-2a Phosphorylation in the Reticulocyte Translation System-The effects of pK3, on activation state of the eIF-2a-specific protein kinases were not measured directly, but its effects on eIF-2a phosphorylation state in the reticulocyte translation system are shown in Fig. 4. The phosphorylation state of eIF-2a was determined by quantitative immunoblotting after vertical slab isoelectric focusing (VSIEF), which separates phosphorylated and nonphosphorylated forms of eIF-2a (44). The ratio of intensities of these two bands gives a direct measure of the steady state phosphorylation state of eIF-Sa, as previously demonstrated (44). In a reticulocyte translation system incubated with hemin and without dsRNA (lane 1 ), eIF-20 was not significantly phosphorylated. After incubation with 125 ng/ml poly(1). poly(C) for 30 min (lane 2), eIF-2a became phosphorylated to a steady state level of approximately 25%. This could be prevented (lane 3) by the addition of pK3, (5 pmo1/100 pl). Similarly, incubation in the absence of hemin led to phosphorylation of eIF-2a (lune 4 ) , and although pK3, at 5 pmol/ 100 p1 was insufficient to prevent this (lane 5 ) , at 50 pmol/ 100 p1 pK3, was able to significantly reduce this phosphorylation (lane 6).
pK3, Prevents Activation and Inhibits Activity of Immunopurified PKR-PKR can be immunopurified from cell extracts using monoclonal antibodies covalently bound to CL-4B Sepharose (28, 46). Fig. 5 (panel A) shows the effect of pK3, on PKR immunopurified from 293 cells. The kinase activity of PKR so purified from interferon-treated 293 cells was assessed by the level of autophosphorylation, as well as by its ability to phosphorylate its substrate, eIF-2a, by measuring the incorporation of [32P]phosphate from [Y-~~PIATP into PKR. Fig. 5 (panel A, lanes 1 and 2) shows the autophosphorylation of 293 cell PKR in the presence and absence of 0.2 pg/ml poly (1) .poly(C). The autophosphorylated kinase appeared as a band of 68 kDa on the autoradiographs. There was little stimulation in the level of autophosphorylation of PKR incubated with dsRNA. This has been observed by other 2a-PKh. Samples were fractionated by VSIEF, transferred to Immobilon-P, and subjected to immunoautography, as described (44). In Fig. 5 (panel B ) the effect of poly(1) -poly(C) and pK3, on the function of PKR from interferon-treated HeLa cells was examined by the autophosphorylation of PKR. In contrast to PKR from 293 cells, PKR from HeLa cells does respond to dsRNA by increased autophosphorylation. As shown in lune 1, PKR from HeLa cells was not maximally autophosphorylated, and autophosphorylation could be stimulated in vitro by incubation with poly(1) .poly(C) from 0.05-0.2 pg/ml, as seen in lunes [2][3][4]. The dsRNA-stimulated autophosphorylation was significantly reduced by the inclusion of 5 or 20 pmol pK3,, as shown in lanes 5 and 6, respectively. Lune 7 shows that extracts from bacteria transformed with vector not containing the K3L insert had no effect on kinase autophosphorylation. Similarly, lune 8 shows that PKR autophosphorylation could not be reduced by pK3, inactivated by heat denaturation.
pK3, Also Inhibits the Activity of Partially Purified eIF-2a-PKh- Fig. 5 (panel C ) shows the effect of pK3, on the activity of partially purified eIF-2a-PKh, which has a molecular weight of 90,000. Lune 2 shows that in the absence of pK3,, added eIF-2a was phosphorylated. Notice in lune 1 that the eIF-2a-PKh preparation contained some contaminating eIF-2a. Lune 4 shows that the addition of 50 pmol of pK3, significantly reduced the phosphorylation of eIF-2a, although kinase autophosphorylation was not appreciably affected.

DISCUSSION
The vaccinia virus K3L gene product, pK3, is a small polypeptide that is homologous to the a subunit of eIF-2. In this investigation, we have shown that recombinant pK3 prevents the activation of PKR in uitro, either in cell-free translation systems, or using immunopurified PKR, thereby preventing eIF-2a phosphorylation. pK3, exerts its effects by tightly binding to the kinase. pK3 is presumed to bind within the eIF-2a binding domain of PKR and these studies suggest that the amino-terminal 30% of eIF-2a contains structures critical for interaction with the kinase. Since pK3, also exerts its effects on eIF-2a-PKh, it seems likely that the two kinases have eIF-2a binding domains that share some sequence homology, but are sufficiently distinct to give different affinities for pK3. The effects of pK3, on autophosphorylation of PKR

Virus eIF-Bar Homologue Prevents Activation of PKR
suggest that the eIF-2a binding site and the autophosphorylation site are closely associated in the three-dimensional structure of this kinase.
The characteristics of pK3 demonstrated in this work are clearly distinguishable from the activity previously described as SKIF (9), currently thought to be the vaccinia virus E3L gene product. Unlike SKIF, pK3 is unable to shift the concentration dependence of PKR for dsRNA. Also in contrast to SKIF, pK3 is able to inhibit the activity of eIF-2a-PKh. In the preparation of SKIF previously described (91, it is probable that most of the pK3 present in the cell extract was lost during dialysis prior to use, due to its small size. Recent findings identify PKR not only as a component of the host defense system, but also as a potent anti-oncogene (50,51). Little is currently known about this anti-oncogenic activity. Furthermore, little is known about how PKR interacts with its substrate(s). The ability of pK3 to bind tightly to the kinase presents the possibility of developing a binding assay to examine the functional domains of eIF-2a-specific protein kinases as well as pK3. By extrapolation, this should also offer some insight into the structure function relationships of eIF-Sa. Such studies should increase our understanding of the role of PKR in preventing or reversing cellular transformation.
The studies presented in this paper show that pK3 has potent effects on PKR in uitro. Earlier studies have shown that the K3L gene transiently expressed in COS cells also has potent effects on PKR. However, the significance of the gene on the ability of vaccinia virus to grow in interferon-treated cells remains unclear. The K3L gene is expressed at early times in vaccinia virus infection (35). However, the E3L gene, another vaccinia virus early gene, also has potent effects on PKR activity, both in vitro and in transfected COS cells (34).
The possibility arises that vaccinia virus has evolved two mechanisms to counteract the interferon-induced antiviral host cell response to infection.