Characterization of Transgenic Mice with Targeted Disruption of the Catalytic Domain of the Double-stranded RNA-dependent Protein Kinase, PKR*

The interferon-inducible, double-stranded RNA-dependent protein kinase PKR has been implicated in anti-viral, anti-tumor, and apoptotic responses. Others have attempted to examine the requirement of PKR in these roles by targeted disruption at the amino terminal-encoding region of the Pkr gene. By using a strategy that aims at disruption of the catalytic domain of PKR, we have generated mice that are genetically ablated for functional PKR. Similar to the other mouse model of Pkr disruption, we have observed no consequences of loss of PKR on tumor suppression. Anti-viral response to influenza and vaccinia also appeared to be normal in mice and in cells lacking PKR. Cytokine signaling in the type I interferon pathway is normal but may be compromised in the erythropoietin pathway in erythroid bone marrow precursors. Contrary to the amino-terminal targeted Pkr mouse, tumor necrosis factor α-induced apoptosis and the anti-viral apoptosis response to influenza is not impaired in catalytic domain-targetedPkr-null cells. The observation of intact eukaryotic initiation factor-2α phosphorylation in these Pkr-null cells provides proof of rescue by another eukaryotic initiation factor-2α kinase(s).

PKR overexpression in HeLa cells induces apoptosis (28) by mechanisms that are inhibitable by Bcl-2 (27). Antisense ablation of PKR conferred resistance to tumor necrosis factor ␣ (TNF-␣)-induced apoptosis in U937 cells (36) indicating the requirement for PKR in the apoptotic response to TNF-␣ in these cells. It is thought that dsRNA is a trigger for apoptosis in vaccinia virus-infected cells (37), and influenza-mediated apoptosis is suppressed in cells expressing inactivated PKR (26).
Much of the work elucidating the role of PKR in growth control and apoptosis utilized mutated PKR in tissue culture settings. Previous work on targeted disruption of Pkr by homologous recombination in mice focused on interruption of two exons including one that encodes the initiating methionine (25). Surprisingly, analysis of this PKR-defective mouse model revealed no evidence of tumors and normal anti-viral responses in untreated animals. The mice were defective in IRF-1 and NF-B signaling and showed diminished stress-induced apoptotic responses (20,25,38). To examine whether PKR is essential in anti-viral response, anti-proliferative functions of cellular growth control, and in apoptotic response to various stimuli, we generated mice devoid of PKR function by targeted disruption of the PKR catalytic domain using homologous recombination that interrupts exon 12. Mice homozygous for Pkr disruption (Pkr 0/0 ) develop normally and are fertile with average sized litters. IFN-␣ and -␤ induction of transcription is intact, and the mice show normal hematopoiesis. Pkr 0/0 mice show responses to vaccinia and influenza infection comparable to control animals or cells. Apoptotic response to influenza infection or TNF-␣ was not impaired. Our data indicate that catalytic disruption of Pkr is not sufficient to ablate eIF-2␣ phosphorylation and that unappreciated members of the eIF-2␣ kinase family must compensate for loss of PKR function.

MATERIALS AND METHODS
Construction of the PKR Gene-targeting Vector-A 28.8-kb region of isogenic DNA encoding murine PKR was isolated and mapped. A 2.9-kb SacI fragment that encodes exons 10 and 11 (39) as determined by sequence analysis and a 1.9-kb XhoI/BamHI fragment that bears intron XII sequence just distal to exon 12 were cloned into the MC1 neo poly(A) vector (Stratagene). The herpes simplex virus thymidine kinase expression cassette was introduced into the 3Ј end of the vector at the NotI site to produce the PKR catalytic domain replacement vector designated pTV65TK. Insertion of the neomycin (neo) cassette results in replacement of a 3.7-kb BamHI to XhoI fragment of the Pkr gene including complete removal of the 180-nucleotide exon 12.
Generation of PKR-deficient Mice-The targeting vector was introduced into the J1 embryonic stem (ES) cell line (strain 129/terSv (40)) by electroporation, and cells were selected with neomycin (200 g/ml) and 1-(2Ј-deoxy-2Ј-fluoro-1-␤-D-arabinofuranosyl)-5-iodouracil (0.2 M; kindly provided by Dr. Michael A. Rudnicki) as described previously (41). Targeted disruption of the Pkr gene was determined by EcoRI or PstI digestion followed by Southern hybridization with probe A or probe B, respectively (see Fig. 1). To verify that only one copy of each construct was integrated in each targeted ES clone, XhoI-digested DNA was probed with the Neo probe. 3 out of 750 ES clones screened were true homologous recombinants. ES cells that were homozygous for the targeted Pkr allele were obtained by selection of the heterozygous targeted clone 7A8.3 in high neomycin selection (2 mg/ml) as described previously (42). Of 20 resistant colonies picked, one clone (7A8.3.7) was homozygous mutant.
Chimeric mice were generated by microinjection of ES cells with a disrupted Pkr gene into Balb/c donor blastocysts and subsequent implantation into CD-1 foster mothers. Chimeric animals were tested for germ-line transmission of the agouti coat phenotype of 129/Sv-derived ES cells by crossing with Balb/C mice. Heterozygote individuals were identified by genotyping and then intercrossed to generate homozygous lines.
Northern Blot Analysis-Total RNA was isolated from various ES cells that were untreated or treated with 500 units/ml interferon ␣ and ␤ (Cytimmune) for 16 h. Poly(A) ϩ RNA was selected on an oligo(dT)cellulose (Life Technologies, Inc.) column. 6 g of mRNA was resolved on an agarose-formaldehyde gel, blotted, and hybridized with a murine PKR cDNA StuI/SspI fragment corresponding to nucleotides 290 -1621 that encode residues 98 -515. The blot was washed at high stringency and imaged on a Molecular Dynamics PhosphorImager. The blot was stripped and reprobed with the mouse glyceraldehyde phosphate dehydrogenase cDNA, washed, and exposed. After stripping the blot again, it was hybridized with an MluI/BamHI fragment of the neomycin gene, washed, and exposed as above. IFN inducibility of 2Ј,5Ј-oligoadenylate synthetase was assessed by rehybridizing this blot with the 2Ј,5Ј-oligoadenylate synthetase cDNA.
Derivation of Primary Mouse Embryo Fibroblasts (MEFs)-Primary MEF cultures were established from E13 embryos as described previously (40). Cells were genotyped as above.
Immunoblot and Immune Complex Kinase Analysis of Pkr-null Cells-Wild-type and homozygous-null ES and MEF cells were lysed in Lysis buffer (10 mM Tris (pH 7.5), 150 mM sodium chloride, 5 mM EDTA, 1% Triton X-100), lysates denatured, resolved by SDS-PAGE, and blotted onto nitrocellulose membranes. PKR polypeptides were analyzed by immunoblotting with antisera (anti-TikMAP2) raised against a multiple antigen peptide (MAP) bearing mPKR residues 101-114 and visualized by enhanced chemiluminescence. Alternatively, cell lysates were incubated with poly(I)⅐poly(C) (pIC)-agarose (Amersham Pharmacia Biotech) at 4°C with rotation to allow binding, washed in lysis buffer, eluted and resolved by SDS-PAGE, and blotted. dsRNA binding PKR polypeptides were visualized with antisera (anti-Tik100) raised against His-tagged mPKR residues 1-98 and ECL.
Immune complex kinase reactions were performed by incubating lysates from various cells with the anti-Tik100A antisera and recover-ing the immune complex with protein A-Sepharose beads. The complexes were resuspended in reaction mixtures of kinase buffer with 5 Ci of [␥-32 P]ATP and 10 ng/l reovirus dsRNA and incubated at 25°C for 30 min. Autophosphorylation activity of PKR was visualized by SDS-PAGE and phosphorimaging.
Clonogenic Assays and Hematopoietic Development-Bone marrow from age-matched mice was harvested by flushing tibias under sterile conditions. An aliquot of cells was obtained, and erythrocytes were removed by lysis in hypotonic solution, and viable nucleated marrow cells were counted using a hemocytometer and trypan blue. Cells were resuspended in Iscove's modified Dulbecco's medium/methyl cellulose supplemented with 20% fetal calf serum, 0.09% bovine serum albumin, and cytokines as noted below. 8 ϫ 10 4 cells were plated in 1 ml of media in 35-mm 2 Petri dishes and incubated at 37°C and 5% CO 2 for 10 days. Erythroid and myeloid colonies were counted, and colony-forming units (CFU) were determined after 10 days. Recombinant human erythropoietin (Epo) was used at 2 units/ml, recombinant murine IL-3 at 10 g/ml, human Kit ligand (KL) from conditioned media at 3% (v/v), and human granulocyte-colony-stimulating factor at 100 ng/ml.
Vaccinia Virus Growth in Cell Lines-CV-1, L-929, wild-type, or Pkr-null cells were plated in duplicate at 2 ϫ 10 5 cells per well in 6-well tissue culture plates. The next day, cells were infected with trypsintreated wild-type vaccinia virus (WR strain) at an m.o.i. of 10 for 1 h at 37°C. Following incubation, monolayers were rinsed and the inoculum replaced with media. At 4,8,12,15, and 18 h after infection, cells were frozen at Ϫ80°C. Cells were scraped off in 0.3 ml of serum-free media and freeze-thawed twice before trypsinization and serial dilution. Plaque assays were performed, and after 2 days of incubation, plates were fixed, stained with crystal violet, and plaques counted to determine the number of pfus/ml of cell lysate as a function of time after infection.
Influenza Infection of PKR-null Mice-Five-week-old Pkr ϩ/O heterozygous control, Clk1-null, 2 and Pkr-null mice were anesthetized with halothane and infected by intra-nasal inoculation with 50 l of serial dilutions of the W29 strain of influenza in phosphate-buffered saline. Each group contained 8 -12 animals that were monitored daily over a 10-day period. The W29 strain is pneumovirulent for mice due to mutations selected on mouse adaptation of the prototype human influenza A strain, A/FM/1/47 (43). Influenza virus stocks were prepared in chicken embryo allantoic cavity, and infectivity titers were assessed by plaque assay on Madin-Darby canine kidney cells (43). The LD 50 value of the virus in the control and Pkr-null animals was assessed by Karber-Spearman analysis. Influenza virus growth in mouse lung was determined from pools of three mice from each group.
Influenza-mediated Apoptosis in Cell Lines-Wild-type and Pkr-null MEFs were plated at 2 ϫ 10 5 cells per well in 6-well tissue culture plates and either left untreated or treated with 500 units/ml IFN-␣ and -␤ overnight. Monolayers were rinsed in phosphate-buffered saline and then infected with influenza A/HK/1/68 (H3N2) virus in 100 l of phosphate-buffered saline at multiplicity of infection (m.o.i.) of 5, 10, or 25 versus mock-infected for 30 min at 37°C. Twenty-four hours later, cells were harvested and assayed for apoptosis by Annexin V-FITC staining (CLONTECH). After incubation of cells with the Annexin V reagent for 10 min, stained cells were fixed in 2% paraformaldehyde and analyzed on a Becton Dickinson FACS instrument for percentage cells with elevated FITC fluorescence indicating apoptosis.
TNF-␣, Lipopolysaccharide (LPS), and dsRNA-mediated Apoptosis in PKR-null Cells-4 ϫ 10 5 MEFs were plated in 60-mm 2 tissue culture plates and treated the next day with 50 -2500 ng/ml actinomycin D for 24 h to establish minimal cytostatic dose and ensure Pkr-null cells responded comparably to control cells. Cell viability was determined by trypsinizing cells and trypan blue staining. Actinomycin D toxicity was not as profound as described previously (38) and was comparable in the two cell types. TNF-␣, LPS, and poly(I)⅐poly(C) synthetic dsRNA (pIC) induction of apoptosis in wild-type MEFs was examined by treating these cells as above with 20 ng/ml TNF-␣, 100 ng/ml LPS, and 100 g/ml pIC with increasing amounts of actinomycin D for 24 h. Cell viability was measured by trypan blue staining and verified by TUNEL assay (Boehringer Mannheim) or morphology under Hoechst staining. TNF-␣-induced apoptosis in these MEFs required actinomycin D cotreatment. The apoptotic dose response to TNF-␣ was determined in PKR-null cells by treating cells with actinomycin D (500 ng/ml) and between 0.1 to 20 ng/ml TNF-␣ for 24 h. Experiments were conducted at least in triplicate.
eIF-2␣ Phosphorylation State in Pkr-null Cells-Immortalized pools of Pkr-null and wild-type cells were washed and lysed in RIPA buffer with protease and phosphatase inhibitors. After clearing the lysate, 20 g of protein were resolved by SDS-PAGE and transferred to nitrocellulose. Phosphorylation of eIF-2␣ was determined by immunoblot with a polyclonal anti-eIF-2␣ phosphoserine 51 antibody, whereas protein levels of eIF-2␣ were determined with a monoclonal antibody to eIF-2␣. Normalization of protein levels was with anti-actin antibody.

Generation of Mice with Mutated Pkr
Gene-We designed a targeting vector such that it would replace exon 12 of PKR (following the nomenclature of Kuhen et al. (39)) with an HSV TK gene promoter-driven neomycin resistance cassette in an orientation opposite to the Pkr promoter (Fig. 1A). Exon 12 encodes subdomains V and VI of the catalytic domain and is essential for enzyme activity (30). Transfected ES cells were grown in selection media, picked, and screened for true homologous recombinants ( Fig. 1B and data not shown) by Southern blot analysis. Targeted cell lines showed the predicted sized fragment for single site integration as detected by the internal neomycin probe (Fig. 1B). Selected ES cells were microinjected into blastocysts from BALB/C mice, implanted into foster mothers, and gave rise to chimeric animals. These were bred with wild-type BALB/C mice to produce heterozygotes which were then interbred to yield progeny with wild-type, heterozygous, and homozygous null genotypes. The frequency of wildtype, heterozygous, and homozygous animals was 0.24, 0.57, and 0.19, respectively (Table I). As with the amino-terminal targeted Pkr mouse (25), no tumors have arisen in homozygous mice targeted at the PKR catalytic domain after a year.

Characterization of Targeted Disruption of the PKR Catalytic
Domain-An ES cell line (7A8.3.7) that was homozygous Pkr 0/0 was selected from the heterozygous mutant cell line, 7A8.3, as described previously (42), using higher dose G418 selection (Fig. 1B). These cell lines and primary mouse embryo fibroblasts (MEFs) derived from homozygous null animals were assessed for PKR activity. Northern blot analysis of poly(A) ϩselected RNA from wild-type ES cells showed the normal pattern of murine PKR transcripts of 6, 4, and 2.5 kb that are induced 2.5-3-fold in IFN-treated cells (Fig. 2 (44)). The 4and 6-kb and larger minor mPKR transcripts appear to be incompletely spliced since they hybridize to an intron-spanning riboprobe (data not shown). Heterozygous Pkr ϩ/0 cells showed similar expression of transcripts. Homozygous Pkr 0/0 FIG. 1. Targeted inactivation of the Pkr gene in ES cells and mice. A, schematic drawing of wild-type Pkr locus, replacement-type targeting vector pTV65TK, and targeted PKR locus after homologous recombination (not drawn to scale). Shaded boxes represent Pkr exons. pTV65TK contains an HSVTK-neo cassette in the opposite transcriptional orientation to the Pkr gene and an HS-VTK-tk expression cassette on the 3Ј end of the construct. Homologous recombination results in introduction of new EcoRI and PstI sites in the targeted allele. Probe A was used in initial screens with EcoRI digestion. Wild-type and recombinant alleles produce 9.4-and 6.9-kb fragments, respectively, with this screen. Established mouse lines were screened with PstI digestion and probe B. This screen produces 4.6-and 2.3-kb fragments from wild-type and recombinant alleles, respectively. The Neo probe was used with an XhoI digest to screen for integration of the targeting vector. R, EcoRI; B, BamHI; S, ScaI; P, PstI; Xh, XhoI; neo, neomycin expression cassette; HSVTK, herpes simplex virus-thymidine kinase promoter; tk, herpes simplex virus-thymidine kinase expression cassette. B, ES cell clones bearing homologous recombinant targeted alleles. Electroporated ES cell colonies were picked and analyzed by Southern screening with EcoRI digest and probe A (left panel). 3 in 750 G418 -1-(2Ј-deoxy-2Ј-fluoro-1-␤-D-arabinofuranosyl)-5-iodouracil-resistant cell lines were homologous recombinants. Single copy integration of the targeting vector was checked with a XhoI digest with the Neo probe showing the predicted 16-kb fragment (right panel). The homozygous Pkr-null cell line 7A8.3.7 was derived by high dose G418 selection. ϩ/ϩ, wild-type; ϩ/0, heterozygous for Pkr-null allele; 0/0, homozygous for Pkr-null allele.

TABLE I
Genotype of progeny of heterozygote and homozygote self-crosses Several mating pairs of either heterozygotes (upper panel) or homozygotes (lower panel) were bred and their offspring genotyped by Southern analysis. The number of offspring of a particular genotype is denoted above along with the sample size and the average litter size arising from these matings. ϩ/ϩ, wild-type; ϩ/0, heterozygotes; 0/0, homozygotes. cells completely lacked the fully spliced mature 2.5-kb transcript as well as the incompletely spliced 6-kb transcript (Fig.  2). However, a transcript of about 4 kb and which appeared IFN-inducible was found in all cells. Reprobing of this blot with the neomycin probe showed this transcript in Pkr 0/0 cells bore neomycin sequence (Fig. 2, right panel) not found in the 4-kb transcript in wild-type cells and thus represented a novel transcript that was derived from the targeted allele. Since the 4-kb transcript in wild-type cells does not hybridize with the neomycin probe, the transcript seen in Pkr 0/0 cells arises from the targeted allele and coincidentally comigrates with the 4-kb wild-type transcript. The HSV TK promoter-derived neomycin transcript is also detected by the neomycin probe as a smaller transcript of about 900 base pairs. The IFN inducibility of the 4-kb novel transcript in Pkr 0/0 cells suggests it is likely Pkr promoter-derived with read-through into the neomycin cassette. This was confirmed by reverse transcriptase-polymerase chain reaction analysis which detects the proximal intron XI and strand-specific riboprobe hybridization which identifies the distal intron XII in this transcript (data not shown). If splicing variants that splice over the neomycin cassette exist, such transcripts would encounter frameshifts in all downstream exons (39,45). The above observations show that targeted disruption of Pkr with our strategy generates alleles that do not produce the normal PKR transcripts and could at most produce message with premature stop codons in the region encoding the catalytic domain of PKR.

Heterozygous-heterozygous crosses
Western blot analysis of cell lysates from wild-type and Pkr 0/0 ES cells and primary MEFs using the antibody raised against murine PKR residues 101-114 (anti TikMAP2) show no mature murine PKR of 65 kDa (Fig. 3A) detectable in the homozygous targeted cell lines. No truncated PKR polypeptide is apparent either, even with 10-fold excess loadings (data not shown). We attempted to increase the sensitivity of detection by enriching for PKR polypeptides with intact RNA binding capacity by incubating lysates with pIC-agarose then immunoblotting for bound PKR polypeptides. Wild-type and Pkr ϩ/0 cells show murine PKR binding to dsRNA that is detectable by another antibody raised to murine PKR residues 1-98 (anti-Tik100; Fig. 3B). Pkr 0/0 cells, however, have no detectable mature or truncated PKR polypeptides. An immune complex kinase autophosphorylation assay using this same antibody, with dsRNA as an activator, was performed on cell extracts from Pkr 0/0 cells. No functional PKR activity was detectable in Pkr 0/0 cells of either ES or MEF origin (Fig. 4). Analysis by cell extract kinase assay without use of our antibodies also revealed no IFN-inducible, dsRNA-activated 65-kDa phosphoprotein in Pkr-null cells (data not shown). These results provide convincing evidence that the strategy of catalytic disruption of Pkr by targeting exon 12 leads to a null allele with no catalytic function retained.
IFN and Cytokine Signaling in PKR 0/0 Cells-Increasing evidence points to a role for PKR in transcription control and cytokine signaling (20 -24). PKR is implicated in regulating IRF-1 and NF-B activation (20,21) and is found in complex with STAT1 in an inverse relation to STAT1 activation (24). We examined the responsiveness of Pkr 0/0 cells to the type I IFNs by assessing the induction of the 2Ј,5Ј-oligoadenylate synthetase transcript in these cells. IFN-␣/␤ signaling through STAT1 appears to be unimpaired since induction of 2Ј,5Ј-oligoadenylate synthetase was normal in Pkr 0/0 cells (data not shown). Furthermore, IFN induction of the 4-kb transcript in Pkr-null cells (Fig. 2) supports this conclusion.
We assessed other cytokine signaling pathways by determining whether hematopoietic development in Pkr 0/0 animals was affected. Bone marrow from animals of wild-type and homozygous null backgrounds were isolated, normalized, and plated in colony forming assays in the presence of the indicated cytokines (Fig. 5). Colony-forming units (CFU) were scored 10 days later for color and counted to determine the response of each hematopoietic precursor population to cytokine stimulation. Myeloid bone marrow precursors showed comparable response to IL-3, Kit ligand (KL), and granulocyte-colony-stimulating factor treatment (data not shown). Erythroid CFU growth from PKR 0/0 animals, however, demonstrated a reproducible diminished response to erythropoietin (Epo). IL-3 and KL co-treatment appear to be able to overcome this PKR-dependent Epo block in erythroid precursors from Pkr 0/0 animals (Fig. 5). These precursors show a higher fold stimulation of CFU in response to IL-3 and KL than in wild-type bone marrow. This is consistent with a study implicating PKR in IL-3 stimulation of protein synthesis in an IL-3-dependent cell line (46). This block appears to have little physiological impact since hematocrit volumes from Pkr-null animals were unaffected (data not shown).
Pkr 0/0 Animals Retain Anti-viral Responsiveness-We wished to determine if loss of PKR would affect the ability of null mice to counter viral challenge. Vaccinia maintains resistance to IFN despite producing large amounts of dsRNA in late infection (47). This is thought to be due to the production of K3L and E3L, two virally encoded PKR inhibitors (48,49). The growth curve of vaccinia (WR strain) in wild-type and Pkr-null primary MEF cells in culture was examined (Fig. 6). The production of infectious vaccinia virus particles in control CV-1 and L-929 cells closely matched the kinetics of growth of this virus strain in HeLa cells (50). The virus replicated equally well in wildtype primary MEFs and, surprisingly, showed no significant difference in growth curve in PKR-deficient MEFs ( Fig. 6 and  inset). No enhancement of vaccinia yield was evident in the absence of PKR, in contrast to the repressive effect of overexpression of PKR (50).
In order to determine if replication of an RNA virus might be more affected by the loss of PKR, we examined the LD 50 of the W29 mouse-adapted strain of influenza virus by intranasal infection of control and Pkr 0/0 animals. As shown in Table II, heterozygous and Pkr-null animals showed an LD 50 of 10 4.3 and 10 4.0 infectious particles, respectively. This 2-fold difference in apparent susceptibility of Pkr 0/0 animals does not seem significant when compared, for example, to the consequences of loss of STAT1 function upon anti-viral response (51,52). STAT1-deficient mice completely succumbed to doses of vesicular stomatitis virus challenge that were 10 4 to 10 6 lower than doses that were sublethal to wild-type animals. Furthermore, these animals had elevated susceptibility to a bacterial pathogen, Listeria, and opportunistic infections by murine hepatitis virus. We also measured the level of influenza growth in mouse lung 2 days post-infection for pools of three mice from each group. Pkr-null mice had 3.8 ϫ 10 7 versus 9.0 ϫ 10 7 pfu/ml for control mice. To date, we have found no sign of impaired antiviral response or opportunistic infections in these animals. Our data demonstrate that PKR is not essential for IFN type I responses or for countering vaccinia or influenza infections and may be redundant in function.
Virus-induced Apoptosis in PKR 0/0 Cells Is Unimpaired-The role of PKR in regulating apoptotic responses has been shown by various groups ( (26 -28, 36, 53) see Introduction), and we wished to determine if PKR may curtail viral infections by inducing apoptosis.
To this end, Pkr 0/0 MEFs were challenged with a strain of influenza A/HK/1/68 (H3N2) (54) at various m.o.i. Cells undergoing apoptosis were monitored using Annexin V-FITC (Boehringer Mannheim) staining, an early apoptosis marker (55). FACS analysis of MEFs infected with virus (Fig. 7) show a dose-dependent increase in cells undergoing apoptosis as determined by the elevated number of FITC-positive cells. The percentage of these cells also showing propidium iodide uptake, indicating incidental necrotic death, was small (typically 10 -20% of total cells) (data not shown). Pkr 0/0 cells were indistinguishable from wild-type cells in triggering apoptosis upon influenza infection. This contrasts with the effect of dominantnegative mutants of PKR that reduced influenza-mediated cell death (26). IFN-␣/␤ pretreatment partially abrogated influenzainduced apoptosis in both normal and Pkr-null cells up to an m.o.i. of 10 ( Fig. 7). At higher m.o.i., IFN pretreatment was unable to rescue cells from virally induced apoptotic pathways. These data indicate that PKR is not essential for influenza to trigger apoptosis pathways in cells and that type I IFNs can mount anti-apoptotic responses at low m.o.i. independent of PKR.
Stress-induced Apoptotic Response in Pkr 0/0 Cells Is Normal-Apoptosis may be induced by a number of signals including cellular stress. PKR itself has been implicated in responses to various forms of stress (56,57). Der et al. (38) describe the abrogation of stress-induced apoptosis in the Pkr-null cells they generated (25). The cells had impaired apoptotic responses to TNF-␣, pIC, and lipopolysaccharide (LPS). We addressed the role of PKR in these pathways by determining whether our Pkr-null cells were likewise impaired. We first assessed the dose response of our wild-type and Pkr 0/0 MEFs to actinomycin D (Fig. 8A) which is required to prime cells to respond to these stimuli (38,58,59) (Fig. 8B). By having found no significant differences, we then optimized TNF-␣, LPS, and pIC treatments of wild-type MEFs to trigger cell death using trypan blue exclusion (Fig. 8B) as described by Der et al. (38) and verified by TUNEL assay (data not shown). We found that at the cytostatic dose of actinomycin D (50 ng/ml) described previously (38), none of these treatments elicited much response from the MEF cells, even at the maximal doses of TNF-␣, LPS, or pIC (Fig. 8B). It required higher doses of co-treatment with actinomycin D (500 ng/ml) before any effect was seen on cell viability, and this was restricted to TNF-␣ treatment only. Finally, we determined the dose response of wild-type and Pkr-null MEFs to TNF-␣-induced apoptosis and found that Pkr 0/0 cells with disruption of the catalytic domain had no impairment of cell death response (Fig. 8C) as described in the other Pkr-null mouse model (38). This was apparent throughout the dose range of TNF-␣ (0.1-20 ng/ml) used ( Fig. 8C and  inset).
Intact eIF-2␣ Phosphorylation in Pkr 0/0 Cells May Indicate the Presence of a Pkr Homolog-The best known mammalian Pkr homolog, the heme-regulated inhibitor, is restricted to cells of erythroid lineage (60 -62). The only other known homolog of Pkr is the yeast GCN2 kinase that is activated by uncharged tRNA and regulates amino acid biosynthesis by control of translation of the GCN4 transcript (63). Despite apparent specialization in function with unique regulatory domains, both mammalian eIF-2␣ kinases, heme-regulated inhibitor, and PKR can rescue GCN2-defective yeast (64). Since our knock-out strategy effectively renders the PKR catalytic domain inactive, any eIF-2␣ phosphorylation in homozygous null cells must result from a previously unappreciated kinase activity. We used an antibody specific to phosphorylated eIF-2␣ ( Fig. 9) (65) and determined the phosphorylation status of eIF-2␣ in both wild-type and Pkr-null animals. As shown in Fig. 9, both cell populations contain similar levels of eIF-2␣, and there is little or no difference in eIF-2␣ phosphorylation status between wildtype and Pkr-null cells.

DISCUSSION
Targeted disruption of a Pkr, catalytic domain exon was performed to explore fully the consequences of loss of PKR function at the organismal level. Our targeting strategy inactivates the catalytic domain by replacing an exon which encodes 60 amino acid residues including 6 amino acids that are known to be required for catalytic function (30). If any splicing variants exist that bypass the disrupted exon and splice into any of the distal exons 13-16 (66), all such transcripts will encounter frameshifts that terminate prematurely within the catalytic domain. PKR cDNA hybridizes to two mRNA species in humans (2.5 and 6 kb) (67), one transcript in rats (4.4 kb) (68), and three transcripts in mouse (2.5, 4, and 6 kb) (44). The 4-kb transcript derived from the targeted allele is coinciden-tally of a similar size to the 4-kb partially spliced transcript in normal cells. However, our data show that this 4-kb transcript contains intron sequence and antisense neomycin sequence that would terminate the reading frame. No mature or truncated polypeptide was observed in Pkr 0/0 cells, and PKR catalytic activity was undetectable. Mice homozygous for this Pkr disruption were bred and appeared to have no obvious defects in gross anatomy and were fertile with average sized litters. No evidence of susceptibility to opportunistic infections or signs of increased tumorigenesis have been observed.
The maintenance of proper control of translation is critical to cellular growth control. Indeed, it has been reported that translation restrictive elements are found in a disproportionate number of the untranslated regions of transcripts of protooncogenes, growth factors, hormone receptors, and transcription factors (69). The absence of any tumorigenic phenotype in either our Pkr-null mice or that of Yang et al. (25) was puzzling given the evidence of the consequences of translation deregulation. For instance, when the restriction of translation at the Cap binding step is circumvented by overexpression of the Cap-binding protein, eIF-4E, cellular transformation occurs (70). PKR restricts growth when expressed in yeast and causes morphological transformation when inactive mutated versions are expressed in mammalian cells. Furthermore, expression of non-phosphorylatable eIF-2␣ mutated at the serine 51 position transforms cells (35) as does overexpression of the PKR inhibitors (p58 ipk (71,72)) and TAR-binding protein (33,34). Thus, the surprising absence of tumors in Pkr-null mice suggests a redundancy of PKR function that has been previously unappreciated.
The anti-viral function of PKR has been studied with analyses of virus mutants lacking anti-PKR inhibitors as well as Since actinomycin D co-treatment is required to trigger apoptosis with various stimuli, wild-type (ϩ/ϩ) and PKR Pkrnull (0/0) MEFs were first assessed for toxicity response to actinomycin D. Cells were treated with increasing doses for 24 h and stained with trypan blue to determine viability. These determinations were performed in triplicate, and error bars represent standard error of the mean. B, the optimal actinomycin D dose for various apoptosis triggers was determined. Wild-type MEFs were co-treated with TNF-␣, LPS, or pIC and increasing amounts of actinomycin D for 24 h. Cell viablity was determined as above and confirmed with TUNEL assay and morphology with Hoechst staining. LPS and pIC fail to show apoptosis above actinomycin D alone levels. Actinomycin D cotreatment is required to potentiate TNF-␣-induced cell death, and the minimal actinomycin D dose for such effect is 500 ng/ml. C, TNF-␣ dose response of Pkr-null cells indicates unimpaired apoptotic response in the absence of PKR. Wild-type (ϩ/ϩ) and Pkr-null (0/0) MEFs were cotreated with actinomycin D (500 ng/ml) and increasing amounts of TNF-␣ for 24 h. Cell viability was determined as above and confirmed with TUNEL assay and morphology with Hoechst staining. TNF-␣ treatment induces cell death in a dose-dependent manner that is unaffected by the absence of PKR. Error bars represent standard error of the mean. Inset, TNF-␣ dose response plotted on a log scale. overexpression of Pkr. Strategies utilized by viruses to overcome PKR activation include inactivating RNAs, pseudosubstrates, endogenous inhibitor mobilization, Pkr expression down-regulation, and dsRNA sequestration. Certain viruses have also developed means to prevent cell death responses to infection such as cowpox crmA (73) and baculovirus IAP (74). A few studies now implicate PKR in controlling apoptosis in response to some triggers. Pkr is itself a candidate "death gene" where overexpression causes apoptosis (28). This has been shown to require the third basic region implicated in PKR autoregulatory regulation (53). PKR is located upstream of Bcl-2 function since Bcl-2 abrogates PKR-mediated apoptosis (27). Expression of inactivated PKR suppresses influenza-mediated apoptosis (26) and antisense ablation of PKR provides immunity to TNF-␣ (36).
Both viral growth and virus-induced apoptosis is normal in Pkr 0/0 cells further supporting the idea that the biological roles of PKR are assisted by parallel pathways. Recently, Taniguchi and co-workers (75) have made similar observations with null fibroblasts derived from the Pkr-null mouse engineered by Yang et al. (25).
We did not find any evidence in Pkr 0/0 cells of defective cytokine signaling (20) or apoptotic response to TNF-␣ (38). This contrasts with the observation that cells expressing the catalytically inactive PKR⌬6 mutant are unable to induce a STAT1-inducible gene in response to IFN-␣/␤ (24). The defect in erythropoietin-induced erythroid CFU development may indicate one developmental signaling pathway that requires PKR, although this defect appears readily compensated by other cytokines in vitro and in vivo.
The discrepancy between our results and those of Yang et al. (25) may be accounted for by mouse strain differences. The mice with targeted Pkr deletion in exons 2 and 3 have a 129/ Sv(ev) ϫ C57BL/6J background (25), whereas our Pkr-null mice have a 129/terSv ϫ BALB/C background. Indeed, strain differences in p53 targeted mice (76,77) have been shown to account for phenotypic differences. Alternatively, our divergent observations could be resolved if the knock-out model described by Yang et al. (25) in fact encodes a truncated gene product that functions as a transdominant protein. Interestingly, Barber et al. (32,78) describe a dsRBD I deletion mutation of human PKR which would closely resemble the putative polypeptide encoded by the Yang et al. (25) PKR knock-out mouse. High level expression of the dsRBD I deletion mutant is an effective dominant-negative polypeptide that causes both cellular transformation and reduced eIF-2␣ phosphorylation (32,78). We speculate that low level expression of a potent transdominant mutant could produce the mouse phenotype observed by Yang et al. (20,25,38). In contrast, we would argue that our Pkr knock-out model produces a null allele that can be compensated for by redundant pathways. This is consistent with our observation of unimpaired eIF-2␣ phosphorylation in null fibroblasts. It will be of interest to determine the in vivo state of eIF-2␣ phosphorylation in cells derived from the Yang et al. (25) mouse with Pkr targeted at the 5Ј end.
The nature of PKR redundancy is not known but could be attributed to the recently discovered human homolog of the yeast GCN2 gene. Indeed, Zhu and Wek (79) have shown that yeast GCN2 contains a previously unidentified motif that functions to target this kinase to ribosomes independently from its response to amino acid starvation. We are pursuing the disruption of this putative eIF-2␣ protein kinase in the mouse to examine its function in mammals. Studying the mouse GCN2 homolog would allow further elucidation of the importance of the family of eIF-2␣ protein kinases, and their role in control of cell growth and apoptosis.