The Anti-apoptotic Function of Hsp70 in the Interferon-inducible Double-stranded RNA-dependent Protein Kinase-mediated Death Signaling Pathway Requires the Fanconi Anemia Protein, FANCC*

Proteins encoded by five of the six known Fanconi anemia (FA) genes form a heteromeric complex that facilitates repair of DNA damage induced by cross-linking agents. A certain number of these proteins, notably FANCC, also function independently to modulate apoptotic signaling, at least in part, by suppressing ground state activation of the pro-apoptotic interferon-inducible double-stranded RNA-dependent protein kinase (PKR). Because certain FANCC mutations interdict its anti-apoptotic function without interfering with the capacity of FANCC to participate functionally in the FA multimeric complex, we suspected that FANCC enhances cell survival independent of its participation in the complex. By investigating this function in both mammalian cells and in yeast, an organism with no FA orthologs, we show that FANCC inhibited the kinase activity of PKR bothin vivo and in vitro, and this effect depended upon a physical interaction between FANCC and Hsp70 but not on interactions of FANCC with other Fanconi proteins. Hsp70, FANCC, and PKR form a ternary complex in lymphoblasts and in yeast expressing PKR. We conclude that Hsp70 requires the cooperation of FANCC to suppress PKR activity and support survival of hematopoietic cells and that FANCC does not require the multimeric FA complex to exert this function.

Certain FA proteins, particularly FANCC, also function independently to suppress apoptotic responses of hematopoietic cells to a variety of environmental cues. Down-regulation of FANCC expression inhibits clonal growth of normal erythroid and granulocyte-macrophage progenitor cells, and the disruption of the FANCC gene in mice renders hematopoietic progenitor cells hypersensitive to the apoptotic effects of interferon (IFN)-␥ and tumor necrosis factor (TNF)-␣ (18 -21). Conversely, FANCC expression suppresses apoptosis in human hematopoietic progenitor cell lines (22), in CD34 ϩ cells from FA-C patients (23), and in hematopoietic progenitor cells from fancc knock-out mice (24). Consequently, inactivating mutations of the FANCC gene account for the near universal development of bone marrow failure in FA-C patients.
Recently, we found that the interferon-inducible doublestranded RNA (dsRNA)-dependent protein kinase (PKR) was constitutively activated in FA cells (25) and that complementation of this defect required FANCC to physically interact with the molecular chaperone Hsp70 (26), a cytoprotective factor known to suppress PKR activation (27)(28)(29). Therefore, both FANCC and Hsp70 protect cells against oxidative stress, chemotherapeutic agents, radiation, and TNF-␣ (26, 30 -33) and share the capacity to suppress the activity of PKR and other eIF-2␣ kinases (25,(27)(28)(29). Consequently, we sought to formally test the influence of these two anti-apoptotic proteins on PKR activity in an in vivo context not confounded by the presence of any other FA proteins. We report here that FANCC and Hsp70 function cooperatively to suppress the kinase activity of PKR and do so independently of the FA protein complex.

EXPERIMENTAL PROCEDURES
Yeast Plasmids and Transformation-A high copy multigenic yeast vector that co-expresses human PKR (or its dominant negative mutant K296R) (for review see Ref. 34) under the control of a galactose-inducible (GAL) promoter and FANCC (or a FA-C patient-derived mutant L554P) under the yeast ADH1 promoter was created by subcloning the cDNA encoding full-length PKR (or K296R) and FANCC (or L554P) into a modified form of the yeast expression vector, pGST (35), in which the DNA sequence coding for the GST fusion protein has been removed. The resulting plasmids, pGAL-PKR, pGAL-K296R, pGAL-PKR-FANCC, and pGAL-PKR-L554P, were used individually to transform the Saccharomyces cerevisiae strains H1816 and H1817 (36) using a Frozen-EZ yeast transformation kit (ZYMO Research). Strain H1816 carries a wild-type eIF-2␣ gene, whereas strain H1817 has a non-phosphorylatable mutant eIF-2␣(S51A). Thus, H1816 cells carrying pGAL-PKR do not grow on synthetic galactose (SGAL) medium because of the inhib-itory effect of PKR kinase phosphorylation on eIF-2␣, but H1817 does grow on SGAL. Transformants of both strains were streaked on a glucose synthetic dextrose (SD)-agar plate and incubated for 3 days at 30°C and replica-plated to both SD and SGAL-agar plate and incubated for 4 days at 30°C.
Protein Extract Preparation-The H1817 transformants, which allow for high expression of both active and inactive PKR, were grown in 10 ml of SGAL medium for ϳ15 h, harvested by centrifugation, washed once with ice-cold sterile water, and resuspended in 600 l of Yeast Breakage Buffer (Qbiogene). Resuspended cells were added to a prechilled FastProtein RED tube (Qbiogene) and homogenized using a FastPrep FP120 Instrument (Qbiogene) at a speed of 6.0 ϫ g for 20 s at 4°C. Aliquots of the resulting extracts containing 50 g of total proteins were then treated with or without 100 units of -protein phosphatase (New England BioLabs) at 30°C for 30 min prior to gel mobility analysis using a monoclonal antibody specific for PKR (Santa Cruz Technology).
Immunoprecipitation and Immunoblotting-Whole cell lysates (1 mg of total protein) were pre-cleared with 50 l of 50% protein A-Sepharose suspension (Amersham Biotech) for 1 h at 4°C. After separation of the protein A-Sepharose, the lysate was incubated with a monoclonal PKR antibody conjugated to agarose (Santa Cruz Technology) or a rabbit polyclonal eIF-2␣ antibody (Santa Cruz Technology), respectively, for 3-5 h at 4°C. Immune complexes were bound to protein A-Sepharose beads, recovered by centrifugation, and washed with Nonidet P-40 lysis buffer. For immunoblotting, samples were heated in Laemmli SDS sample buffer, separated by SDS-PAGE, and transferred onto nitrocellulose membranes. Immunoblots were incubated with primary antibodies for 2 h at room temperature. After washing, the blots were incubated with second antibodies for 30 min at room temperature and developed using an enhanced chemiluminescence kit (Amersham Biosciences).
Recombinant FANCC Purification and PKR Kinase Assays-GST-FANCC and GST-FANCC S249A have been described elsewhere (26). These plasmids were transformed into yeast strain Sc334 (37) using a Frozen-EZ Yeast Transformation kit (ZYMO Research). Expression and purification of the GST fusion proteins were carried out as described previously (38). The indicated amounts of the FANCC or recombinant human Hsp70 (StressGen) suspended in kinase buffer (20 mM Tris-HCl, pH 7.5, 50 mM KCl, 2 mM MgCl 2 , 2 mM MnCl 2 , 0.1 mM ATP, 5% glycerol, and a mixture of protease inhibitors) were incubated with FANCC-and Hsp70-depleted whole cell lysates (200 g) prepared from normal lymphoblasts at 30°C for 10 min. 1 g/ml of poly(I)⅐poly(C) (Amersham Biosciences) and 10 Ci of [␥-32 P]ATP were then added, and the mixtures were incubated for additional 10 min. The phosphorylated PKR and eIF-2␣ were isolated by immunoprecipitation with a monoclonal PKR antibody conjugated to agarose (Santa Cruz Technology) and a rabbit polyclonal eIF-2␣ antibody (Santa Cruz Technology), respectively and analyzed by SDS-PAGE and autoradiography.

Retroviral Transduction of Lymphoblasts, IFN-␥/TNF-␣ Treatments, and Metabolic
Labeling-Normal (JY) and mutant FA-C (HSC536N) lymphoblasts were maintained in RPMI 1640 media (Invitrogen) supplemented with 15% heat-inactivated fetal calf serum. Retroviral vectors expressing FANCC, a mutant FANCC containing a serine to alanine substitution at position 249 (S249A), and an antisense Hsp70 cDNA (asHsp) have been described previously (26). These vectors were transduced into lymphoblasts as described previously (38). Sets of isogenic lines were selected for G418 (300 g/ml) resistance. Cells were stimulated with recombinant human IFN-␥ (10 ng/ml) and TNF-␣ (10 ng/ml) (R & D Systems) for 24 h. For in vivo phosphate labeling, 20 ϫ 10 6 lymphoblasts were treated with IFN-␥ and TNF-␣ (10 ng/ml each) in phosphate-free RPMI 1640 medium containing 15% dialyzed fetal bovine serum for 2 h. Labeling was then performed in the same medium containing [ 32 P]orthophosphate (150 Ci/ml) at 37°C for 3 h. Whole cell lysates were prepared as described previously (20) and were subjected to immunoprecipitation with an agarose-conjugated anti-PKR monoclonal antibody and analyzed by SDS-PAGE followed by autoradiography. Western blot analysis was performed on these precipitates to determine the quantity of total PKR proteins precipitated by the antibody.

RESULTS AND DISCUSSION
FANCC Inhibits PKR Activity in Yeast-Overexpression of human PKR in yeast cells results in autophosphorylation, phosphorylation of eIF-2␣, inhibition of translation, and cell death (40,41). To test the effect of FANCC on PKR activity in vivo, we performed genetic analysis using a yeast (S. cerevisiae) strain in which the sole eIF-2␣ kinase GCN2 has been deleted (36). We used a high copy multigenic yeast vector that coexpresses human PKR (conditionally expressed by a galactoseinducible GAL promoter) and FANCC (constitutively expressed under the control of the yeast ADH1 promoter). Two isogenic strains were used to quantify the impact of human PKR and FANCC on cell survival: strain H1816 (which expresses a wildtype eIF-2␣ gene and is sensitive to PKR activation) and strain H1817 (which expresses a non-phosphorylatable mutant eIF-2␣(S51A) and thus permits the survival of cells containing activated PKR) (36). H1816 yeast transformed with PKR-expressing plasmids grew well on SD minimal medium but as expected (36) was unable to grow on SGAL medium (Fig. 1A). Yeast transformants expressing a kinase-deficient PKR mutant, PKR(K296R) (34), exhibited normal growth on both SD and SGAL media (Fig. 1A). Co-expression of FANCC in H1816 cells reversed PKR-mediated growth inhibition as evidenced by the survival of colonies on SGAL medium (Fig. 1A, lower plate). The co-expression of PKR with a patient-derived inactivating mutant L554P (a leucine-to-proline substitution at residue 554 of FANCC) (42) failed to rescue yeast growth (lower plate). All of the H1817 transformants showed similar growth rates on either SD or SGAL medium, confirming that the expression of these human proteins did not generally perturb normal yeast growth.
PKR-mediated yeast cell death results from autophosphorylation of the kinase (43,44). Indeed, PKR immunoblots of H1817 lysates revealed an expected slower migrating band (45) in yeast overexpressing PKR (Fig. 1B, lane 2), a band eliminated by the treatment of the extract with -phosphatase (lane 3). The kinase-deficient mutant K296R was not phosphorylated when overexpressed in yeast (lanes 4 and 5) as expected (46). The co-expression of wild-type FANCC but not the inactive L554P mutant reduced PKR autophosphorylation as evidenced by a minimal PKR shift (also eliminated by -phosphatase treatment (lanes 6 -9)). We conclude that FANCC inhibits PKR phosphorylation in yeast cells.
We examined the interaction between FANCC and PKR to determine whether these proteins exist in the same protein complex. Immunoblot analysis of anti-PKR immunoprecipitates from H1817 yeast transformants demonstrated a detectable albeit weak protein complex formed between FANCC and PKR when co-expressed in yeast (Fig. 1C, top panel, lane 9). However, higher amounts of mutant FANCC (L554P) protein were found in the anti-PKR immunoprecipitates (compare lane 9 with lane 10), indicating that the binding of FANCC per se is insufficient to modulate PKR activity. Because FANCC interacts with Hsp70 to protect hematopoietic cells from apoptosis induced by IFN-␥ and TNF-␣, both of which function as PKR activators (47,48), and because Hsp70 has been implicated in the repression of PKR and other eIF-2␣ kinases (26,(27)(28)(29), we reasoned that the binding of FANCC with yeast Hsc70 might represent the functional entity for PKR suppression. Indeed, PKR-Hsc70 association was detected in PKR transformants co-expressing the normal FANCC protein (Fig. 1C, middle panel, lane 9) but not the mutant L554P FANCC protein (lane 10).
FANCC and Hsp70 Inhibit PKR Activation in Vitro-To test biochemically the notion that FANCC acts in concert with Hsp70 to suppress PKR activation, we performed in vitro kinase assays using purified GST-FANCC proteins and cell extracts from normal lymphoblasts as sources of PKR and Hsp70 proteins. We first incubated extracts that were depleted of FANCC prepared using antibody affinity chromatography adsorption with increasing amounts of either GST-FANCC or a FANCC mutant GST-S249A known to be defective in interacting with Hsp70 (26) in the presence of dsRNA and [␥-32 P]ATP.
The reaction mixtures were immunoprecipitated with anti-PKR and anti-eIF-2␣ antibodies, and immunoprecipitates were analyzed by SDS-PAGE followed by autoradiography. We noted a dose-dependent inhibition of the phosphorylation of both PKR ( Fig. 2A) and eIF-2␣ (Fig. 2B) by GST-FANCC (lanes 3-6) but not by mutant GST-FANCC S249A (lanes 7-10). Control reactions without GST-FANCC (lane 1) had no effect on PKR activation nor did GST alone (data not shown), indicating that the observed inhibition was mediated by exogenous GST-FANCC. Because FANCC-S249A, which fails to bind to Hsp70, failed to inhibit PKR activity, we suspected that the interaction between FANCC and Hsp70 is required for the suppression of PKR activity and phosphorylation of eIF-2␣.
To confirm the role of Hsp70 in mammalian cells, we performed a second kinase assay by incubating cell extracts de-pleted of both FANCC and Hsp70/Hsc70 with increasing amounts of recombinant Hsp70 and limiting amounts of GST-FANCC or GST-S249A proteins in the presence of dsRNA and [␥-32 P]ATP. PKR was immunoprecipitated, and 32 P incorporation was detected by SDS-PAGE and autoradiography. As shown in Fig. 2C, GST-FANCC had no effect on PKR phosphorylation in the absence of Hsp70 (Fig. 2C, upper panel, lane 1). In the presence of normal but not the mutant S249A FANCC, recombinant Hsp70 reduced the phosphorylation of PKR in a dose-dependent manner (compare lanes 2-5 with lanes 6 -9). We conclude that the formation of an Hsp70⅐FANCC complex is required to suppress PKR activity.
FANCC, Hsp70, and PKR Form a Ternary Complex in Vivo, and Inhibition of PKR Activity Requires Interaction between FANCC and Hsp70 -To demonstrate functional Hsp70- Wild-type (eIF2␣ WT; top half of both plates) or mutant (eIF2␣ S51A; bottom half of both plates) yeast transformed with indicated plasmids were grown on SD medium, replica-plated onto SD (upper plate) or SGAL (lower plate) medium, and incubated at 30°C for 3-4 days. B, H1817 transformants were grown in SGAL medium, and whole cell extracts (30 g) either untreated (Ϫ) or treated (ϩ) with -phosphatase (-PPase) were subjected to immunoblot analysis for PKR. C, H1817 transformants were grown in SGAL medium, and 1 mg of whole cell extracts was immunoprecipitated with anti-PKR antibodies and immunoblotted with antibodies specific for FANCC (top), Hsc70 (middle), or PKR (bottom). 30 g of whole cell extracts (WCE) were run as input controls (lanes 1-5). IP, immunoprecipitation.
FANCC interactions in hematopoietic cells, we quantified IFN-␥/TNF-␣-induced PKR phosphorylation and interactions among FANCC, Hsp70, and PKR in normal and FA-C patientderived HSC536N lymphoblasts. We have already shown that the suppression of Hsp70 expression reduced FANCC⅐Hsp70 complex and further sensitized normal lymphoblasts to IFN-␥ and TNF-␣ (26). As expected, PKR phosphorylation was significantly increased in normal lymphoblasts, expressing an antisense HSP70 cDNA both in ground state (Fig. 3A, lane 3) and in response to IFN-␥ and TNF-␣ (lane 4). PKR was constitutively phosphorylated (lane 5) and hyperphosphorylated (lane 6) in untreated and IFN-␥/TNF-␣-treated HSC536N lymphoblasts, respectively. However, the suppression of Hsp70 expression in HSC536N cells augmented phosphorylation of PKR neither in the ground state nor after treatment with IFN-␥ and TNF-␣ (lanes 7 and 8), findings that are consistent with our previous observation that repression of Hsp70 expression did not further sensitize the mutant FA-C lymphoblasts to IFN-␥ and TNF-␣ because interaction between Hsp70 and mutant FANCC (L554P) protein was already fully compromised (26). The re-introduction of normal FANCC but not the defective Hsp70 interactor FANCC (S249A) into the mutant HSC536 cells fully suppressed PKR phosphorylation mediated by IFN-␥ and TNF-␣ (lanes 9 and 10).
Both Hsp70 and PKR proteins were detected in anti-FANCC immune complexes in the mutant HSC536N and complemented (HSC536N/FANCC) lymphoblasts (Fig. 3B, lanes 1-4), but as observed in the anti-PKR immunoprecipitates of the yeast transformants (Fig. 1C), we found significantly more PKR associated with mutant FANCC than with normal FANCC protein (Fig. 3B, compare lanes 1 and 2 with lanes 3  and 4). Conversely, as was the case in yeast, more Hsp70 was found bound to normal FANCC (lanes 3 and 4) than with the L554P FANCC mutant protein (lanes 1 and 2). Because the L554P substitution severely compromises the association of FANCC with Hsp70 (26), a functional interaction between the two proteins is a prerequisite for the formation of a ternary FANCC⅐Hsp70⅐PKR complex. The results suggested that the suppression of PKR activation requires physical association of Hsp70 with the kinase and that FANCC-Hsp70 interaction is required to facilitate the physical association of Hsp70 with PKR.
To determine more precisely the PKR suppressive role of the FANCC-Hsp70-PKR interaction, we examined the formation of the ternary complex in lymphoblasts expressing antisense HSP70 cDNAs that dramatically reduced the expression of the Hsp70 protein (26). Suppression of Hsp70 reduced the formation of PKR⅐Hsp70 complexes in normal (JY) lymphoblasts (Fig. 3C, compare lanes 1 and 2 with lanes 3 and 4). Treatment of these lymphoblasts with IFN-␥ and TNF-␣ increased the interaction between Hsp70 and PKR (compare lane 1 with lane 2). However, there was little detectable Hsp70 co-precipitated with PKR in extracts from the FA-C mutant HSC536N lymphoblasts, Hsp70 expression, and IFN-␥/TNF-␣ treatments (lanes 5-8) notwithstanding. Re-introduction of the normal FANCC (lane 9) but not the mutant FANCC-S249A known to be defective in interaction with Hsp70 (lane 10) restored the formation of the Hsp70⅐PKR complex to a level observed in normal (JY) lymphoblasts (lane 2). Levels of Hsp70 in PKR immunoprecipitates inversely correlated with phosphorylation (activation) of PKR as shown in Fig. 3A. Interestingly, in normal lymphoblasts expressing antisense HSP70 cDNA, the amount of FANCC co-precipitated with PKR varied inversely with Hsp70 levels (Fig. 3C, middle panel, compare lanes 1 and  2 with lanes 3 and 4) and was actually increased in the mutant  HSC536N cells (lanes 5-8). Expression of normal FANCC in these mutant lymphoblasts reduced the FANCC-PKR association (lane 9) but the expression of the FANCC-S249A mutant did not (lane 10).
We conclude that a physical interaction of Hsp70 with PKR is required for the suppression of PKR kinase activity and that through a direct interaction with Hsp70, FANCC plays a role in

FIG. 2. FANCC and Hsp70 inhibit PKR activation in vitro.
A, suppression of PKR autophosphorylation by the normal but not the mutant (S249A) FANCC proteins. In vitro kinase assays were performed by incubation of increasing amounts of GST-FANCC or mutant GST-S249A proteins with FANCC-depleted cell extracts from normal lymphoblasts (200 g) at 30°C for 10 min followed by the addition of 5 ng/ml dsRNA as PKR activator, and 10 Ci of [␥-32 P]ATP and the mixtures were incubated for an additional 10 min. The phosphorylated PKR was isolated by immunoprecipitation with a monoclonal PKR antibody conjugated to agarose and analyzed by SDS-PAGE followed by autoradiography (top) and immunoblotting with anti-PKR (bottom). NS, not specific; P-PKR, phosphorylated PKR. B, inhibition of eIF-2␣ phosphorylation by the normal but not the mutant (S249A) FANCC proteins. Kinase assays were performed as in A, and the phosphorylated eIF-2␣ was isolated by immunoprecipitation with a rabbit polyclonal eIF-2␣ antibody and analyzed by SDS-PAGE followed by autoradiography (top) and immunoblotting with anti-eIF-2␣ antibody (bottom). P-eIF2␣, phosphorylated eIF-2␣; IgG(H), Ig heavy chain; IgG(L), Ig light chain. C, suppression of PKR autophosphorylation requires both FANCC and Hsp70. Kinase assays were performed as in A but with limiting amounts of GST-FANCC or GST-S249A and increasing concentrations of recombinant Hsp70 and cell extracts that had been depleted of both FANCC and Hsp70.
facilitating the formation of the Hsp70⅐PKR complex. Promiscuous binding of PKR by FANCC in normal lymphoblasts with reduced Hsp70 expression (Fig. 3C, lanes 3 and 4) and in both yeast and human cells expressing mutant FANCC-L554P (Fig.  1C, lane 10, and Fig. 3C, lanes 5-8 and 10) suggests that FANCC binds to PKR and then recruits Hsp70 to the molecule, resulting in the displacement of FANCC and inhibition of PKR. Because the release of FANCC from the ternary complex depends upon an initial association of Hsp70 with FANCC and because the inhibition of PKR is hsp70-dependent, abnormalities of either Hsp70 or FANCC that perturb this interaction would be expected to lead to the kind of constitutive PKR activation documented in FA-C cells (Fig. 3A) (5).
Based on the results described here and on prior work clarifying the role of Hsp70 in regulation of PKR and other eIF-2␣ kinases (27)(28)(29), we propose a model for the role of FANCC and Hsp70 in PKR signaling in hematopoietic cells (Fig. 4). In response to environmental cues such as viral infection, growthinhibitory cytokines (IFN-␥ and TNF-␣) or oxidative stress (e.g. reactive oxygen species), the pro-apoptotic kinase PKR becomes activated by autophosphorylation, leading to the phosphorylation of eIF-2␣ and cell death (49). Uncontrolled activation of PKR causes excessive cell death, leading to the universal hallmark of Fanconi anemia, bone marrow failure (50). These studies also resolve a current debate on the capacity of FANCC to function independently of the other FA proteins. That is, we observe an Hsp70-dependent anti-apoptotic function of FANCC that does not require the participation of any other FA protein, findings that align with the anti-apoptotic influence of FANCC in cells exposed to IFN-␥, TNF-␣, and dsRNA (19,20,25,39). Whether other classical functions of hsp70 require cooperative interactions of FANCC is unknown at this time, but the unexpected observation that high level binding of FANCC to PKR occurs in contexts in which Hsp70 is either absent or unattached to FANCC provides opportunities for probing PKR activation states in the context of this type of Fanconi anemia, one that can be now modeled in yeast. This model may facilitate the development of high throughput screening for prospective therapeutic agents.

FIG. 4. A model for the role of FANCC and Hsp70 in PKR signaling in hematopoietic cells.
In stressed cells, FANCC binds to PKR (1) and recruits Hsp70 to form a ternary complex (2). FANCC is released from the complex (3), and Hsp70 prevents phosphorylation of PKR and eIF-2␣ (4), which allows protein synthesis to continue (5) and cells to survive (6). Mutations in FANCC or down-regulation of Hsp70 result in promiscuous binding of PKR by FANCC and constitutive PKR activation.