Interferon α/β Promotes Cell Survival by Activating Nuclear Factor κB through Phosphatidylinositol 3-Kinase and Akt*

Interferons (IFNs) play critical roles in host defense by modulating gene expression via activation of signal transducer and activator of transcription (STAT) factors. IFN-α/β also activates another transcription factor, nuclear factor κB (NF-κB), which protects cells against apoptotic stimuli. NF-κB activation requires the IFN-dependent association of STAT3 with the IFNAR1 chain of the IFN receptor. IFN-dependent NF-κB activation involves the sequential activation of a serine kinase cascade involving phosphatidylinositol 3-kinase (PI-3K) and Akt. Whereas constitutively active PI-3K and Akt induce NF-κB activation, Ly294002 (a PI-3K inhibitor), dominant-negative PI-3K, and kinase-dead Akt block IFN-dependent NF-κB activation. Moreover, dominant-negative PI-3K blocks IFN-promoted degradation of κBox α. Ly294002, a dominant-negative PI-3K construct, and kinase-dead Akt block IFN-promoted cell survival, enhancing apoptotic cell death. Therefore, STAT3, PI-3K, and Akt are components of an IFN signaling pathway that promotes cell survival through NF-κB activation.

kinases and generate cytoplasmic signals by the tyrosine phosphorylation of STAT proteins (4 -6). IFN-activated STATs (STAT1, STAT2, and STAT3) dimerize and translocate into the nucleus to induce gene transcription. STAT proteins function in the gene activation pathway induced by many other cytokines (7,8). Therefore, the IFN-activated JAK/STAT signaling pathway serves as a paradigm for understanding cytokine signal transduction in general. However, serine phosphorylation events are also critical for the biological response to IFN-␣ (5, 9 -12). IFN-␣ activates PI-3K, an upstream element in a serine kinase transduction cascade, by inducing the rapid tyrosine phosphorylation of its regulatory 85-kDa (p85) subunit (13,14). Furthermore, the IFN-␣-dependent recruitment of PI-3K to the IFNAR1 chain of the type I IFN receptor requires the tyrosine phosphorylation of the STAT3 docking site on the intracellular domain of IFNAR1 (14,15).
We recently reported that IFN activates another transcription factor, NF-B (16). Under most circumstances, NF-B lies dormant in the cytoplasm through the binding of IB inhibitory proteins. Stimulating agents such as viruses, cytokines, and lipopolysaccharides promote dissociation of inactive NF-B/IB complexes, allowing NF-B to enter the nucleus and bind DNA. NF-B binds to cis-acting B sites in the promoters and enhancers of key cellular genes. Active, DNA-binding forms of NF-B are dimeric complexes, composed of various combinations of members of the Rel/NF-B family of polypeptides (p50, p52, c-Rel, v-Rel, RelA (p65), and RelB). In addition to regulating immune and inflammatory responses, NF-B suppresses apoptosis (17)(18)(19)(20). Mice that lack the RelA/p65 gene die embryonically from extensive apoptosis within the liver (17). Activation of NF-B by IFN protects cells from killing through the apoptotic pathway (16). In contrast, inhibition of NF-B nuclear translocation and activation by the introduction of a mutant form of IB that acts as a super-repressor enhance apoptotic killing by IFN.
We evaluated the role that PI-3K and its downstream target, the serine-threonine kinase Akt/PKB, play in IFN-signaling. We identify an IFN signaling pathway that protects cells against proapoptotic agents. IFN receptor signaling through the STAT3 docking site on the IFNAR1 chain leads to NF-B activation, which involves the sequential activation of PI-3K and Akt. STAT3, PI-3K, and Akt are indispensable for IFN-dependent cell survival (antiapoptotic) signals generated through NF-B.

MATERIALS AND METHODS
Biological Reagents and Cell Culture-Recombinant human IFN-␣ (IFNCon1), provided by Amgen, was assayed by protection against the cytopathic effect of vesicular stomatitis virus on human fibroblasts, using the National Institutes of Health human IFN-␣ standard for reference. Anti-Rel and IB␣ antibodies were generously provided by * This work was supported by National Institutes of Health Grants CA73753 (to L. M. P.), CA73023 (to D. B. D.), and CA67891 (to D. B. D.) and by funds from the Department of Pathology, University of Tennessee. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Dr. N. Rice (National Cancer Institute, Frederick, MD). Human Daudi cells were maintained in static suspension cultures at 2-15 ϫ 10 5 cells/ml in RPMI 1640 medium supplemented with 10% defined calf serum (HyClone Laboratories, Logan, UT). For experiments, cells were suspended at 0.5-1 ϫ 10 8 cells/ml in medium before the addition of IFN or other agents.
Transfection Conditions and Constructs-High-efficiency, transient transfection of cells (10 7 ) was accomplished by electroporation (capacitance, 300 microfarads; 250 V) with 500 g of salmon sperm DNA and 20 g of plasmid DNA for each sample. Using a green fluorescent protein construct, we found that transfection efficiency of Daudi cells is Ͼ90%, as determined by confocal microscopy. ⌬p85 is a p85 mutant in which 35 amino acids from residues 479 -513 are deleted, and 2 amino acids (Ser-Arg) are inserted (21). p110* is a constitutively active mutant of the p110 subunit of PI-3K (22). CA-Akt is a constitutively active mutant of Akt in which the Src myristoylation sequence was added to the NH 2 terminus of Akt (23). KD-Akt is a kinase-dead mutant of Akt in which a point mutation K197M was introduced at a site required for kinase activity (23). The pUX-CAT 3XHLAB CAT reporter construct contains three tandemly repeated copies of the NF-B site from the HLA-B7 gene (24).
NF-B Activity Measurements-Nuclei were extracted with buffer (20 mM Tris-HCl, pH 7.85, 250 mM sucrose, 0.4 M KCl, 1.1 mM MgCl 2 , 5 mM ␤-mercaptoethanol, 1 mM NaF, 1 mM Na 3 VO 4 , 1 mM phenylmethylsulfonyl fluoride, 5 g/ml soybean trypsin inhibitor, 5 g/ml leupeptin, and 1.75 g/ml benzamidine), and extracts were frozen and stored at Ϫ80°C (15). For EMSA, the nuclear extracts were incubated with a 32 P-labeled B probe (5Ј-AGTTGAGGGGACTTTCCCAGG-3Ј) derived from a NF-B binding sequence in the immunoglobulin gene promoter (25). To define the presence of specific Rel proteins, nuclear extracts were preincubated with a 1:50 dilution of anti-Rel antibodies at 25°C for 0.5 h and then subjected to EMSA. Gels were quantitated by Phos-phorImage autoradiography. For reporter gene assays, COS-7 cells were transiently cotransfected by electroporation with the pUX-CAT 3XHLAB CAT reporter construct (24) and the appropriate expression vector. After 48 h, the cells were treated with IFNCon1 (5,000 units/ml) for 15 min and assayed for CAT activity. After thin-layer chromatography, radioactivity was measured by PhosphorImage autoradiography.
Introduction of Phosphopeptides into Permeabilized Cells-Daudi cells were permeabilized with streptolysin O as described previously (26). Phosphopeptides (5 M) corresponding to the amino acids surrounding intracellular tyrosine residues of IFNAR1 ( IB␣ Degradation-At various times after IFN-␣ treatment, 1 ϫ 10 8 cells were lysed directly in Laemmli buffer, and equivalent amounts of protein were subjected to SDS-polyacrylamide gel electrophoresis. Proteins were transferred to polyvinylidene difluoride membranes, immunoblotted with specific affinity-purified rabbit anti-IB␣, and visualized by chemiluminescence with the ECL reagent (Amersham Pharmacia Biotech).
Akt in Vitro Kinase Assays-At various times after IFN-␣ treatment, 1 ϫ 10 7 cells were lysed and incubated overnight at 4°C with sheep anti-Akt (Upstate Biotechnology, Inc.) bound to protein G-Sepharose. The immunoprecipitates were collected by centrifugation and subjected to immune complex kinase assays performed with histone H2B as a substrate (27). The reaction was stopped by the addition of 3ϫ Laemmli buffer, and proteins were separated by 12% SDS-polyacrylamide gel electrophoresis and transferred onto polyvinylidene difluoride membranes for autoradiography. Phosphorylated H2B was quantitated on a PhosphorImager. The blot was probed with anti-Akt to quantitate protein levels.
Antiviral and Apoptosis Assays-To determine antiviral activity, cell cultures (5 ϫ 10 5 cells/ml) were preincubated overnight with IFNCon1, followed by infection with vesicular stomatitis virus for 1.5 h at 0.1 plaque-forming units/cell. At 24 h after infection, the virus yield in the medium was assayed by plaque formation on indicator Vero cells (28). To determine apoptosis, cells were cytospun onto glass slides, fixed with 4% formaldehyde, permeabilized with 0.2% Triton X-100, and processed for terminal deoxynucleotide transferase-mediated dUTP nick end labeling according to the manufacturer's recommendations (Boehringer Mannheim). Alternatively, lysates of control and IFN-treated (1,000 IU/ml; 24 h) cells were analyzed for apoptotic DNA by modification of a chemiluminescence-based assay (29). In brief, cells (5 ϫ 10 6 ) were lysed in hypotonic buffer and sequentially digested with RNase and protein-ase K. Low molecular weight DNA was extracted and subjected to nonisotopic labeling of 3Ј ends with digoxygenin-11-dUTP and Taq DNA polymerase. Labeled DNA was separated by electrophoresis on 1.6% agarose and transferred to nitrocellulose, and fragmented DNA was visualized by chemiluminescence detection with alkalinephosphatase-conjugated anti-digoxygenin and CDP-Star substrate (Boehringer Mannheim).

RESULTS
The Role of Tyrosine Motifs in the IFNAR1 Chain in IFNpromoted NF-B Activation-The IFNAR1 chain of the human IFN-␣/␤ receptor acts as a species-specific transducer for type-1 IFN action when transfected into heterologous mouse cells (30,31). Expression of the human IFNAR1 chain of the type I IFN receptor in murine cells confers sensitivity to human IFN induction of NF-B activation, suggesting that this chain of the receptor is critical to IFN-induced NF-B activation (16). To determine whether intracellular tyrosine residues of IFNAR1 are involved in IFN-dependent NF-B activation, peptides corresponding to the amino acids surrounding the four intracellular tyrosine residues in the IFNAR1 chain were introduced into streptolysin O-permeabilized Daudi cells. The permeabilized cells were treated with IFN-␣ and assayed for NF-B activation. A phosphopeptide corresponding to Tyr 527 (YSSQ motif) blocked IFN-induced NF-B activation (Fig. 1). In contrast, the nonphosphorylated Tyr 527 peptide and phosphopeptides corresponding to Tyr 466 , Tyr 481 , and Tyr 538 had no effect on IFN-dependent NF-B activation. These results indicate that NF-B activation is directed through the tyrosine phosphorylation of the conserved YSSQ motif in the IFNAR1 chain. This motif is responsible for bringing PI-3K into a complex with the IFN receptor (14). The general requirement for IFN-dependent tyrosine phosphorylation is demonstrated by the finding that the tyrosine kinase inhibitor genistein blocks IFN-dependent NF-B activation (Fig. 1B). Because the docking domain of PI-3K on IFNAR1 was necessary for IFN-induced NF-B activation, we examined whether IFN-induced NF-B activation involved PI-3K.
PI-3K Is Required for IFN-dependent NF-B Activation-To determine whether PI-3K plays a role in IFN-dependent NF-B activation, we transfected into Daudi cells a mutant p85 subunit (⌬p85) of PI-3K, which functions as a dominant-negative inhibitor for PI-3K-mediated events (21), or a constitutively active PI-3K (p110*, catalytic subunit of PI-3K). IFN treatment of cells transfected with an empty vector induced a prominent NF-B complex ( Fig. 2A). Expression of ⌬p85 blocked IFNinduced NF-B activation, demonstrating that PI-3K is involved in NF-B activation by IFN. In contrast, p110* promoted NF-B activity with the NF-B complex comprised of p50 and c-Rel because the complex was supershifted by antisera to either p50 or c-Rel. The p110*-induced NF-B complex was indistinguishable from that promoted by IFN (Fig. 2B).
To determine the functional importance of PI-3K in NF-B activation, cells were cotransfected with a NF-B-CAT reporter plasmid and p110*, ⌬p85, or an empty vector and assayed for CAT activity 2 days after transfection (Fig. 2C). COS-7 cells were used for these assays because they are IFN-responsive in reporter assays, whereas Daudi cells are not (16). Moreover, COS-7 cells are IFN-responsive as demonstrated by NF-B activation in gel shift assays (data not shown). p110* expression stimulated B-dependent transcription ϳ6-fold as compared with that seen in cells transfected with empty vector. Moreover, IFN stimulated B-dependent transcription ϳ8-fold in cells transfected with empty vector. In contrast, expression of ⌬p85 suppressed IFN-activated B-dependent transcription by Ͼ90%. ⌬p85 expression had no effect on IFN-stimulated response element-dependent transcription (data not shown).
These results indicate that NF-B-activation by IFN via the PI-3K pathway is distinct from the well-established IFN-stimulated response element-dependent mechanism in regulating gene expression.
The activity of NF-B is tightly controlled by inhibitory IB proteins that bind to NF-B complexes and thus sequester NF-B in the cytoplasm. Cytokines, such as interleukin 1 and tumor necrosis factor, promote the serine phosphorylation of IB and its polyubiquitination and proteosome-mediated degradation and thereby induce NF-B translocation to the nucleus. We reported previously that IFN induced a progressive decrease in cellular levels of IB␣, indicating that IFN stimulated NF-B activation by promoting IB␣ degradation (16). Moreover, IB␣ proteins with mutations or deletions of serine phosphorylation sites function as super-repressors of IFN-induced NF-B activation. As shown in Fig. 2D, whereas IFN promoted IB degradation in cells transfected with the empty vector, expression of ⌬p85 blocked IFN-promoted IB␣ degradation. These results indicate that IFN-mediated NF-B activation is PI-3K-dependent.
PI-3K Is Required for IFN-mediated Cell Survival-IFN antagonizes apoptosis by protecting cells against a variety of proapoptotic stimuli, such as virus infection, and antibodymediated cross-linking. In addition, a NF-B-dependent pathway protects cells against the apoptotic action of IFN itself (16). To determine whether PI-3K plays a role in biological activities mediated by IFN, we examined the effects of the selective PI-3K inhibitor LY294002 on IFN action in Daudi cells. LY294002 produced a dose-dependent reduction in IFN-induced NF-B activation (Fig. 3A). Moreover, LY294002 increased IFN-induced apoptotic cell death with an IC 50 of ϳ1 M (Fig. 3B) but did not induce apoptosis by itself. A hallmark of the biological activities of IFN is its ability to inhibit viral replication. LY294002 produced a dosedependent reduction in the antiviral action of IFN (Fig. 3C), indicating that PI-3K is involved in the pathway leading to antiviral activity by IFN.
We examined whether expression of ⌬p85 that blocks IFNinduced NF-B activation would sensitize Daudi cells to IFNinduced apoptosis. In empty vector-transfected cells, IFN only slightly increased apoptosis (from 0.1% to 0.6%), as determined by terminal deoxynucleotide transferase-mediated dUTP nick end labeling assays. However, as shown in Fig. 4A, expression of ⌬p85 markedly sensitized Daudi cells to IFN-induced death (ϳ50%). A prominent feature of apoptosis is the formation of DNA ladders, which reflects DNA cleavage into discrete multimers of ϳ200 base pairs. When cell lysates of IFN-treated Daudi cells expressing ⌬p85 were examined by a highly sensitive chemiluminescence-based DNA fragmentation assay, the formation of the telltale DNA ladder was clearly evident when compared with lysates of IFN-treated Daudi cells transfected with an empty vector (Fig. 4B). These results indicate that the PI-3K-dependent pathway leading to NF-B activation protects cells against the proapoptotic action of IFN. Thus, IFN generates a strong cell survival signal through PI-3K.
IFN Activates Akt, Which Is Required for IFN-dependent NF-B Activation-An important target of PI-3K is the serinethreonine kinase Akt/PKB (32,33), which is known to promote cell survival. Thus, we examined whether IFN activates Akt. Akt was immunoprecipitated from lysates of Daudi cells and assayed for enzymatic activity using histone H2B as a substrate. Although similar amounts of Akt were immunoprecipitated from IFN-treated cells and untreated cells, IFN rapidly increased Akt enzymatic activity (within 5 min) with an ϳ5fold increase at 15 min (Fig. 5A). The IFN-dependent increase in Akt activity was blocked by LY294002, demonstrating that Akt activation is downstream of PI-3K.
To determine the importance of Akt for IFN-promoted NF-B activation, we transfected into Daudi cells a catalytically inactive mutant of Akt called KD-Akt or a constitutively active Akt mutant (CA-Akt). As shown in Fig. 5B, KD-Akt blocked IFNpromoted NF-B activation, as compared with IFN-treated cells transfected with empty vector (EV). In contrast, CA-Akt promoted NF-B activity in Daudi cells (Fig. 5C), and the NF-B complex formed was indistinguishable from that promoted by IFN. DISCUSSION IFN elicits pleiotropic biological effects by regulating gene expression through signals generated upon its binding to a distinct surface receptor on target cells. Type I IFNs bind to a ubiquitously expressed cell surface receptor composed of the IFNAR1 and IFNAR2 subunits (1,3,34,35). Whereas IFNAR2 is the ligand-binding subunit, IFNAR1 acts as a species-specific transducer for the actions of type I IFN (30,31). We have demonstrated previously that the STAT3 transcription factor binds to the tyrosine-phosphorylated IFNAR1 chain through its SH2 domain, and this binding is directed to a YSSQ motif. This motif is perfectly conserved in the cytoplasmic tails of IFNAR1 homologues from diverse species, suggesting that it may play a critical role in type I IFN signaling by specifically docking important SH2 domain-containing cytoplasmic proteins (31,36). A YXXQ motif in the cytosolic tail of the shared signaltransducing gp130 chain of the interleukin 6 receptor family is required for cytokine-dependent STAT3 activation (37).
In the present study, we report that a phosphopeptide corresponding to the YSSQ motif blocked IFN-induced NF-B activation. These results indicate that NF-B activation is directed through the tyrosine phosphorylation of the conserved YSSQ motif in the IFNAR1 chain. This motif is responsible for bringing PI-3K into a complex with STAT3 and the IFN receptor (14). The two SH2 domains of p85 mediate the association of PI-3K with tyrosine-phosphorylated proteins containing a pYXXM consensus sequence. This motif is perfectly conserved in STAT3 homologues from several species. Transfection with a Phe 656 -STAT3 point mutant (in which a Phe was substituted for Tyr 656 of the YXXM motif) blocks IFN-dependent NF-B activation, but not IFN-stimulated gene factor 3 activation, which depends on the formation of STAT1/STAT2 dimers. Therefore, NF-B activation by IFN requires STAT3 phosphorylation at tyrosine residue 656 (a YXXM motif), i.e. the PI-3K docking site. We recently showed that STAT3 is responsible for bringing PI-3K into a complex with the IFN-␣/␤ receptor (14). Expression of STAT3 in an IFN-resistant Daudi cell line complemented defective NF-B activation, as well as several other signaling defects (25). These results suggest that STAT3 may act upstream of PI-3K in NF-B activation by IFN-␣/␤.
These results led us to examine the role of PI-3K in NF-B activation by IFN. By using constitutively active and dominant-negative mutant proteins and pharmacological inhibitors, we have shown that PI-3K is necessary for NF-B activation by IFN, as demonstrated in gel shift assays and in B-dependent gene reporter assays. Moreover, we found in preliminary studies that PI-3K plays a role in the up-regulation of genes by IFN that are dependent on NF-B, such as MHC class I and IFN regulatory factor 1. 2 Cell survival induced by IFN after virus infection in vivo may be due to both direct inhibition of viral replication and protection against virus-induced apoptosis. Thus, our results show that IFN utilizes PI-3K to augment antiviral activity and to promote cell survival via NF-B activation. Because we show that the distinct actions of IFN on viral replication, apoptosis, and cell survival can be modulated, it may now become possible to enhance clinically useful IFN actions or, alternatively, attenuate undesirable IFN actions under specific pathological conditions.
Although there are several downstream targets of PI-3K, the serine-threonine kinase Akt has attracted much attention because of its role in cell survival. Akt was discovered as the product of the oncogene v-akt that transforms lymphoid cells (33). Based on homology to the PKA and PKC family of protein kinases, Akt was also named protein kinase B and RAC-PK (32). The PI-3K/Akt pathway provides cell survival signals in response to nerve growth factor, insulin-like growth factor 1, platelet-derived growth factor, interleukin 3, and the extracellular matrix (38). Akt apparently promotes cell survival by phosphorylating multiple targets, including the Bcl-2 family member BAD (39), the apoptosis-inducing enzyme caspase-9 (40), and the Forkhead transcription factor FKHRL1 that regulates Fas ligand gene expression (41). Our results show that IFN activates Akt enzymatic activity and that kinase-dead Akt blocks IFN-promoted NF-B activation, indicating that Akt is important for IFN-promoted NF-B activation. Moreover, a constitutively active Akt construct promotes NF-B activation. To our knowledge, these results are the first to place Akt in an IFN signaling pathway. Because we found that IFN activates Akt, it will be important to establish which possible substrates for Akt undergo IFN-dependent phosphorylation and determine their physiological significance in IFN-promoted cell survival. IB kinases are a likely substrate because Akt regulates these kinases to activate NF-B in response to tumor necrosis factor and platelet-derived growth factor (42,43).
A STAT3-dependent pathway and a PI-3K/Akt-dependent pathway are known to promote cell survival (38,(42)(43)(44)(45)(46)(47). Interleukin 6 generates cell survival signals to prevent apoptosis through STAT3 activation (48). In the present report, we have defined a mechanism by which IFN promotes cell survival by showing that it activates NF-B through a PI-3K/Akt pathway. This pathway apparently requires STAT3, which acts as an adapter for PI-3K (14). Thus, STAT3, PI-3K, and Akt are all components of the cell survival signaling pathway used by IFN. It will be important to determine whether other cytokines similarly generate cell survival signals by NF-B activation. How STAT3 and PI-3K/Akt specifically relate to one another is not yet known. Recent findings indicate that IFN activates PI-3K through STAT3-independent pathways (49) and that PI-3K is required for STAT3-dependent and STAT3-independent signaling (50,51). Thus, STAT3 and PI-3K are complexly related in IFN signal transduction.
IFN-␣/␤ promotes the survival of activated T cells (52), protects CD4 ϩ cells from human immunodeficiency virus-induced cell death (53), and protects lymphoblastoid cells from cell death induced by virus infection or cross-linking of surface immunoglobulins (16). The clinical efficacy of IFN in the treatment of cancer and viral diseases is often limited by its inability to efficiently induce cell death (54). In contrast, the therapeutic action of IFN-␤ in multiple sclerosis may reflect its ability to protect neuronal cells against proapoptotic cytokines. Our results suggest that the ability of IFN to promote apoptosis is counterbalanced by the induction of potent cell survival signals through signaling dependent on STAT3, PI-3K, and Akt that leads to NF-B activation.