Catalytically Active TYK2 Is Essential for Interferon-β-mediated Phosphorylation of STAT3 and Interferon-α Receptor-1 (IFNAR-1) but Not for Activation of Phosphoinositol 3-Kinase*

TYK2, a Janus kinase, plays both structural and catalytic roles in type I interferon (IFN) signaling. We recently reported (Rani, M. R. S., Gauzzi, C., Pellegrini, S., Fish, E., Wei, T., and Ransohoff, R. M. (1999) J. Biol. Chem. 274, 1891–1897) that catalytically active TYK2 was necessary for IFN-β to induce the β-R1 gene. We now report IFN-β-mediated activation of STATs and other components in U1 (TYK2-null) cell lines that were complemented with kinase-negative (U1.KR930) or wild-type TYK2 (U1.wt). We found that IFN-β induced phosphorylation on tyrosine of STAT3 in U1.wt cells but not in U1.KR930 cells, whereas STAT1 and STAT2 were activated in both cell lines. Additionally, IFN-β-mediated phosphorylation of interferon-α receptor-1 (IFNAR-1) was defective in IFN-β treated U1.KR930 cells, but evident in U1.wt cells. In U1A-derived cells, the p85/p110 phosphoinositol 3-kinase isoform was associated with IFNAR-1 but not STAT3, and the association was ligand-independent. Further, IFN-β treatment stimulated IFNAR-1-associated phosphoinositol kinase activity equally in either U1.wt or U1.KR930 cells. Our results indicate that catalytically active TYK2 is required for IFN-β-mediated tyrosine phosphorylation of STAT3 and IFNAR-1 in intact cells.

The role(s) of the tyrosine kinase TYK2 in the type I IFN 1 signaling pathway has been demonstrated through studies carried out in the TYK2-minus cell line U1A (1,2). U1A cells are completely refractory to IFN-␣, yet retain a partial responsiveness to IFN-␤ (2), suggesting that IFN-␤, but not IFN-␣, activates both TYK2-dependent and -independent signaling pathways (3). Reconstitution of U1A cells with wild-type or mutant forms of TYK2 has revealed a surprising diversity of structural and catalytic functions for TYK2 in the type I IFN receptor (IFNAR-1/2) complex. For example, the N region (residues 1-591) of TYK2 was required for stable cytoplasmic accumula-tion of IFNAR-1 protein (4). Individual Janus kinase homology domains within the N region were specifically implicated in IFNAR-1/TYK2 interaction and signaling in response to IFN-␣ (4). Interestingly, the TYK2 kinase domain was found to be dispensable for some aspects of IFN-␣/␤ signaling, such as the IFN-dependent induction of many classical IFN-stimulated genes (5,6). Recent studies using alanine substitutions on the extracellular domain of IFNAR-2 has revealed the type I IFNs interacted differently with the two receptor subunits IFNAR-1 and IFNAR-2 (7). TYK2 has been implicated in the direct phosphorylation of IFNAR-1 in vitro (8,9), and we (10) recently reported that catalytically functional TYK2 was required for induction of the ␤-R1/I-TAC gene by IFN-␤.
In addition to the above functions, TYK2 may also play a role in the recruitment of other signaling molecules to the IFNAR1/2 complex (11). STAT3, which has been reported to "dock" on phosphorylated IFNAR-1, has also been described as an adaptor for coupling PI3K to the IFN pathway in Daudi cells (12). Overexpression of STAT3 restored antiviral and antiproliferative responses in an IFN-resistant Daudi cell line (13). Although TYK2 is believed to mediate IFNAR-1 phosphorylation, which in turn is considered essential for recruiting STAT3 to the IFNAR-1/2 complex, the role of TYK2 in STAT3 activation has not been examined directly.
PI3K designates a family of enzymes that phosphorylate the D3 position of phosphatidylinositol. PI3K consists of a 110-kDa catalytic subunit (p110) that associates with an 85-kDa regulatory subunit (p85). Ligand-dependent interactions between the SH2 domains of the p85 subunit and the phosphotyrosine containing YXXM motif present on several cytokine/growth factor receptors have been reported (14). The phosphorylated lipid products of this enzymatic reaction may act as second messengers to activate protein kinases such as the Akt gene product (15) or certain forms of protein kinase C (16). p85/p110 PI3K exhibits enhanced lipid kinase and protein serine kinase activity in type I IFN-treated cells. A major substrate for PI3K activity in the IFN pathway appears to be insulin receptor substrate-1 as evidenced by the detection of IFN-␣-dependent serine phosphorylation of insulin receptor substrate-1 in U-266 cells and inhibition of such phosphorylation by the semi-selective PI3K inhibitor wortmannin (17).
In these experiments no interactions between p85 and TYK2, JAK1, IFNAR-1, or IFNAR-2c were detected suggesting PI3K did not interact with IFN-␣ signaling components upstream of insulin receptor substrate-1 (17). To date, the role of serine/lipid kinases such as PI3K in the biological response to IFN remains poorly understood.
In this report we demonstrate that catalytically active TYK2 is essential for IFN-␤-dependent phosphorylation of STAT3 and IFNAR-1 in U1A-derived cell lines. We found PI3K to co-immunoprecipitate with IFNAR-1, but not STAT3. Furthermore, IFNAR-1-associated PI3K activity was markedly elevated by IFN-␤ in the presence or absence of catalytically active TYK2, consistent with the observation that PI3K association with the IFNAR-1 receptor component was phosphotyrosine-independent. Thus, it appears that in these cells TYK2 and PI3K activation are not interdependent, suggesting that these molecules have distinct downstream signaling functions.
Cell Extracts and Genomic DNA Affinity Chromatography (GDAC)-Cells were treated with 15,000 units/ml of IFN for 15 min, and nuclear extracts were prepared as described previously (18). Briefly, cells were washed twice with ice-cold phosphate-buffered saline that contained 1 mM Na 3 VO 4 and 5 mM NaF and once with hypotonic buffer. Cells were lysed in hypotonic buffer containing 0.2% Triton X-100 and nuclear extracts collected by differential centrifugation. GDAC has been previously described (18). 50 g of nuclear extract was incubated for 20 min with 25 g of poly(dI-dC) (Amersham Pharmacia Biotech) in binding buffer as described previously (18). This mixture (200 l) was incubated for 2 h with 100 l of bovine genomic DNA-cellulose (Sigma). The DNAbinding proteins were eluted in high salt buffer, concentrated, and resolved by SDS-PAGE for analysis in Western immunoblotting experiments (18).
Western Immunoblot-Cells were treated with recombinant IFN-␤ at 37°C for 10 -15 min before preparation of cell extracts. The antibodies used were: anti-STAT1 and anti-STAT2 (Transduction Laboratories, Lexington, KY), anti-STAT3 (D. Levy, NY University), anti-p85 (Upstate Biotechnology, Inc., Lake Placid, NY). STAT proteins were immunoprecipitated from extracts (19,20), separated by SDS-polyacrylamide gel electrophoresis, and adsorbed to polyvinylidene difluoride (PVDF) membranes (Stratagene, La Jolla, CA). Incubation with primary antibody was for 2 h at room temperature. Blots were washed thoroughly followed by incubation with secondary antibody for 1 h at room temperature. Immunoreactive bands were visualized with the ECL Western blotting system (Amersham Pharmacia Biotech). Tyrosine phosphorylation was monitored by Western blotting using anti-phosphotyrosine monoclonal antibodies PY20, (Transduction Laboratories) and 4G10 (Upstate Biotechnology, Inc.).

IFN-␤-mediated Activation of STAT3 Requires Catalytically
Active TYK2-The availability of sibling cell lines expressing equal amounts of catalytically active (U1.wt) or kinase-deficient (U1.KR930) TYK2 (6) provided an opportunity to examine the requirement for catalytic TYK2 in IFN-mediated STAT activation. This question was addressed using GDAC, a technique that monitors activated STATs without selection bias for DNA binding elements (18). U1.wt and U1.KR930 cells generated activated STAT1 and STAT2 upon IFN-␣ (results not shown) or ␤ treatment (Fig. 1a). GDAC failed to recover STAT3 from lysates of IFN-treated U1.KR930 cells (Fig. 1a). The activation of STAT1 and STAT2 was 30 -40% lower in the U1.KR930 cells compared with U1.wt cells as determined by densitometry analyzed using NIH Image v1.65. IFN-␥ as expected mediated activation only of STAT1, equally in both U1.wt and U1.KR930 cells (results not shown).
Phosphorylation of tyrosine on STAT1, -2, and -3 was assayed by immunoprecipitation Western blotting experiments. STAT1 and STAT2 were activated in U1.wt and U1.KR930 cells in response to IFN-␤ (Fig. 2a). Consistent with results obtained by GDAC, STAT1 and STAT2 phosphorylation was decreased by 45-50% in the U1.KR930 cells compared with U1.wt cells as determined by densitometry with NIH Image v1.65 (Fig. 2c). No activation of STAT3 was seen in U1.KR930 cells treated with IFN-␤ (Fig. 2b) although equivalent amounts of STAT3 protein were present in both the cell lines. The absence of activated STAT3 from IFN-induced DNAbinding complexes was confirmed by using electrophoretic mobility shift assays with an oligonucleotide probe for the c-sis-inducible element. The c-sis-inducible element is a growth factor-responsive element, originally identified in the c-fos promoter, that binds homo-and heterodimers of STAT1 and STAT3 (23). No STAT3 containing DNA-binding complexes were detected in the extracts from U1.KR930 cells (results not shown). These results, along with the observations defined by GDAC, indicate that STAT3 activation requires kinase function of TYK2.
Catalytically Active TYK2 Is Essential for IFN-␤-mediated Phosphorylation of IFNAR-1 in Intact Cells-Binding of type I IFN to its receptor induces rapid tyrosine phosphorylation of the receptor subunits IFNAR-1 and IFNAR-2. In vitro studies had revealed that TYK2 directly binds and phosphorylates IFNAR-1 (9, 24). STAT3 has also been shown to associate with IFNAR-1 in a tyrosine phosphorylation-dependent manner (11). To examine the phosphorylation status of the IFN-␣ receptor subunits in the presence and absence of catalytically active TYK2, cell lysates from IFN-treated U1.wt and U1.KR930 cells were immunoprecipitated with antibodies to IFNAR-2c and IFNAR-1 and subjected to Western blot assay. As shown in Fig. 3, IFNAR-2c was phosphorylated in response to either IFN-␤ or IFN-␣ in both cell lines. IFNAR-1 immunoprecipitates from IFN-treated U1.wt cells, but not U1.KR930 cells, contained tyrosine-phosphorylated IFNAR-1 (Fig. 3). We concluded that catalytically active TYK2 is essential for tyrosine phosphorylation of IFNAR-1 in intact U1A-derived cells.
IFN-␤-mediated PI3K Activation in U1A-derived Cells-The absence of STAT3 and IFNAR-1 phosphorylation in IFN-␤treated U1.KR930 cells provided an opportunity to examine IFN-␤ signaling to PI3K in the absence of these activated components. PI3K has been shown to be activated by IFN-␣ (17,25). Studies using Daudi cells provided evidence that STAT3 can act as an adaptor to couple PI3K to the IFN pathway (12). In that study, tyrosine phosphorylation of STAT3 was proposed to be essential for association with PI3K. In contrast, experiments using IFN-treated U266 cells showed no associa- tion of p85 with Tyk2, Jak1, IFNAR-1, or IFNAR-2c (17). In UIA-derived cells, we found PI3K to co-immunoprecipitate with IFNAR-1 in the presence or absence of IFN-␤ in both U1.wt and U1.KR930 cells (Fig. 4a). IFN-inducible phosphatidylinositol kinase activity was detected in anti-IFNAR-1 immunoprecipitates, regardless of the presence of catalytically active TYK2 (Fig. 4b). A 4 -5-fold IFN-inducible increase in lipid kinase activity was observed in both cell lines.
We also examined the role of catalytically active TYK2 in association of STAT3 with PI3K in the presence and absence of IFN-␤. The p85 regulatory subunit of PI3K failed to co-precipitate with anti-STAT3 (as shown in Fig. 4c), and failed to associate with phosphopeptides derived from STAT3 (results not shown). Further, STAT3 tyrosine phosphorylation did not induce PI3K association in U1.wt cells. PI-specific lipid kinase activity was not detected in STAT3 immunoprecipitates by in vitro kinase assay. We concluded that activated STAT3 does not mediate adaptor function for recruitment of PI3K to the IFN-␣/␤ receptor in U1A-derived cells. DISCUSSION We addressed the function of type I IFN receptor signaling components and TYK2 kinase by studying human fibrosarcoma cells that possess all structural elements of the type I IFN receptor but lack the kinase function of TYK2. We found that catalytically active TYK2 is essential for IFNAR-1 phosphorylation by type I IFNs. In vitro studies had indicated that IFNAR-1 was a substrate for TYK2. Results reported here (Fig. 3) demonstrate that catalytically active TYK2 is required for IFN-␤-mediated phosphorylation of IFNAR-1 in intact cells. However, phosphorylation of IFNAR-2 by IFNs was not dependent on kinase activity of TYK2. Further, despite the absence of IFNAR-1 phosphorylation in U1.KR930 cells, we did see transcriptional induction of IFN-stimulated genes although the levels were lower than those for the wild-type UIA.wt cells (10). Mutational analysis of type I IFNs (26) and detailed structure-function studies of the ligand-binding regions of IFNAR-2 (7) have also shown that IFN-induced transcription does not require a direct interaction of ligand with IFNAR-1.
Catalytically active TYK2 was essential for STAT3 phosphorylation by type I IFNs. Our data suggest that failure to phosphorylate IFNAR-1 abrogates STAT3 phosphorylation in U1.KR930 cells. The requirement for STAT1 and STAT2 to generate the transcription factor ISGF3 in response to type I IFN is well characterized (27)(28)(29)(30). The role of STAT3 in the transcriptional response to IFN is less certain. STAT3 was initially characterized as a transcription factor that mediates cellular responses to IL-6 and EGF (31). STAT3 is activated by a number of growth factors, oncogenes, and nonmitogenic stimuli, and STAT3-null mice are not viable (32). Recent studies indicated that STAT3 associates with the IFNAR-1 chain of the type I IFN receptor in a tyrosine phosphorylation-dependent manner upon IFN-␣ addition (11). Over-expression of STAT3 in a Daudi cell line resistant to antiviral and antiproliferative effects of IFNs restored an IFN-sensitive phenotype (13). In our studies, U1.wt and U1.KR930 cells clearly manifested IFN-␤regulated antiviral competence. 2 However, in U1.KR930, growth inhibitory responses to IFN-␤ were not observed. 3 Therefore, STAT3 activation may be required for induction of genes that mediate antiproliferative effects of type I IFNs in certain cell types. Further insight into the role of STAT3 in IFN signaling and function may be obtained through the identification of IFN-inducible genes whose expression requires activated STAT3. By comparison, PI3K activity, readily induced by IFN-␤ in U1.KR930 cells, does not appear to be sufficient for FIG. 4. IFN-␤-mediated activation of PI3K in TYK2 expressing (U1.wt) and TYK2 kinase-dead (U1.KR930) cells. Panel a, ligandindependent association of PI3K p85 and IFNAR-1. Cells were serumstarved (0% fetal bovine serum) for 4 h, reserved as controls or treated with IFN-␤-1b (2,500 units/ml) for 10 min before preparation of cell extracts, and immunoprecipitated with antibodies to IFNAR-1. Proteins were resolved by SDS-PAGE, blotted onto PVDF membranes, and probed with anti-p85 antibodies. IFNAR-1 and p85 co-immunoprecipitated regardless of the presence of ligand. Panel b, IFN-␤-mediated activation of PI3K does not require catalytically active TYK2. Cells were serum-starved (0% fetal bovine serum) for 4 h and reserved as controls or treated with IFN-␤-1b (2,500 units/ml) for 5 or 10 min as indicated. Cell extracts containing 250 g of protein were immunoprecipitated with antibodies to IFNAR-1, and lipid kinase activity assayed. Autoradiogram of reaction products is shown. Treatment with IFN-␤-1b induced IFNAR-1-associated phosphatidylinositol kinase activity equally in U1.wt and U1.KR930 cells. ϩC represents immunoprecipitation of IFN-␤-treated cell extracts with anti-p85. ϪC represents immunoprecipitation of IFN-␤-treated cell extracts with pre-immune sera. Panel c, lack of association between PI3K and STAT3 in U1A-derived cells. Aliquots of the cell extracts used in panel a were immunoprecipitated with antibodies to STAT3. Proteins were resolved by SDS-PAGE, blotted onto PVDF membranes, and probed with anti-p85 antibodies (top panel) and anti-STAT3 antibodies (bottom panel). No association between STAT3 and p85 was detected. In whole cell extracts of U1.wt and U1.KR930 (right hand panel), p85 was readily detected by Western analysis.
IFN-␤-mediated antiproliferative effects in the absence of activated STAT3. Genes whose maximal induction by IFN-␤ appears to require PI3K action have been reported (12). 4 Our results extend current understanding of the mechanism by which IFN-␤ activates PI3K in U1A-derived cell lines. Catalytically active JAK1 is known to be required for PI3K activation in response to IFN-␣ (33). Our studies indicate TYK2 kinase activity is not required for IFN-induced activation of P13K. We also found that p85 did not co-precipitate with anti-STAT3 nor associate with phosphopeptides derived from STAT3 (data not shown). Rather, we describe a novel pathway for IFN-␤-mediated PI3K activation, through a ligand-independent association between p85 and IFNAR-1. In these cells, PI3K activation by IFN-␤ proceeded equally in the presence or absence of catalytically active TYK2. This result suggests that TYK2 and P13K signal downstream in parallel, rather than through an obligate kinase cascade. The physiological function(s) of PI3K activation by IFN-␤ remain to be defined through experiments in which both TYK2 and PI3K-dependent signaling are blocked.