INTRODUCTION

The Jak-Stat signal transduction pathway was first discovered for interferon  (IFN-)¹ and interferon  (IFN-) by the complementation of mutant cell lines defective in response to IFN- and/or IFN- (Velazquez et al., 1992; Watling et al., 1993; Müller et al., 1993a, 1993b; Darnell et al., 1994; Leung et al., 1995). It has subsequently been shown that the same general pathway is activated by most cytokines and some growth factors (for review, see Ihle and Kerr (1995) and Taniguchi (1995)). This pathway is activated predominantly through receptors which do not possess intrinsic intracellular kinase domains and belong to the class I or class II cytokine receptor superfamily. The lack of inherent catalytic activity in these receptors is overcome through the use of receptor-associated kinases of the Janus kinase (Jak) family. Upon ligand binding, the receptor chains oligomerize allowing the associated kinases to interact and likely cross-activate each other by tyrosine phosphorylation. Subsequently, the activated Jaks directly phosphorylate the intracellular domains of the receptors on specific tyrosine residues. This phosphorylation allows the selective recruitment of SH2-domain containing proteins, particularly Stats (signal transducers and activators of transcription), through a specific interaction between the Stat SH2 domains and the phosphotyrosines within the Stat recruitment sites of the intracellular domains of the receptor chains. These receptor-associated Stats are then rapidly phosphorylated, likely by the activated Jaks (Quelle et al., 1995). The phosphorylation of the Stats is followed by Stat dimerization, translocation to the nucleus, and activation of cytokine inducible genes.

The Jak and Stat families are growing rapidly. The Jak family consists of four members so far: Jak1, Jak2, Jak3, and Tyk2 (Wilks et al., 1991; Silvennoinen et al., 1993b; Firmbach-Kraft et al., 1990; Witthuhn et al., 1994; Kawamura et al., 1994; for review, see Ziemiecki et al. (1994), Ihle et al. (1994, 1995), and Ihle and Kerr (1995)). The Stat family now includes seven different members, which have been cloned: Stat1, Stat1, and Stats2-6 (Schindler et al., 1992; Fu et al., 1992; Zhong et al., 1994a, 1994b; Yamamoto et al., 1994; Akira et al., 1994; Wakao et al., 1994; Hou et al., 1994) and several others, which were identified by electrophoretic mobility shift assays, but have not been cloned yet (Meyer et al., 1994; Finbloom et al., 1994; Tian et al., 1994; Finbloom and Winestock, 1995; Frank et al., 1995).

The Jak kinases are characterized by seven conserved domains: two PTK-related domains and five domains with unknown functions (Ziemiecki et al., 1994). The main difference between Jaks and other protein tyrosine kinases (PTK) is that along with a kinase domain, shown to be active (Wilks et al., 1991), they also contain a PTK-like domain with substitutions of several residues essential for kinase activity. Thus, the second domain is expected to be inactive as a PTK and probably has some other function. Another feature of this family is the lack of any detectable SH2 or SH3 domains. The functions of the other five regions of homology are also unknown.

Stats represent proteins containing SH2, SH3, and DNA-binding domains (for reviews, see Darnell et al. (1994) and Fu (1995)). The highly selective and specific interaction between Stat SH2 domains and the phosphotyrosine containing Stat recruitment sites on the intracellular domains of the cytokine receptors determines which Stats are to be recruited to a particular receptor complex (Heim et al., 1995; Stahl et al., 1995).

The intracellular domain of each cytokine receptor specifically associates with one or more distinct Jaks. While some Jaks and Stats participate in several cytokine signaling pathways, others are more restricted. In the case of Jaks, Jak2, and, especially Jak1, associate with receptors participating in different apparently unrelated cytokine-receptor systems (for reviews, see Ziemiecki et al. (1994), Ihle et al. (1994, 1995), Taniguchi (1995), and Ihle and Kerr (1995)). Jak3 appears to be restricted to the ligand receptor systems through its association with the IL-2R_C chain (Johnston et al., 1994; Witthuhn et al., 1994; Russell et al., 1994; Miyazaki et al., 1994; Tanaka et al., 1994). Tyk2 was shown to be activated during IFN- signaling (Velazquez, et al., 1992; Barbieri et al., 1994) and also during ciliary neurotrophic factor-related cytokine signaling, albeit only in certain cell types (Stahl et al., 1994; Lütticken et al., 1994). Recently, the activation of Tyk2 by IL-10 and IL-12 was shown (Finbloom and Winestock, 1995; Bacon et al., 1995; Ho et al., 1995). Thus, the Jaks may contribute to the specificity of signal transduction at some level.

In the case of IFN- and IFN- signal transduction, the activation of Jak1 and Stat1 occurs in both pathways, however, Tyk2 and Stat2 are activated only during IFN- signaling; and Jak2 only in the IFN- signal transduction pathway. Thus, it was proposed that there may be substrate specificity of Jaks for Stats (Ihle et al., 1994).

To assess whether the specificity of signal transduction resides in the Jaks, we prepared a variety of chimeric receptors based on the human IFN- receptor complex. The extracellular domain of the second human IFN- receptor chain, designated Hu-IFN-R2 or AF-1 (Soh et al., 1994a), was fused to the transmembrane and intracellular domains of various receptors. We then determined which chimeric receptors could support signal transduction. The result of these studies are reported here.

EXPERIMENTAL PROCEDURES

Taq polymerase and all restriction endonucleases were from Boehringer Mannheim Biochemicals or New England Biolabs; Sequenase 2.0 and T4 DNA ligase were from U. S. Biochemical Corp. The [-³²P]dATP and [-³²P]ATP were from DuPont NEN. The cross-linker bis(sulfosuccinimidyl)suberate was from Pierce Chemical Co. All other chemical reagents were analytical grade and purchased from U. S. Biochemical Corp.

The vector pR2 expressing the Hu-IFN-R2 chain under control of cytomegalovirus promoter was constructed as described previously (Soh et al., 1994a; Kotenko et al., 1995). The vector pR1 expressing the Hu-IFN-R1 chain was constructed as follows. The pHu-IFN-R8 plasmid (Kumar et al., 1989) was digested with HindIII restriction endonuclease, incubated with the large fragment of DNA polymerase I and dNTPs to fill in the ends, and then digested with BamHI restriction endonuclease. The fragment containing the Hu-IFN-R1 cDNA was ligated into EcoRV and BamHI sites of the pcDNA3 vector (Invitrogen).

The Hu-IFN-R1 cDNA was recloned from HuIFNAR1/pcDNAI (Lim, 1995) into pcDNA3 with EcoRI and XbaI restriction endonucleases. The expression vector was designated pR1. The PCR products for Hu-IFN-R2 cDNA were obtained by a nested PCR procedure with a pCEV15 phage DNA mixture isolated from about 5 × 10⁶ phage clones from the human M426 cell library (Miki et al., 1989) as template. The primers for the first round of PCR were: 5-GCCGCAAGGCGAGAGCTGC-3 specific for the 5 end of Hu-IFN-R2 cDNA (Novick et al., 1994); and pCEV15 vector primer 5-AGATCTAAGCTTGGCCGAGG-3. For the second round we used the same 5 primer and primer 5-GCGGAATTCTTAATCACTGGGGCACAG-3 specific for the 3 end (bases 1204-1222) of Hu-IFN-R2 cDNA (boldface) containing an EcoRI site within the primer. The PCR product was digested with EcoRI restriction endonuclease and ligated into EcoRI and blunt ended BamHI sites of the pcDNA3 vector. The expression vector was designated pR2.

To isolate the Hu-CRFB4 cDNA an oligonucleotide 5-GTCCATGGCGTGGAGCCTTGGGAG-3 homologous to the 5 end of Hu-CRFB4 cDNA (Lutfalla et al., 1993) was labeled with [-³²P]dATP with terminal deoxynucleotide transferase to a specific activity of 5 × 10⁶ cpm/µg. It was used to screen about 10⁶ phages from the human M426 cell library (Miki et al., 1989). Two positive clones were purified and plasmids were rescued from these two  phages. The CRFB4 cDNA was released from one of the rescued plasmids designated pCEV15-CRFB4-1 by digestion with ThaI and SalI restriction endonucleases and cloned into EcoRV and XhoI sites of the pcDNAIneo vector (Invitrogen). The expression vector for the Hu-CRFB4 chain was designated pCRF.

The vector expressing the chimeric receptor Hu-IFN-R2/Hu-IFN-R1 (R2/R1) was constructed as followed. The asymmetric PCR reaction was performed with a specific primer 5-GATGCCTCCACTGAGCTTCAGCAACTTTGGATTCCAGTTGTTGC-3 and 100-fold excess of T7 primer with plasmids pHu-IFN-R8 (the Hu-IFN-R1 cDNA) and pR2 as templates in the same reaction mixture. The second round of the PCR was performed with the set of primers 5-GACCCTCTTTCCCAGCTGC-3 and 5-GCCACACATCCTCTTTACGC-3 with 1 µl of PCR reaction mixture from the first round of PCR as template. The final R2/R1 PCR product was digested with BstEII restriction endonuclease and cloned into BstEII and blunt-ended XbaI sites of the pR2 plasmid. The expression vector was designated pR2/R1.

To introduce an NheI site in the beginning of the transmembrane domain of the Hu-IFN-R2 cDNA clone, the PCR reaction was performed with two primers 5-GCCTTTTTTAGTTATTATGTC-3 and 5-ATCGCTAGCCATTGCTGAAGCTCAGTGGAGG-3 and plasmid pR2 as a template according to a standard protocol (Sambrook et al., 1989). The PCR product was digested with BstXI restriction endonuclease and ligated into BstXI and EcoRV sites of the pR2. The plasmid was designated pR2NheI.

To construct chimeras Hu-IFN-R2/Hu-IFN-R1, Hu-IFN-R2/Hu-IFN-R2, and Hu-IFN-R2/CRFB4 the PCR reactions were performed with SP6 primer and 5-GTGGCTAGCTATAGTTGGAATTTGTATTGC-3, 5-GTGGCTAGCATAATTACTGTGTTTTTGAT-3, or 5-GTGGCTAGCCGTCATCCTCATGGCCTCG-3 primers with plasmids pR1, pR2, or pCRF, respectively, as templates. The Hu-IFN-R1 and Hu-IFN-R2 PCR products were digested with NheI and ApaI restriction endonucleases and ligated into NheI and ApaI sites of the pR2NheI. The CRFB4 PCR product was digested with NheI and XbaI restriction endonucleases and ligated into NheI and XbaI sites of the plasmid pR2NheI. The plasmids were designated pR2/R1, pR2/R2, and pR2/CRF, respectively.

To create chimera Hu-IFN-R2/IL-2R_C, the _C chain of IL-2R (Takeshita et al., 1992) was obtained by reverse transcriptase-PCR as follows. The first strand cDNA synthesis in reverse transcriptase-PCR was performed with poly(dT)₁₈ primer with Moloney murine leukemia virus reverse transcriptase with total RNA isolated from peripheral blood leukocytes as template. Two PCR rounds were performed. The primers 5-CGGTTCAGGAACAATCGG-3 and 5-CAAGCGCCATGTTGAAGCC-3 were used for the first round. For the second round the PCR product from the first round was diluted 100-fold and used as a template for the second round with primers 5-GTTAGTACCACTTAGGGC-3 and 5-GTGGCTAGCATGGGAAGCCGTGGTTATC-3. The IL-2R_C PCR product was digested with NheI restriction endonuclease and ligated into the NheI and blunt ended XbaI sites of plasmid pR2NheI. The resultant expression vector was designated pR2/_C. The nucleotide sequences of the modified regions of all the constructs were verified in their entirety.

The 16-9 hamster × human somatic cell hybrid line is the Chinese hamster ovary cell (CHO-K1) hybrid containing a translocation of the long arm of human chromosome 6 encoding the HUIFNGR1 (Hu-IFN-R1) gene and a transfected human HLA-B7 gene (Soh et al., 1993). The 16-9 cells were maintained in F-12 (Ham's) medium (Sigma) containing 5% heat-inactivated fetal bovine serum (Sigma) (complete F-12 medium). HEp-2 cells, a human epidermoid larynx carcinoma cell line, and COS-1 cells, a SV40 transformed fibroblast-like simian cell line, were maintained in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) with 10% heat-inactivated fetal bovine serum.

The 16-9 cells were stably transfected with the expression vectors (1-3 µg of super-coiled plasmid DNA per 10⁵-10⁶ cells) with LipofectAMINE^TM Reagent (Life Technologies) according to the manufacturer's instructions for stable transfection of adherent cells. For cotransfection we used 1-3 µg of plasmid DNA with the neo^R gene and a 10-fold excess of plasmid DNA without neo^R gene per 10⁵-10⁶ cells. All cell lines transfected with plasmids carrying the neo^R gene were selected and maintained in complete F-12 medium containing 450 µg/ml antibiotic G418. COS-1 cells were transiently transfected with the expression vectors by the DEAE-dextran procedure with dimethyl sulfoxide shock (Seed and Aruffo, 1987; Sussman and Milman, 1984).

Cytofluorographic analysis of cells for expression of the HLA-B7 surface antigen was performed as described previously (Jung et al., 1988; Cook et al., 1992; Hibino et al., 1992). Hu-IFN-A/D, a chimeric human interferon active on hamster cells (Rehberg et al., 1982), was used as a control to demonstrate the integrity of the HLA-B7 gene in various cell lines.

Cross-linking of IFN- to Receptors

Recombinant Hu-IFN- with a specific activity of 2 × 10⁷ units/mg was phosphorylated as reported (Rashidbaigi et al., 1986; Mariano and Pestka, 1991). The [³²P]Hu-IFN- was bound to cells and then cross-linked as described previously (Kotenko et al., 1995).

Rabbit anti-Jak1, anti-Jak2, and anti-Jak3 antibodies were developed against synthetic peptides KTLIEKERFYESRCRPVTPSC, DSQRKLQFYEDKHQLPAPKC, and AKLLPLDKDYYVVREPG corresponding to the end of the kinase-like domains of murine Jak1 and Jak2, and to a sequence within the kinase domain of murine Jak3, respectively. Rabbit anti-Tyk2 antibody was from Santa Cruz Biotechnology (catalog number SC-169). Rabbit anti-Stat1 antibody, raised against the C terminus of Stat1, was a gift from James Darnell. Monoclonal anti-phosphotyrosine antibody was from Sigma (catalog number P3300). Rabbit anti-Hu-IFN-R2 antibody, prepared with the extracellular domain of Hu-IFN-R2 as antigen, was a gift of Gianni Garotta.

Cells were starved overnight in serum-free media and subsequently stimulated with Hu-IFN- (1000 units/ml) for 10 min at 37 °C. Preparation of cell lysates, immunoprecipitations, blottings, and in vitro kinase activation assay were performed as described (Kotenko et al., 1995).

Electrophoretic mobility shift assays were performed with a 22-base pair sequence containing a Stat1 binding site corresponding to the GAS element in the promoter region of the human IRF-1 gene (5-GATCGATTTCCCCGAAATCATG-3) as described (Kotenko et al., 1995).

RESULTS

To investigate the specific requirements for the intracellular domain of Hu-IFN-R2 we created chimeric receptors with the extracellular domain of Hu-IFN-R2 attached to the transmembrane and intracellular domains of different human receptors as shown in Fig. 1. In our studies we used the following receptors: two chains of the Hu-IFN- receptor complex, Hu-IFN-R1(R1) (Aguet et al., 1988), or human IFN-R1t₄₅₆ (R1t₄₅₆) the truncated R1 with the intracellular domain terminated by premature stop codon after amino acid 456 (Cook et al., 1992), and Hu-IFN-R2 (R2) (Soh et al., 1994a); two chains of human IFN- receptor complex, Hu-IFN-R1 (R1) (Uzé et al., 1990) and Hu-IFN-R2 (R2) (Novick et al., 1994); Hu-CRFB4, a class II cytokine receptor with unknown function (Lutfalla et al., 1993); and Hu-IL-2 receptor _C chain (Takeshita et al., 1992). To confirm that all chimeric receptors can be expressed properly we first determined the ability of expression vectors encoding chimeric receptors to express the proteins. All plasmids were transiently transfected into COS-1 cells and the expression of the receptors were evaluated. The cellular lysates from COS-1 cells transiently transfected with the expression vectors were resolved on SDS-PAGE, transferred to membrane and probed with antibodies to the Hu-IFN-R2 extracellular domain as all these chimeric receptors contain the Hu-IFN-R2 extracellular domain. In all cases specific bands were detected (Fig. 2). Thus, all vectors encoding the chimeric receptors express these proteins.

It was previously shown that the Hu-IFN-R2 chain is a part of the human IFN- receptor ligand binding complex, although by itself does not bind Hu-IFN-; nevertheless, the Hu-IFN-R2 chain can be detected by cross-linking to Hu-IFN- (Kotenko et al., 1995). Therefore, we used cross-linking to ascertain that the chimeric receptors were expressed on the cell surface and were able to participate in the human IFN- receptor ligand binding complex (Fig. 3). All chimeric receptors were stably expressed in 16-9 cells, hamster cells expressing the Hu-IFN-R1 chain (Soh et al., 1993), and antibiotic G418-resistant cell populations were used in cross-linking experiments. The cell lines were designated according to the extracellular/intracellular domains (e.g. R2/R1) of the chimeric receptors expressed (Fig. 1). It was previously shown that the cross-linking of labeled IFN- to the parental 16-9 cells results in formation of a single cross-linked band on the SDS-PAGE migrating in the region of 120 kDa and corresponding to the Hu-IFN-:Hu-IFN-R1 (IFN-:R1) complex (Kotenko et al., 1995). In the R2 cells, in addition to the IFN-:R1 band the appearance of another cross-linked band migrating in the region of 60 kDa was observed and it was shown that this additional band corresponds to the Hu-IFN-:Hu-IFN-R2 (IFN-:R2) complex (Kotenko et al., 1995; Fig. 3). In cross-linking experiments with cells expressing chimeric receptors the IFN-:R1 cross-linked complex was observed in all cell lines (Fig. 3). Depending on the chimeric receptor expressed in the 16-9 cells, the mobility of the second band corresponding to the Hu-IFN-:Hu-IFN-R2/X (IFN-:R2/X) complex (where X represents the transmembrane and intracellular domains of the chimeric receptor expressed) was different for each chimeric receptor (Fig. 3, left panels). However, the IFN-:R2/X complex was observed for all chimeric receptors, indicating that they all were expressed on the cell surface and formed ternary ligand-receptor complexes (R1:IFN-:R2/X) in all cell lines. Because the IFN-:R2/R1 complex has almost the same mobility as the IFN-:R1 complex formed from the endogenous Hu-IFN-R1, in R2/R1 cells we observed a single cross-linked band in the region of 120 kDa (Fig. 3, right panels). However, in the 16-9 cells transfected with the R2/R1t₄₅₆ (R2/R1 chimera, containing a truncated R1 intracellular domain, R1t₄₅₆) we observed the appearance of an additional band migrating faster than the IFN-:R2/R1 complex and representing the IFN-:R2/R1t₄₅₆ complex (Fig. 3, right panels). The additional complexes observed in all cell lines migrating slower than the IFN-:R1 complex were formed by cross-linking the IFN- homodimer to one or two molecules of R1 and one or two molecules of R2 as shown previously (Marsters et al., 1995).

P]IFN- to the receptors.

Class I MHC antigen (HLA-B7 surface antigen) induction was measured to evaluate the ability of the chimeric receptors to support signal transduction upon Hu-IFN- treatment in the 16-9 cells transfected with different chimeric receptors. The 16-9 cells, expressing only the Hu-IFN-R1 chain of the Hu-IFN- receptor complex, exhibited little or no response to Hu-IFN- (Fig. 4A). The 16-9 cells were transfected with an expression vector encoding the intact Hu-IFN-R2 (R2) or the chimeric receptors Hu-IFN-R2/Hu-IFN-R1 (R2/R1), Hu-IFN-R2/Hu-IFN-R1t₄₅₆ (R2/R1t₄₅₆), Hu-IFN-R2/Hu-IFN-R1 (R2/R1), Hu-IFN-R2/CRFB4 (R2/CRF), Hu-IFN-R2/IL-2R_C (R2/_C) to obtain stable transformants. In addition, 16-9 cells were cotransfected with an expression vector encoding Hu-IFN-R2/IL-2R_C and an expression vector encoding murine Jak3 (R2/_C+Jak3). These stable transformants exhibited a significant response to Hu-IFN- (Fig. 4, B, C, D, E, G, H, and I). For all responsive cell lines the histograms represent the data for clonal cell populations. The Hu-IFN- did not induce MHC class I antigens in 16-9 cells stably transfected with the expression vector encoding the Hu-IFN-R2/Hu-IFN-R2 (R2/R2) chimera (Fig. 4F). As a control, it was shown that all cells responded to Hu-IFN-A/D demonstrating that the MHC class I antigen could be induced in all cell lines (data not shown).

The Recruitment of Different Jaks into the IFN- Receptor Complex

It was reported that different receptors are associated with different Jak family members. Particularly, Hu-IFN-R1 associates with Tyk2 (Barbieri et al., 1994; Colamonici et al., 1994a, 1994b), Hu-IFN-R2 associates with Jak2 (Kotenko et al., 1995), and IL-2R_C associates with Jak3 (Russell et al., 1994; Miyazaki et al., 1994; Tanaka et al., 1994). Jak1 was shown to associate Hu-IFN-R1 (Igarashi et al., 1994; Sakatsume et al., 1995) and reported to associate with Hu-IFN-R2 (Novick et al., 1994). In addition, Hu-IFN-R1 probably weakly associates with Jak2 after oligomerization upon ligand binding (Kotenko et al., 1995). We tested the ability of the intracellular domains of different receptors fused to the Hu-IFN-R2 extracellular domain to recruit different members of the Jak family into the Hu-IFN- receptor complex and to determine if the recruited Jaks can be activated upon Hu-IFN- treatment. First, we determined the phosphorylation of Jak2 upon Hu-IFN- treatment in the cell lines expressing various chimeric receptors (Fig. 5A). The phosphorylation of Jak2 in response to Hu-IFN- in cells was examined by immunoprecipitation with specific anti-Jak2 antibodies, followed by a Western blot visualized with anti-phosphotyrosine antibodies. We were able to detect the phosphorylation of Jak2 only in R2 cells (Fig. 5A).

Tyrosine phosphorylation of Jaks upon IFN- treatment.

Inability to detect Jak2 activation in 16-9 cell lines responsive to Hu-IFN- containing the chimeric receptors R2/R1 and R2/CRF suggested involvement of other Jaks in these cells. Since the Hu-IFN-R1 associates with Tyk2 (Barbieri et al., 1994; Colamonici et al., 1994a, 1994b), we evaluated Tyk2 activation in cells expressing the chimeric R2/R1 chain. With anti-Tyk2 antibodies which were developed against human Tyk2, we were able to detect weak phosphorylation of a protein of ~130 kDa which was precipitated with anti-Tyk2 antibodies in the cells expressing R2/R1 and R2/CRF (Fig. 5B). When the same blot was reprobed with anti-Tyk2 antibodies, these phosphorylated proteins corresponded to a band recognizable by anti-Tyk2 antibodies and likely represented hamster Tyk2 (Fig. 5B, lower panels), but these proteins migrated slightly slower than human Tyk2 from lysates of COS-1 cells transiently transfected with a plasmid encoding human Tyk2 (data not shown). We observed the phosphorylated proteins of the same size in R2/R1 and R2/CRF cells when immunoprecipitation was performed with anti-phosphotyrosine antibodies and the blot was probed with anti-Tyk2 antibodies (Fig. 5C). The same differences in mobility of hamster Tyk2 from human Tyk2 from control lysates of COS-1 cells transiently transfected with a plasmid encoding human Tyk2 were observed (Fig. 5C).

To confirm that the phosphorylated protein in R2/R1 and R2/CRF cells was activated Tyk2, we stably cotransfected a plasmid encoding R2/R1 or R2/CRF with a plasmid encoding human Tyk2 into the 16-9 cells. The new cell lines were designated R2/R1+Tyk2 and R2/CRF+Tyk2, respectively. As controls we also cotransfected a plasmid encoding R2, R2/R1, or R2/R2 with a plasmid encoding human Tyk2 into the 16-9 cells. The resultant cell lines were designated R2+Tyk2, R2/R1+Tyk2, and R2/R2+Tyk2, respectively. First, we performed immunoprecipitation with anti-Tyk2 antibodies from cellular lysates prepared from R2/R1+Tyk2 and R2/R1 cells to evaluate the expression of exogenous human Tyk2 and endogenous hamster Tyk2. After blotting with anti-Tyk2 antibodies we observed that human Tyk2 was expressed at a much higher level or was able to be precipitated with the antibodies to a greater extent than hamster Tyk2. We also observed the same differences in mobility of human Tyk2 expressed in the 16-9 cells and hamster Tyk2 as we observed above with Tyk2 expressed in COS-1 cells (Fig. 5D). With these cell lines, we observed weak phosphorylation of Tyk2 in untreated R2/R1+Tyk2 and R2/CRF+Tyk2 cells as well as in untreated and treated R2+Tyk2, R2/R1+Tyk2, and R2/R2+Tyk2 cells. This is in agreement with the observation that overexpression of members of the Jak family causes a low spontaneous level of phosphorylation of kinases in the absence of ligand (Watling et al., 1993; Müller et al., 1993a; Silvennoinen et al., 1993a). However, only in R2/R1+Tyk2 and R2/CRF+Tyk2 cells did we observe an enhancement of Tyk2 phosphorylation after Hu-IFN- treatment (Fig. 5E). Thus, we showed that the intracellular domains of Hu-IFN-R1 and Hu-CRFB4 linked to the extracellular domain of R2 causes activation of Tyk2 instead of Jak2 which is regularly observed during IFN- signaling.

The chimeric receptor R2/_C was able to render 16-9 cells responsive to human IFN- to a small extent as measured by class I MHC antigen induction (Fig. 4H). It was shown that the IL-2R_C chain associates with Jak3 (Russell et al., 1994; Miyazaki et al., 1994; Tanaka et al., 1994). Thus, we examined whether Jak3 participates in the IFN- receptor complex in R2/_C cells. We hypothesized that failure of Hu-IFN- to induce strong MHC class I antigen induction is due to low level of endogenous Jak3 expression in the 16-9 hamster ovary cells since Jak3 is normally only expressed in hemopoietic cells. To test the hypothesis that the low level of Jak3 limited the IFN- signaling in R2/_C cells, we stably cotransfected a plasmid encoding R2/_C with a plasmid encoding Jak1, Jak2, Jak3, or Tyk2 into the 16-9 cells. Cell lines were designated R2/_C+Jak1, R2/_C+Jak2, R2/_C+Jak3, and R2/_C+Tyk2, respectively. Upon Hu-IFN- treatment, only R2/_C+Jak3 cells showed strong enhancement in MHC class I antigen induction upon Hu-IFN- treatment (Fig. 4I). All other cell lines showed the same level of responsiveness as R2/_C cells (Fig. 4H). We confirmed the participation of Jak3 in IFN- signaling in the R2/_C+Jak3 cells by immunoprecipitation experiments. Only in R2/_C+Jak3 cells did we observe enhancement in Jak3 phosphorylation after Hu-IFN- treatment. After longer exposure, a low spontaneous level of Jak3 phosphorylation was observed in untreated R2/_C+Jak3 cells and in treated and untreated R2+Jak3 cells used as a control similar to Tyk2 phosphorylation observed in cells overexpressing Tyk2 (see above). Thus, we showed that the R2/_C chimeric receptor recruits Jak3 into the IFN- receptor complex due to association of the IL-2R_C intracellular domain with Jak3, and that IFN- induces phosphorylation of Jak3 in R2/_C+Jak3 cells instead of phosphorylation of Jak2. Therefore, the substitution of the intracellular domain of _C for R2 accordingly substitutes Jak3 for Jak2 in the functional IFN- receptor complex.

Jak1 was shown to be activated upon IFN- treatment by the in vitro kinase activation assay only in cell lines positive in MHC class I antigen induction.²

IFN- Activates Stat1 through Jak Family Members Other Than Jak2

It was proposed that Jaks can contribute to the specificity of signal transduction by different IFNs, particularly, Tyk2 (which is active only during Type I IFN signaling) and Jak2 (which is active only during IFN- signaling) (Ihle et al., 1994). To evaluate this hypothesis, we performed electrophoretic mobility shift assays with cell lysates prepared from the various transformants before and after treatment with Hu-IFN- (Fig. 6). Since Stat1 homodimer formation occurs during IFN- signaling and Stat1 homodimers bind the GAS element with high specificity, we used oligonucleotides corresponding to the GAS element in the promoter region of the human IRF-1 gene as the phosphorylated probe (Yuan et al., 1994). We observed the formation of Stat1 DNA binding complexes (Fig. 6) in all cell lines responsive to Hu-IFN- as determined by induction of MHC class I antigen (Fig. 4). The Hu-IFN- induced activation of Stat1 in R2/_C+Jak3 cells was increased to the level comparable to the level of Stat1 activation in all other Hu-IFN- responsive cell lines (data not shown). Because it was shown that other Stats can bind the same GAS element with different affinity (Seidel et al., 1995), we performed supershift assays with specific anti-Stat1 antibodies to determine whether the GAS binding complexes observed in the tested cells are formed by Stat1 homodimers. The GAS binding complexes in all tested cells were shifted, indicating that Stat1 was activated in all Hu-IFN- responsive cells (Fig. 6).

DISCUSSION

Type I interferons (IFN- and IFN-) activate Jak1, Tyk2, Stat1, and Stat2 during signal transduction; Type II interferon (IFN-) uses Jak1, Jak2, and Stat1 for signaling (Velazquez et al., 1992; Watling et al., 1993; Müller et al., 1993a, 1993b; Leung et al., 1995). It was proposed that the utilization of different kinases by a particular receptor complex could contribute to the specificity of signaling by different IFNs (Ihle et al., 1994). To evaluate this hypothesis, we undertook a series of experiments with chimeric receptors.

As a model we used the IFN- receptor complex whose components and signal partners are now well defined. It has been shown that the active receptor complex consists of two chains of IFN-R1 (R1) and likely two chains of IFN-R2 (R2) oligomerized upon primary binding of the IFN- homodimer to the IFN-R1 (Greenlund et al., 1993; Langer et al., 1994; Marsters et al., 1995; Walter et al., 1995; Kotenko et al., 1995). IFN-R1 primarily associates with Jak1 (Igarashi et al., 1994; Sakatsume et al., 1995) and, perhaps, weakly with Jak2 after IFN-R1 homodimerization (Kotenko et al., 1995). The intracellular domain of IFN-R2 associates with Jak2 and brings Jak2 into the complex upon ligand binding (Kotenko et al., 1995; Sakatsume et al., 1995). The expression of a kinase negative Jak1 mutant in a Jak1 negative (U4A) cell line can sustain an IFN- response, indicating that Jak1 predominantly plays a structural role in the normal functional IFN- receptor complex rather than catalytical role (Briscoe et al., 1995). Thus PTK activity of Jak1 is dispensable for IFN- signaling. In contrast, the expression of a kinase negative Jak2 mutant in a Jak2 negative (2A) cell line cannot sustain an IFN- response (Briscoe et al., 1995), indicating that the PTK activity of Jak2 is absolutely necessary for IFN- signaling. Thus, we tested whether substitution of Jak2 by other kinases would change the specificity of signal transduction by IFN-. For this purpose we substituted the intracellular domains of other receptor chains for the intracellular domain of Hu-IFN-R2 to recruit kinases other than Jak2 into the IFN- receptor complex. The extracellular domain of Hu-IFN-R2 was fused to the transmembrane and intracellular domains of either the Hu-IFN-R1 (R2/R1), Hu-IFN-R1 (R2/R1), Hu-IFN-R2 (R2/R2), Hu-CRFB4 (R2/CRF), or Hu-IL-2R_C chain (R2/_C) (Fig. 1). The chimeric chains were expressed in 16-9 cells, Chinese hamster ovary cells expressing Hu-IFN-R1. By cross-linking (Kotenko et al., 1995) we showed that all chimeric receptors were expressed on the cell surface and were able to participate in formation of the extracellular IFN- receptor complex (Fig. 3). We then determined biological responsiveness of the cells to Hu-IFN- as measured by MHC class I antigen induction (Fig. 4). The chimeric receptors R2/R1, R2/R1, R2/CRF, and R2/_C rendered the 16-9 cells responsive to Hu-IFN- (Fig. 4, C, E, G, and H). In contrast, the chimeric receptor R2/R2 did not support signal transduction by Hu-IFN- (Fig. 4F). Finally, we investigated the activation of Jaks and Stats to determine whether the specificity of signal transduction is altered in cells expressing these chimeric receptors.

In R2/R1 cells we detected phosphorylation of Tyk2, but not Jak2 (Fig. 5, A, B, C, and E). This is in agreement with observation that the IFN-R1 (R1) intracellular domain associates with Tyk2 (Barbieri et al., 1994; Colamonici et al., 1994a, 1994b). Similarly, we showed that Hu-IFN- induces phosphorylation of Tyk2, but not Jak2 in R2/CRF cells (Fig. 5, A, B, C, and E). Thus, the intracellular domain of CRFB4 or Hu-IFN-R1 fused to the extracellular domain of Hu-IFN-R2 recruits Tyk2 into the IFN- receptor complex and allows Hu-IFN- to signal through activation of Tyk2 instead of Jak2, as usually occurs during IFN- signaling. Thus, the activation of Jak2 per se is not necessary for signal transduction by IFN- and Tyk2 can substitute for Jak2.

The R2/R1 chimeric chain brings an additional molecule of Jak1 into the IFN- receptor complex. Although we could not detect the phosphorylation of Jak2 in R2/R1 cells (Fig. 5A), there may be some involvement of Jak2 in signal transduction by IFN- in R2/R1 cells, arising through a weak interaction between dimerized IFN-R1 and Jak2 (Kotenko et al., 1995). There are two regions of the IFN-R1 intracellular domain important for signaling, the membrane proximal region, with a possible Jak1 association site (Farrar et al., 1991), and the membrane-distal region around Tyr-457 (Cook et al., 1992; Farrar et al., 1992), the Stat1 recruitment site (Greenlund et al., 1994). We showed that the presence of the Stat1 recruitment site on this chimeric receptor chain is not required for activity because R2/R1t₄₅₆ cells that also express the normal Hu-IFN-R1 chain are still responsive to Hu-IFN- (Fig. 4D).

To investigate further the ability of Jaks to substitute for each other we examined R2/_C cells. The IL-2R_C chain is a common chain for a broad range of receptor complexes for such cytokines as IL-2, IL-4, IL-7, IL-9 (for review, see Kishimoto et al. (1994)), and IL-15 (Giri et al., 1994). The intracellular domain of the IL-2R_C chain associates with Jak3 (Russell et al., 1994; Miyazaki et al., 1994; Tanaka et al., 1994). The expression of both the IL-2R_C chain and Jak3 are restricted to certain cell types (Minami et al., 1993; Johnston et al., 1994; Witthuhn et al., 1994). We observed MHC class I antigen induction in R2/_C cells in response to Hu-IFN- (Fig. 4H), but the induction was weaker than in all other cell lines expressing chimeric receptors (Fig. 4, B, C, E, and G). However, by co-expression of R2/_C together with Jak1, Jak2, Jak3, or Tyk2 we showed that only Jak3 was able to restore the biological responsiveness of R2/_C cells to Hu-IFN- (Fig. 4, H and I). We further showed that Jak3 is phosphorylated upon Hu-IFN- treatment of these cells (Fig. 5F). Thus, Jak3 can functionally substitute for Jak2 in the active IFN- receptor complex.

Since the formation of the Stat1 homodimeric DNA-binding complex as the only binding complex is specific for IFN- signaling, we investigated whether substitution of other kinases for Jak2 would change the specificity of the Jak-Stat signal transduction pathway. The formation of Stat1 DNA binding complexes was observed in all cell lines positive in MHC class I antigen induction (Fig. 6). By the supershift assay with specific anti-Stat1 antibodies we further showed that the GAS binding complexes consisted of Stat1 homodimers (Fig. 6). Thus, we concluded that other members of the Jak family recruited to the IFN- receptor complex can substitute for Jak2 without changing the specificity of IFN- signaling in the Jak-Stat pathway. That is, the activation of Stat1 during IFN- signaling does not require the specific participation of Jak2.

It was shown recently that there are two splice variants of the IFNAR2 gene: a short form, Hu-IFN-R2b (R2b) and a long form, Hu-IFN-R2c (R2c). They share the same extracellular and transmembrane domains, but the stop codon of the R2b is spliced out in the R2c cDNA resulting in a longer intracellular domain for R2c. Only the R2c variant can complement the mutation in U5A cells defective in IFN- signaling and sustain the IFN- response (Lutfalla et al., 1995); and together with the Hu-IFN-R1 chain can reconstitute an active human IFN- receptor complex in mouse cells (Domanski et al., 1995). Since only the R2b variant was known at the time of this work, we used the R2b intracellular domain. After we became aware of the existence of the R2c chain, we constructed chimera R2/R2c and expressed it in 16-9 cells. Hu-IFN- induced MHC class I antigens in R2/R2c cells to the same extent as in R2 cells.³ Thus, because we showed that the R2/R2b (R2/R2) chimera was unable to support IFN- signaling (Fig. 4F), we conclude that the intracellular domain of the short form of R2, R2b, is unable to bring PTK activity to the IFN- receptor complex in contrast to the R2c intracellular domain. The box 1 motif (proline-rich sequence) which is required for association with Jaks (Lebrun et al., 1995; Tanner et al., 1995) is not present in the short R2b chain (Domanski et al., 1995). This can explain the failure of R2b to complement U5A cells (cells lacking R2b and c) for IFN- signaling (Lutfalla et al., 1995), as well as the failure to render mouse cells responsive to Hu-IFN- (Domanski et al., 1995).

In addition to demonstrating that other Jaks can substitute for Jak2 in IFN- signal transduction, we provide here evidence that the intracellular domain of CRFB4, a class II cytokine receptor with unknown function (Lutfalla et al., 1993) associates with Tyk2. Therefore, CRFB4 is likely a component of a ligand-receptor system which activates Tyk2 during signal transduction. Tyk2 was shown to be activated by IFN-, ciliary neurotrophic factor-related cytokines, IL-10 and IL-12 (Velazquez et al., 1992; Barbieri et al., 1994; Lütticken et al., 1994; Stahl et al., 1994; Finbloom and Winestock, 1995; Bacon et al., 1995). Since the components of a given cytokine-receptor complex belong to the same class of the cytokine receptor superfamily, it is most likely that CRFB4 is involved in IFN- or IL-10 receptor complexes, as all cloned subunits of these receptors are members of the same class as CRFB4 (Uzé et al., 1990; Liu et al., 1994; Novick et al., 1994) unlike the components of the ciliary neurotrophic factor-related cytokine receptor subunits and the IL-12 receptor. Other evidence supports this hypothesis. The reconstitution of an active IFN- receptor has not been reported (Cohen et al., 1995).⁴ However, the yeast artificial chromosome containing CRFB4, Hu-IFN-R1, and Hu-IFN-R2 genes (Emanuel, 1995)⁵ encodes a functional Hu-IFN- receptor complex (Soh et al., 1994b; Cleary et al., 1994). It is thus enticing to consider that these three genes are all involved in Type I interferon receptor function. In the case of IL-10, only one chain (IL-10R) has been cloned (Liu et al., 1994). The activation of two different members of the Jak family during signal transduction usually requires the involvement of two different receptor components associated with two