The WD repeat-containing protein receptor for activated protein kinase C (RACK)-1 has been linked to a variety of signaling systems including protein kinase C, growth factors, and IFNs. In the IFN system, RACK-1 functions as an adaptor recruiting the transcription factor STAT1 to the receptor complex. However, RACK-1 should play a broader role in type I IFN signaling because mutation of the RACK-1 binding site in the IFN-α receptor 2/β subunit of the type I IFN receptor abrogates not only STAT1, but also STAT2, activation. In this study, we demonstrate that RACK-1 serves as a scaffold protein for a multiprotein complex that includes the IFN-α receptor 2/β-chain of the receptor, STAT1, Janus kinase 1, and tyrosine kinase 2. In vitro data further suggest that within this complex tyrosine kinase 2 is the tyrosine kinase responsible for the phosphorylation of STAT1. Finally, we provide evidence that RACK-1 may also serve as a scaffold protein in other cytokine systems such as IL-2, IL-4, and erythropoietin.
Cytokines and IFNs (1, 2, 3) activate kinases of the Janus kinase (Jak) 4 family and transcription factors designated as STAT (4, 5, 6, 7). Although there is some promiscuity in STAT activation by different cytokines (i.e., STAT1 is activated by IFN-α, IFN-γ, IL-6, LIF, IL-10, etc.), specific knockout mice models have demonstrated that the biological role of a distinct STAT is restricted to specific systems (for review see Ref. 8). For example, STAT1 is only specifically required for type I (IFN-α, -β, or -ω) and type II (IFN-γ) signaling (9, 10).
In most cytokine systems, activation of STATs through tyrosine phosphorylation requires their previous recruitment to distinct phosphotyrosines within the receptor subunits (reviewed in Refs. 7 , 8 , and 11). In the case of the type I IFNR, STAT2 is constitutively associated with the βL subunit (also designated as IFN-α receptor 2) in a phosphotyrosine-independent manner and has additional phosphotyrosine-dependent docking sites on the α and βL chain (12, 13, 14, 15, 16). Interestingly, full activation of STAT2 by type I IFNs requires the presence of at least two of these three docking sites (13). Activation of STAT1 by type I IFNs differs significantly from its activation by IFN-γ and the activation of STATs in general by other cytokines. For example, the adaptor protein receptor for activated protein kinase C (RACK)-1 (17, 18, 19) recruits STAT1 to the receptor complex and no specific STAT1-docking tyrosine has been identified within the α or β subunits of the receptor (13, 15, 20, 21). Moreover, the tyrosine phosphorylation of STAT1 requires the previous phosphorylation of STAT2 (22). Interestingly, mutation of the RACK-1 binding site of βL has an impact on type I IFN signaling that goes beyond activation of STAT1 because it also impairs activation/phosphorylation of STAT2. This finding raises the question as to whether RACK-1 recruits other signaling components to the receptor complex.
RACK-1 functions are not restricted to protein kinase C (PKC) or IFN signaling because RACK-1 interacts with Src homology 2 (SH2)-containing proteins such as src, phospholipase C γ, and ras-GTPase-activating protein (GAP) (23, 24), β integrins (25), PDE4D5 (26), the β common subunit of the GM-CSF/IL-3/IL-5 receptors (27), and insulin-like growth factor (IGF) receptor (28). Because RACK-1 is a WD repeat-contained protein with no enzymatic activity it has been proposed that it functions as a scaffold protein that recruits specific signaling elements. For instance, scaffold proteins bring together multiple components of the mitogen-activated protein kinase signaling (29, 30).
We sought to determine whether RACK-1, in addition to serving as an adaptor between the βL chain of the receptor and STAT1, was required for docking other components of the type I IFN receptor system such as Jak1 and tyrosine kinase (Tyk)2. We also endeavored to determine whether Jak1 and Tyk2 could be responsible for tyrosine phosphorylation of STAT1 and STAT2. Our findings indicate that RACK-1 directly interacts with Tyk2 and Jak1, however, only Tyk2 can phosphorylate STAT1 in vitro. Finally, we provide evidence that RACK-1 associates with other cytokine receptors such as erythropoietin, IL-2Rβ, and IL-4Rα, suggesting that RACK-1 could play a role in signaling by other cytokines.
Materials and Methods
Cell lines, reagents, and antiviral assays
U-266 and L-929 cells were grown in RPMI 1640 supplemented with 10% (v/v) FBS. Human IFN-α2 (specific activity 2.2 × 108 U/mg) was a gift of R. Bordens (Schering-Plough, Kenilworth, NJ). The anti-phosphotyrosine Ab 4G10 was purchased from Upstate Biotechnology (Lake Placid, NY) and the anti-RACK-1 (IgM), -STAT1, and -Jak1 mAbs were purchased from BD Transduction Laboratories (Lexington, KY). The anti-RACK1 Ab (IgG) used for the experiments described (see Fig. 2⇓C) was obtained from Santa Cruz Biotechnology (Santa Cruz, CA). The anti-STAT1 sera were kindly provided by Dr. A. Larner (Cleveland Clinic, Cleveland, OH).
Immunoprecipitation and immunoblotting
Cells (1 × 107 cells/immunoprecipitation) were treated as indicated, and then lysed in lysis buffer (20 mM Tris-HCl, pH 6.6, containing 1% Nonidet P-40, 50 mM NaCl, 1 mM EDTA, 2.5% glycerol v/v, 1.0 mM sodium fluoride, 1.0 mM sodium orthovanadate, 1.0 mM PMSF, 0.5 μg/ml leupeptin, and 5.0 μg/ml trypsin inhibitor) for 30 min at 4°C. Immunoprecipitations were performed as previously described (13). Proteins were transferred to polyvinylidene difluoride membranes, immunoblotted with the indicated Abs, and developed using a chemiluminescent detection method (Pierce, Rockford, IL).
GST-fusion proteins and mammalian expression construct
The different GST fusion proteins encoding the cytoplasmic domain of the α and βL subunits of the type I IFNR, IL-2Rβ, IL-4α, and GST-RACK-1 have been described previously (31, 32). GST-STAT1 corresponds to the full-length STAT1 sequence subcloned into pGEX-2T. For in vitro kinase assays, GST-STAT1 was eluted from the GSH-Sepharose beads using 10 mM glutathione and ∼2–4 μg/assay used as substrate. The amount of GST used per pull-down was estimated from gels stained with Coomassie blue and compared to BSA standards. Pull-down experiments and immunoblotting were performed using the same procedure described above for immunoprecipitations.
In vitro translation and in vitro kinase assays
For in vitro translation assays, proteins were produced using a T7 wheat germ in vitro transcription/translation kit (Promega, Madison, WI) following manufacturer’s procedure. [35S]Methionine proteins were incubated with the indicated GST fusion proteins overnight, washed, and analyzed by SDS-PAGE as described for immunoprecipitations. In vitro kinase assays were performed after immunoprecipitation with the indicated Abs as previously described (32).
RACK-1 interacts with Jak1 and Tyk2
We have previously reported that RACK-1 interacts with the nonphosphorylated form of STAT1 (21). Fig. 1⇓ confirms this finding because GST-RACK-1 pulls down only nonphosphorylated STAT1 from lysates of either control or cells treated with 5000 U/ml of IFNα2 (Fig. 1⇓, lower panel, lanes 3 and 7), while the anti-STAT1 Ab detects the phosphorylated (Fig. 1⇓, upper panel, lane 4) and nonphosphorylated (Fig. 1⇓, lower panel, lanes 4 and 8) STAT1. As expected, a GST encoding the long form of the β subunit of the type I IFNR also pulls down the nonphosphorylated form of STAT1 (Fig. 1⇓, lower panel, lanes 2 and 6) through its interaction with RACK-1. Interestingly, this experiment also demonstrated that GST-RACK-1 and GST-βL pulled down tyrosine-phosphorylated proteins (Fig. 1⇓, upper panel, asterisks) with electrophoretic mobility similar to Jak1 and Tyk2 from lysates obtained from IFN-treated cells. Additionally, only GST-RACK-1 interacted with a tyrosine-phosphorylated protein with an approximate molecular mass of 100 kDa (lane 3, arrow) that resembles the electrophoretic mobility of the β-chain of the receptor. However, the identity of this protein has not been confirmed due to heavy reactivity of the anti-βL serum (raised using GST-βL as Ag) with the GST part of GST-RACK-1 (data not shown).
To determine whether two of the phosphoproteins associated with RACK-1 corresponded to Jak1 and Tyk2, we performed new pull-down experiments using cell lysates obtained from control and IFN-β-treated U-266 cells. Fig. 2⇓A shows that the protein pulled down by GST-RACK-1 (Fig. 2⇓A, middle panel, lanes 2 and 6) is recognized by an anti-Jak1 Ab and has the same electrophoretic mobility as the Jak1 protein immunoprecipitated with an anti-Jak1 Ab (Fig. 2⇓A, middle panel, lanes 4 and 8). Unlike the interaction with STAT1, the association of Jak1 and RACK-1 is not affected by IFN-β treatment. The level of tyrosine-phosphorylated Jak1 associated with RACK-1 was slightly lower in this experiment but consistently present. As expected, GST-βL also pulled down Jak1 as previously reported (31). As observed in Fig. 1⇑, GST-RACK-1, GST-βL, and in some experiments the anti-Jak1 Ab (Fig. 2⇓A, lane 8) pulled down a tyrosine-phosphorylated protein with a slightly slower electrophoretic mobility that resembled Tyk2 (Fig. 2⇓A). Fig. 2⇓B shows that indeed the tyrosine phosphorylated and nonphosphorylated forms of Tyk2 were pulled down by GST-RACK-1 (lanes 3 and 8).
We next determined whether an interaction between endogenous RACK-1 and Jak1 or Tyk2 could be detected by coimmunoprecipitation. Fig. 2⇑C shows that indeed the anti-Jak1 and -Tyk2 Abs (Fig. 2⇑C, lanes 2 and 3) can coprecipitate RACK-1 indicating that the interaction also occurs in vivo. These data suggest that RACK-1 not only interacts with βL and STAT1 as previously reported (21), but also with Jak1 and Tyk2.
RACK-1 interacts with different cytokine receptors
RACK-1 has been reported to associate with other cytokine and growth factor receptors such as the β common chain of the GM-CSF/IL-3/IL-5 receptors and IGF receptors (27, 28). Because RACK-1 interacts with the βL chain, STAT1, Jak1, and Tyk2, we next examined whether it could also interact with the α-chain of the type I IFNR as well as other cytokine receptors. GST fusion proteins for type I IFNR α-chain, IL-2Rβ and γ, IL-4Rα, and erythropoietin receptor (EPOR) were used to pull-down RACK-1 from cell lysates. Fig. 3⇓ shows that RACK-1 interacts with the α-chain of the type I IFNR (Fig. 3⇓, lane 7), although the interaction appears to be weaker than with the β-chain (Fig. 3⇓, lane 8). Interestingly, RACK-1 also associated with the IL-2Rβ chain (Fig. 3⇓, lane 5), a GST-IL-4Rα that encompasses residues 283–429 (Fig. 3⇓, lane 3), and with the EPOR (Fig. 3⇓, lane 14). However, two other regions of the IL-4Rα (amino acids 209–288 and 429–561, Fig. 3⇓, lanes 2, and 4), IL-2Rγ (Fig. 3⇓, lanes 6 and 11), or GST control (Fig. 3⇓, lane 10) failed to interact with RACK-1. These data demonstrate that RACK-1 interacts specifically with both subunits of the type I IFNR and also suggest that it may play a role in signaling by other cytokine receptors such IL-2, IL-4 and EPORs.
RACK-1 interacts directly with Tyk2, Jak1, and IFN-αRβL, and indirectly with the α-chain of the type I IFNR
The finding that RACK-1 associated with proteins that are part of a large receptor complex (both receptor subunits, STAT1, Tyk2, and Jak1) raised the possibility that some interactions may not be direct but rather mediated by other components of the complex. For example, because both Jak1 and RACK-1 associate with the βL chain, GST-RACK-1 could pull-down Jak1 through the interaction of the kinase with βL. To address this question, RACK-1 was produced using a wheat germ in vitro translation system, which did not contain Jak kinases or IFNR homologs, and pull-down experiments were performed with GST-RACK-1, GST-α, and GST-βL fusion proteins. As previously reported (21), in vitro-translated RACK-1 associates with GST-βL (Fig. 4⇓A, lane 3) but fails to interact with GST-α or GST control (Fig. 4⇓A, lanes 1 and 2). These results suggest two possibilities: 1) the interactions of the α-chain with RACK-1 (Fig. 4⇓A) may not be direct but rather mediated by other proteins within the receptor complex, or 2) the α-chain has a lower affinity for RACK-1 that masks the detection of the association under these experimental conditions.
We also studied the nature of the interaction between RACK-1 and Jak1 and Tyk2 using a wheat germ in vitro translation system. Fig. 4⇑B shows that GST-RACK-1 interacts with in vitro translated Jak1 and Tyk2 (Fig. 4⇑B, lanes 4 and 9) suggesting that the interaction between RACK-1 and both Jaks is direct.
Surprisingly, the GST-α chain, which did not interact with RACK-1, also failed to pull-down significant amounts of Tyk2 produced in wheat germ in vitro translation systems (Fig. 4⇑B, lanes 2 and 7). The association of the α-chain and Tyk2 was previously described using as sources of Tyk2 mammalian cell lysates, rabbit reticulocyte in vitro translation systems and baculovirus extracts (33, 34, 35). To determine whether there were differences in the interaction between the α-chain and Tyk2 that depended on the system used, we compared in parallel the association between the α-chain and Tyk2 using as sources of the kinase either cell lysate or wheat germ or rabbit reticulocyte lysate in vitro translation systems. Fig. 4⇑C shows that indeed GST-α associates with Tyk2 produced by in vitro translation in rabbit reticulocyte lysates and with Tyk2 present in cell lysates (Fig. 4⇑C, lanes 7 and 12), but not when the kinase is produced in a system likely to be devoid of other proteins involved in cytokine signaling such as wheat germ lysates (Fig. 4⇑C, lane 2). These data suggest that the association of the α-chain of the type I IFNR with Tyk2 is mediated by unidentified proteins that are not present in wheat germ systems (see Discussion).
Of note, Fig. 4⇑ demonstrates that there is a direct interaction between βL and Tyk2 (Fig. 4⇑, B and C, lanes 8 and 3, respectively). This suggests that there is some degree of promiscuity in the Jak binding site of βL similar to that previously reported for the gp130 family of receptors (36). Alternatively, βL may independently bind either of these two kinases as previously reported for the IL-2Rβ chain (37) (see Discussion). Altogether, these data suggest that RACK-1 may be the central docking element in a complex that involves Jak1, Tyk2, STAT1, and βL.
Tyk2 phosphorylates STAT1 in vitro
The finding that RACK-1 can associate with STAT1, Tyk2, and Jak1, raises the question as to which kinase phosphorylates this STAT factor. To address this question, purified GST-STAT1 was used as a substrate in immunocomplex in vitro kinase assays after immunoprecipitation with either anti-Tyk2 or anti-Jak1 Abs. We added to the reaction a STAT2 peptide containing phospho or nonphosphotyrosine 691. The rationale for this approach was 2-fold: first, phosphorylation of STAT1 is preceded by phosphorylation of STAT2 and may require the previous binding of phosphotyrosine 691 of STAT2 to the SH2 domain of STAT1. Second, RACK-1 interacts with a region of STAT1 that involves part of the SH2 domain of STAT1 and the phosphopeptide may participate in exposing the target tyrosine of STAT1 (data not shown). Fig. 5⇓ shows that GST-STAT1 is phosphorylated after IFN treatment only in immunocomplexes that include Tyk2. Phosphorylation was not clearly improved by the addition of the phospho-STAT2 peptide in this experiment, although in others there was a slight increase (data not shown). This result demonstrates that at least in in vitro systems Tyk2, but not Jak1, can phosphorylate STAT1, raising the possibility that Tyk2 may also phosphorylate STAT1 in vivo. Unfortunately, we have not succeeded in producing a suitable GST-STAT2 to determine which kinase was responsible for STAT2 phosphorylation.
Adaptor or scaffold proteins have been shown to have an important role in signaling cascades by contributing to the assembly of specific signaling complexes (29, 30). Although RACK-1 was originally described as a receptor for activated PKC-β, it has recently become clear that this scaffold protein has functions that exceed PKC signaling. RACK-1 appears to be important in a diverse group of signaling pathways such as src kinases, cAMP, integrins, and cytokines including type I IFNs (20, 21, 25, 26, 27). In the type I IFN system, RACK-1 associates with the β-chain of the receptor and recruits STAT1 to the receptor complex. However, there is evidence supporting a broader role for RACK-1 in type I IFN signaling. The most compelling argument is that mutation of the region of the βL chain containing the RACK-1 binding site not only abolishes STAT1 tyrosine phosphorylation, but also phosphorylation of STAT2. This finding suggests that RACK-1 is either necessary for the appropriate receptor configuration that allows STAT2 phosphorylation by the Jaks or that RACK-1 may recruit other signaling proteins to the receptor complex that are required for STAT2 activation. We explored here the hypothesis that the scaffolding function of RACK-1 is not limited to the recruitment of STAT1 but rather involves other components of the type I IFNR complex. Our results demonstrate that in addition to the βL chain and STAT1, RACK-1 interacts with Tyk2, Jak1, and the α-chain of the receptor present in mammalian cell lysates. However, unlike the interaction between RACK-1 and STAT1, RACK-1 still associates with phosphorylated Jaks.
The finding that the α and β subunits interact independently with the Jaks and RACK-1 raised the question as to whether RACK-1 was serving as an adaptor protein or simply as one more protein within the multiprotein complex. If RACK-1 indeed functions as a scaffold protein, it must directly interact with different components of the complex directly. Experiments performed with the receptor subunits and Tyk2 and Jak1 proteins produced in wheat germ in vitro translation system, which should be devoid of IFNR subunits and Jaks, demonstrated that RACK-1 interacts directly with Tyk2 and Jak1. However, we did not detect a direct interaction between RACK-1 and the α-chain of the receptor using a wheat germ lysate system. This would explain that GST-RACK-1 pulled down a lower amount of the α-chain than βL from cell lysates (Fig. 3⇑) and that we have not been able to detect RACK-1 in coimmunoprecipitations with the α-chain (20). Alternatively, RACK-1 could have lower affinity for the α-chain than for βL, and the low amounts of RACK-1 produced in the wheat germ in vitro translation system are not enough to detect the association.
The scaffolding function of RACK-1 within the receptor complex may explain why disruption of the interaction between RACK-1 and βL results in deficient activation of not only STAT1, but also STAT2 (21). In this scenario, βL, which directly and through phosphotyrosines docks STAT2, may not make STAT2 available for phosphorylation by Jaks. Interestingly, specific measles virus proteins impede IFN-α signaling by preventing the association of RACK-1 with STAT1, and the activation of Jak1, supporting the concept that RACK-1 is critical for the phosphorylation of STAT1 and STAT2 by the appropriate kinases (38, 39).
The finding that RACK-1 associated with Jak1 and Tyk2 also raised the question as to which kinase was responsible for the phosphorylation of STAT1. Immunocomplex kinase assays using as a substrate GST-STAT1 revealed that Tyk2, but not Jak1, was able to phosphorylate STAT1 in vitro. Although the conditions in in vitro kinase assays do not always reflect the in vivo scenario, the finding that Jak1 was unable to phosphorylate STAT1 in vitro weakens a possible role of this kinase in STAT1 phosphorylation. Unfortunately, we have not been able to produce a form of STAT2 suitable for in vitro kinase assays to determine whether Jak1 is responsible for the phosphorylation of this factor. However, it was previously reported that mouse embryonic fibroblasts null for the α-chain that were stably transfected with the human βL chain were able to show low levels of phosphorylation of Jak1 and STAT2, but not Tyk2 or STAT1 suggesting that Jak1 could be responsible for STAT2 phosphorylation (13). Altogether, these results would support a model in which Tyk2 and Jak1 are responsible for the phosphorylation of STAT1 and STAT2, respectively.
An unexpected finding was that the α-chain of the receptor did not show a significant association with Tyk2 produced in wheat germ lysate systems. This lack of interaction is not due to folding problems of either in vitro-translated Tyk2 or bacterially produced GSTα chain because the same wheat germ preparation of the kinase was able to interact with the β-chain and RACK-1 (Fig. 4⇑, B and C), and the same GSTα chain pulled down Tyk2 produced in rabbit reticulocyte lysates or mammalian cells (Fig. 4⇑C). These data suggest that the interaction between Tyk2 and the α-chain requires an intermediary protein. These results do not necessarily contradict previous reports describing the association between the α-chain and Tyk2 because those experiments were performed using Tyk2 expressed in mammalian cells or produced in baculovirus systems (33, 34, 35).
Interestingly, our results also revealed that the β-chain itself could interact with Tyk2. A possible explanation could be that there is some degree of flexibility between the kinase binding sites of βL and the conserved regions of the kinases that interact with cytokine receptors. This scenario would not be very different from that reported for the gp130 subunit of the IL-6R which can associate with Jak1, Tyk2, or Jak2 (36). Alternatively, the βL subunit has the ability to interact independently with both kinases as previously reported for the β-chain of the IL-2R (37).
Finally, it was not surprising that RACK-1 also interacted with other cytokine receptors such as the IL-2Rβ, IL-4Rα, and EPOR that it has been previously reported to participate in signaling by other growth factor receptors such as the IGF-1 and IL-3/IL-5/GM-GSF receptors (27, 28). Therefore, RACK-1 may represent a modular protein whose function as a scaffold goes beyond its initial description as a receptor for activated PKCβ. There are many questions regarding the role of RACK-1 that have not been addressed. For instance, do all these systems absolutely require RACK-1 for signaling? Are there other WD repeat-containing proteins that can substitute or function redundantly and therefore, complement the absence of RACK-1? So far, experiments using siRNA have not produced a decrease large enough in RACK-1 levels that would allow us to answer these questions (our unpublished observation).
↵1 This work was supported by National Institutes of Health Grants CA55079 and GM54709 (to O.R.C.).
↵2 A.U. and X.T. contributed equally to this manuscript.
↵3 Address correspondence and reprint requests to Dr. Oscar R. Colamonici, Department of Pharmacology (M/C868), University of Illinois, 835 South Wolcott Avenue, Room E403, Chicago, IL 60612. E-mail address:
↵4 Abbreviations used in this paper: Jak, Janus kinase; RACK, receptor for activated protein kinase C; SH2, Src homology 2; IGF, insulin-like growth factor; Tyk, tyrosine kinase; PKC, protein kinase C; EPOR, erythropoietin receptor.
- Received April 14, 2003.
- Accepted July 16, 2003.
- Copyright © 2003 by The American Association of Immunologists