A single STAT recruitment module in a chimeric cytokine receptor complex is sufficient for STAT activation.

We established a system of receptor chimeras that enabled us to induce heterodimerization of different cytoplasmic tails. Fusion constructs were created that are composed of the extracellular parts of the interleukin-5 receptor α and β chains, respectively, and the transmembrane and intracellular parts of gp130, the signal transducing chain of the interleukin-6 receptor complex. In COS-7 transfectants we observed a dose-dependent interleukin-5-inducible STAT1 activation for which the presence of both the α and the β chain chimera was needed. No STAT activity was detected if one of the cytoplasmic tails of the receptor complex was deleted, indicating that STAT activity resulted from a receptor dimer rather than from higher receptor aggregates. We further investigated whether dimerization of STAT1 depends on the juxtaposition of two STAT recruitment modules in a receptor complex. We show that a receptor dimer with only a single STAT1 docking site was still able to lead to STAT1 activation. This indicates that the formation of a paired set of STAT binding sites in a receptor complex is not the prerequisite for STAT factor dimerization. Our findings are discussed in view of alternative STAT dimerization models.

A detailed analysis of interferon (IFN) 1 signaling events first provided insight into a general signaling mechanism, the Jak-STAT pathway, by which many cytokines lead to an altered pattern of gene expression. Jak (Janus kinase) refers to a family of cytoplasmic tyrosine kinases that comprises four known members in mammals: Jak1, Jak2, Jak3, and Tyk2 (1). STAT (signal transducers and activators of transcription) refers to a family of transcription factors with seven known members (2). Cytokine-induced dimerization of receptor components leads to the activation of Jaks, which are constitutively associated with the cytoplasmic parts of the respective receptors. One substrate of the Jaks is the receptor itself. Upon phosphorylating specific tyrosine residues of the cytoplasmic tail of the receptor STAT factors and other proteins with "matching" SH2 domains can be recruited to the receptor where they become activated by tyrosine phosphorylation. Subsequently, the STATs dissociate from the receptor and translocate as homo-or heterodimers to the nucleus where they bind to enhancer elements of target genes and influence transcriptional activity (2,3).
The events leading to STAT factor dimerization are not well understood. Due to receptor dimerization, many cytokines lead to the formation of a paired set of STAT docking sites, e.g. IFN␥ signals through a homodimer of STAT1 that binds to ␥-interferon-activated sequence elements of IFN␥-regulated genes (4,5). A single tyrosine residue (Tyr-440) in the ␣ chain of the IFN␥ receptor serves as a docking site for STAT1 (6 -8). Since IFN␥ receptor ␣ chains dimerize upon binding of IFN␥ (9), phosphorylation of Tyr-440 forms two juxtaposed docking sites for latent STAT1. Dimerization of STAT1 monomers phosphorylated on a specific tyrosine residue (Tyr-701) might be favored by the presence of another phosphorylated STAT1 monomer in the near proximity (7). In the present study we tested whether the formation of a STAT1 homodimer depends on the presence of two STAT1 docking sites in a dimerized receptor. For this reason we established a system of receptor chimeras based on the extracellular parts of the human interleukin-5 (IL-5) receptor ␣ and ␤ chains that enabled us to induce heterodimerization of different cytoplasmic tails. We show that a receptor dimer with only a single STAT1 docking site is still able to lead to activated STAT1 dimers, indicating that close juxtaposition of STAT binding sites in a receptor complex is not the prerequisite for STAT factor dimerization.

MATERIALS AND METHODS
Reagents-Human IL-5 was expressed in Sf9 insect cells and purified as described previously (10). 125 I-hIL5 was prepared with the IODO-GEN iodination agent (Pierce) as described (11). The eukaryotic expression vector pSVLgp130 was generously supplied by Dr. T. Taga (Osaka, Japan). For flow cytometry the monoclonal antibody 16-4 specific for the human IL-5R␣ chain 2 and the monoclonal antibody S-16 specific for human ␤c chain (Santa Cruz Biotechnology, Santa Cruz, CA) were used. Phycoerythrin-labeled goat anti-mouse Ig (FabЈ) 2 was obtained from Dianova (Hamburg, Germany).
Construction of the Chimeric Molecules-For amplification of the cDNA encoding the extracellular part of the human IL-5R␣ chain and human ␤c the primers ccgctcgagccaccATGgTCATCGTGGCGCATG (␣ sense), cggaattcATTTCCCACATAAATAGGTTG (␣ antisense), gctcta-gagccaccATGGTGCTGGCCCAGGGGCTG (␤ sense), and cggaattcGGT-GTCCCAGGAGCGCGCC (␤ antisense) were used. Uppercase letters indicate cDNA sequence; underlined are restriction sites for XhoI, EcoRI, and XbaI, which were added to facilitate cloning. The IL-5R␣ * This work has been supported by the Deutsche Forschungsgemeinschaft, Bonn, Germany and by Fonds der Chemischen Industrie, Frankfurt am Main, Germany. 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.
‡ To whom correspondence should be sent: Inst. of Biochemistry, RWTH Aachen, Pauwelsstrasse, 52057 Aachen, Germany. chain sense primer was designed to provide a "better" Kozak consensus sequence. Therefore a point mutation was introduced, leading to a valine instead of an isoleucine residue at position 2 of the signal peptide of IL-5R␣/gp130. Polymerase chain reactions were performed with 20 ng of plasmid DNA, 20 pmol of each primer, and 1 unit of Vent polymerase (New England Biolabs). The polymerase chain reaction products were digested with XhoI/EcoRI or XbaI/EcoRI, respectively, and inserted into pSVLgp130 cut with the same enzymes, thus replacing the cDNA for the extracellular region of gp130 with the one of the IL-5R ␣ or ␤ chain. The resulting plasmids were named pSVL-IL-5R␣/gp130 and pSVL-IL-5R␤/gp130, respectively. Deletion constructs IL-5R␣/ gp130⌬cyt and IL-5R␤/gp130⌬cyt lack the gp130 cytoplasmic region apart from the first three residues (NKR). Due to the cloning procedure, they contain additionally three vector-encoded residues (IQT) before termination. The construction of chimeric molecules Eg (EpoR/gp130), Eg/⌬B, and Eg/Tyr-␥440 has been described (12). Eg/⌬B includes only the box 1 and 2 regions of gp130 and therefore lacks the five carboxylterminal tyrosine residues. Eg/Tyr-␥440 contains additional 7 amino acid residues from the human IFN␥ receptor (including Tyr-440). Eg/⌬B and Eg/Tyr-␥440 end with a FLAG epitope. From those expression constructs EcoRI/BamHI fragments were inserted into EcoRI/BamHIdigested pSVL-IL-5R␣/gp130 and pSVL-IL-5R␤/gp130, thereby replacing the intracellular part of gp130. Cell Transfections-Simian kidney cells (COS-7, ATCC CRL 1651) were cultured in Dulbecco's modified Eagle's medium (Life Technologies, Inc., Eggenstein, Germany) supplemented with 10% fetal calf serum, 100 mg/liter streptomycin, and 60 mg/liter penicillin. Cells were grown at 37°C in a water-saturated atmosphere with 5% CO 2 . 10 7 COS-7 cells were transiently transfected with 20 -30 g of plasmid DNA by electroporation (Gene Pulser, Bio-Rad). Electroporations were performed at 960 microfarads and 230 V. Cells were harvested 48 -72 h post transfection.
Flow Cytometry-COS-7 cells were released from the dishes by treating them with phosphate-buffered saline, 10 mM EDTA at 37°C for 10 min. Approximately 10 6 cells were incubated with monoclonal antibody 16 -4 or S-16 for 30 min, followed by treatment with the secondary antibody (goat anti-mouse IgG phycoerythrin-conjugated F(abЈ) 2 fragment, Dianova, Hamburg, Germany). Fluorescence was measured on a FACScan (Becton Dickinson).
Binding Studies-Upon transfection approximately 10 5 COS-7 cells were seeded into 24-well plates and incubated for 2 days until they reached 80 -90% confluence. Cells were washed twice with cold binding medium (0.2% bovine serum albumin, 20 mM HEPES buffer (pH 7.0) in Dulbecco's modified Eagle's medium). 200 l of binding medium with 125 I-IL-5 (10 nM and 2-fold dilutions thereof) were added for 2 h on ice in the presence or absence of 100-fold excess of cold IL-5. Cells were washed three times with cold phosphate-buffered saline containing 1 mM MgCl 2 , 0.1 mM CaCl 2 , and 0.2% bovine serum albumin. They were lysed in 1 M NaOH overnight before cell-associated radioactivity was measured in a ␥ counter. Specific binding was calculated as the difference between the binding in the absence (total binding) and presence
(nonspecific binding) of unlabeled IL-5. The dissociation constant (K d ) was calculated by Scatchard analysis of the binding data.

IL-5 Stimulates STAT Activation in COS-7 Cells
Coexpressing IL-5R␣/gp130 and IL-5R␤/gp130 -For the establishment of the system of heterodimeric receptor chimeras, we first made hybrid molecules composed of the extracellular parts of the human interleukin-5 receptor (IL-5R) ␣ and ␤ chains, respectively, and the transmembrane and the intracellular regions of gp130, the signal transducing receptor chain for IL-6-type cytokines. Upon transfection of COS-7 cells with expression plasmids for IL-5R␣/gp130 and IL-5R␤/gp130, surface expression of the chimeric molecules could be observed in a fraction of the cells. Mock-transfected COS-7 cells were not stained with antibodies recognizing human IL-5R ␣ or ␤ chains (Fig. 1A). Affinity of these chimeric constructs is similar to that seen with wild type IL-5 receptors expressed in COS-1 cells (15,16); IL-5R␣/gp130 bound IL-5 with a low affinity (K d ϭ 1 nM), and coexpression of IL-5R␤/gp130 led to a 2-fold increase of binding (Fig. 1B).
Dimerization of gp130 induced by cytokines (17) or agonistic antibodies (18) leads to rapid activation of STAT factors in a variety of cell lines (19,20). We now tested whether the heterodimeric chimeric receptors have signaling ability. COS-7 transfectants expressing IL-5R␣/gp130 and IL-5R␤/gp130 were stimulated (30 min, 37°C) with increasing amounts of recombinant human IL-5; nuclear extracts were prepared and tested in an EMSA. Stimulation with 4 ng of IL-5/ml already led to a signal, and the intensity of the gel shift bands increased dosedependently until at 80 ng/ml a maximum was reached (Fig.  1C). As shown in Fig. 1D, the shifted band had the same mobility as but a lower intensity than the one observed upon homodimerization of EpoR/gp130 chimeras which have previously been shown to lead to STAT1 activation in COS-7-transfectants (12,21). Importantly, expression of either the IL-5R␣/ gp130 or the IL-5R␤/gp130 chimera alone did not lead to IL-5inducible STAT activation (Fig. 1D). Therefore, both the ␣and the ␤-chain chimera contribute to the signaling ability of the receptor complex.
No Evidence for a Further Ligand-induced Receptor Aggregation-The stoichiometry of the components of the IL-5 receptor is not clear (see "Discussion"). The simplest model of a functional IL-5 receptor system is a dimer comprising one ␣ chain and one ␤ chain, but further receptor chain multimerization cannot be excluded (e.g. Fig. 2A, left panel, see also Ref. 22). Therefore, we investigated whether the observed STAT activity might have resulted from receptor aggregates (see Fig.  2A, right panel).
A Single Tyrosine Module in a Cytokine Receptor Complex Is Sufficient for STAT Activation-We have shown previously that a "tyrosine module" from the interferon ␥ receptor mediated a specific activation of STAT1 when fused to the membrane proximal box 1/2 region of gp130 (12,21). In that study chimeric receptors with the extracellular part of the erythropoietin receptor were applied which homodimerized upon stim-ulation with erythropoietin.
We then replaced the extracellular region of the erythropoietin receptor by the IL-5R ␣ and ␤ chains resulting in constructs IL-5R␣/Tyr-␥440 and IL-5R␤/Tyr-␥440 carrying a 7-amino acid tyrosine module of the IFN␥ receptor distal from the box 1/2 region of gp130. The constructs IL-5R␣/⌬B and IL-5R␤/⌬B contain the gp130 box 1/2 region but no STAT recruiting tyrosine module. IL-5 stimulation of the two Tyr-␥440 constructs led to a strong STAT1 activation (Fig. 3A), whereas no STAT1 activation was observed upon dimerization of the two ⌬B constructs (Fig. 3D). Importantly, dimerization of a single Tyr-␥440 with a ⌬B construct resulted in a clear STAT1 activation (Fig. 3, B and C). Therefore, a single tyrosine module in a cytokine receptor complex is sufficient for STAT1 activation. This observation indicates that close contact of STAT factors by juxtaposition of STAT binding sites at dimerized receptors is not necessary for dimerization.
FIG. 2. STAT activation results from receptor dimers rather than from higher aggregates. A, schematic representation of the experimental approach. If the STAT1 activation observed in cotransfectants with the full-length IL-5R/gp130 constructs resulted in part from further receptor aggregations (e.g. ␣␤␤, ␣␣␤, ␣␣␤␤) signaling could be expected to occur even upon deletion of one of the two cytoplasmic chains, because in such a case still two gp130 cytoplasmic tails would be brought into close proximity what is generally believed to be necessary for signal transduction. B, abrogation of signaling upon deletion of one cytoplasmic region in the heterodimeric receptor complex. COS-7 cells expressing the indicated combinations of chimeric receptors were stimulated with IL-5 or left untreated. STAT activity in nuclear extracts was tested in an EMSA using the 67mSIE probe.

DISCUSSION
It was our intention to study whether a single STAT module in a receptor complex is sufficient to achieve STAT activation. For this purpose, an "asymmetric" receptor complex was created with chimeric receptors based on the extracellular parts of the IL-5 receptor. The IL-5 receptor complex is composed of the ligand-binding IL-5R ␣ subunit and the common ␤ receptor chain (␤c) that is shared with the ␣ receptors for IL-3 and GM-CSF and functions as an affinity converter (Ref. 15; reviewed in Ref. 23). Members of this receptor family have already been successfully employed for the construction of chimeric molecules by others (24 -26). In this study we generated fusion constructs of the extracellular parts of the IL-5R ␣ and ␤ chain, respectively, and the transmembrane and intracellular region of gp130, the signal transducing chain of the IL-6 receptor. In COS-7 transfectants expressing the IL-5R␣/gp130 and the IL-5R␤/gp130 chimera, we observed an IL-5-inducible STAT activation (Fig. 1). Although "STAT activation" is certainly only one of several possible experimental read-outs to characterize the chimeras functionally, it indicates that they can be used to mimic signaling events involving gp130 dimers just like chimeric receptors that homodimerize upon ligand binding (e.g. via the extracellular region of receptors for erythropoietin, granulocyte colony-stimulating factor, epidermal growth factor, or neurotrophin-3; Refs. 12, 27, and 28).
Active receptor complexes depend on the presence of both IL-5R␣ and ␤ chimeric molecules. Expression of the IL-5R␣/ gp130 alone did not yield an IL-5-inducible signal. This was expected since cross-linking analyses as well as studies in solution indicated that IL-5, although itself a homodimer, binds only to one IL-5R ␣ chain (29 -32). Similarly, a chimeric molecule consisting of the extracellular domain of murine IL-5R␣ and the transmembrane and intracellular regions of ␤c could mediate proliferation of transfectants only in the presence of a functional ␤c chain (25).
The simplest model of the active IL-5 receptor complex consists of a heterodimer of one ␣ and one ␤ chain, but there are some reports that members of the IL-3/IL-5/GM-CSF receptor family might undergo further multimerization. 1) The cytoplasmic tail of the ␤c chain has signaling capacity when dimerized upon triggering of chimeric receptor constructs (25,33,34); 2) constitutively active mutants of the ␤c chain could be isolated (35)(36)(37); 3) it was suggested that a ligand-bound active receptor complex might normally contain a ␤c chain dimer (37); and 4) the occurrence of ligand-independent ␤c homodimers has recently been reported (38). In addition, unusual affinities were observed when GM-CSF receptor ␣ and ␤ subunits were expressed in COS-1 cells. These data have been discussed in view of possible complex variability (39).
We therefore expressed IL-5R␤/gp130 together with an IL-5R␣/gp130 construct lacking the whole cytoplasmic region (and vice versa) in COS-7 cells. If multimerization would occur, the proximity of at least two cytoplasmic tails could be expected to result in the activation of Jaks and STATs. However, since such receptor complexes did not signal, we conclude that, at least under our experimental conditions, the STAT activity we observed upon IL-5 stimulation is very likely to result from an ␣/␤ receptor dimer rather than from higher receptor aggregates. We cannot exclude, however, the possibility that potential multimers do not signal due to an "incorrect" orientation of their cytoplasmic tails.
The measured affinities of the IL-5R/gp130 chimeric receptors are comparable to those of the wild type IL-5 receptor in COS-1 transfectants (15,16). However, this affinity is lower than the one observed on HL-60 eosinophilic cells (11). Maybe additional, so far unknown receptor components contribute to the high binding affinity in these cells. A potential "␥"-subunit of the GM-CSF receptor has recently been suggested to be present in certain cell types (37). We cannot rule out the possibility that the lack or the presence of unknown receptor components in the COS-transfectants might also have influenced the signals we observed. Additionally, we cannot exclude a potential cross-talk between different receptor systems (such as described for the stem cell factor receptor and the erythropoietin receptor; Ref. 40) that might contribute to the responses we observed.
The mechanisms leading to STAT activation are not well understood. Although STAT activation independent of receptor tyrosine residues has been described (41,42), the importance of tyrosine residues within the cytoplasmic tail of cytokine receptors has been demonstrated in many reports (e.g. Refs. 6, 12, 28, and 43-45).
The following models for STAT activation at receptors phosphorylated on tyrosine residues have been discussed (Fig. 4).
Cytokine receptors, usually containing multiple potential binding sites for STAT factors, dimerize upon ligand stimula- FIG. 3. A single tyrosine module in a receptor complex is sufficient for STAT activation. COS-7 cells expressing the indicated combinations of Tyr-␥440 and ⌬B constructs were stimulated with IL-5 or left untreated. STAT activity in nuclear extracts was tested in an EMSA using the m67SIE probe.
tion. After STAT binding to their receptor docking sites followed by their tyrosine phosphorylation, the dimerization process might be favored by the presence of other phosphorylated STAT factors in close proximity. The "second" STAT monomer of the dimer could be recruited from a phosphotyrosine residue nearby on the same receptor chain. Alternatively, it might have been bound to the other chain of the receptor dimer before (Fig.  4, upper part). Support for this model comes from the observation that IFN␥ receptor ␣ chains point-mutated at the tyrosine residue critical for STAT1 binding (Tyr-440 in the human or Tyr-420 in the murine receptor) act as dominant-negative mutants when overexpressed in homologous cells (46). This effect could be explained by a model in which the IFN␥-induced formation of dimerized STAT1 binding sites was the prerequisite for consequent STAT factor dimerization (see also discussion in Ref. 7).
From our studies, however, we conclude that the presence of dimerized STAT recruitment sites in an active receptor complex is not the prerequisite for STAT activation. We rather show that one "tyrosine module" is sufficient to achieve STAT activation.
This finding can be reconciled with other models for STAT activation. One possibility would be that the second STAT factor binds to a phosphorylated STAT monomer already bound to the tyrosine phosphorylated receptor and then becomes phosphorylated (Fig. 4, middle part). The finding that there is no STAT1 phosphorylation in STAT2-deficient fibrosarcoma cells upon stimulation with IFN␣ could be explained by this model (47). However, the formation of the STAT1/2 heterodimer might be a special case, since it has been demonstrated that STAT1 has a rather high affinity for the phosphorylated peptide of STAT2 (7). According to analogous peptide binding studies, it seems unlikely that STAT1 is also recruited by a receptor-associated tyrosine-phosphorylated STAT1 molecule.
It is also possible that phosphorylated STATs dissociate from the receptor, a process that in analogy to the situation of p56 lck (48) might be encouraged by alteration of the selectivity of the SH2 domain to target tyrosine phosphopeptides (see discussion in Ref. 49). Phosphorylated monomers would then form dimers in the cytoplasm (Fig. 4, lower part).
A recent study with proteins expressed in insect cells suggested that phosphorylation of the STATs by Jaks might include a transient STAT-Jak association via the STAT-SH2 domain. It was further shown that only one STAT monomer of a dimer needed to be phosphorylated for complex formation in vitro (50) (not included in Fig. 4). It remains to be shown to what degree this reflects signal transduction events in a cytokine-stimulated cell.
Our study demonstrated that ligand-induced STAT1 activation can occur even if a receptor complex contains only a single STAT recruitment module. Based on these results it will be of interest to find out whether there is a general mechanism of STAT activation and to delineate its molecular details.