Oligomerization and Scaffolding Functions of the Erythropoietin Receptor Cytoplasmic Tail*

Signal transduction by the erythropoietin receptor (EPOR) is activated by ligand-mediated receptor homodimerization. However, the relationship between extracellular and intracellular domain oligomerization remains poorly understood. To assess the requirements for dimerization of receptor cytoplasmic sequences for signaling, we overexpressed mutant EPORs in combination with wild-type (WT) EPOR to drive formation of heterodimeric (i.e. WT-mutant) receptor complexes. Dimerization of the membrane-proximal portion of the EPOR cytoplasmic region was found to be critical for the initiation of mitogenic signaling. However, dimerization of the entire EPOR cytoplasmic region was not required. To examine this process more closely, we generated chimeras between the intracellular and transmembrane portions of the EPOR and the extracellular domains of the interleukin-2 receptor β and γc chains. These chimeras allowed us to assess more precisely the signaling role of each receptor chain because only heterodimers of WT and mutant receptor chimeras form in the presence of interleukin-2. Coexpression studies demonstrated that a functional receptor complex requires the membrane-proximal region of each receptor subunit in the oligomer to permit activation of JAK2 but only one membrane-distal tail to activate STAT5 and to support cell proliferation. Thus, this study defines key relationships involved in the assembly and activation of the EPOR signal transduction complex which may be applicable to other homodimeric cytokine receptors.

ation, and terminal differentiation of erythroid progenitor cells (1)(2)(3). The erythropoietin receptor (EPOR) belongs to the cytokine receptor superfamily and as such contains the structural motifs of four cysteine residues with canonical spacing and the sequence WSXWS in the extracellular domain (4,5). Although protein tyrosine phosphorylation is required for cytokine signaling (6), the cytoplasmic regions of these receptors lack intrinsic enzymatic activity. Signal transduction instead is mediated by cytoplasmic protein tyrosine kinases of the JAK family (7). These kinases are constitutively associated with receptor chains; residues in the membrane-proximal region of the EPOR cytoplasmic tail are required for this association. Ligand binding results in homodimerization of EPOR chains, which is critical for the initiation of intracellular signaling (8 -12).
The role of homodimerization in the initiation of signaling is highlighted by constitutively active variants of the EPOR chain (10,13). In the three mutants described to date, R129C, E132C, and E133C, amino acids in the extracellular domain are replaced by cysteines, leading to the formation of intermolecular disulfide bridges and constitutive signaling. In contrast, a number of other cytokine receptors signal through the heterodimerization of asymmetric receptor chains. For example, interleukin-2 (IL-2) binds and heterodimerizes the IL-2R␤ and ␥ c receptor chains (for review, see Refs. 14 and 15).
One of the earliest detectable signaling events induced by the EPOR is the phosphorylation of cellular proteins on tyrosines. This phosphorylation is thought to be mediated largely by JAK2, although other protein tyrosine kinases may also be involved (16). Eight tyrosine residues are present in the membrane-distal portion of the EPOR cytoplasmic tail, and several of these have been identified as docking sites for signaling molecules with phosphotyrosine binding motifs, including STAT5, phosphatidylinositol 3-kinase p85 subunit, and SHP-1 (17)(18)(19)(20)(21). Recruitment of these molecules to the EPOR/JAK2 complex leads to the activation of several signaling cascades, most likely through a change in subcellular localization, tyrosine phosphorylation, and/or oligomerization. In addition, tyrosine-phosphorylated JAK2 may interact directly with certain signaling molecules, including the recently described CIS/JAB/ SOCS family of proteins which may be important for the downregulation of signaling (22)(23)(24)(25)(26).
Although it is clear that ligand-induced dimerization of the EPOR extracellular region is required to initiate signal transduction (9 -12, 27), the functional requirements for signaling with regard to the oligomerization of cytoplasmic tails are unknown. To address this question, we coexpressed wild-type (WT) and mutant EPORs in the 32D cell line. We selected cell lines in which the mutant receptors were expressed at 10-fold higher levels than WT receptors to favor the formation of mu-tant-WT heterodimers for further analysis (10). This strategy allowed us to examine the relative capacities of heterodimeric EPORs to support EPO-dependent cellular proliferation. The roles of these cytoplasmic tails were then established definitively using a chimeric receptor system in which the identity of the extracellular domains forces only heterodimerization of cytoplasmic receptor chains. These two assay systems demonstrated that dimerization of the membrane-proximal region of the EPOR cytoplasmic tail is required for signal transduction, whereas a monomer of the membrane-distal region is sufficient for receptor-specific signaling. Thus, the distal portion of the activated receptor can function as a monomeric scaffold to recruit signaling molecules that are critical for inducing cell proliferation.

EXPERIMENTAL PROCEDURES
Plasmids-pME18S STAT5A was the generous gift of Dr. A. Mui. The pEFBOS JAK2 expression vector was the kind gift of Dr. D. Wojchowski. The EPOR mutant 1-321 was generated by exonuclease digestion using the Erase-a-base kit (Promega Corp., Madison, WI). Dideoxy sequence analysis demonstrated that the EPOR 1-321 mutant contains residues 1-320 of the WT EPOR plus an additional 14 amino acids (EINNCTALTGAGGQ) at the carboxyl terminus which are not encoded by the WT EPOR. These residues are the result of the mutagenesis procedure and are followed by a termination codon. To construct the EPOR mutant W282R (28), EPOR cDNA was digested with PvuII and BamHI and ligated to a synthetic, double-stranded oligonucleotide encoding the W282R point mutation. EPOR mutants were subcloned into the mammalian expression vector pMEX as described previously (10).
A modified form of the pEFBOS vector (29) containing the murine IL-3 signal sequence and the FLAG epitope tag was generously provided by Doug Hilton and Clare McFarlane (Walter and Eliza Hall Institute, Melbourne, Australia). To generate ␤EPOR, the extracellular region of the human IL-2R␤ chain (encoding residues 1-214 of the mature protein) was amplified by the polymerase chain reaction and subcloned into the modified pEFBOS vector, downstream of the FLAG epitope tag. The transmembrane and cytoplasmic domains of the murine EPOR (encoding residues 226 -483 of the mature protein) were amplified by polymerase chain reaction; this fragment was subcloned into the pEFBOS vector 3Ј to the IL-2R␤ chain extracellular region. The resulting construct (pEFBOS.S.FLAG␤E) contains the IL-3 signal sequence at the amino terminus, followed in-frame by the FLAG epitope tag, the IL-2R␤ extracellular region, and the EPOR transmembrane and cytoplasmic regions. Additional amino acids (TR or VD) were introduced in between the epitope tag and the IL-2R␤ extracellular domain, and the IL-2R␤ extracellular region and EPOR transmembrane domain, respectively, by the subcloning procedure. ␤EPORs containing the 1-321 or W282R mutations were generated by swapping restriction fragments within the EPOR cytoplasmic domain. Sequences of the ␤EPOR constructs were confirmed by dideoxy sequence analysis. The murine ␥ c cDNA was the kind gift of Dr. T. Kono. To construct the murine ␥ c -EPOR chimera, the transmembrane and intracellular domains of the EPOR were amplified by polymerase chain reaction (from amino acid residue 252 of the mature protein to the stop codon) with oligonucleotides inserting unique 5Ј (BsmI) and 3Ј (XbaI) restriction sites. This fragment was used to replace the corresponding transmembrane and intracellular domains of murine ␥ c , and the construct was verified by DNA sequencing. To generate the ␥EPOR 1-321 chimera, the KpnI-BglII fragment of ␥EPOR and the BglII-XbaI fragment of ␤EPOR 1-321were cloned into the KpnI and XbaI sites of pCMV4neo. The sequence of this construct was confirmed by automated DNA sequencing. Expression of chimeric proteins and EPOR was confirmed by COS-7 cell transfection and immunoprecipitation experiments (see below).
Cell Lines, Culture Conditions, and Transfections-32D is a well described IL-3-dependent myeloid cell line (30). Stable transfectants expressing WT or mutant EPORs were maintained in RPMI 1640 medium supplemented with 10% heat-inactivated fetal calf serum (FCS) and 5% conditioned medium from the WEHI 3B cell line (RPMI/ FCS/WEHI), the latter as a source of IL-3. 32Dn20 cells express the WT EPOR from the pXM vector (10). To establish cell lines coexpressing WT and mutant EPORs, 32Dn20 cells were electroporated (using a Bio-Rad Gene Pulser apparatus) with pMEX containing mutant EPOR cDNAs, using conditions described previously (10). Transfected cells were selected by growth in RPMI/ FCS/WEHI containing 500 g/ml G418 (Life Technologies), and clonal cell lines were isolated by limiting dilution. Expression of the mutant EPORs was verified by EPOR immunoprecipitations from metabolically labeled cells (11). 32D cell lines coexpressing ␤EPOR and ␥EPOR were established by two rounds of electroporation and selection. First, cells were electroporated with pEFBOS.S.FLAG␤E and pSV 2 NEO, and stably transfected cells selected by growth in RPMI/FCS/WEHI containing 500 g/ml G418. G418-resistant cells were then electroporated with pBABE/Puro and pCMV/␥EPOR or pCMV/␥EPOR 1-321. Stably transfected cells were selected by growth in RPMI/FCS/WEHI containing 5 g/ml puromycin and clonal cell lines were isolated by limiting dilution.
␥2A, a somatic mutant of the 2C4 fibrosarcoma cell line that specifically fails to express JAK2 (31), was the generous gift of Drs. G. Stark and I. Kerr. These cells were cultured as described previously in Dulbecco's minimal essential medium supplemented with 10% FCS, 100 IU/ml penicillin, 100 g/ml streptomycin, and 700 g/ml G418. These cells were transfected using calcium phosphate as described previously (32) and according to the manufacturer's instructions.
Growth Factor-dependent Proliferation Assays-The ability of the EPOR-expressing cell lines to confer EPO-dependent proliferation was tested by short term proliferation assays as described previously (10). To examine IL-2-dependent proliferation in cell lines coexpressing ␤EPOR and ␥EPOR or ␥EPOR 1-321, cells were washed three times with RPMI/FCS and then plated in RPMI/FCS supplemented with IL-2 at the indicated concentrations (see figure legends). Cells were passaged for up to 2 weeks in IL-2-containing media. The culture density was maintained under 10 6 cells/ml by dilution in fresh medium. Viable cells, as judged by trypan blue dye exclusion, were counted every 2-3 days, and total cell numbers were graphed to determine doubling times. All cell lines used in this study were strictly growth factor-dependent and died after 24 h in RPMI/FCS.
Antibodies, Immunoprecipitations, SDS-Polyacrylamide Gel Electrophoresis, and Immunoblots-Antiserum specific for the amino-or carboxyl-terminal region of the murine EPOR has been described previously (33). Anti-JAK2 serum and the anti-phosphotyrosine antibody 4G10 were obtained from Upstate Biotechnology (Lake Placid, NY). Antisera specific for STAT5 was obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Cell lysates were prepared for immunoprecipitation as described previously (10), using lysis buffer containing 1% w/v Triton X-100, 150 mM NaCl, 50 mM Tris-HCl, pH 7.4, 1 mM EDTA, supplemented with 1 mM sodium vanadate, 2 mM phenylmethysulfonyl fluoride, and 10 l/ml aprotinin. Lysates were cleared by centrifugation and incubation with protein A-agarose beads and immunoprecipitations performed as described previously (10).
For IL-2 stimulation, cells were washed and cultured in serum-free RPMI for 4 h at 37°C and then stimulated with 10 nM IL-2 for 8 -10 min at room temperature. Proteins were immunoprecipitated from cell lysates, separated by electrophoresis on 7.5% SDS-polyacrylamide gels, and transferred to nitrocellulose filters. The filters were blocked in buffer containing 1% bovine serum albumin (BSA), 1 ϫ phosphatebuffered saline (PBS), and 0.2% Tween 20 (BSA/PBS/Tween) for 60 min at room temperature. The filters were incubated with 4G10 antibody (1:1,000 dilution in BSA/PBS/Tween) at 4°C overnight, washed twice (15 min each) in 1 ϫ TBS containing 0.2% Tween 20. Horseradish peroxidase-conjugated anti-mouse secondary antibody (Amersham Pharmacia Biotech) (1:10,000 dilution in BSA/PBS/Tween) was incubated with the filters for 60 min at room temperature. Filters were washed as described above and were developed with chemiluminescent reagents (Pierce Chemical Co.) according to the manufacturer's instructions.
To label proteins metabolically, cells were washed with methionine/ cysteine-free medium and then incubated in methionine/cysteine-free medium supplemented with 500 Ci/ml [ 35 S]methionine and -cysteine for 2 h. Proteins were immunoprecipitated, separated by SDS-PAGE, and the gels were fluorographed, dried, and exposed to x-ray film.
Nuclear Extracts and Electrophoretic Mobility Shift Assay-48 h after transfection, cells were stimulated with the appropriate cytokine at the indicated concentration for 10 min at 37°C and then harvested. Nuclear extracts were prepared as described previously (34,35) in the presence of the following protease inhibitors: 0.5 mg/ml antipain, 0.5 mg/ml aprotinin, 0.75 mg/ml bestatin, 0.5 mg/ml leupeptin, 0.05 mg/ml pepstatin A, 0.5 mM phenylmethylsulfonyl fluoride, 1.4 mg/ml phosphoramidon, and 0.5 mg/ml soybean trypsin inhibitor as well as 1 mM sodium orthovanadate. Oligonucleotide probes encoding the Fc␥RI or IRF-1 STAT response elements were end labeled with polynucleotide kinase (Boehringer Mannheim) and [␥-32 P]ATP (Amersham Pharmacia Biotech). DNA binding studies were performed with 10 5 cpm probe, 2 g of poly(dI-dC) and 10 g of nuclear proteins as described (34,35). For supershift analyses, the nuclear extract was preincubated with antibody for 45 min on ice before the addition of the radiolabeled probe. Purified rabbit IgG specific for STAT5 (C-17) was obtained from Santa Cruz Biotechnology.

RESULTS
To determine the functional requirements for oligomerization of the EPOR cytoplasmic tails, we coexpressed WT and cytoplasmic tail mutant EPORs in 32D cells. In previous studies, truncation mutants lacking the majority of sequences in the cytoplasmic domain of the EPOR dominantly inhibited EPO-induced cellular proliferation when coexpressed in excess with the WT receptor, indicating that these mutants heterodimerize with the WT receptor and block signal transduction from the cell surface (10,27). Therefore, the coexpression of WT EPOR and less severe truncation mutants should clarify whether a dimer of the full-length EPOR cytoplasmic tail is required for receptor functions involved in cellular proliferation, or if growth signaling can be activated through asymmetric configurations of the EPOR intracellular domains.
IL-3-dependent 32D cells that express the WT EPOR proliferate in response to stimulation with either EPO or IL-3 (10). We constructed the EPOR mutants shown in Fig. 1A, and the ability of these mutants to support EPO-dependent proliferation alone or in combination with the WT receptor was assessed in 32D cells. When expressed singly, the EPOR 1-257, EPOR 1-321, and EPOR W282R mutants were displayed on the cell surface at levels comparable to WT EPOR (600 -1,700 sites/ cell), indicating that the biosynthesis of these mutants is relatively normal. In addition, these mutants have WT affinity for EPO (K d 200 -400 pM) (data not shown). The EPOR 1-257 and EPOR W282R mutants were unable to support EPO-dependent proliferation when expressed alone (Fig. 1A). Expression of the EPOR 1-321 mutant maintained the viability of cells and permitted weak proliferation in only saturating concentrations of EPO (Figs. 1A and 2).
To generate coexpressing cells, a 32D cell line that expresses low levels of WT EPOR (32Dn20) was transfected with the mutants shown in Fig. 1A. Multiple clonal lines coexpressing WT and mutant EPORs were isolated. To determine the relative levels of WT and mutant EPOR expression in each cell line, proteins were metabolically labeled with [ 35 S]methionine and -cysteine, immunoprecipitated with antiserum specific for the EPOR, and separated by SDS-PAGE. We selected cell lines in which the mutant receptors are expressed at 5-10-fold higher levels than the WT EPOR for further analysis (Fig. 1B). At this ratio of expression, EPOR 1-257 functions readily as a dominant inhibitor of WT EPOR-mediated proliferation (10).
To test the ability of these mutants to inhibit dominantly proliferation mediated by the WT receptor, cells coexpressing WT and mutant EPORs were cultured in media containing 10 units/ml EPO or IL-3 for 3 days. Interestingly, the truncation mutant EPOR 1-321, which was unable to support a significant level of EPO-dependent proliferation when expressed on its own, failed to block proliferative signaling (Figs. 1A and 2). However, a full-length form of the EPOR containing the single point mutation W282R in the cytoplasmic tail exerted potent dominant inhibitory effects (Figs. 1A and 2). To determine the EPO dose-response profiles of clonal cell lines coexpressing the WT receptor and EPOR 1-321 or EPOR W282R, we assayed the growth of these cell lines in various concentrations of EPO. For each mutant receptor, three independently isolated clonal lines were characterized, and the results from representative cell lines are shown in Fig. 2. The EPOR W282R mutant dominantly inhibited cellular proliferation at all concentrations of EPO tested, whereas the EPOR 1-321 mutant did not affect the dose-dependent proliferation of these cells in response to EPO (Fig. 2).
Together, these results suggest that dimerization between WT and EPOR 1-321 receptors can stimulate the signaling pathways leading to cellular proliferation, whereas dimerization of WT and EPOR W282R is insufficient for mitogenic signaling. However, by this approach the precise molecular configuration of the cell surface receptor complexes could not be regulated definitively. Therefore, we sought to explore these findings further in a system in which the cell surface receptor configuration could be defined specifically. We replaced the extracellular domain of the EPOR with the extracellular domains of IL-2R␤ and ␥ c because it has been demonstrated definitively that IL-2-induced heterodimerization of these two chains is necessary and sufficient to initiate signal transduction (36 -38). A ␤EPOR chimera, containing the extracellular domain of the IL-2R␤ chain fused to the transmembrane and cytoplasmic tail of EPOR, was constructed; for ease of detection, this construct was tagged with the FLAG epitope at its NH 2 terminus. A WT ␥EPOR chimera was similarly constructed (although lacking the epitope tag) (for a schematic of these chimeras, see Fig. 3A). Transient transfection assays in COS-7 cells were used to confirm protein expression (data not shown). The signaling function of these chimeras was assessed initially by transient cotransfection in the ␥2A cell line. This cell line is a somatic mutant of the 2C4 fibrosarcoma cell line which is specifically deficient in JAK2 protein expression (31); this cell line also does not express IL-2R␤, ␥ c , and STAT5 (data not shown). Because there are two functional isoforms of STAT5 (STAT5A and STAT5B) which seem to be almost completely interchangeable with regard to cytokine-dependent activation (39 -42), for simplicity we used one of these isoforms, STAT5A, in our transfection experiments. Reconstitution of ligand-inducible STAT5 activation required the simultaneous transfection of cDNAs encoding ␤EPOR, ␥EPOR, JAK2, and STAT5A as well as stimulation of the cells with IL-2 (Fig. 3B). The identity of the induced DNA-binding complex was confirmed by supershift analysis using an antibody specific for STAT5 (data not shown). Elimination of either receptor chain or the associated kinase resulted in a loss of signal, demonstrating that each of these components was required for STAT5 activation (Fig. 3B). Furthermore, the requirement for JAK2 demonstrated that these chimeras couple appropriately to the kinase associated with the EPOR (43).
Next, we introduced two of the mutant EPOR tails we have described, EPOR 1-321 and EPOR W282R, into the context of these chimeras. Replacement of ␤EPOR WT with the ␤EPOR 1-321 truncation did not significantly alter STAT5 induction by IL-2 through the ␤EPOR-␥EPOR heterodimer (Fig. 4). In contrast, dimerization of the ␤EPOR W282R mutant and ␥EPOR led to barely detectable activation of STAT5. We believe that the minimal level of STAT5 activation observed with the ␤EPOR W282R chimera reflects the extreme sensitivity of the assay system. Nonetheless, the level of STAT5 activation is To extend these results, we examined the effects of the ␥EPOR 1-321 truncation mutant on signaling. The pairing of ␤EPOR WT with ␥EPOR 1-321 resulted in STAT5 induction that was indistinguishable from the pairing of two WT receptors. In contrast, the homodimerization of EPOR 1-321 cytoplasmic tails resulted in barely detectable STAT5 activation, in agreement with results obtained from 32D cells expressing only EPOR 1-321 (data not shown). Finally, the pairing of ␥EPOR 1-321 with ␤EPOR W282R resulted in virtually no STAT5 activation (Fig. 4).
To analyze the proliferative potential of heteromeric EPOR complexes, clonal 32D cell lines coexpressing ␤EPOR and ␥EPOR, or ␤EPOR and ␥EPOR 1-321, were established. To verify that a proliferative response was caused by transfected receptor subunits and not by the murine ␥ c chain that is expressed endogenously in 32D cells, clonal lines expressing ␤EPOR alone were also derived. These cells were unable to proliferate or survive (data not shown), even in saturating concentrations (1.0 or 10 nM) of IL-2. Coexpression of ␤EPOR and ␥EPOR or ␤EPOR and ␥EPOR 1-321 rendered 32D cells responsive to IL-2 (Fig. 5A), and the doubling time of each clonal line was relatively unchanged over a 100-fold range of ligand concentrations (e.g. media supplemented with 0.1, 1.0, or 10.0 nM IL-2; data not shown). Both JAK2 and STAT5 were efficiently tyrosine-phosphorylated in IL-2-stimulated cells coexpressing ␤EPOR and ␥EPOR or coexpressing ␤EPOR and ␥EPOR 1-321, but not in cells expressing ␤EPOR alone (Fig. 5,  B and C). Thus, IL-2-mediated heterodimerization of ␤EPOR and ␥EPOR, or ␤EPOR and ␥EPOR 1-321, generates functional receptors that are capable of delivering long term proliferative signals. In contrast, ␤EPOR alone, or ␤EPOR and the endogenous murine ␥ c chain, does not form functional receptor complexes in response to IL-2. These results demonstrate de-finitively that cellular proliferation can be stimulated by heterodimers containing full-length EPOR and EPOR 1-321 cytoplasmic regions. DISCUSSION The results presented herein demonstrate that dimerization of the entire EPOR cytoplasmic region is not required for the signal transduction pathways leading to cellular proliferation. When expressed on its own, the EPOR 1-321 mutant bound JAK2 and stimulated EPO-dependent JAK2 tyrosine phosphorylation (data not shown). This result indicates that oligomerization of the membrane-proximal portion of the EPOR cytoplasmic tail, which contains the binding site for JAK2 (43), is sufficient for JAK2 activation. JAK2 activation may occur via a trans-autophosphorylation mechanism, similar to the activation mechanism of other tyrosine kinases (45)(46)(47)(48)(49). Despite activation of JAK2, the EPOR 1-321 mutant, which lacks all eight cytoplasmic tyrosine residues, failed to support cell proliferation at physiological concentrations of ligand. This result is in agreement with the findings of others, who have demonstrated that sequences in the distal portion of the cytoplasmic tail play a critical role in receptor signal transduction for growth (19,20,28,50,51). However, the EPOR 1-321 mutant did not dominantly inhibit signal transduction by the WT EPOR. These findings suggest that the primary role of EPOR dimerization is to oligomerize and activate EPOR-associated JAK2, leading to the initiation of intracellular signaling events that require the presence of elements within the distal portion of the EPOR cytoplasmic tail. In this context, a single fulllength receptor tail is sufficient within the dimeric receptor complex to engage this signaling machinery.
The initial findings using coexpressed EPOR variants were strongly suggestive but were complicated by our inability to define precisely the receptor complexes formed by the binding of EPO. To drive receptor subunit heterodimerization in the absence of the potential artifact of receptor homodimers, we generated two asymmetric chimeras in which the extracellular domain of EPOR is replaced with the equivalent portions of IL-2R␤ and ␥ c . Coexpression of these two molecules along with JAK2 permitted IL-2-mediated STAT5 induction, highlighting the importance of cytoplasmic tail dimerization in signal transduction. Substitution of either receptor chain (but not both) with the truncation mutant 1-321 led to levels of STAT5 induction which were indistinguishable from the levels induced by a receptor complex containing two WT receptor chains. These data demonstrate that only one full-length receptor chain is required to initiate an EPOR-specific signaling program. Furthermore, the fact that a receptor complex containing two truncated receptors fails to activate STAT5 efficiently emphasizes the functional importance of the membrane-distal portion of the EPOR tail. Finally, the observation that EPOR 1-321 and EPOR W282R heterodimers fail to signal demonstrates that two functional JAK2 binding sites are required in addition to a single full-length cytoplasmic tail.
The IL-2-dependent proliferation of 32D cells expressing ␤EPOR and ␥EPOR, or ␤EPOR and ␥EPOR 1-321, confirmed that asymmetric EPOR complexes are functional. The downstream EPOR signaling molecules, JAK2 and STAT5, likewise were activated efficiently by IL-2 in coexpressing cell lines but not in cells expressing ␤EPOR alone. Therefore, dimerization of the EPOR is required to activate JAK2, which directly or indirectly leads to phosphorylation of EPOR cytoplasmic tyrosine residues. Once phosphorylated, a monomer of the distal portion of the EPOR cytoplasmic region can function to recruit and activate the signal transduction proteins essential for cellular proliferation.
However, several questions remain regarding the structure FIG. 3. Recapitulation of EPOR signal transduction by ␤EPOR and ␥EPOR chimeras. Panel A, schematic diagram of the parental receptors and these chimeras. Although not diagrammed, the ␤EPOR chimera is epitope tagged at the NH 2 terminus with the FLAG tag; this construct is subsequently referred to as ␤EPOR. Panel B, reconstitution of EPOR signaling by chimeras encoding ␤EPOR and ␥EPOR is dependent upon cotransfection of ␤EPOR, ␥EPOR, JAK2, and STAT5A cDNAs. ␥2A cells were transfected as indicated. 48 h post-transfection, cells were incubated in medium alone (Ϫ) or in medium containing 10 nM recombinant human IL-2 (2) for 10 min at 37°C. Nuclear extracts were prepared and subjected to electrophoretic mobility shift assay with a 32 P end-labeled Fc␥RI oligonucleotide, GTATTTCCCAGAAAAAGGAC. of the functional EPOR complex. Previous studies have shown that dimerization of the EPOR stimulates the signals leading to erythroid cell survival, proliferation, and differentiation (8,10,11,44). Crystallographic structural analysis of the EPOR extracellular region complexed to a peptide agonist demonstrated directly the dimeric nature of the receptor, and biochemical experiments have shown that EPO promotes dimerization of the EPOR extracellular region (9,12,52). However, a dimer of the EPOR may be the minimal complex required for signaling, and higher order oligomers may form the functional receptor complexes on the cell surface. Further biochemical direct analysis of the EPOR on the membrane of living cells is required to determine the oligomeric nature of the activated receptor complex.
In addition, several lines of evidence suggest that EPOR subunit dimerization is necessary but not sufficient to stimulate signal transduction. First, we show here in two systems that W282R blocks signaling from the WT EPOR even though W282R is expressed on the plasma membrane, binds ligand, associates with JAK2, and contains all eight tyrosines in the cytoplasmic region (data not shown (28,43)). Second, a double mutant EPOR, R129C/W282R, which contains W282R as well as the R129C substitution that promotes EPO-independent disulfide-linked receptor dimerization on the cell surface (11), is inactive in cell proliferation assays, 2 whereas cells expressing an EPOR with the R129C mutation alone proliferate in the absence of any added growth factor (11,33). These findings imply that the nonfunctional W282R mutant must alter either the relative orientations of the receptor cytoplasmic regions or the alignment of JAK2 molecules in the receptor dimer. In addition, EPO-mimetic peptides that bind to and dimerize the EPOR extracellular domain are not always sufficient to activate intracellular signal transduction (53). Thus, it appears that JAK2 molecules and/or the cytoplasmic regions of the EPOR must be brought into the correct orientation in the receptor dimer to create a competent, activated receptor complex. Some degree of flexibility may be allowed in the structure of the EPOR dimer, however, because recent studies suggest that there may be structural differences when the EPOR is activated by a peptide agonist, by the R129C mutation, or by EPO (9, 54).
We have described previously a model for heteromeric cytokine receptor signaling which distinguishes receptor chains based on two essential functions (55). One receptor chain appears to specify the signaling program that is unique to the particular receptor complex. The second receptor subunit does not contribute significantly to the specificity of signal transduction but rather triggers receptor activation. This structure/ function arrangement applies to several members of the receptor family which share the ␥ c chain as a trigger chain (55)(56)(57). The present findings indicate that signaling through the homodimeric EPOR is governed by similar structure/function principles. In the EPOR configurations generated for the present studies, the full-length EPOR drives specificity, and the EPOR 1-321 mutant functions as the trigger chain. Thus, in this homodimeric receptor as in heteromeric receptors, a single phosphorylated receptor tail is sufficient for the engagement of the downstream signaling machinery. In the native EPOR complex, each receptor chain likely functions as a trigger chain for its partner subunit, leading to the efficient initiation of signal transduction by the homodimeric complex. Further studies will be required to test the generalizability of this model to other homodimeric receptor complexes. FIG. 4. Characterization of the functional requirements of ␤EPOR and ␥EPOR chimeras for downstream STAT5 induction. ␥2A cells were transfected with the indicated cDNAs. Cells were treated with IL-2 as described in the legend to Fig. 3; nuclear extracts were prepared and subjected to electrophoretic mobility shift assay as described.
FIG. 5. IL-2-dependent proliferation and signal transduction in stably transfected 32D cell lines. Panel A, cells coexpressing ␤EPOR (␤E) and ␥EPOR (␥E) or ␤EPOR and ␥EPOR 1-321 (␥1-321) were grown in medium containing 1.0 nM IL-2 for 12 days. Cultures were maintained under a density of 10 6 cells/ml, and total cell numbers were determined as described under "Experimental Procedures." The results from analysis of two independent clonal cell lines are shown for each pair of receptor constructs. Panels B and C, IL-2 stimulations were performed as described under "Experimental Procedures" (10 nM IL-2). Proteins were immunoprecipitated from detergent cell extracts with antisera specific for JAK2 or STAT5, separated by SDS-PAGE, and transferred to nitrocellulose. The filters were probed with anti-phosphotyrosine antibody 4G10 and developed with chemiluminescent reagents. cDNA, Dr. Daniel Wojchowski for the JAK2 cDNA, Dr. Bing Su and Chiron Corp. for the gifts of IL-2, and Dr. Doug Hilton and Clare McFarlane for the modified pEFBOS vector. We acknowledge the generous gift of cell lines from Drs. George Stark and Ian Kerr. We thank Drs. Harvey F. Lodish and Rebecca G. Wells for advice during the course of these studies and for comments on the manuscript, Dr. Bradley W. McIntyre and members of his laboratory for assistance with scanning and image analysis, and Dr. Joan Egrie (Amgen Corporation) and Dr. Francis Farrell (R. W. Johnson Pharmaceutical Research Institute) for generous gifts of recombinant EPO. We thank John Carroll, Heather Livesay, Stephen Gonzales, and Neile Shea for assistance in the preparation of this manuscript.