Physical and Functional Interaction between p72 syk and Erythropoietin Receptor*

Erythropoietin (Epo) regulates the proliferation and differentiation of erythroid cells through interaction with a cell surface receptor (EpoR) that belongs to the cytokine receptor family. The Jak2 tyrosine kinase was previously shown to bind to the EpoR, to be activated upon Epo stimulation, and to play a critical role in Epo-induced proliferation. However, little is known about the role of other tyrosine kinases in Epo signaling. In this paper, we examined whether Syk was involved in EpoR activation. Coimmunoprecipitation experiments showed that the phosphorylated EpoR was associated with the Syk kinase in activated UT7 cells. The interaction of Epo with its receptor led to an increased kinase activity. The use of recombinant Syk Src homology 2 (SH2) domains expressed in tandem or individually revealed that both N- and C-SH2 domains of Syk partici-pated in EpoR binding with a major contribution of the C-terminal SH2 domain. Far Western blotting further indicated that Syk directly binds to the EpoR and that the interaction of Syk with EpoR only occurred after Epo activation. These data suggest that phosphorylation of EpoR on tyrosine residues may mediate Syk binding to the receptor through interaction between the two SH2 domains of Syk and tyrosines of the receptor. We propose that ATP plus 10 m g of GST-HS1 peptide for 5 min at 23 °C. The reaction was stopped by the addition of 20 m l of 4 3 sample buffer and 40 m M EDTA. The samples were boiled at 95 °C for 5 min, and proteins were separated by 12.5% SDS-PAGE. Phosphotyrosine- containing proteins were detected by autoradiography after gel treat-ment with 1 N KOH at 55 °C to reduce the levels of phosphoserine and phosphothreonine. Quantification was done using a Molecular Dynam-ics PhosphorImager and ImageQuant Software. Immunoblotting and Far Western Analysis— Immunoblots were car-ried out as described previously (9). Proteins were transferred to nitro- cellulose membranes and immunoblotted with a mixture of 4G10 (0.3 m g/ml) and PY72 (1 m g/ml) or with anti-Syk antibodies (0.3 m g/ml polyclonal, 0.03 m g/ml monoclonal). Bound antibodies were detected by incubation with horseradish peroxidase-coupled secondary antibodies, and the enhanced chemiluminescence system (Amersham Pharmacia Biotech), as detailed by the manufacturer. For Far Western blotting, GST fusion proteins were eluted from glutathione-Sepharose beads with a buffer composed of 100 m M Tris-HCl (pH 8), 20 m M reduced glutathione, 100 m M NaCl, 2 m M dithiothreitol, and 1 m M phenylmethylsulfonyl fluoride. Immunoprecipitates were separated by SDS-PAGE and electrotransferred. The membrane was first incubated 3 h at 4 °C with 1 m g/ml of GST fusion protein followed by detection with anti-GST immunoblotting.

Erythropoietin (Epo) 1 acts primarily on late erythroid progenitors to maintain cell viability and to promote proliferation and terminal differentiation (1,2). Epo binds to a cell surface receptor (EpoR) that, like other cytokine receptors, lacks intrinsic kinase activity (3)(4)(5)(6). Activation of cells following binding of Epo leads to the tyrosine phosphorylation of a number of cellular proteins including the EpoR itself (7)(8)(9)(10). After ligand binding, EpoR transiently activates the receptor associated protein-tyrosine kinase Jak2 (11,12). Based on EpoR mutant studies, Jak2 was suggested to participate in EpoR signaling (12)(13)(14). Very recently, Jak2-deficient mice revealed that Jak2 is absolutely required for definitive erythropoiesis (15,16). However, Jak2 may not be the only tyrosine kinase involved in Epo activation. Truncated receptor mutants, which still activate Jak2, have lost the ability to induce the phosphorylation of most tyrosine kinase substrates (17), suggesting the implication of one or more tyrosine kinase(s) in addition to Jak2.
Several other tyrosine kinases have been shown to associate and/or be activated by EpoR or cytokine receptors, but their role in signaling is still largely unknown. The tyrosine kinase c-Fes has been reported to be tyrosine-phosphorylated and activated by Epo in the TF-1 erythroleukemia cell line (18). The Src family tyrosine kinase Lyn associates with and activates the granulocyte colony-stimulating factor (G-CSF) receptor (19). Lyn has also been shown to associate with EpoR and to be essential for erythropoietin-induced differentiation of JE2 erythroid cells (20). The cytoplasmic tail of the ␤-subunit of the interleukin-2 (IL-2) receptor associates with and activates another Src kinase, p56 lck , and interaction occurs within the acidic region ("A region") of the ␤-chain (21). However, IL-2mediated proliferation involves the serine-rich region ("S region") of the ␤-chain, which has also been shown to associate with Syk protein tyrosine kinase (22). Antibody cross-linking of a transmembrane chimera protein bearing CD16 in its extracellular domain and Syk in its cytoplasmic domain results in c-myc gene induction, an event critical for cell proliferation (22), which suggests that Syk may be an important component of IL-2 signaling. Syk also associates with the G-CSF receptor and is activated upon activation of this receptor (19). Both Lyn and Syk kinases are essential for the antiapoptotic pathway activated through the IL-3/IL-5/GM-CSF receptor ␤-subunit in human eosinophils (23). So far, the involvement of Syk kinase in EpoR activation has not been investigated.
The Syk family of nonreceptor protein-tyrosine kinases comprises two known members, Zap-70 and Syk. Syk is a 72-kDa cytoplasmic protein expressed in a variety of hematopoietic cells (24). The activation of Syk has mainly been studied for multichain immune recognition receptors (25). Syk is activated by the B-cell antigen (26), Fc receptors (27)(28)(29)(30), and T-cell antigen receptor (31)(32)(33). Syk activation depends on an immunoreceptor tyrosine activation motif (ITAM), made of two YXX(L/I) cassettes separated by 6 -8 amino acids and present in intracytoplasmic tails of receptor subunits (34). Receptor stimulation allows the rapid phosphorylation of an ITAM motif, which constitutes a binding site for the two Src homology 2 (SH2) domains of Syk (35). This event is seemingly mediated by members of the Src family of tyrosine kinases. Subsequently Syk is autophosphorylated (36) and/or phosphorylated by Src kinases (37,38) and its intrinsic kinase activity increases.
In the present study, we examined whether the Syk tyrosine kinase was involved in Epo signaling in the UT7 cell line. We show that following Epo activation, Syk and the EpoR are directly associated through the region of Syk that includes the * This work was supported by grants from the Association pour la Recherche sur le Cancer and from the Ligue Nationale Contre le Cancer. 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  two SH2 domains and that both N-SH2 and C-SH2 participate in EpoR binding with a major contribution of C-SH2. Furthermore, we show that Epo increases the intrinsic activity of Syk kinase.

EXPERIMENTAL PROCEDURES
Antibodies and Reagents-Rabbit polyclonal anti-Syk antibody LR and monoclonal anti-Syk antibody 4D10 were purchased from Santa Cruz Biotechnology, Inc. Anti-phosphotyrosine monoclonal antibodies 4G10 and PY72 were, respectively, gifts of B. Drucker (Portland, OR) and B. Sefton (La Jolla, Ca). Peroxidase-conjugated anti-rabbit and anti-mouse secondary antibodies were purchased from Amersham Pharmacia Biotech. Anti-EpoR antiserum was produced against a fusion protein between glutathione S-transferase (GST) and the cytoplasmic portion of human EpoR. Anti-GST antiserum was produced against purified recombinant GST. Purified recombinant human Epo (specific activity of 120,000 units/mg) was a gift of Dr. M. Brandt (Boehringer Mannheim).
GST Fusion Proteins-A cDNA fragment encoding rat Syk residues 1-260 including the tandem N and C SH2 was isolated by polymerase chain reaction. The amplified fragments were cloned into the pGEX-2TK expression vector and sequenced. Constructs encoding human Syk residues 1-265 (N ϩ C)-SH2, 1-117 (N)-SH2 and 163-265 (C)-SH2 in the pGEX-2TK expression vector were a generous gift of Dr. Brugge (Boston, MA). The GST fusion proteins were purified by batch absorption on glutathione-Sepharose 4B (Amersham Pharmacia Biotech) using standard procedures.
The GST-HS1 fusion protein, which serves as a specific substrate for Syk was constructed by joining a BamHI-EcoRI polymerase chain reaction fragment containing the HS1 peptide motif (EQEDEPEG-DYEEVLE-Stop) in frame to GST into the polylinker of the pGEX-2TK vector.
Cell Lines and Stimulation-The human leukemic cell line UT7 (39) was maintained in ␣-minimal essential medium supplemented with 5% fetal calf serum, penicillin, streptomycin, and 2 unit/ml Epo. MO7E-ER cell line corresponds to the MO7E cell line (40) infected with an amphotropic virus encoding a murine EpoR and was cultured in the same medium supplemented with 10% fetal calf serum. Before stimulation, cells were deprived of growth factor by washing in phosphate-buffered saline and by incubating overnight in Iscove's modified Dulbecco's medium containing 0.4% bovine serum albumin, 20 g/ml transferrin, penicillin, and streptomycin. For stimulation, cells were centrifuged and incubated (1 ϫ 10 7 /ml) at 37°C in deprivation medium with or without addition of Epo. The reaction was stopped by the addition of ice-cold phosphate-buffered saline.
Epo Binding Assays-Epo was iodinated using IODO-GEN (Pierce), as described previously (41) with specific radioactivities ranging from 30 to 60 ϫ 10 6 cpm/g. Epo was biotinylated according to Wojchowski et al. (42). Epo binding experiments were performed as described previously (43,44). In brief, cells were deprived of growth factor (see above) to ensure optimal expression of surface EpoR. Cells (2 ϫ 10 7 /ml) were incubated in Iscove's modified Dulbecco's medium containing 0.4% bovine serum albumin for 6 min at 37°C with 1 nM 125 I-Epo and washed in ice-cold phosphate-buffered saline. Nonspecific binding was determined by adding a 100-fold molar excess of unlabeled Epo.
For in vitro kinase assay, immunoprecipitates were washed an additional time with 25 mM Hepes (pH 7.4), 2 mM MnCl 2 , 10 mM MgCl 2 , 1 mM Na 3 VO 4 and incubated in 60 l of 5 mM Hepes, 2 mM MnCl 2 , 10 mM MgCl 2 , 1 mM Na 3 VO 4 , 10 Ci of [␥-32 P]ATP (3000 Ci/mmol; NEN Life Science Products), 10 M ATP plus 10 g of GST-HS1 peptide for 5 min at 23°C. The reaction was stopped by the addition of 20 l of 4ϫ sample buffer and 40 mM EDTA. The samples were boiled at 95°C for 5 min, and proteins were separated by 12.5% SDS-PAGE. Phosphotyrosinecontaining proteins were detected by autoradiography after gel treatment with 1 N KOH at 55°C to reduce the levels of phosphoserine and phosphothreonine. Quantification was done using a Molecular Dynamics PhosphorImager and ImageQuant Software.
Immunoblotting and Far Western Analysis-Immunoblots were carried out as described previously (9). Proteins were transferred to nitrocellulose membranes and immunoblotted with a mixture of 4G10 (0.3 g/ml) and PY72 (1 g/ml) or with anti-Syk antibodies (0.3 g/ml polyclonal, 0.03 g/ml monoclonal). Bound antibodies were detected by incubation with horseradish peroxidase-coupled secondary antibodies, and the enhanced chemiluminescence system (Amersham Pharmacia Biotech), as detailed by the manufacturer. For Far Western blotting, GST fusion proteins were eluted from glutathione-Sepharose beads with a buffer composed of 100 mM Tris-HCl (pH 8), 20 mM reduced glutathione, 100 mM NaCl, 2 mM dithiothreitol, and 1 mM phenylmethylsulfonyl fluoride. Immunoprecipitates were separated by SDS-PAGE and electrotransferred. The membrane was first incubated 3 h at 4°C with 1 g/ml of GST fusion protein followed by detection with anti-GST immunoblotting.

Activated Epo Receptor and Syk Tyrosine Kinase Associate in
Vivo-We used the UT7 cell line to examine whether Syk associates with the EpoR. UT7 cells were stimulated for various times at 37°C with 10 units/ml Epo, and Syk was immunoprecipitated. Similar amounts of Syk protein were detected in such conditions in unstimulated and stimulated cells (Fig.  1A, right part); a control antibody did not precipitate Syk protein (data not shown). Following Epo stimulation, EpoR is transiently tyrosine-phosphorylated reaching an optimum a few minutes after stimulation (9). However, since anti-EpoR are poorly reactive in immunoblot analysis (45), 2 and since phosphorylated EpoR migrates at a position very similar to Syk in SDS-PAGE (72 kDa), the two proteins cannot easily be distinguished. To identify EpoR in anti-Syk immunoprecipitates, we therefore took advantage of the fact that EpoR can be detected as a phosphorylated species following immunoprecipitation with anti-EpoR antibodies in stimulated cells and antiphosphotyrosine blotting. Thus, to assess the presence of activated EpoR in Syk precipitates, immune complexes were denatured by boiling in 1% SDS, and EpoR was further immunoprecipitated. Affinity-isolated proteins were separated by SDS-PAGE and analyzed by immunoblotting with anti-phosphotyrosine antibodies. A 72-kDa tyrosine-phosphorylated protein corresponding to the EpoR was clearly detected in Epostimulated cells, with a maximum 2-5 min after stimulation (Fig. 1A, left part). Reprobing the blot with anti-Syk antibodies did not allow the detection of Syk in the second anti-EpoR immunoprecipitation, showing that the association of Syk with the EpoR was disrupted by the denaturation step. Also, when control antibodies were used in the first immunoprecipitation, no 72-kDa phosphorylated protein was precipitated with anti-EpoR antibodies in the second immunoprecipitation (data not shown). We conclude that the 72-kDa tyrosine-phosphorylated protein, coprecipitated with Syk and reimmunoprecipitated with anti-EpoR in the second immunoprecipitation, is the activated EpoR. The association of Syk tyrosine kinase with the activated EpoR can also be shown directly following stimulation of UT7 cells for 10 min at 37°C with biotinylated Epo and precipitation of solubilized biotinylated Epo⅐EpoR complexes with streptavidin-agarose. This method allows the precipitation of the EpoR (Ref. 46; see Fig. 1B, lane 1). A 72-kDa protein identified as Syk by immunoblot analysis was coprecipitated with the EpoR (Fig. 1B, lane 3) and was not detected when the incubation with biotinylated Epo was omitted (data not shown), indicating that Syk is associated with the EpoR. To further confirm the association of Syk with the activated EpoR, 125 I-Epo was bound to cell surface receptors at 37°C using a saturating concentration of 1 nM. After cell solubilization with Brij 96, 65% of the radioactivity specifically bound to EpoR was precipitated with anti-EpoR antiserum, and a significant fraction was also precipitated with polyclonal anti-Syk antibodies (Fig. 1C). Using an optimal concentration of anti-Syk antibodies (10 g/ml), 2.3% of 125 I-Epo⅐EpoR complexes precipitated with anti-EpoR antibodies were coprecipitated with anti-Syk. These experiments show that Syk is associated with the membrane EpoR in Epo-activated cells.
Activated EpoR Associates with Syk Tyrosine Kinase via Syk SH2 Domains-We extended our analysis to determine whether the interaction between Syk and activated EpoR detected in vivo could be evidenced in vitro. Since the interaction between Syk and Fc⑀RI involves the tandem SH2 domains of Syk (47), we hypothesized that this region could also mediate Syk-EpoR interaction. We generated a Syk-GST fusion protein containing both SH2 domains. UT7 cell proteins binding to the tandem SH2 domains of Syk were isolated from cell lysates prepared before and after activation, using the GST-(N ϩ C)SH2-Syk bound to glutathione-agarose. When bound proteins were analyzed by immunoblotting with anti-phosphotyrosine antibodies, a 72-kDa protein was transiently detected after Epo stimulation ( Fig. 2A). This protein did not bind to GST alone, and it comigrated with the major tyrosine-phosphorylated protein present in cell lysates previously identified as the EpoR (9). Elution of proteins bound to GST-(N ϩ C)SH2-

p72 syk in Epo Receptor Signaling
Syk with glutathione and reimmunoprecipitation of eluted proteins with anti-EpoR antibodies followed by anti-phosphotyrosine immunoblotting identified this protein as the EpoR (Fig.  2B). The results show that the interaction between Syk and the activated EpoR occurs through the region of Syk that includes the tandem SH2 domains. To examine the relative contribution of the N-and C-terminal SH2 domains of Syk to EpoR binding, GST fusion protein containing either the N-or C-terminal SH2 domain was used for precipitation. The protein that contains the C-SH2 domain avidly bound to EpoR (Fig. 2C, left), whereas the protein containing the N-SH2 domain bound to EpoR with less efficiency (Fig. 2C, right). A shorter exposure of the autoradiogram revealed a lower binding of C-SH2 compared with the recombinant protein including both SH2 domains (not shown). These data show that both SH2 domains of Syk participate in the binding to the EpoR with a major contribution of the C-SH2 domain.
Syk SH2 Domains Bind Directly to the Activated EpoR-Recruitment of Syk to the EpoR may be direct or involve an intermediate component. Such a signal-transducing adapter molecule containing an ITAM motif has recently been identified (48, 49) but does not seem to associate with Syk (49). To determine if the interaction between the SH2 domains of Syk and EpoR was direct, GST fusion proteins were used in a Western blot of EpoR isolated from resting or Epo-activated UT7 cells. GST-(N ϩ C)SH2-Syk recognized a 72-kDa protein present in EpoR immunoprecipitates from activated cells, but GST alone did not (Fig. 3). The interaction did not occur when EpoR were isolated from resting cells. To confirm that the 72-kDa protein recognized by GST-(N ϩ C)SH2-Syk was the EpoR, the same experiment was performed after denaturation of precipitated proteins and reimmunoprecipitation with anti-EpoR antibodies. Again, the same 72-kDa protein was recognized by GST-(N ϩ C)SH2-Syk (data not shown). These data demonstrate that Syk can directly interact with the activated EpoR via its SH2 domains. They also strongly suggest that following Epo activation and tyrosine phosphorylation of EpoR Syk SH2 domains can interact with tyrosines present in the cytoplasmic domain of the EpoR.
Enhancement of Syk Kinase Activity in Response to Epo-Recently Brunati et al. (50) analyzed the phosphorylation efficiency of Syk for various peptide substrates derived from naturally occurring phosphoacceptor sites. They identified a peptide from HS1 protein, a target of protein kinases activated by the B-cell antigen receptor (51), as one of the best substrate for Syk (50) with a K m value around 4 M. To monitor Syk kinase activity, we produced a recombinant GST fusion protein containing the HS1 peptide. This fusion protein can be used for the specific monitoring of Syk, since other Src-related kinases (50) as well as Jak2 kinase 3 do not phosphorylate it. Syk tyrosine kinase was immunoprecipitated from lysates of untreated control or Epo-treated cells and assayed for its capacity to phosphorylate the HS1 fusion protein. As shown in Fig. 4A (top), Epo enhanced the Syk kinase activity against the exogenous substrate in UT7 cells with a 2-fold increase at 5 min and a 1.8-fold increase at 20 min after activation. Syk did not induce any detectable phosphorylation of a control GST protein, which indicates that HS1 peptide is the Syk substrate (data not shown). Epo also enhanced the Syk kinase activity in MO7-ER, a human cell line infected with an amphotropic virus encoding a murine EpoR (2-fold increase at 2 min and 2.4-fold increase at 15 min). Western blotting analyses with anti-Syk antibodies confirmed that equivalent amounts of the kinase were present in the anti-Syk immune complexes prepared from untreated or Epo-stimulated cells (Fig. 4A, bottom). In addition to the enhancement of Syk kinase activity against HS1 peptide, Epo also induced an increase in Syk autophosphorylation in vitro (Fig. 4B). A maximum increase was observed about 15 min after Epo activation, and then Syk autophosphorylation returned to basal level. These data indicate that Epo induces a transient activation of Syk kinase. DISCUSSION In this report, we investigated the possible involvement of Syk tyrosine kinase in signal transduction through the Epo receptor. Syk and activated EpoR were shown to associate both in vivo in coprecipitation experiments and in vitro by the ability of Syk SH2 fusion proteins to bind the EpoR. The interaction between EpoR and Syk SH2 occurred directly and only following Epo activation. Moreover, we found that the Syk kinase 3 V. Duprez and P. Mayeux, unpublished data. p72 syk in Epo Receptor Signaling activity increased as a functional consequence of Epo binding to cells.
The association between Syk and tyrosine-phosphorylated EpoR involves the N-terminal region of Syk that includes the tandem SH2 domains. Optimal binding was obtained by using fusion proteins containing both the N-and C-terminal SH2 domains. The relative contributions of Syk N and C domains to EpoR binding were not equivalent. The C-SH2 domain bound the EpoR efficiently whereas the N-SH2 domain showed only weak binding. A similar selectivity and cooperation of Syk SH2 domains binding has been shown for the high affinity IgE receptors (Fc⑀RI) (47,52). Syk associates with the ␥-chain of Fc⑀RI through an interaction between the tandem SH2 domains of Syk and the doubly phosphorylated ITAM motif of the receptor (47). ITAM motif is based on two repeated YXX(L/I) separated by 6 -8 amino acids (34). The receptor of the cytokine G-CSF contains a potential ITAM motif and also associates with and activates Syk (19). Syk has recently been shown to associate with the ␤-subunit of the GM-CSF receptor in eosinophils (23). Interestingly, four YXX(L/I) sequences with no consensus ITAM motif are also present in the ␤-chain. However, these motifs have not yet been explored for their implication in Syk activation by these cytokines. Syk and EpoR interaction could involve a tyrosine-containing motif present in the rapidly and transiently phosphorylated EpoR (7)(8)(9)(10). In accordance with this hypothesis, we showed that Syk directly associates with EpoR and that Epo activation seems to be required for the interaction. Although a consensus ITAM is not present in the EpoR, three tyrosines (Tyr 343 , Tyr 401 , and Tyr 429 in the membrane proximal domain of the intracellular region of the EpoR) present a leucine or isoleucine residue at the ϩ3position and thus represent potential candidates for Syk binding.
In addition to demonstrating the physical association of Syk tyrosine kinase with the EpoR, our data show that Syk tyrosine kinase activity transiently increases following Epo stimulation in UT7 cells as well as in MO7E cells expressing the EpoR, as demonstrated by the increased phosphorylating capacity toward the Syk-specific HS1 peptide. The 2-3-fold increase in Syk kinase activity is reminiscent of the increase observed in B lymphocytes after B cell receptor activation (53,54). These results suggest that Syk, in addition to Jak2, contributes to EpoR signaling. However, the relative contributions of Jak2 and Syk tyrosine kinases in the phosphorylation of EpoR substrates are presently unknown. Jak2 is more likely involved in EpoR phosphorylation. Indeed, studies of mutated EpoR have shown a correlation between receptor phosphorylation and the ability of the EpoR to bind Jak2 (11). In addition, Jak2 autophosphorylation and EpoR phosphorylation can be detected in vitro in anti-EpoR precipitates in 32D cells expressing EpoR (11) and in UT7 cells (data not shown). Furthermore, when EpoR and JAK2 cDNA were introduced into COS cells, we observed an increase in EpoR phosphorylation after Epo stimulation, whereas introduction of Syk with EpoR cDNA had no effect. 3 This suggests that EpoR phosphorylation by Jak2 precedes Syk kinase activation. Thus, JAK2 may phosphorylate EpoR, and phosphorylated tyrosines of the receptor may then constitute binding sites for Syk kinase. Syk recruitment to the receptor following Epo activation would then result in an increase in kinase activity and phosphorylation of its substrates. The increase in Syk kinase activity was observed between 5 and 30 min after Epo stimulation in UT7 cells. Signaling molecules recruited by EpoR such as mitogen-activated protein kinase and STAT5 were activated with similar kinetics in these cells (17,55), but further studies are needed to identify Syk substrates.
Other tyrosine kinases such as Src family protein kinases have been reported to participate in cytokine receptor activation and could also participate in EpoR signaling. p56 lck and Syk are both involved in IL-2 receptor ␤-chain activation (22). G-CSF receptor signaling involves the recruitment of both Lyn and Syk (19). Furthermore, Lyn is required for Syk activation through the B cell receptor (37), the ␤-chain of the GM-CSF receptor (23), and the Fc receptor (38). Whether Lyn is also required for Syk activation by EpoR remains to be determined. Interestingly, Lyn was recently shown to associate with EpoR and to activate STAT5 in erythroid cell lines (20,56) and to be essential for Epo-induced differentiation of J2E erythroid cells (20).