Phosphorylation of tyrosine 503 in the erythropoietin receptor (EpR) is essential for binding the P85 subunit of phosphatidylinositol (PI) 3-kinase and for EpR-associated PI 3-kinase activity.

We recently reported that phosphatidylinositol (PI) 3-kinase becomes associated with the activated erythropoietin receptor (EpR), most likely through the Src homology 2 (SH2) domains within the p85 subunit of PI-3 kinase and one or more phosphorylated tyrosines within the EpR. We have now investigated this interaction in more detail and have found, based on both blotting studies with glutathione S-transferase-p85-SH2 fusion proteins and binding of these fusion proteins to SDS-denatured EpRs, that this binding is direct. Moreover, both in vitro competition studies, involving phosphorylated peptides corresponding to the amino acid sequences flanking the eight tyrosines within the intracellular domain of the EpR, and in vivo studies with mutant EpRs bearing tyrosine to phenylalanine substitutions, indicate that phosphorylation of Tyr within the EpR is essential for the binding of PI 3-kinase. The presence of PI 3-kinase activity in EpR immunoprecipitates from DA-3 cells infected with wild-type but not Y503F EpRs confirms this finding. Our results demonstrate that the SH2 domains of p85 can bind, in addition to their well established Tyr-Met/Val-X-Met consensus binding sequence, a Tyr-Val-Ala-Cys motif that is present in the EpR. A comparison of erythropoietin-induced tyrosine phosphorylations and proliferation of wild-type and Y503F EpR-infected DA-3 cells revealed no differences. However, the PI-3 kinase inhibitor, wortmannin, markedly inhibited the erythropoietin-induced proliferation of both cell types, suggesting that PI 3-kinase is activated in Y503F EpR expressing cells. This was confirmed by carrying out PI 3-kinase assays with anti-phosphotyrosine immunoprecipitates from erythropoietin-stimulated Y503F EpR-infected DA-3 cells and suggested that PI 3-kinase has a role in regulating erythropoietin-induced proliferation, but at a site distinct from the EpR.

Erythropoietin is the principal in vivo stimulator of mamma-lian erythropoiesis (1) and exerts its action by binding to receptors on the surface of relatively mature erythroid progenitors (2)(3)(4). These erythropoietin receptors (EpRs) 1 belong to a family of hematopoietin receptors whose members are characterized by the presence of conserved cysteines and Trp-Ser-X-Trp-Ser motifs in their extracellular domains and the absence of any known catalytic activity in their intracellular regions (5,6). Nonetheless, although the EpR lacks tyrosine kinase activity, it, along with a number of other cellular proteins, becomes transiently phosphorylated on tyrosine residues within minutes of binding erythropoietin (7)(8)(9)(10)(11)(12)(13). Recent studies suggest that this rapid phosphorylation is carried out, at least in part, by the cytoplasmic tyrosine kinase, Jak 2 (14,15). Following this activation of the EpR, a number of intracellular proteins apparently become physically associated with it (8, 16 -19). These proteins include Grb2 (8,18), Shc (8,18), and the enzyme, phosphatidylinositol (PI) 3-kinase (16,17,19). All three of these proteins contain at least one Src homology 2 (SH2) domain, consisting of approximately 100 amino acids (20) that may direct their physical association with specific tyrosine phosphorylated regions within the activated EpR. In the case of PI 3-kinase, a heterodimeric enzyme complex that phosphorylates PI, PI-4 phosphate, and PI-4,5-bisphosphate at the D-3 position of the inositol ring (21), it is now generally believed that its 85-kDa subunit, which contains two SH2 domains (22), functions as an adaptor molecule that targets its associated catalytic 110-kDa component (23) to activated growth factor receptors (24).
In a previous report we demonstrated that erythropoietin stimulates the association of PI 3-kinase with the activated EpR, most likely through the SH2 domains of p85 and either the activated EpR itself or a phosphorylated protein intermediate (16). In the current study we have further investigated the nature of the interaction between p85 and the activated EpR. Our results suggest that this interaction is direct and that p85 binds to the phosphorylated COOH-terminal tyrosine within the cytoplasmic region of the EpR, i.e. Tyr 503 . This finding reveals a new recognition motif for both the NH 2 -and COOH-terminal SH2 domains of p85, i.e. Tyr-Val-Ala-Cys.

MATERIALS AND METHODS
Reagents-Recombinant human erythropoietin was purified from culture supernatants of baby hamster cells expressing an erythropoietin cDNA and was biotinylated as described previously (4). Both the free and agarose-conjugated forms of the anti-phosphotyrosine (anti-PY) monoclonal antibody, 4G10, and the anti-PI 3-kinase, specific for the * This work was supported by the National Cancer Institute of Canada and the Medical Research Council of Canada with core support from the BC Cancer Foundation and the BC Cancer Agency (to G. K.) and by American Cancer Society Grant DB-74554 and American Heart Association Grant NEO-94 -074-GIA (to T. Y.). 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  p85 subunit of PI 3-kinase, were purchased from Upstate Biotechnology Inc (Lake Placid, NY). The anti-EpR antibody used for immunoprecipitations was generated by immunizing rabbits with a glutathione Stransferase (GST) fusion protein containing amino acids 248 -332 of the mature mouse EpR (18). The anti-EpR antibody used for immunoblotting, generated by immunizing rabbits with a GST fusion protein containing the COOH-terminal 18 amino acids of the EpR, was kindly provided by Dr. Alan D'Andrea (Dana Farber Cancer Institute, MA). The p85 SH2 GST fusion proteins, consisting of the 27-kDa aminoterminal of GST linked to either the NH 2 -terminal SH2 (residues 312-444 of bovine brain p85) or the COOH-terminal SH2 (residues 612-722 of p85) of PI 3-kinase were purified, using glutathione-agarose (Pharmacia LKB Biotechnology, Baie d'Urfe, Quebec, Canada), from Escherichia coli expressing these cDNAs in pGEX-2T plasmids (generously provided by Dr.Tony Pawson, Mount Sinai Hospital, Toronto, Canada) as described previously (16). The tyrosine-phosphorylated peptides used in the phosphopeptide competition studies were synthesized at the Cleveland Clinic Research Institute (Cleveland, OH). They corresponded to regions flanking the eight tyrosines within the cytoplasmic domain of the EpR, i.e. QDTY 367 LVLDKWL, SFEY 425 TILDPSSQ, HLKY 453 LYLVVSD, KYLY 455 LVVSDSG, STDY 467 SSGGSQG, DGPY 484 -SHPYENS, SHPY 488 ENSLVPD,and HPGY 503 VACS. The positive control peptide used was a mixture of 11-mer peptides with the sequence XXXYXXXXXXX, where X is any amino acid and was synthesized using an amino acid mixture. Horseradish peroxidase-conjugated second antibodies were purchased from Jackson Immunoresearch (West Grove, PA) and protein-grade Nonidet P-40 from Calbiochem. The enhanced chemiluminescence Western blotting reagents were obtained from Amersham Corp. Wortmannin, PI, PI-4-P, PI-4, 5-P 2 , L-␣-phosphatidyl-Lserine and all other reagents were purchased from Sigma unless otherwise indicated.
Cells and Proliferation Assays-The murine interleukin-3-dependent DA-3 cell line (generously provided by Dr. J. Ihle, St. Judes Children's Hospital, Memphis, TN) was retrovirally infected with a JZen TKneo vector, as described previously (9), containing either the wild-type murine EpR cDNA or site-directed mutants of the murine EpR in which the individual eight tyrosines in the cytoplasmic domain of the EpR were converted to phenylalanines. Clones expressing similar levels of cell surface EpRs, based on FACS analysis with biotinylated erythropoietin (4), were selected and grown in RPMI containing 10% fetal calf serum, 1.8 mg/ml G418, and 5 ng/ml of COS cell-derived mouse interleukin-3.
For proliferation assays, the various EpR-infected DA-3 cells were grown to near confluence with interleukin-3, washed once with RPMI, and resuspended in RPMI containing 10% fetal calf serum. The cells were then aliquoted into 96-well U-bottom microtiter plates (Linbro, ICN, Mississauga, Ontario) to give a final volume of 0.1 ml/well and the plates incubated for 20 h at 37°C in a humidified atmosphere of 5% CO 2 , 95% air. Twenty l of a 50 Ci/ml solution of [ 3 H]thymidine in RPMI was then added to each well to give a final concentration of 1 Ci/well. After another 2 h at 37°C, the contents of the wells were harvested onto filtermats and counted using an LKB Betaplate Harvester and Liquid Scintillation Counter (LKB Wallac, Turku, Finland).
Blotting with 32 P-Labeled SH2 Domains of p85-GST fusion proteins of the NH 2 -and COOH-terminal SH2 domains of p85, bound noncovalently to glutathione-agarose beads, were prepared as described previously (16) from the pGEX-2TK plasmid containing the protein kinase A catalytic subunit recognition sequence. The beads were washed once in HMK buffer (20 mM Tris, pH 7.5, 100 mM NaCl, 12 mM MgCl 2 ) and resuspended in a 2-3 bead volume of HMK buffer con-taining a 1 unit/ml concentration of the catalytic subunit of cAMPdependent protein kinase (Sigma), 1 mC/ml [␥-32 P]ATP (3000 Ci/mmol, DuPont), and 1 mM dithiothreitol. The kinase reaction was allowed to proceed for 30 min and then terminated by the addition of 1 ml of stop buffer (10 mM sodium phosphate, pH 8, 10 mM sodium pyrophosphate, 10 mM EDTA, 1 mg/ml bovine serum albumin) and the GST fusion beads washed five times with phosphate-buffered saline. The labeled fusion proteins were then eluted in 10 bead volumes of 20 mM reduced glutathione, 100 mM Tris, pH 8, 120 mM NaCl for 30 min. The labeled supernatant was counted and diluted in Western blocking buffer (PSB containing 5% bovine serum albumin) at a concentration of 100,000 counts/min/ml. Blotting was carried out with polyvinylidene difluoride membranes (Millipore, MA), blocked previously for 1 h with PSB, 5% bovine serum albumin at 23°C, and then incubated for 3 h with the 100,000 counts/min/ml 32 P-labeled fusion protein. Blots were washed and exposed overnight to Kodak X-Omat film (Eastman Kodak).
Site-directed Mutagenesis-A 1.7-kilobase Kpn fragment of the mEpR cDNA was cloned into M13mp18 and was mutated using a modification of the method of Kunkel (26). Briefly, uracil-containing single-stranded DNA template was generated by growth of the phage in E. coli strain CJ236 (dut, ung, thi, relA; pCJ105 (Cm r )) and 0.3 pmol was annealed to 6 pmol of the mutagenic oligonucleotide prior to synthesis of the complementary strand with unmodified T7 DNA polymerase and ligation with T4 DNA ligase to form covalently closed circular DNA. The conditions for the extension and ligation reactions were 23 mM Tris, pH 7.9, 5 mM MgCl 2 , 35 mM NaCl, 1.5 mM dithiothreitol, 0.4 mM dATP, dCTP, dGTP, and dTTP, 0.75 mM ATP, 1 unit of T7 DNA polymerase (Pharmacia) and 2 units of T4 DNA ligase (Life Technologies Inc.L) at 23°C for 4 h in a final volume of 13 ml. The mutagenesis mixture was then diluted with water to an appropriate level for transformation into E. coli strain MV1190 (⌬ lac-pro AB), thi, supE, (⌬ sr1-recA) 306:Tn10(tet r ) (FЈ:traD36, proAB, lacI q Z⌬M15)). Plaques were picked and phage minipreps prepared and screened for mutants by dot-blot hybridization with 32 P-labeled mutagenic oligonucleotide. All mutations were verified by DNA sequencing and subcloned into the JZen TKneo retroviral vector for infection of DA-3 cells as described previously (9).
SH2 Binding and Phosphopeptide Inhibition Assays-Nonidet P-40 lysates from DA-ER cells, incubated in the presence or absence of erythropoietin, were incubated at 4°C for 12 h with glutathioneagarose beads bearing GST fusion proteins (0.5 ml of lysate with 20 l of packed beads containing approximately 5 g of fusion protein) containing the COOH-or NH 2 -terminal SH2 domain of p85. For phosphopeptide inhibition studies, the beads were first exposed to tyrosinephosphorylated peptides (50 M) corresponding to the eight tyrosinephosphorylated regions within the intracellular domain of the EpR for 20 min at 4°C before adding the cell lysates. The beads were then washed three times in PSB containing 0.5% Nonidet P-40, boiled in SDS-sample buffer, and the eluted proteins subjected to Western blot analysis with anti-PY antibodies.
In Vitro PI 3-Kinase Assay-This assay was performed as described previously (16).

RESULTS
Studies from both our laboratory (16) and that of Miura et al. (19) recently established that the activated EpR is capable of binding in vitro to the SH2 domains of the p85 subunit of PI 3-kinase. However, one issue that remained unresolved at the completion of these studies was whether this binding was direct or through an intermediary protein. This was an important question since none of the eight tyrosines within the intracellular domain of the EpR possess the canonical Tyr-Met/Val-X-Met binding motif for the SH2 domains of p85 (27,28). To address this, DA-ER cells (i.e. DA-3 cells expressing approximately 12,000 EpRs/cell (8)) were incubated in the presence and absence of erythropoietin for 5 min at 37°C, and anti-EpR or anti-p85 immunoprecipitations carried out with the cell lysates. An anti-EpR immunoblot of the anti-EpR immunoprecipitates demonstrated equal loading of EpRs and revealed, as has been shown previously (9), that erythropoietin stimulates an increase in the apparent molecular mass of a fraction of the total EpRs from approximately 66 to 72 kDa (Fig. 1A). An anti-PY immunoblot of identical aliquots of anti-EpR immunoprecipitates demonstrated, as has been shown previously (9,10,12), that this 72 kDa species represents the tyrosine phosphorylated form of the EpR (Fig. 1B). An anti-EpR immunoblot of anti-p85 immunoprecipitates from these same cells revealed that p85 associates only with this 72 kDa form of the EpR (Fig.  1C) and blotting studies carried out with anti-EpR immunoprecipitates (using a mixture of 32 P-labeled GST fusion proteins containing the NH 2 -and COOH-terminal SH2 domains of p85) showed that the SH2 domains of p85 bound directly to the 72-kDa tyrosine-phosphorylated EpR (Fig. 1D). Direct binding of each individual p85 SH2 domain was confirmed by incubating SDS-denatured EpRs with bead-bound NH 2 -and COOHterminal p85 SH2-GST fusion proteins. As can be seen in Fig.  2, both the NH 2 -and COOH-terminal SH2 of p85 recognized SDS-denatured, tyrosine-phosphorylated EpRs.
To determine which of the eight tyrosines within the intracellular region of the tyrosine-phosphorylated EpR was responsible for p85 binding, synthetic phosphopeptides corresponding to these eight regions of the EpR were tested for their ability to inhibit, in vitro, the binding of the activated EpR to agarosebound GST fusion proteins containing either the NH 2 -or COOH-terminal SH2 of p85. As can be seen in Fig. 3, A and B, only one phosphopeptide was capable of inhibiting this binding. This phosphopeptide corresponded to the region flanking the most COOH-terminal tyrosine within the EpR, i.e. Tyr 503 .
From our in vitro data to this point, it appeared that the phosphorylated Tyr 503 within the EpR might be playing a role in the binding of PI 3-kinase. However, to test if this was actually the case in vivo, we converted each of the eight tyrosines within the cytoplasmic domain of the EpR to phenylalanines, using site-directed mutagenesis. Clones bearing similar cell surface EpRs were selected using biotinylated erythropoietin and FACS analysis as described previously (4). Western blot analysis of these selected clones, using anti-EpR antibodies, revealed less than a 4-fold variation in EpR numbers among the nine clones (Fig. 4A). To address the more relevant question of the EpR level at the cell surface, 125 Ierythropoietin binding studies were carried out with the nine clones and the relative levels were found to be similar to those observed when total EpR numbers were assessed (data not shown). These clones were treated with erythropoietin for 5 min at 37°C and half of the cell lysates incubated with agarose beads containing the NH 2 -or COOH-terminal SH2 domain of p85. Bound proteins were eluted and subjected to Western analysis with anti-PY antibodies to monitor the level of activated EpRs. As can be seen in Fig. 4B, the only EpR mutant that did not appear to bind to either the NH 2 -or COOHterminal SH2 domain of p85 was Y503F. However, this conclusion would not be valid if Tyr 503 was the only tyrosine normally phosphorylated following erythropoietin stimulation. To test this, one-quarter of the cell lysates were immunoprecipitated with anti-EpR antibodies and subjected to Western analysis with anti-PY antibodies. As can be seen in Fig. 4B (lower  panel), this immunoblot clearly shows that the Y503F EpR mutant is tyrosine-phosphorylated. The reduced level of tyrosine phosphorylation of this mutant, compared, for example, to a mutant with similar EpR surface expression (e.g. Y453F and Y455F) also indicates, as our results to this point would suggest, that Tyr 503 is a site of tyrosine phosphorylation.
To examine if p85 actually binds the activated EpR in vivo at position Tyr 503 , the remaining one-quarter of the cell lysates were subjected to immunoprecipitation with anti-p85. As can be seen from the anti-PY immunoblot shown in Fig. 5, the only mutant EpR that did not co-precipitate with p85 was Y503F. The anti-PY immunoblot of the anti-EpR immunoprecipitates shown in Fig. 4B (lower panel) demonstrates that this mutant was tyrosine phosphorylated in response to erythropoietin at residues other than Tyr 503 and that these phosphorylated tyrosines were not involved in binding p85 in vivo.
To ensure that our results with the selected EpR cell line containing Y503F were not due to a cloning artifact but were representative of this mutant, anti-p85 immunoprecipitates were carried out with lysates from two additional, independently isolated Y503F clones. As shown in Fig. 6, no co-precipitation of activated EpRs was observed with these clones either.
We (16) and others (17,19) have shown previously that PI 3-kinase activity becomes associated with the EpR following ligand binding. To determine if this association occurs solely through the interaction of its p85 subunit with the activated EpR, PI 3-kinase assays were carried out with wild-type and Y503F EpR expressing DA-3 cells, using clones expressing identical cell surface EpR levels. As expected, the generation of PIP 3 was observed in EpR immunoprecipitates from erythropoietin-stimulated wild-type but not Y503F EpR containing DA-3 cells (Fig. 7). On the basis of this correlation, we conclude that the major binding site within the EpR for PI 3-kinase is the phosphorylated sequence Tyr 503 -Val-Ala-Cys.
Having established that PI 3-kinase binds to the activated EpR via the SH2 domains of the former with the phosphorylated Tyr 503 of the latter, we set out to determine the biochemical and biological consequences of this interaction by first comparing tyrosine phosphorylation events stimulated by erythropoietin in DA-3 cells expressing the same number of cell surface wild-type and Y503F EpRs. Interestingly, anti-PY immunoblots of anti-PY immunoprecipitates revealed that there were no obvious differences in the intensities of the major phosphoproteins, aside from the expected difference in the EpR itself (Fig. 8A). Anti-PY immunoblots of anti-Shc (Fig. 8B) and anti-Jak 2 (Fig. 8C) immunoprecipitates confirmed this finding.
We next compared the erythropoietin-responsive proliferation of DA-3 cells expressing wild-type and Y503F EpRs. Because of previous studies in which we found that erythropoietin-responsive proliferation was highly dependent on cell surface EpR levels (9), clones expressing identical numbers of cell surface EpRs, based on 125 I-erythropoietin binding studies, were again used. As can be seen in Fig. 9A, erythropoietininduced proliferation of the two cell types was very similar, suggesting that EpR-associated PI 3-kinase activity does not play a role in erythropoietin-induced proliferation. However, since PI 3-kinase has recently been shown to be directly activated by Ras GTP in COS cells (29) we investigated whether PI 3-kinase might play a role in erythropoietin-induced proliferation, but at a locus other than the EpR. Specifically, we examined the effect of the PI-3 kinase inhibitor, wortmannin (30), on erythropoietin-induced proliferation of DA-3 cells expressing wild-type and Y503F EpRs. Interestingly, wortmannin reduced the erythropoietin responsiveness of both wild-type (Fig. 9B) and Y503F EpR (Fig. 9C) expressing DA-3 cells to approximately the same degree, i.e. between one and two logs, suggesting, perhaps, that PI 3-kinase is activated by erythropoietin in Y503F EpR expressing DA-3 cells and that PI 3-kinase plays a FIG. 4. A, EpR levels in biotinylated Ep/FACS selected DA-3 cells expressing wild and mutant EpRs. Lysates from 2 ϫ 10 6 unstimulated DA-3 cells expressing wild-type (WT) EpRs or the eight tyrosine to phenylalanine points were subjected to SDS-PAGE and Western analysis using anti-EpR antibodies. Numbers above the autoradiogram refer to relative intensities of the EpR bands, determined by densitometry. B, a Y503F EpR point mutant does not bind to p85 SH2 domains in vitro. Upper panels, lysates from erythropoietin-stimulated DA-3 cells expressing tyrosine to phenylalanine point mutants of the EpR were incubated with bead-bound GST-SH2 fusion proteins containing the NH 2 -or COOH-terminal SH2 domain of p85 and tightly bound proteins were eluted by boiling in SDS-sample buffer and subjected to Western analysis with the anti-PY antibody 4G10. Lower panel, identical lysates were subjected to immunoprecipitation with anti-EpR antibodies and the precipitates analyzed by Western blotting with 4G10 in order to detect activated EpRs.
FIG. 5. Endogenous p85 does not associate with the activated Y503F EpR in vivo. Lysates from erythropoietin-stimulated DA-3 cells expressing the wild-type and each of the tyrosine to phenylalanine EpR point mutants were immunoprecipitated with anti-p85 (upper) or anti-EpR antibodies (see Fig. 4B, lower panel). Tightly bound proteins were eluted with SDS-sample buffer and subjected to Western analysis with 4G10 in order to detect activated EpRs. The (Ϫ) lane represents a lysate from unstimulated DA-3 cells expressing the wild-type EpR. significant role in erythropoietin-induced proliferation but at a site distinct from the EpR.
To confirm this, PI 3-kinase assays were carried out with anti-PY immunoprecipitates from both wild-type and Y503F EpR expressing DA-3 cells, treated with or without erythropoietin. As can be seen in Fig. 10, almost as much PIP 3 was generated in response to erythropoietin in the Y503F EpR as in the wild-type EpR expressing DA-3 cells, consistent with the notion that PI 3-kinase is activated by erythropoietin both at the level of the EpR and also at a locus distinct from the EpR.

DISCUSSION
Within minutes of binding erythropoietin, the EpR becomes phosphorylated on tyrosine residues and attracts a number of intracellular proteins to its intracellular domain (8,16,18,31). As we (16) and others (17,19,31) have shown, one of these proteins is PI 3-kinase. In a previous report we demonstrated that PI 3-kinase becomes rapidly associated with the EpR following erythropoietin stimulation of Ba/F3 cells expressing high levels of the murine EpR and that this interaction might be mediated by the SH2 domains of p85 and tyrosine-phosphorylated motifs within the EpR (16). This was subsequently confirmed by both Miura et al. (19) and He et al. (17). Miura et al. (19) went on to show that PI 3-kinase did not associate with a mitogenically active EpR which lacked the COOH-terminal 108 amino acids and therefore concluded that p85 binds to the carboxyl terminus of the EpR and that this association is not required for transducing a mitogenic signal. He et al. (17), on the other hand, using various EpR mutants with cytosolic truncations and deletions concluded that p85 binds to a more membrane proximal region of the EpR, i.e. between Pro 329 and Glu 374 in the extended box-2 region. In the current study we have tried to resolve this controversy and further explore the nature of this interaction. For this study we used DA-3 cells since they do not express any endogenous EpRs (12). The data presented herein suggest that the binding between p85 and the activated EpR is direct and not through an intermediate protein, such as Syp, which has recently been shown to bind Grb2 to the platelet-derived growth factor receptor (32,33). The data also suggest that the NH 2 -and COOH-terminal SH2 domains of p85 bind specifically to the phosphorylated form of the EpR at Tyr 503 . This not only demonstrates for the first time that Tyr 503 within the EpR is phosphorylated in vivo in response to erythropoietin binding (and that at least one other tyrosine within the EpR is phosphorylated as well) but reveals a previously unreported recognition motif for the SH2 domains of p85, i.e. Y-V-A-C. This is not inconsistent with the exhaustive synthetic phosphopeptide studies carried out by Songyang et al. (28) since they did not test cysteine at position 3ϩ and since cysteine and the bulkier methionine, which is the established amino acid at position 3ϩ, both provide a hydrophobic side chain with sulfer atoms positioned in a similar orientation. We also found that p85 COOH-terminal SH2 beads bound more activated EpRs, in repeated experiments, than an equivalent amount of bead-bound NH 2 -terminal SH2 (as shown in Fig. 2). This was consistent with peptide inhibition studies where we found that approximately 10-fold less phosphopeptide Y503 was needed to completely block activated EpR binding to COOH-terminal SH2 (data not shown). This is in keeping with the difference in affinity reported for the NH 2 -and COOHterminal SH2 domains for the phosphorylated, established con- FIG. 7. PI 3-kinase activity is not associated with the activated Y503F EpR in DA-3 cells. DA-3 cells expressing the wild-type or Y503F EpR were incubated for 5 min at 37°C with erythropoietin and the lysates immunoprecipitated with anti-p85 or anti-EpR antibodies. Immunoprecipitated protein A beads were washed, PI 3-kinase assays carried out, and thin layer chromatography performed as described previously (16). sensus sequence, Tyr-Met/Val-X-Met (28).
As far as the ramifications of PI 3-kinase binding to the activated EpR are concerned, our studies with the Y503F EpR mutant suggest that neither erythropoietin induced prolifera-tion nor tyrosine phosphorylation events are affected by the absence of an EpR-associated PI 3-kinase. This is interesting, given that in many cases there is a good correlation between growth factor stimulation and a rapid increase in the levels of the lipid products of PI-3 kinase, particularly PI 3,4-bisphosphate and PI 3,4,5-trisphosphate (34,35). Since these products are poor substrates for any known phospholipase C (36,37), it has been postulated that they may represent a new class of second messengers involved in mitogenic signal transduction, perhaps, in part, via activation of S6 kinase (38,39). Very recent studies, however, suggest that not all growth factors require PI 3-kinase for cell growth (40) and that this enzyme might function in some cells to stimulate events unrelated to mitogenesis, such as glucose transport (41,42), membrane ruffling (43), differentiation (44,45), cell survival (46), or endocytic trafficking of activated receptors (47). This last finding prompted us to investigate whether the activated Y503F EpR was internalized at a slower rate than the wild-type EpR in DA-3 cells, using 125 I-erythropoietin, but no difference was observed (data not shown). An extra complication in studies designed to investigate the role of the lipid products of PI 3-kinase is that this enzyme has both lipid and serine kinase activity (48,49), and inhibitors like wortmannin inhibit both activities. Thus the substantial reduction in responsiveness to erythropoietin that we observe with both our wild-type and Y503F EpR expressing DA-3 cells could be attributed to the loss of serine kinase activity. Consistent with this possibility, Lam et al. (50) recently reported that PI 3-kinase serine phosphorylates IRS-1 in response to insulin in rat adipocytes. From our studies with wortmannin and our PI 3-kinase assays demonstrating activation of PI 3-kinase in response to erythropoietin in Y503F EpR expressing cells, we tentatively conclude that PI 3-kinase is involved in erythropoietin-induced proliferation at a site other than the EpR. However, the greater than 10-fold drop in erythropoietin responsiveness we detect with wortmannin might represent an underestimate of the role of PI 3-kinase involvement in erythropoietin-induced mitogenesis since eight PI 3-kinases have been found in mammals to date and it is not yet known how many of these are wortmanninsensitive (51). On the other hand, the observed inhibition could be an overestimate since wortmannin, even at the low concentrations used in our studies, might be exerting its effects through molecules other than PI 3-kinase (51,52). Based on both our previous observations that erythropoietin stimulates the tyrosine phosphorylation of Shc and its association with Grb2, (8) and recent reports that Ras can activate PI 3-kinase by directly interacting with its catalytic p110 subunit (29), we  10. PI 3-kinase activity is stimulated by erythropoietin in Y503F EpR expressing DA-3 cells. DA-3 cells expressing the wildtype or Y503F EpR were incubated for 5 min at 37°C with or without erythropoietin and the lysates immunoprecipitated with anti-PY antibodies. Immunoprecipitated protein A beads were washed, PI 3-kinase assays carried out, and thin layer chromatography performed as described previously (16). An anti-p85 immunoprecipitate was also assayed as a positive control (first lane).
hypothesize that this site could be immediately downstream from Ras.
In summary, we have demonstrated that PI 3-kinase becomes directly associated with the EpR, following ligand binding, at Tyr 503 within the EpR and that this binding is not critical for erythropoietin-induced mitogensis. We are currently exploring whether this association may play a role in erythropoietin-induced glucose uptake or erythroid differentiation.