Erythropoietin Induces the Tyrosine Phosphorylation of Insulin Receptor Substrate-2

In this report, we demonstrate that insulin receptor substrate-2 (IRS-2) is phosphorylated on tyrosine following treatment of UT-7 cells with erythropoietin. We have investigated the expression of IRS-1 and IRS-2 in several cell lines with erythroid and/or megakaryocytic features, and we observed that IRS-2 was expressed in all cell lines tested. In contrast, we did not detect the expression of IRS-1 in these cells. In response to erythropoietin, IRS-2 was immediately phosphorylated on tyrosine, with maximal phosphorylation between 1 and 5 min. Tyrosine-phosphorylated IRS-2 was associated with phosphatidylinositol 3-kinase and with a 140-kDa protein that comigrated with the phosphatidylinositol-3,4,5-trisphosphate 5-phosphatase, SHIP. Moreover, IRS-2 was constitutively associated with the erythropoietin receptor. We did not observe the association of IRS-2 with JAK2, Grb2, or PTP1D. Using BaF3 cells transfected with mutated erythropoietin receptors, we demonstrate that neither the tyrosine residues of the intracellular domain nor the last 109 amino acids of the erythropoietin receptor are required for erythropoietin-induced IRS-2 tyrosine phosphorylation. Altogether, our results indicate that erythropoietin-induced IRS-2 tyrosine phosphorylation could account for the previously reported activation of phosphatidylinositol 3-kinase mediated by erythropoietin receptors mutated in the phosphatidylinositol 3-kinase-binding site (Damen, J., Cutler, R. L., Jiao, H., Yi, T., and Krystal, G. (1995) J. Biol. Chem. 270, 23402–23406; Gobert, S., Porteu, F., Pallu, S., Muller, O., Sabbah, M., Dusanter-Fourt, I., Courtois, G., Lacombe, C., Gisselbrecht, S., and Mayeux, P. (1995)Blood 86, 598–606).

Insulin receptor substrate-1 (IRS-1) 1 is a major substrate of the IGF-1 and insulin receptors (1). IRS-1 is a hydrophilic protein with a theoretical molecular mass of 131 kDa that migrates between 160 and 185 kDa on SDS-polyacrylamide gel electrophoresis partially because of a high serine phosphorylation state (2,3). IRS-1 contains a pleckstrin homology domain, a PTB domain, and at least 20 potential tyrosine phosphorylation sites including nine YXXM motifs that are consensus binding sites for the SH2 domains of the regulatory subunit (p85) of PI 3-kinase (4). IRS-1 binds to the tyrosine-phosphorylated IGF-1 and insulin receptors through its PTB domain and becomes a docking protein for signaling proteins such as PI 3-kinase, Grb2, PTP1D, and Nck after tyrosine phosphorylation (5). Moreover, a recent report shows that the pleckstrin homology domain could also be involved in the association of IRS-1 with the insulin receptor (6). More recently, a second IRS protein was identified that was designated IRS-2 (7). This protein corresponds to the previously identified 4PS protein (for IL-4-induced phosphotyrosine substrate) (8,9). IRS-2 exhibits a high structural similarity to IRS-1, with a strong conservation of the PTB and pleckstrin homology domains. Moreover, tyrosine phosphorylation sites shown to bind PI 3-kinase, Grb2, and PTP1D in IRS-1 are also conserved in IRS-2.
Not only insulin and IGF-1 receptors, but also several cytokine and interferon receptors can induce tyrosine phosphorylation of IRS-1 or IRS-2 (9 -18). IRS-1 and IRS-2 activation by IL-4 is well documented. Indeed, it has been shown that IRS-1 and IRS-2 bind to a peptidic sequence of the activated IL-4 receptor with a typical NPXY PTB domain-binding motif (19), and IRS-1 or IRS-2 expression appears to be required for IL-4induced mitogenesis (10). However, other cytokine receptors that induce the tyrosine phosphorylation of IRS-1 and IRS-2 do not possess typical PTB domain-binding motifs, and the mechanism of IRS-1 and IRS-2 activation by these receptors remains unknown.
The erythropoietin (Epo) receptor also belongs to the cytokine receptor family (20,21). Epo binding to its receptor activates the receptor-associated JAK2 tyrosine kinase (22) and induces the tyrosine phosphorylation of the Epo receptor (23)(24)(25)(26) and other proteins. Several intracellular signaling pathways are subsequently activated, including mitogen-activated protein kinases (27,28), STAT5 (29,30), and PI 3-kinase (31)(32)(33)(34). PI 3-kinase was shown to associate with phosphorylated Tyr 479 of the Epo receptor (35), and removal of this tyrosine residue abrogates PI 3-kinase association with the Epo receptor (34 -36). However, binding of PI 3-kinase to the Epo receptor does not appear to be required for Epo-induced PI 3-kinase  activation since Epo receptors devoid of Tyr 479 still activate PI  3-kinase (35,36). These results strongly suggest that Epo could activate PI 3-kinase by several mechanisms. However, these alternate pathways for Epo-induced PI 3-kinase activation have not been identified up to date. We have previously shown that one of these mechanisms required only the first 127 amino acids of the Epo receptor intracellular domain and was not dependent on the phosphorylation of the single tyrosine residue (Tyr 343 ) present in this region (36).
In this report, we show that erythropoietin induces the tyrosine phosphorylation of IRS-2. Epo-induced IRS-2 tyrosine phosphorylation does not require the tyrosine residues of the intracellular domain of the Epo receptor, and IRS-2 appears to be constitutively associated with the Epo receptor. After Epoinduced tyrosine phosphorylation, IRS-2 associates with PI 3-kinase and with a tyrosine-phosphorylated protein comigrating with SHIP (p140). In contrast, we did not detect the association of Grb2 with IRS-2 in Epo-stimulated cells. Thus, our results strongly suggest that IRS-2 binding could be an alternate mechanism for Epo-induced PI 3-kinase activation.

EXPERIMENTAL PROCEDURES
Materials-The highly purified recombinant human Epo (specific activity of 120,000 units/mg) used throughout this study was a generous gift from Dr. M. Brandt (Boehringer Mannheim). Recombinant SCF was kindly provided by Dr. A. Shimosaka (Kirin Brewery Co., Tokyo). Anti-IRS-1 (catalog number 06-248), anti-IRS-2 (catalog number 06-506), and anti-JAK2 (catalog number 06-255) antibodies were from Upstate Biotechnology, Inc. Anti-PI 3-kinase (p85 subunit; catalog number P13030) and anti-SHC (catalog number S14630) antibodies were purchased from Transduction Laboratories, and anti-Grb2 antibodies (catalog number C-23) were from Santa Cruz. Anti-phosphotyrosine antibodies (4G10) were a generous gift from Dr. B. Drucker. Anti-Epo receptor antibodies were produced by immunizing rabbits with a recombinant protein composed of the full intracellular domain of the human Epo receptor fused to GST. Anti-GST antibodies were also prepared in our laboratory using the same protocol. Whole anti-GST and anti-Epo receptor sera were used.
Cell Lines and Cell Culture-A subclone of the human leukemic cell line UT-7 (37) able to grow in SCF, GM-CSF, or Epo was established. These cells were cultured in ␣-minimal essential medium containing 5% fetal calf serum complemented with 2 units/ml Epo, 2.5 ng/ml GM-CSF, or 50 ng/ml SCF. Before each experiment, the cells were serum-and growth factor-deprived by incubation overnight in Iscove's modified Dulbecco's medium (Life Technologies, Inc., catalog number 31980-022) containing 0.1% deionized bovine serum albumin. HCD57 cells (38), Mo7E cells (39), and TF-1 cells (40,41) were cultivated in ␣-minimal essential medium containing 5% fetal calf serum and complemented with 2 units/ml Epo (HCD57) or 2.5 ng/ml GM-CSF (Mo7E and TF-1). BaF3 (42) and FDCP-1 (43) cells were cultivated in the same medium complemented with 2.5% WEHI-conditioned medium as a source of IL-3. WEHI-conditioned medium is the supernatant of end-logarithmic phase cultures of WEHI-3B cells (ATCC TIB-68). BaF3 cells were transfected with expression vectors encoding wild-type or mutated Epo receptors and selected for their ability to grow in Epo as described previously (36). Before each experiment, these cells were serum-and growth factor-deprived by incubation for 5 h in Iscove's modified Dulbecco's medium.
IRS-1 and IRS-2 Detection in Whole Cell Extracts-Exponentially growing hematopoietic cells were washed twice with phosphate-buffered saline and solubilized by boiling in electrophoresis sample buffer at a cell density of 5 ϫ 10 6 cells/ml. Samples corresponding to 5 ϫ 10 5 cells (or 2.5 ϫ 10 5 cells for NIH 3T3 control cells) were separated by SDS-polyacrylamide gel electrophoresis using 6.5% polyacrylamide gels and analyzed by Western blotting.
Immunoprecipitation and Western Blotting-Immunoprecipitations and Western blotting were done as described previously (25) except that 1% Brij 96 was used to solubilize the cells instead of 1% Nonidet P-40. ECL (Amersham Ltd., Les Ullis, France) was used for Western blot visualization. Each experiment was performed at least twice with similar results.

Epo-sensitive Hematopoietic Cell Lines Express IRS-2, but
Not IRS-1-To test for the possible involvement of IRS-1 or IRS-2 in the Epo mode of action, we first investigated whether these proteins were expressed in erythroid and other hematopoietic Epo-responsive cells. As a positive control, we used an NIH 3T3 cell lysate that was previously shown to express both IRS-1 and IRS-2 (18). Total cell lysates from these cells were analyzed by Western blotting. As shown in Fig. 1, all hematopoietic cells tested expressed IRS-2, although IRS-2 expression appeared to be relatively low in Mo7E cells. IRS-2 migrated as diffuse bands with apparent molecular masses of 160 -175 kDa in the murine cells (T3Cl2, BaF3, FDCP-1, HCD57, and NIH 3T3) and 170 -190 kDa in the human cells (UT-7, Mo7E, and TF-1). In contrast, IRS-1 expression was not detected in these cells. (The faint band in Fig. 1 (lower panel) was exactly at the same position as IRS-2, and its intensity varied as did the IRS-2 signal in Fig. 1 (upper panel). Thus, it most likely corresponds to some cross-reactivity of the anti-IRS-1 antiserum with IRS-2.) In another experiment, we immunoprecipitated UT-7 cell extracts with anti-IRS-1 antibodies, and we analyzed these immunoprecipitates by Western blotting using anti-IRS-1 antibodies. Again, no IRS-1 protein was detected in these cells (data not shown).
Epo Induces the Tyrosine Phosphorylation of IRS-2 in UT-7 Cells-Serum-and growth factor-deprived UT-7 cells were stimulated for 2.5 min with 25 ng/ml GM-CSF, 10 units/ml Epo, or 100 ng/ml SCF. To ensure maximal response to the cytokine, each cell population was previously cultured for at least 3 weeks in the corresponding growth factor. Cell lysates were immunoprecipitated with anti-IRS-2 antibodies and analyzed by Western blotting using anti-phosphotyrosine antibodies. As shown in Fig. 2A, Epo (but not GM-CSF or SCF) induced the tyrosine phosphorylation of two proteins of 180 and 140 kDa immunoprecipitated by anti-IRS-2 antibodies. Reprobing the blot with anti-IRS-2 antibodies confirmed that the 180-kDa protein was IRS-2. To control the efficiency of cytokine stimulation, the same extracts were immunoprecipitated with anti-JAK2 (Epo-or GM-CSF-stimulated cells) or with anti-c-Kit (SCF-stimulated cells) antibodies and analyzed by anti-phosphotyrosine Western blotting. As shown in Fig. 2B, JAK2 was activated by GM-CSF, although with a lower intensity than by Epo. Moreover, a strong tyrosine phosphorylation of c-Kit was detected in SCF-stimulated UT-7 cells. Thus, although GM-CSF and SCF receptors were also efficiently stimulated, only Epo induced the tyrosine phosphorylation of IRS-2 in UT-7 cells.
The tyrosine phosphorylation of IRS-2 was maximal after 1-5 min of Epo stimulation and then started to decrease, although it remained detectable after 1 h of stimulation (Fig. 3). The tyrosine phosphorylation of the IRS-2-associated 140-kDa protein seemed to be more transient. Indeed, it appeared to be induced also very quickly, but to disappear earlier, and it was only barely detectable after 30 min of Epo stimulation. In addition, another faintly stained band was observed in Fig. 3. It exhibited a molecular mass of 70 kDa and could correspond to the tyrosine-phosphorylated Epo receptor. A typical Epo doseresponse curve is shown in Fig. 4. Epo-induced IRS-2 tyrosine phosphorylation exhibited a strong correlation with Epo binding to its receptor. The dose-response curve for Epo-induced tyrosine phosphorylation of IRS-2 was similar to the previously reported dose-response curves for Epo-induced tyrosine phosphorylation of its own receptor (25) and Epo-induced PI 3kinase activation (33). Epo-induced IRS-2 tyrosine phosphorylation in UT-7 cells was consistently observed in 11 independent experiments. Moreover, Epo-induced IRS-2 tyrosine phosphorylation was observed in HCD57 cells (data not shown) and in BaF3 cells transfected with the Epo receptor (see below).
IRS-2 Associates with PI 3-Kinase in Epo-stimulated UT-7 Cells-We then tested the association of tyrosine-phosphorylated IRS-2 with signaling proteins after Epo stimulation of UT-7 cells. The blot presented in Fig. 3 was reprobed with anti-PI 3-kinase regulatory subunit (p85) antibodies. As shown in Fig. 5A, p85 was clearly detected in IRS-2 immunoprecipitates from Epo-stimulated cells. To confirm that PI 3-kinase associates with tyrosine-phosphorylated IRS-2, lysates from Epo-stimulated UT-7 cells were immunoprecipitated with anti-p85 antibodies and analyzed by Western blotting using antiphosphotyrosine antibodies (Fig. 5B). The tyrosine phosphorylation of a band comigrating with IRS-2 was significantly increased in Epo-stimulated cells. Most likely, this band corresponded to IRS-2, although only a very faint band was detected by anti-IRS-2 antibodies in anti-p85 immunoprecipitates due to the relatively poor sensitivity of the anti-IRS-2 antiserum (data not shown). We then tested whether Grb2, SHC, or PTP1D also associated with IRS-2 in Epo-stimulated UT-7 cells. Lysates from Epo-treated UT-7 cells were immunoprecipitated with antibodies specific for these proteins and analyzed by Western blotting using anti-phosphotyrosine and anti-IRS-2 antibodies. No IRS-2 protein either tyrosine-phosphorylated or not was detected in these immunoprecipitates (data not shown). Moreover, Grb2, SHC, and PTP1D were not detected in anti-IRS-2 immunoprecipitates analyzed by Western blotting with antibodies specific for these proteins (data not shown). In contrast, IRS-2 was associated with two tyrosine-phosphorylated proteins of 140 and 70 kDa, respectively, in Epo-stimulated cells (see Fig. 3). In many cells, Epo induces the tyrosine phosphorylation of a 140-kDa SHC-associated protein that was identified as the PI-3,4,5-trisphosphate 5-phosphatase, SHIP (44 -47). Since anti-SHIP antibodies are not yet commercially available, the ability of this protein to associate with SHC was used to identify SHIP (48). As shown in Fig. 6, the 140-kDa protein that was immunoprecipitated with anti-IRS-2 antibodies comigrated with the SHC-associated SHIP and most likely  (49), receptor occupancy was 2% for 10 milliunits/ml Epo, 18% for 100 milliunits/ml, 70% for 1 unit/ml, 95% for 10 units/ml, and 99.5% for 100 units/ml. corresponds to this protein. The apparent molecular mass of the other IRS-2-associated protein suggests that it could be the tyrosine-phosphorylated Epo receptor. To test this hypothesis, anti-Epo receptor immunoprecipitates were probed with antiphosphotyrosine antibodies. As shown in Fig. 7, only a faint band comigrating with IRS-2 was detected in extracts from Epo-stimulated cells. We next probed the same blot with anti-IRS-2 antibodies. Surprisingly, the IRS-2 protein was easily detected in anti-Epo receptor immunoprecipitates from both resting and Epo-stimulated UT-7 cells (Fig. 7, lower panel), showing that IRS-2 was constitutively associated with the Epo receptor. Control anti-GST antibodies did not immunoprecipitate any protein revealed by either anti-phosphotyrosine or anti-IRS-2 antibodies.
Tyrosine Phosphorylation of the Epo Receptor Is Not Required for Epo-induced IRS-2 Phosphorylation-BaF3 cells were transfected with the wild-type Epo receptor or with the Epo receptor devoid of tyrosine residues in its intracellular domain (ZERO) and selected for growth in the presence of Epo. These cells were stimulated for 2.5 min with 10 units/ml Epo. Cellular extracts were immunoprecipitated with anti-IRS-2 antibodies and analyzed by Western blotting using anti-phosphotyrosine antibodies (Fig. 8). As shown in Fig. 8, Epo induced the tyrosine phosphorylation of IRS-2 in BaF3 cells expressing each type of Epo receptor, demonstrating that the tyrosine residues of the intracellular domain of the Epo receptor are not required for Epo-induced tyrosine phosphorylation of IRS-2. Reprobing the blot with anti-IRS-2 antibodies showed equal loading of each line (Fig. 8, right panel). Epo-induced tyrosine phosphorylation of IRS-2 was significantly higher in cells expressing the Epo receptor devoid of intracellular tyrosine residues. This result was consistently obtained in several experiments and could be related to a higher number of Epo receptors at the cell surface of BaF3 ZERO cells. Indeed, as previously reported, BaF3 ZERO cells express ϳ6000 Epo-binding sites/ cell, whereas BaF3 cells transfected with the wild-type Epo receptor express only ϳ1500 Epo receptors/cell (36). DISCUSSION Expression of IRS-1 and IRS-2 in the hematopoietic system is relatively poorly documented. Both IRS-1 and IRS-2 were reported to be expressed and activated by IL-4 and IL-2 in human T lymphoblasts (13). In contrast, mast cells (18) or the murine myeloid progenitor cell line 32D (10) expressed neither IRS-1 nor IRS-2. Expression of IRS-2, but not IRS-1, was also detected in other IL-3-dependent myeloid progenitor cell lines such as FDCP-1 and FDCP-2 (10), murine macrophages, and murine B and T lymphocytes (18). Here, we studied the expression of these proteins in cells with erythroid and megakaryocytic characteristics. UT-7 cells express markers from different differentiation lineages depending on the stimulatory cytokines. Indeed, it has been shown that erythroid or megakaryocytic differentiation markers are expressed in UT-7 cells stimulated by Epo (49) or by thrombopoietin (50), respectively. TF-1 and HCD57 cells seem to be strictly committed in the erythroid differentiation pathway, whereas Mo7E cells exhibit megakaryocytic characteristics (38,39,41,51). T3Cl2 cells are Friend virus-transformed cells that correspond to erythroid cells blocked at the colony-forming unit erythroid/proerythroblast stage (52). All these cell lines express IRS-2, whereas we did not detect IRS-1 expression in any hematopoietic cell line tested. Thus, our results confirm that hematopoietic cells express IRS-2 rather than IRS-1 and extend this observation to cells of the erythroid lineage.
We observed that Epo stimulation of UT-7 cells induced the rapid tyrosine phosphorylation of IRS-2 since maximal tyrosyl phosphorylation was detected between 1 and 10 min and decreased after this time. This kinetics was slightly different from that reported for growth hormone or prolactin, which maximally induced the tyrosine phosphorylation of IRS-1 and IRS-2 after 10 -20 min (12, 15, 17). The time course of Epo- induced IRS-2 tyrosine phosphorylation in UT-7 cells closely paralleled the kinetics of JAK2 activation in these cells (data not shown) and was superimposable to the activation kinetics of other intracellular signaling pathways such as PI 3-kinase (33) and mitogen-activated protein kinases (28). The rapid IRS-2 phosphorylation and its association with the Epo receptor (see below) strongly suggest that Epo-induced IRS-2 tyrosine phosphorylation is a direct event of Epo receptor activation.
Although UT-7 cells are sensitive to GM-CSF and SCF, we did not detect the tyrosine phosphorylation of IRS-2 in UT-7 cells stimulated with these cytokines. Using another cell line, Welham et al. (18) previously reported the stimulation of IRS-2 tyrosine phosphorylation by IL-3 and GM-CSF. This discrepancy could be due to the low number of high affinity GM-CSF receptors expressed in UT-7 cells. Indeed, we detected only a few hundred high affinity receptors for GM-CSF, whereas these cells express ϳ7000 Epo receptors (49). Moreover, as shown in Fig. 2B, JAK2 activation was also much less efficient using GM-CSF than Epo in these cells. In contrast, the inability of SCF to induce IRS-2 tyrosine phosphorylation is not due to a low number of SCF receptors since these cells express ϳ35,000 receptors for this cytokine. 2 It should be noted that CSF-1, whose receptor shares strong similarity with c-Kit, also does not induce the tyrosine phosphorylation of IRS-2 (18).
Many cytokines whose receptors belong either to subclass 1 or 2 (interferon receptors) have now been reported to induce the tyrosine phosphorylation of IRS-1 and/or IRS-2, suggesting that this relay could be a common signaling pathway for this class of receptors. All these receptors mediate intracellular signaling through the activation of JAK kinases (reviewed in Ref. 53). Moreover, mutated ZERO receptors that essentially conserved few receptor sequences downstream of the region required for JAK2 activation can efficiently mediate Epo-induced IRS-2 activation (Fig. 8). The corresponding region of the growth hormone receptor was previously shown to be responsible for IRS-1 and IRS-2 activation (12,15). Thus, an attractive hypothesis would be that IRS-1 and IRS-2 proteins could directly interact with the JAK kinases, although JAK2 does not contain NPXY motifs (54). However, our results do not sustain this hypothesis. Indeed, we did not detect JAK2 in IRS-2 immunoprecipitates (Fig. 3), and no IRS-2 protein was detected in anti-JAK2 immunoprecipitates (data not shown). In contrast, we observed the constitutive association of IRS-2 with the Epo receptor in both Epo-stimulated and unstimulated UT-7 cells, strongly suggesting the direct association of IRS-2 with the Epo receptor. The Epo receptor is not tyrosine-phosphorylated in unstimulated UT-7 cells, and it does not contain an NPXY PTB domain-binding sequence. Moreover, the tyrosine residues of the Epo receptor intracellular domain are not necessary for Epo-induced IRS-2 tyrosine phosphorylation (Fig. 8). Taken together, these results indicate that the association between IRS-2 and the Epo receptor probably does not involve the IRS-2 PTB domain.
Although the role of IRS-1 and IRS-2 in normal erythropoiesis was never directly addressed, some reports have evidenced a role in erythropoiesis for IGF-1, which mainly uses IRS-1 and IRS-2 for intracellular signaling. Indeed, IGF-1 has been shown to stimulate erythroid colony formation in vitro even in the absence of Epo (55). Moreover, erythroid progenitors from polycythemia vera patients have been shown to be hypersensitive to IGF-1, leading to Epo independence (56). Thus, signaling through IRS-1 and IRS-2 could sustain the proliferation of erythroid progenitors. Our results show that PI 3-kinase seems to be the main target of IRS-2 in Epo-stimulated cells. Previously published data suggest that PI 3-kinase seems to be involved in the control of Epo-induced cell proliferation. Indeed, the PI 3-kinase inhibitor wortmannin inhibits Epo-induced proliferation of Epo receptor-transfected DA3 cells (35) and of UT-7 cells. 3 PI 3-kinase activation by Epo can be achieved using several mechanisms. One of these mechanisms involves the binding of PI 3-kinase to the tyrosine-phosphorylated Epo receptor (31)(32)(33)(34). However, Epo receptors without tyrosine residues in the intracellular domain are able to mediate Epo-induced PI 3-kinase activation, although PI 3-kinase association with the Epo receptor is not detected (35,36), demonstrating the presence of alternate pathways for Epoinduced PI 3-kinase activation. Our results show that IRS-2 could be one of these pathways and could be responsible for Epo-induced PI 3-kinase activation in cells expressing Epo receptors devoid of intracellular tyrosine residues such as BaF3 cells transfected with the ZERO Epo receptor mutant. Interestingly, a region of the Epo receptor conserved in the ZERO Epo receptor mutant and located between the JAK2-binding sequence and amino acid 329 was previously reported to be required for Epo-induced mitogenesis in the context of a truncated (57), but not a full-length (58), Epo receptor. This result suggests that an intracellular relay could be activated by this region of the receptor, but that other signaling pathways activated by Epo receptor sequences located downstream could substitute for this signal. Our results show that PI 3-kinase could be this relay since it could be activated through the C-terminal part of the Epo receptor by direct binding to the Epo receptor and through the region of the Epo receptor close to the transmembrane domain by using IRS-2. Although under experimental conditions these pathways appear to be redundant, it should be kept in mind that normal erythroid progenitors express lower levels of Epo receptors that the cell lines used as experimental models and that they have to respond in vivo to low Epo concentrations. Under these conditions, PI 3-kinase activation by these different pathways could be additive and required to allow a cellular response to a low level of Epo 2 P. Mayeux, unpublished results. 3 S. Gobert and P. Mayeux, unpublished results.
FIG. 8. Tyrosine residues of the intracellular domain of the Epo receptor are not required for Epo-induced IRS-2 tyrosine phosphorylation. BaF3 cells expressing either the wild-type Epo receptor (WT) or the Epo receptor devoid of tyrosine residues in its intracellular domain (ZERO) were serum-and growth factor-deprived for 5 h. Starved cells were then incubated for 2.5 min with 10 units/ml Epo (ϩ) or with vehicle alone (Ϫ). Cleared lysates were immunoprecipitated (IP) with anti-IRS-2 antibodies, and immunoprecipitated proteins were analyzed by Western blotting with anti-phosphotyrosine antibodies (anti-PY) (left panel). The blot was then stripped and reblotted with anti-IRS-2 antibodies to ensure equal loading of each lane. The structures of the wild-type and ZERO Epo receptors are indicated. To construct the ZERO Epo receptor, a stop codon was inserted at the HindIII restriction site (codon 374), and Tyr 343 was mutated to Phe (36). stimulation. One of the signaling relays downstream of PI 3-kinase was recently shown to be the serine/threonine kinase AKT (see Ref. 59 for review). Interestingly, AKT is activated by PI 3,4-bisphosphate, but not by PI 3,4,5-trisphosphate (60,61). One way to produce PI 3,4-bisphosphate is the dephosphorylation of the main product of PI 3-kinase, PI 3,4,5-trisphosphate, which could be performed by the PI-3,4,5-trisphosphate 5-phosphatase, SHIP. Our results show that a 140-kDa protein that most likely corresponds to SHIP also associates with IRS-2. Thus, the same protein complex appears to contain both PI 3-kinase and SHIP, and these associations could increase the efficiency of PI 3,4-bisphosphate production.